Available online at www.sciencedirect.com

Journal of Human Evolution 54 (2008) 43e77

Theropithecus and ‘Out of Africa’ dispersal in the Plio-Pleistocene John K. Hughes a,1, Sarah Elton b,*, Hannah J. O’Regan c a

Bristol Research Initiative for the Dynamic Global Environment, BRIDGE, School of Geographical Sciences, University of Bristol, University Road, Bristol, BS8 1SS, UK b Functional Morphology and Evolution Unit, Hull York Medical School, University of Hull, Cottingham Road, Hull, HU6 7RX, UK c Research Centre in Evolutionary Anthropology and Palaeoecology, School of Biological and Earth Sciences, Liverpool John Moores University, Liverpool, L3 3AF, UK Received 29 April 2006; accepted 25 May 2007

Abstract Theropithecus oswaldi was one of the most widely distributed Plio-Pleistocene primates, found in southern, East, and North Africa, as well as in Spain, India, and possibly Italy. Such a large geographic range for a single primate species is highly unusual. Here, the nature and timing of its dispersal is examined using the Stepping Out cellular automata model. A hypothetical dispersal of T. darti is also modelled to assess whether the late Pliocene might have been a more favorable period for Afro-Eurasian dispersal than the early Pleistocene. Stepping Out draws on climatic and biome reconstruction to provide the paleovegetative and climatic background necessary for the simulations, and model parameters for T. oswaldi and T. darti were set a priori on the basis of their fossil records and paleobiologies. The simulations indicate that T. darti could have readily left Africa in the Pliocene, and that it swiftly reaches Asia. A European T. darti colonization was less certain and less rapid. The simulated T. oswaldi dispersal out of Africa was slower, but nonetheless T. oswaldi arrived at Mirzapur within the time period indicated by the fossil record. Using the a priori parameters, T. oswaldi did not arrive at the European sites of Cueva Victoria and Pirro Nord. It cannot be discounted, therefore, that some of the European fossils are a result of an earlier T. darti dispersal. The simulations also showed that in order for Theropithecus to reach Europe, it needed to be tolerant of a relatively wide range of habitats. In addition, our finding that Asian colonization was more rapid and more probable parallels the information from the hominin fossil record, in which the fossils from Asia predate those from Europe by several hundred thousand years. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Homo; Hominin; Biogeography; Modelling; Cercopithecid

Introduction Theropithecus oswaldi was one of the most abundant and widely distributed monkey species in Plio-Pleistocene Africa, identified from localities as far removed as Ternifine in Algeria and Hopefield in South Africa (Delson et al., 1993). Cercopithecid fossils argued to be best attributed to T. oswaldi or T. cf. oswaldi have also been recovered outside Africa, from India (Gupta and Sahni, 1981; Delson, 1993; Pickford, 1993), Spain * Corresponding author. E-mail addresses: [email protected] (J.K. Hughes), sarah. [email protected] (S. Elton), H.J.O’[email protected] (H.J. O’Regan). 1 Present address: Met Office, FitzRoy Road, Exeter, Devon, EX1 3PB, United Kingdom. 0047-2484/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.jhevol.2007.06.004

(Gibert et al., 1995; Martinez-Navarro et al., 2005), and Italy (Rook et al., 2004) (Fig. 1). In addition, a specimen assigned to Theropithecus sp. has been identified from the Middle East (Belmaker, 2002). Although T. oswaldi is much less abundant in Eurasia than in Africa, such a large area of colonization is highly unusual for a single primate species. Recent descriptions of European fossil material (Rook et al., 2004; MartinezNavarro et al., 2005) have highlighted the presence of Theropithecus in Eurasia, prompting questions about how and why it was so widespread. Here, we explore the nature and timing of AfroEurasian dispersal of Theropithecus in the Plio-Pleistocene using the computer modelling program Stepping Out. From the late Miocene onwards, several cercopithecid taxa dispersed from Africa to Eurasia. Colobines probably left

44

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

Fig. 1. Map of the localities discussed in the text. Hadar and Makapansgat and the Turkana Basin and Swartkrans were calibration localities for T. darti and T. oswaldi, respectively. The North African and Eurasian sites were the focus of the dispersal model.

Africa in the late Miocene, with the Eurasian colobine Mesopithecus pentelicus appearing in the fossil record around 8.5 Ma (Delson, 1994; Delson et al., 2000). Colobines speciated both within Asia and Africa, and it has been suggested that some African colobine taxa originated from a Eurasian ancestor, ‘returning’ to Africa in the late Miocene (Jablonski, 1998). Macaca is likely to have moved into Asia from a North African point of origin during the terminal Miocene or very early Pliocene (Jablonski, 2002). Once in Asia, Macaca, like the colobines, underwent an extensive adaptive radiation that is still evident today, and probably also gave rise to the extinct Eurasian genera Procynocephalus and Paradolichopithecus (Jablonski, 2002). By the late Pliocene and the Pleistocene, however, the major cercopithecid dispersals into Asia had declined. Other than Theropithecus, the only African monkey to disperse out of continental Africa during the Plio-Pleistocene was Papio, spreading across the Bab-El-Mandeb straits to colonize a restricted area in Arabia (Wildman et al., 2004; Winney et al., 2004). The pattern of Theropithecus and Papio dispersals out of Africa contrasts with those of the macaques and colobines, as neither Theropithecus nor Papio then appeared to speciate within Eurasia. The other large-bodied primate to disperse out of Africa during the Plio-Pleistocene was Homo. Inevitably, the pattern and timing of early Homo dispersals within and out of Africa is of immense interest to paleoanthropologists (Anto´n et al., 2002; Mithen and Reed, 2002) and vital to the understanding of human evolutionary history. The overall pattern of other mammalian dispersals around this time may also have a bearing on our understanding of early human dispersal. Alongside primates, the late Pliocene saw movement of several mammalian groups both into and out of Africa. Immigrants included the genus Equus, which is first found in East Africa at 2.3 Ma (Bobe´ and Behrensmeyer, 2004), the racoon dog, genus Nyctereutes, first found at 2.5 Ma (Geraads, 1997), and several members of the Caprinae (wild sheep, goats, and relatives) that arrived in Africa at 3.2 Ma and again between 2.7e2.5 Ma (Vrba, 1995). In the same period, six bovid taxa of African origin, including Hippotragus, appear in the Siwaliks (Vrba, 1995). In addition to dispersal from and to Africa, there was

also movement of Asian bovids into Europe, following a freshening of the Black Sea at the end of the Pliocene (Spassov, 2003). In this paper we focus on Theropithecus, not simply because of its unusual colonization range, but also because a comparison of its movement patterns with those of Homo may give useful insights into the conditions that influenced large-bodied primate dispersals in the Pliocene and Pleistocene. One obvious question that arises, for example, is why Theropithecus experienced such a severe range contraction when other primates, including Homo and Macaca, did not. Theropithecus has been used to contextualize human evolutionary history in a number of ways (Elton, 2006), and it has been suggested that African Plio-Pleistocene hominins and theropiths may have been commensal (Benefit, 1999a). Even if they were not truly commensal, theropiths and hominins both underwent adaptive radiations in the Plio-Pleistocene, are found in many of the same deposits (Foley, 1993), and interacted at least via predation (Shipman et al., 1981), demonstrating that they were sympatric and subject to some of the same environmental pressures (Elton, 2006). Although aspects of Homo and Theropithecus biology and behavior are undeniably dissimilar, they have several convergent features (Foley, 1993; Elton, 2000, 2006). The most pertinent to this study are similar body mass and terrestriality, as these may influence habitat tolerances, ranging behavior, and predation pressure. Due to the complex sociality that is a defining feature of haplorhines, large-bodied primates might also be expected to behave in more similar ways to each other in response to external pressures than to other mammals in their community, including those that are dispersing in the same time period. The Stepping Out model Computer modeling is an important tool, currently underused in paleoanthropology. Its strength lies in allowing us to question and evaluate received wisdom on a variety of topics by forcing us to address the underlying assumptions and making them explicit in a mathematical model. Here, we have used the Stepping Out computer model, which was originally created to examine Homo erectus dispersals out of Africa (Mithen and Reed, 2002), to examine two separate scenarios of Theropithecus movement, the first a T. darti dispersal in the middle/ late Pliocene, and the second a movement of T. oswaldi out of Africa in the Plio-Pleistocene. Stepping Out models the probability of arrival of a particular taxon at a given point in time based on colonization rate and the taxon’s ecological requirements, both of which are defined a priori. It adopts the modelling paradigm of cellular automata (CA; Toffoli and Margolus, 1987) to simulate the dispersal of fauna and to estimate the probability of arrival at a given point (in this case, localities with T. oswaldi or T. cf. oswaldi fossils). The region in which dispersal occurs is divided into grid boxes, or cells, with each cell either occupied or unoccupied. Previously unoccupied cells adjacent to occupied cells can become occupied, simulating colonization. Such colonization occurs probabilistically in Stepping Out,

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

depending on the predetermined colonization parameters (the Pcr value). An occupied cell may also become unoccupied, according to a pre-specified probability of local extinction, Pext. The colonization (Pcr) and extinction (Pext) parameters are set by the user, with Pcr having a single, global value that is corrected for orography, and Pext incorporating patterns of different land surface types, continental configuration, and coastal proximity. The state of the cells is updated with discrete time stepping, according to a pre-specified continental-scale environmental model, the details of which are described under Materials and Methods (see also Hughes et al., 2007). Like any process-based modelling approach that attempts to consider the complexities of real-world heterogeneity, Stepping Out has limitations, some of which are specifically associated with the use of CA. One such drawback is the computational expense involved in testing parameter sensitivity, as it is not possible to perform mathematical analysis of the heuristic properties of cellular automata models. Since CA models are probabilistic, they require ensembles of simulations to be completed and the results amalgamated (Sato and Iwasa, 2000: 343). Local population density is implied in the model by using the probability of extinction in any given cell as a proxy for how ‘occupied’ that cell is likely to be, or the expectation of finding the cell occupied. This then influences the likelihood of the adjacent cell being occupied, and so on. Colonization rate is determined by using the first appearance date (FAD) of the species alongside its first appearance in another region (outside the target area for modelling) to calibrate movement speed. Dispersal rates in Stepping Out are directly constrained to fit species occurrences at specific localities, but clearly the dates of such localities may not be representative of the first appearance of a particular taxon in that region. Adding demographic data to the model may improve its sensitivity to uncertainties over true first appearance dates, but deducing dispersal rates in this way has other limitations (e.g., Anto´n et al., 2002), not least because demographic data are hard to characterize for extinct species. Thus, in Stepping Out, demographic factors for the species being modelled are not taken into consideration, and the impact of uncertainties in calibrating with dated localities is quantified through a suite of sensitivity experiments, detailed below. Similarly, as Stepping Out is designed to examine the dispersal of one species, it does not take into account ecological interactions with other animals. Estimating such interactions is problematic, especially when considering a number of extinct taxa, whose behavior and ecology may vary over time, as well as from place to place. Speciation events and extinctions are also not included in the model. We are thus considering only a sub-set of environmental factors to ask whether we can gain insight into the continentalscale movements of a species, in this case T. oswaldi, as well as a separate set of simulations considering T. darti movement. No single model will account for all factors implicated in dispersal, and all models, however sophisticated, have limitations. Applying sensitivity analyses, reporting how the model results change when the parameters are altered, and examining the range of given colonization dates for a specific

45

locality facilitate independent assessment of the reliability of the simulations. Stepping Out is therefore a valid tool for investigating the dispersal of T. oswaldi in the Plio-Pleistocene, and allows important questions about Theropithecus movements out of Africa to be addressed in ways that would not be possible by examining fossil evidence alone. Theropithecus chronology, biogeography, and taxonomy Clearly the taxonomy and chronology of fossils and fossil sites are critical to models of species distributions and possible dispersal patterns. Molecular data indicate that around 5 Ma Papio, Theropithecus, and Lophocebus speciated rapidly (Harris, 2000), and although much work on the evolutionary relationships of fossil papionins remains to be done, it is possible that Parapapio was ancestral to them (Szalay and Delson, 1979). The oldest known Theropithecus specimen, dated to around 3.7e3.5 Ma, is a M2 recovered from the Kalochoro Member of the Nachukui Formation at Lothagam (Leakey et al., 1996; Delson et al., 2000). Theropithecus is subsequently well-represented in Omo and Turkana Basin deposits throughout the Pliocene and early to middle Pleistocene (Leakey, 1993), and the region has the greatest Theropithecus species diversity of anywhere in Africa, containing five of the six commonly-recognized fossil species. The genus Theropithecus can be divided into two distinct lineages, the subgenera T. (Omopithecus) and T. (Theropithecus) (Delson, 1993; Jablonski, 2002). T. (Omopithecus) includes T. brumpti, T. baringensis, and T. quadratirostris (Jablonski, 2002). The latter two species have also been included in Papio (Leakey, 1969; Iwamoto, 1982; Delson and Dean, 1993), although there is now tentative acceptance that T. baringensis is better placed in Theropithecus (Delson and Dean, 1993; Delson et al., 2000). We follow Eck and Jablonski (1984) in including T. quadratirostris in Theropithecus partly because of its distinctive postorbital constriction (Eck and Jablonski, 1984), as the reasoning for labelling small brain size symplesiomorphous for Papio and Theropithecus (Delson and Dean, 1993) has not been made explicit. T. baringensis has been identified from Angola in southern Africa as well as from East Africa (Jablonski, 1994, but see Delson, 1984, for an alternative perspective). Nonetheless, the fossil record of T. (Omopithecus) is largely confined to East Africa, specifically the Turkana Basin and Chemeron (Leakey, 1993), and does not yet show evidence for dispersal out of southern and eastern Africa. This contrasts with the other Theropithecus subgenus, T. (Theropithecus), which is much more widespread, and thus is the focus of this Stepping Out model. Although T. (Theropithecus) includes the highly geographically-restricted extant species, T. gelada, it also contains the fossil species T. darti (recovered from East and southern Africa) and T. oswaldi (identified from North, East, and southern Africa and Eurasia) (Jablonski, 2002). In addition, we include the North African species T. atlanticus in T. (Theropithecus), due to its probable close relationship with T. darti and T. oswaldi (Alemseged and Geraads, 1998). Various subspecies are recognized within T. oswaldi. Most workers agree that T. oswaldi sensu stricto can be subdivided

46

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

into at least three subspecies: T. o. oswaldi, represented by the smaller, earlier material from Koobi Fora, the Shungura Formation, Kanjera, Olduvai Bed I plus lower Bed II, and Swartkrans; T. o. leakeyi from Olorgesailie, Ternifine, Thomas Quarries, Olduvai upper Bed II, Beds III, IV and the Masek Beds; and T. o. delsoni, from Mirzapur (Delson, 1993; Leakey, 1993; Jablonski, 2002). However, it is less clear whether the older and more primitive T. darti deserves full species status (as argued by Eck, 1993) or whether it is a chronosubspecies, T. o. darti, within T. oswaldi (Leakey, 1993; Frost and Delson, 2002). We concur with Eck (1993) that features such as overall smaller size, the complex curvature of the muzzle dorsum, and the narrow, oval piriform aperture distinguish T. darti from T. oswaldi at the species level, and also note that there are a number of metric and non-metric differences between T. darti and T. oswaldi in the postcranium (Krentz, 1993; Elton et al., 2003). In line with Eck (1993), we also treat the Hadar and Makapansgat T. darti samples as conspecific (but see Frost and Delson, 2002, for an alternative view and a discussion of this topic). The problems associated with demarcating species within the T. oswaldi lineage are further emphasized by the lack of agreement over the status of T. atlanticus, a relatively smallbodied species from North Africa, dated to 2.5 Ma (Alemseged and Geraads, 1998). It bears the dental and mandibular morphology characteristic of Theropithecus, with a putative ancestor in T. darti (Alemseged and Geraads, 1998). Some workers prefer not to view T. atlanticus as a separate species, indicating that it fits within T. oswaldi sensu lato alongside T. o. darti and the other recognized T. oswaldi subspecies (Frost and Delson, 2002). We argue that the T. atlanticus material described by Alemseged and Geraads (1998) differs sufficiently from T. oswaldi sensu stricto in size and aspects of craniodental morphology to be classed as a separate species. However, we note that in several features, such as size, depth of the mandibular corpus fossa, the inclination of the medial malleolus of the tibia, and male P3 versus M3 length, there are similarities with T. darti (Alemseged and Geraads, 1998). Thus, although we accept Alemseged and Geraads’ (1998) diagnosis for the time being, it is possible that the material from Ahl al Oughlam represents late T. darti. Probably the most important taxonomic assumption for the Stepping Out model is that the theropiths outside Africa represent T. oswaldi sensu stricto. Delson (1993) demonstrated that the Mirzapur fossil from India was best included in T. o. delsoni. Since then, more Eurasian Theropithecus material has been recognized from Spain (Gibert et al., 1995; MartinezNavarro et al., 2005), the Middle East (Belmaker, 2002), and Italy (Rook et al., 2004). Two Theropithecus specimens have been identified from Cueva Victoria, Murcia, Spain. The first, a RM1 or RM2, bears the hallmark Theropithecus molar morphology, and lies within the size range observed in Olorgesailie T. o. leakeyi (Gibert et al., 1995). The other Cueva Victoria specimen, CV-0, a phalanx previously attributed to Homo, closely resembles intermediate pedal phalanges of T. o. leakeyi from Olorgesailie (Martinez-Navarro et al., 2005). The manual and pedal

phalanges of Theropithecus are short and broad, quite distinct from those of other large cercopithecids (Jablonski, 1986; Strasser, 1988), so the designation of CV-0 as Theropithecus appears accurate. The molar and CV-0 have both been referred to T. oswaldi (Gibert et al., 1995; Martinez-Navarro et al., 2005), although others are more cautious about the Cueva Victoria material (Frost and Delson, 2002). Like the molar, CV-0 falls within the size range of later African T. oswaldi. It is therefore plausible that the two Spanish specimens came from the same population, and can be assigned with reasonable certainty to T. oswaldi. The most appropriate subspecies for the Murcia material is unresolved. Jablonski (2002) referred it to T. o. oswaldi, surprising given its clearer size affinities with T. o. leakeyi, whereas Delson et al. (2000) tentatively assigned it, like the Mirzapur material, primarily on the basis of geography, to T. o. delsoni. There is more difficulty in assessing the taxonomic affinities of the other recently published Eurasian Theropithecus specimens. Theropithecus from ‘Ubeidiya is identified only from a juvenile calcaneus (Belmaker, 2002), and a full description with justification and adequate comparative sample is yet to be published. We therefore reserve judgement on the taxonomic status of this specimen pending such a description, and do not use ‘Ubeidiya as a known Theropithecus locality in the model. Theropithecus from Pirro Nord is currently represented by three cervical vertebrae (C3, C5, and C6), elements that are poorly preserved in the cercopithecid fossil record (Rook et al., 2004) and also relatively poorly studied in modern papionins. Due to this, they were attributed to Theropithecus sp. indet., although the material is within the size range expected for T. oswaldi (Rook et al., 2004). When compared to Presbytis, Colobus, Macaca, Papio, Mandrillus, and the extant gelada, Rook et al. (2004) argue that the Pirro Nord fossil vertebrae, although larger, are most similar to T. gelada cervical vertebrae, particularly in body shape, lateral process morphology, and orientation of the superior and inferior articular processes. Very little is known about the functional morphology of the cervical region in cercopithecids, but as the foraging strategy of Theropithecus is distinct enough to influence other parts of the postcranium (Jablonski, 1986; Krentz, 1993; Elton et al., 2003), it is possible that this is also reflected in the neck. However, a recent morphological study has indicated that the vertebrae, although cercopithecid, may not be Theropithecus (Patel et al., 2007), despite Rook et al. (2004) clearly demonstrating that the material has few affinities with Macaca, a papionin genus found in many European Plio-Pleistocene deposits. Based on size, it has been suggested that the fossils may belong to Paradolichopithecus (Patel et al., 2007). Nonetheless, until further taxonomic studies of the Pirro Nord material are made, it should not be ruled out as Theropithecus. As with the Murcia material, the large size of the vertebrae seems consistent with their referral to T. cf. oswaldi, although this cannot be ascertained on current evidence. Much of the African Theropithecus material, particularly in East Africa, has been dated with confidence, but this is not the case for the Eurasian material. Delson (1993) indicated that

