ThesisPDF Available

HOLOCENE ARCHAEOLOGY OF THE COASTAL GARCIA STATE FOREST, SOUTHERN CAPE, SOUTH AFRICA Dissertation submitted to the University of Cambridge for the degree of Doctor of Philosophy

Dissertation submitted to the University of Cambridge for the degree
of Doctor of Philosophy
Christopher Stuart Henshilwood
Clare Hall, University of Cambridge
Department of Archaeology
January 1995
Page No.
List of Tables
List of Figures
Chapter 1: Introduction 1
Chapter 2: Environment & Palaeoenvironment
2.1: Introduction 9
2.2: Physiography 10
2.3: Geology of the Riversdale Plain 15
2.4: Bathymetry of the Agulhas Bank 21
2.5: Sea level changes 21
2.6: Climate and palaeoclimate 23
2.7: Vegatation 27
2.8: Fauna 34
2.9: Summary 37
Chapter 3: Ethnohistory of the southern Cape
3.1: Introduction 41
3.2: Ethnohistory of the Riversdale Plain 43
3.3: Observations at Garcia State Forest: 1928-9. 48
3.4: Summary 49
Chapter 4: Holocene Archaeology of the southern Cape
4.1: Introduction 51
4.2: History of LSA archaeology 52
4.3: The Holocene period (8000 - 2000 B.P.) 54
4.4: The Late Holocene (2000 B.P. - Historic period) 57
4.5: Summary 62
Chapter 5: Site Descriptions and Radiocarbon Dates
5.1: Introduction 64
5.2: Survey methods and site selection 64
5.3: Excavation procedures 65
5.4: Site descriptions and stratigraphy 66
5.5: Site classification and taphonomy 85
5.5: Analysis methodology 91
5.6: Textural analysis of sediment samples 92
5.7: Radiocarbon dating 95
5.8: Summary 96
Chapter 6: Shellfish Analysis
6.1: Introduction 98
6.2: Analysis methodology 101
6.3: Shellfish analysis by site/unit 104
6.4: Inter-site/unit shellfish analysis 108
6.5: Shellfish exploitation of the intertidal zone 120
6.6: Comparison of mean shellfish size by species per site 125
6.7: Palaeoenvironmental factors and predation pressure 132
6.8: Comparison of GSF shellfish patterns with S. Cape sites 136
6.9: Summary 139
Chapter 7: Fauna: Mammals, Reptiles & Fish
7.1: Introduction 144
7.2: Mammals 145
7.3: Domestic animals 151
7.4: Micromammals 155
7.5: Reptiles 157
7.6: Fish 161
7.7: Summary 166
Chapter 8: Cultural Artefacts
8.1: Introduction 170
8.2: Classification scheme and inventory 173
8.3: Raw materials 175
8.4: Unmodified artefacts 179
8.5: Utilised/modified class 183
8.6: Retouched lithic artefacts 187
8.7: Bone artefacts 198
8.8: Marine shell artefacts 199
8.9: Ostrich egg-shell artefacts 200
8.10: Pottery 201
8.11: Summary 202
Chapter 9: Seasonality and Oxygen Isotope Analysis
9.1: Introduction 207
9.2: Basic principles of oxygen isotope analysis 208
9.3: Current problems in seasonality research in S. Africa 209
9.4: Objectives of the GSF seasonality study 210
9.5: Methodology & results: Turbo sarmaticus 211
9.6: Methodology & results: Patella tabularis 222
9.7: Summary 226
Chapter 10: Summary and Discussion
10.1: Introduction 228
10.2: Site synopses 232
10.3: Discussion 243
Bibliography 250
During 1992/3 nine coastal midden sites, ranging in age from 6960 B.P. to 480
B.P., and representing 28 depositional units, were excavated in the Garcia State
Forest, a 3.5 sq. km nature reserve situated on the coast near Still Bay. The
broad objectives of the research project were to examine and interpret the
temporal and spatial variation in coastal technological and subsistence
strategies during the Holocene. The results show considerable variation exists
in the location and use of coastal sites situated within a circumscribed area.
Open middens predate 2000 B.P. and later sites are located in coastal caves.
Site location appears to have been influenced by changes in palaeovegetation
and by the arrival of the first herders. Sheep bone, radiocarbon dated at 1960 ±
50 B.P., stratigraphically coincides with the earliest pottery.
The range of shellfish exploited remains similar during all the GSF occupations
suggesting that the species available were unaffected by changes in sea levels,
although shellfish from the lower intertidal zones predominate in the earlier
sites. Evidence from the cave sites show a wide range of mammals, fish and
reptiles were exploited, however bone is poorly preserved at the earlier, open
Stone tools, both in range and timing, broadly coincide with the Wilton Tradition
in sites predating 2000 B.P. but the high numbers of backed elements from GSF
suggests that hafted tools are a distinct feature of sites located west of the
Gouritz River. Post 2000 B.P. sites reflect a change in lithic raw materials and
formal tools are absent.
Oxygen isotope analysis of marine carbonates from Patella tabularis and Turbo
sarmaticus shells indicates that GSF sites were occupied mainly during the
winter months suggesting that the utilisation of coastal resources during
summer was rare during most of the Holocene.
Excavations at Garcia State Forest provide a unique opportunity to examine
variations in site use and subsistence strategies practised by indigenous people
during the Holocene in a coastal environment.
Coastal environments, located at the interface of two major ecosystems,
namely marine and terrestrial, provide unique advantages for the investigation
of past human behaviour (Bailey & Parkington 1988). On the one hand a suite
of marine resources ranging from sessile intertidal organisms, fish, marine
mammals and plants are available for human exploitation while on the other
hand a choice of terrestrial resources are also accessible. This boundary zone,
or ecotone, provides the archaeologist with an opportunity to study the different
possibilities and limitations provided by the available range of resources and
the way that human subsistence strategies were organised.
Prehistoric shell middens are ubiquitous along the coasts of many parts of the
world, for example in northwest and southeastern America (e.g. Stein 1992;
Stein et al 1992; Moss 1992; Claassen 1986, 1991a,b), Australia (e.g. Coleman
1982; Bowdler 1982, Beaton 1985), southern Africa (e.g. Avery 1976;
Robertshaw 1977, 1979; Buchanan et al 1984; Parkington et al 1988) and in
Europe and Britain (e.g. Bailey 1978; Mellars 1987; Deith 1988). By the turn of
the century archaeological investigations were being conducted into shell
middens, for example in California (Stein 1992) and South Africa (Leith 1898;
Colson 1905), leading to a tradition in which the size of a shell midden became
an estimate of prehistoric population size, diet and site age (Stein 1992).
Ambrose (1967) classified the current state of shell midden research in the
1960s into four categories: examining faunal remains in terms of food supply
and calculating meat weights and calories; interpreting changing ecological
conditions based on plotting shellfish species acquired from column samples;
establishing the positions of ancient shorelines from the locations of shell
middens; constructing cultural historical sequences from the recovered
artefacts only and ignoring the shellfish remains. Increasingly, since then, the
range of research questions and the value of shell midden studies is being
realised. The broader objectives of the Oronsay project (Mellars 1987), for
example, address a wide range of issues. Basic questions deal with essential
subsistence strategies, i.e. establishing the range of resources exploited,
strategies of optimisation, and potential variations in the productivity, availability
and edibility of these resources on a seasonal basis. In addition, the
relationship between the locations of the various middens, whether and why
they were occupied once only, or repeatedly, and what factors, environmental
or other, influenced these decisions are also questioned as are the functions of
the Oronsay middens - were they 'occupation' or 'extraction' sites and why did
Mesolithic groups colonise and exploit the resources of Oronsay Island?
Ethnohistoric evidence from Tasmania (e.g. Lourandos 1968; Vanderwal &
Horton 1984), Australia (e.g. Coleman 1982; Hall 1982) and South Africa (e.g.
Deacon 1969, 1976) suggest that coastal resources were used by hunter
gatherers on a seasonal basis and that, in some cases, midden sites may
represent the accumulation of only a few days occupation. The contents of shell
middens are often well suited to testing for seasonality. Many species of fish
and marine mammals have strongly migratory patterns, the bony structure of
some sea mammals and fish indicate the season of death as do the growth
increments in some species of molluscs (Bailey & Parkington 1988). Oxygen
isotope analysis of shell carbonates has positively identified the season of
occupation at some sites (e.g. Shackleton 1973; Deith & Shackleton 1986).
Other current issues in shell midden research are the effects of post-
depositional factors, e.g. groundwater (Stein et al 1992), the effects of
competition by neighbouring groups of people on coastal subsistence patterns
(Bailey & Parkington 1988; Parkington et al 1988) and the advantages of
excavating a number of coastal sites situated in close proximity to one another
(Vanderwal & Horton 1984). Moss (1992) has pointed out that despite the
conspicuous nature of shell middens, few archaeologists have explicitly
addressed the activities represented by the shell in such deposits, particularly
the traditional role of women in shell collecting.
Shorelines are dynamic zones which are reshaped by factors such as rising or
falling sea levels, land tectonics, changes in the direction and strength of
currents and by climate (Bailey & Parkington 1988). Environmental changes
may affect the range and mix of marine resources available for human
exploitation leading in some cases to the abandonment of sections of the
coastline for long periods, for example in the western Cape (Parkington et al
1988), or to the relocation of populations to new coastal sites, as was the case
in Northern Queensland at around 6000 B.P. (Beaton 1985). However, the
direction of economic change may not be solely predetermined by
environmental factors but, as Bailey and Parkington (1988) point out, it is likely
to be a critical factor with regard to the timing of the change.
The 400 km of southwestern and southern Cape coastline, between the well
excavated and radiocarbon dated sites of Byneskranskop 1 (Schweitzer &
Wilson 1982) and Nelson Bay Cave (Inskeep 1987) has an abundance of shell
middens, both in cave and open sites, most of them unrecorded and only a
handful excavated. In the area between Pearly Beach and Cape Agulhas alone
(Fig. 2.1), Avery (1976) counted 205 open shell middens and 17 coastal cave
sites. Interestingly 92 % of the open station shell middens are located along
rocky shorelines, while only 8 % are found near sandy beaches. In the area
between the Breede River mouth and Still Bay the author noted over 80 shell
middens most of which were situated above sections of rocky shore. Although a
few coastal sites between Cape Agulhas and Cape Seal have been excavated
none of these have been radiocarbon dated. The few site reports which have
been published contain little more than a list of the cultural elements of the
middens, the faunal component being largely ignored.
Within the coastal Garcia State Forest (GSF), a 3.5 sq. km nature reserve
situated 20 km to the west of Still Bay, the author recorded 21 Later Stone Age
sites; 9 of these sites were regarded as suited to the objectives of this project
and during excavation 28 discrete depositional units were recorded (cf. Ch. 5).
The range of site types at GSF, their generally high standard of preservation
and diversity in midden content provided a unique opportunity to study various
aspects of human behaviour on this section of the coast during the period from
around 7000 B.P. up untll 290 B.P. (Table 5.5). Seven of the sites are open
station shell middens, six are elevated at above 90 m a.s.l.and located on a
coastal foreland, one is directly adjacent to the coast and all the open sites
predate 3000 B.P.; two sites are in shelters located in the coastal cliffs to the
south of GSF and postdate 2000 B.P.(Fig. 2.3). Only sites which contained in
situ assemblages were selected for excavation, no site was further than 1.5 km
from the coast and all were located within a 1.5 km radius (cf. Ch. 5).
GSF is therefore the first area on the southern Cape coast in which a number of
Holocene sites in close proximity have been excavated allowing for a detailed
inter-site comparative study to be made. The advantages of studying sites
which share a common site catchment area have been detailed by Jarman et al
(1972) and suggests that the extent of the area around GSF exploited by the
sites inhabitants is unlikely to have exceeded a 2 hour walk or about 10 km.
This approach allows for a direct comparison of different site types and
locations and provides an opportunity to examine changes in subsistence
strategies and cultural patterning, both spatially and temporally, given that the
extent of the area exploited was likely to be similar in all cases, although the
effects of environmental change need to be considered in such a study. One of
the disadvantages of comparing sites which are spatially linked is that the time
gap between occupations, for example at GSF, may be over 1000 years. This
means that inter-site comparisons can only be made in a general sense as we
are unable to determine whether there was an hiatus in occupation in the
interim periods or whether we were simply unable to locate these sites.
The objectives of the Garcia State Forest project were to specifically investigate
the diversity present among a suite of Holocene coastal sites which were
located in close proximity to one another. Fundamental to this objective was a
consideration of environmental factors which may have influenced or limited
human choice as to the range of resources exploited, the location of sites and
the activities represented at each site. An integral part of the project was to
excavate sufficient sites, within the time available, which could demonstrate the
degree of diversity within the GSF area. The area and extent of each
excavation was carefully planned to collect as wide a representation of the
assemblage as was possible. Shellfish were ubiquitous at all sites and were
regarded as one of the key factors in examining subsistence strategies over
time but also in determining seasonality through the application of oxygen
isotope analysis to shell carbonates. Cultural artefacts, in particular stone tools,
provided vital clues in tracing cultural change and allowed comparisons to be
made with excavated sequences from other sites in the southern Cape and
further afield. A further key objective in the GSF project was to compare
assemblages which fell within the pre- and post-2000 B.P. period, and to
determine, firstly, whether the arrival of the first herders at the Cape could be
detected in the later assemblages and what effects, if any, the arrival of the first
herders at the Cape had on hunter gatherers in the GSF region. Careful
consideration was given to the possibility of identifying units, in the younger
excavated sites, which may have represented occupation by pastoralists.
The broader objectives of the GSF project are presented according to chapter
below. Contained within each chapter are the detailed objectives relating to that
specific section of the project.
In Chapter 2 a broad perspective is given of the ecology and geology of the
GSF area. As environmental shifts are likely to have affected the coastal and
terrestrial zone of the GSF area during the last 7000 years, and hence the
opportunities for human exploitation of the available resources, it is useful to
examine how these changes affected the palaeo-landscape over time. This
chapter deals with the physiography of the region, climatic change, variations in
sea levels, dune stabilisation and reactivation, and the vegetation and flora of
the Riversdale Plain.
Attempting to translate the archaeological record into social statements, or
establishing the demographics of an area from site densities, distributions or
size is the subject of much debate in the archaeological literature (e.g. Deacon,
H.J. 1976). One method of gaining some insight into the social workings of
prehistoric societies is by reference to ethnohistoric literature. The last
occupation at GSF dates to 1651 A.D., a year prior to Dutch settlement at the
Cape. Chapter 3 examines the accounts of travellers and seafarers who visited
this region of the southern Cape during the 15th - 18th C. and who recorded
meeting indigenous people in the Riversdale/Mossel Bay area. Most describe
meeting with Khoi herders although some accounts provide limited insight into
the social relationships between herders and hunter gatherers. Although it is
evident that by the 17th C. the Khoi were widely distributed across the
Riversdale Plain, the demographics of the area in previous centuries or
millennia is unknown although it seems likely that the Riversdale area was used
intermittently by herders and hunter gatherers over the last two millennia.
Archaeological evidence from the GSF sites supports this contention (cf. Ch. 7).
Chapter 4 provides a review of our understanding of the Holocene archaeology
of the southern Cape and sets the background for interpreting the cultural and
economic trends observed in the GSF area. In applicable chapters,
comparisons are made between the GSF assemblages and those from other
sites in the southern Cape and other Cape regions.
Prior to the selection of the GSF area for this project, specific objectives were
drawn up regarding the methods to be employed in surveying the coast for
suitable sites and the criteria for site selection. Site sampling and excavation
procedures were recorded prior to the commencent of the GSF excavations.
Details of the above are presented in Chapter 5 together with descriptions of
each GSF site excavated, site sediment analyses and a list of radiocarbon
Chapter 6 provides a detailed examination of the shellfish species recovered
from the GSF sites and Chapter 7 reviews all the other faunal remains
excavated. The faunal data raised a number of interesting questions
concerning changes in the range of organisms exploited over time. At the most
basic economic level this meant determining which faunal resources, both
marine and terrestrial, were exploited by the occupants of each site from the
range of resources that were available. The size and extent of the sample
excavated at each midden was therefore critical in order to recover as broad a
range of the faunal contents of a site as was possible (cf. Ch. 5).
