The climate and vegetation backdrop to hominin evolution in Africa

Abstract

The most profound shift in the African hydroclimate of the last 1 million years occurred around 300 thousand years (ka) ago. This change in African hydroclimate is manifest as an east-west change in moisture balance that cannot be fully explained through linkages to high latitude climate systems. The east-west shift is, instead, probably driven by a shift in the tropical Walker Circulation related to sea surface temperature change driven by orbital forcing. Comparing records of past vegetation change, and hominin evolution and development, with this breakpoint in the climate system is challenging owing to the paucity of study sites available and uncertainties regarding the dating of records. Notwithstanding these uncertainties we find that, broadly speaking, both vegetation and hominins change around 300 ka. The vegetative backdrop suggests that relative abundance of vegetative resources shifted from western to eastern Africa, although resources would have persisted across the continent. The climatic and vegetation changes probably provided challenges for hominins and are broadly coincident with the appearance of Homo sapiens (ca 315 ka) and the emergence of Middle Stone Age technology. The concomitant changes in climate, vegetation and hominin evolution suggest that these factors are closely intertwined.

This article is part of the theme issue ‘Tropical forests in the deep human past’.

1. Introduction: climate-vegetation-evolution linkages

Climatic and vegetation change have long been invoked as drivers of the evolution of hominins in Africa [1–3], but the climate mechanisms that are important for driving change remain debated [4–9]. The link between climate and hominin evolution is founded on the principle that resources essential for the survival of hominins, such as water availability and food, fluctuated in tandem with climate parameters and, consequently, so did their dispersal and evolution [10,11]. While the configuration of resources required by different hominin species varied through the process of evolution [12], diverse vegetation mosaics, including different forest types, are thought to have been preferred habitats because of the greater range of vegetative resources available [13–15]. Furthermore, it is likely that changes in the spatial distribution of these complex vegetation mosaics drove evolutionary changes in hominin populations [5,16]. Through, for example, driving patterns of dispersal and mixing in, and between, hominin groups, or by presenting new selective pressures. Therefore, to understand any linkage between hominin evolution and climate it is important to first establish how climate change modified vegetation, and thus instigated transitions in resource availability within the landscapes in which the hominins lived. In this review, we focus on the last ca 1 million years (Myr) as this timeframe covers the divergence of Homo sapiens from its last common ancestor with Neanderthals and Denisovans [17–19], the emergence of complex culture ca 125 thousand years (ka) ago [20], and the dispersal out of Africa of the ancestors of all present-day non-African human populations ca 60–50 ka [17]. We first establish a framework of past climatic change within Africa in which we set out the dominant phases of pan-African climate change based on evidence from the available marine and terrestrial climate archives. We then compare this climate framework to patterns of vegetational change, and key features of hominin evolution and development. Through structuring our review around the well dated, long-term, continental scale, climate framework, and focusing on broad scale trends related to vegetation and hominins, we aim to negate some of the challenges presented by the paucity of study sites and uncertainty surrounding their dating. Therefore, this review should be regarded as one concerned with seeking overarching patterns rather than trying to identify specific drivers or causal relationships.

2. Pan-African climate history of the last 1 million years

The processes of evolution that ultimately led to the emergence of our species coincided with environmental change (e.g. global cooling and drying) over the course of the Plio-Pleistocene epochs (5.4 Ma to 11.7 ka ago). Several climatic mechanisms have been proposed to have mediated species turnover, and dispersion, of our ancestors in Africa [6,7,9]. One hypothesis suggests that Northern Hemisphere climate change during the Plio-Pleistocene led to a stepwise increase in African aridity, and thus facilitated species turnover (e.g. [6,11]). This would suggest African climate trajectories were linked to the gradual build-up of Northern Hemisphere continental ice sheets; however, this climate teleconnection has not been unequivocally substantiated either by climate or archaeological records within Africa [21]. Alternatively, recent findings strongly suggest that the multi-stage establishment of the tropical Walker Circulation (WC) could have been the dominant driver of long-term climatic change in Africa [8,22,23]. During the last 1 Myr terrestrial records from western and eastern Africa, as well as continental margin marine records, suggest a three-phase climate history linked to distinct changes in WC: (i) ca 1 Ma to 300 ka relatively wetter western Africa, (ii) ca 300 to 150–50 ka relatively wetter eastern Africa, and (iii) ca 150–50 to 0 ka relatively wetter western Africa (figure 1b).

