PALEVO 1631-0683 Elsevier S1631-0683(14)00104-3 10.1016/j.crpv.2014.06.003 Research article Human palaeontology and prehistory Human palaeontology and prehistory Spatial behaviours of Early Oldowan toolmakers in the Shungura Formation (Lower Omo Valley, Ethiopia): Proposal for an integrated approach Comportements spatiaux des tailleurs de pierre de l’Oldowayen ancien dans la Formation de Shungura (basse vallée de l’Omo, Éthiopie) : proposition pour une approche spatiale intégrée Maurin Tiphaine tiphaine.maurin@u-bordeaux.fr a Delagnes Anne anne.delagnes@u-bordeaux.fr a b Boisserie Jean-Renaud jean.renaud.boisserie@univ-poitiers.fr c d Université de Bordeaux, UMR 5199 PACEA, bâtiment B8, allée Geoffroy-Saint-Hilaire, CS 50023, 33615 Pessac cedex, France Université de Bordeaux, UMR 5199 PACEA, bâtiment B8 allée Geoffroy-Saint-Hilaire, CS 50023 Pessac cedex 33615 France Evolutionary Studies Institute, University of the Witwatersrand, 1, Jan Smits Avenue, Braamfontein 2000, Johannesburg, South Africa Evolutionary Studies Institute, University of the Witwatersrand 1, Jan Smits Avenue Braamfontein Johannesburg 2000 South Africa Université de Poitiers, UMR CNRS 7262, UFR SFA, bâtiment B35, TSA 51106, 6, rue Michel-Brunet, 86073 Poitiers cedex 9, France Université de Poitiers, UMR CNRS 7262, UFR SFA, bâtiment B35, TSA 51106 6, rue Michel-Brunet Poitiers cedex 9 86073 France Centre français des études éthiopiennes/French Centre for Ethiopian Studies, CNRS USR 3137, Ambassade de France en Éthiopie, P.O. Box, 5554 Addis Abeba, Ethiopia Centre français des études éthiopiennes/French Centre for Ethiopian Studies, CNRS USR 3137, Ambassade de France en Éthiopie P.O. Box Addis Abeba 5554 Ethiopia 13 8 S1631-0683(14)X0008-4 737 746 © 2014 Académie des sciences. Published by Elsevier B.V. All rights reserved. 2014 Académie des sciences Full (PDF)

The spatial relationships of the Early Oldowan toolmakers with their environment have been so far addressed through raw material procurement analyses and the characterization of hominid habitat. This paper proposes to integrate these two approaches into a broader spatial analysis encompassing archaeological and environmental data (palaeontological, geological and isotopic data) from Member F and lower Member G of the Shungura Formation (Lower Omo Valley, Ethiopia). Heterogeneity in data resolution induces a multiscale approach with three levels of analysis. The level of occurrence complex allows focusing on the characterization of archaeological occurrences and on their environmental settings. The level of “study area” allows working on hominid habitats and on their raw material procurement behaviours. Finally, at the Shungura Formation scale, we can address temporal issues related to the evolution of spatial behaviours between Member F and the lower part of Member G, ca. 2.3 to 2 million years (Ma).

Les relations spatiales des tailleurs de pierre de l’Oldowayen ancien avec leur environnement ont fait jusqu’à présent l’objet de recherches axées sur l’acquisition de la matière première et sur l’habitat des hominidés. Ce papier développe les fondements d’une approche spatiale intégrant ces deux aspects au sein d’un travail croisant les données archéologiques et environnementales (paléontologiques, géologiques et isotopiques) du Membre F et de la partie inférieure du Membre G de la Formation de Shungura (basse vallée de l’Omo, Éthiopie). Ces données se situant à différents niveaux de résolution spatiale, une approche multi-scalaire est proposée : l’échelle du complexe d’occurrences archéologiques pour la caractérisation des occurrences archéologiques et de leur cadre environnemental, celle de « l’aire d’étude » pour laquelle vont être apportées des précisions sur l’habitat des hominidés et leurs approvisionnements en matière premières, et enfin celle de la Formation de Shungura pour confronter entre elles les données spatiales acquises dans le Membre F et dans la partie inférieure du Membre G, entre environ 2,3 et 2 millions d’années (Ma), dans une optique temporelle.

 Early Oldowan, Spatial behaviours, Environment, Stone tools, Eastern Africa, Lower Pleistocene, Shungura Formation Oldowayen ancien, Comportements spatiaux, Environnement, Outils en pierre, Afrique de l’Est, Pléistocène inférieur, Formation de Shungura presented Handled by Marcel Otte Introduction

Over most of their evolutionary history, the role of hominids within the African Neogene ecosystems remains elusive, because they are documented only by rare and mostly fragmentary remains. With the appearance of the Early Oldowan stone tool record prior to 2 Ma, it becomes possible to track down hominid activities within fossil landscapes, enhancing our capacities in understanding key features of hominid adaptive success, such as behaviours, ecology, responses to environmental changes.

Questions about the taxon (or the taxa) of the Early Oldowan toolmakers are still open. Between 2.3 and 2 Ma, hominid fossils from the Omo-Turkana basin (Howell et al., 1987), whether related to the palaeo-Omo river vicinity (Shunugura Formation) or from the palaeo-Turkana vicinity (Nachukui Formation), correspond to robust Australopithecus and early Homo (Brown and Feibel, 1988, Coppens, 1970 and Prat et al., 2005).

So far two types of approaches have been favoured in the analysis of the spatial relationships of the early hominids to their environment. One of them focuses on the exploitation of raw material sources, mostly in the vicinity of the archaeological occurrences in Early Oldowan contexts (Bishop et al., 2006, Braun et al., 2008, Delagnes et al., 2011, Goldman-Neuman and Hovers, 2012, Harmand, 2005, Plummer et al., 1999 and Stout et al., 2005). The other approach aims at characterizing the landscapes inhabited by hominids at larger scales (Aronson et al., 2008, Bonnefille, 2010, Brugal et al., 2003, López-Sáez and Domínguez-Rodrigo, 2009, Plummer et al., 1999, Plummer et al., 2009, Quinn et al., 2013 and Reynolds et al., 2011). Although these studies deal with spatial behaviours, no spatial approach integrating all data related to resource procurement mobility (i.e. water, food and raw material) and habitat features have been developed so far for this time period. Such a holistic approach can bring significant insights into the relationships between early toolmakers and their environment, and on the evolution of these relationships.