Unfortunately, few identifiable animal bones were preserved in the earlier, open
site middens which predated 3000 B.P. However, bone preservation in the post
2000 B.P. shelter sites was excellent allowing for a comparative study to be
made of the range of fish, reptile and mammal species exploited during this
period. Of particular significance was the recovery of sheep bone in Layers 5 &
6 at the GSF 8 site which was directly dated by the accelerator radiocarbon
method to 1960 ± 50 B.P. and 1880 ± 55 B.P. These are the earliest dates
obtained from sheep bone in the southernmost area of the Cape. The results
and implications of the early presence of sheep in this area, and of the faunal
study, are detailed in Chapter 7.
As is the case in most coastal middens, shellfish preservation was generally
excellent in all the GSF sites. Based on the shellfish analysis an inter-site/unit
comparison was possible which examined the range of molluscs exploited and
questioned the differences in species composition between sites/units. A
number of factors were considered in the shellfish study. These included the
range of species exploited from each of the intertidal zones; whether the
selection of certain species was temporally related; what changes, if any, could
be observed over time in the mean sizes of certain species and whether sites
could be typified into 'occupation' or 'extraction' sites based on their shellfish
and cultural components. An extension of this part of the study was a
comparison of the shellfish component in open sites which pre-dated 2000 B.P.
and the shelter sites which post-dated 2000 B.P. to assess whether the arrival of
the first pastoralists in the southern Cape (cf. Ch. 7) impacted on the
subsistence strategies of hunter gatherers in the GSF area.
Details of the cultural artefacts recovered from the GSF sites are given in
Chapter 8. One of the primary objectives of the GSF project was to compare
the lithic traditions at GSF with those from other Cape sites. Although the
trajectory, over time, in stone tool production was broadly similar to that
described for the Wilton tradition and later Pottery Wilton (cf. Deacon, J. 1984a,
b), there are also distinct differences between the GSF lithics and those found
in sites east of the Gouritz River, particularly in the pre-2000 B.P. period.
Backed tools, specifically backed scrapers which were probably hafted, feature
strongly at GSF, Brakfontein and at Byneskranskop 1 and the implications of
this trend are discussed in Chapter 8. Few other cultural artefacts, apart from
ostrich egg shells beads, were recovered from open sites but this is probably a
factor of poor preservation. Decorative artefacts, such as beads and pendants
recovered from the later sites are similar to those from other southern Cape
coastal sites. Pottery occurs in all sites which post-date 2000 B.P. and in Layer
5 at the GSF 8 site pottery is associated with sheep bone dated to 1960 ± 50
Ethnohistoric and archaeological evidence from a number of sites situated on
the Cape coast suggests marine resources were exploited by hunter gatherer
bands on a seasonal basis, and in most cases during the winter months, while
the coastal plains and intermontane regions were utilised during summer (Klein
1973; Avery 1974; Deacon, H.J. 1969, 1976; Buchanan et al 1978; Parkington
et al 1988; Parkington 1991). Testing this paradigm was considered one of the
fundamental objectives of the GSF project and two approaches were adopted.
Firstly, some mammalian fauna and fish species are known to be useful
indicators of seasonality (Klein pers. comm.; Poggenpoel pers. comm.; Best
pers. comm.). In Chapter 7, the dental eruption patterns from a small sample of
juvenile hyraxes from GSF 8 & 9 suggest winter occupation, as does the
presence of the whale barnacle, Coronula diadema, at the GSF 9 site.
Identifying seasonality from the presence of certain migratory fish species at
GSF sites was less convincing. The excellent preservation of shells at the GSF
sites suggested that oxygen isotope analysis of shell carbonates could be
applied to certain shell samples from all the sites. Two species were selected
for analysis, Patella tabularis and Turbo sarmaticus. The results are presented
in Chapter 9 and convincingly demonstrate, with the possible exception of three
stratigraphic units, that all the GSF occupations occurred during autumn, winter
or spring.
In Chapter 10 the relevant evidence from each of the chapters is interpreted
and the conclusions of the Garcia State Forest project are presented, both by
time period and by site type.
The point is that we simply cannot assume that Holocene environments in South
Africa were unchanging or identical. Instead we must allow that the environment
was a variable parameter in any model of Later Stone Age economy or ecology
(Butzer 1974:37)
2.1: Introduction
Most hunter gatherers, worldwide, are known to move their camps, often
seasonally, to make optimal use of a variety of resources within their territorial
range (e.g. Lee 1968; 1984; Marshall 1976; Tanaka 1976; Draper 1977;
Meehan 1982). The size and content of the excavated sites at GSF suggests
that these sites were occupied for relatively short periods, and that the GSF
area may have formed only a part of a larger subsistence strategy. Based on
the limited evidence from the GSF sites we cannot even be sure that any of the
inhabitants utilised the area more than once. Setting up camp at GSF may have
been an opportune decision taken en route to another location, or it may have
been specifically targetted for its known resources. In either event it is likely that
prehistoric visitors to GSF also had a knowledge of the resources which were
currently available in the surrounding Cape coastal plain, the inter-montane
regions, and even beyond.
However, the landscape of the Riversdale Plain, coast and inter-montane
region was not static during the Holocene. Climatic variations affected wind,
rainfall and temperature patterns and this, in turn, caused changes in the
distribution of vegetation and fauna. Higher and lower sea levels covered or
exposed sections of the coastal peneplain and affected the location and nature
of the shoreline. Variations in ocean temperature provided optimum conditions,
at different times, for a diverse range of marine organisms.
An understanding of ecological variation during the Holocene is essential in
order to interpret at least some of the differences observed in the excavated
assemblages at GSF which date to between 6960 ± 70 B.P. and 290 ± 20 B.P
(cf. Section 5.7, Chapter 5). Those aspects of the environment which were
subject to change are discussed in this chapter. A review of the modern
physiography of the region (Section 2.2) and the geology of the Riversdale
Plain (Section 2.3) provides an introduction to the physical features of the
landscape. The formation of calcretes and the activation and stabilisation of the
GSF dunefield were ongoing processes during the occupation of the GSF sites
and the latter aspect may have influenced the surrounding vegetation and
consequently the choice of location of a campsite.
Shellfish are ubiquitous at all the GSF sites indicating that the opportunities for
gathering marine molluscs were apparently available for most of the Mid- to
Later Holocene period. Table Mountain Sandstone (TMS) outcrops occur
directly below GSF and provide ideal conditions for the establishment of
colonies of shellfish. The extent and depth of the TMS outcrops, as well as the
formation of beaches in sigmoidal bays, is discussed in the section on
bathymetry of the Agulhas Bank (Section 2.4). Changes in sea levels can
impact significantly on the availability of shallow marine resources, but the
extent of the impact may depend on the offshore bathymetry. Various sea level
stands for the Holocene are predicted in Section 2.5 and the possible effects of
these changes on the availability of shallow marine organisms discussed.
Present climate of the Riversdale region, palaeoclimatic models and the affect
of coastal ocean temperatures on terrestrial climate and rainfall are presented
in Section 2.6. A palaeoclimatic synthesis for the Holocene shows variations in
rainfall and temperature affected the ratios of grassland to forest taxa in the
southern Cape.
Fynbos has been the dominant vegetation type in the southern Cape during the
Holocene. The possible impact of changes in the vegetation for the animals and
people in the region are discussed in Section 2.7. A gradual change from
grassland, to forest and scrub, during the Early Holocene resulted in an
increase in browsing animals and a decrease in grazers in the southern Cape.
These changes are reflected in the faunal assemblages from a number of Cape
sites. More than fifty species of large mammals were recorded historically in the
Cape ecozone but by the Late Holocene the numbers and distribution of these
animals had been affected by human predation, the impact of domestic
animals, and by deliberate burning of the veld (cf. Section 2.8).
2.2: Physiography
Garcia State Forest (GSF) is a small coastal nature reserve of approximately
350 hectares situated on the extreme southern end of a coastal foreland at 34o
25' S and 21o 13' E (Fig. 2.2). Unconsolidated dune sands, calcrete hardpans
and calcarenites form the exposed surfaces in the reserve. Once a sparsely
vegetated, active dune field the GSF land surface has now been stabilised as a
result of a re-forestation programme started in the 1960s.
The average elevation of GSF is around 150 m a.s.l.; the highest point,
Vlakrant, is 167 m a.s.l. and the lowest point, to the west, is 90 m a.s.l. To the
north the reserve is bordered by dense scrub and arable land, mostly used to
graze cattle. Steep, wave cut and eroded coastal cliffs composed of
calcarenites, calcretes and Table Mountain Sandstone form the southern
boundary. GSF is separated from the sea by the Blombos Nature Reserve (Fig.
Originally, Garcia State Forest formed a part of the Blomboschfontein land
grant made to Hendrik van de Graaff in 1808 by the Earl of Caledon. It is now
under the control of the Department of Nature and Environmental Conservation.
Riversdale is the closest major town situated 38 km to the north. Still Bay, a
coastal village, located at the mouth of the Kafferkuils River lies 19 km to the
north-east. The Duivenhoks River mouth and nearby village of Vermaaklikheid
is situated 19 km to the north-west (Fig. 2.2). Cape Town is 250 km due west
(Fig. 2.1).
The southern Cape physiographic region includes three distinct elements: the
Cape Folded Belt mountains, the Coastal Foreland and the Great Karoo Basin,
which lies between the Folded Belt and the South African Plateau. GSF is
located on a coastal foreland, here named the Riversdale Plain, which extends
40 km inland and lies mostly below the 300 m contour line. This section of the
southern Cape coastal plain is naturally bound by the perennial Kafferkuils and
Duivenhoks rivers. To the north the Riversdale plain is bordered by the
Langeberg mountains, a part of the Cape Folded Mountain range (Fig. 2.2).
Perdeberg at 1340 m is the highest peak. Beyond the 15 km wide Langeberg
lies the Little Karoo.
2.3: Geology of the Riversdale Plain
Bedrock Topography
Between Cape Agulhas and Mossel Bay, Palaeozoic deposits comprise a
basement of folded and faulted sediments of the Cape Supergroup. Table
Mountain Group sandstones dipping in a southerly direction outcrop on the
coast between Odendaals Point and Still Bay. These are on the southern limb
of an overturned syncline and are succeeded by Bokkeveld shales a short
distance north of the coast. Contact between the Table Mountain Group and
Bokkeveld shales is consistently in a west-south-west direction. Table Mountain
Group sandstones re-occur in the anticline of the Langeberg to the north of
Riversdale. An east-west downfault in the Palaeozoic sediments occurs at
around 26 km from the coast on the Riversdale Plain and sediments of the
Enon Formation fill the graben. The Enon and Cape Supergroup sediments are
truncated by a marine peneplain (Deacon et al 1992; Rogers 1984,1988).
Generally the bedrock is surf-cut to form a smooth, and in places stepped,
surface plain that dips seawards by approximately 6 m per km (Whittingham
1971; Marker 1986). Echograms show that while the Bokkeveld shales are
easily plained to a relatively smooth surface, sandstones of the Table Mountain
Group may produce highly irregular topography and major bedrock depressions
(Rogers 1984,1988).
Cenozoic Lithostratigraphy
A suite of Cenozoic sediments, known as the Bredasdorp Group, overly the
marine peneplain and extend for approximately 15 km inland. These facies
record a number of changes in sea levels in this area and reflect the variations
in the volumes of sea water locked up in the ice caps during glacial and inter-
glacial periods (Deacon et al 1992). The Bredasdorp Group consists of a
succession of limestones, sandy limestones, sandstones and conglomerates
and can be divided into six different formations (Table 2.1) (Malan 1986;
Rogers 1984, 1988; Wickens (pers.comm)). Soil classification of the Cenozoic
sequence by Schloms et al (1983) has divided the calcareous, chiefly aeolian
deposits into three sets ranging from A3 (oldest) to A1 (youngest).
Outliers of iron stained, quartzose sands of the Knysna Formation which overly
bedrock are exposed at a number of locations on the Riversdale Plain
(Thwaites & Jacobs 1987). Red sandy rock was recovered from two boreholes
at a depth of around 30 m at Blombos (Whittingham 1971), and at
Jongensfontein and Still Bay red sandstone is visible beneath the overlying
aeolianite and calcarenite (Whittingham 1971; Rogers 1984).
In many places the De Hoop Vlei Formation lies directly on bedrock and forms
the base unit of the calcrete-capped calcareous A3 type sediments (Malan
1986; Rogers 1984; Schloms et al 1983). A 3.3 m thick sequence overlying
Bokkeveld shale is exposed near to Still Bay. The Mio/Pliocene De Hoop Vlei
Formation is characterised by a shelly conglomerate layer, low-angle marine
sands and bioturbated glauconite sand. A feature of this sequence is rare
sharks teeth.
The Wankoe Formation is an aeolian facies of Mio/Pliocene age which
volumetrically comprises the bulk of the Bredasdorp Group and may be up to
300 m thick in places. It is typified by regressional coastal calcarenite dunes
made up of broken shelly material and well rounded quartz grains with
glauconite visible in the cross-bedded aeolianite. Advanced pedogenesis of this
A3 sub-unit produces the terra rossa soils which are widely exposed on the
Riversdale Plain, particularly on dissected slopes and in gorges. To the west of
Still Bay the distinctive and well grassed slopes of terra rossa soils form the
foundation of small farming units.
Sediments of the younger Rooikrans Formation, termed A2 by Schloms et al
(1983), are comprised of a basal pebble layer and an overlying shelly quartzose
sand and are of Late Pleistocene age, possibly around 125,000 years old.
Vegetated and semi-consolidated dune sands which form a narrow band along
the Riversdale Plain coastline form the Waenhuiskrans Formation. These
aeolianites are attributed to increased aeolian activity during the Late
Pleistocene transgression about 20,000 years ago when the coastline was
situated at the present 130 m isobath on the Agulhas Bank. Outcrops may be
up to 60 m thick.
Partially-vegetated unconsolidated dunes, often mobile, mappable as A1
sediments (Schloms et al 1983), occur on the coast at Witsands, near the
Duiwenhoks River, at Garcia State Forest and at Jongensfontein and Still Bay.
Mapped as A1 sediments (Schloms et al 1983) these Holocene coastal dunes
have been designated as the Witzand Formation (Rogers 1984).
Formation Description Age
Witzand (A1) Unconsolidated wind-blown
dunes Holocene
Waenhuiskrans (A1) Semi-consolidated aeolianite Pleistocene
Rooikrans (A2) Shelly quartzose sand and
conglomerate Pleistocene
Wankoe (A3) Consolidated aeolianite Mio/Pliocene
De Hoopvlei (A3) Shelly quartzose sand and
oyster-bearing conglomerate Mio/Pliocene
Table 2.1: Geological sequence of the Bredasdorp Group Cenozoic sediments (after
Rogers 1984; Malan 1986)
Older calcretes and calcarenites of the Bredasdorp Formation underlie most of
the aeolianites of the Holocene Witzand Formation at GSF. At the seaward
edge of the reserve, areas of separate depositional facies of calcrete are
exposed on the cliff tops. In places these calcrete layers are up to 20 m thick.
Similar thick calcrete capped deposits are recorded at Rooikrans east of Still
Bay, and their formation is attributed to strong winds during the Late
Pleistocene (Rogers 1984). Mobile coastal dunes also overly calcretes at
Melkbos, near Cape Town (Netterberg 1974a) at Elands Bay (Butzer 1979) and
at Swartklip (Netterberg 1974a; Barwis & Tankard 1983).
