Hominins, sedges, and termites: new carbon isotope data
from the Sterkfontein valley and Kruger National Park
Matt Sponheimera,b,*, Julia Lee-Thorpb, Darryl de Ruiterc, Daryl Codronb,
Jacqui Codronb, Alexander T. Baughd, Francis Thackeraye
aDepartment of Anthropology, University of Colorado at Boulder, Boulder, CO 80309
bDepartment of Archaeology, University of Cape Town, Rondebosch, WP 7701
cDepartment of Anthropology, Texas A&M University, College Station, TX 77843
dInstitute for Neuroscience, University of Texas at Austin, Austin, TX 78712
eDepartment of Palaeontology, Transvaal Museum, Pretoria, GP 0001
Received 22 August 2004; accepted 27 November 2004
Stable carbon isotope analyses have shown that South African australopiths did not have exclusively frugivorous
diets, but also consumed significant quantities of C4foods such as grasses, sedges, or animals that ate these foods. Yet,
these studies have had significant limitations. For example, hominin sample sizes were relatively small, leading some to
question the veracity of the claim for australopith C4consumption. In addition, it has been difficult to determine which
C4resources were actually utilized, which is at least partially due to a lack of stable isotope data on some purported
australopith foods. Here we begin to address these lacunae by presenting carbon isotope data for 14 new hominin
specimens, as well as for two potential C4foods (termites and sedges). The new data confirm that non-C3foods were
heavily utilized by australopiths, making up about 40% and 35% of Australopithecus and Paranthropus diets
respectively. Most termites in the savanna-woodland biome of the Kruger National Park, South Africa, have
intermediate carbon isotope compositions indicating mixed C3/C4diets. Only 28% of the sedges in Kruger were C4, and
few if any had well-developed rhizomes and tubers that make some sedges attractive foods. We conclude that although
termites and sedges might have contributed to the C4signal in South African australopiths, other C4foods were also
important. Lastly, we suggest that the consumption of C4foods is a fundamental hominin trait that, along with
bipedalism, allowed australopiths to pioneer increasingly open and seasonal environments.
? 2004 Elsevier Ltd. All rights reserved.
Keywords: Hominins; Paleodiet; Carbon isotopes; Sedges; Termites; Kruger National Park
* Corresponding author. Matt Sponheimer, Department of Anthropology, University of Colorado at Boulder, Boulder, CO 80309.
Tel.: C1 303 735 5774.
E-mail address: firstname.lastname@example.org (M. Sponheimer).
0047-2484/$ - see front matter ? 2004 Elsevier Ltd. All rights reserved.
Journal of Human Evolution 48 (2005) 301e312
Over the past decade, stable carbon isotope
analysis of tooth enamel has been used to study
the diets of early hominins in South Africa (Lee-
Thorp et al., 1994; Sponheimer and Lee-Thorp,
1999a; van der Merwe et al., 2003). The premise of
these studies is that you are what you eat, and that
the stable carbon isotope composition of your
food is ultimately reflected in your tooth enamel,
even at several million years remove. Previous
research, using non-isotopic techniques, had sug-
gested that australopiths ate diets dominated by
fleshy fruits or hard objects most likely originating
from trees or bushes (e.g., Kay, 1985; Grine, 1986;
Grine and Kay, 1988). These types of plants use
the C3 photosynthetic pathway which discrim-
inates markedly against
depleted13C/12C ratios (about ?27&). In contrast,
plants that utilize the C4photosynthetic pathway,
such as tropical grasses and some sedges, discrim-
inate less against
depleted (about ?12&) (Smith and Epstein, 1971).
These distinct isotopic signatures are passed down
into the tissues of animals that eat these plants.
For instance, the tissues of zebra, which eat C4
grass, are more enriched in13C than the tissues of
giraffe, which eat leaves from C3 trees. Conse-
quently, it was expected that early hominins, like
the modern frugivorous chimpanzee, would have
a C3isotopic signature (Schoeninger et al., 1999;
Carter, 2001). Unexpectedly, however, all of the
carbon isotope studies to date have shown
australopiths to be rather enriched in13C, suggest-
ing that foods other than fruits were also
important dietary components (Lee-Thorp et al.,
1994; Sponheimer and Lee-Thorp, 1999a; van der
Merwe et al., 2003).
While these isotopic studies were significant in
providing evidence that our understanding of
australopith diets was too narrow, they had
limitations. First, the number of individual hom-
inins analyzed was relatively small. While this has
been substantially remedied through recent publi-
cation (van der Merwe et al., 2003), the total
number of published hominin carbon isotope
ratios is still small compared to the number from
13C, leading to very
13C and are consequently less
C3- and C4-consuming fauna to which they are
Thorp, 2003). This limitation in sample size,
together with concerns that diagenesis may have
affected some results, has led some to question
whether or not australopiths really differed from
their C3 plant consuming coevals or modern
chimpanzees (Schoeninger et al., 2001).
Another limitation of these studies was that,
although they suggested that non-C3foods were
consumed in significant quantities, they were not
able to identify what these foods might have been.
Thus, C4grasses, C4sedges, and animals that ate
these foods were all offered as possible australopith
foods (Lee-Thorp et al., 1994; Sponheimer and
Lee-Thorp, 1999a; van der Merwe et al., 2003;
Peters and Vogel, in press). Amongst the reasons
for this inability to identify the C4dietary source
was a lack of data available on the carbon isotope
compositions of potential C4 foods other than
grasses and large vertebrates. For instance, a recent
investigation of bone tools at Swartkrans suggested
that they have wear formed by digging in termite
mounds, leading the researchers to hypothesize
that consumption of C4grass-eating termites might
explain the13C-enriched isotopic signature of the
australopiths (Backwell and d’Errico, 2001). Al-
though this possibility was both intriguing and
plausible given that some termite taxa (e.g.,
Trinervitermes, Hodotermes) consume grass, it
remained speculative as there were no published
data onthe carbon isotope compositions of African
savanna termites excepting a single species Macro-
termes michaelseni (Boutton et al., 1983).
