Abstract: |
Stable carbon and nitrogen isotope measurements of bone collagen, together with carbon isotope measurements of bone mineral, were compared for two sets of prehistoric human skeletons from different parts of the South African coastline. One set was from the southern Cake, where terrestrial vegetation includes both C3 and C4 species. The second set was from the south-western Cape, a region with predominantly C3 terrestrial flora. One therefore expects carbon isotopes to discriminate between marine and terrestrial foods in the western, but not in the southern Cape. The southern Cape is well-watered, receiving around 1000 mm of rain per year. The western Cape is relatively arid with, in parts, less than 400 mm per year. Hence [15]N[14]N ratios should discriminate between marine and terrestrial foods in the southern but not the western Cape. In fact, 13C/12C and [15]N[14]N ratios correlated well with one another in both areas. The explanation probably lies in the nature of peoples' diets, and the likelihood that their dietary selection did not simply reflect the isotopic patterns of their environments. [13]C/[12]C analyses of bone mineral were compared with bone collagen, and the differences between the two considered in relation to the differential channelling of components of the diet to particular body tissues. Keywords: Carbon, Coastal diets, Later stone age, Nitrogen, Stable isotopes INTRODUCTION One of the more successful applications of stable isotope analysis for dietary reconstruction has been the study of marine versus terrestrial food consumption by coastal populations (e.g. Tauber, 1981
Chisholm et al., 1982, 1983
Hobson and Collier, 1984
Walker and DeNiro, 1986
Keegan and DeNiro, 1988
Yesner, 1988
Lubell et al., 1994). In this paper, I present 13C/12C and [15]N/[14]N measurements of bone collagen and 13C/12C measurements of bone mineral from two series of Holocene human skeletons from South Africa. All are from coastal or near-coastal sites, and the degree of reliance upon marine food is a question of archaeological significance. Most of the skeletons are those of hunter-gatherers, although in the last 2000 years sheep and later cattle were kept by some groups, so that the most recent skeletons may be the remains of pastoralists. Agriculture was not practised in this region until after European colonisation in the 17th century. Skeletons more recent than 400 BP (uncalibrated radiocarbon dates on bone collagen) have been excluded from this study. Skeletons are drawn from two areas: first, the southern Cape coast between the town of George and the Tsitsikamma National Park, a distance of about 130 km (Fig. 1). The second area is the stretch of south-western Cape coast extending from Cape Town about 200km northwards to Elands Bay. Results for the southern Cape are reported here for the first time. The bulk of the results for the southwestern Cape coast have been published previously (Sealy et al., 1987
Sealy and van der Merwe, 1988
Lee-Thorp et al., 1989), although the data set presented here is slightly enlarged. Comparison of the two sets of results reveals some aspects of the use of isotopes for dietary reconstruction which are not generally apparent. The focus in this paper is on the relationships between different dietary indicators delta[13] Ccollagen, delta[15] Ncollagen and delta[13] Capatite), both within and between the two regions. ISOTOPIC PATTERNING IN THE TWO REGIONS Southern Cape The section of the southern Cape under discussion here is a temperate, well-watered area. Rainfall varies from about 600 to more than 1200 mm per annum, depending on topography, and falls year-round, although rather more rain is received in summer than winter (Weather Bureau, 1950). Parts of this region are covered with dense forest, other parts with thick scrub and bush. Grass species include many C4 varieties (Vogel et al., 1978). Recent human activities, especially clearing land for agriculture and wood-cutting in the forests, have had a substantial impact on the ecology. Although the prehistoric vegetation was qualitatively similar to that in the area today, there is debate about the areal extent of different vegetation types in the past. Analyses of pollen from cores from near-coastal peat deposits in the George district (Fig. 1) have shown an increase in grass pollen c. 2600 to 1400 BP, compared with pollen reflecting more forested environments both before and after this period (Scholtz, 1986). In order to monitor possible variations in terrestrial isotopic ecology through the Holocene, a series of archaeological animal bones from Nelson Bay Cave was analysed. The