Holocene vegetation and climate variability in the winter and summer rainfall zones of South Africa

Abstract

To better understand Holocene vegetation and hydrological changes in South Africa, we analyzed pollen and microcharcoal records of two marine sites GeoB8331 and GeoB8323 from the Namaqualand mudbelt offshore the west coast of South Africa covering the last 9900 and 2200 years, respectively. Our data corroborate findings from literature that climate developments apparently contrast between the summer rainfall zone (SRZ) and winter rainfall zone (WRZ) over the last 9900 years, especially during the early and middle Holocene. During the early Holocene (9900–7800 cal. yr BP), a minimum of grass pollen suggests low summer rainfall in the SRZ, and the initial presence of Renosterveld vegetation indicates relatively wet conditions in the WRZ. Toward the middle Holocene (7800–2400 cal. yr BP), a rather moist savanna/grassland rich in grasses suggests higher summer rainfall in the SRZ resulting from increased austral summer insolation and a decline of fynbos vegetation accompanied by an increasing Succulent Karoo vegetation in the WRZ, which possibly suggests a southward shift of the Southern Hemisphere westerlies. During the last 2200 years, a trend toward higher aridity was observed for the SRZ, while the climate in the WRZ remained relatively stable. The ‘Little Ice Age’ (ca. 700–200 cal. yr BP) was rather cool in both rainfall zones and drier in the SRZ while it was wetter in the WRZ.

Keywords

Holocene marine sediments microcharcoal pollen South Africa vegetation change

Introduction

South Africa, located at the interface between subtropical and warm-temperate climate zones and between the Indian and Atlantic oceans, is a critical region for Quaternary environmental research and understanding of Holocene climatic history in the Southern Hemisphere (Chase and Meadows, 2007; Scott et al., 2012). The specific atmospheric and oceanic circulation systems across the region (Shannon and Nelson, 1996; Tyson and Preston-Whyte, 2000) shape the precipitation patterns of southern Africa where three rainfall zones are identified (Figure 1): a summer rainfall zone (SRZ) in the north and east of southern Africa, a winter rainfall zone (WRZ) at the southwestern tip of the continent extending northward from the Cape Peninsula along the west coast to about 28°S, and a transitional year-round rainfall zone (YRZ) between these two regions (Chase and Meadows, 2007; Tyson and Preston-Whyte, 2000). The change in vegetation from the northeastern SRZ to southwestern WRZ along the gradient of rainfall seasonality is expected to be very sensitive to climate change.

Figure 1.

Modern atmospheric and oceanic circulation systems, as well as mean monthly precipitation for the period 1981–2010 (http://www.esrl.noaa.gov/psd/data/gridded/tables/precipitation.html) over southern Africa during austral (a) summer and (b) winter. Major representations of atmospheric and oceanic circulation, South Atlantic Anticyclone (SAA), South Indian Anticyclone (SIA), Benguela Current (BC), and Agulhas Current (AgC) are shown (Tyson and Preston-Whyte, 2000). Three rainfall zones of southern Africa are indicated: summer rainfall zone (SRZ), year-round rainfall zone (YRZ), and winter rainfall zone (WRZ). For the color figure, the reader is referred to the web version of this article.

Although South Africa is known to have experienced phases of significant climate change during the Quaternary, palaeoenvironmental evidence based on terrestrial records is often difficult to obtain due to the paucity of lake or swamp archives in which sedimentation is often discontinuous (Neumann et al., 2008, 2011), having weak age models and/or poor temporal resolution (Meadows et al., 2010) or only covering short time periods (Ekblom et al., 2012; Meadows et al., 1996). Longer, more continuous records (Figures 2 and 3) are widely spaced over South Africa and have been recovered from distinct local environments, which sometimes leads to contradictory results, whereby the use of different proxies impede comparison between them (Baker et al., 2014; Kristen et al., 2007; Norström et al., 2014). Nevertheless, a compilation of results from literature clearly indicates different developments in different regions during the Holocene.

Figure 2.

Map of southern Africa showing the location of study sites GeoB8331 (gravity core GeoB8331-4 and multicore GeoB8331-2) and GeoB8323 (gravity core GeoB8323-2 and multicore GeoB8323-1) (red stars), as well as gravity core GeoB8332-4 (dot) (Weldeab et al., 2013). Other terrestrial records mentioned in the text (crosses, triangles, and dots) are shown. The crosses denote the sites from Chase and Thomas (2006), the triangles denote the sites from Klein (1991), and the dots denote the sites from a wider area over South Africa (MFP = Mfabeni peatland, LE = Lake Eteza, VE = Verlorenvlei, KS = Klaarfontein Springs, PP = Pakhuis Pass, DV = Driehoek Vlei, DR = De Rif, TK = Truitjes Kraal, KP = Katbakkies Pass). The major biomes of South Africa (Mucina and Rutherford, 2006) are shown. For the color figure, the reader is referred to the web version of this article.

Figure 3.

Holocene hydrological records from South Africa discussed in the text. This is not an exhaustive data set but includes records covering the whole or at least most period of Holocene. SRZ: (1) Venda (Scott, 1987); (2) Makapansgat (Holmgren et al., 2003); (3) Wonderkrater 1 (Scott et al., 2012); (4) Wonderkrater 2 (Truc et al., 2013); (5) Tswaing Crater 1 (Kristen et al., 2007); (6) Tswaing Crater 2 (Metwally et al., 2014); (7) Wonderwerk (Brook et al., 2010); (8) Blydefontein (Scott et al., 2005); (9) Graskop & Versailles (Breman et al., 2012); (10) Braamhoek 1 (Norström et al., 2009); (11) Braamhoek 2 (Norström et al., 2014); (12) Mahwaqa (Neumann et al., 2014); (13) Mfabeni peatland (Baker et al., 2014); (14) Lake Eteza (Neumann et al., 2010). WRZ: (15) GeoB 8332-4 (Weldeab et al., 2013); (16) West coast (Chase and Thomas, 2006); (17) Western Cape (Klein, 1991); (18) Verlorenvlei (Meadows et al., 1994, 1996); (19) Klaarfontein Springs (Meadows and Baxter, 2001); (21) Pakhuis Pass (Scott and Woodborne, 2007b); (22) Driehoek Vlei (Meadows and Sugden, 1991); (23) De Rif 2010 (Valsecchi et al., 2013); (24) De Rif (Quick et al., 2011); (25) TK (Truitjes Kraal) (Meadows et al., 2010); (26) Katbakkies Pass (Chase et al., 2015). YRZ: (20) Groenvlei (Martin, 1968); (27) Seweweekspoort (Chase et al., 2013). For the color figure, the reader is referred to the web version of this article.

Within the WRZ, the palaeoenvironmental records are relatively consistent (Figure 3). The climate of the WRZ is characterized by moist conditions during the early Holocene (ca. 10,000–8000 cal. yr BP) (Chase and Thomas, 2006; Klein, 1991; Meadows and Sugden, 1991; Meadows et al., 2010; Quick et al., 2011; Weldeab et al., 2013), except for an arid period between 10,000–9000 cal. yr BP recorded at Pakhuis Pass turning to moist conditions after 9000 cal. yr BP and a brief dry period between 8500–8000 cal. yr BP recorded at De Rif 2010 (Valsecchi et al., 2013). Most studies found conditions gradually becoming drier during the middle Holocene (ca. 8000–3000 cal. yr BP) (Chase and Thomas, 2006; Martin, 1968; Meadows and Baxter, 2001; Meadows and Sugden, 1991; Meadows et al., 2010; Scott and Woodborne, 2007a, 2007b; Weldeab et al., 2013), while generally moist conditions were recorded in Seweweekspoort and Katbakkies Pass between 8000 and 1700 cal. yr BP interrupted by drier periods (8000–7300, 6600–6200 cal. yr BP, 5700–5200 cal. yr BP, and 3200–2700 cal. yr BP). Valsecchi et al. (2013) also found a brief moist period at De Rif 2010 between 7100 and 6700 cal. yr BP. During the late Holocene (3000 cal. yr BP–present), variable conditions were recorded: increasing humidity such as at Driehoek Vlei, Klaarfontein Spring, De Rif, Pakhuis Pass, and Seweweekspoort (Chase and Thomas, 2006; Chase et al., 2013; Klein, 1991; Meadows and Baxter, 2001; Meadows and Sugden, 1991; Scott and Woodborne, 2007b; Valsecchi et al., 2013) but aridity at Katbakkies Pass (Chase et al., 2015).

In the SRZ (Figure 3), two different palaeoenvironments developed from northeast to southeast of South Africa especially during the early Holocene (10,000–8000 cal. yr BP). Generally dry conditions characterized the northeast (savanna) region as recorded at Venda, Wonderkrater, Makapansgat, and Tswaing Crater (Holmgren et al., 2003; Kristen et al., 2007; Metwally et al., 2014; Scott, 1987; Scott et al., 2012; Truc et al., 2013). To the south, in central South Africa, studies from Wonderwerk and Blydefontein also indicate aridity (Brook et al., 2010; Scott et al., 2005), while at Braamhoek, the early Holocene started relatively moist turning more arid later (ca. after 9000 cal. yr BP) (Norström et al., 2009, 2014) compared with northeast South Africa. Along the east coast, moist conditions were recorded at Lake Eteza, Mahwaqa, and Mfabeni peatland between 10,000 and 8000 cal. yr BP (Baker et al., 2014; Neumann et al., 2010, 2014). During the middle Holocene, climate became variable, in particular, along the east coast and records of the same region such as Wonderkrater (Scott et al., 2012; Truc et al., 2013) and Tswaing Crater (Kristen et al., 2007; Metwally et al., 2014) show inconsistent results. However, generally an increase in humidity after 8000 cal. yr BP is indicated for the northeast region, whereas to the central South Africa, dry conditions persisted until ca. 4000 cal. yr BP. Different climate conditions in the two regions were also observed by Chevalier and Chase (2015), which was interpreted to be attributable to different modes of variability with the SRZ. The climate was variable and relatively moist during the late Holocene in both northeastern and southeastern South Africa except at Wonderkrater (Truc et al., 2013), Lake Eteza (Neumann et al., 2010), and Graskop and Versailles (Breman et al., 2012), where gradually drier conditions were recorded.

