5. HOLOCENE CLIMATIC TRENDS AND RHYTHMS
IN SOUTHERN AFRICA
LOUIS SCOTT (Scottl.firstname.lastname@example.org)
Department of Plant Sciences
University of the Free State
P.O. Box 339, Bloemfontein 9300
JULIA A. LEE-THORP (email@example.com)
Department of Archaeology and Quaternary Research Centre
University of Cape Town
Private Bag Rondebosch 7701
Keywords:Aridity, Rainfall seasonality, Grassland savanna,Isotopes, Pollen, Human response, Holocene, South-
Southern Africa’s unique mid-latitude oceanic position invites broad comparisons with
palaeoclimatic records across the PEP III transect including long-distance thermohaline
circulation teleconnections. Atmospheric and oceanic circulation systems around Southern
Africa (Fig. 1) are linked (Lutjeharms et al. 2001). They interact to inﬂuence distribution
of biomes including prominence of C4and C3grasses in the summer-rain and winter
rain regions respectively (Vogel et al. 1978; Cowling et al. 1997) (Fig. 2). Climate is
dominated by two systems, the westerlies and easterlies (Fig. 1) and shifts in these systems
undoubtedly affected the climate history during the Holocene. The generally moderating
effect of the oceans, in particular the warm Agulhas western boundary current on the east
coast (Lutjeharms et al. 2001), and the semi-arid nature of the region suggests that moisture
rather than temperature changes is the more important climate parameter, at least on the
Holocene time scale. Furthermore, the cold Benguela upwelling zone and its associated
atmospheric circulation system in the South Atlantic is strongly linked to coastal aridity on
the west coast.
Asymmetry in the continental conﬁguration of Africa introduces a number of contrasts
with the north of the continent, in climate patterns, biome distributions and orbital param-
eters. In contrast to equivalent regions in Northern Africa that border Eurasia, Southern
Africa lacks the inﬂuence of other continents. Symmetry of biomes on both sides of the
equator in Sub-Saharan Africa is strong near the equator but weakens towards the northern
R. W. Battarbee et al. (eds) 2004. Past Climate Variability through Europe and Africa. Kluwer
Academic Publishers, Dordrecht, The Netherlands.
70 SCOTT AND LEE-THORP
Figure 1. Map showing the global oceanic position of Southern Africa. The climate of the summer rain region is
under inﬂuence of easterly winds and the Agulhas current, and the western winter rain and arid regions under the
inﬂuence of westerlies frontal systems and the Benguela upwelling system respectively. L and H = areas of low
and high pressure, T = Taylor Dome.
10˚ 20˚ 30˚ 40˚ 50˚E
HOLOCENE CLIMATIC TRENDS AND RHYTHMS IN SOUTHERN AFRICA 71
Figure 2. Map showing the approximate positions of major biomes in Southern Africa and the location of sites
mentioned in the text. The dashed lines indicate the intermediate zone between summer and winter rainfall regions.
and southern extremes of the continent. The northern parts of Southern Africa with sub-
tropical and tropical grasslands and savannas have equivalents elsewhere in Africa but the
Southern African Desert, Succulent Karoo, Nama Karoo, and Fynbos biomes (Fig. 2) are
unequaled (Cowling et al. 1997). The southern tip of Africa, which receives winter rain,
has a Mediterranean type of environment (the Fynbos Biome) but it has Gondwanaland
afﬁnities unlike its northern African counterpart, which shows a Eurasian structure.
Biome and climate patterns on both sides of the equator probably ﬂuctuated asymmetri-
cally in the past following orbital forcing with long-term moisture development in Southern
Africa out of phase with the northern tropical parts of the continent. (Street-Perrott and
Perrott 1993). Lake Malawi is the northernmost site that conforms to the southern pattern
(Finney et al. 1996). The 8.2 kyr (1000 years) drought event that occurred in Northern
Africa near the peak of the precession cycle (Gasse 2000) is an example of a rapid event
for which inter-hemispheric connections should be investigated further in Southern Africa
in order to understand possible connections and anomalies in the broad pattern.
Continuous lake sequences are unfortunately very rare in SouthernAfrica; therefore a di-
verse range of archives has been used in palaeoclimate reconstruction. Several factors inhibit
accumulation of good-quality, high-resolution records, and this is a particular disadvantage
* Tswaing Crater
* Rietvlei Dam
* Moreletta River
Tropic of CapricornTropic of Capricorn
Southern KalahariSouthern Kalahari
Elands Bay *
Cave Cango Caves* *Uitenhage
Norga * * Nelson Bay Cave
Elands Bay *
Cave Cango Caves* *Uitenhage
Norga * * Nelson Bay Cave
Deelpan* Rose *
Deelpan* Rose *
* Clarens * Lake Teza
* Kosi Bay
* Lake Teza
* Kosi Bay
Windhoek *Windhoek *
* Scot's Farm
* Tswaing Crater
* Rietvlei Dam
* Moreletta River
* Scot's Farm
*Kuiseb River*Kuiseb River
Limpopo RiverLimpopo River
72 SCOTT AND LEE-THORP
for determination of Holocene climates where greater resolution is required. The underlying
uplifted geological structure and ancient erosional regime of the interior and a lack of recent
volcanics do not favour formation of basins like small craters and marshes. Hence, organic-
rich accumulations are rare; furthermore evaporation and desiccation in the region have
lowered the quality of deposits where they do occur. Continental data are spatially patchy
and obtained largely from relatively coarsely dated, and often pulsed, pollen sequences
and faunal abundance, isotope and other biological data from archaeological sites, and
more rarely from aquifers. Some continuous stalagmite records are capable of providing
information on long-term trends and high frequency variability with good chronological
control. They can address key climatological questions, such as if the Little Ice Age (LIA)
and Medieval Warm Period (MWP) were manifested on the southern tip of Africa, and how
climate changes affected human settlements and biota.
Evidence for millennial scale climate and vegetation change during the Holocene
By pairing more abundant but lower resolution records from pollen, sediments, etc., with
rarer high-resolution, chronologically-controlled data from stalagmites and well-resolved
archaeological sites, palaeoenvironmental reconstructions can be optimised. Evidence of
Holocene environmental conditions is presented here according to time-slices that can be
broadly correlated over a wide area of the sub-continent although they show considerable
internal variability and do not by any means represent uniform climatic phases. As far as
possible the information from each time-slice is presented starting from the northernmost
region and ending with the south. Site localities are shown in Figure 2.
In general, relative aridity is indicated for this period, excepting in the southwestern winter-
rainfall region, which was relatively mesic.
A pollen sequence from Atlantic marine sediments offshore of the Kunene River Mouth
(Fig. 1) suggests relative aridity in the area bordering Namibia and Angola before 7.5 kyr
(Shi et al. 1998). Geomorphological data from Northern Namibia also suggest dry condi-
tions although not as dry as the preceding terminal Pleistocene (Eitel et al. 2001). Currently
submerged speleothems from Aikab, representing lower water tables than at present, may
suggest aridity ca. 10 kyr (Brook et al. 1999).
In the central region south of the Limpopo River (Fig. 2), pollen sequences at Won-
derkrater (Fig. 3), Rietvlei, Tate Vondo, and Tswaing Crater show an extension of Kalahari
Thornveld elements between ca. 10 and 7.3kyr, indicating dryness together with rising
temperatures (Scott 1993; 1999). In the Southern Kalahari, micromammal data at Wonder-
werk Cave (Avery 1981) suggest arid, open conditions. Carbon isotope data from grazing
herbivores at Equus and Wonderwerk Caves show that C4grasses remained dominant in this
region throughout the Holocene (Lee-Thorp and Beaumont 1995), indicating no changes
in rainfall seasonality.
In the eastern highlands of the Free State and Lesotho, rapid, large temperature shifts of
several degrees Celsius magnitude are indicated by carbon isotope data from large-bodied
grazers in the well-dated Rose Cottage Cave sequence, reﬂecting changes in the proportions
HOLOCENE CLIMATIC TRENDS AND RHYTHMS IN SOUTHERN AFRICA 73
Figure 3. (a) Selected calibrations of radiocarbon dates (Scott 1982a; Scott et al. 2003), (b) temperature (solid) and moisture (dashed) indications for theWonderkrater Borehole
3 based on principal components analysis of indicator pollen types (excluding taxa in c) (Scott 1999), (c) percentages of prominent pollen types (Scott 1982a), and (d) δ13C
values of the organics in the spring deposits (L. Scott et al., in prep).
