HOLOCENE PALEOECOLOGY AND LATER STONE AGE HUNTER-GATHERER
ADAPTATIONS IN THE
SOUTH AFRICAN INTERIOR PLATEAU
A Dissertation Presented to the Graduate Faculty of
Dedman College
Southern Methodist University
in
Partial Fulfillment of the Requirements
for the degree of
Doctor of Philosophy
with a
Major in Anthropology
by
Charles Britt Bousman
(B.S., Southern Methodist University, 1974)
(B.A., Cambridge University, 1976)
(M.A., Cambridge University, 1980)
(M.A., Southern Methodist University, 1987)
May 18, 1991
HOLOCENE PALEOECOLOGY AND LATER STONE AGE HUNTER-GATHERER
ADAPTATIONS
IN THE SOUTH AFRICAN INTERIOR PLATEAU
Approved by:
セZエ@
Bousman, Charles Britt
B.S.,
B.A.,
M.A.,
M.A.,
Southern Methodist University, 1974
Cambridge University, 1976
Cambridge University, 1980
Southern Methodist University, 1987
Holocene Paleoecology and Later Stone Age
Hunter-Gatherer Adaptations in the
South African Interior Plateau
Advisor: Professor C. Garth Sampson
Doctor of Philosophy degree conferred May 18, 1991
Thesis completed February 21, 1991
Excavations at Blydefontein Rockshelter and Meerkat Rockshelter are used to
test models of hunter-gatherer technological organization. Climatic and ecologically
driven models that predict differential use of weapons and tool kits among huntergatherers were constructed from modern hunter-gatherer studies from throughout the
world. Ethnographic, historic and archaeological observations on tools made by stillliving Bushmen from the 19th and 20th century were used to predict the specific
technological changes that would occur under varying climatic circumstances. Local
paleoenvironmental and modern botanic studies are used to predict past huntergatherer behavior through the reconstruction of past climates. Tests of these models
were conducted with Later Stone Age artifacts from the rockshelter excavations.
COPYTRIGHT
BY
C. BRITT BOUSMAN
1991
TABLE OF CONTENTS
LIST OF ILLUSTRATIONS.........................................................................................
vi
LIST OF TABLES .. .. . .. .. ... .. .. .. .. .. .. .. .. . .. .. .. . .. ... .. .. .. .. .. .. .. .. .. . ... ... .. .. .. .. . .. .. .. .. .. .. .. .. .. .. .. .. .. .
Xi v
ACKNOWLEDGEMENTS .. .. .. . .. .. .. .. .. .. .. .. .. .. . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . .. .. .. .. .. .. .. .. .. . x v i i
Chapter
I.
INTRODUCTION .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . .
1
II.
THE MODERN ENVIRONMENT ............ ... .. .. .. .. ...................... .... .....................
The Physical Setting .. .. .... .. .. .. .. .. .... .. ............................... .... .. .. . .. .. .. .. .. .. .
The Botanical Setting .................. .... ................. .... .. .... ........ ..... .... .. .... ...
Animals.................................................................................................
European Impacts .................................................................................
Summary ..............................................................................................
8
8
20
28
33
39
Ill.
A REGIONAL MODEL OF LATE QUATERNARY AND RECENT CLIMATIC
CI-W\K3E .. .. .. .. .. .. .. .. .... .. .. .. .. ... .. .... .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . .. .. .. .... .. .. .. .. .... .. .. .. .. .
IV.
V.
The General Circulation Systems .........................................................
Circulation Systems and Rainfall Patterns in Southern Africa .... .......
The Computer Simulations....................................................................
Conclusions .. .. .. .. .. .. .. .. .... .. .. .. . .. .. .. .... ...... .. .. .. .. .. .. .. . . .. .. ... . . .. .. .. .. .. .. .. .. ... . .. .
41
41
45
47
58
CONSTRUCTINGAREGIONALMODELOFVEGETATIONCHANGE....................
Analysis of the Modern Plant Mosaic.....................................................
A Model of Botanical Change for the Eastern Cape .................. .... ...... ....
61
61
77
A TEST OF THE BOTANICAL MODEL .. .. .. .. .... .... .. .. .. .. ............................... ...... 8 2
Fossil Pollen Sites ................................................................................ 8 2
Pollen Analysis .. .. .. . .. .. .. .. .. .......................... ..... .. .. .. .. .. ... .. .. .. .. .. .. ............ 1 0 6
The Test Results ...... .. .. ............................................ .... .. .. .. ........ ............ 1 1 3
VI.
STABLE CARBON ISOTOPE ANALYSIS...........................................................
Principles and Techniques ....................................................................
Blydefontein Soil o13C and Pollen Spectra Compared .............. ............
Stable Carbon Isotopes in Ostrich Eggshell ..........................................
Conclusions .... ........ .... ............................. ....................... .......................
12 0
12 0
12 4
13 0
13 5
VI.
RECONSTRUCTION OF THE BLYDEFONTEIN PALEOCLIMATE ........................
Palynological Estimates of Climate.......................................................
Comparison of Pollen Rainfall Estimates ............ .................................
Pollen Rainfall Estimates Compared to Historic Precipitation Record
Comparison Between Pollen Rainfall Estimates and the o13c Sequence
A Test of the Climatic Model. .................................................................
Conclusions .. .. .. .. .. .. .. .. .. .. .. .. .. .................................................................
13 6
13 6
145
14 9
1 51
1 53
1 55
v
VIII.
THE ROCKSHELTERS: STRATIGRAPHY, CHRONOLOGY, AND SEDIMENTS ....
Blydefontein Rockshelter Excavations ......................... .. .... .......... .. ......
Meerkat Rockshelter Excavations.........................................................
Summary ...............................................................................................
1 56
1 57
175
183
CHRONOLOGICAL MARKERS OF THE INTERIOR WILTON SEQUENCE ....... ... ..
The Artifact Samples ... .... .. .. .. .. .. .. .... .. ... .. .. .. .. .... .. .. .... .. .... .. .. .. .. ..... .... .. .. .
The Artifact Analysis ... .. .. .. .. .. .. .. ........ ... .. .. .. .. .. .. .. .. .......... .. .... .. . ........ .. .. .
Conclusions .. .. .. .. .. .. .. .. .. .. .. .. .. ... .. ...... .. .. .. .. .. .. ...... .. .... ......... .. .. .. .. ........ .. ...
185
186
187
205
X.
EXPLANATORY MODELS OF LATER STONE AGE LITHIC VARIABILITY ...........
Models for Explaining Assemblage Composition .. .... .. .. .. ..... ...... .. .. ........
Models for Explaining Artifact Style....................................................
Conclusions ...........................................................................................
21 3
21 3
226
23 0
XI.
ECOLOGICAL AND BEHAVIORAL BASIS FOR A NEW MODEL ... .. .. .. .. ..... .... ... .. .
Design and Structure ............................................................................
Responses to Risk Posed by Resource Variability ....... ...... ..... .. .. ...... .. ..
Food Selection as a Determinant of Mobility Pattern ...... .. ......... ...... .. ..
The Determinants of Range Size............................................................
Global Trends in Hunter-Gatherer Risk Reduction ......... ....... .......... .. ..
23 1
23 2
233
241
246
2 53
X II.
THE ROLE OF TECHNOLOGY AMONG FORAGERS AND COLLECTORS .. .. .. .. . .... ..
Responding to Risk with Technology .....................................................
Tool Design Decisions Among Foragers and Collectors ........ .. ....... .. ......
Kinds of Curation Used by Foragers and Collectors..............................
Global Trends in Hunter-Gatherer Technology ....................................
Distribution of Assemblages (Settlement Pattern) ..............................
256
2 56
2 59
260
272
276
X Ill.
BLYDEFONTEIN BASIN: A TEST OF THE NEW MODEL .. .... ........ ..... .. .. .. ........
Introduction ..........................................................................................
Blydefontein's Place in the Ecological Model ..... .. .. .. .. .. .. ... .... .. .... .. .. ......
Blydefontein's Range and Holocene Population Density........................
Blydefontein Mobility in the Ecological Model. .....................................
Stone Tool Design Decisions Among Foragers and Collectors ................
Changes in Stone Tool Design Decisions at Blydefontein .. .. ...................
Isolation of the Variables that Reflect Curation Decisions...................
Comparisons between Tool Design and Paleoenvironment Change........
Conclusions ........................................................................................,
278
278
SUMMARY AND CONCLUSIONS .....................................................................
Holocene Paleoecology .. .. .. .. .. .. .. .. .. .. .. .. ..... .. .. .. .. .. .. .. .. .. .. .. .... .. .. .. ... .. .. .. .. .. .
The Model .............................................................. ....... ...... ...................
The Test .................................................................................................
Blydefontein in the Prehistory of the Interior Plateau .......................
31 6
3 16
3 19
320
325
IX.
XIV.
277
283
288
292
294
309
3 12
31 4
APPENDIX
1.
SEASONAL RAINFALL AND TEMPERATURE CORRELATIONS......................... 3 3 0
vi
APPENDIX
2.
BLYDEFONTEIN ROCKSHELTER AND MEERKAT ROCKSHELTER ARTIFACT
TABLES ....................................................................................................... 33 6
BIBLIOGRAPHY......................................................................................................... 3 4 5
vii
LIST OF ILLUSTRATIONS
Figure
1.
Page
Map of Blydefontein and Meerkat Rockshelters and paleoenvironmental
sites in the Kikvorsberg Range.......................................................
9
Mean monthly temperatures in oc with one standard deviation bars at
Grootfontein over an average twenty month period........................
18
Mean rainfall and percent of significant rainfalls by month over a
20 month period ...........................................................................
20
4.
Map of Veld Types within 140 km radius of Blydefontein Shelter..........
25
5.
Raux and Vorster's (1983) model of vegetation change in the eastern
Cape due to overgrazing ............................................................... 3 7
6.
General north-south circulation systems in Southern Hemisphere;
a: circulation cells and summer convergence zone positions;
b: monthly movements of oceanic high pressure cells....................
44
7.
Walker Circulation in High and Low Phases over Southern Hemisphere...
45
8.
Wet and dry climatic conditions and zonal circulation in southern Africa 4 8
9.
Wet and dry climatic conditions and meridonal in southern Africa ........... 4 9
2.
3.
Map ofworld computer grids used in COHMAP simulations...................
51
11 . COHMAP Project high resolution NCAR-CCM model simulated annual
temperature and annual rainfall differences from modern for
southern Africa............................................................................
52
10.
12.
13.
COHMAP Project high resolution NCAR-CCM model simulated annual,
winter and summer temperature differences from modern for
southern Africa ...........................................................................
52
COHMAP Project high resolution NCAR-CCM model simulated annual,
winter and summer rainfall differences from modern for
southern Africa ............................................................................
53
viii
14.
Eastern Cape rainfall variability from 19th Century historic records.... 55
15.
Five year moving averages of Aliwal North and Grapevale rainfall........
56
16.
Measured and observed rainfall estimates for the eastern Cape............
57
1 7.
Yearly rainfall at Grapevale from 1921 to 1984 .... ............ .................
58
1 8.
Relative frequency of Compqsitae plotted against relative frequency of
Gramineae for eleven Veld Types..................................................
64
19.
Cluster analysis of Roux-Biom botanical surveys.................................
69
20.
Plot of mean annual rainfall (mm) and degree of longitude for region
surrounding Blydefontein ..............................................................
73
21.
Seasonal growth cycles for major plant groups in the northeastern Cape 76
22.
Blydefontein Section stratigraphy........................................................
85
23.
Blydefontein Section pollen diagram....................................................
86
24.
BFS radiocarbon dates by sample depth................................................
89
25.
Blydefontein Stream Mouth geological profile.......................................
90
26.
Linear regression of BSM radiocarbon dates by depth..........................
91
27.
Diatom diagram for BSM.............. ..................... ... ............. .......... .......... 91
28.
Pollen diagram for BSM.......................................................................
92
29.
Geological profile for Channel 2...........................................................
93
30.
Pollen diagram for Channel 2................................ ................. ...... ........
94
31 .
Diatom diagram for Channel 2............................................... ........... ....
94
32.
Geological profile for USP ..................................................................... 95
33.
USP radiocarbon dates by depth...........................................................
96
34.
Pollen diagram for USP........................................................................
96
35.
Diatom diagram for USP......................................................................
97
ix
36.
Geological profile from Hughdale Section.............................................
37.
Linear regression of HDS radiocarbon dates by depth........................... 100
38.
Pollen profile from Hughdale Section .................................................... 101
39.
Linear regression Meerkat hyrax midden radiocarbon dates ................. 104
40.
Pollen diagram from Meerkat hyrax dung midden ................................. 105
41.
Curvilinear regression Oppermanskop hyrax midden radiocarbon dates 106
42.
Oppermanskop hyrax dung midden pollen diagram ................................. 107
43.
Graph of Gramineae versus Compositae and Artemisia relative
frequencies in the Older and Younger Fills at BFS ...................... ..... 111
44.
Graph of composite/grass relative frequencies for all Holocene pollen
spectra .. .. .. .. .. .. .... .. .. .. .. . . . .. .. .. . .. .. .. .... .. ... .. .. .. .. .. . .... .. . .... .. . .. . .. .. .... .. . 114
45.
Temporal distribution of paleo-plant communities at geological sites
and hyrax middens .......... .................. ........... ............... .... ........ ..... 118
46.
a13C ratios and relative frequency of grass pollen from matched
samples . . . .. . . . . . . .. . . . . . . .. . . .. . . .. . .. . . .. . .. . . . .. . .. . ... . .. . . .... . . . . . .. . .. . . . . . . . . . . .. . . . . . 125
4 7.
Curvilinear correlations of Cs plant estimates based on a13c ratios and
relative frequency of grass pollen for matched samples................. 128
48.
Histogram of ostrich eggshell a13c values .......................................... 132
49.
Individual and mean ostrich eggshell (OES) a13c values grouped by
excavation unit............................................................................. 133
50.
Corrected mean ostrich eggshell (OES) and radiocarbon dated sediment
a1sc ............................................................................................ 134
51.
Rainfall estimates from Oppermanskop and Meerkat Hyrax Middens ..... 146
52.
Combined rainfall estimates from hyrax middens and pond deposits ..... 148
53.
5-yr moving averages for Vogel's Eastern Cape and Grapevale rainfall
records, and Meerkat hyrax midden rainfall estimates .................. 150
54.
Pollen rainfall estimates and average ostrich eggshell a13c values ....... 152
X
99
55.
Comparison of palynologically estimated rainfall record and COHMAP
simulated rainfall record .. .............. ...... .... .... .......................... ...... 154
56.
Map of Blydefontein Rockshelter ......................................................... 157
57.
Map of current and previous excavation units at Blydefontein
Rockshelter ...... ........ ... . ......... ..... . . ..... . .. ..... .............. ...... . . .. . ... ..... .. 158
58.
Stratigraphy at Blydefontein Rockshelter .......... .... ...................... ....... 161
59.
Sediment texture analysis of layer SD, Blydefontein Rockshelter ........ 164
60.
Sediment texture analysis of layer HG, Blydefontein Rockshelter......... 164
61 .
Sediment texture analysis of layer CPB, Blydefontein Rockshelter....... 165
62.
Sediment texture analysis of layer TS, Blydefontein Rockshelter......... 165
63.
Sediment texture analysis of modern swallows nest, Blydefontein
Rockshelter ... . .. ....... ....... .............. ......... ............................... ... ... . 166
64.
Sediment texture analysis of layer GAC, Blydefontein Rockshelter ...... 166
65.
Sediment texture analysis of Upper TG, Blydefontein Rockshelter........ 167
66.
Sediment texture analysis of Lower TG, Blydefontein Rockshelter ....... 167
67.
Ostrich eggshell d13C values from Layers TG and CAC ........................ 168
68.
Sediment texture analysis of layer CAC, Blydefontein Rockshelter ...... 168
69.
Sediment texture analysis of Upper CY, Blydefontein Rockshelter ........ 169
70.
Sediment texture analysis of Middle CY, Blydefontein
Rockshelter .................................................................................. 169
71.
Sediment texture analysis of Brown 2 in CY, Blydefontein Rockshelter 170
72.
Sediment texture analysis of Lower CY,
73.
Analytical Units at Blydefontein Rockshelter ....................................... 176
7 4.
Combined Analytical Units at Blydefontein Rockshelter.. ........ ...... ..... .. . 177
75.
Map of Meerkat Rockshelter ... .. ..... ........ ........ ........... ................. ......... 178
xi
Blydefontein Rockshelter ...... 170
76.
Stratigraphy at Meerkat Rockshelter ... ......... ............. ..... ..... .. ... . ........ 179
77.
Sediment texture analysis of layer SD, Meerkat Shelter...................... 181
78.
Sediment texture analysis of layer UB, Meerkat Shelter .............. .... .. . 181
79.
Sediment texture analysis of layer Gray, Meerkat Shelter.................. 182
80.
Sediment texture analysis of layer YS, Meerkat Shelter ..................... 182
81 .
Backed bladelets and projectile points from Blydefontein and
Meerkat Rockshelters ................................................................... 189
82.
Mean endscraper lengths and standard error bars for Blydefontein
Rockshelter Blocks C-D ................................................................ 198
83.
Mean endscraper lengths and standard error bars for Meerkat
Rockshelter AUs ........................................................................... 198
84.
Histogram of Blydefontein and Meerkat ostrich eggshell bead maximum
diameters .................................................................................... 200
85.
Distribution of individual ostrich eggshell measurements by analytical
unit at Meerkat Rockshelter .......................................................... 200
86.
Mean OES bead diameters from Glen Elliot Shelter ............................... 201
87.
Ethnographic projectile point with straight backed bladelets mounted in
mastic .......................................................................................... 206
88.
Hypothetical mounting and function of crescents as projectiles ............ 207
89.
Mean endscraper lengths and standard error bars for Blydefontein
Rockshelter Combined AUs ............................................................ 208
90.
Linear regression of mean end scraper lengths for Combined AUs ........ 209
91.
Ecological and behavioral model's four components.............................. 234
92.
Changes in average number of forager camp moves with changing plant
primary production ...................................................................... 238
93.
Scatterplot of effective temperature and percent of hunted resources
contributed to diet........................................................................ 239
94.
Regression between annual precipitation and large herbivore biomass.. 240
xii
95.
Scatterplot of annual rainfall and percent of hunted resources in diet... 240
96.
Scatterplot of overall mobility (average number of camp moves *
average length of special task trip) and percent of hunting............. 242
9 7.
Scatterplot of percent of fish in diet and winter camp occupation
length ........................................................................................... 243
98.
Curvilinear regression between hunter-gatherer range size and
primary productivity.................................................................... 24 7
99.
Scatterplot of territory area (km2) and population density as
measured by persons per km2 ............ ......... ... . .... ......... ... . .. . . ........ 248
1 00.
Linear regression between percent contribution of hunting to diet and
natural logarithm of total area exploited ....................................... 250
1 01 . Scatterplot of total exploited area (km2) and total distance moved
between forager residential camps ........... .............. ...... .. ... .. ........ 250
102. Scatterplot of total exploited area (km2) by number of forager
residential camp moves . .. . . . . .. . . . .. . . . . .. .. .. . . . .. . . . ... . . . . . . . . .. . . . . . . . . .. . . . . . . . . . 251
1 03.
Distances between hxaro partners and spouse births ........................... 252
1 04
Diagram of model variables on world scale.......................................... 255
105.
Ecological and behavioral model with technological component ............. 257
106. Tri-polar plot of a hypothetical tool's design goals .............................. 261
1 07.
Histogram of Dobe !Kung and lngalik artifact and facility active
use-lives ..................................................................................... 263
1 08.
Linear regression between natural logarithm of use-life and
manufacturing time for Dobe !Kung artifacts ................................. 266
1 09.
Linear regressions of lngalik tool manufacturing time and active
use-life ........................................................................................ 266
110.
Linear regressions between manufacturing time and use-life for Dobe
!Kung and lngalik tools .................................................................. 267
111 . Linear regression between total maintenance time and use-life for
Dobe !Kung artifacts ..................................................................... 268
xiii
i 1 2.
Linear regressions between total manufacturing and repair time by
use-life for the !Kung tools and manufacturing time by use-life
for lngalik tools ............................................................................ 269
113. Assemblage composition and settlement pattern model for hypothetical
bands ........................................................................................... 273
i 1 4.
Modern and Pleistocene estimates of percent of hunted resources in
diet at Blydefontein, based on temperature .................................... 279
i 1 5.
Modern estimate of hunted resources in diet at Blydefontein, base
on rainfall . ............ ..... ..... . ........................................ ............ .... ... . 279
116. Changes in the amount of hunted resources in the Blydefontein diet,
based on estimated changes in annual rainfall .. .. .. .... .. .. .. .. .. .. .. .. .... ... 281
117. Hunter-gatherer range size estimated from amount of hunted
resources in diet ......... ..... .... ... .......... ...... .. .............. ........... .......... 284
118. Changes in range size predicted by the ecological model ...................... 284
119.
Histogram of recorded San ranges ...................................... .... .. .. .. ...... 285
120.
Percent of agate-jasper use through Combined Analysis Units............. 288
i 21 . Scatterplot of percent of hunted foods in diet by average number of
residential camp moves ................................................................ 290
122. Estimated changes in the overall mobility of Blydefontein occupants .... 291
1 23. Mean endscraper lengths for analytical units at Blydefontein
Rockshelter . . .... ............... ....... .................................... .... .. . .. . ....... 298
1 24. Mean lengths of hornfels endscrapers for analytical units at
Blydefontein Rockshelter .............................................................. 299
125.
Percent of complete backed bladelet tools by Combined Analytical Unit
at Blydefontein Rockshelter .......................................................... 302
1 26
Percent of blade lets among lithic debris by CAU .... .. ........ .......... ......... 304
127.
Percent of blade let cores among lithic debris by CAU .......................... 305
128.
Linear regression between straight backed bladelets and bladelets ....... 306
129. Mean cortex rank for flakes and bladelets in Combined Analytical Units 309
xiv
130.
Factor 1 and Factor 2 Scores for assemblages from Combined
Analytical Units ... .. ... .. .. . . . . .... . ..... . .. . ............ .......... ........ .... . . ... .. . . . . 312
1 31 . Ostrich eggshell stable isotope curve and agate-jasper percents for
individual Analytical Units at Blydefontein Rockshelter .................. 313
13 2.
Ostrich eggshell stable isotope curve and end scraper mean lengths for
individual Analytical Units at Blydefontein Rockshelter.......... ......... 314
133. Plotting of factor II scores for each year from the Winter-Spring factor
analysis, and SST anomalies from Peruvian coast ...... .. .... .. ........... 334
XV
LIST OF TABLES
Table
Page
1.
Discriminant analysis scores of individual Roux-Biom botanical surveys
by Veld Type ........ ..... ..... ..... . . . . . .. . ........... ...... ... ......... .................... 6 6
2.
Mean and standard deviations of Gramineae, Compositae,
Chenopodiaceae/Amaranthaceae (Cheno/Ams), and Aizoaceae/
Ruschia relative frequencies for Roux-Biom botanical survey
clusters.......................................................................................
70
Stepwise regressions with relative frequencies of Gramineae and
Compositae as dependent variables and elevation, degree of
longitude, and degrees south as independent variables....................
74
4.
List of radiocarbon dates from geological sites in Blydefontein Basin...
84
5.
USP molluscan fauna counts (%) by sample.........................................
98
6.
Radiocarbon dates from hyrax middens in Blydefontein Basin .............. 103
7.
Middelburg pollen rain relative frequencies, Middelburg plant
community composition, and correction coefficients...................... 109
8.
Comparison of Compositae/Gramineae and Artemisia/Gramineae linear
regression slopes, intercepts and correlation coefficients between
Younger and Older Fills at BFS .. .. .. .. .. .. .. .. . .. .. .. .. .. .. .. .... .. .. .. .. .. .. .. .. .. . 11 1
9.
Discriminant analysis canonical scores for transformed and corrected
pollen samples . . . . . . .. .. .. .... . .. . .. .. .. .. .. .. . .. .. .. .. ... .. . .. . .. .. .. .... .. .. .. .. .. . . 1 15
i 0.
BFS Younger Fills relative frequencies of selected pollen taxa and
estimated climatic parameters ..................................................... 140
11 .
BSM relative frequencies of selected pollen taxa and estimated
climatic parameters .. .... . . .. .. . . .. .. .. . .. .. .. .. . .. .. . . . . .... .. . .. .. .. .. .... .. .. .. . .. .. . 141
i 2.
CH2 Younger Fills relative frequencies of selected pollen taxa and
estimated climatic parameters .. .................... .......................... ..... 141
13.
USP relative frequencies of selected pollen taxa and estimated
climatic parameters . . ..... . ..... ... ......... .... . . .......... ... .. . ... ................... 142
3.
xvi
14.
HDS relative frequencies of selected pollen taxa and estimated
climatic parameters .. . ... . ...................... ...... .............. ... ...... ... ...... .. 143
15.
Meerkat Hyrax Midden relative frequencies of selected pollen taxa and
estimated climatic parameters ........ ....... .......... .... .............. .......... 144
1 6.
Oppermanskop Hyrax Midden relative frequencies of selected pollen
taxa andestimated climatic parameters......................................... 145
17.
Radiocarbon dates from Blydefontein Rockshelter ... . . .. . . . . . .. . .. . .. . . . . . .. .. .. 171
18.
Phosphate fractions (0 /oo) of selected sediment samples from
Blydefontein Rockshelter............................................................... 17 4
19.
Radiocarbon dates from Meerkat Rockshelter ..... .......... .... ..... ... ... ...... .. 183
20.
Backed tools in Combined Analytical Units at Blydefontein Rockshelter
21 .
Percentages of backed tools in Combined Analytical Units Blydefontein
Rockshelter . . .. . . . . .. . . . . . .. . . . . . .. . . . . .. . . . . . .. . .. . . .. . . . . .. . .. . . .. . . . . . .. . . . . . .. . . . . .. . . . . . 191
22.
Distribution of backed tools and projectile points in analytical units
from Meerkat Rockshelter...... ......... ........ ............ ................ .... ...... 1 94
23.
Ceramic sherd counts at Blydefontein Rockshelter, Block B ........ .. . ...... 203
24.
Ceramic sherd counts at Blydefontein Rockshelter, Blocks C-D ............. 203
25.
Ceramic sherd counts at Blydefontein Rockshelter, Block B ................. 203
26.
Ceramic sherd counts at Blydefontein Rockshelter, Blocks C-D............ 204
27.
Number of groups tabulated by categories of nomadism and
environment ................................................................................. 237
28.
Resource temporal and spatial patterns and optimal associated huntergatherer mobility patterns ........... ......... ......................... .... .......... 244
29.
Average use-life, and production and maintenance costs of !Kung
artifacts by material ....... ..... ........ ......... ................... .... .. .... ... ....... 264
30.
Average use-life, and production costs of hand-held lngalik artifacts
by material .............. ....... ............. ........ ........ ........... ....... .... .. . ....... 264
xvii
191
31.
Linear regression statistics for !Kung tool use-life by manufacture
costs plus repair costs, and lngalik tool use-life by manufacture
costs ............................................................................................ 269
32.
Extractive and maintenance tool mean use-life (years) for foragers
(!Kung) and collectors (lngalik) .................................................... 270
33.
Backed artifact discards, broken fragments and complete tools at
Blydefontein Rockshelter.. ..... ......... ...... . ... . ... ............. ..... ..... ......... 302
34.
Factor loadings and proportion of artifact type variance
(communality) explained by factor analysis .... .............. .... .... .... .... 311
35.
The Winter-Spring Factor Analysis ..................................................... 332
36.
The Spring-Summer Factor Analysis ................................................... 333
37.
Correlations between factor scores and climatic variables.................. 334
38.
Blydefontein Rockshelter, Block B Artifacts ....................................... 337
39.
Blydefontein Rockshelter, Blocks C-D Artifacts .................................. 339
40.
Meerkat Rockshelter Artifacts........................................................... 343
xviii
ACKNOWLEDGEMENTS
I would like to dedicate this work to two people: my daughter, Suzanna Blythe
Bousman and my father, Kurth Pennington Bousman, Sr. One came and the other
passed during the course of this dissertation, and it is fitting that they both share this
dedication.
Most importantly, my wife, April Goldsmith, supported me through the project.
Without her help and cajoling the project would have taken years longer. She made
many sacrifices through the years it took to complete this study. My mother, Peggy
Bousman, provided no less encouragement and aid through the years. She sustained and
inspired me from my undergraduate years to this conclusion. My brother, Kurth
Bousman, Jr. facilitated the final paper work.
Grants from the National Science Foundation, Wenner-Gren Foundation, L. S. B.
Leakey Foundation, the Institute for the Study of Earth and Man at Southern Methodist
University, and Southern Methodist University's Graduate Student Assembly funded
this research.
I thank them for their financial support.
Dou Lessing and his family, the owners of Blydefontein, graciously allowed the
archaeological and geological work on their farm, and assisted the field project in
many and numerous ways. Few recent visitors to the Blydefontein house, a hay barn
for the last twenty or more years, would call it a home, but I did and happily so for
eight months. Without the support of the Lessing family no work could have occurred.
I am deeply indebted to them.
Dr. C. Garth Sampson, director of the Zeekoe Valley Project and chairman of my
dissertation committee, assisted at every level and at every stage of this
xix
research project.
My appreciation of his invitation to join the Zeekoe Valley
Project is matched by my appreciation of the academic latitude, freedom and
encouragement he offered during this study. His attitude that graduate students
should select and develop their own research problems, obtain their own funding,
and seek outside collaborators for multidisciplinary studies, is to be highly
commended and should serve as a model for other graduate programs.
I wish to thank Dr. Fred Wendorf, also on my dissertation committee, for
faithfully reading drafts of my chapters, making many useful comments, and most
importantly for always encouraging graduate students, including myself, to publish
our research. Dr. David Meltzer has cheerfully and ably advised a graduate student
doing research in a geographic area which he has no first hand experience. His
criticisms of my dissertation were not always pleasant, but they were always
thoughtful and usually right. I thank Dr. Herold Hietala for participating on my
committee, and for providing emergency statistical advice over my years as a
graduate student, even if unheeded, at all times of day and night.
The members of the Hantam Tennis Club centered at Grapevale were extremely
helpful and provided much needed company during the many long months of
excavation at Blydefontein. Norman and Clare Biggs shared in their knowledge of
the region, supplied the Grapevale rainfall records, and always were happy to have
the company of a rather homesick archaeologist. Peter and Clare Barnes-Webb, the
owners of Hughdale, allowed a paleoenvironmental study of various sections on
their farm, and supplied a beautiful house for the crew during March of 1987.
Dr. P. van Rensberg, his late wife, Di, and their daughter, Clare, opened their
house to me for the entire length of the field project.
I will never forget our first
meeting; I had a stomach virus, and went to the Noupoort Doctor for treatment.
That Doctor happen to be Phillip. I walked into his office, and began to describe my
XX
symptoms, when he said "Hey wait a minute, you're not from around here." He then
made me give him the usual 20 minute spiel on the Blydefontein Project. All the
while I fought off throwing up in his office. The van Rensbergs' hospitality and
friendship is well remembered and missed.
As the Blydefontein Project is really a continuation and offshoot of the Zeekoe
Valley Project, many people in the Zeekoe Valley have assisted toward the
completion of this project often in ways they probably did not realize at the time.
wish to extend my thanks and friendship to all the families in the Zeekoe Valley, and
especially to Mr. & Mrs. Bruce Maskell, Mr. & Mrs. Glanton Maskell, Mr. & Mrs
Gerald Naude, Mr. & Mrs. Quintin Naude, Mr. & Mrs. Tinus Naude, Mr. & Mrs. Fred
Rubidge, Mr. & Mrs. Neil Sheard, Mr. & Mrs. Peter Watermeyer,
Mr. & Mrs.
George van Schalkwyk, Mr. and Mrs. Neville Vimpany, and Mr. Hugo Vimpany.
Dr. Louis Scott (pollen), Dr. Tim Partridge (geology), Dr. John Vogel (geological
sites radiocarbon dating and stable carbon isotopes), Dr. Sarah Metcalfe (diatoms),
Dr. Richard Klein and Dr. Kathy Cruz-Uribe (archaeological sites fauna), Dr. Nick
Shackleton (ostrich eggshell stable carbon and oxygen isotopes), Mr. James Brink
(geological sites fauna), Dr. Margaret Avery (microfauna), Dr. Maitland Seaman
(molluscan fauna), Dr. Herbert Haas (archaeological sites radiocarbon dating), Dr.
Robert Eidt and Dr. Robert Brinkmann (soils) have proved to be extremely able
collaborators. Their willingness to take on the various tasks and the quality of their
research results have made the overall project a success.
Dr. Janette Deacon, has more than any local archaeologist, sought the Southern
Methodist University team out, involved us in archaeology in southern Africa, and
on occasions too numerous to mention, provided published and unpublished data,
acted as a stable backboard for ideas, and happily shared information throughout the
entire project.
I greatly appreciate her willingness to interact as both a
xxi
professional archaeologist and a friend. Drs. Hilary Deacon and John Parkington, as
well, have willingly shared their ideas and views on the prehistory of southern
Africa, and Dr. Hilary Deacon graciously allowed me to spend a short break at his
Klasies River Mouth excavations. Also through the years Mr. Mike Taylor, Dr.
Mannie Opperman, Ms. Madelon Tusenius, Dr. Tom Huffman, Mr. Anton Scholtz, Dr.
Aron Mazel, Dr. Simon Hall, Dr. Francis Thackeray, Dr. Anne Thackeray, Mr. Johan
Binneman and Mr. Peter Beaumont have graciously discussed their own research,
and it is exactly this type of interaction that has made conducting research in a
distant country so rewarding and fulfilling.
Dr. Piet Roux, now retired Director of Grootfontein Agricultural College,
Middelburg, CP, freely shared in his knowledge of the eastern Karoo, and opened my
eyes to the great potential the modern environment has for gaining an understanding
of the past environment. He supplied the extremely valuable modern botanical
surveys, temperature and rainfall records from Grootfontein, and plant growth
information used in Chapters I and Ill. He also provided a transit for site mapping.
Mr. Tim Hart assisted in the first months of excavation and proved extremely
valuable in establishing an orderly progression of work;
his services as interpreter
between me and the crew shall never be forgotten, however it was his departure
that forced me to learn Afrikaans, thus enabling me to talk to my crew and gain an
immeasurable amount of satisfaction from that interaction.
The excavation crew,
especially Jan Meyer, Koppie Meyer and Popo Meyer, were extremely helpful and
sympathetic, although not always interested in the work.
Nonetheless, the crew
labored for many hot and cold hours, and found the Megalotragus bones in the Older
Fills and the burnt reeds in the upper soil at Blydefontein Section. I hope the future
of South Africa will include a safe, secure and respectable place for these people
and others of their calibre.
xxii
Southern Methodist University graduate students Ms. Chris Christopher, Mr.
Tim Dalbey, Mr. Chris Hill, Mr. Les Peters and Dr. Joe Saunders have read drafts of
grant proposals, patiently listened to long discussions, and offered an untold number
of suggestions and helpful comments that I have shamelessly included in this
dissertation.
Originally Les Peters intended to excavate Blydefontein Rockshelter,
and I thank him for abandoning the project. Also I thank Dr. Bonnie Fine Jacobs,
Shuler Museum of Paleontology at Southern Methodist University for reviewing the
environmental and climatic chapters, and making a number of insightful
observations and recommendations. Mrs. Beatrix Sampson helped a great deal in
organizing the logistical aspects of the project in the field and at home, and her help
through the years cannot be measured.
At Prewitt and Associates, Inc., Austin, Texas, Mr. Elton Prewitt, Mr. Ross
Fields and the remaining staff have provided mental and physical support and
supplies, but especially Mr. Steve Tomka has provided the academic interaction that
made the development of the final portion of this dissertation possible. Often, when
overnight I had "discovered" a new relationship between modern hunter-gatherers
and their artifacts, I found the next morning that Steve had realized months or
years before that this relationship was false or problematic. Ms. Sandra Hannum at
Prewitt and Associates, Inc. drafted a number of the maps and figures in a
herculean effort in the final days of preparing the manuscript, and her assistance is
gratefully acknowledged. I also want to express by gratitude to Dr. Steven L. Kuhn,
University of New Mexico, for providing reprints of important research he has
conducted.
Finally, I would like to thank Mars, Inc. for making M&M's candy. They
provided the desperately needed sugar rush on those last few late nights that
xxiii
allowed me to finish (I'm sure my dentist, as soon as I can find one, wishes to thank
them too).
xxiv
CHAPTER I
INTRODUCTION
This study has three objectives. The first is to present a case study of
Holocene paleoenvironmental changes in the heartland of the interior plateau of
southern Africa. The second is to examine world-wide processes that contribute to
the formation, composition, and appearance of hunter-gatherer material residues.
The third goal is met by combining both investigations to show how a series of Later
Stone Age hunter-gatherers adapted their material culture to the changing
circumstances of one harsh and demanding habitat in the interior plateau.
The central thesis of this research is that changes in population density
influence all aspects of Later Stone Age organization and behavior. This approach
was prompted by the suggestion of Janette Deacon (1974: 5-9) that Later Stone
Age populations were not stable, and that those in the interior plateau had suffered
a major decline to minimal levels in the early to mid-Holocene. Her hypothesis was
based on an apparent absence of radiocarbon dates from Later Stone Age sites in the
interior plateau, in sharp contrast to coastal and Transvaal Later Stone Age sites
that evidently span the entire Holocene. Although she proposed (ibid: 8) the
purported change in population density could be linked to a deterioration in the
interior plateau environment, no direct evidence for this was available.
Furthermore, J. Deacon's model was based on a small sample of dates from
three rockshelters in the middle Orange River valley, plus a few outlying shelters.
The reliability of the dates was further clouded because the original samples were
widely scattered charcoal flecks gathered from thin arbitrary spits of homogeneous
1
2
unstratified deposit. Virtually all the radiocarbon dates could be challenged on
contextual grounds, therefore. By 1979, when we began the Zeekoe Valley
Archaeological Project in a small tributary adjacent to the middle Orange River, we
were ready to dismiss all but one of the previously acquired radiocarbon dates and
start anew.
Thus a search was on for a new rockshelter with visible stratigraphy and
charcoal-bearing features from sound contexts. This quest became more urgent as our
foot survey began to pile up an inventory of some 5,000 surface lithic scatters
attributable to those very levels whose radiocarbon dates we had decided to reject! All
these surface sites, with unmistakable affinities to various phases of the Interior
Wilton Industry, could never be cross-dated unless we found the right rockshelter
sequence to use as a chronological standard. Without reliable cross-dating, their great
potential for studying settlement pattern changes could never be realized. After 15
months of searching some 5,000 square kilometers, no suitable rockshelter was
discovered that could crack the Interior Wilton chronology. The many we discovered
all shared the same shortcomings as those in the Orange River valley: thin, leached
deposits without visible stratigraphy or features, all poor in organic materials but
stuffed full of microlithic stone tools and debris. We were forced to leave the strict
confines of the valley, and turned to a nearby but superior alternative.
Blydefontein Rockshelter, located only a few kilometers east of the Zeekoe River
basin had been tested by Garth Sampson in 1967, and its rearmost, best protected
deposits had exactly what we needed: a clear stratified sequence of charcoal-packed
hearths and shelter fills with highly visible microstratigraphy.
loaded with
Even though it was
lithic materials and relatively well preserved fauna, Blydefontein
Rockshelter did have two apparent shortcomings: the last several hundred years was
missing from the sequence, and apparently it had no Early Wilton assemblage at its
3
base (Sampson 1974: 325). Thus the site offered only a partial test of J. Deacon's
"mid-Holocene abandonment" model. Nevertheless we knew that interlocking roof-fall
had prevented Sampson from exploring the lower deposits that might reveal the needed
Early Wilton assemblages below, so it was worth the risk, and plans were laid to reexcavated the shelter.
Upon arrival to Blydefontein in early 1985 the landowner, Mr. Dou Lessing, told
us that he had discovered a second, smaller rockshelter upstream and in the opposite
cliff face from Blydefontein Rockshelter. This was named Meerkat Rockshelter after a
decayed and desiccated meerkat carcass found on its floor. During preliminary test
excavations here, it became clear that this would provide a detailed view of the last
1000 years or so, including the later segment missing from Blydefontein Rockshelter.
But the biggest surprise was the discovery of deep stratified alluvial deposits in front
of Blydefontein Rockshelter. These sediments have yielded a superb record of Late
Pleistocene and Holocene sediments that contain pollen and other types of
paleoecological data.
This dissertation attempts to wring information from the data recovered from all
three of these sources in Blydefontein Basin. The basin's geographical setting is
described first (Chapter II) in a conventional format. Its bedrock geology, the
geomorphological forces that shaped it and its more recent sedimentary infilling are
treated first. Annual and seasonal variations of local climate are reviewed, followed
by an extended summary of the complex way in which four regional vegetation
communities, and the smaller and more local plant communities (the so-called Veld
Types) overlap in the basin's vicinity. This is followed by an overview of the relict
fauna of the basin and it surrounds, and the chapter closes with a brief evaluation of
European impacts on both plants and animals during the past two centuries.
4
The mechanics of regional climatic circulation systems are briefly reviewed in
Chapter Ill as a background for understanding how climate simulations models are
designed. The COHMAP simulation is then reviewed in detail as this provides the best
available model of Holocene climate change against which to compare my field data.
With a predictive model of temperature and rainfall fluctuations in place, it
becomes feasible to derive another model of plant-community changes (Chapter IV)
that is directly testable through pollen analysis. By analyzing a massive unpublished
phytogeographic data-base using discriminant function analysis, it is made clear that
the previously defined Veld Types are invalid and cannot be used in such a study. This
botanical data-base is then reclassified into new plant communities using cluster
analyses which are double checked with a second discriminant analysis. These new
plant communities form the backbone of the botanical model, which predicts climatedriven shifts in Blydefontein Basin.
The botanical model is tested in Chapter V by pollen analysis of the extensively
dated basin fills and buried soils, and these results are cross-checked with molluscan
and diatom analysis from the same sediments. A triple check on the younger parts of
these sequences is provided by closely spaced pollen spectra from dated hyrax dung
middens that provide high resolution for the last 1200 years. The combined fossil
pollen record is classified to the new scheme of modern plant communities, executed in
the previous chapter. The combined Holocene botanic record is now seen to oscillate
between five different plant communities over the last eight millennia, but the
terminal Pleistocene pollen spectra have no modern analogs.
These combined data are compared with a sequence of stable carbon isotope ratios
in Chapter VI. Bulk organic carbon samples were obtained from the dated alluvial
sediments that yielded pollen, and fluctuations in C3 and C4 grasses (not accessible by
palynology) become apparent. Furthermore, an independent cross-check on the o13c
humate curve is provided by a second
o13c
5
curve obtained from from a vertical
sequence of ostrich eggshell samples excavated from Blydefontein Rockshelter.
In Chapter VII modern linear transform functions, developed in Chapter IV, are
used for predicting temperature and rainfall from botanical associations, and the
pollen sequences are used to estimate climatic conditions during their accumulation.
The fine-tuned hyrax midden pollen sequences are checked against historical rainfall
records with remarkable success. Thus armed, it is finally possible to combined all
the Holocene pollen-derived climatic estimations, and compare the estimates with the
computer simulations of past climates as presented in Chapter Ill. The fit between the
two climatic estimations is shown to be good, although the Blydefontein data show far
superior resolution and many more oscillations that the simulated curves.
The Blydefontein Rockshelter stratigraphy and matrix granulometry are
described in Chapter VIII, together with the radiocarbon chronology for the sequence.
Phosphate values for the sequence clearly isolate a thin layer of decomposed dung,
possibly derived from penned livestock. There is a notable unconformity in the
sequence that adds significant support to J. Deacon's "mid-Holocene abandonment"
scenario. The suspected terminal Pleistocene layer at the base of the excavation cannot
be dated yet by radiocarbon. Meerkat Rockshelter contains a thin and poorly stratified
fill that overlaps with the upper part of the Blydefontein Rockshelter sequence.
In Chapter IX, I turn to the temporal changes in artifact designs at Blydefontein
and Meerkat Rockshelters, and present evidence for six temporal marker-attributes.
It is proposed that these might be used in a future project to cross-date Interior
Wilton surface assemblages already mapped in the adjacent Zeekoe Valley.
Chapter X reviews the historical development of models designed by South African
archaeologists for explaining variability in Later Stone Age artifact assemblages.
Most of these are found to be mono-causal models, so other non-local models that
6
suggest processual approaches are reviewed as possible frameworks for the multicausal modelling that follows.
In Chapter XI, I outline the basis of a new, four-system model with a central focus
on risk-reducing strategies used by dozens of ethnographic hunter-gatherer groups
from different habitats around the world. A sliding scale of global habitat types is
erected as a framework on which to cross-refer biomass changes, changes in the
plant/animal mix of the habitat, and changes in the spacing and seasonal availability of
foodstuffs. This cluster of variables is linked to sliding scales of hunter-gatherer
risk-buffering responses: changes in overall group mobility, mixed patterns of
forager versus collector mobility, and changes in range size.
A six-part technology system is welded on to the model in Chapter XII. A sliding
scale of mobility pattern options ranging from extreme forager to extreme collector
patterns provides the framework. This is used to organize various hunter-gatherer
options for tool design. These include the mix of reliable versus maintainable
qualities, the amount of use-life, decisions to replace or repair, decision to transport
or cache, the value placed on repair and production kits, and frequency/timing of use
and repairs. The chapter ends with some test implications for archaeology and for
settlement pattern analysis derived from this model.
A partial test of the model with Blydefontein data is presented in Chapter XIII.
Faunal data are not yet available, but the dynamic relationship between the ecological
changes in Blydefontein Basin, and changes in Interior Wilton artifact design at
Blydefontein Rockshelter can be addressed. The time-calibrated vegetation record is
used to predict changes in local population density, mobility, range, territorial
packing, and in between-band reciprocity. From these predicted fluctuations, test
implications are spelled out in terms of predicted fluctuations in exotic raw material
frequencies, endscraper designs, arrow armature (microlith) replacement strategies,
7
and bladelet technology. A factor analysis is used to isolate one cluster of design trends
that fit well with the expectations of the multi-causal model.
Finally, in Chapter XIV the significance of these results is reviewed in terms of
their potential for explaining regional and supra-regional trends in Later Stone Age
lithic design choices.
Obviously a robust test of the model will require a larger spatial data base such as
that available in the adjacent Zeekoe Valley. That, however, is for the future, when I
hope to put my general model through a more exacting set of tests than can be brought
to bear by the data presented in this dissertation. Nevertheless, the model is
presented here in the belief that it helps to explain several large-scale, inherently
puzzling patterns long known to characterize the Later Stone Age archaeology of
southern Africa.
CHAPTER II
THE MODERN ENVIRONMENT
The Physical Setting
Blydefontein and Meerkat rockshelters occur in a perched basin in the western
portion of the Kikvorsberg Range which is drained by the Oorlogspoort River (Figure
1). Kikvor means frog and berg is the word for mountain in Dutch. Since the late
1700s this area has been known as the Bo-Hantam or the upper Hantam. These grassy
mountains can be considered as the westernmost extension of the alpine Drakensberg
Range into the semi-desert interior plateau of southern Africa. The veld (pron.: felt)
in the Bo-Hantam is dominated by tall, tough bunch grasses that are not palatable to
small stock, and do not retain their nutrients during the winter, while the grasses in
the flat Karoo plains below are shorter, much more nutritious and easier for small
stock to eat. The Bo-Hantam grasses are called sourveld and the plains grasses are
known as sweetveld. Today this area is in the summer rainfall region of southern
Africa and, during the summer, clouds driven by westerly and northwesterly winds
blow across the flat Karoo plains.
On many hot summer afternoons these rain-bearing
clouds accumulate on the Kikvorsberg Range and drop their loads on the cooler
mountains. Occasionally heavy rains flood the gullies (colloq.: dongas) that drain the
mountains. During the winter Blydefontein often remains cold and fogged-in until
afternoon while Noupoort, at the base of the Kikvorsberg, only 10 kilometers away is
warm and sunny. Heavy snowfalls are not unusual and in the 1940s a group of people,
caught in a Kikvorsberg blizzard, froze to death. These topographic and orographic
effects result in greater rainfalls and cooler temperatures in the Kikvorsberg as
8
9
opposed to the flat Karoo and grassland plains below. Many of the animal species that
typified the southern African plains are no longer present. Today springbok, steenbok,
meerkats, jackals, bat-eared foxes, and ostriches are still common, blesbok and
wildebeest occur in small numbers, but many of the other animals are gone. The region
is unique, and this distinctiveness is due the combination of its geology, climate, and
the biota.
•
II.
(
Hyrax Middens
Sediment Sites
Excavated Rackshelter
\ .... .........
'·., .............
\i
'
!
;
i \
\
((··
KIKVORSBERG
0
0
)
2000
m
i
/
/
/
i
\,
;\
I
./
\·.
\
. /"\
\·.
["-..1 \\
Figure 1. Blydefontein and Meerkat Rockshelters, and paleoenvironmental sites in the
Kikvorsberg Range.
10
Geology and Geography
Blydefontein Basin Topography and Hydrology
Blydefontein Basin covers approximately 15 square kilometers, and is created by
a large north-south trending dolerite ridge on the east through which a narrow
constricted gap (colloq.: QQ.Qil), Diepkloof (1660 msl), provides the only drainage
out of the basin (see Figure 1). The Basin is flanked by high peaks such as
Oppermanskop (2049 msl) to the north and others to the south, and has rolling
topography with a few small dolerite ridges on its floor. The basin forms a small
cuesta-like feature with the steep, western edge dropping off into the Zeekoe Valley.
The Basin's only drainage divides 1.25km upstream from Diepkloof. One stream flows
from the north running in front of Blydefontein Rockshelter, and the other drains the
southern portion of the Basin and flows in front of Meerkat Rockshelter. Small gorges
occur in the lower reaches where the streams flow year round. These gorges have
valley-in-valley cross sections with numerous overhangs and shelters.
A more
detailed discussion of the bedrock geology and geomorphology is warranted in order to
understand the formation dynamics of the landscape.
Bedrock Geology and Geomorphology
Three major processes have formed Blydefontein Basin's geological landscape:
plate tectonics, deposition and erosion. Each process has played a major part in
creating the landscape as we now see it, and an understanding of the geologic and
geomorphic setting can only be gained by discussing the individual roles of these
processes.
Most of the surficial bedrock sediments in Blydefontein Basin are from the Karoo
Sequence which is capped by the volcanic Drakensberg Formation. Beginning in the late
Carboniferous, 290 million years ago (mya), through the early Jurassic, 190mya,
southern Africa was in the middle of Gondwanaland, a single ancient continent composed
11
of most known southern hemisphere landmasses (Brink 1983: 18). Southern Africa
was a huge basin into which enormous amounts of sediments were deposited. These
sediments, up to eight kilometers thick, are known as the Karoo Sequence. The Karoo
Sequence, from older to younger, is made up of the Dwyka Formation, the Ecca Group
(twelve formations), the Beaufort Group (eight formations) and the Molteno, Elliot and
Clarens formations.
At the beginning of the Karoo Sequence glaciers covered much of southern Africa
because it was situated near the South Pole (Visser 1986: 14-15). In time
Gondwanaland drifted north toward the equator and the glaciers melted leaving a large
shallow periglacial lake with lush vegetation
(Brink 1983: 20).
Lacustrine and
deltaic sediments were deposited in this lake to form the Ecca Group (Visser 1986:
14). Eventually the the lacustrine deposits and prograding deltas filled the shallow
basin, and floodplains formed. These alluvial deposits are known as the Beaufort Group.
By this time central Gondwanaland was drier and warmer due to further northward
movement of the continent, and evidence of life, especially animal life, is abundant.
Most of the bedrock in Blydefontein Basin are of the Beaufort Group.
Biostratigraphers have divided the Beaufort Group into five zones, however
lithostratigraphic units will be discussed here as the lithology is more important to
this discussion. Lithostratigraphers have divided the Beaufort Group into two
subgroups (the Adelaide and Tarkastad) and eight formations (Brink 1983: 22).
Bedrock at Blydefontein Basin is in the Tarkastad Subgroup and probably the Katberg
Formation rather than the Burgersdorp Formation, although detailed geological
mapping is not available. The Katberg Formation, composed of sandstones with
subordinate mudrock lenses, represents a distal alluvial fan and/or a braided fluvial
system, while the Burgersdorp Formation, composed of mudrocks with subordinate
sandstones, represents a meandering fluvial system (ibid: 22). The sequence of
12
alluvial fan to braided stream to meandering stream in a floodplain suggests that during
the deposition of the Tarkastad Subgroup the overall stream gradient was diminishing.
In any case moderately thick sand bodies were deposited with subordinate mudrock
layers, and these sandstone bodies form rockshelter overhangs, while the backwalls of
the shelters are composed of more easily eroded mudrock layers.
Throughout the time of the Karoo Sequence the super-continent continued to move
north and become drier and warmer. Termination of the Karoo Sequence is marked by
the breakup of Gondwanaland. This enormous rifting is associated with volcanic events
that exhumed massive amounts of basaltic lavas, the Drakensberg Formation.
These
basalts covered much of the ancient Karoo sedimentary basin between 190 and
150mya. This basalt layer was so hard and impenetrable that later magma intrusions
were confined to the Karoo Sequence, and formed the numerous dolerite dikes and sills
seen on the surface today (ibid: 19, 177). The dikes followed weaker sedimentary
bedding planes, and thus usually intrude next to the argillaceous sediments such as
shales or mudrocks (ibid: 177-178). Intrusive magma baked the adjoining deposits,
and in extreme cases, metamorphic alteration produced hornfels from mudrocks and
quartzites from sandstones (ibid: 180). Today these dolerite dikes crisscross the
Karoo surface, and the hardened deposits baked by the dikes are an impediment to
erosion.
With the breakup of Gondwanaland in the early Jurassic erosion becomes the
major controlling process on what had finally become an "African landscape" (ibid: 26;
King 1978: 3-17). Erosion of steep scarp faces and the development of flat peneplains
dominate later geological processes. A major erosional cycle, stimulated by the
breakup of the super-continent and changing base levels in the Early Cretaceous
period, removed most of the original surface to form the Post-Gondwanaland Planation
Surface (ibid: 11-12).
13
The most extensive erosional cycle, the African cycle, lasted approximately 100
million years from the Late Cretaceous to the mid-Tertiary (Brink 1983: 23),
resulting in the African Surface (ibid: 27; King 1978: 12). In the eastern Cape much
of the African Surface remains as mesas or buttes (colloq.: tafelbergs) where resistant
Beaufort sandstones and dolerite sills have protected it from later erosion (Brink
1983: 27). The Kikvorsberg is probably African Surface, although the highest hills
(colloq.: koppies) or tafelbergs such as Oppermanskop may be erosion features on the
African Surface. Thus its uppermost surfaces possibly represent the PostGondwanaland Planation Surface (T. C. Partridge, personal communication).
The
African Surface is usually deeply weathered in places where a pedocrete armor has
formed. On African surfaces like the Kikvorsberg these pedocrete remnants are often
silcretes (ibid: 27).
Uplift in the Miocene and early Pliocene then stripped most of the Karoo basin
(ibid: 27, King 1978: 14). Locally the subdued topography and medium-sized upland
remnants suggest that this post-African erosion is extensive and continues today (ibid:
28). The shallow weathering (except for some calcretes) attests to a relatively
shorter duration (ca. less than 20 million years) while elsewhere the African cycle
lasted some 100 million years (Brink: 28).
The active erosional scarp of the
Kikvorsberg (::;; 585m in height) was probably initiated by the Post-African event in
the Miocene and Pliocene.
This is part of the incision limited to a 10-15km wide band
on both sides of the Orange and Vaal Rivers. Narrower bands occur on their major
tributaries which include the Oorlogspoort and the Zeekoe rivers (ibid: 28; Butzer
1971; Helgren 1979).
In Blydefontein Basin, the Diepkloof poort was created by headward erosion of
the Oorlogspoort River which penetrated the resistant dolerite dike that bounds the east
side of the Basin. The many small streams which drain the floor of the basin have cut
14
small gorges with numerous overhangs in their lower reaches. These shelters were
created by stream erosion and weathering of mudrock layers between more resistant
sandstone bodies that form the roofs of both excavated shelters and many other shelters
in the Basin. The cross-sections of Blydefontein and Meerkat gorges have a valley-invalley configuration due to the presence of various resistant sandstone layers.
Thus the modern landscape was created by dynamic forces that were local, regional
and worldwide in scope. Although its configuration has been here for a relatively short
time in geological terms, in human prehistoric terms it evolves with imperceptible
slowness.
Lithic Resources
A number of local knappable lithic resources were identified: hornfels or
lydianite, quartzite and silcrete. The most important is hornfels. Hornfels occurs
where intrusive dolerite dikes have baked shales or mudrocks for prolonged periods.
Hornfels is extremely common in the region, and it was used throughout all of
prehistory as the dominant material for chipped stone artifacts. The texture varies
from coarse to fine grained, but flaking quality is related to many other characteristics
also. Here hornfels occurs in huge sheets that cover the tops of some tafelbergs,
adjacent to dolerite dikes in thin linear strands, as fractured and weathered cobble
debris below in situ occurrences, or as ancient fluvially deposited gravels on terraces
overlooking stream floodplains. Blydefontein Basin has only a few dolerite dikes, and
no hornfels outcrops. The closest hornfels outcrop and prehistoric quarry is 5
kilometers northeast, at Hughdale.
Quartzites are sandstones metamorphosed by intrusive dolerite dikes. These are
not common, and the quartzites are not used extensively for stone tool manufacturing.
The silcretes are a product of weathering associated with the African Surface. They are
15
found occasionally in situ on the higher mountain ridges in the Sneeuwbergen and
presumably the Kikvorsberg, although none has been discovered in or near
Blydefontein Basin.
Two non-local raw materials, agate and jasper, were sometimes used at
Blydefontein although they are known to occur only in Orange River gravel deposits,
apparently originating in the highlands of Lesotho. Orange River terraces extend from
five to ten kilometers from the river (Butzer 1971; Helgren i 979; Partridge,
personal communication) so agate and jasper gravels should be obtainable from these
terraces as well.
Blydefontein Basin Quaternary Sediment Accumulation
The Diepkloof constriction has created a situation, probably due to cyclical
floodplain overbank deposition (Patton and Schumm 1981) and changes in local base
level because of masswasting events in the narrow gorge (T.C. Partridge, personal
communication), whereby moderately thick alluvial deposits of Pleistocene and
Holocene age have accumulated in the gorge above the constriction at Diepkloof. King
(1942: 52) named these types of sediments rock-defended terraces. This type of
perched basin in the mountains above narrow constricted gorges provide more detailed
sedimentary records, at least for the Pleistocene and Holocene, than the open broad
floodplains in the Karoo, e.g. the Oorlogspoort or Zeekoe valleys. This is because
bankfull stage is reached more rapidly in constricted channels and is confined to a
smaller floodplain (Magilligan 1985). Also rapid flushing of water off the mountains
transports greater sediment loads that are then deposited in the lower reaches of
Blydefontein and similarly positioned basins.
16
Very little detailed soil mapping and classification is available for the Blydefontein
region. However a general soil map is available for the entire Karoo region (Ellis and
Lambrechts 1986). Three soil types are mapped for Blydefontein Basin. The lower
portion of the Basin has Duplex soils. These are characterized by loams to loamy sand A
horizons with massive to platy structure above a red B horizon with clay content at
least twice as high as the overlying horizons and moderately to strongly structured. In
the Blydefontein region the B horizons are often known as dorbank (MacVicar et al.
1977: 126) which probably represent multiple cycles of pedogenic development over
at least the Late Pleistocene. Also present in the lower reaches of the Basin are soils on
alluvium with Melanic A horizons (ibid: 126). Eluviated horizons can occur between
the A and B horizons. Parent materials consist of shales and dolerites. On the
relatively flat area of the Basin above the gorges and below the koppies shallow soils of
pedologically young landscapes are mapped. Often these are soils on bedrock. The soils
of the koppies and mountains have not been classified, but on the slopes erosion is a
dominant process and bedrock exposures without soils are common.
Climate
I have first hand experience with Blydefontein weather between late January and
late September (1985).
With the beginning of fall in March, frosts or freezes
occurred almost every night until the end of the excavations in September. The
streams were almost always frozen during the winter. Also light snows occurred in
April and September, but the Kikvorsberg, along with the Sneeuwbergen to the
southwest, are known for heavy snows so 1985's snows were light. While working in
the Zeekoe Valley during early September of 1981, I observed heavy snows in the
17
Kikvorsberg and temperatures were low enough for the snows to remained on the
ground for 10 to 14 days (Jim Meyer, personal communication).
Most rain in Blydefontein Basin falls in the summer and early fall, and during
1985 good general rains occurred in February. The ground was saturated with water
and the grass turned a lush green. Often the rains are due to convectional uplift and
orographic effects as saturated clouds are blown across the plains in the Zeekoe Valley
and uplifted by the Kikvorsberg. General rains are associated with a tropical easterly
jet and usually occur in late spring through early fall (Tyson 1986: 127-128).
Normally if a good rains occur during the summer or fall then that year is a high
rainfall year.
The climate of the region is classified in the Thornthwaite system as mesothermal,
semi-arid, cooler-temperate with severe frosts (Schulze and McGee 1978: 43-49),
and as dry hot arid steppe (mean annual temperature < 18 oc), BSk, in the Koppen
classification system (ibid: 37-39).
Modern rainfall records were available through Mr. Norman Biggs from the
nearby farm of Grapevale. It is important to note that Grapevale is 200m lower in the
Kikvorsberg, where the orographic effects are not as great as at Blydefontein.
Blydefontein receives more rain, but the differences are of degree rather than of kind.
Monthly rainfall records were available from March of 1920 until the conclusion of
the excavations in September of 1985. Unfortunately no temperature records were
available from Grapevale, but a 20 year record (from 1962 to 1982) was available
from Dr. Piet Roux, Grootfontein Agricultural College, Middelberg.
Temperature
At Grootfontein the mean annual temperature for the period 1962-82 was 14.6 °C
(58.3 °F) with the lowest yearly average of 13.5 °C (56.3 °F) and the highest yearly
18
average of 15.4 °C (59.7 °F). Grootfontein is 405 meters below Blydefontein and if
one can assume that the normal lapse rate holds (i.e. 6.5 °C change per 1000 meters,
Trewartha 1968: 46), then the mean annual temperature at Blydefontein would be
some 2.6 °C (4.7 °F) below Grootfontein or 12 °C (53.6 °F). One should realize that
this estimation probably errs on the warm side, because the orographically produced
cloud-cover increases albedo and rainfall, which would act to reduce the mean annual
temperature more that the normal lapse rate.
24
0
(/)
(J)
22
セ@
0)
(J)
20
セ@
18
ii1
....
16
セ@
:::J
(J)
0.
E
(J)
1セ@
.r::.
"E
0
:2
c
res
(J)
:2
14
12
10
8
6
0
2
4
6
8
10
12
Month
14
16
18
20
22
Figure 2. Mean monthly temperatures in °C with one standard deviation bars at
Grootfontein over an average twenty month period.
The Grootfontein average monthly temperatures (with one standard deviation
bars) have been plotted in Figure 2, over a 20 month period in order to view better a
full season's cycle. The average January temperature at Grootfontein is 20.8 °C (69.5
OF) and the average temperature for June is 7.9 °C (46.2 °F). This is a 12.9 °C (23.2
°F) range in the average monthly temperatures. These data show a clear summer high
temperature plateau (in December, January and February), and a winter low plateau
19
(in June and July). This is mostly due to seasonal variation in the solar radiation
budget which is twice as great in the summer as in the winter (Schulze and McGee
1978: 23). The most variable months are May (no.s 5, 17) and September (no.s 9,
21) The minimum recorded temperature for Grootfontein between 1962 and 1982
was -10.3 °C (13.5 °F) in August 1975 and the maximum recorded temperature was
39.2 °C (1 02.6 °F )in January 1965.
Rainfall
The mean annual rainfall for the period from 1921 until 1984 at Grapevale is
366 mm
(14.4 inches) with the driest year, 1949, receiving only 139.5 mm (5.5
inches) and the wettest year, 1974, receiving 865.5 mm (34.1 inches). The range is
726 mm (28.6 inches).
Rainfall from the years 1920 and 1985 were not used
because observations were not available for all months in those years.
Figure 3 shows the average distribution of rainfall by month at Grapevale over an
average twenty month period. Both the mean monthly rainfall and the percent of
months receiving significant rainfall by month are illustrated.
defined as
セ@
Significant rainfall is
25.4 mm (i.e. 2::1 inch) of rainfall during a single month.
The seasonal rainfall pattern peaks in late summer (February and March) then
drops to a four month winter low (June-September).
Not surprisingly these four
winter months are also much less likely to receive any significant rain. It is
important that strong positive correlations exist between mean monthly rainfall,
percent of significant rainfall and maximum monthly rainfall.
Rainfall varies the
most in January, February and March, and March is the only month that has had rain
every year from 1920 until 1985. These rainfall patterns result in a significant
seasonal climatic pattern that affects plant growth (Vorster and Roux 1983: 19).
Omean monthly rainfall (mm)
90
,...._
80
E
E
""-/
70
-セ@
20
D%Months with significant rains
60
c
·:;
50
1...
-:c
:7\
40
c
0
30
E
c
20
lf1
<II
E
10
0
2
4
6
8
10
12
14
Month
16
18
20
22
24
Figure 3. Mean rainfall and percent of significant rainfalls by month over a 20 month
period.
The Botanical Setting
The modern vegetation mosaic of the Blydefontein region is diverse and complex.
As will be shown, the background history of its development is of central concern to
this dissertation because fossil pollen make up a significant part of the new data to be
presented. It is appropriate, therefore, that the regional flora be reviewed in some
detail.
Major Floral Groups in the Blydefontein Region
Southern Africa has four major floral groups that impinge in varying degrees on
the Kikvorsberg:
1) Capen sis flora; 2) the Karoo-Namib flora; 3) the Sudano-
Zambezian flora and; 4) the Afromontane flora (Werger 1978a). Each group has
genera and families that distinguish and characterized its flora, and each is reviewed in
turn.
21
1) Capensis or Fynbos flora are uncommon but nonetheless important in the
Kikvorsberg.
Taylor (1978: 173-229} distinguishes the Fynbos flora of the
southwestern Cape by the presence of three families: Restionaceae, Ericaceae and
Proteaceae. Not all plant species are of these families but those from different families
represent physiognomically similar forms.
Today the Fynbos flora occurs in the
winter rainfall region of the southwest Cape. Intensification of the Benguela Current
off the southwest coast of Africa by 1Omya provides a minimum age for cold water
upwelling critical to the development of the modern Mediterranean-type climate in the
southwest Cape (Shackleton and Kennett 1975). Presumably the establishment of
these climatic conditions allowed for the formation of a Fynbos-type of plant
community. Palynological evidence demonstrates that a Fynbos-type of community was
well developed by the early Pliocene and appears to be associated with a shift toward a
cooler and drier climate in the southwest Cape (Coetzee 1978, 1983, 1986; Coetzee
and Rogers 1982).
2} Karoo-Namib flora (Werger 1978a: 231-299} is a xeric flora characterized
by the families Compositae, Gramineae (esp. the tribe Stipeae), Aizoaceae,
Mesembryanthemaceae, Liliaceae, and Scrophulariaceae. Succulents are common, but
trees are rare, and many genera are endemic to the Karoo-Namib flora. It shares many
taxa with the Sudano-Zambezian flora, but few with the Capensis flora. Nonetheless
Acocks (1975) suggests that the Karoo-Namib flora developed from the Capensis flora.
Levyns (1964) argues that the origin of the Karoo-Namib flora is the SudanoZambezian flora because of stronger associations. However, as aridity in the region has
existed since the late Cretaceous 80 million years ago (Ward, Seely, Lancaster 1983),
we can assume that the origins of the Karoo-Namib flora are ancient and complex.
Certainly by 10 million years ago the increased intensity of the Benguela Current
22
would have resulted in extreme arid conditions in the western portion of the
subcontinent, and the Namib Desert and Karoo have probably existed ever since.
3) Sudano-Zambezian flora (Werger and Coetzee 1978: 301-462) is composed of
vast areas of woodland, savanna or thornveld, and grassveld vegetation. Plant
communities of this group cover most of the inland plateau of southern Africa. The
woody taxa are as a whole poorly adapted to frosts and most do not occur in regions with
significant cold weather. The grassveld is the only Sudano-Zambezian vegetation form
to adapt to cold conditions. Characteristic species include Themeda, Cymbopogon and
Eragrostis.
This floral region intermingles with the Karoo-Namib flora. The origins of
some taxa characteristic of the Sudano-Zambezian flora extends back to at least the late
Cretaceous (Axelrod and Raven 1978), but other important elements did not evolve
until much later (e.g. grasses in the Eocene). At the Miocene/Pliocene boundary the
shift toward a cooler Agulhas Current contributed to reduced rainfall in the interior
(Martin 1981) and this favored the expansion of grasslands over the savanna
especially in the cooler inland plateau (Brain 1980; Vrba 1980).
4) Afromontane flora, which in this discussion includes the Afroalpine flora,
occurs on the high mountains throughout Africa (Killick 1978: 514-560;
1978: 463-513).
White
This floral form is distributed on elevated 'islands' in an arc from
Ethiopia to South Africa. The similarity of species between two South African areas
implies that a continuity or exchange path existed between the Drakensberg and the
Knysna Forest in the past which must have extended across the highland areas of the
eastern Cape that could include the Kikvorsberg. Little data are available concerning
the origins of this flora, but Axelrod and Raven (1978: 88, 92, 109) suggest that it
evolved in the Saharan highlands in the Paleocene and with the creation of highlands in
East Africa it spread south so that by the Miocene it was present as far south as Knysna.
23
Much more research is needed before the questions of origin and chronology of
major floras of southern Africa can be resolved. The modern distributions of plant taxa
offer some insights into the evolution, migration and development of plant taxa and
plant communities, and if linked with paleobotanical studies significant insights can be
gained.
Veld Types in the Blydefontein Region
In 1953 J.P.H. Acocks published his Veld Types of South Africa, a nation-wide
survey of plant communities. The monograph assesses the grazing potential of different
biota, so it is not strictly a botanical classification. Veld Type means "a unit of
vegetation whose range of variation is small enough to permit the whole of it to have the
same farming potentialities" (Acocks 1975: 1).
The Veld Types in the area of Blydefontein Basin can be divided into three major
groups: grassveld, bushveld, and Karoo. Even though all have grasses, the grassvelds
are distinguished from the Karoo and savanna by denser plant cover, fewer of the small
woody Compositae bushes (colloq.: bossies) that distinguish the Karoo Veld Types, and
none of the dense shrub and small trees that characterize the savanna. Most of the
plants that occur today on Blydefontein, both annuals and perennials, have short
generation spans and rapid germination periods, and this is important when
considering their response to changing climatic conditions. All the Veld Types within
an arbitrary 140km radius of Blydefontein are described below (Figure 4).
Grassvelds
Five grassvelds are recognized by Acocks in the Blydefontein area: 1) Karroid
Merxmuellera Mountain Veld, 2) Themeda-Festuca Alpine Veld, 3) Dry CymbopogonThemeda Veld, 4) Cymbopogon-Themeda Veld, and 5) Stormberg Plateau Sweet Veld.
