holocene boundary in the caledon river southern africa

CLIMATIC CHANGE ACROSS THE PLEISTOCENE/

HOLOCENE BOUNDARY IN THE CALEDON RIVER SOUTHERN AFRICA: RESULTS OF A FACTOR

ANALYSIS OF CHARCOAL ASSEMBLAGES*

A.B. ESTERHUYSEN

DepartmenT of Archaeology, University of the Wirwarersrmul, P. 0. Wirs, 2050, Sourh Africa

P ,J. MITCHELL

Piu-Rivers-Museum, Oxford, England and

J.F. THACKERAY

Transvaal Museum, P. 0. Box 413, Pretoria, Sourh Africa

*Accepted for publication April 1999

ABSTRACT

This paper presenrs the results of a multivariate analysis of late Pleistocene and early Holocene charcoal assemblages from Rose Cottage Cave in South Africa, and Tloutle in Lesotho. The results of a factor analysis suggest that cool conditions prevailed between 12 000 and 10 000 BP, and at c. 7000 BP. Mesic conditions in the early Holocene were followed by a drier period dated between 8500 and 7200 BP, but that wetter conditions resumed circa 6500 BP. These data are discussed in the light of palaeoenvironmental evidence from Africa south of the Limpopo.

INTRODUCTION

Recent excavations at archaeological sites in the Caledon Valley have yielded evidence relevant to a wide range of palaeoenvironmental studies (Wadley 1991; Mitchell 1994). Analysis of charc<•al assemblages comprises one such study, and results fr•)m sites in both the Free State (Wadley et al. 1992; Esterhuysen 1996) and the Phuthiatsana Basin of western Lesotho have already been reported (Esterhuysen & Mite hen 1996) _ Here we use factor analysis as a means of investigating these data further. This !kind of multivariate analysis has been previously applied to both micro mammalian and pollen assemblages fr,om a number of southern African sites (Scott & Thackeray 1987: Thackeray 1987, 1988, 1994;

Thackeray & A very 1990).

the eastern Free State near Ladybrand at

±

1660 m asl (Fig. 1). It preserves a long sequence of Upper Pleistocene and Holocene occupation deposits, most recently excavated by Wadley (1991, 1997). For the period 13-5000 BP nine layers have been recognised and three distinct artefact traditions identified. Level Pt yielded a Classic Wilton assemblage and overlies assemblages from layers JaG, Ph, Cm, H and 0 that belong to the Oakhurst Industrial Complex. Beneath these layers DCM, LB and DB present a Robberg-like industry with occupation dated to both the earliest Holocene and the late Pleistocene. Details of relevant radiocarbon dates are given in Table 1.

Both an initial study of charcoals from Rose Cottage Cave (Wadley et al. 1992) and a more thorough second investigation (Esterhuysen 1996) indicated that the vegetation in the site's vicinity had responded to the climatic changes marking the Pleistocene-Holocene boundary. Charcoal was analysed from all layers except 0, DCM and the unit in Pt dating to c. 76fXJ BP. A total of 1491 pieces was examined (Esterhuysen 1996). The THE SITES

Rose Cottage Cave

Rose Cottage Cave (RCC) (29.13S; 27.28E) lies in

NORTHERN CAPE

FREE STATE

0 200km

Fig. l. Africa south of the Limpopo, showing the locations of Rose Cottage Cave, 1loutJ:e and other sites mentioned in the text. Abbreviations: BWE Bonawe; COL Colwinton; CR Craigrossie; HM Ha Makotoko; JS jubilee Shelter; MS Mashai River; RAV Ravenscraig; RC Rose Cottage Cave; RV Rietvlei; TL Tloutle.

second analysis confirmed a marked change in the kinds of vegetation represented in Late Pleistocene and early Holocene deposits. Cliffortia spp., Protea spp., Leucosidea sericea and Rhus spp. predominated in terminal Pleistocene samples, while Rhus, Olea and Buddleia dominated the early Holocene assemblages. The second analysis also suggested changes in vegetation during the first half of tht: Holocene, possibly associated with climatic change that affected the availability of plants suitable for wood collected by human agents of accumulation. In order to investigate this further, charcoal from the Lesotho site, Tloutle (Fig. 1), was analysed for comparison with the Rose Cottage Cave charcoal assemblages.