Comparing the Holocene terrestrial records, it becomes clear that the climate developed differently in the SRZ and WRZ. Hahn et al. (2015) found anti-phased climate variations during the last ca. 3000 years between the SRZ and WRZ, which were already proposed by Tyson (1986) and Cockcroft et al. (1987) based on contemporary climatic patterns. To better understand millennial to centennial mechanisms of regional vegetation and climate change in South Africa, more continuous and high-resolution environmental records are required. Here, marine sediments can offer continuous and well-dated archives. Sediment cores from the continental shelf with terrigenous material mainly from the adjacent continent areas have the potential of providing high-resolution Holocene records. Although the interpretation of marine pollen records is complex due to potentially extensive source areas and different modes of transportation and preservation, the pollen distribution in modern marine sediments of the Namaqualand mudbelt along the west coast of South Africa (Zhao et al., 2015) indicates distinctive pollen spectra reflecting vegetation communities on the adjacent continent and demonstrates that pollen records from marine sediment cores of the Namaqualand mudbelt have the potential to be a tool to reconstruct vegetation development in this region and better understand late Quaternary climate change.

In this study, pollen and microcharcoal records from two marine sites, namely, GeoB8331 and GeoB8323, retrieved from the Namaqualand mudbelt (Figure 2) offshore the west coast of South Africa were analyzed for the following objectives: (1) to reconstruct Holocene vegetation and climate changes in the SRZ and WRZ of South Africa, (2) to compare the regional climate developments during the late Holocene based on two marine records, and (3) to address the issue as to if and when the climate trends in the two rainfall zones developed in different directions. We aim at a better understanding of the spatial and temporal climate dynamics of South Africa during the Holocene associated to the position and intensity of atmospheric and oceanic circulation systems.

Regional setting

During the austral summer (Figure 1(a)), the rainfall in the SRZ is generated mainly by warm and moist easterly winds associated with the South Indian Anticyclone (SIA), and more than 66% of the annual precipitation falls between October and March. In the WRZ, precipitation is minimized in summer when the westerlies are located south due to alongshore winds resulting from a southern position of the South Atlantic Anticyclone (SAA). In contrast, during the austral winter (Figure 1(b)), dry conditions and high pressure prevail over the SRZ. The southern westerlies are at their northernmost position over the southwestern Cape during austral winter supplying rainfall to the WRZ, and more than 66% of the annual precipitation falls between April and September. Between these two regions, the YRZ receives both summer and winter rainfall all year through, where the summer rainfall progressively increases eastward along the south coast of South Africa (Tyson and Preston-Whyte, 2000).

There are nine biomes in southern Africa, which reveal a marked east to west gradient driven by rainfall amount and seasonality (Figure 2) (Bredenkamp et al., 1996; Cowling et al., 1997; Mucina and Rutherford, 2006; White, 1983). The Savanna Biome, which is the most extensive in southern Africa occupying 46% of its area, is characterized by a grassy ground layer and a distinct upper layer of woody elements associated with mean annual rainfall varying from 235 to 1000 mm/yr. The Grassland Biome, situated on the cooler and higher interior plateau, is dominated by a single layer of grasses, and trees are largely absent except in a few localized habitats. It spans a large rainfall gradient from 400 to 1200 mm/yr. The Forest Biome, which is restricted to areas with mean annual rainfall of more than 725 mm in the SRZ and more than 525 mm in the WRZ (Mucina and Rutherford, 2006) comprises mostly evergreen trees in multi-layered canopies, while the ground layer is often poorly developed because of the dense shade. The Nama Karoo Biome is a semi-desert grassy and dwarf shrubland on the central plateau and is dominated by members of the Asteraceae, Poaceae, Aizoaceae, Liliaceae, and Scrophulariaceae families. The rain in this biome falls mainly during austral summer varying between 100 and 520 mm/yr. The Succulent Karoo Biome, located in a narrow strip along the west coast and south of the escarpment, is characterized by dwarf leaf-succulents of which Aizoaceae (including Mesembryanthemaceae) and Crassulaceae are particularly prominent, where Asteraceae, Amaranthaceae, Euphorbiaceae (Euphorbia), and Zygophyllaceae (Zygophyllum) (Wheeler, 2010) are relatively common. Grasses are rare except in some sandy areas. This biome is primarily determined by low winter rainfall and extreme summer aridity with rainfall varying between 20 and 290 mm/yr. The Desert Biome, found under very harsh environmental conditions, is characterized by dominance of annual plants (often annual grasses like Stipagrostis sabulicola). The climate is characterized by occasional summer rainfall, but high levels of summer aridity with mean annual rainfall from approximately 10 mm in the west to 70 or 80 mm on the inland margin of the desert. The Fynbos Biome, which is an evergreen shrubland in the southwest Cape, is typified by the presence of Restionaceae, Ericaceae, and Proteaceae with rainfall usually varying from 600 to 800 mm/yr. Renosterveld is another type of shrubland occurring within the Fynbos Biome and is dominated by Asteraceae, in particular by one species – renosterbos (Elytropappus rhinocerotis), together with other important shrub families including Rubiaceae (Anthospermum), Thymelaeaceae (Passerina), Rosaceae (Cliffortia), Boraginaceae, Fabaceae, and Malvaceae (Goldblatt and Manning, 2002). The rainfall in the Renosterveld is between 250 and 600 mm/yr of which at least 30% falls during austral winter. Other two biomes (Albany Thicket Biome and Indian Ocean Coastal Belt) are located in the southeastern South Africa, which are not relevant for this study considering the pollen sources in the Namaqualand mudbelt (Zhao et al., 2015).

The Namaqualand mudbelt stretches over 500 km on the inner continental shelf along the western coast of southern Africa from 20 km north of the Orange River mouth to south of St Helena Bay (Rogers and Bremner, 1991). Sedimentation, source material, and texture of the mudbelt have been extensively discussed in several papers (Compton et al., 2010; Hahn et al., 2015; Mabote et al., 1997; Meadows et al., 1997, 2002; Rogers and Rau, 2006). The pollen distributions in modern marine sediments of the Namaqualand mudbelt along the west coast of South Africa indicate that vegetation communities on the adjacent continent are reflected in the marine pollen assemblages (Zhao et al., 2015). Three groups can be distinguished: Group 1 (Poaceae, Cyperaceae, Phragmites-type, Typha, Tribulus) consisting of pollen that is mainly contributed by the Orange River to the northern mudbelt, Group 2 (Restionaceae, Ericaceae, Anthospermum, Stoebe/Elytropappus-type, Cliffortia, Passerina, Artemisia-type, Pentzia-type) consisting of typical fynbos elements and dominating in the southern mudbelt, and Group 3 (Aizoaceae, Cheno/Am, Asteroideae) consisting of pollen contributed mainly from the nearshore Succulent Karoo and Nama Karoo vegetation by the so-called berg winds and local ephemeral Namaqualand rivers to the central mudbelt (Mabote et al., 1997).

Materials and methods

Two gravity cores, GeoB8331-4 (29°08.12′S, 16°42.99′E, 887 cm long, Figure 2) and GeoB8323-2 (32°01.70′S, 18°12.21′E, 285 cm long, Figure 2), were retrieved in January–February 2003 during Meteor cruise M57/1 (Schneider et al., 2003) at 97 m water depth from the northern Namaqualand mudbelt off the Holgat River (just south of the estuary of the Orange River) and at 92 m water depth from southern Namaqualand mudbelt off the Olifants River, respectively. The GeoB8331-4 sedimentary sequence consists of olive brown mud from 887 to 15 cm and dark laminated layers within olive brown mud above 15 cm. The GeoB8323-2 sedimentary sequence consists of dark greenish gray sandy mud with bilvalve shell fragments and shell layers from 182 to 178 cm. Additionally, considering the disturbance and compaction of the uppermost parts of the gravity cores during coring, two multicores, namely, GeoB8331-2 (10 samples, same site as GeoB8331-4) and GeoB8323-1 (12 samples, same site as GeoB8323-2), were processed to extend the coverage of both records toward the present.

A total of 65 samples were analyzed from GeoB8331-4 and GeoB8331-2, providing a high temporal resolution of ca. 170 years on average (range: 4–770 years). In all, 31 samples were analyzed from GeoB8323-2 and GeoB8323-1, providing an average resolution of ca. 110 years (range: 8–200 years). The samples were decalcified with diluted HCl (~12%) and two Lycopodium spore tablets (each containing 18,584 ± 372 markers) were added during the decalcification step. After washing, the samples were treated with HF (~40%). The samples were shaken for 2 h, and then kept standing for 2 days to remove silicates. Samples were sieved ultrasonically to remove particles smaller than 10–15 µm. Samples were stored in water, mounted in glycerol, and identified under a light microscope (magnification 400 and 1000×) for pollen, spores, fresh-water algae, dinoflagellate cysts, and microcharcoal. Pollen grains were identified using the African pollen reference collection of the Department of Palynology and Climate Dynamics of the University of Göttingen, African Pollen Database, and after Scott (1982) and Bonnefille and Riollet (1980). At least 300 pollen grains (including terrestrial pollen taxa, sedges, and aquatic taxa but excluding those unidentified and broken) were counted per sample and percentages related to the total number of terrestrial pollen taxa (excluding sedges and aquatic taxa). Pollen diagrams were constructed with TILIA 1.7.16 (Grimm, 2011), using the terrestrial pollen taxa as the pollen sum. Pollen zonation was conducted by Constrained Incremental Sum of Squares Cluster Analysis (CONISS, TILIA 1.7.16) including all counted taxa. In addition, 95% confidence intervals were calculated following Maher (1972). Samples volumes were measured using water displacement to calculate concentration values. Pollen accumulation rates were calculated by multiplying the pollen concentration (grains/cm³) by the sedimentation rate (cm/yr) for each sample. Microcharcoal particles (5–150 µm) were counted on the same slides for pollen analysis using the 202 touch point count method (Clark, 1982) to calculate the microcharcoal concentration in square centimeter/cubic centimeter. At least 225 fields were analyzed to improve the statistical reliability of the results.

