Quaternary Science Reviews 268 (2021) 107132 Contents lists available at ScienceDirect Quaternary Science Reviews journal homepage: www.elsevier.com/locate/quascirev Pliocene to Middle Pleistocene climate history in the Guadix-Baza Basin, and the environmental conditions of early Homo dispersal in Europe  Juha Saarinen a, *, Otto Oksanen a, Indre_ Zliobait e_ a, Mikael Fortelius a, b, c, d Daniel DeMiguel , Beatriz Azanza , Herve e  Bocherens f, Carmen Luzo  n g,  Solano-García , Jose Jose h  Yravedra , Lloyd A. Courtenay , Hugues-Alexandre Blain k, l, i j nchez-Bandera k, l, Alexia Serrano-Ramos g, Juan Jose Christian Sa  Rodriguez-Alba m, Suvi Viranta a, Deborah Barsky k, l, Miikka Tallavaara a, Oriol Oms n, Jordi Agustí k, l, o, Juan Ochando p, Jose  S. Carrio  n p, Juan Manuel Jime nez-Arenas q a University of Helsinki, Finland b Visiting Scholars Program, University of Granada, Granada, Spain c ARAID / Universidad de Zaragoza. Departamento Ciencias de La Tierra, Zaragoza, Spain d  de Paleontologia Miquel Crusafont, Universitat Auto Institut Catala noma de Barcelona, Barcelona, Spain e Universidad de Zaragoza, Departamento Ciencias de La Tierra, Instituto Universitario de Investigacio n en Ciencias Ambientales de Arago n (IUCA), Zaragoza, Spain f €t Tübingen, Germany and Senckenberg Centre for Human Evolution and Palaeoenvironment, Tübingen, Germany Universita g Universidad de Granada, Programa de Doctorado en Historia y Artes, Spain h Department of Prehistory and Archaeology, University of Seville, Spain i Departamento de Prehistoria, Historia Antigua y Arqueología Universidad Complutense de Madrid, Spain j  Department of Cartographic and Terrain Engineering, Higher Polytechnic School of Avila,  University of Salamanca, Hornos Caleros 50, 05003, Avila, Spain k  de Paleoecologia Humana I Evolucio Institut Catala  Social (IPHES-CERCA), Zona Educacional 4, Campus Sescelades URV (Edifici W3), 43007, Tarragona, Spain l ria I Histo Universitat Rovira I Virgili, Departament D’Histo ria de L’Art, Avinguda de Catalunya 35, 43002, Tarragona, Spain m Departamento de Paleontología, Facultad de Ciencias Geolo gicas, Calle Jos e Antonio Novais, 2, Universidad Complutense de Madrid, 28040, Madrid, Spain n Universitat Autonoma de Barcelona, Spain o  Catalana de Recerca I Estudis Avançats, Pg. Lluís Companys 23, 08010, Barcelona, Spain ICREA, Institucio p University of Murcia, Faculty of Biology, Department of Plant Biology, Murcia, Spain q Universidad de Granada, Departamento de Prehistoria y Arqueología / Instituto Universitario de La Paz y Los Conflictos, Granada, Spain a r t i c l e i n f o a b s t r a c t Article history: The Guadix-Baza Basin (GBB) in Andalucía, Spain, comprises palaeontological and archaeological sites Received 20 May 2021 dating from the Early Pliocene to the Middle Pleistocene, including some of the earliest sites with evi- Received in revised form dence for the presence of early humans (Homo sp.) in Europe. Thus, the history of climate and envi- 30 July 2021 ronments in this basin contributes significantly to our understanding of the conditions under which early Accepted 30 July 2021 humans spread into Europe during the Early Pleistocene. Here we present estimates of precipitation and Available online 13 August 2021 primary productivity in the GBB from the Pliocene to the Middle Pleistocene based on dental ecometrics in fossil communities of large herbivorous mammals, and perform an ecometrics-based distribution Handling Editor: Mira Matthews modelling to analyse the environmental conditions of Early and Middle Pleistocene human sites in Europe. Our results show that Early Pleistocene humans generally occupied on average relatively diverse Keywords: habitats with ecotones, such as woodlands and savannas, but avoided very open and harsh (cool or dry) Pleistocene environments. During the Middle Pleistocene in Europe, humans occupied a comparatively much Palaeoclimatology broader range of environments than during the Early Pleistocene, but were on average more concen- Vegetation dynamics trated in environments where the dental ecometric of mammals indicate wooded palaeoenvironments. Hominin environments In the earliest human occupation sites of the GBB, Barranco Leo n and Fuente Nueva 3, the mean annual Large mammals Ecometrics precipitation and net primary production estimates indicate climatic conditions close to modern Med- Mesowear iterranean sclerophyllous woodland environments, but with slightly higher primary productivity, * Corresponding author. E-mail address: juha.saarinen@helsinki.fi (J. Saarinen). https://doi.org/10.1016/j.quascirev.2021.107132 0277-3791/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 Europe indicating some similarity with East African woodlands. On the other hand, the environments did not Guadix-baza basin resemble African grassland savannas. The browse-dominated diets of ungulates from Barranco Leo  n and Fuente Nueva 3 further suggest palaeoenvironments where grasses were a minor component of the vegetation. In the slightly older site of Venta Micena that has no evidence for the presence of hominins, dental ecometric estimates indicate climate and environments similar to Mediterranean “forest steppe” environments existing in the surroundings of Baza today. Grasses were prevalent in the diet of some taxa, especially equids, in Venta Micena, but most of the species show browse-dominated diets even there. © 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). 1. Introduction Moreover, the earliest human occupation sites in Eastern Asia do not appear to have been consistently savanna-like. For example, the Understanding the environmental conditions under which early palaeoenvironments in the Nihewan basin (ca. 1.7e1 Ma) included humans dispersed into Iberia during the Early and Middle Pleisto- elements of temperate woodlands and grasslands (Peterson et al., cene is a key issue for understanding what enabled them to expand 2003; Teague and Potts, 2007), while the palaeoenvironment of and how they interacted within different environments. Dental Gonwangling (ca. 1.2 Ma) has been interpreted as subtropical ecometrics and body size of large herbivorous mammals have been montane forest (Teague and Potts, 2007). associated with various aspects of vegetation and climate (Fortelius et al., 2002, 2016; Liu et al., 2012; Saarinen, 2014; Saarinen et al., 1.1. Research history of climate and palaeoenvironments in the  2016; Zliobait  e_ et al., 2016; Zliobaite_ et al., 2018; Oksanen et al., Guadix-Baza Basin during the pleistocene 2019). Here we use dental ecometric variables and mean body size in fossil ungulate communities to predict presence of humans The Guadix-Baza Basin (GBB) (Andalucía, Spain) is a sedimen- (species distribution of humans) during the Early and Middle tary basin, with an extensive record of palaeontological and Pleistocene in Europe. The model is calibrated on fossil and archaeological sites ranging from the Pliocene to the Middle archaeological evidence from Pleistocene localities of Europe. Pleistocene (ca. 4.5e0.4 Ma), including Barranco Leo  n and Fuente Much of the debate about the environmental conditions during Nueva 3 (ca. 1.4e1.2 Ma), which are among the earliest sites with these episodes of earliest human dispersions has centered around solid evidence of the presence of humans (Homo sp.) in Europe the question of whether they occurred in the context of grasslands (Toro-Moyano et al., 2013). Somewhat contradictory palae- or wooded grasslands, similar to African grassland savannas (the oenvironmental information from the Pleistocene sites in the GBB “savannahstan”-hypothesis; Dennell and Roebroeks, 2005), or has been published based on different methods. For example, rather in woodlands or even closed forests (Belmaker, 2010a,b; Palmqvist et al. (2003) interpreted the palaeoenvironment of Venta Dennell, 2010; Levin, 2015; Patterson et al., 2017a,b; Belmaker and Micena to have been similar to East African savannas on the basis of O'Brien, 2018). Also, the importance of water-edge habitats for early faunal analysis, whereas a Mediterranean woodland environment humans has been discussed (Cohen et al., 2012). Early human was suggested by Belmaker (2006). Most analyses indicate a pre- evolution in the Early Pleistocene of East Africa took place in the dominance of Mediterranean woodland and open environments in broad context of opening landscapes and the spread of C4 grasses, the basin during much of the Early Pleistocene (e.g., Alberdi et al., but in fact the environments where early humans occurred were 2001; Belmaker, 2006, 2010a; Agustí et al., 2010; Blain et al., diverse, with fine-scale spatial and temporal variation in woody 2011; Sa nchez-Bandera et al., 2020; Altolaguirre et al., 2020, cover (Carrio  n et al., 2011, 2019; Levin, 2015; Patterson et al., 2021). The presence of the Baza palaeolake, and changes in its 2017a,b). It has been suggested that the earliest human expansion water-level, have been key elements affecting the palaeoecology of out of Africa into Southern and Eastern Asia occurred in grassland, the basin, by providing a constant supply of water and palustrine wooded grassland and woodland habitats similar to modern Afri- environments around the lake (Oms et al., 2011). Furthermore, the can savannas (Dennell and Roebroeks, 2005; Dennell, 2010). “Galerian turnover pulse” in mammal faunas (ca. 1.0e0.9 Ma) has However, this “savannahstan-hypothesis” sensu stricto is not been associated with the arrival of early humans in Europe (Azanza consistently supported by multi-proxy analyses of environmental et al., 2000), although many of the earliest European sites, including conditions in the Early Pleistocene human occupation sites in Barranco Leo  n and Fuente Nueva 3, now seem to predate this event. Eurasia. Based on indicator species and ecological diversity analysis, Nonetheless, the arrival of some new species, such as the very large it has been suggested that the earliest human occurrence sites in cervid Praemegaceros verticornis (that was proposed as the marker Eurasia, such as Dmanisi (Georgia), and ‘Ubeidiya (Israel), had of the Galerian Mammal Age; Gliozzi et al., 1997) and the large palaeoenvironments similar to African savannas (Dolukhanov, equid Equus suessenbornensis, may be associated with the arrival of 2000; Palmqvist, 2002; Martínez-Navarro, 2004). However, taxo- early humans, suggesting perhaps either a common dispersal event nomic diversity measurements using correspondence analyses or a change of environmental conditions that favoured the dispersal indicate that their palaeoenvironments differed from African sa- of these taxa in Europe. vannas, comprising instead a mixture of temperate forest and In the northern margin of the GBB, on the Baza palaeo- Mediterranean woodland elements (Belmaker, 2006, 2010a, b). lakeshore, swampy conditions are reconstructed at the earliest Other palaeoenvironmental proxies from Dmanisi (such as pollen, Pleistocene site of Hue lago (ca. 2.6 Ma) as indicated by the depo- phytoliths and carpological remains) indicate landscapes that could sitional environment, taphonomy and the abundance of browsing be interpreted to be savanna-like in a physiognomic or structural large herbivores in the mammal fauna. All this evidence suggests sense, but not taxonomic sense, being instead dominated by abundant wooded or bushy environments in the surroundings temperate oak forest elements with grassy vegetation (Messager (Alberdi et al., 2001). Similarly, both the late Early Pleistocene site of et al., 2010). Mesowear analysis of the ungulate fauna in ‘Ubeidiya Hue scar 1 (ca. 1.1 Ma according to Alberdi et al., 2001; 900e800 ka further supports a wooded palaeoenvironment, being dominated according to Rodríquez et al., 2015, and 574e410 ka according to by ungulates with browsing diets (Belmaker and O'Brien, 2018). (Demuro et al., 2015)) and the early Middle Pleistocene site of 2  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 Cúllar-Baza 1 (0.7 Ma) have palustrine depositional environments mammalian fauna and the possible evidence of human occupation and diverse faunas of browsing ungulates. This indicates a swampy can be shown to be associated, as far as can be told from the stra- lake-shore environment, surrounded by woodlands or shrublands tigraphy, 3) the fossil assemblage does not show obvious irregu- in Hue scar 1, but the lack of closed-forest mammal taxa indicates larities due to sorting by human or carnivore activities, or due to open, sparsely forested environments with dry periods (Alberdi other taphonomic processes. et al., 2001). Near the village of Orce, the sites of Venta Micena (VM), Bar- 2.2. Dental ecometric trait data ranco Leo  n (BL) and Fuente Nueva 3 (FN3) provide essential data for contributing to discussions about