Figure 1.

Comparison of between pan-African climate driver, vegetation changes and hominin evolution transitions during the last 1 Ma. (a) Orbital eccentricity (ecc; right) and precession (prec; left) [24]. (b) Detrended sea surface temperature gradient between eastern Pacific Ocean Drilling Program (ODP) Site 806 [25] and western Pacific Ocean ODP Site 846 [26]. Designation of palaeo-El Niño– and La Niña–like conditions follows [8,27]. (c) Spearman correlation coefficients for the months October to April from years 1891 to 2016 of the NINO3.4 index and the Global Precipitation Climatology Centre (GPCC V2018 land) precipitation data in a 0.5° grid. Blue shaded areas indicate positive correlation coefficients = humid conditions; red shaded areas indicate negative correlation coefficients = arid conditions. The spatial correlation shown is significant with p > 10%. Analysis and visualization: https://climexp.knmi.nl/start.cgi. Pie charts indicate proportion of samples within the time window indicating higher, or lower, abundance of grass than the average of the whole record: dark green = lower than average, mid-green = higher than average, and light green = barren samples (less than 10 pollen grains recovered); site Poaceae averages and proportions calculated from the following datasets (black dots): a, offshore Morocco, GIK 15672-3 [28]; b, Lake Bosumtwi, Ghana [29]; c, Lake Bambili, Cameroon [30]; d, off shore Gabon, GIK 16867 [31]; e, Congo fan, ODP 1075 [32]; f, Lake Magadi, Kenya [33]; g, Lake Malawi, Malawi [34]; h, offshore Mozambique, MD96-2048 [35]; i, Vankervelsvlei, South Africa [36]. Break point (mean and 2-sigma uncertainty) follows [8]. (d) Schematic diagram of the inferred age ranges of hominin lineages and tool development in Africa during the last million years [17,37,38]; MSA = Middle Stone Age. Photo credits in (d): Ryan Somma/Wikimedia Commons, [37], and José-Manuel Benito Álvarez and Vincent Mourre/Wikimedia Commons.

For the time interval between 1 Ma and 300 ka, marine dust records off the western African coast (Mauritania, ODP 659, 18°04.63' N, 21°01.57' W, 3069 m below sea level (m.b.s.l.) [39], and Liberia, ODP 663, 1°11.87' S, 11°52.71' W, 3716 m.b.s.l. [40]) indicate a lower median dust input into the subtropical Atlantic compared to the subsequent late Middle Pleistocene time frame. Relatively lower dust levels have been interpreted to indicate more humid conditions, suppressing dust production in the sub-Saharan hinterland [39–41]. Humidity change has, in turn, been linked to monsoonal changes, and thus precipitation changes, in western Africa [6,40–44]. By contrast, terrestrial and marine records tracing moisture changes in eastern and southeastern African regions indicate a different development. Wet–dry phases have been reconstructed on the basis of records tracing the river run-off of the Limpopo River (South Africa, 34°01.00' E, 26°10.00 N, 660 m.b.s.l. [45]), and the Nile (ODP967, 34°04.098' N, 32°43.523' E, 2705 m.b.s.l., [46]). These records indicate increasingly humid conditions from 1 Ma to 300 ka in eastern and southern Africa relative to the subsequent late Middle Pleistocene. These Middle Pleistocene results strongly suggest the existence of an east-west trending pan-African moisture gradient with a relatively wetter western Africa. However, lake level reconstructions from Lake Malawi (Malawi/Tanzania, 11°17′ S, 34°26′ E, 478 m above sea level (m.a.s.l.) [47]) show strong fluctuations between wet and dry phases without a clearly directional trend, highlighting the heterogeneous nature of change across landscapes and the potential for local factors to modulate broader climate signals.