animals chosen were buffalo (Syncerus caffer) and hippotragine antelope, including roan (Hippotragus equinus) and the now-extinct bluebuck (Hippotragus leucophaeus). Hippotragines occur mostly in the late Pleistocene and early Holocene levels (Inskeep, 1987). These species were predominantly grazers, so the carbon isotope values of their bones are determined by the proportions of C3 and C4 grasses in their diets. [15]N/[14]N analyses of the same specimens serve as a check on the consistency of nitrogen isotope values through the Holocene. For all animals from the last 11 000 years, mean delta[13] Ccollagen was -12.2 d: 1.9 %0 (n = 32). These results demonstrate the presence of a large proportion of C4 grass: they are closer to values obtained for grazers in pure Ca grasslands (c. -8 to -6%0: Vogel, 1978
Ambrose and DeNiro, 1986a
LeeThorp, 1989) than to those of C3 feeders (c. -21%o). Nitrogen isotope values were uniformly low (i.e. below 10%o): mean delta 15 Ncollagen is 5.0-t-1.2 %0. These values are typical of terrestrial animals from well-watered environments (Heaton et al., 1986
Sealy et al., 1987). Details of these results are presented elsewhere (Sealy, 1996). It seems, from these analyses, that there were significant quantities of Ca grass in this area throughout the Holocene. Since marine foods are, on average, also enriched in [13]C, [13]C/[12]C measurements of human bone will not clearly differentiate marine and terrestrial foods in the diet. Most marine food items are enriched in [15]N (Schoeninger and DeNiro, 1984
Sealy et al., 1987), so that nitrogen isotope ratios will distinguish between marine and terrestrial foods in this region. South-Western Cape The isotopic ecology of the south-western Cape has been described at length elsewhere (Sealy, 1986
Sealy and van der Merwe, 1986, 1992). The terrestrial flora is predominantly C3, so that the terrestrial environment is depleted in [13]C relative to the marine system. Some terrestrial C4 grasses are present. These are enriched in [13]C, and if present in large quantities will blur the distinction in carbon isotope ratios between terrestrial and marine systems. It has been suggested that the Ca component has been underestimated (Parkington, 1991), but the rarity of grazing animals in both contemporary and archaeological faunas attest to vegetation with relatively little grass (Sealy and van der Merwe, 1992). [13]C/[12]C ratios in this region differ in terrestrial and marine systems, and carbon isotopes can be used to distinguish terrestrial from marine food intake by prehistoric people. The nitrogen isotope ecology is complicated. The southern part of the area receives more than 400 mm of rain per year, and nitrogen isotope ratios of terrestrial animals here are below 10%o. The mean delta15N value for eight animals from the Cape Peninsula and surrounds (various species, excluding carnivores) was 4.2 + 1.2%o. North of Saldanha Bay rainfall decreases to below 400 mm and the nitrogen isotope ratios of terrestrial animals rise above 10%o, into the range shown by marine organisms. The mean delta 15N for twelve animals from Saldanha Bay northwards to Elands Bay was 13.6 d- 1.8%o (see Sealy et al., 1987). In this area, therefore, nitrogen isotopes do not reliably differentiate marine and terrestrial systems. In summary, therefore, in the southern Cape nitrogen isotopes are expected to be the best discriminant of marine and terrestrial foods. In the south-western Cape, carbon isotopes are the most reliable indicator. RESULTS AND DISCUSSION Collagen Southern Cape delta 15Nconagen values for the southern Cape skeletons ranged from 8.3 to 17.9%o, and delta 13Ccollagen from -11.1 to -17.9%o (Table I). delta 15Ncollagen values below 10%o indicate little, if any consumption of marine foods (Schoeninger et al., 1983). Values above 10%o result from seafood, and the most enriched values reflect consumption of a great deal of marine food. delta 13Ccollagen results are influenced both by consumption of marine food and terrestrial foods derived from Ca grasses, as discussed above. Fig. 2 shows delta 15 Ncollagen plotted against delta 13Ccollagen. The fitted regression equation is delta 15N = 27.96 + 1.08 delta 13C, with r2 -- 0.49 (n = 80). A hypothesis test of no 'correlation between the two variables was carried out and the hypothesis rejected at the 1% level (Zar, 1984: 268). The extent of the correlation was somewhat unexpected, given the isotopic patterning outlined above. How close would we expect the correlation to be in an 'ideal' situation, in which coastal humans ate a mixed marine/terrestrial diet without isotopic 'complications' due to arid terrestrial environments or terrestrial C4 flora? Figure 3 shows delta 5N plotted against delta 13C for 61 skeletons from the literature, together with some unpublished results. They come, as far as it is possible to ascertain, from environments with low terrestrial delta 15N values and C3 flora. It would obviously be best if these comparative data all came from a single area