We encountered a similar problem with sedges.
Conklin-Brittain et al. (2002) recently proposed
that underground storage organs (USOs) in wet-
lands and river margins, such as the starchy
rhizomes of some sedges, were important foods
for australopiths; and since many sedges utilize C4
photosynthesis, sedges represent a potential source
of the non-C3signal observed in early hominins
(Sponheimer and Lee-Thorp, 1999a, 2003; Lee-
Thorp et al., 2003; van der Merwe et al., 2003).
Unfortunately, however, although an estimated
33% of the world’s sedges utilize C4photosynthe-
sis (Sage et al., 1999), relatively little is known
302M. Sponheimer et al. / Journal of Human Evolution 48 (2005) 301e312
about the isotopic compositions of sedges in
modern South African environments that are most
similar to those associated with australopiths (but
see Stock et al., 2004). Thus, it was difficult to
evaluate the likelihood that C4sedges were even
available for consumption by South African
The aim of this paper is to begin to address
these limitations in two ways. Firstly, we provide
new carbon isotope data from 14 australopith
specimens that greatly increase the previously
published sample size. Secondly, we proffer novel
carbon isotope data from a study of modern
termites and sedges in Kruger National Park,
South Africa. As Kruger contains a variety of
environments that may be similar to those
inhabited by australopiths (Reed, 1997; Spon-
heimer et al., 1999, 2001), we believe that these
data represent a reasonable first step towards
understanding the isotopic compositions of their
We sampled a total of 14 hominin permanent
molars housed at the Transvaal Museum in
Pretoria, South Africa for this study: these in-
cluded six w2.5 Ma Australopithecus africanus
teeth from Member 4 at Sterkfontein and nine
w1.8 Ma Paranthropus robustus teeth (8 from
Member 1 at Swartkrans and 1 from Member 3 of
Kromdraai B) (Table 1). The three cave sites from
which the teeth originated are within 3 km of each
other in the dolomites of the Sterkfontein Valley
(Brain, 1981). Specimens without heavy staining or
mineral inclusions were sampled and pretreated
based upon protocols outlined in Sponheimer
(1999). All hominin teeth had been previously
fractured and enamel samples (w2 mg) were
acquired from between the enamel-dentine junc-
tion and the outer surface using a rotary drill with
a diamond-tipped burr. Within this constraint, we
sampled as extensive an area as possible in order to
obtain enamel formed over a significant period of
time. Importantly, with this method our sampling
did not alter the external morphology of the teeth.
The enamel powder was pretreated with 1.5%
NaClO for ten minutes to remove organic
contaminants, and then rinsed to neutrality. It
was then subjected to 0.5 ml of 0.1 M CH3COOH
for another ten minutes to remove diagenetic
carbonates, and again rinsed to neutrality. Sam-
ples were lyophilized and placed in individual
reaction vessels and analyzed for
a Kiel II autocarbonate device coupled to a Fin-
nigan MAT 252 mass spectrometer. Carbon iso-
tope ratios (13C/12C) are expressed as d13C values
in parts per thousand (&) relative to the PDB
standard. The standard deviation of working
Specimen numbers, tooth identifications, provenience information, and d13C for hominin specimens analyzed in this study.
Specimen Tooth Taxon Provenience
303M. Sponheimer et al. / Journal of Human Evolution 48 (2005) 301e312
standards analyzed concurrently with the homi-
nins was 0.1&.
Kruger National Park sampling
The Kruger National Park sampling was
carried out seasonally (dry and rainy seasons)
between June 2002 and September 2004. In order
to gain an idea of the potential isotopic variety
available to early hominins, we sampled termites
from the northernmost and southernmost regions
of the park, from closed riverine to open grassland
environments, and from a variety of substrates
(e.g., mounds, feces, logs). All termites were placed
in ethanol inside microcentrifuge vessels and then
dried when we returned to the University of Cape
Town laboratory. Sedges were collected from four
10-meter circular transects at riverine sites, two in
northern, and two in southern Kruger Park. All
available sedge taxa were sampled from each site,
but the number of samples acquired of each taxon
roughly reflected its local abundance. The sedges
were dried at 60(C in the Kruger Park, and upon
arrival in Cape Town were ground using a Wiley-
Mill with a 1 mm screen. Termite and sedge
samples were then weighed, placed in tin capsules,
and analyzed for
elemental analyzer coupled to a Finnigan MAT
252 mass spectrometer. The standard deviation of
working standards run in conjunction with the
Kruger National Park samples was 0.1&.
13C/12C using a Carlo-Erba
New HomininsdThe carbon isotope results for
the australopiths analyzed in this study are
presented in Table 1 and Fig. 1. Australopithecus
(x Z ?6.8&, s.d. Z 2.1, nZ 5) and Paranthropus
(x Z ?7.0&, s.d. Z 0.7, n Z 9) are not signifi-
cantly different (P Z 0.87 t-test; PZ 0.79 Mann-
Whitney U), as was the case in previous studies
(Sponheimer and Lee-Thorp, 1999a, 2003). Not
surprisingly, the new Australopithecus data are
statistically indistinguishable from previous anal-
yses (P Z 0.74 t-test; PZ 0.85 Mann-Whitney U).
They do, however, slightly extend the previously
published range of d13C for confidently identified
Australopithecus samples, as one specimen was
extremely enriched in13C (?4.0&). Nonetheless,
the mean and great range (5.7&) for Australopi-
thecus is exactly what was anticipated from
previous work (see Discussion for more on this
variability). The situation for Paranthropus is
somewhat different. The new Paranthropus d13C
values are slightly (1.3&), but significantly en-
riched compared to previous analyses (P Z 0.01
t-test; P Z 0.01 Mann-Whitney U). Furthermore,
they extend the previously known range for this
taxon, with one specimen being quite enriched in
13C (?5.9&). Despite these differences, however,
the general character of the Paranthropus data is in
accord with previous results in that their mean is
indistinguishable from Australopithecus yet they
are less variable (Fig. 1).