Moving into the late Middle Pleistocene (ca 300 ka) the distinct east-west moisture gradient across Africa reverses, and eastern Africa becomes relatively wetter. Marine and terrestrial climate archives demonstrate a distinct climatic shift between 450 and 200 ka [8]. Specifically, Saharan (ODP 659 and 663) and Namib dust records (GeoB1028–5, 9°11.15′ E, 20°06.24' N, 2209 m.b.s.l. [48]), as well river run-off records offshore Congo (ODP 1075, 4^o 79′ S, 10^o 8′ E [49]), demonstrate a shift from previously wetter conditions to drier climate conditions during this time frame in western Africa. While in eastern Africa, lake level records (Ethiopia, Chew Bahir, 36° 05.00' E, 4° 01.00' N [50]; Kenya, Lake Magadi, 1^o54' S 36^o 14' E 606 m.a.s.l. [51]), and run-off reconstructions (along the Nile [46] and Limpopo River [45]), signal increasingly wetter conditions. A profound climatic shift has also been identified across Africa between 150 and 50 ka [8] when the east-west moisture gradient returns to its pre-300 ka state, leading to a relatively wetter western Africa. This shift is particularly traceable in the progressive drying trend captured in the eastern African Chew Bahir Lake level record (Ethiopia [50]) while dust records off the western African coast (ODP 659 and ODP 663), alongside lake level reconstructions from Lake Bosumtwi, signal an increase in moisture availability around the same time.

Changes in atmospheric pCO₂, and global ice volume, have been linked to African climate change through imposing meridional shifts and/or expansion in the tropical rainbelt (i.e. [7,45,52]). These changes are anticipated to produce a north-south pattern of climate change that mirrors the modern-day seasonal climate migration [53]. The observed changes in the east-west moisture gradient across Africa during the past ca 1 Ma therefore cannot be fully explained by these mechanisms [7,44,45,54–57]. Instead, insolation driven changes in the WC have recently been invoked as a climate mechanism capable of determining the climate gradient across Africa on geological time scales (i.e. [8,41,44]). The WC is an east-west trending band of atmospheric circulation cells along the equator that today exerts strong control on heat and moisture transport in the tropics [58,59]. WC variability is closely linked to sea-surface temperature changes across the Pacific, the Indian, and the Atlantic Oceans, and can have pronounced effects on the humidity regimes of the adjacent continents, including Africa [60–63]. Today the most prominent manifestation of WC variability is the El Niño Southern Oscillation, with its two end-members La Niña and El Niño [53,63] (figure 1c). WC variability therefore seems the most parsimonious explanation for changes in the moisture balance gradient across Africa at ca 300 ka and between 150 and 50 ka (figure 1b). Thus, we hypothesize that shifts in WC at these times drive vegetation change, and thus indirectly, hominin evolution, across Africa. We therefore anticipate that changes in WC, rather than Northern Hemisphere glaciation, provide a better climate framework to test concepts of vegetation and human evolution against. We next test our hypothesis by exploring evidence for vegetation and hominin evolutionary development within this climate change framework.

3. Vegetation change and hominins in Africa

Evidence for the vegetation conditions associated with hominin finds from the last 1 Myr have been directly obtained from the archaeological and palaeontological sites where fossils have been found. These datasets include palaeoenvironmental proxies from the East African Rift (e.g. [64,65]) and southern Africa (e.g. [66,67]). Find-specific studies allow the association of particular environmental conditions with particular stages in hominin development. However, they do not provide insights into the vegetation dynamics that were the backdrop to the evolutionary events that were unfolding. Evidence for vegetation change, or stability, through the last 1 Myr can be gleaned from a relatively small number of sedimentary records recovered from lakes [68] and the continental margins [69]. The terrestrial records provide the closest understanding of the vegetation within the landscapes that hominins inhabited, i.e. these large lakes (10–100s km in width-length) probably source pollen predominantly from greater than 100s m from the lake edge [70]; while the marine records generally provide a wider overview of change at a broader spatial scale, i.e. the pollen recovered from the marine sediments are probably sourced from 100s to 1000s km away [71]. However, the incomplete nature of both the vegetation and hominin records make direct inferences impossible. We therefore focus on reviewing the key evidence for vegetation dynamics and hominins (evolution and cultural development), and then relate these data back to the climate framework already established (figure 1b). Furthermore, owing to ambiguity, and the likely changing nature, of which vegetative resources, from where, were useful to hominins we choose to only discuss in the broadest possible terms ‘vegetative resource’ availability. While lacking in specificity this generalization is underpinned by an incontrovertible principle, i.e. that in an arid, desert like, landscape few vegetative resources would be available to hominins, while in a landscape containing a lush vegetation mosaic a wide range of resources would be potentially obtainable.

