Paleoenvironment and Archaeology of Drotsky’s Cave: Western Kalahari Desert, Botswana

Abstract

Test excavations conducted at Drotsky’s Cave have provided important new information on the paleoenvironment and archaeology of the western Kalahari desert during the late and terminal Pleistocene. An occupation layer dated to the terminal Pleistocene was rich in Late Stone Age artefacts, pieces of ostrich egg shell, the remains of carnivorous bullfrogs, springhare, and other fauna. A detailed sediment study, along with the evidence of Angoni vlei rat, climbing mouse, an aquaticXenopusfrog, and side neck turtle confirms that conditions were for the most part, substantially more moist than at present between approximately 30,000 and 11,000 years ago. Analysis of a diatom assemblage dated to the terminal Pleistocene implies that the currently dry Gcwihaba Valley was most likely flowing for much of the year. Our evidence supports findings made at other localities in the Kalahari documenting the existence of especially moist conditions during the terminal Pleistocene.

Full text

Journal of Archaeological Science (1996) 23, 7–22
Paleoenvironment and Archaeology of Drotsky’s Cave:
Western Kalahari Desert, Botswana
L. H. Robbins and M. L. Murphy
Department of Anthropology, Michigan State University, East Lansing, MI 48824, U.S.A.
N. J. Stevens
Department of Anthropology, Suny Stony Brook, NY 11794-4364, U.S.A.
G. A. Brook and A. H. Ivester
Department of Geography, University of Georgia, Athens, GA 30602, U.S.A.
K. A. Haberyan
Department of Biology, Troy State University, Troy, AL 36082, U.S.A.
R. G. Klein
Department of Anthropology, Stanford University, Stanford, CA 94305-2145, U.S.A.
R. Milo
Department of Anthropology, University of Chicago, Chicago, IL 60637, U.S.A.
K. M. Stewart
Canadian Museum of Nature, P.O. Box 3443, Station D, Ottawa, Ontario, Canada
D. G. Matthiesen
Department of Zoology, University of Florida, Gainsville, FL 32611, U.S.A.
A. J. Winkler
Shuler Museum of Paleontology, Southern Methodist University, Dallas, TX 75275, U.S.A.
(Received 23 February 1994, revised manuscript accepted 11 August 1994)
Test excavations conducted at Drotsky’s Cave have provided important new information on the paleoenvironment and
archaeology of the western Kalahari desert during the late and terminal Pleistocene. An occupation layer dated to the
terminal Pleistocene was rich in Late Stone Age artefacts, pieces of ostrich egg shell, the remains of carnivorous
bullfrogs, springhare, and other fauna. A detailed sediment study, along with the evidence of Angoni vlei rat, climbing
mouse, an aquatic Xenopus frog, and side neck turtle confirms that conditions were for the most part, substantially
more moist than at present between approximately 30,000 and 11,000 years ago. Analysis of a diatom assemblage dated
to the terminal Pleistocene implies that the currently dry Gcwihaba Valley was most likely flowing for much of the year.
Our evidence supports findings made at other localities in the Kalahari documenting the existence of especially moist
conditions during the terminal Pleistocene. ?1996 Academic Press Limited
Keywords: DROTSKY’S CAVE, SOUTHERN AFRICAN PREHISTORY, TERMINAL
PLEISTOCENE/HOLOCENE CLIMATE CHANGE, KALAHARI DESERT.
7
0305-4403/96/010007+ 16 $12.00/0 ?1996 Academic Press Limited

Introduction
Natural rock shelters and caves with long paleo-
environmental and Stone Age archaeological
sequences are relatively rare in the Kalahari in
comparison to neighbouring areas of Zimbabwe and
South Africa. For this reason our overall knowledge of
paleoenvironments and Stone Age archaeology in the
Kalahari is generally based on composite views drawn
from various open sites, which often lack stratigraphy,
or are surface sites, rather than from single localities
with deep deposits. In fact, major rock shelter excava-
tions have only been carried out at the Depression and
White Paintings sites in the Tsodilo Hills and the only
excavated caves are the lower Male Hill Cave at
Tsodilo and Drotsky’s Cave located in the Gcwihaba
Hills (Robbins, 1990; Robbins & Campbell, 1989;
Robbins, 1991; Campbell & Robbins, 1993; Yellen,
Brooks, Stuckenrath & Welbourne, 1987; Pickford,
1990), (Figure 1).
In 1969, John Yellen, accompanied by a group of
!Kung San from the Dobe area, conducted archaeo-
logical research in the northeast entry chamber of
Drotsky’s Cave (Figure 2). He excavated a series
of four 5 ft square test pits (Yellen et al., 1987). One of
the test pits (A) yielded 61 stone artefacts as well as
numerous pieces of ostrich egg shell, fauna and char-
coal. The artefacts were found between 28 and 40 in.
below a datum that was established at the base of the
letter Y in Drotsky’s name, which was carved by
Martinus Drotsky into the side of a large boulder
in the entry chamber in 1932. (Drotsky’s name was
still clearly visible when we visited the cave in
1991). Charcoal recovered from 31 in. below this
datum in square A yielded a radiocarbon date of
12,200&150 . This represented the first discovery of
in situ terminal Pleistocene archaeological material in
the western Kalahari, and as such was important to the
basic question of the length of time Khoisan ancestors
had lived in the area. The other excavation units
apparently did not yield artefacts.
Over 90% of the lithic material recovered by Yellen
was débitage, mainly consisting of silcrete. The absence
of microliths and lack of evidence of blade technology
suggested that the Drotsky’s cave area was part of
the ‘‘non-microlithic’’ complex described by Deacon
(1984) for interior parts of southern Africa during the
terminal Pleistocene/early Holocene.
Drotsky's
Cave
Orange
R.
25°S
Etosha
Pan
Botswana
30°S
20°S
35°S
Namibia
Windhoek
5
4
3
21
8
9
5
610
Gaborone
Africa
Molopo R.
Johannesburg
Swaziland
Lesotho Durban
South Africa
Cape Town
500
0
km
Figure 1. Southern African locality map.
8 L. H. Robbins et al.

Yellen also noted that his !Kung San field assistants
reported traditions of using the cave to obtain honey,
but they did not camp in it. Finally, it was suggested on
the basis of faunal remains that a ‘‘generalist’’ strategy
of food procurement was likely, similar to that
ethnographically observed among the !Kung.
1991 Research
Drotsky’s Cave is in the dolomite marbles of the
Gcwihaba Hills, that consist of six outcrops straddling
the now dry Gcwihaba Valley. The hills are sur-
rounded by low, longitudinal sand dunes, aligned
WNW–ESE, that have been stabilized by vegetation.
Drotsky’s Cave is confined within one of the hills, the
west-facing slope of which is a steep rocky face where
the two entrances to the cave are located (Cooke &
Baillieul, 1974). The present climate of the area is
semiarid. Annual precipitation at Maun 230 km east,
Ghanzi 200 km south, and Shakawe 150 km north is
491, 401, and 520 mm respectively and occurs almost
exclusively in the Austral summer months. The first
paleoenvironmental investigations at the cave were
begun in 1972, and over the past 20 years the cave has
been an extremely valuable source of paleoenviron-
mental data for the Ngamiland region (Cooke, 1975;
Grey & Cooke, 1977; Cooke & Verhagen, 1977; Cooke
& Verstappen, 1984; Shaw & Cooke, 1986).
Following the encouragement of Yellen, another test
pit was excavated at Drotsky’s Cave in 1991. We hoped
to enlarge the artefact and faunal sample by sieving,
which was not done in the original study. The ultimate
aim of our work at Drotsky’s Cave is to provide a data
base that can be compared to ongoing multidisci-
plinary research into the paleoenvironment and pre-
history of the Tsodilo Hills, which are located about
125 km to the north (Robbins, Murphy, Stewart,
Campbell & Brook, 1994).
Our test pit, limited to 1 m
2
, was located next to
Yellen’s pit D not far from the rear wall of the entry
chamber. Figures 2 and 3 show the location and
stratigraphy of our test pit. This pit was excavated to a
depth of 130 cm below the surface, or 186 cm below the
datum used by Yellen. The base of the deposit has not
yet been established and there may well be deeper
occupation levels. The sediments consist largely of
Kalahari sand described in detail below. The most
striking result of our excavation was the discovery of a
continuous layer of charcoal between 50 and 80 cm
below the surface (Figure 3). This charcoal layer was
Figure 2. Drotsky’s Cave: view of northeast entry chamber showing 1991 test pit and remnants of pit D.
Paleoenvironment and Archaeology of Drotsky’s Cave 9