All HomininsdGiven the greatly expanded
hominin dataset, we now briefly address the
question: are hominins really different from known
C3consumers? The answer is an unequivocal ‘‘yes’’
(Fig. 2). To address this question, we merged
hominin and non-hominin data from Swartkrans,
Sterkfontein, and Makapansgat for statistical
analysis. Such conflation would not be warranted
in all contexts, as vegetation d13C can differ in small
but meaningful ways over time and space. The d13C
of C3vegetation, in particular, can change signif-
icantly between sere, open environments and
n = 5
n = 9
Fig. 1. d13C for Australopithecus africanus and Paranthropus
robustus specimens analyzed for this study. The box represents
the 25the75thpercentiles (with the median as a horizontal line)
and the whiskers show the 10the90thpercentiles.
304M. Sponheimer et al. / Journal of Human Evolution 48 (2005) 301e312
damp, closed canopy forests (Ehleringer and
Cooper, 1988; van der Merwe, 1989; Cerling
et al., 2003; Codron, 2003). However, the d13C of
C3consumers at these sites does not vary signifi-
cantly (ANOVA, PZ 0.09), nor does linear re-
gression reveal any trend in their d13C over time
(P Z 0.38, R2Z .01). Thus, merging data from
these sites is justified.
Both Australopithecus (x Z ?7.1&, s.d. Z 1.8,
nZ 19) and Paranthropus (x Z ?7.6&, s.d. Z 1.1,
nZ 18) are strongly different from the C3con-
(x Z ?11.5&,
Scheffe ´ ; P ! 0.0001) with which they are associated
(Tables 1 & 2). Both hominin taxa are also highly
distinct from associated C4-grazing bovids, equids
and suids (x Z ?0.6&, s.d. Z 1.8, nZ 60)(ANO-
VA, Scheffe ´ ; P ! 0.0001), but cannot be distin-
guished from each other (ANOVA, Scheffe ´ ;
P Z 0.62). This new larger dataset confirms our
previous work showing that early hominins were
distinct from both C3and C4consuming fauna.
This disparity cannot be ascribed to diagenesis, as
there is no evidence that browser or grazer d13C has
(ANOVA, s.d. Z 1.3,
been irrevocably altered, and diagenesis should
affect hominins and non-hominins alike. It ap-
pears, however, that we previously underestimated
the importance of non-C3foods to australopiths
(Lee-Thorp et al., 1994; Sponheimer and Lee-
Thorp, 1999a). We previously estimated that about
25% of australopith diets came from non-C3
sources. We arrived at this estimate by comparing
the mean australopith d13C values to the mean
values for associated C3-browsers and C4-grazers
which were taken as indicators of ‘‘pure’’ C3and C4
diets (endmembers) respectively. This comparative
method is the most appropriate for estimating C4
intake in fossil fauna as it requires no assumptions
about the carbon isotope composition of ancient
vegetation (Lee-Thorp, 1989). For example, it
makes no difference in the percent C4estimate if
the d13C of C3vegetation was ?29& or ?25&, as
these differences would be reflected in the d13C of
the C3-consuming fauna to which the hominins are
being compared. Thus, our percent C4estimates
are likely to be robust. Nevertheless, it should be
understood that these are rough estimates (rounded
to the nearest 5%) that have been included
primarily to allow discussion of hominin d13C
without constant reference to biogeochemical
With this caveat in mind, the data suggest that
Australopithecus and Paranthropus ate about 40%
and 35% C4-derived foods respectively. Such
a significant C4contribution, whatever its origin,
is very distinct from what has been observed for
modern chimpanzees (Pan troglodytes). Schoe-
ninger et al. (1999) found no evidence of C4foods
in chimpanzee diets even in open environments
with abundant C4-grass cover. Similarly, Carter’s
(2001) analysis of mammals in Kibale National
Park showed the d13C of chimpanzees and C3-
eating duiker antelope (Cephalophus spp.) to be
indistinguishable. Thus, unlike our closest living
relatives, both hominin taxa appear to have
extensively utilized non-C3derived foods.
Sedges and termites in Kruger National Park
TermitesdThere are two particularly conspic-
uous results from the carbon isotope analyses of
the Kruger National Park’s termites (x Z ?20.1&,
n = 60
n = 19
n = 18
n = 61
Fig. 2. d13C for new and previously analyzed Australopithecus
africanus and Paranthropus robustus specimens, as well as C3
plant consumers (browsing/frugivorous bovids and giraffids)
and C4plant consumers (grazing bovids and equids). The box
represents the 25the75thpercentiles (with the median as
a horizontal line) and the whiskers show the 10the90th
percentiles. Given the size of this dataset, there can be no
doubt that australopith d13C is highly distinct from that of
305M. Sponheimer et al. / Journal of Human Evolution 48 (2005) 301e312
s.d. Z 3.6, n Z 40; Fig. 3). First, they are highly
variable, ranging from nearly pure C3to pure C4
consumers, which is quite similar to the range of
variation that has been observed in Australian
savanna termites (Tayasu et al., 1998). Second,
despite this variability, most have a mixed C3/C4
signal regardless of substrate (mounds or logs).