(a) . One million to 300 000 years ago

(i) . Vegetation change

Three terrestrial and three marine pollen records from Africa have so far yielded substantial insights into past vegetation change stretching back towards 1 Ma. Three of these records reflect changes in western and three in eastern Africa. These in general indicate a relatively more forested vegetation in western Africa during this period than eastern Africa (figure 1c).

The three pollen records related to western Africa are: (i) Lake Bosumtwi (Ghana, 6°30′ N, 1°24′ W, 97 m.a.s.l.) ca 540 000 years [29,72], (ii) marine core ODP Site 1075 (Congo Fan) ca 1.35 Ma to 600 ka [32,73], and (iii) marine core GIK 16867 (offshore Gabon; 2°12′ S, 5°6′ E) ca 700 ka [31,74]. The Lake Bosumtwi sediment record in total spans ca 1 Myr [75–77], however, the current pollen records span only the most recent ca 540 ka [72]. During this period, the composition of the woody vegetation is relatively consistent with Celtis and Moraceae being important taxa [72], and there is an overall greater degree of woody cover and taxonomic diversity relative to the period post-300 ka [29]. Within the period ca 540–300 ka some rapid (100–1000 year) fluctuations in vegetation openness have been tentatively attributed to variations in fire and herbivory within the landscape [78]. Fire and herbivory could have been modulated by hominins through burning and hunting, however, the degree to which this might have occurred remains unknown. The marine records receive pollen, via wind and water transport, from the western African coastline which includes rainforest, montane forest, open ecosystems and coastal mangrove swamps [74]. The southernmost, older, record was obtained from the Congo Fan and so probably contains a significant proportion of pollen sourced from within the Congo River catchment [73]. The Congo Fan record, between ca 1 million and 600 ka, shows large (±30%) fluctuations in the abundance of Podocarpus typical of Afromontane forests; while Poaceae (grass), indicative of more open environments, remains relatively stable (ca 10%), and is lower than earlier in the sequence [73]. The more northerly marine record indicates that from ca 700 ka the vegetation was mainly composed of Afromontane (Podocarpus) and lowland (rain)forest (Alchornea, Uoaca, Tetrorchidium and Celtis), and that vegetation associations varied in their major components between forest stages. In contrast to the Lake Bosumtwi record the abundance of grasses in the marine records is relatively low, and the transitions are not so dramatic. This lower abundance of the grasses is probably owing to the more southerly, more tropical, location of the marine cores, while the more muted changes in pollen abundances are probably owing to the marine pollen signal being ‘smoothed’ over a much larger area. Furthermore, the marine pollen signal is also biased towards the vegetation growing along the rivers. The relatively more wooded landscape during this period suggests enhanced water availability in western Africa could have provided vegetative resources useful to early hominins.