Wonderkrater, Borehole 3
20 40 60 80
20 40 60 80
ab c d
74 SCOTT AND LEE-THORP
of C3and C4grasses (Smith 1997; Smith et al. 2002). In this highland area the proportion
of C3grasses increases with altitude and decreasing temperatures, while that of C4grasses
increases with decreasing altitude and warmth. Cool conditions with C3grasses pertained
at 10, 9.5 and 8 kyr, interspersed with warmer conditions (more C4grass) at 9.9, 9, and
ﬁnally after 7.6 kyr (Fig. 4a). Indices from charcoal abundance data of species adapted
to different temperatures, from the same sites (Esterhuysen et al. 1999), also show rapid
early Holocene variation but suggest that associations between temperature and moisture
conditions were not ﬁxed. Gas solubility data in the Uitenhage aquifer (Fig. 4b) shows
broadly similar temperature shifts (Heaton et al. 1986) but the record lacks the resolution
to resolve even millennial scale shifts at this time. Dominance of Asteraceae pollen at
Blydefontein and prominent Euryops (Asteraceae) charcoal in the Southern Drakensberg
indicate aridity from 11 to 8.7 kyr and again at ∼7.1 kyr (Tusenius 1989; Bousman 1991).
Vegetation typical of the current arid Succulent Karoo biome (Fig. 2) developed at Ek-
steenfontein between ca. 11 and 9.4 kyr (Scott et al. 1995).Along the south coast relatively
high rainfall occurring in both winter and summer currently promotes coastal forests or
thicket; both C3and C4grasses are present where grass occurs (Cowling 1983). Relatively
more fynbos and less forest at ∼8.7 to 7.7 kyr is suggested by pollen at the currently forested
site of Groenvlei (Martin 1968). At the coastal archaeological site of Nelsons Bay Cave,
however, consistently low nitrogen isotope ratios (15N/14N) from grazing fauna through
the Holocene, suggest that little change occurred in rainfall receipts in the region (Sealy
In contrast, the southwestern winter rainfall region experienced relatively mesic con-
ditions during the early Holocene. Archaeological evidence shows a higher abundance of
mesic woody elements in charcoal assemblages until ca. 8.5 kyr (Cartwright and Parkington
1997; Cowling et al. 1999) and lower δ18O in ostrich eggshell at coastal Elands Bay Cave
(Lee-Thorp and Talma 2000). Pollen in a peat deposit in the adjacent Cederberg mountain
range, however, shows little change during this period (Meadows and Sugden 1991).A hyrax
dung pollen sequence slightly further northwest in the Cederberg (Pakhuis Pass) closer to the
Nama Karoo boundary, suggests slightly more mesic conditions with prominent Dodonaea
and other woody elements (Scott 1994).
Numerous proxies show that the temperature optimum beginning shortly before 7.5 kyr
(Scott 1982a; 1993; Lee-Thorp and Talma 2000) also showed moisture improvement, at
least in the summer rainfall region.
On the western side of the subcontinent, desert pollen declined in offshore (windblown)
sediments near the Kunene River Mouth at 7kyr in favour of more mesic afro-montane
elements (Shi et al. 1998). In Central Namibia near Windhoek, a warmer, moister phase
with grass and woodland pollen is indicated between 7.9 and 6.4 kyr (Scott et al. 1991).
Dates for submerged speleothems in Northern Namibia, however, indicate low water tables
(Brook et al. 1999), seeming to contradict the pollen indicators.
In the savanna region of the northern interior pollen data from Wonderkrater (Fig. 3)
suggest generally rising temperatures between ∼9 and 6 kyr, with a delayed moisture in-
crease ∼6.5 kyr (Scott 1982a). Higher humate levels (recorded in the grey scale) and lower
HOLOCENE CLIMATIC TRENDS AND RHYTHMS IN SOUTHERN AFRICA 75
Figure 4. (a) Proportions of C4grasses indicated by δ13C in grazer tooth enamel from Rose Cottage Cave and
three nearby sites in the eastern grassland that provide indications of temperature shifts (Smith et al. 2002).
(b) Temperature curve derived from dissolved gases for the Uitenhage Aquifer, shown with the δ18O series (based
on data from Heaton et al. (1986)).
δ13C in the nearby Makapansgat Valley T7 stalagmite (Fig. 5) indicate wetter, warmer
conditions with enhanced C3vegetation density (woody, shrubby and herbal forms) at
6.4 kyr to 5.1 kyr (Lee-Thorp et al. 2001). In the Makapansgat stalagmites δ18O values are
interpreted as reﬂecting primarily isotopic composition of rainfall, and hence, hydrological
2000 4000 6000 8000 10000 120000
Age (calibrated years BP)
2000 4000 6000 8000 10000 120000
= Rose Cottage Cave
76 SCOTT AND LEE-THORP
conditions (Holmgren et al. 1999; Lee-Thorp et al. 2001). Following this interpretation,
higher δ18O values in this period (6.4–5.1 kyr) indicate moister, warmer conditions. The
highly resolved T7 sequence, recently augmented by similar results for another stalagmite
from the same site (Holmgren et al. 2003), shows that climate variability occurred on short
time-scales including interruptions of the “optimum”, which are blurred in all of the less
resolved records. Warm conditions were continuous until 5.7 kyr, where after 2 reversals
occurred before 5.1 kyr (Lee-Thorp et al. 2001).
Figure 5. δ13 C, δ18O and grey scale series for the T7 stalagmite, Cold Air Cave in the Makapansgat Valley,
plotted against age after interpolation to an average age interval of 12 years (from Lee-Thorp et al. (2001)).
In the highveld grassland to the south, high δ13C in grazers at Rose Cottage Cave and
other sites show that C4grass was as abundant as today by ∼7.6 kyr. Warmer temperatures
300 0 1000 2000 3000 4000 5000 6000
Age (years BP)
δ18O PDB (‰)δ13C PDB (‰)
HOLOCENE CLIMATIC TRENDS AND RHYTHMS IN SOUTHERN AFRICA 77
supporting C4grasses prevailed from this period onwards, or at least for those periods
represented in the archaeological sites (Smith 1997; Smith et al. 2002). Earlier interpreta-
tions of sediments at Florisbad and surroundings (Butzer 1984; Visser and Joubert 1991)
suggested semi-arid conditions despite higher water levels in pans. Recent pollen data
suggests that high seasonal evaporation rates may account for the apparent contradiction,
and that wetness increased after 7.2 kyr at Florisbad (Scott and Nyakale 2002). As today,
frost on the ‘highveld’ continued to prevent the spread of trees at Florisbad, but at ∼7.2 kyr
a brief southward expansion of savanna, and hence less frost, is indicated in the Rietvlei
Dam pollen sequence (Scott and Vogel 1983). Savanna expansion is also suggested by
limited pollen evidence from pan sediment sites associated with stone tools belonging to
the Oakhurst Complex (Horowitz et al. 1978) nearly 100 Km to the west of Florisbad nearer
the boundary of the Savanna Biome.
From the eastern limits of the Fynbos Biome, the dissolved gas temperature record
from the Uitenhage aquifer (Heaton et al. 1986) shows a temperature increase of ∼4◦C
at 6.3 kyr, above the 8.5 kyr temperature determination, followed by marked cooling at
∼5.5 kyr (Fig. 4b). This pattern is in broad agreement with the higher resolution Maka-
pansgat and Cango Cave stalagmite δ18O records (Fig. 5). Although it is not possible to
determine the amplitude of temperature depression from the Makapansgat stalagmites,
for Cango a mean annual temperature depression of ∼1.5 ◦C occurs at 5 kyr (Talma and
Vogel 1992; Lee-Thorp and Talma 2000). On the coast, δ18O in marine molluscs from
Nelson’s Bay Cave indicate warmer sea surface temperatures (SST) at ca. 6.8 kyr (Cohen
and Tyson 1995). Higher SST was interpreted to be a result of intensiﬁed westerlies and
their associated frontal disturbances by Cohen and Tyson (1995), but other inﬂuences, for
instance slightly higher Agulhas current inﬂuence or temperatures could also be considered.