also quite clearly the main artefact and faunal layer
revealed by the excavations. The base of the charcoal
layer was marked by a white/grey ash zone, underlain
by a red fire-oxidized area. Levelling from the datum
used by Yellen established that the top and base of this
charcoal layer correspond closely to the two charcoal
layers discovered by Yellen and his pit D; (i.e. the top
of Yellen’s upper layer matches the top of our layer
and the bottom of ours matches the bottom of his
lower layer.)
The charcoal layer’s general nature and shape
suggests to us that it did not correspond to a single
hearth, but may, instead, represent intensive super-
imposed fires. The charcoal was exceptionally well-
preserved with many pieces measuring about 4 cm.
Some pieces had sub-pointed and rounded ends sug-
gestive of deliberate cutting, but inspection by a
paleoethnobotanist did not reveal cut-marks or other
definite evidence of work (K. Egan, pers. comm.).
Dates
The following radiocarbon dates were obtained from
the 1991 test pit:
(1) Between 20 and 30 cm below the surface (dispersed
charcoal scatter found in sand) dates to 5470&90 
(Beta 50162);
(2) 50 cm (top of the above-mentioned charcoal layer)
dates to 11,240&60  (Beta 50163);
(3) Between 70 and 80 cm (base of the charcoal layer)
dates to 12,450&80  (Beta 47862).
Deposits between 80 and 130 cm yielded very little
charcoal and have not been radiocarbon dated. Esti-
mated ages for the deposits beneath the charcoal layer
will be presented below in a discussion of the
sediments.
The above dates reveal that Drotsky’s Cave was
utilized over an extensive period. However, the main
artefact and charcoal bearing deposits between 50 and
80 cm appear to have accumulated over approximately
1200 years. Although the date obtained from the base
of the charcoal-rich layer is in remarkably good agree-
ment with Yellen’s date, it should be noted that his
dated charcoal sample from Square A was recovered
from approximately 47 cm above our sample in terms
of our comparative measurements to the site datum
(the Y in Drotsky’s name). This difference may result
from sloping deposits, greater buildup of deposits
toward the entrance, or other unknown factors. What-
ever the case may have been, the similarity in the dates
suggests that terminal Pleistocene material is widely
distributed in the entrance area of the cave. As will be
shown, it is significant that the terminal Pleistocene
dates are in close agreement with the age of calcrete
recovered from the Gcwihaba Valley within 2 km of
the cave. The calcrete contains fresh water diatoms
which suggest significant stream flow at the time they
were deposited.
Artefacts
Late Stone Age artefacts from Drotsky’s Cave are
discussed below relative to the terminal Pleistocene
charcoal layer. Most of the interpretations are tentative
until extensive systematic excavations are conducted.
Although detailed reports on the fauna follow the
discussion of the artefacts, we refer to some of the
general information on the fauna in the artefact
discussion for the purpose of enhancing the cultural
interpretation.
The deposits above the charcoal layer (0–50 cm):
Casual use of the cave during the Holocene
Nothing was found in the excavation of the first 50 cm
that was clearly indicative of an occupation surface,
nor was there a concentration of material suggesting
that the cave was used as an actual camp. In fact, very
little cultural material or fauna was recovered from any
of the individual 10 cm levels above the charcoal layer.
For this reason, we present an overview of the infor-
mation while the specific lithic data by levels is shown
in Table 1.
It may be significant, given the current interest in the
issue of the degree of contact between early Iron Age
peoples and Kalahari foragers, that no potsherds or
other traces of contact with Iron Age peoples were
found in the uppermost deposits (Wilmsen & Denbow,
1990). This lack of early Iron Age material might
1m
cm
0
100
50
Red/brown sand
Red/brown sand
Light brown sand
Ash lens
Red oxidized sand
Calcite layer
Grey/brown/black sand
130
Figure 3. Drotsky’s Cave: cross section of the east face of the 1991
test pit.
10 L. H. Robbins et al.

support the view that foraging peoples in the interior of
the Kalahari were little influenced by Iron Age peoples.
However, this negative evidence could easily be a
product of the small area excavated and/or a situation
where only a few people, who rarely lost or discarded
their tools or belongings, used the cave.
Fifty one stone artefacts were recovered in the upper
50 cm, of which 33 were made from the locally avail-
able travertine (flowstone). This casual use of the
flowstone in the cave is suggestive of an ‘‘expedient’’
technology, which would be in keeping with the inter-
pretation of brief visits by a few individuals for honey
collecting as mentioned by Yellen et al. (1987).
Noteworthy finds included an edge damaged chert
flake from the dated 20–30 cm level, along with a chert
burin and chert scaper from the 40–50 cm level. The
latter two finds were the only retouched tools found in
the entire upper 50 cm. The source of the chert is
unknown, but is most likely not local.
In addition to the stone artefacts, there were two
ostrich eggshell beads and 11 unburned ostrich egg
shell fragments. Other signs of human activity included
three mongongo nut shell fragments found in the
10–20 cm level. Yellen et al. (1987) mention the pres-
ence of mongongo groves in the area and that these
were exploited by the !Kung. Elsewhere (Robbins,
1990), we have shown that mongongo nuts, an impor-
tant staple food of the San, have been exploited since at
least the early Holocene in the western Kalahari.
The Terminal Pleistocene charcoal layer (50–80 cm):
intensive use of the cave
There was a substantial increase in cultural remains in
the charcoal layers, especially in chert artefacts and
fauna (Table 1). The increase in artefacts and bones,
mainly of large African bullfrogs, together with an
abundance of large, burned ostrich eggshell fragments
within the dense charcoal layer, represents a significant
change in the pattern of human use of Drotsky’s Cave.
Both fauna and other paleoenviromental data, dis-
cussed in detail later, reveal that it was substantially
wetter in the Drotsky’s Cave area during the late
terminal Pleistocene, with stream flow in the Gcwihaba
Valley possibly for lengthy periods during the year.
The setting certainly was very different in comparison
to today. We suggest that the nearby Gcwihaba River
and valley were being systematically exploited and that
the mouth of the cave was inhabited as a camp.
We recovered 146 pieces of debitage (56 were chert),
six cores (five were chert), 10 retouched pieces (eight
were chert) and eight with edge damage that most
likely resulted from use. In addition, there was a
fragment of what is most likely a cobble-sized grind-
stone. The debitage included 58 side struck flakes
(55·2% of the total from the charcoal layer) and 47 end
struck flakes (44·8%). Although the retouched tools
did not include any backed microliths, the presence
of well-made chert bladelets and four cores with nega-
tive bladelet scars clearly demonstrates the use of a
sophisticated microlithic technology in the terminal
Pleistocene (Figure 4). This interpretation contrasts
with Yellen’s speculation that Drotsky’s Cave was part
of the non-microlithic tradition of the interior of
southern Africa (Yellen et al., 1987).
Unfortunately, the lack of diagnostic tools prevents
any close comparison with any of the well-known
terminal Pleistocene assemblages from southern
Africa, but in a general way, it may represent the end
of the late Pleistocene microlithic period described
by Deacon (1984). These late Pleistocene microlithic
assemblages are recognized archaeologically by ‘‘the
systematic production of small bladelets from stan-
dardized single-platform bladelet cores and by the
occurrence in some quantity of bipolar cores that have
sometimes been so reduced by flaking that they are
classed as . . . scaled pieces’’ (Deacon, 1984: 227). In
such assemblages the frequency of formal tools is less
than 1% of the total, and in general, they are less
standardized than those found in Holocene assem-
blages. This characteristic of terminal Pleistocene
assemblages may explain why comparatively few for-
mal tools were found in the charcoal layer. Other late
and terminal Pleistocene samples from the western
Table 1. Stone artefacts from the 1991 test excavations at Drotsky’s Cave
cm Débitage Cores Edge damage Scraper Notch Burin Bifacial pt Miscellaneous Total
0–10 3 2 5
10–20 8 8
20–30 7 2 9
30–40 19 1 1 21
40–50 15 15
50–60 82 4 1 2 1 1 91
60–70 49 2 3 2 1 57
70–80 15 1 16
80–90 1 1
90–100 2 2
100–110 10 10
110–120 2 2
120–130 3 3
Note: The raw materials for all levels in the test pit include travertine (47%), chert (27%), chalcedony (1%), quartz (14%), and silcrete/other
(10%).
Paleoenvironment and Archaeology of Drotsky’s Cave 11

