The stable isotope setting of Australopithecus sediba at Malapa, South Africa

HOW TO CITE: Holt E, Dirks P, Placzek C, Berger L. The stable isotope setting of Australopithecus sediba at Malapa, South Africa. S Afr J Sci. 2016;112(7/8), Art. #2015-0351, 9 pages. http://dx.doi.org/10.17159/ sajs.2016/20150351 We report δ13C and δ18O results from carbonate-cemented cave sediments at Malapa in South Africa. The sediments were deposited during a short-period magnetic reversal at 1.977±0.003 Ma, immediately preceding deposition of Facies D sediments that contain the type fossils of Australopithecus sediba. Values of δ13C range between -5.65 and -2.09 with an average of -4.58±0.54‰ (Vienna Pee Dee Belemnite, VPDB) and values of δ18O range between -6.14 and -3.84 with an average of -4.93±0.44‰ (VPDB). Despite signs of diagenetic alteration from metastable aragonite to calcite, the Malapa isotope values are similar to those obtained in two previous studies in South Africa for the same relative time period. Broadly, the Malapa δ13C values provide constraints on the palaeovegetation at Malapa. Because of the complex nature of the carbonate cements and mixed mineralogy in the samples, our estimates of vegetation type (C4-dominant) must be regarded as preliminary only. However, the indication of a mainly C4 landscape is in contrast to the reported diet of A. sediba, and suggests a diverse environment involving both grassland and riparian woodland.


Introduction
Palaeoclimate and palaeoenvironment studies are important for identifying the drivers of hominin evolution. 1,2 Long-term shifts and variability in climate are linked to changes in floral and faunal assemblages, which have been matched with significant points in hominin development. [1][2][3] Studies that investigate the link between climate and hominin evolution have focused on Africa because of its relatively extensive hominin fossil record, which extends back millions of years and coincides with significant palaeoenvironmental changes. 2 Caves are prevalent in the Cradle of Humankind (CoH) area of South Africa. These caves are important because they contain records of changes in climate 4 and they host hominin fossils 5 . The caves act as natural traps for floral and faunal remains, capture wind-and water-borne sediments, and are host to carbonate cave deposits. 6,7 In a number of cave settings, fossils and sediments have been preserved in a definable stratigraphic sequence, and owing to the relatively stable nature of the cave these layers have remained undisturbed over long periods of time. 6 Thus, CoH caves are invaluable for analysing the links between changes in climate and terrestrial environments with hominin evolution. 8 Stable isotope proxies in carbonate cave deposits have been used to investigate late Pliocene and early Pleistocene climate and environmental conditions in South Africa. 9,10 In the Makapansgat Valley, an area rich in hominin fossils, Hopley et al. 9 used oxygen isotope values to time-constrain cave carbonate (flowstone) deposition at Buffalo Cave, and to determine the major orbital cycles influencing climate and vegetation patterns. In addition, the flowstone carbon isotope values and organic matter contained therein were used to ascertain changes in the dominant vegetation type (i.e. C 3 versus C 4 ) during the early Pleistocene. A major finding by Hopley et al. 9 was an increase in C 4 vegetation in samples less than 2 million years (Ma) old, with the most significant increase in C 4 vegetation after 1.7 Ma. 9 Similarly, Pickering et al. 10 used carbon isotope values from cave carbonate deposits at Gladysvale Cave in the CoH. They reported a mixed C 4 -dominant CoH landscape and a cool dry environment during the early Pleistocene (~1.8±0.7 Ma).
The ability of cave carbonates to reveal information about Plio-Pleistocene climate and vegetation has been established. However, studies from the modern summer rainfall region of South Africa are lacking. Limestone mining has displaced key stratigraphic sequences, which has led to dislocation of cave deposits and the proxies they contain, from important faunal fossils. 9,11 Challenges in defining stratigraphy are compounded by the difficulty in forming time-depth series because of the uneven rates of sedimentation and mineral growth, and the shortage of appropriate techniques for absolute dating. 7,12 Furthermore, in dolomitic areas such as the CoH, interpretations of oxygen and carbon isotope values from carbonates are affected by post-depositional diagenetic alteration of metastable aragonite to calcite. 13 Despite the difficulties presented by the setting, cave carbonate studies in the CoH remain important for linking climate and environmental change with hominin evolution. Expansion of stable isotope studies to carbonatecemented cave sediments associated with hominin fossils in the CoH provide a valuable source of information for palaeoclimate studies. In this paper, we present stable oxygen and carbon isotope data from Malapa in the CoH. The isotope data were obtained from a thin flowstone drape and carbonate cements, deposited within a cave setting in close proximity to the type fossils of A. sediba. 7,14 Based on a combination of uranium-series dating and palaeomagnetic methods, 7,14 the depositional age of the tested flowstone drape coincides with the age obtained for the A. sediba fossils at 1.977±0.003 Ma.
We compared the raw stable isotope results with the results from studies at Gladysvale 10 and Buffalo Cave 9 . Modelling was applied to the carbon isotope values in an effort to place constraints on the dominant vegetation type and past environment overlying the cave. Our results provide an initial assessment of possible vegetation conditions during the time when A. sediba was alive. In addition, important textural and mineralogical constraints are defined, providing guidance for future studies.