Hence, they are highly distinct from both C3trees
(x Z ?26.4&, s.d. Z 1.8, n Z 550) and C4grasses
(x Z ?12.2&, s.d. Z 1.1, nZ 777) collected dur-
ing the course of the two-year Kruger study
(ANOVA, Scheffe ´ ; P ! 0.0001). As expected,
termites from open environments with abundant
C4grasses (x Z ?15.3&, s.d. Z 2.7, n Z 10) are
enriched in13C compared to those in more closed
environments (xZ ?21.7&, s.d. Z 2.1, nZ 30)
(P ! 0.001 t-test; P ! 0.001 Mann-Whitney U),
indicating that they consumed greater quantities of
C4vegetation. Unexpectedly, however, termites in
closed riverine environments also ate significant
amounts of C4 vegetation, despite the local
dominance of C3 woody vegetation. Indeed,
assuming a diet-termite spacing of about C1&
(following Boutton et al., 1983), the closed
environment termites consumed about 25% C4
vegetation, while termites in open environments
ate around 70% C4 vegetation, resulting in an
average of 35% for the entire park. Thus, C4plant
consumption appears to be important for the vast
majority of termites in the park, or at least for
those that are most readily accessible (in mounds
and logs) by modern hominin gatherers. In
contrast, termites in the tropical forests of
nearly pure C3diets (Tayasu et al., 1997).
Sedgesd The Kruger sedges fall neatly into two
distinct groups, those using the C3(x Z ?27.2&,
s.d. Z 1.4,x Z 54) and those
(x Z ?11.7&, s.d. Z 1.1, nZ 21) photosynthetic
pathway (P ! 0.001 t-test; P ! 0.001 Mann-Whit-
ney U)(Fig. 4). Surprisingly, however, despite the
common assumption that most African sedges in
using the C4
Specimen numbers, taxon, provenience, d13C, and publication date for all other australopith specimens. In the provenience column,
site abbreviations (SK Z Swartkrans, MAK Z Makapansgat, ST Z Sterkfontein) are followed by the appropriate Member number.
The 1994 publication is Lee-Thorp et al., 1994; the 1999 publication is Sponheimer and Lee-Thorp, 1999a; the 2000 publication is Lee-
Thorp et al., 2000; and the 2003 publication is van der Merwe et al., 2003.
Specimen Tooth TaxonProvenience
STW 309b (409)
306M. Sponheimer et al. / Journal of Human Evolution 48 (2005) 301e312
summer rainfall zones utilize C4photosynthesis,
72% of the specimens we analyzed were C3plants
(which reflects the local abundance). This un-
expected result has been affirmed by a recent
herbarium survey of South African sedges which
found that 65% use C3 photosynthesis (Stock
et al., 2004). These authors found that the
distribution of C4sedges differs markedly from
that of C4 grasses, and is likely controlled by
different climatic factors. In contrast, a study of
Kenyan sedges found that 65% use C4photosyn-
thesis (Hesla et al., 1982). Thus, it appears that C4
sedges are much less common in South African
woodlands than they are in previously studied East
The new hominin d13C data demonstrate two
things. First, with a total of 37 australopiths now
analyzed, there can be no question that their
carbon isotope compositions are highly distinct
from those of their C3-consuming contemporaries.
This is in stark contrast to modern chimpanzees
and gorillas, both of which have essentially pure
C3 signatures (Schoeninger et al., 1999; Carter,
2001; Sponheimer, unpublished data). This is not
to say that we will never find a chimpanzee that
deviates from this pattern. What is clear, however,
is that a chimpanzee with a non-C3 signature
would be an exception, whereas non-C3signatures
represent the norm for australopiths. Thus,
australopiths and chimpanzees clearly ate different
disparate dietary adaptations or habitat differ-
ences. The latter possibility seems less likely, as
even in woodland savanna with nearly continuous
C4-grass cover, chimpanzee hair retains no evi-
dence of C4 consumption (Schoeninger et al.,
1999). Moreover, as this environment is similar to
the woodland/bushland environments believed to
have been inhabited by the South African austral-
opiths (Reed, 1997), it seems most parsimonious to
conclude that australopiths and chimpanzees,
when in similar environments, would have utilized
the available resources in different ways. Simply
put, chimpanzees largely ignore the available C4
resources that constituted a major component of
australopith diets. This is, of course, not terribly
surprising given the marked differences in the
craniodental morphology of australopiths and Pan
(e.g., Grine, 1981; Kay, 1985; Ungar, 2004).
Second, these data demonstrate enormous
variability in the d13C of South African hominins,
especially within the taxon Australopithecus afri-
canus. This high degree of variability is very
unusual among both modern and fossil faunas in
South Africa (Lee-Thorp et al., 1994, 2000;
Sponheimer et al., 1999, 2001, 2003; Codron,
n = 777
n = 40
n = 550
Fig. 3. d13C for termites, trees, and grasses in Kruger National
Park. The box represents the 25the75thpercentiles (with the
median as a horizontal line) and the whiskers show the
10the90thpercentiles. Note that while termites have highly
variable carbon isotope compositions, the vast majority of
specimens have mixed C3/C4signatures.
n = 54
n = 21
Fig. 4. d13C for C3and C4sedges in Kruger National Park. The
box represents the 25the75thpercentiles (with the median as
a horizontal line) and the whiskers show the 10the90th
percentiles. Note that a wide majority of the sedges (72%)
307M. Sponheimer et al. / Journal of Human Evolution 48 (2005) 301e312
2003; van der Merwe et al., 2003), and might be
ascribed to environmental heterogeneity. There is
considerable evidence that South African austral-
w3.0 Ma and 1.8 Ma (Vrba, 1980, 1985; Reed,
1997; Luyt and Lee-Thorp, 2003), and in fact the
d18O data for the hominins in this study are
consistent with increasing aridity in the Sterkfon-
tein valley as 1.8 Ma Paranthropus (x Z ?0.6&,
s.d. Z 1.1, n Z 9) is significantly enriched in
d18O comparedto2.5 Ma
(x Z ?2.6&, s.d. Z 0.8, nZ 5; Fig. 5)(P! 0.01
t-test; P ! 0.01 Mann-Whitney U). Factors other
than the d18O of meteoric water are important
determinants of mammalian d18O, however, so we
cannot rule out the possibility that this difference
is the product of ecological rather than climatic
differences (Kohn et al., 1996; Sponheimer and
Lee-Thorp, 1999b; Sponheimer and Lee-Thorp,
2001). Regardless, given the abundant evidence of
environmental change through time, one might
expect that it would explain some of the observed
variability in hominin carbon isotope ratios. Yet,
linear regression demonstrates that there is no
relationship between hominin d13C and time
(P Z 0.63, R2Z 0.01; Fig. 6), and there are no
significant differences in hominin d13C between
3.0 Ma Makapansgat Member 3, 2.5 Ma Sterk-
fontein Member 4, or 1.8 Ma Swartkrans Member
1 (ANOVA, P Z 0.14).