Results

Chronology

Lithology, sediment texture, and other physical property data (including reflectance, magnetic susceptibility, porosity, and density) indicate an undisturbed and continuous sedimentation of core GeoB8331-4 (Schneider et al., 2003). The age model of GeoB8331-4 is based on small well-preserved Nassarius vinctus shells of which seven radiocarbon ages have been published by Herbert and Compton (2007) and two new radiocarbon ages are presented in Hahn et al. (2015) (Table 1). In this study, a local ΔR of 146 ± 85 ¹⁴C years (Dewar et al., 2012) and the Marine13 calibration curve (Reimer et al., 2013) was used for age calibration. The age–depth model is established using linear interpolation to describe the relationship between calendars ages and sediment depth, showing a basal age of 9900 cal. yr BP (Hahn et al., 2015). Sedimentation rates vary from 0.09 cm/yr near the base of the core to approximately 0.24 cm/yr near the top of the core (Table 1). The chronology of multicore GeoB8331-2 has been established by Leduc et al. (2010) based on ²¹⁰Pb_ex measurements suggesting coverage of the period 1940–2000 AD.

Table 1.

Radiocarbon ages for gravity cores GeoB8331-4 and GeoB8323-2.

Core site	Depth (cm)	Material	Uncalibrated ¹⁴C age	Calibrated ¹⁴C age 95.4% (2 δ) median		Rate (cm/yr)	Source
GeoB8331-4	37.5	Gastropod	675 ± 30	320–	159	0.24	Hahn et al. (2015)
GeoB8331-4	195	Gastropod	1420 ± 35	999–650	819	0.24	Herbert and Compton (2007)
GeoB8331-4	211.5	Gastropod	1535 ± 30	1146–736	937	0.14	Hahn et al. (2015)
GeoB8331-4	271	Gastropod	1870 ± 35	1484–1073	1277	0.18	Herbert and Compton (2007)
GeoB8331-4	475	Gastropod	3145 ± 35	3005–2509	2780	0.14	Herbert and Compton (2007)
GeoB8331-4	615	Gastropod	5020 ± 35	5430–4909	5166	0.06	Herbert and Compton (2007)
GeoB8331-4	704	Gastropod	6215 ± 35	6696–6290	6495	0.07	Herbert and Compton (2007)
GeoB8331-4	775	Gastropod	8410 ± 35	9027–8536	8798	0.03	Herbert and Compton (2007)
GeoB8331-4	858	Gastropod	9130 ± 40	9996–9547	9675	0.09	Herbert and Compton (2007)
GeoB8323-2	39.5	Gastropod	740 ± 70	421–	222	0.18	Hahn et al. (2015)
GeoB8323-2	111	Gastropod	1700 ± 30	1276–925	1108	0.08	Hahn et al. (2015)
GeoB8323-2	124	Gastropod	1790 ± 35	1369–981	1195	0.15	Herbert and Compton (2007)
GeoB8323-2	227	Gastropod	2730 ± 35	2514–1992	2254	0.10	Herbert and Compton (2007)

The sandy basal sediment of core GeoB8323-2 was interpreted to be a high-energy lag deposit formed by sea level rise during the last deglaciation, resulting in a hiatus between 285 and 234 cm (Herbert and Compton, 2007). Therefore, in this study, only the material above 227 cm was analyzed and the age model is based on small well-preserved Nassarius vinctus shells of which two radiocarbon ages have been published by Herbert and Compton (2007) and two new radiocarbon ages are presented in Hahn et al. (2015) (Table 1). The age–depth model shows a basal age of 2254 cal. yr BP. Sedimentation rates vary markedly: ca. 0.09 cm/yr around 227–124 cm and 111–39.5 cm, and higher values at 124–111 cm (ca. 0.15 cm/yr) and 39.5–0 cm (ca. 0.18 cm/yr) (Table 1). ²¹⁰Pb_ex measurements of 69, 67, 82, 77, 62, 69, and 39 Bq/kg were performed on multicore GeoB8323-1 at depths of 3, 4, 5, 8, 9, and 12 cm, respectively. The calculated mean sedimentation rate for this core is 0.31 cm/yr, assuming a constant initial ²¹⁰Pb_ex concentration and a constant sedimentation rate. Hence the multicore covers the period 1870–2003 AD.

Pollen and microcharcoal

Pollen appears to be well preserved within sites GeoB8331 (including GeoB8331-4 and GeoB8331-2) and GeoB8323 (including GeoB8323-2 and GeoB8323-1) and pollen concentrations vary from 1400 to 15,000 grains/cm³ and 6100 to 14,000 grains/cm³, respectively. The following two subsections describe the pollen and microcharcoal data from the two sites.

Site GeoB8331

The frequency distribution of selected pollen taxa over the last 9900 years is provided in Figure 4 (see Appendix 1 for the list of identified pollen taxa). The record has been divided into three pollen assemblage zones (PZ) using the CONISS calculation (Grimm, 2011).

Figure 4.

Pollen percentages of selected taxa, microcharcoal concentration (cm²/cm³), pollen concentration (grains/cm³), and accumulation rates (grains/cm²/yr) from gravity core GeoB8331-4 and multicore GeoB8331-2 (upper 10 samples covering the period 1940–2000 AD). Lines denote five times exaggeration of percentage curves with low values, and the dots denote presence of taxa with percentage values less than 1%. Pollen assemblage zones (PZ) were conducted using the CONISS calculation. For the identified pollen taxa of each group, see Appendix 1.

Zone PZ GeoB8331a (ca. 9900–7800 cal. yr BP, 878–745 cm of GeoB8331-4) is characterized by pollen percentage maxima of fynbos elements such as Stoebe/Elytropappus-type (4%), Restionaceae (7%), and Anthospermum (4%) at the base of the zone. These percentages decline gradually toward the top of the zone, as do the pollen percentages of Asteroideae (from 15 to 9%). The decline is accompanied by an increase in Cheno/Am (Chenopodiaceae and Amaranthaceae), and a maximum in Aizoaceae pollen percentages characteristic of Succulent Karoo between 847 and 823 cm, which thereafter decrease toward the end of the zone. The microcharcoal concentration increases from 0.1 cm²/cm³ to a maximum value of 0.3 cm²/cm³ at 823 cm and then decreases again toward the end of the zone. Fluctuations in microcharcoal concentration are matched by those in pollen concentrations and accumulation rates, which also have maxima at 823 cm and then decrease toward the end of the zone.

Zone PZ GeoB8331b (ca. 7800–2400 cal. yr BP, 745–418 cm of GeoB8331-4) begins with a pollen percentage increase of Poaceae and Phragmites-type (grass pollen <25 µm; Bonnefille and Riollet, 1980). Poaceae values increase from 35% at the beginning of this zone to 52% at 589 cm, and then fluctuate around 47%, decreasing toward the end of the zone. Pollen percentages of fynbos elements, Stoebe/Elytropappus-type in particular, decline compared to PZ GeoB8331a, with the exception of Restionaceae. The tree and shrub elements of Acacia, Celastraceae, Diospyros, Lannea-type, and Rhus-type are more prominently represented here than PZ GeoB8331a. Aizoaceae pollen percentages increase to a maximum value (7%) at 464 cm and decrease afterward. Neuradaceae pollen percentages reach the highest value (2%) of the record between 614 and 588 cm, corresponding with the maximum value of Cyperaceae (41%) and the presence of other aquatics and swamp elements (Gunnera, Typha). The spores including Trilete, Ophioglossum, Phaeoceros, and Polypodium occur more often than in PZ GeoB8331a. Concentricystes (possibly a fresh-water algae) (Rossignol, 1962) increases to a maximum value (4%) at the beginning of PZ GeoB8331b and fluctuates around 2% for the rest of the zone. The microcharcoal concentrations in this zone are initially at a low value (0.1 cm²/cm³) in comparison to the previous zone and generally increase with two peaks at 538 and 457 cm. The similar trend is observed in the pollen concentrations and pollen accumulation rates.

Zone PZ GeoB8331c (ca. 2400 cal. yr BP–2003 AD, 418–0 cm of GeoB8331-4 and all samples (30–0 cm) of GeoB8331-2) is characterized by the increased presence of arboreal pollen such as Celastraceae, Combretaceae, and Euclea. Rhus-type, and Acacia percentages also increase in this zone, together with the presence of neophytic trees (Pinus and Quercus). Grass pollen percentages are relatively stable in comparison to PZ GeoB8331b between 418 and 199 cm and then decline from 51% to 30% followed by a sharp increase in the uppermost sample. The percentages of Restionaceae pollen decline toward the end of the zone and also the other fynbos elements are less represented. The pollen percentages of Cheno/Am and Aizoaceae both reach maximum values at the top of the zone with increased representation of Crassulaceae, Solanaceae, and Euphorbia. The percentages of Asteroideae and Pentzia-type first fluctuate around 8% and 3%, respectively, and increase abruptly from 49 cm to the top of the sequence. Percentages of Phragmites-type pollen increase to a maximum at the middle of the zone corresponding to a decline in Asteroideae percentages, and again rise to another maximum at the end of the zone parallel to the decline in grass pollen. Ophioglossum, Phaeoceros, and Polypodium spores are much more prominent than in PZ GeoB8331b, while Concentricystes decreases to low values. The microcharcoal concentration fluctuates around 0.1 cm²/cm³ with maxima at 392 cm and 200 cm, in parallel with the pollen concentration. Pollen accumulation rates increase to higher values in comparison to PZ GeoB8331a and PZ GeoB8331b.

Site GeoB8323

The frequency distribution of selected pollen taxa over the last 2300 years is presented in Figure 5 (see Appendix 1 for the list of identified pollen taxa). Although the record has been divided into three pollen assemblage zones (PZ) using the CONISS calculation, the overall pollen assemblage shows little variability except for the upper zone (PZ GeoB8323c). The pollen sequence is dominated by typical fynbos elements such as Restionaceae, Anthospermum, Stoebe/Elytropappus-type, and Ericaceae. Restionaceae pollen percentages show small fluctuations except for two maxima at 137 cm and 65 cm. Stoebe/Elytropappus-type and Passerina values remain constant and increase to the maximum values only at the end of the record when the representation of Anthospermum declines. The succulent elements (Aizoaceae and Cheno/Am) also contribute a large proportion to the assemblage, as well as other Asteraceae pollen (Asteroideae, Artemisia-type and Pentzia-type) whose percentages are comparable to those of site GeoB8331 but show very little variation throughout the record. Poaceae pollen also represents a significant proportion of the assemblage slightly decreasing toward the end of the record. However, Poaceae pollen percentages are much lower (9–22%) than in site GeoB8331 (28–52%) where they dominate the record. Tree and shrub elements such as Diospyros, Euclea, Myrica, Podocarpus, and Rhus-type are little represented throughout the record. The neophytic pollen, Pinus and Quercus, are present in the upper 17 cm of core GeoB8323-2 and throughout core GeoB8323-1 (35–0 cm). Aquatics and swamp elements such as Cyperaceae and Phragmites-type are also well represented with percentage values between 6–13% and 1–7%, respectively. Monolete spores occur mainly in PZ GeoB8323b, while the other spores (Trilete and Ophioglossum) are not continuously present. Pollen concentrations are relatively stable around 9800 grains/cm³, while microcharcoal concentration varies showing maxima at 209 cm and 145 cm. Pollen accumulation rates remain constant between 227 and 49 cm, and then increase to higher values toward the end of the record probably due to the effect of high sedimentation rates of the multicore GeoB8323-1 (0.31 cm/yr) compared to the gravity core GeoB8323-2 (0.18 cm/yr).