24
1) The Karroid Merxmuellera Mountain Veld is the vegetation type mapped for
most of the Kikvorsberg Mountains including Blydefontein Basin and is usually
restricted to the higher mountains of the eastern Karoo and adjacent fringes (Acocks
1975: 98). It is dominated by the grass Merxmuellera with Themeda and Tetrachne
as co-dominants on high mountain tops. A number of other grass taxa are present
including Eragrostis, Melica, Festuca, Pentaschistis, Brachypodium, Bromus and
Cymbopogon. It is unfortunate that pollen analysis cannot distinguish between grass
taxa as significant differences exist in the modern plant communities. In the
grasslands of the Basin the spectacularly colored red-hot poker (Kniphofia sp.) is
scattered in amazing profusion during years of good rainfall.
A number of Fynbos taxa
are often present, especially Elytropappus rhinocerotis, Euryops , Erica caffra,
Cliffortia , Passerina montana, and Anthospermum . Karoo related composite genera
include Chrysocoma, Helichrysum, Eriocephalus, Nest/era, and Felicia bossies.
Another major group of plants are the semi-succulents such as Ruschia, Aloe,
Euphorbia and Crassula.
On the few dolerite koppies and ridges in the Basin the
shrubs, Rhus and Diospyros, are common. Rhus and Diospyros have edible berries as
does a small rose bramble (Rubus sp.) observed growing in the gorge near
Blydefontein Shelter. The Diospyros berries are very pungent. Other potential plant
foods include springbok (Senecio radicans ), veldpatat (species unknown) and the
vinkel (species unknown). Many of the edible plants occur in the plains below the
Kikvorsberg, and one farm in the Zeekoe Valley is named Vinkelfontein (colloq.:
fennel-spring).
Often along the streams hardbees reeds (Cyperus marginatus) and
swamp grass (Phragmites australis) line the banks in such dense stands that it is
impossible to reach the stream. Hardbees reeds were used by early European settlers
(colloq.: trekboers) to construct huts. Abundant diatom-producing algae are also found
25
in the streams and pools in the Basin. These same riverine plants are found in the
plains below the Kikvorsberg.
140 km radius
false upper karroo
karroid Merxmuellera
mountain veld
succulent mountain scrub
false karroid broken veld
karroid Merxmuellera mountain veld
replaced by karroo veld
Figure 4. Veld Types within 140km radius of Blydefontein Rockshelter (after Acocks
1975}.
26
2) Themeda-Festuca Alpine Veld grows in small patches on the highest peaks of the
Kikvorsberg(Acocks 1975: 95-96).
It is dominated by the grass Themeda with
Festuca, Merxmuellera, Eragrostis, Andropogon, Cymbopogon, Microchloa,
Diheteropogon, Trachypogon and others. Sometimes associated with the ThemedaFestuca Alpine Veld on the high peaks in the Kikvorsberg is the Karroid False Fynbos of
Acocks (1975: 97), or the Subalpine Fynbos of Killick (1978: 538).
The
largest
occurrence of the Themeda-Festuca Alpine Veld/Karroid False Fynbos is found well to
the east of Blydefontein in the Drakensberg, but patches are found between the
Drakensberg and Blydefontein on the tops of the higher mountains, and even 350km
west of Blydefontein (Acocks 1975: 95). This distribution suggests that relict patches
survive from a time when Themeda-Festuca Alpine Veld/ Karroid False Fynbos enjoyed
a much greater distribution.
3) Dry Cymbopogon-Themeda Veld can be found today in small stands in the plains
below the Kikvorsberg. These are dominated by the grasses: Themeda, Tetrachne,
Tragus, Eragrostis, Digitaria and Cymbopogon.
Also present, but not numerous, are
typical Karoo composit bossies: Helichrysum, Felicia, Pentzia and Chrysocoma.
On
dolerite hills and ridges the shrubs, Rhus and Diospyros, are common. Acocks (1975:
78, 91) argues that the Dry Cymbopogon-Themeda Veld is almost totally altered to
Karoo due to overgrazing and poor veld management by Europeans.
4) The Cymbopogon-Themeda Veld is dominated by Themeda, Setaria, Microchloa,
Elionurus, Heteropogon, Eragrostis, Tristachya, Helichrysum, Brachiaria,
Cymbopogon, Harpochloa and Hermannia. Other less common genera include Digitaria,
Senecio, Anthospermum, Felicia, Aristida and Andropogon. The Cymbopogon-Themeda
Veld usually occurs on sandy soils, receives a high amount of rainfall, and usually
occurs at elevations higher than Blydefontein.
27
5) The Stormberg Plateau Sweetveld is transitional to the Karroid Merxmuellera
Mountain Veld and the grassvelds of the higher Drakensberg range such as the
Themeda-Festuca Alpine Veld. The Stormberg Plateau Sweetveld enjoys greater
rainfall than the Karroid Merxmuellera Mountain Veld, slightly higher elevation, and
is on flat plateau areas rather that steep mountain slopes. This grass veld is dominated
by Themeda and Elionurus. Other grass genera are Pennisetum, Tetrachne, Festuca,
Eragrostis, and Digitaria.
Karoo Velds
Open dwarf shrub Karoo Veld Types include 1) False Upper Karoo, 2) Central
Upper Karoo, 3) False Karroid Broken Veld, and 4) Succulent Mountain Scrub.
1) The False Upper Karoo represents the easternmost Karoo Veld Type and is
transitional with the grassvelds.
Werger (1978b: 446-454, 1980) has studied this
region in detail and recognizes two main communities: 1a) dwarf shrub PentzioChrysocomion communities on the peneplains, and 1b) the Rhoetea erosae community
composed of small trees, shrubs and grasses on the rocky ridges and koppies. The
latter communities have two major forms with eight recognizable associations
governed by bedrock type and aspect.
2) The Central Upper Karoo occurs west and northwest of the False Upper Karoo in
areas receiving 200-250 mm of rainfall and ranging from 1050m to 1700m above
sea level (Acocks 1975: 63-64). Karoo bossies are dominant but some grasses, such
as Eragrostis, Aristida , Stipagrostis and Merxmuellera in the hills, do occur. The
major taxa of Karoo composites include Eriocephalus, Pentzia, Pteronia, Nest/era,
Chrysocoma, and Osteospermum. Other important taxa are Euphorbia, Rhus, Ruschia,
and Sa/sola. Although this is a Karoo Veld Type, stands of pure grassveld do occur in the
moist floodplains.
28
3) The False Karroid Broken Veld characteristic genera are Euclea, Pappea,
Cussonia, Acacia, Schotia, Aloe, Pentzia, Becium, Chrysocoma, Asparagus,
Drosanthemum and Eragrostis. Acocks (1975: 79) suggests that invasions of the
Karroid Broken Veld and the Central Lower Karoo into a region once occupied by Dry
Cymbopogon-Themeda Veld created this veld type. The closest portion of this Veld Type
is in the Upper Fish River Basin 60km south of Blydefontein.
4) Succulent Mountain Scrub or Spekboomveld (colloq.: fattree veld) is the only
bushveld that is within 140km of Blydefontein (Acocks 1975: 58-59).
Dense shrub
and small trees dominate the plant taxa.
These diverse plant communities provide the ultimate food source to which the
animals and prehistoric human inhabitants of the northeastern Cape were adapted. The
animals were as diverse as the plants, and prehistoric hunter-gatherers of the BoHantam had a rich bountiful supply of game.
Animals
Historically Blydefontein Basin and the surrounding plains had a very abundant
fauna. Today after 200 years of European hunting and some 170 years of European
stock farming most of the wild animals have been affected significantly, and many are
locally extinct.
It is surprising how many wild animals remain in either a free
ranging state or as managed herds on the farms of the area. The area has few nature
preserves, however, Mountain Zebra National Park located 145km south of
Blydefontein Basin, was repeatedly visited during 1980 and 1981 in order to observe
as many animals as possible under natural (or near natural) conditions. Mountain
Zebra National Park is too small, 65 square kilometers, to support a complete
ecosystem with a full host of predators. It has no large cats, wild hunting dogs, or
hyenas, and the herds are not free ranging. However, it is 50 years old and its higher
29
elevations support a Merxmuellera Mountain Veld in very good condition.
Topographically the higher elevations are also similar to Blydefontein Basin and it is
the best natural system we can use for comparison. The following discussion uses the
taxonomy presented in Bigalke (1978: 981-1048) and is based primarily on
discussions with local farmers and on my own observations.
Large Mammals
The Kikvorsberg today have retained a variety of larger mammals even though the
area is completely utilized as sheep farms and fenced into small camps, i.e. pastures.
Mountain reedbuck (Redunca fulvorufula), vaal rhebuck (Pelea capreolus), steenbok
(Raphicerus campestris), rare klipspringers (Oreotragus oretragus), and a small
herd of managed springbok (Antidorcas marsupia/is) were all observed on the farm of
Blydefontein. The Merxmuellera Mountain Veld is a sourveld with hard tough grasses,
and the masticatory architecture of a springbok or any small grazer is not robust nor
strong enough to efficiently eat these grasses (Dou Lessing, personal communication).
Thus, without managed herds it is questionable whether springbok would live in any
great numbers in the Bo-Hantam. However, during the late 1700s and 1800s
springbok herds were observed by the thousands in the plains below the Kikvorsberg,
and large springbok herds, known as trekbokken, moved en masse to escape
particularly harsh droughts.
A more complete montane fauna is found at Mountain
Zebra National Park in the Merxmuellera Mountain Veld, where additional species
include black wildebeest (Connochaetes gnou), blesbok (Damaliscus dorcas), red
hartebeest (Aicelaphus buselaphus), eland (Taurotragus oryx), and mountain zebra
(Equus zebra).
The only primate in the Bo-Hantam is the chacma baboon (Papio ursinus), and on
Blydefontein during the winter of 1985 two troops were competing over the control of
30
a sheep feeder, one of their most reliable winter food sources. A third troop is known
to habitually raid the farm of Grapevale. Numerous other troops can be found scattered
in the Kikvorsberg, although population numbers are unknown. Remote rockshelters
are often used as protected sleeping areas by baboons as evidenced by abundant baboon
scats in many shelters. However, no baboon dung was found in Blydefontein or Meerkat
shelters probably because these shelters are too close to human dwellings.
The carnivores are still amazingly abundant despite intensive trapping and
hunting. Canine species in the Kikvorsberg include the black-backed jackal (Canis
mesomelas) which is similar in size to the North American coyote, the silver fox
( Vulpes chama), and the insectivorous and fructivorous bat-eared fox( Otocyon
mega/otis). In the mountains, the most common cats are the wild cat (Felis libyca) and
the rooikat or caracal (Felis
」。イセN@
The rooikat is midway in size between the North
American bobcat and the North American mountain lion. During the eight month field
season five rooikats were trapped on Blydefontein and surrounding farms in the
Kikvorsberg. The nocturnal rooikat can kill considerable numbers of sheep even in one
night and the males travel great distances. Today trapping for these cats as well as
jackals is intensive and often conducted by professional trappers.
Small carnivores include the Egyptian mongoose (Herpestes ichneumon), the
suricate (Suricata suricatta), and the yellow mongoose or rooimeerkat (Cynictis
penicillata). Along stream banks, especially in Diepkloof, clawless otter (Lutra
maculico/lis) scats full of freshwater crab exoskeletons are often seen, but the otters
themselves were never observed.
Other large mammals that are found in the region include porcupines (Hystrix
africae-australis) which are still fairly numerous, the nocturnal and rare aardvaark
or antbear (Orycteropus afel), rare honey badgers (Melivora capensis), and the very
rare pangolin or scaly anteater (Manis temminckil).
31
Small and Micro-mammals
Dassies or hyraxes (Procavia capensis) are small diurnal woodchuck-sized
animals (weighs up to 5.5 kg.) that live in groups known as colonies in rocky outcrops.
One colony lives in the cliff (colloq.: .!s.r.an.s.) that forms Blydefontein Shelter.
Abundant colonies can be found all along the gorges carved by the Basin's streams and in
almost any large rock outcrop in the plains below the mountains where they seem to be
as common as in the Basin. The dassie population in Mountain Zebra National Park was
recently estimated at 195 dassies per square kilometer (Swart et. al. 1986).
Two
species of hare are known from the region, Lepus capensis, and Pronolagus sp.
Morphologically and ecologically Lepus capensis , the Cape hare, resembles the
jackrabbit of North America; the smaller bunny Pronolagus sp., the red hare or
rooihaas, lives on rocky or boulder strewn koppies.
The long-hind-legged, nocturnal
spring hare (Pedetes capensis) is common and still hunted by farm workers.
Significant micromammals in rockshelter deposits include the vlei rats, Otomyinae,
and molerats, Bathyergidae (Avery 1988).
Other Animal Life
A number of additional species occur in the Kikvorsberg and surrounding plains
that have archaeological or ethnographical significance. These include birds, reptiles,
amphibians, molluscs, crustaceans and insects.
Reptiles. Amphibians. Molluscs and Crustaceans
The most common amphibians are frogs (Pyxicephalus sp.} which are abundant in
streams.
Other significant reptiles include the large leguaan( Veranus albigularis)
which is often found under large flat rocks near water in the plains below Blydefontein,
but not in the Basin itself where it is too cold.
Cape and yellow cobras (Naja spp.} are
the most common snakes, and during the first month of excavation a large yellow cobra
32
was killed in Blydefontein Shelter. Although not seen during 1985, another venomous
snake is the puff adder (Bitis arietans arietans). Puff adders are more aggressive than
cobras, and thus more dangerous even though their venom is less toxic. A number of
tortoises are present at Blydefontein, including the geometric tortoise (Psammabates
tentoria), the mountain tortoise (Testudo pardalis), and the padloper (Homopus
femora/is). Freshwater molluscs (Unio caffra) and freshwater crabs (Potamonautes
perlatus) represent significant aquatic animals in Blydefontein Basin.
Birdlife
The avifauna is very diverse with at least 198 species present today in the
Blydefontein Basin area (Sinclair 1984). The most significant species for
archaeological studies is the ostrich, Struthio came/us. Common birds include the
terrestrial and gregarious guinea fowl (Numida meleagris), the secretary bird
(Sagittarius serpentarius), korhaans (Eupodotis spp.), bustards (Neotis spp.), and
the blue crane (Anthropoides paradisea). Also present are numerous birds of prey that
range from falcons and kestrels ( Falco spp.) to hawks (Accipiter spp.), harriers
(Circus spp.}, buzzards (Buteo spp.}, osprey (Pandion sp.), owls (Tyto spp. and Bubo
spp.) and eagles (Circaetus spp., Aquila spp., Polemaetus spp., Hieraaetus spp.). The
owl Tyto is believed to be a major contributor of microfauna! remains to rockshelter
deposits. Scavengers such as the Cape vulture (Gyp coprotheres) are rare today but
were commoner in the not too distant past. Another bird that inadvertently is
archaeologically significant is the swallow (Hirundo spp.). These insectivores build
nests of saliva-cemented dirt pellets stuck to the roofs of rockshelters and overhangs.
The roof of Blydefontein Shelter has numerous swallow nests.
Through time these
nests break or dislodge, and become incorporated in rockshelter sediment matrix.
Significant Insects
33
Termites, especially the ant hill genera such as Trinervitermes and harvester
termites of the genera Hodotermes, are certainly present in the Kikvorsberg and
surrounding region, but not in great densities. It is possible that termite nests were
used as grass temper sources by prehistoric potters. The red and black brightly
colored brown locusts (Locustana pardalina) are known to form great swarms after
drought-breaking rains. Clouds of these locusts descend on the veld, eating everything
in their path. The trekboers were terrified by the locust swarms, and modern
farmers still dread the locust plagues. Between swarming periods we now know that
the locusts change morphologically into dusk brown nongregarious grasshoppers. Near
to water and limited to warm days is the carnivorous stickfly (Diptera). Camping near
a water source with active stickflies would be very unpleasant today and in the past.
Lastly the Karoo caterpillar (Loxostege frusta/is) can be very abundant and may have
served as a food source to prehistoric inhabitants.
European Impacts
A detailed review of historical records will not be attempted here as a new project
dealing with 18th and 19th century European activity in the eastern Cape is planned
(Sampson 1987). However, a few words that briefly outline early European history
and possible environmental effects is required.
Historical Background
The Dutch established settlements in Cape Town in 1652 and by the 1760s pioneer
farmers or trekboers, who lived in reed huts or camped by their wagons, began to
colonize the Sneeuwbergen.
By the late 1770s the Sneeuwbergen trekboers began to
graze their stock in the Zeekoe Valley below the mountains during the winter, and these
trekboers founded the first town in the region, Graaff Reinet, 100 kilometers
34
southwest of Blydefontein in 1786. By the 1790s the Hantam was prized by trekboers
as good horse country. Farmer populations increased so that by 1830 the town of
Colesberg was established north of Blydefontein. Other farming villages sprang up in
the area during the mid to late 1800s, and use (and abuse) of the veld increased
accordingly.
Before the veld was fenced in the early 1920s the sheep herding pattern was very
different then than today (N. Biggs, personal communication). A farmer's herd was
divided into a number of flocks. Each flock was the charge of a shepherd.
A shepherd
moved his flock to and from grazing areas as the quality of grazing dictated. Numerous
small kraals and associated huts were constructed of dry walled stone and these are still
abundant throughout the landscape. The kraals were used to confine sheep at night in
order to protect them from predators, especially cats and hyenas. Rockshelters were
perfect places for kraals, and both Blydefontein and Meerkat shelters have stone walled
kraals.
At the close of the Boer War in 1902 an organization called the Rhodes, Beit &
Bailey Syndicate began to purchase farms in the Hantam in a grand plan to introduce
modern farming techniques such as irrigation and crop farming. Farms, such as
Oorlogspoort and Grapevale, were bought by the Syndicate, but with the death of Rhodes
and Beit eventually the Syndicate was split into private individual farms. Many were
sold to the Syndicate's own managers. With the Fencing Act of 1912 these managers
introduced fencing in the early 1920s. This greatly altered the old system of sheep
farming. No longer were shepherds needed for each flock and no longer could predators
move about so freely. Also the Drought Investigation Commission of 1923 argued that
fenced camps would increase the carrying capacity of the veld (Roux and Opperman
1986: 96). In response to these suggestions sheep stocking levels reached an all time
35
high in i 933 which also was a period of extremely low rainfall, and the veld was more
seriously overgrazed than ever before.
In the last 30 or 40 years a more systematic and research-founded approach to
sheep farming has been gaining popularity in the northeastern Cape. Acocks and Roux
are major advocates of this approach. The philosophy is one of conservation, and the
technique is through rotational grazing of clustered camps. The general strategy is that
the veld must be 'rested' for long periods between grazing sessions. In this way the
heavily grazed, more palatable and more nutritious plants have a chance to grow back
after being eaten. When a camp is managed properly plant diversity and grass density
increase. It is believed that this rotational system is similar to the grazing patterns of
the endemic fauna.
Impacts on the Vegetation
The False Upper Karoo, according to Acocks (1975: 78), represents the most
dramatic of all the European induced vegetation changes in South Africa. Basing his
argument on 18th century travelers' accounts from in the plains below Blydefontein,
and the presence of supposed relict stands of Dry Cymbopogon-Themeda Grassveld in
what is now Karoo, he believed that most of the area now covered by the False Upper
Karoo (e.g. 32,200 square kilometers) was grassveld 170 years ago. He argued that
overgrazing by European stock allowed this area to be invaded by Karoo plants.
First
the grasses were destroyed by overgrazing, then erosion removed the top soil, and
finally the Karoo bossies invaded (Acocks 1975: 78).
Werger (1978b: 446-454,
1980: 27-29} generally agrees with the Acocks' model except that he suggests that
many of the bossies were already present in reduced numbers and were less visible in
the luxuriant northeastern Cape grassland before its destruction by overgrazing. As a
distinct group of supposed Karoo invaders have a distribution restricted to both sides of
36
the eastern Karoo/western grassland boundary, Werger (1978b: 447, 1980: 29}
argues that this is an area long influenced by fluctuations due to climatic changes. Thus
in Werger's model Karoo plants became the dominants by default and this seems to have
occurred from domestic stock overgrazing and climatic fluctuations. Without doubt the
severe erosion of top soil that followed overgrazing has retarded grass recolonization.
Today not all botanist agree with the Acocks-Werger model, but few other models are
available in its stead.
Roux and Vorster (1983) have provided more detail, and elaborated on the
Acocks-Werger model for the eastern Cape (Figure 5). Their study is based on
observed changes in the modern conditions of the Veld Types in the eastern Karoo from
east to west (assuming a gradient of progressive degradation). Five phases have been
identified: 1) primary degradation, 2) primary denudation, 3} re-vegetation, 4)
secondary degradation, and 5) desertified.
The first phase marks the rapid destruction and thinning of the primary plant
communities. The second phase is characterized by extreme devegetation with
increased runoff, high erosion rates, and high sedimentation in catchment basins;
carrying capacity is greatly reduced. Phase 3 represents the closing of the
devegetation lag that created Phase 2, however a slightly different suite of plants is
established. If Werger is correct, then most of these plants were present in the
eastern Cape before destruction by overgrazing.
Different grasses, Eragrostis and Aristida, and other bossies such as Augea,
Chrysocoma, Eriocephalus, Galenia, Lycium, Pentzia, Pteronia, Ruschia and
Zygophyllum, now dominate the landscape. In the mountains the most successful Phase
3 invaders are Elytropappus and Euryops. Phase 3 is characterized by increased plant
density and reduced erosion and runoff. However the loss of soil during Phase 2 would
prevent the primary vegetation communities from returning.
Most of the eastern
37
Karoo is (circa 1985) in a Phase 3 state according to Roux and Vorster (1983), but
veld conditions in the western Karoo suggest further potential deterioration to these
researchers. Phase 4 marks a second episode of degradation due to continued veld
mismanagement, i.e. overgrazing. Taxa typical of this phase are Acacia, Augea,
Asparagus, Eriocephalus, Galenia, Lycium, Pentzia, Rhigozum, Pteronia and Ruschia.
In the mountains, plants such as Dodonaea, Elytropappus, Euclea, Euryops,
Merxmuellera and Rhus gain dominance in Phase 4. The final Phase, 5,
desertification occurs with Aloe, Euphorbia, and Mesembryanthemum as the most
important endemic taxa.
セ@
Phose 5
Phose 4
Phose 3
Phose 2
80
セ@
0
1910
1880
1970
1940
2000
Years AD
[:':'::·::J Perennial sweet gross
セ@
セ@
rェヲセ[@
D
Palatable dwarf shrubs
Unpalatable dwarf shrubs
D
Annual and biennial gross
Annual herbs and ephemerols
Woody shrubs
Figure 5. Roux and Vorster's (1983) model of vegetation change in the eastern Cape
due to overgrazing.
Certainly the Dry Cymbopogon-Themeda Veld was at the edge of the Karoo before
overgrazing by European domestic stock affected the area, and climatically induced
fluctuations in the distribution of Veld Types occurred in the past. Droughts in the
38
northeastern Cape are well known historically (C. Vogel 1988: 11, 1989), and the
response of the grasslands to these droughts was a definite reduction in the quantity of
grass. It is extremely likely that a transition from grassveld to Karoo is a natural
response in the ecosystem to drought conditions or to changes in rainfall seasonality
(Bousman et. al. 1988; Coetzee 1967; Meadows et al. 1987; Meadows and Sugden
1988; Scott 1988a, 1988b; Scott and Bousman 1990; van Zinderen Bakker 1957,
1967, 1982a), and that these two modern plant biomes can be used to model past
Karoo-grassland fluctuations in the Kikvorsberg and surrounding region. At least one
palynologist has suggested that Karoo patches in eastern grasslands represent relict
stands and not invasion (Van Zinderen Bakker 1967: 147). Additionally one problem
with the Roux and Vorster (1983) model is that the predicted phase 4 and phase 5
occur only in areas of the Karoo that are much drier and warmer than the eastern
Karoo, and it is questionable whether the more xeric forms could ever dominate the
eastern Karoo plant communities without a significant change of climate. Today the
Karoo is expanding at an amazing rate, up to 2-3.5km per year in some areas (Werger
1978b: 44 7), but the verdict on the cause(s) is still undecided and the jury may be
forever hung!
Impacts on the Fauna
Although never present in Blydefontein Basin, one of the first species to become
locally extinct was the hippo (Hippopotamus amphibius ). Gutsche (1968: 158)
records that by the 1830s no hippos remained in the Zeekoe River, and one of the last
hippos in the Orange River was killed in 1868.
Large carnivores were removed from
the area as quickly as possible because of their danger to humans and their killing of
stock. Numerous references (Skead 1987) are made about lions (Panther leo) in the
early travelers' accounts, but also present in the mountains and koppies were leopards
39
(Panther pardus ). Cheetahs (Acinonyx jubatus ) were known historically in the
plains below Blydefontein as well as brown hyenas (Hyaena brunnea), spotted hyenas
(Crocuta crocuta), wild hunting dogs (Lycaon pictus ), and aardwolves (Proteles
cristatus ) . During the 1800s many species of antelope were virtually exterminated
locally, and these include wildebeest ( Connochaetes gnou), red hartebeest (Aicelaphus
buselaphus), blesbok (Damaliscus dorcas), and eland (Taurotragus oryx).
Quagga
(Equus quagga ), a horse with fewer stripes than zebras, became extinct by the 1870s
and is now known only through early explorer descriptions and paintings. Mountain
zebra were almost totally exterminated, but a few remained in the western portion of
southern Africa and these were reintroduced to the northeastern Cape with the
establishment of Mountain Zebra National Park in 1937. Other locally extinct fauna
are black rhinoceros (Diceros bicornis ), warthog (Phacochoerus aethiopicus), and
buffalo (Syncerus caffer), although these animals were never very numerous in
historic times.
Summary
Even though wild animals remain in numbers and the vegetation is beginning to
respond to modern management practices, the last 200 years of European land use has
had lasting and possibly irreversible consequences. Many animals are locally extinct,
one is completely extinct, and all remaining animals have had to adapt to the
overwhelming presence of humans. Loss of soil due to erosion is a significant factor
affecting modern plant distributions. The above conditions present serious problems
for using modern environmental data to reconstruct past environments. Nonetheless,
as is argued in the following chapters, certain approaches may be used to gain an
understanding of past botanical and faunal communities, and the climates that
influenced and controlled these communities. The environmental chapters that follow
40
will present a paleoclimatic model, attempt to form bridges between present
environmental patterns and modern climate, and then test the climatic model by using
the modern climate/plant relationships to reconstruct paleoclimates with the fossil
pollen data. Finally the adaptations of prehistoric human inhabitants can be assessed
against the dynamic past environmental background.
CHAPTER Ill
A REGIONAL MODEL OF LATE QUATERNARY AND RECENT CLIMATIC CHANGE
Although paleoecological data for the Upper Pleistocene and Holocene of southern
Africa are accumulating rapidly (e.g. Deacon and Lancaster 1988; Vogel 1984), these
are still insufficient to provide a regional framework within which to fit new, local
sequences like the one to be presented in this dissertation. This is especially true of
the Interior Plateau where the paleoecological record is almost nonexistent (Deacon and
Lancaster 1988: 56-59).
In addition, the inductive fitting of local data to such
chrono-stratigraphic frameworks is of somewhat limited use since it only describes
rather than explains the nature and timing of the climatic changes being examined.
In this dissertation I have opted for a deductive approach to the analysis by
starting with a mechanistic model of climate change, to be tested in the chapters that
follow. This general model is derived from existing computer simulations. They in
turn were built from current models purporting to explain the global mechanics that
drive the modern climate of the world and thus of southern Africa. What follows,
therefore, is a brief review of the modern circulations patterns now thought to control
today's climate in southern Africa, so that the logic behind the simulation programs can
be better understood. After the simulation results are described, historically recorded
rainfall fluctuations are briefly reviewed.
The General Circulation Systems
These are the basic building-blocks of the computer simulation. South African
climate is controlled by the interaction of two circulation systems, one flowing northsouth and the other flowing east-west (Hurry and van Heerden 1984; Trewartha
41
42
i 968; Tyson 1986). Both are complex chains of interacting convection cells, linked
in turn to oceanic circulation (Cane 1983; Rasmusson and Wallace i 983; Tyson
1986). The north-south system is called Hadley and Ferrel circulation, for its
constituent cells of those names. The east-west system is called Walker Circulation.
North-south circulation is caused by the fact that the equator receives more solar
energy than the south pole, so heat is wind-transfered from low to high latitudes.
Transfer takes place in a series of circulation cells. Starting in the tropics, where the
earth's spin has less effect, circulation is fueled by low pressure cells that link
together to form the low trough called the Inter-Tropical Convergence Zone (ITCZ).
Over Africa this grows a southwesterly extension called the za·ire Air Boundary where
warm air rises and is pushed southwards (Figure 6a). Once cooled, it subsides again to
create a high pressure ridge called the Subtropical Ridge. Over the oceans on either
side of southern Africa are high pressure cells that are linked to the Hadley-Ferrel
circulation.
These two oceanic high pressure cells shift their positions seasonally, following
the seasonal, north-south movements of the ITCZ. The oceanic cell on the east side of
southern Africa (Figure 6b), called the South Indian Ocean High (SIOH), is more
mobile than its Atlantic counterpart, especially in its east-west wanderings toward and
away from Africa (Bryson and Murray 1977: 102; Tyson 1986: 96). Feeding into
these oceanic high pressure cells is upper level circulation from the circumpolar
Ferrel cells linked to form a low pressure trough called the Antarctic Trough (see
Figure 6a). As the effect of the earth's spin increases pole-wards, these cells create
strong westerly air flow known as the Westerlies.
The Westerlies have semi-stationary meanders that control polar storm paths, and
any changes in meander position, strength or degree have an impact on South African
weather (Tyson 1986: 147-158).
During winter the Westerlies track expands away
43
from the pole and their meander amplitudes grow, pushing cold fronts northwards from
their normal polar positions.
In the summer the entire circular Westerlies path
contracts around the south pole, drawing moist tropical air down during summer.
The east-west (Walker) circulation is linked to the patchy distribution of land and
sea across the southern hemisphere, and is more complex in that it has two states, one
is normally in position and the other is episodic. Both states are summarized in Figure
7. The normal state, called High Phase, causes subsiding air in the West Indian Ocean
to feed into the South Indian Ocean High to the east of southern Africa. When
southeasterly trade winds from the SIOH reach the subcontinent they are forced
upwards over the Za"ire Air Boundary, and made to recurve south. Thus on the surface
Interior Plateau summer rains often appear to be delivered from the northwest. The
high pressure cell on the Atlantic side of southern Africa feeds moist air into lows over
the African interior, bypassing the southwest Cape. The interior low pressure cells
fuel summer rainfall over the eastern parts of southern Africa only.
The episodic state of Walker Circulation, called Low Phase, is also known as the
Southern Oscillation and linked to El Nino events off the west coast of South America. It
occurs about every 3.5 years and lasts about 18 months (Barber and Chavez 1983;
Cane 1983; Rasmusson and Wallace 1983; Tyson 1986: 168). When the Pacific
cells divides into two, a chain of cell reversals occurs with the end result that over
Africa the interior low pressure cells on the eastern side change to highs, and new lows
form over the west. This has a notable effect on rainfall patterns, and Tyson (1986:
166-175) has argued that Walker Circulation accounts for about 20 percent of
rainfall variance in the interior, and there is partial evidence linking interior
droughts with El Nino events, but the mechanics of the connection are still not
understood well.
44
(
0
Hadley
Cell
I
\
0
a.
'-
b.
0 Atlantic Ocean High Pressure Cell
D Indian Ocean High Pressure Cell
!::::. Blydefontein Basin
UKMセ@
0 ............................l .................................................................................................................................................
-5
<II
セ@
-10
"'0
......
j
-15
-20
5
a=f
セZ@
9
V
12
セQR@
3
MTPKセイN@
-20
0
20
40
60
80
Longitude
Figure 6. General north-south circulation systems in Southern Hemisphere; a:
circulation cells and summer convergence zone positions; b: monthly movements of
oceanic high pressure cells (after Tyson 1986).
100
45
HIGH PHASE
t???Z????????A
South
t?????????l
Africa
Indonesia
Africa
Indonesia
America
LOW PHASE
South
America
Figure 7. Walker Circulation in High and Low Phases over Southern Hemisphere
(after Tyson 1986).
Now that the working parts of the mechanism are in place, the consequences of its
motions for southern African weather can be reviewed.
Circulation Systems and Rainfall Patterns in Southern Africa
The dynamics of a mirror-image relationship are diagrammed in Figures 8 and 9.
Wetter conditions in the Interior Plateau occur when the South Atlantic High Pressure
Cell is strengthened and also when the upper level Westerlies display greater meander
46
amplitude. Under these conditions, moisture flowing with the South East Trades is
steered south over central and southern Africa where cloud bands often form. When
this happens, a low pressure cell forms south-center of the cloud band (about 20°S),
aided by the ascending arm of the Walker Circulation, and a high pressure cell forms
off the southern coast at about 30°S. This in turn displaces the upper level Westerlies
southwards although it increase their velocity, thus winter storms are denied access to
the coast and the southwest Cape experiences drier winters while the Interior Plateau
usually is wetter.
Inversely, drier conditions in the interior are induced by weakening the
Westerlies and the South Atlantic High Pressure Cell. Now, a descending arm of the
Walker Circulation is positioned over the interior, thus impeding the formation of any
interior low pressure cells. The amplitude of meanders in the Westerlies are reduced,
and this leads to the Westerlies meander being shifted to the east, causing cloud bands
over the South Indian Ocean rather than the subcontinent. These in turn inhibit the
growth of strong highs off the southern coast. With weaker lows inland and weaker
highs offshore, the amount of interior rain is dramatically reduced. The weakened
Westerlies are displaced north, however, so winter storms reach the southern coast
and increase winter rain in the southwest Cape with some even reaching the normally
dry interior.
Tyson (1986) has proposed that the causal mechanics of these modern weather
patterns could be used to model climatic variations in the past. Since the inverse dry
or wet conditions of the Interior Plateau and the southwestern Cape are controlled by
the changes in positions and strengths of these two general circulation systems,
especially in the upper troposphere, it would be interesting to see what might happen
to those systems during episodes of global cooling and rewarming.
The Computer Simulations
47
It is now apparent that three atmospheric circulation systems influence the modern
climate of southern Africa, including the local climate of the Blydefontein region (see
Appendix 1 for details of the latter). The Westerlies meander path and the Southern
Oscillation both control winter temperature and rain, while the Southeastern Trades
control summer temperature and rain. Computer simulations allow these three
systems to respond to global temperature changes computed from the principles of
Milankovitch forcing. The first step in simulation is to model the above system
interactions, then to superimpose them on a baseline of global temperature
fluctuations.
It is well known that annual and seasonal solar radiation budgets have changed
significantly during the Late Pleistocene and Holocene (Berger 1978; Hopkins 1981
in Kutzbach 1981; Milankovitch 1930). Several computer simulations of past
climates based on different solar radiation conditions have been attempted (see
Schneider 1987 for general discussion; and Gates 1976; Kutzbach 1981; Kutzbach
and Guetter 1984, 1986; Man abe and Hahn 1977; Williams et al. 1974 for more
detailed studies). Most of the climatic simulations are based on changes in the radiation
budget due to variations in three of Earth's orbital parameters: orbital eccentricity,
axial tilt, and axial precession (see Imbrie and Imbrie 1979 for a history of
discoveries). The respective periodicity of these three cyclic variables are given as
100,000 years, 41 ,000 years and 22,000 years. The interaction of these three
orbital parameters with their different periodicities are now thought to account for the
major climatic changes observed in the Pleistocene and Holocene (COHMAP Project
Members 1988; Hays, Imbrie and Shackleton 1976; Kutzbach 1981 ).
WET
48
La-ger crnplitude
upper-level
Atlantic wave
Claud bands aver
southern Africa
r--"...
'---V
DRY
Southward shift of storm
tracks; stronger storms;
south-western Cope
winters drier
smaller crnplitude
upper-level
Atlantic wove
Cloud bonds over
Modogosccr
]セァオエィ・イョN@
t::::> C> c:>
. ure 8
セ@
セ@
t::::> L:::>
セ@
Northward shift of storm
tracks; weaker storms;
south-western Cope
winters wetter
..
Wet and dry climatic conditions
an d meridonial atmospheric circulation in
Africa (after Tyson 1986}.
49
WET
Ascending limb of
Walker cell over
tropical Africa
Strengthened subtropical
jet particularly to the east
DRY
Descending limb of
.\-.-----.4----. Walker cell over tropical Africa
1--1.----\
Weakened subtropical
jet particularly to the west
Figure 9. Wet and dry climatic conditions and zonal atmospheric circulation in
southern Africa {after Tyson 1986).
50
Of computer simulations that cover the last glacial maximum (COHMAP Project
Members 1988; Gates 1976; Kutzbach and Guetter 1986; Manabe and Hahn 1977;
Williams 1974}, all show significant decreases in temperature in southern Africa, but
they differ in details about the amount of rainfall. The last glaciation was stimulated by
a reduction in obliquity, thus reducing summer temperatures, an increase in
eccentricity, and perihelion (the point where the earth is closest to the sun} occurred
during the Northern Hemisphere's winter (Berger 1978}. These factors acted in
concert with the negative feedback mechanism of increased reflection of solar radiation
by snow to reduce the amount of summer snow melt which led to the growth of glaciers.
Kutzbach and Guetter (1986} present a series of climatic simulations calculated
every 3000 years from 18 thousand years ago (kya} so that simulations are available
for 18kya, 15kya, 12kya, 9kya, 6kya, 3kya and Okya. They suggest that for the
Holocene at 9000 B.P., orbital parameters differed most from modern conditions.
When compared to today, at 9000 B.P obliquity or tilt was 0.38 degrees greater,
eccentricity was slightly greater or more elongated, but most importantly perihelion
was during the Southern Hemisphere's winter (30 July} while today it is in the
summer (3 January}. Given these differences winter would have received more solar
energy and summer less at 9000 B.P. than today. Even though the annual radiation
budget at 9000 B.P. does not differ significantly from today's, all simulations (both a
low resolution General Circulation model and a high resolution National Center for
Atmospheric Research-Community Climatic Model} show a seasonal climate in
southern Africa at 9000 B.P. that was more equitable than today with warmer winters
and cooler summers (COHMAP Project Members 1988; Kutzbach 1981; Kutzbach and
Guetter 1984, 1986}.
Furthermore, increased monsoonal circulation in North Africa,
caused by stronger pressure cells in the Northern Hemisphere, actually sucked
51
Southern Hemisphere wind and water across the equator (Kutzbach and Street-Perrott
1985).
COHMAP Temperature and Precipitation Simulations
John Kutzbach and a group of collaborators on the Cooperative Holocene Mapping
(COHMAP) Project have run simulations on the high resolution NCAR-CCM model for
each 3000 year point from 18,000 B.P. until present with run times of 450 days and
global ice distributions based on CLIMAP (1981). The NCAR-CCM model covers the
world with grids, and five grids cover the southern portion of Africa (Figure 10).
These five grids include most of Namibia and Botswana, southern Zimbabwe and
Mozambique, and all of South Africa. This is a large heterogeneous area that has deserts
and rain forests, and winter and summer rainfall areas. However the NCAR-CCM model
is of global scale so that isolation of individual environments is not possible, and
greater spatial resolution would sacrifice the reliability of the predictions.
Dr.
Kutzbach has generously made available these unpublished simulated data for Africa
south of 20°s latitude (Figures 11, 12 and 13).
Figure 10. Computer simulation grids used in COHMAP simulations.
52
Annual
Temperature
Difference
from modern
(degrees C)
Annual Precipitation
Difference from
Modern (mm)
0
-1 . o
_ .
1 2
-50
-1.4
-1 00
-1 . 6
o--
MQNXKイセ@
0
9
12
Years BP (x1 000)
6
3
·0- Annual Temperature
-150
18
15
·•- Annual Precipitation
Figure 11. COHMAP Project high resolution NCAR-CCM model simulated annual
temperature and annual rainfall differences from modern for southern Africa.
Annual
Temperature
Difference
from modern -1 . 0
(degrees C)
-1.5
·---·
-2.0
0
3
6
9
12
15
18
Years BP (x1 000)
·•- Summer
·0- Annual
·•- Winter
Figure 12. COHMAP Project high resolution NCAR-CCM model simulated annual,
winter and summer temperature differences from modern for southern Africa.
53
·-
200
Rainfall
Difference
from modern
(mm)
·---o
-·
-1 00
0
3
9
6
12
Years BP (x1 000)
1·0- Annual
·Ill- Winter
15
·•- Summer
18
I
Figure 13. COHMAP Project high resolution NCAR-CCM model simulated annual,
winter and summer rainfall differences from modern for southern Africa.
The simulated annual temperatures indicate that the climate was coldest from
18kya to 15kya, and rapidly warmed between 15kya and 12kya (see Figure 11). The
estimates for 18kya and 15kya are only 2 degrees Celsius lower than the modern
annual temperature, and many researchers would argue that the height of the
temperature during the Last Glacial Maximum was much colder in southern Africa
(Talma and Vogel in Deacon and Lancaster 1988; Talma et al. 1974; Vogel 1983).
After 12kya the rate of warming declines, and the point of maximum temperature is
6kya.
From 6kya to 12kya winters are significantly warmer than modern winters,
and summers are significantly cooler than modern (see Figure 12). Thus the climate
is more equitable than the modern climate, and it is possible that conditions were
similar to the dry climate conditions presented by Tyson (1986). The simulations
54
suggest that at 3kya annual temperature dropped, but the seasonality of the climate
approaches modern levels with colder winters and warmer summers.
Annual rainfall simulations suggest that very high levels occurred at 18kya and
15kya, but a dramatic drop followed 15kya (see Figure 11 ). Between 12kya and 6kya
rainfall increased slowly, but estimates for 6kya are still well below modern levels.
The simulations suggest that a significant increase between 6kya and 3kya, and that the
rainfall levels at 3kya was well above the modern level. Simulated rainfall seasonality
is revealing (see Figure 13). The two periods of high rainfall (15-18kya and 3kya)
also have high summer rains. One period (6kya) demonstrates a significant drop in
summer rains which increases the importance of winter rains, however summer rains
are still higher than winter rains at this point. At no point are winter rains as high as
the modern simulations.
It is a pity that these simulations have such low spatial resolution, but at least
they are based on modelled atmospheric processes (reviewed above) and on known
variations in solar radiation.
Even though they cannot predict short-term
fluctuations, they do provide a testable sketch of major climatic changes for the Late
Pleistocene and Holocene.
The Short-term Rainfall Oscillations
The computer simulations do not model short term climatic fluctuations on the
order of 1,000 years or less, however, if present, some of these fluctuations might be
expressed in proxy paleoclimatic data such as pollen. A brief review of three rainfall
records provides local evidence of short term fluctuations. The first record is derived
from historic observations from the Eastern Cape spanning the years of 1821-1900
(C. Vogel 1988, 1989). A second record is a measured record from Aliwal North and it
spans the years 1884-1940. A third record is the nearby Grapevale record (see
55
Chapter II) and it spans the years 1920-1985. Taken together these records provide
evidence of rainfall fluctuations over 165 years.
C. Vogel {1988: 11, 1989) published a rainfall record for the Eastern Cape that
dates from 1821 until 1900 (Figure 14). In her rainfall histogram each year is
classified as above or below normal, and three rainfall 'states' were recognized in each
category.
4
3
Rainfall
above
or
below
norma 1
2
0
KキN[mョエセイヲゥwia⦅L@
-1
-2
-3
MTKセイ@
1820
1830
1840
1850
1860
1870
1880
1890
1900
Year A.D.
Figure 14. Eastern Cape rainfall variability from 19th Century historic records
(after C. Vogel 1988: 11).
The Grapevale and Aliwal North rainfall records are best viewed as moving
averages (Figure 15). Five year periods were selected and these show the short term
variations, while reducing the large year to year fluctuations. The period of overlap
between Aliwal North and Grapevale is 1920-1940, and they both indicate low
rainfall amounts in the late 1920s.
56
1000
'E
.s
(/)
.& Aliwal North
OGrapevale
900
800
Q)
セ@
.....
700
Ol
600
Q)
セ@
c
·;:;
セ@
0
500
.....
ro
Q)
>1.0
400
300
200
1880
1900
1920
1940
1960
1980
200(
Years AD
Figure 15. Five year moving averages of Aliwal North and Grapevale rainfall.
Aliwal North is near the foothills of the Drakensberg, and approximately 165 km
east of Blydefontein and Grapevale. Rainfall at Aliwal North appears to be
systematically greater because of its more eastward location (see Chapter IV). The
average difference between Grapevale and Aliwal North (337.7 mm) was used to
standardize the Aliwal North record to the Grapevale amounts. Then 5-year moving
averages of the two records were combined and averaged for the years of overlap,
i.e.1920-1940 (Figure 16). In order to compare the East Cape record to the
combined measured rainfall records and eventually to the proxy paleoclimatic data
(see Chapter VII), a single year's rainfall classification was treated as an actual
rainfall measurement. This was coded by the length and sign of the bar on C. Vogel's
histogram so that each year was assigned a positive or negative 1 , 2 or 3. Then 5year moving averages of these numbers were calculated and plotted as rainfall scores
(see Figure 16).
The combined Aliwal North/Grapevale record shows an amazing amount of
agreement with the East Cape historic record by a high rainfall estimate in the mid-
57
1880s followed by a rapid drop in rainfall thereafter. Taken as a whole these records
can be used to suggest that long and short, albeit irregular, cycles do exist in rainfall
patterns, and that long cycles might be as long as 90 years and span the period from
1890-1980.
It might be possible to calibrate the historic rainfall records with the
Aliwal North and Grapevale records, and the two scales on Figure 16 provide a rough
indication of the approximate conversion. The patterns presented here are supported
by another historic rainfall record where a major drop in rainfall is documented for
the 1890s further to the west on the edge of the Karoo region (Venter et al. 1986).
2.0
• II
1 .5
0.5
Eastern
Cape
0.0
Rainfall -0.5
Scores -1.0
-1 .5
-2.0
セ@
Ut,
1 .0
!fl
a·-"Il.
o
I
Oil
O<QO 0 0
0
0
0
co\c<&
0'1
o1セッ@ o
l0o<0<>/
セ@
600
..•
0
00.
l0po!I
b セ\。u「@
II Yセ@
<•
• •
.,j
\aBGセエ@
iセ@
I
550
500
I
450
\.
400
t(,,. .,.
I
o セM@ (•,, t.
I
N@セ '
セD@ I\•
•I セ@N It
• • ᄋセ@ • ..
•
•
i
P@ oセ@ セ@ j .( セ@
I
C<O •
r,o o.o1
a0
f
°
セ
\itt
セᄋ@
セN@
1 o.. •
0II
••
0
l!
セL@
•
350
300
250
Aliwal
North &
Grapevale
Rainfall
(mm)
2oo
150
0
-2.5
100
1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000
Years AD
·0- East Cape-5yr Mev. Ave.
·•- Aliwal North & Grapevale-5yr
Mev. Ave
Figure 16. Measured and observed rainfall estimates for the eastern Cape.
An even shorter oscillations show up clearly in the Grapevale annual rainfall data
(Figure 17). As the minima show, there are drought cycles of 17-18 years duration.
This 18-year cycle is well documented throughout southern Africa, and in many areas
they account for 20-30 percent of rainfall variance (Tyson 1986: 68-80).
When the
58
summer rainfall totals, i.e. the high ends of oscillations, are extracted from the
Grapevale record they magnify the 18-year oscillation. Tyson (ibid: 160) suggests
that the high ends may be associated with a situation in which there is a trough of lows
along the west coast and the South Atlantic High is about 2500km to the southwest near
Gough Island.
Clearly, forcing mechanisms other than those incorporated in the simulations
must be sought to explain these phenomena. These are interesting because high
resolution proxy paleoclimatic evidence might capture some of these short-term
oscillations. Additionally, one should remember these short fluctuations when
considering hunter-gatherer adaptations to past environments.
900
800
,...,.
700
E
E
'-/
600
.....
セ@ c
·:;;
0::
セ@
.....
0
1-
500
-
400
300
200
100
1910
ll1
1920
II
1930
I
II
1940
II II
I
1950
Year
1960
1970
1980
1990
Figure 17. Yearly rainfall at Grapevale from 1921 to 1984.
Conclusions
In this model, rainfall estimates for 18kya and 15kya are much higher than those
suggested by some paleoecological studies (Deacon 1984; Deacon et al. 1984; Scott
59
1984; Tyson 1986 and many others), but they are similar to those of other
simulations (Gates 1976; Williams et.al. 1974) and some proxy paleoclimatic data
(Grindley 1979; Kent and Gribnitz 1985; Lancaster 1979). As the temperature
gradient in the Southern Hemisphere is greater than that for the Northern Hemisphere
(Pittock 1978: 4), it is reasonable to presume that during the Last Glacial Maximum
this temperature gradient was great enough to drive wind circulation systems with
enough force to actually increase rainfall. Also, the simulated temperatures are
warmer than is generally believed (Deacon 1984; Deacon and Lancaster 1988;
Deacon et al. 1984; Scott 1984; Tyson 1986; Vogel 1983).
The drop in summer rainfall in the terminal Pleistocene, and in early to mid
Holocene times suggests that reduced temperature gradients (caused by warmer
winters and cooler summers) were too weak to drive the summer rainfall system at
modern levels. The implication is that the the South Indian Ocean High Cell and the
za·ire Air Boundary did not deliver the same quantity of moisture to southern Africa as
it does today. It is possible that the Low Phase of the Walker Circulation was the
normal condition during these times.
The shift to modern seasonal dynamics (i.e. warmer summers and cooler winters)
by at least 3kya and probably by Skya would increase the strength of circulation
systems and thus elevate rainfall, especially summer rainfall (Kutzbach 1981 ).
Simulated winter rainfall amounts do not change significantly from 18kya until Okya.
Support of the simulated shift to more marked seasons in the Late Holocene is that the
length of the Antarctic ice season and thus the amount of Antarctic pack ice is strongly
related to solar insolation in late spring and early summer (Kukla 1978: 127).
Increases in summer solar radiation would reduce the amount of Antarctic ice. World
wide Late Holocene sea level high stands suggest that warmer Southern Hemisphere
summers caused increased Antarctic pack ice breakup and melting which raised ocean
60
levels throughout the world (Simmons et. al. 1981: 83-89; Yates et. al. 1986: 164165). Cooler temperatures are reconstructed for the Northern Hemisphere and would
not have been responsible for the Late Holocene sea level rises (Gribbin and Lamb
1978: 68-82;
Simmons et. al. 1981;
Wright 1983).
Although over-generalized in both space and time, the COHMAP simulation
nonetheless affords the best available explanatory model for paleoclimatic change in the
southern African interior. Before it can be made truly testable, however, it must be
manipulated into an alternative form that makes it suitable for direct comparison with
proxy paleoenvironmental data.
CHAPTER IV
CONSTRUCTING A REGIONAL MODEL OF VEGETATION CHANGE
In this chapter the modern botanical environment is searched for suitable analogs
by which to interpret Holocene pollen spectra from the Blydefontein region. Using
new, quantified botanical survey data, it is shown that the widely used Acocks Veld Type
system is fatally flawed. By re-grouping the new quantified data with cluster analysis,
eight revised plant communities for the Blydefontein area are presented. These have
the advantage that they are dominated by plant taxa with identifiable pollens. Next, the
climatic implications of dominant taxa are explored, thus making it possible to model
botanical changes based on the COHMAP Simulation model of climate change.
Analysis of the Modern Plant Mosaic
Today, the major dichotomy in plant communities of the region is one between
Karoo scrub (Compositae) and grassveld (Gramineae). Altogether, 84 modern plant
surveys have been completed in the region (Roux and Blom 1979). These surveys use
the Acocks Veld Types (see Chapter II) as a starting point, then measure the basal cover
percent of species by three similar methods: wheel-point, chain-point and descending
point. The sample population includes some 228,750 plants from the 84 individual
surveys, making this one of the largest quantified botanical studies in southern Africa.
This analysis begins by using all 84 Roux-Biom botanical surveys, then 73
surveys (from seven Veld Types) near Blydefontein are selected for detailed analysis.
First, species were regrouped into palynologically diagnostic taxa. Once grouped into
the palynological taxa, the percent-of-basal-cover measurements for each taxa in each
individual survey were transformed into relative frequencies of basal cover.
61
Relative
62
frequencies based on number of plants was not possible. However, it seems reasonable
that a higher correlation would exist between basal cover and pollen production than
between number of plants and pollen production. An analysis of the relative
frequencies can provide an assessment of how well pollen analysis can identify the
Acocks Veld Types. Descriptions of Acocks Veld Types near Blydefontein are given in
Chapter II.
The Acocks Veld Type System Re-examined
In southern Africa virtually all archaeologists plus some botanists and
palynologists rely heavily on the Acocks Veld Type system as a framework for
predictive modelling. It has become so entrenched that users tend to forget that Acock's
pioneer work was a brilliant and intuitive effort, but not based on quantitative data.
Now that the Roux-Biom surveys make those data available for the first time in the
eastern Karoo, it becomes possible to evaluate the quantitative reality of these intuitive
groupings. Dr. Piet Roux, Grootfontein Agricultural College and Research Station,
generously provided me with the raw data gathered from 12 of the Acocks Veld Types.
There are eight Karoo scrub types: False Upper Karoo (FUK), Central Upper Karoo
(CUK), Karroid Broken Veld (KBV), Arid Karoo/False Desert Grassveld (AK/FDG),
Central Lower Karoo (CLK), False Succulent Karoo (FSK), Noorsveld (NV) and Orange
River Broken Veld (OBV). The four grassvelds are: Karroid Merxmuellera Mountain
Veld (KMMV), Stormberg Plateau Sweetveld (STM), Dry Cymbopogon-Themeda Veld
(DCT), Cymbopogon-Themeda Veld (CTV) and the Cymbopogon-Themeda Veld/ThemedaFestuca Alpine Veld (CTV/TFAV). Their compositions and backgrounds are reviewed in
Chapter II. As discussed in Chapter II, the reader should be aware that the modern
vegetation patterns are a result of complex interactions between domestic stock
overgrazing, edaphic, and climatic patterns. Nevertheless, the Acocks Veld Types are
63
based on the modern distribution of plants, and the Roux-Biom botanical suNeys
provide quantitative data on these plant communities. It is suggested here that the
Roux-Biom botanical suNeys are an adequate representation of the range and
associations of plants that occurred during the Holocene in the eastern Cape.
The most basic assay of these Veld Types is to plot of relative frequencies of all
scrub species (Compositae) against all grasses (Gramineae) for each suNey. When
this is done (Figure 18) it is obvious that a rough binomial relationship exists
between these two plant taxa: as the relative frequency of one increases, the other
decreases in proportion. This is especially true for the False Upper Karoo and the
Central Upper Karoo versus the grassvelds. However, this binomial characterization
is not as accurate for the remaining Karoo Veld Types. It emerges, therefore, that the
Orange River Broken Veld, the Arid Karoo, the Noorsveld, the False Succulent Karoo,
the Central Lower Karoo and the Karroid Broken Veld form a single western group of
plant communities with an ecology that is structurally different from an eastern group
of Veld Types.
Clearly, there are relatively fewer Karoo composites in the western group than
there is in the eastern group, and grass percentages decrease at a slower rate in the
western group. These relationships are represented by linear regression formulas for
each group:
Eastern Group Compositae %
r2 =
-0.724 (Gramineae %) +71.123
0.762
Western Group Compositae %
r2 =
=
=
-0.494 (Gramineae %) +45.774
0.789
I should state at this point that these statistics are not intended to be used for
predicting, but rather only describing the differences between the two groups. One
64
100
90
Compositae90
80
70
60
50
40
X
0
Eastern
セON@
エセク@
セ@
30
20
10
X X
0
0
•
II.
-
10
20
30
Or.:mge
River
Broken
Veld
0
Arid Karoo
Cenkal
Upper
Karoo
セ@
False
Upper
Karoo
False
Succulent
Karoo
Noorsveld
•
40 50 60
Gr amineae 90
•
70
80
90
100
Central
Lower
Karoo
D
Karroid
Broken
Veld
X
Karroid
Merxmuellera
Mt. Veld
X
Dry
Cymbopogon
Themeda
0
Cymbopogon
Themeda
Veld
Figure 18. Relative frequency of Compositae plotted against relative frequency of
Gramineae for eleven Veld Types (data from Roux and Blom, 1979}.
might expect the relationships to be reversed, i.e. more composites in the drier
western Karoo, but it is the greater number of plant forms adapted to more xeric
conditions, that distort western Compositae/ Gramineae ratios. Although, Gramineae
can occur in high frequencies in the western group, when they decrease, the composites
and xeric forms increase at more or less equal rates. Also many grasses in the western
65
Karoo group are annuals that germinate rapidly after unpredictable rains, while many
of the grasses in the eastern group are perennials. The latter presumably respond to
changes in rainfall more slowly (Acocks 1975; Seely 1978).
Obviously, there are scrub-to-grassland continua in both east and west, and not a
western and eastern Karoo-to-grassland dichotomy. This causes one to wonder if the
Veld Types themselves are not parts of the same continua rather than discrete entities,
as originally conceived. Of course the same question has been aired before by botanists
(Campbell et al. 1981: 4), but this is the first time that quantitative data have been
brought to bear on the problem. The new data are sufficiently promising to warrant
further analysis.
Analysis of Local Plant Communities
The next step was to check the integrity of local Acocks Veld Types through
discriminant function analysis.
First, the relative frequencies of four taxa
(Gramineae, Compositae, Chenopodiaceae/Amaranthaceae, and Aizoaceae/Ruschia) from
the Roux-Biom surveys were grouped by Veld Type. These taxa were selected because
they are the most common taxa in both the Roux-Biom surveys and also in the fossil
pollen spectra, to be examined later. Also these are the only taxa for which modern
pollen rain data are available. The data were entered by Veld Type into a discriminant
analysis which reclassified the individual Roux-Biom surveys based on the groups'
discriminant functions. A BIOSTAT II Version 2.0 discriminant analysis for Macintosh
computers was used with an association matrix based on euclidean distances (Pimentel
and Smith 1985). At the outset it was expected that the discriminant analysis would
support the Veld Type groupings. Seven predefined groups, (KBV, CLK, CUK, FUK, CTV,
STM, and MER), were used. The reclassification of the same individual Roux-Biom
surveys resulted in 43 errors out of 73 classifications (Table 1). This means that
Table i.--Discriminant analysis scores of individual Roux-Biom {1979) botanical 6 6
surveys by Veld Type. Highest score of each underlined and * indicates a classification
error. KBV: Karroid Broken Veld, CUK: Central Upper Karoo, CLK: Central Lower
Karoo, FUK: False Upper Karoo, CTV: Cymbopogon Themeda Velds, STM: Stormberg
Plateau Sweetveld, MER: Karroid Merxmuellera Mountain Veld.
Individual
Surveys
KBV 10
KBV 11
KBV 33
CUK15
CUK16
CUK17
CUK18
CUK19
Errors
*
*
*
CLK 12
CLK 13
CLK27
CLK28
CLK29
CLK 31
*
*
FUK 21
FUK22
FUK35
FUK36
FUK38
FUK45
FUK46
FUK47
FUK48
FUK49
FUK50
FUK 51
FUK52
FUK53
FUK54
FUK55
FUK58
FUK 60
FUK62
FUK 64
FUK 65
FUK 71
FUK77
FUK 78
FUK 81
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
KBV
CUK
Veld Type Discriminant Scores
CLK
FUK
CTV
STM
1.00
1.00
0.00
0.00
0.00
0.03
0.01
0.00
0.01
0.00
0.00
0.77
0.17
o..:u
0.00
0.00
0.00
0.00
0.00
0.00
0.09
0.17
0.00
0.01
0.01
0.03
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.03
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.05
0.00
0.00
0.37
0.14
0.06
0.06
0.20
0.37
0.10
0.11
0.07
0.05
0.00
0.08
0.05
0.10
0.20
0.07
0.29
0.06
0.45
0.22
0.04
0.05
0.36
0.05
.Q....2.l
0.21
0.08
0.00
0.00
MER
Q....2.2.
0.00
0.00
0.14
0.00
0.00
0.03
0.00
0.00
0.01
0.00
0.00
0.13
0.01
0.02
0.01
0.03
0.02
0.12
.Q...2l
0.12
0.31
0.19
0.03
0.17
0.03
0.06
0.24
0.02
0.16
0.02
0.04
0.25
0.05
0.21
0.05
0.21
0.25
0.05
0.05
0.99
0.30
0.27
0.00
0.02
0.09
0.09
0.12
0.13
0.00
0.00
0.03
0.05
0.08
0.10
0.00
0.00
0.01
0.03
0.36
0.29
0.00
0.01
0.10
0.05
0.33
0.30
0.26
0.24
Q...ll
0.29
0.30
0.23
0.20
0.14
0.00
0.18
0.14
0.22
0.30
0.22
0.04
0.04
0.13
0.19
0.13
0.09
0.15
0.21
0.24
0.29
0.00
0.25
0.28
0.22
0.13
0.06
0.06
0.27
0.03
0.06
0.24
0.30
0.03
0.08
0.23
0.02
0.02
0.08
0.17
0.11
0.07
0.13
0.22
0.25
0.34
0.00
0.26
0.33
0.23
0.11
0.03
0.04
0.30
0.02
0.03
0.24
. 0.36
0.02
0.05
0.25
0.19
0.33
0.40
0.23
0.24
0.16
0.28
0.22
0.23
0.17
0.00
0.22
0.19
0.21
0.23
0.32
0.23
0.20
0.15
0.28
0.28
0.16
0.09
0.27
0.22
.Q.JlQ.
0.76
0.75
0.05
0.18
0.06
0.10
0.02
0.02
0.05
0.01
0.01
0.01
QJ!l.
0.01
0.01
0.01
0.02
0.30
0.04
0.01
0.05
0.06
0.02
0.01
0.02
0.04
0.01
セ@
0.16
0.30
0.36
0.18
0.13
0.15
0.35
0.20
67
Table 1. (continued).
Individual
Surveys
CTV 59
CTV 80
CTV82
CTV 83
CTV 84
STM
STM
STM
STM
STM
STM
STM
STM
Errors
*
*
*
*
*
66
67
68
72
74
75
76
79
MER20
MER23
MER24
MER25
MER 26
M/F 30
MER32
MER37
MER39
MER40
MER41
MER42
MER43
MER44
MER 56
MER 57
MER 61
MER 63
MER69
MER 70
MER 73
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
KBV
CUK
Veld Type Discriminant Scores
CLK
FUK
CTV
STM
0.00
0.00
0.00
0.00
0.00
0.05
0.05
0.07
0.09
0.05
0.01
0.01
0.02
0.01
0.01
0.14
0.13
0.22
0.21
0.16
0.29
0.29
0.21
0.23
0.27
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.05
0.05
0.05
0.05
0.06
0.05
0.05
0.07
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.13
0.12
0.13
0.14
0.17
0.13
0.15
0.17
0.30
0.30
0.30
0.28
0.26
0.29
0.28
0.26
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.33
0.01
0.06
0.21
0.05
0.07
0.10
0.03
0.08
0.09
0.12
0.27
0.67
0.24
0.06
0.05
0.05
0.05
0.06
0.05
0.05
0.07
0.77
0.01
0.02
0.01
0.01
0.04
0.47
0.12
0.04
0.06
0.03
0.02
0.03
0.01
0.02
0.01
0.01
0.01
0.01
0.01
0.34
0.07
0.18
0.32
0.13
0.18
0.30
0.08
0.28
0.30
0.31
0.34
0.19
0.32
0.14
0.21
0.14
0.14
0.17
0.13
0.15
0.04
0.04
0.26
0.11
0.29
0.25
0.13
0.12
0.08
0.12
0.09
0.07
0.03
0.09
0.28
0.22
0.29
0.28
0.26
0.29
0.28
2
8
8
13
3
5
6
25
MER
0.33
0.19
0.17
0.20
.!l:.21!
0.31
0.23
0.20
セ@
セ@
0.36
0.37
0.36
0.33
0.29
.o.....3.5.
0.32
0.29
0.02
0.03
0.28
0.09
.o.....3.5.
0.27
0.09
0.04
0.04
0.09
0.06
0.05
0.01
0.07
0.33
0.20
0.17
0.16
0.17
0.18
0.21
0.17
0.19
0.20
0.21
0.08
0.21
0.24
0.17
0.22
0.34
0.27
0.40
0.35
.o....3..2
0.33
0.30
0.36
0.32
0.23
0.07
0.25
0.17
0.30
0.18
0.18
0.20
0.17
0.19
0
30
12
5
8
21
セ@
TOTALS
Predicted
Actual
68
about 60 percent of the Roux-Biom surveys were actually in Veld Types other than the
ones to which Acocks had intuitively assigned them. The classification errors cut
across most Veld Types. It follows that taken on whole Acocks Veld Types are not
homogeneous units in terms of the most common taxa diagnostic to pollen analysis. If
the Roux-Biom survey data cannot be used to identify consistently Veld Types using
these taxa, then one would not expect an analysis of pollen samples to discriminate Veld
Types either!
An Alternative System of Plant Community Classification
Obviously, an alternative classification system must be developed that will be of
direct use for comparisons between modern pollen rains and fossil pollen spectra. Now
that the Roux-Biom data allow a quantifiable basis for reclassification, to double check
these results a cluster analysis (Pimentel and Smith 1985) was run on the data using
the same four taxa as was used in the discriminant analysis. The association coefficient
was euclidean distance, and the clustering method was group average (i.e. unweighted
pairs grouped using arithmetic averages with a BIOSTAT II Version 2.0 cluster
analysis). The resulting dendrogram produced eight clusters of varying homogeneity
(Figure 19).
It should be reiterated that such clustering techniques will force
groupings on any data set, and these groups will not necessarily be homogeneous. As a
cross-check on the accuracy on the homogeneity of the clusters, and for a comparison
to the previous discriminant analysis of Veld Types, a second and methodologically
similar, discriminant analysis was run on the Roux-Biom data using these eight
clusters as the pre-defined groups. In the second discriminant analysis all the
individual sample surveys were reclassified correctly with no mistakes. These
statistical analyses suggest that the botanical environment is, after all, significantly
patterned in terms of palynologically diagnostic taxa. The mean and standard deviations
69
of the four taxa used in the cluster analysis were calculated for each cluster along with
the total average percent of the four taxa in the Roux-Biom botanical surveys. This
gives an indication of the relative frequency not included by these four taxa (Table 2).
ALK
-------------------------I---------I
-------------------------I
I----------I
Miセ@
I-----------------------------------·
------------I
I
----------------------------------------------I
--I--I
I-I I-I
I
I I
-----I I----I
GUK
-----I-I
I
-----I I
I----I
-------I
I
------------I
I
I---------I
-----I--I
I
I
-----I
I
I
I-----I
--------I
I--I
I----I
-------I------I
I
I
-------I
I
I----I
---------------------------I
I
I
--------------------------------I I----------I
MセQi@
GLK
I
I
------------------I
I------I
I
------------I---I
I
------------I
I-------------I
----------------I
I------I
TUK
I
I---I
-------I
I
I-------------------I
I
I
----I
I----I
----I-I
I
I
----I I----I
I---------------I
------I
I
I
--------I-------1
I
I
I
I
I
I
I
I
I
I
I
I---------------I
I
--------I
CUK
----I-I
I
----I I---------I
I
------I
I-----I
I
----------------I
I---------I
----------------------I
I
I
I
I
-I-I
I-I
-I
---I I
CGV
HGV
I
I-------------·
-I
I----I
III I
I
I
IIII I
I
-III-I
--II
I--I
I
I
I
I
I
I
I
I
I
II
I-------I
II
-I
I
I
I
I
II-I
II I
II I
I
I
I
I
II I
III
I
I
I
I
I
I------I
I
I
I
I I
II I
I
I
II I
I
II-I
I
-I
-I
I
I
----II I
--I-II-I
--I
I
I
I
I-------------------------------------------------I
--I----I
I
--I
I----------I
LGV
I
I
I
I
I
I
I
I
---I-I
---I
Figure 19. Cluster analysis of Roux-Biom botanical surveys.
Table 2.--Mean and standard deviations of Gramineae, Compositae, Chenopodiaceae/ 7 0
Amaranthaceae (Cheno/Ams), and Aizoaceae/Ruschia relative frequencies for RouxBlom botanical survey clusters.
gイ。ュゥョセ@
qャオセエイ@
qィセョlaュ@
qュァセゥエ。@
AizQLRLJ.fJ.Q.hia
TQTALAV.%
1 (GLK)
39.8±9.1
21.5±9.3
1.0±0.7
17.3±8.5
79.6
2(GUK)
59.6±7 .9
24.5±8.6
0.7±1.1
1.2±2.7
86.0
3(TU K)
40.9±6.2
4 7 .3±3.2
3.0±3.7
0.4±0.8
91.6
4(CUK)
21.3±2.8
63.5±1 0.2
0.0
0.2±0.3
85.0
5(CGV)
89.5±2.1
6.1±1.6
0.2±0.3
0.2±0.4
96.0
6(HGV)
97 .0±1 .6
1 .3±1 .1
0.03±0.09
0.03±0.08
98.36
7(LGV)
82.5±2.2
12. 7±3.5
0.03±0.08
0.1±0.3
95.33
4.7±3.5
35 .3±9.1
8 (ALK)
14.0±11.9
36.1±17.3
90.1
The spatial distributions of these clusters are significant. Most surveys classified
as either Cluster 1 or Cluster 8 occur below the Great Escarpment near Graaff Reinet
or near Beaufort West. They are Karoo plant communities with high Aizoaceae/Ruschia
relative frequencies, and will be termed Lower Karoo for the purposes of this study.
The major difference between Clusters 1 and 8 is the low Gramineae and high
Chenopodiaceae/Amaranthaceae relative frequencies in Cluster 8. Generally, Cluster 1
occurs in the east, while Cluster 8 is found to the west.
Three grass-dominated communities were identified: Clusters 5, 6 and 7. The
major botanical difference between these Grassveld communities is the ratio between
Gramineae and Compositae. Cluster 6, the most grassy, on the average occurs furthest
to the east and at higher elevations. Cluster 7 also occurs in the east but at lower
elevations, and has the highest relative frequencies of Compositae. Cluster 5 has the
broadest longitudinal distribution of the grass plant communities, but occurs at
medium elevations usually in montane settings.
Lastly, Clusters 2, 3 and 4 represent a group of plant communities that are
between the the Lower Karoo group (Clusters 1 and 5) and the Grassvelds (Clusters 5,
71
6 and 7). The main distinguishing criterion is the relative amount of Gramineae
versus Compositae. In general the distribution of the clusters ranges from east to west
starting with Cluster 2 and ending with Cluster 4. These three clusters are called
Upper Karoo.
In summary, the distributions and relative frequencies of the eight clusters
suggest that three major groups are present: Lower Karoo, Upper Karoo, and
Grassvelds (see Table 3). The Lower Karoo can be subdivided into grassy and
nongrassy groups. Cluster 1 is called here Grassy Lower Karoo (GLK), and Cluster 8
is termed Aizoaceae Lower Karoo (ALK). The Upper Karoo can be subdivided into three
groups that range from west to east. Cluster 2 is Grassy Upper Karoo (GUK), Cluster
3 is a general or Typical Upper Karoo (TUK) and Cluster 4 is Compositae Upper Karoo
(CUK). The Grassvelds are composed of three groups (Clusters 5, 6 and 7). Cluster 6
is a high and eastward lying grassveld (High Grassveld or HGV), Cluster 7 is a low and
eastward grassveld (Low Grassveld or LGV), while Cluster 5 is a widely distributed
grassveld usually in montane settings (Central Grassveld or CGV). These names are not
intended for use outside of the area, or beyond palynological analysis. A more detailed
botanical analysis based on species most likely would distinguish different groups,
however for palynological analysis the above eight groups do appear to be significant.