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

the date of the Mirzapur Theropithecus was controversial, giving a conservative estimate of between 1.0 and 0.1 Ma (Delson, 1993; Delson et al., 1993). Gupta and Sahni (1981), who first described the material, stated that it came from the Lower Boulder Conglomerate, above the Pinjor Formation, and suggested it was middle Pleistocene in age. Unfortunately, none of the subsequent dating studies expanded as far northwards as the coordinates of the recovery site provided by Gupta and Sahni (1981), complicating age estimates. A positive paleomagnetic signal at the base of the Lower Boulder Conglomerate has been correlated with the Olduvai event, dated to between 1.79 and 1.95 Ma (Azzaroli and Napoleone, 1982). Recent reconsideration of faunal, geological, and magnetostratigraphic data from the area suggests that the contact between the Pinjor Formation and the Lower Conglomerate varies from 1.72 to 0.6 Ma in different parts of the region (Nanda, 2002). At the Patiali Rao section, some 10 km SW of Mirzapur and the closest studied section to the probable Theropithecus site, the change between the faunas has been dated at 0.63 Ma (Nanda, 2002). Despite this, Rook et al. (2004) suggested a date of 1.0 Ma for Mirzapur, based on the earlier paleomagnetic study by Azzaroli and Napoleone (1982). It is clear that there is yet no consensus over its age, but given that the closest section is dated to the middle Pleistocene (Nanda, 2002), this currently appears to be the most likely age of the Mirzapur Theropithecus material. Cueva Victoria also has somewhat controversial dating, with published suggestions of its age ranging from early to middle Pleistocene. Gibert et al. (1995) suggest that the site is early Pleistocene based on the mammalian fauna, with Martinez-Navarro et al. (2005) giving an estimate of around 1.0 Ma. Studies of the Megaloceros remains, on the other hand, have indicated a middle Pleistocene age of around 0.5 Ma for at least part of the fauna (van der Made, 2004). Mixing of fauna from different periods in the deposit cannot be ruled out, with a maximum estimated age range of 1.4e0.5 Ma, but a late early Pleistocene date of 1.4e0.9, chronologically between the French sites of Sainzelles and Le Vallonnet, seems preferred (Agusti et al., 1986). The other Eurasian locality included in the Stepping Out model, Pirro Nord, is not dated absolutely. The biochronology of the Pirro Nord assemblage, of which the Theropithecus specimens are part, indicates an age of 1.6e1.3 Ma (Rook et al., 2004). It is worth noting that ‘Ubeidiya, although not investigated here, also falls within this time range, at c.1.4 Ma (Belmaker, 2002). It is apparent that there was at least one Theropithecus dispersal out of Africa into Eurasia during the late Pliocene or early Pleistocene.1 Since the exact chronology of this is unknown, two alternative theropith models have been examined in this study. It has been suggested that paleoenvironmental 1

In this paper, the Pliocene - Pleistocene boundary is regarded as being at 1. 8 Ma, as marked in the Vrica type section of southern Italy (Pasini and Colalongo, 1997). However, it is worth noting when discussing Afro-Eurasian dispersal that some pollen specialists and researchers in Asia place this boundary earlier, at 2.5 Ma, corresponding to the Praetiglian in the northwest European pollen chronology (Zagwijn, 1998), and the beginnings of the loess accumulations in central and eastern Asia (Dodonov and Baiguzina, 1995).

47

conditions and the availability of the Bab-el-Mandeb Straits crossing in the late Pliocene may have made this time period more suitable than the early Pleistocene for hominin and other faunal movement (Turner, 1999; Dennell, 2003). The first model therefore simulates the movement of T. darti from Hadar at 3.4 Ma to test the probability of Theropithecus darti dispersal out of Africa in the Pliocene. The second models dispersal of Theropithecus oswaldi (the species to which most of the Eurasian fossil material has been assigned) from the Turkana Basin at 2.4 Ma, examining movement out of Africa in the latest Pliocene/early Pleistocene. In the T. oswaldi simulation, a Turkana Basin origin and point of dispersal is assumed partly because the earliest material attributed to T. oswaldi sensu stricto, with a FAD of 2.4 Ma, was recovered from this region (Delson et al., 1993; Leakey, 1993). T. oswaldi is also highly abundant in this area, which has evidence of long, continuous theropith occupation alongside remarkable species diversity. These factors all support the notion that there was a high probability of origin in and subsequent dispersal from the Turkana Basin. However, as the earliest fossils that can be assigned unequivocally to T. darti are found in relative abundance at Hadar (Eck, 1993), and following Anto´n et al. (2002), Hadar is used as the point of origin and dispersal in the T. darti simulation. The cercopithecid fossil record in Africa is obviously biased towards the regions that have excavated sediments of the correct ages, and this has inevitably influenced our assumptions about T. (Theropithecus) origins. Given current fossil evidence, an alternative scenario would be a southern African origin of T. darti and T. oswaldi. This is difficult to justify. Makapansgat, a major T. darti locality, is not radiometrically dated, but it is highly likely to postdate Hadar (Eck and Jablonski, 1987), and despite a large cercopithecid assemblage, Theropithecus is absent from Sterkfontein Member 4 (Kuman and Clarke, 2000; Elton, 2001). T. oswaldi has recently been identified from the StW 53 breccia at Sterkfontein, but the dating of this breccia is problematic (Kuman and Clarke, 2000). No full description of the StW 53 T. oswaldi material has been published to date, so the earliest confirmed southern African T. oswaldi sensu stricto material is still from Swartkrans Member 1, dated to 1.8e1.7 Ma (Brain, 1993), significantly postdating the T. oswaldi FAD in the Turkana Basin. In consequence, it seems most parsimonious to use existing fossil evidence as the basis for the model but, in recognition of the fact that there might have been a very different origin of Theropithecus within Africa than is suggested from examination of the fossil record, two additional T. darti simulations are also reported, one with an origin in Angola and the other starting in Morocco, as each of these regions has a PlioPleistocene cercopithecid fossil record. The general environmental and geographic contexts of the model During the Pliocene warm period three million years ago, sea level was approximately 35 meters higher than at present, with relative warming greatest in high latitudes (Dowsett et al.,

48

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

1999). Europe was more densely forested, with many Asian warm temperate tree species present in the vegetational assemblage from Reuver in Holland (Zagwijn, 1998). The Pliocene and Pleistocene are characterized by a global cooling trend and increased aridity, which is well-documented in many regions of the world, and a number of transition periods in global climate have been identified (Marlow et al., 2000). The first transition relevant to Plio-Pleistocene dispersals occurred around 2.5 Ma, and is associated with climatic cooling and the initial build up of ice sheets (Marlow et al., 2000). The boundary between the Pliocene and the Pleistocene, at c.1.8 Ma, represents a second period of transition, whilst another event, at c.1.0e0.8 Ma, is linked to the intensification of global ice volumes, with the formation of much larger ice sheets and a shift from 41 ky to 100 ky cycles in climate variability. The first transition at 2.5 Ma witnessed the first stages of fragmentation in the widespread forests of Pliocene Europe (Agusti and Anto´n, 2002). In Italy, a transition from warm to cooler faunas began at w2.6 Ma, with a slow transition from warm-humid forest animals to temperate grass and woodland faunas, interspersed with more abrupt faunal turnovers at 2.5 Ma, 1.8 Ma, and 0.9 Ma (Sardella et al., 1998). In northern China, where vegetation patterns have been inferred from micromammal assemblages, it appears that the grasslands with broadleaf and conifer forests that were present before 2.5 Ma were replaced by open grasslands that persisted well into the Pleistocene (Jin et al., 1999). The cooling event at 2.5 Ma is also seen as the beginning of the loess accumulations, indicative of aridity, in Central and eastern Asia (Dodonov and Baiguzina, 1995) and the development of widespread grasslands in these regions. In deposits at Karamaidan, Tadjikistan, from which the cercopithecid Paradolichopithecus has been recorded, the loess layers dated to between 2.5 and 2.0 Ma are associated with xerophytic pollen (Dodonov and Baiguzina, 1995). However, paleosols at the same site contain forest plant taxa, indicating cyclicity in the deposition of loess (Dodonov and Baiguzina, 1995). Indeed, it is only after w0.85 Ma that Central Asia is characterized by a growing signal of C4 grasses (Yang and Ding, 2006) related to changes in the precipitation regime linked to the larger glacial cycles, although C4 grasslands have been present in parts of Asia since the Miocene (Quade and Cerling, 1995). In Africa, the 2.5 Ma event forms the start of the general progression towards greater aridity, cooler temperatures, and increased climate variability (DeMenocal, 2004), and this trend also corresponds to a shift from closed, C3 pathway vegetation towards more open, C4 pathway grass, especially after the mid-Pleistocene transition (Cerling and Hay, 1988). A three million year record of aeolian dust north of the Sahara demonstrates an increase in dust production around 0.95 Ma, suggesting the beginning of the modern extent of the Sahara (Larrasoa~ na et al., 2003). At the same time, however, the transition periods of 2.7e2.5 Ma, 1.9e1.7 Ma, and 1.1e0.9 Ma were also periods of increased moisture in eastern Africa, demonstrated by fossil lake levels (Trauth et al., 2005). In Israel, at the boundary between Africa and Eurasia, pollen data

indicate an overall trend towards steppic vegetation punctuated by periods of tree cover during the Pleistocene (Horowitz, 1989), although these changes are not well-dated. Topography and geography will inevitably have a major impact on patterns and routes of dispersal. Two of the most significant topographical features of modern-day Africa that might have influenced Afro-Eurasian mammalian dispersals are the Nile River and the Sahara Desert. These may have been present in the Pliocene and Pleistocene, but in different forms to those observed today. The evolution of the Nile has been the subject of intense debate. It has followed its present course only since the late Pleistocene (Goudie, 2005), and prior to that it probably drained into the Red Sea (Goudie, 2005). The Sahara in North Africa was likely to have been smaller than at present (Thompson and Fleming, 1996). The shifting extent of the Sahara in the past is correlated with changes in global ice volume (Leroy and Dupont, 1994), and the size of the desert has grown over time, with a relatively abrupt increase around w0.95 Ma (Larrasoa~na et al., 2003), the time of the mid-Pleistocene transition. Between 2.8 and 2.5 Ma, tropical forest occurred sporadically above 21 N, now the heart of the modern-day desert (Leroy and Dupont, 1994). The environment of the Saharan region has also undergone great fluctuations, at least since the end of the Pleistocene, with extreme reductions of the size of the desert at 6,000 years ago, associated with the appearance of higherprecipitation biomes (Jolly et al., 1998). Holocene sites in southern Libya contain African elephant (Loxodonta africana), buffalo (Syncerus sp.), waterbuck (Kobus ellipsiprymnus), and hippopotamus (Hippopotamus amphibius), suggesting that the area contained environments suitable for a wide variety of species that today would find the desert uninhabitable (Peters and Po¨llath, 2004). The longest paleotributary of the Nile, the Wadi Howar, ran northeast from Chad for over 1,000 km to join the Nile at Debba (Sudan), and in the Holocene climatic optimum, it may have been a perennial river, possibly then dwindling to small lakes before disappearing with the return to hyperaridity 3,000 years ago (Peters and Po¨llath, 2004). Regional variations in the climate of the Sahara probably also occurred, related both to elevation (Jolly et al., 1998) and geographic location, with southern Libya being much more humid between 8,000 and 5,000 BP than it is today, in contrast to the aridity of the desert in Sudan (Peters and Po¨llath, 2004). Assuming that the early Holocene is representative of previous interglacials, the Pliocene and Pleistocene desert may well have been passable for, geologically, relatively brief periods. One drawback of the present Stepping Out model is that, as described above, it uses broad climate patterns, with the result that paleoenvironments cannot be resolved on a site by site basis. The resolution of the model also makes it difficult to include even relatively large landscape features such as the Nile. However, given the intense debate over the course of the Nile and its presence during different periods, the model may actually be more robust in using broad climate patterns than it would be if individual landscape features were erroneously included.

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

Materials and methods The Stepping Out model Stepping Out was first described in Mithen and Reed (2002), and the version used in the present study is a modification of the original model. Its spatial domain, in this case the Old World, is divided into rectilinear grid boxes, with a resolution of 1 longitude by 1 latitude, with a time-step size of 250 years. Each grid cell is defined as either occupied or unoccupied, and changes in the grid box geometry (a function of latitude) are corrected for, even though the effects of changing geometry were not significant for this study. Probability of extinction (Pext) and colonization rates (Pcr) are determined a priori through reference to evidence from the fossil record. Pcr has a single, global value for a particular species in all vegetation and habitat types but is altered by local orography. The Pcr value is unaffected for elevations below 1,000 meters, but it is set to one quarter of its original value for elevations between 1,000 and 3,000 meters. Colonization of regions higher than 3,000 meters is prohibited in the model. For each set-up of Stepping Out an ensemble of 50 simulations are completed. Each simulation produces slightly different results due to the stochastic processes inherent in Stepping Out. Thus, the average properties of the arrival dates must be calculated from the 50 simulations. Patterns of vegetation and climate variability provide the background for dispersal, and a unique feature of Stepping Out relative to other CA models is the level of detail incorporated into the representation of the environment. The version of Stepping Out used in the present study models only one taxon (for example, T. oswaldi) and includes an explicit representation of changes to vegetation patterns. This means that environmental tolerances can be represented as a single probability of local extinction, Pext, for each land surface type, increasing the transparency of results and helping to improve understanding of the model. A further advantage of using Pext values is that they can be manipulated to confer advantage (or disadvantage) as appropriate in specific regions, such as along the coast. Pext parameters for the different land types are calculated through a series of sensitivity analyses based on the inferred environmental requirements of the species being modelled. This aspect of the model represents a significant addition to the previous version of Stepping Out, in which the dispersing species was separated into different ‘types’ based on dietary and climatic adaptations, and the environmental tolerances were harder to understand (Mithen and Reed, 2002). In developing the revised form of Stepping Out, land surface types and vegetation patterns were determined using proxy global climate data and models of spatial patterns of climate and vegetation. The proxy climate data (benthic foraminifera) were used to infer the chronology and baseline climatic variability of the glacial/interglacial cycles. Benthic foraminifera data offer a quasi-continuous record of global ice volume, and are commonly regarded as a measure of global climate variability. However, they do not directly provide information about spatial patterns of climate, so General Circulation

49

Models (GCMs) based on the known laws of atmospheric and oceanic physics, are also used to simulate spatial patterns of climate. The simulated climate from the GCM is then used to drive a global vegetation model that predicts global patterns of vegetation. This process is repeated for a range of climate states covering a spectrum of possible patterns between glacial and interglacial conditions and are combined with the benthic foraminifera data to produce vegetation patterns throughout the simulations. The HadCM3 GCM (Gordon et al., 2000) is the source of the environmental simulation used here. We have used HadCM3 simulations of Pliocene climate (Haywood et al., 2002) and Pleistocene climate (Joos et al., 2004) and the associated vegetation patterns predicted using BIOME4 (Kaplan et al., 2003) to provide the environmental background for the Stepping Out dispersal simulations. BIOME4 uses climatological fields of temperature, precipitation, sunlight hours, and atmospheric carbon dioxide concentration to predict global vegetation patterns. Its ability to simulate vegetation patterns during the Plio-Pleistocene has been the subject of a number of studies (Prentice and Webb, 1998; Joussaume et al., 1999; Haywood et al., 2002), and BIOME4 is therefore well-established. In a test of the validity of the predicted vegetation patterns from HadCM3 and BIOME 4, Haywood et al. (2002) compared the model output to vegetation distributions in the PRISM2 dataset (Thompson and Fleming, 1996), and the model was shown to be in good agreement with the observations, especially in regions where PRISM2 had a high density of data sites. Figure 2aec shows example patterns of land surface types used in Stepping Out, as predicted by HadCM3 and BIOME4. The vegetation divisions used here largely correspond to the mega-Biome classifications used in BIOME4 (Kaplan et al., 2003), and the only difference is that in Stepping Out the mega-Biome classifications Tundra and dryTundra are contained in a single classification, Tundra. The land surface types used in Stepping Out are therefore: 1) Tropical forest, 2) Warm-Temperate forest, 3) Temperate forest, 4) Boreal forest, 5) Savannah and dry woodland, 6) Grassland and dry shrubland, 7) Desert, 8) Tundra, 9) Ice. At some point during the period of time examined in this study, the vegetation patterns shift from a ‘Pliocene’ configuration (Haywood et al., 2002) to ‘Pleistocene’ (Joos et al., 2004). However, there is uncertainty as to when the major shift took place in land surface properties, for example, in the aridification of the Sahara (Larrasoa~na et al., 2003). As a result, three different climate scenarios have been adopted in the model, corresponding to three possible climatic transition periods, 2.5 Ma, 1.8 Ma, and 1.0 Ma. Separate simulations can then be run for each transition period. Once the transition has been made to Pleistocene vegetation, the climatic state (as in the original study of Mithen and Reed, 2002) is characterized in Stepping Out by an integer between 0 and 10, where 0 corresponds to full glacial conditions, and 10 corresponds to full inter-glacial conditions. Unlike the original Stepping Out, Pleistocene climate variability is here constrained to a benthic foraminifera proxy of global sea level (Shackleton and Hall, 1989). This climate proxy is then scaled onto the

50

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

Fig. 2. Examples of vegetation patterns used in Stepping Out, illustrating the range of variability included in the model. These correspond to (a) mid-Pliocene, (b) interglacial and (c) glacial climates. Theropithecus was given no ability to survive in tundra, ice, water, and boreal forest and excellent survival ability in savannah and dry woodland and grassland and dry shrubland.