Figure 5.

Pollen percentages of selected taxa, microcharcoal concentration (cm²/cm³), pollen concentration (grains/cm³), and accumulation rates (grains/cm²/yr) from gravity core GeoB8323-2 and multicore GeoB8323-1 (upper 12 samples covering the period 1870–2003 AD). Lines denote five times exaggeration of percentage curves with low values, and the dots denote presence of taxa with percentage values less than 1%. Pollen assemblage zones (PZ) were conducted using the CONISS calculation. For the identified pollen taxa of each group, see Appendix 1.

Discussion

Interpretation of the pollen record

Interpretation of pollen and spore records (as those of other particles in the marine sediments) should take into account the wide and complex source area, transport, and sedimentation processes (Dupont, 1999). Many studies of the pollen distribution in marine surface sediments around Africa and elsewhere have shown that the lateral displacement of pollen by ocean currents is relatively small, even in deep-sea sediments (e.g. Hooghiemstra et al., 2006). The sediments used in our study are far shallower being all retrieved on the inner-shelf with water depths not exceeding 120 m further limiting the effect of ocean currents. Moreover, Mabote et al. (1997) showed that the poleward countercurrent is not competent to transport medium to fine silt as far south as the Buffels River mouth. Absolute dating of several multicores and gravity cores from the mudbelt (Herbert and Compton, 2007; Leduc et al., 2010; Taylor, 2004) asserted that vertical mixing within the Holocene sediment is not prominent.

Site GeoB8331 from the northern mudbelt is located approximately 17 and 60 km away from the mouths of the Holgat and Orange Rivers, respectively. Comparing this record with the pollen distribution of the mudbelt indicates that the pollen in site GeoB8331 is mainly fluvially transported and probably originates mostly from the karroid vegetation in the middle catchment area of the Orange River (Zhao et al., 2015). We do not exclude that single pollen grains may have been transported over long distances, even as far as the Drakensberg region, as is the case for much of the Orange River sediment load (Hahn et al., 2015). However, the pollen sources lying between the Drakensberg and the Orange River mouth would have overwhelmed any signal from eastern South Africa being so much closer to deposition in front of the Orange River. Also the sparse occurrences of Podocarpus and Pinus pollen, which highly productive sources grow far from the western coast, argue that long-distance transport of pollen is only of minor importance.

The pollen of grasses (Poaceae, dominant taxa in Group 1; see Regional setting), probably having its main origin in the savanna, grassland, and Nama Karoo vegetation, dominate the entire record of GeoB8331 (28–52%). Our Poaceae pollen percentages compare well with the records of Blydefontein (Scott et al., 2005) and Braamhoek wetland (Norström et al., 2009) situated in the Orange River basin. Other abundant pollen types belonging to Group 1 originate from Cyperaceae species (sedges etc.) (8–41%) and Phragmites-type (1–18%), probably growing in the Orange River riparian zone and swamps (e.g. halophytic swamps in the savanna biome). Arboreal pollen originates from trees and shrubs such as Acacia, Olea, Diospyros, Euclea, Myrica, and Rhus-type growing in diverse habitats of savanna, woodland, and dry forest. Pollen of the fynbos elements (Group 2) is present throughout the record of GeoB8331 with low percentages indicating a small but continuous pollen supply from the Cape region. In contrast, the pollen assemblage of site GeoB8323 is dominated by Group 2 and has a relatively poor representation of Group 1, while the proportion of Group 3 shows no significant differences with that of site GeoB8331.

In summary, the pollen in site GeoB8331 mainly originates from fluvial transport by the Orange River, as well as local runoff and berg winds (pollen from Succulent Karoo, Nama Karoo, possibly Namib Desert) and fynbos elements from the Cape region, while the pollen in site GeoB8323 are dominated by fynbos elements from the Cape region with much less contribution by the Orange River. Therefore, we use the record of pollen types mainly coming from the Orange River catchment area (Group 1, including Poaceae, Cyperaceae, Phragmites-type, and Typha) to infer vegetation change in the SRZ and pollen types from the fynbos elements (Group 2, including Restionaceae, Anthospermum, Ericaceae, Cliffortia, Stoebe/Elytropappus-type, Passerina, Artemisia-type, and Pentzia-type) for interpretation of vegetation change in the WRZ. Tribulus was excluded from Group 1 to avoid the possible effect of pioneer vegetation from disturbance areas.

Grass pollen are mainly derived from grassland, savanna, and Nama Karoo, which are vegetation types strongly associated with predominantly summer rainfall regions, while Asteraceae (excluding Stoebe/Elytropappus-type) pollen mainly originating from the Nama Karoo is favored by weaker rainfall seasonality or drier and cooler summers (Cooremans, 1989; O’Connor and Bredenkamp, 1997). Thus, the Poaceae/Asteraceae ratio (Poac/Ast) can be used to offer a perspective on total rainfall and degree of seasonality (Norström et al., 2009; Scott et al., 2005). Additionally, pollen taxa growing along the Orange River and swamps (Cyperaceae, Phragmites-type, Typha) may provide some indication of a high water availability in the Orange River floodplain.

The microcharcoal concentration is expressed as area of fragments per cubic centimeter. It is difficult to distinguish between the microcharcoal sources from the SRZ and WRZ. Therefore, in this study, the microcharcoal concentration was interpreted based on the vegetation change in the SRZ and WRZ inferred from the pollen record.

Holocene climate history and regional comparison

Analysis of sites GeoB8331 yields a high-resolution record of Holocene climate and vegetation change of South Africa. In order to test the contention that the climate variability in the SRZ and WRZ are contrasted, our data are compared in Figure 6 with the pollen record of Wonderkrater (Figure 6e) (Scott et al., 2012) and the δ¹³C record of the Makapansgat stalagmite (Figure 6d) (Holmgren et al., 2003) from the SRZ, the pollen record of Pakhuis Pass (Figure 6h) (Scott et al., 2012) from the WRZ, and the alkenone-derived sea surface temperature record from the same core GeoB8331-4 (Figure 6i) (Leduc et al., 2010). We also compare our results with iron concentrations from the Chilean continental margin at 41°S (Figure 6j) (Lamy et al., 2001), which might provide a record of the global latitudinal position of the southern westerlies as its position is mainly determined by the global temperature gradient.

Figure 6.

Comparison of the percentage of (b) Group 1, (g) Group 2, and (c) ratio of Poaceae over Asteraceae counts from gravity core GeoB8331-4 and multicore GeoB8331-2 (upper 10 samples) (a) with austral summer (December) insolation at 30°S (Berger and Loutre, 1991), (d) δ¹³C record from Makapansgat (Holmgren et al., 2003), (e) humidity index based on pollen records from Wonderkrater (Scott et al., 2012), and (h) Pakhuis Pass (Scott et al., 2012), (f) precipitation stack of the wettest quarter in northern South Africa, and (k) central and eastern South Africa (Chevalier and Chase, 2015), (i) alkenone-derived sea surface temperature (SST) records from gravity core GeoB8331-4 (Leduc et al., 2010), and (j) the latitudinal position of the southern westerlies based on iron concentrations from the Chilean continental margin at 41°S (Lamy et al., 2001). For the color figure, the reader is referred to the web version of this article.

Early Holocene (9900–7800 cal. yr BP)

The low representation of Poaceae in the pollen record at the beginning of the sequence indicates a reduced extension of grasslands/savanna and less summer rainfall in the interior of South Africa (SRZ) during the early Holocene, which is also reflected in the low Poac/Ast ratio (Figure 6c). Many shrubs and trees such as Rhus-type, Diospyros, and Euclea are represented with few pollen grains only, although Acacia does appear and sustains its contribution to the pollen sum. Both the Makapansgat δ¹³C record and the Wonderkrater pollen record indicate a decreased C₄ grass cover and considerably drier environments between 10,200–8400 cal. yr BP. After 8500 cal. yr BP, aridity persisted at Wonderkrater as indicated by the increased grass pollen record, while no clear trend was found at Makapansgat. Drier conditions with low summer rainfall in the SRZ before 7800 cal. yr BP are corroborated by the pollen and geochemistry records from the Tswaing Crater (Kristen et al., 2007; Metwally et al., 2014). Early Holocene aridity concurs with a minimum for the austral summer insolation (Figure 6a) (Berger and Loutre, 1991).

Maximum percentages of Stoebe/Elytropappus-type pollen in combination with high relative abundance of Anthospermum pollen during the earliest Holocene indicate the presence of Renosterveld vegetation, which is part of the fynbos biome. However, the declining trend in fynbos elements accompanied by a brief expansion of Aizoaceae and Cheno/Am between 9600–9300 cal. yr BP suggests a short period of warmer and drier climate with the expansion of Succulent Karoo in the WRZ. This is supported by the pollen record of Pakhuis Pass indicating a brief phase of humid conditions prior to 9500 cal. yr BP and followed by a marked shift to drier conditions favoring succulent and scrub vegetation between 9500–9000 cal. yr BP (Scott and Woodborne, 2007a, 2007b). Other records from the WRZ also indicate humid conditions between 10,000–8000 cal. yr BP (Klein, 1991; Meadows and Sugden, 1991; Meadows et al., 2010; Quick et al., 2011; Valsecchi et al., 2013; Weldeab et al., 2013). During this period, the high microcharcoal concentrations indicate enhancement of fires, probably attributable to lower summer rainfall in the SRZ and subsequent expansion of Succulent Karoo accompanied by relatively abundant fynbos vegetation, which is fire prone (Mucina and Rutherford, 2006).