the early human occupation of We use dental ecometric traits within large mammal commu- Europe. Despite intensive excavations, the oldest of these sites, VM, nities to model mean annual temperature (MAT), minimum annual dated to ca. 1.6e1.4 Ma (Duval et al., 2010), has not yielded any temperature (MINT), mean annual precipitation (MAP) and net conclusive evidence for the presence of early humans. In contrast, primary production (NPP) of GBB localities and other Early and abundant Oldowan stone tools and bones with anthropogenic Middle Pleistocene European localities. Such an approach is taxon- modifications have been discovered in BL (1.81e1.05 [1.43 ± 0.38], free (Damuth, 1992) in the sense that the inference of past condi- Barsky et al., 2010; Toro-Moyano et al., 2013; Blain et al., 2016; tions is based on generalised dental ecometric traits rather than Espigares, 2019; Titton et al., 2020; Titton et al., 2021) and FN3 individual adaptations of taxa, even if the traits are coded at (1.4e0.98 [1.19 ± 0.21], Barsky et al., 2010; Duval, 2008; Duval et al., species-level and are in this sense based on species lists per local- 2012; Espigares, 2019). In addition, BL yielded a fossil human tooth ities. It has been demonstrated that dental ecometric traits are not (Toro-Moyano et al., 2013). Palustrine depositional environments significantly influenced by non-ecological processes such as near the Baza lakeshore have been evidenced for all these sites phylogenetic concervatism or spatial autocorrelation, and are (Oms et al., 2011). The palaeoenvironment in VM was interpreted as instead primarily driven by climatic variables (Lawing et al., 2017). similar to African savannas by Palmqvist et al. (2003), based on Dental ecometric traits for the large herbivorous mammal species analyses of feeding ecology and locomotion types of large mam- present in the GBB and other locations in Early and Middle Pleis- mals. However, taxonomic correspondence analysis of large tocene Europe were determined following the coding of Liu et al. mammals (Belmaker, 2006, 2010a) indicates that the palae- (2012) and Zliobaite et al. (2018). The dental ecometric traits oenvironment comprised elements of Mediterranean woodlands were observed from second upper molar in ungulates, or from the and temperate forests and was thus quite different from African third upper molar in pigs and proboscideans. The traits were scored savannas. Rodent faunas indicate that the precipitation and tem- at the species level. They include: perature in VM were roughly similar to present (Agustí et al., 2010). - Hypsodonty, HYP (tooth crown height): 1 (brachydont), 2 The reconstructions of palaeoclimate and palaeoenvironments for (mesodont) or 3 (hypsodont). Hypsodonty is an adaptation for BL and FN3 indicate temperature conditions similar to present, but increasing functional durability of dentition. Mean ordinated hyp- higher precipitation, with Mediterranean woodland/shrubland sodonty in large mammal communities is correlated with precipi- -type vegetation (Agustí et al., 2009, 2010; Blain et al., 2011, 2016; tation through the link of high abrasion on teeth in dry climatic nchez-Bandera et al., 2020). The vegetation type in level D1 at Sa conditions, possibly because of multiple factors, including aeolian Barranco Leo  n, containing most of the faunal remains and evidence dust accumulation and prevalence of grasses with phytoliths in dry of human presence in BL, is described as a wooded environment. In environments (e.g., Fortelius et al., 2002; Liu et al., 2012). contrast, the environmental conditions at FN3 are now recon- - Longitudinal loph count, LOP: the number of elongated cutting structed as more open and dryer than at BL, based on the edge structures that are oriented in the direction of the tooth row amphibian and reptile fauna (Sa nchez-Bandera et al., 2020). (for example rhinoceroses have one (the ectoloph), ruminants and In this study we provide further information about the palae- horses have two (the buccal and lingual “lophids” formed by the oenvironmental conditions in the GBB from the Early Pliocene (ca. flattened cusps), and proboscideans have none (because all the 4.5 Ma) to the Middle Pleistocene (ca. 0.4 Ma), based on dental lophs are transverse)). LOP is correlated with temperature (high ecometrics, average body size and diet of large herbivorous mam- LOP value indicating cool climate), because longitudinal lophs are mals. We further discuss the palaeoenvironments of the earliest most abundant in ungulates feeding seasonally on bark and human occupation sites, BL and FN3, in the wider context of early branches of deciduous trees that drop their leaves during cold human palaeoenvironments in Europe, based on distribution months (e.g., Liu et al., 2012). modelling of preferred environments during the Early and Middle - Horizodonty (HOD): the amount of transverse cusp pairs or Pleistocene of Europe. lophs; the categories are: 1 (brachyhorizodont: two cusp pairs or lophs, this includes most herbivorous mammals), 2 (mesohor- 2. Materials and methods izodont: three cusp pairs or lophs, this happens e.g. in some pigs with enlarged last molars), 3 (hypsohorizodont: more than three 2.1. Materials cusp pairs or lophs, this includes mostly mammutids, gompho- theres and elephants, and some derived pigs such as warthog). The mammal fossils from Venta Micena, Barranco Leo  n and Horizodonty is associated with increase in tooth volume by Fuente Nueva 3 used in this study are housed in the Archaeological increasing lophs or cusp pairs, and is associated with functional Museum of Granada, Andalucía, Spain. Large mammal faunal lists of durability of dentition (Zliobaite et al., 2016). European Early and Middle Pleistocene fossil mammal localities - Presence of acute lophs (AL): 1 (present), 0 (absent); these are and archaeological sites were collected from literature (references sharp, cutting lophs that maintain sharp facets when worn, such as in Supplementary Table S1). Stratigraphic and sedimentological the ectoloph of most rhinos, the chisel-like lophs of tapirs, and the data for the GBB were also collected. In total, dental ecometric sharp buccal lophs of most cervids, but e.g. horses and almost all values and mean body size indices from 12 localities from the GBB (hypsodont) bovids lack these. Acute lophs are typical for browsers and altogether 92 Early and Middle Pleistocene European localities in forested environments (Zliobaite et al., 2016; 2018) If in doubt, AL were used in this study (Supplementary Tables S1 and S2). The for selenodonts is assigned if mesowear is sharp. The proportion of localities were chosen based on the following criteria: 1) the fauna acute lophs in a community is correlated with temperature today, include at least 5 species of large herbivorous mammals, 2) the as it is highest in deciduous forest communities where most 3  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 ungulate species cut branches when leaves are dropped during cold localities. For the body mass estimates, we used primarily post- season (see also Oksanen et al., 2019). cranial bone measurements utilising the regression equations of - Presence of obtuse lophs (OL): 1 (present), 0 (absent); these are Scott (1990) for bone measurements and body mass in modern “blunt” lophs that become basin-like without maintaining sharp ungulates, and Christiansen (2004) for proboscideans. These were facets when worn (e.g., the transverse lophs in rhinos, hippos, complemented by body mass estimates based on second upper and gomphotheres and elephants, and most of the “lophs” in bovids and lower molar lengths for taxa for which postcranials were not equine horses). available, using the equations of Janis (1990). - Structural fortification of cusps (SF): 1 (present), 0 (absent); Relative body size indices were calculated for each of the 92 these are any morphological features of the cusps that make the Early and Middle Pleistocene localities. This was done using the cusp stand out from the occlusal surface (by resisting wear) in worn following method: 1) Measurements of postcranial bones or molar teeth. The most typical examples of such structures are: buccal teeth were obtained from the literature (or measuring in museums cusps with bulbous, thickened lingual enamel loops in some ru- by JS) and used for calculating body mass estimates (preferably for minants (e.g. reduncine antelopes and bovines), cusps with heavily populations of the species within localities, but if measurement folded enamel (e.g. in many pigs such as Sus scrofa), and molars data of a species was lacking for a locality, total mean body mass with raised central ridges formed by conules or thickened enamel estimate for the species was used instead). 2) Mean body mass was edges (e.g. in Hippopotamus). Structural fortification occurs in calculated for each locality as the mean value of the body mass modern herbivorous mammal communities in fresh-grass grazers estimates of each ungulate species in the locality (palae- and mixed-feeders in wooded habitats, primarily in forests and opopulations). 3) Relative body mass indices (BMloc) were calcu- wetlands, and are thus indicators of such habitats (Zliobaite et al., lated for each locality by dividing their mean body mass value with 2018). the largest mean body mass value in the total set of 92 localities. - Occlusal topography (OT): 1 (flat), 0 (non-flat); this refers to Thus, the locality with the largest BMloc has the value 1, and all the the overall shape of the worn occlusal surface of the teeth in bucco- other localities have values between 0 and 1. The mean body mass lingual direction. Flat occlusal topography is typical for grazing indices of the localities were then used in the human distribution ungulates such as horses. modelling analysis as an additional ecometric trait, as average body -Plain selenodonty (PS): 1 (present), 0 (absent): this is a com- size of ungulates correlates with habitat openness on one hand, and bination of traits where longitudinal loph count is 2, only obtuse resource quality and availability on the other hand (see Saarinen, lophs are present, there is no structural fortification, no coronal 2014; Saarinen et al., 2016). Proboscidea was omitted from the cement and the occlusal topography is non-flat. This dental body size categories because of lack of resolution in the body mass morphology type is particularly abundant in ruminants and camels estimate data. The normalised body mass indices for the localities inhabiting relatively harsh environments, such as deserts and are composite values that take into account the relative size of the montane environments. species (whether on average small- or large-sized species are -Bunodonty (BUN): 1 (present), 0 (absent); bunodonty refers to dominant in the localities) and for abundantly sampled species, dental morphology with separate cusps that are not fused into body mass variation within species. elongated, loph-like structures. This dental morphology type is typical for omnivores and frugivores such as most suids and pri- 2.4. Mean annual precipitation, mean annual temperature and net mates. High proportion of bunodont taxa indicates relatively warm primary production estimates for the pliocene e pleistocene of and humid forest environments. Guadix-Baza Basin We used mesowear analysis (Fortelius and Solounias, 2000) for reconstructing the dietary composition of herbivorous mammals in Mean annual precipitation and net primary productivity in the an attrition-abrasion (browse vs. grass) continuum for the three GBB localities ranging from the Pliocene (ca. 4.5 Ma) to the Middle localities from the vicinity of the village of Orce, Venta Micena, Pleistocene (ca. 400 ka) were estimated based on hypsodonty, Barranco Leo  n and Fuente Nueva 3. Average mesowear signal in the longitudinal loph count and presence of acute lophs, averaged in ungulate communities from the Orce sites provides further infor- the large herbivorous mammal assemblages, using the equations by mation about the vegetation type at the Orce sites. The mesowear Oksanen et al. (2019). Based on these estimates, the climate history method is based on scoring the relief and shape of worn ungulate in the GBB from the Early Pliocene to the Middle Pleistocene is molar teeth, which reflects the degree of abrasion caused by grass discussed. The equations for the estimation are: on the molar surface, and allows comparing the dietary composi- tion in the fossil ungulate populations with known diets of extant Mean annual temperatureðMATÞ ¼ 27  :28:5 AL ungulates by hierarchical clustering (Fortelius and Solounias, 2000). Mesowear was scored at the specimen-level. The hierar- Mean temperature of coldest monthðMINTÞ ¼ 18  :42:9 AL chical clustering was performed using Ward's method