Climates of the Eight Revised Plant Communities
The eight newly defined plant communities will be used as analogs for the
interpretation of pollen spectra to be examined in the next chapter. Ultimately,
however, those spectra will have to be fitted to the simulated climate-change model
presented in Chapter Ill. Direct comparisons are impossible unless the pollen data can
be converted to climatic analogs. In this section I explore the climatic parameters of
key plant taxa in the revised plant communities.
72
To this end, forward stepwise regressions were calculated using the relative
frequency of key plant taxa in the new plant communities as the dependent variable, and
proxy estimates of mean annual temperature and precipitation as the independent
variables. Mean annual temperatures of the interior plateau decrease from north to
south (Venter et al. 1986: 46), but the Roux-Biom survey samples were taken in a
narrow east-west transect which limits the effect of latitude on temperatures. On the
other hand, elevation strongly affects local temperatures. The average change in
temperature with elevation is known as the normal lapse rate (Trewartha 1968: 46).
Through increases in elevation and thus decreases in temperature, elevation is accepted
in this analysis as a second temperature variable.
As average annual rainfall in southern Africa systematically declines from east to
west (Venter et al. 1986: 40),the longitudinal reading (degree of longitude)
approximates mean annual precipitation (Figure 20).
Only Roux-Biom surveys from
summer rainfall areas near Blydefontein were included in this analysis, because
Blydefontein is now in the summer rainfall region and it is highly unlikely it would
have received significantly more winter rains during the Holocene. Surveys from
areas with winter or year-round rainfall patterns were omitted.
The forward stepwise regression for Gramineae entered elevation first, then
degree of longitude (Table 3). Correlations with degrees south were not high enough to
be entered. This suggests that both temperature (elevation) and rainfall (longitude)
influence the relative frequency of grass. Numerous studies have shown that, for a
single location, precipitation is very important in affecting grass growth (Seely 1978;
Vorster and Raux 1983), however this analysis suggests that over large areas
temperature is also important. Thus it is the combined affect of temperature and
rainfall that jointly influence the growth of grasses in southern Africa.
73
750
+
700
'E 650
E.
600
セ@
c
"ii1
550
a:
500
""ii1
::J
c
c
450
<(
c
Ill
Q)
セ@
400
350
+
300
+
250
200
23.5
24
.t
+
24.5
25
25.5
26
26.5
27
27.5
28
Degree Longitude
Figure 20. Plot of mean annual rainfall (mm) and degree of longitude for region
surrounding Blydefontein (data from Werger 1980). Third order polynomial fitted
curve (r2 0.918, p value = 0.0001)
A forward stepwise regression between Karoo composites (as the dependent
variable) and the independent variables of elevation, degree of longitude, and degree of
latitude resulted in only degree of longitude (precipitation) being entered into the
regression (see Table 3). A negative correlation exists between precipitation and
relative frequency of composites. A number of other taxa produced significant
correlations which are included in Table 3.
The stepwise regressions indicate that elevation and/or degree of longitude
significantly affect the distributions of all major taxa. As the Roux-Biom botanical
surveys are scattered along an east-west trending swath, the latitudinal influence on
mean annual temperature was not great enough to influence plant relative frequencies.
Apparently elevation more strongly influence temperatures than latitudinal position in
the sampled region. Taxa that are not in this table occurred in such low frequencies
that their calculations were not acceptable. This is unfortunate because many of these
74
excluded taxa could be highly indicative of climatic parameters such as low
temperatures associated with high elevations. Other taxa had no linear correlation
with the variables. Those taxa that are most strongly correlated with temperature
include Gramineae, Chenopodiaceae and Amaranthaceae (Cheno/Am), Aizoaceae,
Zygophyllum, Crassula, Elytropappus/ Stoebe, Lycium, Acanthaceae, Ruschia, and
Euphorbia. Taxa that have the greatest correlation with precipitation are Gramineae,
Compositae, Chenopodiaceae and Amaranthaceae, and Ruschia. Only Gramineae and
Ruschia were significantly influenced by both temperature and rainfall, and
temperature was the most important variable in each case.
Table 3.--Stepwise regressions with relative frequencies of Gramineae and Compositae
as dependent variables and elevation, degree of longitude, and degrees south as
independent variables. Statview 512+ stepwise regressions for a Macintosh computer
were used (Abacus Concepts, Inc. 1986).
Independent Variables Correlation
DeQendent Variable
ElevatiQn
Gramineae
0.279(R2)
Compositae
insignificant
Cheno/Am
-0.226(partial
Aizoaceae
Degrees セqオエィ@
Degrees East
0.380 (partial
R2)
insignificant
-0.237(r2)
insignificant
R2) -0.175(R2)
insignificant
-0.208(r2)
insignificant
insignificant
Lycium
-0.287(r2)
insignificant
insignificant
Zygophyllum
-0.130(r2)
insignificant
insignificant
Acanthaceae
-0.249(R2)
insignificant
-0.293(partial
Ruschia
-0.261 (R2)
-0.309 (partial
Euphorbia
-0.150(r2)
insignificant
insignificant
Crassula
-0.094(r2)
insignificant
insignificant
Elytropappus/Stoebe 0.062(r2)
insignificant
insignificant
R2)
insignificant
These relationships can be reversed so that the relative frequencies of the taxa
predict the environmental variables through the use of multiple regressions and the
R2)
75
strength assessed through multiple correlations. In the first calculation elevation was
chosen as the dependent variable with the relative frequencies of Gramineae,
Chenopodiaceae and Amaranthaceae (Cheno/Am), Acanthaceae, Aizoaceae, Ruschia,
Lycium, Zygophyllum, Euphorbia, Elytropappus!Stoebe, and Crassula as the
independent variables. These are the variables that were best predicted by elevation.
The multiple R2 = 0.871, and the least-squares-fit equation is:
Elevation = 1710.204 - 1.027(Gramineae) - 17.856(Cheno/Am) 97.575(Acanthaceae) - 15.379(Aizoaceae) - 49.448(Zygophyllum) +
299.120(Crassula) - 64.775(Ruschia) - 13.139(Lycium) +
3.720(Eiytropappus/Stoebe) + 13.909(Euphorbia)
Degree of longitude was also used as a dependent variable with the relative
frequencies of the Gramineae, Compositae, Chenopodiaceae and Amaranthaceae and
Ruschia as independent variables. Gramineae and Ruschia were included in both
regressions because elevation and degree of longitude both entered into their stepwise
regression. The multiple R2 = 0.485, and the least-squares-fit equation is:
Degree of longitude = 24.686 + 0.013(Gramineae) -0.023(Compositae) 0.053(Cheno/Am) + 0.031 (Ruschia)
In theory, any local pollen spectrum could be plugged directly into these
formulae to predict the elevation and degree of longitude, thus temperature and
rainfall parameters for an individual pollen sample.
In practice, however, this
assumes that pollen production, dispersion, deposition, and preservation are all equal
for all plants. Methods for coping with these distortions are discussed later. Another
important variable, which may further hinder a direct use of these formulae is that
rainfall seasonality also affects plant growth.
76
Seasonal Rainfall and Plant Growth
Vorster and Roux (1983) demonstrate that plant groups in the eastern Karoo and
adjacent grassvelds have different growth cycles, and that variations in rainfall season
can select for or against plant growth. Plant growth was measured by month for
different plant groups and the relative percent of growth was plotted (Figure 21 ). The
most important pattern is that grass growth varies seasonally more than Karoo
composite growth. Karoo bossies retain a higher level of growth during the winter
when grasses virtually cease growing. Also the season of highest growth activity of
both climax and subclimax grasses is late summer/early fall, while the greatest
growth activity of Karoo scrub is during the fall and spring. Trees and shrubs attain
their highest growth activity during summer.
Except for trees and shrubs, all plant
groups exhibit a decline in growth activity during mid summer due to high
temperatures and associated water stress, and this decline is more marked in the Karoo
composites than in the grasses.
HIGH
セ@
a:
CJ
w
>
j
w
a:
AUG
NOV
FEB
MAY
AUG
Figure 21. Seasonal growth cycles for major plant groups in the northeastern Cape
(after Vorster and Roux 1983).
77
These growth patterns can be used to predict environmental changes that might
occur with changing seasonal rainfall patterns. Obviously grassvelds are better
adapted to summer rainfall than are Karoo scrub communities, and this is supported
by observations that good summer rains in the eastern Karoo are known to produce
blanketing grass cover (Roux and Vorster 1983: 26}. However grasses generally
require more water than Karoo bossies, and summer droughts can significantly hinder
grass growth while Karoo scrub is capable of suffering through summer droughts and
making use of rain in any season because they grow throughout the year. Thus
superficially it appears that significant drought or a change from summer to winter
rainfall could produce similar patterns in plant distributions and abundances,
however, this is not the case. In the beginning of this chapter I demonstrated that an
eastern and western group of plant communities can be isolated with the Roux-Biom
botanical surveys, and it is the the increase of winter rainfall in the region of the
western group that most likely causes this difference. Thus an increase in winter
rainfall by itself could be identified by an increase of succulents that now characterize
the western portion of the Karoo and the Namib, while a simple change in rainfall that
retains a summer maximum pattern would most likely affect the relative frequencies
between composites and grasses.
A Model of Botanical Change for the Eastern Cape
The relationships between climate and plant taxa identified in the above sections
are used to model vegetation changes under the various climatic conditions suggested by
the COHMAP simulations. Three climatic variables are considered: annual
temperature, annual precipitation and seasonal precipitation. These variables can be
viewed as axes on a three dimensional chart (x,y, z), and, for the purposes of this
discussion, are viewed as independent variables.
78
As temperatures increase the model predicts that Aizoaceae, Lycium,
Zygophyllum, Acanthaceae, Ruschia, Euphorbia, Cheno/Ams and Crassula to increase,
while grasses would decrease. Elytropappus!Stoebe would only be present with cool
temperatures and abundant with cold temperatures. As annual precipitation increases
the relative frequencies of the Compositae, Cheno/Ams and Ruschia would decrease, and
the relative frequencies of Gramineae would increase. As summer rainfall declines and
as winter rainfall increases the grasses are less able to survive from year to year. At
that point Karoo composites, already in the grassvelds in low frequencies, increase
because of reduced competition with the grasses.
The COHMAP simulations of mean annual temperature and precipitation are
coldest and wettest for 18kya and 15kya when summer rainfall was greater than today
(see figures 3.7-3.9 in Chapter 3). The model predicts plant communities to have
been dominated by grasses and Elytropappus!Stoebe. These grassy communities could
be represented by the HGV community, however, modern plant communities may not
provide appropriate analogs for the Pleistocene, because ecological conditions seem so
different from today.
By 12kya this changes to much warmer annual temperatures with cooler
summers and warmer winters than the annual mean. Precipitation drops to its lowest
point. Now, the model predicts a plant community dominated by Karoo composites with
cool temperature indicators like Elytropappus/Stoebe. A CUK or a TUK community
could have existed at this time and possibly a GUK if conditions became wetter.
The climatic situation at 9kya is similar to 12kya except mean annual
temperature is similar to modern levels. The plant communities at 9kya would be
similar to those hypothesized for 12kya except with an increase in thermophilous or
warm-loving taxa. Plant communities at this time could be GLK or CUK.
79
By 6kya mean annual temperature is at its highest with warmer winters and
cooler summers than before. Mean annual precipitation is still low, but winter rains
increase. The model predicts a plant community dominated by Karoo composites, but
the cold-hardy taxa would be replaced by thermophilous plants. Either CUK, GLK or
ALK communities could have colonized Blydefontein at this time.
Major climatic changes occur between 6kya and 3kya, so that by 3kya climatic
conditions are more similar to the modern patterns than ever before. Mean annual
temperature is only slightly cooler than today, and warmer summers and cooler
winters are established.
Rainfall dramatically increases with summer rains much
greater than today. The implications for modelling vegetation are that by 3kya we
would have grass dominating plant cover with warm indicators such as Aizoaceae,
Lycium, Zygophyllum (Tribulus), Acanthaceae, Ruschia, Euphorbia, Cheno/Ams and
Crassula.
Karoo composites would certainly be present, but dominant only during
brief droughts. Cool temperature indicators like Elytropappus!Stoebe would be
present only on the highest peaks of the Kikvorsberg. This translates to expect TUK,
GUK or even LGV communities at 3kya.
Historical Vegetation Change as an Analog of Holocene Changes
Any attempt to use historical trends in the local vegetation as a first test of the
Botanical Model will be complicated by additional factors even though Roux and Vorster
(1983: 26} suggest that droughts have contributed to the change from grassveld to
Karoo in the northeastern Cape. The Grapevale record for the last 60 years shows that
the drought lows in the 18-year cycle have actually become less severe in the later
portion of the 20th century, along with an increase of average annual rainfall.
One line of argument could say that this implies that drought is not a significant
contributing factor in the continuing spread of Karoo plants in this portion of the
80
eastern Cape during most of the 20th century. Today Karoo scrub is invading apparent
healthy as well as overgrazed grasslands. In healthy grassveld with Karoo scrub bare
halos are observed around the plant Chrysocoma tenuifolia, Bitter Karoo (Squires and
Trollope 1979). These bare areas quickly re-vegetate with grasses and forbs if the
above-ground portion of Karoo plants are destroyed by fire or some other mechanism
(ibid: 88), especially when followed by rain (pers. obs. 1985 ). Squires and
Trollope discovered that Chrysocoma emits an allelopathic chemical that inhibits the
germination and seedling growth of plants. Rain washes this chemical to the ground
below the plant and a small barren patch is formed where it is absorbed.
Based on these observations an alternative hypothesis to explain historical Karoo
expansion into grassveld is that the slight increase in winter rainfall in this region
over the last 60 years may be coupled with with the production of allelopathic
chemicals or as yet other undiscovered inhibiting mechanisms by Karoo scrub
(however see above discussion on effects of winter rainfall on plant communities).
When reduced veld burning, especially after the fencing of farms in the 1920s, this
provided just enough selective advantage for the Karoo scrub to overtake healthy
grassveld and overrun poorly managed land. Without doubt some 200 years of
overgrazing has been a major factor governing eastward Karoo expansion, but no one
single factor completely explains the modern rapid spread of Karoo plants.
Returning to climatic explanations it is important to realize that in Roux and
Vorster's (1983) model of historic impacts on eastern Karoo vegetation the shift from
Phase 1-destruction of natural plant communities to Phase 2-devegetation and on to
Phase 3-re-vegatation corresponds to extremely high rainfall amounts in the 1880s
that decline to drought conditions in the 1920's, and thereafter begin to recover to
nondrought conditions in the later portion of the 20th century. It seem clear that the
way the plant communities responded in the Phase 3 re-vegetation was influenced by
81
grazing pressures from domestic stock as this was a period of high stock densities, but
it also seems very likely that the Phase 1 destruction of "natural" plant communities
and the Phase 2 devegetation are responses to climatic changes. Today many farms,
including Blydefontein, have not been overgrazed for many years (others are and can
easily be identified), and plant distributions and plant community compositions on
these well managed farms may be approaching their natural limits as imposed by
climatic conditions.
CHAPTERV
A TEST OF THE BOTANICAL MODEL
Five alluvial exposures and two hyrax dung middens in the vicinity of
Blydefontein Rockshelter were sampled for fossil pollens and dating materials.
Altogether 22 radiocarbon dates were obtained from these sequences, so that an
unusually complete chronometric framework is available. Thus all eight pollen
diagrams can be correlated. Together, these overlapping sequences present a
composite picture of vegetation changes in and around the Blydefontein Basin from late
Pleistocene times through most of the Holocene.
When analogs for past pollen spectra are sought among the eight newly defined
plant communities found in the Blydefontein region today, the pollen data allow a
fairly robust test of the vegetation-change model presented in Chapter IV.
Fossil Pollen Sites
Two types of fossil pollen traps were documented in the Kikvorsberg mountains:
alluvial sites and stratified hyrax dung middens (see Figure 1). The following
summaries are based in part on Bousman et al. (1988), and Scott and Bousman
(1990). The analyses were carried out by a number of scientists who generously
made their data available: Dr. L. Scott analyzed the pollen, Dr. T. C. Partridge studied
the sediments in the geological sites, Dr. J. C. Vogel dated the sites, Dr. S. E. Metcalfe
investigated the diatoms, Dr. M. Seaman analyzed the molluscan fauna, and Mr. J. S.
Brink identified the large mammalian fauna.
82
83
Geological Sites
Five geological sites have been studied from Blydefontein and Hughdale Basins in
the Kikvorsberg Range: Blydefontein Section (BFS), Blydefontein Stream Mouth
(BSM), Channel 2 (CH2), Upper Section-Pond (USP), and Hughdale Section (HDS).
Most geological sites occur in the lower reaches of Blydefontein and Meerkat streams
at or above their confluence, but Hughdale Section is in a perched basin on the farm of
Hughdale approximately 5km northeast of Blydefontein Rockshelter (see Figure 1).
The alluvial sediments in Blydefontein Basin have been divided into two major groups:
the Older and Younger Fills. Both are present at most geological sites in the basin. The
single profile from the nearby Hughdale Basin has both Older and Younger fills, but
the lowest sediments differ from the Older Fills at Blydefontein Basin and are
identified as dorbank, i.e. a hard cemented subsoil horizon such as a B horizon
(MacVicar 1977: 126;
Partridge, personal communication 1989).
Radiocarbon Dates and Alluvial Chronology
Seventeen radiocarbon dates were run by Natural Isotopes Laboratory, DEMAST,
CSIR, Pretoria on material from the geological sites (Table 4). All dates are based on
the 5568 year half-life, corrected for 13c fractionation, but uncalibrated.
These
dates were obtained on organic humates in buried soils and stream deposited silts,
carbonates, and charred reeds.
All dates from the Younger Fills in Blydefontein Basin and Hughdale Basin are of
Holocene age and most are restricted to the Late Holocene. Radiocarbon samples Pta4459 and Pta-4465 represent radiocarbon ages on carbonates and humates from the
same sample at USP, and a comparison of these two dates indicates a small enrichment
of more recent carbonate has made Pta-4459 younger.
The age of the Older Fills can be inferred from an extinct giant hartebeest
(Megalotragus priscus) metacarpal from BFS. Also the dorbank deposit in Hughdale
Table 4.--List of radiocarbon dates from geological sites in Blydefontein Basin
LAB NO.
Pta-4259
Pta-4417
DATE AND RELATIVE
13C CONTENT
290±40 B.P.
o13c= -22.7°/oo
MATERIAL
charred reeds
1360±100 B.P.
humates
PROVENIENCE
BFS, upper buried soil, 60cm below
surface (BS)
o13c= -19.0°/oo
BFS, organic silts in Channel 4,
140cm BS
Pta-4792
2080±50 B.P.
o13c= -21.0°/oo
humates
BFS, middle buried soil, 130cm BS
Pta-4390
3290±60 B.P.
o13c= -16.8°/oo
humates
BFS, lower buried soil, 320cm BS
Pta-4392
4010±60 B.P.
o13c= -20.8°/oo
humates
4430±70 B.P.
o13c= -21.1°/oo
humates
5080±70 B.P.
o13c= -19.8°/oo
humates
Pta-494 7
5270±70 B.P.
o13c= -23.3°/oo
humates
BSM, organic clay in pond deposit,
302-312cm BS
Pta-4458
4260±60 B.P.
humates
CH2, organic silts, top of channel
Pta-4237
Pta-4273
7790±90 B.P.
410±40 B.P.
1870±50 B.P.
BSM, black silty clay loam in pond
deposits, 230cm BS
fill, 240cm BS
humates
CH2, organic silts, bottom of channel
fill, 340cm BS
humates
o13c= -3.1°/oo
Pta-4459
BSM, black silty clay in pond deposits,
110cm BS
o13c= -24.0°/oo
Pta-4796
BSM, black silty clay in pond deposits,
50cm BS
o13c= -17.8°/oo
Pta-4461
84
USP, clay loam in pond deposit,
161cm BS
carbonate
o13c= -3.1°/oo
USP, clay loam in pond deposit,
181cm BS, same sample as Pta-4465
Pta-4465
2000±60 B.P.
o13c= -18.7°/oo
humates
USP, clay loam in pond deposit,
181cm BS, same sample as Pta-4459
Pta-5126
840±50 B.P.
humates
HDS, 1st buried soil, 20cm BS
o13C= -16.8°/oo
Pta-4977
2520±60 B.P.
o13c= -15.6°/oo
humates
HDS, 2nd buried soil, 11 Ocm BS
Pta-4950
3990±70
o13c= -18.3°/oo
humates
HDS, 3rd buried soil, 270cm BS
Pta-5131
4750±100 B.P.
o13c= -15.2°/oo
humates
HDS, 4th buried soil, 365cm BS
SFS
_._.
(I)
ii:
UNCONFORMITY
0::
w
0
z
セ@
liaGRAVEL
セcoluvim@
0
ALLUVIUM
Ill FLUVIAL
P9735
5m
::>
LOAMS
a
SILTY CLAYS
P : POUEN SAMPLE
S: SOIL SAMPLE
Figure 22.
Blydefontein Section stratigraphy.
OJ
C.11
セ@
:J
w
0
z
w
セ@
::;
"'""
セ@
0
セ@
u
....
セ@
w
セ@
1!
(/)
.
w
0
セ@
.•
I
1.4.
•
•
セ@
...
(/)
I-
0
• ......
-
P9738
3.4
I
I
I
-
P9607
55
OLDER P9736
5:7
P'377B
5.7
I
-
P9735
6.1
I
..
.,..
,..
セ@
UJ
m
セ@
セ@
u
I(/)
::;
セ@
z
UJ
0:
w
...
II II
250
250
II
263
1111111
276
Ill
250
I
II
----·
..J
セ@
....I0 I'!
I
II
SCALE
""
N
u
•
[
セ@
-. -...
I
11111111
I
I
....
.;.,
w
0
II
...
1.55
'
0:
w
I
@セ
';-
Ill
32908P
FILLS
"'w
0
P9605
P9739 1.3
::;
- -- (/)
290 BP P9740 0.4
P9606 0.45
1360 BP P9777
0.
セ@
I
,..
w
セ@
セ@
w
セ@
li:
w
""
:§
I-
86
zw
BFS
•
I
I
I
I
1111
249
250
194
-255
II
250
Figure 23. Blydefontein Section pollen diagram.
Basin had Megalotragus priscus remains as well as an immature wildebeest
( Connochaetes sp.) tooth. The youngest radiocarbon dated Megalotragus priscus is ca.
7500 B.P. at Wonderwerk Cave in the Northern Cape (Klein 1980, 1984; Thackeray
1981), however the oldest date for the Younger Fills at Blydefontein Basin is
7790±90 B.P. (Pta-4461) from CH2.
As the Older Fills in Blydefontein Basin are
stratigraphically below this date they must be older, possibly terminal Pleistocene in
age. The age of the dorbank deposit in Hughdale Basin is even less secure, but two
Middle Stone Age sites were found in dorbank a few kilometers down stream of the
geological site of Hughdale Section and a few kilometers upstream from Blydefontein
Section in a side tributary. The youngest chronometric ages for nearby Middle Stone
Age occupations are the thermoluminescence assays of 39,700±4300 B.P. and
26,300±3000 B.P. on sediment samples from Driekoppen Rockshelter in the Zeekoe
Valley, and 30,840±480 B.P. and 38,900±1200 B.P. from radiocarbon assays on
charcoal from Highlands Rockshelter (Deacon 1976: 222;
Wallsmith 1990: 14).
87
Thus it is possible that the dorbank at Hughdale Section also dates to the latter half of
the Late Pleistocene.
Blydefontein Section (BFS)
BFS provides the most complete section documented in Blydefontein Basin. BFS is
6.5 meter deep terrace, informally named the Blydefontein Terrace. It consists of
Older Fills in the bottom 2.5 meters with approximately 4 meters of Younger Fills
above (Figure 22). The Older Fills consist of stratified layers of brown and mottled
yellowish brown alluvial sandy loams, sandy clay loams with pollen and mammalian
fauna, and a few lenses of sand and gravel. Generally the Older Fills are of coarser
texture than the Younger Fills, and this suggests that either sediment supply increased
because of reduce plant cover or erosive energies were higher during the accumulation
of the Older Fills. The clay mineral halloysite is only present in the Older Fills. This
implies that a period of time separates the Older Fills from the Younger Fills, and that
a humid period occurred during or shortly after the accumulation of the Older Fills
(Birkeland 1974: 238). This humid period may be responsible for the erosional
unconformity between the Older and Younger Fills. Two bones were discovered
together in the Older Fills at BFS: the distal portion of a Megalotragus priscus
metacarpal, and a complete Syncerus caffer radius. These bones were submitted for
radiocarbon dating, but no collagen was preserved. Four pollen samples were analyzed
from the Older Fills, and these are characterized by high pollen percentages of
Compositae, Artemisia, and Stoebe-type (i.e. either Stoebe or Elytropappus plants),
and low Gramineae relative frequencies (Figure 23). Dr. Louis Scott (1987 personal
communication) suggests that these samples are characteristic of Pleistocene pollen
spectra.
In the Younger Fills at BFS is overbank alluvium with three buried soils and one
paleochannel with pollen, as well as a non-pollen bearing buried channel (see Figure
88
22). Younger Fills sediments are much more variable than those documented in the
Older Fills. Bulk humate samples from the lowest soil dated to 3290±60 B.P.,
similar material from the middle soil dated to 2080±50 B.P. and charred reeds in the
uppermost buried soil dated to 290±40 B.P. Humic rich silts in Channel 4 dated to
1360±100 B.P. The clayey soils are separated by sandy loam alluvial overbank
deposits, and the youngest buried soil is covered by a thin surface mantle of colluvial
deposits which are slightly altered by modern pedogenesis. At BFS the Younger Fills
are characterized by three cycles of alluviation and relative stability marked by
pedogenesis and non-deposition. A linear regression between soil radiocarbon ages and
mean depth of sample demonstrates a fairly regular accumulation of sediments at BFS
(radiocarbon age = 10.771 (sample depth) + 32.857; r2 = 0.917), which averages
0.093cm/yr (Figure 24). A close look at the radiocarbon dates and depths suggests
that the accumulation rate between the first and second soil was slower than the
overall mean. One possible explanation is that it could be due to the removal of
sediments by stream erosion as two channels occur stratigraphically between these
two soils (see Figure 22).
The pollen samples from the Younger Fills at BFS have varying amounts of Gramineae,
Compositae and Cyperaceae (see Figure 23), and it is suggestive of a grassy karroid
plant community with increases in Compositae during dry conditions or possibly
during periods of increased winter rainfall (Scott, personal communication).
Blydefontein Stream Mouth (BSM)
BSM is in the Blydefontein Terrace, and consists of 6.8 meters of dark organic
clayey silts and diatomaceous lenses representing a buried pond stratified below the
three buried soils documented from BFS (Figure 25). The pond deposits have been
dated to 4010±60 B.P. at 50cm below the top of the pond, 4430±70 B.P. at 111cm,
89
3500
3000
2500
c.. 2000
co
セ@
ro 1500
Q)
>-
1000
500
0
-500
-50
0
50
100
150
200
250
300
350
Depth (em)
Figure 24 BFS radiocarbon dates with one sigma bars plotted by sample depth.
5080+70 B.P. at 236cm, and 5270±70 B.P. at 295cm (Figure 26).
The regularity
and strength of the relationship between depth and age (r2 = 0.988), as well as a
consistent depositional environment allows an extrapolation of ages for undated pollen
samples at BSM between and below the radiocarbon samples. Extrapolating the ages of
pollen samples below the radiocarbon samples is risky as it is based on the untested
assumption that sediment accumulation rates have not changed. With this assumption
in mind, sample age can be estimated through the formula:
Radiocarbon Years B.P. = 5.158(cm below top of pond deposits) + 3805.242
The lower three meters of pond deposit is below the modern water table, and
samples were obtained by coring. However, the core never reached the bottom of the
pond. Diatom samples are available only from the upper 1.8 meters of pond deposit,
but pollen samples are available throughout. In the upper portion of the pond black
organic silts interfinger with overbank alluvium which finally transgresses the pond
90
deposits, and this represents normal pond infilling that does not appear to be
indicative of climatic change. Four diatom samples show a clear fall in water level
with a shift from euplanktonic to epiphytic and aerophilous species up the profile
(Figure 27). Diatom preservation was too poor in the highest sample to allow
quantification, plus an increase in chrysophyte cysts suggest water depths became
unsuitable for diatoms. Pollen spectra indicate fluctuating grassy karroid plant
communities (Figure 28).
BSM
0 SANDY LOAM 0
l2J COLLUVIUM [ill
BLACK POND DEPOSIT
DIATOMS
P : POLLEN SAMPLE
D : DIATOM SAMPLE
S : SOIL SAMPLE
t P97!54
4010 BP
!50 em
I P97!55
Sl+
I P9756
4430 BP
IP9758
P97591
08
S2l
I
P9760
IP9761
I P9763
UNEXPOSED
SIDE WALL
!5080 BP
BACK WALL
Figure 25 Blydefontein Stream Mouth geological profile.
91
5400
5200
5000
c.. 4800
co
....
ro 4600
If)
Q)
>-
4400
4200
4000
3800
0
50
100
150
200
250
300
Depth (em)
Figure 26 Linear regression of BSM radiocarbon dates with one sigma bars plotted by
sample depth.
SSM
HABITAT
ALKALINITY
DEPTH IMI.
Dl
D8
1.57
D9
1.9
oa)-.......,......,.
DB,__ _ __
,.......,
0
"1.
10
Based on spp. セ@ 1'r.
,..--,
0
"1.
Based on spp. セ@ 1'Vt
10
Figure 27 Diatom diagram for BSM.
92
セ@
..... .....,..,.
セ[BG@
-·-- -
...
""""
...,.,.,
.
JY],.,..,
Ill
..
Ill
-
-
......
Ill
-
1111
-
Ill
111111
II
..
Ill
II
Ill
II
I
Ill
I
Ill
Ill
エャZゥセudi@
t'NIOCWol
"""""""'
]\']"""""""'"
........ lDlll
..............
'"""""""'
I
...,..,.,
...,.,..,_,.,
nGdヲャセ@
- -
-
I 111 I I I
I
I I
I
Ill
"'
I
I
Mセ@
I
--
....,..UTIXM
....,.. ].17lQU
,...,..
...,..,.,.
....,..,...
.......,
セ@
....,._..,.
....,.....,
......,..,.,
....,.,..,_
....,.,..._
...,....,..._.
-
..
I
--- -
-
セ@
'5
A
セ@
-
-••
I
II
..
II
-
;
Ill
I.
Ill
I
Ill
..
1.1 I
I Ill
Ill
I
I I II I
35,
3
•
Ill
I
..
I
..
Ill
I
!
I
.. I
. セ@
I..
I
""""
""
......
:on:>nUIW
""'"""""'
_,.,..,.,
C411111&Al1M
'lot
.... ..._.
:t
セ@
- I
H
'Ji
セ@
セr@
セ@
I!
.I
§
I
I I 'Ill I .I I
.
3
セ@
lR
I
'
I
I
...
セ@
I
I
I
I
c:
0
Q)
"'0
>-
i1i
....
0
E
ctS
....
Ol
I
YnOnJ
,..,..,_
-Mセ@ Q)
I"
'-
............,.,
-
Ci5
- - -
---
--- -
セ@
セᄋNオ。ョッ@
ctS
....
Q)
s
I
•
-
""""
.....,..,....,
.........
E
I
セ@
...,..,....
セ@
:::J
0
:::?!
.L
..... m<UT1liiW1
セ@
-
..c
セ@
........,.
セ@
:w.u
II
I
I
- ..
l'O<JI<>:lOJ3r1
I
I
ctS
c:
0
a..
co
C\1
Q)
....
:::J
Ol
u:::
93
Channel 2 (CH2)
CH2 deposits sit unconformably on the Older Fills (Figure 29). The channel
deposits consist of of two units separated by an unconformity, and CH2 provides the
only radiocarbon dated Early Holocene pollen sample. The lower channel fill is dated to
7790±90 B.P., and the upper channel fill is dated to 4260+60 B.P. The pollen
spectra in the lower channel fill is dominated by Compositae, and is most similar to
pollen spectra from the Older Fills (Figure 30). The upper channel fill pollen
spectra are similar to samples from the Younger Fills in BFS and BSM. Three diatom
samples from the upper channel fill suggest a shift from flowing water in the bottom
sample to a marsh habitat with abundant epiphytic plants in the upper two samples
(Figure 31 ).
P
D
POLLEN SAMPLE
DIATOM SAMPLE
CH 2
CHANNEL FILL
DIATOMS
GRAVEL
THICK REEDS ON
BLACK SILT POND DEPOSITS
ROCK
----- -------------------,,,
----------BLACK SILT POND DEPOSITS WITH ROar CASTS
TRYP⦅セM
:_ P 9
I ZMセ]⦅ェ@
P9752 I
Figure 29 Geological profile for Channel 2.
lm
Ed
I]]
セ@
0
WLl
94
CH2
セ@
en
1::
セ@
セ@
w
...
0
a:'
::E
iA
tw
セ@
0
I
- --Mセ
-
1111
I
11111111111
Ill
4.1
Ill
SCALE
•
セ@
;w
セヲAゥ@
4290 BP P 9751 2.4
p 97!!3 3
7T90BP P 9663 3.5
p 97!12
..
lr.
t
!
-
-11111111111111111111111111 Mセ@
ID
!
::
en
J:
0
w
@セ
w
w
I
1111
,セ@ _
セ@
I
-
w
L5
g
,_
gf!i
\{;!:
:- セ@ ==]
YOUNGER FlUS
•
1111111111
250
Ill
1111 II
200
OLDER FlUS
セ@
Figure 30 Pollen diagram for Channel 2.
CH2
セG@
.......,
0
.
AU<AUNITY
OEPTHUIU
HABITAT
BaMd on app.
セ@
1'!1.
2.8
D10
..
D11
3.3
D12
..Gセ@ iLセ@
Figure 31
......
1382"'1
(37.7nj
,..-,
0
10
_,
... 10
Diatom diagram for Channel 2.
Upper Section Pond (USP)
USP is 70cm of carbonaceous and diatomaceous pond deposits that sit
unconformably on the Older Fills, and are buried by ca. 150cm of overbank alluvium
with a thick surface soil (Figure 32).
Bulk humates from the lower portion of the
pond have been dated by radiocarbon to 2000±60 B.P. Carbonates extracted from the
same sample yielded an age of 1870±50 B.P., however the 2000 B.P. date is used here
for consistency. The thick surface soil at USP previously was correlated to the middle
buried soil at BFS (Bousman et al. 1988), but that soil has recently been dated to
2080±50 B.P. at BFS and it is unlikely that this correlation is correct. The top of the
95
pond deposit has been dated at 41 0±40 B.P ., but this date is questionable as it is
unlikely that the overlying soil could have formed in so short of a period. A plotting of
radiocarbon age by depth can be used to suggest that the 410 B.P. sample was
contaminated by more recent carbon that moved down profile through illuviation
(Figure 33). The pollen spectra from the pond are fairly homogeneous and dominated
by grass pollen, except for one sample {9748a) which has high Compositae and
Cyperaceae relative frequencies (Figure 34). Pollen spectra in the overlying
alluvium and soil exhibit a fairly consistent decrease in Gramineae pollen and
increases in Cyperaceae and Compositae pollen.
I P9743,S2
ALLUVIUM
I P9744
50 em
Figure 32 Geological profile for USP.
96
2200
2000
1800
1600
0..
OJ
セ@
1400
1200
I'd
>- 1000
Q)
800
600
400
200
.
0
0
20
40
.
60
80
100
120
140
160
180
200
Depth (em)
Figure 33 USP radiocarbon dates with one sigma bars plotted by sample depth.
USP
セᄋ@
<(
セ@
0
0.3
P9741
0.6
P9742
0.9
P9743
1.2
P9744
P9745
1.5
P9746
1-tl)
iii:<
00
セ@
oo
2000
OJ:
BP
セAZ@
a..3:
1-
0
<(
(/)
0: 1-
:!u
(/)
>-
(ill;"
- w
::<
m
w w
w
セᄃエ@
1.8
P9749
2
P9750
- -·- ·111111
-- ··-·-·
IIIII
-
SCALE
L..-.1
0%' 20%
1-
- ·-
I
-·
1111250
-
1111
250
1111111111111111 111111
265
250
•
1111
I
111111
111111
II
-250
111111
11111111250
1111
-
II IIIII 250
1111111111250
Ill 1111 250
Ill
II -305
II
Figure 34 Pollen diagram for USP.
97
Diatoms show a significant drop in epiphytic species, and increases in aerophilous
species as well as alkaliphilous/alkaibiontic species in samples below 9748a which
suggest that water depths were dropping (Figure 35). 9748a is a pollen sub-sample,
and the diatom sample that corresponds to 9748a was not analyzed. However diatoms
samples above this pollen sample show a return to deep water conditions with an
increase in euplanktonic species and a dramatic decrease in alkaliphilous/alkaibiontic
species. It is important that the greatest frequency of euplanktonic species occurs at
the top of the pond. This suggests that deposits representing the final infilling of the
pond have been eroded, and this is substantiated by the fact that the overlying alluvial
deposits sit unconformably on pond deposits.
USP
DEPTHIMI
HABITAT
(21.15'4)
(24.8"4)
(31.15'4)
DO
D1
,.,.--,
0
.
10
Baaed on
・ーNセQG@
1.4
1.!55
1.6
(19,0'4)
1.69
(45,0'4)
1.8
(13.05'4)
2
,.-.,
0 10
Baaedonspp.!:t'f.
...
Figure 35 Diatom diagram for USP.
Molluscan samples from the lower portion of the pond deposits (samples SH5 and
SH6) are dominated by Burnupia, and show deep water conditions (Table 5). The
molluscan sample that matches 9748a indicates a major shift to the semi-aquatic
Succinea which favors marshy conditions, and the terrestrial Trachycystis (Seaman,
personal communication).
SH1-3 demonstrate a return to Burnupia dominated
98
samples, and deep water conditions. Thus the diatoms and the molluscs show the pond
was deep, then dried to a marsh, and then became a pond again. Cyperaceae is at its
greatest relative frequency when the pond dries to a marsh, however in the past
palynologists have used Cyperaceae pollen percentages as an indication of relative
wetness. At USP, dogmatic adherence to this logic would result in an incorrect
climatic and micro-habitat reconstruction.
Table 5.--USP molluscan fauna raw counts (%) by sample
SamQie
BurnuQia
SH1
55(93)
SH2
26(72)
SH3
14(52)
SH4
0
lセョュ。・@
Suc;c;inea
2(3.5)
2(3.5)
0
59
1(3)
2(6)
1 ( 3)
6 ( 1 6)
36
6(22)
0
6 ( 2 2)
1 ( 4)
27
0
8 ( 6 7)
4 ( 3 3)
12
0
1 ( 1)
1(1)
1 (3)
7(22)
2(6)
Pisidium
0
0
SHS
98(87)
12(11)
SH6
2Q(63)
2(6)
Total
213
21
5
25
tイ。」[ィセウエゥ@
TQtal
14
1 12
32
278
Hughdale Sec;tiQn (HDS)
One geological section documented on the farm of Hughdale in 1987 was discovered
by searching aerial photographs and topographic maps for topographic situations
similar to that in the lower reaches of Blydefontein Basin, i.e. rock defended alluvial
terraces.
Hughdale Section displayed a fairly simple alluvial stratigraphy. This
section is similar to the section at BFS except that it has more buried soils, lacks
infilled stream channels, plus the basement alluvial deposit is dorbank and probably
is older than the Older Fills in Blydefontein Basin (see discussion above). Five buried
soils were documented in a 6 meter thick alluvial section (Figure 36). The soils are
numbered in sequence from top to bottom, and the first (1 0-30cm), second (80130cm), third (250-290cm) and fourth (360-375cm) soils were dated by
99
radiocarbon to 840±50 B.P., 2520±60 B.P., 3990±70 B.P. and 4750±1 00 B.P.,
respectively.
The fifth (505-525cm) soil had too little organic carbon for a
radiocarbon assay, but probably dates to the early or middle Holocene. The lowest soil
sits unconformably on a firm red loam with yellowish brown and grey mottles. This is
dorbank (MacVicar et al. 1977: 126). A similar dorbank deposit occurs on another
feeder stream lower in the drainage near the Hughdale farm house. These two dorbank
deposits are very similar in texture, structure and weathering, but not necessarily
correlated stratigraphically. Nevertheless, the deposit near the farm house has MSA
artifacts in situ, and it is possible that the Hughdale Section dorbank may be of similar
age.
0
セcッャオカゥュ@
F.·;:;;::·:;:::::::··;::;::·;::;:::·;:;::;·:;:;:·::;;:l
Z[ᄋNRセ@
-840±50
-2520±60
2
Q)
セ@ ::J
-3990±?0
(/)
5:
0
Q3
..0
(/)
..._
Q)
+-'
Q)
4
-Alluvium
2
·.·::::::::2::::2::::::::::::::::::::::;:;:2:. -Buried Soil
0
0
0
0
- Dorbank
....._______, - Water Level
0
6
0
0
0
0
0
Figure 36. Geological profile from Hughdale Section.
The sequence of buried soils and alluvium represents various episodes of surface
stability and deposition. The four radiocarbon dates plotted by age by mean depth
100
below the modern surface in 1987 shows a variable accumulation rate (Figure 37).
Nevertheless a linear regression between age and depth adequately models the
accumulation rate, and the resulting linear regression (r2 = 0.968) is:
estimated radiocarbon age = 933.4 + 10.937 (em depth)
8000
7000
6000
5000
0..
セ@
....
4000
til
-:. 3000
2000
1000
0
-1 000
-50
0
50
1 00
150 200 250
300
350
400
450
500
550
600
Depth (em)
Figure 37. Linear regression of HDS radiocarbon dates with one sigma bars by sample
depth.
The pollen sequence from HDS was obtained entirely from the five buried soils and
no samples were collected from the intervening alluvium (Figure 38).
Pollen
frequencies from the uppermost soil are dominated by Compositae, Gramineae,
Aizoaceae, Cheno/Ams and Artemisia. The second soil is dominated by Gramineae and
lacks significant numbers of xeric adapted taxa such as Aizoaceae and Cheno/Ams, but
does have moderate amounts of Compositae. The third soil is dominated by Graminaea
1 01
I-
I I II i I I i
I
I
I
I
I
I
---·1
-
II
81
I
Ill
lllllllllllillllllil
iセM
II
-·
-
--
1111
I
1111111111111111111111111111111111 I
I
-1111111111
II I
IIIII I I
I
I I I
I
1111
I
I
I .
I
I
I
I
I
- I l l
Figure 38 Pollen profile from Hughdale Section.
and Anthroceros, and it has the lowest relative frequencies of composits from the
entire section. Pollen spectra from the lowest three soils are more similar the the
first soil. These soils are dominated by Compositae and other xeric adapted taxa, have
low percentages of Gramineae, and have the only occurrence of Stoebe-like pollen,
which indicates cooler conditions. The lower three soil pollen spectra are similar to
the Early Holocene spectra from CH2. If the accumulation rate for the overlying
sediments is relevant for the age of the fifth soil at Hughdale, then the regression
formula can be used to provide an estimate of 6500-6600 B.P. for this bottom soil.
At this point there is no independent way to assess the accuracy of this age estimate,
but it is possible that the lower soil at HDS is roughly coeval with the Early Holocene
sample from CH2.
102
Hyrax Dung Middens
Two hyrax (Procavia) dung middens were collected from Blydefontein Basin:
Meerkat Midden and Oppermanskop Midden (see Figure 1). Hyrax live in colonies
among large rocks and overhangs which provide easy escape from predators. Hyrax
defecate and urinate in selected spots known as latrines. In these latrines orderly dung
and urine piles (middens) accumulate, and chronological samples can be extracted for
pollen analysis. Hyrax urine is very concentrated and sticky, and it dries to hard
amber-like hyracium which traps and preserves airborne pollen (Scott 1988a).
Preliminary analysis of fossil pollen trapped in the dung middens indicates that hyrax
diet contributes little if any pollen to the midden, and the majority of pollen extracted
from the middens is probably derived from atmospheric pollen rain but dietary
additions cannot be excluded completely (Scott and Bousman 1990). The middens
appear to be one of the least biased fossil pollen traps known in Africa. The first
palynological analysis of a hyrax midden was by Pons and Quezel (1958). They
analyzed a midden from the Haggar Mountains in the Sahara Desert, but thirty years
passed since anyone else extracted pollen from a hyrax midden (Scott 1988a).
Additionally, macrobotanical rests are found in small numbers in hyrax middens, and
this research demonstrates further lines of investigation for hyrax middens
(Lindquist and Fall 1987).
Radiocarbon Dates
Five radiocarbon dates from the two hyrax dung middens are available (Table 6).
Natural Isotopes Laboratory, DEMAST, CSIR, Pretoria ran four dates, and Professor
Paul. S. Martin, University of Arizona, funded the fifth date at Beta Analytic, Inc. All
dates are based on the 5568 year half-life, corrected for 13C fractionation, but
uncalibrated.
103
Table G.--Radiocarbon dates from hyrax middens in Blydefontein Basin
LAB NO.
Pta-5026
DATE AND RELATIVE
13C CONTENT
MATERIAL
hyrax dung
200±45 B.P.
s13c= -22.7°/oo
Pta-4403
300±35 B.P.
PROVENIENCE
Meerkat Midden, 15cm bs
hyrax dung
Meerkat Midden, 18-22cm bs
s13c= -?o/oo
Pta-4571
460±45 B.P.
s13c= -?o/oo
hyrax dung
Oppermanskop Midden,
4.3-7.3cm bs
Pta-5003
1070±50 B.P.
s13c= -23.0°/oo
hyrax dung
Oppermanskop Midden,
8-12cm bs
Beta-14658
1130±80 B.P.
s13c= -?o/oo
hyrax dung
Oppermanskop Midden,
15-18cm bs
Meerkat Midden
Meerkat Midden was found in an overhang adjacent to the Later Stone Age
rockshelter, Meerkat Shelter. It is on an east facing valley wall in the lower reaches
of the Basin. The midden was 22cm thick and two radiocarbon dates are available. If
one assumes that the midden has continued to accumulate at a constant rate until the
time of collection (1985) then a linear regression of radiocarbon ages and depth (r2
= 0.997) can be used to estimate the ages of undated pollen samples (Figure 39) by
the formula:
Years B.P. = 16.5 (depth em) -37.5
The pollen record from Meerkat Midden suggests that in the last 300 years
Gramineae, Cyperaceae and Ranunculaceae pollens have declined, while Compositae and
Cheno/Am pollen have increased, especially in the last two samples (Figure 40).
Oppermanskop Midden
Oppermanskop Midden was located in a small rockshelter in a very narrow and
steep ravine on the north facing slope of the Kikvorsberg. Oppermanskop Midden was
104
18cm thick, and three radiocarbon dates were obtained: 1130±80 B.P. (15-18cm),
1070±50 B.P. (8-12cm), 460±45 B.P. (4.3-7.3cm).
It appears that the midden
350
300
250
0..
200
ctl
150
OJ
セ@
セ@
100
50
0
-50
-2.5
0
2.5
5
7.5
10
12.5
15
17.5
20
22.5
Depth (em)
Figure 39 Linear regression Meerkat hyrax midden radiocarbon dates with one sigma
bars plotted by sample depth.
accumulated at varying rates (Figure 41 ), and it is very likely that the midden was a
secondary accumulation that flowed off of a primary latrine area. A curvilinear
regression can be used to estimate the ages of undated pollen samples. The formula is:
Years B.P. = -74.549 + 140.714 x - 3.984 x2
The Oppermanskop Midden has a Gramineae peak in the oldest sample at ca. 1200
B.P. (Figure 42). Gramineae declines sharply and Compositae and Scrophulariaceae
increase equally at 17 and 14cm (ca. 1000-900 B.P.). The Gramineae increases
again and remain dominate until the upper Scm when Compositae, Euclea, Rhus,
Cheno/Am and Scrophulariaceae increase. This last grass decline in the Oppermanskop
Midden is roughly coeval with the Meerkat Midden grass decline at ca. 300 B.P.
8ᄋセ@
14
c DATE
!YR BPI
"'"'
セ@ セ@
セ@
i
"'
m セ@
..
セ@ セ@ セ@
..
w セ@
"'
=o
セ@ セ@
.
セ@
セ@
...
.
o •
.. ..
-
0
セ@
]OEPTH!CMJ'
JJ
l>
..
0
セ{⦅M
X
SAMPLE NO•
;;::
a0
POOOCARPUS
MYRICA
fTI
RHUS
.z
EUCLEA
;;::
fTI
fTI
TARCHONANTHEAE
セ@
LYCIUM
!:i
tfJ
:X:
.fTI
Cj
fTI
GRAMINEAE
JJ
.,
<0'
CYPERACEAE
c:
セ@
ANTHOSPERI.U.4
CD
セ@
RANUNCULACEAE
0
GUNNERA
OXYOONUM
"'0
I
ALOE-TYPE
0
CD'
ASPARAGUS
::I
LIUACEAE !OTHER!
a.
iii'
SCROPHULARIACEAE TYPE
<0
セ@
Ill
CARYOPHYLLACEAE
CRASSULA
3
0
GERANIACEAE
セ@
!111!3
セ@ ,e
:5::
CD
CD
-"
セ@
Ill
LINACEAE
SCABIOSA
CAMPANULACEAE
セ@
POLYOALACEAE
GALIUM
TRILETE SPORES
: :r
MONOLETE SPORES
'<
セ@
RICCIA
Ill
X
UMBELLIFERAE TYPE
a.
LABIATAE
c:
::I
NYCTAGINACEAE
3
0.:
a.
COMPOSITAE
<0
CD
::I
"411
ARTEMISIA
STOEBETYPE
PASSERINA
,.
•
RESTIONACEAE
CHENOjAMARANTHACEAE
AIZOACEAE TYPE
RUSCH IA
TRIBULUS
TYPE A
OTHER
ゥァャセᆳ_. . gl
SOL
]POU.ENIORAM
106
1400
1200
1000
c..
Ill
セ@
セ@
liS
800
600
400
200
0
-200
-2
0
2
4
6
8
10
12
14
16
18
Depth (em)
Figure 41 Curvilinear regression Oppermanskop hyrax midden radiocarbon dates
with one sigma bars plotted by sample depth.
Pollen Analysis
The pollen data used in the following analysis were collected and analyzed by Dr.
Louis Scott, who very generously has made available all the raw pollen data on which
the following analysis is based.
Preliminary Manipulations of the Pollen Samples
If pollen production, dispersion, deposition, and preservation was equal for all
plants, then pollen relative frequencies could be plugged directly into the climatic
formulae (Chapter IV) to predict temperature and rainfall parameters for each pollen
sample. However, it is well known that pollen taphonomy is complex, and direct
pollen relative frequencies cannot be used. A first step toward investigating pollen
taphonomy is to document the different atmospheric pollen rain of the taxa in question.
This controls for differential production and dispersion, but not deposition and
preservation.
ii ::f 5 ii i. id N :
i5
c
ca
0 _,. a ._,
セ@
CJI
u
•
w
N
Nセッキ@
CJIUI
o
-
!DEPTH ICM,
セ@セ@
samplenoNセ@
r.:-
セ@
.......
セi@
loHuo
E""""
-
NACEAE
.,
セ@
"
Iii
LLA
I
DtOIPYROS
セ@
c
r
c:
....
CD
.j:>.
"'
T
'"0
""""'ActA£
Q
-
ANTHOSP£AWM
-
POOOSTEMACEAE
ffi"
RANUNCLLACEAE
::J
Q.
--
ii)"
--
(Q
Q)
.._..
-
....
----
-
......us TYPo:
-
LILIACE:AE IOTHER)
-
-
ALOE: TYPE
-
EUf'HOROIA
3
0
...
-
....
0
セ@ CD
-
セ@
セ@
...,
セ@
--
..,
セ@
SCR<ll't<UI.ARIACEA£
"'
----
I
l!
TYPo:
セ@
CRASSULA
」セ@
pセylacem@
--
::J
--
-- --=-
--
セ@
""C
----
-
::J"'
ZNセ{D@
...,.
__.,.
'<
OALIUM
UAHTHJS TYPf
UMBELUFERA£ TYPE
_,..........
....
LABIATAE
t\)
-
PLANTAGO
X
0..
§
I
,....- -
ICOMPOSITAt:
<0
3
c:
ART"EMSIA
---
Q..
LACT\JCOIO(AE
--
セ@
-
STOEBE TYPE
-
CLIP"FORTIA
-
-
ERICACEAE
PASSfRINA
CHfNO/AMARANTHActAE
;AJZOACEAE TYPE
AUSCHIA
ACANTHAC[Af
TYP£ A
OTHER
loセ@
セエ@
セ@
QMセ@
108
It has long been known that the number of pollen grains in a sample does not
correctly reflect the numerical representation of individual plants or plant cover
(Erdtman 1943; von Post 1918). In southern Africa some plants produce great
amounts of pollen that are widely dispersed by wind, while others produce very small
amounts that are transported by insects and their distributions are limited (Rebelo
1987). Correction coefficients or R values of Davis (1963) were one of the first
strategies palynologists devised to help correct taphonomic biases due to differences in
pollen production and dispersal among various taxa.
At present only a small number of modern surface pollen samples have been
collected from the research area. These data are insufficient to estimate the effects of
differential pollen production, transportation and preservation for all taxa, but
preliminary assessments of Gramineae, Compositae, Cheno/Am and Aizoaceae pollen
are possible. These four taxa always equal over 50 percent, and usually much more,
of any single fossil pollen sample from the Kikvorsberg and, unless stated otherwise,
the following analysis is restricted to these taxa. More information is needed on pollen
production, and many more modern plant communities are in need of quantitative
survey. Also significant temporal gaps exist in the fossil pollen record from the
Kikvorsberg, and only more research can bridge these gaps.
Coetzee (1967: 116) presents relative frequencies of pollen rain from the plant
communities at Middelburg, C.P. collected in 1952 and 1953 (Table 7). Fortunately,
a number of the modern botanical surveys were conducted at Middelburg from the late
1940s through the 1960s so a rough match between pollen rain and plant composition
is possible. These calculations use the available data which were not collected for the
objective to which they are now used. In a properly designed study of modern pollen
rain and vegetational correlates, multiple modern botanical surveys and matching
surface pollen samples should be collected. In this way more statistically rigorous and
109
accurate estimates can be derived that avoid many of the pitfall of A-values (Birks and
Gordon 1985: 182-204). Nevertheless, with the data at hand, only correction
coefficients can be calculated, and these should be considered preliminary. Relative
frequencies from all the Middelburg modern botanical surveys were averaged and
compared to the average pollen rain data, and then correction coefficients estimated
(Table 7). The calculation is straightforward for each taxa:
correction coefficient = mean botanical survey % + pollen rain %.
These correction coefficients can be used to adjust each taxon's palynological
relative frequency to estimate its composition in the veld. Unfortunately pollen rain
data for only four taxa were published. However, these are the dominant taxa in all
pollen samples. Pollen samples are "corrected" by multiplying a fossil pollen
sample's transformed relative frequencies by the appropriate correction coefficients.
Table 7 .--Middelburg pollen rain relative frequencies from Coetzee(1967),
Middelburg plant community composition(Roux and Blom 1979) and correction
coefficients
Taxa (relative frequencies)
Compositae
Cheno/Am
Aizoaceae
Middelburg plant surveys71.23%
16.81%
1.41%
6.53%
Middelburg pollen rain
70.65%
12.41%
2.25%
3.90%
Correction Coefficients
1.008
1.354
0.627
1.674
Sample
Gramineae
One criticism of this approach is that it uses modern plant relative frequencies
without regard for distortion by overgrazing and veld mismanagement by European
farmers. However it has been argued above (see Chapter IV) that what has been
presumed to be vegetation change due only to overgrazing may also be due to climatic
change.
11 0
Recently Fall (1987) has shown that different environments of deposition can also
bias alluvial pollen records, however preliminary assessment of the Blydefontein and
Hughdale pollen samples does not reveal strong patterns in pollen relative frequencies
corresponding to depositional environment. However, biases introduced by differential
deposition and preservation cannot be fully assessed with the data at hand, and should be
the focus of future research.
Finally, the Roux-Biom surveys did not sample marsh or riverine plant
communities, and a comparison with the pollen samples requires that pollen types
restricted to these environments be omitted from the relative frequency calculations in
the fossil pollen samples. Relative frequencies of fossil pollen samples minus marsh
pollen (i.e. Cyperaceae, Typha, Gunnera, Monolete and Trilete spores) are called
marsh-adjusted samples in this study.
Palynology of the Older Fills and Younger Fills
In this comparison the pollen frequencies are marsh-adjusted but not corrected
because certain taxa do not have pollen rain correction coefficients. In most pollen
samples the Compositae/Gramineae ratio is the most clear cut relationship, however a
comparison of Younger Fills and Older Fills pollen provides evidence that a major
difference exists between the plant communities that grew during the accumulation of
both sedimentary units. It was also discovered that Artemisia had a linear
relationship with Gramineae in the Older Fills at BFS, but not in the Younger Fills
(Figure 43). In both cases grasses are represented by the x-axis and composites and
Artemisia are on the y-axis. The linear relationships are ecological relationships and
do not represent temporal sequences. The composite/grass differences are clearly
illustrated by a comparison of the regression slopes between Older and Younger Fills
111
at BFS (Table 8). These regression statistics are presented for descriptive purposes
and they are not intended for predictive purposes in a statistical sense.
80
Bl ydefontei n Section
70
60
Compositae
and
aイエ・ュゥウ。セ@
50
40
30
20
10
Gramineae
I 0 Artemisia
DComposit
60
70
Olders Fills
I
セ@
0 Younger Fills
•
Figure 43 Graph of Gramineae versus Compositae and Artemisia relative frequencies
in the Older and Younger Fills at BFS.
Table B.--Comparison of Compositae/Gramineae and Artemisia/Gramineae linear
regression slopes, intercepts and correlation coefficients between Younger and Older
Fills at BFS
s・、ゥュョエ。イセ@
Younger Fills
Older Fills
Unit Taxa
Comp/Grass
ArtemLGrass
Comp/Grass
Artem/Grass
SIQQe
-1.000
I nterQeQl
78.322
-Q.Q11
-4.864
Q.717
86.984
2.054
-5.336
Grass relative frequencies in the Older Fills are never very high, but much
higher in the Younger Fills. The Younger Fills composite/grass regression slope of 1.0 is similar to the regression slope of the eastern Karoo group (-0.724) from the
112
Roux-Biom botanical surveys presented in Chapter IV. However it is clear that the
composite/grass regression slope for the Older Fills (-4.864) is much steeper than
the regression slope for the Younger Fills or eastern Karoo group. Also the regression
slopes for the Older Fills and the western Karoo group (-0.494) diverge in opposite
directions from the Younger Fills and eastern Karoo regression slopes. This suggests
that the eastern Karoo plant communities and the plant communities from the Younger
Fills at BFS may share similar distributional relationships among Compositae and
Gramineae taxa, but that the plant communities reflected by the Older Fills pollen
spectra have no modern counterparts sampled by the Raux and Blom botanical surveys.
The Artemisia/Gramineae patterns also highlight the differences between the
Older Fills and the Younger Fills. A strong positive linear correlation is present
between Artemisia and Gramineae in the Older Fills at BFS (r2 = 0.999}, but not in
the Younger Fills (r2 = 0.095). These correlation coefficients and graphs support the
thesis that the herbaceous and grass components in plant communities that grew in the
Basin during the accumulation of the Older Fills are different in terms of spatial
distributions and ecological structure from those in the Younger Fills. In addition,
Artemisia only occurs in two of the modern botanical surveys, and a correlation
coefficient between Artemisia and Gramineae indicates a lack of correlation in the
modern botanical surveys (r2 = 0.007). The Older Fills pollen spectra are not
similar to any modern plant community sampled by Raux and Blom (1979), and
possibly distinct from any modern plant community extant in southern Africa today.
The high composite percentages in Older Fills suggests dry conditions, and high
Artemisia, Passerina and E!ytropappus/Stoebe relative frequencies suggest cool
conditions. The most similar modern plant community is probably some form of
Alpine Fynbos present in high elevations in the Drakensberg, but for which no
quantitative modern botanical survey was available. Similar conditions are known at
113
7790 B.P. from Channel 2 where pollen sample no.9663 has high composite, low
grass, low Artemisia, high Elytropappus/Stoebe, and high Aizoaceae percentages.
This probably still represents a dry climate, but the high Aizoaceae relative
frequencies suggests that it occurred when conditions were warmer than during the
accumulation of the Older Fills sediments. This Early Holocene sample may date to a
transitional period when the plant communities were switching from some type of
Alpine Fynbos to Karoo/Grassland community. Coarser sediments in the Older Fills
suggests that greater erosive energies or a less effective vegetation mat existed during
their accumulation as opposed to the Younger Fills (Partridge, personal
communication). Some sort of Alpine Fynbos would be a less effective vegetation cover
when compared to the grassy karroid plant communities present in the post 8000 B.P.
sediments.
In order to assess the hypothesis that the plant communities of the Holocene are
represented by the modern botanical surveys, composite/grass ratios for all
Blydefontein Basin fossil pollen samples dating within the last 8000 years have been
plotted (Figure 44), and linear regressions calculated. The regression slope of the
post 8000 B.P. pollen spectra is -0.915 (r2 = 0.743), which compares even more
favorably to the eastern Karoo regression slope of -0.724. These data suggest that the
modern biota was established by approximately 7-8000 B.P. in Blydefontein Basin,
but distinctly different plant communities were present in the Late Pleistocene.
The Test Results
A comparison between the plant communities defined by the cluster analysis of the
Roux-Biom botanical surveys and the post 8000 B.P. pollen samples is possible with
marsh-adjusted and corrected pollen samples used in Figure 44. The discriminant
functions used to assess the accuracy of the cluster analysis of the Roux-Biom
114
70
60
50
40
Compositae セ@
30
20
•
10
0
PKMセイ@
0
10
20
30
40
50
Gramineae
I•
Meer
0
Opper B
CH2
D
USP
.&.
60
70
BFS
X
eo
セ@
BSM
A
HDS
I
Figure 44. Graph of composite/grass relative frequencies for all Holocene pollen
spectra.
botanical surveys in Chapter IV, were saved and then used to classify the transformed
and corrected pollen samples from Blydefontein Basin (Table 9). The procedure
assigned pollen samples to the newly defined plant communities. A similar study west
in the Nuweveldberg used modern pollen rain samples grouped by Acocks' Veld Types
assigned fossil pollen samples to Veld Types with multiple discriminant analysis
(Sugden and Meadows 1989). Unfortunately in this study the modern pollen rain
samples were not submitted to a cluster analysis first so the homogeneity of the groups
could be assessed with discriminant analysis.
The totals indicate that only five of eight modern plant communities defined by
cluster analysis can be identified in the last 7900 years. The Meerkat 10 sample can
be considered as modern, and it was classified as Grassy Lower Karoo. The Aizoaceae
Lower Karoo, Central Grass Veld, and High Grass Veld do not occur.
11 5
Table 9.-- Discriminant analysis canonical scores for transformed and corrected
pollen samples, samples arranged in chronological order from younger to older
Site &
Sample
Aizoa.
Lower
Karoo
Grassy Composite Typical
Lower Upper Upper
Karoo
Karoo
Karoo
Grassy
Upper
Karoo
Low
Grass
Veld
Central
Grass
Veld
High
Grass
Veld
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.65
0.00
0.00
0.00
0.00
1.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.13
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
BFS
BFS
BFS
BFS
BFS
BFS
9740
9606
9777
9739
9605
9738
0.00
0.00
0.00
0.00
0.00
0.00
0.14
0.00
0.17
0.00
0.00
0.00
0.06
0.90
0.00
.1.Jl.Q.
1.00
0.00
0.79
0.10
0.23
0.00
0.00
1.00
BSM
BSM
BSM
BSM
BSM
BSM
BSM
BSM
BSM
BSM
BSM
BSM
BSM
BSM
BSM
9754
9755
9756
9757
9758
9759
9760
9761
9763
10430
10434
10438
10444
10449
10456
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0. 00
0.00
0.00
0.20
0.99
0.07
0.03
0.49
0.06
0.08
0.00
0.00
0.00
0.00
0.00
0.00
0.19
0.10
0.00
0.00
0.90
0.00
0.06
0.00
0.01
0.35
1.00
1.00
1.00
1.00
0.00
0.81
0.70
0.29
0.01
0.03
0.97
0.45
0.80
CH2
CH2
CH2
9751
9753
9663
0.00
0.00
0.00
1.JlQ
0.60
0.00
0.00
0.00
1.00
0.00
0.00
0.00
0.00
0.39
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
USP
USP
USP
USP
USP
USP
USP
USP
USP
USP
9741
9742
9743
9744
9745
9746
9748
9748a
9749
9750
0.00
0.00
0.00
0. 00
0.00
0.00
0.00
0.00
0.00
0.00
0.55
0.76
0.99
0.01
0.44
0.18
0.03
0.01
0.81
0.22
0.03
0.00
0.00
0.00
0.00
0.00
0.00
0.11
0.00
0.00
0.42
0.24
0.01
0.00
0.00
0.00
0.00
0.23
0.00
0.97
0.00
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
!l..11
Q.Jll
Q....9.a
0.33
0.82
0.00
0.87
0.19
0.00
M.D.
.Q...1.8.
Table 9
11 6
Continued.
Aizoa.
Lower
KarQQ
0.01
0.00
0.00
0.00
0.00
0.00
Grassy
Lower
KarQQ
0.00
0.00
0.00
0.00
0.00
0.00
OPPER 1
OPPER 2
OPPER 3
OPPER 4
OPPER 5
OPPER 6
OPPER 7
OPPER 8
OPPER 9
OPPER10
OPPER 11
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
1.00
MEERK10
MEERK 9
MEERK 8
MEERK 7
MEERK 6
MEERK 5
MEERK 4
MEERK 3
MEERK 2
MEERK 1
Site
セ。ュqi@
HDS
HDS
HDS
HDS
HDS
HDS
&
10419
10420
10421
10422
10423
9781
TOTALS
Composite Typical
Upper
Upper
KarQQ
KarQQ
0.00
0.99
0.99
0.00
0.99
0.01
1.00
0.00
0.00
1.00
0.50
0.50
0.85
0.00
0.80
0.00
0.00
0.93
0.91
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.04
0.00
0.33
0.00
0.00
0.00
0.00
0.00
0.01
0.04
0.00
0.00
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.75
Q..A..2
0.22
0.01
0.91
0.00
0.03
0.00
0.00
0.20
0.00
0.19
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.25
0.35
0.00
0.00
0.07
0.00
0.18
0.00
0.00
0.00
0
19
10
セ@
セ@
11
Grassy
Upper
KarQQ
0.00
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.33
0.15
0.99
0.20
QJl.9.
QJM
0.06
0.04
0.00
0.00
0.00
Q.11t
0.99
0.02
1.Q.Q.
Q...Z.9.
.1...Q.Q.
0.02
Q...Z.9.
13
Low
Grass
Central
Grass
High
Grass
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.01
0.06
0.00
0.00
0.66
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.33
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.89
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.09
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
2
0
vセiq@
vセiq@
vセiq@
0
Each site provides a sequence of alternating plant communities. The
classifications have been arranged chronologically in Figure 45.
In Figure 45
samples with good chronological control have horizontal lines marking shifts from one
plant community to another, otherwise classifications are plotted in their
approximate chronological position.
11 7
BSM begins with Composite Upper Karoo in the lower four samples, and then shifts to
Typical Upper Karoo. It briefly changes to Grassy Lower Karoo, and then to Typical
Upper Karoo. This is followed by a shift to Composite Upper Karoo which drops to
Grassy Lower Karoo and then shifts back to Typical Upper Karoo and the sequence ends
with a Grassy Upper Karoo plant community. Channel 2 only has Grassy Lower Karoo
communities. Blydefontein Section (BFS) begins with a Typical Upper Karoo
community that shifts to Composite Upper Karoo, then to Grassy Lower Karoo which is
followed by a shift back to Typical Upper Karoo and the sequence ends with a Grassy
Upper Karoo plant community. Upper Section Pond (USP) begins with a Grassy Upper
Karoo plant community which shifts to Grassy Lower Karoo and is followed by Typical
Upper Karoo, then Grassy Upper Karoo back down to Typical Upper Karoo that drops to
Grassy Lower Karoo then back to Typical Upper Karoo and ending with a Grassy Upper
Karoo plant community. Hughdale Section (HDS) begins at approximately 7900 B.P.
with Central Lower Karoo. This community is present once more at approximately
5500 B.P., but by 4000 B.P. a Typical Upper Karoo community is established, and it
is again present at about 2500 B.P. The last sample at HDS is roughly 500 B.P. in
age, and a Central Lower Karoo community is present. Oppermanskop Hyrax Midden
(OPPER) begins with the extremely grassy Low Grassveld, and this is followed by
several cycles between Grassy Lower Karoo and Grassy Upper Karoo plant
communities. The beginning plant community in Meerkat Hyrax Midden (MEERK) is
Grassy Upper Karoo which changes to Low Grassveld, and is followed by several cycles
between Grassy Upper Karoo and Grassy Lower Karoo.
The modern plant community, as identified in the final pollen sample in Meerkat
Midden, is Grassy Lower Karoo. These comparisons, agree with the predictions of the
botanical model presented in Chapter IV, and also with the general discussion on Older
and Younger Fills pollen spectra. Clearly, vegetation has not remained stable for long
11 8
Years BP
(x 1000)
BSM
BFS
USP
CH2
HDS
0
lUKfCUK
GLK
CUI:
GLK
GLK
-
lUI:
OPPER
.._cut:
GLK
GLI< /lUI: I GUI:
GLK
GUI<
GLK
GUI<
Gttc.=::=
3
-
CUK
セ「]@
GUI<
GUK
GLI<
LG\1
Ill ...
GUK
2
MEER
GUK
lUI:
GUK
lUI:
GLK
GUK
lUI:
lUK
4
-
lUI:
GUK
GLK
GLK
llJK
GLK
CUK
"Till!'
5
-
GLI<
lUK
CUI:
CUI:
6
-
7
-
CUI:
CUI:
8
OLK : oイ。セZウケ@
ャッキセイ@
CUK : CDmpD:sit uーセイ@
1UK: Typical uーセイ@
l<aroo
ャ\セイpッ@
QUK: Or=:ll:l uーセイ@
l<arDtt
LO'r' : Lvweor Ora:5:5 \leolo:l
l<aroo
Figure 45. Temporal distribution of paleo-plant communities at geological sites and
hyrax middens.
11 9
periods of time in the Holocene. Also important to the assessment of plant
communities is the length of time represented by a pollen sample. Pollen spectra
from the Meerkat Midden demonstrate that significant changes can take place over
very short periods. Apparently, then, the time span represented by an individual
sample can affect the relative frequencies in the sample, and thus the mix of plant
communities represented by the sample.
CHAPTER VI
STABLE CARBON ISOTOPE ANALYSIS
The use of the stable carbon isotopes 13c and 12c for correcting radiocarbon
dates is well known, but for the last 13 years these isotopes have served as analytical
tools for a variety of purposes in archaeology (van der Merwe i 982; van der Merwe
and Vogel 1983). Vogel and van der Merwe (1977) pioneered the use of stable
isotopes in archaeology for deciphering prehistoric diets. Other analyzed materials
include inorganic and organic fractions of bone, plant material, pedogenic calcium
carbonate, speleothems, soil humates, snail shells and even ostrich eggshell. While
many African applications in archaeology still focus on prehistoric human diets
(Ambrose and DeNiro 1986; Rightmire and van der Merwe 1976; Sealy 1986 and
1989}, analysis of 13c;12c ratios promises to provide important information on past
environments (Heaton 1987; Heaton et al. 1986; Stuiver 1975; Vogel 1983; von
Schirnding et al. 1982}. The application discussed here involves an analysis of 13c
and 12c ratios from bulk radiocarbon humate samples with matching pollen samples
and from ostrich eggshell in Blydefontein Rockshelter.