The three pollen records related to eastern Africa are: (i) Lake Magadi (Kenya) ca 1 Myr [33,47], (ii) Lake Malawi (Malawi/Tanzania) ca 600 000 years [34,79], and (iii) marine core MD96–2048 (Indian Ocean, offshore Mozambique; 26°10′ S, 34°01′ E) [35,80]. The Lake Magadi record [81,82] has yielded insights into past vegetation back as far as ca 1 Myr [33,51]. However, owing to poor preservation, significant temporal gaps exist which probably represent periods when the lake dried out, causing the pollen to degrade (barren samples), and thus represent the driest portions of the record (figure 1c). Vegetation change from ca 740 ka is interpretated as indicating progressive drying, culminating in an especially arid period ca 500–400 ka [33]. Complementary phytolith records of local (less than 100 m source area) vegetation change from the Koora Basin (ca 20 km east of Lake Magadi) demonstrate variance in the vegetation composition and dynamics at the landscape scale [16]. Heterogeneity of resource availability and distribution across the complex topography of the Rift Valley would probably have presented a range of opportunities for early hominins. Around 800 km south, the Lake Malawi pollen record spans the last ca 600 ka without a significant hiatus, and reveals shifts in the abundance of woody taxa [34]. The longest period of more open, grass-dominated, ecosystems was between ca 600 and 475 ka (abundant Poaceae). From ca 475 to 300 ka the abundance of grass pollen within the Lake Malawi record is relatively stable (ca 40%); however, there are significant changes in the woody taxa. The most notable shift in woody taxa occurs around 350 ka when the abundance of Podocarpus (Afromontane forest) declines, while Celtis (tropical seasonal forest) increases. Through this period, around Lake Malawi, the abundance of open ecosystems, miombo woodlands, tropical seasonal forests and Afromontane forests shifted; although elements of all persist throughout the record. The Indian Ocean record spans the last ca 800 ka and contains pollen representative of the vegetation of a portion of southeastern Africa derived from airborne and fluvial transport (Limpopo River) [80]. Between ca 800 and 300 ka, heathland (Cyperaceae and Ericaceae), mountain forest (Podocarpus, Celtis and Olea) and shrubland (Asteraceae and Poaceae) are major components of the vegetation, with the shrublands expanding at the expense of the heathlands after ca 600 ka [80]. In contrast to the Lake Malawi record, the marine record contains relatively low proportions of grass pollen [34,80]. This discrepancy probably reflects the differences in the source areas for the pollen, and suggests that more wooded (wetter) vegetation persisted along the watercourses (marine record), while Lake Malawi (and Lake Magadi to the north) were subject to greater aridification. It therefore seems likely that while relatively arid internal basins would probably have been challenging environments for hominins to inhabit, water courses may have provided a relatively more stable and resource rich habitat in during this period.

(ii) . Hominin evolution and cultural development

Neanderthal and Denisovan genomes have suggested that H. sapiens split from their last common ancestor at least between 700 and 500 ka, and possibly as far back as 1 Ma [17–19]. Whether variation in the abundance of vegetation resources drove the fragmentation of hominin populations during this time is unclear, as the fossil record is very fragmentary between the 1 Ma and 300 ka time frame (table 1). The hominin fossils do, however, demonstrate widespread dispersal; with high concentrations of finds to date in Morocco, Algeria, the eastern African Rift and southern Africa. Furthermore, archaeological sites featuring forms of Acheulean technology, which is associated with the period 1 Ma to 300 ka, are fairly widespread in Africa (figure 1d). The pan-African spread of hominins indicates that habitable landscapes were present across the continent during this period. The record of past vegetation change shows a dynamic and diverse vegetation within regions through time (figure 1c) despite the overarching climatic trends (figure 1b).

Table 1.

Key African fossils dating to between 1 Ma and 300 ka. (Ages are chronometric (radiometric or palaeomagnetism) unless otherwise indicated. Fossils marked in bold are not thought to be on the Homo sapiens lineage.)

region	site	age	remains	reference
north Africa	Air Maarouf	faunal associations suggest approximately 700 ka	left femur shaft	[83]
	Tighenif	faunal correlations suggest approximately 780 ka	various, including three mandibles	[84]
	Thomas Quarries	600–400 ka	various, including cranial fragments	[84]
	Jebel Irhoud	approximately 315 ka	various remains of different individuals inc. a cranium	[85]
	Kebibat	faunal correlation indicates Mid-Pleistocene	subadult fragmentary cranial vault + jaw	[86]
	Salé	approximately 400 ka	fragmentary skull	[83]
	Dar es Soltan II		immature calvaria, an adolescent mandible and the anterior part of a skull	[87]
	Al Aliya		maxilla and teeth	[88]
	Temara		vault fragments
eastern Africa	Buia	1 Ma	various, including near complete skull	[89]
	Daka	1–0.78 Ma	various, including partial skull	[90]
	Melka Kountouré	1.3–0.78 Ma	cranial fragment	[83]
	Olorgesaillie	970–900 ka	cranial fragments	[91]
	Olduvai, Lower Masek beds	970–780 ka	partial mandible	[83]
	Bodo	640–550 ka	various, including partial skull with face	[92]
	Kapthurin (Baringo)	543–509 ka	two mandibles	[83]
	Eliye Springs	undated cranium, ‘Middle Pleistocene’	cranium	[93,94]
	Ndutu	undated, possibly 500–300 ka based on artefacts	skull	[83]
southern Africa	Elandsfontein	approximately 600 ka based on faunal associations	skullcap and mandible fragment	[83]
	Florisbad	259 ka	cranium	[95]
	Cave of Hearths	undated, 400–200 ka based on faunal associations	right mandible fragment	[83]