unfortunately such a set is not available, so results from several areas and time periods have been combined. Most of the scatter on this plot occurs at the lower left-hand side, where individuals with depleted delta 13C values fall into two groups: some have low delta 15N values (below 10%o) while others are more positive. These clusters are formed partly by Neolithic (low delta 15N) and Mesolithic (higher delta 15N) skeletons from Portugal (Lubell et al., 1994)
the variation thus represents real spread in the relationship of delta 15N to delta 13C in coastal populations, and is not a function of the diverse nature of the sample. The fitted regression equation is /515N=27.92+ 0.84 delta 13C, with r2--0.72 (n=61). A hypothesis test of no correlation between the two variables was carried out and the hypothesis rejected at the 1% level. The reason for the unexpectedly close correlation between delta 15 Ncollagen and delta 13Ccollagen for the human skeletons in the southern Cape probably has to do with the nature of the diets which people chose to eat, which did not necessarily mirror the isotopic composition of the environment in which they were living. In the southern Cape, we expect that nitrogen isotopes will provide a reliable indicator of marine foods consumed. We anticipate that carbon isotopes will not provide such information, because of the presence of terrestrial Ca grasses which mimic the marine carbon isotope signal. But how would grasses have found their way into human diets? In Namibia and the Northern Cape region of South Africa, 1500km or more north of the southern Cape, there are ethnohistoric records of hunter-gatherers and contemporary pastoralists, who still collect wild foods, eating grass seeds. Ants collect seeds and store them in underground caches, which people rob(bed) for their useful concentrations of nutritious plant material. Among the Dama of Namibia, the seeds are then ground to a flour and boiled with water into porridge. They may also be used in brewing beer (Steyn and Du Pisani, 1984/5). Ants' or termites' eggs were also eaten (Schapera, 1930). These northern regions are, however, much more arid than the southern Cape, with a correspondingly lower capacity to provide food for hunter-gatherers. There is no indication, from the archaeology of the southern Cape, that these practices were important there. The principal route through which Ca grasses contributed to the diets of humans in the southern Cape is likely to have been through the meat of grazing animals. Grazers are a noticeable, though not a dominant component of the faunas from archaeological sites in this region. The best-documented archaeological sequence is that from Nelson Bay Cave, which contains deposits from the late Pleistocene and the whole of the Holocene (Inskeep, 1987). Faunal lists for the Holocene levels of Nelson Bay Cave include the buffalo and hippotragine antelope discussed above. Other species were, however, mostly browsers, especially the small antelope and hyrax which are a recurrent feature of later Holocene faunal assemblages from cave sites all over South Africa, In the early Holocene grazers were indubitably more common, but after c. 8000BP most of the meat eaten may have come from browsing, rather than grazing animals. In spite of their presence, C4 grasses probably did not make a major contribution to human diets. Skeletons which date to between 11 000 and 8000BP are marked as plus signs on Fig. 2
they do not cluster towards the lower, right-hand side of the group, as would be expected if they really reflected a larger Ca grass component. As mentioned above, skeletons yielding radiocarbon dates more recent than 400BP have been excluded from this study (only two). Three skeletons date to the second millennium AD
they are plotted in Fig. 2 as filled triangles and cluster towards the lower right-hand side of the plot, including the two outliers. Interpretation of this phenomenon should be approached with caution, since the sample is so small. One of the skeletons (NMB 1704) was coated with preservative