Indeed, what is most striking about these data is
the lack of change in hominin d13C in the face of
pronounced environmental change. Somewhat
paradoxically, however, within any given time
period (Member) hominin d13C is highly variable.
This might be attributed to changing environments
during the accumulation of these faunas, yet this
would run counter to the fact that the mean
hominin values do not change over a period of at
least a million years during which local environ-
ments become more open. On the other hand, the
variability might simply be an indication of the
australopiths being extremely opportunistic pri-
mates with wide habitat tolerances that always
inhabited a similarly wide-range of microhabitats
regardless of broad-scale environmental flux. This
would be consistent with Wood and Strait’s (2004)
recent suggestion that early hominins, including
the robust australopiths, were eurytopic rather
than ecological specialists.
In the case of A. africanus, the variability is so
great that one might be excused for asking if there
are not two ecologically distinct taxa presently
commingled within its hypodigm. In fact, if one
includes the numbers for three teeth (STW 236,
STW 213i, STW 207) that are possibly, but not
definitively, attributed to A. africanus (van der
1.6 1.82 2.22.4 2.62.83 3.2
Fig. 6. d13C of South African hominins through time. No
temporal trend is evident, despite abundant evidence that South
African hominin environments changed during this time.
n = 5
n = 9
Fig. 5. d18O of the new hominin specimens analyzed for this
paper through time. The box represents the 25the75th
percentiles (with the median as a horizontal line) and the
whiskers show the 10the90thpercentiles. The relatively depleted
d18O of Australopithecus at 2.5 Ma compared to Paranthropus
at 1.8 Ma may indicate either increasing aridity in the
Sterkfontein Valley over time, or an ecological distinction
between the taxa.
308 M. Sponheimer et al. / Journal of Human Evolution 48 (2005) 301e312
Merwe et al., 2003), then this taxon would range
from nearly pure C3to nearly pure C4diets. Stated
otherwise, the range of d13C within A. africanus
(?1.8& to ?11.3&) would be nearly as great as
the entire range for ecologically disparate Papio
and Theropithecus combined (C0.4& to ?12.6&)
(Lee-Thorp et al., 1989). Stable isotopes in and
of themselves cannot address the question of
A. africanus unity, but numerous researchers have
suggested that A. africanus might demonstrate
more morphological variability than would be
expected for a single taxon (Kimbel and White,
1988; Clarke, 1994; Lockwood, 1997; Moggi-
Cecchi et al., 1998). Hence the possibility of two
taxa, one highly dependent on C4foods and the
other much less so, cannot be dismissed. Further
work addressing this hypothesis is warranted.
Nonetheless, we continue to work under the
assumption that the specimens currently assigned
to A. africanus represent a single species.
We now return to the question: what were the
C4foods exploited by the australopiths? Grasses,
sedges, and animal foods have all been considered
as possible C4-derived food sources (Lee-Thorp
et al, 1994; Sponheimer and Lee-Thorp, 1999a,
2003; van der Merwe et al., 2003; Peters and
Vogel, in press). Our data bear directly on two of
these possibilities: sedges and termites. Termites
are a favored food of chimpanzees (e.g., Goodall,
1986; McGrew, 1992), and from our study of
termites in Kruger National Park, it is clear that
they could have contributed to the australopiths’
13C-enrichment (see Backwell and d’Errico, 2001).
Even in densely wooded riverine environments,
almost all of the termites we sampled consumed
significant proportions of C4foods, and termites
throughout Kruger Park ate 35% C4vegetation
on average. Thus, termite consumption by aus-
tralopiths in woodland savanna and even in
riverine forest would be expected to impart some
C4carbon to consumers. It seems very unlikely,
however, that termites alone could account for the
large non-C3 signal in these hominins, because
a diet of nearly 100% termites would be necessary
to explain the 35%e40% C4 component of
australopith diets. Alternatively, if the hominins
(Trinervitermes, Hodotermes) with virtually pure
C4 diets, a diet of about 35%e40% termites
would be sufficient to produce the observed
hominin carbon isotope ratios. This scenario,
however, is highly unlikely. Our opportunistic
foraging in woodland environments showed these
termites to be much less common than those with
mixed C3/C4diets (and see Braack and Kryger,
2003). And while these harvester termites are
more abundant in open grasslands and during
acute droughts (Braack and Kryger, 2003), there
is no reason to believe that such open environ-
ments were frequented by australopiths or that
drought conditions were so preponderant. More-
over, while Trinervitermes builds highly visible
above ground nests (mounds), Hodotermes does
not, making it much less conspicuous on the
landscape (Carruthers, 1997; Stuart and Stuart,
2000). Thus, it is possible and even likely that
termites contributed in some way to the unusual
d13C valuesof australopiths,
resources were almost certainly consumed in
We now return to the case for sedges. Sedges
are another potentially attractive C4 source for
a number of reasons. For one, sedges like Cyperus
esculentus and C. papyrus have long served as
foods for modern humans (Tackholm and Drar,
1973; Defelice, 2002). Western lowland gorillas
(Gorilla gorilla gorilla) have also been observed
consuming the pith of aquatic plants (including
sedges), although in small quantities (Doran and
McNeilage, 1998). Additionally, the underground
portions of sedges (as well as other plants) would
be relatively inaccessible to most mammals, yet
readily accessible to hominins with crude digging
implements (Hatley and Kappelman, 1980). And,
perhaps most importantly, the USOs of certain
sedges would represent a relatively low-fiber
resource (compared to
Conklin-Brittain et al., 2002) that would still be
available during the dry season when preferred
dietary resources were scarce.