with Euclidean distances. The clustering was performed in two- dimensional space, based on the shape and relief scores of the Mean annual precipitationðMAPÞ ¼ 2:491  289 HYP traditional mesowear scoring system of Fortelius and Solounias  841 LOP (2000). Net primary productionðNPPÞ ¼ 2601  144 HYP  935 LOP 2.3. Locality mean body size indices The climatic variables (MAT, MINT and MAP) for the modern In addition to the dental ecometric traits (crown types and surroundings of Baza were obtained from long-term weather mesowear), body mass estimates were calculated for a large set of observation data (https://en.climate-data.org/europe/spain/ large herbivorous mammals from the 92 Early and Middle Pleis- andalusia-252/). NPP for the modern Baza area was estimated tocene European localities. Relative body size indices were calcu- from dental ecometrics of a combined species assemblage of wild lated for each locality by calculating the mean body mass of ungulates and domestic ungulates that are herded in that region ungulate species within each locality and dividing these locality today. This approach is justified based on the observation that means with the highest mean body mass in the total set of adding feral domestic ungulates in dental ecometrics -based 4  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 estimates of climate and vegetation type compensate for the bias in We performed principal component analyses (PCA) to further the ecometric signal caused by the loss of ungulate species by hu- explore the effect of ecometric traits on the presence/absence of man interference, for example as has been observed in the Sahel humans in Early Pleistocene and Middle Pleistocene localities today (Zliobaite et al., 2018). separately, and the presence of Early Pleistocene, early Middle The MAP, MINT and NPP estimates of the GBB localities based on Pleistocene and late Middle Pleistocene human sites in the com- dental ecometrics were compared with both estimated and plete set of localities. This was done in order to compare the relative measured values for modern biomes and ecosystems using hier- and common associations of the ecometric traits with human archical clustering. For the comparisons, dental ecometric esti- presence. Partial per cent contribution of the ecometric traits on the mates of MAP, MINT and NPP were calculated for selected modern principal components were used for evaluating the significance of IUCN sites from Eurasia, North America and East Africa, and several associations of the constituent ecometric traits on human presence, Early and Middle Pleistocene early human sites from Africa, in order and generalised conclusions about the environmental conditions of to evaluate the similarity of the GBB sites with modern environ- most Pleistocene European human presence sites are discussed ments and African early human sites (Supplementary Table S3). In based on this analysis. addition, measured MAP, MINT and NPP values for the modern biomes were calculated from the values reported by Olson et al. 3. Results (2013) and compared with the dental ecometric estimates of MAP, MINT and NPP of the GBB sites (Supplementary Table S4). The 3.1. Pliocene to Middle Pleistocene history of climate and net hierarchical clustering was performed using Ward's method, with primary production in the Guadix-Baza Basin the Euclidean distances. The MAP, MINT and NPP values were normalised for the hierarchical clustering analyses. Mean annual precipitation (MAP) and net primary production The IUCN sites for the comparison of ecometric estimates from (NPP) estimates based on dental ecometrics in large herbivorous the Pliocene and Pleistocene GBB localities were chosen to repre- mammal communities indicate a drastic decline from ca. 870 to ca. sent a wide range of modern environments that are relatively un- 450 mm/a in MAP and from ca. 1130 to ca. 700 g(C)/m2/a in NPP from disturbed by human activities and contain diverse large the Early Pliocene of Baza 1 (ca. 4.6 Ma) to the Plio-Pleistocene herbivorous mammal faunas, including boreal forests (Nahanni NP boundary of Hue lago (ca. 2.6 Ma) in the GBB (Table 1, Fig. 2). After and British Columbia), boreal forest-grassland mosaic (Yellow- this, Fuente Nueva 1 (ca. 2.2 Ma) and Fonelas P-1 (ca. 2 Ma) as well as stone), European temperate forest (Bialowieza NP), Mediterranean Venta Micena (ca. 1.6 Ma) have the lowest estimated MAP and NPP, high-altitude forest (Sierra Nevada NP), Mediterranean low- while Barranco del Paso (ca. 1.8 Ma) with intermediate age has altitude woodland/shrubland (near Baza), and a series of tropical relatively high MAP and NPP estimates. After the drop at ca. 1.6 Ma in African environments from rainforest (Nouabale-Ndoki NP) to Venta Micena, the MAP and NPP estimates become more homoge- grassland savanna (Serengeti) and dry shrubland/woodland (near nous and equable, ranging from ca. 600e700 mm MAP and Nairobi). In addition, dental ecometric estimates from a wide range 900e1050 g(C)/m2/year NPP. The MAP in the GBB today is around of African early human sites were chosen for comparison, including 450 mm (ca. 470 mm in Guadix (https://en.climate-data.org/europe/ Lake Turkana region, Kenya, Olduvai, Tanzania and Swartkrans, spain/andalusia/guadix-57014/), ca. 430 mm in Baza (https://en. South Africa (Supplementary Table S3). climate-data.org/europe/spain/andalusia/baza-56902/)), which is To complement the hierarchical clustering analysis of the GBB similar to the estimated value for Fuente Nueva 1 and Venta Micena, sites, we use MAP ranges of modern Mediterranean biomes (Que zel while the rest of the Pleistocene localities in the GBB are estimated to and Me dail, 2003; Suc and Popescu, 2005) as a further reference have higher mean annual precipitation values. Also the NPP value of point for discussing the palaeoenvironments of the GBB sites. modern surroundings of Baza, including semi-feral domestic un- gulates, is similar to the estimated values for Fuente Nueva 1 and 2.5. Modelling of the environmental conditions under which early Venta Micena (Table 1). humans occurred in the early e Middle Pleistocene in europe In the hierarchical clustering of dental ecometric-based MAP, MINT and NPP estimates, Venta Micena and Fuente Nueva 1 For the modelling of early human habitats in Early and Middle consistently cluster together with modern surroundings of Baza Pleistocene in Europe, we used a tailored species distribution (Fig. 3 a and b), and with modern Mediterranean oak and pine modelling approach. The modelling task was to predict the prob- woodland biomes (Fig. 3 b). Hue lago and Fonelas P-1, together with ability of presence or absence of humans based on ecometric con- Dmanisi, consistently form a cluster of their own next to the cluster texts of large mammal communities on sites. The model itself was a containing modern Baza and Mediterranean forests in Fig. 3 a and b. logistic regression fit via generalised linear modelling. The pres- Baza 1, Solana del Zamborino and Barranco del Paso cluster ence/absence of humans was coded for 92 Early and Middle Pleis- together with Bialowieza forest and Sierra Nevada National Park, tocene palaeontological and archaeological localities from Europe, indicating temperate forest-dominated palaeoenvironments. The and dental ecometric traits of large herbivorous mammals were rest of the GBB sites, including BL and FN 3, form a cluster of their used as the predictive variables for the probability of human own together with modern African woodland/shrubland and some presence. We experimented on the analysis using different com- Pleistocene African sites, and this cluster is separate from modern binations of traits in the analysis (but with a maximum of 5 traits at African grassland savanna (Serengeti plains) and most of the time to avoid overfitting) and ended up with a model using hyp- Pleistocene East African hominin sites (Fig. 3 a). Also ‘Ubeidiya, sodonty (HYP), acute lophs (AL), structural fortification (SF) for Israel, clusters together with these sites. Early Pleistocene and Middle Pleistocene localities, and HYP, AL, SF A further comparison of the MAP values of the GBB sites with and locality mean body size indices (BM) for the complete set of MAP ranges of modern Mediterranean biomes (Que zel and Medail, localities. We evaluated how well the model predicts presence of 2003; Suc and Popescu, 2005) indicates that Hue lago, Fuente Nueva humans by applying a leave-one out approach to compute the 1, Fonelas P-1, Mencal 9 and Venta Micena, together with modern average Area Under the Roc Curve (Fielding and Bell, 1997). In environment of Baza, fall within the range of modern “forest addition, we used mean absolute errors and confidence statistics steppe” with Artemisia, Pinus halepensis and scarce evergreen over the wrights og input variables in the models to compare the Quercus (Table 1). In contrast, Barranco del Paso, Fuente Nueva 3, performance of the models. Barranco Leo  n, Huescar 1 and Cúllar-Baza 1 are within the range of 5  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 Table 1 Mean annual precipitation (MAP), minimum annual temperature (MINT) and net primary production estimates for the GBB sites, based on mammalian dental ecometrics (Oksanen et al., 2019). Modern MAP and MINT values for the vicinity of Baza are from climate observation data (https://en.climate-data.org/europe/spain/andalusia/baza- 56902/) and the modern NPP value is estimated from dental ecometrics of large wild and domestic ungulates occurring in the area today. The Mediterranean vegetation zones (Que zel and Me dail, 2003) indicated in the last column are: 1) Forest-steppe with Artemisia, Pinus halepensis, Juniperus and scarce evergreen Quercus (250 < MAP < 600 mm), 2) Evergreen forest with sclerophyllous oaks, Pinus pinaster and P. pinea (600 < MAP < 800 mm) and 3) Mixed forest with deciduous oaks, Fagus and conifers (MAP > 800 mm). Age Locality presence of MAP MINT NPP (g(C)/ vegetation type estimated from MAP range of modern Mediterranean vegetation zones (Ma) humans (mm) ( C) m2/a) zel and Me (Que dail, 2003) 4.6 Baza 1 absent 873 [0] 1131 3 2.6 Huelago absent 447 0.5 691 1 2.2 Fuente Nueva 1 absent 384 10 713 1 2 Fonelas P-1 absent 543 1.5 797 1 1.8 Barranco del Paso absent 678 1 975 2 1.7 Mencal 9 absent 538 13 882 1 1.6 Venta Micena absent 449 11 767 1 1.4 Barranco Leon present 602 8.5 914 2 1.2 Fuente Nueva 3 present 602 8.5 914 2 0.9 Huescar 1 present? 708 12 1049 2 0.7 Cúllar Baza 1 present 602 8.5 914 2 0.4 Solana del present 945 2 1221 3 Zamborino 0 Modern (around present 433 7 683 1 Baza) Fig. 1. Map of the Guadix-Baza Basin with the localities. evergreen forests dominated by sclerophyllous Quercus and Pinus, indicate purely grazing diet similar to extant zebras and white and Baza 1 and Solana del Zamborino have higher MAP values rhinoceros, and in BL and FN3 grass-dominated but mixed diet indicating affinity with mixed forest of deciduous Quercus, Fagus similar to extant saiga antelope and rock hyrax. Abundance of sharp and conifers (Table 1). We consider these vegetational in- cusps but low relief in E. altidens molars from BL and FN3 indicates a terpretations as further reference to the vegetation at the GBB sites, significant dietary element of some sort of tough vegetation that but note that these are likely to be oversimplified as they do not was not very abrasive. Somewhat surprisingly, the larger species of take into account temperature and NPP (see Discussion). horse, Equus suessenbornensis, from BL and FN3 have mesowear signals similar to extant browsers such as moose and Javan rhi- noceros, although they are represented by relatively small sample 3.2. Mesowear analysis of the large herbivorous mammals from the sizes. There is a slight but interesting difference between Venta orce sites Micena and the rest of the Orce sites, especially in the estimated diet of the horses: the browsing large species of horse (E. suessen- The results show that most of the Orce ungulate species from all bornensis) is absent from Venta Micena and the diet of the smaller the three localities (Venta Micena, Fuente Nueva 3 and Barranco n) are clustered with extant ungulates that have predominantly species, E. altidens, is more grass-dominated in VM than in the other Leo sites. browsing and browse-dominated diets (Fig. 4, see Supplementary Table S5 for a summary of mesowear results of the Orce un- gulates). Only Equus altidens shows mesowear signals that in VM 6  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 Fig. 2. Mean annual precipitation and net primary production in the Guadix-Baza Basin from the Early Pliocene to the Middle Pleistocene as estimated from dental ecometrics of large herbivorous mammals using the equations published by Oksanen et al. (2019). 