Kalahari include assemblages from Gi, an open-air
pan margin site near Dobe, and the Tsodilo Hills
Depression and White Paintings Shelters. Interestingly,
the artefact assemblage recovered from levels dated
to between 10,900&420 and 13,060&280  at the
Depression shelter, the same approximate period as the
Drotsky’s Cave charcoal layer, is similar to the lithic
material from the charcoal layer in that very few
retouched tools occur (Robbins, 1990). Further re-
search at Drotsky’s Cave may confirm whether the
same tradition of artefacts was widely distributed in
the western Kalahari during the terminal Pleistocene.
The charcoal layer also included two ostrich eggshell
beads and 197 ostrich eggshell fragments, of which 53
were burned. As mentioned above, similar large pieces
of ostrich eggshell were recovered by Yellen, but most
of them were not burned. It is interesting that large
pieces of unworked and unburned ostrich eggshell were
also common in terminal/late Pleistocene levels at the
Tsodilo Hills White Paintings Shelter. As far as we can
determine, the ostrich eggshells at Drotsky’s Cave were
not being used to manufacture beads. Other sites where
beads were being made, such as the more recent levels
at White Paintings Shelter, clearly show evidence of the
stages of manufacture in terms of overall shape
changes, edge grinding and the drilling of holes.
Although ostrich egg shells are also used as canteens,
the charring of the eggs most likely indicates that the
eggs were being cooked, but we are uncertain about the
method of cooking. Hitchcock (pers. comm.) describes
how Kalahari San pour the contents of the egg into a
pot or pan in order to cook it. Presumably, late
Pleistocene peoples in the Kalahari lacked containers
which could be placed directly over a fire to cook eggs.
However, some recent observations may be pertinent.
In 1992, one of us (Robbins) noticed San youths at
Tsodilo cooking chicken eggs directly on smoldering
coals by wrapping the eggs in wet dung. Whether
similar methods were used in the past for cooking
ostrich eggs is, of course, unknown.
Judging from the rest of the fauna described below,
it appears as if the terminal Pleistocene inhabitants of
the cave had a reasonably varied diet, including a range
of small to medium mammals, tortoises, birds (includ-
ing ostrich eggs), and bullfrogs. The comparative
abundance of ostrich eggs and bullfrogs in the charcoal
layer hints that the diet had some specialized aspects to
it as well.
The deposits below the charcoal layer (80–130 cm):
casual use of the cave
The story of the human use of the cave, as suggested by
our test pit, once again shifts to a pattern of casual
usage consistent with the ethnographic observations
reported by Yellen et al. (1987). The marked concen-
tration of artefacts, and animal remains that were
likely to have been used as food, is no longer evident,
and is replaced by bones of rodents and birds that are
most likely the result of owl predation and other
natural factors.
Artefacts were rarely found, but include 18 pieces of
probable débitage, of which 15 were in flowstone or
travertine. A small grindstone was recovered from the
110–120 cm level. There was no chert and formal
retouched tools were lacking. However, we did recover
20 ostrich eggshell fragments, one of which was
burned, and two ostrich eggshell beads (Figure 3).
The presence of the beads, below the terminal
Pleistocene dated levels, confirms that one of the
standard items of decoration used by people in the
Kalahari has ancient roots that extend at least to
the late Pleistocene. While the overall collection of
beads recovered from all levels of our excavation at
Figure 4. Drotsky’s Cave artefacts. Upper row-chert bladelet core and chert/silcrete bladelets from the charcoal layer; lower row-chert hollow
scraper and chert utilized flake from the charcoal layer, ostrich egg shell beads from 10–20, 90–100 and 100–110 cm. Photo by Susan Eyde.
12 L. H. Robbins et al.

Drotsky’s Cave is limited to six specimens, it is inter-
esting to observe that the size of the beads increases at
the base of the deposits. Bead diameter is as follows:
0–10 cm level is 4·6 mm, 10–20 cm is 4·3 mm, 50–60 cm
is 4·6 mm, 60–70 cm is 4·2 mm, 90–100 cm is 5·3 mm,
100–110 cm is 5·5 mm. This increase in size in the
oldest deposits is the opposite to size trends observed
at more recent sites in Namibia where there are larger
samples (Jacobson, 1987).
Fauna
Faunal remains were well-preserved and, as already
noted, included mammals, birds, amphibians and rep-
tiles. We present the basic analysis below in the form of
individual specialist reports. Some general comments
are offered as an interpretive ‘‘preview’’ to the reports.
There were no fish and this lack could be seen as
paleoecological evidence that the Gcwihaba River was
not connected to fish-bearing waters such as the
Okavango Delta, which was larger during the wetter
periods of the late Pleistocene. On the other hand, it
could be that the occupants of the cave did not catch or
eat fish, even though they quite clearly utilized the
river. We think that this interpretation is unlikely given
the prevalence of catfish remains found at the Tsodilo
White Paintings Shelter throughout the Late Stone
Age (Robbins et al., 1994).
At present, leopards and hyaenas occasionally visit
the cave. In the past such predators could have been
responsible for the introduction of some of the bovid
remains as well as the single equid bone that was found
below the charcoal/main occupation layer. The hyaena
bones found above and below the charcoal layer could
represent natural deaths of these animals.
As will be shown, the majority of the birds and
rodents, with the exception of springhares, were recov-
ered from below the charcoal layer in the levels that
had little cultural material. Most of them probably
represent natural deaths. In great contrast, 88% of the
springhare elements, an important food animal for
the ethnographically observed San, along with the
majority of the macromammals and bullfrogs, were
recovered from the charcoal layer. Most of these
animals were likely food refuse or, in a few cases,
notably the fox and caracal or serval, could have been
exploited for furs.
Macromammals
The macromammals are listed in Table 2. As far as we
know, all the species occurred in the area historically.
Small sample size may explain the absence of lechwe
(Kobus leche), reedbuck (Redunca arundinum), or other
wetland mammals that are represented in late/terminal
Pleistocene layers in the Tsodilo Hills. Sampling
error may also account for the absence of blesbok
(Damaliscus dorcas) which Yellen et al. (1987) reported
from the previous excavation at Drotsky’s Cave. If this
identification was correct, it could imply a time when
the environs of Drotsky’s Cave were grassier, since
blesbok is a grassland species that was restricted to
South Africa historically. A possible expansion of
highveld-type grassland northwards has been inferred
from the presence of blesbok in the late Pleistocene
layers of RedcliffCave, Zimbabwe (Cruz-Uribe, 1983).
Small sample size almost certainly does not explain the
relative abundance of springhare, which are also com-
mon at the White Paintings Site in the Tsodilo Hills
and in later Stone Age rock shelters in the northern
Cape Province of South Africa, on the southern mar-
gin of the Kalahari (Klein, 1979). The archaeological
data imply that springhare were as important to pre-
historic Kalahari hunter–gatherers as they were to the
!Kung and other historic San groups (Lee, 1979).
Sampling bias is also unlikely to explain the relative
abundance of small and small–medium antelope,
which similarly dominate Later Stone Age samples at
Tsodilo, in Namibia, and on the southern Kalahari
margin (Cruz-Uribe & Klein, 1983; Klein, 1979). The
relative abundance of smaller antelope at so many sites
probably reflects their live abundance nearby and the
relative ease with which they could be transported
from kill sites to camp sites.
Micromammals: Rodentia
The rodent remains from Drotsky’s Cave include a
minimum of eight genera (Table 3). With the exception
of the springhares, discussed above under macro-
mammals because of their large size (up to 4 kg),
rodent specimens were most common in the lower part
of the section (100–130 cm). Most of the rodent re-
mains (springhares are a likely exception) were prob-
ably deposited as food refuse by non-human predators,
in particular by owls as regurgitated pellets. The rodent
bones are well preserved and evidence of digestion and
transport is rare. Most specimens are nearly intact:
long bones are usually missing only the very proximal
and/or distal ends, teeth are often present in the alveoli,
and many mandibles are essentially complete except
for the ascending rami. In addition, all the identifiable
taxa are almost exclusively nocturnal and all (for
whom the predators are known) are known prey for
owls, in particular the barn and grass owls. As
discussed below, bones of what is most likely barn
owl (0–10 cm and 70–80 cm) and also other raptors
(70–80, 100–110, 120–130 cm) have been recovered
from Drotsky’s Cave.
As was previously noted, springhares found in the
greatest abundance in the terminal Pleistocene char-
coal layer from 50 to 80 cm are reasoned to have been
an important food resource. The fat mouse is also
considered a delicacy by indigenous African people (de
Graaff, 1981), but this genus may only be present at
110–120 cm, and it is not found in the units (50–80 cm)
where occupation was most evident.
Paleoenvironment and Archaeology of Drotsky’s Cave 13