efforts were made to remove this in the laboratory, but it is difficult to know whether these were entirely successful. Results for the other two skeletons, however, were not compromised in this way. This part of the plot (positive delta 3C but low 615N values) almost certainly reflects increased intake of terrestrial Ca foods. Where did these come from? At this time, and indeed during the first millennium AD, people began to keep sheep and cattle in the southern Cape, and Iron Age farmers about 600 km to the northeast were growing Ca crops: millets and sorghum. Could the most recent southern Cape individuals have been pastoralists, with Ca input into their diets from heavy reliance on animal products from sheep and/or cattle? The nitrogen isotope values are towards the lower end of the range of this set of analyses: 9.3 and 10.7 for the outliers, and 11.4 for the point on the edge of the main group. These values are similar to those reported for pastoralist/farmers from East Africa (Ambrose, 1986
Ambrose and DeNiro, 1986b). Mean delta 15 N values for Kikuyu male and female skeletons were 10.4 and 10.3 respectively, for Kalenjin 10.4, for Pokot 14.2, for savannah pastoral Neolithic 12.6, and for Elmenteitan skeletons 11. These East African people do not, however, have access to seafood. It seems unlikely that the three Cape coastal individuals all avoided seafood, although they died (and probably lived) close to the coast. The only skeleton from the Cape coast for which there is good evidence of pastoralist identity is a juvenile, about nine years old, from the site of Kasteelberg (Smith, 1987). It had very high delta 15 N (17.6
see Table I), presumably as a result of the individual having eaten a diet rich in both seafood and domestic animal products (Sealy, 1989). The relatively low delta 15N compared with delta 13C values of the two outlying skeletons is puzzling. Could these people have eaten cultivated Ca grains traded from the farmers to the east? South-Western Cape delta 15Ncollagen values for the 77 south-western Cape skeletons ranged from 10.2 to 17.7%0 and delta 13Ccollagen from --10.6 to --17.9%o (Table I). These ranges are more or less the same as those for the southern Cape, except that there were no delta 15Ncollagen values below 10%o in the south-western Cape sample. Figure 4 shows delta 15Ncollagen plotted against delta 13 Ccollagen. The fitted regression equation is delta 5 N = 24.0 + 0.69 delta 13C, with r2-- 0.51 (n = 77). A hypothesis test of no correlation between the two variables was carried out and the hypothesis rejected at the 1% level (Zar, 1984). The slope of the regression for the south-western Cape (0.69 with a standard error of 0.08) is lower than that for the southern Cape (1.08 with a standard error of 0.12). This lower slope is expected if there is less difference in delta 15 N between marine and terrestrial resources in the south-western compared with the southern Cape. Again, the correlation is better than might be predicted from first principles. In an earlier discussion of some of these data, we suggested that the correlation might have to do with trophic level. Terrestrial diets probably included large quantities of plant foods (containing some protein, but with low delta 15 N) whereas marine diets were based on animal foods, with much higher protein contents and delta 15 N (Sealy et al., 1987). 15N/14N values were somewhat elevated compared with the southern Cape, with no values below 10, contributing to the lower slope of the regression line. Arid terrestrial conditions mentioned earlier may contribute to this elevation, though many of the skeletons came from the southerly part of the research area, where this is not the case. Skeletons dating to the last 1000 years fell within the general cluster. Skeletons dating to the last 2000 years tended to lie towards the left-hand side of the plot, as a result of their more negative delta 13C values and, to a lesser extent, lower delta 15 N values. This chronological patterning has been discussed elsewhere (Sealy and van der Merwe, 1988). Briefly, skeletons which post-date 2000BP have isotopic values reflecting a greater intake of terrestrial foods compared with strongly marine-derived signals for skeletons dating between 3000 and 2000 BP. This pattern is consistent with the picture formulated from excavation of coastal sites, especially at Elands Bay (Parkington et al., 1988), where excavated food-waste demonstrated especially intensive collection of shellfish between 3000 and 2000BE Isotopic analyses of skeletons have shown that these dietary shifts occurred over an area wider than Elands Bay alone. There was no tendency for southwestern coastal skeletons from the first millennium BP to show a stronger Ca signal, as in the southern Cape. Apatite (Bone Mineral) Southern Cape delta 13 C of bone apatite was determined for 58 skeletons. Apatite values were more positive than those for bone collagen, reflecting both the dietary origin of the carbon and the diettissue spacings. Apatite delta 3C values probably best reflect an individual's diet as a whole, whereas collagen delta 13C is believed to represent mostly the protein component of the diet (Ambrose and Norr, 1993). delta 13C values for the southern Cape skeletons ranged from -6.8 to -12.6 (Table I). The magnitude of the spacing between collagen and apatite delta 13C within individuals is also of interest