The δ13C record from large grazing herbivores indicates a short episode with fewer C4
grasses at ∼5.5 kyr (Sealy 1997), suggesting greater inﬂuence of winter rainfall or possibly
also lower temperatures, which is apparently not reﬂected in the marine mollusc δ18O
record. Unfortunately, dating resolution in the archaeological site and especially the aquifer
is not high enough to determine whether the 5.5 kyr and 5kyr “events” are the same or not.
Numerous coastal records show that sea-levels were about 2m higher in the middle
Holocene. These levels are recorded at Groenvlei (Martin 1968), the Verlorenvlei and
along the western Cape coast near Elands Bay (Fig. 2) (Miller et al. 1993; Meadows
et al. 1996), and at Lake Teza and surrounding coast line, before 5.6 kyr (Ramsay 1995;
Scott and Steenkamp 1996). The change in pollen trapping between marine and freshwater
marshy situations complicates estimates of moisture conditions during this phase. A greater
abundance of grasses occurred until shortly after 5 kyr atVerlorenvlei after which relatively
dry conditions may be indicated by Asteraceae pollen (Meadows et al. 1996).
A climate sequence for the Western Cape winter rainfall region is difﬁcult to establish
owing to a lack of indicators during this period, at least partly due to the rarity of human
Various proxy data sequences show that an overall aridiﬁcation trend began across the
subcontinent soon after 5 kyr. There are, however, some contradictions in these records,
78 SCOTT AND LEE-THORP
some of which may simply be due to poor chronological resolution that confuses different
episodes, others too subtle, imperfectly understood, differences in the perspectives offered
by different proxies. For instance, δ13C in the geographically widely separated Makapansgat
and Cango stalagmites (Fig. 2) show decided increases in C4grass after 4 kyr, reaching peak
abundances at 2–2.4 kyr (Figs. 5 and 6). Pollen data, however, suggest increasing abundance
of Asteraceae (Fig. 3), a C3shrub characteristic of summer rain areas incorporating also
some winter rain, as typically occurs in the dry Karoo region today.
Figure 6. δ13 C, δ18O and the derived temperature sequence of the Cango Caves stalagmite (from Talma and
Vogel (1992), Lee-Thorp and Talma (2000)).
A variable, modestly moist/warm period is indicated by the Makapansgat stalagmite
δ18O and grey series (showing relatively high humate levels) for the interior summer
rainfall region from 4.3 to 3.2 kyr (Fig. 5) (Lee-Thorp et al. 2001). Termination of this
event corresponds with a vegetation change evident in the δ13C and grey index; lower
δ18O values thereafter suggest increasing aridity. Two relatively modest δ18O minima cor-
responding to drying and/or cooling are clustered at 3.1 and 2.6 kyr. The Cango Cave δ18O
sequence (Talma and Vogel 1992) indicates variably lower temperatures between 3.3 and
2.5 kyr, while principal component analysis of selected pollen at Wonderkrater (excluding
Asteraceae), and enhanced Ericaceae pollen in the Blydefontein Shelter deposits ∼3.1 kyr
also suggest cooler conditions (Scott and Thackeray 1987; Scott 1999; C.B. Bousman and
L. Scott, unpublished). Two brief episodes of higher δ18O in the Makapansgat stalagmite,
suggesting moister conditions, at 2.1 and 2.9 kyr correspond to two distinct peaks of swamp
forest growth at Scot’s Farm on the northern foot of the Soutpansberg (Scott 1982b).
The values of δ13C in the Makapansgat stalagmites peak at ∼2.2 kyr, indicating a
stepwise trend towards higher C4grass abundance (Fig. 5). This matches the pattern
observed in the Cango stalagmite in the southern Karoo/fynbos transition (Fig. 6) (Talma
and Vogel 1992), suggesting those conditions supported more C4grasses. In contrast,
the Wonderkrater spring pollen sequence (Fig. 3) shows a strong local representation of
fern spores after 4.1 kyr, and little Poaceae (grass) pollen. Asteraceae remained relatively
0 1000 2000 3000 4000 5000 6000
Age (calibrated years BP)
δ18O PDB, δ13C PDB (‰)
HOLOCENE CLIMATIC TRENDS AND RHYTHMS IN SOUTHERN AFRICA 79
prominent but since no other indications of aridity are observed, their presence could
reﬂect cooler conditions similar to adjacent high-lying surroundings like the Waterberg
(Scott 1982a). Development of cooler conditions is supported by independent PCA that
exclude Asteraceae pollen (Scott and Thackeray 1987; Scott 1999) (Fig. 3). Sediment δ13C
values at Wonderkrater suggest that, apart from locally abundant C3Phragmites (Vogel
et al. 1978) the remaining grasses must have included C4types (Fig. 3d).
Pollen data from Florisbad suggests relatively mesic, or grassy, conditions in the central
highlands after 4.9 kyr (Scott and Nyakale 2002), also observed at Deelpan and Blydefontein
(Scott 1993). Shrubby Asteraceae pollen became gradually more prevalent between 4.4 and
2 kyr at a number of highveld grassland and Karoo sites suggesting karroid vegetation but
the pattern is not replicated at Florisbad, where increased Cyperaceae pollen by 3.9 kyr at
the site indicates local wetness. In Equus Cave in the Southern Kalahari, ostrich eggshell
δ15N and δ18 O eggshell apparently do not suggest dry conditions from 5–2 kyr (Johnson
et al. 1997). Although these data are few, they are supported by relatively low δ15N values
from grazing fauna at Wonderwerk Cave, and a micro-mammmal-based moisture index,
suggesting relatively moist conditions for this region between ∼4.2 and 2 kyr (Thackeray
and Lee-Thorp 1992).
On the east coast, Podocarpus pollen ﬂuctuations in lake and swamp deposits suggest
extensive mesic forests during the middle Holocene that began to diminish in size after
3.1 kyr (Scott and Steenkamp 1996; Mazus 2000).Vessel diameter of fossil wood charcoal
in the higher altitude Drakensberg mountain region indicates relatively wetter conditions
ca. 2.4 kyr (February 1994).
Marked climatic change is indicated in the southwestern Cape from 4.7 kyr. A con-
spicuous arid episode is indicated by high δ18O ostrich eggshell values at Elands Bay
Cave between 4.7 and 3 kyr (Lee-Thorp and Talma 2000). At the same time, increased
Asteraceae shrubs are recorded in charcoal assemblages from this site, indicating general
and xeric thicket but no mesic thicket like Dodonaea (Cartwright and Parkington 1997;
Cowling et al. 1999). Meadows et al. (1996) note increasing Asteraceae at Verlorenvlei
(adjacent to Elands Bay). Sea surface temperatures off the west coast were cooler ca. 3 kyr
(Cohen et al. 1992; Jerardino 1995), suggesting increased upwelling intensity, or reduced
Agulhas water inﬂuence, or possibly inﬂux of cooler waters from the Southern Ocean.
The Elands Bay ostrich eggshell series suggests the return of more mesic conditions from
2.7–1.2 kyr (Lee-Thorp and Talma 2000), while pollen from hyrax dung middens in the
Cederberg show more Cyperaceae, “renosterbos” (Stoebe/Elytropappus type) and declining
Dodonaea thicket ca. 2 kyr (Scott 1994).
In contrast to the Western Cape, in the Southern Cape region, increased forest pollen in
a peat sequence at Norga points to relatively moist conditions at 3.1kyr (Scholtz 1986).
2 kyr - present
There is good evidence for a number of environmental shifts during the last 2000 years, but
the most noteworthy event may be the climatic ﬂuctuation during the Little Ice Age (LIA)
period. Evidence for a lower sea level on the southern coast during its earlier phase (1520
or 1570 AD) may be a reﬂection of widespread anomalous conditions during this phase
80 SCOTT AND LEE-THORP
High numbers of grass pollen in hyrax dung from the Namib Desert (Kuiseb River)
suggest relatively moist but ﬂuctuating conditions ∼2 kyr, while δ13C analysis points to
slightly higher C4dietary contributions (Scott and Vogel 2000). Lake-deposits with pollen
from Otjikoto further to the north in Namibia also seem to suggest moist conditions around
this time although there are some uncertainties about the dates (Scott et al. 1991). C4grass
contribution to diets of hyraxes in the Kuiseb River decreased soon after 2 kyr and by
ca. 0.9 kyr very little grass pollen is recorded in the dung sequence. The results show some
agreement with Makapansgat especially around the time of the LIA but the temperature
interpretations based on PCA curves of frost sensitive forms in the Kuiseb do not seem to
correspond well with regional interpretations extrapolated from Makapansgat and Cango.