Site location
Malapa is located approximately 40 km northwest of Johannesburg at 25°52'S, 27°48'E (elevation 1440 masl) in the Grootvleispruit catchment, and is hosted by chert-free dolomite of the Palaeoproterozoic Lyttelton Formation in the Malmani Group ( Figure 1). 7 The site comprises two small pits excavated by miners in the early part of the 20th century. 14 The larger pit, Pit 1, measures approximately 20 m 2 on the ground and 4 m deep, and hosts the type fossils of A. sediba. 15 Pit 1 was the source of samples obtained in our study. A second smaller pit, Pit 2, which is roughly 12 m 2 in size and 1 m deep, is situated nearby and also hosts hominin remains. 7,14 Current climate and vegetation Pretoria (elevation 1330 masl) is the closest monitored weather station to Malapa, and demonstrates a clear seasonal pattern in temperature and monthly rainfall. The highest temperatures occur from December to February (local summer), with February recording an average daytime temperature of 23.5 °C (1999-2013). 16 July (local mid-winter) is the coldest month, with an average daytime temperature of 12.6 °C. 16 The mean annual temperature (MAT) is 19.3 °C and the average annual rainfall is 700 mm, with most rain occurring in the summer months. 16 Vegetation cover in the CoH is a consequence of precipitation, geology, MAT, and location within the landscape. [17][18][19] Malapa is situated within the Carleton Dolomite Grassland, which is dominated by a single layer of C 4 photosynthetic process grasses. 19 However, this area also has a high level of species richness because of the heterogeneous nature of the dolomitic landscape. 19 A range of woodland communities occurs in specific areas, 18,19 such as near cave entrances 18 and at spring sites or along streams. In addition, C 3 -type grasses are common at relatively high and cool elevations, and in areas that experience winter rainfall. 17,19 Depositional setting and stratigraphy The depositional history and stratigraphy of the Malapa site has been well described. 7,14 The type fossils of A. sediba (MH1 and MH2) 15 were found within poorly-sorted sandstone of Facies D, which has been interpreted as a mass flow deposit. 7 A flowstone unit from Pit 1, measuring 5 cm to 20 cm in thickness (Flowstone 1), has been dated to 2.026±0.021 Ma 7 and presents an important chronostratigraphic marker for Facies D. Flowstone 1 consists of a single sheet in the southeast corner of the pit, and splits first into two and then into three separate sheets towards the centre of the pit, where the top sheet directly underlies the A. sediba fossils contained in Facies D. Overlying Facies Da sediments are other sediments that belong to the finer-grained, weakly layered topmost section of Facies D (which were originally described as Facies C). 7 The sediments of Facies D also record intermediate and normal polarity, indicating that Facies Da and Facies D were both deposited during a brief magnetic reversal (the Pre-Olduvai event) at 1.977±0.003 Ma. 14 Facies D sediments are overlain by horizontally laminated, poorly sorted muddy sandstone of Facies E, which is also fossiliferous. 7 Along the northwest wall of Pit 1, the sediments of Facies Da, D and E overlap an erosion remnant of peloidal fine-grained muds, which remain undated and belong to Facies C. 7,14 Textural evidence indicates that the sedimentary units composed of Facies C, Da, D and E were deposited in a water-saturated environment. 7