Thanks to the rich corpus of geolocated archaeological and contextual data (palaeontological, geological and isotopic data) available for the Shungura Formation, an original integrated approach is proposed. This article defines the methodology and highlights the perspectives of this new approach. This research is developed as part of the Omo Group Research Expedition (OGRE; Boisserie et al., 2008). It also includes the data brought by earlier field works carried out by the International Omo Research Expedition (IORE) between 1967 and 1976 (Boisserie et al., 2013, Chavaillon, 1976, Coppens, 1970, Coppens, 1971, Coppens, 1977, Coppens et al., 1973, de Heinzelin, 1983a, Gautier, 1976, Howell and Coppens, 1985, Howell et al., 1987, Jaeger and Wesselman, 1976, Johanson et al., 1976 and Merrick and Merrick, 1976).

 Brief review of the available spatial data

Only five in situ Early Oldowan site complexes have been documented so far in eastern Africa: Gona, Hadar, and Shungura in Ethiopia, Lokalalei and Kanjera in Kenya (Chavaillon, 1976, Harris, 1983, Merrick and Merrick, 1976, Plummer et al., 1999, Roche et al., 2003 and Semaw et al., 1997) (Fig. 1), although a large number of Early Pleistocene hominid fossils have been found in many localities from eastern to southern Africa (Reed et al., 2013). The available spatial data are limited to these archaeological complexes.

 Raw material procurement and spatial data

A majority of studies has focused on the location of the closest available resources for assessing the distance of raw material transport. The general availability and diversity of raw material at a broader scale was usually not described, except at Kanjera South (Braun et al., 2008), where the distribution of two different raw materials has been mapped in the context of the local river drainage system. The distances of the raw material sources vary significantly from one site to another: within a few meters at Gona (Stout et al., 2005) and Lokalalei (Harmand, 2005), within a kilometre at Hadar (Goldman-Neuman and Hovers, 2012), over 10 kilometres at Kanjera South (Braun et al., 2008). Despite this low mobility, the selection of specific types of raw material and/or specific morphologies has been documented in all the Early Oldowan site complexes (Bishop et al., 2006, Braun et al., 2009b, Delagnes et al., 2011, Goldman-Neuman and Hovers, 2012, Harmand, 2005, Plummer et al., 1999 and Stout et al., 2005). Such behaviours reflect an accurate knowledge of raw material knapping properties and availability.

 Reconstructing hominid environments

Hominid environments are usually described through various proxies. These environments can be considered at a very large-scale, e.g. eastern Africa, in order to compare climatic changes with human evolution. Data from non-hominid sites with a regional representativeness, such as pollen from marine cores, and data from hominid sites with a more local representativeness, as pollen or soil carbonates, are both considered and compared mostly in a diachronic perspective (Bonnefille, 2010). At the more local scale of the hominid-bearing palaeolandscapes, the most common proxies relate to faunal assemblages (Brugal et al., 2003 and Plummer et al., 2009), pollen (López-Sáez and Domínguez-Rodrigo, 2009), and soil carbonate isotopes (Aronson et al., 2008, Plummer et al., 1999 and Quinn et al., 2013).

At Gona, hominids inhabited a mosaic landscape with a large proportion of wooded areas, as indicated by palynological data (López-Sáez and Domínguez-Rodrigo, 2009). In the Nachukui Formation, the palaeontological record related to the Early Oldowan occurrences also indicates a mosaic landscape (Brugal et al., 2003). Soil carbonate isotopes suggest that the archaeological occupations took place in the wooded part of this landscape (Quinn et al., 2013). Geological data are consistent with settlements in the vicinity of a lake, on the banks of lagoonal ponds (Tiercelin et al., 2010) or in the floodplain and in the vicinity of a secondary channel (Roche et al., 2003). By contrast, at Hadar (A.L. 666, Aronson et al., 2008) and at Kanjera South (Plummer et al., 1999 and Plummer et al., 2009). Isotopic and faunal analyses indicate that Early Oldowan activities took place in drier settings.

These environmental interpretations mostly come from spatially restricted records, associating few concentrations of archaeological occurrences with large-scale environmental data. They usually result in a vague “mosaic” or “grassland-dominated” assessment of the spatial structure of palaeo-ecosystems, and the distribution of hominid activities in relation to this structure remains unclear. The investigation of local variations in palaeolandscapes and hominid activities requires coeval archaeological and environmental records of significant geographical extension at a landscape scale. Currently, only the Shungura Formation displays such an Early Oldowan archaeological and spatial record, extending along a transect of at least 8 km in association with rich geological and palaeontological records.

 The Shungura Formation, a context adapted to a spatial approach

In the Shungura Formation, our approach aims to combine archaeological locations and distributions of water, lithic, vegetal, and animal resources in the framework of micro-regional spatial analyses. The broad array of available data, including archaeological, palaeontological, geomorphological, sedimentological, isotopic, and those from dental microwear, retrieved from an extended research area has been integrated into a single GIS database. It must however be kept in mind that we cannot access the entire landscape for a given period, insofar as what we can analyse are only “observation windows” that are physically delimited by the outcrops and that represent a sampling of the palaeolandscapes. We assume that this sampling is random and that all variations in the palaeolandscape at a multi-kilometer scale are represented.