Indeed sensitivity to temperature seems to vary widely across the subcontinent.
The highest Asteraceae values occur in the Wonderkrater pollen sequence at ∼2 kyr,
followed by a decline (Scott 1982a). This evidence continues the apparent contradictory
pattern between the pollen and Makapansgat δ13C values, which indicate high proportions
of C4grasses in the vegetation overburden. Both δ13C and δ18 O in the Makapansgat
stalagmites show a decreasing trend after 2 kyr, indicating a decline in C4grasses in step
with episodic sharp δ18O minima, suggesting that rain formed at markedly higher, cooler
altitudes, and in turn, that stormier and cooler conditions pertained (Holmgren et al. 1999).
The series of oscillations culminated in a marked δ18O minimum at AD 1750, which seems
to represent the regional manifestation of the LIA peak (Fig. 5). Corresponding variations in
pollen data include a decline of frost sensitive arboreal species in the bushveld at Moreletta
River (Pretoria) near the northern highveld boundary (Fig. 2) (Scott 1984) while the detailed
pollen diagram of Wonderkrater (Scott 1982a) suggests a similar decline in trees.
In the eastern summer-rain region in coastal KwaZulu-Natal, Podocarpus pollen in lake
and swamp deposits began to retreat northward to the Kosi Bay area after by 1.3kyr (Mazus
2000) suggesting increasing dryness.
At Florisbad, a phase with summer rain and strong evaporation indicated by prominent
grass and Chenopodiaceae pollen at ∼2 kyr (Scott and Nyakale 2002) is followed by moister
conditions by 1.5 kyr with higher grass and Cyperaceae pollen. Near Florisbad, in similar
grassland to that surrounding the site, the spring sequence at Deelpan (Meriba I), shows
a marked increase in halophytes indicating strong evaporation and warm temperatures
prior to ca. 0.7 kyr (Scott and Brink 1992). Both δ15 N and δ18O from ostrich eggshells in
Equus Cave suggest warm and arid, but ﬂuctuating, conditions from 1.5kyr (Johnson et al.
In the Karoo, dry conditions are indicated by presence of Asteraceae pollen and isotope
ratios in ostrich eggshell and soil humates from Blydefontein ca. 2 kyr (Scott and Bousman
1990; Bousman 1991). After 2kyr, grass pollen shows a recovery, conﬁrmed by isotope
ratios in ostrich eggshell and soil organics at Blydefontein (Bousman 1991). Pollen in hyrax
dung suggests that grassy vegetation occurred at Blydefontein ca. 900–1500AD (Bousman
and Scott 1994). Karoo shrubs increased at the cost of grass cover since ca. 1700 AD,
possibly as a result of generally decreasing summer rainfall receipts.
Slightly further south, a decline towards present levels of C4grass is registered in the
Cango stalagmite after 2 kyr. Abundance of a typical summer rainfall, frost resistant tree
species, Acacia karroo, increased in nearby Boomplaas Cave at 2 kyr (Deacon et al. 1984;
Scholtz 1986). Because of its position at the transition between the winter and summer rain
regimes, this area is highly sensitive to a decline in summer rainfall.
HOLOCENE CLIMATIC TRENDS AND RHYTHMS IN SOUTHERN AFRICA 81
Dietary δ13C values in hyrax dung from Pakhuis Pass (Cederberg) decline sharply
ca. 1–0.9 kyr (Scott and Vogel 2000), suggesting cooler winter-rain conditions. Following
a moister episode, δ18O values of ostrich eggshell in Elands Bay Cave register arid and/or
more evaporative conditions for most of the last 1000 years (Lee-Thorp and Talma 2000).
Evidence for sub-millennial variability
Long continuous records capable of yielding information on short term variability are ex-
tremely scarce. Spectral analysis of the Makapansgat δ18O series in the T7 stalagmite shows
signiﬁcant periodic components in this rainfall-linked proxy at 57 years (with associated
peaks at 53 and 65 years) reminiscent of the North Atlantic Oscillation, while the δ13C
and grey scale parameters peak at 540 and 330 years respectively (Lee-Thorp et al. 2001).
When these records are subjected to wavelet analysis, however, it can be seen that dominant
periodic components shifted over time (Tyson et al. 2002). Wavelet analysis of the δ18O
series shows periodicity at 500 years, and a cluster near 100 years between AD 0–1000,
which shifts to 70–80 years after AD 1000 (Tyson et al. 2002).
Pollen data did not produce the same high resolution or continuity as the stalagmites
but a 20th century pollen record from hyrax dung at Clarens suggests that this material
is capable of recording decadal variations (Carrion et al. 1999). Quasi-centennial scale
moisture cycles can be observed in some pollen records, for example, in the grass pollen
abundance in desert hyrax dung accumulations from the Kuiseb River (Scott 1996).
Mechanism of environmental change on different temporal scales
In order to understand the mechanism of climatic change over Southern Africa during the
Holocene we need to establish the rapidity of shifts in moisture and temperature patterns,
whether shifts appeared simultaneously over large parts of the continent, and how these
changes relate to events elsewhere. Spatially separated biomes can be expected to respond
in different ways to forcing of seasonal and moisture regimes, as well as to temperature.
The winter rainfall area stands in contrast with the northern, central and eastern parts
in view of more mesic conditions on the southwestern coastal region during the early
Holocene (Cartwright and Parkington 1997; Cowling et al. 1999), and drier conditions in
the mid-Holocene when conditions in the interior were more mesic.
On a broader scale the rapid event of ca. 8.2 kyr widely observed in North Atlantic
sediments and also in the northern and equatorial parts of Africa (Gasse 2000; Stager
and Mayewski 1997), can at present only be tentatively linked with any anomaly in the
Southern Hemisphere. A climate transition does seem to have occurred at 8 kyr, with a
marked temperature depression observed on the eastern highlands (Smith et al. 2002),
but whether or not this shift is linked to the 8.2 kyr “event”, or rather forms part of large
millennial-scale climate shifts is uncertain. A mechanism which could link the 8.2 kyr event
in the North Atlantic regions with a marked temperature depression in the Southern African
highlands is in any case, not clear.The general moisture pattern seems to show the expected
anti-phase response of monsoon-like (ITCZ) moisture distribution to the precessional cycle
(Street-Perrott and Perrott 1993).
82 SCOTT AND LEE-THORP
Two alternative explanations have been proposed for the development of moister con-
ditions during the middle Holocene in the central interior of Southern Africa during the
middle Holocene (Scott 1993). In one, higher incidence of summer rain spread gradually
from the savanna in the north to the Karoo in the south, based on the ﬁrst appearance of
early Holocene pollen moisture indicators. The second explanation points to moist pulses
as indicated by geomorphic evidence (Butzer 1984). The new Florisbad data support the
‘pulses’ proposal (Scott and Nyakale 2002), but the later arrival of wetness at Blydefontein
to the south perhaps supports the gradual spreading of moisture (Scott 1993). In view of the
high amplitude, rapid changes demonstrated in the Makapansgat stalagmites, “gradualism”
from pollen diagrams, or “pulses” from geomorphic evidence may merely be related to poor
chronological control and resolution. The stalagmite data provide the strongest evidence
for a pattern of extremely rapid changes — on the order of a couple of decades or less
in many cases (Lee-Thorp et al. 2001). This rapidity and scale suggests that apparently
long-term climate or environmental states noted from archaeological, geomorphological,
and palynological evidence may be crude averages of a variety of conditions.
Periodic elements evident in the Makapansgat data differ for all three variables of δ18O,
δ13C and grey scale, suggesting separate inﬂuences on their respective time evolution (Lee-
Thorp et al. 2001). Different processes associated with moisture cycles on centennial scales
and millennial scales (moisture, seasonality, or temperature-dependent evapo-transpiration
rates) are likely superimposed on each other. The long-term pattern seems to be reﬂected in
the general δ13C curves from Makapansgat and Cango, and to a lesser degree in the δ18O
curves, while strong shorter-term oscillations superimposed on these trends are observed.
Given the constraints of chronological control, there is no discernible lag between the
patterns of the two δ13C curves, both reaching the highest values at ∼2.2 kyr (Figs. 5
and 6) (Talma and Vogel 1992; Lee-Thorp and Talma 2000). This observation argues for
coincident climate responses from north to south. The rise and recent partial decline in C4
grasses during the last 3 kyr may also be a seasonal feedback of incoming solar radiation,
induced by orbital forcing in the precessional cycle.