Calcretes are formed by the mobilisation of calcium carbonates from overlying
calcareous or beach sand during alternating wet-dry-wet conditions. They
represent a period of slow deposition and stability of the overlying soils. In
some cases, for example at Sambio in Namibia, hardpan calcretes may provide
the parent material for the overlying calcrete (Netterberg 1974b). The formation
of hardpan calcretes, exposed in some areas at GSF, are generally associated
with periods of rainfall below 550 mm and may also relate to a period of change
from wet to dry (Netterberg 1974b). Climatic conditions strongly influence the
physical and chemical composition of calcrete. However, the exact processes
of calcrete formation are imperfectly understood and palaeoclimatic information
derived from calcretes should be treated with caution (Deacon & Lancaster
At Swartklip the deposition of the last facies of calcrete may date to the start of
the Last Glacial when sea levels dropped and atmospheric circulation was more
intense (Barwis & Tankard 1983). It is not clear when the upper layers of the
hardpan calcretes at GSF were formed, although a Late Pleistocene/ Early
Holocene date is not unlikely. Calcrete formation is an ongoing process at GSF
and at two sites the archaeological deposits are surrounded by a heavily
brecciated matrix.
Coastal Dunes: Activation and Stabilisation
An aerial photograph of GSF taken in 1954 shows unvegetated mobile dune
fields covering the whole reserve. In the 1960s an extensive fynbos planting
programme was undertaken by the Department of Forestry to stabilise the dune
field. An aerial photograph of the same area taken in 1983 shows vegetation
extending over most of the reserve.
Admiralty charts for 1860, 1867 and 1869 show no evidence of extensive
stretches of open sand along the Riversdale coast, although surveys carried out
at the time did not note minor details and existing dune fields may not have
been recorded. Heese (n.d.) reports that as recently as 1917 extensive and
dense forest covered the coastal tract from Still Bay to GSF. In order to get rid
of marauding predators, including leopards, local farmers set fire to the forest
causing destabilisation of the dunes. However, denudation and re-activation of
the GSF dune fields may have started at least 130 years ago due to a
combination of trampling, veld burning and overgrazing (Walsh 1968). Cattle
were brought to the coast from the acidic soils further inland to graze on the
alkaline dune fields. The coastline between GSF and Still Bay was particularly
attractive for this purpose as, at the interface of the basal TMS and overlying
Bredasdorp Group Limestones, perennial fresh water springs could be found
above the high water mark. The general absence of fresh water away from the
main rivers of the Riversdale Plain meant that in order to utilise the coastal
dunes for grazing, cattle had to be watered at the coastal springs every second
day. By 1960 there were 3500 acres of major drift-sands in the area between
Still Bay and GSF (Heese n.d.; Walsh 1968).
The aeolianite facies of the Holocene Witzand Formation at GSF shows
episodes of dune stabilisation alternating with renewed aeolian activity. Some
palaeoenvironmental reconstruction is possible by associating the geo-
stratigraphy of GSF sites with the radiocarbon dates obtained from these sites.
This aspect is discussed more fully in Chapter 5.
A major accumulation of aeolian sands and dune ridges took place in the
southern Cape some time during the -130 m sea levels of the Last Glacial
(Butzer & Helgren 1972). Coastal dune formation is a function of a plentiful
supply of sand and close proximity to beaches (Walsh 1968; Deacon &
Lancaster 1988). Renewed accumulation of dune ridges in the Late Pleistocene
and early Holocene was probably a response to a massive input of fluvial and
aeolian deposits exposed during the Last Glacial transgression (Dingle &
Rogers 1972). The dominant winds causing the deposition of dunes at Robberg
during the Late Pleistocene were from the south-east and south-west. Strong
winds in this area now come from the west and north-west, suggesting changes
in storm winds since the Pleistocene (Deacon & Lancaster 1988).
Blombos Beach, stretching 15 km in a west-north-west direction, is the modern
source of the aeolianites at GSF. The mean axial alignment of the dunes at
GSF is in a north-south direction and at right angles to the coast. Prevailing
winds determine dune alignment but other factors such as natural barriers,
vegetation and wind deflection cause variations in the alignment of major dunes
and also cause continual realignment of the axes of the lesser interior dunes. At
Cape Agulhas, west of GSF, the greatest sand displacement for the summer
months (September - April) of 1940-43 was caused by westerly and easterly
winds and in winter by westerly winds (Walsh 1968). At Cape Agulhas the
westerly wind is the main rain-bearing wind but, being very stormy, can cause
considerable sand movement prior to the main rainfall periods in spring and
autumn. A westerly wind striking the coast to the west of GSF would be partly
deflected by the coastal cliffs above Blombos Beach. Beach aeolianites carried
by the wind are then funnelled upwards at the eastern end of the beach thereby
feeding the GSF dune fields. Periods of lowered sea level would lead to
increased aeolian activity and coastal dune instability. High sea levels during
the Mid-Holocene may have resulted in periods of dune stabilisation, increased
vegetation cover and the formation of humic palaeosols and valley peat floors.
Similar areas of unconsolidated sands and weathered aeolianites are
widespread in the Cape coastal areas. In the Wilderness area, to the east of
GSF, Butzer & Helgren (1972) distinguished at least six generations of
aeolianites based on palaeosol stratigraphy, and Martin (1968) recognised four
successive aeolian dune belts in the same region. Based on geomorphological
and palynological data from the Wilderness Lakes area, Butzer (1974) records
episodes of dune stabilisation, pedogenesis and local peat accumulation during
the Holocene. The first period of dune stabilisation with attendant pedogenesis
at about 8000 B.P. is followed by renewed aeolian activity and inland soil
erosion from 7000 - 4200 B.P. Humic soils and valley peat floors develop until
about 1000 B.P. followed by another phase of gullying and aggradation. After
200 B.P. Butzer (1974) reports renewed gullying with dune re-activation.
Palaeosols mapped on the Robberg Peninsula by Butzer & Helgren (1972) also
indicate a period of dune stabilisation prior to 7030 B.P. Renewed sand
mobilisation occurs from 7000 B.P., followed by stabilisation and peat formation
during the period 4850 - 1300 B.P. and again at 500 B.P. The Mid-Holocene
period of stabilisation, increased vegetation cover and higher precipitation
would have coincided with a +2 m sea level (Butzer & Helgren 1972). At Nelson
Bay Cave, Butzer (1984a, b) notes a strong input of aeolian dune sands at
5500 B.P. but minimal sedimentation after 4000 B.P. Renewed deposition of dry
shelly sands is also reported from Die Kelders at around 6000 B.P. (Tankard
Pedogenesis, peat accumulation and the development of humic soil horizons
on aeolian sands are the result of a combination of reduced sediment supply
and/or increased vegetation cover (Butzer & Helgren 1972; Butzer 1984a, b).
Specific climatic conditions may be reflected by the degree and type of
pedogenesis. Increased precipitation is also likely to contribute to humic
palaeosol formation although Deacon & Lancaster (1988) point out that even
without an increase in precipitation, vegetation may develop on dunes as a
result of a regressing shoreline.
Incised Rivers: The Kafferkuils and Duivenhoks Rivers.
The Kafferkuils and Duiwenhoks Rivers, respectively 54 km and 48 km in
length, border the Riversdale Plain to the east and west. Both rivers originate in
the Langeberg Mountains of the Cape Folded Belt and from here cut through a
peneplain plateau at 100 - 300 m which slopes to the coast.
Both the Kafferkuils and Duiwenhoks rivers cut through highly erodible
Cretaceous sedimentary rocks of the Enon Formation. For the next 40 km and
20 km, respectively, the rivers flow over Bokkeveld shale. The surroundings at
the mouth of the Kafferkuils River are aeolianites and coastal sands with some
outcrops of Table Mountain sandstone. From Vermaaklikheid, the Duiwenhoks
River cuts through 7 km of Bredasdorp Formation calcarenite overlain by
Quaternary sands to emerge at the coast. Both the Kafferkuils and Duivenhoks
estuaries are permanently open with constricted tidal inlets and have a tidal
reach of 19 km and 14 km respectively (Carter & Brownlie 1990).
After the Late Pliocene uplift the surfaces above the marine peneplain were
incised by the two rivers. The degree and intensity of erosion of the river valleys
would have increased during the Late Pleistocene (Marker 1986; Rogers 1984,
1988). The cooler world climate at the time inhibited the hydrological cycle but
also caused less evaporation and any rainfall would have more effectively
eroded the landscape. Seismic surveys of the continental shelf show that as a
result of the Holocene transgression the lower reaches of the two rivers were
drowned (Friedinger 1985; Rogers 1984). It is possible that the extent of the
marine estuaries at the present mouths of the Kafferkuils and Duivenhoks
Rivers may have been substantially greater during the Mid-Holocene (Rogers
1984, 1988).
2.4: Bathymetry of the Agulhas Bank
Resistant outcrops of the Cape Supergroup quartzites have led to the formation
of a series of eastwards facing sigmoidal bays along the southern Cape coast
(Deacon & Lancaster 1988). The continental shelf, known as the Agulhas Bank,
begins as a relatively shallow topographical feature south of Port Elizabeth and
extends to the south and west beyond Cape Agulhas. At its widest point, south
of Cape Infanta, the Agulhas Bank extends more than 200 km and at Cape
Barracouta the edge of the continental shelf is around 100 km to the south (van
Andel 1989).
The bathymetry of the Agulhas Bank between the Duivenhoks and Kafferkuils
Rivers is characterised by three major zones: Table Mountain Group outcrops
form a narrow shelf 4 - 15 km wide, the base of which lies at a depth of 20 - 80
m; to the west of Cape Barracouta, and farther offshore, the middle shelf is
smooth and lies at a depth of between 60 - 100m; the middle shelf is underlain
by south to south-east dipping Cretaceous strata covered by unconsolidated
sediment; on the middle shelf east of Cape Barracouta outcropping Tertiary
rocks are responsible for less regular isobaths (Friedinger 1985:143; van Andel
2.5: Sea level Changes
The effects of changing sea levels can have a significant impact on the
availability of shallow marine resources (Yates et al 1986; Jerardino 1993) and
the responses of hunter-gatherers to these environmental changes are global
issues in the study of coastal prehistory. Examples include Australia
(Lourandos et al n.d.), southeastern United States (Claassen 1986), parts of
northwestern Europe (Bailey & Parkington Vol. 1988) and South Africa (Klein
1973, Avery, G. 1975, Yates et al 1986; Parkington et al 1988, van Andel 1989,
Jerardino 1993).
During the Last Glacial Maximum, between ca. 20000 and 17000 B.P., sea
levels on the southern Cape coast were between -100 m and -165 m (Dingle &
Rogers 1972; Shackleton 1977; Cronin 1982; van Andel 1989). As the sea
regressed a coastal plain more than 100 km wide was exposed to the south of
GSF (van Andel 1989).
Increasing evidence provides clues to the post-glacial recovery of sea level.
Most authors have assumed a more or less steady rise from the Late
Pleistocene until the Mid-Holocene when sea levels were similar to the present.
The general rise curve rests on data selected for stability from Bloom (1979)
and takes account of the recently established two-step course of northern
hemisphere deglaciation (Ruddiman & Duplessy 1985).
Although Tankard (1976b) could find no evidence for higher Holocene sea
levels on the western Cape coast there is firmly dated evidence which suggests
a mid-Holocene sea level +2 m above present. A +2-3 m sea level is reported
from Langebaan Lagoon at between 6500 and 5000 B.P. (Birch 1976; Flemming
1977; Miller et al 1993). A raised cobble beach at 2.8 m above present sea
levels at the Verlorenvlei mouth is dated at 3820 B.P. (Yates et al 1986).
Based on an analysis of shellfish remains and certain types of marine
sediments present in the archaeological deposit at Tortoise Cave on the Cape
West coast, Jerardino (1993) suggests a series of transgressions and
regressions occurred during the Mid- to Late Holocene. A +2 - 3 m sea level is
indicated at 6000 B.P. followed by a rapid regression to present sea level by
4200 B.P.; a further 2 m transgression until 3800 B.P. and modern sea levels by
ca. 3500 B.P. A small but temporary rise may have occurred at around 1800 B.P.
(Jerardino 1993).
A raised gravel beach on the southern Cape coast near the Gouritz River is
interpreted by Rogers (1986) as a storm beach associated with a Holocene sea
level high of +2 - 3 m dating to between 3000 - 2000 B.P. A Mid to Late
Holocene high sea level of +1.5 m based on diatom analysis from Groenvlei
near Wilderness is dated at between 6870 B.P. and 1905 B.P. (Martin 1968).
Evidence for a shore at this elevation on the southern Cape coast has also
been cited by Martin (1962), Maud (1968), Butzer and Helgren (1972) and
Singer and Wymer (1982). Evidence for a tectonic origin for higher sea levels is
ruled out and evidence presented by Clark and Lingle (1979) suggests that
isostatic compensation was responsible for the Holocene high sea stand in
some regions, including southern Africa.
Available sea level data suggest that during the early Holocene the sea was
less than 1 km from the present coastline at GSF and at its current elevation by
Mid-Holocene (van Andel pers. comm.). The impact of a predicted +2 m sea
level (Lambeck pers. comm.) rise on the 120 m high coastal cliffs to the south
of GSF may have been minimal, although wave erosion of calcarenites at the
cliff base is evident. A raised cobble beach below GSF 8 at Skuinsbaai may be
a relic of a previous Holocene sea level higher than at present. Although some
of the presently exposed TMS outcrop to the south of GSF would have been
inundated by a +2 m transgression, a considerable portion of the rocky coast
would still have been exposed during low tides and the impact on shallow
marine resources would have been minimal. The quantity and variety of
shellfish recovered from the middens at GSF dating to between ca. 6000 B.P.
and 500 B.P. supports this hypothesis. The impact of a higher sea level on the
sandy Blombos Beach, situated to the west of Cape Barracouta, may have
caused extensive erosion of the littoral dune margin above the beach. Scouring
of sections of the beach down to bedrock level may also have occurred, as is
evidenced at present during high seas caused by winter storms.
2.6: Climate and Palaeoclimate
Present Climate
Southern Africa, dominated by strongly seasonal precipitation and dry climates,
lies almost entirely within the subtropical high pressure belt. The land mass of
Africa divides this belt into two cells: the South Atlantic and Indian Ocean
anticyclones causing a mean anticyclonic circulation above the atmospheric
boundary layer throughout the year. These two anticyclonic cells are centred on
30o S with 5 - 6o of seasonal latitudinal movement. To the south of the
subcontinent lie the circumpolar westerlies and their associated temperature
low pressure disturbances. As a result the climate in southern Africa is affected
by the interaction of tropical and temperate circulations during all seasons
(Deacon & Lancaster 1988).
The general anti-cyclonic nature of circulation patterns in southern Africa
strongly influences the pattern of surface winds, those in coastal areas
generally being stronger than inland areas. Strong winds, generally parallel to
the coast, are a feature of the weather along the Cape coast. Atmospheric
moisture content is highest in the east, as the major moisture source for most of
southern Africa is the Indian Ocean. A gradient in moisture content from east to
west is especially true in summer and is evidenced by pan evaporation rates
being least along the southern Cape coast (Deacon & Lancaster 1988; Cohen
The Riversdale Plain is classified as Climatic Region A (Schulze 1965) and
receives rain almost equally in all seasons with peaks in spring and autumn.
The orographic effects of the Cape Folded Mountains leads to higher
precipitation in the intermontane regions than at the coast. The mean annual
precipitation (MAP) for the Kafferkuils River Catchment is 482 mm and 634 mm
for the upper catchment area. In the Duivenhoks River Catchment the MAP is
480 mm, and 750 mm in the upper catchment (Carter & Brownlie 1990). The
minimum and maximum MAP recorded at Still Bay for the period 1926 - 1950
was 380 mm and 470 mm (Walsh 1968), and the mean annual rainfall for 1980
- 1992 was 430 mm. During winter and spring the predominant winds are from
the west and southwest and have an average daily strength of 54 km/h.
Easterly and southeasterly winds predominate in the summer months with an
average speed of 64,8 km/h but are less frequent than the winds of the winter
months (Carter & Brownlie 1990; Rebelo et al 1991).