Yet, while attractive, there are several potential
problems with this scenario. Firstly, while sedges
are relatively abundant around watercourses in
Kruger National Park and South Africa in
general, those that we encountered in Kruger’s
riverine woodlands (e.g., C. textilis) rarely if ever
309 M. Sponheimer et al. / Journal of Human Evolution 48 (2005) 301e312
had well-developed rhizomes or tubers. Thus, while
extensive stands of sedges with well-developed
USOs are common in extensive wetlands like the
Okavango Delta (Ellery et al., 1995), they are not
abundant in South African riverine woodlands
today. Similarly, the South African australopiths
are generally associated with woodland/bushland
environments (albeit with some evidence of edaph-
ic grasslands) (Reed, 1997), making it unlikely that
large quantities of edible sedges with well-de-
veloped USOs were readily available for austral-
opith consumption (Peters and Vogel, in press).
Most significantly, only 28% of the sedges we
encountered utilized the C4photosynthetic path-
way. So unless hominins were deliberately seeking
out C4sedges, or the distribution of C3and C4
sedges was markedly different in the Pliocene,
australopiths would have had to have had a diet of
100% sedges to come close to producing the
observed 35%e40% C4signature. We think this
degree of sedge specialization to be unlikely,
especially given the high predation risk around
South African watercourses today. On the whole,
while it is certainly possible that sedges contributed
to the C4signal of South African australopiths, we
find it highly improbable that sedge consumption
alone was responsible. It is worth noting, however,
that early hominin habitats such as the wetlands of
the Eastern Lacustrine Plain at Olduvai Gorge
(Hay, 1976; Deocampo et al., 2002) might have
been better sources of C4sedges. Puech et al. (1986)
have also suggested that the dental microwear of
early East African hominins is consistent with the
consumption of such foods.
We had two primary goals in this paper: first, to
present new data that should erase any doubts that
hominin carbon isotope ratios are fundamentally
different from those of associated C3and C4plant
consumers; and second, to proffer data showing
that while two of the proposed foods for South
African australopiths (termites and sedges) could
have contributed to their C4 signal, they were
unlikely to be solely responsible. We still must
consider the possibility that grasses (seeds or roots)
and animal foods made important contributions to
early hominin diets. Grass roots, grasshoppers,
bird’s eggs, lizards, rodents and young antelope
might have been important resources, particularly
during the dry season when little other food was
readily available. Succulent plants like euphorbias
(Euphorbiaceae) and aloes (Aloaceae) (which are
rare in most woodlands but have d13C values that
are sometimes indistinguishable from those of C4
grasses) are also possibilities; for although they are
often poisonous to humans (and presumably
chimpanzees) they are occasionally utilized by
baboons and humans (Codron, 2003; Peters and
Vogel, in press). Determining the nature of the C4
resources consumed by australopiths is important,
as it could have profound physiological, social, and
behavioral implications. For instance, if australo-
piths were consuming large quantities of C4grass
like modern geladas (Theropithecus gelada), this
would indicate that their diets were less nutrition-
ally dense than those of extant chimpanzees, and
possibly place important limitations on burgeoning
hominin brains and sociality. Alternatively, if they
were consuming large quantities of animal foods, it
extant apes, ultimately relaxing nutritional con-
straints on encephalization and social complexity
(Aiello and Wheeler, 1995; Milton, 1999). Further
work on dental microwear and morphology, Sr/Ca
analysis, and the potential
nutritional properties of foods, may make it
possible to identify these C4resources with greater
Despite the uncertainty as to the exact resources
that australopiths consumed, it is clear that many
australopiths heavily utilized C4 foods that are
typically overlooked by African apes. Moreover, it
is probable that, in conjunction with adaptations
like bipedalism, utilization of these resources
allowed hominins to not only cope with dwindling
forests, but pioneer new, more open environments
in their increasingly arid and seasonal world. We
believe it likely that diets containing significant
quantities of C4-derived foods are fundamental
hominin traits which will be found in all species
with clear adaptations for bipedalism. This hy-
pothesis can be tested by analysis of earlier East
African hominin specimens.
310 M. Sponheimer et al. / Journal of Human Evolution 48 (2005) 301e312
Teresa Kearney, and Stephany Potze of the Trans-
vaal Museum and Phillip Tobias, Ron Clarke,
Bruce Rubidge, and Lee Berger of the University of
Witwatersrand for their help and access to speci-
mens. Ethan Codron, Yasmin Rahman, and Karim
Sponheimer all provided invaluable support during
We thank Sandi Copeland, Kaye Reed, and three
anonymous reviewers who provided valuable com-
ments on the manuscript. We also thank Charles
Peters for access to an unpublished manuscript and
Nikolaas van der Merwe for the many discussions
we have had with him on this topic over the years.
This work was funded by the National Science
Foundation (USA), National Research Founda-
tion (RSA), Leakey Foundation (USA), Wenner-
Gren Foundation (USA), and the University of
Aiello, L.C., Wheeler, P., 1995. The expensive tissue hypothesis.
Curr. Anthropol. 36, 199e221.
Backwell, L.R., d’Errico, F., 2001. Evidence of termite foraging
by Swartkrans early hominids. Proc. Natl. Acad. Sci.
U.S.A. 98, 1358e1363.