The evidence from the middle Holocene (ca. 6 kyr) suggests a scenario for dominant
inﬂuence from moist easterlies and reduced inﬂuence from the westerlies, and accords
well with warmer temperatures and a reduced Antarctic circumpolar vortex recorded in the
Taylor Dome ice core as higher δ18O (Fig. 7) and lower sodium concentrations respectively
(Steig et al. 2000). The regional vegetation pattern in the interior of Southern Africa over
the late Holocene between 5 and 2 kyr likely responded to generally rising summer season-
ality, promoting C4grasses, as reﬂected in stable carbon isotopes. The near-simultaneous
general increase in shrubby Asteraceous vegetation seems at ﬁrst glance to contradict this
scenario; but higher Asteraceae abundance can, apart from lower summer rain, also be
related to cooling. Hence, in combination the data may reﬂect interplay between cycles of
drying, cooling and more seasonally restricted rainfall. Tree-cover also declined markedly
according to pollen evidence from Wonderkrater and as implied by the stalagmite data.
This development might be related to increased incidence of frost (Scott 1982a; Scott and
Thackeray 1987; Lee-Thorp et al. 2001), or ﬁre, but ﬁre disturbance at Wonderkrater is
not suggested until recently (Scott 2002). In the case of the karroid Cango region where
no savanna trees are present, a similar argument can be advanced if mountain scrub is
considered instead. An increase in charcoal of a typically summer rainfall and relatively
frost resistant species, Acacia karroo, in nearby Boomplaas Cave at 2 kyr (Deacon et al.
HOLOCENE CLIMATIC TRENDS AND RHYTHMS IN SOUTHERN AFRICA 83
1984; Scholtz 1986) may reﬂect a replacement of mountain scrub as ﬁrewood, by valley
material. More open Asteraceous vegetation in the Western Cape (Scott 1994; Cartwright
and Parkington 1997; Cowling et al. 1999) would support a regional pattern of more open
Figure 7. Makapansgat stalagmite δ18O compared with δ18 O values in the Taylor Dome (Antarctica) and GISP
II (Greenland) ice core (Grootes et al. 1993; Steig et al. 2000), all plotted with a Gaussian low-pass ﬁlter (∼100
year running mean). (From Lee-Thorp et al. (2001)).
A comparison of the Makapansgat δ18O records with the near-coastal Antarctic Taylor
Dome ice core δ18O, sodium (Na) and other trace element concentration records (Steig
et al. 2000) indicates strong correspondences despite the large distance between them
(Fig. 1). Warming (18 O enrichment) at Taylor Dome prior to ∼5000 years is matched
at Makapansgat (Fig. 7). A weakened Antarctic circumpolar vortex is indicated by the
lowest sodium concentrations for the entire Holocene in Taylor Dome at this time (Steig
et al. 2000). Concurrent reduction in wind-driven mixing in Lake Victoria on the equator
was used to infer reduced inﬂuence of the circumpolar vortex far to the north (Stager
and Mayewski 1997). Warming also occurs in Taylor Dome (indicated as higher δ18O) at
about the same time as higher δ18O at Makapansgat during a variable, modest warm/humid
phase at 4.3–3.2 kyr. A shift towards lower δ18O in Taylor Dome occurs at ∼3 kyr as
seen in Makapansgat, while both records show declining δ18O trends thereafter, observed
also in the Greenland GISP 2 ice core (Grootes et al. 1993) (Fig. 7). The drier, cooler
conditions inferred from the Makapansgat stalagmite record are reﬂected in various more
poorly resolved proxy data across Southern Africa (Tyson and Lindesay 1992). Moreover
higher Na levels in Taylor Dome indicate enhanced Antarctic circumpolar atmospheric
circulation at this time. Elsewhere in Africa, trends towards aridiﬁcation are observed in
African lake levels including Lake Victoria, and Lake Malawi fell shortly after 1.5 kyr
(Stager and Mayewski 1997; Johnson et al. 2001).
Taylor Dome -38
Age (cal. yr. BP)
δ18O (‰PDB)δ18O (‰SMOW)
84 SCOTT AND LEE-THORP
Some correlation between major cultural manifestations and climatic or environmental
changes can be drawn, but the nature of these relationships is complex and difﬁcult to
determine, and very often several explanations for cultural shifts, including social causes,
must be considered. Farmers who arrived shortly after 2kyr with their crops and domestic
animals and their attendant requirements for water and food probably had fewer mechanisms
for dealing with variable conditions than the highly ﬂexible hunter-gatherer societies before
Deacon (1974) ﬁrst observed that archaeological visibility of Later Stone Age people
in Southern Africa might be related in some way to climatic conditions as illustrated by a
comparison of frequency of radiocarbon dates associated with archaeological ﬁnds in the
interior. In the earlier part of the Holocene, Oakhurst Complex stone tools of the Later
Stone Age appeared over wide parts of Southern Africa (Mitchell 2000). Development of
ﬂuctuating and often arid conditions during this phase presented challenges for hunter-
gatherers, which may be at least partly reﬂected in the low density of sites occupied, and
the nature of resource exploitation and of stone tool manufacture. The lowest frequency
of dates for the sub-continent as a whole occurs at ca. 8.7 kyr during the early Holocene
dry phase. A recent update of the pattern shows a more complex picture but supports the
pattern (Mitchell 2000). One of the shifts was in hunting strategy: small solitary bovids
were taken instead of the large, mobile grazers typical of the Pleistocene (Deacon et al.
1984). Assemblages of the Oakhurst Complex disappeared over a wide region in South
Africa ca. 8.7 kyr.
Expansion of microlithic Wilton-type industries after ca. 7.5kyr (Mitchell 2000) took
place in conditions of increasing moisture availability. Deacon’s (1974) radiocarbon fre-
quencies show that Later Stone Age sites, and hence archaeological ‘visibility’ of hunter-
gatherers in the interior landscape, peaked at ∼3.1 kyr. In the Eastern Cape, popula-
tion densities begin to rise after 6.8 kyr, the range of sites occupied had increased after
∼5 kyr, and between 5 and 2 kyr resource exploitation intensiﬁed, including a concen-
tration on riverine resources (Hall 2000). This evidence, together, suggests a signiﬁcant
increase in population density of hunter gatherers. It appears that the slightly cooler
conditions pertaining by this time were favourable for people over wide parts of the
The pattern differs in the Western Cape where archaeological visibility of coastal hunter-
gatherers was low between ∼7.8 and 4.2 kyr. The possibility has long been entertained that
marked regional aridity rendered the Western Cape region unattractive to hunter-gatherers
(Parkington et al. 1987), although other factors including changes in rocky shorelines
associated with a higher sea-level at this time, might also have made it less attractive (Miller
et al. 1993; Jerardino and Yates 1996). New δ18O mollusc data from the Steenbokfontein
Cave, where human occupation was more continuous, suggest that sea-surface temperatures
were not different from those before or after the occupation ‘hiatus’, implying that coastal
Benguela upwelling, usually associated with coastal aridity, was not unusually high at
this time (A. Jerardino, pers. comm.). The period between 4 and 2 kyr, is known as the
“megamidden” period because of the accumulation of enormous shell middens along the
coast (Parkington et al. 1987). A very high proportion of the human burials found in coastal
caves, shelters or simply along the dune cordon, have been assigned to this period based
HOLOCENE CLIMATIC TRENDS AND RHYTHMS IN SOUTHERN AFRICA 85
on their radiocarbon dates (Parkington et al. (1987), Lee-Thorp et al. (1989), Sealy, pers.
Agriculture arrived relatively late in Southern Africa, shortly after 2 kyr, and therefore
its impact has generally not been as prolonged as in other African regions to the north.
The Iron Age farmers who penetrated to the eastern and central regions of the subcontinent
manufactured iron implements and brought the C4crops sorghum and millet, as well as
domestic sheep and later, cattle. These cereals require summer rain. Climate and environ-
mental patterns over the last 2 kyr may be compared with a relative wealth of archaeological
data from areas occupied by these semi-sedentary farmers. Agriculture based on C4crops
was never possible in the drier, westerly winter rainfall regions. Transhumant Khoisan
pastoralists with sheep and later cattle entered the region by about 2 kyr (Sealy and Yates
1994), probably when conditions were relatively moister and grassier (Scott et al. 1991;
Scott 1996; Bousman 1998).