The Shungura outcrops stretch out over a length of ca. 60 km and a mean width of ca. 5 km, favouring a micro-regional spatial approach thanks to their physical configuration. Deposits are divided into members designated by letters, i.e. fluvio-lacustrine cycles interbedded between major extended tuff deposits that define the limits of each member (except for the bottom of Basal Member and the top of Member L (see de Heinzelin and Haesaerts, 1983). Members were tilted westward 10° to 20°, or sometimes more (Fig. 2), as a result of a tectonic phase dated to between 800 ka and 100 ka (Brown and de Heinzelin, 1983). Outcropping members currently form series of long and narrow strips oriented north-south. Each member also outcrops several times along an east-west axis as a result of the faulting system (de Heinzelin, 1983a). Temporal information is therefore closely related to spatial information. Previous palaeontological and archaeological studies have mostly focused on the northern part of the formation (e.g. Boisserie et al., 2010, Coppens, 1977, de Heinzelin, 1983b, Delagnes et al., 2011 and Howell et al., 1987), where the fossiliferous sequence extends from the Basal Member to lower Member J, between ca. 3.6 Ma and 1.76 Ma (McDougall and Brown, 2008 and McDougall et al., 2012).

The former IORE investigations suggested an archaeological occupation extending over Member E and Member F (Chavaillon, 1976, de Heinzelin, 1983a, Howell et al., 1987 and Merrick and Merrick, 1976). The OGRE archaeological investigations have evidenced a quite distinct picture with an archaeological presence starting with Member F and occurring up to lower Member G (Delagnes et al., 2011). Member E is dated between 2.40 ± 0.05 and 2.324 ± 0.020 Ma (Feibel et al., 1989 and McDougall and Brown, 2008), while Member F is dated between 2.324 ± 0.020 Ma and 2.271 ± 0.041 Ma (McDougall and Brown, 2008 and McDougall et al., 2012), and lower Member G between 2.271 ± 0.041 Ma and 2.077 ± 0.012 Ma or 2.059 ± 0.013 Ma (Kidane et al., 2007 and McDougall et al., 2012). The time span covered by the archaeological occurrences in the Shungura Formation would thus extend from ca. 2.3 to 2 Ma, instead of 2.4 Ma to 2.2 Ma, as previously assumed. Although all members below and above Member F and lower Member G have provided abundant collections of vertebrate remains including hominids (Boisserie et al., 2008, Coppens, 1977 and Howell et al., 1987), they have not yet yielded any significant archaeological record. This absence of artefact is unlikely to result from a methodological bias, insofar as careful surveys have been performed on extensive portions of their outcrops. While discrete archaeological occurrences have been recorded outside Members F and G, it clearly appears that significant toolmaking-related activities are limited to these two members. The total surfaces of Member F and lower Member G northern outcrops are respectively about 3.28 km2 and 8.86 km2.

 Contextual data

A geological map of the northern Shungura outcrops was drawn by IORE geologists (de Heinzelin, 1983a). We have geolocated and re-drawn this map with the help of satellite images that allow us to observe the sedimentary units within the members at a more precise chronological and environmental resolution than the geological map. In addition, sedimentological data related to a number of type sections are available (de Heinzelin, 1983a). They help us to characterize the landscape (as lacustrine, deltaic or fluviatile) during each deposition phase and to define the specific location of the archaeological occurrences within fluviatile landscapes (as channel, point-bar, levee, floodplain, fluviatile swamp [Haesaerts et al., 1983]). At a wider scale, isotopic analyses made on soil carbonates (Levin et al., 2011) and on large mammal enamel (Bibi et al., 2012 and Souron et al., 2012) are also informative for reconstructing the palaeo-environments. They can be combined with palaeo-vegetation related-data such as fossil wood and pollen (Bonnefille and Deschamps, 1983).

The faunal data are particularly abundant. Some 9037 fossils have been collected from Member F and 16395 in lower Member G (combining the IORE and OGRE collections). The fossils mostly consist of isolated teeth, cranio-mandibular fragments, and post-cranial elements. They are at least taxonomically attributed at the family level, and often at the genus or species level. Each specimen collected by the OGRE since 2006 is geolocated with an individual GPS point, whereas only the GPS points related to each locality are available for the fossils collected by the IORE. There are 252 localities in Member F and 509 in lower Member G, a majority of them being positioned in a single geological unit, while others are more widespread and encompass different units (de Heinzelin, 1983b).

The collected taxa belong mostly to large mammals. Bovids predominate, followed by hippopotamids, cercopithecids, and suids. Proboscideans (elephantids and deinotheres), giraffids, perissodactyls (equids and rhinocerothids), hominids and carnivores (hyenids, felids, viverrids, mustelids) are also present. Micromammals and non-mammalian fossils such as actinopterygians, chondrichthyans, crocodilians, chelonians or molluscs are also documented.

This important corpus resulted from various sampling methods. Most of the IORE fossils were collected by two independent teams (French and American) working on two different, arbitrarily defined territories, although in practice some overlap occasionally occurred. These two teams did not use the same sampling methods (Alemseged et al., 2007 and Eck, 2007), introducing biases in the corpus composition and limiting the possibilities of comparison between both collections. Conversely, the OGRE, composed of only one multidisciplinary team, collected specimens with homogeneous sampling methods in all parts and in all members of the formation. This new reference corpus will be used as a standard for assessing the IORE corpus biases. This condition is a prerequisite for comparing the faunal spectra from the different areas and time periods.

 Archaeological data

Many terms have been used to designate the archaeological evidence in the Shungura Formation: “sites” and “occurrences” are commonly used, but “patches” (Merrick and Merrick, 1976), “gisements (à éclat)”, “points” (Chavaillon, 1974, Chavaillon, 1980 and Chavaillon and Boisaubert, 1977), and “localities” (Chavaillon, 1976, Delagnes, 2012 and Delagnes et al., 2011) are also mentioned. In this article we will use the term “archaeological occurrence” as a general term that includes all the located archaeological evidence, whether or not in situ. Each archaeological occurrence is individually geolocated and refers to a point that is the basic archaeological spatial entity used in the GIS. To date, 99 archaeological occurrences have been found in Member F and 51 in the lower part of Member G.

