A great deal of interest has been shown in issues related to the decline of large-scale forests and opening up of the African landscape. Some years ago, the "Savanna hypothesis'' held that forest shrinkage/savanna expansion in Africa was a primary driver of hominin bipedalism because it was thought that these two trends occurred at about the same time. This hypothesis fell from favor once it became apparent that bipedalism emerged before 5 Ma among relatively wooded habitats (WoldeGabriel et al. 2001). Nevertheless, there have undoubtedly been important and influential changes in the vegetation structures across Africa. Vrba's turnover-pulse hypothesis relied on the observed radiation of open-country grazing bovids to draw inferences about shifts to more open landscapes in the Pliocene and Pleistocene, which could be linked to hominin evolutionary changes (Vrba 1980, 1985, 1988). This approach is not entirely straightforward because the habitat preferences of extinct bovids, or indeed of any extinct animals, are not always clear. Vrba proposed that a shift to bovid lineages with open-country preferences between about 2.4 and 2.6 Ma was associated with the onset of Northern Hemisphere glaciation observed from marine oxygen isotopes in marine sediment core records (Shackleton et al. 1984). But the nature of the climate trigger is not clear. The idea of a pulse in faunal change has been challenged, although it is acknowledged that faunal changes occurred between 2 and 3 Ma (Behrensmeyer et al. 1997; Bobe et al. 2002).
Further perspectives on the appearance of savanna landscapes have emerged from several isotope studies. Although C4 grasses appeared in the Late Miocene and persisted with fluctuations through the Pliocene (Cerling et al. 1988; Cerling 1992; Kingston et al. 1994; WoldeGabriel et al. 2000; Wynn, 2004), importantly, they did not become a consistently dominant part of the floral biomass until ~1.8 Ma (© Figure 9.2). A shift in 818O also occurs at about 1.8 Ma, suggesting that the source area and vapor transport pathway changed (O Figure 9.2) (Cerling
Isotopic indicators of open, grassy environments from the Koobi Fora Basin and from the South African hominin sites, showing small but variable presence of grasses prior to 1.8 Ma, and a distinctive shift to open habitats after this time. The sequences are as follows: (a) pedogenic carbonate 813C, (b) 81SO from the Koobi Fora Basin, shown alongside, and (c) calculations for the percentages of browsers and grazers from Makapansgat and Sterkfontein. The arrows provide some guidance about the predicted ranges for pure C3 and C4 for (a) and (c), and some pointers to interpretations of the record in (b). Data for (a) and (b) are from Cerling et al. (1988), and they are expressed relative to an age model calculated relative to depth data in the same source. The data for (c) comes from Sponheimer and Lee-Thorp (2003), and Luyt and Lee-Thorp (2003)
et al. 1988). The 813C and 818O sequences for the Turkana basin shown in O Figure 9.2 also illustrate the variability of the results for each period, indicating a rather broad view of vegetation structure. Additional data sets from a wider range of regions and periods have been completed (Kingston et al. 1994; Sikes et al. 1999; WoldeGabriel et al. 2000). These data show a rather complex and variable overall mix of wooded terrain that tends to deemphasize the contributions of C4 grass to the overall mix.
Isotopic studies of fossil faunas, on the other hand, can be designed to investigate the various components of ancient flora. A number of isotopic data sets are now available for Pliocene and Pleistocene faunas that demonstrate presence or absence of C4 grasses in East and South Africa, (Morgan et al. 1994; Cerling et al. 1997; Sponheimer and Lee-Thorp 1999; Lee-Thorp et al. 2000; van der Merwe et al. 2003), and in Chad (Zazzo et al. 2000). In the Chad case, the faunal isotopic data demonstrated significant grassy vegetation from the Late Miocene through the Pliocene (Zazzo et al. 2000). In East Africa, faunal isotope data have in some cases run counter to the perspective afforded by pedogenic studies. At Olduvai, for instance, the isotope data from fauna suggest a greater C4 biomass (van der Merwe et al. 1999) than suggested by the soil carbonates. Another contradiction occurs in the Awash of Ethiopia, where de Heinzelin et al. (1999) suggest a wooded environment at ~2.5 Ma based on presence of certain key fauna (Colobus and Tragelaphus) and soil isotope values, but faunal isotope analyses suggest a significant proportion of C4 grazers (Levin et al. 2004). These discordances suggest that different parts of the ecosystem are being sampled. The advantage of faunal isotope data is that they are abundant in the actual sites associated with hominin activity, and because we can follow first principles to establish abundance of C4 without assumptions about dietary preferences.