Table 2. Macromammals from the 1991 test excavations at Drotsky’s Cave
Taxon 0–10
cm 10–20
cm 30–40
cm 40–50
cm 50–60
cm 60–70
cm 70–80
cm 80–90
cm 90–100
cm 100–110
cm 110–120
cm 120–130
cm All
Pedetes capensis (Springhare) — — 1/1 1/1 21/2 4/1 4/1 — — 1/1 1/1 — 33/2
Otocyon megalotis or vulpes chama (Fox) — — 1/1 — — 1/1 — — — — — — 2/1
Viverridae gen. et sp. indet. (Indeterminate mongoose) — — — — 1/1 — — — — — — — 1/1
Hyaena brunnea or Crocuta crocuta (Hyaena) 1/1 ——————— — — — 1/12/1
Felis serval or caracal (caracal or serval) — — — — 4/1 — — — — — — — 4/1
Equus burchelli (Plains zebra) — ——————— — — 1/1 — 1/1
Diceros bicornis or Ceratotherium simum (Rhinoceros) — ————1/1—— — — — — 1/1
Tragelaphus strepsiceros (Greater kudu) — 2/1 —————— — — — — 2/1
Sylvicapra grimmia (Gray duiker) — 1/1 — 1/1 1/1 — — —. — — — 1/1 4/1
Raphicerus campestris (Steenbok) 1/1 1/1
Bovids—general
Small (e.g. steenbok) — 1/1 — — 12/1 4/1 1/1 — — — — — 18/2
Small medium (e.g. gray duiker, springbok) — 1/1 — 1/1 1/1 1/1 — — — — — 1/1 5/1
Large medium (e.g. greater kudu, hartebeest) — 2/1 — 1/1 1/1 2/1 1/1 1/1 — — — 1/1 9/1
Large bovid (e.g. eland, buffalo) — ——————— — — — — —
Note: The larger mammal species found in successive 10 cm spits in the 1991 excavations at Drotsky’s Cave. The number before the slash in each case is the Number of Identified Specimens
(NISP). The number afterwards is the Minimum Number of Individuals (MNI) from which the specimens must have come. Most of the bones come from a charcoal layer 50–80 cm below
the surface, which accumulated between roughly 12,500 and 11,000 radiocarbon years ago.
Table 3. Rodents (micromammals) from the 1991 test excavations at Drotsky’s Cave
Taxon 0–10
cm 10–20
cm 20–30
cm 30–40
cm 40–50
cm 50–60
cm 60–70
cm 70–80
cm 80–90
cm 90–100
cm 100–110
cm 110–120
cm 120–130
cm All
Mystromy albicaudatus (white-tailed rat) — ———————— — — 2/1 3/2 5/3
Gerbillurus paeba (Hairy-footed gerbil) — — 1/1 — — 1/1 — — — — — 3/1 1/1 6/4
Tatera cf. T. leucogaster (Bushveld gerbil) — ————1/1——— — — — — 1/1
Tatera cf. T. brantsii (Highveld gerbil) 1/1 —————1/15/7— —34/13 11/5 20/7 72/34
Tatera sp. 1/1—2/1————2/2— — 1/1 3/3 1/110/9
Dendromus sp. (Climbing mouse) — ———————— — — — 1/1 1/1
Steatomys sp. (Fat mouse) or Malacothrix sp.
(Large-eared mouse) — ———————— — — 2/2 — 2/2
Otomys cf. O. angoniensis (Angoni vlei rat) — ——————3/3— — 5/2 12/4 3/2 23/11
Muridae — — 2/1 —————— — — — — 2/1
Unidentified muroids — ——————1/1— — 1/1 2/2 4/3 8/7
Note: All of the identifiable fauna that was recovered is reported with the exception of the remains of bats and insectivores. The bat remains are almost certainly the result of natural deaths.
Bats are abundant at present in the cave.
14 L. H. Robbins et al.