3.3. Early human (Homo sp.) habitat modelling based on dental high values of acute lophs (AL), structural fortification (SF), high ecometrics and body size of large herbivorous mammals mean body mass index of ungulates in localities (BM_loc), mean bunodonty (BUN) and mean horizodonty (HOD), moderate values We selected the following model for predicting human occur- of hypsodonty (HYP) and low values of plain selenodont ecomorphs rence for the complete set of Early and Middle Pleistocene localities, (PS) (Appendix,Fig. A.1). The positive correlation with the presence based on estimated coefficients of dental traits, their p-values, ac- of Early Pleistocene Homo sp. is dominated by the effects of high curacy as area under ROC and the mean absolute error: percentage of structural fortification and high locality mean body P (occurrence) ¼ 1/(1 þ e-Z), where size index, while the correlation with the Middle Pleistocene Z ¼ 15.852 þ 3.542*HYP þ 10.662*SF þ 9.526*AL þ 5.327*BMloc. humans varies more is more strongly driven by high percentage of In the model for all Early and Middle Pleistocene localities, the acute lophs (Fig. 5). Furthermore, within the Middle Pleistocene, strongest significant positive effect for the probability of human the Early Middle Pleistocene human sites are more associated with presence is given by percentage of the presence of acute lophs (AL), high percentage of acute lophs (AL) than the Late Middle Pleisto- followed by percentage of structural fortification (SF) in the large cene localities (Appendix,Fig. A.1). mammal communities (Table 2). When the model is divided into Human habitat suitability scores were calculated based on the Early Pleistocene and Middle Pleistocene separately, hypsodonty total dental ecometric model prediction for all the 92 Early and (HYP), structural fortification (SF) and acute lophs (AL) remain the Middle Pleistocene European localities included in this study and most significant constituents of the model, but the model is compared with the presence or absence of humans in the localities. different for Early Pleistocene and Middle Pleistocene. In the Early The human habitat suitability scores for the localities, ranked from Pleistocene, the most significant positive correlation for the prob- highest to lowest, are presented in Supplementary Table S6. Since ability of human presence is provided by structural fortification, the model was fit on observed presence-absence data, the sites whereas in the Middle Pleistocene the most significant correlation where the evidence for human presence is lacking generally have is with acute lophs (Table 2.) the lowest habitat suitability scores and sites where humans are Component 1 in the principal component analysis of human present have the highest scores, but there are some notable ex- presence based on the traits corresponds to 46.5% of variation in ceptions to this pattern. Most notably, several localities with evi- human presence and it is contributed by the opposite effects of dence for human presence have lower-than-average estimated SF þ AL þ BMloc and PS þ HYP (Fig. 5). Component 2 corresponds to habitat suitability scores (Atapuerca TE9c, Bisnik Cave, Visogliano, 38.6% of the variation in human presence and is mainly driven by Orgnac 3, Caune de l’Arago, Pakefield, Swanscombe, Pirro Nord, the opposite effects of PS and SF þ BMloc. Hoxne, Vallparadis, Dmanisi and Cúllar-Baza 1). Of these, Orgnac 3, In general, human occurrence is concentrated in localities with cool stage levels of Caune de l’Arago and Bisnik Cave are 7  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 Fig. 3. Hierarchical clustering of the Pleistocene Guadix-Baza sites based on MAP, MINT and NPP estimated from dental ecometrics with a) dental estimates of MAP, MINT and NPP for modern IUCN sites and Pleistocene African hominine sites, and b) reported MAP, MINT and NPP for modern biomes (Chirici et al., 2007; Olson et al., 2013). The Euclidean distance values marking branch lengths are indicated for 11 first branches. The GBB localities in separate clusters are marked with different colours. Dmanisi and ‘Ubeidiya are included as examples of earliest Eurasian human sites outside the GBB. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) characterised by the presence of cold-climate taxa such as Coelo- sites Barranco Leo  n and Fuente Nueva 3 have a higher habitat donta and Rangifer tarandus, indicating harsh climatic conditions in suitability score for early humans than Venta Micena where evi- this respect. In contrast, some sites with no evidence for human dence of human presence is lacking. presence have higher-than-average habitat suitability scores, sug- The colour-interpolated maps of human habitat suitability gesting that conditions there could have been suitable for humans, scores in Europe through time indicate geographical patterns of or more specifically, that condition there were quite similar to those habitat suitability that reflects the history of early human dispersals where humans have found. Such sites without evidence for human in Europe (Fig. 6). Some of the earliest indication of increase in presence but with high suitability estimates are the Early Pleisto- habitat suitability for humans (as predicted by the model) occur in cene sites Colle Curti, Cesi, Castagnone and Monte Riccio in Italy the first phase of Early Pleistocene (2.6e1.8 Ma) in Central Italy and and the early Middle Pleistocene sites of Mosbach and Prezletice in Dmanisi, Georgia. During the second phase of Early Pleistocene Central Europe. According to the model prediction, the Orce human (1.8e0.7 Ma) the areas of elevated human habitat suitability scores 8  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 Fig. 4. Hierarchical clustering of mesowear of the Orce ungulates with extant ungulate species, based on percentage of high relief, percentage of sharp cusps and percentage of blunt cusps (following the methodology of Fortelius and Solounias, 2000). Dietary interpretations are based on known diets of the modern ungulate species in the clusters (see Fortelius and Solounias, 2000). The ungulate taxa from Barranco Leo n (BL) are marked in green colour, the ones from Fuente Nueva 3 (FN3) in red, and the ones from Venta Micena (VM) in blue. The numbers in brackets indicate number of dental specimens used for each ungulate palaeopopulation from the Orce sites. Half square Euclidean distances are indicated for the first 10 clusters. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) are centered in Israel, Spain, Northern Italy and in the Black Sea 4. Discussion coast. During the early Middle Pleistocene, Central Europe and Southern England show a notable increase in the estimated human The Guadix-Baza Basin in Andalucía, Spain, has particular rele- habitat suitability scores. In the late Middle Pleistocene, most of the vance for the question of early human habitats and dispersals, as European site have relatively high habitat suitability scores, except n and the archaeological and palaeontological sites of Barranco Leo for a few sites indicating cold climatic conditions, such as Orgnac 3 Fuente Nueva 3 (both ca. 1.4 Ma in age) have provided some of the (see Fig. 6). earliest evidence for the presence of early humans (Homo sp.) in Europe (Toro-Moyano et al., 2013). Understanding the history of climatic and environmental conditions in the GBB and elsewhere in 9  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 Table 2 Coefficients for the dental ecometric traits from the generalised linear model prediction for the probability of human occurrence in Early and Middle Pleistocene Western Eurasian sites. Positive coefficients indicate positive correlation with the probability of human occurrence. Hypsodonty, acute lophs, structural fortification and locality mean body mass index have positive effects for the probability of human occurrence. Sample variables coefficient standard error z value Pr (>|z|) Mean absolute error Accuracy as area under ROC Early Pleistocene intercept 12.812 8.934 1.434 0.15 0.20 0.76 HYP 3.834 3.016 1.252 0.21 SF 11.035 4.997 2.209 0.03 AL 3.705 6.69 0.554 0.56 Middle Pleistocene intercept 7.502 6.108 1.228 0.219 0.34 0.63 HYP 2.112 2.100 1.006 0.315 SF 5.792 4.612 1.256 0.209 AL 7.095 5.157 1.376 0.169 Total intercept 15.852 5.04 3.145 0.002 0.29 0.76 HYP 3.542 1.603 2.21 0.03 SF 10.662 2.926 3.644 0.0003 AL 9.526 3.631 2.624 0.009 BMloc 5.327 3.965 1.347 0.18 the Early Pleistocene of Europe is highly important for under- consistently cluster next to the cluster containing estimated MAP, standing the underlying environmental conditions under which MINT and NPP for modern Baza and measured values for modern early human dispersal to Europe happened, and ecometric and Mediterranean woodlands, but they differ mainly in having cooler palaeoecological analyses from fossil mammals provide new, pre- estimated mean temperatures of coldest month (MINT) than at viously understudied insights into these. present. Comparison of the MAP estimates of the GBB sites with Fossil molar teeth of large herbivorous mammals are abundantly MAP ranges of modern Mediterranean biomes (Que zel and Me dail, available for the Pliocene to Middle Pleistocene fossil localities in 2003; Suc and Popescu, 2005) provides an additional reference for the GBB, and thus the dental ecometrics of large herbivorous comparing the vegetation types in the GBB localities. According to mammals provide, to our knowledge, thus far the most compre- this classification, MAP range between 250 and 600 mm/a corre- hensive record of climate and environmental history in the spond with forest steppe habitats with Artemisia, Pinus halepensis, archaeological and palaeontological sites in GBB and elsewhere in Juniperus and scarce evergreen Quercus. The modern environment Europe during the Early and Middle Pleistocene. around Baza, as well as the fossil sites of Huelago, Fuente Nueva 1, Fonelas P-1 Mencal 9 and Venta Micena fall within this MAP range. 4.1. Pliocene to pleistocene climate and palaeoenvironments in the The rest of the Early and Middle Pleistocene localities in the GBB, GBB e high-productivity woodland ecosystems in the including the earliest human occupation sites of Fuente Nueva 3 mediterranean climatic realm (FN3) and Barranco Leo n (BL), are clustered together with African woodland or shrubland environments, the Early Pleistocene Natoo MAP, MINT and NPP estimated from dental ecometric traits in Formation from West Turkana, Kenya, and the Early Pleistocene site large herbivorous mammal communities indicate a prevalence of of Swartkrans, South Africa, based on dental ecometric estimates of palaeoenvironments that are most closely clustered with modern MAP, MINT and NPP. Furthermore, these sites form a cluster of their Mediterranean forests than other extant biomes for the GBB sites. own that is more closely clustered with the modern Mediterranean The high MAP and NPP estimates for the Pliocene site of Baza 1 (ca. forests than other modern biomes when compared with reported 4.5 Ma(Pin~ ero et al., 2017; Rodríquez et al., 2015), the Early Pleis- MAP, MINT and NPP for modern biomes (data from Chirici et al., tocene Barranco del Paso (ca. 1.8 Ma; Gibert et al., 1992) and the 2007; Olson et al., 2013). In these environments, the estimated Middle Pleistocene Solana del Zamborino (ca. 0.4 Ma; Janis, 1990; precipitation varies from similar to somewhat higher than today in  (Alvarez-Posada et al., 2017); although see (Scott and Gibert, 2009)) the GBB and is lower than in temperate forest biomes, but NPP indicate a much more humid climate and more high-productivity estimates are higher than in modern Mediterranean and temperate palaeoenvironments than in the GBB today. Hierarchical clus- forests. On the other hand, the estimated precipitation and NPP are tering of these sites with modern IUCN sites indicates a close lower than in modern tropical forests and humid savannas, and similarity to the temperate mixed forests of Bialowieza and Sierra estimated NPP is significantly higher than in C4 grassland savannas. Nevada National Parks. This interpretation is further supported by After the relatively dry conditions at Venta Micena, precipitation comparison of estimated MAP with the MAP range of modern and NPP increased around 1.4 Ma at BL and FN3, the oldest human Mediterranean biomes (Que zel and Medail, 2003; Suc and Popescu, sites in the GBB. The dental ecometrics-based estimates of MAP 2005), where Baza 1 and Solana del Zamborino fall within the range (602 mm) and NPP (914 g(C)/m2/a) are higher than today in the of mixed forests with deciduous oaks, Fagus and conifers. According GBB, while estimated minimum annual temperature is similar to to Fauquette et al. (1999), the taxonomic composition of the present (8.5  C). Herpetofauna from these sites indicates that vegetation in the Mediterranean environments was largely similar temperature and precipitation in BL and FN 3 were higher than at to present already in the Pliocene, with the exception of additional present (Blain et al., 2011, 2016), and the vegetation would have presence of subtropical elements such as Taxodiaceae, Sequoia, been dominated by relatively open woodland type (Blain et al., Engelhardia and Symplocos. 