Principles and Technigues
Before a discussion of the present applications can proceed, however, both the
technique and the occurrence of carbon isotopes in the natural environment of
southern Africa is required. 12C and 13C are stable isotopes that do not decompose
into other elements (as does 14C). 12C and 13C occur in relatively constant amounts
on earth, and the ratio is approximately 100 to 1.1, respectively (van der Merwe and
Vogel 1983).
However, the ratio varies slightly between materials, or from
120
1 21
environment to environment depending on a variety of factors discussed below. The
measurement of 12c and 13c represents the ratio of 13Cf12c in the sample as
compared to the ratio of these isotopes in a piece of marine limestone, known as the
PDB standard. The ratio, written as o13c, is measured in parts per mil (expressed as
Ofoo).
The PDB standard has a very high amount of 13c and as its ratio of 13Cf12c is
set arbitrary at zero, most terrestrial o13c measurements register as negative
values. Increased amounts of 13C in a sample cause a o13c value to become less
negative. The formula for calculating o13c values is:
6
13 c =
<13c;12c)
sample[ ( 13 C1 12 C) standard
-1
Jx
1000
As 12C and 13C have different atomic weights, they react differently in chemical
and physical reactions such as photosynthesis (ibid).
This is because lighter isotopes
have higher vibrational frequencies, form weaker bonds, and are more reactive in
chemical processes than heavier isotopes (Faure 1986}. The differential use of these
two isotopes in chemical reactions cause the ratios of the isotopes to change, and this is
known as isotopic fractionation. It is now well known that photosynthesis causes the
most significant fractionation of stable carbon isotopes.
Photosynthesis uses carbon atoms from the atmosphere, and the unpolluted
atmospheric o13c value is approximately -7.0 °/oo (van der Merwe and Vogel 1983).
Photosynthesis begins at this fractionation starting point. Three forms of
photosynthesis are identified in plants. These three photosynthesis types or pathways
are known as Cs, C4, and CAM, and each fractionates carbon isotopes to produce its own
characteristic average o13c (Mooney et al. 1977; Vogel et al. 1978).
122
Natural Occurrences of Stable Carbon Isotopes
C3 plants, which include all trees, most shrubs, and a number of grasses that
have adapted to temperate or shaded conditions, strongly fractionate carbon isotopes in
favor of 12c, and have an mean o13c value of -26.5 °/oo (Vogel et al. 1978). C4
plants are more efficient at using 13c than C3 plants, and the mean o13c value of C4
plants is much higher, -12.5 °/oo, than C3 plants (ibid). C4 plants include a wide
variety of grasses, but also species in the Cyperaceae, Chenopodiaceae, Aizoaceae and
Amaranthaceae. C4 plants are normally those adapted to hot and arid conditions. CAM
species are plants that can switch from a C4 to a C3 pathway as the environmental
conditions allow. Most CAM plants are succulents, and the mean o13C is -16.5 °/oo.
CAM plants can confuse the dichotomous signal between Cs/C4 plants, but normally
CAM plants do not occur in great enough numbers in this portion of southern Africa to
be a serious problem.
Vogel (1978), Vogel et al. (1978), and Ellis et al. (1980) have mapped the
distribution of C3/C4 grasses in southern Africa, and the grass communities in the
Kikvorsberg are dominated by over 90 percent C3 grasses, while the surrounding
Karoo plains is dominated by over 90 percent C4 grasses. Thus a clear C3/C4
dichotomy exists today in the grass species in and near Blydefontein Basin, which
should be translated to unambiguous o13C measurements if a fossil collector can be
found.
Stable Carbon Isotopes in Soil
Haas et al. (1986) submit that the o13C from bulk soil humate radiocarbon
samples can be used to infer C3-C4 botanic changes. Dr. John Vogel measured o13c
from Blydefontein Basin's geological radiocarbon humate samples and these samples
are used below. A comparison of grass pollen relative frequency, and o13C
123
measurements for matched humate and pollen samples is an improvement on using
only humate
s13c,
as discussed for Lubbock Lake (ibid).
However some problems do exist. The first concern is the addition of new carbon
through illuviation. In radiocarbon dating this tends to make radiocarbon dates too
young, so that the resulting age is generally believed to represent the mean residence
time (MRT) of the deposit (Taylor 1987). It seems reasonable to expect that stable
isotopes may also reflect MRT averages, but analysis by Hillaire-Marcel et al.
(1989) suggest that MRT problems are limited for aquatic depositional environments
such as lakes and ponds, although this is not necessarily the case for buried soils.
Future research should address this problem directly. As long as each sample is
independently dated by radiocarbon, the major issue is how much time span is
represented in a single
s13c value
and how much bias is created by looking at samples
representing greatly varying time spans. A second consideration is an apparent
fractionation of carbon isotopes in sediments. This fractionation process is believed to
be the result of microbial decomposition in sediments and soils, and estimates of the
effect range from +3°/oo to +4°/oo (Dzurec et al. 1985; Natelhoffer and Fry 1988),
and I have accepted the fractionation effect at +3°/oo with the realization that
fractionation in sediments may not be consistent. The third problem is one of equality
of catchments between isotopes and pollen. In other words it is assumed in the
following analysis that the source areas (and source plants) for pollen and isotopes
are equal. Intuitively this assumption is difficult to accept, nevertheless on a general
level the differences may not be significant. It is important in this analysis to include
pollen percentages based on all pollen, including marsh pollen, and the following
analysis does this. Finally one radiocarbon age estimate from BSM does not have
matching pollen sample (Pta-4947), and it has not been used in the following
comparisons.
Blydefontein Soil o13C and Pollen Spectra Compared
124
Starting with the oldest date and working up through time in Figure 46, at 7790
B.P. the grass pollen percent and the associated o13C value are both very low. No
matter how much or how little the stable carbon isotope ratios have been fractionated
due to microbial decomposition, a complete or almost complete C3 plant environment,
including grasses, must have existed at this time. The slopes of the two lines leading to
5080 B.P. both increase. The pollen spectra indicate that the major change is a large
increase in grasses, and significant decreases in C3 Compositae and Stoebe-type pollen
(either Stoebe spp. or Elytropappus sp.). This implies that C4 grass species increased
between 7790 B.P. and 5080 B.P., otherwise the younger o13C value would have
remained the same or nearly so. One radiocarbon date, 5270±70 B.P. (Pta-4947)
from BSM, has a low o13C value (-23.3°/oo), but unfortunately it is not associated
with an analyzed pollen sample and cannot be plotted on this graph. Nevertheless, this
low o13C value suggests that few or no C4 plants were present at that time either;
thus implying that the increase in C4 grasses occurred between 5300 B.P. and 5000
B.P.
The slopes between 5080 B.P. and 4750 B.P. indicate that grass pollen drops
dramatically while the o13C slope shoots up. The pollen spectra in the 4750 B.P.
sample shows that Compositae, mainly C3 plants, increase and Cheno/Am, Aizoaceae
and Ruschia remain stable, thus it is likely that the grasses in this sample are
dominated by C4 grasses otherwise the o13c value would have been much lower.
Between 4750 B.P. and 4430 B.P. grass pollen percentage increases, but the o13C
value drops significantly. Pollen spectra show that the most significant changes are an
increase of grass, Cyperaceae (C3 species), Aizoaceae (CAM species) and Cheno/Am
(mostly C4 species) pollen percentages, and the decrease of Compositae. However, the
increase in Cyperaceae and decrease of Compositae are not great enough to account for
125
the o13c drop without a shift from C4 grasses to C3 grasses by 4430 B.P. In the next
sample, 4290 B.P., grass pollen percentages change very little (ca. 1 percent), while
the o13c value increase is quite large (>3°/oo). Cheno/Am increases a little, but
other pollen taxa likewise remain fairly stable.
This suggests that C4 CHENO/AMs
increased, as well as C4 grasses in relation to C3 grasses, with little change in overall
grass relative frequency at this time.
-1 5
50
0
-1 6
45
-1 7
40
-1 8
d13C
o/oo
35
-1 9
30
-20
25
-21
20
•
-22
-23
Grass
Percent
•
15
-24
10
0
1000
2000
3000
4000
5000
6000
7000
Years BP
1·0- Humate d13C
·•- %Grass Pollen
I
Figure 46 o13c ratios and relative frequency of grass pollen from matched samples.
The slopes between 4290 B.P. and 4010 B.P. show a marked change. Here, grass
pollen increases, but the o1 3 c slope drops sharply. The pollen spectra indicate that
the most significant difference is the increase of Gramineae at the expense of
Compositae and Cheno/Am. If grasses were dominated by C4 species then the o13c
value would not have dropped, thus a shift back to C3 grasses is the most reasonable
interpretation.
126
The slopes leading to the next sample, at 3990 B.P., show another C3-C4 grass
exchange. Pollen spectra indicate that the only significant change is a reduction of
grass and an increase in composite pollen, but the o13c value sharply increases. Again
this suggests an increase in C4 grasses over C3 grasses. This trend continues to the
next sample at 3290 B.P., which has even less grass pollen, more composites, and
higher o13C values. The values 2520 B.P. demonstrate the highest grass pollen
percentages and one of the highest o13c values. This must reflect an increased C4
grass dominance. Between 2520 B.P. and 2080 B.P. the o13c value and grass pollen
percent decline markedly. Composites increase equally rapidly, in fact this sample
has the highest composite percentages in the dated samples. This most likely
represents a slight change in C3/C4 grass relative frequencies but the addition of
other C3 species might be enough to pull down the o13c value at 2080 B.P.
The next shift to 2000 B.P. shows an increased o13c value, but much less so than
the grass pollen increase. The pollen spectra indicate that the major change is a
trade-off between grasses and composites accompanied by a minor increase in
CHENO/AMs. This implies a slight increase of C3 grasses versus C4 grasses,
otherwise the o13c value would have increased at a greater rate. The slope between
2000 B.P. and 1360 B.P. suggests a further increase in C3 grasses since grass
percent increased while the o13c values decreased. This pattern reverses between
1360 B.P. and 840 B.P. when composite and Cheno/Am pollens increase at the expense
of grass pollen, and o13c values increase. While the increase in CHENO/AMs might be
responsible for a portion of the o13c rise, it is suggested that a significant shift from
C3 to C4 grasses also occurred. A reversal appears to occur between 840 B.P. and 410
B.P., when C3 grasses appear to increase but possibly not to the level of dominance.
It is possible to estimate the relative amount of C3 plants represented by the
humate isotope ratios by the formula:
127
Percent of C3 Plants= (&13c value- 3.0 + 12.5)/ -0.14
This formula assumes that the humate &13c values had been fractionated by
microbial decomposition by 3°/oo, and that the mean value for C4 plants is -12.5°/oo
and the average difference between C3 and C4 plants is 14°/oo.
Hypothetically from a
C3 plant community with no grass, as C4 grasses begin to colonize Blydefontein Basin
then one would expect the C3 plant estimates to decrease with higher grass pollen
percentages, i.e. they would have a negative relationship. Additionally, as C3 grasses
take over from C4 grasses or other C4 plants then C3 plant estimates should begin to
increase as grass percentages increase, and the relationship between grass percent and
C3 plant estimates should have a positive slope.
The second order polynomial fitted cuNe (r2 = 0.233) in Figure 47 shows a
slope that changes from negative to positive and strongly suggests that the relationship
between C3 plants and percent of grass pollen switches as predicted by a shift between
C3 and C4 grasses. It appears that C4 grasses are dominant when grass occurs in very
low percentages (less than 30-33%). This is interpreted as an increase in C4
grasses versus non-grass C3 plants such as the Compositae. At approximately 30-33
percent grass pollen it appears that C3 grasses begin to increase in relation to C4
grasses, and thus the C3 plant estimates increase. It should be noted that a single
sample from Hughdale Section, pollen sample 10420 (2520 B.P., 50% grass pollen,
C3 plant estimate = 0.7%) , was dropped from the regression. This single sample
seems to have an unusually large amount of C4 plants considering the overall
abundance of grass pollen, and it anomalous position on this scatterplot will be
considered below. The diagonal line shows the expected C3 plant estimates assuming
that all grasses were C4 plants and that no other C4 plants occurred. This theoretical
function can be used to suggest that samples that fall above the line may have had C3
grasses, and that samples that fall below this line must have had other C4 plants in
128
addition to grass. Additionally this theoretical function implies that the grass in the
pollen sample dated to 2520 B.P. is less of an anomaly that the curvilinear regression
implies.
A comparison between modern botanical survey data (see Chapter IV) and climatic
records shows a negative relationship between mean annual rainfall, composite
percent and C4 grass percent, while overall grass percent, C3 grass percent and
rainfall have a positive relationship. One implication of Figure 47 is when C4 grasses
are believed to dominate over C3 grasses, as grass pollen decreases it is replaced by C3
plants, ie. composites. This is logical because the local C4 grasses are drought
resistant-warm weather species (Vogel et al. 1978, Ellis et al. 1980), and as
conditions become too dry for C4 grasses, they would be replaced by C3 composites.
However, as conditions become more moist C3 grasses appear to replace C4 grasses and
11 0
1 00
0
7790
90
% C3
Plants
80
70
60
50
40
30
10
0 4750
15
20
25
30
35
40
45
50
% Grass Pollen
Figure 47 Curvilinear correlations of C3 plant estimates based on o13c ratios and
relative frequency of grass pollen for matched samples, minus pollen sample dated to
2520 B.P. from Hughdale Section. Numbers represent radiocarbon years B.P.
Theoretical linear function for percent of C3 plants given all grass is C4.
55
129
other C3 non-grass species. Thus warmer and drier conditions seem to increase the
relative frequency of composites, and they cause a shift from C3 to C4 grasses. During
wetter conditions, not only do grasses increase in overall relative frequency, but C3
grasses appear to replace C4 grasses. These interpretations agree with a number of
modern studies on the distribution of C3 and C4 grasses in Africa and, in fact,
throughout the Southern Hemisphere (Cavagnaro 1988; Cowling 1983; Ellis et al.
1980; Tieszen et al. 1979; Vogel et al. 1978; Young and Young 1983).
It should be noted that the general pattern does not seem to hold for the pollen
sample dated to 2520 B.P., but the reason for this disjunction is not entirely clear
(see Figure 47). This sample may represent a plant community which was colonized
by the C4 grass Themeda trianda, rooigrass, (Cowling 1983: 123) commonly found
today at higher elevations to the east such as in the Stormberg and other areas covered
by grass velds (Acocks 1975:88-98). Many other grasses found at these elevations
are C3 species including the now dominant Merxmuellera (Vogel et al. 1978: 213).
Thus it is possible that Themeda trianda becomes dominant only after a significant
amount of grass is established.
The stable isotope value associated with the 7790 B.P. sample suggests that few
C4 plants were present at that time and, if that was the case, then no matter what
changes occurred in the pollen spectra the o13c based C3 plant estimates would remain
low because no C4 plants were available to raise the 13c t12c ratios. It is possible
that different seasonal rainfall and temperature patterns existed during the Early
Holocene (Kutzbach and Guetter 1986), and the manner in which plant taxa associated
with each other differed in comparison to Middle and Late Holocene plant communities.
Vogel and Talma's (in Deacon and Lancaster 1988: 144) carbon isotope analysis of the
Cango Cave speleothem has o13c values that are indicative of an almost exclusive C3
plant biota in the southern Cape during the Early Holocene as well. These data have
130
been used to suggest that a sharp increase in o13c values and apparently C4 plants
occurred only after 5500 B.P. with the establishment of modern plant communities
(Deacon and Lancaster 1988: 144). The Cango Cave, Blydefontein Basin and other
carbon isotope data from southern Namibia and Lesotho (Vogel 1983) might indicate
the existence of an Early Holocene C4 grass reservoir in southern Namibia. As more
data are collected it might be possible to actually delimit the boundaries of an Early
Holocene C4 grass reservoir from which climatically induced migratory pulses
originated.
Stable Carbon Isotopes in Ostrich Eggshell
Ostrich eggshell is ubiquitous in paleolithic archaeological sites throughout Africa
and western Asia. Often, because of poor preservation, these sites lack more
traditional sources of data used for reconstructing past environments such as other
faunal remains or pollen. If present, these traditional sources are often biased by
human selection in ways that limit their usefulness in reconstructing past
environments. If palaeoenvironmental information can be extracted from ostrich
eggshell, then analysis of past human behavior in a more complete environmental
context will be possible. Moreover, the data are not biased by human selection.
A preliminary study stable of carbon isotopes in ostrich eggshell by von
Schirnding et al. (1982) suggests that the organic fraction in eggshell rapidly
decomposes within 1000 years. Thus carbon isotope analysis of the organic fraction
has very limited applicability to many archaeological sites. However carbonate from
well preserved eggshell appears to be free from diagenetic chemical changes over, at
least, the last 1 0,000 years or more. von Schirnding et al. (1982) argue that
eggshell carbonate o13C values are enriched by approximately 16.2°/oo in relation to
diet during metabolism (carbonate o13c values are enriched by 14.1 °/oo in relation
131
to the organic fraction, and the organic fraction is enriched by 2.1 °/oo in relation to
diet), and they argue this fractionation effect is constant. Their published data (ibid)
can be used to show that this model of isotopic fractionation is not correct, at least for
the fractionation between the organic and inorganic fractions of shell. But a complete
study that links the two fractions of eggshell to diet is lacking, and their (ibid) model
of fractionation is used here in absence of a better controlled study. As ostriches are
thought to be indiscriminate generalized feeders, it is expected that carbon isotope
ratios in eggshell carbonate represent a fairly accurate estimate of the Cs/C4 plant
composition in the environment.
Fluctuations in Blydefontein Ostrich Eggshell s13.Q_
A sample of ostrich eggshell (n=121) from Blydefontein Rockshelter was
measured for stable carbon isotopes, and a histogram of the s13C values shows that
most of the samples form a slightly skewed (skewness = -0.306) unimodal
distribution (Figure 48).
However two individual measurements above -4.0 are
suspected as outliers. Reinspection of these two samples showed they were taken from
weathered shell, and we believe that oxidation of the shell enriched the s13C values.
These two measurements have been omitted from further analysis.
In Figure 49 mean and individual ostrich eggshell s13C values from twenty
four superimposed excavation units from Blydefontein Rockshelter are plotted. A
single Early Holocene sample has been omitted for clarity, but it will be included
below. The chronological sequence is based on radiocarbon dates from associated
excavation units, termed analytical units and described in the following chapter, and
ages of intervening excavation units without radiocarbon dates were estimated by
assuming a constant accumulation rate between the two nearest radiocarbon dates. As
the top and bottom of most sedimentary units in the rockshelter were dated, each
132
30
Frequency
25
20
f-
r-
15
10
r-
5
rf-
0
-11
nrl
-10
-9
-a
-7
hn .
-6
-5
n
-4
n
-3
-2
-1
s13c
Figure 48. Histogram of ostrich eggshell o13c values.
major layer has its own estimated accumulation rate. Nevertheless, an assumption of
a constant accumulation rate is obviously not correct in every case or even most cases,
but it is a necessary starting point. Also one must assume that the period of time
represented by some analytical units probably spans multiple, diverse climatic
episodes, even though the excavation units represent relatively short periods of time
(ca. 70-150 years). This cautionary note is supported by analyses of modern and
historical climates which demonstrate that significant climatic and biotic changes can
be very rapid in the Karoo (Cowling et al. 1986; C. Vogel 1988, 1989).
The mean o13c curve (see Figure 49) has at least three major dips during the
last 4300 years: ca. 4100 B.P., 2000 B.P. and 900-1300 B.P. Also a dip at ca.
3200 B.P. may have palaeoenvironmental significance.
If ostriches are truly
indiscriminate feeders then lower o13C values reflect periods when ostriches ate more
C3 plants because the coeval plant communities had more C3 versus C4 plants. This is
0
133
a difficult assumption to test without conducting a modern controlled experiment.
However, other data are available that reflect on the validity of this assumption.
-5
s 13c
0
/oo
-6
-7
-8
-9
-10
500
1000
1500
2000
2500
3000
3500
4000
4500
Years BP
0 Individual Measurements
•
Average Measurements
Figure 49. Individual and mean ostrich eggshell (OES) o13C values grouped by
excavation unit. Lines connect mean values between each excavation unit and
individual values within excavation units.
Blydefontein Basin Soil o13C and Rockshelter Ostrich Eggshell o13C Compared
The independent o13c curve from dated sediment humates in the geological sites
discussed above can be used. A comparison between radiocarbon dated humate and
ostrich eggshell o13c values shows that the two curves are similar, even though many
fewer points occur on the humate curve (Figure 50). The ostrich eggshell o13c
values are corrected (ie. reduced by 16.2°/oo) according to the data presented in von
Schirnding et. al. (1982), and the fit of the humate curve to the ostrich eggshell
curve suggests that the humate o13c values are too high. These elevated humate values
are probably due to fractionation caused by microbial decomposition in soils (Dzurec
et al. 1985; Flexor and Volkoff 1977; Natelhoffer and Fry 1988). Also the ostrich
134
eggshell curve is flatter. This could be due to dietary selection, ostrich metabolism
and incorrect understanding of fractionation between diet and eggshell carbonate, or
averaging the values in either the eggshell or soils. Two major points of disagreement
are the high o13c values for humate samples dating to 3290 B.P. and 2520 B.P., and
low values of the contemporaneous ostrich eggshell samples. However fairly high
individual o13c values (> -6.0) are present in the ostrich eggshell samples estimated
to date to 2855 B.P. and 3135 B.P. (see Figure 50). These could represent a short
period with increased amounts of C4 plants which is more or less coeval with the high
sediment o13c values discussed above, or an alternative possibility is that the o13c
values in these sediment samples reflect a local plant community that has significantly
more C4 plants than the regional biota as registered by the diet of ostriches. The
remaining dips and peaks especially at ca. 4000 B.P. and 2000 B.P. are much closer
to the ostrich eggshell curve, and can easily be considered as coeval and reflecting the
same local environmental changes.
0
0
Humate /oo
Ostrich Eggshell /oo
-21
-15
-16
-22
-17
-18
-19
-20
-21
-22
-23
-23
-24
-25
-26
MRTKセイ]W@
0
1000
2000 3000 4000 5000 6000 7000
Years BP
1·0- Soil Humate &'13C
8000 9000
·•- Corrected DES &'13C
I
Figure 50. Corrected mean ostrich eggshell (OES) and C-14 dated sediment o13C.
135
Conclusions
In this chapter stable carbon isotope evidence from two separate sources, dated
sediments and ostrich eggshell, provide additional insights on paleoenvironmental
changes in the Kikvorsberg range. Variations in carbon isotope ratios provide
evidence of fluctuations in C3 and C4/CAM plants over the last 8500 years. The o13Q
samples from sediments have matching pollen samples, and when both sets are
compared fluctuations in C3 and C4 grasses can be identified. Similarities between
sediment and ostrich eggshell o13Q sequences show that these two different sources are
recording the same past botanical changes. This analysis suggests that few, if any, C4
plants were in the Kikvorsberg before 5500 B.P., and significant decreases in C4
plants occur at ca. 4000 B.P., 2000 B.P. and 1300-1000 B.P. As the ostrich
eggshells are from Blydefontein Rockshelter, this o13Q sequence provides an unbiased,
high resolution record of paleoenvironmental change that unquestionably can be
associated directly with the archaeological record from this rockshelter.
Rarely in
Stone Age archaeology are such clear cut paleoenvironmental associations possible.
CHAPTER VII
RECONSTRUCTION OF THE BLYDEFONTEIN PALEOCLIMATE
In this chapter Blydefontein palynological and stable isotope results are used to
reconstruct climatic conditions through the Holocene. The reconstruction is then
compared with the predictive model for rainfall and temperature, outlined in Chapter
Ill.
Palynological Estimates of Climate
A number of studies have reconstructed past climatic conditions from pollen data,
using a variety of statistical methods known cumulatively in pollen studies as transfer
functions (Adam and West 1983; Birks and Birks 1980; Birks and Gordon 1985.:
252-259; Bryson and Kutzbach 1974; Cole and Bryson 1968; Guiot 1987; Guiot
et al. 1988; Hooghiemstra 1984; Huesser et al. 1980; Huesser and Streeter 1980;
Webb and Bryson 1972; Webb and Clark 1977). In general transfer functions use
quantitative modern plant or pollen-climate relationships to statistically estimate
past climatic parameters.
As shown in Chapter IV, the relative frequencies of various plant taxa in the
eastern Karoo can be used to estimate approximate elevation and degree of longitude.
Marsh-adjusted and corrected fossil pollen frequencies can be used in multiple
regressions to reconstruct temperature (i.e. elevation) and rainfall (i.e. degrees of
longitude). This assumes (1) fossil pollen assemblages accurately reflect the eight
modern plant communities defined in Chapter IV, and (2) that these reconstructed
plant communities respond to climatic parameters in the same way as living plant
communities do today.
136
137
Elevation, Degree of Longitude, Climate and Taxa
Multiple regression estimates for elevation and degree of longitude were
calculated with the Roux-Biom survey data. The estimates of elevation and degree of
longitude can be used to predict expected mean annual temperature and rainfall.
Gramineae, Cheno/Am and Aizoaceae were used as independent (predictor) variables
for elevation. In these equations Aizoaceae includes Ruschia. Compositae was not
included in this equation, as elevation was not a good predictor of Compositae percent
(See Table 3). However, all four taxa are independent variables for estimating degree
of longitude.
elevation
=
1680.7279+ 1.489(Gramineae) - 7.58(Cheno/Am) 21.817(Aizoaceae)
The associated R2 = 0.582.
degree of longitude
= 24.7373
+ 0.029(Gramineae) -0.005(Compositae)
-0.058 (Cheno/Am) +
0.039 (Aizoaceae)
The associated R2 = 0.497.
The mean annual temperature at 1275 msl in Middelburg is 14.6 degrees C, and
temperature can be estimated for Blydefontein (1680 msl) at 12° C using the normal
lapse rate ( 6.5° C per 1000 meters). However it is the difference from the modern
temperature that is the unit of study. The formula estimating temperature difference
is:
estimated temperature difference = ((1680 - elevation) * 0.0065)
An estimate of the correlation between rainfall and degree of longitude can be
gained by analyzing the annual rainfall and degree of longitude for 49 stations in the
upper Orange River drainage (Werger 1980: 12-13), and calculating a regression
curve from the data (see Figure 20). A third order polynomial curve provided the
138
best fit and the associated r2 was 0.918, p value = 0.001. This equation estimated
Grapevale's mean annual rainfall as 352mm, but the actual mean annual rainfall is
14mm above this estimate (i.e. 366mm). The discrepancy is probably due to
orographic effects. The Grapevale data were used to fine-tune this equation by
increasing the intercept by 14mm. With X representing degree of longitude, the
resulting third order polynomial equation for rainfall difference from the modern
average is:
mm rainfall difference estimate
=
-150700.4759
+ 18362.547*X
-745.587*X2 + 10.111*X3- 366.
Temperature and Rainfall Estimates
Four geological sites and two polleniferous hyrax dung middens are used to
estimate climatic parameters: Blydefontein Section, Blydefontein Stream Mouth,
Channel 2, Upper Section Pond, Oppermanskop Hyrax Midden, and Meerkat Hyrax
Midden. Pollen frequencies for these four taxa with pollen rain data were adjusted for
marsh taxa and corrected. The resulting values were assumed to accurately represent
those taxa in the ancient vegetation. These frequencies were then entered into the
above multiple regression formulas.
The multiple regression formulas estimated the elevation and degrees longitude of
the pollen sample, and these two variables were used to estimate the difference from
modern rainfall and temperature. The multiple regression formulas have large
standard deviations, so these estimates should not be accepted with a high degree of
confidence. Nonetheless, the technique offers a promising approach for estimating
climatic parameters using fossil pollen frequencies. From the outset the influence of
European veld mismanagement, and its consequences on plant distributions has been a
concern in this study. This problem is believed to be at least one cause of the large
139
standard deviations of the regressions, but the general relationships indicated by the
regressions are thought to be due to climatic restrictions on the distributions of
plants. No doubt European influences have acted to weaken the plant-climate
relationships. Also, if European influences have affected the multiple regressions,
then the estimates will err on the warm and dry side of the scale and not the cool and
wet end, because it is the composites and warm/xeric forms that have expanded their
ranges to the east and thus distort the plant-climate relationships.
Blydefontein Section
Blydefontein Section (BFS) temporally is the longest geological section. The
sediments are divided into two major units: Older Fills and Younger Fills. The pollen
spectra from the Older Fills are distinctly different from the pollen spectra from the
Younger Fills. The plant communities represented by the pollen from the Older Fills
are not represented by any of the plant communities sampled by the modern botanical
survey, thus these methods are not applicable to Older Fills pollen spectra. Pollen
frequencies from the Younger Fills are presented below along with the temperature
and precipitation estimates shown as differences from modern levels (Table 10).
Samples 9740, 9739 and 9738 are buried soils, and sample 9777 is from a silt lens
in a buried stream channel. The general pattern shifts from cool-moist to warm-dry
back to cool-moist with a return to warm-dry. The upper two soils seem to have
formed under warm-dry conditions, while the lower soil formed under cool-moist
conditions.
140
Table 10.--BFS Younger Fills relative frequencies of selected pollen taxa and
estimated climatic parameters
Mean Annual
Taxa
Climate Estimates from
modern
sample Gramineae Compositae Cheno/Am Aizoaceae
Temperature oC
Rainfall
9740
29.3
45.0
11.4
3.1
0.1
- 7
9606
25.5
53.7
5.9
4.3
-0. 1
-2
9777
48.8
28.9
5.0
2.9
-0.3
52
9739
17.9
59.2
7.1
5.8
0.0
-11
9605
14.8
65.3
9.7
1 .9
0.2
-29
9738
37.6
44.0
3.0
2.1
-0.3
19
bャセ、・ヲqョエゥ@
Stream MQUlh
Blydefontein Stream Mouth (BSM) represents the upper 1.8m of pond deposits.
It is not unreasonable to accept a constant sedimentation rate for these deposits
because they are from a single depositional environment. An estimated age of each
sample was calculated by a linear regression using sample depth and radiocarbon age
(see Figure 27). The general temperature pattern changes from slightly cooler
temperatures to a warm period from 4300 to 4700 B.P. with a short cool interval at
4500 B.P ., and after 4300 B.P. back to cooler temperatures (Table 11 ). Rainfall is
less than modern levels between 4700 and 4100 B.P. with a short moist period
coincident with the cool temperatures at approximately 4500 B.P.
Channel2
Pollen was recovered from both Older and Younger Fills at Channel 2 (CH2), but
the Older Fills sample could not be used in these equations because its pollen spectrum
does not represent a plant community sampled by the modern botanical survey.
Sample 9751 was dated to 4290 B.P., and it is roughly coeval with Sample 9757 from
BSM. It is difficult to estimate the age of Sample 9753, but as it is stratified in
channel fill a few centimeters below Sample 9751, it is probably no more than a few
hundred years older.
1 41
Sample 9663 was dated to 7790 B.P., and represents the only
radiocarbon dated early Holocene sample.
Table 1 1.--BSM relative frequencies of selected pollen taxa and estimated climatic
parameters
Mean Annual
Climate Parameters from
Taxa
modern
Temperature oC
Rainfall
sample Gramineae Compositae Cheno/Am Aizoaceae
-0.4
22.4
2.9
2.4
74
9754
53.2
-0.2
9755
29.3
48.2
4.5
2.3
1
45.0
10.4
5.2
0.0
0
9756
28.9
32.1
7.6
5.1
-0.2
9757
42.2
37
14.0
4.1
0.1
28.1
9758
33.5
3
0.1
-11
45.0
8.3
5.5
9759
16.5
-0.3
9760
40.4
37.7
4.0
2.2
26
5.3
8.0
-0.1
9761
28.3
41.6
22
33.2
4.2
1.2
-0.3
9763
45.6
34
6.3
5.4
0.2
10430 35.2
31.8
27
42.2
0.1
10434 34.2
6.8
3.4
10
50.2
5.5
2.6
-0.1
10438 27.2
-3
4.6
2.5
0.1
10444
8.8
64.6
-2 7
47.2
8.9
4.3
0.0
10449 22.6
- 9
58.2
5.6
1.2
0.1
-2 7
10456 11.7
Table 12.--CH2 Younger Fills relative frequencies of selected pollen taxa and
estimated climatic parameters
Mean Annual
Taxa
Climate Estimates from
modern
Temperature oC
Rainfall
sample Gramineae Compositae Cheno/Am Aizoaceae
0.1
29.5
10.0
5.5
9751
25.5
2
15.5
6.0
2.5
-0.3
9753
46.0
46
52.9
3.1
10.2
-0.1
9663
16.0
9
Upper Section Pond
Ten pollen samples were collected from a diatomaceous pond deposit and overlying
alluvium.
Samples 9750, 9749, 9748a, 9748 and 9746 are from pond sediments,
and 9749 is dated to 2000 B.P. Samples 9745, 9744, 9743 and 9742 are from
alluvium. The uppermost pond sample (9746) has a radiocarbon age of 410±40 B.P.,
142
but it difficult to accept this assay as correct (see Chapter V). The uppermost pond
deposits were truncated by erosion so that the depositional record preserving the
desiccation of the pond was destroyed by an erosional unconformity. The pond samples
show a shift from wetter conditions to much drier conditions without increased
temperature just after 2000 B.P. (9748a), and then a return to greater
precipitation and cooler temperatures (Table 13).
Initially the rainfall estimated in
the lowermost alluvial sample (9745) shows no great difference from the previous
precipitation level from the pond deposit, but very quickly estimated rainfall drops
and estimated temperature increases.
The uppermost sample indicates even drier-
warmer conditions.
Table 13.--USP relative frequencies of selected pollen taxa and estimated climatic
parameters
Mean Annual
Taxa
Climate Estimates from
modern
sample Gramineae Compositae Cheno/Am Aizoaceae
Temperature oC
Rainfall
9741
29.1
2.1
43.9
18.0
0.3
-19
9742
42.7
34.1
14.1
3.8
0.0
15
26.1
20.8
9743
40.8
3.3
0.2
- 1
9744
38.6
42.3
6.5
1 .9
-0.2
13
29.1
7.7
9745
48.3
2.6
-0.2
40
9746
43.9
35.5
7.9
-0.2
3.5
30
9748
56.3
20.3
7.7
4.1
-0.3
80
9748a
29.3
46.1
5.4
-0. 1
3.0
2
9749
45.5
30.4
11.2
-0. 1
3.6
28
9750
51.1
19.2
6.9
3.6
-0.3
62
Hughdale Section
Six pollen samples were collected from five buried soils in Hughdale Basin
(HDS). The lowest soil was sampled twice for pollen. All but the lowermost soil have
been dated and the dates, in sequence from top to bottom are: 840 B.P., 2520 B.P.,
3990 B.P., and 4750 B.P. When first discovered it was believed that the lower soil
143
might date to the Pleistocene, but this seems unlikely considering the radiocarbon ages
of the soils above and the general stratigraphy in nearby Blydefontein Basin. HDS
rainfall estimates indicate a dry middle Holocene, and then conditions begin to become
more moist by 3990 B.P. (Table 14).
Table 14.--HDS relative frequencies of selected pollen taxa and estimated climatic
parameters
Mean Annual
Taxa
Climate Estimates from
modern
sample Gramineae Compositae Cheno/Am Aizoaceae
Temperature oC
Rainfall
10419 18.8
46.5
10.2
11.4
0.1
6
10420 50.6
34.8
3.6
0.0
-0.4
40
10421
35.3
2.1
0.0
34.4
-0.3
12
1 0422 20.3
58.6
5.4
1.4
0.0
-17
10423 27.6
0.3
-0.2
53.5
1.2
-4
44.2
2.8
0.0
9781
33.5
-0.2
4
Meerkat Hyrax Midden
Ten pollen samples were extracted from a 22 em thick consolidated hyrax midden
(Table 15). This midden was discovered under a very small overhang adjacent to
Meerkat Rockshelter. A radiocarbon age of 300 B.P. was obtained on dung from the
bottom 4 em of the midden and an age of 200 B.P. was obtained on dung from 15 em
below the top of the midden. The top of the midden was assumed to date to 0 B.P., and
using these three points a linear regression was calculated to estimate the age of
individual pollen samples (see Figure 40).
The rainfall estimates show a sharp decrease from a high mean annual amount
relative to today during the interval 300 to 200 B.P., which is followed by a
fluctuating plateau until approximately 86 B.P., i.e. AD 1864 . After 86 B.P. rainfall
estimates drop dramatically. The age estimate for this point is 30 B.P. (AD 1920).
Rainfall estimates increase after AD 1920. The temperature estimates are a negative
144
mirror image of the rainfall estimates except for the oldest sample which is warmer
than might be expected.
Table 15.--Meerkat Hyrax Midden relative frequencies of selected pollen taxa and
estimated climatic parameters
Taxa
Climate Estimates from
modern
sample Gramineae Compositae Cheno/Am Aizoaceae
Temperature oC
Rainfall
10
38.91
32.30
12.45
0.39
set at 0
set at 0
9
24.80
46.80
20.00
1.20
0.4
-2 9
1.29
8
43.10
15.09
1.29
-0.4
48
17.20
1.20
7
47.60
0.80
-0.4
56
-0.3
25.52
1.26
1 .26
6
34.73
22
-0.4
5
51.03
17.28
2.06
1.64
68
-0.4
4
44.49
29.92
0.00
0.39
41
1.22
-0.5
0.82
3
57.96
11 .43
95
0.44
2
73.57
9.69
0.00
-0.7
174
1
60.09
8.33
2.19
14.91
-0.5
284
Oppermanskop Hyrax Midden
Eleven pollen samples were extracted from a consolidated 18 em thick hyrax
midden discovered in a small rockshelter in a protected kloof below Oppermanskop.
Age estimates of individual pollen samples uses the second order polynomial
regression in Figure 42.
Rainfall estimates are generally higher and temperature estimates somewhat
lower than modern levels between 1200 and 450 B.P. However within this period a
significant reduction in rainfall and increase in temperature is estimated to have
occurred between 1200 and 900 B.P. At or shortly after 450 B.P. rainfalls are
estimated to have dropped and temperatures climbed (Table 7.7). Since the Europeans
had a significant influence on the environment only in the last 170 years, and mostly
in the last 100 years, sample 1 is the only sample that might reflect this disruption.
Comparison of the Pollen-Derived Rainfall Estimates
145
Rainfall estimates for the sites and hyrax middens were compared by plotting
them together. Even though the sites do not overlap greatly in time, such a comparison
allows for a further assessment of the consistency of the estimations. It also provides
Table 16.--0ppermanskop Hyrax Midden relative frequencies of selected pollen taxa
and estimated climatic parameters
Mean Annual
Taxa
Climate Estimates from
modern
Temperature oC
sample Gramineae Compositae Cheno/Am Aizoaceae
Rainfall
-0 .1
19.3
45.9
2.9
3.3
1
- 6
20.1
5.7
4.1
-0.2
2
33.6
24
42.4
28.4
3.1
0.8
-0.3
3
30
10.7
0.8
0.4
-0.3
4
36.1
30
8.2
1.2
1.2
-0.6
60.5
112
5
-0.4
42.5
17.0
1.6
4.5
6
64
2.5
1.6
-0.5
7
60.4
6.9
11 0
7.7
1 .2
1.2
-0.6
63.4
125
8
18.7
2.0
2.0
-0.3
37.8
9
34
33.0
23.0
0.0
0.0
-0.3
10
18
-0.7
11
77.7
3.1
1.6
0.4
195
a first step toward constructing a long term estimate of Holocene rainfall fluctuations.
Temperature estimates have not been presented because the range of temperature
fluctuation estimated was not great, and because a strong correlation exists between
rainfall and temperature so that one mirrors the other at least as modeled here and for
other quantitative Holocene climatic estimates (Scott and Thackeray 1987: 93-98).
First both hyrax middens rainfall curves are compared, and then these curves are
plotted along with the remainder of the sites.
A comparison of the Meerkat and Oppermanskop hyrax midden estimates shows
some similarity between the two sites in terms of rainfall decline over the last 300
years (Figure 51). However, the Meerkat estimates appear to be consistently higher
than the Oppermanskop estimates, and the Meerkat estimates suggest greater
146
fluctuations. As the accumulation rate is faster in the Meerkat Midden, the resolution
of the Meerkat samples is finer and samples span short periods. If the biotic
environment can respond rapidly to climatic changes then the higher estimates may be
correct as they reflect less temporal averaging. Another potential influence is that
today the micro-habitats of the two sites differ, and this may be reflected in the
rainfall estimates. Oppermanskop Midden is in a very steep and narrow north-facing
ravine with abundant trees species, and Meerkat Midden, in a shallow gorge in the
middle of the Basin, has an eastern aspect and more open vegetation (Scott and
Bousman 1990). Nevertheless, both records produce the same general pattern.
200
150
Rainfall
Difference 100
from
Modern
50
(mm)
0
MUPKイセ@
-100
100
300
500
700
900
110013001500
Years BP
1·•-
Oppermanskop ·.&.- Meerkat
Figure 51. Rainfall estimates from Oppermanskop and Meerkat Hyrax Middens.
A comparison of geological site rainfall estimates allows a further assessment of
the technique, and, if accepted, produces a more lengthy, although still patchy,
estimate of rainfall for the Holocene. Ages for the five USP pond samples were
147
calculated by using the sedimentation rate of the BSM pond deposits. The margin of
error of this calculation is totally unknown, but the importance of the USP samples
warranted their inclusion. The USP alluvium and soil samples were not included.
Even with the gaps and lack of overlap the sites appear comparable except for the
samples from the soils at BFS and HDS. A comparison between the BFS sample and
contemporaneous estimates from other sites suggests that the soil pollen sample
estimates are too low. Two upper soils at BFS exhibit low grass percentages, however
the lower soil has high grass relative frequencies. If one looks at the remaining and
unplatted samples from overbank alluvium or buried soils from any of the sites the
rainfall estimates appear to be low in these contexts. This is probably a preservation
problem due the alternative wetting and drying of pollen in soils, and it probably
biased the relative frequencies of the samples (Holloway 1989). The estimate for the
BFS upper buried soil could be 'calibrated' to an average of the contemporaneous hyrax
midden estimates, and a coefficient calculated. However this does not really address
the problem, and goes over the edge as too much data manipulation. It appears that the
hyrax middens, the stream deposits, and the pond deposits trap and preserve pollen
reasonably well. However, in water laid sediments pollen is deposited as any clast or
particle, and the strength of the current influences the size of pollen trapped in a
stratum (Fall 1987). This suggests that a bias might occur against the deposition of
very small pollen in flowing water.
More research is needed to investigate
taphonomic problems, and develop strategies for dealing with biased samples. In
Blydefontein Basin it appears that hyrax middens and pond deposits are the best pollen
traps and preservers. Soils appear to be the worst context and this is probably a
preservation problem The taphonomic characteristics of alluvium and stream
deposits require more study. Even considering these problems it is worthwhile to
148
inspect the rainfall estimations for pollen samples from hyrax middens and pond
deposits (Figure 52).
The resultant rainfall record has some large gaps and inconsistencies (see Figure
52). Meerkat Midden estimates were omitted due clarity as all the samples crowd
together near the 0 B.P. mark. Quantitative estimates were not possible for
Pleistocene pollen samples from the Older Fills, because of significant differences in
200
•
150
Rainfall
Difference 1 00
from
Modern
50
(mm)
0
Jru
0
MUPKイセ@
0
1·•-
1000 2000 3000 4000 5000 6000 7000 8000
Years BP
Oppermanskop ·0- USP
·•- BSM & CH2
Figure 52. Combined rainfall estimates from hyrax middens and pond deposits except
Meerkat Midden.
the structure of Pleistocene plant communities (see above and Chapter V). The
J
earliest rainfall estimate dates at 7790 B.P. from CH2, and indicates conditions about
the same as present. Between 6500-5500 the estimates from BSM suggest that,
except for this century, this is the driest period recorded in the Holocene at
Blydefontein Basin. By 5000 B.P. the estimates from BSM indicated that the Basin
was much better watered. Between 5000-4000 B.P. another dry period occurs, and
the youngest sample could indicate a wetter period although more samples are needed
to be sure. Approximately 2000 years pass before another acceptable documented
149
record exists. The only samples from this intervening period are soils from BFS and
HDS, and these samples appear to be biased by preservation. The 2000 B.P. USP pond
sequence shows a moderately moist sequence that is split by a short drought. This
drought is well recorded by diatoms and molluscs from USP (Bousman et. al. 1988).
The Oppermanskop Midden estimates show a sharp reduction in rainfall at
approximately 1300 B.P. which is followed by an equally sharp rainfall increase soon
thereafter.
Then there is a general decline in precipitation from approximately 900
B.P. until the present century.
Pollen Rainfall Estimates Compared to Historic Precipitation Records
The Meerkat rainfall estimates overlap with the composite rainfall records
discussed in Chapter Ill: the local Grapevale rainfall record spans 1920-1985, the
rainfall record from Aliwal North
extends from 1884-1940, and the historic record
from Eastern Cape that spans the years of 1821-1900 (C. Vogel 1988, 1989). In
order to gauge the accuracy of the pollen derived estimates, comparisons with these
records are required.
The comparisons between the smoothed Eastern Cape, Aliwal North, and Grapevale
rainfall records, and the Meerkat pollen derived rainfall estimates are reasonably
close (Figure 53). In this comparison it was assumed that the East Cape historic
records could be converted to actual rainfall estimates by using the scales on Figure
15 (rain estimate= (111.111 *East Cape Score)+ 27.778).
Obviously the pollen
samples reflect a much grosser sampling interval than the rainfall records, and a
great amount of variability is not reflected in the pollen derived estimates. These
longer sampling intervals also would tend to smooth the curve, and it is why the pollen
rainfall estimates are more constricted than the historic or measured rainfall moving
averages. The hyrax middens were sampled at irregular intervals, and thus the
150
samples represent different time spans. An improvement of detail and accuracy would
be to date a midden first, and calculate accumulation rates. Then, based on the
accumulation rates, select samples of given time spans, and compare these to rainfall
measures averaged by a similar time interval. This would entail much more detailed
sampling and chronological control than achieved by the middens used in the analysis
presented here.
250
250
200
200
150
150
100
50
0
100
50
•
0
-50
-50
-100
-1 00
-150
-150
-200
-200
-250
-250
1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000
Age
·•- Meerkat Pollen
Rainfall Estimates
-
Grapevale & Aliwal
North
-
East Cape Estimated
Rainfall
Figure 53. 5-yr moving averages for Vogel's Eastern Cape and Grapevale rainfall
records, and Meerkat hyrax midden rainfall estimates.
In Figure 53 a number of important patterns are present in the pollen derived
rainfall estimates from Meerkat Midden.
Firstly, the 20th century increase in
rainfall is very clear. Also an inspection of the 19th century portion of the record
suggests that a slight decline is present as well. When considering the pollen
estimates along with the historic records it seems reasonable to suggest that at least
one and possibly two long rainfall cycles are present. The pollen data would suggest
Rainfall
(mm) from
Modern
Average
1 51
that it begins ca. 1850, and declines to the late 1920s, and then rapidly rises until
the 1970s, but the rainfall records suggest two cycles. The pollen estimates provide
another link between the two drastically different types of historic records.
The similarity between the pollen rainfall estimates and the historic rainfall
records, both measured and estimated, provides credence to the technique of estimating
climatic parameters with pollen data. The contention that vegetation response to
European veld abuse overwhelmed or nullified vegetation responses to climatic
fluctuations in the last 200 years is brought into question by the degree of agreement
between the pollen estimates and the historic rainfall records. The possibility that
man and climate acted in tandem to dramatically alter the indigenous vegetation in the
19th and 2oth centuries appears to be a more accurate assessment of changes in
vegetation observed over the same period. Lastly, in the future pollen records may
help to provide a bridge for calibrating other 19th century quasi-quantitative rainfall
record based on documentary evident with actual rainfall measurements collected
systematically in the 20th century.
Comparison Between Pollen-Derived Rainfall Estimates and o13Q Seguences
In order to see more clearly the similarities and differences between the eggshell
o13Q values and the rainfall estimates, rainfall estimates from two pond deposits (USP
and BSM) and the Oppermanskop hyrax dung midden were used as these depositional
contexts have the most reliable and continuous pollen samples (Bousman et al. 1988;
Scott and Bousman 1990). As stated above dry periods in the rainfall estimates date to
approximately 4100-4200 B.P. at BSM, 3200 B.P., 2000 B.P. at USP, and 1300
B.P. as well as the modern 20th century drought at Oppermanskop. Drier rainfall
estimates are closely matched by negative dips in the eggshell s13Q curve (Figure
54). Generally samples with high rainfall estimates are characterized by pollen
152
spectra with high grass relative frequencies, while pollen spectra producing low
rainfall estimates are dominated by composits. In this portion of southern Africa
composits consist mostly of small, drought resistant, mainly C3 bushes typical of the
Nama-Karoo biome, and grasses include both C3 and C4 species and especially the C4
species Themeda trianda (Cowling 1983; Rutherford and Westfall i 986; Sealy
1986; Vogel et al. i 978). During period when rainfall is predicted to be high
because of high grass relative frequencies the o13c curve is also elevated, and during
periods of low estimated rainfall associated with increased composits and other Karoo
flora the o13c values are low. These data can be used to suggest that the biotic changes
reflected in the pollen spectra are also reflected in the eggshell o13c values. The two
are linked to the same phenomena.
200
-3
0
150
-4
0
Rainfall
Difference
from
Modern
(mm)
100
0}1 i!
50
0
I
011
.. /. ..
\;
-1 00
0
-6
..
セBGヲj@
JD
•"\
•, _,_..r"
1
1•1'
•r
••
o
-50
d|ゥアLセ@
-5
/
...
t
•
D
bo/
'a 'o
-7
-8
-9
1000 2000 3000 4000 5000 6000 7000 8000
Years BP
1·0-
Oppermanskop ·•- USP
·D- BSM & CH2
·•- CES
Figure 54 Pollen derived rainfall estimates and average ostrich eggshell d13c values.
As argued in Chapter VI, similarities between ostrich eggshell and sediment
humate o13C curves imply that major shifts in relative frequencies of C3 and C4
153
plants are registered by both mediums. Different degrees of isotopic fractionation
occurs in both materials, but, nevertheless, reasonable correlations are possible. The
s13c changes registered
in ostrich eggshell and sediment humates are supported by
fluctuations in contemporaneous pollen spectra as well. Based on these data it appears
that Holocene biotic changes in the upper Oorlogspoort drainage were registered by all
three independent sources of evidence. Thus carbon isotopes in ostrich eggshell can
provide important proxy palaeoenvironmental data from a wealth of archaeological
sites, and often when no other evidence is available. Ostrich eggshell has the added
benefit of being directly associated with archaeological remains which allows a direct
comparison between biotic and human behavioral changes.
A Test of the Climatic Model
The palynologically estimated rainfall record can be compared to the COHMAP
simulations for a visual assessment of the similarity between the two (Figure 54).
The temporal resolution of the simulated rainfalls is very low by comparison to the
pollen samples. Nonetheless, Figure 54 demonstrates a reasonable fit between the
rainfall estimates derived from the pollen spectra and the COHMAP simulations. No
major conflict is apparent between the simulated rainfall record and the pollen data,
but the Early Holocene COHMAP estimates appear to overestimate the drought.
Without doubt this is a period when a continuous and complete record could add a great
deal. It should be remembered that the computer simulations are for all of southern
Africa below 20°8 latitude, and the Blydefontein rainfall estimates apply to only a
very small portion of this region.
Also short term climatic events of 100-1000
years duration are not factored into the climatic model, and events of this scale
probably dominate the pollen derived rainfall estimates and single floating samples
could provide a very biased view of the general rainfall during a period. This
154
comparison provides good circumstantial evidence that seasonal variations in solar
insolation influenced climates during the Holocene, and that the simulations may be
roughly correct for the Pleistocene when pollen samples are very scarce. The pollen
samples from the Older Fills all reflect cold and dry conditions, and if the COHMAP
simulations are correct for the Pleistocene then these samples could date more
recently than 15kya, or considering the clay mineralogy, the Older Fills could predate
the wet period predicted for the Last Glacial Maximum. However, these samples must
be dated by radiocarbon before they can be used to test the Pleistocene segment of the
COHMAP simulations.
200
•
150
Rainfall
Difference
from
Modern
(mm)
MQUPKセ@
0
1000 2000 3000 4000 5000 6000 7000 8000 9000
Years BP
·•- Oppermanskop ·0- USP
·•- BSM & CH2
·D- COHMAP
Simulations
Figure 55 Comparison of palynologically estimated rainfall record and COHMAP
simulated rainfall record.
155
Conclusions
Climatic controls on modern taxa can be used to estimate temperature and rainfall
for each pollen sample. These rainfall estimates indicate a major drought between
6500 B.P. and 5500 B.P., and another occurred between 5000 B.P. and 4000 B.P.
Throughout the remainder of the Late Holocene rainfall appears to be high except for a
very short period at about 2000 B.P., another period at 1300 B.P., and possibly a
minor one at 3200 B.P. This long period of high rainfalls ends in the later part of the
19th century according to historic and measured rainfall records. The 20th century
is dry. The timing and intensity of these climatic estimates can be used to understand
at least some of the causes behind prehistoric human behavior in the region during the
Late Stone Age.
Thus the comparison between the rainfall estimates derived from pollen samples
and COHMAP simulated rainfalls shows a reasonably good fit. This suggests that the
long term climatic changes of the Quaternary are influenced by variations in solar
insolation caused by perturbations in the orbit of the earth around the sun. This has
been well documented for the comings and goings of glaciers during the Pleistocene
(Hays, Imbrie and Shackleton 1976), and strongly implicated for Holocene climates
in other regions as well (Kutzbach 1981 ). The data presented here provide unusually
clear evidence that the system is detectable in unglaciated regions. It also shows
clearly that the system continues today.
Short variations are not well understood (Berger 1988), but considering the
amazing overriding influence of solar radiation one should not rule out short term
astronomical effects (Gribbin 1978: 150-154).
Other factors such as major
volcanic eruptions (Bryson 1989) or rapid surges in glacial meltwaters at the end of
the Pleistocene (Emiliani et al. 1975; Jones and Ruddiman 1982; Schneider 1987)
could also have short term effects on worldwide climates.
CHAPTER VIII
THE ROCKSHELTERS: STRATIGRAPHY, CHRONOLOGY, AND SEDIMENTS
Two archaeological sites were excavated in Blydefontein Basin in 1985:
Blydefontein Rockshelter and Meerkat Rockshelter. Blydefontein Rockshelter had been
excavated twice previously. The first excavation was by A. C. Hoffman and D. J.
Esterhuyse in the 1957, and the second excavation was by C. G. Sampson in 1967.
Hoffman and Esterhuyse kept no field records and never published their results.
Sampson reanalyzed their artifacts and published them along with the results of his
excavations (Sampson 1970: 87-1 05). Based on the results of Sampson's Orange
River Scheme research (Sampson 1970, 1972 and 1974), and on the Zeekoe River
Valley survey (Sampson 1985}, it became obvious that Blydefontein Rockshelter was
one of the largest and potentially deepest rockshelters in the entire region that had the
added benefits of good faunal preservation and intact stratigraphy (Klein 1979;
Sampson 1970: 88-89}.
In order to obtain a time-calibrated settlement pattern
from the Zeekoe Valley surface sites, it was clear that a cross-dating scheme with
tight chronological controls was needed for the region. Blydefontein Rockshelter, with
its good stratigraphy, could be the corner stone of such a cross-dating scheme.
The shelter was revisited in 1980 during the first season of the Zeekoe Valley
Project, and excavation plans were begun at that time.
Funds were secured in 1984
for renewed excavation of Blydefontein Rockshelter, and in early 1985 C. G. Sampson
revisited Blydefontein in order to liaise with the landowner, Mr. D. Lessing. At that
time Mr. Lessing showed Sampson an unknown smaller rockshelter, later named
Meerkat Rockshelter. Meerkat Rockshelter had never been investigated.
156
157
Blydefontejn Rockshelter Excavations
Blydefontein Rockshelter is situated in a sandstone overhang with two main lobes:
north and south (Figure 56). The deposits in the northern lobe are shallow, but the
deposits in the southern lobe are at least one meter or more in depth. The shelter
talus deposits slope down to the edge of a terrace where a historic stone wall stands.
Beyond the stone wall is the 6 meter thick alluvial terrace known as Blydefontein
Section (see Chapter V).
Three areas were excavated at Blydefontein Rockshelter in
1985 (Figure 57). The major excavation effort was placed adjacent to Sampson's
excavations in the south lobe, while a single one-by-one meter test pit was placed in
60
I
/
cliff edge
40
8> 20
.§.
excavation
block
.c
t::
0
z
/
セ@
Q)
cu
2
(
0
terroce
セZ|G@
::oc::1
ァ・ッャセZ@
.,..'-.
-20
·'-. .. """"-.
-··-...... ..
-60
-40
-20
0
Meters East (mag.)
Figure 56. Map of Blydefontein Rockshelter and surrounding terrace area.
0
158
geological
trench
-5
talus
1x1m unit
D
-10
0)
0
0
.s
..c
t
Blockm
Sampson's
pit
Sampson's
backdirt
セ@
z0
p
Block C
BlockS
Block D
0 •
0
•
0
Oo o
0 0
o0 o
-15
0
Ul
._
stone wall
Q)
......
Q)
2
-20
""'shelter and cliff wall
-25
-30
-15
-5
0
5
10
15
Meters East (mag.)
Figure 57. Map of current and previous excavation units at Blydefontein Rockshelter.
-10
front of the south lobe excavations in the talus deposits. A large deep geological trench
was placed in front of the one-by-one meter test trench on the terrace surface. The
geological trench had the same stratigraphy as the alluvial cutbank in front of the
shelter known as Blydefontein Section.
Excavation Methods
Sampson's pit was relocated, and permanent datum established on a large flat
boulder laying midway between the north and south lobes of the shelter. A grid system
aligned with magnetic north and centered on the datum was established over the site.
159
Then the site was mapped with a transit. Even though Sampson had backfilled the pit
with large sandstone slabs, erosion had slumped the uppermost sediments, especially
on the outside (west) side of the pit. The slabs were removed and the profiles cleaned.
I placed a two-by-two meter excavation block (Block A) over Sampson's five-by-five
foot (5 feet = 1.524 meter) pit. Three additional two-by-two meter blocks (B
through D) were laid out on the south and east side of Sampson's pit with blocks C and
D on the inside of the shelter on the east, and blocks A and B on the out (west) side.
The two-by-two meter blocks were subdivided into 25-by-25 em excavation units.
Each two-by-two meter block had 64 smaller excavation unit blocks. The sediments
were excavated by arbitrary unit, numbered from top to bottom in sequence. When
possible, arbitrary units were subdivided by natural stratigraphic layer.
The
notation system used can best be described by example: B53.2 indicates Block B,
excavation unit 5, arbitrary level 3, and natural layer 2. The sediments west of
blocks C and D in blocks A and B are heavily leached and few natural stratigraphic
boundaries can be deciphered. Excavation proceeded out from Sampson's pit walls.
Stratigraphy
Seven major stratigraphic layers were observed in the main excavation block at
Blydefontein Rockshelter. These layers were defined in the field by Munsell soil
colors and sediment texture. At least one sediment sample from each of the layers
were submitted to the Soils and Physical Geography Lab, University of WisconsinMilwaukee for quantitative textural analysis and phosphate analysis. The textural
analysis calculated the relative frequency (i.e. percentages) by weight of seven
sediment texture classes. The size classes are gravels, very coarse sand, coarse sand,
medium sand, fine sand, very fine sand, and silt and clay. A phosphate fraction
analysis was undertaken to identify the nature of two of the layers in Blydefontein
160
Rockshelter (Eidt 1977).
Every major stratigraphic layer contained minor
stratigraphic units which usually consisted of ash or charcoal lenses. These were not
analyzed individually. As shown in Figure 58 and from top to bottom the major layers
are: Surface Dust (SD), Hard Grey (HG), Tan Sand (TS), Gray Ash & Charcoal (GAG),
Tan GriVCharcoal Ash & Crab (TG/CAC) and Compact Yellow (CY). Figures 59-66
and Figures 67-72 provide histograms of the sediment texture analysis .
.L.aw Descriptions
SD is a loose dark brown to dark grayish brown (1 OYR 3/3 to 4/2) loam on the
surface. Sediment textures are weakly bimodal with approximately 60 percent of
sediments are smaller or equal to fine sands (Figure 59). Modern sheep dung pellets
are found throughout SD and this sediment is the surface deposits that are churned by
modern sheep or other animals trampling.
The second layer, HG, is a slightly more firm dark grayish brown to dark gray
(1 OYR 4/2 to 4/1) loam composed mostly of fine sand, and silts and clays (Figure
60).
CPS occurs in the northeast corner of the main excavation block at Blydefontein,
and artifactually it is almost sterile. This is a very light and fluffy, but firm dark
gray (1 OYR 4/1) loam composed of clays, silts, very fine sand, fine sand, and medium
sands (Figure 61 ). This layer is sandwiched between HG.
TS is a discontinuous, almost sterile yellowish brown to light brownish gray
(1 OYR 5/4 to 6/2) loose sand composed of medium sand, fine sand, very fine sand,
silt and clay (Figure 62). Fragments of partially unconsolidated swallow nest were
found in this layer, and the textural analysis of a modern swallow nest is more like TS
than any other layer in Blydefontein Rockshelter (see Figure 63). A number of
swallow nests were observed on the roof of the rockshelter during excavation and it is
!-
-7.0 N
-6.0 N
-9.0 N
-8.0 N
844 BP
セ@
セZNLG@
/
TG
3135 BP
セ@
-4101 BP
CAC
......:
CY
4286 BP
Figure 58.
Stratigraphy at Blydefontein Rockshelter.
en
_._
162
likely that swallows also used the rockshelter in prehistoric times. It is suggested
that TS is mainly decomposed swallow nest, and represents a stratigraphic marker
between Layers HG and GAC in this portion of the rockshelter deposits.
GAC is a dark gray (1 OYR 4/1) sandy loam with a bimodal distribution (Figure
64). Distribution peaks occur in gravel sized particles and very fine sands. Abundant
charcoal from prehistoric hearths characterized this layer, and clearly GAC
represents a period of intensive prehistoric occupation.
TG is a brown to grayish brown (1 OYR 5/3 to 5/2) loam with a single modal peak
in the fine sands range near the top of the layer and an increase in very fine sands and
silt/clay toward the bottom of the layer (Figures 65 and 66).
CAC is a hearth complex in the bottom of and interfingering with Layer TG.
Originally it was thought to be a separate layer, but radiocarbon dates (discussed
below), and stable carbon isotope analysis of ostrich eggshell from excavation units in
both sedimentary units all point to Layer CAC as overlapping and coeval with the
bottom of Layer TG (Figure 67). The sediment analysis indicates that CAC is very
similar to the lower TG sample as well with almost equal amounts of fine sands, very
fine sands and silt/clay (Figure 68).
CY is a thick yellow silt loam layer that was divided into subunits. Three brown
loam sub-layers (Brown 1, Brown 2 and Brown 3 from top to bottom) were
discovered in the lower half of this unit, but only the Brown 2 layer extended across
the entire east profile. The yellow silts (Upper Yellow) were sampled above the
Brown 1 layer, between the Brown 1 and 2 layers (Middle Yellow), and between the
Brown 2 and 3 layers (Lower Yellow). Upper Yellow is a brownish yellow to
yellowish brown (1 OYR 6/6 to 5/6) silt loam with the highest percentage of material
in the silt/clay texture clay and stepwise reductions of material in each larger texture
class (Figure 69). The large amounts of silt/clay may be due to decomposition, but as
163
the bedrock is sandstone, it is more likely that the greater amount of silt/clay in this
layer represents an increased amount of wind deposited materials. Neither this layer,
nor any sub-layer in CY has evidence of intense human occupation.
Middle Yellow is a very pale brown to light yellowish brown (1 OYR 7/4 to 6/4)
silt loam (Figure 70).
It's textural distribution is very similar to that in the Upper
Yellow with the greatest percentage of material occurring within the silt/clay size
range and decreasingly less material in each larger size class.
Brown 2 is a brown (1 OYR 5/3 to 7.5YR 5/2) loam. The sediment texture of
Brown 2 is slightly more coarse than the texture of Upper or Middle Yellow (Figure
71 ). It is assumed that the textures of Brown 1 and Brown 3 would be of similar
texture to Brown 2.
Lower Yellow is a reddish yellow to brown (7.5YR 6/6 to 5/4) loam. The texture
distribution shows a weakly bimodal distribution with a major peak of very fine sands
and a weak gravel peak (Figure 72).
The sediments in CY, except for the Brown sub-layers, are believed to represent
natural depositional processes with most of the material occurring in the silt/clay
range. These sediments probably were deposited through aeolian processes. In layers
with more intensive human occupations coarser grained materials occur. Usually the
fine sands form the largest mode and a second mode occurs in the gravels. As the
analysis of the swallow's nest indicates, at least one source of the fine sands could be
decomposed swallow's nests, and it is possible that humans dislodged nests stuck to the
overhang roof incorporating swallow nests into the sediment matrix at a faster rate
that would normally occur. It should be mentioned that in some layers, eg. CAC and
GAC, charcoal, ash, artifacts, and faunal remains formed a significant amount of the
matrix, and that true sediment, ie. gravels, sand, silt and clay, was not the dominant
constituent of these layers. Another texture size that sometimes forms a secondary
164
mode in the texture analysis is gravels. One obvious source is from the walls and roof
of the shelter itself. It is also possible that the prehistoric inhabitants also tracked
in a significant amount of this material as gravel modes seem to occur in layers with
more intensive occupations.
70
Blydefontein, SD
60
50
Relative
Frequency
40
30
20
10
0
GRVL
vcs
cs
MS
FS
セsilt@
VFS
Figure 59. Sediment texture analysis of layer SD, Blydefontein Rockshelter.
70
Blydefontein, HG
60
50
Relative
Frequency
40
30
20
10
0
GRVL
VCS
cs
MS
FS
VFS
セsilt@
Figure 60. Sediment texture analysis of layer HG, Blydefontein Rockshelter.
165
70
bャケ、・ヲッョエセゥL@
CPB
60
50
40
rセャ。エゥカ@
fイセアオョ」ケ@
30
20
10
0
GRVL
vcs
cs
MS
FS
VFS
s.SIL T
Figure 61. Sediment texture analysis of layer CPB, Blydefontein Rockshelter.
70
bャケ、セヲッョエ・ゥL@
TS
60
50
fイセアオョ」ケ@
rセャ。エゥカ@
40
30
20
10
0
GRVL
VCS
cs
MS
FS
VFS
セsilt@
Figure 62. Sediment texture analysis of layer TS, Blydefontein Rockshelter.
166
70
Blydefontein, Modern Swallow Nest
60
50
Relative
Frequency
40
30
20
10
0
GRYL
YCS
cs
MS
FS
YFS
iSILT
Figure 63. Sediment texture analysis of modern swallows nest from Blydefontein
Rockshelter.
70
Blydefontein, GAC
60
50
Relative
Frequency
40
30
20
10
0
GRYL
YCS
cs
MS
FS
YFS
iSILT
Figure 64. Sediment texture analysis of layer GAC, Blydefontein Rockshelter.
167
70
Bly de-fonte-in, TG 1
60
50
Re-lativeFre-que-ncy
40
30
20
10
0
GRVL
VCS
cs
MS
FS
VFS
iSILT
Figure 65. Sediment texture analysis of layer upper TG, Blydefontein Rockshelter.
70
Bly de-fonte-in, TG2
60
50
Re-lativeFreque-ncy
40
30
20
10
0
GRVL
VCS
cs
MS
FS
VFS
iSILT
Figure 66. Sediment texture analysis of layer lower TG, Blydefontein Rockshelter.
168
22
&.CAC
DTG
20
18
16
BJ
14
12
10
8
6
-9.5
-8.5
-9
-7.5
-8
-7
-6.5
-6
d13C
Figure 67. Ostrich eggshell d13C values from Layers TG and CAC plotted by excavation
units (EU). Excavation units plotted in reverse order from top to bottom.
70
Blydefontein, CAC
60
50
Relative
Frequency
40
30
20
10
0
GRVL
VCS
cs
MS
FS
VFS
セsilt@
Figure 68. Sediment texture analysis of layer CAC, Blydefontein Rockshelter.
169
70
Blydefontein, CY, upper yellow
60
50
Relative
Frequency
40
30
20
10
a•-·
GRVL
VCS
cs
MS
FS
VFS
セsilt@
Figure 69. Sediment texture analysis of layer upper CY, Blydefontein Rockshelter.
70
Blydefontein, CY, middle yellow
60
50
Relative
Frequency
40
30
20
10
0
GRVL
VCS
cs
MS
FS
VFS
セsilt@
Figure 70. Sediment texture analysis of layer middle CY, Blydefontein Rockshelter.
170
70
Bly defontein 1 CY, brown 2
60
50
Relative
Frequency
40
30
20
10
0
GRVL
VCS
cs
FS
MS
VFS
iSILT
Figure 71. Sediment texture analysis of layer Brown 2 in CY, Blydefontein Rockshelter.
70
Bly defontein 1 CY, lower yellow
60
50
Relative
Frequency
40
30
20
10
0
GRVL
VCS
cs
MS
FS
VFS
iSILT
Figure 72. Sediment texture analysis of layer lower CY, Blydefontein Rockshelter.
1 71
Radiocarbon Chronology
Nine radiocarbon assays were run on charcoal from Blydefontein Rockshelter
(Table 17). All samples were charcoal and collected from discrete occurrences within
a single 25-by-25cm excavation unit. The dates were calculated with the 5568 year
half-life, corrected for carbon isotope fractionation, and calibrated with the Stuiver
and Pearson (1986} and Person and Stuiver (1986} calibration. An attempt was
made to date the tops and bottoms of layers as well as select samples with abundant
charcoal remains.
Table 17.--Radiocarbon dates from Blydefontein Rockshelter. Only approximate
calibrations are possible for dates older than 8100 B.P.
l。セ・イ@
ExQavatiQn Unit Lab Number
QQrreQted B.P.
Calibrated B.P.
1-G
C22
SMU-1902
844±119
768±105
CPB
C10s
SMU-1925
1255±109
1187±112
1-G
C42s
SMU-1850
1305±31
1256±36
GlC
C50s
SMU-1853
2292±117
2318±141
GlC
D18s
SMU-1849
3135±33
3372±33
lG
C1814
SMU-1901
4101±273
4613±365
CIC
02616
SMU-1851
4066±55
4578±108
CIC
D172o
SMU-1852
4286±149
4855±193
()(
C5723
SMU-1823
8541±417
ca. 9440
The results of the radiocarbon dates suggest that the accumulation of deposits
represents a coherent chronological succession. No datable charcoal was recovered
from the bottom of CY. The single radiocarbon date is from Upper Yellow sub-layer
and it indicates that this portion of Layer CY dates to the Early Holocene. Three
radiocarbon ages are available for TG/CAC. Unfortunately at the time the radiocarbon
samples were selected it was believed that TG and CAC were separate layers, and
samples from the top and bottom of Layer CAC were dated by radiocarbon but only the
middle of Layer TG. This resulted in no radiocarbon date for the top of Layer TG.