Samples
As shown in Figure 2, four samples from Pit 1 were studied for stable isotope analysis: UW88-PM09 (PM09), PM09-2 (a duplicate of PM09), UW88-PM04 (PM04) and PM04-2 (a duplicate of PM04  The studied samples exhibit a number of relatable features, including an intercalated flowstone-sediment unit, which represents the top of Facies Da and the abrupt transition to Facies D. The inferred stratigraphic relationships between the samples indicate that they comprise a full profile of deposition that directly preceded (Facies Da) or coincided (Facies D) with the burial of A. sediba.
Feigl's staining of thin sections of sample PM09 revealed mixed carbonate mineralogy, evidence of post-depositional diagenesis, and a varying depositional environment ( Figure 3). Evidence of primary acicular aragonite (CaCO 3 ) is common in all samples; however, the aragonite has largely been recrystallised to or replaced by sparry columnar calcite (CaCO 3 ). Thin carbonate layers in Facies Da, void fillings (laminated) in Facies Da and Facies D, and the flowstone drape separating Facies C and Facies Da show calcite-preserving relics of aragonite crystals. These crystals take various forms, including botryoidal, acicular or micrite crystals. Where aragonite relics are preserved, the diagenetic process is interpreted to be the result of calcitisation by thin water films. 13 However, complete dissolution of primary aragonite and replacement by secondary calcite 13 has also occurred, as evidenced by the lack of aragonite relics in parts of all facies within the samples. Crystal form confirms that Malapa was subject to fluctuating water flow and sediment input, resulting in a mixed phreatic-vadose depositional environment that was conducive to both carbonate deposition and post-depositional changes.

Sample preparation for stable isotope analysis
A total of 99 carbonate sub-samples were collected for stable isotope analysis from flowstone layers and carbonate cement within the samples described above. Samples were obtained by hand, using a standard engraving drill with removal bits measuring between 1 mm and 3 mm. Cross-contamination between sites was reduced by discarding surface material and rinsing the drill bits in 5% nitric acid, deionized water and ethanol, respectively. Carbonate powders were collected in PCR tubes, then weighed and transferred into clean glass sampling tubes.
Stable isotope analyses were performed at the Advanced Analytical Centre at the James Cook University in Cairns, Australia. The equipment used was the Thermo Scientific Delta V gas source isotope ratio mass spectrometer (IRMS) together with GasBench III and Conflo IV interfaces. Carbonate samples were digested in 99% phosphoric acid at 25 °C, and the resulting CO 2 gas was analysed after equilibrating for 18 h. Results were normalised to Vienna Pee Dee Belemnite (VPDB) using the calibrated reference materials of NBS 19 and NBS 18, and were reported in parts per mil (‰) with delta (δ) notation. Mean analytical precision of repeat reference materials is ±0.1‰ for both δ 18 O and δ 13 C.

Replication
Flowstones and carbonate-cemented sediments are not normally subject to Hendy criteria tests. This is because of the variable nature of growth, including difficulty in defining vertical versus lateral advances over time. 20 For these types of samples, replication has been suggested as a more purposeful and valuable method in stable isotope studies. 21 In our study, sample PM09-2 was chosen for replicate analysis of the primary sample UW-PM09. The methodology for sample collection and data analysis of the replicate was the same as for the original samples.

