New site at Olduvai Gorge (AGS, Bed I, 1.84 Mya) widens the range of locations where hominins engaged in butchery

Outstanding questions about human evolution include systematic connections between critical landscape resources—such as water and food—and how these shaped the competitive and biodiverse environment(s) that our ancestors inhabited. Here, we report fossil n-alkyl lipid biomarkers and their associated δ13C values across a newly discovered Olduvai Gorge site (AGS) dated to 1.84 million years ago, enabling a multiproxy analysis of the distributions of critical local landscape resources across an explicit locus of hominin activity. Our results reveal that AGS was a seasonally waterlogged, largely unvegetated lakeside site situated near an ephemeral freshwater river surrounded by arid-adapted C4 grasses. The sparse vegetation at AGS contrasts with reconstructed (micro)habitats at the other anthropogenic sites at Olduvai Gorge, suggesting that central-provisioning places depended more heavily on water access than vegetation viz. woody plants as is often observed for modern hunter-gatherers. As hominins at AGS performed similar butchering activities as at other Bed I sites, our results suggest they did not need the shelter of trees and thus occupied a competitive position within the predatory guild.

www.nature.com/scientificreports/ mid-(C 14 to C 22 ) and long-chained (C 24 to C 32 ) with a strong predominance of even-numbered homologues (Fig. 2B). Most samples' n-alkanol distributions are dominated by nC 18 OH that is characteristic of bacteria and freshwater algae inputs [24][25][26] . However, several distributions are rather dominated by phytoplankton-derived nC 24 OH 24,25 or commonly plant-derived nC 28 OH 25 , further clarifying the complex mixture of both aquatic and terrestrial lipid inputs. Common occurrences of nC 26 OH, typical of freshwater microalgae (Eustigmatophyceae) 24,26 , further indicates regular inundation with flowing potable water 18 at AGS. Previous studies suggest the ratio of n-hexacosanol (nC 26 OH) to n-nonacosane (nC 29 )-called the alcohol preservation index (API)-reflects changes in oxygenation at the sediment depositional interface 27,28 , wherein smaller values indicate more oxic conditions. At AGS, API values range from 0.2 to 0.4 ( Fig. 6A) that fall between average values of hypoxic (API > 0.4) and oxic (API < 0.2) conditions 27 . Based on these values, we interpret API values at AGS to indicate seasonal flooding pulses across a riparian wetland 29 that drove cyclic organic matter oxidation via oxygen exposure at the sediment-water interface after inundation 30 . Moreover, all samples from AGS lack detectable concentrations of the isoprenoidal lipids pristane and phytane. Because pristane/phytane are dominantly produced from chlorophyll degradation 31,32 , and (photo)degradation rates are rapid in most fluviolacustrine systems 33 , lacking pristane/phytane at AGS is consistent with (sub)seasonal redox oscillations, aerobic conditions and high-light intensities characteristic of barren medial floodplain channel deposits with little standing plant biomass 34 . n-Alkanoic acids. The polar fractions of AGS lipids show an archetypal bimodal distribution of saturated mid-and long-chain n-alkanoic acids (i.e., n-docosanoic acid [nC 22:0 ] to nC 26:0 and nC 28:0 to nC 32:0 , respectively) ( Fig. 2C) with an even-over-odd predominance that is indicative of mixed aquatic and terrestrial higher vegetation inputs, respectively 35 . Even so, shorter-chained homologues viz. nC 16:0 and nC 18:0 dominate among polar moieties at AGS. These shorter-chain acids are ubiquitous among plants, animals and fungi, but are especially prominent in algae and bacteria 21 . Therefore, we used the so-called terrigenous to aquatic ratio (TAR FA )-the ratio of long-chain versus shorter-chain alkanoic acids 35 -for determining proportional lipid input from disparate organic matter sources. Low calculated TAR FA values of 0.05 (Fig. 6F) at AGS indicate aquatic or algal lipids dominated biomass input 35 despite its shoreside setting. The diagnostic occurrence of mono-and polyunsaturated nC 22 alkanoic acids suggest there was a major input from microbes and freshwater phytoplankton 26 , with further supports our interpretation of AGS as a riverside wetland that experienced frequent flooding.