2011; Sanchez-Bandera et al., 2020). MAP range between 600 and During the Early Pleistocene, the localities of Fuente Nueva 1 (ca. 800 mm/a in modern Mediterranean realm indicates the presence 2.2 Ma) and Venta Micena (ca. 1.6 Ma) show MAP and MINT esti- of evergreen forests with sclerophyllous oaks (e.g. Quercus ilex- mates that are similar to the present MAP and MINT in the GBB, and coccifera) and pines, and the fossil sites of Barranco del Paso, FN3, NPP estimates that are similar to values estimated from dental BL, Huescar 1 and Cúllar-Baza 1 fall within this vegetation category ecometrics of large mammals around Baza today and measured by modern analogue. However, note the affinity of these sites with values for modern Mediterranean woodland or forest steppe en- African woodland habitats in the hierarchical clustering which also vironments. Hue lago (ca. 2.6 Ma) and Fonelas P-1 (ca. 2 Ma) also uses MINT and NPP values in addition to MAP, indicating perhaps a 10  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 Fig. 5. Principal component analysis of human and non-human sites in Western Eurasia during the Early and Middle Pleistocene. The red dots mark localities where evidence of the presence of humans is lacking, and the open blue diamond symbols mark localities where humans were present. During the Early Pleistocene, most of the human occurrence sites are characterised by relatively high values of structural fortification and mean body mass, while sites with high hypsodonty and PS values suggesting “harsh” environments show an almost complete absence of human occurrence. In the Middle Pleistocene, the human sites are more evenly distributed over the ecometric space but are more concentrated towards sites with relatively high values of acute lophs, structural fortification and mean body mass, suggesting on average relatively wooded palaeoenvironments. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) slightly higher-productivity environment than modern Mediterra- Members of Koobi Fora Formation in East Turkana, Nariokotome nean sclerophyllous forests. In summary, the climatic and produc- Member in West Turkana, Olduvai Bed IV, and Kanjera. tivity estimates for the earliest human occupation sites in the GBB The interpretation of woodland or shrubland environments, (FN3 and BL) were climatically within the range of modern Medi- without a significant grassy component of undergrowth, at BL and terranean evergreen oak and pine forest environments, but had FN3 is further supported by the mesowear analysis of the ungulate relatively high NPP, indicating similarity to African dry woodlands/ communities. In both sites, vast majority of the ungulate species shrublands. However, these environments differed from African show mesowear values that cluster together with modern browsers grass-dominated savannas. In this regard, BL and FN3 were similar such as Alces and Rhinoceros sondaicus (most of the species) or with to some African early human sites (Swartkrans and Natoo Forma- browse-dominated mixed-feeders such as Taurotragus oryx (the tion), as well as ‘Ubeidiya, Israel, but were different from most of large cervid, cf. Praemegaceros verticornis, from BL) or Tragelaphus them that indicate palaeoenvironments similar to extant grass- strepsiceros (the large bovines from BL and FN3). Most notably, even dominated savannas of Serengeti, such as Upper Burgi and KBS the larger species of horse (cf. Equus suessenbornensis) from both of 11  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 Fig. 6. Colour-interpolated maps of probability of human presence (human habitat suitability estimated from dental ecometric variables and locality body mass indices) for Early and Middle Pleistocene Western Eurasian sites. Blue indicates poor suitability for humans and red indicates the highest suitability, with the spectrum from light blue to green and yellow indicating increasing habitat suitability. In the earliest phase of Early Pleistocene, the earliest human site in Eurasia (Dmanisi, Georgia), Tegelen (Netherlands) and especially Central Italian sites have the highest suitability scores. In the later phase of Early Pleistocene, areas of highest suitability are centered around the Black Sea coast, Andalucia and Central Italy. In the early Middle Pleistocene, there is high variation in the habitat suitability, with many Central European and Italian sites getting high suitability scores. In the late Middle Pleistocene, habitat suitability is mostly high, being highest in the Mediterranean coast, Central Europe and Southern England, but poor in a few localities that indicate cold climate. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) the sites clusters with the browsers, while the smaller species of relatively high NPP. In the surviving natural temperate forests horse (E. altidens) is the only species from BL and FN3 that clusters today, such as Bialowieza, the large herbivorous mammals such as with the extant grass-dominated feeders saiga antelope (Saiga European bison, moose, red deer and feral horses, do not create tatarica) and rock hyrax (Procavia capensis). Such a prominence of significant open environments, but instead affect the species browsing diets in the ungulate communities indicates a palae- composition of the trees (Kujper et al., 2015). Unfortunately, we are oenvironment that could not have been grass-rich, and provided currently lacking temperate and Mediterranean environments with rather rich foliage of trees, shrubs or dicotyledonous herbs. Also, a diversity of large mammals even nearly as high as during the Early the ungulate assemblage from Venta Micena, with estimated dryer Pleistocene, making it challenging to estimate the effect of large climatic conditions, shows an abundance of browsing and browse- herbivores on vegetation. For example, the effect of elephants on dominated feeders, but a higher proportion of mixed-feeders and tree cover can be significant in some environments (Guldemond only one species of horses (Equus altidens), which has the only truly et al., 2017), and this effect could have been prominent in the grazing mesowear signal of all the ungulates in the Orce sites, Pleistocene but is currently lacking in temperate and Mediterra- clustering with modern grazers such as zebras (Equus burchelli and nean environments. The difficulty of assessing how drastic the ef- E. grevyi), tsessebe (Damaliscus lunatus) and prairie bison (Bison). fect of large herbivores on vegetation was in the past, especially as The mesowear data thus indicates a more open and grass-rich reliable estimates of herbivore population sizes are lacking, was palaeoenvironment in Venta Micena than in BL and FN3. acknowledged by Bradshaw et al. (2003). However, they conclude The climatic range estimated for most of the Pleistocene GBB that the balance of paleoenvironmental evidence indicates mostly a sites (including the Orce sites) falls within a temperature and rather local effect of herbivory on otherwise mostly forested precipitation regime under which the biome cannot be estimated landscapes during the Late Pleistocene and especially Holocene in directly from the climate, because other factors such as soil fertility, Northern Europe. fire and the top-down influence of large mammal herbivory have a The similarity of Venta Micena to modern Mediterranean “forest profound impact on the vegetation type (Bond, 2005; Kujper et al., steppe” and the present climatic conditions and productivity 2015). The NPP estimated by large mammal dental ecometrics around Baza, and the affinity of FN3 and BL with Mediterranean represents primary productivity edible by the herbivorous mam- evergeen forests with lack of significant grass undergrowth, prob- mals, and the high diversity and body size range of the Early ably relate to the difference in seasonality of rainfall patterns in the Pleistocene large mammal communities is expected to reflect Mediterranean realm and East Africa. The seasonality of 12  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 precipitation in East Africa includes two rainy seasons, whereas the environments in BL and FN3, which would more likely corre- Mediterranean climate is characterised by one rainy season in the spond with the woodland-type vegetation in the Palominas core. winter and hot and dry summers. Drought-tolerant trees and The inferred vegetation of such phases of the Palominas core is shrubs have an advantage during the hot and dry summer in the generally forested with some degree of openness and dominated by Mediterranean realm, while grasses that require more evenly deciduous and evergreen Quercus, with a diversity of mesophytes distributed annual rain are in disadvantage there especially due to such as Carpinus, Acer, Juglans, Corylus, Populus, Salix, Ulmus, Cas- extended autumn drought (Dallman, 1998; Nogueira et al., 2017). tanea, Fagus and Buxus, and thermophytes such as Olea, Pistacia, On the other hand, grass-dominated savannas in East Africa thrive Coriaria, Tamarix, Phlomis, Myrica, in addition to genera today under conditions with two rain seasons (e.g., Hesla et al., 1985). extinct in the western Mediterranean such as Eucommia, Parrotia, Such dissimilarity in the rainfall pattern could largely explain the Cathaya, Aralia, Zelkova, Pterocarya, Tsuga and Nyssa (Altolaguirre grass-poor environments in the GBB and the on average more et al., 2019, 2020). The sediments at BL and FN3 were deposited grass-dominated environments of the hominin sites of East Africa during a phase when the Baza Lake in the GBB was relatively large during the Pleistocene. due to basin dynamics (Oms et al., 2011). Sedimentological evi- Whatever the case, it is evident from the scarcity of grazers that dence indicates a palustrine depositional environment for both BL the vegetation in the Orce sites was not dominated by grasses and and FN3 (Oms et al., 2011 and references therein), and the presence would thus not likely have been particularly similar to modern of wetlands is further attested by faunal elements such as the African grasslands or grass-dominated savannas. Thus, our results abundance of hippopotamus (Hippopotamus antiquus) in these lo- do not support the “savannahstan” hypothesis (Dennell and calities. Freshwater sources at the periphery of the main Baza saline Roebroeks, 2005; Dennell, 2010) for the GBB sites, at least if the“- palaeolake would have provided a constant supply of groundwater savannahstan” is understood in the strict sense to be similar in (Anado  n et al., 2015) that may have overcome the limits imposed by environmental condition to African savannas with grassy under- rainfall on the vegetation, and supported high-productivity palus- growth. Although never dominant, Poaceae is relatively frequent in trine environments (Oms et al., 2011). the pollen record from Palominas lake sequence in the GBB These wooded phases are intersected in the Palominas core by (Altolaguirre et al., 2020), but its fluctuations do not parallel those phases during which the vegetation became steppe-like, domi- of the xerophytic component. This leaves open the possibility that a nated by pollen of Artemisia, Ephedra, Chenopodioideae, Asteraceae, portion of the total count is due to local overrepresentation of grass and Poaceae. The conditions during those stages were climatically pollen (e.g. marginal and/or basinal vegetation). Furthermore, it probably similar to present conditions, possibly allowing the early cannot be discarded that grasses could have increased regionally humans to survive in the area through the glacial stages and to during wet interglacials. Today, hard-leaved perennial grasses such thrive during the warm, humid phases such as those at BL and FN3 as Lygeum spartum (on marls and gypsum soils) and Stipa tena- (Altolaguirre et al., 2019). cissima (on limestones) dominate the local vegetation, but this was clearly not the case during the Early and Middle Pleistocene. In summary, the climate and productivity estimates from the 4.2. Early human palaeoenvironments in europe e savannas, dental ecometrics of the large herbivorous mammal community, forests or wetlands? and the comparison of those with modern biomes, indicate palaeoenvironments similar in vegetation structure to Mediterra- In a broader perspective, the dental trait distribution of large nean woodlands for the earliest human occupation sited in the GBB