More specifically, it is interesting to note that Acheulean sites are widespread in the Sahara region that is dominated by desert today, indicating that at least some hominin groups inhabited a ‘Green Sahara’ for at least part of this period. However, the Sahara region endured dramatic climate and vegetation fluctuations primarily influenced by glacial and interglacial cycles, unlike sub-Saharan Africa [43]. The wide geographical spread of hominin fossils, including presence in the Sahara, is perhaps linked to the relatively greater expanse of wetter, and thus vegetative resource rich, land in western Africa (figure 1c, phase 1). This, coupled with low population size, makes it seem possible that the morphological diversity observed in the limited hominin fossil record was, at least in part, driven by their spread, i.e. populations were geographically isolated, by climate and vegetation, which limited interbreeding [4,68].

(b) . Three hundred thousand to 50 000 years ago

(i) . Vegetation change

Eight pollen records (five terrestrial and two marine) are available that cover substantial portions of the period ca 300–50 ka. In addition to those introduced above these include two records with more affinity to western African climate (Lake Bambili, Cameroon, 6°30′ N, 10°15′ E, 2273 m.a.s.l., ca 90 ka [30,96]; offshore Morocco, GIK15627-3, 29°10′ N, 12^o05′ E, ca 250 ka) [28,97], and one eastern (Vankervelsvlei South Africa, 34°0′ S, 22°54′ E, 153 m.a.s.l., ca 110 ka [36,98]).

The two terrestrial pollen records, from western and central Africa (Lake Bosumtwi and Lake Bambili) show dramatic landscape scale vegetation change (forest-savannah) during the period ca 300–50 ka [29,96]. During the latter part of this period (90–50 ka), the Lake Bambili pollen record also shows significant changes in tree line, typified by a decline in the overall abundance of Podocarpus (Afromontane taxa) and palynological diversity, and a rise in Ericaceae [96]. The equatorial Atlantic marine record shows a progressive increase in Podocarpus pollen after ca 300 ka [74]. The record from offshore Morocco is the closest pollen record to the earliest H. sapiens fossils found to date (Jebel Iroud, Morocco; ca 315 ka) [85,99]. The offshore Moroccan pollen record is interpreted to be mainly comprised of wind transported pollen delivered by the northeasterly trade winds [97]. This record indicates a landscape containing desert (Chenopodiaceae), steppe (Stipia and Artemisia), ‘transitional’ forest (Agania, Acacia, and Euphorbia) and oak forest (Quercus) ecosystems, with the greatest expansion of woody vegetation occurring between ca 240–190 ka and ca 130–80 ka [97]. This suggests that during these periods there would have been relatively greater availability of vegetative resources within the region for H. sapiens. However, the long persistence of grass-dominated ecosystems and arid conditions around Lake Bosumtwi could have inhibited hominin activity in this area, i.e. few vegetative and water resources. However, it should be noted that vegetation resources along ecotones would have persisted in western Africa through this period although their location probably shifted further northwards relative to today [71]. Therefore, although some sub-regions could have been less hospitable for hominins, diverse vegetative resource would have been available within western Africa that could have offered opportunities for early H. sapiens.

The three terrestrial records that come from the eastern and southern portion of Africa (Lake Magadi, Lake Malawi and Vankervelsvlei) all record major fluctuations in vegetation cover during the period ca 300–50 ka. Around Lake Magadi, the closest record to early eastern Africa H. sapiens finds at Omo (table 2) an increase in the diversity/abundance of woodland taxa (e.g. Acacia and Commiphora) and a decrease in the diversity/abundance of Afromontane taxa (e.g. Podocarpus) is observed [51], which has been associated with the emergence of Middle Stone Age (MSA) technologies [33]. At Lake Malawi, the Afromontane vegetation (Podocarpus) reaches its maximum abundance and is generally abundant and persistent, however, its dominance is punctuated by five episodes where open ecosystems expand and contract (Poaceae pollen oscillating between 40 and 80%) [34]. For the latter part of this period, the Vankervelsvlei record shows a general rise in Fynbos taxa (Ericaceae) and decline in Afrotemparate taxa (Podocarpus). The vegetation change observed in these records do not occur in concert between records [68], although, it is interesting to note that the trend in the Afromontane forest abundance between Lake Magadi and Lake Malawi seems to be opposed, in line with expectations based on the spatial pattern anticipated by our WC climate model (figure 1c). The Indian Ocean record shows more frequent high amplitude fluctuations between wooded vegetation, and more open vegetation, than in the preceding period. Furthermore, a relatively lower proportion of the woody taxa are associated with montane forests associations than was found in the preceding period (lower Podocarpus and increased Alchornea, Spirostachys africana and Myrsine africana) [80]. This evidence suggests that the vegetative resources within this landscape were both diverse and dynamic through this period.