Stable carbon isotope signatures among plant biomarkers. Alkane δ 13 36 . Macrophytes with aquatic (e.g., Hydrilla), floating (e.g., Potamogeton), and emergent (e.g., Cyperus and Typha) habits are common in recent East African wetlands 19 , and have similar δ 13 C values as we observe at AGS 19,37,38 . However, macrophytic carbon sources are often influenced by partial incorporation of dissolved HCO 3 − (which causes increased, C4-like δ 13 C values) in aquatic environments, and likewise are affected by physiological differences among plant growth forms and functional types 19,39,40 that lead to interpretational difficulties of δ 13 C 31 values vis-à-vis differentiating plants with bicarbonate uptake mechanisms versus cooccurring graminoids with C4 photosynthetic pathways.
To determine source and input differences among plant-lipid biosignatures, we applied basic analysis of variance (ANOVA) to our measured δ 13 C values for odd-numbered homologues (i.e., nC 23 , nC 25 , nC 27 , nC 29 , nC 31 ) in sediments at AGS. In this context, the mid-chain n-alkanes (nC 23 and nC 25 ) (Fig. 6G) are derived from submerged/floating macrophytes, and longer-chain n-alkanes (> nC 25 ) are dominated by plants with terrestrial habitats 39 . Resultant δ 13 C ANOVA data demonstrate a statistical difference (p-value < 0.01) between δ 13 C 31 as compared to contemporaneous mid-chain and longer-chain n-alkanes. Our results suggest that nC 31 had a discrete source with more 13 C-enriched lipids-such as arid-adapted C4 graminoids 41,42 -as compared to other n-alkanes. In contrast, the other mid-and longer-chained homologues do not show statistical differences in average isotopic composition. With this in mind, we suggest that the algae and submerged/floating aquatic plants at AGS incorporated higher amounts of dissolved HCO 3 − as a source of carbon, resulting in heavier δ 13 C values in mid-chain n-alkanes 40 . Complementary proxies viz. n-alkane ratios (Figs. 3A,B, 6A-E) further suggest that AGS functioned like a seasonal-to-permanent freshwater-fed floodplain or river margin dominated by graminoids and C3-like macrophytes with sparse woody plants and limited C4 grasses. This is in accordance with the geological and sedimentological description that suggested a playa lake margin with small rivers with low-energy 11,16 .
The Zinj Paleolandscape and hominin evolution. At least three hominin taxa-including Homo cf.
ergaster, Paranthropus boisei and Homo habilis 11,46,47 -have been recovered from the Zinj Paleolandscape. Across this contiguous archaeological horizon, artifacts and fossil bones are embedded within a distinctive silty-clay layer covered by airfall Tuff IC 11 , which suggests that the remains were captured in situ under volcanic ash fallout. Rapid burial fostered the exceptional preservation of fine-sediment features (e.g., root ichnofossils) 16 , macrofossils and biomarkers across its surface 11,16 . Sedimentological features at AGS further reveal that this site was situated at littoral-supralittoral interface, where alluvial fans and floodwater river runoff interposed the www.nature.com/scientificreports/ mudflat littoral zone 16 . Altogether, such multidisciplinary proxies reveal that Olduvai, and the Zinj Paleolandscape especially, presented hominins with a variegated landscape of heterogeneous resource distributions, and thus impacted hominin diets and behaviour.
Diet is considered to serve as a major selective force amid hominin encephalization 48 , and the intake of 'essential' lipids (e.g., polyunsaturated fatty acids), which are uniquely prevalent in wetland flora 49,50 , is critical to cerebral development in modern infant humans 51 . Aquatic macrophytes proliferate in wetlands today in East Africa 19,52 and likely provided hominins with high-quality provision viz. rootstocks and leaves all year long 53,54 . Contemporary pan-African macrophytes rarely use dedicated C4 photosynthesis; but, these (semi)aquatic plants can incorporate bicarbonate when in alkaline waters and, in doing so, create C4-like δ 13 C isotopic signature 40 . Mid-chain n-alkanes from AGS reveal 13 C-enriched biomass (Fig. 4), and that their source vegetation (i.e., macrophytes) can account for high δ 13 C signatures in hominin diets, as reflected through bioapatite isotopes and dentition analyses [55][56][57][58] .