T1outle

Tloutle rock-shelter (29.28S; 27.46E) is situated at an altitude of 1879m asl in the Phuthiatsana Basin of western Lesotho. The main sequence of deposits here is

of early and middle Holocene age, although sporadic Later Stone Age occupation in the second millennium AD is also attested (Mitchell 1993a). Preceding this three distinct artefact assemblages were identified: a Robberg- like assemblage of probable terminal Pleistocene age at the base (layers BC and BS); an industry in layer GS reminiscent of Goodwin and Van Riet Lowe's (1929) original definition of the Smithfield B and now recog- nised as an evolved component of the Oakhurst Industrial Complex; and lithic assemblages representing the early and classic phases of Sampson's (1974) Interior Wilton Industry in layers CSL-LR, CSL-UP and CCL.

Assemblages from layers WC, BGL and SS, which lie just below the surface of the deposit, are palimpsests of Classic Wilton and much more recent material (Mitchell 1993a).

Charcoal was present throughout the sequence, but in layers WC, BGL and SS consisted of tiny flecks that could not be retrieved for analysis. In general, all the

Layer Lab. No. Date BP

Pt Pta-5934 5970

±

70

Pt Pta-6783 7630

±

80

JaG Pta-5600 8380

±

70

H Pta-5560 8610

±

40

0 Pta-5599 9250

±

70

DB Pta-5593 12690 ± 120

Table 2. TiouUe: radiocarb•m dates from the Pleistocene and early Holocene level~.

Layer Lab. No. Date BP

BGL OxA-4070 375

±

65

BGL OxA-4068 5080

±

80

CCL Pta-5158 6140

±

100

CSL-UP Pta-5162 6910

±

80

CSL-LR Pta-5171 7230

±

⁸⁰

GS Pta-5172 8680

±

70

charcoal pieces from th•s site were small and frag- mentary. For this reason all pieces (totalling 2271) that could be sectioned were analysed. The taxa identified indicated that the Tloutl! inhabitants were harvesting similar woods to their RCC counterparts (Esterhuysen 1996).

F ACfOJt ANALYSIS

Factor analysis can be used to detect underlying structures in large matrice~ of data (Gould 1992:238). In this study the technique v.as employed in an exploratory study to identify possible factors that contributed to variability in the relative abundance of taxa represented by charcoal in the Rose Cottage and Tloutle assemblages.

As in principal component analyses, interpretation of the dichotomies between taxa with high or low loadings on the principal factors (~{. components) relies on a knowledge of the modern distribution and abundance of those taxa (Scott & Thackeray, 1987).

Taxa not well represented by charcoal at Rose Cottage Cave and Tloutle were excluded in order to minimise statistical noise. The ren1aining taxa were re-expressed as percentages of their own sum. These percentages were then incorporated into a factor analysis. Summary statistics were calculated hy summing the products of the loadings (Table 3) on the first factor (F1) and the percentage occurrences of the corresponding taxa. These indices (factor scores based on F1) were designated FSl, and were then standardized for each assemblage between 0 and 100 (i.e., SSFl, summary statistics based on F1).

Similarly SSF2 values wne calculated by summing the products of F2 loadings and the respective percentages for the corresponding. Figs. 2 and 3 show these indices plotted for each of tht· late Pleistocene and early Holocene layers at Rose Cottage Cave and Tloutle.

high loadings on Fl include Rhus, Olea, Celtis, Buddleia and Diospyros. Rhamnus, Protea and Leucosidea flourish at cool, high-altitude conditions (Bews 1925; Palmer &

Pitman 1972; Palgrave 1983), while Rhus, Olea, Celtis, Diospyros and Buddleia often form the main components of 'warmer' montane woodlands (Palmer & Pitman 1972;

Henderson 1987). We infer that the dichotomy between taxa with high and low loadings on F1 relates primarily to temperature.

Taxa with the most negative loadings on the second principal factor (F2) are Protea, Rhamnus and Euphorbia. Leucosidea, Olea and Erica, on the other hand, display the most positive loadings on F2. An environmental factor which appears to be primarily responsihle for this dichotomy appears to be related to moisture. Protea spp. possess xerophytic characteristics (Bews 1925: 158-160) and a typical summer-rainfall species such as Protea ca.ffra is found most frequently on dry slopes and rocky ridges where drainage is exceptionally good (De Winter et al. 1966:35). Rhamnus has a wide modern distribution, but displays a high tolerance for dry conditions (Pooley 1993:300), while Euphorbia is generally adapted to dry conditions (Bews 1925:42). Leucosidea, on the other hand, prefers moister conditions; although it has been known to occupy dry areas, this is usually at high altitudes where conditions are cooler and evapotranspiration low (Scholes pers.

comm.).