Table 2.

Key African fossils dating to between 300 and 50 ka. (Ages are chronometric (radiometric or palaeomagnetism) unless otherwise indicated. Fossils marked in bold are not Homo sapiens.)

region	site	age	remains	reference
south Africa	Florisbad	259 ka	cranium	[95]
	Klasies River Mouth	115–160 ka	various remains of mandibular, maxillary, facial, cranial vault and postcranial fragments	[83]
	Border Cave	74 ka	various fragmentary remains including partial skull found in spoil heap	[101,102]
	Die Kelders I	75–65 ka	isolated teeth and phalanges	[83]
	Blombos Cave	approximately 75 ka	teeth and teeth fragments	[103]
north Africa	Dar es Soltan II	85–75 ka	immature calvaria, an adolescent mandible and the anterior part of a skull	[87]
	Al Aliya	undated, possibly MIS 5	maxilla and teeth	[88]
	Haua Fteah	73–65 ka	fragmentary mandibles	[104]
	Taramsa Hill	as old as 69 ka but possibly younger/intrusive burial	fragmentary child's body, skull cap	[105]
	El Harhoura I	>40 ka	mandible and tooth	[83]
	Singa	>135 ka	calvaria, secondary context	[106,107]
eastern Africa	Kabwe	299 ± 25 ka	near complete skull and fragmented post-crania	[108]
	Omo 1	155–187 ka, 233 ± 22 ka	partial skeleton in member 1 of the formation	[109,124]
	Omo 2	not directly dated, assumed to be the same age as Omo 1	surface calvaria	[37,109–111]
	Omo 3		frontal fragment from member III	[110]
	Guomde KNM-ER 999 Guomde	300–100 ka	proximal femur fragment	[112,113]
	KNM-ER 3884	>180 ka	partial skull	[113–115]
	Herto	160	several dental and cranial remains	[115]
	Ngaloba	approximately 120 ka	partial cranium	[83,116]
	Mumba	130–109 ka	three isolated molars	[83]
	Aduma	105–70 ka	fragments of four adult skulls	[117]

(ii) . Hominin evolution and cultural development

Around 300 ka, the first physical and behavioural manifestations of H. sapiens begin to appear (figure 1d). There is some debate over which traits should, and should not, be used to ‘diagnose’ very early H. sapiens fossils, an issue exacerbated by the fact that there is no agreement on what early members of the H. sapiens clade should look like. Extant H. sapiens share species specific traits, such as cranial globularity, a small face retracted under the frontal bone, a chin and features of pelvic shape and the inner ear [93]. However, these features do not appear together in the same individuals until sometime between 100 and 40 ka [4,100]. This indicates that, owing to the diverse evolutionary processes involved, all the behavioural and physical traits that define us today did not appear as a package, and therefore were did not appear at the same time or place [4,17]. Furthermore, although the early H. sapiens fossils found to date are typically associated with open ecosystems (table 2), given the diverse and dynamic nature of vegetation across Africa (figure 1c), it seems likely that early populations experienced, and exploited, a range of different environments. This is perhaps reflected in the technological shift (shaping to core flaking), and the increasing diversity of tools, that emerges with the MSA during this period as H. sapiens equipped themselves to deal with a range of environmental challenges.