The rodents found at Drotsky’s Cave represent two
main ecological niches. The majority (hairy-foooted,
Highveld, and Bushveld gerbils, and springhare, to-
gether comprising 80% of the total number of speci-
mens identifiable to genus) can be characterized as
inhabitants of dry, open areas with light vegetation
cover and sandy soils. The fat mouse may also be
found in this environment, but the large-eared mouse
prefers harder ground. The other important ecological
niche represented is a more heavily vegetated area
along a major water source such as a river. This niche
is suggested by the Angoni vlei rat and the climbing
mouse. The recovery of the latter taxa supports other
evidence (e.g. from calcrete samples) indicating the
former presence of a river in the Gcwihaba Valley
below the cave. If owls deposited the remains of the
vlei rats in the cave, we can estimate from the range of
the modern barn owl that the wetland habitat of the
rats was probably located within a few kilometres of
the cave itself. However, vlei rats are also eaten by
people (Smithers, 1971) and it is possible that they were
brought in from a more distant area.
Most of the fossil rodent genera may have living
counterparts in the general area of Drotsky’s Cave.
However, in Botswana, the Angoni vlei rat is found to
the north of the study area along the Okavango River,
and the large-eared mouse is found to the south of
about 22)S longitude. The vlei rat has also been found
at the Tsodilo Hills White Paintings Shelter about
50 km to the west of the Okavango (Robbins et al.,
1994). It is probably significant that the vlei rat is not
found above the terminal Pleistocene deposits in more
recently deposited sediments. As discussed below, sedi-
mentological and other evidence suggests that the
late/terminal Pleistocene was much wetter than today.
The most notable extralimital taxon is the white-tailed
rat (Mystromys albicaudatus) recovered between 110
and 130 cm, which is currently not found in Botswana.
This relatively rare rodent is confined mainly to high-
veld and montane grasslands in the Transvaal, north-
western Natal, the Orange Free State, Lesotho, and the
eastern region and western Cape of South Africa. It is
found in the Southern Savanna Grassland and the
South West Cape biotic zones (de Graaff, 1981).
Nowak and Paradiso (1983) describe it as inhabiting
grassy flats and dry sandy areas.
Birds
The preliminary identifications of the birds from
Drotsky’s Cave are presented in Table 4. At the present
time, in the absence of other sources of water, birds are
attracted to the entrance of the cave to drink small
droplets of water emerging from the stalactites. The
cave is also used by owls, as evidenced by castings
(pellets). Two barn owls were observed in the cave in
1993. They were seen in both entrances.
Most of the bird remains were found below the main
occupation layer and are believed to represent natural
deaths rather than human food refuse. Although con-
ditions were wetter in the late Pleistocene, with fre-
quent and possibly at times perennial river flow in the
Gcwihaba Valley, no aquatic or marsh birds were
recovered. In this regard, it should be noted that
aquatic birds, such as ducks, do frequent the western
Kalahari pans at present when they contain water
during the rainy season. The owl remains are most
likely referable to the barn owl (Tyto alba). Further
comparative study of the buttonquail (Turnix), bee-
eater (Merops) and probably the doves (Columbidae)
could result in identifications to the species level.
Bullfrogs
The most striking aspect of the faunal sample from the
point of view of human subsistence and paleoenviron-
mental conditions was the comparative abundance of
bullfrog (Pyxicephalus adspersus) bones recovered
from the deposits. A total of 129 bullfrog elements
were recovered representing a minimum of 40 individ-
uals. While they were distributed in every 10 cm level
between 40 and 130 cm, 26 of the individuals (65%)
were recovered from the terminal Pleistocene charcoal
layer between 50 and 80 cm. These frogs most likely
were obtained from the terminal Pleistocene river in
the Gcwihaba Valley or from nearby pans. Pyxicepha-
lus adspersus is one of the largest amphibians in
central/southern Africa. The elements represent large
adults, estimated to range in size from approximately
13 to 20 cm in length.
The skeletal element representation was strongly
biased towards cranial elements, with under-
representation of appendicular elements as compared
to an average frog skeleton (Table 5). Part of this bias
is undoubtedly due to better preservation of the bull-
frog cranial elements, which are ridged and very
robust. However, because the limbs are among the
most edible parts of the frog, these bones may have
been infrequently preserved because they were ingested
along with the meat. Two of the elements (1·8% of the
total) showed evidence of burning, suggesting that at
least some of the frogs were roasted before eating.
Pyxicephalus adults are large, fat bullfrogs, which
inhabit arid/semiarid regions throughout much of the
African continent. They are a food source for several
modern human populations, including groups of
Kalahari San (Stewart, 1967; Lee, 1979). The frogs are
fossorial, hibernating underground during the dry
season, often for 10 months. They could be captured at
this time, being sluggish, by digging them out of their
burrows. In this regard, an historical reference is
pertinent. Livingstone (1859: 49) learned from the
‘‘Bushmen’’ that the ‘‘matlametlo (bullfrog) makes a
hole at the root of certain bushes and there ensconces
himself during the months of drought. As he seldom
emerges, a large variety of spider takes advantage of
the hole, and makes its web across the orifice. He is
thus furnished with a window and screen gratis.’’
Paleoenvironment and Archaeology of Drotsky’s Cave 15

Table 4. Birds from the 1991 test excavations at Drotsky’s Cave
0–10
cm 10–20
cm 20–30
cm 30–40
cm 40–50
cm 50–60
cm 60–70
cm 70–80
cm 80–90
cm 90–100
cm 100–110
cm 110–120
cm 120–130
cm
All
NISP/
MNI
Passeriformes (songbird): 4(+ )spp. 1/1 3/2 5/2 11/4 13/4 33/8
Turnicidae: Turnix sp. (buttonquail) 1/1 1/1 1/1 1/1 3/1 7/2
Falconidae: Falco sp. (falcon) 2(+ )spp. 1/1 3/3 4/3
Columbidae (pigeon/dove) 1/1 1/1 2/1
Tytonidae: Tyto cf. T. alba (barn owl) 1/1 1/1 2/1
Accipitridae: Neophron peronopterus (Egyptian vulture) 1/1 1/1
Phasianidae: Coturnix sp. (quail) 1/1 1/1
Apodidae (swift) 1/1 1/1
Meropidae: Merops sp. (bee-eater) 1/1 1/1
‘‘Raptores’’ (raptorial bird claw) 1/1 1/1
Aves indeterminate 1/1 1/1 2/1
Total NISP/MNI 2/2 1/1 2/2 8/5 7/4 13/8 24/13 55/21
Note: The number before the slash is the Number of Identified Specimens (NISP) and the number after the slash is the Minimum Number of Individuals (MNI).
16 L. H. Robbins et al.

Livingstone then noted that the ‘‘Bushmen’’ would
look for the webs to locate the frogs. Also possible is
their capture during the rainy season when they con-
gregate in groups around waterholes. Because this is
their breeding period, they are highly aggressive and
can inflict serious bites. Silberbauer (1981) reports that
frogs are considered to be a delicacy among the G/wi of
the central Kalahari where they are caught between
November and March. The highest totals are in the wet
season months of January, February, and March.
The presence of bullfrog bones consistently over
several thousand years of occupation at Drotsky’s
Cave indicates the importance of these animals as a
food source. It also indicates long-term familiarity with
their seasonal behaviour.
Interestingly, in addition to the bullfrogs, one right
ilium of another frog, Xenopus sp. was recovered
from between 120–130 cm. This specimen was either
Xenopus laevus petersii or Xenopus muelleri. This frog is
significant because it is highly aquatic. While it can
survive in desert regions, a constant water source is
required.
Reptiles
The distribution of reptile elements is presented in
Table 6. The majority of remains were tortoise, with
elements clustering around 50 to 60 cm and between 60
and 70 cm. The clear association with the charcoal
layer implies use as food by humans. Two species
of tortoise were identified: (1) Gechelone pardalis
babcocki, represented by five costal bones from
between 50 and 60 cm and one peripheral from 110 to
120 cm, and (2) Kinixys belliana, including two left
scapulae from the 50–60 cm level and one peripheral
from the 110–120 cm level. In addition to the above,
the left scapula of a side neck turtle, Pelusios sp, was
recovered from the 50–60 cm level. The pelomedusid
turtles are not very tolerant of saline or alkaline waters.
Although the species could not be identified, the most
likely possibilities based on current distributions are
Pelusios subniger or Pelusios bechunanicus. The latter
form, known as the Okavango Hinged Terrapin, is the
geographically nearest one. Interestingly, it is described
as ‘‘a deep-water terrapin restricted to the clear waters
of the greater Okavango system on the Kalahari Sands
(Auerbach, 1987: 74).’’
Lizards were represented by one vertebra of Varanus
sp. recovered from between 50 and 60 cm in the
charcoal layer. This monitor lizard was either Varanus
exanthematicus or Varanus niloticus. Monitor lizards
are a common food resource in Botswana and are
found at the White Paintings shelter, as are the tor-
toises mentioned above (Stewart, Stevens & Robbins,
1991).
The final reptile specimen recovered was the vertebra
of a snake belonging to the Family Colubridae, Genus
indeterminate. It was found in the 120–130 cm level.
While snakes are eaten by Kalahari peoples, it is
uncertain if this one was eaten, or whether it was the
result of predation or a natural death. The snake,
lizard, and tortoise elements are all from families that
would be expected for this region (Auerbach, 1987).
Cave Sediments and Valley Calcretes: the
Paleoenvironmental Record
The faunal record and archaeological information
described above have provided one body of evidence
bearing on late/terminal Pleistocene environmental
conditions. The long record of wet and dry periods
preserved in the cave clastic sediments and in the
Gcwihaba Valley calcrete sequence provides another
unique body of evidence that can be used to elucidate
past climates/environments in the Kalahari.
Cave clastic sediments
A column of sediment was removed in increment of
7·5 cm from the east wall of the excavation. Sampling
extended to 135 cm providing 18 samples of about
160 g in weight. Sediment colour was determined on
dry, untreated samples using a Munsell colour chart.
Each sample was split to obtain a representative 25 g
subsample. Macro-organic matter and human artefacts
were first removed and then the sample was sieved to
obtain particles coarser than "1ö. These particles
were manually separated into weathered dolomitic
marble and soft nodules of secondary calcite. Both
fractions were weighed and then the secondary calcite
was returned to the sample. The samples were then
treated with 0·5 N hydrochloric acid to remove
Table 5. Bullfrog elements from the 1991 test excavations at Drotsky’s
Cave (identified by K. Stewart)
Drotsky’s Cave Average frog skeleton
%N%
Cranial 49 53·3 17.8
Axial 19 20·7 17·0
Appendicular 24 26·0 65·2
Table 6. Reptile elements from the 1991 test excavations at Drotsky’s
Cave
Level (cm) Tortoise Lizard Snake
0–40 0 0 0
40–50 1 0 0
50–60 47 1 0
60–70 19 0 0
70–80 4 0 0
80–110 0 0 0
110–120 3 0 0
120–130 1 0 1
Paleoenvironment and Archaeology of Drotsky’s Cave 17