In this analysis Olea and Erica have the highest F2 loadings grouped together with other taxa associated with moist conditions. In a separate factor analysis, Ericaceae pollen from the eastern Free State sites of Craigrossie, Elim and Cornelia, were associated with taxa that also related to moist conditions (Scott 1989).

DISCUSSION

The data from the two sites compare well. The SSFl 'temperature' indices suggest low temperatures around 12 690 BP. Higher temperatures prevailed around 11 000 BP. Thereafter there appears to be a dramatic increase in temperature as reflected by assemblages GS (Tloutle), Ha, Cm. Ph and Ja (Rose Cottage Cave) all dated between about 9000 and 8000 BP. Sometime around 7200 BP (CSL-LR, Tloutle) conditions became cooler.

The SSF2 'moisture' index indicates that conditions around 12 690 BP were moist. This could be attributable to more effective atmospheric moisture due to cooler conditions, and/or an increase in rainfall. Layer LB reflects drier conditions possibly associated with global warming during the Pleistocene/Holocene transition.

Early Holocene mesic conditions dating c. 8600 BP seem to have been followed by a drier episode between about 8500 and 7200 BP. Between 6900 and 5500 years ago conditions once again became wetter.

Table 3. Rose Cottage Cave and Tloutle: loadings of charcoal taxa on the first and second principal components of the factor analysis.

Celtis a.fricana Rhus spp.

Olea a.fricana Buddleia spp.

Diospyros spp.

Euryops spp.

Euphorbia sp.

Podocarpus sp.

Erica spp.

Rhamnus sp.

Protea spp.

Leucosidea sericea Pittosporum viridiflorum

F1

0,78 0,61 0,57 0,45 0,35 0,25 -0,12 -0,13 -0,27 -0,31 -0,55 -0,65 -0,73

Comparison with the re~ults of faunal and sediment analyses

Faunal evidence from mid-Holocene levels at Rose Cottage Cave are consistent with interpretations based on the charcoal data. Sable antelope (Hippotragus cf.

equinus) occur in the Pt layer and kudu (Tragelaphus strepsiceros) and vervet monkey (Cercopithecus aethiops) in the Oakhurst levels, which "suggests that the valley of the Caledon River, or perhaps the area in general, could have been more wooded in the past than it is at present"

(Plug & Engela 1992: 19). Further evidence for a mesic close-canopy vegetation at this time may be given by the overwhelming presence of vleirats (Otomys irroratus) in the associated micro mammalian assemblages (Avery 1982). Analysis of the representation of the ungulate taxa at Rose Cottage Cave ac<"ording to their preference for wet or dry environments also indicates that conditions were wettest at the end of the Pleistocene and again in level Pt (Engela 1995).

Butzer's (1984) sediment analysis of the stratigraphic sequence at Rose Cottage Cave is compatible with these observations based on fauna and charcoal. His identification of two 'wett~r· periods, the first associated with a date of8600

±

100 BP (Pta-2067) and the second with one of 6850

±

45 BJ> (GrN-5299) fits the results of the factor analysis, without necessarily implying that rainfall remained high throughout the intervening period.

Isotopic analysis of faunal remains from RCC and Tloutle likewise indicated a warmer phase c. 8400-8100 BP followed by a cooler phase c.7200 BP, and a return to warmer conditions between 6900 and 5900 BP (Smith 1997: 129).

At Tloutle three additional lines of evidence support the results described above. First, the occurrence of sinter deposited from springs in the GS layer (which is associated with a date of 8680

±

70 BP, Pta-5172) implies that conditions were relatively humidl at this time.

This is supported, both in GS and for the period 6900 - 5000 BP, by the presence of Otomys irroraws, a rodent species usually indicative •)f a high percentage of canopy

Erica spp.

Olea a.fricana Leucosidea s.ericea Podocarpus sp.

Pittosporum viridiflorum Celtis africaHa

Euryops spp.

Diospyros spp.

Buddleia spp.

Rhus spp.

Euphorbia sp.