Middle Holocene (7800–2400 cal. yr BP)

The gradual increase of Poaceae pollen after 7800 cal. yr BP suggests increasing grass cover especially between 6900 and 3500 cal. yr BP, which might imply increased summer rainfall in the SRZ. Representation of local riparian vegetation (Phragmites-type, Cyperaceae, and Typha) increased somewhat. The presence of pollen of other aquatics and swamp plants as well as spores is suggestive of moister conditions in the Orange River catchment since 7800 cal. yr BP. The middle Holocene trend toward relatively higher δ¹³C values at Makapansgat parallels the increase in grass pollen of site GeoB8331, and the Wonderkrater humidity index also indicated wetter conditions. Between 6900 and 3500 cal. yr BP, the maximum grass cover and high Poac/Ast ratio indicate high summer rainfall corresponding to a humidity maximum at Wonderkrater around 5500 cal. yr BP. A similar conclusion has been drawn from Blydefontein where the transition from dominantly Karoo shrubs to more grassy vegetation after 5400 cal. yr BP has been interpreted as a response to increased summer rainfall (Scott et al., 2005). A rather humid climate during the middle Holocene has been supported by the pollen record of Tswaing Crater showing increased representation of mesic woodland, savanna trees, and local swamps (Metwally et al., 2014). The geochemical record of Tswaing Crater also suggests that conditions gradually became wetter after 6000 cal. yr BP (Kristen et al., 2007). The enhanced moisture availability in the SRZ of the middle Holocene appears to be associated with increasing austral summer insolation (Figure 6a).

Fynbos elements (Group 2) fluctuate around a very slightly decreasing trend, while the representation of Aizoaceae gradually increases suggesting a warmer and drier climate in the WRZ, which corresponds with a dry period at Pakhuis Pass ca. 8000–5000 cal. yr BP (Scott and Woodborne, 2007a, 2007b). Most records from the WRZ (Klein, 1991; Meadows and Sugden, 1991; Meadows et al., 2010; Quick et al., 2011) indicate relatively dry conditions between 8000 and 3000 cal. yr BP except that of De Rif 2010, which indicates more variable conditions (Valsecchi et al., 2013). The drier climate during this period might be related to a southward shift of the southern westerlies (Lamy et al., 2001) especially between 8000 and 5000 cal. yr BP, resulting in less winter rainfall in the WRZ. A southward shift of the westerlies would also have resulted in more upwelling in the northern mudbelt area decreasing the SSTs at GeoB8331 (Leduc et al., 2010). Pollen and isotope records of Katbakkies Pass (Chase et al., 2015) indicate increasing moisture ca. 7000–3000 cal. yr BP interpreted to be a result of increased tropical easterly flow, which could explain the variable moisture conditions at Seweweekspoort, which is located in the YRZ (Chase et al., 2013). The low values in microcharcoal concentration between 7600 and 5900 cal. yr BP could be attributed to a moister climate trend in the SRZ leading to less frequent fire. The gradual increase in microcharcoal concentration between 5900 and 2600 cal. yr BP could be associated with accumulated availability of fuel supplied by the abundant grasses under relatively humid conditions in the SRZ as well as gradually drier climate in the WRZ.

Late Holocene (2400 cal. yr BP–modern)

Expansion and a change in the composition of the savanna since the middle Holocene is indicated by the regular occurrence of tree and shrub pollen, notably that of Combretaceae, in the record after 2400 cal. yr BP. Combretaceae species are mostly constituents of the savanna and are missing in the southern and western parts of South Africa today (Dyer, 1975). Development of savanna woodlands is also recorded at Tswaing Crater between ca. 7200 and 1800 cal. yr BP (Metwally et al., 2014). The expansion of savanna suggests a southward shift or an intensification of the SRZ. The Poac/Ast ratio increases toward a maximum at about 850 years ago, indicating more summer rainfall in the SRZ corresponding with the austral summer insolation maxima. Gradually increasing representation of Phragmites-type until ca. 1500 cal. yr BP indicates the development of a riparian reed belt along the Orange River corresponding with the increase of Cyperaceae and Typha pollen in Tswaing Crater (Metwally et al., 2014). The δ¹³C record at Makapansgat reaches a maximum between 3000 and 2000 cal. yr BP, indicating a grassy environment that turns to a bushy environment after 2000 cal. yr BP, while the humidity index of Wonderkrater increases to the maximum at about 500 cal. yr BP. The sharp decrease of moisture at Wonderkrater and higher Asteraceae pollen in site GeoB8331 corresponding to the decline in grass pollen since the last 850 years could be a consequence of a drier climate with less summer rainfall in the SRZ. Additionally, grasses could have been reduced at least in part by the grazing of domestic stock since the arrival of the Iron Age people, which were present in at least some parts of the SRZ as early as ca. 1700 cal. yr BP (Evers, 1975; Klapwijk, 1974).

The continuing low representation of fynbos elements and strong increase of Succulent Karoo elements (Aizoaceae, Crassulaceae, Euphorbia, and Cheno/Am) suggest that the climate was becoming more arid in the WRZ. The humidity index of Pakhuis Pass first declines to minimum values and after that rises again with large variations to the middle Holocene level. However, general moist conditions were indicated in other records from the WRZ (Klein, 1991; Meadows and Sugden, 1991; Valsecchi et al., 2013) except at Katbakkies Pass where drier conditions were indicated after 1700 cal. yr BP. The low values in microcharcoal concentration during the late Holocene are difficult to explain. On the one hand, the decline in grasses in turn diminished the fuel supply for fires in the SRZ; on the other hand, the drier climate in the WRZ should have been conducive to fires. Considering the history of Khoikhoi people, which are the native pastoralist people of southwestern Africa since ca. 2300 cal. yr BP, the increase of Asteroideae and Aizoaceae pollen since ca. 1400 cal. yr BP, suggesting the occurrence of more shrubby vegetation, could indicate drier conditions but also could be (partly) attributable to the activities of Khoikhoi people (Bousman, 1998; Bousman and Scott, 1994).

The neophyte Pinus is common and widespread in the southwestern mountains of the Cape region for the last ca. 100 years (Shaughnessy, 1986). Pollen of neophytes associated with the arrival of Europeans such as Pinus and Quercus are found in low percentages (no more than 1%) and only for the last 100 years (which helps to confirm the upper part of the age model). The percentage of neophytes pollen recorded in the modern marine sediments is also very low (no more than 2%, Zhao et al., 2015). This is supported by most records in both the WRZ and SRZ where Pinus pollen is only present in low frequencies (in percentages of less than 5%) during the last 300 years, such as in Driehoek Vlei, Pakhuis Pass, De Rif, Klaarfontein springs, Verlorenvlei, and Spring Cave Shelter in the WRZ (Baxter, 1996; Meadows and Baxter, 2001; Meadows and Sugden, 1991; Meadows et al., 1996; Quick et al., 2011; Scott and Woodborne, 2007a), and Wonderkrater, Tswaing Crater, Braamhoek, Blydefontein, and Mahwaqa in the SRZ (Metwally et al., 2014; Neumann et al., 2014; Norström et al., 2014; Scott et al., 2005, 2012). Only Neumann et al. found high representation (percentages of up to 60%) of Pinus pollen in Princess Vlei of the WRZ (Neumann et al., 2011), Lake Eteza, and Lake Sibaya in the SRZ (Neumann et al., 2008, 2010). The low percentages of Pinus in our samples indicate that long-distance transport of pollen is not an issue, otherwise high percentages of Pinus pollen would have been found in the upper sediments dated after pine trees have been planted along the south coast.

Climate variations as deduced from two sites for the last 2200 years

A comparison between the pollen and microcharcoal records of the two sites, GeoB8331 and GeoB8323, has been carried out for the last 2200 years. To facilitate the comparison, we have interpolated the pollen and microcharcoal values at regular time intervals of 100 years between 2200 and 100 cal. yr BP and of 10 years between 100 cal. yr BP and modern (Figure 7).

Figure 7.

Comparison of the percentages of (a) Group 1, (b) Group 2, (c) Group 3, and (d) microcharcoal concentrations from the sites GeoB8331 (including gravity core GeoB8331-4 and multicore GeoB8331-2 in green) and GeoB8323 (including gravity core GeoB8323-2 and multicore GeoB8323-1 in blue) during the last 2200 years. Samples dated between 2200 and 100 cal. yr BP have been interpolated every 100 years and those dated thereafter have been interpolated every 10 years. For the color figure, the reader is referred to the web version of this article.

The same pollen groups are present in both cores but vary in percentage, especially Groups 1 and 2 (Figure 7a and b). These percentage variations are most likely the consequence of different pollen sources for the two sites GeoB8331 and GeoB8323 i.e. being drawn largely from the SRZ and WRZ, respectively. This pattern is consistent with the pollen distribution in modern marine sediments from the mudbelt (Zhao et al., 2015). The contrasting temporal trends between the two sites in Group 1 pollen types emphasize the different source areas contributing to the pollen assemblage at each site. Pollen of Group 1 at GeoB8323 may be mainly locally derived from the west coast rather than from the SRZ, while transport via the Orange River might be the case at GeoB8331. The recovery of grasses at GeoB8331 during the last 150 years might be attributed to human control of grazing intensity in the SRZ. Additionally, the greater fluctuations of Group 2 at GeoB8323 (Figure 7b) suggest that it more likely reflects climate changes in the WRZ. On the other hand, the obvious increase of Group 3 accompanied by the decline of Group 1 at GeoB8331 after 1400 cal. yr BP (Figure 7a) suggests drier conditions with less summer rainfall in the SRZ. The drier conditions may be associated with the event of ‘Little Ice Age’ (LIA) in the SRZ during which the cold and dry climate is not conducive to the growth of grasses which are more favored under the warm and moist conditions. This is supported by the δ¹⁸O records from Makapansgat stalagmite in which the LIA is expressed as around the last 700–200 years (Tyson et al., 2000). Huffman (1996) reported that it was colder in northeastern South Africa for the last 650–150 years. The maximum percentages of Group 2 and minimum percentages of Group 3 (Figure 7b and c) at GeoB8323 between 700 and 200 cal. yr BP, however, suggest a wetter climate for the LIA in the WRZ. Wet conditions during the LIA would corroborate the interpretation of the sediment record of core GeoB8332-4 (located near the Orange River, Figure 2) as a wet phase in the WRZ during the LIA (Weldeab et al., 2013). The contrasting climate conditions during the LIA between the SRZ and WRZ were also detected based on the geochemical and isotopic records from the same gravity cores, namely, GeoB8331-4 and GeoB8323-2 (Hahn et al., 2015). On the other hand, the obvious increase of Group 3 accompanied by the decline of Group 1 suggests an increased pollen contribution from the lower Orange River catchment resulting in less input of grass pollen from the middle catchment. The source shift to the lower catchment is supported by the distinctive Sr and Nd isotopic compositions of gravity core GeoB8331-4 (Hahn et al., 2015). The greater contribution of Group 3 at GeoB8331 appears to also be recorded in the microcharcoal concentrations (Figure 7d), which show a comparable variation between two cores following the decline of Group 1 and increase of Group 3 after 1400 cal. yr BP. The vegetation characterized by Group 3 is mainly Nama Karoo and Succulent Karoo, which are distributed in the areas acting as pollen sources for the site GeoB8323.

