Julia Lee-Thorp, Andrew Sillen and Nikolaas van der Merwe*

Archaeometry Research Unit, Department of Archaeology, University of Cape Town, Private Bag, Rondebosch 7700, South Africa

* also Depts of Anthropology and Earth & Planetary Sciences, Harvard University, Cambridge 02138, USA.

Swartkrans has seen the introduction of many new approaches in paleoanthropology, including cave taphonomy and use of fire and tools. Here we describe a set of tools for examining diet and habitat niche of hominids based directly on chemical tracers in the fossils themselves, which provide on first principles a different perspective to those based on comparative morphology of fossils, or circumstantial evidence from abundances of fauna or sediments. These approaches are particularly useful in karstic sites like Swartkrans which lack conventional stratigraphy, and where more than one hominid species is often found in the same Members. We measured carbon isotope (¹³C/¹²C), strontium isotope (⁸⁷Sr/⁸⁶Sr), and trace element (Sr/Ca) ratios in enamel and bone of hominids and associated fauna at Swartkrans. Using results from a large suite of fauna as a framework, we find that ¹³C/¹²C for both Paranthropus and Homo indicate the incorporation of a small but significant contribution from C₄ grass, probably in the form of grass-eating animals, while Sr/Ca values fall inbetween those typical for herbivores and carnivores. These results are most consistent with omnivory, providing a new view of Paranthropus diets and raising questions about the extent of niche separation between this species and Homo. Faunal ¹³C/¹²C values indicate the dominance of C₄ grassland at that time, consistent with interpretations derived from faunal abundances. ⁸⁷Sr/⁸⁶Sr in fossils hold promise as a habitat indicator since values in the environment and foodweb they are directly derived from the local geology with the exception of plants and animals living in or near the stream. These differences are reflected in the fossils, and thus may be used either as a riverine niche-indicator, or detection of immigrants from geologically different areas.

The foundation for reconstructing the diets of fossil taxa using stable carbon isotope (¹³C/¹²C) analysis lies in isotopic discrimination by plants during photosynthesis. In African savannas, trees, bushes, shrubs, and herbs (C₃ plants) discriminate more markedly against ¹³C during CO₂ fixation than do grasses and sedges (C₄ plants), resulting in distinctly lower ¹³C/¹²C ratios (expressed as d values in parts per thousand relative to the PDB standard) in C₃ plants than in C₄ plants. ¹³C/¹²C ratios in the tissues of herbivores reflect the relative amounts of C₃ and C₄ vegetation eaten, with some further fractionation, and in turn ¹³C/¹²C ratios in predators reflect the diets of their main prey. In studies of fossil ecosystems tooth enamel is used because it has been shown to preserve original isotopic signals most reliably. Small differences (2-3 ^o/_oo) between ¹³C/¹²C of modern and fossil animals with similar or known diets may exist. But this difference has been ascribed mainly to isotopic depletion of the entire modern ecosystem as a result of the "fossil fuel effect", and to a lesser extent to isotopic alteration of biogenic values; some workers now ignore diagenesis as a significant interference. As shown here, enamel of browsing herbivores is isotopically depleted while that of grazers is enriched, mixed-feeders fall between these extremes, and carnivores are similarly distinguished depending on the diets of their principal prey.

The d¹³C results for a range of species, from Members 1 and 2, are shown on the right. An age of about 1.8My is estimated for Mbr 1 based on biostratigraphic comparison with East African sites, and while Member 2 is younger, the age gap is not clear. Several last and first appearances distinguish the faunal assemblages of the two members. An unusually high number of primate taxa occurs in Mbr 1, and various hypotheses have been advanced about which of the predators might have been the main accumulating agent(s). Browsing and grazing herbivores show expected C₃=96 (depleted) and C₄-dominated (enriched) patterning in both units. Amongst the primates Papio robinsoni and P. Dinopithecus ingens (shown as Papio sp.) clearly ate C₃-based foods, while Theropithecus oswaldi relied somewhat variably on greater amounts of C₄ foods. Procavia (Hyrax), Paranthropus robustus, Homo and Panthera pardus (leopard) clearly all reflect carbon of mixed C₃ and C₄ origin. The false sabre-tooth, Dinofelis, relied more on C₄-based prey and it may have filled the same kind of niche as Panthera leo in Member 2. Furthermore, it was unlikely to have been a specialist hominid predator as Brain (1981) suggested. In Member 2, leopard diets shifted towards C₄ prey, lending support to Brain=92s proposal that leopards shifted their attention to Antidorcas bondi, a small grazing antelope common in Mbr 2 (included in the grazing class). d¹³C for Paranthropus diets did not change. Surprisingly values for Homo and Paranthropus are almost identical, suggesting that both hominids had a similar mix of C₃- and C₄-based foods in their diets (~75% and 25% respectively). The C₄ contribution may come from direct consumption of grasses, or from grass-eating animals.

Because relatively more calcium (Ca) than strontium (Sr) is absorbed by animals, bone Sr/Ca ratios are lower than those in the food consumed, a phenomenon known as =91biopurification of calcium=92. Hence Sr/Ca should be reduced in higher trophic levels. It is now recognised that considerable variability in Sr/Ca occurs within each trophic level, including variability within and between individual plants and plant parts. Roots, rhizomes and stems have elevated Sr/Ca compared to leaves, likely due to transport phenomena of Ca and Sr in plants. Hence leaf-eating herbivores have lower Sr/Ca in bone compared to herbivores which eat mainly stem material, and animals which eat underground material have significantly higher Sr/Ca. This complicates the trophic picture significantly, because carnivores have reduced Sr/Ca compared to their particular prey species. In this example from the Western Cape, where total Sr/Ca in bone has been analysed, one identified prey-predator pair is hyrax/leopard; the latter is clearly reduced relative to hyrax, but a leaf-eating herbivore (steenbok) overlaps the leopard.

In order to avoid problems of diagenesis in fossil bone material, Sr/Ca ratios were measured in a series of solution products of bone and buffered acetate, known as =91solubility profiles"; only data from washes 11-15 which demonstrate proportions of [Ca] and [PO4] typical of biogenic apatites are used. The scale between this figure and the previous one differs, nevertheless similar patterns are observed. Highest Sr/Ca occurs in Procavia (Hyrax) and Hystrix (Porcupine) as is the case for the modern foodweb. The two carnivores shown here, P. pardus and Hyena show reduced Sr/Ca compared to most of the other animals including the primates. T. cf. strepsiceros (kudu), a leaf-eating browser, is also low, but these animals are unlikely to have been prey of the leopard and hyena. Of the hominids, Paranthropus robustus has fairly low Sr/Ca, and the Homo individuals slightly higher ratios. The data are most consistent with an interpretation of omnivory. Not all the animal data is shown in this diagram. Taking the d¹³C and Sr/Ca data together, leopards are a likely predator of Procavia, Papio and Paranthropus, but not of Lepus and Theropithecus.

Strontium isotopes as a habitat indicator