Rhamnus prinoides Prorea spp.

F2

0,87 0,58 0,54 0,47 0,39 0,27 0,22 0,07 -0,04 -O,I5 -0,31 -0,32 -0,52

cover accompanied by a moist streamside element (Plug I993). Thirdly, the presence of ostrich (Strurhio came/us) in the fauna from the 7230-6IOO BP layers (Plug 1993) is consist-ent with this having been a relatively warm period as ostrich are not found in western Lesotho today.

Regional Evidence

Pollen studied from Craigrossie (Fig. I), eastern Free State, shows the development of Podocarpus forest and moist conditions prior to 10 700 BP (Scott I989: 107;

Scott 1990). Similar moist, cool conditions at 10 200

±

100 BP are suggested by a high percentage of Leucosidea sericea in charcoal assemblages at Ravenscraig (Fig .I), northern Eastern Cape (Tusenius I989). Based on EDXA measurements of Olea and Rhus charcoals from Bonawe (Fig. I) in the same area dry conditions are thought to have occurred at 8040 ± 100 BP (Pta-1709; Tusenius 1989). A period of cooling c. 9000 - 8300 BP is also evident from the relative increase in Leucosidea in charcoal assemblages at Ha Makotoko, western Lesotho (Fig. 1) (Mitchell 1993b; Esterhuysen & Mitchell 1996).

Carter's (1976) dating of peat layers interbedded with fluviatile material in the Upper Mashai River catchment of the eastern Lesotho highlands shows an increase in precipitation from 7280

±

150 BP (Q-1165).

Studies carried out at sites on the margins of the interior plateau, in the Northern Province bushveld zone and along the coast exhibit similar trends (Partridge er al.

1990; Partridge 1993). In general the palaeoclimatic proxy evidence for several southern African localities south of the Limpopo suggests that conditions at the end of the Ple.stocene c. 12000 BP were cool and moist, that a dry phase occurred c.8000 BP and that there was a resurgence of wet conditions after 7000 BP (Partridge er al. 1990; Partridge 1993). Two localised incidences of contradictory evidence need to be noted. Charcoals from Colwinton, northern Eastern Cape Province (Fig. 1), indicate dry conditions c. 6270 BP (Tusenius 1989) and micro mammalian data from Jubilee Shelter, North West Province (Fig. I), suggest that tree cover there also

100- PT Cf 90-

60- ^UP-CSL

~

_70-

0 ~

...

CCL

w _{60 -}

a: ~

~

^50-

UJ

0.. 40 -

:::E

UJ t-

l.R-CSL

•

-

^30-

u..

C/)

(/) 20-

10 -

-

I I

5 6 7

Q JA

/_../

~

^Ph

... t •

^TLOlJTl.E

... ~ 0 ROSE COTIAGE CAVE

I

8

1GJ Ha :! :$

::: iH

HI

:::

y!

Cml 6\

l LB

•··· · ·· ···[)

GS ··· ....

I I

·~

.. .

9

I

10 11

·.

_··~

. .

·· ...

· · .. _··.

12

'

llME (x 1000 YEARS B.P.)

Fig. 2. Rose Cottage Cave and Tloutle: plot of values of SSFI 'temperature' index obtained from factor analysis of charcoal data from the late Pleistoc.,ne and early Holocene layers.

declined at this time (/\very 1993). Though higher resolution models need to be deve'loped which can take account of such local anomalies (Partridge 1990), in general the evidence supports the case for a wet Altithermal across Africa south of the Limpopo (Partridge 1993).

The high degree of consensus between results of the charcoal analysis and <•ther palaeoclimatic evidence suggests that the patterning detected by factor analysis in the Rose Cottage Cave and Tloutle charcoal data is:

related primarily to temperature and moisture fluctuation.

It would appear that palaeoenvironmental factors have played a major role in contributing to variability in the relative abundance of charcoal fragments representing taxa selected by human <•ccupants of the Rose Cottage and Tloutle cave sites.

CONCLUSION

Charcoals were analysed from two different sites in the Caledon River Valley. The application of factor analysis reported here has facilitat~ a comparison of the results of the charcoal analysis from the eastern Free State site, RCC and the penecontemporaneous western Lesotho site of Tloutle. It has allowed a more refined index of palaeoenvironmental change to be produced (Figs 2 and 3). The results ofthe fact<•r analysis confirm cool tempe-

rature regimes between 12 000 and 10 000 BP. An increase in temperature occurred within the period between 9000 and 8000 BP, followed by a cool phase c.