climate scale of Stepping Out (0e10). Figure 3aec shows the three different climate scenarios associated with this environmental model. Calibration and sensitivity analyses of dispersal rates Dispersal rate for each of the two species was calculated by initializing the populations at their respective FADs in East Africa and calibrating them with the earliest-known presence of each species in southern Africa. T. darti was initialized at Hadar at 3.4 Ma and set to arrive at Makapansgat (Member 3) for 3 Ma. T. oswaldi was started in the Turkana Basin at 2.4 Ma, arriving at Swartkrans (Member 1) at 1.75 Ma. Calibration with these dates, and assuming travel through mainly savanna-like vegetation, gives a colonization rate (Pcr) of 0.07 for T. darti and 0.06 for T. oswaldi. It is these rates, corrected for orography as necessary, that are used in the main simulations. Arrival dates in southern Africa for other Pcr values are given in Table 1. This shows that colonization rates between 0.09 and 0.05 overlap (or very nearly overlap) a 3 Ma arrival date at Makapansgat for T. darti, and that Pcr values between 0.07 and 0.05 overlap (or nearly overlap) a 1.75 Ma arrival at Swartkrans for T. oswaldi. Thus, in the main simulations, arrival dates based on differing values of Pcr are reported in addition to data generated from the preferred colonization rates. The Stepping Out simulations were tested for sensitivity to differing parameters by running several identical scenarios in which only one variable was changed, in order to understand the effects of changing that variable on the overall predictions of the model. Analyses were performed on T. darti populations (with supplementary data shown in Appendix 1), using only the 2.5 Ma global climate transition model. The implications of the sensitivity results are applicable to any other dispersing population modelled, including T. oswaldi; the raw data for

a more limited number of T. oswaldi sensitivity analyses are provided in Appendices 2ae2c. The first sensitivity analysis was designed to observe the effects of distance on the arrival dates at the Eurasian sites of interest. A T. darti population, initialized in East Africa and allowed to disperse out of the region after 3.4 Ma, was given an equally low probability of extinction (Pext ¼ 0.01) in all habitats except tundra, boreal forest, and ice. This showed that, as might be expected, the distance between a pair of sites is a strong influence on the time taken to disperse between them. The distribution of arrival dates for North Africa and Eurasia is shown in Fig. 4aee. The second analysis quantified the sensitivity of the model to the dispersal rate constant, Pcr, based on the calibration experiments. Pcr was calculated between 0.04e0.12 for all eight scenarios shown in Table 2 and for the standard vegetation tolerances of the two Theropithecus species (Table 3). The complete results of this exercise are shown in Appendix 1. To summarize, using the standard set of T. darti environmental tolerances, Table 4 shows the effect that altering Pcr values from 0.12e0.04 has on T. darti arrival dates in North Africa and Eurasia (Pcr rates of 0.10 and 0.06 are the equivalent of calibrating the arrival at Makapansgat at 3.2 Ma and 2.91 Ma, respectively). In comparison with the standard Pcr of 0.07, a Pcr value of 0.10 causes the Mirzapur arrival to occur around 600,000 years earlier, whilst a Pcr value of 0.06 delays arrival at Mirzapur by some 500,000 years. As might be expected, this analysis shows that Stepping Out requires a reliable dispersal chronology. The analysis also demonstrates that the model is sensitive to variability in Pcr values, although there is no ‘step change’ in arrival times when Pcr values between 0.10 and 0.06 are altered. With Pcr values below 0.06, the changes in arrival times become less linear, there are more fluctuations, and there is an increased standard error about the mean. The Pcr rates of 0.07 for T. darti and 0.06

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

51

Table 1 Calibration of dispersal rates. T. darti was initialized at Hadar at 3.4 Ma and calibrated with a FAD of 3 Ma at Makapansgat to give a Pcr of 0.07. T. oswaldi was initialized in the Turkana Basin at 2.4 Ma and calibrated with a FAD of 1.75 Ma at Swartkrans to give a Pcr of 0.06. These values are listed in bold, and were used in the main simulations T. darti arrival date (Ma) at Makapansgat

T. oswaldi arrival date (Ma) at Swartkrans

Pcr

Mean  SE

Range

Mean  SE

Range

0.10 0.09 0.08 0.07 0.06 0.05

3.20  0.01 3.17  0.01 3.13  0.01 3.02  0.02 2.91  0.03 2.63  0.08

3.29e3.10 3.28e3.03 3.27e3.01 3.19e2.92 3.11e2.51 3.08e1.57

e e 2.01  0.01 1.88  0.02 1.76  0.02 1.38  0.04

e e 2.20e1.96 2.14e1.78 2.11e1.39 2.01e0.26

for T. oswaldi, calculated from the fossil record and used in the main Stepping Out simulations, are the most reasonable proxy for actual movements of Theropithecus that can be constructed based on the current fossil evidence. These estimates are, of course, open to the problems relating to dating that are inherent in all paleontological studies. However, by running all scenarios for Pcr rates between 0.04 and 0.12, we have data for a variety of colonization rates, allowing changes in these parameters to be explored and new simulations generated if new evidence, either fossil or chronological, becomes available. Model parameters and sensitivity analysis: Theropithecus environmental tolerances

Fig. 3. Climate variability scenarios used in Stepping Out, with the transition from Pliocene vegetation to modern glacial-interglacial variability taking place at a) 2.5 Ma, b) 1.8 Ma, c) 1.0 Ma. Climate Step 12 ¼ Pliocene vegetation distributions; Climate Step 10 ¼ modern interglacial; Climate Step 0 ¼ glacial maximum conditions.

The range of environments tolerated by a species is an obvious factor in its ability to disperse from its region of origin and colonize new areas. Environmental tolerances are quantified in Stepping Out through a probability of extinction (Pext) value. These are determined through examination of the known paleoenvironmental contexts of the dispersing taxa, and from paleobiological inferences. The one extant member of the genus Theropithecus, T. gelada, is highly geographically restricted and associated with a montane grassland habitat (Napier and Napier, 1967). Extinct theropiths share adaptations for graminivory with the modern gelada (Jolly, 1972), but it is unlikely that they were biologically, behaviorally, and ecologically identical. T. darti is associated with relatively open, dry grassland habitats at Hadar (Eck, 1993) but more mosaic woodland habitats at Makapansgat (Cadman and Rayner, 1989; Reed, 1997; Sponheimer et al., 1999), and certain postcranial features indicate that it may have been more arboreal than T. gelada or T. oswaldi (Krentz, 1993; Elton et al., 2003). Although T. oswaldi is often considered to be indicative of highly open habitats, it is found in a range of paleoenvironments from the relatively wooded and wet Upper Burgi Member at Koobi Fora (Reed, 1997) to the much more open C4 habitat evident at Olorgesailie (Sikes et al., 1999). Ecomorphic analysis of T. oswaldi postcrania indicates that it probably used its habitat in a similarly flexible manner to that seen in modern common baboons (Elton, 2002). It is likely to have been dietarily dependent on

52

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

Fig. 4. Sensitivity analysis of arrival date distributions for T. darti at (a) Ternifine, (b) Makapansgat, (c) Cueva Victoria, (d) Mirzapur, and (e) Pirro Nord without the influence of land surface heterogeneity (Pcr ¼ 0.07).

grass (Jolly, 1972), possibly supplementing with leaves (Teaford, 1993) or even fruit at some localities (Benefit, 1999b). Thus, T. darti and T. oswaldi are associated with open or semi-open habitats but are unlikely to have been confined to the types of grassland associated with the modern gelada. Table 3 lists the ‘standard’ vegetation tolerances used for each species of Theropithecus in the main simulations. Of the nine habitat categories used in Stepping Outd1) Tropical forest, 2) Warm-Temperate forest, 3) Temperate forest, 4) Boreal forest, 5) Savannah and dry woodland, 6) Grassland and dry shrubland, 7) Desert, 8) Tundra, and 9) Icedboth Theropithecus species are given a 100% probability of extinction (a Pext value of 1.00) in ice, tundra, and boreal forest due to the temperature (beyond the adaptive limits of the majority of primates) and the paucity of grass and other suitable foodstuffs in these habitats. Based on paleoenvironmental evidence from African localities, both species are given a low probability of extinction (Pext equal to 0.01) in the savannah and dry woodland and grassland and dry shrubland categories. These

five ‘standard’ Pext values were kept constant in all simulations including the sensitivity analyses. As mentioned above, some postcranial and paleoenvironmental evidence indicates that T. darti might have had a slightly wider habitat tolerance than did T. oswaldi. In the standard scenarios it has therefore been assigned a Pext value of 0.03 in the remaining habitat categories (which are hereafter called ‘non-constant’ categoriesdtropical forest, warm-temperate forest, temperate forest, and desert). T. oswaldi has been given a slightly higher probability of extinction in these non-constant habitats, with a Pext value of 0.04. The desert category is given the same Pext value as the other ‘non optimal’ Theropithecus habitats because of the possibility that the Sahara underwent periodic ‘greening’ phases, at least in certain regions. To further assess the influence of desert conditions on dispersal, however, other Pext values for desert occupation are the subject of a set of sensitivity analyses, reported below. Sensitivity analyses were performed to assess the influence of varying Pext values on migration rates. A number of scenarios were constructed for T. darti (Table 2), each with different

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

53

Table 2 The range of scenarios describing different habitat tolerances used in the sensitivity analyses. Pext is extinction probability, where 0.01 is high probability of survival, and values above this confer progressively higher extinction risks. Boreal forest, tundra, and ice are all classed as 1.0 (no chance of survival) in every simulation (including the main results as well as the sensitivity analyses), and are thus not listed Pext Scenario 1: 2: 3: 4: 5: 6: 7: 8:

Tropical forest

Warm-Temperate forest

Temperate forest

Savanna and dry woodland

Grassland and dry shrubland

Desert

0.01 0.10 0.07 0.04 0.02 0.03 0.03 0.03

0.01 0.10 0.07 0.04 0.02 0.03 0.03 0.03

0.01 0.10 0.07 0.04 0.02 0.03 0.03 0.03

0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01

0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01

0.01 0.10 0.07 0.04 0.02 0.05 0.07 1.00

ability equal in all habitats little chance of survival in non-grassland habitats marginal survival in non-grassland habitats reasonable survival in non-grassland habitats very good survival in non-grassland habitats quite reasonable survival in desert marginal survival in desert extinction in desert

permutations of Pext in the vegetation categories. In these sensitivity analyses, the dispersal rate (Pcr) was set to the calibrated value for T. darti, 0.07. T. darti and T. oswaldi are sometimes argued to be exclusively grassland dwelling taxa, based on the ecology of the only living theropith, the gelada. In order to test the effect that restricting T. darti to grassland habitats had on dispersal, two simulations were run (Tables 2 and 5). In the first, Pext values for all vegetation types except the five constants (ice, tundra, boreal forest, savannah, and grassland) were set to 0.10 (Scenario 2: little chance of survival). In the second, Pext was 0.07 (Scenario 3: marginal survival) for all the nonconstant vegetation types. In both cases T. darti failed to arrive in Europe. In the first simulation, with the Pext value of 0.10 in the non-constant vegetation categories, the mean arrival date at Mirzapur was 0.88 Ma (range 2.56e0.17 Ma). In the second (with the Pext value of 0.07), the mean arrival date at Mirzapur was 1.29 Ma (range 2.60e0.17 Ma). Arrival at Ternifine was also affected by variations in Pext in the non-constant vegetation categories, with a 10% probability of arrival in the first simulation and a 36% probability of arrival in the second simulation. Therefore, it is unlikely that the extinct T. darti and T. oswaldi were restricted purely to grassland environments; in the model, this scenario resulted in non-arrival in Europe, which is contradicted by the fossil evidence. To test the validity of the species standard T. darti Pext value of 0.03 (good survival in non-grassland habitats), the non-constant Pext values were varied to 0.04 (Scenario 4: reasonable survival) and 0.02 (very good survival) for the forest categories and desert (Table 5). When Pext was set to 0.04, T. darti did not arrive at either Pirro Nord or Cueva Victoria. In summary, the results of the sensitivity analyses demonstrate that the lower the Pext,

the higher the probability of arrival at a given site, and the tighter the range of arrival dates (see Appendix 1 for full details). The presence and size of the Sahara are likely to have influenced the dispersal patterns of Pliocene and Pleistocene mammals. However, it is difficult to pinpoint the exact extent of the desert at different periods, and thus, the degree to which it would have acted as a barrier. In the standard analyses, the ability to survive in desert is included as one of the main parameters (with a Pext value of 0.03 for T. darti and 0.04 for T. oswaldi). To simulate the Sahara as a barrier to T. darti dispersal, three simulations were run, each with increasingly large values of Pext (0.05, 0.07, and 1.0). The effects of differing survival abilities in desert (which could be argued to be equivalent to differing intensities of the Sahara ‘barrier’) on arrival dates are given in Table 6. With desert Pext set to 0.05 (Scenario 6: quite reasonable survival), T. darti has a mean arrival date at Ternifine of 2.99 Ma (range 3.20e 2.51 Ma) and at Mirzapur of 2.11 Ma (range 2.90e1.21 Ma), both with high probabilities of arrival. At Pirro Nord, the mean arrival date is 1.14 Ma (range 1.86e0.24 Ma) and at Cueva Victoria it is 1.24 Ma (range 1.76e0.67). When the Pext value is increased to 0.07 (Scenario 7: marginal survival), the mean arrival date for Ternifine decreases slightly to 2.96 Ma (range 3.16e2.49 Ma), and there was little change in the mean arrival dates at Pirro Nord and Cueva Victoria, although the arrival probabilities decreased. The Mirzapur mean arrival date was most clearly influenced by the extinction rate change, at 1.28 Ma (range 2.79e0.29 Ma). No survival ability (Scenario 8: extinction) in desert (a Pext value of 1.0) did not prevent T. darti from reaching Eurasia, but it did lower the probabilities of arrival. Once again, the Ternifine mean arrival

Table 3 Standard vegetation tolerances for T. darti and T. oswaldi, based on paleoenvironmental and paleobiological evidence as discussed in the text. Both species are given a high survival probability in grassland habitats. T. darti is given good survival in other habitats, whereas T. oswaldi is given reasonable survival. Pext is extinction probability. Pext

T. darti T. oswaldi

Tropical forest

Warm-Temperate forest

Temperate forest

Savanna and dry woodland

Grassland and dry shrubland

Desert

0.03 0.04

0.03 0.04

0.03 0.04

0.01 0.01

0.01 0.01

0.03 0.04

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

54

Table 4 Sensitivity of the model to the dispersal rate constant, showing T. darti arrival times in North Africa and Eurasia with varying Pcr values (0.12e0.04) Arrival date (Ma) Site

Pcr

Mean  SE

Range

Mirzapur

0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04

3.08  0.01 3.04  0.01 2.97  0.01 2.87  0.01 2.74  0.03 2.37  0.05 1.76  0.09 1.22  0.69 0.00  0.00 3.31  0.00 3.28  0.00 3.25  0.01 3.20  0.01 3.14  0.01 3.03  0.02 2.68  0.06 1.92  0.09 1.69  0.21 2.91  0.01 2.76  0.02 2.50  0.05 1.83  0.11 1.64  0.11 1.53  0.09 1.20  0.00 0.00  0.00 0.00  0.00 2.78  0.02 2.63  0.02 2.44  0.02 2.22  0.02 1.80  0.04 1.33  0.06 0.00  0.00 0.00  0.00 0.00  0.00

3.21e2.97 3.21e2.82 3.25e2.71 3.09e2.60 3.04e2.14 2.90e1.29 2.82e0.11 2.20e0.25 0.00e0.00 3.36e3.24 3.34e3.22 3.32e3.12 3.30e3.01 3.24e2.80 3.18e2.47 3.11e1.54 2.91e0.49 2.88e0.13 3.12e2.69 3.01e2.18 2.90e1.52 2.76e0.14 2.49e0.57 2.05e0.95 1.20e1.20 0.00e0.00 0.00e0.00 3.06e2.57 2.82e2.38 2.73e1.95 2.56e1.76 2.39e1.08 2.03e0.47 0.00e0.00 0.00e0.00 0.00e0.00

Ternifine

Pirro Nord

Cueva Victoria

Probability of arrival 100% 100% 100% 100% 100% 98% 90% 4% 0% 100% 100% 100% 100% 100% 100% 100% 100% 30% 100% 100% 94% 80% 40% 22% 2% 0% 0% 100% 100% 100% 100% 98% 84% 0% 0% 0%

date remained relatively constant, at 2.92 Ma (range 3.10e 2.57 Ma), with a 94% chance of arrival. Sensitivity analysis: point of origin Based on current evidence, East Africa is the most likely region of origin for Theropithecus, and thus the main simulations originate there. However, it is possible that the bias towards the fossiliferous East African sites masks the true region of origin and therefore the dispersal patterns of Theropithecus. The effects of different geographic origins on dispersal were examined in two T. darti simulations, both of which used the standard parameters of Pcr and Pext, and a starting date of 3.5 Ma. In the first simulation, the population was seeded in Morocco. In the second, the population was initialized in Angola. The results (Fig. 5aee, Table 7) show that the specific location of the African point of origin has relatively little effect on the speed and likelihood of colonization of any of the four sites of interest except Ternifine in

Algeria, where starting the simulation in Morocco produces a virtually instantaneous arrival. Colonization probabilities are reduced slightly for the Eurasian sites, but the overall effect is very small. Thus, it appears legitimate, based both on the evidence from the fossil record and the robusticity of the model, to start the populations for the main simulations at Hadar and in the Turkana Basin, especially as these analyses show that the position of the seeding population within Africa does not greatly effect the overall outcomes of the model. Results T. darti main simulation: Pcr ¼ 0.07; Pext values in Table 3 Table 8 gives the mean arrival dates and ranges plus probability of arrival for the four sites of interest (Mirzapur, Pirro Nord, Cueva Victoria, Ternifine), with a climate transition model based on the 2.5 Ma transition (the only one applicable to T. darti). The mean arrival date at Mirzapur is 2.37 Ma (range 2.90e1.29 Ma), with an arrival probability of 98%. At Ternifine it is 3.03 Ma (3.18e2.47 Ma; 100%). The probability of arrival at Cueva Victoria is slightly lower, 84%, with a mean arrival date of 1.33 Ma (range 2.03e0.47 Ma), and at Pirro Nord it is much lower, 22%, with a mean arrival date of 1.53 Ma (2.05e0.95 Ma). It was shown in the sensitivity analysis (Table 4) that these arrival probabilities rise with an increased Pcr value, which also gives an earlier arrival. When Pcr is reduced to 0.06 (Table 4), however, the probabilities of arrival drop to 2% at Pirro Nord and 0% (no arrival) at Cueva Victoria, but little effect is observed on the probabilities of arrival at either Mirzapur (90%) or Ternifine (100%). In both the main simulation and the sensitivity analyses, dispersal via the Gibraltar Straits was prevented. When the Gibraltar Straits were included in the model as a plausible dispersal route, there was a noticeable impact on European colonization (Table 8). First arrival dates and mean arrival dates for Pirro Nord and Cueva Victoria increased, as did the probabilities of arrival, although at 26% arrival at Pirro Nord was still not guaranteed. Allowing crossings at the Gibraltar Straits to take place had little effect on arrival dates at Mirzapur and Ternifine. T. oswaldi main simulation: Pcr ¼ 0.06; Pext values in Table 3 and global environmental conditions based on the 2.5 Ma transition Table 9 gives the mean arrival dates and ranges plus probability of arrival for the four sites of interest (Mirzapur, Pirro Nord, Cueva Victoria, Ternifine), with a climate transition model based on the 2.5 Ma transition. The mean arrival date at Mirzapur is 0.80 Ma (range 1.55e0.29 Ma), with a 28% probability of arrival. There is a 94% chance of arrival at Ternifine, with a mean arrival date of 1.08 Ma (2.09e0.29 Ma). The probability of arrival at both Pirro Nord and Cueva Victoria is 0%, and it is only when Pcr is set much higher, at 0.08, that arrival occurs at Cueva Victoria (in 6% of cases; mean

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

55

Table 5 Sensitivity of the model to differing Pext values in the various habitat categories. Pext is extinction probability Arrival date (Ma) Site

Scenario

Mean  SE

Range

Mirzapur Ternifine Pirro Nord Cueva Victoria Mirzapur Ternifine Pirro Nord Cueva Victoria Mirzapur Ternifine Pirro Nord Cueva Victoria Mirzapur Ternifine Pirro Nord Cueva Victoria

2: 2: 2: 2: 3: 3: 3: 3: 4: 4: 4: 4: 5: 5: 5: 5:

0.88  0.18 2.69  0.06 0.00  0.00 0.00  0.00 1.29  0.15 2.56  0.12 0.00  0.00 0.00  0.00 2.06  0.07 2.62  0.05 0.00  0.00 0.00  0.00 2.65  0.02 3.14  0.01 1.52  0.10 1.96  0.02