Boutton, T.W., Arshad, M.A., Tieszen, L.L., 1983. Stable
isotope analysis of termite food habits in East African
grasslands. Oecologia 59, 1e6.
Braack, L., Kryger, P., 2003. Insects and savanna heterogene-
ity. In: du Toit, J.T., Rogers, K.H., Biggs, H.C. (Eds.), The
Kruger Experience: Ecology and Management of Savanna
Heterogeneity. Island Press, Washington, pp. 263e275.
Brain, C.K., 1981. The Hunters or the Hunted? University of
Chicago Press, Chicago.
Carruthers, V., 1997. The Wildlife of Southern Africa. Southern
Book Publishers, Halfway House.
Carter, M.L., 2001. Sensitivityof stable isotopes (13C, 15N, and
18O) in bone to dietary specialization and niche separation
among sympatric primates in Kibale National Park,
Uganda. Ph.D. Dissertation, University of Chicago.
Cerling, T.E., Harris, J.M., Passey, B.H., 2003. Diets of East
J. Mammal. 84, 456e470.
Clarke, R., 1994. Advances in understanding the craniofacial
anatomy of South African early hominids. In: Corruccini,
R., Ciochon, R. (Eds.), Integrative Paths to the Past.
Prentice Hall, Englewood Cliffs, pp. 205e222.
on stable isotopeanalysis.
Codron, D.M., 2003. Dietary ecology of Chacma Baboons
(Papio ursinus (Kerr, 1792)) and Pleistocene Cercopithecoi-
dea in Savanna Environments of South Africa. M.Sc.
Thesis, University of Cape Town.
Conklin-Brittain, N.L., Wrangham, R.W., Smith, C.C., 2002.
A two-stage model of increased dietary quality in early
hominid evolution: the role of fiber. In: Ungar, P.S.,
Teaford, M.F. (Eds.), Human Diet: Its Origin and
Evolution. Bergin & Garvey, Westport, pp. 61e76.
Defelice, M.S., 2002. Yellow nutsedge Cyperus esculentus
L.dsnack food of the Gods. Weed Technol. 16, 901e907.
Deocampo, D.M., Blumenschine, R.J., Ashley, G.M., 2002.
Wetland diagenesis and traces of early hominids, Olduvai
Gorge, Tanzania. Quat. Res. 57, 271e281.
Doran, D.M., McNeilage, A., 1998. Gorilla ecology and
behavior. Evol. Anthropol. 6, 120e131.
Ehleringer, J.R., Cooper, T.A., 1988. Correlations between
carbon isotope ratio and microhabitat in desert plants.
Oecologia 76, 562e566.
Ellery, W.N., Ellery, K., Rogers, K.H., McCarthy, T.S., 1995.
The role of Cyperus papyrus L. in channel blockage
and abandonment in the northeastern Okavango Delta,
Botswana. Afr. J. Ecol. 33, 2549.
Goodall, J., 1986. The Chimpanzees of Gombe. Cambridge
University Press, Cambridge.
Grine, F.E., 1981. Trophic differences between gracile and
robust australopithecines. S. Afr. J. Sci. 77, 203e230.
Grine, F.E.,Kay, R.F., 1988. Early hominid dietsfrom quantita-
Grine, F.E., 1986. Dental evidence for dietary differences in
Hatley, T., Kappelman, J., 1980. Bears, pigs, and Plio-
Pleistocene hominids: case for exploitation of belowground
food resources. Hum. Ecol. 8, 371e387.
Hay, R.L., 1976. Geology of the Olduvai Gorge. Univ. of
California Press, Berkeley.
Hesla, A.B.I., Tieszen, L.L., Imbaba, S.K., 1982. A systematic
survey of C3and C4photosynthesis in the Cyperaceae of
Kenya, East Africa. Photosynthetica 16, 196e205.
Kay, R.F., 1985. Dental evidence for the diet of Australopithe-
cus. Annu. Rev. Anthropol. 14, 315e341.
Kimbel, W.H., White, T.D., 1988. Variation, sexual dimor-
phism, and taxonomy of Australopithecus. In: Grine, F.E.
(Ed.), Evolutionary History of the ‘‘robust’’ Australopithe-
cines. Aldine de Gruyter, New York, pp. 175e192.
Kohn, M.J., Schoeninger, M.J., Valley, J.W., 1996. Herbivore
tooth oxygen isotope compositions: effects of diet and
physiology. Geochim. Cosmochim. Acta 60, 3889e3896.
Lee-Thorp, J.A., 1989. Stable carbon isotopes in deep time: the
diets of fossil fauna and hominids, Ph.D. thesis, University
of Cape Town.
Lee-Thorp, J.A., van der Merwe, N.J., Brain, C.K., 1989.
Isotopic evidence for dietary differences between two extinct
Lee-Thorp, J.A., van der Merwe, N.J., Brain, C.K., 1994. Diet
of Australopithecus robustus at Swartkrans from stable
carbon isotopic analysis. J. Hum. Evol. 27, 361e372.
311M. Sponheimer et al. / Journal of Human Evolution 48 (2005) 301e312
Lee-Thorp, J.A., Thackeray, J.F., van der Merwe, N., 2000. Download full-text
The hunters and the hunted revisited. J. Hum. Evol. 39,
Lee-Thorp, J.A., Sponheimer, M., van der Merwe, N.J., 2003.
What do stable isotopes tell us about hominin diets. Int. J.
Osteoarchaeol. 13, 104e113.
Lockwood, C.A., 1997. Variation in the face of Australopithe-
cus africanus and other African hominoids. Ph.D. disserta-
tion, University of the Witwatersrand, Johannesburg.
Luyt, J., Lee-Thorp, J.A., 2003. Carbon isotope ratios of
Sterkfontein fossils indicate a marked shift to open environ-
ments ca. 1.7 Ma. S. Afr. J. Sci. 99, 271e273.