The archaeological record of the Shungura Formation includes data from in situ excavated archaeological occurrences, in situ non-excavated archaeological occurrences, and surface archaeological occurrences. The in situ archaeological occurrences are mostly concentrated into occurrence complexes that correspond to locations grouping together several archaeological occurrences that are close enough for establishing direct stratigraphic correlations. In Member F, 22 of the 26 in situ recorded occurrences are found in complexes (Chavaillon, 1976, Delagnes et al., 2011 and Merrick and Merrick, 1976 and recent work by the OGRE). Eight of these 21 occurrences are located in fine-grained sediments (clay or silt; see Table 1). The richest lithic assemblages (FtJi 1 and FtJi 2, OMO 57, OMO 84, OMO 123) have been first studied by Merrick and Merrick (1976) for the FtJi archaeological occurrences and by Chavaillon, 1974 and Chavaillon, 1976 for the OMO archaeological occurrences. The OMO archaeological collections have been reassessed recently by De la Torre et al. (2004), and by Delagnes et al. (2011) with the addition of the material from OMO 1/E. All these studies point out that lithic artefacts are mainly flakes, made from quartz pebbles in an overwhelming majority of cases. Furthermore, the anthropic origin of the lithic tools can be rejected only for OMO 71 located in Member E (De la Torre et al., 2004 and Delagnes et al., 2011).

Since 2008, the OGRE has carried out a large-scale archaeological survey together with a systematic inventory of the archaeological occurrences and raw material sources. More than 151 ha have been surveyed in Member F and more than 161 ha in lower Member G (Fig. 3). There are 97 archaeological occurrences recorded so far in Member F and 50 in lower Member G (Fig. 4; Table 2). An archaeological occurrence was created when at least three artefacts were found with a maximal distance of 10 m one from each other, in a primary or secondary position, excluding the lithics found in recent channels (Delagnes, 2012). The abundance of the assemblages is assessed on the field using ranges of class values.

The artefacts discovered in silts and clays give indications of the type of landscape occupied by the hominids. The related occurrences can be considered geologically in situ (sensu Sitzia et al., 2012). The assemblages discovered in coarser sediments, i.e. river sands, indicate at least that the hominids settled in the area, but it is not possible to determine their exact habitat. Finally, large spaces devoid of artefacts between occurrences located within the same geological unit reflect the lack of significant remobilisation that would have scattered artefacts over large areas. We thus assume that the spatial distribution of the artefacts can bring relevant information on the relationships between the hominids and their environment, provided that their depositional context is taken into account.

 A multiscale approach

A three-scale approach has been developed to address a number of questions. Such a multiscale approach is required for processing data with varied levels of spatial accuracy (Pillot and Saligny, 2012) and for getting results that are consistent with variations in data resolution.

 Occurrence complex level

The first level is the local scale and focuses on the occurrence complex. It gives the opportunity to better understand the hominid behaviours at a high-resolution level along both horizontal and vertical axes. The horizontal axis allows studying the spatial and environmental relationships between penecontemporaneous occurrences. On a vertical axis, the fine chronostratigraphical placement of occurrences provides clues to settlement timing. Several occurrence complexes, e.g. OMO 123, OMO 1/E, OMO 79, show repetitive occupations of the same area at different time periods (Delagnes et al., 2011). Local-level palaeoenvironmental data (including fauna; see Table 1) help us to understand why these specific locations were preferentially occupied by hominids. In addition, the technological and raw material data can be compared between occurrences to better characterize patterns of human occupation within each complex.

 “Study area” level

Understanding hominid spatial behaviours as part of a landscape archaeological approach requires focusing on a larger micro-regional scale. This scale will deal with penecontemporaneous data related to different areas that have to be precisely defined. The uncertainty of some data, such as the precise provenience of the archaeological artefacts from surface occurrences, will be offset because data are gathered by “study area” and the comparisons will be performed between the different sets of data issued from each “study area”. In order to figure out the artefact density within each member, a test was made with a square mesh area of 350 meters side. The size of the mesh was chosen for fitting as much as possible with the outcrop width. An estimate of the number of artefacts for each archaeological occurrence was used to calculate this density, based on the median of the field class value for the class “3 to 10” and “11 to 50”, and on the average of the available numbers of surface artefacts for the class “> 50”. The resulting map shows a contrast between high and low densities of artefacts per hectare in each surveyed area (Fig. 5). The high density of occurrences in Member F seems to be associated with the presence of occurrence complexes and with a significant number of occurrences preserved in fine sediments (clay, silty clay, silt, except for FtJi5 complex for which occurrences are located in sandy tuffite).

The faunal data will provide information about the local environments, but territory sizes vary according to the mobility and the behaviour of each taxon. This scale of study aims at highlighting differences between some areas that reflect different bioenvironmental landscapes within the Shungura Formation. The location of the available resources in the landscapes is fundamental for understanding hominid interactions with their environment. A very fine mapping of the sedimentary units provides two types of information. It will first allow us to assess the representativeness of the observable landscapes in the outcrops. It should also help us to specify the position of the river system that constitutes the supplying area for water and raw materials. The raw material availability in the landscape will be studied thanks to the exhaustive survey of raw material sources. Such reconstruction of the landscape allows us to virtually correct the effects of the strata tilting.

 Formation level

The third scale relates to the whole Shungura Formation. At this level, only the temporal variations will be addressed. Data should be compared between sedimentary units that refer to a particular time slice. The successive sedimentary units and members provide a time frame that allows us to characterize only major palaeoenvironmental and behavioural changes. A preliminary comparison of the artefact density from Member F and lower Member G (Table 2) shows that the density of artefacts is higher in Member F than in lower Member G for the sedimentary windows under study. This gives us a preliminary picture of the evolution of the hominid spatial behaviours over time that should be developed in our future work.