Holocene vegetation and climate dynamics of South Africa

The high-resolution pollen record of site GeoB8331 corroborates contrasting climate developments between the SRZ and WRZ during the early and middle Holocene (see Introduction, Figure 3). For the SRZ, reduced grass cover indicates low summer rainfall during the early Holocene, and the expansion of grasses toward the middle Holocene suggests increasing levels of moisture that are associated with higher summer rainfall. The opposite climate trend is found for the WRZ, where a greater prominence of Renosterveld vegetation suggests relatively humid conditions during the earliest Holocene followed by gradually drier conditions toward the middle Holocene accompanied by the expansion of Succulent Karoo and the decline of fynbos vegetation. The late Holocene comparison between the two sites of GeoB8331 and GeoB8323 also reflects different climate conditions during the LIA being drier in the SRZ but wetter in the WRZ.

What mechanisms can be considered to explain the emerging pattern? The driving forces of continental vegetation and hydrology during the Holocene are debated and four different main drivers of regional climate changes have been proposed: (1) strength and latitudinal position of the southern westerlies, (2) effects of sea surface temperature on continental hydrology associated with the subtropical anticyclones over the southeast Atlantic Ocean, (3) insolation forcing, and (4) intensification of human activities.

The latitudinal position of the southern westerlies directly influences the continental hydrology of the Southern Hemisphere (Reason and Rouault, 2005). At present, only the southwestern tip of South Africa is affected by the southern westerlies when they shift equatorward during austral winter (Figure 1b) resulting in winter rainfall in the Cape region. Further north, a strengthened South Atlantic Anticyclone would lead to intensified southeast trade winds, intensification of Benguela upwelling, and decrease of sea surface temperature along the southwestern coast causing aridification of the coastal region (Shannon and Nelson, 1996). The initial presence of Renosterveld vegetation and the subsequent declining trend of fynbos elements accompanied by the expansion of Succulent Karoo in the WRZ could be explained by two different hypotheses. One is a northward shift of the southeast trade winds during the course of the Holocene which would lower the effectivity of wind transport and to bring fynbos pollen to site GeoB8331. However, a northward shift of the southeast trade winds would be expected if the westerlies are located further north allowing more humidity to penetrate into the continent, which is inconsistent with other records from the WRZ (Meadows et al., 2010; Quick et al., 2011; Scott and Woodborne, 2007a, 2007b). Also the low SSTs between 8000 and 5000 cal. yr BP (Leduc et al., 2010) indicate rather more upwelling in the northern mudbelt area, suggesting trade winds affected the southern Benguela Upwelling System. Another hypothesis is the southward shifting of the southern westerlies toward the middle Holocene, which would have arrived at its southernmost position between 8000 and 5000 cal. yr BP (Lamy et al., 2001), resulting in less winter rainfall in the WRZ and drier climate during this period. Reduced winter rainfall negatively impacts fynbos but induces an expansion of the Succulent Karoo. Therefore, we interpret that the decrease in fynbos pollen at site GeoB8331 is probably an effect of retreat of the WRZ resulting from a poleward shift of the southern westerlies.

Fluctuations in orbital variation may have a strong influence on glacial–interglacial climate variability (Hopley et al., 2007; Partridge et al., 1997), while Chevalier and Chase (2015) demonstrated that it was not the primary driver of climate variability in northeastern South Africa except during the Holocene when insolation forcing became significant (Figure 6f). The insolation forcing on shorter time scales during the Holocene has also been proposed by other studies (Chase et al., 2009, 2010; Schefuß et al., 2011). Higher insolation intensifies atmospheric convection, leading to high rainfall and vice versa (Partridge et al., 1997). The trend in the Poac/Ast ratio (Figure 6c), which is interpreted as the result of increasing summer rainfall, corresponds well with the early Holocene increase in Southern Hemisphere summer insolation (Figure 6a) (Berger and Loutre, 1991). The drier conditions with low summer rainfall in the SRZ during the early Holocene concur with a minimum for the austral summer insolation. The middle Holocene period with enhanced moisture availability in the SRZ appears to be associated with increasing austral summer insolation, and the maximum Poac/Ast ratio at about 850 years ago indicates more summer rainfall in the SRZ corresponding with the austral summer insolation maximum. The marine sedimentary leaf wax δ¹³C records off the coast of Namibia (MD08-3167) suggest increase in C₄ vegetation during the austral summer insolation maximum interpreted as a result of increased summer rainfall (Collins et al., 2014; Daniau et al., 2013). On the other hand, a marine pollen record off the coast of Namibia (GeoB1711-4) suggests that strengthened southeast trade winds were coeval with austral summer insolation maxima (Shi et al., 2001). This is supported by the hyrax midden records from the Namib Desert (Chase et al., 2009), suggesting aridification toward the middle and late Holocene during austral summer insolation maxima. The results of Shi et al. (2001) and Chase et al. (2009) seem to be in contradiction with those of Collins et al. (2014), Daniau et al. (2013), and our study. One possible explanation is associated with the source areas of these four records. The core site of MD08-3167 was interpreted to receive terrestrial material from a large source area from the Namib Desert to the east of Kalahari Desert stretching far into the interior of the SRZ (Collins et al., 2014). The pollen taxa of Poaceae and Asteraceae that are used to interpret the summer rainfall variability in our record also mainly come from the middle Orange River catchment (interior of the SRZ). However, the pollen source areas of GeoB1711-4 were interpreted to be restricted to the zone west of the Kalahari region (Shi et al., 2001), which is also the case for the hyrax midden record receiving pollen from local to regional sources in the Namib Desert (Chase et al., 2009). On the other hand, changes in seasonality, which are controlled by the precession, might offer another but not excluding explanation. High summer rainfall in the interior of the SRZ related to the austral summer insolation could be coeval with strong southeast trade winds along the west coast of southern Africa during weaker austral winter insolation. Therefore, as Collins et al. (2014) proposed, the inconsistency seems to indicate that the Namib Desert might be more controlled by the southeast trade winds, which induce strong upwelling and cause aridity along the coast, while the rainfall into the interior of the SRZ is mainly brought by easterly winds which are associated with austral summer insolation. Overall, the consistence between the rainfall seasonality of the SRZ and the austral summer insolation suggests that insolation forcing on short time scales is a dominant driver of Holocene rainfall variations in the SRZ but does not extend as far as the west coast of southern Africa.

The influence of the latitudinal position of the southern westerlies might have reached no further than the WRZ. Chevalier and Chase (2015) proposed that precipitation variability in central and southeastern South Africa (Figure 6k) might be linked to the latitudinal position of the southern westerlies rather than the role of insolation forcing. Low precipitation was indeed recorded between ca. 7500 and 5000 cal. yr BP in central and southeastern South Africa (Figure 6k) corresponding to a southward shift of the southern westerlies. No such evidence from the SRZ is found in our records probably because the pollen sources in our records do not grow as far as southeastern South Africa. However, the strong declining trend in precipitation of central and southeastern South Africa after ca. 4000 cal. yr BP (Figure 6k) would not be in line with a relatively stable position of the global southern westerlies (Lamy et al., 2001), suggesting that the complexities and evolution of this driver remain to be further investigated as proposed by Chevalier and Chase (2015).

The effects of human activities such as deforestation, soil erosion, enhanced burning, and overgrazing are not apparent during the Holocene, except for the last 850 years. The decline in grass pollen at site GeoB8331 over the last 850 years could have been partly affected by the grazing of domestic stock, while the recovery of grasses during the last 150 years could possibly be attributed to human control of grazing intensity in the SRZ.

Conclusion

In this study, we reconstructed contrasting vegetation and climate histories in the SRZ and WRZ of South Africa covering the last 9900 years based on continuous high-resolution pollen and microcharcoal records of site GeoB8331 retrieved from the northern mudbelt offshore the west coast of South Africa. Different pollen types were grouped to infer vegetation and climate change in both the SRZ and WRZ based on the distinctive pollen sources and transport processes. Contrasting climate patterns are evident in the SRZ and WRZ corroborating results from literature, especially during the early and middle Holocene. Relatively humid conditions in the WRZ are suggested by the presence of Renosterveld vegetation during the earliest Holocene, which then gave way to gradually warmer and drier conditions inferred from the decline of fynbos vegetation and expansion of Succulent Karoo. Opposing climate developments are observed for the SRZ, where a rather moist savanna/grassland expanded during the middle Holocene implying increased summer rainfall. The drier conditions toward the middle Holocene in the WRZ are attributable to a southward shift of the southern westerlies, bringing less rainfall during the austral winter, while the increase in austral summer insolation during the middle Holocene might present a dominant driver of higher summer rainfall in the SRZ.

Comparing the results of site GeoB8331 from the northern mudbelt with the pollen and microcharcoal records of site GeoB8323 retrieved from the southern mudbelt allowed a more detailed interpretation for the last 2200 years. Pollen composition of the two sites shows marked differences in the proportions of the three pollen groups with distinctive source areas. The climate in the WRZ during the last 2200 years appears to have been more stable in comparison to that of the SRZ. Effects of the LIA were detected around the last 700–200 years with colder and drier conditions in the SRZ suggested by a decline of grasses and an increase of Succulent Karoo, while in the WRZ colder and wetter conditions were indicated by the increased presence of fynbos vegetation and decreased representation of Succulent Karoo.

Footnotes

Appendix 1

Identified pollen taxa from sites GeoB8331 and GeoB8323.