The average daily maximum temperatures are 22 oC in January, and 16 oC in
July. Extreme temperatures reach 42 oC and 32 oC respectively. Average daily
minimum temperatures are about 15 oC in January and 7o C in July. Occasional
extremes of 4o C are recorded in summer and -4 oC in winter. In winter and
spring snow occasionally falls on the Langeberg Mountains. Frost rarely occurs
and hail and thunderstorms are infrequent (Carter & Brownlie 1990; Rebelo et
al 1991).
The Agulhas Current, the Agulhas Bank and Ocean Temperatures
Coastal ocean temperatures affect terrestrial climate and rainfall. Close to the
edge of the continental shelf, the warm Agulhas current flows rapidly in a
southwesterly direction and near the southern tip of Africa is retroflected back
towards the southwest Indian Ocean. The Agulhas current usually has a
minimal influence on the water temperature of the inshore waters of the
southern Cape coast during summer although advective processes, such as
intermittent summer upwelling, can have an effect (Cohen 1993). However, the
Agulhas current can influence the coastal weather by being causative in
establishing a sharp and areally extensive thermocline over the area in the
summer months. During winter, strong southwesterly winds may affect surface
and sub-surface sea temperatures by forcing warm water plumes inshore
(Cohen & Tyson 1995). Solar insolation is the major factor in determining intra-
annual variability in sea surface temperatures (SST) on the Agulhas Bank
(Cohen 1993; Cohen & Tyson 1995).
A strong correlation between phase changes of the southern ocillation and
intra-annual SST's is also present. Anomalously high SST's during summer
months are recorded during the low phase El Niño, and anomalously low
summer SST's during the high phase La Nina. The effects of these phases on
winter SST's are not necessarily synchronic (Cohen 1993).
Strong seasonal signals are evident in sea surface temperatures with an
average amplitude of 5 oC for the period January 1972 - December, 1992,
recorded at Still Bay. The highest SST's recorded during this period are
consistently in the mid-summer months and the lowest SST's in mid-winter.
Cohen (pers. comm.) measured the changes in the stable oxygen isotopes
present in the carbonate of a modern Patella tabularis shell collected at GSF.
The results demonstrate that for the period January 1990 - January 1993 the
average temperature range between mid-summer and mid-winter was 7.8 o C.
Palaeoclimatic Models
Climatic conditions during the Holocene in southern Africa have been
reconstructed from a range of evidence including: pollen (Martin 1968; Avery
1982a; Scholtz 1986; Scott 1984); micromammals (Avery 1982a, b; Thackeray
1987); charcoal and botanical studies (van Zinderen Bakker 1982; Prior & Price
Williams 1985; Scholtz 1986; February 1992); terrestrial fauna (Klein 1973,
1980, 1984); geomorphology (Butzer & Helgren 1972; Tankard 1976a; Helgren
& Butzer 1977; Butzer 1974, 1984a, b); and ocean based models (Cohen &
Branch 1992). Most of the models predict past temperature and/or precipitation
changes suggesting a link between the two variables. Palynological evidence,
suggests that changes in temperature and precipitation were not necessarily in
harmony, particularly during the Mid-Holocene (cf. Cohen 1993) . However,
Cockroft et al (1987) and Tyson & Lindsay (1992) imply that cooler periods are
generally drier and that wet conditions relate to periods of global warming. A
summary of the available data and a suggested reconstruction of the Holocene
climate in southern Africa is contained in Deacon & Lancaster (1988), Tyson
(1986), Cohen (1993) and Cohen & Tyson (1992).
Temperatures in the southern Cape during the Holocene have been markedly
warmer than at any other time during the last 100,000 years. Deacon &
Lancaster (1988:157) point out that the scale of change in terms of fluctuations
in temperature and humidity during the Holocene was considerably lower than
that of the glacial and inter-glacial modes of the Late Pleistocene and
Holocene. Temperature changes for the Holocene, calculated from multivariate
analyses of micromammal species (Thackeray 1987) and from the oxygen
isotope contents of a Cango speleothem, show fluctuations of less than 1o C
around the present day mean, although in some areas variability may have
been up to 2o C (Deacon & Lancaster 1988).
A synthesis of the available palaeoclimatic evidence suggests that some broad,
time-related patterns can be charted for the southern Cape climate during the
Palaeoclimatic Synthesis
10,000 - 7,000 B.P.
Increasingly warm temperatures are recorded with evidence for alternating
cooler/warmer intervals at Byneskranskop and Boomplaas. High sea surface
temperatures in the Indian Ocean are recorded at about 10000 - 9000 B.P.(cf.
Cohen 1993) and an early Holocene thermal rise is reported in isotope data
from Antarctic deep-ice cores between 10000 - 7500 B.P. (cf. Cohen 1993).
Precipitation estimates for the Cape suggest generally dry conditions (Deacon
& Lancaster 1988).
7,000 - 4,000 B.P.
A synthesis of palaeoclimatic data for this period by Cockroft et al (1987),
Deacon and Lancaster (1988) and Tyson & Lindsay (1992) indicates increasing
summer rainfall and a decrease in winter rainfall. Higher air temperatures on
the east and southern Cape coasts meant greater precipitation (Cohen & Tyson
Oxygen isotope analysis of Patella tabularis shells from Nelson Bay Cave show
that prior to 6300 B.P. summer and winter temperatures were lower on the coast
than today. Between 6300 - 5300 B.P. the mean SST was 2 oC higher than at
present. Higher SST's and greater precipitation than at present prevailed in the
region until 3800 B.P. although there is some evidence for a cooling trend after
4300 B.P.(Cohen & Tyson 1995). However, Van Andel (pers. comm.) points out
microclimatic changes, such as that recorded at Nelson Bay Cave, may only
apply to a confined region, and some caution is therefore needed in extending
observations from one region to another.
Pollen analyses from Groenvlei (Martin 1968), and a period of dune
stabilisation at Beacon Island (Helgren & Butzer 1977), suggests that warm and
moist conditions post-date 7000 B.P. in the all-year rainfall region of the
southern Cape. Pollen analysis from Groenvlei shows an increase in forest taxa
in the Mid-Holocene (Martin 1968). The presence of warm water diatoms from
the same site (Martin 1968), and the evidence of warmer temperatures at
Langebaan Lagoon on the west coast (Flemming 1977), may relate to higher
Mid-Holocene sea levels and SST's.
4,000 - 0 B.P.
Oxygen isotope data from shells at Nelson Bay Cave show a mean decrease in
SST of 1 oC after 3300 B.P. Colder sea temperatures, possibly associated with
the start of a Little Ice Age, are also reported at 650 B.P. (Cohen & Tyson 1995).
A lower incidence of forest taxa and a higher grass component in the Cape
ecozone during the last 2000 years (Martin 1968; Scholtz 1986) suggests
cooler conditions in the southern Cape. Optimum forest conditions in the
southern Cape prevailed during the warm, moist period from around 6000 -
2000 B.P. Thereafter, cool and wet conditions prevail along the southern and
southwestern Cape coast (Deacon & Lancaster 1988; Avery 1982a, 1983).
2.7: Vegetation
The vegetation of the Riversdale Plain (defined in this section as the area
between the Duivenhoks River to the east, the Gouritz River to the west and the
Langeberg Mountains to the north) has been described by Muir (1929) and
Acocks (1953; 1988). The flowering plants in the region were surveyed by
Bohnen (1986) and veld plants suitable for grazing by Marais (1937) and van
Breda & Barnard (1991). Despite these publications the Riversdale Plain
remains one of the poorest-studied regions in the Fynbos Biome (Rebelo et al
1991). A recent ecological survey of the Riversdale Plain plant communities has
been undertaken by Rebelo et al (1991) and forms the basis for the resume
The term 'fynbos' requires some clarification as it is used in the context of a
vegetation type, a biome and a flora. Dutch settlers first used the term fijnbos to
describe the fine or small-leafed vegetation they encountered in the
southwestern Cape. Acocks (1953) introduced the term 'macchia' in an attempt
to replace the parochial term 'fynbos' but the latter term is still widely used and
now generally accepted in both scientific and popular literature (Cowling
Within the Cape Floristic Region, fynbos is the dominant vegetation type and
includes elements from two floras, five phytochoria and five biomes (Cowling
1992). Phytogeographically, the Riversdale Plain is a region distinct from that of
the Langeberg Mountains and is unique in the Cape fynbos biome. The region
contains the largest Enon deposits on the coast and the largest development of
Tertiary Limestone in the Fynbos Biome. Within the 2800 square km of the
Riversdale Plain, 1580 plant species have been recorded and 88% of the total
Dune Fynbos found in the Cape occurs here; unique patches of Dune Scrub
Forest patches are also present (Rebelo et al 1991).
Three major vegetation groups occur on the Riversdale Plain consisting of two
non-fynbos groups (Forest and Thicket, and Karroid and Renoster Shrubland)
and one Fynbos Group made up of four series: Grassy, Asteraceous, Restioid
and Proteoid). The GSF environment, consisting of well drained calcareous
sands on recent coastal dunes, supports a Dune Asteraceous Fynbos and
differs from other Asteraceous Fynbos types due to its high cover of non-
ericaceous ericoids and an absence of proteoids. Few Ericaceae are found but
a high cover of Rutaceae and Restionaceae may be present. Dominant plant
species include Agathosma apiculata, Carpobrotus acinaciformis and
Ischyrolepis eleocharis. In the absence of fire, a critical factor in the life-cycle of
fynbos regeneration, Dune Thicket replaces Dune Asteraceous communities
especially in mesic areas. Dune Thicket typically now occurs on the northern
and western boundary of GSF and is characterised by a high cover of fleshy-
leafed shrubs, a high cover of non-proteoids mostly below 2 m and a low cover
of succulents. Typical species of Dune Thicket are Syderoxylon inerme,
Canthium mundianum, Cassina peragua and Salvia africana-lutea. Deep,
organic-rich alkaline sands are preferred by Dune Thicket species but they may
also be found on recent coarse dryish sands with lower organic levels, generally
situated on the littoral fringe (Rebelo et al 1991).
In an attempt to stabilise the shifting aeolian dune field which characterised
GSF during the first half of this century an extensive programme of stabilisation
and re-planting of indigenous plant species was carried out by the Department
of Forestry in the 1960s. During 1993 Prof. Richard Cowling (pers. comm.)
collected samples of the woody plant species found within an approximate 20
km radius of GSF. A list of the species collected is presented in Table 2.2.
The present distribution of vegetation groups in the GSF area suggests there
were periods of Dune Asteraceous Fynbos cover and Dune Thicket type
vegetation during the Holocene. However when the shifting aeolian dunes were
active, ground cover would have been minimal or absent. Vegetation during the
Late Holocene is likely to have alternated between a light Dune Asteraceous
Fynbos cover and unvegetated shifting dunes.
Lower sea levels after the Mid-Holocene high resulted in a wider expanse of
beach to the west of GSF, with a concomitant increase in the amount of aeolian
sand available to feed the dune field. After 2000 B.P. the GSF dune vegetation
may have been utilised by Khoi herders to seasonally graze stock on the 'sweet'
pasture of the alkaline dune sand, just as European settlers later grazed their
stock on the dunes in the 18th and 19th C. (cf. Heese n.d.; Godee-Molsbergen
1932; Blommaert & Wiid 1937). An extreme shortage of water on most of the
Riversdale Plain meant the coastal dunes had to be crossed by stock to gain
access to the fresh water springs found on the coast below GSF, thereby
damaging a fragile ecosystem (Marais 1968). It is also likely that game animals
used the same access route to the coastal spring water and grazed or browsed
in the littoral dunes.
Vegetation may have been periodically burnt in the GSF area (cf. Heese n.d.) to
encourage the growth of new shoots, a practice which was widespread among
the Khoi and the Dutch settlers at the Cape (cf. Marais 1937: Avery, G. 1976;
Wilson & Thompson 1969; Deacon, H.J. 1976; Smith 1986; Parkington 1991).
One, or a combination of any of these actions is likely to have impacted
considerably on the stability of the dune fynbos (Marais 1968). The effects of
poor veld management in the littoral zones of the Cape, mainly due to fires and
overgrazing this century are still evident (Marais 1937; Walsh 1968). The
unstable, shifting aeolian dune field which constituted GSF in the 1950s may
have persisted for much of the last 2000 years and highlights the fragility of this
coastal environment. (Heese n.d.; Marais 1968).
Changes in the vegetation cover at GSF during the Holocene may help explain
the variation in the location of prehistoric campsites. Seven of the radiocarbon
dated sites at GSF are older than 3000 years B.P. and all are situated within the
dune area. Both the later sites, GSF 8 and GSF 9, post-date 2000 B.P. and are
situated in shelters immediately adjacent to the coast. GSF 8 was repeatedly
occupied between 290 B.P. and 1840 B.P., yet there is no evidence of the dunes
being re-occupied after 2000 B.P. Palaeosols underlying the deposits of the
dune sites provide useful pointers to past vegetation cover at GSF and the
significance of these soil horizons is discussed in Chapter 5. A change from
Dune Thicket type vegetation to Dune Asteraceous Fynbos cover on the GSF
dunes would have led to a reduction or absence of milkwood (Syderoxylon
inerme) within the dunefield. Milkwood is a shrub which grows several metres in
height and forms an extensive and dense canopy. Natural hollows form under
these milkwood canopies and make excellent camping spots offering protection
from the elements. A link between the location of coastal shell middens and
clumps of milkwood (melkbosch) was suggested by Martin (1872:55).
I have often been surprised when walking round the coast to come upon
heaped-up mounds of fresh-looking shells, in very bare parts. I now find that
where the mounds are there was a dense clump of melkbosch - a bush so
close, that as all know who ever camped out at Cometjie, is almost as good a
shelter as a house. The bush has long been cut down and burnt, but its roots
can be found running down under the shells. These mounds were kitchen-
middens, without doubt; and the conditions attending their situation invariably
are, shelter, which the bush would give, suitability of sea-shore for successful
fishing and a fresh-water spring near by.
Only a small number of young milkwood plants now grow in the dunes at GSF
but fossilised root-casts, tentatively identified as those of milkwood (Cowling
pers. comm.), are evident in the locality of five sites, GSF 1, GSF 2, GSF 3,
GSF 4 and GSF 5. As the root-casts have not been radiocarbon dated they
cannot be positively correlated with the times these sites were occupied.
However, it is reasonable to argue that the situation of the sites relates to the
protection afforded by suitable vegetation and that stands of milkwood made
the dunes an attractive area for habitation. At the Gordon's Bay shell midden,
dated to around 3000 B.P., carbonised milkwood seeds were recovered in
Layers 1 and 3, although milkwood is now rare in the vicinity of the site (van
Noten 1974).
Ethnographic and ethnohistoric records indicate that a wide range of plants
were used by indigenous people for food, raw materials and medicines. The
importance of plants as food resources for prehistoric people in the Cape can
be traced back to the Middle Stone Age. Plant materials dating to the Late
Pleistocene are preserved in human occupation layers at Klasies River mouth
(Deacon, H.J. 1988). Geophytes - plants with underground tubers, corms or
bulbs - reflect the best collecting opportunities as they offer higher kilojoule
returns than any stems, leaves or inflorescences (Parkington 1991; Deacon,
H.J. 1992). Well preserved plant remains have been recovered from a number
of sites in the southern Cape, most notably at Scott's Cave (Wells 1965),
Melkhoutboom and Highlands Rock Shelter (Deacon H.J. 1976) suggesting that
plant foods were being systematically collected by people throughout the
Holocene as food, but also for a variety of other uses. The most common
geophytes at these sites are Watsonia, Hypoxis, Cyperus usitatus and the
rootstock Bulbine alooides. A key factor in the seasonal movement model of
hunter gatherers between the coast and the interior during the Holocene is the
inland availability of corm bearing plants during the summer months, followed
by a reliance on coastal resources during winter (Wells 1965; Avery 1974;
Deacon H.J. 1976; Buchanan et al 1978; Parkington et al 1988; Parkington
Some of the more common plants recovered from archaeological sites in the
southern Cape include Freesia, Schotia afra sp., Tritonia and a variety of grass
species. Cyperus textilis was used for bedding and also to make rope, matting
and nets (also cf. Grobbelaar & Goodwin 1952). Helichrysum sp. a soft, woolly-
haired shrub is still used for bedding (Bohnen 1986) and in the western Cape
seaweed was used for bedding at Hoffman's Cave and at Eland's Bay Cave
(Deacon, H.J. 1972; Klein 1973).