Modelling palaeovegetation at Malapa
The values of δ 18 O and δ 13 C in cave deposits are governed by a complex set of variables and processes. 22 28 However, the accurate application of equilibrium fractionation factors requires that either the source water or soil-gas stable isotope value, or the temperature at time of deposition must be known. 26 where α (alpha) = fractionation factor, which is temperature-dependent; and the values for stable isotopes are relevant to the same standard.
Constraining all parameters that influence stable isotope fractionation in palaeocave settings such as Malapa is not possible, if the cave environment and source-water chemistry are unknown. 26 Although a number of these factors, including the depositional environment, can be determined using petrographic studies and statistical analysis of stable isotope results (i.e. Hendy criteria tests for kinetic fractionation), others must be estimated. Examples of estimable factors include sourcewater δ 13 C values and cave temperature. Here, we outline the estimable variables and fractionation factors used to calculate source carbon δ 13 C values from the Malapa stable isotope values.

Fractionation factors
The primary equilibrium fractionation factors chosen for carbon source calculations in our study were those of Romanek et al. 32 These factors, which include calcite-CO 2(g) and aragonite-CO 2(g) equilibrium reactions, were obtained experimentally 32 and have been used in travertine studies to determine dissolved organic and inorganic carbon components. 33 In addition, we used re-evaluated factors for stepwise calcite-CO 2(g) fractionation 34 and back-calculated factors for step-wise aragonite-CO 2(g) fractionation. 32,34 Cave temperature Cave interiors have minimal fluctuation in temperature, with cave temperature equalling surface MAT. 21,35 Estimates of temperature at the time of cave carbonate deposition are based on known changes in MAT during glacial and interglacial periods. The minimum temperature in southern Africa during the Last Glacial Maximum (LGM) is consistent with the global temperature decrease, and has been quantified as MAT minus 6 °C. 36 This estimate has been obtained by analysing noble gases in groundwater. 36  The samples in our study were deposited within a short period of 6000 years, at 1.977±0.003 Ma. During this period global temperature might have been relatively cool, as evidenced by changes in benthic δ 18 O values ( Figure 4). 37 Tectonic uplift and local insolation also have a bearing on MAT; however, the quantification of these effects at 1.977±0.003 Ma is currently unattainable. We estimated the minimum temperature at Malapa for the studied period as having been 12.1 °C, using the current MAT (calculated as 18.1 °C) 16,38 and the quantified minimum at the LGM. In addition, a maximum temperature of MAT minus 0.5 °C can be assumed, based on the southern African response to the LGM. 36 Host rock contribution to δ 13

C values
The contribution of host rock δ 13 C to the dissolved inorganic carbon (DIC) pool in source water can be as high as 50%, depending on various factors (e.g. source-water pH, residence time, and host rock mineralogy). 25,30 A greater contribution of host rock δ 13 C to the DIC pool leads to more strongly positive δ 13 C values in the tested carbonate, 26 which skews the results in favour of a C 4 vegetation-dominated source. The dissolution and fractionation of dolomite, such as that which occurs at Malapa, is not welldefined. However, calcite and low magnesium calcite levels are considered to be influential components in the δ 13 C value of source water as affected by dolomite host rock dissolution. 27,39 For this reason, we used the calcite fractionation factor of Mook 34 to calculate the host rock contribution to DIC in source water, in addition to an average δ 13 C value of -0.74‰ that was obtained from two dolomite samples at Malapa. The relative contribution of host rock dissolution to the final δ 13 C source-water value was calculated using a simple percent contribution calculation (i.e. 10% host rock + 90% soil CO 2 = source-water δ 13 C value).

Stable isotope values
The stable isotope values for the four samples are presented in Table 1. These values demonstrated a relatively small range for δ 18  Replicate stable isotope results of Facies Da in UW-PM09 and PM09-2 were examined using the Welch modified 2-sample t-test. 40 The results confirmed that the mean values for δ 18 O and δ 13 C in the two samples differed significantly (α=0.05).