herbivorous mammals across Europe during the Early and Middle (BL and FN3), with temperature close to present and annual pre- Pleistocene indicates on average relatively high-productivity and cipitation and NPP slightly higher than at present in Baza. This wooded environments, and avoidance of the harshest environ- raises the question of what these environments were like in terms mental conditions. This is illustrated by the positive correlation of of resources available for the early humans. The predominantly high values of AL, SF, BMloc and BUN, and negative correlation of browse-dominated mesowear signals of ungulates suggests a high values of PS with human presence throughout the Early and palaeoenvironment where grasses were not a dominant element in Middle Pleistocene in Europe. Large average body size in ungulate the vegetation, but the browse-dominated mesowear does not communities, particularly in the temperate zone such as in Europe further separate whether the browse was mainly obtained from during the Quaternary, indicates moderately open habitats but also trees, shrubs or non-grass herbs. The Palominas pollen core in- high resource quality and availability (Saarinen, 2014; Saarinen dicates a high variation in the vegetation in the GBB during the et al., 2016). These environments, where ungulate communities Early Pleistocene, ranging from open grass and Artemisia -domi- have high proportion of acute lophs, structural fortification and nated during dry phases to diverse deciduous/evergreen Quercus large average body size, would have likely provided diverse re- woodland and wetland during the humid phases (Altolaguirre sources for the early humans. In many cases there is also evidence et al., 2019). Pollen record from a sedimentary section in Pal- of the presence of significant bodies of water and waterside envi- ominas provides a reference to the vegetation cycles in the GBB ronments, as indicated by the frequent co-occurrence of Hippo- during the Early Pleistocene roughly between the MIS 53 and MIS potamus with early humans in Europe. The importance of bodies of 34, suggesting that the levels indicating warm and humid condi- water for ancient human occupation is prominently attested by archaeological data (e.g. Cohen et al., 2012). Furthermore, Deve s tions correspond with the conditions present in BL and FN3 (Altolaguirre et al., 2019). The evidence from mammal ecometrics, et al. (2014) found a suggestive association of early Homo with herpetofauna and ungulate mesowear all point at relatively marshy/lakeside environments, where they suggest relatively low- somewhat more humid than present conditions and non-grassy risk exploiting large animals such as elephants would have been possible, and the use of rougher terrains was a secondary strategy 13  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 used in the absence of large, water-dependent herbivores. (2019) suggested that early hominines in Eurasia would have fav- Despite these interpretations of average environmental condi- oured wooded environments because of safer scavenging oppor- tions in early human occupation sites, open questions remain. One tunities compared to open savanna environments, and argued that of these is how the top-down effect of humans and carnivores on the early dispersal of Homo from Africa to the Caucasus region the herbivore community is reflected in the ecometric signal of the (Dmanisi) would have followed the distribution of such wooded early human environments. The sampling of the Early and Middle habitats ca. 1.8 Ma. However, the continuous presence of humans in Pleistocene sites for the dental ecometric analysis was made with Europe since the Early Pleistocene might indicate that they were caution in regard to sampling biases, including those introduced by able to survive periods of climatic deterioration and the spread of selection of prey animals by carnivores and humans. For this more steppe-like environments, perhaps leaving no traces in the reason, sites with accumulations gathered primarily by human or archaeological record because of diminished population sizes. For carnivore activities were avoided, such as carnivore dens and hu- example, Altolaguirre et al. (2019) suggested that while humans man dwellings. Furthermore, no systematic difference in the spe- first appeared in the GBB under relatively humid and wooded cies composition between human sites and non-human sites was conditions, they could have survived there during the coming observed in the herbivorous mammal communities. However, this glacial stages when the climate became dryer but not much cooler, does not preclude the possibility of human influence in the her- approaching modern conditions. This interpretation is supported bivorous mammal communities. Whether this matters for the by Sanchez-Bandera et al. (2020). Overall, the archaeological re- ecometric estimates of temperature, precipitation and NPP de- cords indicate that Homo sp. Was a frequent but not abundant pends on whether human influence changes the selection of her- species in European Early Pleistocene localities (Rodríquez et al., bivorous mammal species in the localities in a way that obscures 2015). the environmental signal. As we selected all sites with more than The importance of local conditions, often phytogeographically five large herbivorous mammal species present, and the species linked to micro- and mesoclimates and/or strong edaphism within the assemblages typically represent a wide range of eco-  n et al., 2019), should not be underestimated, and while (Carrio metric traits, it is unlikely that a systematic bias from human in- general climates could be cold dry and regional vegetation open fluence would obscure the environmental signal. woodland and steppe, the highly complex orography of the sur- The principal component analysis where Early Pleistocene sites rounding mountains may well have favoured the existence of with human presence (Homo sp.), early Middle Pleistocene sites (cf. biodiversity strongholds where ecologically disparate species Homo heidelbergensis) and late Middle Pleistocene sites (cf. Homo cohered in time and space during hundreds of millennia. The study neanderthalensis) are separated, provides additional information case of the mid-Pleistocene Neanderthal Cova de Bolomor, north- that points at interesting differences in the habitat preference of wards in Valencia, is particularly pertinent (Ochando et al., 2019). these different “chronospecies” of Homo. The particularly high This scenario could explain why Arctotertiary taxa mix up with correlation of Early Pleistocene Homo and late Middle Pleistocene xerothermophilous scrub in Palominas (Altolaguirre et al., 2020). Homo neanderthalensis with sites where the ungulate assemblages Needless to say, these phytodiversity reservoirs would be crucial to show high proportion of structurally fortified cusps in molars and  n et al., 2018). recolonization patterns during interglacials (Carrio large average body size indicate that they were mostly associated with woodland-type environments. On the other hand, the early 4.3. Early human occupation e boundary conditions vs. Middle Pleistocene Homo are particularly associated with sites attractions? where the ungulate communities have a high proportion of acute lophs, indicating forested environments and perhaps cooler cli- Some of the Early Pleistocene human occupation sites with matic conditions (Oksanen et al., 2019). This observation is inter- relatively low predicted human habitat suitability score have lower esting in the light of archaeological discoveries. For example, the mean hypsodonty values, but moderate structural fortification and earliest evidence of human occurrence in Britain comes from the ca. acute loph values (e.g. TE9c of Atapuerca, Vallparadís level 12). 900 ka site of Happisburgh 3, where the proxy evidence indicates Pollen and small vertebrate records indicate the presence of relatively cool climatic conditions and vegetation dominated by abundant wooded vegetation and relatively high precipitation at conifers, possibly similar to the southern edge of boreal forest, such TE9c (Blain et al., 2010; Huguet et al., 2017) and at Vallparadís as today in Southern Sweden (Parfitt et al., 2010; Farjon et al., 2020). (Martínez et al., 2010). Dmanisi, Georgia, which is the earliest of the The earliest human occupation in Europe during the Early human occupation sites (Garcia et al., 2010), has a comparatively Pleistocene thus seems to concentrate in relatively equable and low human habitat suitability score, which is, however, notably wooded but diverse environments, and our analysis also suggests elevated in comparison to most Early Pleistocene sites of compa- that the dispersal of early Homo into Europe roughly followed the rable age distribution of such environments through time, the conditions Humans are opportunistic in terms of diet and habitat re- being best first in the Caucasus region and the Mediterranean quirements and use tools to overcome limitations set by environ- (Spain and Northern Italy). This is in accordance with the ecological mental conditions. It is thus not surprising that the Pleistocene model proposed by Blain et al. (2021), based on herpetofaunal as- human occupation sites indicate a wide range of environments semblages, who suggested that rainfall and environmental hu- based on proxy data and dental ecometric modelling. The presence midity (expressed as forest cover) may have had an impact on the of humans at sites with the lowest human habitat suitability scores occupation of the territory during the Early and Middle Pleistocene based on the dental ecometric model, such as TE9c, Vallparadis and more than other factors as temperature. Agustí and Lordkidpanidze Dmanisi, could reflect other attractions, such as raw material for 14  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 tool making that spurred the hominins to overcome the challenges  contributed to writing the manuscript. - Indre_ Zliobait _ major role e: presented by the less-than-optimal environmental conditions. In in designing the human habitat distribution modelling and the R- TE9c the raw materials for the stone tools were gathered from the code for that analysis; contributed to writing the manuscript. - vicinity of the site at a range of less than 2 km (Huguet et al., 2017). Mikael Fortelius: significant role in collecting mesowear data and At Dmanisi, the range of volcanic and other rock types present in Guadix-Baza locality data, planning the research and writing the the area would have provided a high diversity of lithic raw mate- manuscript. - Daniel DeMiguel, Beatriz Azanza, Herve  Bocherens, rials (Baena et al., 2010). This could also, to some extent, be the case Carmen Luzo n, Jose Yravedra, Lloyd Courtenay, Hugues-Alexandre for the earliest human sites in the GBB (Barranco Leo  n and Fuente Blain, Christian Sa nchez-Bandera, Alexia Serrano-Ramos, Juan Nueva 3) as while the human habitat suitability score for them is  Rodriguez-Alba, Suvi Viranta, Deborah Barsky, Jose Solano- Jose moderately good, the value of the basin for human occupation García, Miikka Tallavaara, Oriol Oms, Jordi Agustí, Juan Ochando could have been enhanced by the availability of lithic raw materials,  S. Carrio and Jose  n: all authors contributed to data collecting and/or especially flint (Barsky et al., 2010; Titton et al., 2021). discussions concerning data analyses, and writing parts of the manuscript. - Juan Manuel Jime nez-Arenas: contributed to direct- 5. Conclusions ing data collection and writing the manuscript. According to our mammalian dental ecometric estimates, mean Declaration of competing interest annual precipitation and primary productivity decreased in the GBB from the Pliocene to the Pleistocene, but showed intensive The authors declare that they have no known competing fluctuation thereafter reaching a higher level again during the financial interests or personal relationships that could have n and Fuente earliest human occupation in the sites of Barranco Leo appeared to influence the work reported in this paper. Nueva 3 (1.4e1.2 Ma). Overall, the estimated environmental con- ditions in the GBB during the Pleistocene ranged from conditions Acknowledgements similar to modern Mediterranean woodlands and shrublands to woodland and forest environments with higher primary produc- This research has been carried out thanks to the permissions tion. The dietary spectrum of the ungulates from the Orce sites, and financial support of the General Research Project “Primeras especially Barranco Leon and Fuente Nueva 3, is heavily dominated  gico a partir de los ocupaciones humanas y contexto paleoecolo by browsers and browse-dominated feeders, suggesting a palae- depo  sitos Pliopleistocenos de la cuenca Guadix-Baza. Zona oenvironment where grasses were not a significant element of the Arqueolo gica de la cuenca de Orce” funded by the Consejería de herbivorous mammals’ diet. In Venta Micena (ca. 1.6e1.4 Ma) Cultura, Junta de Andalucía (Spain). Special thanks are given to the especially the equids show a more grass-dominated dietary signal, Archaeological Museum of Granada staff, headed by its Director although most of the ungulates there were also browse-dominated Isidro Toro, for its availability and kindness. Thanks are extended to feeders. the rest of the research team and to the volunteers who worked Early human distribution modelling in the Early and Middle during the field campaigns. Juha Saarinen was working on the Pleistocene of Europe based on dental ecometrics of large herbiv- Academy of Finland and University of Helsinki -funded post- orous mammals indicates that the environments where humans doctoral research project “Behavioural and morphological adapta- occurred during the Early Pleistocene were mostly woodland tion to environmental change: the example of African Neogene to -dominated environments, whereas in the Middle Pleistocene, Quaternary Proboscidea (Academy of Finland no: 315691) during humans dispersed to a wide range of environments, being some- this work. JS Carrio n and J Ochando acknowledge the grants PID what more concentrated in forest environments. 