From the Last Interglacial (ca 125 ka) onwards, the MSA (figure 1d) based material culture record evidences new features, including the regional diversification of technology, the likely use of ornamental pigments, the collection and intentional perforation of shells (interpreted as for personal ornamentation), and the use of diverse bone tools (e.g. [4,118]). These manifestations of a seemingly more complex material culture are also pan-African, and do not seem to emerge from one region and spread. Various studies have attempted to make sense of this apparent shift in the archaeological record. While older theories have tended to focus on the possibility of unexplained brain mutations, more recent studies suggest that demographic and climate factors probably play a role. For example, climate and vegetative resource change can regulate population size and density which can, in turn, structure changes in cultural innovations and their persistence [119,120]. Similarly, during the MSA in southern Africa innovation in the region has been linked to the formation of refugial populations that emerged in response to increased rainfall and associated vegetative change [121]. A key point is that while these innovations appear in the MSA archaeological record and persist, often for thousands of years, they also disappear for even longer periods of time. This may be linked to a saw tooth pattern of population growth and collapse, fragmentation and coalescence (e.g. [121]). At the same time, the archaeological record begins to evidence the emergence of behaviours linked with ecological managements, ranging from water storage technology [122] to the widespread burning of landscapes [123], which perhaps suggests that humans began to increase the diversity of ecosystems they exploited.

4. Conclusion

The major climatic transition of the last million years in Africa occurred ca 300 ka (figure 1b). This climate shift, driven by a change in the WC, is manifested over Africa as a change in hydroclimate, with western Africa becoming relatively drier, and eastern Africa relatively wetter. Assessing the records of past vegetation change and hominin evolution and development from Africa against this framework is challenging owing to the sparce, fragmentary and complex nature of the available data. The temporal comparison of the different data types is probably the greatest current challenge in furthering our understanding of hominin evolution. The challenge primarily persists owing to the uncertainties in dating approaches and the large spatial and temporal gaps in the records, both of which require significant further research to move our understanding forward. In the light of these current constraints, we consider here simply the two main modes of climate (pre- and post-300 ka) and identify broad changes related to vegetation and hominins around this breakpoint.

In concert with the shifting moisture balance across Africa ca 300 ka, the relative proportions of open ecosystems (abundance of grass) shifts at many sites (figure 1c); with the topographically more complex landscape of eastern Africa increasing in relative abundance, and diversity, of vegetative resources after this time. These shifts probably created new selective pressures and challenges for hominins. We note that the geographical shift in the relative availability vegetative resources across the continent is broadly concomitant with the end of the long phase of Homo erectus and the appearance of new species of Homo and the development of MSA technology. It is also interesting to note that forest phases are recorded around the time of the earliest H. sapiens fossils found in northern Africa (ca 315 ka), but that these phases subsequently diminish. The diverse and widespread nature of the evidence from the archaeological record suggest that the process of evolution throughout this period probably occurred through multiple phases of separation and recombination of populations on the human lineage. This hypothesis is supported by the record of past vegetation change which suggests that, while some regions might become more or less preferable through time, that mosaics of vegetation persisted across the continent that would have provided opportunities for hominins to obtain diverse vegetative resources. However, it remains ambiguous which, and how many, diverse resources would have been used, and how this might have changed through time.

Unravelling the relationship between climate, vegetation and hominin evolution remains a challenge owing to the paucity of data. However, the emergent patterns suggest that these are strongly intertwined, and that a low latitude climate mechanism which changed rainfall, and consequently vegetation, patterns probably had a significant impact on the core habitation areas, population structure, and the evolution of our ancestors. The widespread distribution of hominins across a variety of ecosystems and climates suggests that the full diversity of Africa's varied landscape—from semi-arid regions to tropical forests—should be considered as potentially playing a role in the evolutionary and developmental processes.

Data accessibility

This article has no additional data.

Authors' contributions

W.D.G.: conceptualization, data curation, formal analysis, writing—original draft, writing—review and editing; E.M.L.S.: conceptualization, writing—original draft, writing—review and editing; S.K.-B.: conceptualization, data curation, formal analysis, writing—original draft, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Competing interests

We declare we have no competing interests.

Funding

E.M.L.S. thanks the Max Planck Society for funding. S.K.-B. received funding from an Open Topic Postdoctoral fellowship from the University of Potsdam and acknowledges funding from the Deutsche Forschungsgemeinschaft (DFG) through grant no. KA 4757/3-1.