A synthesis of the vegetation changes in the Cape ecozone can be constructed
from pollen samples and archaeological floral data. An increase in forest taxa is
reflected in pollen profiles from Groenvlei and Norga during warmer and wetter
conditions between 7000 B.P. and 3000 B.P. (Martin 1968; Scholtz 1986).
Thereafter, grassland species increase and forest taxa decrease in the
southern Cape. Prior to the present day renosterveld, grassland was the natural
climax (Deacon & Lancaster 1988). Cowling (1983) suggests that the most
common vegetation in the southern Cape during the last 5,000 years was
grassland rather than renosterveld as suggested by Acocks (1953). An increase
in forest cover is likely during the mid-Holocene high sea levels due to
increased temperatures and precipitation. In the last 2000 years, conditions in
the Cape were somewhat cooler and drier with a high grass component (Martin
1968; Scholtz 1986; Avery, D.M. 1982a, b). Peat deposits from Cape Hangklip
in the southwestern Cape dated to between 6000 and 2000 B.P. show that
fynbos vegetation is represented for the whole period and no significant
changes are reflected (Deacon & Lancaster 1988).
Perhaps the most significant feature of the last 4000 - 5000 years is that, based
on the available biological evidence, it was only during this period that modern
plant and animal community alliances in the southern Cape were formed
(Deacon & Lancaster 1988).
Genus / Species Family Collect
Agathosma ap iculata Rutaceae A C
Azima tetracantha Salvadoraceae B R C C
Canthium cf. mundianum Rubiaceae B C
Carissa bispinosa Ap ocynaceae B C C C
Cassine aethiopica Celastraceae A C D D
Cassine maritima Celastraceae A C C R
Cassine peragua Celastrac eae B R C C
Cassine tetragona Celastraceae A R R R
Chrysanthemoides monolifera Asteraceae A C C R
Clutia daphnoides Euphorbiaceae B C C R
Colpoon compressum Santalaceae A R R
Diospyros dicrophy lla Ebenaceae B R C D
Euclea racemosa Ebenaceae A D C C
Leucadendron salignum Proteaceae B R
Lycium cinereu m Solanaceae A C C C
Maytenus proc umbens Celastraceae A C C R
Maytenus heterophylla Celastraceae B C C
Metalasia muricata Asteraceae C
Myrica cordifolia Myricaceae A C
Myrsine africana Myrsinacea e A R R R
Olea europeae subsp. africana Oleaceae B R C
Olea exasperata Oleaceae A D C R
Olinia cymosa Oliniceae B C
Passerina rigida Thym eleaceae C C
Polygala my rtifolia Polygalacea e A R R R
Pterocelastrus tricus pidatus Celas traceae A C C C
Putterlickia pyracantha Vitaceae A R C R
Rhoicissus digitata Anacardiceae A R R R
Rhus crenata Anac ardiceae C D R
Rhus glauca Anacardiceae D D C
Rhus laevigata Anacardiceae A R R R
Rhus longispina Anacardiceae A R C C
Rhus lucida Anacardiceae B R
Salvia africana-lutea Lamiaceae A R R
Syderoxylon inerm e Sapotaceae A D D D
Zanthoxylum capense Rutacea e B C
Zygophyllum morgsana Zygophyllaceae A C C
R = Rare C = Common D = Dominant
Collect Codes
Comm Data
A: Collected on red sands, Jongensfontein Comm A: Thicket on white dune sand
B: Collected in forest, Heuningbos Comm B: Thicket on older reddish sand
C: Collected on white sands, Jongensfontein Comm C: Forest
Table 2.2: Woody plants collected in the Jongensfontein/ Garcia State Forest Area,
southern Cape, South Africa.
Notes to Table 2.2.
1. Collection date: June, 1993.
2. Collection and Identification: Prof. Richard Cowling, Botany Department., University of Cape
3. The predominant vegetation in the Garcia State Forest area at present is dune thicket on
white, calcareous sand.
4. Myrica cordifolia, Passorina rigida, Rhus crenata and Metalasia muricata currently are the
dominant species in the mobile dune area of Garcia State Forest (possibly as a result of
stabilisation / restoration attempts by Forestry Dept.).
2.8: Fauna
The Cape Ecozone forms one of six ecozones defined within the southern
African sub-continent. Major variables such as climate, topography,
phytogeography and zoogeography characterise each ecozone but an
overlapping of features prevents each region from being sharply divided. Due
largely to the lack of vegetation suitable as animal food, the Cape Ecozone has
the lowest diversity of large mammal species found in southern Africa (Klein
1984; Skinner & Smithers 1990). Fifty species of large mammals are recorded
historically in the Cape (Skead 1980) although even within this ecozone some
of these animals were restricted by their habitat and dietary preferences and
not all were endemic to the coastal areas of the southern Cape.
Elephant, Cape buffalo, hippopotamus and eland were present in the
Riversdale District in the 17th- and 18th C. (Le Vaillant 1790; Raven-Hart
1967:21: Skead 1980). In 1777, William Patterson reported seeing lion spoor at
Riviersonderend and both Anders Sparrman in 1775 and J.W. Moodie in 1820
recorded leopards near Swellendam. La Trobe saw wild dogs (Lycaon pictus),
and hyenas (H.crocuta & H.vilosa) near Mossel Bay in 1816, and the latter were
again reported in the Swellendam area in 1838 by James Backhouse (Skead
1980). By the latter part of the 19th C. most of the larger mammals, in particular
predators, had been eradicated and most of the land on the Riversdale Plain
had been granted for settlement by farmers.
The largest sample of well excavated faunal assemblages in southern Africa
are derived from Holocene sites excavated in the Cape ecozone. After the Last
Glacial Maximum warmer conditions and generally higher precipitation led to a
gradual change from open grasslands to increasing forest or scrub cover in the
southern Cape. Relatively higher numbers of vaalribbok, mountain reedbuck
and roan antelope occur in early Holocene assemblages but are rare after 5000
B.P. (Klein 1980). At Nelson Bay Cave prominent grazers are replaced by small
browsers by around 8000 B.P., although at Melkhoutboom and Elands Bay
Cave the presence of hartebeest, buffalo, wildebeest and zebra in the Early
Holocene implies an environment with more grass and less fynbos than at
present (Klein 1973; Klein 1980; Deacon, J. 1984a, b; Deacon & Lancaster
1988). Based on measurements of jackal, dune mole rats, hyraxes and grysbok
recovered from Cape sites, Klein (1984) has demonstrated a correlation
between past temperature and rainfall, and body size. Similarly, modern dune
mole rats in low rainfall areas are smaller than those found in the high rainfall
areas of the southern Cape (Klein 1984). At all Holocene sites in the southern
Cape the micromammalian community structure was distinctly different to that
observed during the Late Pleistocene (Avery, D.M. 1982a, b).
The effects of human predation, the introduction of herding, and accidental or
deliberate and repeated burning of veld have also been cited as attributable
factors in changes in faunal distributions during the Later Holocene (cf. Marais
1937; Deacon, H.J. 1972; Klein 1973; Avery, G. 1974; Deacon, H.J. 1976;
Smith 1986; Parkington 1991).
The overall picture in the Cape Ecozone is one of a gradual change from
grazers to browsers during the early Holocene and, by 5000 B.P., the large
mammal communities were essentially similar to those of historical times (Klein
1973, 1980, 1984; Deacon, J. 1984b; Deacon & Lancaster 1988).
A recent survey indicates at least 26 species of mammals now occur in the area
between the Kafferkuils and Duiwenhoks Rivers (Stuart & Stuart 1988; Carter &
Brownlie 1990; also cf. Inskeep 1987 for a list of southern Cape mammals).
These include 12 species of rodents; the African wild cat (Felis lybica), African
lynx (Felis caracal); steenbok (Raphicerus campestris), grey rhebuck (Pelea
capreolus); black backed jackal (Canis mesomelas); Cape porcupine (Hystrix
africaeaustralis); Cape clawless otter (Aonyx capensis), water mongoose (Atilax
paludinosus), 2 species of genet (Genetta sp.); 2 species of hare (Lepus sp.);
Smiths red rock rabbit (Pronolagus rupestris); the common mole rat (Cryptomys
hottentotus). Sea mammals found in the southern Cape waters include the
Cape fur seal (Arctocephalus pusillus) and a number of species of dolphins and
whales (cf. Stuart & Stuart 1988; Skinner & Smithers 1990).
Ocean temperature is a critical factor in determining the abundance, distribution
and presence of marine animals and plants. The generally warm waters found
on the southern Cape coast are a combination of solar insolation and the effect
of the Agulhas current originating in the tropical regions of the Indian Ocean .
Periodic low water temperatures at the coast are the result of upwellings of cold
water situated off the Agulhas Bank being driven onshore (Cohen & Tyson
1992; Cohen 1993). The resultant disparate conditions provide for enormous
biodiversity, allowing both cold and warm water faunal communities to co-exist
(Smith 1952; Tietz & Robinson 1974; Branch & Branch 1981; van der Elst
Rocky shores, sandy beaches and offshore reefs occur in the immediate vicinity
of GSF providing conditions suited to a wide range of sea fishes. Additionally, to
the east and west of GSF, estuaries are found at the mouths of the Duivenhoks
and Kafferkuils Rivers. Angling on this section of the coast is generally excellent
although in recent years the number of fish landed has diminished due to
excessive fishing. At least 25 species of edible fish are found in these waters,
some of the more common species landed included galjoen (Coracinus
capensis), elf (Pomotamus saltatrix), kob (Argyrosomus hololepidotus) and
white musselcracker (Cymatoceps nasutus) (Smith 1953; van der Elst 1988).
During the Holocene fish were probably abundant in this area and the bones of
both small and large species were recovered from a number of sites at GSF,
and elsewhere along the southern Cape coast (cf. Goodwin 1946; Inskeep
1972, 1987; Klein 1973; Avery, G. 1975, 1976).
Five distinct inter-tidal zones are recognised on the rocky shores of the
southern Cape coast. Highest on the shore is the Littorina zone, followed by the
Upper Balanoid, the Lower Balanoid, the Cochlear and the Infratidal zone.
Particular species of shellfish are adapted to specific conditions leading to
regular zoning in each of the inter-tidal bands. Factors affecting the density and
size of marine shellfish colonies are a combination of water temperature, light,
salinity, wave action, available food and competition, both intra- and inter-
specific. (Tietz & Robinson 1974; Branch & Branch 1981).
During low neap tides a variety of edible shellfish species can be gathered from
the upper shore. Greater exposure during low spring tides increases the
number and size of species available for collection. The larger species of
gastropods found in the lower intertidal zone include alikreukel (Turbo
sarmaticus) and occasionally perlemoen (Haliotis midae). Turbo can be
collected during neap tides as they also occur in the mid-tidal zone; their shells
are ubiquitous in many middens along the southern Cape coast. A variety of
limpets (Patella sp.) are easily gathered; periwinkles (Oxystele sp.), whelks
(Burnupena sp.), Venus ear (Haliotis spadicea.) and chitons (e.g. Dinoplax
gigas) are common. Brown mussels (Perna perna) are found in large colonies
attached to rocks in the upper- and mid-tidal zones and are easily collected.
Other marine organisms, for instance crabs (e.g. Cyclograpsus sp. and
Plagusia sp.) and octopus (Octopus granulatus), may be opportunistically
Shellfish offer a sessile, dependable source of protein and the large numbers of
shell middens situated along much of the southern Cape coastline attests to the
importance of this resource to prehistoric Holocene people.
Other Fauna
Additional fauna which may been utilised for food by prehistoric people in the
vicinity of GSF includes a wide range of land and marine birds (cf. Sinclair
1987), amphibians and reptiles, including 4 species of tortoise, 22 of snake and
15 types of lizard (cf. Branch 1990). The recovery and identification of faunal
remains from sites at GSF is discussed in Chapters 6 & 7.
2.9: Summary
Garcia State Forest is situated in a once active dunefield on the southern edge
of a coastal foreland. It is bound to the east and west by the Kafferkuils and
Duivenhoks Rivers respectively; to the north of the coastal foreland are the
Langeberg mountains which form a part of the Cape Folded Mountain range.
Underlain by bedrock Palaeozoic deposits, the overlying Cenozoic sediments
are known as the Bredasdorp Group and consists of five sedimentary groups.
From oldest to youngest these are: De Hoopvlei and Wankoe of Mio/Pliocene
age; Rooikrans and Waenhuiskrans of Pleistocene age; Witzand dunes
deposited during the Holocene.
Calcretes and calcarenites of the Bredasdorp formation underlie most of the
Holocene Witzand Formations at GSF and were formed during alternating wet-
dry-wet conditions. Calcrete formation is an ongoing process at GSF and a
heavily brecciated matrix surrounds the deposit at two of the sites.
Episodes of dune stabilisation alternating with periods of renewed aeolian
activity are recorded during the aeolianite facies of the Holocene Witzand
Formation at GSF. During the Late Pleistocene and Early Holocene a re-
accumulation of dune ridges occurred as a response to a massive input of
fluvial and aeolian deposits exposed during the Last Glacial transgression. High
sea levels during the Mid-Holocene may have resulted in periods of dune
stabilisation and increased vegetation cover at GSF. Renewed activation of the
GSF dunefields is recorded during this century although this process may have
started at an earlier date.
Eastwards facing sigmoidal bays, for example St. Sebastian Bay, are formed
by the resistance of outcrops of the Cape Supergroup. A continental shelf, the
Agulhas Bank, extends for more than 100 km to the south at Cape Barracouta.
Table Mountain Group outcrops form an inner narrow shelf 4 - 15 km wide.
Sea levels on the southern Cape coast were between -100 m and -165 m
during the Last Glacial Maximum. A steady rise in sea levels occurs during the
Early Holocene and by the Mid-Holocene a + 2 m sea level is likely. During the
Later Holocene a series of regressions are possible but after about 3500 B.P.
modern sea levels were maintained. During the Early Holocene the sea was
about 1 km from the present coastline at GSF. The impact of a predicted + 2 m
sea level on the coastal cliffs to the south of GSF may have been minimal,
although Blombos Beach to the west is likely to have been inundated, causing
erosion of the coastal dunes. Shallow marine resources are unlikely to have
been severely affected by a + 2 m transgression in the area below GSF as the
irregular TMS outcrops are sufficiently elevated to allow marine organisms to
move to a higher plateau.
Surface winds are strongly influenced by the general anti-cyclonic nature of
circulation patterns in southern Africa and coastal winds are stronger along the
coast than in inland areas. Predominant winds during winter and spring are
from the west and southwest and in summer easterly and southeasterly winds
predominate. Atmospheric moisture content is highest in the east as the Indian
Ocean is the major moisture source for southern Africa. Mean annual rainfall for
Still Bay is around 450 mm but higher precipitation is recorded in the catchment
areas of the Duivenhoks and Kafferkuils Rivers. The Riversdale Plain receives
rain in all seasons with peaks in spring and autumn.
Coastal weather is also influenced by the Agulhas Current which is causative in
establishing an extensive thermocline over the area during the summer months.
Intra-annual variability in sea surface temperatures on the Agulhas Bank are
determined mainly by solar insulation but strong southwesterly winds in winter
may also force warm water plumes inshore. An average amplitude of 5 oC in
sea surface temperatures recorded over a 20 year period at Still Bay provides
evidence for strong seasonal signals with mid-summer highs and mid-winter
Based on a range of evidence including pollen, micrommals, charcoal,
botanical studies, terrestrial fauna and ocean based research a palaeoclimatic
model for the southern Cape has been established. A synthesis of this data
suggests that during the Early Holocene, land and sea temperatures increase
but precipitation is relatively low. Summer rainfall on the southern Cape coast
increases in the Mid-Holocene due to higher air temperatures. Sea surface
temperatures are 2 o C. higher than at present. After 3300 B.P., sea surface
temperatures show a slight decrease at Nelson Bay Cave and again at 650 B.P.,
possibly associated with the start of a Little Ice Age. A higher grass component
and a lower incidence of forest taxa suggests cooler conditions in the Cape
after 2000 B.P. Optimum forest conditions are likely to have prevailed in the
Cape between 6000 - 2000 B.P.