A marked increase in population size and social complexity amongst the farming and
iron-producing groups took place in and around the Shashe-Limpopo region beginning
about AD 800. Based on the temporal distribution of radiocarbon dates for IronAge groups
(Vogel 1995), Huffman (1996) proposed that populations expanded during warmer, moister
phases when crop growing and cattle farming ﬂourished and that they declined or moved
away during drier periods. Social complexity increased rapidly leading to the development
of stratiﬁed societies and large towns most apparent between AD 1000–1270 at the K2 and
Mapungubwe sites (Vogel 1995; Huffman 1996). Since this area currently experiences very
low (<330 mm/yr) rainfall with a high co-efﬁcient of variation less suitable for maintenance
of large domestic herds or cultivation, it is widely considered that the scale of settlement,
and development of wealth, was made possible by more favourable conditions existing
at the time (Huffman 1996; Tyson et al. 2000). The recent data from Makapansgat and
Wonderkrater to the southeast conﬁrm that between AD 1000 and 1300 regional conditions
were at times slightly moister and warmer than today, although strictly speaking they fall
within a general trend of continental drying (Scott and Thackeray 1987; Holmgren et al.
1999; Lee-Thorp et al. 2001). Nevertheless, even slight increases in moisture, or perhaps
greater predictability, would have promoted a suitable environment for herding of cattle and
sheep, because of the predominance of ‘sweet’ grasses (grasses which remain nutritious
throughout the year). An ‘oversupply’ of rain, although this might seem to be advantageous
in a generally arid region, brings disadvantages in the form of less nutritious grasses, and
worse, diseases such as malaria and tsetse ﬂy.
The subsequent collapse of Mapungubwe at AD 1280, apparent dispersal of the pop-
ulation, and subsequent rise to power of Great Zimbabwe further north, has given rise
to speculation about a connection to worsening conditions associated with emergence
of “Little Ice Age” conditions (Huffman 1996). This scenario now seems unlikely for
several reasons. Firstly, the ﬁrm Makapansgat chronology shows that marked dry, cool
“LIA” conditions emerged only later, about AD 1600 (Lee-Thorp et al. 2001). More recent
excavations do not suggest a population disappearance, merely the demise of Mapungubwe
itself, and some sites post-dating AD 1600 appear in low-lying seasonally inundated areas,
suggesting that they were clustered close to remaining pans and water sources (S. Hall,
pers. comm.). Finally, new nitrogen isotope data from a number of sites suggest that no
climate deterioration took place simultaneously with the collapse of Mapungubwe, rather
drier conditions emerged much later (J.M. Smith, pers. comm).
86 SCOTT AND LEE-THORP
One other relationship remains to be investigated, that is, the effect of intense farming on
a landscape which is often marginal for dryland agriculture and introduced non-indigenous
animals. Pollen preserved in cow dung from Iron Age “kraals” (cattle enclosures) over the
last 1500 years suggests that environments were grassier and more open than the savanna
environments of today (Carrion et al. 2000). The openness wasprobably due to local clearing
of wood near settlements for domestic and manufacturing (iron smelting) fuel, while the
regional woodland was naturally more open than at present.A study of microscopic charcoal
as an indicator of ﬁre in swamp and lake sediments of the savanna, does not, with the
exception of the Wonderkrater site, suggest that the arrival of farmers coincided with higher
ﬁre incidences (Scott 2002). Controlled use of ﬁre and protection of grassland for grazing
and removal of fuel probably account for high grass pollen and low microscopic charcoal
Large parts of the subcontinent remain unexplored with regard to palaeoenvironments and
records are scattered and often incomplete. A great deal more research is therefore needed
to improve our current understanding of Holocene change. Correspondence between rare
available high resolution continental climate data with a near-coastal Antarctic ice core
suggest that the climate of the subcontinent continued to be strongly inﬂuenced by shifts in
intensity and extent of frontal systems associated with the circumpolar vortex. This major
forcing probably holds an important key to synchroneity with global climate trends and
events observed for Southern African climates.
Southward spreading of summer-rain moisture related to the ITCZ during the middle
and late Holocene was characterised by strong centennial scale variability over different
regions of the interior. On a millennial scale the link between moisture and temperature
conditions seems less certain, but at smaller centennial scale variability this link is observed.
A picture is emerging that shows that biota and human settlements across the subcontinent
were strongly inﬂuenced by the relatively mild (compared to Glacial period, Partridge et al.
(2004)) climate ﬂuctuations which occurred during the Holocene.
Although palaeoclimate records across Southern Africa are patchy in time and space, it
seems clear that Holocene climates and environments varied on scales that carried sig-
niﬁcant implications for distribution of fauna and ﬂora, and for human settlement. Large
millennial-scale temperature shifts of up to several degrees Celsius are indicated by changes
in proportions of C3and C4grasses in the sensitive eastern highlands until ∼7 kyr, and by
dissolved gases in the Uitenhage aquifer. In the north, Makapansgat stalagmite records in-
dicate shifts of similar duration and scale in addition to smaller, more rapid shifts.Visibility
of prehistoric human populations was lowest between 8 and 9 kyr. The warming trend from
∼7 kyr was accompanied by spreading summer moisture, culminating in a warm, wet phase
from 6.5–5.1 kyr according to stalagmite and pollen data. The Western Cape, in contrast,
seems to have been relatively arid. The suggestion is that the easterlies circulation system
dominated at this time and the westerlies system was reduced, concurrent with a reduced
HOLOCENE CLIMATIC TRENDS AND RHYTHMS IN SOUTHERN AFRICA 87
circumpolar vortex recorded in coastalAntarctica. Most records after 5 kyr reﬂect an overall
but variable drying trend. High proportions ofAsteraceae pollen are recorded in many sites
from ∼4.4–2 kyr. Stalagmite sequences record increasing development of C4grasslands,
peaking just before 2 kyr.A series of ∼1◦C temperature ﬂuctuations associated with drier,
stormier and cooler conditions culminated at 1750 AD. Climate variability as determined
from the Makapansgat stalagmites also shows periodic components at centennial and multi-
decadal scales. Shifts in intensity and extent of frontal systems associated with the Antarctic
circumpolar vortex likely hold an important key to synchroneity with global climate trends
We thank our colleagues in the Southern African Society for Quaternary Research and the
Quaternary Research Centre for their contributions. Tim Partridge suggested the title of this
paper.Anne Westoby did the original drafting of Figure 1. Jeannette Smith, Siep Talma and
Karin Homgren provided data for Figures 4, 6 and 7, respectively. Two referees, Raymonde
Bonneﬁlle and Lydie du Pont, provided constructive critiques. This work was funded by
the National Research Foundation, the University of the Free State (GUN 2053236), the
University of Cape Town (GUN 2047166) and the Water Research Commission of South
Africa. We thank Françoise Gasse and Rick Battarbee for their efforts in leading the PEP
III project and for organizing an excellent conference in Aix-en-Provence (August 2001).
Avery D.M. 1981. Holocene micromammalian faunas from the Northern Cape, South Africa. S. Afr.
J. Sci. 77: 265–73.
Bousman C.B. 1991. Holocene palaeoecocology and Later Stone Age hunter gatherer adaptations in
the South African interior plateau. Ph.D. thesis, Southern Methodist University, Dallas, Texas,
Bousman C.B. 1998. The chronological evidence for the introduction of domestic stock into Southern
Africa. African Archaeological Review 15: 133–150.
Bousman C.B. and Scott L. 1994. Climate or overgrazing?: the palynological evidence for vegetation
change in the eastern Karoo. S. Afr. J. Sci.: 575–578.
Brook G.A., Marais E., Cowart J.B. 1999. Evidence of wetter and drier conditions in Namibia from
tufas and submerged speleothems. Cimbebasia 15: 29–39.
Butzer K.W. 1984. Archaeology and Quaternary environment in the interior of Southern Africa. In:
Klein R.G. (ed.), Southern African Prehistory and Palaeoenvironments. Balkema, Rotterdam, pp.
Carrion J.S., Scott L. and Vogel J.C. 1999. Twentieth century changes in montane vegetation in the
Eastern Free State, South Africa, derived from palynology of hyrax middens. J. Quat. Sci. 14:
Carrion J.S., Scott L., Huffman T. and Dryer C. 2000. Pollen analysis of Iron Age cow dung in
Southern Africa. Veg. Hist. Archaeobot. 9: 239–249.