 Conclusion

In this paper, we propose an innovative integrative spatial approach dedicated to the study of Early Oldowan hominid spatial behaviours. The wide array of available data in the Shungura Formation allows this type of study by integrating archaeological and palaeoecological data in an extended time frame and at various spatial scales. The multiscale approach aims at adapting the investigations to the potential and limits of the record. The spatial behaviours of the hominids and their changes and similarities over time at a local scale can be drawn with a very fine resolution at the site complex level using local environmental data. At a larger scale, the patterns of landscape occupation and the subsistence behaviours can be assessed thanks to the systematic survey of a wide area for both Member F and lower Member G. At the scale of the Shungura Formation, the temporal variations and continuities provide information about the landscape and the behavioural dynamics. The synthesis of the results issued from these different scales is fundamental for achieving a holistic approach of hominid behaviours. Our preliminary results related to the density of artefacts indicate that some areas were more visited by the Early Oldowan toolmakers than others. The reasons of such preferences still have to be explained using palaeoenvironmental data. Furthermore, both members do not show the same occupation patterns, with a decrease of artefact density in lower Member G. The relation between such decrease and the total lack of archaeological evidence in the Shungura Formation after 2 Ma (upper Member G and subsequent members) still has to be investigated.

The results provided by the three scales of analysis form the backbone of a holistic approach of the hominid behaviours in the specific context of a rift basin environment at various time periods. Some comparisons with other older, contemporaneous or younger contexts might be envisaged, in particular with Hadar (Reed, 2008), Nachukui (Brugal et al., 2003 and Roche et al., 2003), Koobi Fora (Braun et al., 2009a) and Olduvai (Blumenschine et al., 2012).

 Acknowledgments

We are very grateful to all the participants of the IORE and the OGRE, who devoted a lot of energy to collect this amount of data. We would like to thank Sabine Riffaut, Frédéric Prodéo, and Pascal Bertran for their help. This work was authorized by the Authority for Research, Conservation and Cultural Heritage. We would like to thank the ARCCH as well as the bureau of Culture of the Southern Nations, Nationalities and People Region, and of the South Omo Zone.

This work is part of Master and PhD theses (T. Maurin), which have been funded by the LABEX LaScArBx (University of Bordeaux). This research of the OGRE received funding and logistical support from the Agence nationale pour la recherche, the Ministère des Affaires Étrangères (sous-direction de l’archéologie), the Fondation Fyssen, the Centre français des études éthiopiennes, Ambassade de France en Éthiopie, CNRS, PACEA–University of Bordeaux and IPHEP–University of Poitiers (France).

We would like to thank two anonymous reviewers for their suggestions that helped us to improve the manuscript.