Our molecular and isotope data (Figs. 3A, Fig. 6A,C) suggest there was a intermittently-waterlogged floodplain environment at AGS, which is consistent with earlier mineralogical descriptions at the vicinal DS site (Fig. 1C) 59 . We suggest there was a notably perennial large stream or river near AGS that flowed from southto-north 16 into nearby paleolake Olduvai, and brought with it the allochthonous plant lipid from nearby grasslands and forest patches 13,15 . Given overall low relief in the immediate area around AGS, which likely produced lower-energy paludal flow, entrainment of allochthonous materials most likely peaked through seasonal flooding pulses 60 . In any case, associated freshwater fluvial input must have attracted both hominins and large carnivores alike to AGS.
Biomarkers at AGS build upon earlier reconstructions at Olduvai 6,17,22 and the Zinj Paleolandscape 15 that imply plants and freshwater distribution exerted direct influence on hominin behaviour. However, in contrast to earlier reconstructions among penecontemporaneous sites at Olduvai Gorge, our data at AGS indicates hominins took strategic advantage of low-or unvegetated locations in addition to dense woodland thickets, low-visibility www.nature.com/scientificreports/ papyrus stands, and tall grassland 13,61 . AGS itself likely harboured occasional large woody plants such as Hyphaene palms 13 , but the abundance of this vegetation would have been nominal 62 .
The occurrence of discrete plants on otherwise unvegetated fluviolacustrine surfaces might be one explanation for differing paleoecologic reconstructions with biomarkers, which integrate immediate and upstream organic matter inputs, as compared to, e.g., phytoliths, which capture only productive in situ vegetation and do not transport downstream 63 . Regardless, vicinal (over)bank deposits and levee ridges would be dominated by generally groundcover vegetation 29 such as macrophytes, pteridophytes (e.g., ferns) and low woody plants, meaning that hominins at AGS would have an unencumbered view of the surrounding landscape 64 , including precious refuge about 400 m away at FLK Zinj itself 13,15,16 (Fig. 1C). Biomarkers across the Zinj Paleolandscape 15 and from the AGS site show distributions of plants that suggests critical resources had a direct implication on hominin behaviour at Olduvai Gorge.
The AGS archaeological site contains one of the highest densities of faunal remains on the Zinj Paleolandscape after DS, which to date is the biggest window into an Early Pleistocene anthropogenic site 65 . This suggests that the area at AGS must have been occupied for repeated instances of large carcass consumption 65 . To boot, AGS was dominated by (semi)aquatic plants, minimal trees or shrubs, although in all likelihood hominins were occupying the site especially for its vicinity to flowing water. In contrast, FLK NN (500 m west from AGS) contained limited faunal remains and lithics within an environment that featured floating and submerged aquatic plants 15 located nearby freshwater carbonates (tufa 8 ). If AGS was not a terrestrial plant-dominated habitat, because of its barren vegetation, and hominin and carnivore visibility were similar, as in open grasslands, this posits the question of how hominins efficiently avoided carnivore hazards for the prolonged occupation(s) inferred from the intensity of ungulate carcass butchery documented at the site (Fig. 7). This is mostly notable, especially given the paucity of taphonomic evidence signalling carnivore modification of carcass remains. The evidence is suggestive of hominins having carved a competitive niche against other predators by efficiently fending off their hazard.