7 200 BP. The factor analysis on charcoal data also provides an indication of relative changes in moisture through time: mesic conditions were followed by a drier period between 8500 and 7200 BP, while moister conditions prevailed circa 6500 BP. This pattern of moisture fluctuation is evident at other sites in South Africa and Lesotho. The fact that 'wet' spells occurred in both the northern and :southern regions of Africa south of the Limpopo suggests that higher resolution models (sensu Partridge 1993:242) are required to explain these phenomena.

ACKNOWLEDGEMENTS

Our thanks are extended to S. Hall, A. Jerardino, J.

Smith and L, Wadley for commenting on earlier drafts of this paper. We would also like to thank referees J. Lee- Thorpe and F. Prins for providing valuable comment.

We thank Drs J. Vogel and R. Houseley for radio- carbon dates. Funding for this study has been provided by the Foundation for Research Development, the British Academy, and the Centre for Scientific development and the HSRC. Excavation at Tloutle was made possible by a permit issued by the Lesotho Protection and Preser-

100 -

90 -

Cm 80 -

... .o

... / DB

liJ• GS :'

PT ./~~;sL If ··· ... · ..

o• . : Ha ·.

~

70- 0

z

w

^{60 -} a:

...

/

...

· ... / .. / .. /

~L ^{\':.. !} ~

\ ":i

::>

t>

^{50 -}

6

~ 40-

~

\ JA)

~ ... ···~!

(/) 30-

(/)

LB

20-

LR-CSL

\1 u

_::

b

^Ph

10 -

-

I I I I

5 6 7 8

I

9

•

0

I

10

TLOIJTLE

ROSE COlTAGE CAVE

I I

11 12

llME (x 1000 YEARS B.P.)

Fig. 3. Rose Cottage Cave and TJoutle: plot of values of SSF2 'moisture' index obtained from factor analysis of charcoal data from the late Plei'itocene and early Holocene layers.

vation Commission and was supported by the British Academy, the Swan Fund. the Society of Antiquaries, the Sir Henry Strak:osch Memorial Trust, the University of Oxford, the Prehistoric Society and St. Edmund Hall, Oxford.

REFERENCES

Avery, D. M. 1982. Mkromammals as palaeoenviron- mental indicators and an interpretation of the late Quaternary in the southern Cape Province, South Africa. Annals of the South African Museum 85:183- 374.

Avery, D. M. 1993. Last Interglacial and Holocene altithennals in South Africa and Namibia: micro·

mammalian evidence. Palaeogeography, Palaeo- climatology, Palaeoecnlogy 101: 221-22&.

Bews, J. W. 1925. Plant forms and their evolution in South Africa. London: Long mans, Green and Co.

Butzer, K. W. 1984. Archaeology and Quaternary environments in the interior of southern Africa. In: R.

G. Klein (ed.) Southern African prehistory and palaeoenvironments: 1-64. Rotterdam: A. Balkema.

Carter, P. L. 1976. The effects of climatic change on settlement in eastern Lesotho during the Middle and Later Stone Age. World Archaeology 8:197-206.

De Winter, B., De Winter, M. & Killick, D.J.B. 1966.

Sixty-Six Transvaal Trees. Pretoria: Botanical Research Institute.

Engela, R. 1995. Space, Material Culture and Meaning in the late Pleistocne and Holocene at Rose Cottage cave. Unpublished MA thesis, Department of Archaeology, University of the Witwatersrand.

Esterhuysen, A. B. 1996. Palaeoenvironmental recon- struction from Pleistocene to present: an analysis of charcoal from the eastern Orange Free State and Lesotho. Unpublished MA thesis, Department of Archaeology, University of the Witwatersrand.

Esterhuysen, A. B. & Mitchell, P. J.1996. Palaeo- environmental and archaeological implications of charcoal assemblages from Holocene sites in western Lesotho, southern Africa. Palaeoecology of Africa, 24: 203-232.

Goodwin, A. J. H. & Van Riet Lowe, C. 1929. The Stone Age cultures of South Africa. Annals of the South African Museum 27:1-289.

Gould S.J. 1981. The Mismeasure of Man. London:

Penguin Books.