collagen-apatite spacings in the southern Cape skeletons varied from 1.8 to 6.3, with a mean of 3.65 + 0.98. There do not appear to be significant shifts in collagen-apatite spacings through time (Fig. 5). The mean value for skeletons older than 4000 BP was 3.82 4-0.96 (n = 18), for those between 4000 and 3000BP it was 3.86 4-1.2 (n = 11), for those between 3000 and 2000BP it was 3.614-1.00 (n = 20) and for those post-dating 2000 BP it was 3.06 + 0.44 (n = 8). The mean for the last, most recent group was smaller than the others
the sample, however, was small and the range of values was within the ranges of the preceding groups. Comparison of the distribution of collagen-apatite spacings in the post-2000 BP group with those in the 2000 to 3000 BP group revealed that they do not, on the available data, show themselves to be significantly different (MannWhitney approximate Z-value=l.55, 0.06< p < 0.1). Values for the three skeletons from the second millennium AD were not unusual: at 3.4, 3.0 and 3.0 they fell within the main cluster of points in Fig. 5. There were no significant differences between collagen-apatite spacings for male and female skeletons. Twenty nine skeletons could be identified with a reasonable degree of confidence as either male or female
the mean collagen-apatite spacing for females was 3.62 4- 0.74 (n = 13), for males it was 3.45 4- 0.94 (n = 16). Western Cape delta 13C of bone apatite was determined for 32 adult skeletons. Results for three sub-adult individuals reported in Lee-Thorp et al. (1989) are not included here. Values ranged from -7.7 to -13.5 (Table I). Collagen-apatite spacings ranged from 0.9 to 4.4 with a mean of 2.5O 4- 0.88. In the western Cape sample, there was chronological variation in collagen-apatite spacings. Skeletons dating between 3000 and 2000BP had the smallest spacings, and also the most enriched delta 13Ccollagen values. Those older than 3000BP and younger than 2000BP had somewhat larger spacings, and less enriched delta 13Ccollagen values (Fig. 6). Collagen-apatite spacings for the south-western Cape were unusually small. Earlier, we interpreted these results to indicate diets in which protein came mainly from seafood (with relatively positive delta 13C values) and carbohydrates and fats came mainly from terrestrial plant foods and the fat of seals and whales, all with depleted delta 13C values. It has been argued that if collagen delta 13C reflects primarily the protein part of the diet, while apatite delta 13C derives from fats and carbohydrates, then the combination of enriched delta 13C values for protein, plus a relatively small (c. 5%0) diet-collagen spacing, and depleted delta 13C values for fats and carbohydrates plus a large (c. 12%o) diet-apatite spacing, would result in very small collagenapatite spacings (Lee-Thorp et al., 1989). This pattern would be most marked in individuals who ate a great deal of seafood, consistent with the observation that skeletons with very enriched delta 13Ccollagen values also had the smallest collagen-apatite spacings. Recently, Ambrose and Norr (1993) have conducted a series of controlled-feeding experiments with rats, in order to ascertain the extent to which particular fractions of the diet may be channelled to specific compartments of bone. They concluded that "dietary protein is routed to collagen, while all nutrient biochemical fractions are scrambled and integrated in bone carbonate" and that the magnitude of the dietcollagen spacing depends on the delta 13C value of dietary protein, on the proportion of protein in the diet and on the difference in delta 13C between protein and energy components of the diet. Both their study and a controlled feeding study on mice (Tieszen and Fagre, 1993) have yielded diet-apatite spacings of about 9 to 9.5%0. This may be a better estimate for the spacing in humans than the 12%0 used by Lee-Thorp et al. (1989), which was derived mostly from measurements of ruminant herbivores (see also Metges et al., 1990). What do these results mean for the interpretation