carbonates and further organic matter was removed by
oxidation with hydrogen peroxide. What remained of
each sample was then dispersed in a solution of sodium
hexametaphosphate and wet sieved through a 4 ösieve
to remove silt and clay. Particles coarser than 4 ö,
including the dolomitic marble gravel removed earlier,
were dry-sieved at whole öintervals in the range from
"2to1öand at half öintervals from 1 to 4 ö. The
<4 ösilt and clay fraction was then subjected to pipette
analysis. Withdrawals were taken to divide the sedi-
ment at the 5, 6 and 9 ösize intervals. Mean grain size,
sorting, skewness and kurtosis were determined by the
graphical method of Folk and Ward (1957).
Based on the radiocarbon dates of 5,470&90 and
11,240&60 years , the rate of sediment accumula-
tion in the depth intervals between 0 and 25 cm
(4·57 cm/1000 years) and from 25 to 50 cm (4·33 cm/
1000 years) was very similar, averaging 4·45 cm/1000
year. The radiocarbon age of 12,450&80 years  for
charcoal from 75 cm depth suggests a sediment accu-
mulation rate of 20·66 cm/1000 years in the depth
interval between 75 and 50 cm or 4·6 times the accu-
mulation rate in the upper 50 cm of the sediment
profile. Although such an accumulation rate is
possible, we favour the view that sedimentation rates
remained fairly constant throughout the period of
sediment deposition and that the charcoal dated at
12,450&80 yr  is a cultural feature related to inten-
sive human occupation of the cave, as is much of the
charcoal in the charcoal rich horizon between 50 and
80 cm depth. The chronology attached to the sediment
sequence and archaeological material is therefore
based on an assumed sediment accumulation rate of
4·45 cm/1000 years. We recognize that because organic
matter of low density makes up between 4·5 and 15·7%
(average 8·5%) by weight of the sediments in the 50 to
80 cm depth range, estimated age for sediments below
50 cm may be slightly too old.
Sediment characteristics are shown in Figures 5 and
6. Mean grain size varies from 2·46 to 2·66 ö, which is
85
135
Depth (cm)
5
15
25
35
45
55
65
75
95
105
115
125
2.45–2.7
5,470±90
11,240±60
12,450±80
Mean
0.85–1.5
0.15–0.4
1.5–2.25
0.0–2.0 3.5–16.5
80604020
5/4
5/5
5/4
4/4
4/3
7.5 yr
10 yr 4/3.5
4/3
4/4
4.5/4
4/4
7.5 yr
Cumulative percent
Grain size characteristics (phi units)
Sorting
Skewness
Kurtosis
Calcrete nodules %
Munsell color
Organic matter %
Coarse to very coarse sand plus gravel (<1 φ)
Medium sand (1–2 φ)
Fine sand (2–2.5 φ)
Fine sand (2.5–3 φ)
Very fine sand (3–4 φ)
Medium and coarse silt (4–6 φ)
Fine to very fine silt plus clay (>6 φ)
Layers containing abundant charcoal
Figure 5. Analysis of sediments from the 1991 test pit.
18 L. H. Robbins et al.

in the range of fine sand. Sorting varied from 0·87 to
1·42 indicating moderately to poorly sorted sediments.
The sediments all exhibited a fine tail and so were
positively to very positively skewed (skewness between
0·16 and 0·36) and were very to extremely leptokurtic
(kurtosis from 1·51 to 2·55) reflecting relatively peaked
particle size distributions. Organic matter varied from
3·21 to 15·69% by weight and, not surprisingly, was
highest between 50 and 80 cm where the sediments
were rich in charcoal and faunal remains. Calcrete
nodules coarser than "1ömade up 0·01 to 1·7% of
untreated sample weight.
As Cooke (1975) has pointed out, clastic sediments
in Drotsky’s Cave are produced by processes operating
within the cave but are also derived from outside the
cave, being carried in by wind and water. Cooke found
that windblown sand had in the past entered Drotsky’s
Cave in large quantities being particularly thick
close to the two entrances and being finer and better
sorted than sediments derived from within the cave
itself.
Logically, finer-grained, better-sorted sediments at
the archaeological site should equate with drier con-
ditions when more material was transported into the
cave by wind. Under slightly wetter, possibly semiarid,
conditions we might expect an increase in coarse and
very coarse sand and gravel derived from within and
outside the cave and having a sizable dolomitic marble
component. Some of these fragments would be derived
from the interior walls and ceiling of the cave, others
would be washed into the cave from outside by runoff
after heavy rain. We might also expect a reduction in
the input of aeolian sand. Under conditions of even
greater wetness, the input of aeolian sand should be
reduced still further due to a denser vegetation cover
on the nearby dunes. Also, there should be a greater
accumulation of coarse material produced by increased
breakdown within the cave and by increased runoff
into the cave carrying coarser particles not transported
easily by wind. The particle size most easily entrained
by running water is 0·2 mm (fine sand, diameter
between 2 and 3 ö) while that most easily entrained by
wind is 0·08 mm (very fine sand, diameter from 3 to
4ö) (Bagnold, 1941; Sundborg, 1956). Furthermore,
with increased moisture in the cave there is every
likelihood that fragments of dolomitic marble dis-
lodged from the walls and ceiling of the cave will be
reduced in size or totally removed by dissolution.
Under this scenario, there should be an increase in
sediment grain size under moist conditions because of
an increase in very coarse and medium sand rather
than because of an increase in very coarse sand and
gravel.
Figures 5 and 6 show that there was a major
decrease in the mean grain size of sediments accumu-
lating at the site shortly after about 11,000 years  and
that this was accompanied by an increase in sorting
and kurtosis. In addition, material coarser than 0 ö
e
135
Depth (cm)
5
Estimated age of Drotsky's Cave
sediments (yr. B.P.)
500
15
25
35
45
55
65
75
85
95
105
115
125
123456789
g
f
e
d
c
b
a
12345
g
f
d
c
a
67891011
5,470±90
11,240±60
12,450±80
1.5φ
0φ1φ
369
Angoni vlei rat
Climbing mouse
Xenopus sp. (frog)
Bullfrog
2500
4500
6500
8500
10 500
12 500
14 500
16 500
18 500
20 500
24 500
26 500
28 500
30 500
22 500
b
Figure 6. Late Quaternary Paleoclimatic records from the region of Drotsky’s Cave and neighbouring region of southern Africa. /, wet
periods indicated by high lake levels, periods of speleothem and calcrete deposition, or cave clastic sediment texture; /, low lake levels;
-, radiocarbon ages; a–g, possible matching peaks in columns 1 and 4.
1: Cumulative percent by weight of grains <0ö,<1öand <1·5ö, 2: Drotksy’s Cave clastic sediments (this paper), 3: Drotksy’s Cave faunal
remains indicating past wetter conditions (this paper), 4: Number of
14
C dates indicating past wetter conditions in the summer rainfall zone of
southern Africa (Deacon & Lancaster, 1988), 5: Makgadikgadi Basin lake levels (Cooke & Verstappen, 1984; Shaw & Cooke, 1986; Thomas
& Shaw, 1991), 6: Ngami, Mababe and Caprivi Basin lake levels (Shaw, 1986; Shaw & Cooke, 1986; Shaw & Thomas, 1988; Thomas & Shaw,
1991), 7: Drotsky’s Cave speleothem ages (Cooke & Verhagen, 1977; Shaw & Cooke, 1986; Brook et al. 1990), 8: Gcwihaba Valley calcrete ages
(Cooke & Verhagen, 1977; this paper), 9: Dry valleys of the middle and southern Kalahari (Shaw et al., 1992), 10: Tsodilo Hills lake (Brook
et al., 1992), 11: Dobe Valley lake at Gi (Helgren & Brooks, 1983).
Paleoenvironment and Archaeology of Drotsky’s Cave 19