The modern vegetation of the Riversdale Plain has, until recently, been poorly
documented. Recent surveys show three major vegetation groups occur on the
Riversdale Plain. A Dune Asteraceous Fynbos is supported in the GSF
environment with a high cover of Rutaceae and Restionaceae. To the north and
west of GSF, Dune Thicket replaces Dune Asteraceous communities and is
characterised by a high cover of fleshy-leaved shrubs. Typical among Dune
Thicket species are milkwood, Syderoxylon inerme, which has a spreading
canopy and is well suited for use as a campsite. Dune Thicket species are likely
to have occurred within GSF in the past when the dunefields were stable.
Ground cover, on the other hand, would have been minimal or absent during
periods when the aeolian dunes were active. Vegatation during the Late
Holocene at GSF is likely to have alternated between a light Dune Asteraceous
Fynbos cover and unvegetated shifting dunes. In the latter period possible
burning of vegetation in the GSF area by herders, and later by Dutch farmers,
may have impacted considerably on the stability of the Dune fynbos.
Seven of the excavated sites at GSF predate 3000 B.P.and all are located within
the dunefield area. Sites postdating 2000 B.P. are located in shelters to the
seaward side of the coastal cliffs. One of the reasons for the variation in the
location of GSF campsites may be changes in vegetation cover in this area
during the Late Holocene.
Based on available biological evidence it seems that modern plant and animal
community alliances were only formed in the Cape during the last 4000 - 5000
years. The Cape Ecozone, one of six ecozones defined within the southern
African subcontinent, has the lowest diversity of large mammal species. A
number of well excavated faunal assemblages are derived from Holocene sites
located in the Cape Ecozone. Prior to 8000 B.P. grazers are dominant but are
gradually replaced by small browsers. Animals such as vaalribbok, mountain
reedbuck and roan antelope are rare in assemblages which postdate 5000 B.P.
Small mammals, such as dune mole rats and hyraxes are more common in
later assemblages in the southern Cape and indicates a greater use of small
food parcels during the Mid- and Later Holocene. Faunal assemblages from
GSF sites provide evidence that a diverse range of animals were collected,
hunted or opportunistically scavenged. Various species of shellfish provided a
dietary staple and may have been one of the principal reasons for coastal visits.
Fish were also caught; sea birds trapped; and seals hunted or scavenged.
There is evidence of a whale at GSF 9, probably scavenged after a wash-up.
Bovids, ranging from eland to grysbok, were hunted or trapped. Evidence of a
variety of smaller animals including tortoises, snakes, lizards, birds, dune mole
rats, hares, and hyraxes illustrates the diverse tastes of the GSF inhabitants.
3.1: Introduction
During the 15th C. Portuguese explorers sailed around the tip of Africa and
landed on the southern Cape coast. Vasco da Gama and Bartholomew Diaz
both stopped at Mossel Bay and reported contact with indigenous people
(Axelson 1954). In 1576, Manuel de Mesquita Perestrello arrived at the Breede
River mouth and named the bay between Cape Infanta and Cape Barracouta
after St. Sebastian. A party of seamen were landed at the Breede River mouth
and walked overland to Cape St. Bras (Mossel Bay), presumably crossing close
to the present GSF area (Kruger & Beyers 1977).
Calibrated radiocarbon dates for some of the excavated sites at GSF overlap
with those of the early explorers; GSF 8, Layer 8\1 dates to 1651 A.D. and GSF
9, Layer OH to 1493 A.D. One method of broadening our interpretation of the
GSF archaeology for this period is to examine the ethnohistorical records for
this region.
Warnings abound as to the uncritical use of ethnohistoric accounts in the
interpretation of archaeological evidence (e.g. Avery 1976; Klein 1986;
Henshilwood 1990). The same criticisms often levelled at archaeologists in their
imprudent use of ethnographic analogy can equally be applied in the field of
ethnohistory. Rigorous standards of empirical observation, developed in the
archaeological field, are frequently not applied to the evidence afforded by
ethnography, leading to weak and ambiguous analogies (Gould 1977:362).
Furthermore, the uncritical use of analogy may mask an understanding of past
behaviour by imposing on it, and limiting it to the categories of the ethnographic
or ethnohistoric present (Ascher 1961; Gould 1977; Yellen 1977; Wylie 1985).
Hodder (1985) has pointed to the inadequacies of examining the relationship
between statics (the archaeological record) and dynamics (human behaviour)
without understanding the generation of that relationship by individuals in an
active social context.
Compilations of ethnohistoric accounts of indigenous people in the Cape (eg,
Axelson 1954; Thom 1954; Raven-Hart 1967) clearly illustrate the problems of
the relationship between archaeology and historic observation. Eurocentric
viewpoints surface in the use of words such as 'savage' and 'barbaric' to
describe the local inhabitants, implying the use of a value system heavily
loaded in favour of European cultural norms and which frequently fails to grasp
the complexity of interaction amongst indigenes (cf. Raven-Hart 1967).
Observations frequently lack ethnographic objectivity and thoroughness and as
such are not directed at the questions which concern archaeologists (Klein
The instigation of a trade in livestock with local herders by the Europeans, the
formation of alliances, and the later establishment of a permanent Dutch
presence at the Cape in 1652, led to gradual changes in local economic and
subsistence strategies, demographics and intergroup dynamics. In addition, a
large body of ethnohistoric data postdates the mid-17th C., and may not
therefore reflect conditions during the prehistoric period.
Early misconceptions arising from attempts to group indigenous people
illustrate the points above. Mendelslo, who visited the Cape in 1639, describes
two distinct groups, Watermen and Solthanimen. The former live by the shore
eating herbs, roots, fishes and dead whales, the latter own cattle and sheep
and being dependent on adequate grazing for their stock move between the
coast and inland pastures. In later accounts the Watermen are also described
as Strandloopers (litt. beach walkers) (Raven-Hart 1967). In a report to van
Riebeeck in the 1650s, Herry, initially himself described as a Waterman, added
a third group called Fishermen who subsist on fishing, own cattle and are
continuously at war with the Saldanhamen (Solthanimen). Herry's group, who
through past misfortunes had been forced to rely on coastal resources, later
acquire cattle from the Dutch and revert to a herding way of life (Thom 1954). A
further cultural group, the Soaqua or Sonqua, who live a 'meagre existence' and
own no stock are met with by Jan Wintervogel on an expedition inland. He also
notes that the so-called Fishermen do not always possess cattle and equates
the Sonqua with the Fishermen. This is later confirmed by Herry who also states
that cattle are only acquired by the Fishermen when stolen from the
Saldanhamen (Solthanimen), hence the constant state of war between the two
groups (Thom 1954).
Confusion in the grouping and naming of the two economic groups persists
throughout the ethnohistoric literature and misnomers, such as Strandlooper,
persist to the present day (cf. Raven-Hart 1967; Avery 1976). The clearest
distinction which characterises indigenous groupings at the Cape can be made
on economic grounds. San hunter gatherers (Bushmen, Soaqua, Sonqua,
Watermen, Fishermen) subsisted by utilising a wide range of resources inland
and on the coast, possibly on a seasonal basis, and did not manage stock
although sheep and cattle may have been opportunistically stolen (e.g.
Schapera 1930; Schapera & Farrington 1933; Raven-Hart 1967; Deacon, H.J.
1976; Buchanan et al 1978; Parkington et al 1988; Parkington 1991). Khoi
groups (Khoikhoi, Hottentots, Saldanhamen) on the other hand were
pastoralists who managed domestic stock, but also hunted and gathered and
were seasonally transhumant (e.g. Schapera 1930; Goodwin 1952; Raven-Hart
1967; Smith 1986; Klein 1986). Transitional categories for people falling in the
grey area between the two may include hunters acquiring stock and becoming
herders, herders losing stock and reverting to hunting and gathering and
various forms of clientship. Social and economic differences between these two
groups, and recognising these differences in the archaeological record, is the
subject of an ongoing and lively archaeological debate (cf. Schrire 1984;
Parkington 1984; Smith 1986; Hall 1986; Klein 1986; Parkington & Mills 1991).
These issues are discussed more fully later in this chapter and in Chapter 7.
Despite the limitations imposed by ethnohistorical analogy there is merit in the
judicious application of some of this data in expanding our understanding of the
later archaeological record at GSF. For instance, the recorded presence of Khoi
huts, cattle and sheep on the Riversdale Plain coincides roughly with the
occupation of Layer 8\1 at GSF 8 and this information may be useful in
understanding the identity and purpose of these coastal visitors. For obvious
reasons attempts at using ethnohistoric data to interpret earlier GSF sites must
be less secure.
3.2: Ethnohistory of the Riversdale Plain
Although Perestrello was possibly the earliest European explorer to make
contact with indigenous people on the Riversdale Plain in 1576 there is scant
record of this meeting, presumably he was more intent on finding a suitable port
for future Portuguese ships. The Breede River mouth was described by
Perestrello as adequate to 'shelter a great fleet', but later explorers preferred
the shelter of Mossel Bay situated 100 kms to the northeast. Both da Gama and
Diaz had sought refreshment in Mossel Bay in the late 15th C., as the Dutch
seafarers did a century later. Although not on the Riversdale Plain, Mossel Bay
ethnohistory contains some interesting early accounts of the regions indigenous
people on the coast. Cornelis de Houtman, arrived at Mossel Bay on the 2nd
August, 1595 aboard the 'Mauritius' and his journal records the crews contact
with locals and their efforts to obtain fresh meat. The extract below is translated
from the Dutch account (Cape Archives VC94 No.1).
The locals were very afraid but after a while they came up to our two people ....
We asked them to bring sheep and cattle in exchange for money but they did
not know of any money but only wanted iron, copper and bracelets. They spoke
in their own way and indicated that they would bring cattle but when we heard
them talk we wondered about their speech as they spoke like turkeys gobbling
through their throat, they had ugly, flat noses and teeth like dogs and had an
animal hide around their waist and in front of their manliness the tail of an ox
and wide pieces of leather tied around their feet. They smelt so much that we
could not stand their company. They were very glad when we gave them a small
knife and within one hour they went and fetched four sheep.
On the fifth instant we went ashore again and bartered with the 'wild ones'
exchanging iron for animals and we shot the best of the animals with our
muskets ... they saw us throw away the stomach and intestines and they
grabbed these. .... soon after they made a knife from a stone and cut the
stomach open with it. They shook the dirt from the stomach and ate the stomach
as if they were dogs.
On the 9th August about forty men armed with muskets went inland to find the
Bushmen's living places. When our people had travelled some distance towards
the mountains, the wild ones ran away and started to make fires everywhere.
Some of our people returned as they could not walk due to scurvy.
This account, and others like it (cf. Thom 1954; Raven-Hart 1967) provide
some clues as to the local economy. Sheep and cattle were in the vicinity of
Mossel Bay, presumably indicating the the presence of Khoi herders. A local
knowledge of iron and copper suggests that bartering had taken place
previously, possibly with earlier seafarers, but the the expedient manufacture of
a stone knife may indicate that iron was not in common use, or at least not by
the people they met. In other respects the account is ambiguous; what was the
identity of the 'Bushmen' people they met, were they coastal hunter gatherers
acting as go-betweens for the Khoi, or were they itinerant stock owners trading
In July, 1601, Paulus van Caerden aboard the 'Hof van Holland' records they
were unable to obtain fresh meat at Mossel Bay as the eight 'natives' they met
did not possess stock. However, a week later the ship is driven close inshore at
Vleesch Baai, 50 kms to the northeast of Still Bay where they are able to trade
with the locals 'as many oxen, calves and sheep as they wanted until the salt
for drying ran out'. On the 23rd July they are forced to anchor in Visch Baai,
slightly to the east of their last location, but are only able to obtain five sheep
(Cape Archives VC94 No.1).
The above accounts suggest that in the late 16th C. Khoi people with sheep
and cattle could be found along the coast in the vicinity of the Riversdale Plain
during the winter months, but that their location was not fixed, possibly due to a
need for fresh pasture. Other coastal dwellers, without stock, were also
encountered who presumably hunted and gathered opportunistically. Whether
they occasionally acted as agents for the pastoralists is not clear.
In the latter half of the 17th C. relations between the Dutch and western Cape
Khoi deteriorated and the large amounts of fresh meat required to provision
passing ships became increasingly difficult to obtain. As a result the Dutch
looked to setting up trade relations with the Khoi in the southern and
southwestern Cape. By 1660 the pastoralists occupying the area between the
Breede and Gouritz Rivers, known as Hessequas, had been contacted and
were reputed to be 'rich in cattle' (Mossop 1931). The Houtniquas occupied the
area to the east of the Gouritz River up to Algoa Bay, and to the west the
Chainoqua were situated between Swellendam and the Hottentots Holland
Mountains (Kruger & Beyers 1977). In 1667, Sergeant Cruijthoff met with
Hessequas at the Breede River and his party returned with 170 cattle and 300
sheep. The following year Hieronymous Cruse sailed to Mossel Bay on the
'Voerman' to set up trading relations with the Khoi (van Wyk 1988). Further
expeditions to obtain sheep and cattle from the Hessequas are recorded in
1685 and 1686 (Resolution of Council, DEIC). In 1687, Gammou Kouchama,
described as a Hessequa 'captain', visited the Castle in Cape Town to invite
further trade with his 'country' described as 'populous and overflowing with
oxen, sheep and goats' (Moodie 1838). There is no evidence that the Khoi kept
goats, historically or archaeologically, and this reference to goats may be an
error in translation.
Ensign Isak Schrijver, sent to trade with the Hessequas in January 1689,
recorded his journey across the Riversdale Plain in considerable detail. At the
Duivenhoks River his party meet thirteen Hessequas who were sent to direct
them to 'the Swart Captains kraal'. The following day they pass 'many kraals of
Hessequas' and outspan at the kraal of the 'Captain', also known as the 'Oude
Heer'. In the vicinity of modern-day Riversdale, Schrijver reports 'we saw on all
sides Hessequa kraals' and the following day more kraals are noted near the
Gouritz River. At this point the Hessequas who accompanied Schrijver are sent
away and 'out of mischief' set fire to the veld. Schrijver returned to the Castle
two months later with over 1000 cattle (Mossop 1931). In 1690 Schrijver returns
to the Riversdale area to barter for livestock, and again in 1692 and 1693.
During the last two expeditions he records that the Hessequa are increasingly
hostile and it is becoming difficult to obtain stock (Beyers 1981).
However, Jan Hartogh meets with a party of Hessequa 'Captains' at the Breede
River in 1707 who indicate they are most willing to trade their sheep and cattle.
One reason for this change in attitude may relate to the deaths of large
numbers of Khoi from war and sickness in 1665, 1672 and 1673. A major
outbreak of smallpox in 1713 is estimated to have killed ninety percent of of the
Khoi population. An entry in the Company diary dated 6th May, 1713 notes that
the 'Hottentots' are almost extinct (Godee-Molsbergen 1916). After the smallpox
epidemic many new 'Captains' are appointed and new 'tribe' names are also
recognised, for example the 'Gouritz Hottentots' (Godee-Molsbergen 1916;
Goodwin 1952; Elphick 1977). Despite these setbacks the trade in livestock
with the Hessequas continues; for example, Lichtenstein visits the Zoetemelks
Valley in 1803 and notes that the Cape government use this area to rest cattle
bought from the Hessequas and Outeniquas (Plumptre 1928).