Neophytic trees	Neuradaceae	Labiateae
Pinus (Pinaceae)	Protea (Proteaceae)	Lankesteria (Acanthaceae)
Quercus (Fagaceae)	Restionaceae	Leguminosae
Trees and shrubs	Passerina (Thymelaeaceae)	Malvaceae
Celastraceae	Succulent	Myrsine (Myrsinaceae)
Combretaceae	Justicia/Monechma (Acanthaceae)	Parkinsonis (Fabaceae)
Diospyros (Ebenaceae)	Aizoaceae	Pelargonium (Geraniaceae)
Dodonaea (Sapinadaceae)	Euphorbia (Euphorbiaceae)	Persicaria (Polygonaceae)
Euclea (Ebenaceae)	Solanaceae	Petalidium (Acanthaceae)
Ilex (Aquifoliaceae)	Crassulaceae	Phyllanthus (Phyllanthaceae)
Myrica (Myricaceae)	Chenopodiaceae/Amaranthaceae	Plantago (Plantaginaceae)
Olea (Oleaceae)	Aquatics & swamp	Portulaceae
Podocarpus (Podocarpaceae)	Cyperaceae	Pterocarpus (Fabaceae)
Rhamnaceae	Gunnera (Gunneraceae)	Polygalaceae
Salix (Salicaceae)	Phragmites (Poaceae)	Rannulaceae
Rhus (Anacardiaceae)	Juncus (Juncaceae)	Rutaceae
Lannea (Anacardiaceae)	Typha (Typhaceae)	Scrophulariaceae
Tilia (Tiliaceae)	Haloragaceae	Tapiranthus (Loranthaceae)
Acacia (Mimosaceae)	Other vegetation	Tetrapterum (Pottiaceae)
Capparaceae	Alchornea (Euphorbiaceae)	Tribulus (Zygophylaceae)
Other various habitats	Alismataceae	Verbenaceae
Artemisia (Asteraceae)	Aloe (Asparagaceae)	Group I
Pentzia (Asteraceae)	Balanties (Zygophyllaceae)	Poaceae
Asteroideae (tubulifloreous Asteraceae)	Bignoniaceae	Cyperaceae
Gazania (Asteraceae)	Campanulaceae	Phragmites-type
Gerbera (Asteraceae)	Caryophyllaceae	Typha
Pacourina (Asteraceae)	Canthium (Rubiaceae)	Group 2
Vernonia (Asteraceae)	Celtis (Ulmaceae)	Restionaceae
Poaceae	Cheilanthes (Adianthaceae)	Anthospermum
Spores	Coccinia (Cucurbitaceae)	Ericaceae
Monolete	Convolunaceae	Cliffortia
Trilete	Cucumis (Cucurbitaceae)	Stoebe/Elytropappus-type
Ophioglossum (Ophioglossaceae)	Detarium (Fabaceae)	Passer na
Phaeceros (Notothyladaceae)	Dichrostoehy (Fabaceae)	Artem s a-type
Polypodium (Polypodiaceae)	Elaeagnaceae	Pentzia-type
Fynbos	Geranium (Geraniaceae)	Group 3
Anthospermum (Rubiaceae)	Hypoestes (Acanthaceae)	Aizoaceae
Stoebe/Elytropappus (Asteraceae)	lcancinaceae	Cheno/Am
Cliffortia (Rosaceae)	Indigofera (Fabaceae)	Asteroideae
Ericaceae	Kirkia (Kirkiaceae)

Acknowledgements

Thanks to the captain, the crew, and scientists of the Meteor M57-1 cruise for recovering the studied material. We also thank Gesine Mollenhauer, Matthias Zabel, and three anonymous reviewers for critical discussion and helpful advice.

Funding

This study was funded by the German Federal Ministry of Education and Research (BMBF). The investigations were conducted within the collaborative project ‘Regional Archives for Integrated Investigations’ (RAiN), which is embedded in the international research program SPACES (Science Partnership for the Assessment of Complex Earth System Process). Funding source is the German Federal Ministry of Education and Research (BMBF) (Grant/Award Number: ‘03G0840A’).

References

Baker

Routh

Blaauw

. (2014) Geochemical records of palaeoenvironmental controls on peat forming processes in the Mfabeni peatland, Kwazulu Natal, South Africa since the Late Pleistocene. Palaeogeography, Palaeoclimatology, Palaeoecology 395: 95–106.

Baxter

(1996) Late quaternary palaeoenvironments of the Sandveld, Western Cape Province, South Africa. PhD Thesis, University of Cape Town.

Berger

Loutre

(1991) Insolation values for the climate of the last 10 million years. Quaternary Science Reviews 10(4): 297–317.

Bonnefille

Riollet

(1980) Pollens des savanes d’Afrique orientale. Paris: Éditions du Centre national de la recherche scientifique.

Bousman

(1998) The chronological evidence for the introduction of domestic stock into Southern Africa. African Archaeological Review 15(2): 133–150.

Bousman

Scott

(1994) Climate or overgrazing?: The palynological evidence for vegetation change in the eastern Karoo. South African Journal of Science 90(11): 575–578.

Bredenkamp

Granger

van Rooyen

(1996) Vegetation of South Africa, Lesotho and Swaziland. Pretoria: Department of Environmental Affairs and Tourism.

Breman

Gillson

Willis

(2012) How fire and climate shaped grass-dominated vegetation and forest mosaics in northern South Africa during past millennia. The Holocene 22(12): 1427–1439.

Brook

Scott

Railsback

. (2010) A 35 ka pollen and isotope record of environmental change along the southern margin of the Kalahari from a stalagmite and animal dung deposits in Wonderwerk Cave, South Africa. Journal of Arid Environments 74(7): 870–884.

10.

Chase

Meadows

(2007) Late Quaternary dynamics of southern Africa’s winter rainfall zone. Earth-Science Reviews 84(3–4): 103–138.

11.

Chase

Thomas

DSG

(2006) Late Quaternary dune accumulation along the western margin of South Africa: Distinguishing forcing mechanisms through the analysis of migratory dune forms. Earth and Planetary Science Letters 251(3–4): 318–333.

12.

Chase

Boom

Carr

. (2013) Holocene climate change in southernmost South Africa: Rock hyrax middens record shifts in the southern westerlies. Quaternary Science Reviews 82: 199–205.

13.

Chase

Lim

Chevalier

. (2015) Influence of tropical easterlies in southern Africa’s winter rainfall zone during the Holocene. Quaternary Science Reviews 107: 138–148.

14.

Chase

Meadows

Carr

. (2010) Evidence for progressive Holocene aridification in southern Africa recorded in Namibian hyrax middens: Implications for African Monsoon dynamics and the ‘‘African Humid Period’’. Quaternary Research 74(1): 36–45.

15.

Chase

Meadows

Scott

. (2009) A record of rapid Holocene climate change preserved in hyrax middens from southwestern Africa. Geology 37(8): 703–706.

16.

Chevalier

Chase

(2015) Southeast African records reveal a coherent shift from high- to low-latitude forcing mechanisms along the east African margin across last glacial–Interglacial transition. Quaternary Science Reviews 125: 117–130.

17.

Clark

(1982) Point count estimation of charcoal in pollen preparations and thin sections of sediments. Pollen et spores 24(5): 23–35.

18.

Cockcroft

Wilkinson

Tyson

(1987) The application of a present-day climatic model to the late quaternary in southern Africa. Climatic Change 10(2): 161–181.

19.

Collins

Schefuß

Govin

. (2014) Insolation and glacial–Interglacial control on southwestern African hydroclimate over the past 140 000 years. Earth and Planetary Science Letters 398: 1–10.

20.

Compton

Herbert

Hoffman

. (2010) A tenfold increase in the Orange River mean Holocene mud flux: Implications for soil erosion in South Africa. The Holocene 20(1): 115–122.

21.

Cooremans

(1989) Pollen production in central southern Africa. Pollen et spores 31: 61–78.

22.

Cowling

Richardson

Pierce

(1997) Vegetation of Southern Africa. Cambridge: Cambridge University Press.

23.

Daniau

A-L

Goñi

MFS

Martinez

. (2013) Orbital-scale climate forcing of grassland burning in southern Africa. Proceedings of the National Academy of Sciences 110(13): 5069–5073.

24.

Dewar

Reimer

Sealy

. (2012) Late-Holocene marine radiocarbon reservoir correction (ΔR) for the west coast of South Africa. The Holocene 22(12): 1481–1489.

25.

Dupont

(1999) Pollen and spores in marine sediments from the east Atlantic – A view from the ocean into the African continent. In: Fischer

Wefer

(eds) Use of Proxies in Paleoceanography: Examples from the South Atlantic. Berlin: Springer, pp. 523–546.

26.

Dyer

(1975) The Genera of Southern African Flowering Plants. Pretoria: Department of Agricultural Technical Services.

27.

Ekblom

Gillson

Risberg

. (2012) Rainfall variability and vegetation dynamics of the lower Limpopo Valley, Southern Africa, 500 AD to present. Palaeogeography, Palaeoclimatology, Palaeoecology 363: 69–78.

28.

Evers

(1975) Recent Iron Age research in the eastern Transvaal, South Africa. The South African Archaeological Bulletin 30: 71–83.

29.

Goldblatt

Manning

(2002) Plant diversity of the Cape region of southern Africa. Annals of the Missouri Botanical Garden 89(2): 281–302.

30.

Grimm

(2011) TILIA Software, version 1.7. 16. Springfield, IL: Illinois State Museum, Research and Collection Center.

31.

Hahn

Compton

Meyer-Jacob

. (2015) Holocene paleoclimatic record from the South African Namaqualand mudbelt: A source to sink approach. Quaternary International. DOI: 10.1016/j.quaint.2015.10.017.

32.

Herbert

Compton

(2007) Geochronology of Holocene sediments on the western margin of South Africa. South African Journal of Geology 110(2–3): 327–338.

33.

Holmgren

Lee-Thorp

Cooper

GRJ

. (2003) Persistent millennial-scale climatic variability over the past 25,000 years in Southern Africa. Quaternary Science Reviews 22(21–22): 2311–2326.

34.

Hooghiemstra

Lézine

A-M

Leroy

SAG

. (2006) Late Quaternary palynology in marine sediments: A synthesis of the understanding of pollen distribution patterns in the NW African setting. Quaternary International 148(1): 29–44.

35.

Hopley

Weedon

Marshall

. (2007) High- and low-latitude orbital forcing of early hominin habitats in South Africa. Earth and Planetary Science Letters 256(3): 419–432.

36.

Huffman

(1996) Archaeological evidence for climatic change during the last 2000 years in southern Africa. Quaternary International 33: 55–60.

37.