 Alemseged Z. Bobe R. Geraads D. Comparability of fossil data and its significance for the interpretation of hominin environments Bobe R. Alemseged Z. Behrensmeyer A.K. Hominin Environments in the East African Pliocene: an Assessment of the Faunal Evidence, Vertebrate Paleobiology and Paleoanthropology Series 2007 Springer Netherlands 159–181 Aronson J.L. Hailemichael M. Savin S.M. Hominid environments at Hadar from paleosol studies in a framework of Ethiopian climate change J. Hum. Evol. 55 2008 532–550 Bibi F. Souron A. Bocherens H. Uno K. Boisserie J.-R. Ecological change in the lower Omo Valley around 2.8 Ma Biol. Lett. 9 2012 20120890 Bishop L.C. Plummer T.W. Ferraro J.V. Braun D. Ditchfield P.W. Hertel F. Kingston J.D. Hicks J. Potts R. Recent research into Oldowan hominin activities at Kanjera South, western Kenya Afr. Archaeol. Rev. 23 2006 31–40 Blumenschine R.J. Stanistreet I.G. Njau J.K. Bamford M.K. Masao F.T. Albert R.M. Stollhofen H. Andrews P. Prassack K.A. McHenry L.J. Fernández-Jalvo Y. Camilli E.L. Ebert J.I. Environments and hominin activities across the FLK Peninsula during Zinjanthropus times (1.84 Ma), Olduvai Gorge, Tanzania J. Hum. Evol. 63 2012 364–383 Boisserie J.-R. Guy F. Delagnes A. Hlukso L.J. Bibi F. Beyene Y. Guillemot C. New palaeoanthropological research in the Plio-Pleistocene Omo Group, Lower Omo Valley, SNNPR (Southern Nations, Nationalities and People Regions), Ethiopia C. R. Palevol 7 2008 429–439 Boisserie J.-R. Delagnes A. Beyene Y. Schuster M. Reconstructing the African background to human expansions in Eurasia: new research in the Shungura Formation, Ethiopia Quat. Int. 223–224 2010 426–428 Boisserie J.-R. Delagnes A. Beyene Y. Travaux passés et actuels dans la basse vallée de l’Omo (Sud-Ouest de l’Éthiopie) Sahnouni M. Actes Du Colloque International : l’Afrique, Berceau de L’humanité : découvertes récentes 2013 Centre national de recherches préhistoriques, anthropologiques et historiques Sétif 87–103 Bonnefille R. Cenozoic vegetation, climate changes and hominid evolution in tropical Africa Global Planet. Change 72 2010 390–411 (Quaternary and Global Change: review and Issues Special issue in memory of Hugues FAURE) Bonnefille R. Deschamps R. Data on fossil flora Heinzelin J. de Archives of the International Omo Research Expedition 1983 The Omo Group Tervuren, Belgique 191–207 Braun D.R. Plummer T. Ditchfield P. Ferraro J.V. Maina D. Bishop L.C. Potts R. Oldowan behavior and raw material transport: perspectives from the Kanjera Formation J. Archaeol. Sci. 35 2008 2329–2345 Braun D.R. Harris J.M. Maina D. Oldowan raw material procurement and use: evidence from the Koobi Fora Formation Archaeometry 51 2009 26–42 Braun D.R. Plummer T. Ferraro J.V. Ditchfield P. Bishop L.C. Raw material quality and Oldowan hominin toolstone preferences: evidence from Kanjera South, Kenya J. Archaeol. Sci. 36 2009 1605–1614 Brown F.H. de Heinzelin J. The Lower Omo basin de Heinzelin J. Archives of the International Omo Research Expedition 1983 The Omo Group Tervuren, Belgique 7–24 Brown F.H. Feibel C.S. “Robust” hominids and plio-pleistocene paleogeography of the Turkana basin, Kenya and Ethiopia Grine F.E. Evolutionary history of the robust Australopithecines 1988 De Gruyter New York 325–341 Brugal J.-P. Roche H. Kibunjia M. Faunes et paléoenvironnements des principaux sites archéologiques plio-pléistocènes de la formation de Nachukui (Ouest-Turkana, Kenya) C. R. Palevol 2 2003 675–684 Chavaillon J. Documents pour servir à l’histoire des civilisations éthiopiennes fasc. 5 1974 CNRS Paris, France Chavaillon J. Evidence for the technical practices of Early Pleistocene hominids, Shungura Formation, Lower Omo Valley, Ethiopia Coppens Y. Howell F.C. Isaac G.L. Leakey R.E. Earliest Man and Environments in the Lake Rudolph Bassin: Stratigraphy, Paleoecology and Evolution 1976 University of Chicago Press Chicago 565–573 Chavaillon J. Le point des recherches archéologiques dans la basse vallée de l’Omo, Éthiopie Leakey R.E. Ogot B.E. Presented at the VIIIe congrès panafricain de pré-histoire et d’étude du Quaternaire Louis Leakey Memorial Institute for African Prehistory, Nairobi 1980 (400 p.) Chavaillon J. Boisaubert J.-L. Prospection archéologique dans le Gemu-Gofa Doc. Pour Serv. A Hist. Civilis. Ethiop. Paris 8 1977 3–10 Coppens Y. Localisation dans le temps et dans l’espace des restes d’hominidés des formations plio-pléistocènes de l’Omo C. R. Acad. Sci. Paris Ser. D 271 1970 1968–1971 Coppens Y. Les restes d’hominidés des séries supérieures des formations plio-villafranchiennes de l’Omo en Éthiopie C. R. Acad. Sci. Paris, Ser. D 272 1971 36–39 Coppens Y. Hominid remains from the Plio/Pleistocene Formations of the Omo Basin, Ethiopia J. Hum. Evol. 6 1977 169–173 Coppens Y. Chavaillon J. Beden M. Résultats de la nouvelle mission de l’Omo (campagne 1972). Découverte des restes d’Hominidés et d’une industrie sur éclats C. R. Acad. Sci. Paris, Ser. D 276 1973 161–163 de Heinzelin J. The Omo group: archives of the International Omo Research Expedition. Musée royal de l’Afrique centrale. Annales 1983 Musée royal de l’Afrique centrale Tervuren, Belgique de Heinzelin J. Catalogue of observations and collected areas de Heinzelin J. Archives of the International Omo Research Expedition 1983 The Omo Group Tervuren, Belgique 241–304 de Heinzelin J. Haesaerts P. The Shungura Formation de Heinzelin J. Archives of the International Omo Research Expedition 1983 The Omo Group Tervuren, Belgique 25–128 De la Torre I. de Beaune S. Davidson I. Gowlett J. Hovers E. Kimura Y. Mercader J. Omo revisited: evaluating the technological skills of Pliocene Hominids 1 Curr. Anthropol. 45 2004 439–465 Delagnes A. The Earliest Stone Age of Ethiopia in the East African context Sanz N. Human Origin Sites and the World Heritage Convention in Africa, World Heritage Papers 2012 UNESCO Paris 101–114 Delagnes A. Boisserie J.-R. Beyene Y. Chuniaud K. Guillemot C. Schuster M. Archaeological investigations in the Lower Omo Valley (Shungura Formation, Ethiopia): new data and perspectives J. Hum. Evol. 61 2011 215–222 Eck G.G. The effects of collection strategy and effort on faunal recovery Bobe R. Alemseged Z. Behrensmeyer A.K. Hominin Environments in the East African Pliocene: an Assessment of the Faunal Evidence, Vertebrate Paleobiology and Paleoanthropology Series 2007 Springer Netherlands 183–215 Feibel C.S. Brown F.H. McDougall I. Stratigraphic context of fossil hominids from the Omo group deposits: northern Turkana Basin, Kenya and Ethiopia Am. J. Phys. Anthropol. 