2.56e0.17 2.93e2.53 0.00e0.00 0.00e0.00 2.60e0.17 3.01e0.57 0.00e0.00 0.00e0.00 2.89e0.80 3.20e1.81 0.00e0.00 0.00e0.00 2.88e2.24 3.23e2.99 2.39e0.12 2.25e1.64

little chance of survival in non-grassland habitats little chance of survival in non-grassland habitats little chance of survival in non-grassland habitats little chance of survival in non-grassland habitats marginal survival in non-grassland habitats marginal survival in non-grassland habitats marginal survival in non-grassland habitats marginal survival in non-grassland habitats reasonable survival in non-grassland habitats reasonable survival in non-grassland habitats reasonable survival in non-grassland habitats reasonable survival in non-grassland habitats very good survival in non-grassland habitats very good survival in non-grassland habitats very good survival in non-grassland habitats very good survival in non-grassland habitats

arrival date 0.47 Ma; range 0.74e0.29 Ma) and Pirro Nord (in 12% of cases; mean date 0.82; range 1.12e0.36 Ma). In the simulation above, dispersal via the Gibraltar Straits was prevented. When T. oswaldi dispersal was modelled with the Gibraltar Straits as an available dispersal route (Table 10), there was still no probability of arrival at Cueva Victoria when Pcr was set to the preferred value of 0.06, but there was a slim probability of arrival when Pcr was 0.07. Cueva Victoria arrival dates were earlier than when dispersal via the Gibraltar Straits was prevented, but only at the higher Pcr values. Allowing Gibraltar Straits dispersal had no appreciable effect on arrival at Ternifine or Mirzapur. At Pcr values of 0.06 and 0.07, there was no arrival at Pirro Nord, and at the higher Pcr values no clear pattern of either increased probability or earlier arrival time emerged; indeed extinction probabilities and chances of non-arrival were slightly higher for some values of Pcr. global environmental conditions based on the 1.8 Ma transition Table 11 gives the mean arrival dates and ranges plus probability of arrival for the four sites of interest, with a climate

Probability of arrival 30% 10% 0% 0% 60% 36% 0% 0% 90% 100% 0% 0% 100% 100% 68% 100%

transition model based on the 1.8 Ma transition. The mean arrival date at Mirzapur is 1.01 Ma (range 1.76e0.28 Ma), with a 66% probability of arrival. There is a 88% chance of arrival at Ternifine, with a mean arrival date of 1.34 Ma (2.09e0.20 Ma). The probability of arrival at both Pirro Nord and Cueva Victoria is 0%, although with a Pcr of 0.08, arrival occurs at Cueva Victoria in 4% of cases (mean arrival date 0.34 Ma; range 0.54e0.13 Ma) and Pirro Nord in 10% of cases (mean date 1.10; range 1.25e0.92 Ma). With the Gibraltar Straits set as a potential dispersal route (Table 12), the patterns for each of the sites were very similar to those seen for the model based on the 2.5 Ma climatic transition. global environmental conditions based on the 1.0 Ma transition Table 13 gives the mean arrival dates and ranges plus probability of arrival for the four sites of interest, with a climate transition model based on the 1.0 Ma transition. The mean arrival date at Mirzapur is 0.93 Ma (range 1.78e0.30 Ma), with an 84% probability of arrival. There is an 82% chance of arrival at Ternifine, with a mean arrival date of 1.36 Ma

Table 6 Sensitivity of the model to differential survival ability in desert. Arrival date (Ma) Site

Scenario

Mirzapur Ternifine Pirro Nord Cueva Victoria Mirzapur Ternifine Pirro Nord Cueva Victoria Mirzapur Ternifine Pirro Nord Cueva Victoria

6: 6: 6: 6: 7: 7: 7: 7: 8: 8: 8: 8:

quite reasonable survival in quite reasonable survival in quite reasonable survival in quite reasonable survival in marginal survival in desert marginal survival in desert marginal survival in desert marginal survival in desert Extinction in desert Extinction in desert Extinction in desert Extinction in desert

desert desert desert desert

Mean  SE

Range

2.11  0.08 2.99  0.02 1.14  0.26 1.24  0.05 1.28  0.14 2.96  0.02 1.35  0.08 1.26  0.10 0.72  0.23 2.92  0.02 1.32  0.00 1.12  0.08

2.90e1.21 3.20e2.51 1.86e0.24 1.76e0.67 2.79e0.29 3.16e2.49 1.52e1.18 1.95e0.84 1.43e0.18 3.10e2.57 1.32e1.32 1.61e0.61

Probability of arrival 86% 100% 12% 62% 60% 98% 6% 22% 10% 94% 2% 28%

56

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

Fig. 5. Sensitivity analysis of arrival date distributions for T. darti (Pcr ¼ 0.07) at (a) Ternifine, (b) Makapansgat, (c) Cueva Victoria, (d) Mirzapur and (e) Pirro Nord when Stepping Out is run with different origins of the dispersing population: Angola (solid, gray markers) and Morocco (thin, black markers).

(2.13e0.44 Ma). The probability of arrival at both Pirro Nord and Cueva Victoria is 0%, although a Pcr of 0.08 gives arrival at Cueva Victoria in 2% of cases (mean arrival date 0.31 Ma; range 0.31 e 0.31 Ma) and Pirro Nord in 66% of cases (mean date 0.59; range 1.36e0.10 Ma). Opening the Gibraltar Straits to dispersal (Table 14) did not result in arrival at Cueva Victoria with a Pcr of 0.06, but it did increase the chances of arrival to 30% at a Pcr of 0.07. Again, the patterns for the other sites were very similar to those observed for the 2.5 Ma and 1.8 Ma models, although Pirro Nord arrival was consistently more probable in the 1.0 Ma model, despite little change in mean arrival dates. Discussion Two main trends can be identified from the simulations of Theropithecus Afro-Eurasian dispersals. First, T. darti colonization is consistently more rapid with extinction en route less likely than is the case for T. oswaldi. Second, arrival

at Mirzapur is more probable than arrival at the other Eurasian sites, for both T. oswaldi and T. darti, and happens earlier. In the simulated Theropithecus darti dispersal with the Gibraltar Straits excluded as a possible dispersal route, populations had a very high probability of reaching both Ternifine and Mirzapur, but a slightly lower probability of arrival at Cueva Victoria and a much lower probability of arrival at Pirro Nord. Arrival at Ternifine is simulated to occur at 3.03 Ma (with the earliest date being 3.18 Ma and the most recent 2.47 Ma). This date range is earlier than is given for the North African middle Pleistocene T. oswaldi specimens, but encompasses the earliest age estimates of the Ain Jourdel Theropithecus material (Delson, 1993; Delson et al., 1993) and the probable date of Ahl al Oughlam, from which T. atlanticus has been recovered (Alemseged and Geraads, 1998). It is thus possible, especially given the morphological similarities between T. atlanticus and T. darti, that these early North African Theropithecus specimens could represent a T. darti

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77 Table 7 Sensitivity of the model to the point of origin of the dispersing species. Standard parameters were used (Pcr ¼ 0.07; Pext as given in Table 3 for T. darti). For an indication of the level of difference compared to a dispersal from Hadar, please see main T. darti results in Table 8

57

Table 9 Results of the main T. oswaldi simulation (2.5 Ma climate transition), with the Gibraltar Straits excluded as a potential dispersal route. The standard Pext values were used (Table 3). Data for a range of Pcr values are reported; results based on the standard value of 0.06 are listed in bold

Arrival date (Ma)

Arrival date (Ma)

Site

Origin

Mean  SE

Range

Mirzapur Ternifine Pirro Nord Cueva Victoria Mirzapur Ternifine Pirro Nord Cueva Victoria

Angola Angola Angola Angola Morocco Morocco Morocco Morocco

1.96  0.05 3.06  0.01 1.51  0.06 1.24  0.05 2.01  0.06 3.50  0.00 1.28  0.14 1.24  0.05

2.64e1.14 3.20e2.89 1.81e1.36 1.87e0.75 2.73e0.63 3.50e3.50 1.97e0.53 1.82e0.64

Probability of arrival 96% 100% 12% 60% 86% 100% 18% 70%

Site

Pcr

Mean  SE

Range

Mirzapur

0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04

2.07  0.01 2.02  0.01 1.88  0.02 1.75  0.02 1.62  0.04 1.10  0.07 0.80  0.11 0.00  0.00 0.00  0.00 2.28  0.01 2.26  0.01 2.19  0.01 2.13  0.02 1.98  0.03 1.72  0.05 1.08  0.08 1.13  0.22 1.00  0.00 1.79  0.01 1.44  0.06 1.14  0.09 0.79  0.14 0.82  0.10 0.00  0.00 0.00  0.00 0.00  0.00 0.00  0.00 1.57  0.02 1.33  0.03 1.00  0.05 0.62  0.04 0.47  0.11 0.00  0.00 0.00  0.00 0.00  0.00 0.00  0.00

2.22e1.88 2.17e1.81 2.10e1.44 2.04e1.13 2.03e0.72 1.82e0.12 1.55e0.29 0.00e0.00 0.00e0.00 2.35e2.09 2.36e2.08 2.27e1.91 2.27e1.78 2.22e1.50 2.20e0.76 2.09e0.29 1.96e0.32 1.00e1.00 2.05e1.58 1.87e0.42 1.92e0.10 1.30e0.18 1.12e0.36 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 1.83e1.31 1.69e0.89 1.83e0.20 1.05e0.10 0.74e0.29 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

Ternifine

expansion into North as well as southern Africa in the middle to late Pliocene. The mean Mirzapur arrival date of 2.37 Ma is significantly outside the date range of 1.72 Mae0.6 Ma given for the Pinjor Formation and the Lower Conglomerate in the region (Nanda, 2002). The most recent Mirzapur arrival date indicated through the T. darti simulation, 1.29 Ma, although falling within the possible overall range of the Formation, is also earlier than the dates preferred by Rook et al. (2004) and Delson et al. (1993) for the Mirzapur T. oswaldi material. In line with suggestions that the late Pliocene might have been a more suitable period for mammalian dispersal than the early Pleistocene (Turner, 1999; Dennell, 2003), the simulations indicate that a T. darti expansion out of Africa into Asia during the Pliocene was possible. Indeed, compared to the results of the T. oswaldi simulations, Eurasian colonization by Theropithecus is more probable in the Pliocene than in the early Pleistocene. However, if there was a Pliocene Theropithecus dispersal into Asia, there is as yet no fossil evidence to support this. To date, no primate material that can be attributed to Theropithecus has been recovered from Asian sediments older than those in the Lower Conglomerate, although there is evidence for other Old World monkeys in the Siwaliks, including a colobine species, probably Presbytis sivalensis, dated to 6.3 Ma (Barry, 1987). P. sivalensis, although smaller, resembles the Eurasian

Table 8 Results of the main T. darti simulation, including data for simulations in which the Gibraltar Straits were a potential dispersal route. The standard parameters were used (Pcr ¼ 0.07; Pext as given in Table 3) Arrival date (Ma) Site

Potential Gibraltar Straits crossing?

Mean  SE

Range

Mirzapur Ternifine Pirro Nord Cueva Victoria Mirzapur Ternifine Pirro Nord Cueva Victoria

No No No No Yes Yes Yes Yes

2.37  0.05 3.03  0.02 1.53  0.09 1.33  0.06 2.40  0.06 3.03  0.02 1.81  0.09 2.67  0.06

2.90e1.29 3.18e2.47 2.05e0.95 2.03e0.47 2.93e0.97 3.21e2.54 2.30e1.24 3.09e1.05

Probability of arrival 98% 100% 22% 84% 100% 100% 26% 100%

Pirro Nord

Cueva Victoria

Probability of arrival 100% 100% 100% 100% 100% 84% 28% 0% 0% 100% 100% 100% 100% 100% 100% 94% 16% 2% 100% 82% 56% 14% 12% 0% 0% 0% 0% 100% 100% 96% 72% 6% 0% 0% 0% 0%

Miocene colobine Mesopithecus (Jablonski, 2002), and so is likely to have evolved in Eurasia from an ancestor that dispersed out of Africa in the late Miocene. Cercopithecines may not have been present in this region, however, until 3.2 Ma or later (Barry, 1987). Two of the three cercopithecine species recognized by Barry in the Siwalik faunas, ?Macaca palaeindica and Procynocephalus subhimalayanus, were very probably the result of the Macaca dispersal out of Africa in the late Miocene or very early Pliocene. We argue that the third species identified by Barry (1987), Theropithecus (oswaldi) delsoni, represents a more recent wave of dispersal of cercopithecids into the region. The cercopithecid fossil record from the Siwaliks is relatively poor, and it is possible that T. darti, if it dispersed, is not found because it was never very abundant. Given that other cercopithecines had dispersed from Africa to this part of Asia during the Pliocene, there is no compelling geographic or environmental reason why T. darti could not also have done so, which is reinforced by the results of the

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

58

Table 10 Results of the main T. oswaldi simulation (2.5 Ma climate transition) with the Gibraltar Straits included as a potential dispersal route. The standard Pext values were used (Table 3). Data for a range of Pcr values are reported; results based on the standard value of 0.06 are listed in bold

Table 11 Results of the main T. oswaldi simulation (1.8 Ma climate transition), with the Gibraltar Straits excluded as a potential dispersal route. The standard Pext values were used (Table 3). Data for a range of Pcr values are reported; results based on the standard value of 0.06 are listed in bold

Arrival date (Ma) Site

Pcr

Mean  SE

Range

Mirzapur

0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04

2.07  0.01 2.00  0.01 1.93  0.01 1.75  0.02 1.57  0.04 1.19  0.06 0.99  0.13 0.00  0.00 0.00  0.00 2.27  0.01 2.25  0.01 2.18  0.01 2.12  0.02 1.98  0.03 1.72  0.05 1.16  0.09 1.16  0.16 0.00  0.00 1.81  0.01 1.60  0.05 1.32  0.08 0.43  0.18 0.29  0.09 0.00  0.00 0.00  0.00 0.00  0.00 0.00  0.00 1.91  0.03 1.73  0.04 1.44  0.06 0.96  0.09 0.93  0.21 0.33  0.00 0.00  0.00 0.00  0.00 0.00  0.00

2.18e1.96 2.16e1.85 2.08e1.74 2.08e1.19 1.99e0.31 1.76e0.50 1.69e0.10 0.00e0.00 0.00e0.00 2.34e2.12 2.36e2.10 2.30e1.92 2.28e1.74 2.21e1.43 2.14e0.62 2.06e0.12 1.83e0.11 0.00e0.00 1.99e1.63 1.96e0.12 1.78e0.17 1.19e0.11 0.51e0.16 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 2.25e1.41 2.21e1.11 2.12e0.37 2.03e0.11 1.85e0.16 0.33e0.33 0.00e0.00 0.00e 0.00 0.00e0.00

Ternifine

Pirro Nord

Cueva Victoria

Arrival date (Ma) Probability of arrival 100% 100% 100% 100% 100% 88% 30% 0% 0% 100% 100% 100% 100% 100% 100% 94% 28% 0% 100% 92% 58% 10% 6% 0% 0% 0% 0% 100% 100% 100% 72% 20% 2% 0% 0% 0%

simulation. However, it should be noted that the absence of T. darti is not necessarily simply due to taphonomic bias against primates, due to the fact that other cercopithecids have been identified from Pliocene and early Pleistocene horizons in the Siwaliks. If ‘absence of evidence’ turns out in the future to be most probably ‘evidence of absence,’ competitive exclusion of Theropithecus from the Siwalik primate community during this period should not be ruled out. The dispersal of T. darti to the sites in Europe (Pirro Nord and Cueva Victoria) is much slower and has a lower probability of arrival in comparison to Mirzapur. The Cueva Victoria mean arrival date of 1.33 Ma is consistent with the possible early Pleistocene age of the site (Agusti et al., 1986; Martinez-Navarro et al., 2005), although the simulated range of arrival dates, from 2.03e0.47 Ma, encompasses a much wider time period. The Pirro Nord faunal

Site

Pcr

Mean  SE

Range

Mirzapur

0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04

2.07  0.01 2.01  0.01 1.93  0.01 1.80  0.02 1.52  0.03 1.33  0.04 1.01  0.06 0.76  0.27 0.00  0.00 2.28  0.01 2.24  0.01 2.19  0.01 2.14  0.02 1.98  0.03 1.69  0.05 1.34  0.07 1.34  0.07 1.19  0.00 1.77  0.01 1.66  0.02 1.46  0.02 1.17  0.02 1.10  0.06 0.00  0.00 0.00  0.00 0.00  0.00 0.00  0.00 1.56  0.02 1.31  0.03 0.88  0.05 0.58  0.05 0.34  0.14 0.00  0.00 0.00  0.00 0.00  0.00 0.00  0.00

2.20e1.87 2.16e1.84 2.09e1.77 2.01e1.43 1.88e0.92 1.82e0.32 1.76e0.28 1.14e0.37 0.00e0.00 2.35e2.16 2.31e1.99 2.30e1.91 2.27e1.80 2.25e1.41 2.10e0.85 2.09e0.20 1.83e0.73 1.19e1.19 2.00e1.58 1.86e1.28 1.83e1.16 1.68e 0.91 1.25e0.92 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 1.81e1.14 1.75e0.67 1.43e0.24 1.28e0.15 0.54e0.13 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

Ternifine

Pirro Nord

Cueva Victoria

Probability of arrival 100% 100% 100% 100% 100% 100% 66% 4% 0% 100% 100% 100% 100% 100% 100% 88% 40% 2% 100% 100% 100% 92% 10% 0% 0% 0% 0% 100% 100% 98% 66% 4% 0% 0% 0% 0%

assemblage is dated biochronologically to between 1.6e1.3 Ma (Rook et al., 2004), and the simulated range of arrival dates (2.05e0.95 Ma), with a mean of 1.53 Ma, includes this period. Thus, simulated dispersals that originate in Africa during the Pliocene prior to 3 Ma do not result in colonization in Italy and Spain until much later, with the dates in the simulation largely concordant with the fossil record. It is therefore possible that the presence of Theropithecus in Europe is the result of a mid-Pliocene dispersal from Africa. However, as is discussed below, the taxonomic affinities of the Theropithecus material at Cueva Victoria, and to a lesser extent at Pirro Nord, do not necessarily support this hypothesis. In the simulated T. oswaldi dispersals, the probability of arrival at Ternifine was high for all climate transition models. The climate shifts incorporated in the model result in the

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77 Table 12 Results of the main T. oswaldi simulation (1.8 Ma climate transition), with the Gibraltar Straits included as a potential dispersal route. The standard Pext values were used (Table 3). Data for a range of Pcr values are reported; results based on the standard value of 0.06 are listed in bold

Table 13 Results of the main T. oswaldi simulation (1.0 Ma climate transition), with the Gibraltar Straits excluded as a potential dispersal route. The standard Pext values were used (Table 3). Data for a range of Pcr values are reported; results based on the standard value of 0.06 are listed in bold

Arrival date (Ma)

Arrival date (Ma)

Site

Pcr

Mean  SE

Range

Mirzapur

0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04

2.07  0.01 2.03  0.01 1.93  0.01 1.78  0.02 1.62  0.03 1.40  0.04 1.01  0.06 0.54  0.08 0.00  0.00 2.28  0.01 2.24  0.01 2.20  0.01 2.07  0.02 2.00  0.02 1.78  0.05 1.37  0.08 1.28  0.10 1.35  0.16 1.81  0.01 1.70  0.02 1.48  0.02 1.15  0.03 0.99  0.09 0.00  0.00 0.00  0.00 0.00  0.00 0.00  0.00 j1.90  0.03 1.74  0.04 1.40  0.07 1.14  0.09 1.34  0.09 1.33  0.08 0.00  0.00 0.00  0.00 0.00  0.00

2.20e1.89 2.19e1.87 2.16e1.68 2.06e1.39 1.89e1.06 1.81e0.78 1.59e0.16 0.66e0.42 0.00e0.00 2.37e2.13 2.30e2.07 2.28e1.99 2.26e1.55 2.21e1.59 2.17e0.66 2.06e0.25 1.82e0.11 1.67e1.01 1.97e1.65 1.97e1.38 1.82e1.15 1.49e0.48 1.14e0.68 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 2.25e1.45 2.18e1.12 2.04e0.25 2.16e0.14 1.94e0.23 1.61e1.05 0.00e0.00 0.00e0.00 0.00e0.00