Cartwright C. and Parkington J. 1997. The wood charcoal assemblages from Elands Bay Cave,
Soutwestern Cape: principles, procedures and preliminary interpretation. S. Afr. Archaeol. Bull.
88 SCOTT AND LEE-THORP
Cohen A.L., Parkington J.E., Brundrit G.B. and van der Merwe N.J. 1992. A Holocene sea surface
temperature record in mollusc shells from the southwest African coast. Quat. Res. 38: 379–385.
Cohen A. and Tyson P.D. 1995. Sea surface temperatures during the Holocene on the south coast of
Africa. The Holocene 5: 304–312.
Cowling R.M. 1983. The occurrence of C3and C4grasses in fynbos and allied shrublands in the
southeastern Cape, South Africa. Oecologia 58: 121–127.
Cowling R.M., Cartwright C.R., Parkington J.E. and Allsop J.C. 1999. Fossil wood charcoal assem-
blages from Elands Bay Cave, South Africa: implications for Late Quaternary vegetation and
climates in the winter-rainfall fynbos biome. J. Biogeogr. 26: 367–378.
Cowling R.M., Richardson D.M. and Pierce S.M. 1997. Vegetation of Southern Africa. Cambridge
University Press, Cambridge, 615 pp.
Deacon J. 1974. Patterning in the radiocarbon dates for the Wilton/Smithﬁeld Complex in Southern
Africa. S. Afr. Archaeol. Bull. 29: 3–18.
Deacon H.J., Deacon J., Scholtz A., Thackeray J.F., Brink J.S. and Vogel J.C. 1984. Correlation of
palaeoenvironmental data from the late Pleistocene and Holocene deposits at Boomplaas Cave,
Southern Cape. In: Vogel J.C. (ed.), Late Cainozoic Palaeoclimates of the Southern Hemisphere.
Balkema, Rotterdam, pp. 339–352.
Eitel B., Blümel W.D., Hüser K. and Mauz B. 2001. Dust and loessic alluvial deposits in Northwest-
ern Namibia (Damaraland, Kaokoveld): sedimentological and palaeoclimatic evidence based on
luminescence data. Quat. Int. 57: 57–65.
Esterhuysen A.B., Mitchell P. and Thackeray J.F. 1999. Climatic change across the Pleis-
tocene/Holocene boundary in the Caledon River Southern Africa: Results of a factor analysis
of charcoal assemblages. S. Afr. Field Archaeol. 8: 28–34.
February E. 1994. Rainfall reconstruction using wood charcoal from two archaeological sites in South
Africa, Quat. Res. 42: 100–107.
Finney B.P., Scholz C.A., Johnson T.C., Trumbore S. and Southon J. 1996. Late Quaternary lake-level
changes in Lake Malawi. In: Johnson T.C. and Odada E. (eds), Climatology and Palaeoclimatology
of East African Lakes. Gordon and Beach, Toronto, pp. 495–508.
Gasse F. 2000. Hydrological changes in the African tropics since the Last Glacial Maximum. Quat.
Sci. Rev. 19: 189–211.
Grootes P.M., Stuiver M., White J.W.C., Johnsen S. and Jouzel J. 1993. Comparison of oxygen isotope
records from the GISP2 and GRIP Greenland ice cores. Nature 366: 552–554.
Hall S. 2000. Burial sequences in the Later Stone Age of the Eastern Cape Province, South Africa.
S. Afr. Archaeol. Bull. 55: 137–146.
Heaton T.H.E., TalmaA.S. and Vogel J.C. 1986. Dissolved gas palaeotemperatures and 18 O variations
derived from groundwater near Uitenhage, South Africa. Quat. Res. 25: 79–88.
Holmgren K., Karlén W., Lauritzen S.E., Lee-Thorp J.A., Partridge T.C., Piketh S., Repinski P.,
Stevenson J., Svanered O. and Tyson P.D. 1999. A 3000-year high-resolution stalagmite-based
record of palaeoclimate for North-Eastern South Africa. The Holocene 9: 295–309.
Holmgren K., Lee-Thorp J.A., Cooper G.R.J., Lundblad K., Partridge T.C., Scott L., Sithaldeen R.,
TalmaA.S. and Tyson P.D. 2003. Persistent millennial-scale variability over the past 25,000 years
in Southern Africa. Quat. Sci. Rev. 22: 2311–2326.
Horowitz A., Sampson C.G., Scott L. and Vogel J.C. 1978. Analysis of the Voigspost site, O.F.S.,
South Africa. S. Afr. Archaeol. Bull. 33: 152–9.
Huffman T. 1996. Archaeological evidence for climatic change during the last 2000 years in Southern
Africa. Quat. Int. 33: 55–60.
Jerardino A. 1995. Late Holocene neoglacial episodes in Southern South America and Southern
Africa: a comparison. The Holocene 5: 361–368.
Jerardino A. and Yates R. 1996. Preliminary results from excavations at Steenbokfontein Cave:
implications for past and future research. S. Afr. Archaeol. Bull. 51: 7–16.
HOLOCENE CLIMATIC TRENDS AND RHYTHMS IN SOUTHERN AFRICA 89
Johnson B.J., Miller G.H., Fogel M.L. and Beaumont P.B. 1997. The determination of Late Qua-
ternary palaeoenvironments at Equus Cave, South Africa, using stable isotopes and amino acid
racemization in ostrich eggshell. Palaeogeogr. Palaeoclim. Palaeoecol. 136: 121–137.
Johnson T.C., Barry Y., Chan Y. and Wilkinson P. 2001. Decadal record of climate variability spanning
the past 700 yr in the Southern Tropics of East Africa. Geology 29: 83–86.
Lee-Thorp J.A., Sealy J.C. and van der Merwe N.J. 1989. Stable carbon isotope ratio differences
between bone collagen and bone apatite, and their relationship to diet. J. Arch. Sci. 16: 585–599.
Lee-Thorp J.A. and Beaumont P.B. 1995. Vegetation and seasonality shifts during the Late Quaternary
deduced from 13C/12 C ratios of grazers at Equus Cave, South Africa. Quat. Res. 43: 426–432.
Lee-Thorp J.A. and Talma A.S. 2000. Stable light isotopes and past environments in the Southern
African Quaternary. In: Partridge T.C. and Maud R.R. (eds), The Cenozoic of Southern Africa.
Oxford Monographs on Geology and Geophysics 40, Oxford University Press, New York, pp.
Lee-Thorp J.A., Holmgren K., Lauritzen S.-E., Linge H., Moberg A., Partridge T.C., Stevenson C.
and Tyson P.D. 2001. Rapid climate shifts in the Southern African interior throughout the mid to
late Holocene, Geophys. Res. Lett. 28 (23): 4507–4510.
Lutjeharms J.R.E., Monteiro P.M.S., Tyson P.D. and Obura D. 2001. The oceans around Southern
Africa and regional effects of climate change. S. Afr. J. Sci. 97: 119–130.
Marker M.E. 1997. Evidence for a Holocene low sea-level at Knysna. S. Afr. Geogr. J. Special Edition:
Martin A.R.H. 1968. Pollen analysis of Groenvlei lake sediments, Knysna (South Africa). Rev.
Palaeobot. Palynol. 7: 107–144.
Mazus H. 2000. Clues on the history of Podocarpus forest in Maputaland, South Africa, during the
Quaternary, based on pollen analysis. African Geoscience Rev. 7: 75–82.
Meadows M.E., Baxter A.J. and Parkington J.E. 1996. Late Holocene environments at Verlorenvlei,
Western Cape Province, South Africa. Quat. Int. 33: 81–95.
Meadows M.E. and Sugden J.M. 1991. AVegetation history of the last 14 000 years on the Cederberg,
South-Western Cape Province. S. Afr. J. Sci. 87: 34–43.
Miller D.E., Yates R.J., Parkington J.E. and Vogel J.C. 1993. Radiocarbon-dated evidence relating to
a mid-Holocene relative high sea level on the South-Western Cape coast, South Africa. S. Afr. J.
Sci. 89: 35–44.
Mitchell P. 2000. The Quaternary Archaeological record in Southern Africa. In: Partridge T.C.
and Maud R.R. (eds), The Cenozoic of Southern Africa. Oxford Monographs on Geology and
Geophysics 40, Oxford University Press, NewYork, pp. 357–370.