Klapwijk

(1974) A preliminary report on pottery from the north-eastern Transvaal, South Africa. The South African Archaeological Bulletin 29(113–114): 19–23.

38.

Klein

(1991) Size variation in the Cape dune Molerat (Bathyergus suillus) and late Quaternary climatic change in the southwestern Cape Province, South Africa. Quaternary Research 36: 243–256.

39.

Kristen

Fuhrmann

Thorpe

. (2007) Hydrological changes in southern Africa over the last 200 Ka as recorded in lake sediments from the Tswaing impact crater. South African Journal of Geology 110(2–3): 311–326.

40.

Lamy

Hebbeln

Röhl

. (2001) Holocene rainfall variability in southern Chile: A marine record of latitudinal shifts of the Southern Westerlies. Earth and Planetary Science Letters 185(3–4): 369–382.

41.

Leduc

Herbert

Blanz

. (2010) Contrasting evolution of sea surface temperature in the Benguela upwelling system under natural and anthropogenic climate forcings. Geophysical Research Letters 37(20): L20705.

42.

Mabote

Rogers

Meadows

(1997) Sedimentology of terrigenous mud from the Orange River delta and the inner shelf off Namaqualand, South Africa. South African Geographical Journal 79(2): 108–114.

43.

Maher

(1972) Nomograms for computing 0.95 confidence limits of pollen data. Review of Palaeobotany and Palynology 13(2): 85–93.

44.

Martin

(1968) Pollen analysis of Groenvlei lake sediments, Knysna (South Africa). Review of Palaeobotany and Palynology 7(2): 107–144.

45.

Meadows

Baxter

(2001) Holocene vegetation history and palaeoenvironments at Klaarfontein Springs, Western Cape, South Africa. The Holocene 11(6): 699–706.

46.

Meadows

Sugden

(1991) A vegetation history of the last 14, 000 years on the Cederberg, south-western Cape Province. South African Journal of Science 87(1): 34–43.

47.

Meadows

Baxter

Adams

(1994) The late Holocene vegetation history of lowland fynbos, Verlorenvlei, southwestern Cape Province, South Africa. Historical Biology 9: 47–59.

48.

Meadows

Baxter

Parkington

(1996) Late Holocene environments at verlorenvlei, Western Cape Province, South Africa. Quaternary International 33: 81–95.

49.

Meadows

Chase

Seliane

(2010) Holocene palaeoenvironments of the Cederberg and Swartruggens mountains, Western Cape, South Africa: Pollen and stable isotope evidence from hyrax dung middens. Journal of Arid Environments 74(7): 786–793.

50.

Meadows

Dingle

Rogers

. (1997) Radiocarbon chronology of Namaqualand mudbelt sediments: Problems and prospects. South African Journal of Science 93(7): 321–327.

51.

Meadows

Rogers

Lee-Thorp

. (2002) Holocene geochronology of a continentalshelf mudbelt off southwestern Africa. The Holocene 12(1): 59–67.

52.

Metwally

Scott

Neumann

. (2014) Holocene palynology and palaeoenvironments in the Savanna Biome at Tswaing Crater, central South Africa. Palaeogeography, Palaeoclimatology, Palaeoecology 402: 125–135.

53.

Mucina

Rutherford

(2006) The Vegetation of South Africa, Lesotho and Swaziland. Pretoria: South African National Biodiversity Institute.

54.

Neumann

Botha

Scott

(2014) 18,000 years of grassland evolution in the summer rainfall region of South Africa: Evidence from Mahwaqa Mountain, KwaZulu-Natal. Vegetation History and Archaeobotany 23(6): 665–681.

55.

Neumann

Scott

Bamford

(2011) Climate change and human disturbance of fynbos vegetation during the late Holocene at Princess Vlei, Western Cape, South Africa. The Holocene 21(7): 1137–1149.

56.

Neumann

Scott

Bousman

. (2010) A Holocene sequence of vegetation change at Lake Eteza, coastal KwaZulu-Natal, South Africa. Review of Palaeobotany and Palynology 162(1): 39–53.

57.

Neumann

Stager

Scott

. (2008) Holocene vegetation and climate records from Lake Sibaya, KwaZulu-Natal (South Africa). Review of Palaeobotany and Palynology 152(3–4): 113–128.

58.

Norström

Neumann

Scott

. (2014) Late Quaternary vegetation dynamics and hydro-climate in the Drakensberg, South Africa. Quaternary Science Reviews 105: 48–65.

59.

Norström

Scott

Partridge

. (2009) Reconstruction of environmental and climate changes at Braamhoek wetland, eastern escarpment South Africa, during the last 16,000 years with emphasis on the Pleistocene–Holocene transition. Palaeogeography, Palaeoclimatology, Palaeoecology 271(3–4): 240–258.

60.

O’Connor

Bredenkamp

(1997) Grassland. In: Cowling

Richardson

Pierce

(eds) Vegetation of South Africa. Cambridge: Cambridge University Press, pp. 215–245.

61.

Partridge

Demenocal

Lorentz

. (1997) Orbital forcing of climate over South Africa: A 200,000-year rainfall record from the Pretoria saltpan. Quaternary Science Reviews 16(10): 1125–1133.

62.

Quick

Chase

Meadows

. (2011) A 19.5 kyr vegetation history from the central Cederberg Mountains, South Africa: Palynological evidence from rock hyrax middens. Palaeogeography, Palaeoclimatology, Palaeoecology 309(3–4): 253–270.

63.

Reason

CJC

Rouault

(2005) Links between the Antarctic Oscillation and winter rainfall over western South Africa. Geophysical Research Letters 32(7): L07705.

64.

Reimer

Bard

Bayliss

. (2013) IntCal13 and Marine13 radiocarbon age calibration curves 0–50,000 years cal BP. Radiocarbon 55(4): 1869–1887.

65.

Rogers

Bremner

(1991) The Benguela ecosystem; Part VII – Marine-geological aspects. Oceanography and Marine Biology: An Annual Review 29: 1–85.

66.

Rogers

Rau

(2006) Surficial sediments of the wave-dominated Orange River Delta and the adjacent continental margin off south-western Africa. African Journal of Marine Science 28(3–4): 511–524.

67.

Rossignol

(1962) Analyse pollinique de sédiments marins Quaternaires en Israël. II. Sédiments Pléistocènes. Pollen et Spores 4(1): 121–148.

68.

Schefuß

Kuhlmann

Mollenhauer

. (2011) Forcing of wet phases in southeast Africa over the past 17,000 years. Nature 480(7378): 509–512.

69.

Schneider

Bleil

Buschhoff

. (2003) Report and Preliminary Results of Meteor Cruise M57-1, Cape Town – Walvis Bay, 20.01.2003–08.02.2003. Bremen: University of Bremen.

70.

Scott

(1982) Late Quaternary fossil pollen grains from the Transvaal, South Africa. Review of Palaeobotany and Palynology 36(3–4): 241–278.

71.

Scott

(1987) Late quaternary forest history in Venda, Southern Africa. Review of Palaeobotany and Palynology 53(1–2): 1–10.

72.

Scott

Woodborne

(2007a) Pollen analysis and dating of Late Quaternary faecal deposits (hyraceum) in the Cederberg, Western Cape, South Africa. Review of Palaeobotany and Palynology 144(3–4): 123–134.

73.

Scott

Woodborne

(2007b) Vegetation history inferred from pollen in Late Quaternary faecal deposits (hyraceum) in the Cape winter-rain region and its bearing on past climates in South Africa. Quaternary Science Reviews 26(7): 941–953.

74.

Scott

Bousman

Nyakale

(2005) Holocene pollen from swamp, cave and hyrax dung deposits at Blydefontein (Kikvorsberge), Karoo, South Africa. Quaternary International 129(1): 49–59.

75.

Scott

Neumann

Brook

. (2012) Terrestrial fossil-pollen evidence of climate change during the last 26 thousand years in Southern Africa. Quaternary Science Reviews 32: 100–118.

76.

Shannon

Nelson

(1996) The Benguela: Large scale features and processes and system variability. In: Wefer

Berger

Siedler

. (eds) The South Atlantic: Present and Past Circulation. Berlin: Springer, pp. 163–210.

77.

Shaughnessy

(1986) A case study of some woody plant introductions to the Cape Town area. In: Macdonald

Kruger

Ferrar

(eds) The Ecology and Management of Biological Invasions in Southern Africa. Cape Town: Oxford University Press, pp. 37–44.

78.

Shi

Schneider

Beug

H-J

. (2001) Southeast trade wind variations during the last 135 kyr: Evidence from pollen spectra in eastern South Atlantic sediments. Earth and Planetary Science Letters 187(3–4): 311–321.

79.

Taylor

(2004) A trace element study of sediments from the Olifants River estuary, the Berg River estuary, and the off-shore Mud Belt. PhD Thesis, University of Cape Town.

80.

Truc

Chevalier

Favier

. (2013) Quantification of climate change for the last 20,000 years from Wonderkrater, South Africa: Implications for the long-term dynamics of the Intertropical Convergence Zone. Palaeogeography, Palaeoclimatology, Palaeoecology 386: 575–587.

81.

Tyson

(1986) Climatic Change and Variability in Southern Africa. New York: Oxford University Press.

82.

Tyson

Preston-Whyte

(2000) The Weather and Climate of Southern Africa. Cape Town: Oxford University Press.

83.

Tyson

Karlen

Holmgren

. (2000) The little ice age and Medieval warming in South Africa. South African Journal of Science 96(3): 121–126.

84.

Valsecchi

Chase

Slingsby

. (2013) A high resolution 15,600-year pollen and microcharcoal record from the Cederberg Mountains, South Africa. Palaeogeography, Palaeoclimatology, Palaeoecology 387: 6–16.

85.

Weldeab

Stuut

J-BW

Schneider

. (2013) Holocene climate variability in the Winter Rainfall Zone of South Africa. Climate of the Past Discussions 9: 2309–2356.

86.

Wheeler

(2010) Impacts of degradation on critically endangered Oudtshoorn Gannaveld. PhD Thesis, University of the Western Cape.

87.

White

(1983) The Vegetation of Africa (Natural Resources Research 20). Paris: United Nations Scientific and Cultural Organization.

88.

Zhao

Dupont

Meadows

. (2015) Pollen distribution in the marine surface sediments of the mudbelt along the west coast of South Africa. Quaternary International. DOI: 10.1016/j.quaint.2015.09.032.