Ternifine

Pirro Nord

Cueva Victoria

59

Probability of arrival 100% 100% 100% 100% 100% 98% 68% 4% 0% 100% 100% 100% 100% 100% 100% 78% 34% 6% 100% 100% 100% 76% 8% 0% 0% 0% 0% 100% 100% 100% 86% 40% 12% 0% 0% 0%

‘Pliocene’ vegetation patterns persisting for longer, so it is likely that the arrival dates will become earlier in the later scenarios. The mean arrival date of 1.08 Ma at Ternifine for the 2.5 Ma climatic model is more recent than the mean arrival dates (1.34 Ma and 1.36 Ma) for the 1.8 Ma and 1.0 Ma transitions, respectively. All these mean dates fall before the likely ages of the North African deposits that have yielded T. oswaldi specimens, indicating that the Pleistocene colonization of North Africa is consistent with expansion of T. oswaldi from East Africa at or even later than 2.4 Ma. The extent of the Sahara Desert is a major environmental factor that may have impacted upon Theropithecus dispersal into North Africa, in both the Pliocene and the Pleistocene. This is taken into consideration in the model through Pext values for desert survival. Altering the Pext value for survival in desert to 1.0 (no survival) reduces the probability of T. oswaldi arrival at Ternifine to 20% for the 2.5 Ma transition, 60% for the 1.8 Ma model,

Site

Pcr

Mean  SE

Range

Mirzapur

0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04

2.08  0.01 2.00  0.01 1.92  0.01 1.81  0.02 1.65  0.03 1.29  0.04 0.93  0.06 0.44  0.03 0.00  0.00 2.29  0.00 2.25  0.01 2.22  0.01 2.09  0.02 1.97  0.03 1.72  0.05 1.36  0.07 0.84  0.09 0.58  0.23 1.78  0.02 1.68  0.02 1.46  0.02 1.12  0.03 0.59  0.05 0.32  0.00 0.00  0.00 0.00  0.00 0.00  0.00 1.54  0.03 1.41  0.03 0.93  0.05 0.55  0.06 0.31  0.00 0.00  0.00 0.00  0.00 0.00  0.00 0.00  0.00

2.19e1.93 2.17e1.81 2.13e1.72 2.08e1.33 1.99e0.98 1.71e0.58 1.78e0.30 0.53e0.36 0.00e0.00 2.35e2.21 2.33e2.08 2.31e2.01 2.26e1.74 2.25e1.59 2.17e0.62 2.13e0.44 1.47e0.11 1.52e0.10 2.04e1.56 2.03e1.47 1.83e1.11 1.64e0.59 1.36e0.10 0.32e0.32 0.00e0.00 0.00e0.00 0.00e0.00 1.94e0.98 1.73e0.85 1.61e0.22 1.10e0.12 0.31e0.31 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

Ternifine

Pirro Nord

Cueva Victoria

Probability of arrival 100% 100% 100% 100% 100% 100% 84% 10% 0% 100% 100% 100% 100% 100% 100% 82% 32% 10% 100% 100% 100% 100% 66% 2% 0% 0% 0% 100% 100% 90% 40% 2% 0% 0% 0% 0%

and 90% for the 1.0 Ma model. However, in no case did all simulated dispersals fail to reach Ternifine, and the range of arrival dates still falls before the North African T. oswaldi FAD. Added to this is the possibility that the Sahara underwent periodic episodes of ‘greening,’ with the extent of the desert shifting in response to global climatic fluctuations which might have further facilitated the movement of large mammals into North Africa. The T. oswaldi arrival probabilities at Mirzapur were lower than in the T. darti simulation, particularly for the 2.5 Ma climate transition model. As was the case for Ternifine, mean arrival date at Mirzapur was more recent, 0.80 Ma, using the 2.5 Ma model than it was with the other two climate transition models, although the overall differences were fairly small and the ranges overlapped. However, the simulated Mirzapur mean arrival dates and ranges correspond well to the potential age of the Mirzapur Pinjor Formation to Lower Conglomerate

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

60

Table 14 Results of the main T. oswaldi simulation (1.0 Ma climate transition), with the Gibraltar Straits included as a potential dispersal route. The standard Pext values were used (Table 3). Data for a range of Pcr values are reported; results based on the standard value of 0.06 are listed in bold Arrival date (Ma) Site

Pcr

Mean  SE

Range

Mirzapur

0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.12 0.11 0.10 0.09 0.08 0.07 0.06 0.05 0.04

2.06  0.01 2.00  0.01 1.93  0.02 1.82  0.02 1.62  0.03 1.28  0.04 0.83  0.05 0.66  0.18 0.00  0.00 2.29  0.01 2.24  0.01 2.20  0.01 2.10  0.02 1.93  0.03 1.63  0.05 1.26  0.08 0.98  0.08 0.98  0.18 1.83  0.02 1.66  0.02 1.48  0.02 1.14  0.04 0.60  0.04 0.00  0.00 0.00  0.00 0.00  0.00 0.00  0.00 1.89  0.03 1.78  0.05 1.49  0.06 1.16  0.09 1.03  0.11 0.89  0.13 0.00  0.00 0.80  0.00 0.00  0.00

2.23e1.87 2.18e1.74 2.12e1.66 2.07e1.43 1.98e0.73 1.85e0.39 1.44e0.10 1.36e0.19 0.00e0.00 2.34e2.17 2.32e2.06 2.30e1.80 2.29e1.63 2.24e1.43 2.10e0.67 2.07e0.19 1.76e0.20 1.35e0.49 2.14e1.62 1.99e1.37 1.76e1.15 1.67e0.37 1.40e0.20 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 2.28e1.32 2.16e0.27 2.11e0.54 2.14e0.11 1.95e0.15 1.64e0.19 0.00e0.00 0.80e0.80 0.00e0.00

Ternifine

Pirro Nord

Cueva Victoria

Probability of arrival 100% 100% 100% 100% 100% 100% 82% 14% 0% 100% 100% 100% 100% 100% 100% 90% 46% 8% 100% 100% 100% 100% 72% 0% 0% 0% 0% 100% 100% 98% 80% 46% 30% 0% 2% 0%

transition (Nanda, 2002) as well as age estimates of the Mirzapur T. oswaldi material (Delson, 1993; Delson et al., 1993; Rook et al., 2004). There is no evidence for T. darti in Asia, so it appears possible, if not probable, that the Mirzapur colonization was the result of a T. oswaldi dispersal from Africa relatively soon after 2.4 Ma. In order to do this, colonizing Theropithecus populations would need to exploit the C3 grasslands and ecosystems found in parts of Asia. The one extant member of the genus, T. gelada, consumes C3 grasses like Danthonia found at higher altitudes in the Simen Mountains (Iwamoto, 1993), and also supplements seasonally with the C3 legume Trifolium (clover; Iwamoto, 1993). This suggests that despite the dominance of C4 vegetation at important African Theropithecus localities such as Olorgesailie (Sikes et al., 1999), members of the genus would not be precluded from incorporating C3 as well as C4 plants into their diets. Results of

stable carbon isotope analyses of T. darti teeth from South Africa also support this, as the range of d13C values obtained indicate that C3 foods were being consumed within a diet based on tropical C4 plants (Lee-Thorp et al., 1989; Codron et al., 2005). Using the colonization rates and extinction probabilities for the different habitat categories that were defined a priori on the basis of paleobiology and the fossil record, the simulated T. oswaldi populations failed to reach both Cueva Victoria and Pirro Nord when the Gibraltar Straits were not used as a dispersal route. With a faster Pcr of 0.08, arrival was more likely at Pirro Nord than at Cueva Victoria, but for both sites arrival occurred in only a very small number of cases under the 2.5 Ma and 1.8 Ma climate transition models. Using the 1.0 Ma climate model, again with a Pcr of 0.08, the probability of arrival at Cueva Victoria was still very small, but the probability of arrival at Pirro Nord increased dramatically, to 66%. For all climate transition models and for both European sites, the ranges of simulated arrival dates were more recent than the probable ages of the sites. The Pcr rate had to be increased to 0.10 under the 2.5 Ma climate model and to at least 0.09 under the other climate models before the arrival ranges corresponded to the probable ages of the sites (Appendices 2ae2c). Using the original calibrated Pcr value of 0.06, Pext values had to be reduced to 0.01 in all habitats except ice, boreal forest, and tundra before simulated arrival occurred at Pirro Nord and Cueva Victoria within the dates indicated by the fossil record. Arrival at Mirzapur, for both the T. darti and the T. oswaldi simulations, was much more likely and occurred faster than arrival at the other Eurasian sites. This was largely due to differences in environment and topography. Travelling from Arabia to Mirzapur, Theropithecus disperses through a mosaic of desert and grassland, and India itself is modelled as containing mainly grassland. This route is therefore particularly suitable for Theropithecus dispersal, as its probability of extinction in grassland and savannah was set to 0.01 (a minimal value). Dispersal to India nonetheless involves crossing mountain ranges in Iran and Afghanistan, but the simulations, as well as the actual evidence from the fossil record, suggest that if Theropithecus gets out of Arabia at all, such a barrier would not have prevented dispersal. Movement into Europe requires crossing, and therefore presumably exploiting, several biomes not usually associated with Theropithecus in Africa. To reach Cueva Victoria from Arabia requires travel through warm-temperate and temperate forest; as the simulation progresses through the Pleistocene and glacial cycles increase, Europe is periodically dominated by boreal forest and tundra, two habitats in which the modelled Theropithecus was given a very high probability of extinction. Colonizing Pirro Nord may have entailed dispersal through some boreal forest, as well as through warm-temperate and temperate forest. To enter Italy via a Eurasian mainland dispersal would also involve crossing the Alps, and although this might have been harder during glacial periods, the Alps would not necessarily have been any more of a barrier to

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

dispersal than were the mountain ranges of Iran and Afghanistan. Since the environment is a key factor in successful dispersal, one explanation for the failure of a simulated dispersal of T. oswaldi from East Africa at 2.4 Ma to colonize Spain or Italy using the parameters set a priori could be that they did not accurately reflect T. oswaldi paleobiology and environmental tolerances. Members of the T. dartieT. oswaldi lineage are sometimes viewed as exclusively grassland and savanna dwelling, despite recovery of fossils from sites at which the paleoenvironment is reconstructed as being relatively closed (Reed, 1997), and postcranial evidence that suggest Theropithecus (Theropithecus) species may have been able to use a mix of arboreal and terrestrial substrates in a manner similar to modern common baboons (Elton, 2002; Elton et al., 2003). The Pext values for extinction risk in particular habitats used in the main simulation were set with this in mind, but given the poor probabilities of arrival and the late arrival dates of T. oswaldi at Pirro Nord and Cueva Victoria, these might have been underestimates of its ability to survive in non-grassland habitats. The potential flexibility of Theropithecus ecology is reinforced by sensitivity analyses indicating that if either T. darti or T. oswaldi were exclusively confined to grassland and savanna habitats, they would be very unlikely to disperse out of Africa into Europe. Indeed, an ability to survive in temperate forest would be necessary if Theropithecus populations were to reach Cueva Victoria without crossing the Gibraltar Straits. Another explanation for the poor probabilities of T. oswaldi arrival at the European sites, particularly Pirro Nord, is that the model imposes constraints on dispersal that might not exist in reality. The Pcr values might be underestimates of the actual dispersal speeds, bearing in mind that the FAD at a particular site will not always accurately represent the true date of first appearance. Coupled with this, colonization is simulated by presence or absence within grid squares. The entry route to Italy is thus restricted, and stochastic factors could have influenced Italian colonization more severely than in other modelled regions, as well as having a greater impact than would have really been the case. A third explanation for the poor colonization of Europe by Theropithecus oswaldi does not have to invoke potential shortcomings of the model. In the simulation, the dates of T. darti arrival in Europe correspond to the likely ages of the sites, so based on our results it cannot be discounted that the Theropithecus material at Cueva Victoria and Pirro Nord is the result of a Pliocene T. darti dispersal. However, the relatively large size of the Eurasian material links it more clearly to T. oswaldi, although larger body masses might have increased convergently in Eurasian and African Theropithecus. The uncertainties related to Theropithecus colonization highlighted by the Stepping Out simulations demonstrate that there is still much to discover about cercopithecid evolution in general and Theropithecus evolutionary history in particular. On current evidence, the fossils at Cueva Victoria are best referred to T. oswaldi sensu stricto, with the Pirro Nord material being more uncertain. The paucity of European material attributed to Theropithecus means that it is possible the taxonomy and

61

relationships of Theropithecus populations in Eurasia might only be resolved satisfactorily through new fossil finds. It is theoretically possible, but by no means probable, that some Afro-Eurasian mammalian dispersal occurred via the Gibraltar Straits. Both in the T. darti and the T. oswaldi simulations, allowing the Gibraltar Straits to be used as a dispersal route had no impact on colonization of Ternifine and Mirzapur. However, it had a positive impact on Cueva Victoria and Pirro Nord colonization probabilities for T. darti. In the T. oswaldi model, Cueva Victoria arrival probabilities and dates were only affected when colonization rates were relatively rapid (Pcr of 0.07 and above), with this being most noticeable under the 1.0 Ma transition model. The effect of allowing a Gibraltar Straits crossing was inconclusive for T. oswaldi colonization of Pirro Nord. These results suggest that although a Gibraltar Straits crossing might make Theropithecus oswaldi arrival at Cueva Victoria easier, allowing such a crossing did not have a dramatic impact on its dispersal into Eurasia more generally. The means by which Theropithecus might have crossed such a body of water are also not clear. This is reinforced by biogeographical and genetic studies of modern mammals which indicate that, with the exception of a few small-bodied taxa, there is little evidence for mammalian dispersal via the Gibraltar Straits since the Messinian (O’Regan et al., 2006). The results of the Stepping Out simulations therefore lend support to the argument (O’Regan et al., 2006) that the Levantine corridor was the most likely route of large-bodied primate dispersal from Africa to Eurasia. Hominins were the other large-bodied primates to disperse out of Africa in the Plio-Pleistocene, and it is a possibility that at least one cercopithecid dispersed into Asia with them. Dennell and Roebroeks (2005: 1100) mention the ‘surprising find’ of Papio sushkini in central Asia, and also argue that it is ‘often regarded as a commensal of Homo.’ The presence of Papio sushkini in Asia is not good evidence for commensal dispersal, however. Papio proper is likely to be confined to Africa and a small part of Arabia, and the cercopithecid specimens named Paradolichopithecus sushkini (Trofimov, 1977) were re-assigned to Papio sushkini by Maschenko (1994) and subsequently referred to Paradolichopithecus arvernensis by Jablonski (2002). The individual represented by these fossils was therefore much more likely to have evolved in situ in Asia from an earlier dispersal of Macaca than it was to have dispersed directly out of Africa. This leaves Theropithecus as the only cercopithecid taxa of indisputably African origin to be found in Eurasia after the late Miocene/early Pliocene dispersal of macaques. The simulations indicate that conditions would have favored movement of Theropithecus darti out of Africa during the Pliocene, but there is no solid evidence for this. The same is true for the hypothesized dispersal of Pliocene australopiths out of Africa (Dennell and Roebroeks, 2005). The Eurasian evidence for T. oswaldi and Homo is much stronger, and as the two were probably sympatric in East Africa (Elton, 2006), it is possible that T. oswaldi could have accompanied Homo out of Africa in the early Pleistocene, or at least, as Rook

62

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

et al. (2004) argue, Theropithecus could have moved in parallel with Homo and other large mammals. The simulations suggest some parallels between the colonization of Eurasia by Homo and Theropithecus. The greater simulated ease of the Theropithecus oswaldi colonization of Asia compared to that of Europe is similar to the pattern seen in the hominin fossil record. Homo is first found in central and eastern Asia (e.g., Dmanisi, w1.7 Ma [Gabunia et al., 2000] and Indonesia, w1.5Ma [Larick et al., 2001]), and only later in Europe (e.g., Ceprano, 0.8e0.9 Ma [Ascenzi et al., 2000] and Atapuerca TD6, w0.8 Ma [Falgue`res et al., 1999]). In the T. oswaldi simulation, arrival at Mirzapur in Asia occurred with greater probability and speed than did the arrival at Pirro Nord and Cueva Victoria, indicating that large-bodied primates might have found colonizing Asia easier than moving into Europe. Several questions remain about the patterns evident in the Eurasian dispersal of Theropithecus, in particular why Theropithecus is not very well-represented in Eurasia, and why it experienced such a severe range contraction, even in Africa, when other primates such as Homo and Macaca did not. Theropithecus oswaldi dominates the primate fauna at several East African Pleistocene sites, including the Koobi Fora Okote Member, Olorgesailie, and Olduvai Upper Bed II (Delson et al., 1993; Leakey, 1993; Jablonski, 2002). These localities are well-studied and have yielded the vast majority of the evidence for Theropithecus paleobiology, leading to potentially skewed interpretations of the ecologies and behavior of dispersed Theropithecus populations. The southern African primate communities are dominated in the Plio-Pleistocene by Parapapio and in the Pleistocene by Papio, neither of which have a very good fossil record in East Africa (Szalay and Delson, 1979; Benefit 1999b; Jablonski, 2002; Elton, in press). Theropithecus is much less well-represented in southern Africa. This is potentially taphonomic bias, but the uneven distribution of Theropithecus fossil densities could reflect the ecology of primate communities, with the apparently incoming Theropithecus in southern Africa having to compete with papionins and other cercopithecids such as Cercopithecoides williamsi in established populations. The small amount of material at the Pleistocene Eurasian sites at which Theropithecus has been identified may reflect a lower density of Theropithecus in Eurasia, where Macaca, as well as the colobines, had speciated (Delson, 1994; Jablonski, 2002). However, more work on Asian deposits is required to confirm this. Studies of extant mammals indicate that dispersal rates can be highly variable, influenced by environmental conditions, and competition with existing species (Anto´n et al., 2002). The pattern that currently emerges from the Theropithecus fossil record of southern Africa and Eurasia compared to that of East Africa suggests that dispersing Theropithecus faced competition from existing resident monkeys, and as a result never reached the high densities observed in East Africa. As a grass-eater, it might also have faced competition from the grazing ungulates that radiated during the Pliocene and Pleistocene. Even if there was no synchronous dispersal into Eurasia, interactions with Homo may have been a factor in

Theropithecus range contraction in Africa and elsewhere. In the Pleistocene, Papio replaced Theropithecus as the dominant cercopithecid in Africa. It is unlikely that Theropithecus was directly out-competed (Elton, in press), but Papio, with its ecological and behavioral flexibility, is certainly adept at living alongside humans (Jablonski, 2002; Elton, in press). Several Macaca species are similarly flexible and manage to survive in habitats with extensive anthropogenic modification. It has been suggested that Theropithecus had greater difficulty living alongside Homo (Elton, in press). Furthermore, the extinction of T. oswaldi in Africa coincides with the onset of the major Pleistocene climatic fluctuations, and it may have found it difficult to ‘ride out’ relatively rapid climatic and environmental changes (Leakey, 1993; Elton, in press). The simulation results also suggest that conditions for Theropithecus dispersal in Eurasia would be less favorable in the Pleistocene than in the Pliocene. A combination of pressure from humans and climate change in environmental conditions that were already less than optimal may therefore have resulted in a massive contraction of the Theropithecus range, far beyond that evident in Macaca during the middle to late Pleistocene.