2019-1049449 GB-I00 funded by FEDER/Ministry of Science and Innovation, 20788/PI/18 Fundacio n Seneca, and Project SUMHAL, Data availability AEI. The Institut Catala  de Paleoecologia Humana i Evolucio  Social (IPHES-CERCA) has received financial support from the Spanish The data used for the analyses in this study are provided in Ministry of Science and Innovation through the “María de Maeztu” Supplementary Tables S1-S5. Further details on raw-data collection program for Units of Excellence (CEX 2019-000945-M). H.-A.B., and sources can be inquired from the corresponding author: juha. C.S.-B. and J.A. were supported by the projects CGL 2016-80000-P saarinen@helsinki.fi. (Spanish Ministry of Economy and Competitiveness) and 2017SGR-859 (Generalitat de Catalunya). We acknowledge the Author contributions funding of the project “Environment of early hominins outside Africa: the Guadix-Baza Basin” by the Leakey Foundation to H. - Juha Saarinen: collected most of the mesowear and ecometric Bocherens. B. Azanza and D. DeMiguel belong to the Gobierno de data, performed the analyses, made the figures (except Fig. 1) and Arago n Research Group E33_20R. J.M. Jime nez-Arenas belongs to wrote most of the text (main role in planning and preparing the the Junta de Andalucía Research Group HUM-607. We thank two research and the manuscript). - Otto Oksanen: major role in data anonymous reviewers and the editor for their constructive com- collecting, especially species lists of Pleistocene localities; ments that were helpful in improving this article. 15  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 Appendix Figure A.1. Principal component analysis of localities with Early Pleistocene Homo, early Middle Pleistocene Homo and late Middle Pleistocene Homo, with effect of individual dental traits on the principal components. Early and late Middle Pleistocene Homo localities are concentrated on environments with relatively high SF, HOD and OT values, which early Midde Pleistocene Homo localities are associated with localities where average AL and BUN values are high. Localities without evidence for the presence of Homo are associated with high mean PS (only mean is indicated). Appendix A. Supplementary data palaeontological site ‘Solana del Zamborino’ (GuadixeBaza Basin, S Spain). Scientific Reports 7, 13495. https://doi.org/10.1038/s41598-017-14024-5. Anado  n, P., Oms, O., Riera, V., Julia, R., 2015. The geochemistry of biogenic car- Supplementary data to this article can be found online at bonates as a paleoenvironmental tool for the Lower Pleistocene Barranco Leo n https://doi.org/10.1016/j.quascirev.2021.107132. sequence (BL-5D, Baza Basin, Spain). Quat. Int. 389, 70e83. Azanza, B., Alberdi, M.T., Prado, J.L., 2000. Large mammal turnover pulses correlated with latest Neogene glacial trends in the northwestern Mediterranean region. References Geological Society, London, Special Publications 181, 161e170. Baena, J., Lordikpanidze, D., Cuartero, F., Ferring, R., Zhvania, D., Martín, D., Shelia, T., Agustí, J., Blain, H.-A., Cuenca-Besco s, G., Bailon, S., 2009. Climate forcing of first Bidzinashuili, G., Roca, M., Rubio, D., 2010. Technical and technological hominid dispersal in Western Europe. Journal of Human Evolution 57, 815e821. complexity in the beginning: the study of Dmanisi lithic assemblage. Quat. Int. https://doi.org/10.1016/j.jhevol.2009.06.005. https://doi.org/10.1016/j.quaint.2010.01.019. Agustí, J., Blain, H.-A., Furio , M., de Marfa , R., Santos-Cubedo, A., 2010. The early Barsky, D., Celiberti, V., Cauche, D., Gre goire, S., Lebegue, F., de Lumley, H., Toro- Pleistocene small vertebrate succession from the Orce region (Guadix-Baza Moyano, I., 2010. Raw material discernment and technological aspects of the Basin, SE Spain) and its bearing on the first human occupation of Europe. Quat. Barranco Leon and Fuente Nueva 3 stone assemblages (Orce southern Spain). Int. 223e224, 162e169. Quat. Int. 223e224, 201e219. Agustí, J., Lordkidpanidze, D., 2019. An alternative scenario for the first human Belmaker, M., 2006. Large Mammalian Community Structure through Time: dispersal out of Africa. L’Anthropologie 123, 682e687. ‘Ubeidiya as a Case Study. Ph.D. Dissertation. The Hebrew University of Alberdi, M.T., Alonso, M.A., Azanza, B., Hoyos, M., Morales, J., 2001. Vertebrate Jerusalem. taphonomy in circum-lake environments: three cases in the Guadix-Baza Basin Belmaker, M., 2010a. Early Pleistocene faunal connections between Africa and (Granada, Spain). Palaeogeogr. Palaeoclimatol. Palaeoecol. 165, 1e26. Eurasia: an ecological perspective. In: Fleagle, J.G., Shea, J.J., Grine, F.E., Altolaguirre, Y., Postigo-Mijarra, J.M., Barro n, E., Carrio n, J.S., Leroy, S.A.G., Baden, A.L., Leakey, R.E. (Eds.), Out of Africa I: the First Human Colonization of Bruch, A.A., 2019. An environmental scenario for the earliest hominins in the Eurasia. Springer, Dordrecht. Iberian Peninsula: early Pleistocene palaeovegetation and palaeoclimate. Rev. Belmaker, M., 2010b. On the road to China: the environmental landscape of the Palaeobot. Palynol. 260, 51e64. Early Pleistocene in Western Eurasia and its implication for the dispersal of Altolaguirre, Y., Bruch, A., Gibert, L., 2020. A long early Pleistocene pollen record Homo. In: Norton, C.J., Braun, D.R. (Eds.), Asian Palaeoanthropology: from Africa from Baza Basin (SE Spain): major contributions to the palaeoclimate and to China and beyond, Vertebrate Palaeobiology and Palaeoanthropology. palaeovegetation of Southern Europe. Quat. Sci. Rev. 231, 106199. Springer, pp. 31e40. https://doi.org/10.1007/978-90-481-9094-2_3. Altolaguirre, Y., Schulz, M., Gibert, L., Bruch, A.A., 2021. Mapping Early Pleistocene Belmaker, M., O'Brien, H.D., 2018. Mesowear study of ungulates from the early environments and the availability of plant food as a potential driver of early Pleistocene site of ‘Ubeidiya (Israel) and the implications for early Homo Homo presence in the Guadix-Baza Basin (Spain). J. Hum. Evol. 155, 102986. dispersal from Africa. Quat. Int. 480, 66e77.  Alvarez-Posada, s, J.M., Sala, M., Viseras, C., Pla-Pueyo, S., 2017. New mag- C., Pare Blain, H.-A., Bailon, S., Cuenca-Besco  s, G., Benna sar, M., Rofes, J., Lo pez-García, J.M., netostratigraphic evidence for the age of Acheulean tools at the archaeo- Huguet, R., Arsuaga, J.L., Bermude z del Castro, J., Carbonell, E., 2010. Climate and 16  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 environment of the earliest West European hominins inferred from amphibian Fauquette, S., Suc, J.P., Guiot, J., Diniz, F., Feddi, N., Zheng, Z., Bessais, E., Drivaliari, A., and squamate reptile assemblages: sima del Elefante Lower Red Unit, Ata- 1999. Climate and biomes in the west mediterranean area during the Pliocene. puerca, Spain. Quat. Sci. Rev. 29, 3034e3044. Palaeogeogr. Palaeoclimatol. Palaeoecol. 152, 15e36. Blain, H.-A., Bailon, S., Agusti, J., Martínez-Navarro, B., Toro, I., 2011. Palae- Fielding, A.H., Bell, J.F., 1997. A review of methods for the assessment of prediction oenvironmental and palaeoclimatic proxies to the Early Pleistocene hominids of errors in conservation presence/absence models. Environ. Conserv. 24, 38e49. Barranco Leo n D and Fuente Nueva 3 (Granada, Spain) by means of their Fortelius, M., Solounias, N., 2000. Functional characterization of ungulate molars amphibian and reptile assemblages. Quat. Int. 243, 44e53. https://doi.org/ using the abrasion-attrition wear gradient: a new method for reconstructing 10.1016/j.quaint.2010.12.031. palaeodiets. Am. Mus. Novit. 3301, 1e36. https://doi.org/10.1206/0003- Blain, H.-A., Lozano-Ferna ndez, I., Agustí, J., Bailon, S., Menendez, L., Patrocinio, M., 0082(2000)301<0001:FCOUMU>2.0.CO;2. Ros-Montoya, S., Manuel, J., Toro-Moyano, I., Martínez-Navarro, B., Sala, R., 2016. Fortelius, M., Eronen, J.T., Jernvall, J., Liu, L., Pushkina, D., Rinne, J., Tesakov, A., Refining upon the climatic background of the early Pleistocene hominid set- Vislobokovo, I., Zhang, Z., Zhou, L., 2002. Fossil mammals resolve regional tlement in western Europe: Barranco Leo  n and Fuente nueva-3 (Guadix- Baza patterns of Eurasian climate change during 20 million years. Evol. Ecol. Res. 4, Basin, SE Spain). Quat. Sci. Rev. 144, 132e144. 1005e1016. Blain, H.-A., Fagoaga, A., Ruiz-Sa nchez, F.J., García-Medrano, P., Olle , A., Jime nez-  Fortelius, M., Zliobait _ I., Kaya, F., Bibi, F., Bobe, R., Leakey, L., Werdelin, L., 2016. An e, Arenas, J.M., 2021. Coping with arid environments: a critical threshold for hu- ecometric analysis of the fossil mammal record of the Turkana basin. Phil. man expansion in Europe at the Marine Isotope Stage 12/11 transition? The case Trans. Biol. Sci. 371, 1e13. https://doi.org/10.1098/rstb.2015.0232. of the Iberian Peninsula. J. Hum. Evol. 153, 102950. raud, G., Falgue Garcia, T., Fe res, C., de Lumley, H., Perrenoud, C., Lordkidpanidze, D., Bond, W.J., 2005. Large parts of the world are brown or black: a different view on 2010. Earliest human remains in Eurasia: new 40Ar/39Ar dating of the Dmanisi the ‘Green World’ hypothesis. J. Veg. Sci. 16, 261e266. hominid-bearing levels, Georgia. Quat. Geochronol. 5, 443e451. Bradshaw, R.H.W., Hannon, G.E., Lister, A.M., 2003. A long-term perspective on Gibert, J., Arribas, A., Teran, J., Palomar, J., 1992. Contexto geolo gico del Barranco del ungulateevegetation interactions. For. Ecol. Manag. 181, 267e280. Paso (Orce, Granada). In: Gibert, J. (Ed.), Presencia Humana en el Pleistoceno  n, J.S., Ochando, J., Ferna Carrio ndez, S., Munuera, M., Amoro s, G., Blasco, R., Rosell, J., Inferior de Granada y Murcia Proyecto Orce-Cueva Victoria (1988-1992. Ayun- Finlayson, S., Giles, F., Jennings, R., Finlayson, G., Giles-Pacheco, F., Rodríguez- tamiento de Orce, pp. 203e218. Vidal, J., Finlayson, C., 2018. Last neanderthals in the warmest refugium of Gliozzi, E., Abbazzi, L., Argenti, P., Azzaroli, A., Caloi, L., Capasso Barbato, L., Di Europe: palynological data from vanguard Cave. Rev. Palaeobot. Palynol. 259, Stefano, G., Esu, D., Ficcarelli, G., Girotti, O., Kotsakis, T., Masini, F., Mazza, P., 63e80. Mezzabotta, C., Palombo, M.R., Petronio, C., Rook, L., Sala, B., Sardella, R.,  n, J.S., Rose, J., Stringer, C., 2011. Early human evolution in the western Carrio Zanalda, E., Torre, D., 1997. Biochronology of selected mammals, molluscs and Palearctic: ecological scenarios. Quat. Sci. Rev. 30, 1281e1295. ostracods from the Middle Pliocene to the late Pleistocene in Italy. The state of  n, J.S., Scott, L., deMenocal, P., 2019. Paleofloras, Paleovegetation and human Carrio the art. Rivista Italiana di Paleontologia Stratigrafica 103, 369e388. evolution. Rev. Palaeobot. Palynol. 267, 32e38. Guldemond, R., Purdon, A., van Aarde, R.J., 2017. Asystematic review of elephant Chirici, G., Barbati, A., Maselli, F., 2007. Modelling of Italian forest net primary impact across Africa. PloS One 12, e0178935. productivity by the integration of remotely sensed and GIS data. For. Ecol. Hesla, B.I., Tieszen, H.L., Boutton, T.W., 1985. Seasonal water relations of savanna Manag. 246, 285e295. shrubs and grasses in Kenya, East Africa. J. Arid Environ. 8, 15e31. Christiansen, P., 2004. Body size in proboscideans, with notes on elephant meta-  Huguet, R., Vallverdú, J., Rodríguez-Alvarez, X.P., Terradillos-Bernal, M., Bargallo , A., bolism. Zoological Journal of the Linnean Society 140, 523e549. Lombera-Hermida, A., Mene ndez, L., Modesto-Mata, M., Van der Made, J., Cohen, K.M., MacDonald, K., Joordens, J.C.A., Roebroeks, W., Gibbard, P.L., 2012. The Soto, M., Blain, H.