(very coarse sand and gravel) also increased substan-
tially, much of this consisting of large fragments of
dolomitic marble. Figure 3 also shows that accom-
panying the increase in coarse material after 11,000
years  was an increase in material finer than 2 ö, that
is fine to very fine sand, silt and clay. Together, these
changes suggest that, overall, the period from 11,000
years  to the present was significantly drier than the
preceding 20,000 years. However, variations in grain
size also indicate that the periods c. 11,500–9000 and
6500–3000 years  were somewhat drier than the
intervals 9000–6500 and 3000–1250 years  when
increased wetness is suggested.
The sediment data suggest that the interval 18,000–
11,500 years  was the wettest of the past 30,000
years. This is indicated by the lowest percentages of
very fine sand, silt and clay and maximum cumulative
percentages of medium sand, coarse sand and
gravel. Prior to 18,000 years  the wettest con-
ditions occurred at 30,500–28,500, 25,000–23,500 and
21,500–20,000 years  with drier conditions centred at
19,500, 22,500 and 27,000 years . However, the
generally higher percentages of particles coarser than
2·5 öin sediments deposited prior to about 11,500
years  indicates that this entire period was wetter
than the Holocene. Finally, the percentage of second-
ary calcrete nodules in the sediments increases sharply
below about 75 cm depth with peaks of 1·15% at about
79 cm and 1·69% at 116 cm depth. We believe that
these nodules are of pedogenic origin and that they
accumulated between 60 and 75 cm below the level of
the cave floor during dry Holocene intervals centred at
9500 and 4500 years .
Valley calcretes
Cooke (1975) reports four calcretes of different age
along the Gcwihaba Valley near Drotsky’s Cave.
Radiocarbon ages indicate that the two oldest deposits
are beyond the range of the method (i.e. they are
>45,000 years old), while the youngest two deposits
were laid down from 11,000–10,000 years  (Cooke &
Verhagen, 1977; Cooke, 1984). Cooke (1984) interprets
the various calcretized sands and gravels in the
Gcwihaba Valley as evidence of ephemeral river flow in
a sub-humid to semi-arid climate. In 1991 we collected
a sample of calcrete from the floor of the Gcwihaba
Valley approximately 2 km up valley of Drotsky’s
Cave. The calcrete was radiocarbon dated to
11,430&60  (Uga 6591) assuming that it contained
15% old, dead carbon at the time of formation (Cooke
& Verhagen, 1977).
The calcrete contained 8·1 million diatom valves per
gram, thus providing important information about
terminal Pleistocene conditions in the Gcwihaba
Valley. Diatom preservation was good and there were
few other microfossils in the sample. Identification and
interpretation of diatoms were based on Gasse (1986)
and Patrick and Reimer (1966, 1975). Six species made
up 78% of the 456 diatoms counted (Cymbella muelleri
25%; Rhopalodia gibberula 24%; Gomphonema gracile
9%; Cyclotella meneghiniana 8%; Navicula species 6%;
Nitzschia cf fonticola 6%). Cymbella muelleri is typi-
cally benthic, but has been found in plankton in
significant numbers. Gasse (1986) considers it broadly
tolerant of alkalinities, pH, and mineral content.
Rhopalodia gibberula is also tolerant of water con-
ditions but Gasse’s (1986) survey of African lake
sediments led her to suggest that it prefers strongly
alkaline waters, growing best in hyperalkaline lakes of
high conductivity, alkalinity and pH. Gomphonema
gracile (variety intricatiformis) is pH-tolerant although
Gasse (1986) found it to be most common in water of
pH 7·0–8·7, pehaps of low nutrient content. Cyclotella
meneghiniana is littoral or planktonic, and seems to
prefer more saline waters (Gasse, 1986).
Cemented into the outside of the calcrete was a shell
of Trachycystis sp., which is a terrestrial snail not
necessarily associated with water (Chris C. Appleton,
University of Natal-Pietermaritzburg (pers. comm.
1991). This confirms that the carbonate cement of the
calcrete formed under conditions of seasonal or even
ephemeral stream flow. Thus, the diatom assemblage,
which is older than the carbonate cement of the
calcrete, indicates that prior to c. 12,000 years  there
was substantial stream flow in the Gcwihaba Valley.
Stream flow may have lasted for a considerable portion
of the year or even have been perennial. The age of the
carbonate cement of the calcrete, and the presence of
the terrestrial snail shell, indicates that by c. 11,500
years  stream flow had become seasonal or even
ephemeral.
Summary and Conclusions
The sediment and faunal data from the Drotsky’s Cave
excavation are compared in Figure 6 with selected
paleoenvironmental records for the Kalahari and with
the frequency of radiocarbon ages indicating wetter
conditions in the summer rainfall zone of southern
Africa. The records are in broad agreement and the
correlation between the sediment and radiocarbon age
data is particularly striking. The evidence indicates that
the late/terminal Pleistocene from about 30,000 to
11,000 years  was substantially wetter and probably
cooler than the ensuing Holocene and that the interval
between 17,500 and 11,000 years  may have been
particularly moist.
Cooler conditions are indicated by the possible
presence of blesbok in the charcoal-rich layer (Yellen
et al., 1987) and by the presence of white-tailed rat at
130–110 cm depth. Neither blesbok nor white-tailed rat
is found in the area today, the latter being confined to
highveld and montane grasslands. Wetter conditions
along the Gcwihaba Valley in the late/terminal
Pleistocene are clearly suggested by the presence of
Angoni vlei rat, climbing mouse, the Xenopus frog, and
20 L. H. Robbins et al.