Henderson, L. 1987. Barrier plants of Southern Africa.

Memoi;rs of the Botanical Survey of South Africa.

No. 55. Botanical research Institute - Department of AgricuLture and Water Supply.

two Lesotho rock-shelters:. the terminal Pleistocene/early Holocene assemblages from Ha Makotoko and Ntloana Tsoana. Proceedings of the Prehistoric Society 59:39-60.

Mitchell, P. l. 1994. The archaeology of the Phuthiatsana -ea-Thaba Bosiu Basin, western Lesotho, southern Africa: changes in Later Stone Age regional demography. Antiquity 68: 83-96.

Palgrave, K.C. 1983. Trees of Southern Africa. Ca.pe Town: C.Struik Publi!-hers.

Palmer, E.& Pitman, N. 1972. Trees of southern Africa. Vol I & 2. Cape Town: A.A. Balkema.

Partridge, T., 1990. Cainozoic e111vironmental changes in southern Africa. South African Journal of Science 86:315-317.

Partridge, T., 1993. Wanning phases in southern Africa during the last 150 000 years: an overview. Palaeo- geography, Palaeoclimatology, Palaeoecology 101:

237-244.

Partridge, T. C., Avery D.M., Botha, G.A., Brink, J.S., Deacon, J.,Herbt·rt, R.S., Maud, R.R., Scholtz, A., Scott, L., Talma. A.S. and Vogel, J.C. 1990.

Late Pleistocene and Holocene climatic change in Southern Africa. South African Journal of Science 86: 302-306

Plug, I. 1993. The macr<•faunal and molluscan remains from Tloutle, a Later stone age site in Lesotho.

Southern African Field Archaeology 2: 44-48.

Plug, I. & Engela, R. 1992.The macrofauna! remains from recent excavations at Rose Cottage Cave, Orange Free State. South African Archaeological Bulletin 47:16-25.

Pooley, E. 1993. The complete field guide to trees of Natal Zululand and Transkei. Natal: Natal Flora Publications Trust.

Sampson, C. G. 1974. The Stone Age archaeology of southern Africa. New York: Academic Press.

Scott, L. 198'9. Late Quaternary vegetation history and climatic change in th<! eastern Orange Free State, South Africa. South African Journal of Botany 55:107-116.

nary warming episodes in Southern Africa. Palaeo- geography, Palaeoclimatology, Palaeoecology 101:

229-235.

Scott, L. & Thackeray, J. F. 1987. Multivariate analysis of Late Pleistocene and Holocene pollen spectra from Wonderkrater, Transvaal, South Africa. South African J'Jurnal of &ience 83: 93-98.

Smith, J. 1997. Stable Isotope Analysis of Fauna and Soils from Sites in the Eastern Free State and Western Lesotho, Southern Africa: a palaeoenvironmental interpretation. Unpublished Msc Thesis. Department of Archaeology: UCT.

Thackeray, J. F. 1987. Late Quaternary environmental changes inferred from small mammalian fauna, southern Africa. Climati~.: Change 10: 285-305.

Thack.eray, J. F. 1988. Quantification of climatic change during the Late Quaternary in southern Africa.

Palaeoecology of Africa 19: 317-324.

Thackeray, J.F. 1994. Letter to the Editor: Comment on Temperature Indices from the Late~ Quaternary Terrestrial Sequence at Wonderkrater. South Africa.

Quaternary Research 42: 354-355.

Thackeray, J. F. & Avery, D. M. 1990. A comparison between temperature indices for Late Pleistocene sequences at Klasies River Mouth and Border cave, South Africa. Palaeoecology of Africa 21:: 311-315.

Tusenius, M. L. 1989. Charcoal analytical studies in the north-eastern Cape, South Africa. South African Archaeological Society Goodwin Series 6: 77-83.

Wadley, L. 1991. Rose Cottage Cave: background and preliminary report on the recent excavations. South African Archaeological Bulletin 46: 125-130.

Wadley, L. 1997. Rose Cottage Cave: Archaeological Work 1987-1997. South African Journal of Science, 93: 439-444.

Wadley, L., Esterhuysen, A. B. & Jeannerat, C. 1992.

Vegetation changes in the eastern Orange Free State:

the Holocene and late Pleistocene evidence from charcoal studies at Rose Cottage Cave. South African Journal of Science 88: 558-563.