78 1989 595–622 Gautier A. Assemblage fossil freshwater mollusks from the Omo Group and related deposits in the Lake Rudolf Basin Coppens Y. Howell F.C. Isaac G.L. Earliest Man and Environments in the Lake Rudolf Basin 1976 University of Chicago Press Chicago 379–382 Goldman-Neuman T. Hovers E. Raw material selectivity in Late Pliocene Oldowan sites in the Makaamitalu Basin, Hadar, Ethiopia J. Hum. Evol. 62 2012 353–366 Haesaerts P. Stoops G. Van Vliet-Lanoe B. Data on sediments and fossil soils de Heinzelin J. Archives of the International Omo Research Expedition 1983 The Omo Group Tervuren, Belgique, Belgique 149–196 Harmand S. Matières premières lithiques et comportements techno-économiques des homininés plio-pléistocènes du Turkana occidental, Kenya 2005 Paris 10 Paris, France Harris J.W.K. Cultural beginnings: Plio-Pleistocene archaeological occurrences from the Afar, Ethiopia Afr. Archaeol. Rev. 1 1983 3–31 Howell F.C. Coppens Y. Les Faunes plio-pléistocènes de la basse vallée de l’Omo, Éthiopie, Cahiers de paléontologie. Travaux de paléontologie est-africaine 1985 Éd. du CNRS Paris, France Howell F.C. Haesaerts P. de Heinzelin J. Depositional environments, archeological occurrences and hominids from Members E and F of the Shungura Formation (Omo basin, Ethiopia) J. Hum. Evol. 16 1987 665–700 Jaeger J.-J. Wesselman H.B. Fossil remains of micromammals from the Omo Group Deposits Coppens Y. Howell F.C. Isaac G.L. Earliest Man and Environments in the Lake Rudolf Basin 1976 University of Chicago Press Chicago 351–360 Johanson D.C. Splingaer M. Boaz N.T. Paleontological excavations in the Shungura Formation, Lower Omo Basin, 1969–1973 Coppens Y. Howell F.C. Isaac G.L. Earliest Man and Environments in the Lake Rudolf Basin 1976 University of Chicago Press Chicago 402–420 Kidane T. Otofuji Y.-I. Brown F.H. Takemoto K. Eshete G. Two normal paleomagnetic polarity intervals in the lower Matuyama Chron recorded in the Shungura Formation (Omo Valley, Southwest Ethiopia) Earth Planet. Sci. Lett. 262 2007 240–256 Levin N.E. Brown F.H. Behrensmeyer A.K. Bobe R. Cerling T.E. Paleosol carbonates from the Omo Group: isotopic records of local and regional environmental change in East Africa Palaeogeogr. Palaeoclimatol. Palaeoecol. 307 2011 75–89 López-Sáez J.-A. Domínguez-Rodrigo M. Palynology of OGS-6a and OGS-7, two new 2.6 Ma archaeological sites from Gona, Afar, Ethiopia: insights on aspects of Late Pliocene habitats and the beginnings of stone tool use Geobios 42 2009 503–511 McDougall I. Brown F.H. Geochronology of the pre-KBS Tuff sequence, Omo Group, Turkana Basin J. Geol. Soc. 165 2008 549–562 McDougall I. Brown F.H. Vasconcelos P.M. Cohen B.E. Thiede D.S. Buchanan M.J. New single crystal 40Ar/39Ar ages improve time scale for deposition of the Omo Group, Omo–Turkana Basin, East Africa J. Geol. Soc. 169 2012 213–226 Merrick H.V. Merrick J.P.S. Archaeological occurrences of Earlier Pleistocene age from the Shungura Formation Coppens Y. Howell F.C. Isaac G.L. Leakey R.E. Earliest Man and Environments in the Lake Rudolph Bassin: stratigraphy, Paleoecology and Evolution 1976 University of Chicago Press Chicago 574–584 Pillot L. Saligny L. L’évolution de l’occupation humaine : l’analyse spatiale exploratoire des données. Le problème de l’incertitude et de l’hétérogénéité des données en archéologie, in: Variabilités environnementales, mutations sociales. Nature, intensités, échelles et temporalités des changements XXXIIes Rencontres internationales d’archéologie et d’histoire d’Antibes 2012 155–175 Plummer T. Bishop L.C. Ditchfield P. Hicks J. Research on Late Pliocene Oldowan Sites at Kanjera South, Kenya J. Hum. Evol. 36 1999 151–170 Plummer T.W. Bishop L.C. Ditchfield P.W. Ferraro J.V. Kingston J.D. Hertel F. Braun D.R. The environmental context of Oldowan Hominin activities at Kanjera South, Kenya Hovers E. Braun D.R. Interdisciplinary Approaches to the Oldowan, Vertebrate Paleobiology and Paleoanthropology 2009 Springer Netherlands 149–160 Prat S. Brugal J.-P. Tiercelin J.-J. Barrat J.-A. Bohn M. Delagnes A. Harmand S. Kimeu K. Kibunjia M. Texier P.-J. Roche H. First occurrence of early Homo in the Nachukui Formation (West Turkana, Kenya) at 2.3–2.4 Myr J. Hum. Evol. 49 2005 230–240 Quinn R.L. Lepre C.J. Feibel C.S. Wright J.D. Mortlock R.A. Harmand S. Brugal J.-P. Roche H. Pedogenic carbonate stable isotopic evidence for wooded habitat preference of Early Pleistocene tool makers in the Turkana Basin J. Hum. Evol. 65 2013 65–78 Reed K.E. Paleoecological patterns at the Hadar hominin site, Afar Regional State, Ethiopia J. Hum. Evol. 54 2008 743–768 Reed K.E. Fleagle J.G. Leakey R.E. The paleobiology of Australopithecus 2013 Springer, Dordrecht New York Reynolds S.C. Bailey G.N. King G.C.P. Landscapes and their relation to hominin habitats: case studies from Australopithecus sites in eastern and southern Africa J. Hum. Evol. 60 2011 281–298 Roche H. Brugal J.-P. Delagnes A. Feibel C. Harmand S. Kibunjia M. Prat S. Texier P.-J. Les sites archéologiques plio-pléistocènes de la formation de Nachukui, Ouest-Turkana, Kenya : bilan synthétique 1997–2001 C. R. Palevol 2 2003 663–673 Semaw S. Renne P. Harris J.W.K. Feibel C.S. Bernor R.L. 2.5-million-year-old Stone Tools from Gona, Ethiopia Nature 385 1997 333–336 Sitzia L. Bertran P. Boulogne S. Brenet M. Crassard R. Delagnes A. Frouin M. Hatté C. Jaubert J. Khalidi L. Messager E. Mercier N. Meunier A. Peigné S. Queffelec A. Tribolo C. Macchiarelli R. The paleoenvironment and lithic taphonomy of Shi’Bat Dihya 1, a Middle Paleolithic site in Wadi Surdud, Yemen Geoarchaeology 27 2012 471–491 Souron A. Balasse M. Boisserie J.-R. Intra-tooth isotopic profiles of canines from extant Hippopotamus amphibius and late Pliocene hippopotamids (Shungura Formation, Ethiopia): insights into the seasonality of diet and climate Palaeogeogr. Palaeoclimatol. Palaeoecol. 342–343 2012 97–110 Stout D. Quade J. Semaw S. Rogers M.J. Levin N.E. Raw material selectivity of the earliest stone toolmakers at Gona, Afar, Ethiopia J. Hum. Evol. 48 2005 365–380 Tiercelin J.-J. Schuster M. Roche H. Brugal J.-P. Thuo P. Prat S. Harmand S. Davtian G. Barrat J.-A. Bohn M. New considerations on the stratigraphy and environmental context of the oldest (2.34 Ma) Lokalalei archaeological site complex of the Nachukui Formation, West Turkana, northern Kenya Rift J. Afr. Earth Sci. 58 2010 157–184