-A., García, N., Cuenca-Besco s, G., Go mez-Merino, G., Perez- earliest occupation of north-west Europe: a coastal perspective. Quat. Int. 271, Martínez, R., Expo sito, I., Allue, E., Rofes, J., Burjachs, F., Canals, A., Benna sar, M., 70e83. Nun ~ ez-Lahuerta, C., Bermude z del Castro, J.M., Carbonell, E., 2017. Level TE9c of Dallman, P., 1998. Plant Life in the World's Mediterranean Climates. University of Sima del Elefante (Sierra de Atapuerca, Spain): a comprehensive approach. California Press, Berkeley, CA. Quat. Int. 433, 278e295. Damuth, J., 1992. Taxon-free characterization of animal communities. In: Janis, C.M., 1990. Correlation of cranial and dental variables with body size in un- Behrensmeyer, A.K., Damuth, J.D., DiMichelle, W.A., Potts, R., Sues, H.D., gulates and macropodoids. In: Damuth, J., MacFadden, B.J. (Eds.), Body Size in Wing, S.L. (Eds.), Terrestrial Ecosystems through Time: Evolutionary Paleo- Mammalian Palaeobiology e Estimation and Biological Implications. Cambridge ecology of Terrestrial Plants and Animals. University of Chicago Press, Chicago, University Press, New York, pp. 255e299. IL, pp. 183e203. Kujper, D.P.J., te Beest, M., Churski, M., Cromsigt, J.P.G.M., 2015. Bottom-up and top- Dennell, R.W., Roebroeks, W., 2005. An Asian perspective on early human dispersal down forces shaping wooded ecosystems: lessons from a cross-biome com- from Africa. Nature 438, 1099e1104. parison. In: Hanley, T.C., La Pierre, K.J. (Eds.), Trophic Ecology: Bottom-Up and Demuro, M., Arnold, L.J., Pare s, J.M., Sala, R., 2015. Extended-range luminescence Top-Down Interactions across Aquatic and Terrestrial Systems. Ecological Re- chronologies suggest potentially complex bone accumulation histories at the views. Cambridge University Press, pp. 107e133. Early-to-Middle Pleistocene palaeontological site of Hue scar-1 (Guadix-Baza Lawing, A.M., Eronen, J.T., Blois, J.L., Graham, C.H., Polly, P.D., 2017. Community basin, Spain). Quaternary International 389, 191e212. functional trait composition at the continental scale: the effects of non- Dennell, R.W., 2010. The colonization of “savannahstan”: issues of timing(s) and ecological processes. Ecography 40, 651e663. patterns of dispersal across Asia in the late Pliocene and early Pleistocene. In: Levin, N.E., 2015. Environment and climate of early human evolution. Annual Re- Norton, C.J., Braun, D.R. (Eds.), Asian Palaeoanthropology: from Africa to China view of Earth and Planetary Sciences 43, 405e429. and beyond, Vertebrate Palaeobiology and Palaeoanthropology. Springer, Liu, L., Puolama €ki, K., Eronen, J.T., Mirzaie Ataabadi, M., Hernesniemi, E., pp. 7e30. https://doi.org/10.1007/978-90-481-9094-2_2. Fortelius, M., 2012. Dental functional traits of mammals resolve productivity in Deves, M., Sturdy, D., Godet, N., King, G.C.P., Bailey, G.N., 2014. Hominin reactions to terrestrial ecosystems past and present. Proc. R. Soc. Lond. B Biol. Sci. 279, herbivore distribution in the lower Palaeolithic of the southern Levant. Quat. 2793e2799. Sci. Rev. 96, 140e160. Martínez, K., Garcia, J., Carbonell, E., Agustí, J., Bahain, J.-J., Blain, H.-A., Burjachs, F., Dolukhanov, P.M., 2000. Microenvironments and the initial hominid settlement in Caceres, I., Duval, M., Falgue rez, C., Gomez, M., Huguet, R., 2010. A new lower Western Asia. In: Lordkipanidze, D., Bar-Yosef, O., Otte, M. (Eds.), Early Humans Pleistocene archeological site in Europe (Vallparadís, barcelona, Spain). Pro- at the Gates of Europe. Lie ge: Etudes et Recherches Arche ologiques de l'Uni- ceedings of the National Academy of Sciences USA 107, 5762e5767. versite  de Liege, pp. 93e98. Martínez-Navarro, B., 2004. Hippos, pigs, bovids, saber-toothed tigers, monkeys, Duval, M., 2008. Evaluation du potentiel de la me thode de la datation par and hominids: dispersals through the Levantine Corridor during Late Pliocene Resonance de Spin Electronique (ESR) applique e aux gisements du Ple istocene and Early Pleistocene times. In: Goren-Inbar, N., Speth, J.D. (Eds.), Human inferieur: e tude des gisements d'Orce (bassin de Guadix-Baza, Espagne) et Palaeoecology in the Levantine Corridor. Oxbow Books, Oxford, pp. 37e52. contribution  a la connaissance des premiers peuplements de l'Europe. PhD. Messager, E., Lordkidpanidze, D., Kvavadze, E., Ferring, C.R., Voinchet, P., 2010. Dissertation. Museum national d'Histoire naturelle de Paris. Palaeoenvironmental reconstruction of Dmanisi site (Georgia) based on Duval, M., Toro-Moyano, I., Falgue res, C., Mestour, B., Perrenoud, C., Patrocinio palaeobotanical data. Quat. Int. 223e224, 20e27. Espigares, M., Ros, S., 2010. Estudio litoestratigr afico del yacimiento Nogueira, C., Bugalho, M.N., Pereira, J.S., Caldeira, M.C., 2017. Extended autumn arqueolo gico de Fuente Nueva 3 (Orce, cuenca de Guadix-Baza, Espan ~ a). In: drought, but not nitrogen deposition, affects the diversity and productivity of a Toro-Moyano, I., Martínez-Navarro, B., Agustí, J. (Eds.), Ocupaciones humanas Mediterranean grassland. Environ. Exp. Bot. 138, 99e108. en el Pleistoceno inferior y medio de la Cuenca de Guadix-Baza. Junta de Ochando, J., Carrio  n, J.S., Blasco, R., Ferna ndez, S., Amoro  s, G., Munuera, M., Andalucía, Consejería de Cultura, pp. 57e76. San~ udo, P., Ferna ndez Peris, J., 2019. Silvicolous neanderthals in the far west: Duval, M., Falgue res, C., Bahain, J.J., Grün, R., Shao, Q., Aubert, M., Toro-Moyano, I., the mid-pleistocene palaeoecological sequence of bolomor Cave (Valencia, 2012. On the limits of using combined U-series/ESR method to date fossil teeth Spain). Quat. Sci. Rev. 217, 247e267. from two Early Pleistocene archaeological sites of the Orce area (Guadix-Baza  Oksanen, O., Zliobait _ I., Saarinen, J., Lawing, A.M., Fortelius, M., 2019. e, basin, Spain). Quat. Res. 77, 482e491. A Humboldtian approach to life and climate of the geological past: estimating Espigares, 2019. The earliest cut marks of Europe: a discussion on hominin sub- palaeotemperature from dental traits of mammalian communities. J. Biogeogr. sistence patterns in the Orce sites (Baza basin, SE Spain). Sci. Rep. 9, 15408. 1e17. https://doi.org/10.1111/jbi.13586, 2019. https://doi.org/10.1038/s41598-019-51957-5. NPP multi-biome: global primary production data initiative products, R2. Data set. Farjon, A., Horne, D., Parfitt, S.A., Buckland, P.I., Lewis, M.D., 2020. Early Pleistocene In: Olson, R.J., Scurlock, J.M.O., Prince, S.D., Zheng, D.L., Johnson, K.R. (Eds.), conifer macrofossils from Happisburgh, Norfolk, UK, and their environmental 2013. from the Oak Ridge National Laboratory Distributed Active Archive Cen- implications for early hominin occupation. Quat. Sci. Rev. 232, 106115. https:// ter, Oak Ridge, Tennessee, USA. https://doi.org/10.3334/ORNLDAAC/617. Avail- doi.org/10.1016/j.quascirev.2019.106115. able on-line. http://daac.ornl.gov. 17  J. Saarinen, O. Oksanen, I. Zliobait e_ et al. Quaternary Science Reviews 268 (2021) 107132 Oms, O., Anado  n, P., Agustí, J., Julìa, R., 2011. Geology and chronology of the con- Saarinen, J., Eronen, J., Fortelius, M., Seppa€, H., Lister, A.M., 2016. Patterns of diet and tinental Pleistocene archeological and palaeontological sites of the Orce area body mass of large ungulates from the Pleistocene of Western Europe, and their (Baza basin, Spain). Quat. Int. 243, 33e43. relation to vegetation. Palaeontol. Electron. 19, 1e58, 3.32A. palaeo-electronica. Palmqvist, P., 2002. On the community structure of the large mammals assemblage org/content/2016/1567-pleistocene-mammal-ecometrics. from Dmanisi (East Georgia, Caucasus). In: De Renzi, M., Alonso, M., nchez-Bandera, C., Oms, O., Blain, H.-A., Lozano-Ferna Sa ndez, I., Bispal-Chinesta, J.F., Belinchon, M., Pen ~ alver, E., Montoya, P., M arquez-Aliaga, A. (Eds.), Current Agustí, J., Saarinen, J., Fortelius, M., Titton, S., Serrano-Ramos, A., Luzo n, C., Topics on Taphonomy and Fossilization. Ajuntament de Valencia, pp. 151e160. Solano-García, J., Barsky, D., Jime nez-Arenas, J.M., 2020. New stratigraphically Palmqvist, P., Groo €cke, D.R., Arribas, A., Farin ~ a, R.A., 2003. Palaeoecological recon- constrained palaeoenvironmental reconstructions for the first human settle- struction of a Early Pleistocene large mammal community using biogeochem- ment in western Europe: the early Pleistocene herpetofaunal assemblages from ical (d13C, d15N, d18O Sr:Zn) and ecomorphological approaches. Palaeobiology Barranco Leo  n and Fuente Nueva 3 (Granada, SE Spain). Quat. Sci. Rev. 243, 29, 205e229. 106466. Parfitt, S.A., Ashton, N., Lewis, S.G., Abel, R.L., Coope, G.R., Field, M.H., Gale, R., Scott, K.M., 1990. Postcranial dimensions of ungulates as predictors of body mass. Hoare, P.G., Larkin, N.R., Lewis, M.D., Karloukovski, V., Maher, B.A., Peglar, S.M., In: Damuth, J., MacFadden, B.J. (Eds.), Body Size in Mammalian Palaeobiology e Preece, R.C., Whittaker, J.E., Stringer, C.B., 2010. Early Pleistocene human Estimation and Biological Implications. Cambridge University Press, New York, occupation at the edge of the boreal zone in northwest Europe. Nature 466, pp. 301e335. 229e233. Scott, G.R., Gibert, L., 2009. The oldest hand-axes in Europe. Nature 461, 82e85. Patterson, D.B., Braun, D.R., Behrensmeyer, A.K., Lehmann, S.B., Merrit, S.R., Suc, J.-P., Popescu, S.-M., 2005. Pollen records and climatic cycles in the North Reeves, J.S., Wood, B.A., Bobe, R., 2017b. Landscape scale heterogeneity in the Mediterranean region since 2.7 Ma. In: Head, M.J., Gibbard, P.L. (Eds.), Ear- East Turkana ecosystem during the okote member. J. Hum. Evol 112, 148e161, lyeMiddle Pleistocene Transitions: the LandeOcean Evidence, vol. 247. 1.56e1.38 Ma. Geological Society, London, Special Publications, pp. 47e158. Patterson, D.B., Braun, D.R., Behrensmeyer, A.K., Merrit, S., Zliobait _ I., Reeves, J.S., e, Teague, R.L., Potts, R., 2007. Faunal and behavioral context of the earliest Pleistocene Wood, B.A., Fortelius, M., Bobe, R., 2017a. Ecosystem evolution and hominin Homo in East Asia. Am. J. Phys. Anthropol. 44, 230. palaeobiology at East Turkana, northern Kenya between 2.0 and 1.4 Ma. Titton, S., Barsky, D., Bargallo, A., Serrano-Ramos, A., Verge s, J.M., Toro-Moyano, I., Palaeogeogr. Palaeoclimatol. Palaeoecol. 481, 1e13. Sala-Ramos, R., García Solano, J., Jime nez-Arenas, J.M., 2020. Subspheroids in Peterson, E.C., Shen, C., Chen, C., Chen, W., Tank, Y., 2003. Taphonomy of an early the lithic assemblage of Barranco Leo n (Spain): Recognizing the late Oldowan in Pleistocene archaeofauna from xiaochangliang, Nihewan basin, north China. In: Europe. PLoS ONE 15, 1e41. https://doi.org/10.1371/journal.pone.0228290 Current Research in Chinese Pleistocene Archaeology, vol. 1179. Bar Interna- e0228290. tional Series, Oxford, pp. 83e98. Titton, S., Oms, O., Barsky, D., Bargallo , A., Serrano-Ramos, A., García-Solano, J., ~ ero, P., Agustí, J., Oms, O., Blain, H.-A., Laplana, C., Ros-Montoya, S., Martínez- Pin nchez-Bandera, C., Yravedra, J., Blain, H.-A., Toro-Moyano, I., Jime Sa nez- Navarro, B., 2017. Rodents from Baza-1 (Guadix-Baza Basin, southeast Spain): Arenas, J.M., Sala-Ramos, R., 2021. Oldowan stone knapping on a raw material filling the gap of the early Pliocene succession in the Betics. Journal of Verte- reservoir deposit 1.4 million years ago at Barranco Leo n (Orce, Spain). Archae- brate Paleontology 37, e1338294. https://doi.org/10.1080/ ological and Anthropological Sciences 13 (108). https://doi.org/10.1007/s12520- 02724634.2017.1338294. 021-01353-w. Quezel, P., Medail, F., 2003. Que faut-il entendre par "fore ^ts mediterrane ennes"? Toro-Moyano, I., Martínez-Navarro, B., Agustí, J., Souday, C., Bermúdez de Fore^t mediterrane enne XXIV, 11e31. Castro, J.M., Martino n-Torres, M., Palmqvist, P., 2013. The oldest human fossil in Rodríquez, J., Mateos, A., Martín-Gonza les, J.A., Rodríguez-Go mez, G., 2015. How Europe, from Orce (Spain). J. Hum. Evol. 65, 1e9. rare was human presence in Europe during the Early Pleistocene? Quat. Int.  Zliobait _ I., Rinne, J., Toth, A., Mechenich, M., Liu, L.P., Behrensmeyer, A.K., e, 389, 119e130. Fortelius, M., 2016. Herbivore teeth predict climatic limits in Kenyan ecosys- Saarinen, J., 2014. Ecometrics of Large Herbivorous Land Mammals in Relation to tems. Proc. Natl. Acad. Sci. U.S.A. 113, 12751e12756. Climatic and Environmental Changes during the Pleistocene. PhD thesis.  Zliobait _ I., Tang, H., Saarinen, J., Fortelius, M., Rinne, J., Rannikko, J., 2018. Dental e, Department of Geosciences and Geography, University of Helsinki, ISBN 978- ecometrics of tropical Africa: linking vegetation types and communities of large 952-10-9468-2. Retrieved from. http://urn.fi/URN. plant-eating mammals. Evol. Ecol. Res. 19, 127e147. 18