172
Radiocarbon samples from the top and bottom of Layer GAC were dated, as well as top
and bottom samples from HG. These dates suggest that Layer TG/CAC began to
accumulate at approximately 4300 B.P. Layer GAC accumulated between 3135 B.P.
and 2292 B.P. and possibly as late as 2000 B.P. as implied by Sampson's radiocarbon
date 1980±120 B.P. (SR-132) that appears to come from the top of Layer GAC.
Layer HG appears to have accumulated between 1305 B.P. and 844 B.P. with the age of
CPB estimated at 1255 B.P. Sediment deposition dramatically slowed after 844 B.P.
as evidence by the thinness of the surface Layer SD. The lack of sediments after 844
B.P. is believed to be due to the lack of human occupation in Blydefontein Rockshelter
during this period.
Phosphate Analysis
Phosphate fraction analysis (Eidt 1977: 1327-1333, 1985: 155-190) was
undertaken in order to understand better the mode of sediment deposition in
Blydefontein Rockshelter. Based on field observations two layers, CPB and TS,
appeared to originate from animal sources.
Analysis of the inorganic phosphates from
these deposits was undertaken to help assess the possibility of contamination from
human or animal sources. Inorganic phosphates can be divided into three types or
fractions. The first fraction is the available or non-occluded phosphates. This
fraction is further divided into phosphates loosely bound to aluminum and iron
(Fraction 1a), and phosphates (re)sorbed by calcium carbonate (Fraction 1b).
Fraction II is the occluded or tightly bound phosphate incorporated with aluminum and
iron oxides, and hydrous oxides. Fraction Ill is apatite or other tightly occluded
calcium phosphates such as that found in the inorganic fraction of bone or in mineral
grains. These individual fractions summed equal the total inorganic phosphates. Eidt
(1977) indicates that Fraction I phosphates change to Fraction II phosphates
173
systematically through time, and thus the ratio of Fraction II to Fraction I is time
dependent.
Unfortunately sediment samples were too small for phosphate fractionation
analysis of all layers at Blydefontein Rockshelter. Four layers and a series of control
samples were selected for analysis. An inspection of total inorganic phosphates in
Table 18 shows that the SO and CPB samples have high levels of total inorganic
phosphate, and these compare well with levels from modern hyrax and sheep dung. As
the SO sample had modern sheep dung mixed with the sediments, its high total
inorganic phosphate level is not surprising. The total phosphate level of CPB strongly
suggests that animal dung is a major constituent of this deposit, and the field
observations support this hypothesis. The two most obvious animal species that could
be responsible are hyrax or domestic sheep.
Prehistorically many rockshelters were used by sheep herders (Hall and Smith
1986; Hart 1989; Schweitzer 1979), and at least one shelter, Boomplaas, has
burned sheep dung that makes up a significant portion of its deposits (H. Deacon et al.
1978). Prehistorically sheep are documented in the coastal zone of southern Africa
from approximately 2000 B.P. (Klein and Cruz- Uribe 1989; Schweitzer and Scott
1973), so the possibility of sheep are responsible for the deposition of CPB is
chronologically feasible. Blydefontein shelter was used by modern herders as a kraal,
but the cliffs surrounding the shelter are also favored by hyrax for habitation.
Unfortunately the phosphate analysis cannot identify which animal species contributed
the phosphates in the CPB layer. However hyrax middens have significant amounts of
well preserved pollen, but CPB had little pollen and it was poorly preserved (Louis
Scott personal communication). Nevertheless no sheep bones were recovered from the
rockshelters (Richard Klein personal communication), so more analysis is required
174
before this layer can be definitely identified as sheep dung and thus indicative of a
herder occupation.
Table 1a.--Phosphate fractions (0 /oo) of selected sediment samples from
Blydefontein Rockshelter. Percent of Fraction II is minus Fraction Ill
Sample
so
Fraction Ia
Fraction lb
Fraction II(%) Fraction Ill
Total
777
2510
767
404(29.1) 1059
2444
33
289
284(46.9)
854
1461
0(
35
360
440(52.7) 1 130
1965
Modern
Swallow nest
26
30
20(26.3)
422
499
Modern
Hyrax Midden 250
472
279(27.9)
272
1273
Modern
Hyrax Midden 434
1412
253(12.1)
512
2611
Modern
Sheep Dung
2800
174(5.5)
93
3280
16
1304
215
TS
CPB
213
41 3(23.8)
Analytical Units at Blydefontein Rockshelter
As Blydefontein was excavated in such a disjointed manner the analysis required
that the site be considered as at least two separate units. The artifacts from Block B
were analyzed as a unit in the arbitrary spits in which they were dug with the
exceptions that B4.5 levels were combined with B4 levels, and Bs.5 levels were
lumped with Bs levels. In addition some levels, usually B12 levels, were split
stratigraphically when Layer CY was encountered. Blocks C, D and a portion of A,
hereafter called Block C-D, were much more complex to correlate. The approach used
was to arrange analytical units by arbitrary spits within major layers, eg. GAC, and
then arrange each stratigraphic layer's stack end-on-end in their correct vertical
order. Artifacts from some of the Block A spits were not included in this analysis.
This was done because these artifacts came from those spits in the northwest corner of
175
Block A where no stratigraphy survived and I attempted to teach the crew to excavated.
This experiment resulted in irregularly excavated arbitrary spits that could not be
correlated with the remainder of the excavations without a dramatic loss of
stratigraphic resolution. As artifact numbers were low in this portion of Block A,
artifacts from these excavation units were dropped from the analysis. The resulting
units, called Analytical Units (AU) were used for all analysis, and these are shown in
Figure 73.
The integration of Block B with Block C-D required a loss of stratigraphic
resolution, however because of small artifact numbers the resulting Combined
Analytical Units (CAU) form the basic units for most of the analysis in the final
chapter. The Combined Analytical Units were consolidated by visually matching the
placement of the excavation units for Block B with Block C-D for the upper layers. In
Layer CY, because of low artifact numbers and sporadic distributions, artifact
numbers among the arbitrary spits were inspected in Block B which did not have any
continuous Brown sub-layers recognized during excavation. The vertically and
horizontally clustered distributions of artifacts were used to define CAUs in Layer CY
(Figure 74).
Meerkat Rockshelter Excavations
Meerkat Rockshelter is approximately 0.75km south of Blydefontein Rockshelter.
It is a much smaller overhang and nearer to the stream channel. Terrace deposits
appear to extend below the overhang, but excavations did not reach fluvial sediments.
Two adjacent one-by-one meter test pits were excavated at Meerkat Rockshelter.
Meerkat was excavated in order to support and confirm the Blydefontein Rockshelter
sequence.
176
.....
Q)
Q)
..c
(f)
..lC
()
0
a:
c:
-
·a;
c:
Q
Q)
u
セ@
CJ
Cl3
(f)
·-=c:
::J
cti
()
:;::::
>.
cti
c:
<(
177
....:
Q)
.:!::::
Q)
..c
U)
セ@
()
0
a:
-
.!:::
Q)
c
.2
Q)
"0
セ@
Ill
セ@
Q)
....
::J
.2>
LL.
178
Excavation Methods
Two one-by-one meter blocks were excavated at Meerkat Rockshelter (Figure
75).
Each one-by-one meter block was subdivided into 25-by-25 em excavation
units producing 16 smaller excavation unit blocks. The sediments were excavated by
arbitrary unit, numbered from top to bottom in sequence. When possible, arbitrary
units were subdivided by natural stratigraphic layer. The notation system is the same
used at Blydefontein Rockshelter and can best be described by example: A6s.2
indicates Block A, excavation unit 6, arbitrary level 5, and natural layer 2 within
level 5.
20
I
I
I
I
I
I
terrace edge
I
I
I
I
I
I
I
I
15
I
I
\
g;
.s
..c
t
0
z
\
\
0
10
\
\
I
t
I
I
excavation units
I
I
....
<ll
(/)
I
/
......
<ll
2
I
shelter bock wall
I
I
I
I
I
I
5
/
I
/drip line
/
," "
0
l⦅セM@
0
5
10
15
Meters East (mag.)
Figure 75 Map of Meerkat Rockshelter.
20
179
Stratigraphy
The stratigraphy at Meerkat Shelter was not as complex as at Blydefontein. Five
major layers were identified (Figure 76). These layers were defined in the field by
soil color (Munsell Soil Color Chart) and sediment texture. At least one sediment
sample from each of the layers were submitted to the State Soils Lab, University of
Wisconsin-Milwaukee textural analysis. The same textural classes applied to the
Blydefontein Rockshelter were used for the Meerkat Rockshelter sediments. Some
major stratigraphic layers contained minor stratigraphic units which usually
consisted of ash or charcoal lenses. These were not analyzed individually. From top to
bottom the major layers are: Surface Dust (SD), White Ash (WA), Brown Silt (BS),
Gray Ash & Charcoal (GAC), and Yellow Silt (YS) .
.Law Descriptions
SD is a surface deposit that has been churned by trampling. In a few excavation
units untrampled deposits were excavated and identified as Brown Above Ash. SD and
Brown Above Ash are a grayish brown (1 OYR 5/2) silty sand (Figure 77). These
surface stratigraphic units are separated from deposits below by the White Ash layer,
which is a thin layer of burnt material that might be dung.
----------------
--------979 BP
---
Metate
----
Surface Oust
White Ash
.Brown Silt
_Gray Ash ッョセ@
Charcoal
.Yellow Silt
Figure 76. Stratigraphy at Meerkat Rockshelter.
ROCK
180
Below the thin White Ash layer is a grayish brown (1 OYR 5/2) silt called Brown
Silt (BS). The major components of this layer are very fine sand and silt (Figure
78). Within the Brown Silt layer is a thin lens of gray ashy sediment.
Below the Brown Silt layer is a relatively thick dark grayish brown (1 OYR 4/2)
silt loam layer that consists of interlocking hearths, and scattered charcoal and ash.
This is called Gray Ash and Charcoal (GAC). These sediments are similar to the Brown
Silt layer above with very fine sands and silts, but the Gray Ash and Charcoal Layers
also has a slightly greater amount of gravels (Figure 79}. In the middle of this layer
is a thin brown zone with less charcoal and ash. This allowed layer GAC to be divided
into three minor layers: Upper Gray, Mid Brown and Lower Gray.
Below the complex of hearths in Lower Gray is a thick layer of brown (1 OYR
4/3) silt loam. In the field this was label Yellow Silt and it consists of very fine sands
and silts (Figure 80}.
The sediment analysis at Meerkat Rockshelter demonstrates less textural
variability between layers as compared to Blydefontein Rockshelter. Except for the
surface sample, all samples are dominated by the fine textural classes, and these
layers are very similar texturally to Layer CY at Blydefontein. However Layer CY was
much firmer, and of different color. It is possible that aeolian processes and
slopewash of aeolian sediments contributed a major source of sediments documented in
the Meerkat Rockshelter deposits. Even though Meerkat is very close to the stream no
evidence of fluvial depositional processes was discovered at the level excavated, and in
fact the shelter is approximately 5-6 meters above stream level.
1 81
Meerkat, SD
70
60
50
Relat;ve
Frequency
40
30
20
10
0
1
GRVL
2
3
4
vcs
cs
MS
5
FS
6
VFS
7
iSILT
Figure 77. Sediment texture analysis of layer SD, Meerkat Shelter.
Figure 78. Sediment texture analysis of layer UB, Meerkat Shelter.
182
Meerkat, GAC
70
60
50
Relat;ve
Frequency
40
30
20
10
0
1
GRYL
2
3
4
YCS
cs
MS
5
6
7
FS
YFS
セsilt@
Figure 79. Sediment texture analysis of layer Gray, Meerkat Shelter.
Meerkat, YS
70
60
50
Relat;ve
Frequency
40
30
20
10
0
GRYL
2
3
4
YCS
cs
MS
5
6
FS
YFS
Figure 80. Sediment texture analysis of layer YS, Meerkat Shelter.
セsilt@
7
183
Radiocarbon Chronology
Three charcoal samples were submitted for dating to the Radiocarbon Dating
Laboratory at Southern Methodist University (Table 19). All samples were charcoal
and collected from discrete occurrences within a single 25-by-25 em excavation unit
although these were not clear identifiable hearths. The dates are calculated with the
5568 year half-life, corrected for carbon isotope fractionation, and calibrated with
the Stuiver and Pearson (1986) and Pearson and Stuiver (1986) calibration.
Unlike
the sampling strategy used in Blydefontein Rockshelter, no attempt was made to date
the tops and bottoms of layers. Natural layers in Meerkat were too few and too diffuse
to allow that type of dating scheme. Rather samples were selected based on their
association with diagnostic tools or depth.
Table 19.--Radiocarbon dates from Meerkat Rockshelter
La;iflr
oc
oc
YS
ExQavatiQn Unit
Lab Numbflr
QQrrflQlfld B.P.
Qalibratfld B.P.
B2s
SMU-1899
979±70
837
SMU-1898
1102±62
1015±73
A1312
SMU-1931
2353±122
2389±159
895±77
Summar;i
The results of the stratigraphic investigations suggests that the sediments in to
two shelters accumulated with a significant contribution from the human occupants.
Blydefontein Rockshelter provides the long term view of occupations in the Basin, but
with little possibility of truly fine chronological separation of assemblages. Chemical
analysis of sediments from Blydefontein Rockshelter can be used to suggest a brief
interval of stock herding at 1255 B.P. It is unfortunate that no charcoal was
recovered from the lower portion of CY at Blydefontein as these sediments could date to
the Late Pleistocene as will be suggested below from the archaeological remains
recovered from these sediments. Meerkat, on the other hand, represents a much
184
shorter period of time and has the possibility of providing a much more detailed view
of prehistoric occupations in the Basin, especially within the last 2500 years.
CHAPTER IX
CHRONOLOGICAL MARKERS OF THE INTERIOR WILTON SEQUENCE
About 5000 Interior Wilton sites were discovered in the adjacent Zeekoe valley
(Sampson 1985). Our survey records include "phase" labels for each site (early,
classic, developed or ceramic), based on the presence or absence of diagnostic stone
tool types and ceramics. This was a direct field application of Sampson's (1972,
1974) four-phase system of classification derived from rockshelter sequences,
including Blydefontein, in the Middle Orange River.
Many of the surface sites which we examined and classified by these criteria
emerged as apparently mixed palimpsests of two or more phases. It remains to be seen
whether this is the correct interpretation. Given the unstated assumptions of the
original four-phase system (unilinear evolution), and given the shortcomings (see
Chapter I) of the excavated sites from which it was built (Sampson 1967a, 1967b,
1970, 1972; Sampson and Sampson 1967), our field classifications are of limited
use at present. For this reason, separate distribution maps of each "phase" of the
Interior Wilton have not been published (Sampson 1985: 69).
One of the objectives of the Zeekoe Valley Archaeological Project is to trace
changes in the Interior Wilton (IW) settlement pattern through time.
These in turn
can be compared with the environmental trends summarized in Chapter VII. The
comparison cannot be made, however, unless the existing IW site distribution map can
be broken down chronologically to show trends in the IW settlement pattern. It is of
the utmost importance, therefore, that an improved system of cross-dating of surface
sites be established. Surface sites lack associated organics, so the only viable method
185
186
is the conventional one: to compare stone tool designs with those from a tightly dated
and stratigraphically secure sequence. Blydefontein and Meerkat Rockshelters were
excavated to establish that sequence.
The Artifact Samples
Blydefontein Rockshelter was divided into four blocks. Block A, covering
Sampson's (1970) excavation, was not used in this analysis. Block B is a larger area
where rain and roof drip has leached out the visible stratigraphy. Here, nothing but
the boundary between CY and all the above deposits can been seen. Except at this
boundary, all of Block B was excavated in thin, arbitrary spits only a few centimeters
thick. Blocks C and D, being the most protected, contained finely stratified sediments.
These were excavated by individual micro-layer and some thicker layers were
arbitrarily split into sub-layers.
Although the artifact assemblages from C and D have the best contextual integrity
of any assemblage in the Karoo, some of them are too small for statistical analysis.
was thus compelled to combine them with material from the adjacent Block B.
Correlations between stratified layer assemblages and adjacent spit assemblages was
based on shared elevations. This procedure undoubtedly detracts from the integrity of
the Combined Analytical Units (CAUs), but the compromise is essential to obtain
statistically valid samples that can provide meaningful results. Artifact counts for
individual layers and/or spits in each block are given in Appendix B.
Meerkat Rockshelter was test excavated with two adjacent, 1-meter square
blocks. Time did not permit larger excavations, and the samples are small. Meerkat
had less visible stratigraphy than Blocks C-D in Blydefontein, and it was excavated in
arbitrary spits subdivided by natural stratigraphic boundaries when possible.
Nevertheless, Meerkat artifact distributions have chronological patterning, but small
187
sample size makes the Meerkat results less reliable than those obtained from
Blydefontein Rockshelter.
The Artifact Analysis
Design changes through time of four Interior Wilton artifact classes are
examined. These include backed microliths and projectile points, endscrapers, ostrich
eggshell beads, and ceramics. These artifacts were chosen because observations could
be made easily on surface site assemblages in the field. Also, a number of other
attributes were also found to have temporal patterning (to be discussed in Chapter
XIII), but these were not included here because their documentation on surface site
assemblages would be too time consuming, or their chronological patterns were too
variable, or their inclusion produced no increased resolution for the chronological
placement of undated surface assemblages. Thus the set of artifacts used here provide
easy-to-recognize attributes that display maximum chronological patterning.
Analytical Procedures
The selection of units-of-comparison has generally followed the procedure
identified by Sackett (1982) as isochrestic style analysis. This approach attempts to
analyze changes through time in functionally similar artifacts (as suggested by
morphology). Collaboration of artifact function would benefit through a wear pattern
study of use-damage on artifacts, but ethnographic observations (Clark 1959, 1977;
Dunn 1880, 1931; Kannemeyer 1890; Rudner 1979) and previous wear pattern
studies (Binneman 1982, 1983, 1984; Binneman and Deacon 1986) conducted
locally can be used to infer stone tool functions with a reasonable amount of confidence.
The four groups of formal artifacts analyzed in this chapter were selected, in part,
because they also provide a reasonable amount of morphological variety within
individual functional classes.
188
Backed Tools and Projectile Points
The Interior Wilton artifacts that are most likely to display some form of
stylistic patterning are the backed microliths and projectile points.
Dunn (1880),
Goodwin (1945) and Clark (1959, 1977) discuss the probable uses of these artifacts
through an analysis of ethnographic collections from southern Africa, and it appears
that they are used, for the most part, as arrow armaments. This may be correct in a
general sense, although they may have had secondary uses for a variety of purposes
including drills for ostrich egg shell beads or inserts in composite cutting tools.
Morphological variations have long been recognized and these will be discussed below,
however it should be noted that double backed tools with tip damage indicative of use as
drills were not used in this analysis.
Eight variations of backed tools have been identified. Diagnostic criteria are
backing, location of trimming and flaking technique. Traditionally a distinction has
been made between straight backed bladelets and crescents, also called lunates or
segments (J. Deacon 1972: 14). High frequencies of crescents are known to be
common in mid-Holocene assemblages and then replaced by straight backed bladelets
in the late Holocene (J. Deacon 1984: 297). Careful examination of the Blydefontein
and Meerkat Rockshelter backed tools has revealed more variability than the simple
straight backed bladelet and crescent dichotomy. The most numerous group at
Blydefontein Rockshelter is simple straight backed bladelets, and the majority of the
ones tabulated in Table 20 are broken medial sections with proximal and distal snap
fractures. Most complete straight backed bladelets are backed points, however it is
not the pointed end that appears to offer the greatest opportunity for cross-dating. A
small number of straight backed bladelets are trimmed on all sides by pressureflaking and these are called pressure-flaked straight backed blade lets (Figure 81 ).
189
b
a
d
c
red ochre
on backing
'
e
9
1 em
Figure 81 Backed bladelets and projectile points from Blydefontein and Meerkat
Rockshelters. a) Pressure-flaked straight backed, b) straight backed with bifacial
trimming, c) straight backed with unifacial trimming, d) truncated straight backed,
e) crescent, f) simple straight backed, g) double backed cresent.
190
Similar tools were found at Glen Elliot Shelter (Sampson 1967: 145), at
Riversmead Shelter (Sampson and Sampson 1967 : 30}, Moshebi's Shelter in Lesotho
(Figure 111-5 in Carter 1969) and in previous excavations at Blydefontein
Rockshelter (Sampson 1970: 90-91 ). Most backed bladelets are not pressureflaked, but a small number do exhibit evidence of curved bifacially-trimmed bases or
butts. Others had only curved unifacially-trimmed butts, and still others had convex,
straight or concave abruptly trimmed bases that could be classified as truncations, i.e.
triangle scalene of Tixer (1963) or other strictly geometric forms.
Tixer's typology
of Maghreb backed tools seems to have little applicability for forms found in southern
Africa, however. Backed tools that do not have straight backing include, in addition to
crescents, curved backed bladelets and backed flakes.
Also shown in Figure 81 are bifacially pressure-flaked tanged and barbed
projectile points. Although none of these is a backed tool, the pressure-flaking
technique on these projectile points is very similar to the pressure-flaking observed
on the trifacial straight backed bladelets. They are included because of the presumed
functional similarity to backed tools.
Table 21 lists the percentages for the backed tools from Blydefontein Rockshelter
listed in Table 20, and the available ages for the CAUs. Straight backed bladelets
remain the dominant form throughout the sequence, however other forms vary in
frequency through time and I will focus on these changes. Starting with CAU9 at the
bottom of Layer CY it is clear that very few backed tools were found, but crescents,
curved backed bladelets and backed flakes occur. This assemblage remains undated,
but it tentatively is identified as an "Early Microlithic" assemblage (sensu Mitchell
1988). Much larger samples are necessary before this assemblage can be
characterized with reliance.
Table 20.--Backed tools in Combined Analytical Units at Blydefontein Rockshelter
Pressure
Straight Flaked
Projectile
Backed
Blade let
Eoint
Combined
AU
Straight
Pressure Backed
Flaked
Bifacial
Backed
Trimmed
Blade let
Base
Straight
Backed
Unifacial
Trimmed
Base
Straight
Backed
on
Truncation
'
2
3
4
5
6
7
8
9
Total
6
19
35
17
58
4
1
140
-
-
-
1
3
1
1
2
1
2
-
5
-
2
5
11
4
11
2
2
3
8
-
-
35
13
Proximal
Curved Double Backed & Distal
Qi§c§!rd
Cre§cent Backed Backed Flake
1
1
4
1
8
1
14
2
1
1
1
3
10
11
4
1
23
3
1
1
-
-
0
15
15
Total
1
15
36
65
33
126
11
1
-
-
3
6
4
54
290
Table 21.--Percentages of backed tools in Combined Analytical Units at Blydefontein Rockshelter
Combined
AU
1
2
3
4
5
6
7
8
9
Straight
Pressure Pressure Backed
Flaked
Bifacial
Straight Flaked
Backed
Projectile Backed
Trimmed
Bladelet
Eoint
Bladelet
Base
1 -
4 o. o
52.8
53.8
51.5
46.0
36.7
33.3
Straight
Backed
Unifacial
Trimmed
Base
-
-
-
-
13. 3
-
8.3
1 .5
3.0
1 3. 3
13.9
16.9
12.1
8.7
18.2
-
1 .5
-
-
-
Straight
Backed
on
Truncation
-
3.1
9.1
6.3
-
Crescent
Curved Double
Backed Backed
Backed
Flake
-
-
-
-
-
2.8
1.5
3.0
8.7
-
6. 7
-
-
3.0
3.2
-
-
33.3
33.3
-
6.1
7.9
27.3
-
3.0
0.8
9.1
-
Proximal
& Distal
Qiscard • 14C Age BE
1 oo. o
26.7
I
844-1305
22.2
1980-2292
21.5
3135
9.1
18.2
I 4066-4286
9.1
8541 _..
I
-
-
I
<.o
.......
192
No backed tools were recovered in CAU8, from the middle of Layer CY, and this
absence is an important attribute used to identify this assemblage as a macrolithic
Lockshoek assemblage (Bousman 1989}. A single radiocarbon date comes from the top
of AU8 so most of the artifacts attributable to this assemblage are slightly older than
the radiocarbon date.
CAU7, from the top of Layer CY, shows that crescents and truncations occur in
high frequencies, while straight backed bladelets with unifacially-trimmed butts
occur for the first time. CAU7 probably represents a long period of time, and it is
reasonable to expect that this CAU provides evidence of a mid-Holocene Interior
Wilton assemblage that could span the period from 8000-4300 B.P., although not
necessarily with continuous occupations throughout this period.
CAU6, the bottom of Layer TG and all of Layer CAC, has crescents, curved backed
bladelets, straight backed truncations and straight backed bladelets with unifaciallytrimmed butts in equal frequencies, and double backed bladelets in slightly lower
frequencies. These assemblages occur between 4000-4300 B.P. and seem to
represent a transition from mid-Holocene to late Holocene Interior Wilton
assemblages, i.e. Classic to Developed, but with larger samples it is possible that
rapid step-like changes rather than general shifts might emerge. It is important to
remember that CAU6 occurs during a period of rapid climatic change, and huntergatherers certainly had to respond to climatic induced changes in resource diversity,
density and availability (see Chapters XI and XII for additional discussions).
In CAU5 straight backed bladelets with unifacially-trimmed butts and truncations
occur in high frequencies, straight backed bladelets with bifacially-trimmed butts
appear, and crescents occur for the last time. It is not clear whether truncations
simply represent variations of unifacial butt trimming or if the truncated backed
193
bladelets are stylistically different. As with CAU6, this backed tool assemblage
appears to be transitional toward later backed tool assemblages.
At the bottom of Layer GAC in CAU4 a significant shift occurs. A single pressureflaked straight backed bladelet occurs, and crescents are found no longer. In CAU3 at
the top of Layer GAC straight backed bladelets with bifacially-trimmed butts occur in
their greatest frequencies, and are not found above this.
In CAU2, i.e. Layer HG, another distinctive change is documented. In CAU2
pressure-flaked bifacial tanged and barbed projectile points occur. Given the young
radiocarbon age of this assemblage (844-1305 B.P .) it is likely that the projectile
points were manufactured to emulate metal Iron Age projectile point styles even
though the nearest coeval Iron Age communities were still restricted to Natal and the
Transvaal (Hall 1990). CAU1 represents artifacts in the mixed surface layer that
apparently post-date 850 B.P. It is clear that only limited occupation occurred at
Blydefontein Rockshelter at this time and that the production of backed bladelets was
not a major activity. The absence of microlithic tools is used to signify the shift from
Interior Wilton to Smithfield assemblages (Sampson 1974, 1985), and the presence
of pressure-flaked bifacial tanged and barbed projectile points in CAU2 indicates that
these represent terminal Interior Wilton assemblages. The shift from Wilton to
Smithfield represents a change, but the significance of this change is hotly debated and
few people are in agreement.
At Meerkat Rockshelter many fewer backed tools were recovered (Table 22).
This is due to the small area excavated, and more rapid sedimentation accumulation.
Meerkat Shelter appears to provide a detailed or exploded view of the termination of
the Interior Wilton Industry and the beginning of the Smithfield Industry and
correlates to the uppermost two or three CAUs at Blydefontein Rockshelter. The
backed bladelet sequence from the two rockshelters are in strong agreement. The
Table 22.
AU
10
11
12
13
14
15
16
17
18
Total
Percent
194
Distribution of backed tools and projectile points in analytical units from
Meerkat Rockshelter
Straight
Backed
Bladelet
QQ!..!D1L0/Q
1 13 3
2133
1 15 0
Pressure
Flaked
Projectile
Point
QQuntL0/Q
1I 1 0 0
1133
1133
41 67
1 I 50
1 I 50
1 15 0
211 00
11100
8
47.0
Pressure
Flaked
Backed
Blade let
QQunU0/Q
2
11 .8
7
41.2
TQtal
1
3
6
2
0
2
2
0
1
17
14Q aァセ@
B.P.
1102±62
2352±122
upper nine AUs at Meerkat Rockshelter correlate to CAU1 at Blydefontein Rockshelter.
These Smithfield assemblages from both rockshelters are characterized by an absence
of backed tools. The backed blade let sequence at Meerkat Rockshelter begins with
straight backed bladelets. Then toward the middle of the deposits, in AU12, pressureflaked straight backed bladelets occur, and above this pressure-flaked tanged and
barbed projectile points occur.
It is possible that with larger sample sizes from
Meerkat Rockshelter pressure-flaked straight backed bladelets will assume a wider
temporal distribution, as at Blydefontein Rockshelter, or alternatively the Meerkat
sequence may be a more accurate representation of the true temporal patterning of
these distinctive backed bladelets. At both sites bifacially pressure-flaked tanged and
barbed projectile points mark the termination of the Interior Wilton and appear to
have a very limited temporal distribution.
One pressure-flaked projectile point at
Meerkat is directly associated with the radiocarbon sample dated to 11 02±62 B.P. It
is possible that the single tanged and barbed bone projectile point from Glen Elliott
Rock Crevice (Sampson 1970: 64) could be contemporaneous to these flaked
195
projectile points. It is tempting to suggest that the termination of the Interior Wilton
is associated with the drought identified in the Oppermanskop pollen profile (see
Chapter V).
Endscraper Length
Numerous archaeologists have documented increases in mean endscraper length
through time during the Later Stone Age (Sampson 1970: 97; Sampson and Sampson
1967: 18-20;
H. Deacon 1976: 61-63;
J. Deacon 1984: 283, 301; Opperman
1987: 176-177), and various arguments have been presented to account for this
change. Sampson (Sampson and Sampson 1967: 6) divided end scrapers into different
types, in part, defined by length. However, nowhere in the literature has length or
other metric attributes been used to statistically define morphological distinct and
naturally occurring groups of end scrapers.
He (ibid: 20; Sampson 1970: 97)
suggests that in the Orange River Scheme area and at Blydefontein Rockshelter an
increase through time in mean endscraper lengths was due to changes in response to
shifts in lithic raw material usage. Now it is known that the size increase occurs over
much of southern Africa with end scrapers made of many different types of raw
material, and this pattern clearly is not just a function of a change in raw material (J.
Deacon 1984).
For a number of years J. Deacon (1972: 15, 1984: 282) has argued that
stylistic "norms of manufacture" can be identified in scrapers through the calculation
of means and standard deviations, however few assemblages actually have scraper
length distributions that can be statistically documented as having normal gaussian
distributions (J. Deacon 1972: 20). This by itself suggests that mean scraper lengths
do not represent stylistic norms that are adhered to closely. Also J. Deacon (1972:
36) observed that small scraper means are associated with small standard deviations,
and she argues that this reflects less stylistic variability. But it should be realized
196
that assemblages with small mean endscraper lengths must have small standard
deviations. This is because small scrapers do not have the same potential metric range
as large scrapers. Thus small end scraper mean values will have small standard
deviations and large end scraper means can have large or small standard deviations.
Obviously standard deviations are not providing a measure of normalized stylistic
choices.
In addition, and most importantly, the extension of Dibble's (1987) Middle
Paleolithic scraper analysis to end scrapers from southern Africa would suggest that
the final form, including length, of end scrapers is a function of intensity of use and
not a reflection of "norms of manufacture". By considering end scrapers from this
viewpoint, it is logical that endscraper length can be strongly conditioned by hafting
or lack of hafting, and that the final form in many cases does not represent a
preconceived shape of the tool upon its manufacture but rather its condition, perhaps
exhausted perhaps not, upon discard. Under these circumstances endscraper lengths
would not be expected to have normal distributions. Recent ethnoarchaeological
observations on chipped stone scrapers used by modern Ethiopian hide tanners shows
that intensity of utilization and condition upon discard have a significant effect on
scraper morphology especially length (Clark and Kurashina 1981: 309; Gallagher
1977: 411). The concept of hafted and nonhafted end scrapers in southern Africa is
not new (Clark 1959; H. Deacon 1966, 1972: 40; H. Deacon and J. Deacon 1980;
Kannemeyer 1890; Sampson and Sampson 1967: 19), but the full implication of
hafted or unhafted end scrapers for analysis of stone scrapers in southern Africa is not
appreciated yet (Humphreys and Thackeray 1983: 10-14, 275-282).
It can be
argued that shorter end scrapers often with trimmed butts probably represent hafted
end scrapers that have been utilized near or to the point of exhaustion and possibly
curated, while long unretouched end scrapers probably represent scrapers that are
197
not intensively used and, in fact, may be expedient tools that are little curated. This
has further implications for assemblage formation processes that will be discussed in
the next chapter.
Figure 82 plots the mean lengths and standard error bars for Blydefontein
endscraper samples (N <:: 5) from analytical units only within Blocks C-D. As Blocks
C-D have the finest stratigraphic control and directly associated radiocarbon dates,
these values show in detail the nature of temporal change for mean endscraper lengths
for samples that have a reasonable number of end scrapers. This line chart shows that
endscraper lengths are small and fairly stable in AU15-20 (Layer CAC and the lower
portion of TG). In the upper portion of TG, i.e. AU12-14, end scrapers begin to
increase in length. Only one analytical unit from the lower portion of Layer GAC has
more than five end scrapers, i.e. AU9. This sample is stratified above a 3135±33
B.P. radiocarbon date so AU9 is approximately 1000 years older than AU17. The
sample from AU9 does not represent a significant increase from the last sample, i.e.
AU13. The next sample is from AU? and is dated by radiocarbon to ca. 700 years later
than AU9. A sharp drop occurs in AUG and this is roughly dated by Sampson's
radiocarbon date of 1980±120 B.P.
Interestingly this is virtually the exact time of
the drought documented at USP. A dramatic increase in mean endscraper length occurs
by AU4 which is dated to 1305±31 B.P., and this is very close in age to the drier
conditions demonstrated by pollen from Oppermanskop hyrax midden (see Chapter V).
Finally a moderate drop in mean endscraper lengths is evident by the AU2 sample dated
to 844±119 B.P. These minor fluctuations demonstrate the complexity of the record,
but in general very consistent patterns are evident. These data can be used to suggest
that end scraper length increases through time, but minor fluctuations may be tied to
paleoenvironmental changes.
198
45
E'
40
£
c,
35
.3....
30
.s
c
(])
0..
....
(.)
<U
C/)
25
"'0
c
UJ
c
<U
20
(])
:::2:
15
10
0
2
6
4
8
10
AU
12
14
16
18
20
22
Figure 82. Mean endscraper lengths and standard error bars for Blydefontein
Rockshelter Blocks C-D.
60
E
55
.s
50
c,
45
.3....
40
£
c
(])
0..
<U
....
35
C/)
30
(.)
"'0
c
UJ
c
25
<U
(])
:::2:
20
15
10
0
2
4
6
8
AU
10
12
14
16
18
Figure 83. Mean endscraper lengths and standard error bars for Meerkat Rockshelter
AUs.
199
Mean endscraper lengths at Meerkat Rockshelter do not change significantly
through time (Figure 83). An ANOVA test which is more sensitive than a
nonparametric test, failed to demonstrate any significant difference between
endscraper lengths among the AUs (p value = 0.926). Also by grouping the AUs into
assemblages with and without backed tools (AU1-9 and AU1 0-18) no difference could
be distinguished between endscraper lengths with an AN OVA test (p value = 0.651 ).
The mean endscraper length for AU1-9 is 34.4±12.1 mm (N = 35) and the mean
endscraper length for AU10-18 is 32.9±13.7 mm (N = 25).
Ostrich Eggshell Beads
Meerkat Rockshelter produced many more ostrich eggshell beads than
Blydefontein Rockshelter especially considering the size of excavations at the two
rockshelters. Both shelters have more ostrich eggshell beads in the upper levels. It
is well known that through time wider and wider ostrich eggshell beads were
manufactured in Namibia (Jacobson 1987), and the data from Meerkat and
Blydefontein Rockshelters suggests a similar trend as well. As complete ostrich
eggshell beads were not extremely numerous a histogram of all measurements of
complete finished beads from Blydefontein and Meerkat Rockshelters was plotted
(Figure 84). This histogram shows that a gap occurs between 4.8-5.0 mm. An
inspection of the distribution of these two size classes demonstrated that most of the
smaller beads are in Interior Wilton microlithic assemblages, and most of the larger
beads do not occur with microlithic tools, i.e. with Smithfield. As Figure 85 shows the
shift to larger ostrich eggshell beads occurs by AU1 0 in Meerkat Rockshelter.
Interestingly this analytical unit lacks backed bladelets, but does have a single
pressure-flaked tanged and barbed projectile point.
It is tempting to suggest that the
small ostrich eggshell beads actually were made with Wilton backed microlithic tools,
200
but this is only a guess. However wear pattern analysis might be capable of
demonstrating this possibility.
5
§
0
4
3
u
2
0
6
5
4
3
7
9
8
OES Bead Maximum dゥ。ュ・セイ@
11
10
(mm)
Figure 84. Histogram of Blydefontein and Meerkat ostrich eggshell bead maximum
diameters.
10
9
0
.....
$
CD
E
8
(lj
i:5
Mセ@
E
::J
E
セ@
7
0
0
6
0
セ@
"0
(lj
CD
Ill
5
0
0
0
0
0
0
0
0
0
0
0
4
0
3
2
.
4
0
0
6
8
0
8
.
10
12
0
8
14
16
18
AU
Figure 85. Distribution of individual ostrich eggshell measurements by analytical
unit at Meerkat Rockshelter.
201
The only published site with good samples of finished ostrich eggshell beads is
Glen Elliot Shelter near the Orange River (Sampson 1967b: 136-137). A plotting of
the mean diameter measurements reported by Sampson shows a clear temporal trend
(Figure 86). Interestingly Sampson's Level 4 assemblage has a pressure-flaked
projectile point, and a few backed bladelets as well.
,.....
E
E
'-/
7
6.5
!...
Cll
.....Cll
E
rq
0
"
rq
Cll
m
6
5.5
(/)
w
0
crq
5
Cll
!:
4.5
4
2
3
4
5
6
Le-ve-l
Figure 86. Mean OES bead diameters and standard error bars from Glen Elliot Shelter
(data after Sampson 1967b).
Ceramics
Neither Blydefontein nor Meerkat Rockshelters produced many ceramic sherds,
however a few stratigraphic relationships are clearly significant for cross dating, and
for the regional ceramic record.
Sampson (1967a: 52-53) original defined two
ceramic types (A and B) occurring in the Later Stone Age occupations at Zaayfontein
Rockshelter. Class A ceramics are most easily distinguished by abundant grass temper
and a variety of punctate or impressed outer surface decorations. Class B ceramics
generally lacks grass temper and often has a light brown (buff) exterior that may or
202
may not be burnished. Sampson (1970: 93-94) classified the 46 ceramic shards
excavated from Blydefontein Rockshelter in 1967 as Class B ceramics. Humphreys
{1979) reanalyzed the ceramics from the Orange River Scheme sites and concluded
that many of Sampson's Class B shards also had grass tempering and only one ceramic
tradition was present. Renewed research in the Zeekoe Valley, south of the Orange
River Scheme area, resulted in a intensive analysis of the ceramics (Sampson 1988;
Sampson et al. 1989). These studies recognized three major groupings: grass
tempered plain ware (GTPW), silt tempered plain ware (STPW) and Khoi. GTPW are
slab-built with fairly thick walls and fired with low heat. The known shapes include
bowls and bag-shaped pots, and surfaces may be burnished or decorated with a variety
of motifs (ibid: 4). STPW are coil-built with thin walls, sand, silt and grass temper.
The shards appear to be well fired, and surfaces may be burnished. Khoi shard are
similar to STPW except they may have an orange-to-red-to-mauve slip on their
surface and Khoi shard lack any evidence of grass temper. Additionally some Khoi
shards do have surface decorations. Many of the GTPW shards from the Zeekoe Valley
are decorated and Sampson {1988) has recently studied the spatial distributions of
the different decorative motifs. Clearly GTPW is the same as Class A, but it is unclear
whether Class B ceramics are Khoi or STPW. The present analysis uses Sampson's
more recent classification scheme, in fact, Sampson classified the shards from both
Meerkat and Blydefontein Rockshelters.
The ceramic shard counts from Blydefontein Rockshelter are presented
individually in Block B AUs and Blocks C-D AUs, and then correlated in Combined AUs
(Table 23, Table 24, Table 25). No decorated ceramics were recovered from
Blydefontein Rockshelter. In Block B STPW and Khoi sherds are restricted to AUs 3
and 4. In AU3 GTPW occurs and it continues to the surface. In Blocks C-D three small
GTPW sherds, called crumbs by Sampson (personal communication 1989) were
203
recovered in AUs 7 and 8, and it is likely that these were not in their original
stratigraphic context.
The sherd in AU5 is thin buff-colored sherd with grass and silt
temper. In AU4 a single Khoi spout fragment was recovered and the radiocarbon date
(1305 B.P) is in direct association with this Khoi sherd.
Except for one STPW sherd
the remainder of the ceramics from Blocks C-D are GTPW sherds. In the Combined
AUs it is clear that Khoi and GTPW begin at about the same time, then Khoi ceramics
stop.
Table 23. Ceramic sherd counts at Blydefontein Rockshelter, Block B
AU
Surface
1
2
3
4
Total
GTPW
1
7
6
4
18
STPW
KHOI
-
-
-
1
1
2
2
1
3
Total
1
7
6
7
2
23
Table 24. Ceramic sherd counts at Blydefontein Rockshelter, Blocks C-D
AU
1
2
3
4
5
6
7
8
Total
セhew@
STPW
1
4
4
Kl::lQI
1
6
1
1
1
17
1
1
TQtal
5
4
0
7
1
0
1
1
19
Table 25. Ceramic sherd counts at Blydefontein Rockshelter, Block B
Combined
AU
GTPW
12
STPW
Kl::lQI
3
2
1
1
1
35
3
4
1
2
3
21
Total
1
TQtal
13
25
4
42
14C Date B.P.
844±119
1255±109
1305±31
1980±120
2292±117
204
At Meerkat Rockshelter 49 sherds were recovered (Table 26), and this
represents a much higher density than at Blydefontein Rockshelter. Nevertheless a
very similar pattern is presented by the Meerkat sherds. Both GTPW and Khoi appear
in the shelter stratigraphy at virtually the same time and this pre-dates the 11 02
B.P. date. Khoi sherds are only found in AU12, and from AU1-11 GTPW occur as the
overwhelming majority. STPW do not appear to reflect chronological patterns.
Lastly, in AU3 two refitted GTPW sherds with a quill-groove decoration were
recovered. Unfortunately this level was not radiocarbon dated, but they are only a
short distance above the 979 B.P radiocarbon date and it seems likely that they are
only a few hundred years, at most, younger than this date. The sherds recovered from
the surface are both decorated. One is covered with large smooth spatulate, oblique
stab-and-lift, parallel entry pattern, and the other is decorated by a large multinotch spatulate, vertical stab-lift, diagonal entry pattern. These sherds were
recovered from the edge of the shelter where slope erosion is beginning to remove the
upper 10-15cm of deposit, and it is impossible to know if these sherds originally
were in the upper layers or on the surface.
Table 26. Ceramic sherd counts at Meerkat Rockshelter
AU
Surface
1
2
3
4
5
6
7
8
9
10
11
12
Total
GTPW
2 *
1
2
4 *
5
4
7
5
8
-
3
2
1
44
* Decorated sherds in sample
STPW
KHOI
-
-
1
-
1
-
-
-
1
-
-
3
-
-
2
2
Total
2
1
3
4
5
5
7
5
14C Date B.P.
979±70
8
1
3
2
3
49
1102±62
205
Conclusions
The chronological sequence of backed tools and projectile points from Blydefontein
and Meerkat Rockshelters demonstrates clear temporal patterning that can be used for
cross-dating. Within the Holocene assemblages a temporal sequence of backed blade let
tools can be demonstrated from (i) crescents and curved backed blade lets, to (2)
truncations and straight backed bladelets with unifacially-trimmed butts, to (3)
straight backed bladelets with bifacially-trimmed butts, to (4) pressure-flaked
straight backed bladelets, to (5) pressure-flaked tanged and barbed projectile points,
and finally to (6) an apparent lack of these tool types. Through most of the sequence,
i.e. from ca. 4300-1300 B.P. plain straight backed blade lets are numerically
dominant, but it is only at the beginning of the Blydefontein sequence, and the end of
the Blydefontein and Meerkat sequences that straight backed bladelets occur in
frequencies that fall below their overall average. These patterns have not been
documented at other sites in southern Africa, and at present it is unclear how large of
a region such patterns can be expected to extend.
A small number of ethnographic collections have San arrow points made with
backed tools (Clark 1977). These often consist of two straight backed bladelets
mounted in a well formed mastic tip on a foreshaft (Figure 87). It seems logical that
straight backed bladelets were shaped, sharp on one end and a corner on the other, so
that they would securely fit into the mastic and not dislodge even upon impact, or at
least not until after penetrating an animal. Crescents, on the other hand, lack a flat
abrupt corner, and, in fact, have a shape that might cause the implement to rock in a
haft.
It is unlikely that they would have been used in exactly the same fashion as
straight backed bladelets. One possibility, suggested to me by Steve Tomka, a lithic
specialist at the University of Texas at Austin, is that crescents were hafted like
straight backed bladelets, i.e. at the tip of a projectile, and that the rocking motion was
206
an intended function of the tool. The rocking motion could increase the effectiveness of
the weapon by forcing the distal ends of crescents to splay upon impact and cause
increased bleeding (Figure 88). This increased destruction would be especially
important if poisons were not used with crescents. The smooth curved surface would
act to force an unstable rocking motion in virtually any haft, and for example this
would reduce a crescent's effectiveness in a composition cutting tool. Also, if
crescents were hafted in a similar manner to the ethnographic examples, then this
might explain why no slotted hafts for microlithic tool have been recovered from
excavations in southern Africa even though a reasonable number of sites have
extremely good preservation of organic materials. Of course, the transverse hafting of
crescents could have occurred as well.
Figure 87. Ethnographic projectile point with straight backed bladelets mounted in
mastic.
207
111
Figure 88. Hypothetical mounting and function of crescents as projectiles.
All end scrapers from Blocks B, C and D are shown by Combined AUs, and the same
general trend is evident (Figure 89}. As these samples are much larger than those
used above form Blocks C-D, the pattern is more reliable. These data demonstrate that
single endscraper in CAU8, the Lockshoek, is long, but beginning in CAU7 and
continuing through CAU1, the Interior Wilton and Smithfield assemblages, a dramatic
shortening and then a consistent increase in mean endscraper length occurs. Assuming
that CAUs represent equal periods of time (which they do not) a linear regression was
208
calculated for estimating CAU with mean endscraper length (Figure 90). The formula
is:
CAU= -0.395 x (mean endscraper length) + 13.85, (r2 = 0.961 ). The reader
should be cautioned that this is a misapplication of linear regression analysis, and is
only included as a tentative and rough estimate of relative age of mean endscraper
samples. Once the range and mean ages of the CAUs are better dated, then it might be
possible to estimate the ages of surface endscraper samples with a similar method, but
more radiocarbon dates are needed that completely bracket the age span of these
samples. The mean length for all end scrapers at Meerkat Rockshelter equals 33.77
mm. Using this value in the linear regression equation for estimating CAUs at
Blydefontein, the resulting product is 0.5 CAU. This is in close agreement with
correlations based on radiocarbon dates and separately on backed bladelets.
90
80
..c
70
Q)
60
0,
c::
...J
.....
Q)
c..
ro
.....
50
(J
C/)
"LUc::
40
30
20
10
0
2
3
4
5
6
7
8
Combined AU
Figure 89. Mean endscraper lengths and standard error bars for Blydefontein
Rockshelter Combined AUs.
9
209
8
7
6
::::>
<(
5
"0
Q)
c
:0
E
0
()
4
3
2
PKMセイ@
18
20
22
24
26
28
30
32
34
Mean End Scraper Length (mm)
Figure 90. Linear regression of mean end scraper lengths and Combined AU.
This analysis suggests that the Interior Wilton is, in part, characterized by small
(< 5 mm) ostrich eggshell beads. At the termination of the Interior Wilton and at the
beginning of the Smithfield Industry ostrich eggshell beads increase significantly in
size. It has been suggested above that the size shift might be due to a change in the
types of tools used to drill the beads, but other interpretations are equally as valid at
this point. It is possible that ostrich eggshell beads were used in the same manner as
glass beads are today by the Kalahari San (Wiessner 1984). In other words ostrich
eggshell beads functioned in assertive style manipulations. Even though the Zeekoe
Valley survey data lack high temporal resolution it is clear that some sort of
population increase occurred in the region during the Late Holocene with the advent of
the Smithfield Industry (Sampson 1985: 104-1 09).
With greater population
densities it is possibly that both types of stylistic behavior identified by Wiessner
(1983, 1984, 1985) would also increase in occurrence. It may also be reasonable
that the shift to larger ostrich eggshell beads was part of this increased stylistic
210
behavior. At the least the larger ostrich eggshell beads would be more noticeable than
the smaller sizes, and this may have played a role in the process. Clearly this
hypothesis will be very difficult to conclusively demonstrate, but the change in
ostrich eggshell beads and the associated diversification in projectile points at this
time strikes similarities to Wiessner's San artifact study.
The ceramics at Meerkat and Blydefontein Rockshelters indicates that ceramics
became part of the assemblage during the later part of the Interior Wilton. Both grass
tempered plain ware and Khoi ceramics appear at approximately the same time. This
event is best dated at Blydefontein Rockshelter at 1305 B.P .. Khoi sherds are no
longer part of the assemblage by 1255 B.P. at Blydefontein Rockshelter and by 1102
B.P. at Meerkat Rockshelter. This is significant as Sampson et al. (1989: 13) and
Hart (1989: 158-163) argue that three radiocarbon dates, 1140±60 B.P. (SMU1790), 560±170 B.P. (SMU-1791), 544±43 B.P. (SMU-1636) mark the ages of
Khoi sherds at two sites, Haaskraal Shelter and Volstruisfontein Shelter, in the
Zeekoe Valley. The tree ring calibrated ages for these dates are 1050 B.P., 560 B.P.,
and 555 B.P., respectively. The calibrated age for the Khoi sherd from Blydefontein
Rockshelter is 1256 B.P., and this is 200 tree-ring calibrated years before the
earliest Khoi ceramics in the Zeekoe valley. Surprisingly Khoi ceramics do not
reappear in either rockshelter sequence at Blydefontein Basin, as occurs at Haaskraal
and Volstruisfontein Shelters in the Zeekoe Valley. However it is not evident that
either Meerkat or Blydefontein Rockshelters were occupied in so recent times and this
may be the reason for the absence of Khoi sherds.
Taken as a whole the temporal shifts in artifacts can be used to cross-date surface
sites in the Zeekoe Valley. Simplistically it appears that increasing mean end scraper
lengths could be the most regular and reliable chronological tool for most Interior
Wilton sites. However, arguments presented below will call this conclusion into
21 i
question. Backed tools and projectile points also provide a extremely useful temporal
patterns that may be more reliable than those presented above for end scrapers. For
the terminal Wilton and Smithfield periods ceramics may be the best for cross-dating,
but the possibly sporadic occurrence of Khoi sherds (and possibly herding as an
economic activity) and effects on the transition from STPW to GTPW as well as the
variations in decorative patterns still requires a much more complete data base than
is presently available for the Zeekoe Valley and Blydefontein region. Ostrich eggshell
beads can also be helpful at diagnosing the difference between Interior Wilton and
Smithfield, but their rare occurrence on surface sites limits the usefulness of ostrich
eggshell beads.
In terms of cross-dating surface sites, it is clear that the temporal span of the
surface sample, as well as the temporal span of the excavated sample, could be a
significant stumbling block in successful cross-dating. It is clear for the calculation
of mean end scraper lengths that the longer the temporal span of the excavated sample
the less variation is shown in the mean end scraper values. This phenomenon would be
reflected in surface samples as well.
An added complication for surface sites is reoccupation, especially in different
temporal periods. Care must be taken to separate assemblages of different temporal
periods on surface sites. Under the assumption that hornfels patina changes color
through time in a regular fashion (Sampson 1985), one possible method for sorting
distinctly different occupations on surface sites is the color coding of patina using a
Munsell Color Chart as used by soil scientists, and statistically separating distinctive
patina colors. A preliminary and limited test has shown that this procedure may be
possible (Bousman, nd). The soils derived from dolerite, sandstone, and shale must be
considered as surface artifacts sit on these soils and these bedrocks can produce soils
with very different chemical properties (Tim Dalbey, personal communication).
212
Obviously much more work is required and care must implemented before reliable
cross-dating of surface assemblages can take place.
CHAPTER X
EXPLANATORY MODELS OF LATER STONE AGE LITHIC VARIABILITY
The cross-dating scheme developed in the previous chapter does not address the
more interesting question of why the various changes in lithic design took place.
Before developing my explanatory model, it is useful to review the development of
pioneer models that also purport to explain variability and change in Later Stone Age
lithic assemblages from southern Africa. In most of them, two sorts of variability are
considered: the relative frequencies of stone tools within an assemblage, otherwise
known as assemblage composition, and the morphological shapes of these tools. These
topics have been debated hotly over the last 20-30 years in archaeology, and still are
as few solid conclusions have been agreed upon. However, in the last 15-20 years
many new approaches have been developed to account for these two types of
variability, and these require consideration as well.
Models for Explaining Assemblage Composition
Indigenous Models
There have been several attempts to explain Later Stone Age assemblage change by
different South African archaeologists. In 1929 Goodwin and van Riet Lowe published
The Stone Age Cultures of South Africa. In this landmark study the authors established
a classification system for South Africa's local Stone Age archaeological record, and
their basic categories still are in use today, including the terms Earlier, Middle and
Later Stone Age. Goodwin and van Riet Lowe clearly associated the Later Stone Age with
modern Homo sapiens, called Neo-anthropic groups, and two coeval traditions were
recognized: the Smithfield and Wilton cultures. The Wilton, at that time known
213
214
mostly from the coastal region, was characterized by microlithic tools, especially
crescents, and small convex scrapers. Goodwin (Goodwin and van Riet Lowe 1929:
150) suggested that the Wilton represented an "offshoot from the Capsian Group of
North Africa".
The Smithfield, discussed by van Riet Lowe (ibid: 187), occurred in the Interior
Plateau of South Africa, where he proposed a sequence of three phases, Smithfield A, B
and C, with Smithfield Band C overlapping in time to an unknown degree. Smithfield A
was characterized by large round and curved (concavo-convex) scrapers. The
Smithfield B lacked these large scrapers but had numerous end scrapers, and
Smithfield C, known from rockshelters, had small convex scrapers and backed
bladelets. Smithfield C was essentially a crescent-less Wilton. Smithfield A and B
artifacts were made from indurated shale known as hornfels, but agate and jasper
were commonly used to produce Smithfield C tools. The Smithfield was seen by van
Riet Lowe as a local development restricted to the Inland Plateau of South Africa, and
believed to be the result of interaction between Middle Stone Age groups and inmigrating modern Homo sapiens. Thus both Goodwin and van Riet Lowe accounted for
the development and existence of Later Stone Age groups by invoking population
movement and immigration. Obviously, they used the presence of stone tools to reflect
ethnic groups in a direct, unambiguous, and unquestioning manner.
From the 1930s into the 1960s numerous new variants of the Smithfield and
Wilton cultures were defined; each with slightly different assemblage compositions.
One variant of the Smithfield, the Smithfield N, was restricted to Natal and
characterized by notched scrapers or spokeshaves. The Smithfield N assemblage was
thought to reflect the intensity of wood working in this region, and this represents one
of the first times, even though in a limited manner, that stone tool function rather that
ethnic identity was used to explain assemblage composition in southern Africa (van
215
Riet Lowe 1936). Clark {1959) in his synthesis of Stone Age archaeology in
southern Africa elaborated on this concept of adaptation. He explained assemblage
differences as reflecting regional specialization in terms of hunting specific animals,
but he continues to see these differences in mostly cultural terms. Thus Clark
suggests that the Wilton and Smithfield cultures represent differences in adaptation to
hunting different game.
The next major research in southern Africa focused on the Interior Plateau was in
the Orange River Scheme area (Sampson 1967a, 1967b, 1970, 1972; Sampson and
Sampson 1967). He presented the most detailed Later Stone Age culture chronology
available at the time for southern Africa, and it represents a refinement of the
research begun by Goodwin and van Riet Lowe, and developed by Clark. Sampson
(1972) suggested that six phases, numbered 1-6, were present in the Later Stone Age
of southern African and all were present in the Orange River Scheme area.
He
provided the first radiocarbon-dated culture chronology, and quantified the range of
variation in the relative frequencies and metric attributes of tool types between the
phases.
In 1974 Sampson presented a new synthesis of Stone Age archaeology in southern
African that was a departure from his 1972 monograph. In this he followed the
recommendations of the 1965 Burg Wartenstein Symposium by dropping the term
"Later Stone Age" and replacing it with three major "Industrial Complexes": the
Oakhurst (formerly Smithfield A), Wilton and Smithfield.
Each Industrial Complex
could have regional industries and temporal phases. In the Interior Plateau the
Oakhurst Complex was called the Lockshoek Industry, and the Wilton in the same area
is termed the Interior Wilton Industry with four phases (early, classic, developed,
and ceramic). The Interior Wilton is followed by the Smithfield Complex but only in
the Interior Plateau. In this synthesis Sampson applied a standard typology for
216
analysis so that the range of "stylistic preferences" selected by a prehistoric group
could be compared through space and time. Thus the culture/style paradigm used by
Goodwin and van Riet Lowe, although now quantified, still was being followed by
Sampson.
Concurrent with Sampson's research was another large research project in the
southeast Cape Province initiated by H. and J. Deacon. J. Deacon (1972) suggested
that the chronological differences documented by her at Wilton Large Rockshelter were
due to the birth, life and death, known as ontongeny, of a single cultural system. J.
Deacon (1972) discusses assemblage variability in terms of relative frequencies of
tool types reflecting activities or function, which is an acceptance of Binford and
Binford (1966) hypothesis of tool kits and assemblage variability in the Middle
Paleolithic, and suggests that metric attributes (means and standard deviations) can
be used to identify cultural norms and deviations.
Two years later J. Deacon (1974) presented a distribution of radiocarbon dates
organized by region in southern Africa. In this study she argued that the Interior
Plateau had no assemblages dated in the mid-Holocene between 9500 BP and 4500 BP,
while the coastal zones had numerous assemblages dated to this period. Assemblages
during the mid-Holocene contain high frequencies of crescents, and these are lacking
in all Interior Plateau artifact assemblages. In this paper she proposed that if the
number of radiocarbon dates could be used as a rough indicator of population density,
then the Interior Plateau supported a very small population of hunter-gatherers
during the mid-Holocene. Deacon attributes this population decline to reduced
carrying capacity because of drier conditions during this period. This model was in
direct conflict with Sampson's chronological scheme for the Interior Plateau which did
not recognized an occupational hiatus in the region.
217
The next major contribution to Later Stone Age archaeology was by Klein (1974).
In his excavations at Nelson Bay Cave, he discovered and dated a bladelet industry that
was older than the Oakhurst Complex assemblages, and the people responsible for
producing this industry exploited large gregarious herd animals. This was named the
Robberg Industry, which is now known to date to the Late Pleistocene, ca. 18,00012,000 BP (Deacon 1984;
Mitchell 1988).
At the same time, H. Deacon was developing a new paradigm for explaining
assemblage variability. By selecting sites for excavation with well preserved plant
and animal food remains, H. Deacon (1972, 1976) was able to develop a model of
culture change based on the concept of adaptation that incorporated the exploitation of
plants, animals and other resources. This differs significantly from Clark's and van
Riet Lowe's approaches by systematically including plant food resources and by
utilizing general systems theory. The argument takes the form of a punctuated
equilibrium model with positive and negative feedback mechanisms to explain
adaptations and artifact assemblage variability. He suggests that long periods of
homeostatic plateaus interrupted by short spurts of rapid change characterize the
known stone tool assemblages and economic adaptations in southern Africa. This model
implies that between the big jumps, a number of small scale readjustments occur, but
directional change is limited because of negative feedback. H. Deacon's contribution
represents one of the most sophisticated uses of the concept of adaptation in southern
Africa.
Another major research approach is represented by the work of Parkington and
his students.
Parkington (1972, 1984), Mazel and Parkington (1978, 1981) and
Manhire (1987) discuss the nature of variability in lithic assemblages.
Their
concern is with the determinants inter-assemblage patterns and assemblage change.
They state that at a local scale much assemblage variability is probably due to the
218
seasonal scheduling of activities by prehistoric groups that utilize stone artifacts.
Parkington (1972, 1976) has argued that a series of excavated coastal and
mountainous rockshelter sites in the western Cape were occupied sequentially on a
seasonal basis. Mazel, Manhire and Parkington elaborate on this pattern with a
discussion of artifact assemblages from coastal surface sites in what became known as
the "Sandy Bay Problem". Maze I and Parkington noted that adzes are common at inland
sites in the mountain range some 50km from the coast, but often adzes are virtually
absent on sites near the coast. In fact, the only coastal areas were adzes are common
occur in or near fynbos plant communities. This pattern had been known for some
time (Rudner and Rudner 1954), and Sampson (1974) had argued that two separate
ethnic groups were responsible for the difference in assemblages with Coastal Wilton
groups in the mountains and marine adapted groups termed "Strandloper" or Sandy
Bay near the shores. Alternatively, Mazel and Parkington (1978, 1981) posit that
only a single group was responsible for this pattern and the artifact differences were
due to certain activities (i.e. woodworking with adzes) occurring at the mountain sites
but not at the coastal sites which had no local sources of wood. Initially Mazel and
Parkington did not consider the possibilities of chronological change, but later
Parkington (1984: 111) and Manhire (1987) indicate that this pattern occurred
only within the last 3000 years.
Parkington (1984) has extended this type of
reasoning to earlier time periods and suggested that the undated Lockshoek Industry in
the Interior Plateau is coeval with the Robberg Industry known along the coast.
Parkington suggests that the Robberg Industry represents marginal Late Pleistocene
hunter-gatherers who make a macrolithic Lockshoek assemblage when they are in an
area with abundant hornfels, but on the margins of their range in the Cape Folded Belt
Range, where only quartz and quartzite are available, these same hunter-gatherers
produce a bladelet dominated, ie. Robberg assemblage, because of the lack of hornfels.
219
Recently Parkington's Late Holocene model of seasonal mobility and transhumance
has been challenged from an unexpected direction. Dietary analysis using stable
carbon isotopes from human burials in the southwest Cape have shown that a
significant dietary difference exists between individuals buried in the mountains and
individuals buried in the coastal zone (Sealy 1986 and 1989; Sealy and van der
Merwe 1985, 1986, 1987 and 1988). Thus the stable carbon isotope studies
support Sampson's (1974) thesis that two distinct groups were present with
dramatically different adaptations and diets. Additionally, Bousman (1989) and
Mitchell (1988) has argued that Parkington's Robberg/Lockshoek model does not fit
the available data as Robberg-like assemblages, coeval with the coastal Robberg,
termed Early Microlithic by Mitchell, are found in the Interior Plateau and Lockshoek
assemblages from Blydefontein are not coeval with any Early Microlithic assemblages.
Mitchell (1988:247-250) goes on to suggest that the major factor that causes the
difference between the Late Pleistocene Early Microlithic Complex and the Early
Holocene Lockshoek is the use of the bow-and-arrow by groups that produced
Lockshoek assemblages.
Recently a shift away from ecological-adaptation explanations toward socially
orientated explanations has emerged. Wadley (1986, 1987) has proposed that
hunter-gatherer bands have two settlement phases. These are aggregated and
dispersed phases, and during these two phases she argues that hunter-gatherers
engaged in different activities which are reflected in artifact assemblages recovered
from archaeological sites. Wadley suggests that "public" activities occur at San
aggregation sites, and this "public" behavior is marked by the production of gifts for
exchange (especially ostrich eggshell beads and arrows), and ritual ceremonies such
as trance dances and marriages. Dispersed sites are characterized by "private"
behavior which lacks evidence of exchange, ritual or the manufacture of gifts. She
220
further argues that formal, curated stone tool assemblages and evidence of symbolic
behavior (art work) are more likely to occur at aggregation sites with "public"
behavior, and that dispersal sites with "private" behavior should have more expedient
tools and a lack of evidence suggestive of symbolic behavior. Wadley (1986, 1987)
suggests that the mid-Holocene "classic" Wilton assemblages from Jubilee Shelter in
the Magaliesberg of the Transvaal reflects aggregation occupations with abundant
evidence of "public" behavior. Contemporary assemblages from Cave James are
considerably different and are the result of dispersed phase occupations reflecting
"private" behavior. Analysis of Later Stone Age settlement patterns and rockart
distributions in Lesotho support some of these conclusions (Bousman 1988).
Mazel (1989) also invokes social processes for explaining changes in the Thukela
River basin Later Stone Age. He argues that the ecological-adaptation arguments are
flawed, and that social processes account for the significant changes observed. Mazel
ascribes fluctuations in artifact densities to variations in the exchange network which
he posits is driven by social needs (mainly mate recruitment) rather than economic
forces associated with risks in the food quest as originally described by Wiessner
(1977, 1982). Also, on purely theoretical grounds, he proposes that the underlying
forces in social change are due to "conflict and tensions in social relations" (Mazel
1989: 46). Maze I steadfastly refuses to acknowledge any role that changing carrying
capacities (due to environmental fluctuations) may have played in population changes,
and social responses to these changes or unpredictable fluctuations in food amounts or
availability thus ignoring an enormous amount of solid research on these topics. He
also did not turn to any of the recorded stories or myths of the Southern and Mountain
Bushmen to strengthen his case (Bieek 1933; Bleek and Lloyd 1912; LewisWilliams 1981; Vinnicombe 1976), which have ample documentation in the
Bushmen's own words on the importance of rain (water, !khwa). Mazel (1989: 119-
221
132) identifies the only possible source of "social tension" in egalitarian huntergatherer societies as the division of labor between the sexes. He presents crosscultural comparisons that seem to suggest that as women contribute more and more
economically, their status within a hunter-gatherer society changes to greater
equality with men. Mazel suggests that from the mid-Holocene to the late Holocene,
the economy shifted from meat oriented to a more equal balance between plant and
meat, and women's status increased accordingly. However, Mazel (ibid: 156)
acknowledges that he ultimately fails to identify gender associated archaeological
correlates (artifacts or any other demonstrated class or pattern of archaeological data
including plant/meat ratios), and this highlights the most serious flaws with his
study. First and foremost, he has not provided a systematic and coherent linkage
between social theory and archaeological data. Mazel has developed the workings of his
theory, but he relied on vague, unexplicit and untested assumptions that traditional
archaeologists often use when theory is translated into archaeological signatures.
Many of his underlying assumptions about the meaning of observations on artifacts
{epistemology) remain unchanged from earlier works going all the way back to
Goodwin and Van Riet Lowe. For example, how do we know that changes in artifact
densities are due to proposed variations in exchange systems, and not simply due to
changes in sediment deposition rates in the rockshelters (low deposition rate = high
artifact density and vice versa)?
Under these circumstances it is difficult, if not
impossible, to arrive at correct conclusions. As will be shown below in this chapter
and in Chapters XII and XIII, much of what archaeologists thought was indicated by
certain artifact patterns is probably not what really is reflected by these patterns.
Both his and Wadley's research would greatly benefit with a careful consideration of
how and in what condition artifacts are discarded, and the ways this affects their
archaeological observations.
222
One additional and unstated assumption by the adherents of the "social process"
approach, is that it represents some type of "ernie" form of analysis or, in other
words, it attributes culture change to the internal workings of that individual society
and it claims to understand this from the point of view of the society. Mazel's claims
of writing a society's history implies this as well as statements like early Thukela
Basin society "perceived its environment as hostile" (ibid: 121 ). However, the social
approach is as hopelessly "etic" as the cultural-adaptational or ecologicaladaptational approaches. This is not to say that social factors do not cause social
change, they can and do, but social process practitioners need to state much more
clearly their assumptions and models under which they work. Through middle range
studies they need to develop clear, concise, unambiguous archaeological signatures of
the social processes they study, because these linking mechanisms are completely
absent in their work in southern Africa. Until this is done their arguments can only
be accepted through special pleading or theoretical faith.
Thus South African archaeologists have through the years employed a number of
different models to account for assemblage variability and composition. The most
common model is the cultural-adaptation model, however, since the mid 1970s many
archaeologists have more intensely stressed, in one form or another, the concept of
adaptation with its overtly functional explanation.
If the cultural-adaptation model
accounted for the majority of assemblage variability then assemblage composition
could be used directly to construct a cross dating scheme. Only Wadley's and Mazel's
research and models might seriously question the utility of using simple assemblage
composition for cross dating purposes, although they do not make the point. Outside of
southern Africa other more general models have been presented that further diminish
the utility of the cultural-adaptation model for explaining assemblage composition.
General Models
223
A more theoretical approach to explain assemblage composition was suggested by
Ammerman and Feldman in 1974. These authors proposed a model that accounts for
relative frequency differences in stone tools without invoking cultural changes or
adaptational changes on the scales implied by H. Deacon (1976) or Sampson (1974),
or the social changes suggested by Wadley (1987) and Mazel (1989). Ammerman and
Feldman (1974: 61 0) suggest that three elements affect artifact relative frequencies
[J.l], and these are (1) the relative frequency of each activity [a] from the total set of
activities performed by a single group within a given period of time, (2) the
'mapping' relations [m] between tools [T] and activities, and (3) the tool droppage
rate [d]. Some of the elements of the model are intuitively understood, but others
require comment. The mapping relations refer to the relations between tools and
activities.
For example, stone tools may be single- or multi-functional, or tasks can
be completed by a variety of different tools. Variations in these mapping relations
will affect the numerical relationship between a single activity and the number of
tools required to complete that task or activity. The droppage rate refers to the
probability at which a tool is abandoned and incorporated into the archaeological
record. Droppage rate reflects the longevity of a tool, and longevity of a tool can be
conditioned by raw material, intensity of use or a number of other factors. This
attribute will be discussed in more detail below. Ammerman and Feldman's model
must be calculated in a series of steps and the first step is to calculate the expected
abundance of each tool type:
Expected Abundance ofT 1 = d1 (m1 a1 +m2a2+ ... mnan)
where
d: droppage rate
m: mapping relation
224
The relative frequency of T 1 is calculated by totalling the expected abundance
values for all tool types as per the formula above and dividing the individual T values
by the total:
Relative frequency of a tool: ll1 = T1/I.Ti
Ammerman and Feldman (197 4: 616) stated a number of years ago that
archaeologists underestimate the effect of droppage rates on the formation of
archaeological assemblages. After 15 years this observation is still pertinent, but
some researchers, especially ethnoarchaeologists, have begun to tackle the problem.
Yellen (1977: 73-84) proposed a model for assemblage composition among
foraging hunter-gatherers. Yellen states that material found on modern San sites can
be classified as maintenance or subsistence related. He argues that subsistence related
artifacts or debris clearly reflect activities that relate to the local resources, but
maintenance activities such as refurbishing tools occur as needed in a quasi-random
or unpredictable cycle. Thus it appears that the San ethnoarchaeological evidence
indicates that assemblage composition is strongly controlled by the need to replenish
exhausted, broken, or lost tools. This is not necessarily related to site location, local
resources, or even subsistence activities that occur at a particular site, but rather to
the need of tool replacement. He states that the likelihood of maintenance activities
(and associated discarded artifacts) to occur on a site is related to the length of time
the site is occupied. Thus site assemblage diversity is related to length of occupation
and possibly the number of occupants.
Recently, Shott (1986, 1989) has suggested that hunter-gatherer mobility and
tool use-life can affect assemblage composition significantly. Shott argues that higher
settlement mobility leads to smaller tools and fewer tools with a wider range of uses.