By comparing our results at AGS to previously published data about biomarker distributions amid the Zinj Paleolandscape 15 , we suggest that AGS functioned as a seasonally waterlogged, low-vegetation environment characterised by dense accumulations of butchery-process debris within a wider mosaic environment that harboured both open 5,15 and dense, closed ecotones, such as at DS [57]. FLK Zinj, DS and PTK are the other three anthropogenic sites occupying the same thin clay stratum as AGS on the Zinj Paleolandscape [63], [107]. These three sites were created in a wooded to forested alluvial environment [57], [108]. The close spatial association between stone tools and fossil bones indicates a functional relationship resulting from butchery. This is further supported by the abundant cut-marked and percussed bones retrieved in the archaeofaunal assemblage, which is currently under analysis. Given that in the other anthropogenic sites, with areas sampled similar to the size of the area sampled at AGS, the paleobotanical signal is of closed vegetation, we assume that the area sampled at AGS is equally representative of the main vegetational characteristics of the setting. In this case, the signal retrieved, in contrast with the other anthropogenic middle Bed I sites is of a fairly open environment. The butchery patterns documented in those sites are identical to those observed at AGS (work in progress), with a similar number of animals and taxa represented and a similar single-round cluster site structure. This latter feature is clearly showing a carcass butchery and consumption pattern and a distinctive use of the space that contains crucial socio-reproductive information [109]. To conclude, we suggest the spatial landscape ecology patterns www.nature.com/scientificreports/ defined both by macro-and molecular fossils reveal hominin engagement in social transport of large resources, such as bringing animal carcasses and freshwater-sourced food to AGS from the surrounding grasslands and lakeside environments.

Conclusions
Our multi-proxy geochemical interpretations of the new site called AGS at Olduvai Gorge ca. 1.84 Ma reveal that it was a sparsely vegetated, high-visibility paludal location on a riverbank within a wider mosaic environment defined by the Zinj Paleolandscape. Standing plant biomass at AGS was dominated by aquatic vegetation viz. macrophytes and C4 graminoids, consistent with its riparian and lake margin setting. Comparison with literature data for coeval localities at Olduvai Gorge suggest that this ancient landscape was rich in ecotones with abrupt transitions between disparate vegetation communities, as is often observed today in East African wetlands. Given the markedly patchy paleoenvironment that defined the vicinal Zinj Paleolandscape and the unique (sub)seasonal floodplain at AGS, we conclude early hominins at Olduvai Gorge selected locations for cooperative resource processing-such as animal butchery-as related to water resources rather than refuge (i.e., closed thickets). This conclusion diversifies the environments in which vertically discrete anthropogenic sites occur and furthermore insinuates that hominins felt equally at ease in such environments. Considered together, new and old data at Olduvai reveal that hominins had reached an adaptive carnivore status by 1.84 Ma that enabled them to cope with terrestrial predation risks and fend off other carnivore competitors.

Materials and methods
Sampling. Samples were collected from sediment 0-2 cm below Tuff IC that is isochronous with Zinj level 22A, in a silty-clay layer where bones and lithic tools occurred in situ (Fig. 1C). Representative sediment samples (~ 50 g; n = 24) from individual 1 m 2 gridding quadrats (Fig. 1D) were collected between fossil remains and stone tools with a clean aluminium spoon. . The GC oven was programmed to: 60 °C injection and hold for 2 min, ramp at 10 °C min to 150 °C, ramp at 4 °C min to 300 °C followed by isothermal hold of 20 min. The transfer line and source were set at 320 °C and 270 °C, respectively. Procedural blanks were run to monitor contamination and background interferences. Compound identifications were made via comparison with authentic standards in conjunction with the NIST20 electron ionization spectral library.

Biomarker extraction and isolation.
Isotopic characterization. Target biomarkers were analysed for their isotopic signatures by gas chromatography combustion isotope-ratio monitoring mass spectrometry (GC-C-irMS) using a TRACE 1310 interfaced to a GC Isolink II connected to a Conflo IV and Delta V Plus. The GC oven was programmed to: 60 °C injection and hold for 2 min, ramp at 4 °C min to 320 °C followed by isothermal hold of 10 min. An aliquot of 1 μL was injected in splitless mode onto a 30-m DB5ms column (0.25 mm × 0.25 μm) before combustion over copper, nickel, and platinum wire with oxygen and helium at 1000 °C. Isotopic values were normalized and corrected using a mixture of n-alkanes (nC 16

Data availability
All data generated or analysed during this study are included in this published article or in the accompanying Supplementary Information file.