(Color online.) Early Oldowan site complexes.

(Couleur en ligne.) Les complexes de sites du début de l’Oldowayen.

(Color online.) View of OMO 1/E. The sediments of the Shungura Formation are tilted westward.

(Couleur en ligne.) Vue d’OMO 1/E. Les sédiments de la formation de Shungura sont inclinés vers l’ouest.

(Color online.) OMO A82 archaeological occurrence (Member F). White circles point out artefacts.

(Couleur en ligne.) Occurrence archéologique OMO A82 (Membre F). Les cercles blancs indiquent les artefacts.

(Color online.) Archaeological occurrences of the Shungura Formation. All the occurrences are located in Member F and in lower Member G (LG). The occurrence complexes are pointed out with their name. Coloured outcrops correspond to the surveyed area.

(Couleur en ligne.) Occurrences archéologiques de la Formation de Shungura. Toutes les occurrences se situent dans le Membre F et dans la partie inférieure du Membre G (LG). Les complexes d’occurrence sont indiqués par leur nom. Les affleurements en couleur sur la carte correspondent aux zones prospectées.

(Color online.) Artefacts density per hectare in Member F (left) and in lower Member G (right). Artefact density was calculated from the sum of occurrences’ estimated artefacts numbers, divided by the amount of the surveyed area within a 350 meter wide mesh.

(Couleur en ligne.) Densités d’artefacts par hectare pour le Membre F (gauche) et la partie inférieure du Membre G (LG, droite). La densité d’artefact a été calculée à partir de la somme des estimations du nombre d’artefacts de chaque occurrence située dans une maille de 350 m de large, divisée par la surface prospectée dans cette même maille.

Inventory of in situ archaeological occurrence of Member F.

Inventaire des occurrences archéologiques in situ du Membre F.

Inventory number Occurrence name Complex Inventor Expedition Year of investigation Texture of sediment Fauna OMO A16 FtJi1 FtJi1 - FtJi3 - FtJi4 H.V. Merrick IORE 1973 Sandy silt Present OMO A17 FtJi3 FtJi1 - FtJi3 - FtJi4 H.V. Merrick IORE 1972 Sandy silt Absent OMO A68 L204-4 FtJi1 - FtJi3 - FtJi4 A. Delagnes OGRE 2010 Sandy silt Present OMO A18 FtJi4 FtJi1 - FtJi3 - FtJi4 H.V. Merrick IORE 1973 Sandy silt Absent OMO A19 FtJi5 FtJi5 H.V. Merrick IORE 1973 Tuffite Absent OMO A95 L182 FtJi5 A. Delagnes OGRE 2011 Coarse sand Present OMO A96 L182 FtJi5 A. Delagnes OGRE 2011 Coarse sand Absent OMO A97 L182 FtJi5 A. Delagnes OGRE 2011 Coarse sand Present OMO A2 FtJi2 – H.V. Merrick IORE 1971–1973 Silty clay Absent OMO A40 OMO 130 – A. Delagnes OGRE 2008 Sand ? OMO A7 OMO 123e OMO 123 J. Chavaillon IORE 1973 Sand Present OMO A11 OMO 123i OMO 123 J. Chavaillon IORE 1973 Silt Absent OMO A12 OMO 123j OMO 123 J. Chavaillon IORE 1973 Silt Absent OMO A13 OMO 123k OMO 123 J. Chavaillon IORE 1973, 1976 Silty clay Absent OMO A14 OMO 123l OMO 123 J. Chavaillon IORE 1973 Silty clay Absent OMO A15 OMO 123m OMO 123 J. Chavaillon IORE 1973, 1976 Sand Present OMO A130 O97 – A. Delagnes OGRE 2011 Tuffic sand Absent OMO A106 L181 – A. Delagnes OGRE 2011 Coarse sand Present OMO A31 OMO 1/E-1 OMO 1/E A. Delagnes OGRE 2008, 2010 Sand Absent OMO A32 OMO 1/E-2 OMO 1/E A. Delagnes OGRE 2008, 2010 Sand Present OMO A33 OMO 1/E-3 OMO 1/E A. Delagnes OGRE 2008, 2010 Clay Absent OMO A41 OMO 79-1 OMO 79 A. Delagnes OGRE 2008, 2010 Coarse sand Absent OMO A82 OMO 79-8 OMO 79 A. Delagnes OGRE 2010 Silt Absent OMO A88 OMO 79 OMO 79 A. Delagnes OGRE 2011 Coarse sand Present OMO A128 OMO 79 OMO 79 A. Delagnes OGRE 2011 Silt Absent OMO A167 OMO 57-5 OMO 57 J. Chavaillon IORE 1972 Coarse sand ?

Artefact density per ha for surveyed parts of Member F and lower Member G (LG). The total of artefacts was calculated from the artefact number estimations for each occurrence. These estimations fall into three classes (number of surface artefacts). For class 3 to 10 artefacts, the estimation is 6.5 artefacts; for class 11 to 50, the estimation is 30.5 artefacts; for class > 50, the estimation is 165.75 artefacts. The class Indet. refers to occurrences for which the number of surface artefacts is unknown.

Densité d’artefacts par hectare pour les zones prospectées du Membre F et de la partie inférieure du Membre G (LG). Le nombre total d’artefacts a été calculé à partir des estimations du nombre d’artefacts par occurrence en fonction de la classe de valeur enregistrée sur le terrain. Les estimations sont de 6,5 artefacts pour la classe 3 à 10 ; de 30,5 artefacts pour la classe 11 à 50 et de 165,75 artefacts pour la classe > 50. La classe Indet. correspond aux occurrences pour lesquelles le nombre d’artefacts de surface est inconnu.

Member OGRE surveyed surface (ha) Archaeological occurrences Number of surface artefacts Estimated number of artefacts Artefact density 0 (only in situ artefact) 3 to 10 11 to 50 > 50 Indet. F 151.8 97 1 48 36 10 2 3067.5 20.20750988 LG 161.35 51 0 40 10 1 0 724.25 4.488689185