Additionally he argues that tool use-life is similar in concept to Ammerman and
Feldman's (1974) tool droppage rate, however Shott realizes that the situation is
225
more complicated that simply dropping tools. Shott relates ethnoarchaeological
assemblages from !Kung San and lngalik to manufacturing costs and curation rate.
Thus, in general, the more time and energy spend manufacturing a tool, and the longer
the tool is curated, maintained or rejuvenated, then the slower individual items are
incorporated into the archaeological record. Shott {1989: 17-26) discusses the
"discard process" and notes that more than one process accounts for tools entering the
archaeological record. These processes, with slight modification, are (1) breakage
during manufacture, (2) abandonment during or after manufacture [normally due to
material flaws],
(3) breakage during use, (4) loss or abandonment during use, (5)
recycling [change in tool form and use through rejuvenation], and (6) exhaustion or
total depletion. Additionally each of these six processes influence discard rates and are
conditioned by activities, manufacturing complexity, material brittleness, and
intensity of use. For example in a Later Stone Age assemblage the discard rate for
ostrich egg shell beads would be different for those broken during manufacture than
for those discarded from use-breakage. Also manufacture discards are probably more
frequent for artifacts that require many stages of production (eg. pressure flaked
backed bladelets versus end scrapers) or for artifacts that are made of more brittle or
breakable material (eg. ostrich eggshell versus skin).
The above discussion suggests that simple artifact relative frequencies do not
provide a robust data set for understanding cultural groupings. Variations of
activities, tool/activity mapping relations, tool use-life can all independently affect
the occurrence of artifacts in an assemblage and these variables may not be significant
to culture history.
Models for Explaining Artifact Style
226
In 1966 Sackett presented one of the first really elaborate applications of
morphometric Paleolithic tool typology. He suggested that an artifact type should
represent formal variation that is culturally meaningful, and he further argues that
an artifact type is a configuration of formal elements that comprise an attribute
cluster. By designing a comprehensive attribute system, formal variation within and
between tool types could be defined. In the first application of an attribute analysis to
South African artifact assemblages, J. Deacon (1972) has suggested that among lithic
artifact assemblages in southern Africa metric attributes within a single artifact
group, e.g. scrapers, can be used to identify "stylistic" norms using the mean moment
measures, while the standard deviation provides an indication of how tightly the
"style" is defined. In Africa, as in much of the world, the attribute analysis approach
has not been entirely successful because too much variation exists within artifact
"types" and well defined attribute clusters do not occur often. The source of much of
this variation has not been identified in southern Africa.
Dibble (1987) and Barton (1990), however, have successfully argued that
Middle Paleolithic stone tool morphology can be altered by the intensity of use, reuse,
maintenance, and rejuvenation before discard. This is an ontongeny model for
individual stone tools, rather than cultural systems. Thus tools classified as side
scrapers can change to transverse scrapers or to other "types" simply through
continued use. Especially for stone tools which easily wear and are quickly exhausted,
this developmental model of changing tool morphology can be used to suggest that the
attribute analysis approach, subjectively or quantitatively applied, does not identify
"types" in terms of preconceived and idealized forms of formal stone tool shapes, but
rather many stone tool attributes best reflect the degree of wear or exhaustion at the
time of discard. Examples of stone scraper exhaustion in ethnoarchaeological contexts
227
are the analysis of hide scrapers by Ethiopian hide tanners (Clark and Kurashina
1981; Gallagher 1977). These studies help to explain why attribute analysis of
stone tools rarely provides clear, concise and interpretable patterns (J. Deacon
1972;
Sackett 1966).
Recently Sackett (1982, 1985, 1989) has taken a slightly different approach to
style and suggests that ethnic groupings can be identified by analyzing "isochrestic"
style. lsochrestic means "equivalent in use". Sackett argues that artifacts that have
identical functions can be made in a variety of forms, and it is the "tradition" of craft
production that seems to govern the morphology of an artifact. Sackett argues that if
functionally equivalent artifacts change in any way through time or space then this is
stylistic change. His approach to style in archaeology is to compare functionally
similar artifacts, eg. end scrapers to end scrapers or burins to burins, and look for
temporal changes. Even if these can be related to functional differences, he argues this
still represents isochrestic style.
lsochrestic style is very tempting for the
paleolithic archaeologist, because generally we are able to define functional similarity
between tool types through wear pattern analysis (Binneman and Deacon 1986;
Keeley 1980; although see Shackley and Kerr 1986), through ethnographic or
historic observations (Dunn 1880; Kannemeyer 1890; Lee 1979; MacCalman and
Grobbelaar 1965;
Rudner 1979; Silberbauer 1980), or by well preserved
archaeological specimens or good archaeological context (Clark 1959, 1977; H.
Deacon 1966; although see Wendorf 1968).
Close (1978, 1989) and Close et al. (1979) have proposed a method for
identifying style in backed bladelet artifacts that generally follows Sackett's (1982)
strategy of studying isochrestic style, and her approach is worth discussing as she
analyzed backed microlithic tools. Close has attempted to chose attributes that would
be unlikely to change through use and thus have little or no functional significance.
228
The first attribute is the side chosen for backing; left or right termed sinister and
dexter, respectively. Secondly she suggests the type of backing (Ouchtata, obverse,
inverse, sur enclume, alternating or any combination) is also reflective of
nonfunctional stylistic choice. The third stylistic attribute chosen is which end was
trimmed to a point; distal or proximal. Close (1989: 12) states that these attributes
were assessed for their functional nature by comparing the distribution of these three
attributes to "functional" parameters such as size, shape, and completeness. It would
be advantageous to combine Close's approach with a wear pattern analysis, but it is
unlikely that a single tool type as limited as backed bladelets would change
functionally. Using both cluster analysis and principal component analysis, Close
(1989: 16-20) has been successful at identifying assemblages that are restricted in
time and, to a limited extend, in space.
Even though Close and other archaeologists have enjoyed success with the
isochrestic approach, it is theoretical deficient in that it offers no true explanation of
why stylistic change occurs.
Alternatively Wiessner (1983, 1984, 1985) has
proposed a new "theory" of style that is more robust. She argues that in many cases
"style" is used as a mechanism for communication in social settings. More specifically
style is used for personal and social identification by comparison with other
individuals or groups. Individuals may not be able to put the message into words, but a
message is communicated nonetheless.
She identifies two types of style, and calls
these assertive and emblemic styles. Wiessner, studying projectile points and beaded
headbands made and used by living Kalahari San, argues that morphological variations
in projectile points reflect social groupings and carry emblemic stylistic information.
Different designs on beaded headbands, on the other hand, do not indicate group
identities. Wiessner argues that these are used by the San to show the uniqueness of
the individual maker or owner, and reflect assertive style. Wiessner's theory of style
229
is dynamic, and it can be postulated that as population densities increase and as social
interactions increase so does the need for both social and personal identification. This
implies that the stylistic "load" carried by an artifact assemblage will increase as
population density increases, and Hodder (1979) has argued that stress (social or
environmental) can also create situations where the stylistic loads in artifact
assemblages may increase. However a major operational problem exists in terms of
identifying emblemic or assertive styles in prehistoric contexts. Obviously an
artifact must be used in a manner so that it is clearly visible, otherwise individual or
group comparisons would not be possible, but Wiessner offers no solution or strategy
to help resolve the problem of style identification. Thus ethnoarchaeologists have a
glimmer of hope, but prehistoric archaeologists have little.
Sampson (1988: 16) now embraces Wiessner's approach to style in his analysis
of decorations on prehistoric Smithfield ceramics, and he suggests a method for
determining the presence of Wiessner's emblemic style from archaeological data. It
should be added that decorative motifs on ceramics are unlikely to have functional
significance in terms of vessel use, or change through use, and they are not obscured
from view as are hafted lithic tools such as scrapers or backed tools. Thus ceramic
motifs have many advantages over stone tools for stylistic studies. However Sampson's
approach requires detailed quantified data from all sites within a large area so that
isopleth maps can be constructed for individual motifs on ceramics. It is especially
helpful if the ages of ceramic motifs are known as well. In this way motif boundaries
presumably representing emblemic style changes along group territorial boundaries
through space and time can be represented by steep drop-off gradients with clear
shoulders. Unfortunately no approach has been developed for identifying emblemic or
assertive styles in lithic assemblages excavated from a few or single sites. It appears
230
that at a minimum detailed spatial information is necessary for the application of
Wiessner's approach.
Conclusions
Theoretically, analysis of assertive or emblemic styles would seem to provide the
most robust patterns for any cross-dating scheme, but because of the difficulty in
identifying assertive or emblemic styles with prehistoric lithic artifact assemblages
the cross-dating scheme (Chapter IX) has, by default, focused on isochrestic styles. It
is possible that assertive or emblemic styles might be registered in the isochrestic
styles analyzed from the rockshelter assemblages, but one cannot know for sure. In a
very narrow sense and solely for the purposes of developing a cross-dating scheme, it
is more important to identify chronological patterns than to understand why these
patterns exist. Although as more refined models are developed an explanation of these
patterns becomes increasingly important for a specific research area. In the
following chapters a model and a series of hypotheses are presented in an attempt to
explain some of the chronological patterns, however a robust test of these hypotheses
must await future research that incorporates larger artifact samples with better
paleoenvironmental associations, comparisons with faunal and other economic
remains, and surface site distributional data.
CHAPTER XI
ECOLOGICAL AND BEHAVIORAL BASIS FOR A NEW MODEL
With the single exception of H. Deacon's (1976) processual model, all other
pioneer models generated by South African archaeologists (Chapter X) offer
monolithic (single-cause) explanations of Later Stone Age assemblage variability and
design change. Here, I present the theoretical underpinning for a more elaborate
processual model that attempts to take into account several interacting forces. The
model, firmly based within the paradigms of Culture Ecology and Foraging Theory, is
essentially deterministic and views environmental changes as the primary causal
agent. It incorporates some social processes and interactions, but it does not give them
a primary causal role as do the models of Wadley (1988) and Mazel (1989).
It focuses instead on problems arising from the ebb and flow of food supplies on
the landscape, and on the ways that hunter-gatherer-foragers go about acquiring food
resources and insuring that they have access to food sources. Unlike the other models
from southern Africa, this one attempts to take full advantage of the available
ethnological data on global hunter-gatherer ecology and behavior. However, it
eschews the use of ethnographic analogies with specific groups. It also avoids much of
the systemic jargon used by hunter-gatherer theorists, although processual terms
will be mentioned parenthetically.
For clarity, the model will be compiled into two stages. In this chapter I outline
the man-land interactions that constitute the model's foundations. In Chapter XII, I
show how the model predicts various patterns of archeological fallout in changing
231
232
circumstances. Finally, the combined ecofact and artifact data from Blydefontein will
be used as a first and partial test of the new model.
Design and Structure
The model is designed to take advantage of the huge spatial data-base offered by the
Zeekoe valley survey (Chapter 1). The goals of the design are to explain assemblage
composition and changes in artifact shape within a framework of settlement pattern
changes. Thus the dominant mechanism within the model will be hunter-gatherer
mobility patterns.
Figure 91 summarizes the model's four basic components. There are two
ecological parts (climate and resources) and two adaptive-behavior parts (risk
response and range). Each component is modelled as a system of interacting variables.
The first system (climate) serves here as a prime mover, thus it acts as an
essentially closed system, not influenced directly by the others. The second system
(resources) directly influences both the third system (risk response}, and the fourth
system (range). The resource system is neither closed not is it fully open. Thus, the
risk and range systems have some feedback effect on resources. Figure 91 shows
influence-arrows of different thicknesses to diagram weaker feedback and the porous
(rather than open) nature of the resource system. Again, the third system (risk
response) has a stronger influence on the fourth (range) than vice versa. Although
range has less feedback-influence on resources, there is a stronger connection to risk
response.
The climate system has two variables. Both rainfall and temperature will be
shown to affect, through resources, several different aspects of hunter-gatherer
adaptive behavior.
233
The resource system has four interacting variables embedded in it: abundance,
the plant/animal ratio, seasonal timing, and patchiness.
Different strengths of
feedback influence exist between the four variables but they cannot be generalized, so
arrows are diagrammed at equal thickness in Figure 91.
Response to risk entails another five interacting variables: camp moves,
collecting trips, plant/meat ratio in the diet, exchange, and storage. Each one is an
adaptive response to uncertainty about future supplies of resources. No attempt is
made to model different strengths of feedback influences between the five variables.
The range system contains three interacting variables. No attempt is made here to
distinguish between range and territory. Population density has a dominant influence
on range size and mobility, with some feedback from the other two. Range size and
mobility are heavily interdependent, however.
Responses to Risk Posed by Resource Variability
Hunter-gatherers calculate daily where they will get their next meal, and the
several meals thereafter. This calculation deals with partly predictable changes in the
total amount (richness, abundance) of food/water, called food hereafter, in their
range. Also, changes will occur in the food's seasonal availability, in its patchy
distribution across the range, and in plant/animal content (diversity). Of course
other items are of concern, but food is so basic that secondary items (resources) play
an insignificant role in the calculation of risk. Thus the food quest is the central drive
mechanism of the model. Throughout, the model assumes that responses to perceived
risk are aimed at reducing uncertainty. Strategies for reducing risk (coping
strategies) may include changes in mobility patterns within the range, different
emphasis on plant or animal inclusion in the diet, changes in food storage patterns,
234
increased sharing/exchange, and increased between-band visits (Cashdan 1983,
1985;
Wiessner 1977, 1982;
Winterhalder 1986).
I. CLIMATE
Rainfall
セ@
Temperature
I
u
II. RESOURCES
Abundance
I
Seasonal Ti mi ng
Plant/Animal Mix
)
(
)
UT
Ill. RESPONSE TO RISK
Camp Moves
1
セ@
Collecting Trips
I
Pate hi ness
ui
IV. RANGE
Size E
> Population
\
/Den•ity
Mobility
Figure 91. Ecological and behavioral model's four components.
Mobility Patterns: Foraging and Collecting
It is useful to distinguish between two options (exploitation strategies) by which
a band can acquire food from its home range: foraging and collecting (Binford 1980).
Foragers gather foods daily, normally return to residential camps at the end of
each day. Daily forays involve visiting a few places within the camp's catchment area,
235
and food is not extensively processed at places outside of the camp. They tend not to
store great quantities of food, but rather foragers transport people to resources by
moving residential camps frequently. Residential sites are abandoned when resources
within local site catchments are depleted. Foraging strategies are designed to exploit
foods that have a fairly even distribution through space and time.
However variations
in the abundance, distribution, availability and predictability of foodstuffs determine
how often a camp is moved. Foragers rarely organize special task groups who leave
camps for extended periods of time in order to exploit distant but rich food patches.
Collectors live in higher-risk, more stressful environments with very limited
and patchy food supplies, such as the arctic, and they depend heavily on mobile animals
and less on plants.
They tend to transport resources to people, and acquire food in
bulk whenever they can. They also store it for seasons when nothing else is available.
For this reason, their mobility pattern involves relatively few camp moves each year,
but organized task groups leave camp on long trips (logistic mobility) and stay at
selected areas (procurement sites) to take advantage of some briefly available plants
or animals. These foods are partly processed here, and brought back to camp. It is in
this light that collectors are seen as moving food to people and not vice versa as with
foragers.
There are few ethnographic examples of "pure" foragers or collectors, and most
hunter-gatherers use a judicious mix of the two.
In fact, Kelly (1983: 299)
comparison of 37 groups shows that as the average number of yearly camp moves per
group decreases, so the average number of task-group trips per year increases
exponentially. This pattern is especially strong for hunter-gatherers who are heavily
dependent on hunting rather than gathering or fishing.
In fact, Winterhalder (1986:
377) has argued that as risk increases there is a tendency for hunters to focus on
fewer prey species. However, where gathering increases in importance, it can be
236
shown that the number of camp moves may decrease without any increase in taskgroup trips. According to Kelly's (1983: 282, 299) data the Dobe !Kung and the G/wi
have some of the highest rates of gathering plant foods, and they are among the groups
with the lowest mobility (residential and logistical) that still would be classified as
mobile hunter-gatherers. The Kalahari San do not need to move a great deal,
presumably because of the abundance and density of plant resources. What determines
the exact mix employed by a particular band is explored next.
Ecological Determinants of the Mobility Pattern Mix
Murdock's (1967) survey of hunter-gatherer societies can be used as a rough
guide to climate/mobility relationships.
Unfortunately Murdock uses a system of
three mobility categories that does not match exactly with Binford's (1980)
forager/collector dichotomy, although an approximation can be made (ibid: 14-15).
In Table 27 the number of ethnographic groups in each of Murdock's mobility classes
is tabulated against six gross habitats distributed from tropics to arctic. Using
standardized deviates (high values underlined) the table suggest that mobility
decreases with decreasing temperature until arctic environments are encountered, at
which point the preferred pattern reverts back to highly mobile.
It is likely that the
most mobile groups in the tropical and semi-tropical zones are foragers who move
camp frequently, while the arctic groups are collectors who take long task-group
trips from a base camp. Murdock's semi-nomads in the warm-temperate and cool
zones probably represent a mix in which both strategies are used. Semi-sedentary
and sedentary groups in boreal environments include the Northwest Coast Indians who
were sedentary hunter-gatherer-fishermen exploiting the highly predictable and
abundant salmon.
237
Table 27.--Number of groups tabulated by categories of nomadism and environment.
Numbers in parentheses represent standardized deviates and unusually high values
underlined (data from Binford 1980:14, Table 2).
EnvirQnment
tropics
semi-tropics
warm temperate
cool temperate
boreal
arctic
fオャセ@
9
9
3
4
5
5
NQmadiQ
(U..)
(a&)
(-1.4)
(-2.1)
(-1.4)
(La)
35
Total
セ・ュゥM、ョエ。イF@
セ・ュゥMnq。、@
2
4
21
32
21
4
(-1.6)
(-1.1)
(.L.a.)
(L1.)
(-0.3)
(-0.8)
84
1
1
8
17
19
3
(-1.3)
(-1.5)
(-0.4)
(0.4)
(La)
(-0.3)
49
Primary Plant Production as a Determinant of Forager Mobility
The number of camp moves by foragers is controlled, in part, by the richness of
the above ground productivity of the exploited plant communities (Figure 92). The
curvilinear regression shows that for foragers, in general, at the low end of the
primary production scale, as plant productivity drops the number of camps moves
increases. The number of camp moves increases at the high end of the scale because
most of the plant biomass is locked up in trees and other inedible plant materials
(Kelly
1983).
Climate as a Determinant of Food Selection
Climate also affects the plant and animal biomass of the band's range, and this in
turn influences decisions about how much hunted food versus plant food will become
incorporated in the diet. Both annual rainfall and mean annual temperature are used.
Kelly's (1980, 1983) uses a temperature measurement, known as effective
temperature (Bailey 1960), that is based on the average annual temperature (T) and
the average annual range of temperature (AR). Effective temperature (ET) is
calculated by the formula:
ET
=
(8T + 14 AR)/(AR+8)
238
55
1/)
(])
セ@
50
E)
45
c..
40
E
res
(,)
.....
(])
g'
.....
0
LL.
0
.....
(])
.0
E
::I
z
0
00
35
30
0
25
20
0
15
10
oO
5
0 0
0
0
0
500
1000
0
1500
2000
2500
Primary Production (g/m2)
3000
3500
400(
Figure 92. Changes in average number of forager camp moves with changing plant
primary production (Number of camp moves = 35.381 - 0.032x + 9.901-6 x2,
r2 = 0.437, p value = 0.0318) (data from Kelley 1983).
Effective temperature considers the average temperature as well as the variation
of temperature through the seasons for any specific region. Thus effective
temperature increases as average annual temperature increases .QL as the annual range
of temperature increases. The average ET value at the equator is 26 and the average
ET value at the poles is 8 (Kelly 1980: 12), thus ET is a fair measure of the
insolation available for plant growth.
A scatterplot between ET and percent of hunted resources in the diet of terrestrial
hunter-gatherers (Figure 93) demonstrates that a curvilinear regression best fits
the data. Between 8 (arctic) and 18 (temperate) the contribution of hunted resources
to hunter-gatherer diets declines with increased ET, but between 18 and 26 (equator)
the regression reverses to a positive relationship. This change must be related to the
239
major shifts in global plant communities and primary production, especially in
tropical forests.
The relationship between mean annual rainfall and hunting is also complex, and it
can be explored with two studies. First the amount of large herbivore biomass in east
and southern African environments increases as rainfall increases (Figure 94) at
least within the range of measurement (Coe et al. 1976: 341-354). This
relationship shows that moderate changes in rainfall can bring about important
increases in the herbivore biomass. Individual species do not all respond in tandem to
changes in precipitation because of interspecies competition and other local factors,
but large herbivores as a single group do. It would be expected that as animal
abundance increases so might its use by hunter-gatherers.
90
Qi
0
80
.9
70
(f)
0
Q)
....
::;,
(.)
0
60
0
0
(f)
Q)
a:
50
""0
$
c
40
0
30
'E
Q)
20
::;,
I
セ@
co
0
0
Q)
0..
10
0
0
8
10
12
14
16
18
20
22
24
26
Effective Temperature
Figure 93. Scatterplot of effective temperature and percent of hunted resources
contributed to diet (Second Order Polynomial, r2=0.407, p value = 0.0198}.
Hunter-gatherer groups include Punan, Mbuti, Vedda, Bihor, Aeta, Semang, Chenchu,
Siriono, Guayaki, Hadza, G/wi, Dobe !Kung, Aranda, Walapai, Maidu, Northern Paiute,
Montagnais, and Ona (data from Kelly 1983: 280-281, Table 1).
240
Large Herbivore Biomass (kg km2)
YPKMセN⦅@
8000
7000
6000
5000
4000
3000
2000
1000
PセMイK@
100
200
300
400
500
600
700
800
900
1000
Annual Precipitation
Figure 94. Regression between annual precipitation and large herbivore biomass in
east and southern Africa (data from Coe et al. 1976). Herbivore biomass = 8.52
(annual precipitation}
- 833.5, r2 = 0.709, p value < 0.0001.
-
.9:1
0
80
70
0
.!::
(/)
Q)
セ@
60
0
0
0
::I
0
(/)
Q)
a:
"0
.l!l
c
-::I
I
50
40
30
0
c
Q)
20
00
0
セ@
Q)
a.
10
0
0
0
500
1000
1500
2000
Rainfall (mm)
2500
3000
350(
Figure 95. Scatterplot of annual rainfall and percent of hunted resources contributed
to diet (Second Order Polynomial, r2=0.324, p value = 0.0785}. Hunter-gatherer
groups include Punan, Mbuti, Vedda, Bihor, Aeta, Semang, Chenchu, Siriono, Guayaki,
Hadza, G/wi, Dobe !Kung, Aranda, Walapai, Maidu, Northern Paiute, Montagnais, and
Ona (data from Kelly 1983: 280-281, Table 1, and Fullard and Darby 1975:
climatology maps}.
241
Rainfall-diet relationships can be further explored with Kelly's (1980, 1983)
data and climatic figures gleamed from world climatic maps (Fullard and Darby
1975). A second order polynomial (Percent hunted resources = 18.168 + 0.037x 1.113-5x2) provides a weak although significant fit to the data (r2 = 0.324, p value
=
0.0785), and a plotting of the data with the regression curve shows (Figure 95)
that a simple linear regression between rainfall and hunted resources does not exist.
Between 0 to 1750 mm rainfall the percent of hunted resources increases as rainfall
increases, then between 1750 to 3500 mm the percent of hunted resources declines
as rainfall increases.
Food Selection as a Determinant of Mobility Pattern
Hunting versus Gathering
The amount of hunting versus gathering practised by a band will have a marked
effect on its selection and mix of mobility patterns. Kelly's (1980, 1983) data cover
many groups, and the data include useful measurements such as the average number of
camp moves per year, average distance of camp moves, total distance of camp moves,
length of winter site occupation, average distance of task-group moves, average length
of time of task-group forays, and total exploited area. In spite of the several gaps in
this data base, particularly for hunter-gatherers in temperate zones, a rough
measure of overall mobility is obtainable. The average number of camp (residential)
moves of a group multiplied by its yearly average duration of special task-group trips
(logistic forays) provides a measure of overall mobility. When the overall mobility
for eleven groups is plotted against the percent of hunted food in hunter-gatherer diets
(Figure 96), it appears that overall mobility increases as the percent of hunting
increases (Overall mobility = -5.283 + 4.229x - 0.11 x2 + 0.001 x3).
One
implication of this is that mobility decreases as the amount of gathered plants in the
diet increases. For example, the Dobe !Kung and the G/wi ranked among the highest
242
percentages of plant food in the diet, and they are among the groups with the lowest
mobility. The Kalahari Bushmen (San) do not have to move a great deal, presumably
because of the abundance and density of plant foods.
Fishing and Mobility
At one extreme in Murdock's array of hunter-gatherers from temperate zones
are the semi-sedentary and sedentary groups in rainy, boreal habitats where they
could exploit highly predictable and abundant resources such as the salmon runs used
by the Northwest Coast Indians. These cases suggest strongly that local, specialized
selection of prey/plant species can distort any global trends in the mix of mobility
patterns used.
..c
0,
450
c:
400
a.
·;:::
350
Q)
_J
f..lJ::
(/)
cu
f-
Iii
·c:;
セ@
(/)
..
0
300
250
200
150
0
(/)
Q)
セ@
::2:
a.
E
cu
(.)
100
50
0
0
-50
0
10
20
30
40
50
60
70
80
90
%Hunt
Figure 96. Scatterplot of overall mobility (average number of camp moves * average
length of special task trip) and percent of hunting (Third Order Polynomial,
r2 =0.501, p value = 0.1591 ). Hunter-gatherer groups include Punan, Vedda, Aeta,
Chenchu, Guayaki, G/wi, Dobe !Kung, Cheyenne, Crow, Sanpoil, Numamiut, and Ona
(data from Kelly 1983: 280-281, Table 1).
243
More useful is Kelly's (1 980, 1983) between-group comparison using various
measurements of hunter-gatherer mobility patterns. For fifteen groups there is a
rough but inverse correlation between length of time spent at a winter camp and the
amount of fish in the diet (Figure 97). As winter residence time is a rough indicator
of overall mobility (the longer the time in camp, the less mobile the group) these data
lend quantitative support to the notion that fishing specialization will override the
effects of climate, especially in temperate zones, on the choice and mix of mobility
patterns.
12
.r::.
0>
c
.3
c
·-;0
10
0
8
c.
::J
(.)
(.)
0
c.
E
ro
0
セ@
セ@
....
6
0
0
0
4
2
0
-1 0
0
0
10
20
30
40
50
60
70
Percent of Fish in Diet
Figure 97. Scatterplot of percent of fish in diet and winter camp occupation length
(r2=0.241 ). Hunter-gatherer groups include Punan, Vedda, Anadmanese, Chenchu,
Siriono, G/wi, Dobe !Kung, Crow, Sanpoil, Numamiut, Mistassini Cree, Montagnais,
Ainu, Maidu, Northern Paiute, Nootka, Klamath, and Ona (data from Kelly 1983:
280-281, Table 1 and 2).
Food Patchiness and Mobility
There are two aspects to patchiness: the one spatial and the other temporal.
Conaty (1987) makes another useful distinction by pointing out that spatial
patchiness can be viewed on a sliding scale from clumped to dispersed (Table 28).
244
Temporal patchiness may be stable to transient. Obviously, habitats characterized by
clumped, transient plants and animals promote collector mobility patterns, while
dispersed, stable food distributions promote foragers. Habitats with dispersed and
transient resource patterns are not viable because they are not rich enough to support
hunter-gatherers.
Table 28. Resource temporal and spatial patterns and optimal associated huntergatherer mobility patterns.
Resource
Temporal
P rn
Stable
Resource Spatial Pattern
Sedentary H-G
Foragers
Hayden (1986: 85-86) makes a further distinction between habitats with
animals having long reproductive cycles (K-selected resources) and those with short
reproductive cycles (r-selected resources). Habitats dominated by K-selected foods
yield prey with with few offspring per generation and large body sizes, and they
promote collector mobility patterns. On the other hand, habitats with r-selected foods
yield animals with high numbers of offspring per generation that provide an abundant,
reliable food base for a short time each year. He argues that they also promote
collector patterns among hunter-gatherers in temperate climates.
Foraging, it is
asserted, is the best option in warmer habitats with dispersed, transient foods, but
this does not take overall habitat richness into account.
In some patchy habitats, hunter-gatherers are likely to switch mobility pattern
in the course of the annual round. When foragers move camp to another location, they
try to select an area where the food supply has been rested and allowed to recover
245
since the previous cropping. They also prefer to move to a site where foods are as
evenly distributed in the the camp's catchment as in the one from which they have just
left. If they are forced to move to an area where vital foods are unevenly distributed,
and clustered such that daily foraging will not produce enough food for group use, then
they may resort to long collecting trips, and store enough for future use.
However, Binford (1980: 5-1 O) notes some interesting cases where collectors
switch to a variant of forager-style camp moves. This pattern occurs among the
Yahgan, Slave and Copper Eskimo who position their camps to take advantage of
predictable foods as they become available sequentially. These "serial specialists"
(ibid: 17) are solving problems like foragers although they live in more stressful and
risk prone environments that often do not promote foraging mobility patterns. Serial
specialists can be expected in any environment with highly predictable animal
migrations and/or patchy plant foods with short, rapid growing seasons.
The type of patchiness involved may also affect storage decisions among huntergatherers. Even though collectors tend to store food, their decisions to form caches is
dictated more by a temporally patchy food supply than by spatial patchiness
(Bamforth 1988: 16). Spatial patchiness, however, is more likely to provoke
decisions to organize long, special task-group trips, and not by itself promote storage.
One final point about food patchiness is that not all of the food in a habitat may be
accessible to humans. This is especially so in tropical environments, particularly in
rain forests where most of the above-ground biomass in either inedible or out of reach
(Kelly 1983).
In fact for tropical hunter-gatherers, the number of camp moves
increases exponentially as the primary biomass increases (ibid: 291 ).
246
The Determinants of Range Characteristics
Carrying Capacity and Range Size
The richness, diversity and distribution of food supplies within a band's range
(territory) determines the number of people in a band, and, in part, the size of the
band's range. As usual the needed detailed environmental information is not available
for hunter-gatherer groups, however estimates of primary productivity in a formula
developed by Rosenzweig (1968} was published by Kelly (1983). The formula is:
log1 oNAAP = (1.66±0.27} log1 oAE - (1.66±0.07)
where NAAP is net annual above-ground productivity of plants in grams per
square meter, and AE is actual evapotranspiration in millimeters (Rosenzweig 1968:
71).
Evapotranspiration (water loss from respiration of plants and evaporation) can
be viewed as the opposite of rainfall, and it is based on precipitation and temperature.
As AE increases net above-ground primary productivity of plants increases. Even
though primary productivity is a crude estimate of plant food availability, it is useful
to see the response that hunter-gatherer range size has in relation to this variable
(Figure 98}.
The curvilinear regression in Figure 98 is weak and the p value shows that the
correlation is a borderline significance level at best (r2
=
0.155, p value
= 0.185).
As stated above primary productivity is not the best estimator of hunter-gatherer
carrying capacities, but unfortunately it is the only one we have. Nevertheless, it
shows that range size increases with low or high primary productivity values. The
increase in range size as primary productivity (carrying capacity) drops is expected,
but the increase in range size with higher primary productivity values is surprising.
This is due to reduced amounts of edible foods in tropical ecosystems where greater and
greater amounts of biomass are locked up in tree trunks and other materials that
247
cannot be eaten by humans (Kelly 1983: 286). The high variability at the low end of
the primary productivity scale is probably due to some desert or semi-desert
environments with fairly high carrying capacities for hunter-gatherers due to an
abundance of edible below ground roots and bulbs. By averaging winter and summer
potential evapotranspiration from Venter et al. (1986: Figure 13a and i 4a) modern
primary production for the Blydefontein area is approximately 100 g/m2.
12
11
0
0
10
C\1
E
.::.:.
CIS
セ@
9
0
8
0
0
<(
"iii
0
7
.£
6
!::..
0
0
5
0
0
4
3
-500
0
0
0
500
0
1000
1500
2000
2500
Primary Production (g/m2)
3000
3500
Figure 98. Curvilinear regression between hunter-gatherer range size and primary
productivity, r2 =0.155, p value = 0.185. Groups include the Pun an, Mbuti,
Semang, Vedda, Anadamanese, Aeta, Siriono, G/wi, Dobe !Kung, Aranda, Walapai,
Crow, Maidu, Micmac, Nootka, Twana, Southern Kwakiutl, Klamath, Ainu, Makah,
Mistassini Cree, and Nunamiut (data from Kelly 1983: 280, 282).
Population and Range Size
Population density also controls the size of the range. Many theoretical studies
have argued that as population density increases, all other factors being equal, range
size decreases due to greater territory constriction and population packing. Thirtyeight observations from Kalahari San groups (Hitchcock 1982; Silberbauer 1981;
400(
248
Wiessner 1977; Yellen 1976) show that this premise is correct for most modern
San (Figure 99). These data consist of multiple and single observations taken during
the 1960s and 1970s, and represent groups experiencing a variety of social
processes that include group fission, transition toward sedentism due to increased
population constriction, and increased interactions with agro-pastoralists Bantus and
Europeans who drew the San into their own economic systems. Notwithstanding all
these complicating factors, population density as measured by the natural logarithm of
persons per square kilometer has a significant {p value = 0.0012) although weak
correlation (r2
=
0.25) with the natural logarithm of territory size (see Figure 99).
Figure 99. Scatterplot of territory area (km2) and population density as measured
by persons per km2, r2 = 0.25, p value = 0.0012.
San groups include Kua
(Khwee1, Khwee2, Diphala, Ana-0, Go/to, Ramokgophane, and Pulenyane), !Kung
(Dobe 1964, Dobe 1968, Dobe 1969, and /Xai/Xai), G/wi (*xade, G!o:sa, Easter
Pan, Kxaotwe, Tsxobe, Piper Pans, G*wi/dom, Dantukwe, Lana, //oege, Sibobane,
//Hue, Kikao, Monatsha, Metse-a-monong, /o*we, and Molapo-Gyem), and !Xo
(N*haite-Hukuntsi-Nwatle, Pepane-Lehututu-Monong, Hukuntsi-Tshotswa, TshaneLotlhake, and Kang) (data from Hitchcock 1982: Table 11.5; Silberbauer 1981:
193; Wiessner 1977: 19, 282; Yellen 1976: 54-60).
249
This demonstrates that as population densities increase, San territory size decreases,
and it is suggested here that greater densities of people on the ground are associated
with tighter territorial packing and greater constriction of territories.
Hunter-
gatherers with lower population densities and larger territories have greater
mobility than groups with smaller territories and greater densities.
Food Selection and Range Size
Kelly (1983: 298) argues that range size (total exploited area) is strongly
related also to the amount of hunted food in the diet. It is unfortunate that there are no
comparable data for the San groups discussed above, but Kelly's data for 23 groups do
show clearly that as the amount hunting increases so does the range size (Figure
100). This makes sense as animals must be less dense on the landscape than the plants
on which they survive. Because energy is lost as it is transferred up trophic levels, a
(hypothetical) group that depends entirely on animals for food is one complete trophic
level above another (hypothetical) group that depends solely on plants for food. Thus
hunters have less available energy per km2 than gatherers.
Mobility Patterns and Range Size
Among foragers, it is axiomatic that the number of camp moves per year will
increase as range size increases. It is again unfortunate that San data are incomplete,
but Kelly's (1983) figures can be used to show the relationship (Figure 101). They
also show total distance moved per year against range size (Figure 102). Of interest
are two groups (the Aeta and Semang) do not move as far in relation to the size of their
ranges, compared to others. Both have clustered, stable and predictable food supplies
that allow long sojourns at individual camps. In spite of these two exceptions, both
regressions show positive relationships between the two variables.
250
12
11
0
0
10
C\1
E
9
ro
Q)
....
8
Ci1
0
7
..s:
6
セ@
<(
t:..
5
4
0
3
10
0
30
20
40
50
60
70
80
90
Percent Contribution of Hunting to Diet
Figure 100. Linear regression between percent contribution of hunting to diet and
natural logarithm of total area exploited, r2 =0.509. Groups include the Punan,
Mbuti, Semang, Vedda, Anadamanese, Aeta, Siriono, G/wi, Dobe !Kung, Aranda,
Walapai, Crow, Maidu, Micmac, Nootka, Twana, Southern Kwakiutl, Klamath, Ainu,
Makah, Mistassini Cree, and Nunamiut (data from Kelly 1983: 280, 282).
400
Pun an
350
,.....
300
'-./
250
E
::.f.
ell
Aeta
!Kung
0
()
c:.
.....VI
0
lf1
r;;
0 Siriono
200
Semang
150
......
0
1-
100
50
0
-500
0
500
1000
1500
2000
Total Area (km2)
2500
3000
Figure 101. Scatterplot of total exploited area (km2) and total distance moved
between forager residential camps. Birhor, Mbuti and Vedda used in both linear
regressions (data from Kelly 1983: 280-282).
3500
251
VI
55
>
0
I:
50
...
セ@
......
...
""iii...
....0::::
...
c:
0
L
.Q
45
40
!Kung
35
0
30
25
E
20
z
15
10
Aeta
:;,
...
...
>
0\
tV
Mbuti
L
<
5
0
-500
0
500
1000
1500
2000
Tota1 Area (km2)
2500
3000
3500
Figure 102. Scatterplot of total exploited area (km2) by number of forager
residential camp moves. Mbuti, Aranda, Birhor and Vedda used in both regressions
(data from Kelly 1983: 280-282).
Range Size and Reciprocity
A change in the productivity of plant or animal foods can influence range size and
even the use of an individual range, but Wilmsen (1989: 180-225) argues that
specific Kalahari San kin groups have long associations with given tracts of land and
this suggests long established home ranges. Wiessner (1977, 1982) argues that
among Kalahari foragers a common way to reduce economic risk is to use social
obligations to pool or share risk among the regional population. One important social
mechanism that the Kalahari !Kung use to bond their social obligations is an exchange
system known as hxaro which provides a tight but nevertheless far-flung network of
complex social obligations that can be called upon by needy families during times of
food shortage.
Wiessner (1977) and Wilmsen (1989) argue that hxaro is intimately
linked with kinship obligations and mate recruitment, and a line chart of the distance
between spouse birth places and the distance between hxaro partners in two !Kung
252
groups shows the spatial similarity between mate recruitment networks and hxaro
networks (Figure 103).
Wiessner (1977:211-214) identifies three ever-increasing spatial scales of
risk (personal, local and regional) which are absorbed by larger and large social
units (extended family, local bands, regional populations). It is expected that regional
responses to uncertainty would be most visible in the archaeological record. Wiessner
(1977: 60) and Yellen (1977: 41-4 7) provide models of individual movements
(band fission) in response to risk which is channelled by hxaro and kin ties. Even
though detailed information is lacking on how exchange systems and individual
movements react to environmental change, it is clear that during periods of low
production and great risk when food sharing among the local band cannot compensate
for food shortages, but before population densities drop, increased levels of exchange
occur (Wiessner 1977: 154-160).
Then with continued shortages individual
families move to areas with kin and/or hxaro partners with more plentiful resources.
70
60
50
40
Percent
30
20
10
0
30
60
90
120
150
180
210
240
270
Distance (km)
1111
/ai/ai-hxaro
mD
'cum!kwe-hxaro ·•- Spouse Births
Figure 103. Distances between hxaro partners in two exchange systems (/ai/ai and
'cum!kwe), and distances between spouse births (data from Wilmsen 1989).
253
Global Trends in Hunter-Gatherer Risk Reduction
An integrated summary of all the variables discussed in the foregoing sections is
best organized on a framework of sliding scales of temperature and precipitation that
combine to produce gross global habitat types. These are flagged in the two columns on
the left of Figure 104 where arctic conditions are arbitrarily divided from plains,
plains from temperate forest, and so on through desert, savannah and tropics.
Intermediate habitats (eg. semi-desert, woodland savannah, etc.) fit between these
arbitrary units, but are omitted from Figure 104 to reduce clutter.
Under Resources, total biomass is diagrammed in the third column to peak in the
temperate arboreal habitats and again in the rain forests at the top of the diagram. The
planVanimal mix in the biomass is diagrammed in the fourth column so that plant
frequencies peak in temperate boreal forests, in deserts, and again in the tropics.
Animal frequencies peak under arctic, semi-desert and savannah habitats.
Appropriate food spacing is generalized in the fifth column to indicate maximum
clumping in arctic and plains habitats, with steadily reduced clumping covarying with
increasing temperature. Food timing (in the last column under Resources) is
modelled to show maximum stability in the temperate forests and again in the tropics.
The first column under Mobility attempts to generalize the appropriate mix of
collector/forager strategies best suited to each habitat. There is a rough decrease in
the frequency of collecting (logistic strategies) until semi-sedentary mobility
patterns are reached in peak temperate forests. Collecting increases somewhat in
semi-deserts, then declines with rainfall until the disperse/transient conditions of
full desert are met and the landscape is essentially abandoned. Collecting increases
again then declines erratically towards full rain forests.
Storage is more frequently
adopted in habitats with erratic timing in the food supply, such as the arctic and
deserts, and decrease again in the tropics with a slight increase in rain forests.
254
Diversions from the trends are built in at random to accommodate niches where
fishing is an alternative option so that total mobility can be reduced. Finally, range
size covaries with mobility, except in rain forests.
Application of Global Trends to the Blydefontein Region
The diagram in Figure 104 also serves to show how the model will be converted so
that it can be applied to regional data. In Figure 104 the layout is designed vertically
to cover global space from pole (bottom) to equator (top). It can be rapidly converted
for regional use by redesigning the vertical layout to represent an elapsing time scale
of a single area. In this layout, older is at the bottom and younger is at the top. In this
scenario, a single region has experienced the whole gamut of global habitat changes.
Although this is possible for the real world during very extended (geological) time,
the model in this form is of limited use in human prehistory.
It will be the central portion of the diagram that will be most useful for modelling
purposes in the Blydefontein area during the Late Pleistocene, Holocene and to recent
times. The base of this portion starts close to the arbitrary arctic/plains boundary,
and extends up to the beginning of the temperate arboreal habitat (represented by the
Fynbos vegetation in South Africa). It starts again above the full arboreal habitat and
extends to full desert conditions. This is the range of habitats with which huntergatherers of the Blydefontein region had to contend. This model predicts in a general
fashion their appropriate risk-reducing responses to each.
255
CLIMATE
RESOURCES
MOBILITY
RANGE
otion Density
Size-
Figure 104. Diagram of the ecological and behavioral model variables on world scale.
CHAPTER XII
THE ROLE OF TECHNOLOGY AMONG FORAGERS AND COLLECTORS
The new model readily can be fitted to frequency data derived from ecofacts, but it
is not useable yet for archaeological purposes. A fifth system, technology, must be
connected to the other four (Figure 105) before the model can be made to produce any
test implications that fit archaeological data. In this extension of the model, the
technology system contains several interacting variables (see Figure 105). These are
six groups of variables that refer to the qualities of tools and weapons made by
hunter-gatherers to help buffer risk.
The pairs of qualities are:
expedient/reliable/
maintainable tools, specialized/generalized tools, cached/carried tools, finished/
unfinished tools, use-life/production cost of tools, and use lives of repair kit/toolsweapons. Each group will be shown to articulate with the sliding scale of
collector/forager mobility ratios outlined in Chapter XI.
It is these -pairs of variables
that allow test implications for archaeological data to be presented.
Responding to Risk with Technology
Not only do hunter-gatherers combine mobility patterns, store food, and adjust
plant food/meat intakes to buffer themselves against risk, but they also adjust their
technology with the express purpose of reducing risk (Torrence 1989: 60). There is
a small body of ethnoarchaeological evidence from which to derive rules about the
ways in which hunter-gatherers make tools (design strategies) in order to stave off
risk. Furthermore, those rules produce global trends that can be compared with the
trends in the adaptive behavior reviewed in Chapter XI. The development of such
trends are essential if the new model is to become testable through archaeological data.
256
257
I. CLIMATE
r。ゥョヲャセ@
Temperature
I
u
II. RESOURCES
Abundance
+-(- - " " ) )
I
Seasonal Timing
セ\@ MセI@
Plant/Ani mal Mix
I
Pate hi ness
ui
Ill. RESPONSE TO RISK
Camp Moves
セ@
ui
IV. RANGE
Collecting Trips
1
Size <
> Population
\
/Density
Mobility
1
un
V. TECHNOLOGY
Reliable /Maintainable /Expedient
I
Cached/Carried
I
Finished/Unfinished
j 1
Specialized /Generalized
I
I
Use-life/Production Costs
t------------7
Repair Kit/Tool-Weapon
Use-life
Figure 105. Ecological and behavioral model with technological component.
258
The Goals of Tool Design
Bleed (1986) points out that modern engineers design tools with one objective in
mind: efficiency. Tool efficiency means a combination of four different things:
increased effectiveness (a sharper knife); increased use-life (a longer lasting
knife); a quicker production time (the knife costs less in energy and material);
increased production volume (more knives per unit of material). These qualities
reflect the manufacturer's view. From the user's viewpoint, the second quality is of
major interest: how long does the tool stay in working order? To make an effective
tool, therefore, modern engineers strike a compromise between these four qualities.
The exact mix depends on their decision either to stress the first two qualities and
make a reliable tool, or to stress the second two qualities and make a maintainable tool.
These decisions do not produce two mutually exclusive types (Torrence 1989: 63), as
both reliability and maintainability can be designed into the same tool.
A reliable tool lives up to its name and functions when needed. It will have these
salient features: move invested (over-design) in its critical parts, eg. the knife edge;
extra sturdy construction; quality fitted parts;
spare parts;
repair kit, including
raw materials. One typical example of a hunter-gatherer's reliable tool would be any
hunting weapon used to kill migratory animals. If the weapon failed, this could result
in partial or complete loss of food with dire consequences. Thus reliability in a
weapon aims to buffer the hunter against the severity of the risk that he faces (ibid:
63).
A maintainable tool can be made to work in part, even if broken, and it can be used
for other tasks than the one for which it was designed. It is light and portable, has a
modular design, and it comes with a specialized repair kit. Also, it can be repaired
quickly and easily by the user in the middle of a job. A good example of a huntergatherer's maintainable tool is the digging stick. If it broke and simply could not be
259
resharpened, then a new one could be made right away or soon afterward, and used on
the same roots or tubers being excavated with the stick that broke. By contrast,
maintainability is designed into a forager's weapon not so much to cope with the
severity of the hunter's risk, but rather to buffer him against the erratic timing of
the risks he faces (ibid: 63).
Tool Desjgn Decisions Among Foragers and Collectors
At first glance, it appears that reliable tools would be more desirable to huntergatherers in all circumstances. However, reliable tools carry hidden costs: they are
over designed, need extra care and require special spares. Some may be bulky or
difficult to transport. Reliable tools really pay off for collectors dealing with clumped
resources, long travel schedules, and little spare time. They pay off because they can
be repaired at predictable times and cached (passive tools) for future use so they do
not have to be schlepped with the equipment on duty (active tools). It is intuitively
reasonable to assume that collectors will have very reliable weapons (Bleed 1986:
744-745). Good examples are the Nunamiut (Binford 1979: 268) who make
elaborate tool preparations before long hunting trips (gearing up sessions), and the
lngalik whose elaborate technology focuses severely on reliable tools (Osgood 1940).
Thus collectors ten to increase their overall investment in technology and also in the
diversity and specialized nature of their tools (Torrence 1989: 60-61 ).
By contrast, foragers who are not under stress tend to invest in artifact designs
without complex specialized tools. They are dependent on evenly distributed and more
stable food supplies, and are more inclined to use maintainable implements and
weapons (extractive tools) because tool failure is not so costly in terms of missed
opportunities. There are no intense bursts of tool making because time is available
260
almost every day (make and mend sessions) for maintaining and replacing tools
(Silberbauer 1981: 243).
As most hunter-gatherers use a mix of both strategies, their tool kits should
reflect a comparable mix .of reliable and maintainable tools. Typical recent examples
I
are the !Kung (Lee 1979: 128-144) and the G/wi (Silberbauer 1981: 206-209)
who use both. It should be reiterated, however, that these are not discrete tool types.
For example Torrence (1989: 63) suggests that all weapons, whether made by
collectors or foragers, are basically maintainable, but the design of a collector's
weapon will incorporate a higher degree of reliability whereas that of a forager will
have less.
Kinds of Curation used by Foragers and Collectors
Bleed's (1986) reliable and maintainable tool design strategy can be merged with
Binford's (1973, 1979) expedientlcurated tool dichotomy.
Curation, as a term in
archaeological literature, has many different meanings (see below), but these two
classifications seem to merge if we accept that the use of uncurated (expedient) tools
is also a design goal. Curated tools could have either maintainable or reliable
objectives in mind. Figure 106 suggests how a hypothetical tool's design might be
fitted into a tri-polar diagram to reflect the appropriate mix of all three goals.
The term curation has come to mean either: (1) the tool was made long before it
was used; (2) the tool was carried around and used for a long time; (3) the tool was
regularly maintained during its life; (4) the tool was designed for many uses or; (5)
the tool was reshaped for other uses (Bamforth 1986). Although there is some
overlap, each is more usefully viewed as a distinct behavior set, driven by different
needs. The first three variants deserve closer scrutiny because they articulate well
with the collector/forager dichotomy explored in the previous chapter. The other two
261
(multi-use and recycling) are more usefully linked to raw material distribution
(Bamforth
1986).
reliable
tool's design goals
maintainable
expedient
Figure 106. Tri-polar plot of a hypothetical tool's design goals plotted against three
design qualities.
However, before turning to a discussion of curation, it is beneficial to look at the
flip-side of curation: discard. In general tools may be discarded because of breakage
during manufacture, expedient tools by definition are thrown away immediately after
use, tools may break during use, tools may be lost during use or transport, tools may
be cached but never retrieved, tools may be replaced if they appear to be near or at
risk of failure, or tools may be discarded when exhausted and no longer maintainable
(Kuhn 1989; Shott 1989; Tomka 1990).
Reason for discard is inextricably linked
with tool curation, and the two must be considered together.
(1) Tools Made Long Before Use: Foragers versus Collectors
Production in advance of use is promoted by what Torrence (1983: 11-13) calls
"time stress" caused by scheduling conflicts. A typical scenario is the sudden arrival
262
(and departure) of a migratory herd. There is no time to make more weapons, so they
must be ready beforehand. Clearly, time stress promotes this kind of curation, which
is practised vigorously by collectors, who are often faced with such situations.
How long it takes to make the tool may also determine how much time elapses
between its manufacture and use.
Yellen (1977: 76) and Lee (1979: 274-275)
show that some !Kung tools take so long to finish that they are carried around in an
unfinished state from site to site. This is quite different from time stress, a feature
virtually unknown in !Kung daily life (Lee 1979: Tables 9.11 and 9.12) nor indeed in
the lives of the rather more stressed G/wi (Silberbauer 1981: 243).
To generalize from these few examples:
Collectors in higher-risk settings are
more likely to carry around (or cache) finished, unused tools as a precaution;
foragers in lower-risk habitats are likely to carry around unfinished tools that
reflect little or no time stress. Discard due to this form of curation would be those
tools broken during manufacture or those lost during transport.
(2) Tools with Long Use-lives: Foragers and Collectors
The best record of tool use-life for a forager band is that from the !Kung, whose
shortest recorded tool use-life is five days and the longest is 16 years. It is not clear
whether Lee (1979: 274-275) recorded details of shorter use-lives of real
expediency tools. Roughly comparable data for a collector group come from the lngalik
that suggest shorter, not longer tool use-lives (Osgood 1940). Although the sample
sizes for both data sets are limited and recording biases may have added more
distortions, these are among the best published data available (Figure 107). lngalik
tools had an average use-life of 2.6±7.9 years while !Kung tools lasted on average
3. 7±3.4 years.
Taken at face value, then, curation (meaning longer use-life, sensu Shott 1989)
is more intensive for foragers than collectors and not the other way around, as
263
predicted for other kinds of curation. One possible explanation is that collectors must
keep their weapons in good working order at all times, and more frequently replace
parts or even complete tools before they wear out (Kuhn 1989). Discard could occur
because an artifact was lost or because of use-breaks, replacement before failure, or
artifact exhaustion.
Also, it may be that the !Kung were in a position to use more durable, less brittle
raw materials (eg. iron wire for arrow points) than the lngalik, but ethnographic data
(Lee 1979: Table 9.10, Osgood 1940) also includes skin artifacts that stay in use for
longer periods of time than seemingly more durable materials like wood or stone
(Tables 29 and 30). This is because the less durable artifacts are not being used as
stressfully as many the more durable ones. Clearly, the intensity of use, not just the
length of use must be considered also (Silberbauer 1981: 223-232).
25
20
15
Number of
Tools
10
5
0
2
•
3
4
5
6
7 8 9 10111213141516
Use-Life (years)
lngalik Tools&Facilities
IIIII
!Kung Tools&Facilities
Figure 107. Histogram of Dobe !Kung and lngalik artifact and facility active use-lives
in years (data from Lee: 1979: Table 9.10 and Osgood 1940).
264
Table 29.--Average use-life, and production and maintenance costs of !Kung artifacts
by material (data from Lee 1979: Table 9.1 0)
Artifact
Materials
Average
Manufacturing
Time (min.)
Ostrich Eggshell
& Tortoise Shell
Skin
(minus oracle disks)
Wood
Metal
Other Plant Material
Number of
Average Total
Maintenance Maintenance
Time (min.)
Episodes
Total
Average
Costs
Use-Life
(min./day)
(day)
180
397
24
33
365
1069
0.75
0.96
730
1528
359
600
600
64
462
12
1084
8303
730
1.80
4.44
3.64
802
2007
365
Table 30.--Average use-life, and production costs of hand-held lngalik artifacts by
material (data from Osgood 1940)
Artifact
Materials
Bone
Stone
Wood
Bark
Grass
Skin
Average
Manufacturing
Time (day)
1 .3±0. 7
3.4±1.0
0.9±1.1
1 .9±2.2
1
3.3±4.6
Average
Use-Life
(day)
2.3±2.2
1 .4±1 . 1
1 .4±2.2
1 .9±1.5
1
2.4±1.3
N
7
5
16
4
1
4
Production Costs and Tool Use-life
Shott (1989) has argued that among foragers, use-life also increases with the
effort and time invested in the tool's manufacture. He demonstrates this relationship
with Lee's (1979) Dobe !Kung data, but he mixes multi-tool production costs with
use-life measurements for single tools, which spoils the comparison for ostrich
eggshell water containers. When this is corrected (Figure 108) the linear regression
(r2 = 0.47, p value = 0.005) suggests a moderately weak, although significant
relationship between production time and use-life. Steel and flint fire kits, with very
265
short use-lives, were omitted from the analysis on the grounds that they are not part
of the indigenous kit. These are discussed later.
Among the lngalik collectors, Shott (1989: 22) argues that there is no such
relationship between tool production time and use-life, but he inexplicably omits 45
tools and facilities from his analysis.
Furthermore, if a tool is periodically retired
from use and cached, its time in cache as a passive tool should not be counted towards
use-life. Only the active portion of a tool's use-life should be considered. Osgood's
(1940) original lngalik data lists the season(s) of use and, for many tools, the total
number of seasons used. When active use-life is plotted (Figure 109) there is, after
all, a weak linear regression (r2 = 0.248, p value = 0.0001 ), implying a weak
correspondence between production time and active use-life. This could be
strengthened if two tools (bone wood scrapers and work boards) that have short
production times and long use-lives were omitted from the analysis. Another item of
this kind (ceremonial hats) with extremely long use-lives, was omitted from the
analysis on the grounds that is not an adaptive part of the food quest. On the other
hand, the correlation would be further strengthened if sleds, which have enormous
production times invested in them, were included in the comparison.
When the !Kung and lngalik data and regressions are combined on the same chart
(Figure 11 0) it becomes clear that marked differences occur between the two. To
generalize from this very limited sample, foragers and collectors may have different
attitudes towards production time and use-life: foragers get much more use-life for
their production effort, collectors do not appear to be terribly efficient overall.
266
9
0
0
8
0
7
セ@
6
d>
(/)
2.£
5
4
3
2
0
2.5
4
3.5
3
4.5
5
5.5
In (Manufacturing Time)
6
6.5
7
Figure 108. Linear regression between natural logarithm of use-life and
manufacturing time for Dobe !Kung artifacts minus European clothing, r2 =0.47 (data
from Lee 1979: Table 9.10).
12
0
10
0
セ@
ro
Q)
.b
セ@
8
6
d>
(/)
:::>
Q)
Mセ@
t>
<(
4
2
0
-2
-5
0
5
15
20
10
Manufacture Time (days)
25
30
35
Figure 109. Linear regressions of lngalik tool manufacturing time and adjusted uselife (data from Osgood 1940).
267
.6 !Kung, ,
0 lngalik, .
QVKMセN⦅@
14
12
10
.
0
8
6
4
2
PセMK@
MRKセイ@
-5
0
5
10
15
20
Manufacture Time (days)
25
30
35
Figure 11 0. Linear regressions between manufacturing time and use-life for Dobe
!Kung and lngalik tools (data from Lee 1979 and Osgood 1940).
(3) Regularly Maintained Tools: Foragers and Collectors
The amount of time invested in a tool through repair and maintenance may also
extend its use-life (Shott 1989). Unfortunately there are no data for the lngalik, but
the !Kung data (minus the European manufactured items) show a positive correlation
(r2 = 0.313) between total maintenance time and use-life (Figure 111 ). One of the
omitted European items is of special interest here: the flint-and-steel fire kit is
maintained almost every day, not because it wears out rapidly but rather out of fear
that it might fail. This anxiety determines that it is kept in working order at all times
(Kuhn 1989).
It is an anxiety very similar to that which drives collectors to replace
weapons more frequently, to assure fail-safe use at short notice at all times.
To generalize, again from very limited data, foragers in low-risk habitats may
increase the maintenance of a tool to save the bother of making a new one. Collectors
in high-risk settings may increase production/maintenance of a tool by replacing a
part or the entire tool to avoid failure at a critical moment.
268
11
10
0
0
9
8
セ@
CIS
(J)
セ@
.g1
::JI
(J)
(/)
:::>
7
6
5
4
3
2
0
0
0
0
0
0
0
0
.5
1
1 .5
2
Total Maintenance Time (days)
2.5
3
Figure 111. Linear regression between total maintenance time and use-life for Dobe
!Kung artifacts minus European clothing and flint-and-steel fire kit, r2 =0.313 (data
from Lee 1979: Table 9.1 0).
Maintenance versus Production Costs
If collectors are more likely to use a reliable technology and foragers a
maintainable technology (Bleed 1986) then it follows that combined production and
maintenance costs for foragers should be equal to or possibly greater than production
costs alone for collectors. If this is so, then combined !Kung production/repair time
should be equal or greater than lngalik production time alone. This is not the case
(Figure 112, Table 31) because lngalik production costs are much higher than
combined !Kung costs. What is actually happening is that !Kung foragers keep
production and maintenance (and transportation) costs down by making their tools last
as long as possible and by discarding them only when completely worn out. lngalik
collectors are investing very heavily in the production of tools which they discard long
before they are worn out (Kuhn 1989). Note that this result contradicts the
expectations of Binford's (1973, 1977) original curation model.
269
Of course neither pattern is exclusive to foragers or collectors who both use a
mix of the two patterns in ratios that seem to covary with the foraging/collecting mix
in the individual bands mobility pattern. Thus the greater the emphasis on collecting
and logistic mobility, the higher the likelihood that investment in tool production will
increase and use-life will decrease, and vice versa.
Table 31.--Linear regression statistics for !Kung tool use-life by manufacture costs
plus repair costs, and lngalik tool use-life by manufacture costs
Group
!Kung
Slope
1.436
Intercept
1.843
r2
0.352
p value
0.0022
lngalik
0.174
0.978
0.248
0.0001
12
+ lngalik
•!Kung
10
8
Use-life
(years)
6
4
2
0
-2
-5
0
5
10
15
20
25
30
!Kung Manufacture & Repair, lngalik Manufacture (days)
Figure 112. Linear regressions between total manufacturing and repair time by uselife for the !Kung tools and manufacturing time by use-life for lngalik tools.
Ceremonial hats and sleds were not included with the lngalik data set. Flint and steel
fire kits were omitted from the !Kung data. !Kung r2 =0.352, lngalik r2 = 0.248
(data from Lee 1979: Table 9.1 0, and Osgood 1940).
35
270
The Use-Life of Repair Kjts versus Tools and Weapons
Bleed's (1986) original dichotomy between maintainable and reliable tools was
formulated for weapons only, not for the tools used to make and repair them.
However, repair kits (maintenance tools) may not have been used and maintained in
the same ways as implements and weapons (extractive tools). Also, foragers may not
imbue them with the same relative worth as collectors do. This will be directly
reflected in the use-life of repair kits versus implements and weapons. As usual, the
available data are limited and there are some distortions. The lngalik used more stone
and bone, while the !Kung use more metal, probably inflating the use-lives of the
latter's equipment in both classes. Nevertheless, Table 32 suggests that the hardpressed collectors may take greater care of their repair kits and keep them longer,
while the understressed foragers are more cavalier with their repair kits (including
a greater use of expedient maintenance tools) but try to get their tools and weapons to
last longer. A small number of ethnoarchaeological studies have documented the
existence of expedient stone tool use (Binford 1986; Gould 1977; Gould et al 1971;
Hayden 1979;
Miller 1979;
Sillitoe 1982;
Strathern 1969;
White 1968;
White
and Thomas 1972) including two cases from southern Africa (MacCalman and
Grobbelaar 1965; Stow 1905: 66). In all cases the expedient tools are repair kit
(maintenance) tools, and in each case lithic raw materials are readily available.
Table 32.--Extractive and maintenance tool mean use-life (years) for foragers
(!Kung) and collectors (lngalik). M/E Ratio equals maintenance tool uselife/extractive tool use-life
Foragers
(!Kung)
Collectors
(lngalik)
Extractive Tool Use-life Maintenance Tool Use-life
4.2±3.2
(n=7)
3.3±2.5
(n=18)
1.4±0.3
(n=3)
1.7±2.0
(n=30)
M/E Ratio
0.79
1 .21
Raw Materials and (4) Multi-Purpose or (5) Recycled Tools
271
For both foragers and collectors it is axiomatic that a scarcity of raw materials to
make tool replacements will promote longer use-lives and repeated refurbishing. But
scarcity will also promote multiple uses of the same tool, and even the recycling of
broken parts or discarded tools (Bamforth 1986). Also, bits of raw material will be
carried around from site to site (Kelly 1988; Parry and Kelly 1985). This applies
with equal force to collectors or foragers so that a single band, of whatever mobility
mix, will start practicing all the above saving behaviors each time they move into a
portion of their range where the raw material in question is scarce or absent, or
access to it may be unpredictable. As soon as they have ready access to the raw
material, they will abandon these practices with alacrity.
Raw materials like stone (lithic resources) are not all acquired in the same way
(procurement strategies). The hunter-gatherer may simply fetch some from an
outcrop (direct procurement) as he/she passes by in the course of the seasonal or
daily round (embedded strategy). Bands with very high mobility rates, particularly
those with frequent camp moves as practised by foragers, will seldom if ever organize
a long trip specifically to gather stone from a quarry. Although Gould and Saggars
(1985) assert that this is not so, they do not develop their argument. Perhaps
logistically organized bands, already inured to long expeditions, would be more readily
inclined to make raw material trips. In this scenario, the distance that raw materials
are transported between quarry and camp where the tool was discarded may not be a
direct reflection of stone material scarcity
(Binford 1979).
Another obvious way to acquire raw materials is through exchange (indirect
procurement) where the hunter-gatherer gets it from somebody else who did the
fetching (McAnany 1988). There is plenty of ethnographic evidence for this among
the Kalahari San (see discussion on hxaro in Chapter XI and Lee 1979: 365-366;
Marshall 1976: 303-311;
Silberbauer 1981: 239-242;
272
Wiessner 1977, 1982)
and many other groups. Other ways (indirect procurement) to acquire raw material
include fetching it from a previously stored cache, or scavenging it from an old camp
or other surface site.
Global Trends in Hunter-Gatherer Technology
The foregoing analyses suggest the presence of several interlocking trends in
hunter-gatherer technology, all aimed at buffering risk.
It follows that these same
trends must articulate with the mobility pattern mix, which also aims to reduce risk.
In the left column of Figure 113 an array of five hypothetical bands (A through E) are
listed. The next column shows that they are arranged on a sliding scale based on their
mobility pattern mix. As the number of annual camp-moves increases from bottom to
top of the second column, so the annual number of long task-specific trips decreases.
Each band is allowed a portion of the sliding scale, to signify that its mobility pattern
mix is not rigidly fixed at a constant ratio.
By scanning the tops of the adjacent columns it emerges that the relatively
stress-free forager bands A and B will make a smaller range of highly maintainable,
general-use tools and weapons with longer use-lives. On average, they will get more
use from a tool for the time and effort they invest in making it, even though they take
their time to finish it. Thus they tend to carry around unfinished tools, and refurbish
their completed tools frequently, to save the bother of making new ones. Consequently,
their repair kits (maintenance tools) wear out quite rapidly and are often replaced.
They seldom cache new, finished tools. Whenever and wherever raw materials become
scarce, the tool(s) made of that material will be refurbished even more often. It will
be used for various jobs and it may be recycled when broken.
ASSEMBLAGE COMPOSITION
....
I
26
セ@
MOBILITY.
MIX
I
AI I
FORAGER
MOBILI1Y
I REPAIR
TOOLS & WEAPONS
Special/
General
GENERALIZED
(Mointoinable)
TOOLS &
WEAPONS
II
DESIGN
GOALS
I
ARTIFACT APPEARANCE
I
Use-life Mix
II
LONGER
USE-LIFE
I
Repair/
Replace
Production
Time
KIT
TOOLS &
WEAPONS
Use·life Mix
Active/
Passive
II
ALL
Timing
MAKE&
MEND
MORE
REPAIRS
(Maintenance)
REPAIR/
REPLACE
'PRODUCTION I REPAIR KIT
TIME
....
SETTLEMENT PATTERN
WHEN
USED
WHEN
MAINTAINED
NO. OF MICROSITE NO. OF
HALOS
TASKCAMPS IN
AROUND
SPECIFIC
RANGE
CAMP
SITES
I\
MORE
CAMP
MOVES
CAMPS
I\
MORE
DIURNAL
FORAGING
I\
FEWER
LONG TRIPS
MICROSITES )SPECIAL TASK
SITES
I\)
-....!
Figure 113. Assemblage composition and settlement pattern model for hypothetical bands.
w
274
When the base of Figure 113 is scanned it is apparent that they time-stressed
collector bands D and E will make a greater variety of highly reliable, specialized
tools and weapons with shorter use-lives. On average, they will get less use from a
weapon for the time and effort they invest in making it, even though they finish it
during limited, intense bouts of tool making. They seldom carry around unfinished
tools and weapons, and replace completed ones frequently, rather than trying to
maintain old ones. Consequently, their repair kits get light use and lasts longer. Not
to overburden themselves, they tend to cache new, finished tools. When or where raw
materials become scarce, the tool(s) made of that material will be refurbished rather
than replaced. It may acquire various uses and it may even be recycled when broken.
Archaeological Implications
Different kinds of material culture fallout can be implied from groups of the
above trends; there are trends affecting artifact appearance, trends that affect overall
assemblage composition, and trends affecting between-assemblage comparisons. Also,
the mobility pattern will directly affect the distribution of assemblages (settlement
pattern) of the band.
Artifact Appearance
The appearance of a new tool or weapon is dictated by a number of factors; first
by the physical properties of the raw material, then by the limits of it's makers
technological know-how, then by his or her personal experience or skill, and finally
by a set of culturally dictated values (style) to which the maker subscribes.
However, the appearance of a tool or weapon by the time it enters the archaeological
record is dictated by still more factors. Figure 113 regroups trends that dictate tool
appearance. By scanning the tops of the adjacent columns, it emerges that the
appearance of a specific tool or weapon made by members of the relatively stress-free
275
forager bands A and B will have generalized design. Furthermore, most specimens
will show signs of greater wear and refurbishment, likely resulting in alternations of
shape. Tools from the repair kit may show less wear and tear before discard. There
will also be some unfinished specimens that show no signs of maintenance.
At the other end of the scale, tools or weapons made by members of the timestressed collector bands E and F are likely to betray a specialized design, and show
fewer signs of wear or repair. Tools of the repair kit will be over-designed also, and
tended to show signs of prolonged use before discard. Unfinished specimens, not
broken during manufacture, will be very rare unless found in caches.
Assemblage Composition
The cluster of trends that dictate the range of tool types in the five bands is shown
in Figure 113. The combined archaeological residues of forager bands A and B will
yield relatively few, generalized types, mostly heavily used and repaired, plus some
unfinished specimens of these types. Lightly used (expedient) repair kits will be
quite common.
Combined artifact residues from collector assemblages produced by band E and F
will contain a greater variety of clearly defined, finished types will little use-wear.
Repair kit types will tend to be worn down (maintainable). Unfinished specimens will
be rare.
Inter-Assemblage Comparisons
An assemblage recovered from one camp site within the range of forager band A
will look pretty much like any other from the same range/territory. Only the size of
the assemblage is likely to vary, depending on the popularity of the camping place.
Very small assemblages with incomplete inventories may occur at task specific
foraging areas. These will contain some of the same tools found in camp.
276
Although assemblages from collector camps like those in the range of band E will
be quite similar to one another, the inventory of types will be incomplete. There will
be rare, unused caches with very restricted inventories, possibly of types that occur
only rarely in camp. There will be assemblages that reflect gearing up activities, in
which fragments of replaced types occur. Some of the cached types will be found in
broken or worn condition also concentrated at distant, task-specific areas that were
repeatedly visited.
Distribution of Assemblages (Settlement Pattern)
Typically, the forager mobility pattern (Chapter XI) produces a residual pattern
with numerous small/medium camp sites distributed widely across the range. There
is a halo of small, task -specific foraging sites/areas within about 1Okm radius
around each camp site. The reflect the accumulated residues of diurnal foraging from
that camp. Specifically absent are caches and large, task-specific sites/areas remote
from other camp sites.
Extreme collector mobility patterns generate a site distribution pattern with
relatively few, large camp sites, without any 1Okm radius halos of micro-sites.
Remote from these camps are several large sites/areas that denote intensive collecting
tasks, such as frequently visited kill sites. Cache sites are present.
Again, these two extreme examples are at opposite ends of the forager-collector
sliding scale. Most bands use a judicious mix of the two mobility patterns, and this
means that both settlement patterns will be overlaid and mixed on the map of a single
band's range. Any bias in favor of one mobility pattern, will be reflected in the band's
site distribution map, however.
Application of the Technology System to the Blydefontein Region
277
As was shown at the close of Chapter XI, a diagram like that in the Figure 113 can
be converted so that it becomes applicable to regional data. In Figure 113 the layout is
designed vertically to cover five, widely separated bands ranked from extreme
collector mobility (bottom) to extreme forager mobility (top).
It is converted for
regional use by redesigning the vertical layout to represent the long term history of a
set of adjacent bands in one region. In this layout, older is at the bottom and younger
is at the top. In this scenario, a band cluster has experienced the whole gamut of
global mobility changes. Although this is unlikely in any single site or even region, it
is a useful device for modeling changes in the archaeological fallout resulting from
such changes in the mobility mix of local hunter-gatherer residents.