Conclusions The Stepping Out simulations demonstrate that there is still much to be discovered about Theropithecus evolutionary history, especially in Eurasia. They indicate that in order for dispersal to occur, fossil Theropithecus species would have needed to be more ecologically flexible than might be assumed based on the ecology of the one living member of the genus, T. gelada. In both the T. darti and the T. oswaldi simulations, colonizing Asia was easier than dispersing into Europe, a result that parallels evidence from the fossil record of Homo. Competition with previously established communities of other large mammals, including primates, might have prevented Theropithecus from attaining the abundance apparently evident in the Plio-Pleistocene fossil record of East Africa. The modelling also indicates that a Plio-Pleistocene dispersal might have been harder and slower than one that occurred earlier in the Pliocene, and although taxonomic evidence points to a Plio-Pleistocene dispersal of T. oswaldi, an earlier T. darti dispersal cannot be completely discounted.

Acknowledgements We thank the Editor and Associate Editor for their helpful and enormously constructive comments on this paper. We are also immensely grateful to three anonymous referees whose detailed suggestions improved this research beyond measure. The modelling was undertaken whilst JH was at the Bristol Research Initiative for the Dynamic Global Environment, and Paul Valdes kindly provided the GCM simulations used in this paper. This work was funded by the NERC EFCHED grants NER/T/S/2002/00431 and NER/T/S/2002/00464.

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

Appendix 1. T. darti mean, standard error, range of arrival dates (Ma), and probabilities of arrival for all scenarios and colonization rates

2.73  0.04 2.69  0.06 2.54  0.00 2.64  0.00 0.00  0.00

2.89e2.56 2.93e2.53 2.54e2.54 2.64e2.64 0.00e0.00

18% 10% 2% 2% 0%

100% 100% 100% 100% 100% 100% 100% 100% 56%

Statistics for Pirro Nord Pcr ¼ 0.12 0.00  0.00 Pcr ¼ 0.11 0.00  0.00 Pcr ¼ 0.10 0.00  0.00 Pcr ¼ 0.09 0.00  0.00 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

0% 0% 0% 0% 0% 0% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 0.00  0.00 0.00  0.00 Pcr ¼ 0.11 Pcr ¼ 0.10 0.00  0.00 Pcr ¼ 0.09 0.00  0.00 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

0% 0% 0% 0% 0% 0% 0% 0% 0%

Scenario 1: ability equal in all habitats Statistics for Mirzapur Pcr ¼ 0.12 3.15  0.01 Pcr ¼ 0.11 3.11  0.01 Pcr ¼ 0.10 3.05  0.01 Pcr ¼ 0.09 2.98  0.01 Pcr ¼ 0.08 2.89  0.01 Pcr ¼ 0.07 2.76  0.02 Pcr ¼ 0.06 2.55  0.03 Pcr ¼ 0.05 2.02  0.07 Pcr ¼ 0.04 1.00  0.12

3.24e3.07 3.23e2.99 3.19e2.95 3.08e2.80 3.08e2.66 2.97e2.49 2.86e1.91 2.65e0.61 2.28e0.10

Statistics for Ternifine Pcr ¼ 0.12 3.33  0.00 Pcr ¼ 0.11 3.31  0.00 Pcr ¼ 0.10 3.30  0.00 Pcr ¼ 0.09 3.28  0.00 Pcr ¼ 0.08 3.24  0.00 Pcr ¼ 0.07 3.18  0.01 Pcr ¼ 0.06 3.11  0.01 Pcr ¼ 0.05 3.02  0.01 Pcr ¼ 0.04 2.86  0.01

3.37e3.27 3.37e3.27 3.35e3.23 3.32e3.21 3.29e3.14 3.27e3.08 3.19e2.91 3.16e2.86 3.03e2.62

100% 100% 100% 100% 100% 100% 100% 100% 100%

Statistics for Pirro Nord Pcr ¼ 0.12 3.04  0.01 Pcr ¼ 0.11 2.99  0.01 Pcr ¼ 0.10 2.93  0.01 Pcr ¼ 0.09 2.78  0.01 Pcr ¼ 0.08 2.64  0.03 Pcr ¼ 0.07 2.19  0.05 Pcr ¼ 0.06 1.81  0.07 Pcr ¼ 0.05 1.19  0.11 Pcr ¼ 0.04 1.32  0.10

3.16e2.91 3.10e2.84 3.11e2.69 2.99e2.56 2.96e2.01 2.89e1.39 2.64e0.23 2.12e0.22 1.45e1.18

100% 100% 100% 100% 100% 100% 98% 52% 4%

Statistics for Cueva Victoria Pcr ¼ 0.12 2.96  0.01 2.90  0.01 Pcr ¼ 0.11 Pcr ¼ 0.10 2.82  0.01 Pcr ¼ 0.09 2.67  0.01 Pcr ¼ 0.08 2.56  0.02 Pcr ¼ 0.07 2.30  0.02 Pcr ¼ 0.06 1.97  0.03 Pcr ¼ 0.05 1.53  0.05 Pcr ¼ 0.04 1.02  0.13

3.09e2.78 3.03e2.76 3.01e2.59 2.92e2.48 2.87e2.25 2.74e1.99 2.52e1.55 2.28e0.55 1.61e0.61

100% 100% 100% 100% 100% 100% 100% 90% 12%

Scenario 2: little chance of survival in non-grassland habitats Statistics for Mirzapur Pcr ¼ 0.12 2.89  0.02 Pcr ¼ 0.11 2.77  0.03 Pcr ¼ 0.10 2.59  0.06 Pcr ¼ 0.09 2.28  0.08 Pcr ¼ 0.08 1.51  0.12 Pcr ¼ 0.07 0.88  0.18 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

3.15e2.51 3.07e1.82 3.04e0.80 2.93e0.99 2.90e0.13 2.56e0.17 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 100% 96% 82% 30% 0% 0% 0%

Statistics for Ternifine Pcr ¼ 0.12 2.86  0.06 Pcr ¼ 0.11 2.85  0.07 Pcr ¼ 0.10 2.59  0.11 Pcr ¼ 0.09 2.72  0.06

3.26e1.21 3.29e0.73 3.11e0.57 3.12e1.49

94% 86% 64% 48%

Pcr ¼ 0.08 Pcr ¼ 0.07 Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04

63

Scenario 3: marginal survival in non-grassland habitats Statistics for Mirzapur Pcr ¼ 0.12 2.99  0.01 Pcr ¼ 0.11 2.93  0.02 Pcr ¼ 0.10 2.73  0.02 Pcr ¼ 0.09 2.58  0.04 Pcr ¼ 0.08 2.06  0.09 Pcr ¼ 0.07 1.29  0.15 Pcr ¼ 0.06 0.77  0.21 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

3.15e2.80 3.19e2.57 2.98e2.25 3.01e1.59 3.02e0.52 2.60e0.17 1.29e0.49 0.00e0.00 0.00e0.00

100% 100% 100% 100% 98% 60% 6% 0% 0%

Statistics for Ternifine Pcr ¼ 0.12 3.09  0.02 Pcr ¼ 0.11 2.98  0.03 Pcr ¼ 0.10 2.84  0.05 Pcr ¼ 0.09 2.39  0.11 Pcr ¼ 0.08 2.34  0.14 Pcr ¼ 0.07 2.56  0.12 Pcr ¼ 0.06 2.66  0.10 Pcr ¼ 0.05 2.45  0.00 Pcr ¼ 0.04 0.00  0.00

3.28e2.79 3.28e2.30 3.23e1.23 3.24e0.28 3.15e0.29 3.01e0.57 3.03e2.43 2.45e2.45 0.00e0.00

100% 100% 100% 96% 72% 36% 10% 2% 0%

Statistics for Pirro Nord Pcr ¼ 0.12 0.00  0.00 Pcr ¼ 0.11 0.00  0.00 Pcr ¼ 0.10 0.00  0.00 Pcr ¼ 0.09 0.00  0.00 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

0% 0% 0% 0% 0% 0% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 0.00  0.00 0.00  0.00 Pcr ¼ 0.11 Pcr ¼ 0.10 0.00  0.00 Pcr ¼ 0.09 0.00  0.00 Pcr ¼ 0.08 0.00  0.00

0.00e0.00 0% 0.00e0.00 0% 0.00e0.00 0% 0.00e0.00 0% 0.00e0.00 0% (continued on next page)

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

64 Appendix 1 (continued) Pcr ¼ 0.07 Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04

0.00  0.00 0.00  0.00 0.00  0.00 0.00  0.00

0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

0% 0% 0% 0%

Scenario 4: reasonable survival in non-grassland habitats

Pcr ¼ 0.09 Pcr ¼ 0.08 Pcr ¼ 0.07 Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04

3.25  0.00 3.21  0.00 3.14  0.01 3.04  0.01 2.89  0.03 2.30  0.08

3.33e3.17 3.28e3.13 3.23e2.99 3.17e2.88 3.10e1.93 2.97e0.63

100% 100% 100% 100% 100% 94%

Statistics for Pirro Nord Pcr ¼ 0.12 2.98  0.01 Pcr ¼ 0.11 2.87  0.01 Pcr ¼ 0.10 2.78  0.02 Pcr ¼ 0.09 2.53  0.05 Pcr ¼ 0.08 2.08  0.07 Pcr ¼ 0.07 1.52  0.10 Pcr ¼ 0.06 1.15  0.13 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

3.10e2.84 3.05e2.75 3.01e2.39 2.99e1.36 2.69e0.62 2.39e0.12 1.89e0.43 0.00e0.00 0.00e0.00

100% 100% 100% 100% 92% 68% 26% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 2.88  0.01 2.77  0.01 Pcr ¼ 0.11 Pcr ¼ 0.10 2.67  0.01 Pcr ¼ 0.09 2.50  0.02 Pcr ¼ 0.08 2.26  0.02 Pcr ¼ 0.07 1.96  0.02 Pcr ¼ 0.06 1.40  0.05 Pcr ¼ 0.05 1.10  0.06 Pcr ¼ 0.04 0.00  0.00

2.99e2.73 2.94e2.63 2.90e2.48 2.88e2.20 2.53e1.99 2.25e1.64 1.96e0.42 1.67e0.79 0.00e0.00

100% 100% 100% 100% 100% 100% 100% 28% 0%

Statistics for Mirzapur Pcr ¼ 0.12 3.07  0.01 Pcr ¼ 0.11 3.01  0.01 Pcr ¼ 0.10 2.92  0.02 Pcr ¼ 0.09 2.83  0.02 Pcr ¼ 0.08 2.50  0.04 Pcr ¼ 0.07 2.06  0.07 Pcr ¼ 0.06 1.66  0.12 Pcr ¼ 0.05 1.46  0.32 Pcr ¼ 0.04 0.00  0.00

3.18e2.90 3.16e2.83 3.12e2.66 3.03e2.40 2.91e1.68 2.89e0.80 2.79e0.38 1.92e1.00 0.00e0.00

100% 100% 100% 100% 100% 90% 48% 4% 0%

Statistics for Ternifine Pcr ¼ 0.12 3.29  0.00 Pcr ¼ 0.11 3.25  0.01 Pcr ¼ 0.10 3.18  0.01 Pcr ¼ 0.09 3.10  0.02 Pcr ¼ 0.08 2.95  0.03 2.62  0.05 Pcr ¼ 0.07 Pcr ¼ 0.06 2.06  0.09 Pcr ¼ 0.05 1.45  0.19 Pcr ¼ 0.04 0.00  0.00

3.34e3.18 3.36e3.08 3.27e2.88 3.24e2.72 3.26e2.38 3.20e1.81 3.08e0.24 2.79e0.19 0.00e0.00

100% 100% 100% 100% 100% 100% 100% 40% 0%

Statistics for Pirro Nord Pcr ¼ 0.12 2.76  0.03 Pcr ¼ 0.11 2.56  0.05 Pcr ¼ 0.10 1.78  0.12 Pcr ¼ 0.09 1.65  0.12 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

3.01e1.41 2.93e1.43 2.70e0.81 2.12e1.32 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

100% 96% 44% 12% 0% 0% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 2.58  0.02 2.40  0.03 Pcr ¼ 0.11 Pcr ¼ 0.10 1.94  0.05 Pcr ¼ 0.09 1.34  0.06 Pcr ¼ 0.08 1.24  0.05 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

Statistics for Mirzapur Pcr ¼ 0.12 3.06  0.01 Pcr ¼ 0.11 2.98  0.01 Pcr ¼ 0.10 2.91  0.02 Pcr ¼ 0.09 2.80  0.02 Pcr ¼ 0.08 2.43  0.05 Pcr ¼ 0.07 2.11  0.08 Pcr ¼ 0.06 1.63  0.17 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

3.23e2.89 3.15e2.68 3.15e2.70 3.05e2.42 2.85e1.37 2.90e1.21 2.72e0.44 0.00e0.00 0.00e0.00

100% 100% 100% 100% 100% 86% 30% 0% 0%

2.92e2.22 2.81e1.80 2.51e0.57 2.15e0.76 1.39e1.15 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 100% 78% 8% 0% 0% 0% 0%

Statistics for Ternifine Pcr ¼ 0.12 3.29  0.00 Pcr ¼ 0.11 3.26  0.00 Pcr ¼ 0.10 3.23  0.01 Pcr ¼ 0.09 3.18  0.01 Pcr ¼ 0.08 3.09  0.01 Pcr ¼ 0.07 2.99  0.02 Pcr ¼ 0.06 2.43  0.12 Pcr ¼ 0.05 2.33  0.17 Pcr ¼ 0.04 2.35  0.04

3.34e3.24 3.33e3.18 3.32e3.03 3.29e3.00 3.22e2.82 3.20e2.51 3.12e0.10 2.90e0.56 2.41e2.30

100% 100% 100% 100% 100% 100% 90% 26% 4%

Statistics for Pirro Nord Pcr ¼ 0.12 2.81  0.01 Pcr ¼ 0.11 2.60  0.05 Pcr ¼ 0.10 2.30  0.07 Pcr ¼ 0.09 1.68  0.13 Pcr ¼ 0.08 1.68  0.06 Pcr ¼ 0.07 1.14  0.26 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

2.99e2.62 2.98e0.41 2.86e0.40 2.62e0.25 1.98e1.15 1.86e0.24 0.00e0.00 0.00e0.00 0.00e0.00

100% 98% 90% 56% 32% 12% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 2.68  0.01 2.54  0.02 Pcr ¼ 0.11 Pcr ¼ 0.10 2.36  0.02 Pcr ¼ 0.09 2.12  0.02 Pcr ¼ 0.08 1.70  0.03

2.95e2.49 2.80e2.08 2.63e2.06 2.51e1.69 2.10e0.99

100% 100% 100% 100% 98%

Scenario 6: quite reasonable survival in desert

Scenario 5: very good survival in non-grassland habitats Statistics for Mirzapur Pcr ¼ 0.12 3.13  0.01 Pcr ¼ 0.11 3.08  0.01 Pcr ¼ 0.10 3.02  0.01 Pcr ¼ 0.09 2.92  0.01 Pcr ¼ 0.08 2.79  0.01 Pcr ¼ 0.07 2.65  0.02 Pcr ¼ 0.06 2.08  0.06 Pcr ¼ 0.05 1.38  0.11 Pcr ¼ 0.04 0.00  0.00

3.26e3.00 3.21e2.89 3.16e2.85 3.14e2.68 3.07e2.57 2.88e2.24 2.79e0.92 2.41e0.19 0.00e0.00

100% 100% 100% 100% 100% 100% 100% 62% 0%

Statistics for Ternifine Pcr ¼ 0.12 3.32  0.00 Pcr ¼ 0.11 3.30  0.00 Pcr ¼ 0.10 3.28  0.00

3.35e3.27 3.34e3.26 3.33e3.20

100% 100% 100%

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77 Appendix 1 (continued ) Pcr ¼ 0.07 Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04

1.24  0.05 0.00  0.00 0.00  0.00 0.00  0.00

1.76e0.67 0.00e0.00 0.00e0.00 0.00e0.00

62% 0% 0% 0%

Scenario 7: marginal survival in desert Statistics for Mirzapur Pcr ¼ 0.12 3.02  0.01 Pcr ¼ 0.11 2.93  0.02 Pcr ¼ 0.10 2.82  0.03 Pcr ¼ 0.09 2.66  0.04 Pcr ¼ 0.08 2.07  0.11 Pcr ¼ 0.07 1.28  0.14 Pcr ¼ 0.06 1.72  0.46 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

3.16e2.82 3.12e2.39 3.07e2.03 3.06e1.86 2.94e0.18 2.79e0.29 2.63e0.28 0.00e0.00 0.00e0.00

100% 100% 100% 98% 94% 60% 8% 0% 0%

Statistics for Ternifine Pcr ¼ 0.12 3.28  0.00 Pcr ¼ 0.11 3.24  0.01 Pcr ¼ 0.10 3.21  0.01 Pcr ¼ 0.09 3.16  0.01 Pcr ¼ 0.08 3.07  0.02 2.96  0.02 Pcr ¼ 0.07 Pcr ¼ 0.06 2.81  0.02 Pcr ¼ 0.05 2.68  0.03 Pcr ¼ 0.04 2.42  0.00

3.34e3.22 3.31e3.12 3.31e3.03 3.28e2.90 3.21e2.68 3.16e2.49 3.02e2.45 3.00e2.42 2.42e2.42

100% 100% 100% 100% 100% 98% 68% 34% 2%

Statistics for Pirro Nord Pcr ¼ 0.12 2.72  0.02 Pcr ¼ 0.11 2.48  0.05 Pcr ¼ 0.10 2.10  0.08 Pcr ¼ 0.09 1.66  0.10 Pcr ¼ 0.08 1.31  0.12 Pcr ¼ 0.07 1.35  0.08 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

2.92e2.03 2.83e1.05 2.91e0.29 2.67e0.26 2.20e0.28 1.52e1.18 0.00e0.00 0.00e0.00 0.00e0.00

100% 98% 78% 80% 38% 6% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 2.63  0.01 2.45  0.02 Pcr ¼ 0.11 Pcr ¼ 0.10 2.26  0.02 Pcr ¼ 0.09 2.02  0.03 Pcr ¼ 0.08 1.58  0.05 Pcr ¼ 0.07 1.26  0.10 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

2.83e2.39 2.67e2.08 2.78e1.88 2.48e1.41 2.21e0.45 1.95e0.84 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 100% 98% 94% 22% 0% 0% 0%

Scenario 8: extinction in desert Statistics for Mirzapur Pcr ¼ 0.12 2.58  0.04 Pcr ¼ 0.11 2.38  0.05 Pcr ¼ 0.10 1.98  0.09 Pcr ¼ 0.09 1.40  0.11 Pcr ¼ 0.08 1.06  0.16 Pcr ¼ 0.07 0.72  0.23 Pcr ¼ 0.06 1.05  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

2.93e1.76 2.90e1.45 2.86e0.38 2.62e0.13 2.57e0.18 1.43e0.18 1.05e1.05 0.00e0.00 0.00e0.00

100% 100% 96% 84% 34% 10% 2% 0% 0%

Statistics for Ternifine Pcr ¼ 0.12 3.25  0.00 Pcr ¼ 0.11 3.21  0.01 Pcr ¼ 0.10 3.18  0.01

3.29e3.17 3.31e3.11 3.25e3.08

100% 100% 100%

Pcr ¼ 0.09 Pcr ¼ 0.08 Pcr ¼ 0.07 Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04

65

3.10  0.01 3.03  0.01 2.92  0.02 2.81  0.02 2.66  0.03 0.00  0.00

3.19e2.79 3.17e2.75 3.10e2.57 2.99e2.59 2.83e2.53 0.00e0.00

100% 100% 94% 74% 28% 0%

Statistics for Pirro Nord Pcr ¼ 0.12 2.57  0.04 Pcr ¼ 0.11 2.24  0.07 Pcr ¼ 0.10 2.07  0.07 Pcr ¼ 0.09 1.59  0.09 Pcr ¼ 0.08 1.39  0.13 Pcr ¼ 0.07 1.32  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

2.87e1.80 2.84e1.18 2.78e0.38 2.22e0.23 2.15e0.15 1.32e1.32 0.00e0.00 0.00e0.00 0.00e0.00