The presence of Khoi kraals on the Riversdale Plain is recorded intermittently
by travellers to the area during the 18th C. A number of Hessequa kraals are
reported near the Gouritz River by Carel Haupt in 1752 (Godee-Molsbergen
1922) and in 1776 Hendrik Swellengrebel records seeing a number of kraals,
most consisting of six to seven huts (Godee-Molsbergen 1932). It seems likely,
though, that the power and population of Khoi people on the Riversdale Plain
had been considerably diminished by the mid 18th C., effectively opening the
land for settlement by Dutch colonists.
During the late 18th C. and early 19th C. most of the land on the Riversdale
Plain was subdivided by the Cape Government and allocated to settlers of
European origin, although scattered groups of Khoi herders were still in
evidence. On a visit to Commandant Lombards farm in 1803, J.W. Janssens
reports that Khoi kraals can still be seen on the Slang River, one hour away
from his hosts farm situated on the Duiwenhoks River (Godee-Molsbergen
1932). The application by H. van de Graaff to the Earl of Caledon for the
granting of the land rights to the farm Blomboschfontein, on which Garcia State
Forest is situated, is dated 1808.
By the mid-19th C. most of the Khoi were no longer stock owners and had been
recruited by the Dutch as servants. The demise of the Khoi is best summed up
by Lt. Henry Pemberton who records visiting a local farmer near the Gouritz
River mouth in the mid-1800s (Kruger & Beyers 1977).
The Dutch farmer at whose house we lodged, Frederic Potskeeter, a native of
the country, informed us he had lived at Croutz Riviere 17 years ... he has a few
slaves, mostly Malays with Hottentot servants ... The original inhabitants, the
Hottentots, are of peacable, mild and laborious dispositions making most
excellent household and farm servants, and free from slavery. The country is at
present but thinly inhabited by Dutch.
There are few historical references to San hunter gatherers in the vicinity of the
Riversdale Plain. O.F. Mentzel refers to 'Bushmen' living in the Swellendam and
Riversdale mountains who rob the other 'tribes' when they have the opportunity.
He also records the presence of San 'Bushmen' in the Mossel Bay area
(Mandelbrote 1944:330)
From Mossel Bay to the Gamtoos River only Bushmen are found who manage
to exist in small numbers, either in the crevices of rocks or in miserable hovels.
They live on game, roots and herbs but occasionally also help themselves to an
animal belonging to one of the colonists who are very scattered and isolated in
these regions. They seldom take more than one at a time, for they never keep a
live animal with them overnight.
In a letter written by C.H. Heese in 1929, he comments on the age of the fish
traps at Still Bay by referring to stories told by the first 'Voortrekkers' who settled
in the area between 1810 - 1820. At the time the 'Voortrekkers' arrived, there
were still 'wild Bushmen' living on the coast and operating fish traps between
the mouths of the Kafferkuils and Duiwenhoks Rivers. Parts of Heese's letter
are translated below.
Whether they belonged to the Hessequa or the Attaqua tribe will only be
determined by academics when they consider the names of the old farms in the
area such as Wankou, Kragga etc. The kitchen middens on the sand dunes
along the coast, also at Platbosch, provide overwhelming evidence that the
Khoisan who lived near Still Bay were very fond of fish; fish, both great and
small and shellfish of all types. Undoubtedly they also gathered fish that washed
up on the coast.
Heese's letter effectively sums up the problems inherent in determining the
origins of these coastal hunter gatherers. It is quite possible they were
Hessequas, dispossessed of their stock and pasture, who chose to hunt and
gather along the coast rather than be taken into service by the Dutch. On the
other hand they may have been the remnants of San hunter gatherer bands
whose territory had been gradually diminishing, both due to the expansion of
the Khoi in the area during the previous 2000 years, and now due to Dutch land
claims. No evidence of such late occupation was found at GSF although further
research in this area may help resolve this issue. At present the identity of
these last coastal hunter gatherers remains enigmatic.
3.3: Observations at Garcia State Forest: 1928-9.
Although not strictly ethnohistoric, this section deals with observations made by
C.H.Heese in the GSF area in the late 1920s. Heese named the GSF area
'Blombosch Sands' after the shifting dunesands which then characterised this
area. The information Heese provides is useful as some of the sites he
describes are covered by sand and no longer visible, suggesting the GSF area
may have been occupied more regularly than is currently apparent. Surface
collections of selected lithics, although not specifically documented, may have
been made from some of the sites currently excavated at GSF.
Numerous bones of elephant, rhino, buffalo and other animals were observed
protruding from the sand along a western route through the dunes and Heese
claims this may have been an ancient game trail. The same route had been
followed for decades by local farmers who regularly brought their cattle from
inland pastures for watering at the numerous 'bakke' or springs located on
Blombos Beach and at the base of the cliffs.
Alongside the protruding animal bones he noted at least a dozen small 'Wilton'
middens containing variously bone, shell, stone pestles, bored stones, a bone
amulet, some sites with pottery and also two skeletons protruding from the
sand. Walking eastwards along the clifftops he came across further 'Wilton'
middens composed of shell, bone, stone chips and tools in quartzite, quartz
and silcrete. In some cases the action of percolating calcium carbonates had
cemented the assemblages into a solid breccia, the effect of which was to
protect the stone artefacts from the erosive actions of the wind and sand.
Implements from the MSA 'Still Bay Culture' were also noted, as were some
ESA or 'Stellenbosch' handaxes. Below the cliff face, and at the eastern edge
of GSF1, a number of rock shelters were reported, one of which contained a
deep layer of shells and which Heese (n.d.) describes as a 'Wilton station'.
From a nearby shelter a skeleton was recovered by the local schoolmaster.
3.4: Summary
At about the same time that early European explorers were recording their
observations at the Cape, GSF 8 and GSF 9 were occupied by indigenous
people. Ethnohistoric documents, if used judiciously, can be a useful adjunct in
the interpretation of the later occupations at GSF.
In the historic period two distinct economic groups can be recognised at the
Cape, namely hunter gatherer egalitarian San who did not possess livestock
and pastoralist Khoi groups, hierarchically structured, who accumulated wealth
by managing sheep and cattle but also hunted and gathered.
Representatives of both groups are recorded historically within the Riversdale
Plain region. Reports of the Hessequa Khoi are more numerous than that of
San, presumably because of greater Khoi/ colonist contact due to an active
trade in livestock in the 17th- and 18th C. Contact was made with Khoi
pastoralists both at the coast and inland suggesting that the Khoi made use of
grazing opportunities in a range of environments. Although the hierarchical
structures existing within the Khoi groups on the Riversdale Plain are only
hinted at, the Dutch recognised that a distinct group, known as the Hessequa,
occupied the coastal foreland between the Breede and Gouritz Rivers. During a
visit by a Hessequa 'Captain', Gammou Kouchama, to the Castle at Cape Town
in 1687, reference is made to the 'king' of the Hessequas, suggesting that some
unity existed at that stage among the Riversdale Plain pastoralists. War and
disease, particularly smallpox in the early 18th C., decimated a large section of
the Khoi population in most areas of the Cape leading to demographic changes
and the emergence of new 'tribal' groupings. At about the same time Dutch
farmers were expanding into the southern and southwestern Cape, filling the
land vacuum left by the diminished and fragmented Khoi survivors. By the early
19th C. only small surviving groups of Khoi pastoralists were scattered in the
southern Cape and many had been taken into service by the Dutch settlers.
Few historic records relate the presence of San in the southern Cape, although
accounts from elsewhere in the Cape are more numerous. It seems likely that
parts of the Riversdale Plain area were inhabited by San but due to the
remoteness of the regions they occupied, contact between San and travellers
or colonists was an infrequent occurrence. However, there are a few incidents
where Dutch and Portuguese explorers, in the 16th and 17th C., report the
presence of indigenous folk at the coast who do not possess stock and live on
shellfish and fish. In the Riversdale and Mossel Bay areas, O.F Mentzel,
indicates that there were still San people living on the fringe of Khoi society in
the late 18th C. In the region of GSF, Heese (n.d.) relates the presence of
coastal hunter gatherers in the early 1800s. The origins of these later
'Bushmen' are not clear, they may have been dispossessed stock owners or
surviving San.
The fate of the hunter gatherer San in the southern Cape is mostly unrecorded.
This is not surprising as San survival depended on escaping detection by Khoi
and colonist, by whom they were frequently hunted. In the Eastern Cape
Mentzel records that on two separate occasions traps were set by Dutch
farmers to trap 'Bushmen'. In one instance in 1775, 122 'Bushmen robbers'
were reported shot. (Mandelbrote 1944).
Based on ethnohistoric accounts of the Riversdale Plain it is clear that the Khoi
were well established in the region by the 17th C. and probably had been for a
considerable time prior to this date. The presence of sheep bone at GSF 8,
radiocarbon dated at 1960 ± 50 B.P,. indicates a long history of pastoralism in
this region. GSF 8 was repeatedly occupied during the last two millennia and it
seems certain that the occupants of this site witnessed some of the major
changes which occurred in the the Riversdale Plain region at this time, firstly
the advent of livestock at around 2000 B.P., and secondly, the arrival of the first
4.1: Introduction
This chapter provides a synopsis of the archaeology of the southern Cape
primarily, and sets the background for the chapters that follow. As the oldest
excavated site at GSF is radiocarbon dated at 6960 ± 70 B.P. and the youngest
at 290 ± 20 B.P. the discussion in this chapter centres around the archaeology
of the Mid- to Late Holocene period.
Assemblages containing ethnographically known items of material culture
appear during the last 40,000 years in southern Africa, although it is only during
the Holocene that characteristic Later Stone Age (LSA) elements commonly
occur at most sites. The similarity between articles typically occurring in LSA
sites (see below and cf. Deacon, J. 1984b:221-222) and the material culture of
the historically known San and Khoi suggests that LSA sites represent the
prehistory of these indigenous people.(cf. Goodwin & van Riet Lowe 1929;
Clark 1959; Inskeep 1967; Deacon H.J. 1972, 1976; Parkington 1972, 1977;
Phillipson 1977; Deacon, J. 1984b; Rightmire 1984). However, the material
culture of the LSA should not be viewed as an unchanging static entity but
rather as a continuing tradition, subject to gradual modification and adaptation
through the effects of diffusion and migration (Deacon, J. 1984b).
For most of the Holocene, San hunter gatherers had exclusive use of the
coastal and interior regions of the Cape. Pastoralism was introduced to the
Cape at around 2000 B.P., possibly by herder groups originally from northern
Botswana. San social organisation and subsistence strategies at the Cape were
disrupted by the economics of herding, possibly only marginally initially, but by
historic times hunter gatherers had been forced into peripheral areas (Smith
1986; Parkington 1987; Parkington et al 1988; Hall 1986).
Iron Age, Bantu-speaking agropastoralists arrived in South Africa about 1600
years ago but were confined by the nature of their crops to the summer rainfall
areas in the eastern and central regions. In the Cape they only penetrated as
far south as the Great Fish River, 800 km east of Cape Town (Maggs 1984).
Although contact was made with local San groups the effects of this interaction
are only archaeologically visible within the immediate area (Hall 1986). More
extensive contact may have occurred between early Khoi pastoralists and Iron-
age agropastoralists in the western part of the sub-continent - at about 1600
B.P. both groups had domesticated sheep and cattle and made pottery (Sealy &
Yates 1994).
As a result of archaeological excavations at a number of sites in the Cape over
the last sixty years a substantial base of information has been built up
documenting the occupation of the region by indigenous people during the
Holocene. The results of this research, particularly in the southern Cape, are
discussed in this chapter. In later chapters the results of the excavations at
GSF are interpreted with reference to our present understanding of the
archaeology of the region.
4.2: History of LSA archaeology
Based largely on recovered lithic implements, the LSA tradition in South Africa
was first defined in the 1920s by Goodwin (1926) and Goodwin & van Riet
Lowe (1929). They saw the LSA as being essentially similar to the Mesolithic in
Europe, divided it into two main industries, the Wilton and the Smithfield,
placing the origins of the Wilton microlithic industry to the north, and proposed a
'successive wave' or invasion theory to account for its arrival in the south
(Goodwin & van Riet Lowe 1929).
In the following decades an expanding data base allowed for more detailed
inter-site comparisons. A number of LSA coastal and near-coastal cave sites in
the Cape were excavated (e.g. Dreyer 1933; Goodwin 1938; Drennan 1937;
Goodwin 1946; Grobbelaar & Goodwin 1952; Fagan 1960; Louw 1960), as well
as open station shell middens, or 'kitchen middens' (e.g. Colson 1905; Goodwin
1946; Rudner & Rudner 1954; Mabbutt et al 1955) (cf. Fig. 2.1). However, the
orientation of most interpretations remained lithocentric (but cf. Deacon &
Deacon 1963; Wells 1965) and the 'invasion' hypothesis was retained. Few
attempts were made to test this 'invasion' model, most explanations resting on
what J. Deacon (1984a) terms 'post-hoc' accomodative arguments (e.g. Clark
1959, 1966; Sampson 1967, 1972, 1974).
Re-excavation and subsequent interpretation of the Wilton rock shelter in the
Eastern Cape demonstrated ongoing stone artefact changes through time and
the Wilton tradition was likened to an ontogenetic model which stresses cultural
change associated with continuity (Deacon J. 1972).
Ecological modelling formed a composite part of most research strategies in the
1970s and hypotheses were proposed which related climatic and environmental
variables with perceived cultural, subsistence and demographic change. Early
examples of this new approach in the Southern Cape were at Nelson Bay Cave
(Klein 1972a,b, 1974) and Melkhoutboom (Deacon, H.J. 1972, 1976), where
changes in artefact manufacture were linked to a shift in hunting patterns, rising
sea levels and environmental changes in the Late Pleistocene and Early
Holocene. Although ecocentric 'cause and effect' modelling is now regarded by
some as too simplistic, ignoring as it does the 'human' factor as a catalyst for
change, it remains an essential determinant in defining the parameters within
which 'human' choice could have been exercised (cf. Deacon 1984; Bailey &
Parkington 1988; Mazel 1989; Wadley 1989).
The contributions of Klein (1972a, b, 1974) and H.J Deacon (1972, 1976) led to
the naming of a three stage system in the Later Stone Age of the Southern
Cape, namely Robberg, Albany and Wilton, with a further Pottery Wilton
subdivision in the post 2000 B.P. period, the latter associated with the arrival of
pastoralist people (cf. Deacon, J. 1984a). These culture-stratigraphic
sequences can be broadly applied to all known LSA assemblages situated
within, and to the seaward side, of the Cape Folded Belt mountains in the Cape
In a later study the correlation between environmental change and stone
artefact manufacture was tested at three sites in the Southern Cape, namely
Nelson Bay Cave, Kangkara and Boomplaas. Although the timing of the
changes for the Robberg, Albany and Wilton Industries are similar at all sites,
some assemblages reflected minor variations in toolkits which were site
specific. On the other hand, macroevolutionary or significant changes in lithic
tool traditions cut across environmental boundaries and subsistence strategies,
and a direct relationship between environmental change and stone tool
manufacture could not be demonstrated (Deacon J. 1984a).
Parkington's (1972, 1976) pioneering study on floral and faunal remains found
at coastal and inland sites in the Western Cape is linked to ecological
modelling. The results suggest that some LSA people followed seasonal
subsistence rounds. Archaeological evidence from other sites in the southern
and south western Cape support a seasonal subsistence model (cf. Shackleton
1973; Deacon 1969, 1976; Klein 1973; Avery 1976; Buchanan et al 1978;
Parkington 1991).
4.3: The Holocene Period (8,000 - 2,000 B.P.):
At around 8000 - 7000 B.P. the Albany Industry is broadly replaced by the
microlithic Wilton Industry in the southern and eastern Cape and is
accompanied by a florescence in the number of occupation sites (Deacon
The type site of the Wilton culture is situated near Alicedale in the eastern
extension of the Cape Folded Belt Mountains and was first excavated by Hewitt
(1921), and later by H.J. Deacon. The lithic assemblage from the Wilton site
was described in terms of an ontogenetic model displaying typical
formative/maturity/decline phases (Deacon, J. 1972). The formative range of
the new Wilton is at around 8000 B.P. with little variety in tool types and a low
artefact density. Scrapers are small and round, mainly in chalcedony. The
climax phase lies between 5000 - 2500 B.P., with a high density of artefacts and
a wide variety of tools recorded. Formal tools are dominated by a selection of
small scrapers less than 20 mm in length with a mean width/length ratio of 1:1.