Two portions of the diagram will be most useful for modelling purposes in the
Blydefontein area during the late Pleistocene, Holocene and to recent times. The late
Pleistocene roughly spans band F, and the remainder spans the upper part of band C up
to the lower part of band A. This model predicts the appropriate risk-reducing
technology used in the region under various habitat conditions.
CHAPTER XIII
BLYDEFONTEIN BASIN: A TEST OF THE NEW MODEL
Introduction
A first and partial test of the new model is attempted in this penultimate chapter.
Although a full and comprehensive test must await new fieldwork on the huge Zeekoe
Valley data base, some preliminary tests are possible now, even though they must be
designed from the perspective of a single site. First, Blydefontein's habitat is fixed at
appropriate points on the global hunter-gatherer ecological framework summarized
in Figure 103.
Blydefontein's Place in the Ecological Model
The estimated percentage of hunted resources in the Blydefontein basin can be
inferred from the local effective temperature and rainfall figures (Chapter XI).
(1962-1982) mean ET at Blydefontein is 14.2.
The
By fitting this value to the
curvilinear regression in Figure 92, we arrive at an estimate of 31 percent of hunted
resources for the area (Percent of hunted resources = 214.481 - 21.196x +
0.584x2). Given Blydefontein's position on this curve, it would take a dramatic
temperature increase (at least 4-5 ETs) before the percentage of hunted resources
would start to climb. On the other hand, even slight reductions in mean ET would
significantly increase the percentage of hunted resources. If temperatures during the
Last Glacial Maximum fell as much as 5°C (Vogel and Talma in Deacon and Lancaster
1988: 143) and the seasonal range remained the same as today, then the mean ET
would be approximately 12.2 and the estimated hunted resources for Blydefontein
equal 43 percent (Figure 114).
278
279
0 Modern Hunter-gatherers
90
Mv'lodern
Ill Pleistocene
0
80
70
Percent
60
of
Hunted 50
Resources
40
0
0
0
30
20
00
0
0
10
0
0
8
10
14
12
18
20
16
Effective Temperature
22
24
26
Figure 114. Modern and Pleistocene estimates of percent of hunted resources in diet
at Blydefontein, based on estimated ET and curvilinear regression from Figure 92.
0
&.
Modern Hunter-gatherers
XPKMセN⦅@
70
Modern Blydefontein
0
Percent
60
of
Hunted
50
Resources
40
0
0
0
0
30
CD
0
20
0
10
PKMイセ@
0
500
1000
1500
2000
Rainfall (mm)
2500
3000
Figure 115. Modern estimate of hunted resources in diet at Blydefontein, base on
modern rainfall and curvilinear regression derived from Figure 94.
3500
280
Today's mean annual rainfall for Blydefontein is 366mm. When this figure is
fitted to the regression curve in Figure 94, the estimated percent of hunted resources
is 30.2. This is in close agreement with the estimate based on effective temperature
given above. No reasonable estimates for Last Glacial Maximum rainfall is available
so a plotting of a Pleistocene estimate is not possible as in Figure 114. Blydefontein's
position on the curve shows that any slight increase or decrease in rainfall would
significantly increase or decrease the percent of hunted resources (see Figure 115).
Because Blydefontein is on the edge of the semi-desert Karoo shrubland and grassveld,
more rain would increase the amount of grass cover, and this would certainly lead to
increased densities of larger herbivores (see Chapter XI).
Changes in the Animal/Plant Food Mix in the Blydefontein Habitat
Temperature estimates were not used to estimate changes in the animal/plant food
mix because their estimated variation was extremely low, which agrees well with the
oxygen isotope temperature estimates from Cango Caves (Vogel and Talma in Deacon
and Lancaster 1988: 144). However, the pollen-derived rainfall estimates (see
Chapter VII) can be used to estimate the percent of hunted resources in the
Blydefontein area. Because of possibly biases in alluvial and buried soil pollen
sequences, only those estimates from Oppermanskop Midden, USP and BSM were used.
Parenthetically, it is worth noting that the Southern, /Xam, and Mountain Bushmen
also gave greater significance to rain (!khwa) in their mythologies and apparently in
their paintings (Bieek 1933;
Bleek and Lloyd 1912;
Lewis-Williams 1981;
Schapera 1930; Vinnicombe 1976). These estimates are only intended to suggest
temporal patterns and should not be considered as exact estimates. Also one should
remember that the pollen-derived rainfall estimates, at best, show general
precipitation trends (averages) compared with the highly variable historic rainfall
281
records (see Chapter VII}, thus these estimated fluctuations in hunted foods are
intended to reflect long-term (ca. 75-100 years or more} averages. Also, as shown
in Chapter XI, the large herbivore biomass fluctuates as rainfall varies, and this
provides the added animals that allow increased hunting. All the estimates of amount of
hunted food fall between 29-36 percent of the diet.
36
•
35
34
Estimated
Percent of 33
Hunted
Foods in
32
Diet
31
30 •
セNヲャL@
I
•
•
¥.
u
0
RYKMセ@
0
1 000 2000
3000
4000 5000
6000 7000
8000
Years BP
1·•-
Oppermanskop ·0- USP
·•- BSM
Figure 116. Changes in the amount of hunted resources in the Blydefontein diet, based
on estimated changes in annual rainfall (see Figure 52}.
As almost no plant food remains survived in neither the Blydefontein Rockshelter
matrix (two Diospyros sp. seeds), nor the Meerkat Rockshelter deposits, a direct test
of this portion of the model is not possible. Hunted resources are accessible through
faunal analysis, but these data are not yet finalized (K. Cruze-Uribe, personal
communication). The faunal analysis from Sampson's test excavation (Klein 1979:
Table 3} indicates the range of hunted resources throughout the Interior Wilton
occupations at Blydefontein Rockshelter. However, the sample is too small and the
range is too large (about 16 different species) so that most cells in the table yield
282
MIND counts of 1-3 only. Consequently it is impossible to detect any significant
changes in the frequency of species through time.
Ignoring for the moment the various shortcomings of MIND counts, Sampson's
combined sample yields these estimates: 11 hares (Lepus spp.), two baboons (Papio
ursinus), a fox ( Vu/pes chama), four yellow mongoose ( Cynictis penicillata), two
wildcat (Felis libyca), two rooikatlcaracal (Felis spp.), a leopard (Panthera
pardus), 17 hyrax (Procavia capensis), seven quagga (Equus ct. quagga), eight
mountain reedbuck (Redunca flulvorufula), ten vaalribbok (Pelea capreolus), seven
black wildebeest or red hartebeest (Connochaetes gnou/Aicelaphus buselaphus), three
klipspringer ( Oreotragus oreotragus), three steenbok (Raphicerus campestris) and
two Cape buffalo (Syncerus caffef). No doubt changes in the frequencies of some
species will become evident when the larger samples from my own excavations has
been completed.
This is by no means the full range of meat in the hunter-gatherer diet, however.
Large amounts of foraged meat is present in the Blydefontein sample, including a
monitor lizard (Varanus sp.), abundant tortoises, amphibians, and crabs. Although
micromammals were also present (Avery 1988: 341-342) it remains uncertain
what proportion of these (if any) were acquired by humans. Barn owl pellets are
assumed to be the major contributor of microfauna to the rockshelter matrix.
Additional protein was obtained from ostrich eggs, and there are a few freshwater
mussels also. There are no known fish bones in the sample and this is not surprising
considering the small size of the local streams and the relatively high position of the
rockshelters in the Oorlogspoort basin. This suggests that fishing played little or no
part in the Blydefontein subsistence pattern, and any distortions to the model (see
Figure 97) can be safely discounted, therefore. It seems likely that fluctuations in the
frequencies of these foraged species can be expected, once the analysis is completed. A
283
partial test of this portion of the model may eventually become possible by comparing
the ratio of hunted to foraged meat in the diet at different layers in the two
rockshelters.
Blydefontejn's Range and Late Holocene Population Density
The (untested) assumption here is that Blydefontein is in the heartland of one
band's range, and is not in a zone of territorial overlap between two or more bands. If
so, then the ecological model predicts the relative changes in range size of huntergatherers who used both rockshelters. Range size fluctuations in accordance with the
estimated amount of hunted resources (ultimately derived from the palynological
rainfall estimates) and estimated changes in primary plant production (also
originating from the pollen derived rainfall estimates using Rosenzweig's (1968)
formula and estimates of primary production where ln{primary production) =
ln(rainfall)*1.09 - 0.94, r2 = 0.701, p value = 0.0001 ).
The range sizes
estimated from amount of hunted resources yields lower estimates for the drier
periods because of presumed increased reliance on plant foods while the range size
estimates derived from the primary production estimates yield larger range estimates
for the dry periods because of reduced carrying capacities (Figures 117 and 118).
These two separate estimates roughly overlap with each other, and they are well
within the limits of known San range sizes (Figure 119). However, these estimates
do not consider range constriction due to growing populations in the late Holocene. As I
demonstrated in Chapter XI this is influential in controlling range size as well.
284
750
•
700
Estimated
HunterGatherer
Range Size
(km2)
650
600
0
Jri?
550
Ill
•...,
0
500
..
.:I.
Ill\ Ill
セ@
Ill
'\. ....
Ill" Ill
セᄋ@
TUPKMセ@
0
1·•-
1000 2000 3000 4000 5000 6000 7000 8000
Years BP
·Ill- BSM
Oppermanskop ·0- USP
Figure 117. Hunter-gatherer range size estimated from amount of hunted resources
in diet.
640
620
0
600
580
Estimated
HunterGatherer
Range Size
(km2)
Qセ@
560
540
セ@
0
52 0
500
1u •
Q
ᄋセQ|@
カNセLj@
• •
I.,/."'
セ@
·'
•
•
480
460
440
Ill
TRPKMセ[@
0
1·•-
1000 2000 3000 4000 5000 6000 7000 8000
Years BP
BSM & CH2
·0- USP
·Ill- Oppermanskop
I
Figure 118. Changes in range size predicted by the ecological model for huntergatherers who incorporated Blydefontein within their subsistence round.
285
5
4
Count
3
2
0
0
500
1000
1500
2000
2500
3000
3500
4000
4500
Range Size (Km2)
Figure 119. Histogram of recorded San ranges (see Figure 98).
Independent proxy evidence for population increase comes from the adjacent
Zeekoe Valley where sub-recent (Smithfield) surface sites greatly outnumber later
Holocene (Interior Wilton) surface sites (Sampson 1985).
Finer time resolution for
overall population increase during the last millennium is available for the upper
Zeekoe Valley. There, sites with later ceramics (rocker-stamped wares) greatly
outnumber those with the earlier (non-rocker) decorations (Sampson 1988;
Sampson et al. 1989). Additionally the rarity of crescent dominated backed tool
assemblages throughout the region points to low population densities in the early and
mid-Holocene as originally suggested by J. Deacon (1974) using radiocarbon dates.
As population rises to meet full carrying capacity, so territories shrink, with overall
reductions in hunter-gatherer mobility. Thus the fluctuations in Figure 117 should
be welded to a longer term trend towards smaller ranges during the later Holocene, in
response to climbing population numbers. At present I am unable to provide a detailed
model of these changes, however it is suggested that changes in carrying capacity as
286
well as constrictions forced by by growing populations would be the most significant
determinants of range size.
Test Implications
A sound test of this portion of the model may be possible with a spatial data base as
large as that for the adjacent Zeekoe Valley (Sampson 1985), and any test derived
from one or two sites is proportionally less sound. However, there are some test
implications that can be built from local circumstances. Most of the stone used to
make artifacts at Blydefontein is hornfels, but there are also rare specimens made of
agate and jasper. These must have been obtained from the Orange River gravels at
least 65km to the north. It is well known that agate or jasper tools decrease in
frequency with distance from the Orange River (Humphreys 1972: 48-51 ; Sampson
1970: 97). While it is widely believed that adherence to a microlithic technology
dictated the demand for agate and jasper, this is without foundation as all types of
backed microliths, including all of the bifacial pressure-flaked tanged-and-barbed
points, and pressure-flaked bladelets recovered from Blydefontein and Meerkat
rockshelters are made of hornfels. It is far more likely that Orange River pebbles
were acquired through contacts with local residents and, more significantly here, by
personal visits to the Orange River valley.
Dunn {1931) observed hunter-gatherers
in Bushmanland, some 400km to the northwest, transporting stone as a matter of
routine, and it is entirely possibly that this occurred in earlier times as well.
The implication is that the amount of agate and jasper in Blydefontein artifact
assemblages is a direct reflection of the size of the inhabitant's range and overall
mobility. At times when groups with small territories occupied the shelters, they did
not have direct access to the Orange River gravels and even their indirect access would
be reduced, consequently the amount of these rock types in the lithic debris would
287
decrease. An increase in range size would bring them periodically closer to the
gravels, and the amount of agate-jasper in the total assemblage would rise
proportionally. This assumes also that risk-buffering gift exchanges (Wiessner
1977) between adjacent bands would increase during times of range increases. As the
latter are modelled to occur during drought episodes, there are grounds for predicting
an increase in overlap of territorial boundaries, more porous boundaries (Sampson
1988: Figure 1-11 ), and increased gift exchange. Orange River pebbles were
potential gifts because of their attractive, colorful appearance and perhaps also
because of the edge-holding qualities of the stone itself, certainly superior to that of
the local hornfels. Estimates of real range sizes and boundary definitions are of course
impossible from raw material ratios alone (pace Hester and Grady 1977; Schiffer
1975; Wilmsen 1973), and large amounts of spatial data (e.g. Sampson 1988) are
needed before this becomes possible.
The test implication derived from the above hypothesis can be compared with the
actual plot of agate-jasper in the lithic debris from Blydefontein (Figure 120).
Meerkat Rockshelter is omitted from the test because samples are too small to be
reliable. The Blydefontein sequence shows that the Early Microlithic assemblage in
CAU9 has the highest frequency of all, and the Lockshoek in CAU8 has none. The
remainder of the early Holocene and the first half of the mid-Holocene is represented
here by the basal assemblage of the Interior Wilton sequence (CAU7), which has
almost as much agate-jasper as the Early Microlithic, but there is a steady decline in
the (always small) percentages thereafter. Similar declines are well known from
sites closer to the agate-jasper source (Humphreys 1972: 48-51; Sampson 1967a:
288
157, 1967b: 79, 1970: 105; Sampson and Sampson 1967: 71) where the overall
frequencies are higher. This is is a regional trend, therefore, not just a local one.
This late Holocene decline fits well with the new model's expectations of a gradual
reduction in hunter-gatherer territorial sizes during the Interior Wilton and
Smithfield industries. The fluctuations about this decline, due to habitat changes is not
reflected, because of the coarse interval of time represented by the samples.
5
4.5
4
3.5
(/)
·;::
.c
Q)
0
.£!l
!tl
Jf
セ@ 0
3
2.5
2
1 .5
.5
0
-.5
0
2
3
4
5
6
7
8
9
10
Combined AU
Figure 120. Percent of agate-jasper use through Combined Analysis Units.
Blydefontein Mobility in the Ecological Model
Obviously, a band's range size and its overall amount of mobility covary (see
Figure 101 and 102). Another approach to calculating total mobility is to use the
estimated amount of hunted food in the Blydefontein diet. Under modern conditions this
is 30-31 percent and ranges between 29-36 percent for the Holocene (see above).
When this is fitted to the regression curve in Figure 95, the resulting mobility score
(number of camp moves X average length of logistic trip) is ca. 50, which is
289
relatively low on the global scale. Any decrease in the frequency of hunted food at
Blydefontein will produce almost no change in overall amount of mobility, but an
increase of 15-20 percent of hunted foods could significantly raise the required
mobility of the band.
Changes in the Blydefontein Collector/Forager Mobility Mix
The relative abundance of foraged meat in the Blydefontein fauna suggests strongly
that it belongs on the forager-dominant end of the collector/forager scale. Most of the
assemblages in the sequence would match somewhere along the gradient marked by
hypothetical bands C-D-E in Figure 113. The only exception is the Early Microlithic
(CAU9} that has no microfauna and is dominated entirely by large animals (R. Klein,
personal communication). As the Early Microlithic assemblage was deposited during a
time of more restricted environmental possibilities (see Chapter VII), it is
reasonable to place this basal part of the sequence on the collector portion of the scale
in the vicinity of band B in Figure 113.
A less intuitive fit can be achieved by looking at the number of camp moves as
predicted by the percentage of the hunted food in the diet and calculating a linear
regression. However, as Figure 121 shows a simple relationship does not exist. At
approximately 45 percent of hunted foods in the diet residential camp moves either
continues to increase or declines (presumably as the number of special task group
trips, i.e. logistic mobility begins to dominate the mobility strategy.
Given this split,
a linear regression was calculated without the four (high percent of hunted foods and
low number of residential camp moves) groups in the lower right hand corner
(number of residential camp moves
=
-4.129 + 0.835 * percent of hunted
resources). As the range of predicted hunted foods in the diet ranges between 29 and
36 percent, the span of
290
70
(/)
Q)
>
:§
60
0
a.
E
!IS
0
セ@
c
50
0
0
40
Q)
0
"0
"(jj
30
セ@
0
z
Q)
g'
....
Q)
セ@
0
20
0
0
10
0
0
0
10
0
0
0
20
0
0
0
30
0
0
40
50
60
70
80
90
Percent of Hunted Foods in Diet
Figure 121. Scatterplot of percent of hunted foods in diet by average number of
residential camp moves (data from Kelly 1983).
estimated camp moves falls between 20-26 moves. This is not a great deal of
variation and it seems likely that for the entire Holocene the Blydefontein huntergatherers would be classified as foragers with slightly lesser or greater amounts of
non-residential camp mobility.
Changes in Forager Mobility Predicted by Plant Productivity
The estimated primary production of plants for the individual pollen samples has
been plugged into the formula for estimating number of camp moves (see Figure 92)
and the resulting values are presented in Figure 122. These plant primary production
estimates are based on the pollen-derived rainfall estimates, and again, I caution the
reader that the actual historic rainfall variation was much greater than that estimated
by the pollen and also that these are only estimates that I use for model building and
not exact observational values. I expect that the actual range of camp moves by
Blydefontein foragers would be greater than those predicted.
Nevertheless the test
291
implications for agate-jasper are the same as those for changes in range-size, so the
test (Figure 120) is also the same. It is clear that the early and mid-Holocene
foragers were highly mobile foragers and it is likely that reduced carrying capacities
were the reason. Between 4000 B.P. and 900 B.P. reduced forager mobility appears
to be a response to richer carrying capacities. The predicted rise in forager mobility
does not show on the agate-jasper frequencies, possibly because population densities
were then restricting mobility.
This figure suggests that the position of Blydefontein's forager/collector mix
should be within the range of bands B and C in Figure 113. Further test implications
for this segment of the model must be derived from its technological system, the
components of which were reviewed in Chapter XII. As several aspects of decisionmaking in artifact design are intimately entwined with the forager/collector mobility
mix, it is in the domain of artifact design that tests for the above prediction must be
sought.
29
.
•il'\
セY@
u....セサ@
• •
OGセ@
••• •
•
I• • •
Estimated
Forager
Camp
Moves
24
23
0
1·•-
ᄋMセK@
1000 2000
3000 4000 5000
Years BP
Oppermanskop ·0- USP
6000 7000
8000
·•- BSM
Figure 122. Estimated changes in the overall mobility of Blydefontein occupants based
on estimates of plant primary production and correlation between plant primary
production and forager camp moves (Figure 92).
Stone Tool Design Decisions among Foragers and Collectors
292
Procedures for identifying the mix of maintainable/reliable design qualities (see
Chapter XII) in a prehistoric stone tool are still in their infancy. Ethnoarchaeological
studies of living stone toolmakers are not sharply focused on the problem, and only
limited data for stone and bone tool use-life, manufacturing costs are available (see
Table 30). To make matters worse these do not include detailed data for chipped stone
tools. Replication/edgewear studies are mainly dealing with problems of usewear
verification (Keeley 1980; Keeley and Newcomer 1977; Moss and Newcomer 1982)
and these types of studies are just beginning to turn their attention to the question of
tool efficiency (Keeley 1991 ).
Dibble (1987) has shown that Middle Paleolithic
scrapers, whether for working hides or wood, are essentially maintainable tools, and
Barton (1990) has argued that almost all tools in Middle Paleolithic assemblages are
really a single tool type used for many different tasks, each specimen being discarded
at a slightly different stage of exhaustion. Thus tools classified as sidescrapers could
be altered through intensive use to become transverse scrapers, concave scrapers, or
sundry other types in the attribute-style approach developed in France (Bordes
1961 ), and elaborated and exported elsewhere (e.g. J. Deacon 1978; Sackett 1966).
Indeed Binford (1973, 1977) has argued that the Middle Paleolithic toolmakers
were foragers who used, in his terminology, expedient (or non-curated) designs, such
that tools would be discarded in a wide variety of shapes. One upshot of this is that
assemblages from different sites would end up looking very different from each other.
Binford further argues that the Upper Paleolithic toolmakers who followed used a
collector mobility pattern that demanded curated tools. These would have standardized
shapes that would not be altered before each tool was broken and discarded. In the
latter scenario, assemblages from different sites would look very similar. Subsequent
research (e.g. Dibble 1987; Hayden 1986; Marks 1988; Marks and Freidel 1978)
293
has not supported the linkage between forager/expedienVMiddle Paleolithic or
collector/curation/Upper Paleolithic, but none of these tests of the model is
particularly rigorous, and the results need not be taken to imply that the linkage
never existed elsewhere or in other periods of prehistory.
Little has been done yet with the effects of mobility on the appearance of stone
tools. Kelly (1988) suggests that highly mobile groups who were at the collector end
of the forager/collector mobility scale carried bifacial cores about and used them for a
relatively long time. He suggests that they were used also as tools that could double as
raw material sources in areas and times when suitable rock sources were not
available. In this case, bifacial cores in different assemblages from different sites
would end up looking quite different from one another.
Stone Tool Design Decisions in the Later Stone Age
In South Africa, the lack of ethnographic stone tool makers is keenly felt. Almost
all Bushmen now make metal arrowheads from fencing wire instead of stone, bone or
wood (Lee 1979: 133, 277;
Schapera 1930: 128-130;
Silberbauer 1981: 206).
The switch took place in the early 20th century when wire was still a scarce
commodity. As the source became more abundant, so the switch to wire arrowheads
became almost complete except where wire remains unobtainable (Wiessner 1983).
This change does point up one apparent axiom about design efficiency, however. More
durable and less brittle raw materials are at all times preferred. As access to these
materials changed, so the frequency of tools in the assemblage changed according.
The role of raw material has been evoked repeatedly in the literature to explain
changes in assemblage composition and stone tool design. In the interior plateau, the
overriding assumption has been that siliceous rocks like agate and jasper are more
flakeable, less brittle and have more durable edge-holding qualities than hornfels.
294
These differences have been used even to explain why there are no microliths in the
Smithfield Industry: the knappers only used hornfels (Fagan 1965: 38). The
underlying assumption is that assemblage composition is tied to the availability of
suitable rock outcrops. A deeper assumption, nowhere made explicit in the South
African archaeological literature, is that a group's access to suitable rocks is strongly
linked to its mobility pattern mix and its range size. For example Parkington (1984:
128-131) suggests that the widely observed shift from an Early Microlithic (ct.
Robberg) assemblage to the macrolithic (Oakhurst) assemblage was a response by
resident hunter-foragers to a drop in the amount of available non-hornfels, but the
implications for reduced range size and mobility are not spelled out. These controlling
factors are given prominence in the technological system of the new model (Chapter
XI) and are the target of the tests that follow.
Changes in Stone Tool Design Decisions at Blydefontein
The stone tool designs from Blydefontein that are most amenable to analysis are
those with large enough sample from each of the CAU's in the sequence. These include
the endscrapers, and the backed bladelets. Other items of technological interest are
the bladelet debitage (and their parent cores) and the flake debitage.
End Scrapers
Later Stone Age endscrapers from the interior plateau have been long known to
occur in a wide variety of forms (Goodwin and van Riet Lowe 1929). They have been
extensively described from the Orange River Scheme area (Sampson 1970: 5-6;
Sampson 1972: 259-261, 322-327, 376-381;
Sampson and Sampson 1967: 6)
and subjected to attribute and metric analysis at Highlands Rockshelter (H. Deacon
1976: 58-69). No systematic study of edgewear has been undertaken yet, although
the potential clearly exists (Binneman 1981, 1983).
The function(s) of the
295
endscraper within the Later Stone Age adaptive system is still a wide open question,
therefore. The only ethnoarchaeological research of immediate use is that on Ethiopian
herdsmen who use hafted obsidian hide scrapers to prepare skins. From these studies
(Clark and Kurashina 1981: 307-311;
Gallagher 1977: 411-413) it is fairly
clear that overall shape and size (especially length) directly reflects the intensity of
use, and also whether the tool was hafted or not. Unremarkably, exhausted scrapers
are shorter than new ones because they have been trimmed (resharpened) so many
times. Also, hafted ones are shorter because one can still grip the tool long after an
unhafted scraper has become too small to hold. The hafted ones are normally trimmed
on their butts to improve their fit with the handle (Clark and Kurashina 1981: 309;
Gallagher 1977: 41 0). Non-hafted scrapers are larger and have much less trimming
(Clark and Kurash ina 1981: 308). The motives for making hafted versus hand-held
scrapers is not recorded, but the lack of suitable large flakes could be a limiting factor
that might favor the use of hafted scrapers.
Short (Hafted) Endscrapers
Many analysts agree with Clark's (1959: 202, 232-234) suggestion that the
differences between shorter (typically Wilton) and longer (typically Smithfield)
endscrapers might be due to the former hafting their scrapers and the latter using
hand-held scrapers (H. Deacon and J. Deacon 1980). Although hafted scrapers and/or
adzes have been recovered from several coastal sites, usually embedded in a wad of
mastic and attached to a wide variety of handles (Clark 1959: 232-234, H. Deacon
1966), there are few typical endscrapers in the sample. However, numerous short
specimens with mastic adhering to the butts come from Melkhoutboom Cave (H. Deacon
1976: 58, H. Deacon and J. Deacon 1980), and it is now widely accepted that these
were hafted and used for hide scraping.
296
These scanty and diverse scraps of data all converge to indicate that the short,
hafted endscraper fits into the new model's technological system thus: it was designed
with a balance (50-50) mix of maintainable and reliable qualities. It was designed
also to have relatively long use-life and was refurbished rather than replaced.
Discard occurred only when the bit was close to exhaustion. Note that there is no
evidence for or against the replacement of LSA short endscraper bits. Production time
was relatively costly because of the three materials that had to be gathered (stone,
mastic, handle), so it is likely that it was carried about for some time before it was
finished. There is no evidence of caches. It would have been in daily use at regular
make-and-mend sessions. All of these qualities make it fit easily into the repertoire
of a group with a forager-dominant mobility pattern (Figure 113).
It misfits at one
point in the system, however, because it falls squarely within the repair kit
(maintenance tool) category, used to produce and repair other tools and weapons. The
new model predicts that foragers have less regard for such tools and replace them
often. Seen from this perspective, it would fit better with a collector's specialized
repair kit, with a long use-life.
Long (Hand-Held) Endscrapers
The longer, unhafted endscraper fits more comfortably into the forager toolmaking repertoire.
Here the maintainable design qualities are utilized more often,
with reliability of less concern. The unhafted endscraper is a very generalized,
maintainable repair kit tool with a relatively short use-life (expedient tool), and
soon discarded, well before exhaustion. It is also replaced when broken, which
happens often. Endscraper tip fragments are a common occurrence. It is in active
daily use at make-and-mend sessions. One residual doubt must be sounded, however.
Kannemeyer (1890) observed Bushmen near Burgersdorp, only 120km from
297
Blydefontein, using unhafted stone tools called "Kuin" as skinning knives for antelope.
His (albeit sketchy) illustration resembles an elongated "Smithfield" endscraper, but
there appears to be heavy, adze-like retouch along the sides that is atypical. If
endscrapers doubled as skinning (extractive) tools, they should have relatively longer
use-lives among foragers.
Test Implications
It is reasonable to predict that abundant suitable rock will encourage the
production of more hand-held, longer endscrapers, while a shortage of suitable stone
will promote the more costly production of hafted, short endscrapers. Length here is
considered to represent a rough index of remaining use-life. This is different from
the Index of Reduction proposed by Kuhn (1990) who is attempting to measure the
consumed portion of a scraper. However, Kuhn's procedure requires fairly
standardized blanks for scraper use, and these are lacking in the artifact assemblages
at Blydefontein. Viewed from a single site, a switch from abundance to scarcity of
suitable rock can occur only when the occupant's range (and mobility) is curtailed so
that they no longer have easy access to the right stone.
Test
The agate-jasper data support a model in which the Blydefontein occupants' range
was gradually shrinking throughout the later Holocene (Figures 119 and 120). It
follows that the test implications for endscrapers are that shorter, hafted endscrapers
will proliferate with time. When mean endscraper lengths are plotted (with 1
standard deviation bars) the test fails (Figure 123). No endscrapers were recovered
in the tiny Early Microlithic assemblage, and the single Lockshoek scraper in CAU8 is
outside the range of the later samples and will not be considered further. Overall,
endscrapers are smallest in CAU7 at the base of the Interior Wilton sequence, and
298
increase in length through time. This implies that curation and hafting of endscrapers
decreased, thus refuting the expectations of the model.
90
80
70
..c
0,
c
セ@
60
....
Q)
a.
50
Ill
....
(.)
CJ)
40
"0
c
UJ
30
20
10
0
2
3
4
5
6
7
8
9
Combined AU
Figure 123. Mean endscraper lengths for Analytical Units at Blydefontein Rockshelter
with one standard error bars. No end scrapers were recovered from CAU9.
If this trend towards increased use of hand-held endscrapers was dictated rather
by a switch in the used of lithic materials to hornfels, as suggested by several authors,
starting with Sampson and Sampson (1967), then the range hypothesis derived from
the new model is redundant. However, Figure 124 plots the mean lengths of hornfels
endscrapers only, and the trend is still present. The shift in endscraper raw material
is too trivial at Blydefontein, and does not control the design change. Nor is it certain
that a shrinking range would have made hornfels shortages inevitable for the
Blydefontein occupants. In the adjacent Zeekoe Valley, hornfels outcrops are abundant
and evenly distributed (Sampson 1984: 17, 1985: 81), and the nearest quarry to
Blydefontein and Meerkat Rockshelters in only Skm away in the neighboring Hughdale
basin.
299
40
..c
0,
c::
セ@
35
.....
セ@
CIS
.....
(.)
30
C/)
"'0
c::
UJ
25
.Sl
"2
CIS
'5
20
3'
c::
セ@
CIS
15
10
0
2
3
4
5
6
7
8
Combined AU
Figure 124. Mean lengths of hornfels endscrapers for Analytical Units at Blydefontein
Rockshelter with one standard error bars. No end scrapers were recovered from
CAU9 and standard errors should not be calculated for the two hornfels end scrapers
from CAU8.
Obviously other factors are at work here, and rival hypotheses must be built and
tested. Possibly the assumptions of similar use for hafted and hand-held scrapers is
incorrect, and this can only be tested by extensive use-wear studies. Another testable
scenario is that shrinking ranges made it more difficult to obtain mastic (tiny drops
adhere to some Blydefontein tools although its source is unknown), while hornfels
shortages were unlikely. The increasing risk of being caught short without mastic
makes it more efficient to carry around (or cache at camps) a few large hornfels cores
from which to strike suitable flake-blade blanks for hand-held endscrapers. This
will require a study of local mastic sources, their distribution on the landscape and
their fluctuations through time. If mastic sources prove to be rare, widely spaced and
clumped on the the landscape, then the test implications would fit the endscraper data.
300
A final factor may be that a change in the mobility (from high to low) influences a
band's ready access to lithic sources during work sessions. This is different from the
first hypothesis. If a band is highly mobile it may be easier to use hafted, long uselife endscrapers in limited, more intensive work sessions when maintenance tools may
be rapidly exhausted. In less mobile bands that work sporadically but often in make
and mend sessions, raw materials may be obtained daily embedded in other
foraging/hunting trips.
Obviously more work is required before this issue will be
settled.
Backed Bladelets
These tools are assumed, on limited grounds to have served exclusively as arrow
armatures, either barbs or points.
Clark (1977) describes a number of 19th
century Bushmen arrows that have backed bladelets mounted in mastic as points. In
1872 Dunn (1880) observed an old woman in Botswana mounting two small
triangular flakes in mastic to make an arrow point. Within the framework of the new
model they should be viewed as artifacts to be replaced, rather than maintained. Their
main design characters should be reliable rather than maintainable, and they will
have relatively short use-lives, possibly with relatively long (passive) spells out of
use. Collectors will replace them frequently in brief gearing-up sessions to buffer
the risk of failure (Kuhn 1989). Foragers will replace them only when broken, and
they will do this during daily make-and-mend sessions.
If overall forager mobility is
forced to increase, however, there may be a collateral increase in collector-like
repairing, i.e. replace-before-failure.
A measure of replace-before-failure planning can be obtained by looking at the
frequency of whole backed armatures versus broken fragments. Fragments are
believed here to reflect replacement after breakage (see Keeley 1991 for intrasite
301
discard patterns of this behavior). Whole arrow armatures may take the form of
backed points, crescents, complete backed bladelets or complete bladelet blanks.
Fragments including proximal, medial or distal pieces of any of the former. Note that
manufacturing discards (proximal, medial and distal) are also present, but as most of
these enter the archaeological record during production (H. Deacon 1976: 141-13;
Movius et al. 1968: 50-51), they are not used in this test.
Test Implications
The implications for backed bladelets are that more whole specimens will occur in
the shelter at times when overall mobility has increased, and also when the frequency
of collector mobility patterns (task-specific trips) has increased.
At times when
forager mobility patterns (frequent camp moves) prevail, there will be more
fragments, reflecting a more relaxed, fix-when-broken approach to arrowhead
maintenance. Thus the relative frequencies will covary with the predictive curves
shown for mobility in Figures 119 through 121.
Only Blydefontein assemblages are considered as assemblages from Meerkat are
too small for reliable comparisons. Table 33 lists all backed artifacts by CAU, and
Figure 125 plots the percentages for the complete backed bladelets. The percentages
suggests that complete backed tools occur in high frequencies in the the bottom of CY
(CAU9), the upper portion of CY (CAU7), and lower portion of TG/CAC (CAU6). A
working hypothesis can now be derived from these results: Interior Wilton hunters
were discarding a number of complete tools and by implication using a significant
amount of replace-before-failure planning in their arrowhead maintenance, a
practice which they all but abandoned in the later Holocene. This fits well with the
overall expectations of the shrinking-range hypothesis tested above. It would also fit
302
well with the still-untested proposition that mastic became a scarcer commodity in
the later Holocene--this could have inhibited frequent armature replacements as well.
Table 33.--Backed artifact discards, broken fragments and complete tools at
Blydefontein Rockshelter. Percentages calculated for fragments and complete backed
blade lets
MajQr QA!..!
1
2
3
4
5
6
7
8
Fragmf!nts
0 17 177.7%
27 I 81.8
46 I 82.1
24 I 82.6
83 172.8
7 158.3
DisQards
1
4
8
13
5
22
1
1
195
Q
9
54
Total
TQtal (- DisQards)
1
(0)
13
(9)
41
(33)
69
(56)
34
(29)
( 1 1 4)
136
13
(12)
0
QQmQiflle TQQIS
0 12 122.3%
6 118.2
10 117.9
5 117.4
31 I 27.2
5 141.7
2 l 66.7
l33.3
61
3
(3)
310
( 2 56)
70
(f)
0
60
0
1"0
50
(])
セ@
(.)
ro
Ill
40
(])
0
..c
セ@
-e
0
30
20
c
(])
10
(])
0..
0
-1 0
0
2
3
4
5
6
7
8
9
10
CAU
Figure 125. Percent of complete backed bladelet tools by Combined Analytical Unit at
Blydefontein Rockshelter.
303
Bladelet Debitage and Bladelet Cores
If we cannot assume that hornfels shortages for the Blydefontein occupants were
real during times of shrunken range and mobility (see caveat above), the model's
expectation of more efficient use of raw materials (Kelly 1988: 719-721;
Parry
and Kelly 1987) may not apply in this case study. However, other related factors may
come into play. If efficiency is defined as getting more cutting edge per weight of
stone, then bladelet technology is a logical response to the need for more efficient use
of stone (Bordaz 1970: 56-57;
1988:
Clark 1976: 13;
Leroi-Gourhan 1943;
Mitchell
262-263).
Although range reduction and quarry losses may not be enough to induce more
bladelet manufacture, the hypothesized decision to adopt more replace-before-failure
planning in arrow maintenance will stimulate more bladelet production.
Because
bladelet shape and size can be better controlled during knapping than flake shapes and
sizes, a greater proportion of pieces will be useable in any reliable technology that
needs bladelet fitted parts like arrow armatures.
It follows that bladelet production is likely to play an important part in a
replace-before-failure pattern of arrow maintenance, because the demand for
armature blanks will be relatively great.
Test Implication 1
The hypothesis predicts higher bladelet production levels in mid-Holocene layers
at Blydefontein and less during the later Holocene levels. The curve for changing
bladelet frequencies in the Blydefontein sequence should resemble the curve for whole
arrow armatures (see Figure 125).
304
Figure 126 plots the percentages of bladelets in the lithic debitage of each CAU at
Blydefontein. The graph shows that bladelets were predictably common in the Early
Microlithic (CAU9), and predictably rare in the Lockshoek (CAU8).
However, the
Interior Wilton in CAU7 has few bladelets. The frequency climbs to high levels on in
CAU6 and it remains high through CAU4, after which it declines steadily.
These
results do not match with the frequency curve generated for the test implications
above, so the test apparently fails to support the hypothesis.
When bladelet cores are plotted against flake cores (Figure 127) there is a rough
covariance with the bladelet frequency curve (see Figure 126), but the high
frequency is sustained through CAU3 although this and the following samples are very
small. Nevertheless this curve also fails to covary with that for whole armatures.
20
18
16
.!9
(])
Qi
14
"0
ctS
co
セ@ 0
12
10
8
6
0
2
3
4
5
6
Combined AU
7
8
9
10
Figure 126. Percent of bladelets among lithic debris by CAU.
Thus bladelet production ratios are not linked to restricted access to quarries, nor
are they determined by the amount of replace-before-broken planning in arrowhead
305
maintenance schedule. A far more basic proposition is that the overall need for
armature blanks, i.e. backed bladelets and their microlith variants, determined how
often bladelets were produced. It is well known that the frequency of backed
microliths is relatively low at the start of an Interior Wilton sequence, it peaks in the
middle, then declines to near zero in the terminal (Smithfield) levels at the top of the
sequence. This is demonstrated at several sites, and Blydefontein is no exception.
70
60
c
Q)
50
セ@
Q)
a.
40
セ@
-
0
()
30
Q)
(i)
"0
!1S
ill
20
10
0
-1 0
0
2
3
4
5
CAU
6
7
8
9
10
Figure 127. Percent of bladelet cores among lithic debris by CAU.
Test Implication 2
As the widespread trend bears a general resemblance to the bladelet curve in
Figure 126 and the bladelet core curve in Figure 127, there are grounds for
supposing they are linked. Thus the hypothesis states that there will be good
correlation between the percentage backed bladelet tools in an assemblage and the
percentage of bladelets in the debitage. Likewise there will be a similar correlation
306
between the percentage of backed bladelet tools in the assemblage and the percentage of
bladelet cores.
A regression analysis displayed in Figure 128 demonstrates a clear relationship,
and the hypothesis is supported. The rate of bladelet production is determined by the
need for backed bladelets for arrow armatures and not by external factors. The early
portion of the Interior Wilton sequence (CAU6-7) has more backed crescents that
could have been made of small flake blanks. Reasons for this decline in backed
armatures toward the end of the sequence is uncertain, but it has been linked at Glen
Elliot Shelter to an increase in the use of cylindrical bone arrow points without
armatures (Sampson 1967a). This relationship cannot be demonstrated at
Blydefontein, however.
(/)
Q)
C)
res
c
e
Q)
Q)
Cl..
70
65
60
55
50
45
40
35
30
25
20
15
10
5
0
-20
,._ Bladelet Cores
0
20
OBiadelets
40
60
80
100
% Straight Backed Bladelets
Figure 128. Scatterplot and linear regression between straight backed bladelets with
bladelets (r2 = 0.391, p = 0.072) and bladelet cores (r2 = 0.673, p = 0.089).
307
Dorsal Cortex on Debitage
The amount of cortex on the dorsal surface of debitage is a rough measure of the
degree to which cores have been reduced (Ahler 1989; Butler and May 1984; Collins
1975; Henry 1989; Johnson and Morrow 1987; Muto 1971; Vehik 1985).
Factors controlling the amount of core reduction are many, and not all of these are
connected to the size and quality of the original block or nodule. One factor of
immediate interest is the effect of source-to-site distance on the amount of core
reduction found at a site. All things being equal, the farther the core must be carried,
the more material the knapper will try to get from it. If quarries at great distances
from the site are being exploited, then cores from those quarries will be more heavily
reduced and less cortex from those cores will occur in the debitage. As huntergatherer mobility, either number of camp moves or length and distance of task group
trips, increases the amount of dorsal surface cortex will decline. Implicit in this
scenario is the na'ive assumption that nodules/blocks were not decorticated at the
source.
Test Implication
The model predicts reduced amounts of cortex on the dorsal surface of debitage in
mid-Holocene Interior Wilton and Early Microlithic assemblages when mobility levels
are thought to be higher.
All lithic debris was classified into six cortex categories and ranked: zero percent
cortex = 0, one to twenty-five percent cortex = 1, twenty-six to fifty percent cortex
= 2,
fifty-one to seventy-five percent cortex
= 3,
seventy-six to ninety-nine percent
cortex = 4, and one-hundred percent cortex = 5. Simple averages of the cortex ranks
were calculated for all flakes and bladelets over 10 mm in length. This eliminated the
308
very small resharpening flakes produced by the retouching of scrapers and other
tools. Figure 129 is a line graph of the resulting mean cortex rank for flakes and
bladelets from each of the nine Blydefontein Rockshelter assemblages.
There is high cortex retention in the Early Microlithic of CAU9, and a significant
drop in CAU8. Cortex retention again peaks in CAU4-5, and then steadily declines to
CAU1. These results are virtually opposite to those expected. For example the very
low retention in the Lockshoek sample of CAU8 appears to be anomalous because cortex
on Lockshoek scrapers is so common that it is a hallmark. This case points up one
flawed assumption mentioned above: that cores were not decorticated at the quarry.
Hundreds of Lockshoek (Goodlands) quarries are known from the middle Orange River
(Sampson 1972) and from the Zeekoe Valley (Sampson 1985). One notable feature of
these quarries is that large, hand-sized flakes were struck directly from the bedrock
outcrops. If these were carried away as raw material, they would retain cortex only
on the dorsal surface, thus cutting in half the potential amount of cortex removed from
the quarry. The smaller flakes were converted directly into typical macro-scrapers,
while the larger ones were removed and further reduced as cores.
By contrast, the many Interior Wilton sources recorded from the Zeekoe Valley
are usually weathered rubble cobbles in terrace gravels or outcrop scree fans.
Refitting of bladelets and flakes on the original cores demonstrates that the flat oblong
rubble cobbles make perfect pre-formed blanks for bladelet removal. Typical
Interior Wilton flaking debris at these sources includes tested cores and some
decertification flakes, but all indications from both quarries and other sites suggest
that small cobbles were taken away whole, and with most of their original cortex
surface intact after only a few test flake removals (Sampson 1985: 29, 57).
Instead, there is an obvious resemblance between the cortex retention curve and
that for bladelet production rates (see Figure 126). It is reasonably to suppose,
309
therefore, that cortex retention is determined indirectly by technical production
decisions, rather than by any external factors. The use of weathered rubble as a
ready-formed source for bladelets simply produces more cortex on debitage than the
use of large flakes struck from bedrock outcrops. At Blydefontein, the cortex
retention curve does not conform to the predicted correlation with the overall
mobility curve, so the test fails to support the hypothesis that cortex retention
reflects range/mobility .
.t. Flake Cortex Rank
.7
0 Blade let Cortex Rank
.6
.:.:.
c
rd
.5
a:
X
(])
t::
0
(.)
c
.4
.3
rd
(])
:2
.2
.1
0
0
2
3
4
5
6
7
8
9
10
Combined AU
Figure 129.
Mean cortex rank for flakes and bladelets in Combined Analytical Units.
Isolation of the Variables that Reflect Curation Decisions
It is now apparent that an interlocking cluster of technological variables is
running interference with the continued testing of the new model. Before proceeding,
it is necessary to isolate this cluster from the other variables so that
interrelationships between the latter can be put in sharper focus. Thus far, each of
the above test compared only a pair of trends at a time. However, the results have
revealed similarities between more than just two pairs. In order to tease out more
310
complex, combined relationships between these trends, all the above data were
subjected to a factor analysis. Factor analysis assumes that the combined trends are
controlled by one or more variable, called factors, for which no direct observations
exist. Then, assuming linear relationships, the factor analysis attempts to identify
the underlying controlling variable(s) and to determine which of the documented
variables are influenced by them.
In this analysis, only the assemblages from the Interior Wilton sequence at
Blydefontein Rockshelter (CAU7 through CAU2) were used, as there are missing data
from all the others. Unfortunately bladelet cores had to be dropped from this test
because of small samples in CAU2 and CAU3. Two factors were identified (Table 34),
and an orthogonal varimax solution was used. The same factors composed of the same
artifacts were identified by oblique methods as well.
The first factor is an uni-polar factor with significant factor loadings
greater than 0.5 or less than -0.5) that were only positive values.
(i.e.
Not surprisingly,
Factor 1 had significant loadings on percent of bladelets, flake cortex rank and bladelet
cortex rank. This effectively isolates the bundle of trends controlled by technological
forces of the bladelet core reduction sequence. No doubt the trend for bladelet cores
would be controlled by Factor 1.
The second factor is bipolar, with significant positive loadings for complete
backed tools and percent of agate-jasper debitage. Factor 2 also has significant
negative loading for mean endscraper length. These are the variables that are more
likely to be determined by the various kinds of curation decisions discussed above.
These in turn are controlled by larger adaptive forces like mobility, and range size,
which are in their turn acted upon by carrying capacity and population density.
Table 34.--Factor loadings and proportion of artifact type variance (communality)311
explained by factor analysis. Factor Analysis uses Principal Components and varimax
orthogonal rotation. Also listed are proportion of common or explained variance for
each factor, and variable complexity. Variable complexity indicates how many factors
account for variable's (artifact type) communality. Significant loadings underlined
Variable
8rtifaQt tセq・@
Complete Backed
Blade lets
FaQtQr 1
-0.437
% Agate-Jasper
Lithic Debris
-0.263
Mean End Scraper
Length
FaQtQr 2
0.863
qュゥ・クエセ@
qュオョ。ャゥエセ@
0.935
1.481
0.949
0.970
1 .153
-0.387
-o. a9 2
0.954
1.363
Mean Flake Cortex
Rank
Q.864
-0.312
0.844
1 .257
Mean Bladelet
Cortex Rank
Q.977
-0.208
0.997
1.090
% Bladelets
Q.974
0.186
0.984
1.073
Proportion of
Common Variance
0.539
0.461
Factor analysis also measures, through the use of scores, the contribution of each
CAU assemblage to the factor analysis. These scores are plotted in Figure 130, and
this clearly shows that Factor 1 has a very similar pattern to the bladelet percentages
and the mean cortex retention patterns. While this resemblance is gratifying, it is the
scores for Factor 2 that are of outstanding interest. The scores decrease from the
bottom to the top of the sequence in a very regular exponential (decay) manner. These
results fit extremely well with the expectations of the model: the combination of stone
tool curation strategies is being gradually altered as an ongoing adjustment to
gradually shrinking range size and mobility due to increasing population density.
312
C Factor 2
OFactor 1
1.5
1.25
セ@
0
()
(/)
.75
.9
()
.5
.....
ctl
u.
i\1
c:
0
C)
.25
0
0
..c:
t::
0
-.25
-.5
-. 75
-1
1
2
3
4
5
6
7
8
Combined AU
Figure 130. Factor 1 and Factor 2 Scores for assemblages from Combined Analytical
Units.
Comparisons between Tool Design and Paleoenvironment Change
A detailed comparison (using individual AUs) between the paleoenvironmental
record as measured by the ostrich eggshell o13C values and non-local raw material
percentages shows that the high resolution situation is complex, but as the
environment becomes drier (and presumably carrying capacity drops} huntergatherers use more non-local raw materials (Figure 131 ).
This pattern is especially
strong for the older portion of the record (AU1 0 to AU24) when I believe that
population densities are low, but a response clearly occurred for the drought at 2000
B.P. in AU6 even though at this time I believe that populations have grown enough to
constrict population ranges. It is possible that the increased use of agate and jasper in
AU4-6 is due to increase exchange (gift-giving) as a response to greater risks and
stress as environmental productivity dropped rather than significant increases in
range size.
313
-6.0
3.0
0
2.5
-6.5
2.0
d13C
Percent
1.5
1.0
0.5
0.0
25
AU
1·0- a:s
·•- Non-local Raw Material
I
Figure 131. Ostrich eggshell stable isotope curve and agate-jasper percents for
individual Analytical Units at Blydefontein Rockshelter.
A similar comparison between the ostrich eggshell stable isotope curve and
endscraper lengths shows that endscrapers are short between AU24 and AU13 (the
relatively large endscraper measurement in AU21 is a single specimen) (Figure
132). This is a period when population appears to be increasing, but territories are
believed to be large. Between AU1 and AU13 populations are probably dense enough to
constrict hunter-gatherer ranges, and it is at this point that environmental change
and endscraper lengths becomes more tightly linked although there is a lag response
between environmental change and endscraper lengths. In general as the environment
becomes wetter and less stressful and with limited risks, endscraper lengths become
longer due to less intensive use. When droughts occur, such as at 2000 B.P. in AU6,
endscraper lengths drop, an indication of more intensive use and greater curation.
314
50
-6.0
-6.5
-7.0
\
.o
fv
1\
l;zy\\ . .0/\1 O-v . 1
0
0
d13C
-7.5
-8.0
-8.5
•
0
/o.,o \ ,..•-•
/0
\
\/
0
•
0
/
\
.,
I ᄋセML@
0 0
•
45
40
Mean
35 Length
30
25
;•
;·
20
o
セ@
-9.0 +----+-----+------·-----f-...::!to..___-+ 1 5
10
15
20
0
5
25
AU
1·0- CES
·•- Endscrapers
I
Figure 132. Ostrich eggshell stable isotope curve and endscraper mean lengths for
individual Analytical Units at Blydefontein Rockshelter.
Conclusions
Clearly, not just one set of factors will control the observable patterns recognized
in lithic assemblages in southern Africa. It has been argued above that population
density, range size, settlement mobility, tool design strategy, variations in core
reduction, and the nature and size of raw materials can influence the well known
patterns recognized as characteristic of the Later Stone Age in the eastern Karoo.
Stylistic changes in lithic artifacts, at least as defined by Wiessner (1983, 1984)
may be particularly difficult to identify, and making assumptions that morphological
variation equals stylistic variation results in incorrect conclusions.
In reference to
Wiessner's usage of style, it is clear that style, either assertive or emblemic,
requires visual comparison, and many of the artifacts most commonly used for
stylistic analysis are tools that would be obscured from view, thus making comparison
impossible. However it is possible that hafts had stylistic designs (Clark 1959; H.
315
Deacon 1966), and it is possible that as hafts changed stylistically the manner in
which stone tools were affixed also changed.
CHAPTER XIV
SUMMARY AND CONCLUSIONS
In this dissertation I undertook three objectives. The first was to reconstruct a
radiocarbon-calibrated record of Holocene climate change for my research area: a
small portion of the South African interior plateau. Because this semi-desert region
is littered with (mainly lithic) residues of ancestral Bushmen hunter-gatherers, and
because it is well known that their stone tools gradually changed shape and frequencies
throughout the Holocene, my second objective was an obvious choice. I would have to
explain why these changes occurred if I was to avoid merely describing another
artifact sequence. Furthermore, I would need an ecological-adaptative model that went
well beyond the limited cultural-stylistic causes already put forth to account for
Later Stone Age tool design changes. While building the model, I took advantage of a
variety of recent data on living hunter-gatherer ecology, and tool-making/using
strategies.
Having built a climate-driven model and integrated Binford's (1979)
notion of hunter-gatherer managed technologies with it, my third task was to test the
model. This I did by comparing trends in stone tool designs and frequencies recovered
from my own excavations with the climatic/ecological changes I had already extracted
from the parallel geological sequence.
Holocene Paleoecology
Blydefontein basin in the Kikvorsberg Range has a restricted outlet that created a
sediment trap during the terminal Pleistocene and Holocene. Recent down cutting has
exposed a long sequence of sediments that reflect changing climatic episodes. Hillwash
316
317
and alluvial sediments interfingered with with humic-rich vlei deposits from which
radiocarbon, pollen, and stable carbon isotope samples could be extracted. Rare pond
deposits, yielding diatoms, pollens, and molluscs, intercalated with other valley fills.
Several exposures were studied and correlated with the help of seventeen radiocarbon
dates. The basal Older Fills, mainly dorbank deposits, were undatable, but contained
isolated Late Pleistocene extinct mammal remains. The Younger Fills span most of the
Holocene, and are correlated with the Blydefontein Rockshelter sequence, from which
another nine radiocarbon dates were obtained. Stable carbon isotopes from ostrich
eggshell in the Blydefontein Rockshelter sequence were correlated with those from the
radiocarbon dated humate samples. Abundant faunal remains from the Rockshelter are
still under analysis, and could not be used in the reconstruction. The last 1200 years
of the archaeological record is poorly represented at Blydefontein Rockshelter, but
this span is represented well at nearby Meerkat Rockshelter. Three further
radiocarbon dates in this second rockshelter sequence allow a tight correlation. Better
still, two finely stratified, pollen-rich hyrax dung middens are correlated with the
last 1300 years of the sequence by means of five more radiocarbon dates.
This rich, composite record reveals that terminal Pleistocene conditions were
colder and drier than the semi-desert, summer-rainfall, cold-winter pattern of
today. The Pleistocene plant communities have no modern analog. The early Holocene
saw a warming trend, but with continued dryness. Modern Karoo plant communities
(as defined by my new quantified analysis) began to establish themselves, but the
typical montane C4 grasses of the area had not yet appeared. After about 6500 B.P. all
pollen spectra are comparable to modern plant communities, i.e. combinations of
semi-desert scrub and grasses. There was a prolonged, major drought episode
between 6500-5500 B.P. that was probably drier than the current dry conditions.
Thereafter conditions improve, and the C4 grasses appeared at about 5000 B.P.
318
Another extended, extreme dry spell occurred between 5000-4000 B.P. This equates
well with the Cango Cave speleothem temperature record that shows a 2°C drop
between 5000-4300 B.P, rising rapidly to 4000 B.P.
Thereafter, rainfall levels
were higher than present for the rest of the late Holocene, except for several short
drought episodes. Cool-dry conditions occurred around 3200 B.P .; warm-dry
conditions at 2000 B.P.; and cool-dry once again at about 1300 B.P. The late Holocene
period of sustained high rainfall ends in the later part of the 19th century, and the
20th century is warm-dry. Modern conditions may not be as severe as the two midHolocene drought episodes, but over-grazing has distorted the pollen record so that
direct comparisons produce blurred results.
This record has been used to test computer simulations of regional climate
changes based on orbital forcing mechanisms. Although the simulations lack the fine
detail of the Blydefontein record, the overall fit is good. Thus the reconstruction
conforms with current models of circulation systems, and a major regional gap in
southern Africa climatic history has been filled. Although archaeological research has
been conducted in the upper Karoo for 25 years, only hints have emerged of a major
mid-Holocene dry spell, derived from radiocarbon dates (J. Deacon 1974) and
micromammal studies (Avery 1988). That episode is now verified and fixed in the
radiocarbon time scale as a two-phase event of which the second phase was apparently
cool-dry. The later Holocene is also established as a relatively wet episode with
oscillations, culminating in modern dry conditions. No other Karoo record is this
detailed, so well calibrated at multiple points of time, nor is any other record so
thoroughly cross-checked by multi-disciplinary methods.
It provides an important
baseline from which to judge prehistoric shifts between hunting and gathering, and the
raising of domestic stock (Sampson, personal communication), it is supported by new
paleoenvironmental research efforts in the area (Scott and Cooremans 1990), and it
319
hopefully will help stimulate future paleoenvironmental research in this largely
ignored region of southern Africa (Hubbard, personal communication). Of wider
significance, this dissertation represents an early use of hyrax dung middens as a
potent source of paleoclimatic data from the last millennium and more (Scott 1988a;
Scott and Bousman 1989), and the continued analysis of hyrax dung middens will
probably alter our current views of past environments dramatically.
The Model
Current models purporting to explain changes in stone tool design during the
Later Stone Age use too few processes to be at all convincing. In this climate-driven
model, carrying capacity and population density are prime movers. Ecological data
from over a dozen hunter-gatherer ethnographies in different habitats are ordered
according to each group's level of risk associated with the food quest. The axis of the
model is a sliding scale of mobility options used to reduce that risk. At one end of the
scale are mobility patterns that used frequent camp moves to secure enough food.
These are called forager patterns. At the opposite extreme are settlement strategies in
which camps are seldom moved, but long, task-specific trips are taken to secure food
by smaller groups, which is often cached or stored. These are collector patterns.
Most hunter-gatherers use a mix of both patterns. The chosen balance is
determined by the spacing and seasonal timing of foodstuffs in the habitat. Collectors
live in habitats with clumped erratic sources; foragers operate in habitats with
dispersed continuous food supplies. One crude measure of the mobility mix employed
by a group is the amount of hunted food in the group's diet, versus plant foods. If the
average temperature and rainfall of the group's habitat is known, the percentage of
hunted foods in their diet can be statistically predicted from the ethnographic data.
Armed with this figure, it becomes possible to predict the collector/forager mobility
320
mix chosen by the group, as well as the total annual mobility of the group. Percentage
of meat in the diet also makes it possible to predict the size of the group's
range/territory, which in turn allows a prediction of the amount of between-band
reciprocity.
The mix of collector/forager mobility choices employed by the band has a
profound impact on the way its member's design their tools. Again, ethnographic data
can be ranked to show that groups with strong collector patterns make a wider range
(relative to forager toolmakers) of specialized, reliable tools and weapons which they
keep in top condition by frequent replacement of parts. Such artifacts tend to have
short use-lives but are likely to be cached. Repair kits are carried around and have
longer use-lives. At the opposite extreme, foragers make fewer, more generalized
tools and weapons, with longer use-lives. They are casually maintained on the move,
and unfinished tools are carried around. The repair kit has no great value and is soon
discarded.
The Test
The stone artifacts from Blydefontein Rockshelter were used as a first and partial
test of the model. At the base of the sequence is a very small sample of Early
Microlithic (ct. Robberg sensu Jato) followed by a small Lockshoek sample. This is
followed by a mid-Holocene decline in occupation. Thereafter the accumulation begins
again with a dense and apparently continuous accumulation of Interior Wilton. These
occupations span most phases of the originally described local sequence. Unlike many
other rockshelter accumulations in the region, the terminal Later Stone Age phase
(Smithfield) is poorly represented.
When the ecological components of the model are applied to Blydefontein,
computations predict that the percentage of hunted food in the diet was 30-35 percent,
321
under modern conditions. On the sliding scale of collector-forager mobility, this
would place the Blydefontein occupants firmly in the class of forager-dominated
mobility patterns, with 20-30 camp moves per year (one of which involved
Blydefontein, itself).
There would be some long, task-specific (collector) trips also,
and these would occasionally include Blydefontein as well. Under modern conditions
the band's range size is estimated at 600km2 with moderate reciprocity and gift
exchange between bands.
Under cold-dry terminal Pleistocene and early Holocene conditions, the Early
Microlithic probably had around 40-45 percent hunted food in their diet, with fewer
camp moves, more collector trips and a much larger range. The high frequency of
macrofauna in the layer gives a subjective hint that these were mainly game hunters
without the generalized, microfauna! component in the diet that appears later. The
range would have been considerably larger than under present conditions.
Early Holocene conditions are estimated to have approached modern
circumstances, so that at ca. 8500 B.P. the Lockshoek diet, mobility mix and range
would be comparable to the figures given above for today.
Population decline marks the earliest Interior Wilton occupations during the
mid-Holocene major drought episode. Dramatically reduced carrying capacities forced
a near abandonment of the region, and the few remaining Interior Wilton groups would
have used much larger ranges, greater overall mobility, possibly tethered to
dependable waterholes, and a more intensive use of collected (versus hunted) foods.
After the mid-Holocene reduction in occupations, the Interior Wilton sequence
dramatically increases at Blydefontein around ca. 4300 B.P. under cool-dry
conditions. The model predicts increased hunted food in the diet over modern
conditions, with more collector trips and fewer camp moves per year. Range size is
larger than modern. Thereafter, three millennia of generally wetter conditions induce
322
a marked population growth. During this time span, late Holocene population density
becomes the prevailing force. Increased density drives down range size and
reciprocity to an all time low at the end of the sequence at ca. 700 B.P. While the
minor drought episodes centered on ca. 3200 B.P., 2000 B.P. and 1300 B.P. should
have expanded the range size, the model predicts that there would have been relatively
minor impacts on the dietary and mobility mix. Population growth dominates the
minor setbacks induced by these oscillations.
Range size and reciprocity changes can be inferred from changing frequencies of
agate-jasper found in the Blydefontein sequence. These came from sources at least
65km distant and are a measure of contact with remote quarters of the landscape.
Agate-jasper percentage is highest in the Early Microlithic, drops to zero in the
Lockshoek, starts high at the base of the Interior Wilton, and declines at ever slower
rates to zero by ca. 700 B.P. This is an excellent fit with the expectations of the
model, and a high resolution test with the ostrich eggshell stable carbon isotopes
shows that increased use of agate-jasper occurs in association with drought episodes at
ca. 2000 B.P. and 1300 B.P., but not 3200 B.P. This last drought episode was the
only late Holocene dry phase that could not be demonstrated with more than one line of
paleoenvironmental evidence, so its accuracy may be questioned.
Habits of composite arrowhead maintenance are a good reflection of the way
hunters approach risk. The typical collector pattern is to replace whole microlithic
armatures frequently to keep arrowheads in top working condition in case a suitable
quarry is encountered. Typical forager patterns evince less anxiety because suitable
quarry are encountered more frequently, and a few missed opportunities will not spell
disaster. Armatures are repaired only when they are broken. Thus the percentage of
whole microliths in the camp residues will increase in periods when quarry
encounters diminish. Declines in encounters are induced by drought. In the
323
Blydefontein sequence, whole armatures are at record frequencies in the Early
Microlithic. There are no data from the Lockshoek, presumably because microlithic
armatures were not in use then. When the Interior Wilton occupies the shelter, whole
armature percentages are again high, then they decrease sharply to ca. 3200 B.P.
Thereafter, percentages are relatively stable until the most recent portion of the
sequence, with a hint of increase at the end. Overall, this fits well with the
expectations of the model, but poor dating resolution and small sample sizes do not
allow me to test the potential effects of the three late Holocene droughts, although the
terminal rise might be associated with one of these droughts.
Endscrapers afford bigger samples throughout the late Holocene sequence, so these
should be more sensitive indicators still. Short hafted, and often trimmed endscrapers
are maintenance-and-repair tools that require three different materials (handle,
mastic and stone bit) to construct. They are relatively costly to produce and will not
be discarded frequently. This design conforms well with the model's requirements of a
typical collector's repair kit.
Long unhafted endscrapers perform similar functions
but could double as knives, and are more generalized tools, therefore. They cost very
little to produce, are ill cared for, often break, and are quickly discarded. This fits
well with the model's requirements for a typical forager's repair kit tool.
At Blydefontein, small hafted endscrapers predominate during the more stressful
conditions at the start of the Interior Wilton sequence, but decline once conditions
improve. Hand-held endscrapers increase rapidly in frequency after conditions
improve. Also, they increase gradually in average length throughout the sequence.
There is a decrease in both frequency and average size at the end of the sequence. The
overall fit with the expectations of the model are good, and a high resolution
comparison with the ostrich eggshell stable isotope record suggests that the predicted
changes associated with the three late Holocene droughts occur as predicted as well.
324
Another set of tests designed around the frequency of microblade production and
cortex retention failed outright, and further analysis suggested strongly that bladelet
production may have been an independent technological variable, driven by
mechanisms not accounted for by the model. Collectors replace arrow armatures more
often than is absolutely necessary, so they should need more bladelet blanks.
However, bladelet production frequencies failed to covary through time with whole
microlith frequencies, so it cannot be demonstrated that bladelet production is tied to
the needs of collectors. Nor can it be demonstrated that flake production (and cortex
retention) increased as they switched from hafted to unhafted endscrapers. Evidently
the models needs to be modified further to take into account the changing role played by
informal tools (untrimmed flakes) in the overall tool-management strategy.
Throughout, these tests have been bedeviled by small sample sizes which force the
collapsing of short-term units into larger ones. This has robbed the tests of the kind
of chronological resolution needed to provide adequate rigor, but the overall trends
revealed thus far are encouraging, to say the least. The multi-causal nature of the
model's design makes it easily adaptable to other local sequences from southern Africa
with good parallel climatic records. This model holds out the first real possibilities
for explaining some of the subcontinent-wide trends in Later Stone Age tool design that
have been observed by many researchers, but not adequately explained by them. With
judicial care the general model might be used for other periods such as the Middle
Paleolithic and Howiesonspoort, where significant changes in tool design strategies are
evident. For example the largely untrimmed Orangian assemblages contrast sharply
with the heavily retouched scrapers and points from Florisbad, and these two
assemblages may represent expedient and maintainable technologies while the backed
crescents in the Howiesonspoort seem to be an early production of a reliable tool. If
the Howiesonspoort does represent the production of reliable technologies then the
325
organizational abilities of Middle Stone Age groups may be more complex than Binford
(1984) seems willing to accept.
Further applications to Acheulian (largely
maintainable) and even Oldowan (expedient) assemblages await consideration.
Blydefontein in the Prehistory of the Interior Plateau
A more sharply focussed picture of Later Stone Age hunter-gatherers in the
interior plateau is beginning to emerge. This picture begins with a somewhat clouded
view, but Blydefontein offers the only available glimpse of very thinly scattered
groups of Early Microlithic hunter-gatherers in the late Pleistocene. They evidently
ranged over large areas and exploited the larger, more mobile herd animals. These
Early Microlithic groups appear to have employed collector-dominant strategies or
they even may have been serial specialists [in Binford's (1979) sense] in that their
own movements could have been planned around the predicted movements of migrant
herds. The little artifactual data that were recovered from Blydefontein Rockshelter
suggest that projectile point armatures were replaced with a before-failure strategy,
and nonlocal knappable stone frequencies suggest that large territories were travelled
by these groups. It is likely that bladelet technology and core transport were used to
insure a regular supply of flakeable stone.
Roughly coeval pollens from the Older Fills suggest that plant communities were
most similar to those of the higher Drakensberg ranges. Other coeval pollen spectra
(Coetzee 1967; Scott and Cooremans 1990} hint at numerous abrupt and rapid
fluctuations between Karoo scrub and grassland communities. It may emerge with
more research that such fluctuations created such unstable and unpredictable habitats
in much of the upper Karoo, that it was virtually abandoned by these Early Microlithic
hunters. When enlarged, Blydefontein will provide the only sealed assemblage with
which to compare the many reported "Driefontein" surface sites in the adjacent Zeekoe
326
Valley (Sampson 1985: 75). Most of these are ephemeral middle (cf. Developed)
Wilton chipping stations, but some are suspected to be Early Microlithic in age.
Blydefontein has provided the first associated radiocarbon date with a Lockshoek
assemblage, thus settling various long standing arguments about the nature and age of
the industry. It is indeed the local expression of the Oakhurst macrolithic complex in
southern Africa. At the start of the Holocene, there appears to have been a significant
increase in population, assuming that most "Driefontein" sites are later than
Lockshoek. Early Holocene pollens from Channel 2 suggest that conditions were wetter
and possibly more stable than Early Microlithic times. This seems likely, as the
adjacent Zeekoe Valley, the middle Orange River Valley, and Blydefontein basin, itself,
all indicate that Lockshoek hunter-gatherers were widespread throughout the region,
as well as in the type area of the western Orange Free State (Goodwin and van Riet
Lowe 1929). Lockshoek territories probably were fairly small, and there may have
been a dietary shift towards a more generalized diet focussing on nocturnal animals,
diverse microfauna and more plants. Although Blydefontein will afford the first
glimpse of the Lockshoek diet, an expanded sample is needed for a better view. The
Lockshoek groups employed typical forager technology focussed on quick production of
rapidly made knives and scrapers that were resharpened but not to the point of
exhaustion, and soon discarded. Projectile armatures are missing, and may have been
made of wood or bone. Expansion of the sample by further excavation should help
determine whether or not backed microliths persist in very low frequencies in
Lockshoek assemblages. The present sample is too small to settle this question, nor
can chronological markers for the Lockshoek be determined yet. Note, however, that
the Blydefontein sequence, along with others (Mitchell 1988) refutes Parkington's
(1984) model in which the Early Microlithic and the Lockshoek are aspects of the
327
same group's technological response to differences in the quality and size of locally
available flakeable stone. They clearly are of different ages.
The mid-Holocene is poorly represented at Blydefontein Rockshelter, and three
other rockshelters in the adjacent Zeekoe Valley have yielded gaps at this time
(Sampson, personal communication), although details have still to be worked out. The
pollen from BSM indicates a two-phase drought episode (6500-5000 B.P. and 50004300 B.P.), but rockshelter evidence cannot corroborate this. Nevertheless, these
combined results lend strong support to J. Deacon's {1974) hypothesis for a midHolocene population decline. Clearly rockshelters were seldom in use, and the very
thin but present mid-Holocene occupations at Blydefontein show that this is not a
completely missing segment of the record. Presumably drier conditions would have
stimulated more intensive use of plant foods, and lower carrying capacities would have
forced hunter-gatherers to range over large areas, although probably not as large as
those of Early Microlithic bands. This contrasts sharply with relatively dense
rockshelter occupations dating between 7000-5000 B.P. in the Drakensberg and
Winterberg Ranges east and south of the Karoo rim only 150km away from
Blydefontein (Hall 1985: 1-50;
Opperman 1987).
Occupation intensity of Blydefontein Rockshelter dramatically increased at ca.
4300 B.P., and comparable dates have been obtained for the first Interior Wilton
occupations from two rockshelters in the adjacent Zeekoe Valley. Five other
rockshelters in the valley were first occupied at about the same time (Sampson,
personal communication). There can be little doubt that improving conditions
stimulated population growth and possibly even influx at the start of the late Holocene.
Interior Wilton site counts in the adjacent Zeekoe Valley are far in excess of Lockshoek
site numbers, and there is a marked switch in settlement patterns away from springeye focus (Lockshoek pattern) to hillsides near springs (Interior Wilton pattern).
328
Conditions remained favorable except for the three brief drought episodes at ca. 3200
B.P., 2000 B.P. and 1300 B.P. Sustained high carrying capacity led to further
population growth which led in turn to constriction of bands with even smaller
territories. Artifact design strategies responded to these changes in range size and
overall mobility.
Maintenance of projectile armatures shifted from a replace-
before-failure to a fix-when-broken strategy. Also, armatures changed shape and
form with relative rapidity, and there may be some covariance with the short-term
climatic oscillations.
For instance the appearance of pressure-flaked bifacial
projectile points corresponds with the 1300 B.P. drought. After this, there is a
notable decline in the overall frequency of backed microlith production. Many other
factors may be involved here, such as penetrations of ancestral Khoi herders into the
area, and/or the introduction of arrow poison on the bone projectile point. Hafted
endscrapers were gradually replaced by hand-held ones used in a wider range of tasks.