Source carbon values
The calculated values of source δ 13 C varied 1‰ according to the different estimated temperatures at deposition and the fractionation factors we used. The minimum temperature estimate (12.1 °C) yielded slightly more negative values of δ 13 C compared with the maximum temperature estimate (17.6 °C) results. In addition, the re-evaluated fractionation factors 34 resulted in more strongly positive calculated δ 13 C values compared with the experimentally obtained factors. 32 The greatest effect on calculated source δ 13 C values was the contribution of host rock dissolution to the DIC pool. The modelling demonstrates that as the estimated host rock contribution increases from 0% to 50%, so the calculated source δ 13 C values move towards the C 3 vegetation range (i.e. becoming more negative). This result can be explained by the model's accounting for the effect of host rock DIC on the δ 13 C values of the tested carbonate. For a host rock contribution to the DIC pool of up to 50%, the calculated source δ 13 C values for Malapa fit predominantly within the C 4 vegetation range ( Figure 5).

Discussion
When carbonate samples are unaltered and are deposited in isotopic equilibrium with the cave environment, variations in the δ 18 O values are linked to changes in palaeotemperature and palaeohydrology. 30,41 The δ 13 C values reflect the vegetation type and interaction with the host rock. 30 Climate change and cyclicity in Plio-Pleistocene southern Africa have been linked to, and primarily attributed to, North Atlantic sea surface temperatures and high-latitude ice volumes. 44 The Earth's orbital cycles also played a role. 9 In particular, the influence of orbital precession on monsoonal patterns during that period is considered to have been a major driver of climate, 9 especially prior to 2.8 Ma when 23 000-year cycles dominated. 45 The vegetation type fluctuated in response to glacial (stadial) and interglacial periods and the associated changes in rainfall and MAT, with shifts in dominant vegetation occurring rapidly. 46 Hopley 9 suggests that after 1.7 Ma, a shift occurred towards grassland (C 4 vegetation) and away from forested landscapes (C 3 vegetation), with increasing aridity. This theory is in keeping with the change towards higher amplitude 40 000-year cycle glacial periods after 1.7 Ma. 45 The suggestion that the landscape in the region of Malapa at 1.977±0.03 Ma was dominated by C 4 vegetation aligns with estimates for palaeovegetation patterns at Gladysvale Cave, albeit for the later time of ~1.8 Ma (Figure 6). 10 9 However, deep vertical convection, ascribed to movement of the intertropical convergence zone and subtropical highs throughout the seasonal year 50 , also lead to a decrease of δ 18 O values in precipitation and therefore also in cave carbonates. Modern data from the Global Network of Isotopes in Precipitation for Pretoria show that as rainfall and temperatures increase, the δ 18 O value in precipitation decreases -known as the rainfall 'amount effect' (Figure 7). [51][52][53] Furthermore, there is a strong correlation between increased rainfall in East Africa as a consequence of a positive Indian Ocean dipole, and significant depletions in δ 18 O values in precipitation. 52

Implications for palaeovegetation-evolution linkages
The suggested palaeovegetation at Malapa, as reconstructed in this paper, provides an important insight into the ecological niche that A. sediba occupied. In studying pollen and phytolith remains recovered from plaque on teeth of A. sediba fossils, Henry et al. 54 found that their diet consisted wholly of leaves, fruits, the bark of trees, and herbaceous plants (C 3 vegetation). The diet of A. sediba is at odds with the diet of other hominin species (such as A. africanus) 55 in the CoH, including those from the same time period. 54 Therefore the location of Malapa, in a relatively sheltered valley near the confluence of two streams, might have provided the ideal riparian environment for A. sediba in an otherwise C 4 -dominated grassland.

Conclusion
Stable isotope values of cave carbonates at Malapa during the early Pleistocene are similar to those from two previous studies conducted in the summer rainfall region of South Africa. The δ 13