100% 98% 88% 68% 36% 2% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 2.53  0.02 2.40  0.02 Pcr ¼ 0.11 Pcr ¼ 0.10 2.22  0.02 Pcr ¼ 0.09 1.94  0.03 Pcr ¼ 0.08 1.57  0.04 Pcr ¼ 0.07 1.12  0.08 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

2.76e2.29 2.69e2.08 2.61e1.86 2.29e1.44 2.05e0.78 1.61e0.61 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 100% 100% 92% 28% 0% 0% 0%

Preferred T. darti model with no dispersal across the Gibraltar Straits Statistics for Pcr ¼ 0.12 Pcr ¼ 0.11 Pcr ¼ 0.10 Pcr ¼ 0.09 Pcr ¼ 0.08 Pcr ¼ 0.07 Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04 Statistics for Pcr ¼ 0.12 Pcr ¼ 0.11 Pcr ¼ 0.10 Pcr ¼ 0.09 Pcr ¼ 0.08 Pcr ¼ 0.07 Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04 Statistics for Pcr ¼ 0.12 Pcr ¼ 0.11 Pcr ¼ 0.10 Pcr ¼ 0.09 Pcr ¼ 0.08 Pcr ¼ 0.07 Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04 Statistics for Pcr ¼ 0.12 Pcr ¼ 0.11 Pcr ¼ 0.10 Pcr ¼ 0.09

Mirzapur         

0.01 0.01 0.01 0.01 0.03 0.05 0.09 0.69 0.00

3.21e2.97 3.21e2.82 3.25e2.71 3.09e2.60 3.04e2.14 2.90e1.29 2.82e0.11 2.20e0.254% 0.00e0.000%

100% 100% 100% 100% 100% 98% 90% 4% 0%

3.31  3.28  3.25  3.20  3.14  3.03  2.68  1.92  1.69  Pirro Nord 2.91  2.76  2.50  1.83  1.64  1.53  1.20  0.00  0.00  Cueva Victoria 2.78  2.63  2.44  2.22 

0.00 0.00 0.01 0.01 0.01 0.02 0.06 0.09 0.21

3.36e3.24 3.34e3.22 3.32e3.12 3.30e3.01 3.24e2.80 3.18e2.47 3.11e1.54 2.91e0.49 2.88e0.13

100% 100% 100% 100% 100% 100% 100% 100% 30%

0.01 0.02 0.05 0.11 0.11 0.09 0.00 0.00 0.00

3.12e2.69 3.01e2.18 2.90e1.52 2.76e0.14 2.49e0.57 2.05e0.95 1.20e1.20 0.00e0.00 0.00e0.00

100% 100% 94% 80% 40% 22% 2% 0% 0%

0.02 0.02 0.02 0.02

3.06e2.57 100% 2.82e2.38 100% 2.73e1.95 100% 2.56e1.76 100% (continued on next page)

3.08 3.04 2.97 2.87 2.74 2.37 1.76 1.22 0.00 Ternifine

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

66 Appendix 1 (continued) Pcr Pcr Pcr Pcr Pcr

¼ ¼ ¼ ¼ ¼

0.08 0.07 0.06 0.05 0.04

1.80 1.33 0.00 0.00 0.00

    

0.04 0.06 0.00 0.00 0.00

2.39e1.08 2.03e0.47 0.00e0.000% 0.00e0.000% 0.00e0.000%

98% 84% 0% 0% 0%

Preferred T. darti model with Gibraltar Straits available as a potential dispersal route Statistics for Mirzapur Pcr ¼ 0.12 3.10  0.01 Pcr ¼ 0.11 3.05  0.01 Pcr ¼ 0.10 2.98  0.01 Pcr ¼ 0.09 2.87  0.02 Pcr ¼ 0.08 2.70  0.02 Pcr ¼ 0.07 2.40  0.06 Pcr ¼ 0.06 1.48  0.09 Pcr ¼ 0.05 0.84  0.32 Pcr ¼ 0.04 0.00  0.00

3.23e2.98 3.23e2.87 3.15e2.72 3.08e2.60 3.01e2.38 2.93e0.97 2.58e0.23 2.52e0.24 0.00e0.00

100% 100% 100% 100% 100% 100% 80% 14% 0%

Statistics for Ternifine Pcr ¼ 0.12 3.30  0.00 Pcr ¼ 0.11 3.28  0.00 Pcr ¼ 0.10 3.25  0.00 Pcr ¼ 0.09 3.21  0.01 Pcr ¼ 0.08 3.14  0.01 Pcr ¼ 0.07 3.03  0.02 Pcr ¼ 0.06 2.64  0.06 Pcr ¼ 0.05 1.83  0.09 Pcr ¼ 0.04 1.72  0.40

3.35e3.24 3.33e3.20 3.31e3.16 3.28e3.07 3.25e2.85 3.21e2.54 3.12e1.68 3.05e0.57 2.86e0.12

100% 100% 100% 100% 100% 100% 100% 94% 12%

Statistics for Pirro Nord Pcr ¼ 0.12 3.03  0.01 Pcr ¼ 0.11 2.95  0.01 Pcr ¼ 0.10 2.84  0.02 Pcr ¼ 0.09 2.61  0.06 Pcr ¼ 0.08 2.04  0.13 Pcr ¼ 0.07 1.81  0.09 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

3.14e2.90 3.13e2.72 3.03e2.16 3.02e0.58 3.01e0.62 2.30e1.24 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 100% 96% 56% 26% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 3.22  0.01 3.19  0.01 Pcr ¼ 0.11 Pcr ¼ 0.10 3.14  0.01 Pcr ¼ 0.09 3.07  0.01 Pcr ¼ 0.08 2.96  0.02 Pcr ¼ 0.07 2.67  0.06 Pcr ¼ 0.06 2.63  0.08 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

3.28e3.11 3.29e3.02 3.27e2.92 3.19e2.90 3.17e2.63 3.09e1.05 2.96e1.02 0.00e0.00 0.00e0.00

100% 100% 100% 100% 100% 100% 44% 0% 0%

Appendix 2a. T. oswaldi mean, standard error, range of arrival dates (Ma), and probabilities of arrival for all scenarios and colonization rates (2.5 Ma climate transition) Ability equal in all habitats, 2.5 Ma transition Statistics for Mirzapur Pcr ¼ 0.12 2.14  0.01 Pcr ¼ 0.11 2.11  0.01 Pcr ¼ 0.10 2.05  0.01 Pcr ¼ 0.09 2.00  0.01 Pcr ¼ 0.08 1.90  0.01 Pcr ¼ 0.07 1.77  0.01

2.21e2.03 2.18e1.99 2.16e1.87 2.11e1.85 2.04e1.72 1.94e1.55

100% 100% 100% 100% 100% 100%

Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04

1.58  0.03 1.06  0.06 0.49  0.09

1.86e0.91 1.75e0.16 0.84e0.17

100% 90% 16%

Statistics for Ternifine Pcr ¼ 0.12 2.33  0.00 Pcr ¼ 0.11 2.31  0.00 Pcr ¼ 0.10 2.29  0.00 Pcr ¼ 0.09 2.27  0.00 Pcr ¼ 0.08 2.24  0.00 Pcr ¼ 0.07 2.18  0.01 Pcr ¼ 0.06 2.13  0.01 Pcr ¼ 0.05 2.04  0.01 Pcr ¼ 0.04 1.86  0.01

2.37e2.29 2.37e2.27 2.34e2.24 2.31e2.17 2.27e2.18 2.27e2.10 2.21e2.01 2.19e1.88 2.09e1.61

100% 100% 100% 100% 100% 100% 100% 100% 100%

Statistics for Pirro Nord Pcr ¼ 0.12 2.04  0.01 Pcr ¼ 0.11 1.98  0.01 Pcr ¼ 0.10 1.89  0.01 Pcr ¼ 0.09 1.79  0.01 Pcr ¼ 0.08 1.58  0.03 Pcr ¼ 0.07 1.20  0.04 Pcr ¼ 0.06 0.89  0.06 Pcr ¼ 0.05 0.64  0.07 Pcr ¼ 0.04 0.00  0.00

2.17e1.91 2.14e1.83 2.06e1.75 2.03e1.67 1.97e0.87 1.71e0.46 1.74e0.12 1.28e0.15 0.00e0.00

100% 100% 100% 100% 100% 96% 84% 30% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.97  0.01 1.90  0.01 Pcr ¼ 0.11 Pcr ¼ 0.10 1.79  0.01 Pcr ¼ 0.09 1.68  0.01 Pcr ¼ 0.08 1.52  0.02 Pcr ¼ 0.07 1.28  0.02 Pcr ¼ 0.06 1.05  0.03 Pcr ¼ 0.05 0.57  0.03 Pcr ¼ 0.04 0.56  0.00

2.11e1.84 2.05e1.74 1.95e1.64 1.89e1.51 1.88e1.33 1.57e1.10 1.53e0.72 1.02e0.10 0.56e0.56

100% 100% 100% 100% 100% 100% 100% 78% 2%

Little chance of survival in non-grassland habitats, 2.5 Ma transition Statistics for Mirzapur Pcr ¼ 0.12 1.91  0.02 Pcr ¼ 0.11 1.80  0.03 Pcr ¼ 0.10 1.58  0.05 Pcr ¼ 0.09 1.20  0.08 Pcr ¼ 0.08 1.21  0.09 Pcr ¼ 0.07 0.88  0.19 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

2.19e1.44 2.13e0.66 1.91e0.41 1.85e0.27 1.73e0.16 1.56e0.29 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 98% 78% 58% 10% 0% 0% 0%

Statistics for Ternifine Pcr ¼ 0.12 1.91  0.04 Pcr ¼ 0.11 1.84  0.06 Pcr ¼ 0.10 1.85  0.03 Pcr ¼ 0.09 1.77  0.04 Pcr ¼ 0.08 1.70  0.02 Pcr ¼ 0.07 1.69  0.05 Pcr ¼ 0.06 1.59  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

2.27e0.41 2.24e0.27 2.07e1.58 2.19e0.85 1.88e1.54 1.81e1.59 1.59e1.59 0.00e0.00 0.00e0.00

94% 72% 64% 56% 42% 6% 2% 0% 0%

Statistics for Pirro Nord Pcr ¼ 0.12 0.00  0.00 Pcr ¼ 0.11 0.00  0.00 Pcr ¼ 0.10 0.00  0.00 Pcr ¼ 0.09 0.00  0.00 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00

0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

0% 0% 0% 0% 0% 0% 0%

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77 Appendix 2a (continued )

1.46  0.04 0.76  0.07 1.32  0.14 0.00  0.00

1.85e0.59 1.75e0.11 1.52e1.13 0.00e0.00

100% 68% 4% 0%

0% 0% 0% 0% 0% 0% 0% 0% 0%

Statistics for Ternifine Pcr ¼ 0.12 2.31  0.00 Pcr ¼ 0.11 2.28  0.00 Pcr ¼ 0.10 2.24  0.00 Pcr ¼ 0.09 2.21  0.01 Pcr ¼ 0.08 2.13  0.01 Pcr ¼ 0.07 1.99  0.03 Pcr ¼ 0.06 1.79  0.04 Pcr ¼ 0.05 1.15  0.09 Pcr ¼ 0.04 0.92  0.26

2.37e2.22 2.34e2.22 2.29e2.16 2.29e2.03 2.24e1.78 2.25e1.42 2.16e1.08 2.06e0.20 1.66e0.12

100% 100% 100% 100% 100% 100% 100% 84% 10%

Marginal survival in non-grassland habitats, 2.5 Ma transition

Statistics for Pirro Nord Pcr ¼ 0.12 1.87  0.01 Pcr ¼ 0.11 1.78  0.02 Pcr ¼ 0.10 1.65  0.03 Pcr ¼ 0.09 1.22  0.07 Pcr ¼ 0.08 0.72  0.08 Pcr ¼ 0.07 0.40  0.13 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.99e1.67 2.00e1.22 1.98e0.90 1.88e0.40 1.21e0.18 1.03e0.11 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 96% 76% 36% 12% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.75  0.01 1.62  0.02 Pcr ¼ 0.11 Pcr ¼ 0.10 1.49  0.02 Pcr ¼ 0.09 1.25  0.03 Pcr ¼ 0.08 0.88  0.03 Pcr ¼ 0.07 0.51  0.04 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.90e1.54 1.87e1.27 1.77e1.07 1.66e0.44 1.22e0.40 0.85e0.16 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 100% 100% 96% 52% 0% 0% 0%

Pcr ¼ 0.05 Pcr ¼ 0.04

0.00  0.00 0.00  0.00

0.00e0.00 0.00e0.00

0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 0.00  0.00 Pcr ¼ 0.11 0.00  0.00 Pcr ¼ 0.10 0.00  0.00 Pcr ¼ 0.09 0.00  0.00 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

Pcr ¼ 0.07 Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04

67

Statistics for Mirzapur Pcr ¼ 0.12 1.99  0.01 Pcr ¼ 0.11 1.91  0.01 Pcr ¼ 0.10 1.79  0.02 Pcr ¼ 0.09 1.49  0.06 Pcr ¼ 0.08 1.27  0.07 Pcr ¼ 0.07 0.97  0.12 Pcr ¼ 0.06 1.33  0.14 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

2.21e1.75 2.12e1.64 2.11e1.39 2.00e0.32 1.86e0.14 1.75e0.12 1.53e1.13 0.00e0.00 0.00e0.00

100% 100% 100% 100% 88% 42% 4% 0% 0%

Statistics for Ternifine Pcr ¼ 0.12 2.08  0.02 Pcr ¼ 0.11 2.03  0.02 Pcr ¼ 0.10 1.82  0.06 Pcr ¼ 0.09 1.50  0.09 Pcr ¼ 0.08 1.71  0.05 Pcr ¼ 0.07 1.60  0.05 Pcr ¼ 0.06 1.64  0.05 Pcr ¼ 0.05 1.54  0.00 Pcr ¼ 0.04 0.00  0.00

2.35e1.70 2.28e1.64 2.31e0.55 2.22e0.13 2.15e0.57 1.99e1.02 1.79e1.45 1.54e1.54 0.00e0.00

100% 100% 100% 92% 52% 30% 12% 2% 0%

Statistics for Pirro Nord Pcr ¼ 0.12 0.00  0.00 Pcr ¼ 0.11 0.00  0.00 Pcr ¼ 0.10 0.00  0.00 Pcr ¼ 0.09 0.00  0.00 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

0% 0% 0% 0% 0% 0% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 0.00  0.00 0.00  0.00 Pcr ¼ 0.11 Pcr ¼ 0.10 0.00  0.00 Pcr ¼ 0.09 0.00  0.00 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

Statistics for Mirzapur Pcr ¼ 0.12 2.12  0.01 Pcr ¼ 0.11 2.08  0.01 Pcr ¼ 0.10 2.02  0.01 Pcr ¼ 0.09 1.93  0.01 Pcr ¼ 0.08 1.81  0.02 Pcr ¼ 0.07 1.60  0.02 Pcr ¼ 0.06 1.12  0.04 Pcr ¼ 0.05 0.77  0.10 Pcr ¼ 0.04 0.00  0.00

2.21e1.99 2.19e1.89 2.18e1.84 2.09e1.76 1.98e1.52 1.96e1.07 1.75e0.27 1.77e0.19 0.00e0.00

100% 100% 100% 100% 100% 100% 94% 38% 0%

0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

0% 0% 0% 0% 0% 0% 0% 0% 0%

Statistics for Ternifine Pcr ¼ 0.12 2.32  0.00 Pcr ¼ 0.11 2.30  0.00 Pcr ¼ 0.10 2.27  0.00 Pcr ¼ 0.09 2.25  0.00 Pcr ¼ 0.08 2.21  0.00 Pcr ¼ 0.07 2.14  0.01 Pcr ¼ 0.06 2.05  0.01 Pcr ¼ 0.05 1.81  0.02 Pcr ¼ 0.04 1.26  0.07

2.37e2.27 2.36e2.25 2.32e2.22 2.30e2.17 2.29e2.14 2.23e1.94 2.18e1.70 2.00e1.22 1.88e0.19

100% 100% 100% 100% 100% 100% 100% 100% 86%

Good survival in non-grassland habitats, 2.5 Ma transition Statistics for Mirzapur Pcr ¼ 0.12 2.10  0.01 Pcr ¼ 0.11 2.03  0.01 Pcr ¼ 0.10 1.98  0.01 Pcr ¼ 0.09 1.85  0.02 Pcr ¼ 0.08 1.71  0.02

2.23e1.99 2.16e1.85 2.19e1.81 2.07e1.46 2.03e1.26

100% 100% 100% 100% 100%

Very good survival in non-grassland habitats, 2.5 Ma transition

Statistics for Pirro Nord Pcr ¼ 0.12 1.97  0.01 Pcr ¼ 0.11 1.90  0.01 Pcr ¼ 0.10 1.77  0.01 Pcr ¼ 0.09 1.57  0.05 Pcr ¼ 0.08 1.24  0.06

2.16e1.81 100% 2.06e1.72 100% 1.98e1.61 100% 1.94e0.19 100% 1.87e0.21 90% (continued on next page)

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

68 Appendix 2a (continued) Pcr ¼ 0.07 Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04

0.84  0.05 0.41  0.07 0.18  0.00 0.00  0.00

1.65e0.24 0.78e0.17 0.18e0.18 0.00e0.00

68% 14% 2% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.87  0.01 Pcr ¼ 0.11 1.79  0.01 Pcr ¼ 0.10 1.65  0.01 Pcr ¼ 0.09 1.52  0.02 Pcr ¼ 0.08 1.30  0.02 Pcr ¼ 0.07 0.94  0.03 Pcr ¼ 0.06 0.49  0.03 Pcr ¼ 0.05 0.17  0.00 Pcr ¼ 0.04 0.00  0.00

2.04e1.73 2.01e1.53 1.94e1.47 1.81e1.16 1.73e0.87 1.50e0.52 0.90e0.11 0.17e0.17 0.00e0.00

100% 100% 100% 100% 100% 100% 94% 2% 0%

Quite reasonable survival in desert, 2.5 Ma transition

Pcr ¼ 0.09 Pcr ¼ 0.08 Pcr ¼ 0.07 Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04

1.63  0.04 1.31  0.07 0.93  0.12 1.03  0.31 0.00  0.00 0.00  0.00

2.13e0.70 1.85e0.19 1.94e0.11 1.48e0.59 0.00e0.00 0.00e0.00

100% 86% 44% 4% 0% 0%

Statistics for Ternifine Pcr ¼ 0.12 2.26  0.01 Pcr ¼ 0.11 2.22  0.01 Pcr ¼ 0.10 2.16  0.01 Pcr ¼ 0.09 2.06  0.02 Pcr ¼ 0.08 1.96  0.02 Pcr ¼ 0.07 1.83  0.03 Pcr ¼ 0.06 1.74  0.04 Pcr ¼ 0.05 1.66  0.04 Pcr ¼ 0.04 0.00  0.00

2.33e2.09 2.31e2.06 2.28e1.88 2.24e1.63 2.19e1.57 2.11e1.25 2.05e1.52 1.89e1.53 0.00e0.00

100% 100% 100% 100% 92% 64% 30% 14% 0%

Statistics for Pirro Nord Pcr ¼ 0.12 1.56  0.05 Pcr ¼ 0.11 1.26  0.07 Pcr ¼ 0.10 0.90  0.08 Pcr ¼ 0.09 0.81  0.09 Pcr ¼ 0.08 0.71  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.88e0.48 1.73e0.41 1.76e0.10 1.27e0.48 0.71e0.71 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