Backed microliths include blades and segments. The predominant raw material
is silcrete, with some quartz and chalcedony. After around 2300 B.P. the
scrapers are more similar to those in the early layers as is the tool inventory.
The artefact density decreases and although silcrete remains dominant, there is
increasing use of shales and quartzite. The Wilton system is in decline after
2000 B.P., scrapers become more variable , the number of tool types declines,
there are no adzes and few backed tools, and a wide variety of raw materials is
evident. These changes coincide with the arrival of pottery makers/ herders and
it is possible that they replaced the existing population at the Wilton site. The
changes in the faunal component at the Wilton site are broadly coincident with
those of the lithics. In the formative phase, small game are mainly present and
marine shells indicate coastal contact. This continues through the climax phase
with evidence of some freshwater mussels. After 2300 B.P. there is a marked
decline in marine shells and a pronounced shift to the freshwater mussel, Unio
The trajectory of lithic change at Wilton is broadly consistent with a number of
southern Cape sites. Amongst others, these include Oakhurst (Goodwin 1938),
Glentyre (Fagan 1960), Melkhoutboom (Deacon, H.J. 1976), Kangkara
(Deacon, J. 1984a) and Boomplaas (Deacon, J. 1984a). Lithic assemblages
from a number of coastal sites, however, reflect considerable variation from the
Wilton pattern. The formal tool component at Nelson Bay Cave (Inskeep 1987)
during the Holocene comprises scrapers, backed scrapers, segments, some
doubtful backed bladelets, tanged points, drills, reamers, bored stones, two
doubtful adzes, sinkers, grooved stones and ochre pencils, but all in relatively
small numbers. This is despite the fact that the total number of stone artefacts
at the site exceed 47,000 units; interestingly 82% of these are of quartzite.
During the period 6000 - 3300 B.P. the majority of scrapers are of chalcedony,
followed by quartz with some of quartzite. The mean size in the former two
categories is <13mm, in the latter >34mm. Only one silcrete scraper is reported.
The large size of the quartzite scrapers led Inskeep (1987) to suggest that they
were used to perform different functions from the smaller scrapers. After 3300
B.P. quartzite scrapers dominate and small scrapers in fine grained materials
are virtually absent. Of the 40 segments recovered all are of chalcedony and
predate 3300 B.P.. One retains traces of mastic suggesting hafting. An unusual
component are the 8 backed scrapers which date to around 4500 B.P.
Similar tools are recorded from only a few other sites all situated to the west of
Nelson Bay. These sites are Matjes River (Louw 1960) and Byneskranskop
(Schweitzer & Wilson 1982), and a small number may be present at Oakhurst
(Schrire 1962). At Matjes River (Louw 1960) the range of tools in Layer C are
more varied than in the mid-levels at Wilton and include more saws, lance
heads, more bone tools, shell ornaments and painted grave stones. The largest
numbers of backed scrapers (>10,000) are reported from undated surface
collections made by C.H.Heese (n.d.) in the 1930s at the open site,
Brakfontein, located 12 km from the coast and 20 km to the northwest of the
Garcia State Forest sites. Over 30,000 LSA artefacts which include large
numbers of small scrapers, segments and adzes were collected but these have
yet to be formally studied.
Broad similarities in the trajectory of artefact change in the Mid/Later Holocene
are evident, but there is considerable variation between inland and coastal
assemblages, and also between sites in the western and eastern zones of the
southern Cape. This agrees with H.J. Deacon's (1976:171) observations
... artefact occurences to the west of the Gouritz River fall outside any concept
of a specific Wilton content unit
Apart from the lithics a wide variety of tools, decorated items, ornaments and
grave goods have been recovered from mid-Holocene sites. Bone tools include
arrow points, foreshafts and linkshafts, 'spoons', spatulas, and fish hooks. A
variety of beads, pendants and other ornaments are made in shell, stone and
bone. Ostrich egg shell containers and fragments are sometimes decorated,
tortoise shells are worked to form bowls and scoops. Due to the excellent
preservation of organic materials at some sites a range of items in leather,
wood and plant fibre have been recovered. These include wooden digging
sticks and pegs, fire sticks, bows, arrows, fragments of sewn leather and
rope/netting. A few artefacts mounted in mastic provide clear evidence of the
hafting of microlithic tools.(Goodwin 1938; Louw 1960; Wells 1965; van Noten
1974; Deacon, H.J. 1976; Deacon, J. 1984a, b; Inskeep 1987).
Burials may be accompanied by a range of grave goods including grindstones/
hammerstones, painted stones, palettes, bone awls, ornaments such as beads
and pendants, ochre pencils and powdered ochre, tortoise carapace bowls (e.g.
Louw 1960; Deacon 1979:69-70 Inskeep 1987). Grave goods are most
commonly found associated with burials in the southern and eastern Cape, and
although the symbolic significance of these is not understood it may illustrate
social differences between people in this area and those in the western Cape.
Subsistence strategies broaden in the Holocene to include a wide range of
'small' food parcels and may be related to both higher population densities and
available resources. At most sites large bovids are rare, one exception is the
Uniondale site in the eastern Cape where large numbers of Cape Buffalo have
been recovered (Brooker 1989). In the Wilton layers at Nelson Bay Cave the
emphasis is on smaller terrestrial animals, especially small bovids (Klein 1973;
Inskeep 1987), a pattern repeated at a number of other sites including Wilton
Rock Shelter, Melkhoutboom, Andrieskraal I, Kangkara, Boomplaas, Matjes
River and Oakhurst (Louw 1960; Deacon & Deacon 1963; Deacon, J. 1972;
Deacon, H.J. 1976; Deacon, J. 1984a).
With the mid-Holocene rise in sea temperature, Perna perna replaces C.
meridionalis at Nelson Bay. A wider range of molluscs from the whole intertidal
zone are also represented including Patella sp., Haliotis sp., Turbo sarmaticus,
Oxystele sp., Donax serra and Burnupena (Inskeep 1987). Klein (1973)
suggests this is due to greater efficiency at cropping marine resources, or
simply not minding getting wet, and not to environmental changes. It is
interesting to note that at the Andrieskraal I site, situated 50km from the coast
in the Gamtoos River Valley, shells from these same species were recovered,
suggesting that coastal resources formed a part of the occupants subsistence
strategy (Deacon, J. 1974). At around 3000 B.P. a similar molluscan species
composition is evident at the Gordon's Bay shell midden with the exception of
the continued presence of C. meridionalis. On the western Cape coast at
Eland's Bay there is an hiatus in occupation between 7800 - 4400 B.P., possibly
due to low rainfall and higher sea levels, but thereafter marine molluscs are
heavily exploited, particularly C. meridionalis but also limpets and whelks
(Parkington 1987; Parkington et al 1988).
By the Mid-Holocene the proportions of marine species, which includes fish,
birds and mammals increases rapidly in relation to terrestrial fauna at Nelson
Bay Cave. Technological innovations include fish gorges and net sinkers and
the application of these is evident in both the species variety and numbers of
fish caught at Nelson Bay (Klein 1973; Inskeep 1987). Large numbers of fish
bones were also recovered at Oakhurst (Goodwin 1938) and fish remains are
mentioned at Matjes River (Louw 1960).
Systematic collection of a range of plant foods, particularly geophytes, is
evident at a number of inland sites in the southern Cape, but are virtually
absent at coastal sites. It seems unlikely this is due to preservation as seaweed
bedding is found at Eland's Bay and at Hoffman's Cave, and milkwood seeds at
the Gordon's Bay shell midden. At coastal sites the emphasis appears to be on
shellfish rather than plant foods. (Parkington 1972, 1976, 1987; Deacon 1972;
Klein 1973). Common species at the inland Melkhoutboom site include the
geophytes Watsonia and Hypoxis. At the time of new growth during May/June
the corms of these plants are unusable and H.J. Deacon (1976) suggests this
period may have coincided with coastal visits (also cf. Parkington 1972, 1976).
4.4: The Late Holocene (2,000 B.P. - Historic Period):
Perhaps the most characteristic feature of this period is the arrival of a new
economic system at the Cape, namely animal husbandry. In the historic period
the widely dispersed Khoi groups in South Africa all spoke similar dialects
which have linguistic affinities to the Central Bush family of languages spoken
by hunter gatherers in northern Botswana and southwestern Zimbabwe
(Westphal 1963). Based on ethnographic data and Westphal's linguistic
evidence, Elphick (1977) proposed two alternative routes for the arrival of
pastoralists in the Cape. The initial movement was from northern Botswana
towards the Orange River. One element continues along the Orange River in a
westerly direction to the Atlantic coast, then south to Namaqualand and finally
to the western Cape; the other continues south through the Central Karoo.
Earlier, Cooke (1965) had proposed similar origins but favoured the western
coastal route. However, there are problems with the linguistic evidence as most
Khoi had been decimated before their language was adequately documented
(Inskeep 1978; Deacon, J. 1984b; Klein 1986; Smith 1986).
An alternative scenario is that a variety of routes were followed at different
times and that consequently there were multiple introductions of livestock to the
Cape. Sheep may have been introduced at around 2000 B.P. via the west coast
route, but cattle could have been obtained from Iron Age people on the east
coast and only brought to the southern and western Cape at a later date (Klein
1986). Archaeologically we would therefore expect earlier dates for sheep on
the west coast but some of the earliest associated radiocarbon dates come
from the southern and southwestern Cape sites of Die Kelders (1960 ± 85 B.P.),
Nelson Bay Cave (1930 ± 60 B.P.) and Byneskranskop (1880 ± 50 B.P.) (cf.
Schweitzer 1974, 1979; Schweitzer & Wilson 1982; Klein 1986). At the west
coast site, Kasteelberg A, sheep may date to 1860 B.P. (Smith 1986).
Sealy and Yates (1994) argue that the dates for sheep at the above sites are
based on indirect dating, that is by association of faunal materials with
recovered charcoal from within stratigraphic layers. Direct accelerator
radiocarbon dating of sheep bone from six sites in the western and
southwestern Cape produced a younger set of dates; Kasteelberg A 1630 ±60
B.P. & 1430 ±55 B.P., Die Kelders 1325 ±60 B.P. & 1290 ±60 B.P. and
Byneskranskop 1370 ± 60 B.P. The oldest dates are from Spoegrivier in the
northwestern Cape at 2105 ± 65 B.P. The results, they argue, point to a lack of
evidence for stock-keeping in the southernmost Cape prior to 1600 B.P. and
tentatively suggest a southerly movement of stock may be detected.
Two samples of sheep bone, excavated from GSF 8, Layers 5 & 6 were
submitted for accelerator radiocarbon dating. The dates obtained, 1880 ± 55
B.P. and 1960 ± 50 B.P. contradict the Sealy & Yates (1994) model for a 1600
B.P. date and indicate that sheep were present in the southern Cape at around
2000 B.P. (cf. Chapter 7).
The wild ancestors of domestic stock, namely sheep and cattle originated in
northeastern Africa or southwestern Asia. Goats were unknown in the southern
and southwestern Cape during the prehistoric period and there is no evidence
of any indigenous species of animals being domesticated by the Khoi. Klein
(1986) suggests this is clear evidence that the herding concept was introduced
to the Cape from elsewhere, but whether this introduction was through a
process of diffusion or through the migration of herders colonising new areas is
unclear. Support for the latter model is provided by Smith (1986) who argues
that pastoralism is a coherent strategy which requires specific social relations
between people and animals and that consequently the egalitarian ethic
practised by hunter gatherers was not consistent with the economics of animal
husbandry. A combination of both migration and acculturation by resident
hunter gatherers is recognised by some archaeologists as a more plausible
explanation (cf. Schrire 1984; Klein 1986). It seems likely, therefore, that by the
historic period Khoi ancestors could be traced to both indigenous people and
migrant groups, but clearly there still were, and possibly always had been,
separations both economically and socially between the San and Khoi (cf.
Parkington 1984; Parkington & Mills 1991).
Recognising the Khoi/San distinction in the archaeogical record has proved
problematic as herders and hunter gatherers, despite their economic
differences, also shared many cultural and physical characteristics, and the
occupation site residues left by both appear very similar (cf. Schweitzer 1974;
Klein 1986). The general range of tools, ornaments and other cultural materials
found in the Mid/Late Holocene persist in the post-pottery period and there is
little evidence of artefacts specifically linked to a herding economy. One
exception may have been wooden or grass bowls and containers used by the
Khoi for storing milk, but no archaeological evidence of these items has been
found (Deacon 1984b).
Stone artefact traditions show broad continuity at a number of sites after the
arrival of the Khoi. At Boomplaas microlithic artefacts continue to be
manufactured during the Late Holocene and are found associated with pottery
(Deacon, H.J. et al 1978). Some changes are evident at coastal shell middens
after 2000 B.P. where flaked quartzite cobbles and large, unretouched flakes
become more common, although interestingly this trend towards larger
quartzite tools is evident at the Gordon's Bay Midden before 3000 B.P. (van
Noten 1974). Formal tools are absent at some coastal sites (e.g. Deacon &
Deacon 1963; Avery 1974; Robertshaw 1977, 1979). A comparison of the pre-
pottery and post-pottery assemblages shows the most common differences are
a reduction in the frequency of microlithic backed tools, especially segments,
and decreasing use of fine grained raw materials (Deacon 1984b).
The introduction of pottery at the Cape is generally associated with the arrival of
the first pastoralists, although some authors note that early pottery dates are
more common than those for sheep (Deacon 1984b, Klein 1986). Based on a
date of 1600 B.P. for the introduction of livestock to the southernmost Cape, and
on secure pottery dates clustered around 2000 B.P. from a number of sites,
Sealy and Yates (1994) speculate that the arrival of pottery may predate
pastoralism. Pottery is found in Layer 5 at GSF 8 in association with sheep
bone radiocarbon dated at 1960 ± 50 B.P. suggesting that the introduction of
pottery to the Cape may coincide with that of stock. Klein (1986) notes that in
every assemblage with domestic stock, pottery is present. Although pottery
does occur in the absence of domestic stock, Klein suggests this is due to
sampling error as a result of the small size of the faunal sample. Distinguishing
Khoi and San sites on this basis would thus seem to be clear cut, but this is not
the case. Historically, it is known that part of the enmity between San and Khoi
was due to hunters stealing cattle and sheep from herders (cf. Raven-Hart
1967), and livestock remains may occur at both herder and hunter gatherer San
sites. Pottery was used, and possibly made by the Khoi and San (cf. Schapera
One method suggested by Schweitzer (1974) for resolving this issue is to
examine the sex and age of domestic stock. At the Die Kelders site, Schweitzer
found that 78% of the 23 sheep in Layer 2 (Radiocarbon dates 1465 B.P & 1600
B.P.) were juveniles and predominantly male, and 3 were more than 4 years old.
Based on the assumption that in order to maintain a successful breeding herd,
only the young males or old animals would be slaughtered by the Khoi, he
concluded that Die Kelders was a Khoi site. If the assemblage had, on the other
hand, been composed mainly of ewes, this may suggest the sheep were stolen
by San people. Stock samples of this size are very rare however, and we have
no record of the sex/age composition of Khoi flocks. However, the dietary
elements in the layers which contain sheep at Die Kelders, are the same as
those found in the lower pre-stock levels, and include dune mole rats, tortoises,
small antelope, seals, marine birds, fish and molluscs (Schweitzer 1974). On
the basis of a lack of clear evidence for a Khoi presence, and the physical
inaccesibility of the site to herders with stock other authors believe that Die
Kelders was occupied by San who were simply proficient at stealing or bartering
sheep from the Khoi (Avery, G. 1976; Wilson pers. comm.).
On the basis of the rarity of domestic animal bones, Klein (1986) suggests that
in the Late