of the apatite-collagen spacings in the skeletons discussed above? Current models have shifted from the protein-collagen, energy-apatite scenario considered by Lee-Thorp et al. (1989). Instead, we should consider a proteincollagen, mixed total diet-apatite model, with variable diet-collagen fractionation factors. Thus the delta 13C values of the starting points become more similar (dietary protein versus mixed diet, rather than dietary protein versus dietary fats and carbohydrates), and the model perhaps slightly less effective as an explanation for the very small collagen-apatite spacings in the western Cape skeletons. The premise that the phenomenon is likely to be due to routing of different dietary ingredients to different components of bone probably holds true. It is not clear to what extent variations in the dietcollagen fractionation factors are likely to be important, nor do we know how closely the experimental rats represent the metabolic patterns in humans. Experiments presently under way involving controlled feeding of pigs - a better analogue for humans - will help to answer this question. Comparison of the Southern and Western Cape Samples The mean apatite-collagen spacing for the southern Cape skeletons (3.65+0.98, n=58) was significantly different from that for the south-western Cape set (2.50+0.88 n=32) (t = 5.52, p < 0.01). The distributions of the two samples were also significantly different (MannWhitney approximate Z-value=4.8, p<0.001). In the south-western Cape, the smallest apatitecollagen spacings were associated with the largest marine food intakes
this was not the case in the southern Cape. Food-waste from archaeological sites reflects local environments but does not indicate substantial differences in the nature of Holocene human diets in these two regions. Why are the apatite-collagen spacings larger in the southern Cape? According to the model above, and assuming for the moment that diettissue spacings hold constant, this might result from more enriched delta 13C values in the energy component of the diet, or perhaps more depleted delta 13C values in the protein component. Changes in the nature of protein foods ought to be reflected in collagen delta 13C, and the ranges of collagen delta 13C values in the two samples were similar (Figs. 5 and 6). What differences might there be between the two regions in carbohydrate/fat foods? The main sources of carbohydrates were probably plants, especially the underground corms of plants in the Iridaceae family, such as Watsonia, Moraea, Ixia, etc. These are, on the whole, found in fairly open habitats
they are not forest plants, and carbohydrate-rich plant foods may have been harder to come by in heavily forested areas of the southern Cape than in the western Cape. In the western Cape "adzes", or stone tools thought to have been used for wood-working, form a prominent part of stone tool assemblages of the last three millennia. Adzes may have been used to fashion digging sticks used to extract underground corms and bulbs
certainly the increase in adzes in the last 3000 or so (radio-carbon) years is paralleled by an increase in identifiable plant food residues in the western Cape. This pattern is less clear in the southern Cape. There are fewer excavated sites in the southern Cape, and the most detailed excavation report, for Nelson Bay Cave (Inskeep, 1987), describes a coastal site, nowhere rich in plant food residues. The only site in the southern Cape (as defined here) which is not directly on the coast and for which a relatively detailed excavation report is available is Oakhurst, which also yielded skeletons analysed in this study. Artefacts from Oakhurst were described in the 1930s (Goodwin, 1938) and again in the early 1960s (Schrire, 1962), using a slightly different system of categorisation from that currently employed. It is difficult to compare the artefact frequencies from Oakhurst directly with those from Western Cape sites