Partridge T.C., Scott L. and Schneider R.R. 2004., this volume. Between Agulhas and Benguela:
responses of Southern African climates of the Late Pleistocene to current ﬂuxes, orbital precession
and extent of the Circum-Antarctic vortex. In: Battarbee R.W., Gasse F. and Stickley C.E. (eds),
Past Climate Variability through Europe and Africa. Kluwer Academic Publishers, Dordrecht, the
Netherlands, pp. 45–68.
Parkington J.E., Poggenpoel C., Buchanan W., Robey T., Manhire A. and Sealy J.C. 1987. Holocene
coastal settlement patterns in the Western Cape. In: Bailey G.N. and Parkignton J.E. (eds), The
Archaeology of Prehistoric Coastlines. Cambridge University Press, Cambridge, pp. 22–41.
Ramsay P.J. 1995. 9000 years of sea-level change along the Southern African coastline. Quat. Int.
Scholtz A. 1986. Palynological and palaeobotanical studies in the Southern Cape. M.Sc.-thesis,
University of Stellenbosch, Stellenbosch, 280 pp.
Scott L. 1982a. A Late Quaternary pollen record from the Transvaal bushveld, South Africa. Quat.
Res. 17: 339–370.
Scott L. 1982b. A 5000-year old pollen record from spring deposits in the bushveld at the north of
the Soutpansberg, South Africa. Palaeoecol. Afr. 14: 45–55.
90 SCOTT AND LEE-THORP
Scott L. 1984. Reconstruction of Late Quaternary palaeoenvironments in the Transvaal region, South
Africa, on the basis of palynological evidence. In: Vogel J.C. (ed.), Late Cainozoic Palaeoclimates
of the Southern Hemisphere. Balkema, Rotterdam, pp. 317–327.
Scott L. 1993. Palynological evidence for Late Quaternary warming episodes in Southern Africa,
Palaeogeogr. Palaeoclim. Palaeoecol. 101: 229–235.
Scott L. 1994. Palynology of late Pleistocene hyrax middens, South-Western Cape Province, South
Africa: a preliminary report. Hist. Biol. 9: 71–81.
Scott L. 1996. Palynology of hyrax middens: 2000 years of palaoenvironmental history in Namibia
Quat. Int. 33: 73–79.
Scott L. 1999. The vegetation history and climate in the Savanna Biome, South Africa, since 190 000
KA: A comparison of pollen data from theTswaing Crater (the Pretoria Saltpan) and Wonderkrater.
Quat. Int. 57-58: 215–223.
Scott L. 2002. Microscopic charcoal in sediments: Quaternary ﬁre history of the grassland and savanna
regions in South Africa. J. Quat. Sci. 17: 77–86.
Scott L. and Bousman C.B. 1990. Palynological analysis of hyrax middens from Southern Africa.
Palaeogeogr. Palaeoclim. Palaeoecol. 79: 367–379.
Scott L. and Brink J.S. 1992. Quaternary palaeoenvironments of pans in Central South Africa:
palynological and palaeontological evidence. S. Afr. Geographer 19: 22–34.
Scott L., Cooremans B., de Wet J.S. and Vogel J.C. 1991. Holocene environmental changes in Namibia
inferred from pollen analysis of swamp and lake deposits. The Holocene 1: 8–13.
Scott L., Holmgren K., Talma A.S., Woodborne S. and Vogel J.C. 2003. Age interpretation of the
Wonderkrater spring sediments and vegetation change in the savanna biome, Limpopo Province,
South Africa. S.A.J. Sci. 99: 484–488.
Scott L. and Nyakale M. 2002. Pollen indications of Holocene palaeoenvironments at Florisbad in
the Central Free State, South Africa. The Holocene 14: 497–503.
Scott L. and Steenkamp M. 1996. Environmental history and recent human disturbance at coastal
Lake Teza, Kwazulu/Natal. S. Afr. J. Sci. 92: 348–350.
Scott L., Steenkamp M. and Beaumont P.B. 1995. Palaeoenvironmental conditions in South Africa
at the Pleistocene-Holocene transition. Quat. Sci. Rev. 14: 937–94.
Scott L. and Thackeray J.F. 1987. Multivariate analysis of Late Pleistocene and Holocene pollen
spectra from Wonderkrater, Transvaal, S. Africa. S. Afr. J. Sci. 83: 93–98.
Scott L. and Vogel J.C. 1983. Late Quaternary pollen proﬁle from the Transvaal highveld, South
Africa. S. Afr. J. Sci. 79: 266–272.
Scott L. and Vogel J.C. 2000. Evidence for environmental conditions during the last 20,000 years in
Southern Africa from 13C in fossil hyrax dung. Global Planet. Change 26: 207–215.
Sealy J.C. 1997. Seasonality of rainfall around the Last Glacial Maximum as reconstructed from
carbon isotope analyses of animal bones from Nelson Bay Cave. S. Afr. J. Sci. 92: 441–444.
Sealy J.C. and Yates R. 1994. The chronology and introduction of pastoralism to the Cape, South
Africa. Antquity 68: 58–67.
Shi N., Dupont L.M., Beug H.-J., Schneider R. 1998. Vegetation and climate changes during the last
21 000 years in SW Africa based on a marine pollen record. Veg. Hist. Archaeobot. 7: 127–140.
Smith J.M. 1997. Stable isotope analysis of fauna and soils from sites in the Eastern Free State
and Western Lesotho, Southern Africa: a palaeoenvironmental interpretation. M.Sc. thesis, UCT,
Smith J.M., Lee-Thorp J.A. and Sealy J.C. 2002. Stable carbon and oxygen isotopic evidence for
late Pleistocene and early — middle Holocene climatic ﬂuctuations in the Caledon River Valley,
Southern Africa. J. Quat. Sci. 17: 683–695.
Stager J.C. and Mayewski P.A. 1997. Abrupt Early to Mid-Holocene climatic transition registered at
the Equator and poles. Science 276: 1834–1836.
HOLOCENE CLIMATIC TRENDS AND RHYTHMS IN SOUTHERN AFRICA 91
Steig E.J., Morse D.L., Waddington E.D., Stuiver M., Grootes P.M., Mayewski P.A., Twickler M.S.
and Whitlow S.I. 2000. Wisconsinian and Holocene climate history from an ice core at Taylor
Dome, Western Ross Embayment, Antarctica. Geogr. Ann. A 82: 213–235.
Street-Perrott A.F. and Perrott R.A. 1993. Holocene vegetation, lake levels, and climate of Africa. In:
Wright H.E., Kutzbach J.E., Webb T., Ruddiman W.E., Street-Perrott F.A. and Bartlein P.J. (eds),
Global Climates since the Last Glacial Maximum. University of Minnesota Press, Minneapolis,
TalmaA.S. and Vogel J.C. 1992. Late Quaternary paleotemperatures derived from a speleothem from
Cango Caves, Cape Province, South Africa. Quat. Res. 37, 203–213.
Thackeray J.F. and Lee-Thorp J.A. 1992. Isotopic analysis of Equid teeth from Wonderwerk Cave,
northern Cape Province, South Africa. Palaeogeogr. Palaeoclim. Palaeoecol. 99: 141–150.
Tusenius M.L. 1989. Charcoal analytical studies in the North-Eastern Cape, South Africa. S. Afr.
Archaeol. Soc. Goodwin Series 6: 77–83.
Tyson P.D., Cooper G.R.J. and McCarthy T.S. 2002. Millennial to multi-decadal variability in the
climate of Southern Africa. Int. J. Climatol. 22: 1105–1117.
Tyson P.D. and Lindesay J.A. 1992. The climate of the last 2000 years in Southern Africa. The
Holocene 2, 3: 271–278.
Tyson P.D., Karlen W., Holmgren K. and Heiss G.A. 2000. The Little Ice Age and medieval warming
in South Africa. S. Afr. J. Sci. 96: 121–126.
Visser J.N.J. and JoubertA. 1991. Cyclicity in the late Pleistocene to Holocene and lacustrine deposits
at Florisbad, Orange Free State. S. Afr. J. Geol. 94: 123–131.
Vogel J.C., Fuls A. and Ellis R.P. 1978. The geographical distribution of Kranz Grasses in South
Africa. S. Afr. J. Sci. 74: 209–215.
Vogel J.C. 1995. The temporal distribution of radiocarbon dates for the Iron Age in Southern Africa.
S. Afr. Archaeol. Bull. 50: 106–109.