The last several hundred years of these long-term trends in tool design strategies
are scarcely represented at Blydefontein Rockshelter and the sample from Meerkat
Rockshelter is too small to throw much light on the terminal (Smithfield) phase of
occupation. By this time hand-held endscrapers and poison-covered bone arrow
points appear to have become standard equipment. Possibly population constriction
prompted the kinds of social interactions that stimulated more group signals through
artifact design (heavier stylistic loads). This would have involved increased
production of ostrich eggshell water beads, decorated ceramics, shell pendants,
engraved ostrich eggshell water containers, and other perishable items. On the other
hand, many other design changes that culminated in the Smithfield can be traced back
to the beginnings of the Interior Wilton, and these two entities appear to be linked in a
developmental sense. Although no systematic comparison has been made yet, the
differences between the coterminous phase of the Coastal Wilton and the Interior
329
Wilton may be equal to or greater than the differences between the later phases of the
Interior Wilton and the Smithfield. Unless a chronological break between the two can
be established there is no cause at present to regard the Smithfield as a separate
prehistoric entity in the sense of a culture, people, or industry. No break has been
established yet, although the reasons why Blydefontein Rockshelter (and Riversmead
Rockshelter on the middle Orange River) were not occupied in Smithfield times
remains enigmatic.
For the present, the Smithfield is an historically established field
label, best used to describe a terminal phase of Interior Wilton development.
APPENDIX 1
Seasonal Rainfall and Temperature Correlations
The three atmospheric circulation systems that influence the modern climate of
southern Africa, also drive the local climate of the Blydefontein region. A formal proof
of this is presented here in the form of two factor analyses. The data are seasonal mean
temperatures from Grootfontein and seasonal rainfall from Grapevale between 1963
and 1981.
By running two factor analyses each season can be associated with its previous
season. As the Southern Hemisphere, summer is December through February,
December from the previous year was combined with the following year's January and
February to comprise a single season summer. Also, within a single calender year
spring follows winter.
Thus to investigate spring-summer relationships, a previous
year's spring was matched with the following year's summer, fall and winter.
Two factor analyses were run on this data set. The first factor analysis uses a
seasonal progression that begins in summer and ends in spring, here called the
Winter-Spring Factor Analysis (Table 35). The second factor analysis uses a seasonal
progression that begins in spring and ends in winter, and this analysis is called the
Spring-Summer Factor Analysis (Table 36). Both factor analyses produced three
factors.
It is worth noting that droughts (mid 1960s) and wet years (early 1960s and
mid 1970s) are represented (see Figure 17), and thus a reasonably full range of short
term climatic conditions are present in the data.
The type of factor analysis used was a Statview 512+ principal components
varimax oblique rotation method so that the factors would not necessarily be
330
331
independent. Three factors were identified in each run, and the factor loadings for each
season's temperature and rainfall are presented in Tables 35 and 36. The factor
loadings for each factor can be used to identify the variables that contribute the most to
the patterns isolated by the factor analysis. If a variable's loading was 0.5 or greater,
then that variable was considered a significant component of the factor. All six factors
represent bipolar relationships, i.e. variables with high positive and high negative
loadings. If a factor has high loadings in seasons that are not adjacent (i.e. summer and
spring in Table 35, or winter and spring in Table 36), then that factor is not
considered further. In the second factor analysis, Factor Ill has high loadings only on
winter temperature and rainfall, but in the first factor analysis, Factor II indicates
that winter temperature and rain, and the following spring temperatures comprise a
more complete association. A full complement of factors are obtainable by using Factor
I and Factor II from the Spring-Summer factor analysis (Table 35), and Factor II from
the Winter-Spring factor analysis (Table 36).
There are obvious relationships between temperature and rainfall within the same
season caused by the interplay between rainfall, cloud cover and temperature, but
these analyses suggest that different, more complex types of interactions are in effect.
Winter-Spring Factor 2 has high positive loadings for winter and spring
temperatures, while winter rainfall has a high negative loading (Table 35).
Temperature and rainfall are inversely related, and in this pattern as winter rains
increase winter and spring temperatures decrease or vice versa. In Spring-Summer
Factor I, fall temperature has high negative loadings, and summer and fall rainfall have
high positive loadings. Thus as summer and fall rains increase (decrease), fall
temperature decrease (increase).
In Spring-Summer Factor II, spring and summer
temperatures have high positive loadings, and spring rainfall has a high negative
332
loading. Spring temperatures appear to be the most complex variable as it is
controlled by two factors (Winter-Spring Factor II and Spring-Summer Factor II).
It can be argued that two summer patterns are represented by Spring-Summer
Factor I and Factor II, and that Winter-Spring Factor II represents a winter pattern.
Spring-Summer Factor I may be linked to increased strength of the South East Trades,
and Spring-Summer Factor II might correspond to variations in convectional and
thunderstorm activity.
Correlations of all three factors with annual rainfall and
average temperature (Table 37) show that Spring-Summer Factor I has the strongest
correlation with annual rainfall, but Spring-Summer Factor II has no clear
Table 35.--The Winter-Spring Factor Analysis. Factor loadings and proportion of a
individual variable's variance (communality) explained by factor analysis.
Factor
Analysis uses Principal Components and varimax oblique rotation. Also listed are
proportion of common or explained variance for each factor, and variable complexity.
Variable complexity indicates how many factors explain a variable's communality
Variable
seasQn
FaQtQr 1*
summer temp.-0.074
FaQtQr 2
FaQtQr 3*
QQmmunalil)£ QQmQiexit)£
0.122
-Q.9Q6Q
0.816
1.050
-Q. 8 9 8
-0.120
0.0004
0.848
1.036
winter temp. -0.4 9 4
Q.628
0.0040
0.562
1.895
spring temp.
0.091
Q.72Q
-0.3930
0.707
1.584
summer rain
Q.789
-0.310
Q.579Q
0.841
2.1 01
fall rain
Q.869
-0.167
0.1970
0.729
1.180
winter rain
0.025
-Q. 91 Q
-0.0930
0.834
1.022
spring rain
Q.533
0.231
Q.646Q
0.674
2.209
Proportion
0.425
of Common
Variance
0.304
0.276
fall temp.
* Factor not used because significant seasons are not adjacent.
333
Table 36.--The Spring-Summer Factor Analysis
Variable
QQmmunalit)l
qュセャクゥエI@
ヲャセqNsョ@
FaQtQr 1
FaQtQr 2
spring temp.
0.396
0.536
-0.108
0.451
1.935
summer temp.-0.274
0.856
0.117
0.836
1.243
-Q.B25
-0.350
0.154
0.828
1.428
winter temp. - 0. 3 7 7
0.189
-Q. 62Z
0.544
1 .851
-0.069
-Q.BQ5
0.206
0.682
1.146
summer rain Q.B6Q
-0.207
0.163
0.799
1 .191
-0.014
0.209
0.802
1.113
0.808
1 .011
fall temp.
spring rain
fall rain
qNbWセ@
FaQtQr 3*
-0.067
0.001
qNbセT@
0.441
Proportion
of Common
Variance
0.320
0.272
winter rain
* Factor not used because full temporal sequence of seasons not available.
associations. As the greatest contribution to a year's rainfall occurs during the
summer and fall the strong association of annual rainfall with Spring-Summer Factor I
is no surprise. It is also clear that it is the South East Trades that delivers the
moisture to the interior during these seasons (Tyson 1986).
A comparison of factor scores for individual years, and sea surface temperature
(SST) anomalies from the Peruvian coast provide a visual comparison between the
three factors and the Walker Circulation (Table 37}. High SST measurements are
correlated to the Low Phase of the Walker Circulation, when reversed wind directions
reduce the amount of convectional uplift over the interior. The SST measurements
were gleaned directly from Barber and Chavez (1983: Figure 7), and can only be
considered approximate estimates of this phenomenon. The closest correspondence of
SST anomalies was to Winter-Spring Factor II scores (r2
= 0.273),
and it suggests
that as SST temperatures increase so do Winter-Spring Factor II scores (Figure 133).
334
Locally this means that a shift to the Low Phase of the Walker Circulation appears to
cause warmer winters with less rain which are followed by warm springs. Rasmusson
and Wallace (1983) provide information which can be used to suggest that the
mechanism is complex and involves the Westerlies and its semi-stationary meanders.
Comparisons of the SST anomaly, and Spring-Summer Factor I and Spring-Summer
Factor II scores show no associations with the SST anomaly, but it appears that other
climatic elements are associated with those factors as discussed above.
Table 37.--Correlations (r2) between factor scores and climatic variables
Annual Rainfall
-0.130
FaQtQr
Winter-Spring Factor II
Average IemQerature
0.429
SST aョqュ。ャセ@
0.273
Spring-Summer Factor I
0.656
-0.319
-0.079
Spring-Summer Factor II
-0.120
0.341
0.077
DFactor II score
0 SST Anomaly
1.5
VI
Cll
!...
0
0
(J)
!...
0
......
0
.5
0
LL
-.5
tQ
-1
tQ
.,c
::;1\
"";V
E
0
-1.5
c
-<t
-2
1-
-2.5
(J)
(J)
-3
1960
1965
1970
1975
1980
Figure 133.--Factor II scores for each year from the Winter-Spring factor analysis,
and SST anomalies from Peruvian coast.
335
These analyses suggest that two summer systems and a winter system are in
operation. The information presented on general circulation systems in Chapter Ill
can be used to suggest that the summer system represented by Spring-Summer Factor
I is linked to tropical circulation patterns and the Southeastern Trades, while the
winter system is associated with the Walker Circulation. The second summer system
is tentatively linked to convectional activity. Yearly fluctuations of trough and ridge
positions, including the ITCZ, are clearly linked to seasonal changes in solar radiation,
and changes in solar radiation, on an annual or seasonal basis, could alter the position
and/or strength of the circulation systems.
APPENDIX2
BLYDEFONTEIN ROCKSHELTER AND MEERKAT ROCKSHELTER
ARTIFACT TABLES
The tables that comprise this appendix provide the raw artifact data on which the
archaeological analyses are based. The artifacts were classified by Sampson's (1974)
system, and presented by Analysis Unit for each site.
336
Table 38.--Biydefontein Rockshelter, Block B Artifacts.
A. U.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22...._
Total
End
Scra1,2ers
5
2
8
9
18
19
8
10
21
7
16
7
4
1
1
4
1 41
Circular
Side
Scra1,2ers Scr;a1,2ers
Scraper
Frag.s
1
1
-
2
1
1
3
2
4
1
-
-
-
1
2
1
-
6
1
T/U
Flake or
Backed
Outils
Core
Core
Backed Flake or Bladelet
Bladelets Adzes Notches Borers Bladelet w/ Wear escailles Cores Frag. l::!ammers
2
8
18
25
20
16
22
29
10
19
7
-
-
1
2
8
2
2
17
195
1
4
2
1
2
2
1
3
-
2
1
1
1
3
1
3
1
4
1
2
2
1
1
4
2
-
17
18
12
6
3
6
6
6
3
11
5
3
8
3
3
-
-
1
4
2
1
2
6
2
5
3
1
1
1
-
2
-
-
1
2
2
1
2
64
1
10
30
-
3
2
w
w
"'
Table 38.-Continued.
A. U.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Total
Burins
-
1
-
GrindStones
1
1
-
Reamer
-
-
-
-
2
1
1
-
-
-
-
-
Palattes
-
2
-
-
Ceramics
7
10
6
2
-
2
1
-
1
7
-
-
1
-
-
-
-
1
3
25
5
1
1
1
-
-
-
-
-
Ochre
1
1
-
-
-
1
1
Other
Ostrich
Bone Worked Eggshell
Points
Bone
Beads
1
1
1
1
1
1
4
1
3
-
-
-
-
-
-
6
9
2
Whole
Lithic
Debitaae
108
323
439
491
615
500
466
545
521
291
420
302
21
11 7
99
32
24
42
14
8
9
5397
Total
121
349
469
536
683
558
500
601
599
320
483
327
30
130
105
39
28
45
17
8
5969
w
w
(X)
Table 39.--Biydefontein Rockshelter, Blocks C-D Artifacts.
A. U.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
End
Scrapers
3
8
1
7
3
7
6
3
14
4
4
6
9
8
11
17
10
5
8
2
3
1
Side
Transverse Scraper
Scrapers Scrapers Frag.s
1
3
T/U
Flake or
Backed
Backed Flake or Bladelet
Outils
Core
Core
Bladelets Adzes Notches Borers Bladelet w/ Wear escailles Cores Frag. Hammers
1
3
1
4
2
2
3
3
6
6
8
3
11
5
4
2
2
4
6
14
2
7
2
2
1
4
1
12
2
4
1
4
3
5
5
5
8
4
2
2
6
1
3
3
2
2
1
2
7
5
2
12
2
1
2
2
4
1
2
4
1
2
3
4
2
2
2
2
w
(tJ
(0
Table 39.-Continued.
End
A. U. Scrapers
3o
1
31
32
33
34
35
36
37
38
Total
1 43
I _
Side
Transverse Scraper
Scrapers Scrapers Frag.s
1
T/U
Flake or
Backed
Flake or Bladelet
Outils
Core
Core
Bladelets Adzes Notches Borers Bladelet w/ Wear escailles Cores Frag. Hammers
1
2
17
4
32
1 01
14
3
25
9
73
39
3
w
セ@
0
Table 39.-Continued.
A. U.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Bifacial
Points
GrindStones
Painted
Rock
Palattes
-
:
1
1
-
-
-
-
2
-
1
-
-
-
-
-
1
-
-
-
-
1
1
-
-
-
-
3
1
1
-
1
-
3
3
2
1
1
1
1
-
-
Ochre
-
5
14
1
4
1
3
1
-
Ceramics
Ostrich
Other
Bone Worked Eggshell
Bone
Beads
Points
1
1
2
3
2
2
1
1
2
1
2
1
-
-
-
-
1
-
-
-
-
2
1
-
-
-
1
1
-
-
Whole
Lithic
Debitage Total
169
251
59
302
155
434
198
290
334
215
69
69
103
430
280
233
220
241
133
1 51
105
77
21
9
1
23
6
11
19
182
284
65
326
170
466
223
328
368
235
75
79
11 2
482
314
262
259
270
148
176
121
85
27
10
1
26
6
11
21
(..)
NセI@
Table 39.-Continued.
A. U.
30
31
32
33
34
35
36
37
Total
Bifacial
Points
GrindStones
Painted
Rock
-
-
-
-
-
-
1
Palattes
Ceramics
Other
Ostrich
Bone Worked Eggshell
Points
Bone
Beads
Ochre
-
-
-
-
-
-
-
-
2
18
1
1
-
27
-
-
-
-
-
9
18
3
-
2
-
-
Whole
Lithic
Debitage Total
31
13
10
5
16
18
9
21
14
4745
32
14
10
6
22
20
10
22
5289
w
セ@
!\)
Table 40.--Meerkat Rockshelter Artifacts.
A. U.
End
Side
Scragers Scragers
1
1
2
3
3
6
4
2
5
8
6
4
7
2
8
8
9
3
1-
8
11
-
12
4
13
4
Pressure
Scraper Flaked
Backed
Bladelets Adzes Notches
Frag.s
Points
-
1
-
1
1
1
1
3
1
1
-
1
1
1
-
1
1
1
1
Borers
-
Trimmed
Flake or
Blade let
Flake or
Bladelet
wl Wear
Outils
・ウ」。ゥeセ@
Cores
-
2
1
2
2
-
-
2
1
-
-
2
1
1
2
7
1
-
3
4
2
-
2
5
19
1
14
15
6
3
16
-
-
3
17
2
1
3
62
2
2
18
Total
8
2
23
7
4
6
1
9
VJ
.g:,.
VJ
Table 40.-Continued.
GrindStones
A. U.
Palattes
Ceramics
I
Bone
Points
Other
Worked
Bone
3
1
2
1
3
2
5
4
-
5
5
1
6
2
-
3
1
8
1
2
3
55
63
1 01
11 3
124
147
56
65
174
229
5
1
1
33
7
1
-
28
85
129
2
62
74
11
199
232
89
1 02
4
154
1 81
2
85
100
2
211
241
2
123
145
22
24
166
182
79
88
66
76
3
3
1886
2227
8
1
-
1
10
2
3
2
1
11
1
3
-
12
2
3
2
13
7
14
-
15
1
16
-
2
Total
-
1
20
Whole
Lithic
Debitage
1
1
1
Total
:
Mussel
Shell
Pendant
2
9
8
Ostrich
Eggshell
Beads
3
4
7
17
Bone
Beads
48
2
-
-
3
1
2
1
-
11
2
-
1
-
-
1
16
1
94
1
w
""'
""'
BIBLIOGRAPHY
Acocks, A. J. P.
1975 Veld types of South Africa, 2nd Edition. Memoirs of the Botanical Survey of
South Africa No. 40: 1-128.
Adam, D.P. and G. J. West
1983 Temperature and precipitation estimates through the last glacial cycle from
Clear Lake, California, pollen data. Science 219: 168-170.
Ahler, S. A.
1989 Mass analysis of flaking debris: studying the forest rather than the tree. In
Alternative approaches to lithic analysis, D. 0. Henry and G. H. Odell (eds.).
Archaeological Papers of the American Anthropological Association, Number 1:
85-118.
Amrose, S. H. and M. J. DeNiro
1986 Reconstruction of African human diet using bone collagen carbon and
nitrogen isotope ratios. Nature 319: 321-323.
Ammerman, A. J. and M. W. Feldman
1974 On the "making" of an assemblage of stone tools. American Antiquity 39:
610-616.
Avery, D. M.
1988 The Holocene environment of central South Africa: micromammalian
evidence. Paleoecology of Africa 19: 335-345.
Axelrod, D. I. and P. H. Raven
1978 Late Cretaceous and Tertiary vegetation history of Africa. In Biogeography
and ecology of Southern Africa. M. J. A. Werger (ed.). Junk, The Hague.
Bailey, H. P.
1960 A method of determining the warmth and temperateness of climate.
Geografiska Annaler 43: 1-16.
Bamforth, D. B
1986 Technological efficiency and tool curation.
American Antiguity 51: 38-50.
Barber, R. T. and F. P. Chavez
1983 Biological consequences of El Nino. Science 222: 1203-1210.
Barton, C. M.
1990 Beyond style and function: a view from the Middle Paleolithic. American
Anthropologist 92: 57-72.
Berger, A. L.
1978 Long term variations of caloric insolation resulting from the earth's orbital
elements. Quaternary Research 9: 139-167.
1988 Abrupt climatic change. D. Reidel Publishing Company, Dordrecht.
345
Bigalke, R. C.
346
1978 Mammals. In Biogeography and ecology of Southern Africa. M. J. A. Werger
(ed.). Junk, The Hague.
Binford, L. R.
1973 lnterassemblage variability: the Mousterian and the 'functional argument'.
In The explanation of cultural change: models in prehistory, pp 227-254, C.
Renfrew (ed.). Duckworth, London.
1977 Forty-seven trips: a case study in the character of archaeological formation
processes. In Stone tools as cultural markers: change. evolution and
complexity, pp 24-36, R. V. S. Wright (ed.). Australian Institute of Aboriginal
Studies, Canberra.
1978 Nunamiut ethnoarchaeology. Academic Press, New York.
1979 Organization and formation processes: looking at curated technologies.
Journal of Anthropological Research 35: 255-273.
1980 Willow smoke and dog's tails: hunter-gatherer settlement systems and
archaeological site formation. American Antiguity 45: 4-20.
1984 Faunal remains from Klasies River Mouth. Academic Press, New Yorkl
1986 An Alyawara day: making men's knives and beyond. American Antiguity 51:
547-562.
Binford, L. R. and S. R. Binford
1966 A preliminary analysis of functional variability in the Mousterian of
Levallois Facies. American Anthropologist. Special Publication, 68: 2: 2: 238295
Binneman, J. N. F.
1982 Mikrogebruikstekens op steenwerktuie: eksperimentele waarnemings en'n
studie van werktuie van Boomplaasgrot. Unpublished M.A. Thesis, Universiteit
van Stellenbosch.
1983 Microscopic examination of a hafted tool.
Bulletin 38: 93-95.
.sru.tlb. African
Archaeological
1984 Mapping and interpreting wear traces on stone implements: a case study
from Boomplaas Cave. In Frontiers: southern African archaeology today.
Proceeding of the conference of the Southern African Association of
Archaeologists. Gaborone 1983., M. J. Hall, G. Avery, D. M. Avery, M. L. Wilson
and A. J. B. Humphreys (eds.). British Archaeological Reports, Oxford.
Binneman, J. N. F. and J. Deacon
1986 Experimental determination of use wear on stone adzes from Boomplaas
Cave, South Africa. Journal of Archaeological Science 13: 219-228.
Birkeland, P. W.
34 7
1974 Pedology. weathering. and geomorphological research. University of Oxford
Press, New York.
Birks, H. J. B. and H. H. Birks
1980 Quaternary palaeoecology. Edward Arnold, London.
Birks, H. J. B. and A. D. Gordon
1985 Numerical methods in Quaternary pollen analysis. Academic Press, London.
Bleek, D. F.
1933 Belief and customs of the /Xam Bushmen. Part V. The rain. Part VI. Rainmaking. Bantu Studjes 7: 297-312, 375-392.
Bleek, W. H. I. and L. C. Lloyd
1912 Specimens of Bushmen folklore. George Allen & Company, Ltd, London.
Bleed, P.
1986 The optimal design of hunting weapons:
American AntiQuity 51: 737-747.
maintainability or reliability.
Bordes, F.
1961 Typologie du paleolithiQue ancien et moyen. lnstitut de Prehistoire de
I'Universite de Bordeaux, 1-2.
Bordaz, J.
1970 Tools of the old and new stone age. Natural History Press, New York.
Bousman, C. B.
1988 Prehistoric settlement patterns in the Senqunyane Valley, Lesotho.
African Archaeological Bulletin 43: 33-37.
Sm.!.th
1989 Implications of dating the Lockshoek Industry from the interior plateau of
southern Africa. Nyame Akuma 32: 30-34.
nd Interior Wilton population and settlement in the Zeekoe Valley, South Africa.
National Science Foundation, Grant Proposal, not submitted.
Bousman, C. B., T. C. Partridge, L. Scott, S. E. Metcalfe, J. C. Vogel, M. Seaman and J.
S. Brink
1988 Palaeoenvironmental implications of Late Pleistocene and Holocene valley
fills in Blydefontein Basin, Noupoort, C. P., South Africa. Palaeoecology of
Africa, 19: 43-67.
Brain, C. K.
1981 The evolution of man in Africa: was it a consequence of Cainozoic cooling?
Alex L. du Toil Memorial Lecture 17, Transactions of the Geological Society of
South Africa 84: 1-19.
Brink, A. B. A.
1983. Engineering geology of southern Africa. Volume 3. the Karoo sequence.
Building Publications, Pretoria.
348
Bryson, R. A.
1989 Late Quaternary volcanic modulation of Milankovitch climate forcing.
Theoretical and Applied Climatology 39: 115-125.
Bryson, R. A. and J. E. Kutzbach
1974 On the analysis of pollen-climate canonical transfer functions.
Research 4: 162-174.
Quaternary
Bryson, R. A. and T. J. Murray
1977 Climates of hunger. mankind and the world's changing weather. The
University of Wisconsin Press: Madison.
Butler, B. M. and E. E. May
1984 Prehistoric chert exploitation: studies from the midcontinent. Center for
Archaeological Investigations Occasional Paper No. 2, Southern Illinois
University at Carbondale, Carbondale.
Butzer, K. W.
1971 Fine alluvial fills in the Orange and Vaal basins of South Africa. Paper
presented at the Association of American Geographers, Boston.
Campbell, B. M., R. M. Cowling, W. Bond and F. J. Kruger
1981 Structural characterization of vegetation in the Fynbos Biome. South
African National Scientific Programmes Report No. 52.
Cane, M.A.
1983 Oceanographic events during El Nino. Science 222: 1189-1195.
Carter, P. L.
1969 Moshebi's Shelter: excavation and exploitation in eastern Lesotho. Lesotho
Notes & Records 8: 13-23.
Cashdan, E.
1983 Territoriality among human foragers: ecological models and an applications
to four Bushman groups. Current Anthropology 24: 47-66.
1985 Coping with risk reciprocity among the Basara of northern Botswana. Man
20: 222-242.
Cavagnaro, J. B.
1988 Distribution of C3 and C4 grasses at different altitudes in a temperate arid
region of Argentina. Oecologia 76: 273-277.
Clark, J. D.
1959 The prehistory of southern Africa. Pelican Books Ltd., London.
1977 Interpretations of prehistoric technology from ancient Egyptian and other 3 4 9
sources. Part II: prehistoric arrow forms in Africa as shown by surviving
examples of the traditional arrows of the San Bushmen. Paleorient 3: 127150.
Clark, J. D. and H. Kurashina
1981 A study of the work of a modern tanner in Ethiopia and its relevance for
archaeological interpretation. In: Modern material culture: the archaeology of
NlセsL@
R. A. Gould and M. B. Schiffer (eds.}, 303-321, Academic Press, New York.
Clarke, D. L.
1976 Mesolithic Europe: the economic basis. In Problems in economic and social
archaeology, G. Sieveking (ed.}, 449-481, Cambridge University Press,
Cambridge.
CLIMAP Members
1981 Seasonal reconstructions of the Earth's surface at the last glacial maximum.
Geological Society of America, Map Chart Series, MC-36.
Close, A. E.
1978 The identification of style in lithic artefacts. World Archaeology 10: 223237.
1989 Identifying style in stone artefacts: a case study for the Nile Valley. In
Alternative approaches to lithic analysis, D. 0. Henry and G. H. Odel (eds.}.
Archaeological Papers of the American Anthropological Association, Number 1:
3-26.
Close, A. E., F. Wendorf and R Schild
1979 The Afian: a study of stylistic variability in a Nilotic Industry. Department
of Anthropology, Institute for the Study of Earth and Man, Southern Methodist
University, Dallas.
Coe, M. D., D. H. Cummings and J. Phillipson
1976 Biomass and productivity of African large herbivores in relation to rainfall
and primary production. Oecologia 22: 341-354.
Coetzee, J. A.
1967 Pollen analytical studies in east and southern Africa. Palaeoecology of
Africa 3: 1-146.
1978 Climatic and biological changes in south-western Africa during the
Cainozoic. Palaeoecology of Africa 10: 13-29.
1983. Intimations on the Tertiary vegetation of southern Africa. Bothalia 14:
345-354.
1986
Palynological evidence for major vegetation and climatic change in the
Miocene and Pliocene of the southwestern Cape. South African Journal of
Science 82: 71-72.
Coetzee, J. A. and J. Rogers
3 50
1982 Palynological and lithological evidence for the Miocene palaeoenvironments
in the Saldanha region (South Africa). Palaeogeography. Palaeoclimatology and
Palaeoecology 39: 71-85.
COHMAP Members
1988 Climatic changes of the last 18,000 years: observations and model
simulations. Science 241: 1043-1052.
Cole, H. S. and R. A. Bryson
1968 The application of eigenvector technigues to the climatic interpretation of
pollen diagrams: an initial study.l. Bog D Pond. Itasca. Center for Climatic
Research, University of Wisconsin, Madison.
Collins, M. B.
1975 Lithic technology as a means of processual inference. In Lithic technology
making and using stone tools, E. Swanson (ed.), pp15-34. Mouton Publishers,
The Hague.
Conaty, G. T.
1987 Comment on Hayden's resource models of inter-assemblage variability.
Lithic Technology 16: 59-61.
Cowling, R. M.
1983 The occurrence of C3 and C4 grasses in fynbos and allied shrublands in the
south eastern Cape, South Africa. Oecologia 58: 121-127.
Cowling, R. M., P. W. Roux and A. J. H. Pieterse (eds.)
1986 The Karoo biome: a preliminary synthesis. Part 1-physical environment.
South African National Scientific Programmes Report No. 124, Pretoria.
Cowling, R. M. and P. W. Roux (eds.)
1987 The Karoo biome: a preliminary synthesis. Part 2- vegetation and history.
South African National Scientific Programmes Report No. 142, Pretoria.
Davis, M. B.
1963 On the theory of pollen analysis. American Journal of Science 261: 897912.
Deacon, H. J.
1966 Note on the X-ray of two mounted implements from South Africa. Man 1:
87-90.
1972 A review of the post-Pleistocene in South Africa. South African
Archaeological Society. Goodwin Series 1: 26-45.
1976 Where hunters gathered: a study of Holocene stone age people in the Eastern
セN@
South African Archaeological Society Monograph Series, No. 1.
Deacon, H. J. and J. Deacon
3 51
1980 The hafting, function and distribution of small convex scrapers with an
example from Boomplaas Cave. South African Archaeological Bulletin 5: 3137.
Deacon, H. J., J. Deacon, M. Brooker and M. L. Wilson
1978 The evidence for herding at Boomplaas Cave in the southern Cape, South
Africa. South African Archaeological Bulletin 33: 39-65.
Deacon, J.
1972 Wilton: an assessment after fifty years. South African Archaeological
Bulletin 27: 10-48.
1974 Patterning in the radiocarbon dates for the Wilton/Smithfield complex in
southern Africa. South African Archaeological Bulletin 29: 3-18.
1978 Changing patterns in he Late Pleistocene/Early Holocene prehistory of
southern Africa as seen from the Nelson Bay Cave stone artifact sequence.
Quaternary Research 10: 84-111.
1984 Later Stone Age of southernmost Africa. pp 441. BAR International Series
213, Cambridge Monographs in African Archaeology 12, British Archaeological
Reports, Oxford.
Deacon, J. , I. N. Lancaster
1988 Late Quaternary palaeoenvironments of southern Africa. Clarendon Press,
Oxford.
Deacon, J. , N. Lancaster and L. Scott
1984 Evidence for Late Quaternary climatic change in southern Africa. In .L..a1e.
Cainozoic palaeoclimates of the Southern Hemisphere, pp 391-404, J. C. Vogel
(ed.). A. A. Balkema, Rotterdam.
Dibble, H. L.
1987 The interpretation of Middle Paleolithic scraper morphology. American
Antiguity 52: 109-117.
Dunn, E. J.
1880 Stone implements of South Africa. Transactions of the South African
Philosophical Society (1879-1880) 2: 6-22.
1931 The Bushmen. Griffin, London.
Dzurec, R. S., T. W. Boutton, M. M. Caldwell and B. N. Smith
1985 Carbon isotope ratios of soil organic matter and their use in assessing
community compoistion changes in Curlew Valley, Utah. Oceologia 66: 17-24.
Eidt, R. C.
1977 Detection and examination of anthrosols by phosphate analysis. Science
197: 1327-1333.
1985 Theoretical and practical considerations in the analysis of anthrosols. In 3 52
Archaeological geology, G. Rapp, Jr. and J A. Gifford (eds.). Yale University
Press, New Haven.
Ellis, F. and J. J. N. Lambrechts
1986 Soils. In R. M. Cowling, P. W. Roux and A. J. H. Pieterse (eds.) The Karoo
biome: a preliminary synthesis. Part 1- physical environment. South African
National Scientific Programmes Report No. 124.
Ellis, R. P., J. C. Vogel and A. Fuls
1980 Photosynthetic pathways and the geographical distribution of grasses in South
West Africa/Namibia. South African Journal of Science 76: 307-314.
Emiliani, C, S. Gartner, B. Lidz, K. Eldridge, D. K. Elvey, T. -C. Huang, J. J. Stipp and
M. F. Swanson
1975 Paleoclimatological analysis of late Quaternary cores from the northeastern
Gulf of Mexico. Science 189: 1083-1088.
Erdtman, G.
1943 An introduction to pollen analysis. Chronica Botanica, Waltham,
Massachusetts.
Fagan, B. M.
1965 Southern Africa. during the Iron Age. Thames and Hudson, London.
Fall, P.
1987 Pollen taphonomy in a canyon stream. Quaternary Research 28: 393-406.
Faure, G.
1986 Principles of isotope geology. John Wiley & Sons, New York.
Flexor, J. M. and B. Volkoff
1977 Distribution de l'sotope stable 13C dans Ia matiere organqui d'un sol
ferrallitique de l'etat de Bahia (Bresil). Comptes Rendus de Seances de
!'Academia des Sciences. Series D. 234: 307-314.
Fullard H. and H. C. Darby (eds.)
1975 The university atlas. George Philip & Son, Limited, London.
Gallagher, J. P.
1977 Contemporary stone tools in Ethiopia: implications for archaeology. Journal
of Fjeld Archaeology 4: 407-414.
Gates, W. L.
1976 The numerical simulation of ice-age climate with a global general
circulation model. Journal of the Atmospheric Sciences, 33: 1844-1873.
Goodwin, A. J. H.
1945 Some historical bushman arrows. South African Journal of Science 41:
429-443.
Goodwin, A. J. H. and C. van Riet Lowe
353
1929 The Stone Age cultures of South Africa. Annals .Q.{ 1.b..e. SQ.u1h African Museum
27: 1-289.
Gould, R. A.
1977 Ethno-archaeology: or, where do models come from? A closer look at
Australian Aboriginal lithic technology. In Stone tools as cultural markers:
change. evolution and complexity, R. V. S. Wright (ed), pp 162-168.
Prehistory and Material Culture Series No. 12. Australian Institute of
Aboriginal Studies, Canberra, Humanities Press, New Jersey.
1980 Living archaeology. Cambridge University Press, Cambridge.
Gould, R. A., D. A. Koster and A. H. L. Sontz
1971 The lithic assemblage of the Western Desert Aborigines of Australia.
American Antiguity 36: 149-169.
Gould, R. A. and S. Saggars
1985 Lithic procurement in central Australia: a closer look at Binford's idea of
embeddedness in archaeology. Amerjcan Antiguity 50: 117-136.
Gribbin, J.
1978 Astronomical influences, short term affects. In Climatic change, J. Gribbin
(ed.), p. 150-154, Cambridge University Press, Cambridge.
Gribbin, J. and H. H. Lamb
1978 Climatic change in historical times. In Climatic change, J. Gribbin (ed.), p.
114-129, Cambridge University Press, Cambridge.
Grindley, J. R.
1986 Pleistocene rains in the north-western Karoo. South African Journal of
Science 82: 178-179.
Guiot, J.
1987 Late Quaternary climatic change in France estimated from multivariate
pollen time series. Quaternary Research 28: 100-118.
Guiot, J., A. Pons, J. L. de Beaulieu and M. Reille
1989 A 140,000-year continental climate reconstruction from two European
pollen records. Nature 338: 309-313.
Gutsche, T.
1968 The microcosm. Cape Town: Howard Timmins.
Haas, H., V. Holliday and R. Stuckenrath
1986 Dating of Holocene stratigraphy with soluble and insoluble organic fractions
at the Lubbock Lake archaeological site, Texas: an ideal case study. Radiocarbon
28: 2A: 473-485.
Hall, M.
354
1990 Farmers. kings. and traders: the people of southern Africa. 200-1860.
University of Chicago Press, Chicago.
Hall, M. and A. B. Smith
1986 Prehistoric pastoralism in southern Africa. The South African
Archaeological Society, Goodwin Series, Vol. 5.
Hall, S. L.
1985 Edgehill and Welgeluk. In Guide to archaeological sites in the eastern and
north eastern Cape, pp 1-38, S. L. Hall and J. N. F. Binneman (eds.), Prepared
for The southern African Association of Archaeologists Excursion, September
18th-21st 1985.
Hart, T. J. G.
1989 Haaskraal and Volstruisfontein, Later Stone Age Events at two rockshelters
in the Zeekoe Valley, Great Karoo, South Africa. Unpublished M. A. Thesis,
University of Cape Town.
Hayden, B.
1979 Palaeolithic reflections: lithic technology and ethnographic excavation
among Australian Aborigines. Australian Institute of Aboriginal Studies,
Canberra. Humanities Press, New Jersey.
1986 Resource models of inter-assemblage variability.
82-89.
Lithic Technology 15: 3:
Hays, J. D., J. Imbrie and N. J. Shackleton
1976 Variations in the earth's orbit: pacemaker of the ice ages. Science 194:
1121-1132.
Heaton, T. H. E.
1987 The 15N/14N ratios of plants in South Africa and Namibia: relationship to
climate and coastal/saline environments. Oecologia 74: 236-246.
Heaton, T. H. E., J. C. Vogel, G. von Ia Chevallerie and G. Collett
1986 Climatic influence on the composition of bone nitrogen. Nature 322: 822823.
Helgren, D. M.
1979 Rivers of diamonds: an alluvial history of the lower Vaal basin. South Africa.
University of Chicago, Department of Geography Research Paper 18.
Henry, D. 0.
1989 Correlations between reduction strategies and settlements patterns. In
Alternative approaches to lithic analysis, D. 0. Henry and G. H. Odel (eds.).
Archaeological Papers of the American Anthropological Association, Number 1:
139-155.
Hester, J. J. and J. Grady
35 5
1977 Paleoindian social patterns on the Llano Estacada. Texas Tech University.
The Museum Journal 17: 78-96.
Hillaire-Marcel, C., A-M Aucour, R. Bonnefille, G. Riollet, A. Vincens and D.
Williamson
1989 13C/palynological evidence of differential residence times of organic carbon
prior to its sedimentation in East African Rift lakes and peat bogs. Quaternary
Science Reviews 8: 207-212.
Hitchcock, R. L.
1982 The ethnoarchaeology of sedentism: mobility strategies and site structure
among foraging and food producing populations in the eastern Kalahari Desert,
Botswana. Unpublished Ph.D. Dissertation, Albuqurque: University of New
Mexico.
Hodder, I.
1979 Economic and social stress and material culture patterning. American
AntiQuity 44: 446-454.
Holloway, R. G.
1989 Experimental mechanical pollen degradation and its application to
Quaternary age deposits. The Texas Journal of Science 41: 131-145.
Hooghiemstra, H.
1984 Vegetational and climatic history of the high plain of Bogota. Columbia: a
continuous record of the last 3.5 million years. J. Cramer, Hisrschberg,
Germany.
Huesser, C. J., L. E. Heusser and S. S. Streeter
1980 Quaternary temperatures and precipitation for the north-west coast of
North America. Nature 286: 702-704
Huesser, C. J. and S. S. Streeter
1980 A temperature and precipitation record of the past 16,000 years in
southern Chile. Science 210: 1345-1347.
Humphreys, A.J.B.
1969 Four bifacial tanged and barbed arrowheads from Vosberg. South African
Archaeological Bulletin 24: 72-74.
1972 Comments on aspects of raw material usage in the Later Stone Age of the
Middle Orange River area. South African Archaeological Society Goodwin Series
1: 46-53.
1979 The Holocene sequence in the Northern Cape and its position in the
prehistory of South Africa. Unpublished Ph.D. Thesis, University of Cape
Town.
Humphreys, A. J. B. and A. I. Thackeray
356
1983 Ghaap and Gariep: Later Stone Age studies in the northern Cape. Cape Town:
The South African Archaeological Society Monograph Series No.2.
Hurry, L. and J. Van Heerden
1984 Southern Africa's weather patterns. a guide to the interpretation of synoptic
セM
Via Afrika Limited, Goodwood.
Imbrie, J. and K. P. Imbrie
1979 Ice ages. solving the mystery. Harvard University Press, Cambridge.
Jacobson, L.
1987 The size variability of ostrich eggshell beads from central Namibia and its
relevance as a stylistic and temporal marker. The South African Archaeological
Bulletin 42: 55-58.
Johnson, J. K. and C. A. Morrow
1987 The organization of core technology. Westview Press, Boulder, Colorado.
Jones, G. A. and W. F. Ruddiman
1982 Assessing the global meltwater spike. Quaternary Research 17: 148-172.
Kannemeyer, D. R.
1890 Stone implements of the Bushmen. Cape Illustrated Magazine 1: 120-130.
Keeley, L. H.
1980 Experimental determination of stone tool uses: a microwear analysis.
University of Chicago Press, Chicago.
1982 Hafting and retooling: effects on the archaeological record. American
Antiquity 47: 798-809.
1991 Tool use and spatial patterns: complications and solution. In The
interpretation of archaeological spatial patterning, E. M. Kroll and T. D. Price
(eds.), pp 257-268. Plenum Press, New York.
Keeley, L. and M. H. Newcomer
1977 Microwear analysis of experimental flint tools: a test case. Journal of
Archaeological Science 4: 29-62.
Kent, L. E. and K. -H. Gribnitz
1985 Freshwater shell deposits in the northwestern Cape Province: further
evidence for a widespread wet phase during the late Pleistocene in southern
Africa. South African Journal of Science, 81: 361-370.
Kelly, R. L.
1980 Hunter-gatherer settlement systems. Unpublished M. A. Thesis, Albuqurque:
University of New Mexico.
1983 Hunter-gatherer mobility strategies. Journal of Anthropological Research
39: 277-306.
1988 The three sides of a biface. American AntiQuity 53: 717-734.
357
Killick, D. J. B.
1978 The afroalpine region. In Biogeography and ecology of Southern Africa. M. J.
A. Werger (ed.). Junk, The Hague.
King, L.
1942 South Africa scenery. Oliver and Boyd, Edinburgh.
1978 The geomorphology of central and southern Africa. In Biogeography and
ecology of Southern Africa. M. J. A. Werger (ed.). Junk, The Hague.
Klein, R. G.
1974 Environment and subsistence of prehistoric man in the southern Cape
Provence, South Africa. World Archaeology 5: 149-184.
1979 Paleoenvironmental and cultural implications of Late Holocene
archaeological faunas from the Orange free State and north-central Cape
Province, South Africa. South African Archaeological Bulletin 34: 34-49.
1980 Environmental and ecological implications of large mammals from Upper
Pleistocene and Holocene sites in southern Africa. Annals of the South African
Museum, 81: 7: 223-283.
1984 Mammal extinctions and stone age people in Africa. In Quaternary
extinctions. a prehistoric revolution, P. S. Martin and R. G. Klein (eds.).
University of Arizona Press, Tucson.
Klein, R. G. and K. Cruz-Uribe
1989 Faunal evidence for prehistoric herder-forager activities at Kasteelberg,
western Cape Province, South Africa. South African Archaeological Bulletin
44: 82-97.
Kuhn, S. L.
1989 Hunter-gatherer foraging organization and strategies of artifact replacement
and discard. In Experiments in lithic technology, Daniel S. Amick and Raymond
P. Mauldin (eds.). BAR International Series 528. British Archaeological
Reports, Oxford.
1990 A geometric index of reduction for unifacial stone tools. Journal of
Archaeological Sciences 17: 583-593.
Kukla, G. J.
1978 Recent changes in snow and ice. In Climatic change, J. Gribbin (ed.), p.
114-129, Cambridge University Press, Cambridge.
Kutzbach, J. E.
1981 Monsoon climate of the Early Holocene: climate experiment with the Earth's
orbital parameters for 9000 years ago. Science, 214: 59-61.
Kutzbach, J. E. and P. J. Guetter
358
1984 The sensitivity of monsoon climates to orbital parameter changes for 9000
years BP: experiments with the NCAR General Circulation Model. In
Milankovitch and Climate, A. L. Berger, J. Imbrie, J. Hays, G. Kukla and B.
Saltzman (eds.), Part 2, p. 801-820, D. Reidel Publishing Company,
Dordrecht.
1986 The influence of changing orbital parameters and surface boundary
conditions on climate simulations for the past 18000 years. Journal of the
Atmospheric Sciences, 43: 1726-1759.
Kutzbach, J. E. and F. A. Street-Parrott
1985 Milankovitch forcing of fluctuations in the level of tropical lakes from 18 to
0 kyr BP. Nature 317:130-134.
Lancaster, I. N.
1979 Evidence for a widespread late Pleistocene humid period in the Kalahari.
Nature 279: 145-146.
Lee, R. B.
1979 The !Kung San: men. women. and work in a foraging society. Cambridge
University Press, Cambridge.
Leroi-Gourhan, A.
1943 L'homme et Ia matere.
Albin-Michel, Paris.
Levyns, R.M.
1964 The evolution of Capensis flora. In Biogeography and ecology of Southern
Africa. M. J. A. Werger (ed.). Junk, The Hague.
Lewis-Williams, J. D.
1981 Believing and seeing: symbolic meanings in southern San rock art. Academic
Press, London.
Lindquist, C. A. and P. L. Fall
1987 Fossil hyrax middens from the Middle East - a new source of palaeobotanical
evidence. International Union for Quaternary Research, XII International
Congress, Ottawa, July 1987, Programme with Abstracts. National Research
Council of Canada, p. 212.
MacCalman, H. R. and B. J. Grobbelaar
1965 Preliminary report of two stone-working OvaTjimba groups in the northern
Kaokoveld of South West Africa. Cimbebasia 134: 1-39.
MacVicar, C. N., R. F. Loxton, J. J. N. Lambrechts, J. Le Roux, J. M. de Villiers, E.
Verster, F. R. Merryweather, T. H. Van Rooyen and J. J. Von M. Harmse
1977 Soil classification: a binomial system for South Africa. Department of
Agricultural Technical Services. Science Bulletin 390, Pretoria.
Magilliagan, F. J.
35 9
1985 Historical floodplain sedimentation in the Galena River basin, Wisconsin and
Illinois. Annals of the Association of American Geographers 75: 583-594.
Manabe, S. and D. G. Hahn
1977 Simulation of the tropical climate of an ice age. Journal of Geophysical
Resources, 82: 3889-3991.
Manhire, A.
1987 Later Stone Age settlement patterns in the Sandveld of the south-western
Cape Province. South Africa. Cambridge Monographs in African Archaeology
21, British Archaeological Reports, Oxford.
Marks, A. E.
1988 The curation of stone tools during the Upper Pleistocene, a view from the
central Negev Desert, Israel, In Upper Pleistocene prehistory of western
Eurasia, H. L. Dibble and A Montet-White (eds.), pp. 275-285. University
Museum Symposium Series No. 1, University Museum Monographs 54,
University of Pennsylvania, Philadelphia.
Marks, A. E. and D. A. Freidel
1977 Prehistoric settlement patterns in the Avdat/Aqev area. In Prehistory and
paleoenvironments in the central Negev. Israel. Volume II. The Avdat/Agev
area. Part 2 and the Har Harif, pp 131-158, A. E. Marks (ed.), Southern
Methodist University Press, Dallas.
Marshall, L.
1976 The !Kung of the Nyae Nyae. Harvard University Press, Cambridge.
Martin, A. K.
1981 Evolution of the Agulhas Current and its palaeo-ecological implications.
South African Journal of Science 77: 547-554.
Mazel, Aron David
1989 People making history: the last ten thousand years of hunter-gatherer
communities in the Thukela Basin. Natal Museum Journal of Humanities 1: 1168.
Mazel, A. and J. E. Parkington
1978 Sandy Bay revisited: variability amongst Late Stone Age tools. South African
Journal of Science 74: 381-382.
1981 Stone tools and resources: a case study from southern Africa. WQ.d.d
Archaeology 13: 16-30.
McAnany, P. A.
1988 Effect of lithic procurement strategies on tool curation and recycling. .Li1hi.Q
Technology 17:
Meadows, M. E., K. F. Meadows and J. M. Sugden
360
1987 The development of vegetation on the Winterberg escarpment. The Naturalist
31: 1: 26-32.
Meadows, M. E. and J. M. Sugden
1988 Late Quaternary environmental changes in the Karoo, South Africa. In G. F.
Dardis and B. P. Moon (eds.) Geomorphological studies in southern Africa. A. A.
Balkema, Rotterdam.
Milankovitch, M.
1930 Mathematische klimalehre und astronomische theorie der
klimaschwankungen. In W. Koppen and R Geiger (eds.) Handbuch der
klimatologie. I (A) GebrOder Borntraeger, Berlin.
Miller, T. 0., Jr.
1979 Stonework of the Xeta Indians of Brazil. In Lithic use-wear analysis, B.
Hayden (ed.), pp. 401-407, Academic Press, New York.
Mitchell, P. J.
1988 The Early Microlithic assemblages of southern Africa.
388, British Archaeological Reports, Oxford.
International Series
Mooney, H. A., J. H. Troughton and J. A. Berry.
1977 Carbon isotope ratio measurements of succulent plants in southern Africa.
Oecologia 30: 295-305.
Morley, J. J. and J. D. Hays
1979 Comparison of glacial and interglacial oceanographic conditions in the South
Atlantic from variations in calcium carbonate and radiolarian distributions.
Quaternary Research 12: 396-408.
Moss, E. H. and M. H. Newcomer
1982 Reconstruction of tool use at Pincevent: microwear and experiments. Studia
Praehistorica Belgica 222: 289-312.
Movius, H. L., Jr., N. C. David, H. M. Bricker and R. B. Clay
1968 The analysis of certain major classes of Upper Palaeolithic tools. American
School of Prehistoric Research, Peabody Museum, Harvard University, Bulletin
No. 26.
Murdock, G. P.
1967 Ethnographic atlas: a summary. Ethnology 6: 109-236.
Muto, G.
1971 A stage analysis of the manufacture of stone tools. In Great Basin
Anthropolgical Conference 1970: selected papers, C. M. Aikens (ed.), pp. 109118. Anthropological Papers 1, University of Oregon, Eugene.
Natelhoffer, K. J. and B. Fry
1988 Controls on natural Nitrogen-15 and Carbon-13 abundances in forest soil
organic matter. Soil Science Society of America Journal 52: 1633-1640.
Opperman, H.
361
1987 The Later Stone Age of the Drakensberg Range and its foothills. Cambrigde
Monographs in African Archaeology 19, BAR International Series 339. British
Archaeological Reports, Oxford.
Osgood, C.
1940 lngalik material culture. Publications in Anthropology, No. 22. Yale
University, New Haven.
Parkington, J. E.
1972 Seasonal mobility in the Late Stone Age. African Studies 31: 223-243.
1976 Follow the San: an analysis of seasonality in the prehistory of the southwestern Cape, South Africa. Unpublished Ph.D. Thesis, University of
Cambridge.
1984 Changing views of the Later Stone Age of South Africa. In Advances in World
Archaeology. Volume 3, pp 89-142, F. D. Wendorf and A. Close (eds.).
Academic Press, New York.
Parry, W. J. and B. L. Kelly
1987 Expedient core technology and sedentism. In The organization of core
technology, J. K. Johnson and C. A. Morrow (eds.), pp. 285-304. Westview
Press, Boulder, Colorado.
Patton, P. C. and S. A. Schumm
1981 Ephemeral-stream processes: implications for studies of Quaternary valley
fills. Quaternary Research, 15: 24-43.
Pearson, G. W. and M. Stuiver
1986 High-precision calibration of the radiocarbon time scale, 500-2500 BC.
Radiocarbon. Calibration Issue, 28: 2B: 839-862.
Pimentel, B. A., and J. D. Smith
1985 BIOSTAT II. Macintosh version. Sigma Soft, Placentia, California.
Pittock, A. B.
1978 An overview. In Climatic change and variability: a southern perspective, A.
B. Pittock, L. A. Frakes, D. Jenssen, J. A. Peterson and J. W. Zillman (eds.), p.
1-8. Cambridge University Press, Cambridge.
Pons, A. and P. Quezel
1958 Premieres remarques sur l'etude palynologique d'un guano fossile du Haggar.
Comptes Rendus de Seances de !'Academia des Sciences 246: 2290-2292.
Rasmusson, E. M. and J. M. Wallace
1983 Meteorological aspects of the El Nino/Southern Oscillation. Science 222:
1195-1202.
Rebelo, A. G.
1987 A preliminary synthesis of pollination biology in the Cape flora. South
African National Scientific Programmes Report, No. 141, Pretoria.
362
Rightmire, G. P. and N.J. van der Merwe
1976 Two burials from Phalaborwa and the association of race and culture in the
Iron Age of southern Africa. South African Archaeological Bulletin 31:73-112.
Rosenzweig, M. L.
1968 Net primary productivity of terrestrial communities: prediction from
climatological data. The American Naturalist 102: 67-74.
Raux, P. W. and C. D. Blom
1979 Vegetation surveys in the Karoo region, quantitative data. Unpublished
manuscript, Grootfontein Agricultural College, Middelburg, South Africa.
Raux, P. W. and D.P. J. Opperman
1986 Soil erosion. In R. M. Cowling, P. W. Raux and A. J. H. Pieterse (eds.) The
karoo biome: a preliminary synthesis. Part !-physical environment. South
African National Scientific Programmes Report No. 124, Pretoria.
Raux, P. W. & M. Vorster
1983 Vegetation change in the Karoo. Proceedings of the Grassland Society of
southern Africa. 18: 25-29.
Rudner, J.
1979 The use of stone artefacts and pottery among the Khoisan peoples in historic
and protohistoric times. South African Archaeological Bulletin 34: 3-17.
Rudner, I. and J. Rudner
1954 A local Later Stone Age development. South African Archaeological Bulletin
9: 103-107.
Rutherford, M. C. and R. H. Westfall
1986 Biomes of southern Africa - an objective categorization. Memoirs of the
Botanical Survey of South Africa, Botanical Research Institute, Pretoria.
Sackett, J. R.
1966 Quantitative analysis of Upper Paleolithic stone tools. American
Anthropologist 68: 365-394.
1982 Approaches to style in lithic archaeology. Journal of Anthropological
Archaeology 1: 59-112.
1985 Style and ethnicity in the Kalahari: a reply to Wiessner. American
Antiguity 50: 154-159.
1989 Statistics, attributes and the dynamics of burin typology. In Alternative
approaches to lithic analysis, D. 0. Henry and G. H. Odel (eds.). Archaeological
Papers of the American Anthropological Association, Number 1: 51-82.
Sampson, C. G.
1967a Excavations at Glen Elliot Shelter, Colesberg District, northern Cape.
Memoirs Q.f 1!:lli National Museum, No. 2, Bloemfontein.
363
1967b Excavations at Zaayfontein Shelter, Norvalspont, northern Cape.
Researches of the Nasionale Museum, 2: 4: 41-124, Bloemfontein.
1970 The Smithfield industrial complex: further field results.
National Museum, No. 5, Bloemfontein.
Memoirs Q! 1.tlii
1972 The Stone Age industries of the Orange River Scheme and South Africa.
Memoirs Q.f 1!:lli National Museum, No. 6, Bloemfontein.
1974
セ@
.stQ.n.e. Ag.a archaeology Qf southern AfriQ.a. Academic Press, New York.
1985 Atlas of Stone Age settlement in the central and upper Seacow Valley.
Memoirs Q.f 1!:lli National Museum, No. 18, Bloemfontein.
1987 European impacts on hunter-forager spatial organization: archaeological
and ethnohistorical enquiries. Proposal for the National Science Foundation.
1988 Stylistic boundaries among mobile hunter-foragers. Smithsonian
Institution Press, Washington.
Sampson, C. G. and B. Bousman
1985 Variations in the size of archaeological surface sites attributed to the Seacow
River Bushmen. SQ..u1b. African Journal Q.f Science 81: 321-323.
Sampson, C. G., T. J. G. Hart, D. L. Wallsmith and J. D. Blagg
1989 The ceramic sequence in the upper Seacow Valley: problems and
implications. South African Archaeological Bulletin 44: 3-16.
Sampson, C. G. and Sampson, M.
1967 Riversmead Shelter: excavations and analysis. Memoirs Q.f 1.tlii National
Museum, No. 3, Bloemfontein.
Schapera, I.
1930 The Khoisan peoples of South Africa. George Routledge and Sons, Ltd, London.
Schiffer, M. B.
1975 Some further comments on the Dalton settlement pattern hypothesis. In The
Cache River archeological project. an experiment in contract archeology, 103112, M. B. Schiffer and J. H. House (eds.). Arkansas Archeological Survey,
Research Series No. 8, Fayetteville.
Schneider, Stephen H.
1987 Climate modeling. Scientific American 256: 5: 72-80.
Schulze, R. E. and 0. S. McGee
1978 Climatic indices and classifications in relation to the biogeography of
southern Africa. In Biogeography and ecology of Southern Africa. M. J. A.
Werger (ed.). Junk, The Hague.
364
Schweitzer, F. R.
1979 Excavations at Die Kelders, Cape Province, South Africa, the Holocene
deposits. Annals of the South African Museum, 78: 10: 101-233.
Schweitzer, F. R. and K. Scott
1973 Early occurrence of domestic sheep in sub-Saharan Africa. Nature 241:
547.
Scott, L.
1984 Palynological evidence for Quaternary paleoenvironments in southern
Africa. In Southern African prehistory and paleoenvironments, R. G. Klein
(ed.). A. A. Balkema, Rotterdam.
1988a Hyrax (Procaviidae) and dassie-rat (Petromuridae) middens in
paleoenvironmental studies in Africa. In P. S. Martin et al. (eds.) セ@
packrat middens: the last 40 000 years of biotic change. University of Arizona
Press, Tucson.
1988b Holocene environmental change at western Orange Free State pans, South
Africa, inferred from pollen analysis. Palaeoecology of Africa, 19: 109-118.
Scott, L. and C. B. Bousman
1990 Analysis of hyrax middens from Africa. Palaeogeography. Palaeoclimatology
and Palaeoecology 76: 367-379.
Scott, L. and B. Cooremans
1990 Late Quaternary pollen from a hot spring in the upper Orange River Basin,
South Africa. South African Journal of Science 86: 154-156.
Scott, L. and J. F. Thackeray
1987 Multivariate analysis of late Pleistocene and Holocene pollen spectra from
Wonderkrater, Transvaal, South Africa. South African Journal of Science 83:
93-98.
Sealy, J. C.
1986 Stable carbon isotopes and prehistoric diets in the south-western Cape
Province. South Africa. Cambridge Monographs in African Archaeology 15,
British Archaeological Reports, Oxford.
1989 Reconstruction of Later Stone Age diets in the south-western Cape, South
Africa: evaluation and application of five isotopic and trace element techniques.
Unpublished D. Phil. Thesis, University of Cape Town.
Sealy, J. C. and N.J. van der Merwe
1985 Isotope assessment of Holocene human diets in the southwest Cape, South
Africa. Nature 315: 138-140.
365
1986 Isotope assessment and the seasonal-mobility hypothesis in the southwestern
Cape of South Africa. Current Anthropology 27:135-150.
1987 Carbon isotopes, Later Stone Age diets and seasonal mobility in the southwestern Cape. In Papers in the Prehistory of the Western Cape. South Africa.
J. Parkington and M. Hall (eds.), British Archaeological Reports International
Series 332, British Archaeological Reports, Oxford.
1988 Social, spatial and chronological patterning in marine food use as determined
by d13C measurements of Holocene human skeletons from the south-western
Cape, South Africa. World Archaeology 20: 87-102.
Seely, M. K.
1978 Grassland productivity: the desert end of the curve. South African Journal of
Science 14: 295-297.
Shackleton, N.J. and J.P. Kennett
1975 Palaeotemperature history of the Cainozoic and the initiation of the Antarctic
glaciation: oxygen and carbon isotope analysis in DSDP sites 277, 279 and 281.
In Initial reports of the DSDP. Leg 29: 743-755, J. P. Kennett et. al. (ed.). U.
S. Government Printing Office, Washington.
Shackley, M. and H. Kerr
1985 Ethnography and experiment in the interpretation of quartz artifact
assemblages from Namibia: an optimistic attempt. Lithic Technology 14: 9597.
Shott, M. J.
1986 Technological organization and settlement mobility: an ethnographic
examination. Journal of Anthropological Research 42: 15-51.
1989 On tool-class use lives and the formation of archaeological assemblages.
American Antiquity 54: 9-30.
Silberbauer, G. B.
1981 Hunter and habitat in the central Kalahari Desert. Cambridge University
Press, Cambridge.
Sillitoe, P.
1982 The lithic technology of a Papua New Guinea Highland people. The Artefact 7:
3-4: 19-38.
Simmons, I. G., G. W. Dimbleby and C. Grigson
·1981 The Mesolithic. In The environment in British prehistory, I. G. Simmons
and M. J. Tooley (eds.), p. 82-124. Cornell University Press, Ithaca.
Sinclair, I.
1984 Field guide to the birds of southern Africa. C. Struik Publishers, Cape Town.
Skead, C. J.
366
1987 Historical mammal incidence in the Cape Province. Volume 2: the eastern
half of the Cape Province. including the Cskei. Transkej and East Grigualand.
The Chief Directorate Nature and Environmental Conservation of the Provincial
Administration of the Cape of Good Hope, Cape Town.
Squires, V. R & W. S. W. Trollope
1979 Allelopathy in the Karoo shrub, Chrysocoma tenuifolia. South African
Journal of Science, 75: 88-89.
Stow, George W.
1905 The native races of South Africa. A history of the intrusion of the Hottentots
and Bantu into the hunting grounds of the Bushmen. the aborigines of the
country. Swan Sonnenschein & Co., Limited, London.
Strathern, M.
1969 Stone axes and flake tools: evaluations from two New Guinea Highland
societies. Proceedings of the Prehistoric Society 35: 311-329.
Stuiver, M.
1975 Climate versus changes in 13c content of the organic component of lake
sediments during the Late Quaternary. Quaternary Research 5: 251-262.
Stuiver, M. and G. W. Pearson
1986 High-precision calibration of the radiocarbon time scale, AD1950-500 BC.
Radiocarbon. Calibration Issue. 28: 2B: 805-838.
Sugden, J. M. and M. E. Meadows
1989 The use of multiple discriminant analysis in reconstructing recent
vegetation changes on the Nuweveldberg, South Africa. Review of Palaeobotany
and Palynology 60: 131-147.
Swart, J., M. R. Perrin, J. W. Hearne and L. J. Fourie
1986 Mathematical model of the interaction between rock hyrax and caracal lynx,
based on demographic data from populations in the Mountain Zebra National
Park. South Africa. South Africa Journal of Science 82: 289-294.
Talma, A. S., J. C. Vogel and T. C. Partridge
1974 Isotopic contents of some Transvaal speleothems and their palaeoclimatic
significance. South African Journal of Science 70: 135-140.
Taylor, H. C.
1978 Capensis. In Biogeography and ecology of Southern Africa. M. J. A. Werger
(ed.). Junk, The Hague.
Taylor, R. E.
1987 Radiocarbon dating. an archaeological perspective. Academic Press, Orlando.
Thackeray, A. I.
1981 The Holocene cultural sequence in the northern Cape Province, South Africa.
Unpublished Ph.D. Dissertation, Yale University.
Tieszen, L. L., M. M. Senyimba, S. K. lmbamba and J. H. Troughton
367
1979 The distribution of C3 and C4 grasses and carbon isotope discrimination along
an altitudinal and moisture gradient in Kenya. Oecologia 37: 337-350.
Tixer, J.
1963 Typologie de I'Epipaleolithigue du Maghreb. Centre de Recherches
Anthropologiques, Prehistoriques et Ethnographiques Memoir 2, Alger.
Tomka, S. A.
1990 Models of late Holocene adaptation in the lower Rio Grande valley. In
Prehistoric archeology and paleoenvironments in Hidalgo and Willacy Counties.
south Texas: results of Phase II test excavations, C. B. Bousman, S. A. Tomka and
G. L. Bailey (eds.), pp. 37-73. Reports of Investigations No. 76, Austin:
Prewitt and Associates, Inc.
Torrence, R.
1983 Time budgeting and hunter-gatherer technology. In Hunter-gatherer
economy in prehistory: a European perspective, pp. 11-22, G. Bailey (ed.).
Cambridge University Press, Cambridge.
1989 Time. energy and stone tools. Cambridge University Press, Cambridge.
Trewartha, G.T.
1968 An introduction to climate. McGraw-Hill Book Company, New York.
Tyson, P.O.
1986 Climatic change and variability in southern Africa.
Press, Cape Town.
Oxford University
van der Merwe, N. J.
1982 Carbon isotopes, photosynthesis, and archaeology. American Scientist 70:
596-606.
van der Merwe, N.J. and J. C. Vogel
1983 Recent carbon isotope research and its implications for African archaeology.
The African Archaeological Review 1: 33-56.
van Riet Lowe, C.
1936 The Smithfield "N" culture. Transactions Qf 1M. BQy,aJ. Society Qf セ@
23: 367-372.
Af.d.Qa
van Zinderen Bakker, E. M.
1957 A pollen analytic investigation of the Florisbad deposits (South Africa). In J.
D. Clark (ed.) Proceedings of the 3rd Panafrican congress on prehistory
(Livingstone, 1955). Chatto & Windhus, London.
1967 Upper Pleistocene and Holocene stratigraphy and ecology on the basis of
vegetation changes in Sub-Saharan Africa. In Background to evolution in Africa,
W. W. Bishop and J. D. Clark (eds.). The University of Chicago Press, Chicago.
1982a Pollen analytic studies of the Wonderwerk Cave, South Africa. Pollen et 3 6 8
Spores XXIV: 235-250.
Vehik, Susan C. (ed.)
1985 Lithic resource procurement: proceedings from the second conference on
prehistoric chert exploitation. Southern Illinois University at Carbondale,
Center for Archaeological Investigations, Occasional Paper No. 4.
Venter, J. M., C. Mocke and J. M. De Jager
1986 Climate. In The Karoo biome: a preliminary synthesis. Part 1- physical
environment, R. M. Cowling, P. W. Roux and A. J. H. Pieterse (eds.). South
African National Scientific Programmes Report No. 124, Pretoria.
Vinnicombe, P.
1976 People of the eland.
Natal University Press, Pietermaritzburg.
Visser, J. N. J.
1986 Geology. In The Karoo biome: a preliminary synthesis. Part 1- physical
environment, R. M. Cowling, P. W. Roux and A. J. H. Pieterse (eds.). South
African National Scientific Programmes Report No. 124, Pretoria.
Vogel, C. H.
1988 Climatic change in the Cape Colony, 1820-1900. South African Journal of
Science 84: 11.
1989 A documentary-derived climatic chronology for South Africa, 1820-1900.
Climatic Change 14: 291-307.
Vogel, J. C.
1978 Isotopic assessment of the dietary habits of ungulates. South African Journal
of Science 74: 298-301.
1983 Isotopic evidence for past climates and vegetation of South Africa. Bothalia
14: 391-394.
Vogel, J. C. (ed.)
1984 Late Cainozoic palaeoclimates of the Southern Hemisphere. A. A. Balkema,
Rotterdam.
Vogel, J. C., A. Fuls and R. P. Ellis
1978 The geographical distribution of Kranz grasses in South Africa. ,S.Q..u1b.
African Journal of Science 74: 209-215.
Vogel, J. C. and N.J. van der Merwe
1977 Isotopic evidence for early maize cultivation in New York State. American
AntiQUity 42: 238-242.
von Post, L.
1918 Skogastrad pollen i sydvenska torvmosselagerfoldjer. Forhandlingar
Skandinavika Naturforskeres 16: 432-465.
von Schirnding, Y., N. J. van der Merwe and J. C. Vogel
1982 Influence of diet and age on carbon isotope ratios in ostrich eggshell.
Archaeometry 24: 3-20.
369
Roux
Vorster, M. & pセwN@
1983 Veld of the Karoo Areas. Proceedings of the Grassland Society of southern
Africa, 18: 18-24.
Vrba, E. S.
1980 Evolution, species and fossils: how does life evolve? South African Journal
of Science 76: 61-84.
Wadley, L.
1986 Private lives and public lives: a social interpretation for the stone age.
Paper presented at The longest record: the human career in Africa, a conference
in honour of J. Desmond Clark, University of California, Berkely.
1987 Later Stone Age hunters and gatherers of the southern Transvaal. BAR
International Series 380, British Archaeological Reports, Oxford.
Wallsmith, Debbie
1990 Driekoppen: a Middle Stone Age rockshelter. Nyame Akuma 33: 13-16
Ward, J. D., M. K. Seely and N. Lancaster
1983 On the antiquity of the Namib. South African Journal of Science, 79:175183.
Webb, T. and R. A.Bryson
1972 Late- and postglacial climatic change in the northern Midwest, USA:
quantitative estimates derived from fossil pollen spectra by multivariate
statistical analysis. Quaternary Research 2: 70-115.
Webb, T., Ill., and D. R. Clark
1977 Calibrating micropaleontological data in climatic terms: a critical review.
Annals of the New York Academy of Sciences 288: 93-118.
Wendorf, F.
1968 Site 117: a Nubian final paleolithic graveyard near Jebel Sahava, Sudan. In
The prehistory of Nubia, Vol. 2, F. Wendorf (ed.), pp. 954-955. Fort Burgwin
Research Center Publications 5, Dallas.
Werger, M. J. A. (ed.)
1978a Biogeography and ecology of Southern Africa. Junk, The Hague.
1978b The Karoo-Namib region. In M.J.A. Werger (ed.) Biogeography and ecology
of Southern Africa. Junk, The Hague.
1980 A phytosociological study of the upper Orange River valley. Memoirs of the
Botanical Survey, No. 46, Pretoria.
Werger, M. J. A. and B. J. Coetzee
1978 The Sudano-Zambezian region. In Biogeography and ecology of Southern
Africa. M. J. A. Werger (ed.). Junk, The Hague.
370
White, F.
1978 The afromontane region. In Biogeography and ecology of Southern Africa. M.
J. A. Werger (ed.). Junk, The Hague.
White, J. P.
1968 Fabricators, outils ecailles, or scalar cores? Mankind 6: 658-666.
White, J.P. and D. H. Thomas
1972 What mean these stones? Ethno-taxonomic models and archaeological
interpretations in the New Guinea Highlands. In Models in archaeology, D. L.
Clarke (ed.), pp. 275-308. Methuen, London.
Wiessner, P.
1977 Hxaro: a regional system of reciprocity for reducing risk among the !Kung
San. Unpublished Ph.D. dissertation, University of Michigan, Ann Arbor.
1982 Risk, reciprocity and social influences on !Kung San economies. In Politics
and history in band societies, pp. 61-84, E. Leacock and R. B. Lee (eds.).
Cambridge University Press, Cambridge.
1983 Style and social information in Kalahari San projectile points. American
Antiguity 48: 253-276.
1984 Reconsidering the behavioral basis of style. Journal of Anthropological
Archaeology 3: 190-234.
1985 Style or isochrestic variation? a reply to Sackett. American Antiguity 50:
160-168.
Williams, J. H., R. G. Barry and W. M. Washington
1974 Simulation of the atmospheric circulation using the NCAR global circulation
model with ice age boundary conditions. Journal of Applied Meterorology, 13:
305-317.
Wilmsen, E. N.
1973 Interaction, spacing behavior, and the organization of hunting bands.
Journal of Anthropological Research 29: 1-31.
1989 Land filled with flies. a political economy of the Kalahari. University of
Chicago Press, Chicago.
Winterhalder, B.
1986 Diet choice, risk, and food sharing in a stochastic environment. Journal of
Anthropological Archaeology 5: 369-392.
Wright, H. E.
1983 The Holocene. University of Minnesota Press, Minneapolis.
Yates, R. J., D. E. Miller, D. J. Halkett, A. H. Manhire, J. E. and J. C. Vogel
3 71
1986 A late mid-Holocene high sea-level: a preliminary report on geoarchaeology
at Elands Bay, western Cape Province, South Africa. South African Journal of
Science, 82: 164-165.
Yellen, J. E.
1976 Settlement patterns of the !Kung: an archaeological perspective. In
Kalahari hunter-gatherers. studies of the !Kung San and their neighbors, pp.
47-72, R. B. Lee and I. DeVore (eds.). Harvard University Press, Cambridge.
1977 Archaeological approaches !Q.l!:!.e. present: models fQr reconstructing
セᄋ@
1!:!.e.
Academic Press, New York.
Young, H. J. and T. P. Young
1983 Local distribution of C3 and C3 grasses in sites of overlap on Mount Kenya.
Oecologia 58: 373-377.