the category "adzes" does not appear in the Oakhurst reports, but some of the artefacts described as hollow, flat or side scrapers are similar to western Cape adzes (J. Parkington, personal communication). It is probably true to say, though, that adzes do not dominate the Oakhurst artefact assemblage of the last 3000 years in the way that they do in many western Cape assemblages, and this may reflect differences in the importance of wood-working, and perhaps underground plant food collection, in the two regions. Inland of the area from which the southern Cape coastal skeletons are derived, in the mountainous belt which divides the coastal plain from the interior plateau of South Africa, is the site of Boomplaas. Extensive excavations have been conducted in this large cave, coupled with surveys of relevant aspects of the modern environment. Adzes reach 18.5% of the formal tool assemblage in c. 2000BP deposits (Deacon, 1984), but are not as common as in western Cape sites of the same age, where they may contribute 50% or more of the retouched artefacts (e.g. Manhire, 1993). Searches for edible plants in the Boomplaas valley have been remarkably unsuccessful, and it has been suggested that the occupants of the cave probably had to forage further afield, into the adjacent mountains, to obtain plant foods (Moffett and Deacon, 1977). CONCLUSIONS delta 13C and delta 15N values for bone collagen, and delta 13C values for apatite were compared within and between two sets of Later Stone Age skeletons from the southern and western Cape of South Africa. Most individuals were huntergatherers and were buried along the coast. Variations in isotope ratios probably reflect primarily differences in the amounts of seafoods these people ate during life. Terrestrial Ca grasses in the southern Cape seem not to influence delta 13Ccollagen to a significant extent (except, perhaps, for three relatively recent skeletons)
similarly, enriched terrestrial nitrogen isotope ratios in the south-western Cape affect delta 15Ncollagen less than might be expected. Collagen-apatite spacings in southern Cape skeletons did not vary significantly through time, or between males and females, but were on average larger than collagen-apatite spacings for skeletons from the south-western Cape. The reasons for this are not well understood, but probably have to do with differences in the delta 13C values of the energy-providing portions of the diet in relation to dietary protein. The results discussed here constitute the largest set of stable carbon and nitrogen isotope measurements for coastal (mostly hunter-gatherer) human skeletons currently available. The size of the data set makes it possible to explore relationships between isotopic indicators of diet in a way that cannot otherwise be done. MATERIALS AND METHODS Southern Cape Eighty adult human skeletons were selected from museum collections. In a few cases radio-carbon dates and stable isotope values were already available
for most specimens samples were submitted to a radiocarbon laboratory (Pretoria) for determination of the date. Stable isotope ratios were measured in the Archaeomerry Laboratory at the University of Cape Town. Samples were prepared and analysed as follows. In order to extract acid-insoluble bone protein, loosely referred to here as "collagen", chunks of bone were surface-cleaned and soaked in dilute (c. 2%) hydrochloric acid until decalcified. They were then rinsed in distilled water and soaked overnight in 0.1M NaOH, then rinsed repeatedly in distilled water until neutral. Finally, the extracted material was freeze-dried. delta 13Ccollagen and delta 15Ncollagen were measured on a Finnigan-MAT 252 ratio mass spectrometer coupled to a Carlo-Erba preparation unit. Samples were loaded into tin capsules, combusted at 1600 degrees C and the resultant gases cleaned on-line in the Carlo-Erba system, before being introduced into the mass spectrometer. Mass spectrometric measurements were made against commercially-available CO2 and N2. Carbon dioxide results were related to the PDB standard and nitrogen results to atmospheric N2, through the running of a variety of secondary standards whose values relative to PDB and atmospheric N2 are well known. The standard deviation of repeated determinations of a homogeneous material was 0.2%o for both carbon and nitrogen. Bone apatite was prepared from 58 southern Cape skeletons according to methods described previously (Lee-Thorp et al., 1989). Carbon dioxide was produced by reaction with 100% H3PO4 in a closed vessel and purified by cryogenic distillation. Gas samples were introduced manually into a VG Micromass 602E ratio mass spectrometer and 13C/12C ratios measured against a calibrated reference gas. Precision of duplicate analyses of homogeneous material was better than 0.2%o. South-Western Cape Results for collagen from 77 human skeletons are considered here: all have been radiocarbon dated and their stable carbon and nitrogen isotope ratios measured. Apatite was prepared from 32 of these skeletons. Results in this paper are for adult individuals only
most were taken from the sample set already published (Sealy et al., 1987
Sealy and van der Merwe, 1988
Lee-Thorp et al., 1989), augmented somewhat by more recent work. Samples subsequently added to the data set were treated in the same
|