No pedogenic carbonate isotope record exists for the regions associated with the South African hominin sites, and the sites are not as old as many in East Africa. The nature of the karstic infill sites present serious challenges for determining the nature of long-term vegetation change. Faunal d13C data from all the more important sites analyzed to date (Makapansgat, Sterkfontein, Swartkrans, Kromdraai) have shown that C4 grasses were present from the earliest periods associated with hominins, about 4 Ma or earlier, onward (Lee-Thorp and van der Merwe 1987; Sponheimer et al. 1999; van der Merwe et al. 2003; Hopley et al. 2006). On their own, these data demonstrate presence of C4 but are not very informative about changes in proportions of woody or grassy cover. A "C3/C4 index'' has been developed, which is essentially an isotopic expression of the "alcelaphine + antilopine'' index developed by Vrba (1980); it is based on the proportions of genera (Sponheimer and Lee-Thorp 2003), or individual specimens (Luyt and Lee-Thorp 2003), falling into one of the grazer, mixed feeder or browser categories as determined by d13C. The underlying idea is that browsers will be favored in a closed habitat with many trees, while grazers will be favored in open, grassy landscapes (Sponheimer and Lee-Thorp 2003). When applied to a series of sites to construct a sequential view, the results suggest that the most significant shift toward open, grassy landscapes occurred about 1.6-1.8 Ma (O Figure 9.2) rather than at ~2.4-2.6 Ma.
Isotopic data from the Buffalo Cave speleothem, in the Makapansgat Valley, provides more detailed information about vegetation change in this important time period. The speleothem is dated by independent means to between 2 and 1.5 Ma (Hopley 2004). The detailed 813C sequence shows cyclical fluctuations in the proportions of C4 grass, with a dominant variability at ~40 Ka, indicating that the major control is the angle of the Earth as it orbits the Sun, and known as the obliquity cycle. Precessional cycles clustered around 20 Ka, due to the Earth's "wobble", are also visible. This sequence shows a ~1.7 Ma shift to higher proportions of C4 as before, but importantly, that this was but part of a cyclical pattern. The entire speleothem shows a slight long-term trend toward grassier conditions. Variability in the d18O (reflecting rainfall) seems to be most strongly controlled by precessional cycles, in good agreement with those shown for the Tswaing Crater Lake Late Pleistocene sequence (Partridge et al. 1997), and for some Pliocene East African lakes (Trauth et al. 2005). Other stalagmite isotope data from Botswana (Holmgren et al. 1995) and Cold Air Cave in the Makapansgat Valley (Holmgren et al. 2003; Lee-Thorp 2004) hint at considerable floral and rainfall fluctuations but these records are too intermittent or too short to demonstrate precessional cycles. Although both 813C (reflecting vegetation) and d18O (reflecting rainfall) in the Buffalo cave speleothem are orbitally controlled, the differences in dominant orbital forcing modes between them emphasize the complexities of climate and environment, and the need for multiple sources of evidence.
The shift to grassier conditions in South Africa about 1.6-1.8 Ma is entirely concordant with the shift to open ecosystems in East Africa at the same time. This might be one occasion in which environmental shifts are in phase in East and South Africa.
These changes in floral composition, which can be documented using iso-topic tools, are important for evaluating competing hypotheses about links between environment and hominin evolution. What does the floral information tell us? For one, the broad correspondence between the emergence (first appearance) of C4 grasses and hominin bipedalism still holds, in spite of observations that locally, environments may have remained relatively closed. The underlying connection might be dietary rather than locomotor. 813C data on hominin diets have, almost without exception, shown an involvement with C4-derived foods
(see Sponheimer and Lee-Thorp Vol. 1, Chapter 17, Lee-Thorp et al. 2003). The implication is that early hominins chose to make use of these new food sources, even if C4 grass remained a relatively minor component of the ecosystem. We need to test how far back in time this dietary flexibility goes.
There is widespread agreement on the appearance of more open, grassy habitats around 1.6-1.8 Ma. Given the increasing evidence for presence of a fully bipedal hominin equipped with more sophisticated stone tools (including handaxes) from about this time, it is tempting to draw links between these two occurrences. In this case, we can suggest with some confidence that there is a real environmental "shift," but the exact nature of linkages to hominin behavioral and evolutionary shifts still require further investigation. Finally, the evidence for cyclical floral and moisture changes from the speleothem data, if upheld, requires reappraisal of several interpretations we have made. These have to do with the issue of time-scale and the nature of the fossil record. Most of the existing data rely on material that can, in principle, reflect small time-windows but the chronologies with which we have to work are quite gross. As a result, what emerges is rather a lot of noise and only the really large-scale environmental trends. The speleothem data also forcibly raise the issues of variability and cyclicity, which are crucial to the "variability selection'' hypothesis advanced by Potts (1996, 1998).
Was this article helpful?