McGrew, W.C., 1992. Chimpanzee Material Culture: Implica-
tions for Human Evolution. Cambridge University Press,
Milton, K., 1999. A hypothesis to explain the role of meat-
eating in human evolution. Evol. Anthropol. 8, 11e21.
Moggi-Cecchi, J., Tobias, P.V., Beynon, A.D., 1998. The mixed
dentition and associated skull fragments of a juvenile fossil
hominid from Sterkfontein, South Africa. Am. J. Phys.
Anthropol. 106, 425e466.
Peters, C.R. and Vogel, J.C. Africa’s wild C4plant foods and
possible early hominid diets. J. Hum. Evol., in press
Puech, P.F., Cianfarani, F., Albertini, H., 1986. Dental
microwear features as an indicator for plant food in early
hominids: a preliminary study of enamel. Hum. Evol. 1,
Reed, K., 1997. Early hominid evolution and ecological change
through the African Plio-Pleistocene. J. Hum. Evol. 32,
Sage, R.F., Wedin, D.A., Li, M., 1999. The biogeography of
C4 photosynthesis. In: Sage, R.F., Monson, R.K. (Eds.),
Schoeninger, M.J., Bunn, H.T., Murray, S., Pickering, T.,
Moore, J., 2001. Meat-eating by the fourth African ape.
In: Stanford, C.B., Bunn, H.T. (Eds.), Meat-eating and
Human Evolution. Oxford University Press, Oxford,
Schoeninger, M.J., Moore, J., Sept, J.M., 1999. Subsistence
strategies of two savanna chimpanzee populations: the
stable isotope evidence. Am. J. Primatol. 49, 297e314.
Smith, B.N., Epstein, S., 1971. Two categories of13C/12C ratios
for higher plants. Plant Physiol. 47, 380e384.
Sponheimer, M., Lee-Thorp, J.A., 1999a. Isotopic evidence for
the diet of an early hominid, Australopithecus africanus.
Science 283, 368e370.
Sponheimer, M., Lee-Thorp, J.A., 1999b. The ecological
significance of oxygen isotopes in enamel carbonate.
J. Archaeol. Sci. 26, 723e728.
Sponheimer, M., Lee-Thorp, J.A., Reed, K., 2001. Isotopic
ecology of the Makapansgat Limeworks Perissodactyla.
S. Afr. J. Sci. 97, 327e329.
Sponheimer, M., Lee-Thorp, J.A., 2001. The oxygen isotope
composition of mammalian enamel carbonate: a case study
from Morea Estate, Mpumalanga Province, South Africa.
Oecologia 126, 153e157.
Sponheimer, M., Lee-Thorp, J.A., 2003. Differential resource
utilization by extant Great Apes and Australopithecines:
towards solving the C4 conundrum. Comp. Biochem.
Physiol. 136, 27e34.
Sponheimer, M., Lee-Thorp, J., DeRuiter, D., Smith, J., van
der Merwe, N., Reed, K., Ayliffe, L., Heidelberger, C.,
Marcus, W., 2003. Diets of southern African Bovidae: stable
isotope evidence. J. Mammal. 84, 471e479.
Sponheimer, M., 1999. Isotopic Ecology of the Makapansgat
Limeworks Fauna. Ph.D. Dissertation, Rutgers University.
Sponheimer, M., Reed, K., Lee-Thorp, J.A., 1999. Combining
isotopic and ecomorphological data to refine bovid paleo-
dietary reconstruction: a case study from the Makapansgat
Limeworks hominin locality. J. Hum. Evol. 34, 277e285.
Stock, W.D., Chuba, D.K., Verboom, G.A., 2004. Distribution
of South African C-3 and C-4 species of Cyperaceae in
relation to climate and phylogeny. Austral. Ecol. 29,
Stuart, C., Stuart, T., 2000. A Field Guide to the Tracks and
Signs of Southern and East African Wildlife. Stuik Publish-
ers, Cape Town.
Tackholm, V., Drar, M., 1973. Flora of Egypt. vol. II. Otto
Koeltz Antiquariat, Koenigstein.
Tayasu, I., Abe, T., Eggleton, P., Bignell, D.E., 1997. Nitrogen
and carbon isotope ratios in termites: an indicator of
trophic habit along the gradient from wood-feeding to soil-
feeding. Ecol. Entomol. 22, 343e351.
Tayasu, I., Inoue, T., Miller, L.R., Sugimoto, A., Takeichi, S.,
Abe, T., 1998. Confirmation of soil-feeding termites
(Isoptera; Termitidae; Termitinae) in Australia using stable
isotope ratios. Func. Ecol. 12, 536e542.
Ungar, P., 2004. Dental topography and diets of Australopi-
thecus afarensis and early Homo. J. Hum. Evol. 46,
van der Merwe, N.J., 1989. Natural variation in the 13C
concentration and its effect on environmental reconstruc-
tion using 13C/12C ratios in animal bones. In: Price,
T.D. (Ed.), The Chemistry of Prehistoric Human Bone.
Cambridge University Press, Cambridge,
van der Merwe, N.J., Thackeray, J.F., Lee-Thorp, J.A., Luyt,
J., 2003. The carbon isotope ecology and diet of Austral-
opithecus africanus at Sterkfontein, South Africa. J. Hum.
Evol. 44, 581e597.
Vrba, E.S., 1980. The significance of bovid remains as
indicators of environment
In: Behrensmeyer, A.K., Hill, A.P. (Eds.), Fossils in the
Vrba, E.S., 1985. Ecological and adaptive changes associated
with early hominid evolution. In: Delson, E. (Ed.),
Ancestors: The Hard Evidence. Alan R. Liss, New York,
Wood, B., Strait, D., 2004. Patterns of resource use in early
Homo and Paranthropus. J. Hum. Evol. 46, 119e162.
312M. Sponheimer et al. / Journal of Human Evolution 48 (2005) 301e312