the side neck turtle. The diatom assemblage in the
calcrete from the Gcwihaba Valley confirms that there
was more permanent river flow in late/terminal
Pleistocene times with calcretization of fluvial sedi-
ments probably taking place as river flow became
markedly seasonal or even ephemeral c. 11,500
years .
It is perhaps significant that Drotsky’s Cave was
used more intensively from c. 12,500 to 11,000 years 
at the time of a major transition from cooler and wetter
to warmer and drier climatic conditions. At this time of
markedly seasonal or ephemeral stream flow in the
valley the cave may have provided a dry shelter for
hunter–forager groups utilizing the resources of the
Gcwihaba Valley. In addition, it is likely that drip-
waters in the cave may have provided a source of water
during dry periods. From c. 30,000 to 12,500 years 
the cave may have offered no special advantages given
a relatively plentiful supply of water in the river
valleys. Furthermore, being wetter and colder than
now it may not have been an ideal shelter. During
the Holocene, with little or no stream flow in the
Gcwihaba Valley, it is not surprising that the cave was
used sparingly, although dripwaters may have pro-
vided water in the dry season. The fact that charcoal
deposits in the Drotsky’s Cave excavation (between 80
and 50 cm, and c. 25 cm) occur at times of transition
from wetter to drier conditions suggests that the cave
was most attractive to hunter–forager groups at these
marginal times when the cave was probably an ideal
shelter, provided water if needed, and was close to the
wet-season resources of the Gcwihaba Valley.
Acknowledgements
We thank the National Science Foundation for fund-
ing and the National Museum of Botswana for facili-
tating this research. We are especially grateful to
A.C. Campbell, N. Walker, T. Mpulubusi, J. Clark,
B. Smith, J.A. Holman, J. Yellen, G. Schneider,
R. Hitchcock, K. Egan and H.J. Cooke. The analyses
in the paper reflect a multidisciplinary effort and are
divided among researchers as follows: Robbins and
Murphy, archaeology and general comments; Brook
and Ivester, sediment analysis; Brook and Haberyan,
diatoms, Klein and Milo, macromammals, Stewart,
bullfrogs, Winkler, rodents, Matthiesen, identification
of the birds; Stevens, reptiles.
References
Auerbach, R. D. (1987). The Amphibians and Reptiles of Botswana,
Gaborone, Botswana: Mokwepa Consultants.
Bagnold, R. A. (1941). The Physics of Blown Sand and Desert Dunes,
London; Methuen & Co.
Brook, G. A., Burney, D. A. & Cowart, J. B. (1990). Desert
paleoenvironmental data from cave speleothems with examples
from the Chihuahuan, Somali-Chalbi, and Kalahari Deserts.
Palaeogeography, Palaeoclimatology, and Palaeocology, 76,
311–329.
Brook, G. A., Haberyan, K. A. & De Filippis, S. (1992). Evidence of
a shallow lake at Tsodilo Hills, Botswana, 17,000 to 15,000 years
: further confirmation of a widespread late Pleistocene
humid period in the Kalahari Desert. Palaeoecology of Africa, 23,
165–175.
Campbell, A. C. & Robbins, L. H. (1993). A decorated bone artefact
from the Tsodilo Hills, Botswana: implications for rock art.
Botswana Notes and Records, 25, 19–28.
Cooke, H. J. (1975). The palaeoclimatic significance of caves
and adjacent landforms in the Kalahari of western Ngamiland,
Botswana. Geographical Journal, 143, 431–444.
Cooke, H. J. (1984). The evidence from northern Botswana of
climatic change. In (Vogel, J. C. Ed.). Late Cainozoic Palaeo-
climates of the Southern Hemisphere. Rotterdam: Balkema,
pp. 265–278.
Cooke, H. J. & Baillieul, T. (1974). The caves of Ngamiland: an
interim report on explorations and fieldwork 1972–74. Botswana
Notes and Records, 6, 147–156.
Cooke, H. J. & Verhagen, B. T. (1977). The dating of cave
development—an example from Botswana. Proceedings of the 7th
International Congress of Speleology, Sheffield, pp. 122–124.
Cooke, H. J. & Verstappen, H. T. (1984). The landforms of the
western Makgadikgadi Basin in northern Botswana, with a
consideration of the chronology of the evolution of Lake Paleo-
Makgadikgadi. Zeitschrift für Geomorphologie, N.F. 28, 1–19.
Cruz-Uribe, K. (1983). The mammalian fauna from Redcliffe Cave,
Zimbabwe. South African Archaeological Bulletin, 38, 7–16.
Cruz-Uribe, K. & Klein, R. G. (1983). Faunal remains from some
Middle and Later Stone Age archaeological sites in South West
Africa. Journal of the South West African Scientific Society, 36/37,
91–114.
Deacon, J. D. (1984). Later Stone Age people and their descendants
in southern Africa. In (R. G. Klein, Ed.). Southern African
Prehistory and Paleoenvironments. Rotterdam: Balkema, pp. 221–
328.
Deacon, J. D. & Lancaster, N. (1988). Late Quaternary Palaeo-
environments of Southern Africa. Oxford: Clarendon Press.
de Graaff, G. (1981). The Rodents of Southern Africa, Durban,
Butterworth.
Folk, R. L. & Ward, W. C. (1957). Brazos River Bar: a study in the
significance or grain size parameters. Journal of Sedimentary
Petrology, 27, 3–26.
Gasse, F. (1986). East African Diatoms. Bibliotheca Diatomologica,
11. Berlin: Cramer.
Grey, D. R. C. & Cooke, H. J. (1977). Some problems in the
Quaternary evolution of the landforms of northern Botswana.
Catena, 4, 123–133.
Helgren, D. M. & Brooks, A. S. (1983). Geoarchaeology at Gi, a
Middle Stone Age and Later Stone Age site in the northwest
Kalahari. Journal of Archaeological Science, 10, 181–197.
Jacobson, L. (1987). The size variability of ostrich eggshell beads
from central Namibia and its relevance as a stylistic and temporal
marker. South African Archaeological Bulletin, 42, 55–58.
Klein, R. G. (1979). Paleoenvironmental and cultural implications of
late Holocene archaeological faunas from the Orange Free State
and north-central Cape Province, South Africa. South African
Archaeological Bulletin, 34, 34–49.
Lee, R. B. (1979). The !Kung San. Cambridge: Cambridge University
Press.
Livingstone, D. (1859). Missionary Travels and Researches In
Southern Africa, New York, NY: Harper & Brothers.
Nowak, R. M. & Paradiso, J. L. (1983). Walker’s Mammals of the
World, 4th edition. Baltimore, MD: The John Hopkins University
Press.
Patrick, R. & Reimer, C. W. (1966). The Diatoms of the United
States, Vol. 1. Philadelphia: National Academy of Sciences,
Monograph 13.
Patrick, R. & Reimer, C. W. (1975). The Diatoms of the United
States, Vol. 2, part 1. Philadelphia, National Academy of Sciences,
Monograph 13.
Pickford, M. (1990). Some fossiliferous Plio-Pleistocene cave systems
of Ngamiland, Botswana. Botswana Notes and Records, 22, 1–15.
Paleoenvironment and Archaeology of Drotsky’s Cave 21

Robbins, L. H., Murphy, M. L., Stewart, K. M., Campbell, A. C. &
Brook, G. A. (1994). Barbed bone points, paleoenvironment, and
antiquity of fish exploitation in the Kalahari Desert, Botswana.
Journal of Field Archaeology, 21, 257–264.
Robbins, L. H. (1990). The Depression site: A Stone Age sequence in
Northwest Kalahari. National Geographic Research, 6, 329–338.
Robbins, L. H. (1991). The White Paintings rock shelter: western
Kalahari Desert, Botswana. National Geographic Research and
Exploration, 7, 494–495.
Robbins, L. H. & Campbell, A. C. (1989). The Depression rock
shelter, Tsodilo Hills. Botswana Notes and Records, 20, 1–3.
Shaw, P. A. (1986). The palaeohydrology of the Okavango Delta:
some preliminary results. Palaeocology of Africa, 17, 51–58.
Shaw, P. A. & Cooke, H. J. (1986). Geomorphic evidence for the late
Quarternary palaeoclimates of the middle Kalahari of northern
Botswana. Catena, 13, 349–359.
Shaw, P. A. & Thomas, D. S. G. (1988). Lake Caprivi—a late
Quaternary link between the Zambezi and middle Kalahari
drainage systems. Zeitschrift für Geomorphologie NF, 32, 329–337.
Shaw, P. A., Thomas, D. S. G. & Nash, D. J. (1992). Late
Quaternary fluvial activity in the dry valleys (Mekgacha) of
the Middle and Southern Kalahari, Southern Africa. Journal of
Quaternary Science, 7, 273–281.
Silberbauer, G. B. (1981). Hunter & Habitat in the Central Kalahari
Desert, Cambridge: Cambridge University Press.
Smithers, R. H. N. (1971). The Mammals of Botswana, Museum
Memoir 4, National Museums of Rhodesia.
Stewart, K., Stevens, N. J. & Robbins, L. H. (1991). Fish and reptiles
from the Tsodilo Hills White Paintings rock shelter, Botswana.
Nyame Akuma, 35, 11–17.
Stewart, M. M. (1967). Amphibians of Malawi. New York, NY: State
University of New York Press.
Sundborg, A. (1956). The River Klaralven, a study of fluvial
processes. Geografiska Annaler, 38, 127–316.
Thomas, D. S. G. & Shaw, P. A. (1991). The Kalahari Environment,
Cambridge: Cambridge University Press.
Wilmsen, E. & Denbow, J. R. (1990). Paradigmatic history of
San-speaking peoples and current attempts at revision. Current
Anthropology, 31, 489–524.
Yellen, J. E., Brooks, A. S., Stuckenrath, R. & Welbourne, R. (1987).
A terminal Pleistocene assemblage from Drotsky’s Cave, western
Ngamiland, Botswana. Botswana Notes and Records, 19, 1–6.
22 L. H. Robbins et al.