94% 66% 48% 16% 2% 0% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.46  0.02 Pcr ¼ 0.11 1.22  0.04 Pcr ¼ 0.10 0.93  0.04 Pcr ¼ 0.09 0.60  0.05 Pcr ¼ 0.08 0.45  0.03 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.73e1.01 1.61e0.57 1.46e0.24 1.11e0.12 0.55e0.40 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 100% 60% 8% 0% 0% 0% 0%

Statistics for Mirzapur Pcr ¼ 0.12 2.04  0.01 Pcr ¼ 0.11 1.96  0.01 Pcr ¼ 0.10 1.90  0.01 Pcr ¼ 0.09 1.70  0.03 Pcr ¼ 0.08 1.41  0.05 Pcr ¼ 0.07 1.09  0.08 Pcr ¼ 0.06 0.97  0.28 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

2.18e1.89 2.12e1.74 2.07e1.70 2.02e0.95 1.90e0.45 1.76e0.28 1.54e0.31 0.00e0.00 0.00e0.00

Statistics for Ternifine Pcr ¼ 0.12 2.28  0.01 Pcr ¼ 0.11 2.25  0.01 Pcr ¼ 0.10 2.18  0.01 Pcr ¼ 0.09 2.09  0.02 Pcr ¼ 0.08 1.96  0.03 Pcr ¼ 0.07 1.61  0.07 Pcr ¼ 0.06 1.40  0.11 Pcr ¼ 0.05 1.34  0.34 Pcr ¼ 0.04 0.00  0.00

2.34e2.12 2.31e2.11 2.29e1.94 2.29e1.75 2.21e0.98 2.18e0.33 2.07e0.24 1.90e0.20 0.00e0.00

100% 100% 100% 100% 100% 94% 54% 8% 0%

Statistics for Pirro Nord Pcr ¼ 0.12 1.72  0.02 Pcr ¼ 0.11 1.43  0.06 Pcr ¼ 0.10 1.03  0.11 Pcr ¼ 0.09 0.82  0.12 Pcr ¼ 0.08 0.60  0.19 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.93e1.25 1.86e0.40 1.73e0.16 1.61e0.32 0.86e0.34 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

96% 90% 40% 18% 4% 0% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.53  0.02 1.32  0.03 Pcr ¼ 0.11 Pcr ¼ 0.10 0.98  0.03 Pcr ¼ 0.09 0.64  0.06 Pcr ¼ 0.08 0.32  0.07 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

Statistics for Mirzapur Pcr ¼ 0.12 1.56  0.06 Pcr ¼ 0.11 1.34  0.06 Pcr ¼ 0.10 1.08  0.06 Pcr ¼ 0.09 0.95  0.11 Pcr ¼ 0.08 0.56  0.13 Pcr ¼ 0.07 0.51  0.05 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

2.01e0.12 1.94e0.25 1.83e0.11 1.88e0.12 1.04e0.17 0.58e0.44 0.00e0.00 0.00e0.00 0.00e0.00

100% 96% 90% 48% 14% 4% 0% 0% 0%

1.83e1.22 1.60e0.59 1.40e0.49 1.20e0.12 0.48e0.19 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 100% 54% 6% 0% 0% 0% 0%

Statistics for Ternifine Pcr ¼ 0.12 2.23  0.00 Pcr ¼ 0.11 2.19  0.01 Pcr ¼ 0.10 2.12  0.01 Pcr ¼ 0.09 2.00  0.02 Pcr ¼ 0.08 1.90  0.02 Pcr ¼ 0.07 1.81  0.03 Pcr ¼ 0.06 1.76  0.05 Pcr ¼ 0.05 1.56  0.04 Pcr ¼ 0.04 0.00  0.00

2.29e2.12 2.26e2.06 2.24e1.94 2.20e1.66 2.11e1.56 2.09e1.45 1.99e1.56 1.68e1.47 0.00e0.00

100% 100% 100% 100% 88% 60% 20% 10% 0%

Statistics for Pirro Nord Pcr ¼ 0.12 1.49  0.04 Pcr ¼ 0.11 0.99  0.08 Pcr ¼ 0.10 0.99  0.09 Pcr ¼ 0.09 0.69  0.08 Pcr ¼ 0.08 0.00  0.00

1.82e0.71 1.66e0.16 1.62e0.46 1.02e0.34 0.00e0.00

88% 60% 28% 20% 0%

100% 100% 100% 100% 96% 66% 8% 0% 0%

Extinction in desert, 2.5 Ma transition

Marginal survival in desert, 2.5 Ma transition Statistics for Mirzapur Pcr ¼ 0.12 2.01  0.01 Pcr ¼ 0.11 1.90  0.02 Pcr ¼ 0.10 1.80  0.03

2.23e1.84 2.13e1.55 2.08e1.12

100% 100% 100%

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77 Appendix 2b (continued ) Pcr ¼ 0.07 Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04

0.00  0.00 0.00  0.00 0.00  0.00 0.00  0.00

0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

0% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.41  0.02 Pcr ¼ 0.11 1.15  0.03 Pcr ¼ 0.10 0.85  0.04 Pcr ¼ 0.09 0.56  0.04 Pcr ¼ 0.08 0.60  0.07 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.65e1.12 1.51e0.61 1.31e0.23 1.04e0.12 0.76e0.51 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 98% 70% 6% 0% 0% 0% 0%

Appendix 2b. T. oswaldi mean, standard error, range of arrival dates (Ma),and probabilities of arrival for all scenarios and colonization rates (1.8 Ma climate transition) Ability equal in all habitats, 1.8 Ma transition Statistics for Mirzapur Pcr ¼ 0.12 2.15  0.01 Pcr ¼ 0.11 2.10  0.01 Pcr ¼ 0.10 2.05  0.01 Pcr ¼ 0.09 2.00  0.01 Pcr ¼ 0.08 1.90  0.01 Pcr ¼ 0.07 1.78  0.02 Pcr ¼ 0.06 1.52  0.03 Pcr ¼ 0.05 1.16  0.03 Pcr ¼ 0.04 0.86  0.07

2.23e2.05 2.18e1.93 2.18e1.93 2.13e1.72 2.05e1.73 1.98e1.43 1.92e0.94 1.63e0.57 1.54e0.38

100% 100% 100% 100% 100% 100% 100% 100% 38%

Statistics for Ternifine Pcr ¼ 0.12 2.33  0.00 Pcr ¼ 0.11 2.32  0.00 Pcr ¼ 0.10 2.29  0.00 Pcr ¼ 0.09 2.27  0.00 Pcr ¼ 0.08 2.23  0.01 Pcr ¼ 0.07 2.19  0.01 Pcr ¼ 0.06 2.12  0.01 Pcr ¼ 0.05 2.02  0.01 Pcr ¼ 0.04 1.85  0.01

2.38e2.30 2.37e2.26 2.34e2.24 2.32e2.20 2.33e2.07 2.28e2.06 2.21e1.99 2.12e1.91 1.97e1.63

100% 100% 100% 100% 100% 100% 100% 100% 100%

Statistics for Pirro Nord Pcr ¼ 0.12 2.02  0.01 Pcr ¼ 0.11 1.97  0.01 Pcr ¼ 0.10 1.87  0.01 Pcr ¼ 0.09 1.80  0.01 Pcr ¼ 0.08 1.68  0.02 Pcr ¼ 0.07 1.48  0.02 Pcr ¼ 0.06 1.16  0.03 Pcr ¼ 0.05 0.67  0.08 Pcr ¼ 0.04 0.95  0.00

2.16e1.91 2.11e1.77 2.07e1.70 2.03e1.48 2.01e1.39 1.89e1.09 1.65e0.45 1.05e0.17 0.95e0.95

100% 100% 100% 100% 100% 100% 92% 32% 2%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.94  0.01 Pcr ¼ 0.11 1.89  0.01 Pcr ¼ 0.10 1.78  0.01 Pcr ¼ 0.09 1.69  0.01 Pcr ¼ 0.08 1.55  0.02 Pcr ¼ 0.07 1.33  0.02 Pcr ¼ 0.06 0.97  0.03 Pcr ¼ 0.05 0.50  0.04 Pcr ¼ 0.04 0.38  0.09

2.10e1.82 2.05e1.73 1.99e1.60 1.95e1.34 1.90e1.30 1.67e0.95 1.52e0.38 0.95e0.14 0.62e0.13

100% 100% 100% 100% 100% 100% 100% 68% 8%

69

Little chance of survival in non-grassland habitats, 1.8 Ma transition Statistics for Mirzapur Pcr ¼ 0.12 1.90  0.02 Pcr ¼ 0.11 1.82  0.02 Pcr ¼ 0.10 1.61  0.04 Pcr ¼ 0.09 1.51  0.04 Pcr ¼ 0.08 1.24  0.04 Pcr ¼ 0.07 0.92  0.08 Pcr ¼ 0.06 0.84  0.04 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

2.15e1.45 2.07e1.34 2.03e0.89 1.90e0.72 1.72e0.44 1.75e0.11 0.96e0.72 0.00e0.00 0.00e0.00

100% 100% 100% 100% 82% 44% 8% 0% 0%

Statistics for Ternifine Pcr ¼ 0.12 1.87  0.05 Pcr ¼ 0.11 1.76  0.05 Pcr ¼ 0.10 1.73  0.04 Pcr ¼ 0.09 1.52  0.05 Pcr ¼ 0.08 1.51  0.06 Pcr ¼ 0.07 1.38  0.07 Pcr ¼ 0.06 1.22  0.10 Pcr ¼ 0.05 1.00  0.12 Pcr ¼ 0.04 0.00  0.00

2.29e0.56 2.26e0.91 2.20e0.91 2.03e1.00 1.96e0.92 1.98e0.74 1.57e0.78 1.19e0.71 0.00e0.00

100% 98% 88% 80% 48% 34% 14% 6% 0%

Statistics for Pirro Nord Pcr ¼ 0.12 0.00  0.00 Pcr ¼ 0.11 0.00  0.00 Pcr ¼ 0.10 0.00  0.00 Pcr ¼ 0.09 0.00  0.00 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

0% 0% 0% 0% 0% 0% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 0.00  0.00 0.00  0.00 Pcr ¼ 0.11 Pcr ¼ 0.10 0.00  0.00 Pcr ¼ 0.09 0.00  0.00 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

0% 0% 0% 0% 0% 0% 0% 0% 0%

Marginal survival in non-grassland habitats, 1.8 Ma transition Statistics for Mirzapur Pcr ¼ 0.12 1.99  0.01 Pcr ¼ 0.11 1.93  0.02 Pcr ¼ 0.10 1.77  0.02 Pcr ¼ 0.09 1.63  0.03 Pcr ¼ 0.08 1.31  0.06 Pcr ¼ 0.07 1.21  0.06 Pcr ¼ 0.06 0.83  0.06 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00 Statistics for Ternifine Pcr ¼ 0.12 2.12  0.02 Pcr ¼ 0.11 2.01  0.03 Pcr ¼ 0.10 1.78  0.04 Pcr ¼ 0.09 1.76  0.04 Pcr ¼ 0.08 1.55  0.06 Pcr ¼ 0.07 1.38  0.06 Pcr ¼ 0.06 1.21  0.05 Pcr ¼ 0.05 1.16  0.07

2.16e1.79 2.15e1.63 2.05e1.33 2.00e0.91 1.86e0.11 1.87e0.49 1.29e0.48 0.00e0.00 0.00e0.00

100% 100% 100% 100% 100% 70% 28% 0% 0%

2.33e1.72 100% 2.29e1.37 100% 2.20e0.99 100% 2.15e0.81 96% 2.11e0.30 80% 1.86e0.74 58% 1.57e0.76 28% 1.41e0.94 14% (continued on next page)

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

70 Appendix 2b (continued) Pcr ¼ 0.04

Pcr ¼ 0.04

0.79  0.00

0.79e0.79

2%

Statistics for Pirro Nord Pcr ¼ 0.12 1.07  0.00 Pcr ¼ 0.11 0.00  0.00 Pcr ¼ 0.10 0.00  0.00 Pcr ¼ 0.09 0.00  0.00 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.07e1.07 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

2% 0% 0% 0% 0% 0% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 0.00  0.00 Pcr ¼ 0.11 0.00  0.00 Pcr ¼ 0.10 0.00  0.00 Pcr ¼ 0.09 0.00  0.00 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

0% 0% 0% 0% 0% 0% 0% 0% 0%

Good survival in non-grassland habitats, 1.8 Ma transition Statistics for Mirzapur Pcr ¼ 0.12 2.10  0.01 Pcr ¼ 0.11 2.05  0.01 Pcr ¼ 0.10 1.96  0.01 Pcr ¼ 0.09 1.87  0.02 Pcr ¼ 0.08 1.73  0.02 Pcr ¼ 0.07 1.40  0.04 Pcr ¼ 0.06 0.96  0.05 Pcr ¼ 0.05 0.71  0.13 Pcr ¼ 0.04 0.00  0.00

2.20e1.91 2.21e1.87 2.15e1.76 2.12e1.61 2.07e1.21 1.76e0.46 1.54e0.25 1.30e0.15 0.00e0.00

100% 100% 100% 100% 100% 100% 94% 18% 0%

Statistics for Ternifine Pcr ¼ 0.12 2.31  0.00 Pcr ¼ 0.11 2.28  0.00 Pcr ¼ 0.10 2.25  0.01 Pcr ¼ 0.09 2.21  0.01 Pcr ¼ 0.08 2.15  0.01 Pcr ¼ 0.07 1.97  0.03 Pcr ¼ 0.06 1.81  0.05 Pcr ¼ 0.05 1.26  0.08 Pcr ¼ 0.04 1.20  0.11

2.35e2.24 2.33e2.20 2.30e2.05 2.31e2.05 2.30e1.78 2.17e1.49 2.15e0.40 1.99e0.16 1.68e0.42

100% 100% 100% 100% 100% 100% 100% 74% 22%

Statistics for Pirro Nord Pcr ¼ 0.12 1.87  0.01 Pcr ¼ 0.11 1.78  0.01 Pcr ¼ 0.10 1.65  0.02 Pcr ¼ 0.09 1.46  0.02 Pcr ¼ 0.08 1.20  0.02 Pcr ¼ 0.07 1.10  0.04 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

2.07e1.73 1.99e1.54 1.91e1.41 1.88e1.19 1.54e0.94 1.26e0.98 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 100% 100% 94% 10% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.76  0.01 1.62  0.02 Pcr ¼ 0.11 Pcr ¼ 0.10 1.46  0.02 Pcr ¼ 0.09 1.17  0.03 Pcr ¼ 0.08 0.76  0.03 Pcr ¼ 0.07 0.32  0.03 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00

1.91e1.58 1.87e1.33 1.78e1.05 1.52e0.73 1.20e0.20 0.63e0.12 0.00e0.00 0.00e0.00

100% 100% 100% 100% 100% 38% 0% 0%

0.00  0.00

0.00e0.00

0%

Very good survival in non-grassland habitats, 1.8 Ma transition Statistics for Mirzapur - India Pcr ¼ 0.12 2.12  0.01 Pcr ¼ 0.11 2.08  0.01 Pcr ¼ 0.10 2.02  0.01 Pcr ¼ 0.09 1.94  0.01 Pcr ¼ 0.08 1.83  0.01 Pcr ¼ 0.07 1.64  0.02 Pcr ¼ 0.06 1.32  0.03 Pcr ¼ 0.05 0.79  0.06 Pcr ¼ 0.04 0.84  0.00

2.20e2.01 2.19e1.95 2.13e1.93 2.11e1.70 2.11e1.63 1.91e1.25 1.66e0.52 1.49e0.18 0.84e0.84

100% 100% 100% 100% 100% 100% 100% 68% 2%

Statistics for Ternifine Pcr ¼ 0.12 2.32  0.00 Pcr ¼ 0.11 2.30  0.00 Pcr ¼ 0.10 2.28  0.00 Pcr ¼ 0.09 2.26  0.00 Pcr ¼ 0.08 2.20  0.01 Pcr ¼ 0.07 2.15  0.01 Pcr ¼ 0.06 2.05  0.01 Pcr ¼ 0.05 1.81  0.03 Pcr ¼ 0.04 1.33  0.07

2.35e2.27 2.35e2.25 2.33e2.20 2.31e2.17 2.30e2.09 2.26e2.02 2.17e1.73 2.04e1.01 1.88e0.12

100% 100% 100% 100% 100% 100% 100% 100% 90%

Statistics for Pirro Nord Pcr ¼ 0.12 1.95  0.01 Pcr ¼ 0.11 1.88  0.01 Pcr ¼ 0.10 1.80  0.01 Pcr ¼ 0.09 1.68  0.01 Pcr ¼ 0.08 1.48  0.02 Pcr ¼ 0.07 1.22  0.02 Pcr ¼ 0.06 0.88  0.07 Pcr ¼ 0.05 0.46  0.00 Pcr ¼ 0.04 0.00  0.00

2.13e1.80 2.06e1.74 2.03e1.64 1.93e1.45 1.77e1.24 1.69e0.81 1.38e0.13 0.46e0.46 0.00e0.00

100% 100% 100% 100% 100% 98% 42% 2% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.86  0.01 1.78  0.01 Pcr ¼ 0.11 Pcr ¼ 0.10 1.67  0.01 Pcr ¼ 0.09 1.52  0.02 Pcr ¼ 0.08 1.29  0.02 Pcr ¼ 0.07 0.92  0.03 Pcr ¼ 0.06 0.53  0.03 Pcr ¼ 0.05 0.31  0.00 Pcr ¼ 0.04 0.00  0.00

2.08e1.70 1.95e1.62 1.94e1.49 1.81e1.28 1.61e0.81 1.55e0.17 1.04e0.12 0.31e0.31 0.00e0.00

100% 100% 100% 100% 100% 100% 80% 2% 0%

Quite reasonable survival in desert, 1.8 Ma transition Statistics for Mirzapur Pcr ¼ 0.12 2.03  0.01 Pcr ¼ 0.11 2.00  0.01 Pcr ¼ 0.10 1.90  0.02 Pcr ¼ 0.09 1.72  0.03 Pcr ¼ 0.08 1.57  0.03 Pcr ¼ 0.07 1.22  0.05 Pcr ¼ 0.06 1.01  0.06 Pcr ¼ 0.05 0.90  0.08 Pcr ¼ 0.04 0.00  0.00

2.22e1.86 2.17e1.84 2.11e1.52 2.04e1.26 1.96e1.11 1.84e0.15 1.57e0.31 1.10e0.76 0.00e0.00

100% 100% 100% 100% 100% 94% 54% 6% 0%

Statistics for Ternifine Pcr ¼ 0.12 2.28  0.01 Pcr ¼ 0.11 2.23  0.01 Pcr ¼ 0.10 2.18  0.02 Pcr ¼ 0.09 2.13  0.02 Pcr ¼ 0.08 1.92  0.03 Pcr ¼ 0.07 1.62  0.05

2.35e2.16 2.32e1.92 2.29e1.84 2.28e1.54 2.21e1.50 2.13e0.48

100% 100% 100% 100% 100% 96%

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77 Appendix 2b (continued ) Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04

1.55  0.05 1.18  0.07 1.27  0.11

2.01e1.02 1.88e0.48 1.42e1.12

66% 46% 4%

Pcr ¼ 0.05 Pcr ¼ 0.04

0.00  0.00 0.00  0.00

71 0.00e0.00 0.00e0.00

0% 0%

Extinction in desert, 1.8 Ma transition

Statistics for Pirro Nord Pcr ¼ 0.12 1.73  0.01 Pcr ¼ 0.11 1.62  0.02 Pcr ¼ 0.10 1.40  0.02 Pcr ¼ 0.09 1.09  0.02 Pcr ¼ 0.08 1.11  0.03 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.90e1.47 1.92e1.29 1.75e1.01 1.41e0.47 1.16e1.04 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 100% 78% 6% 0% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.46  0.02 Pcr ¼ 0.11 1.33  0.03 Pcr ¼ 0.10 0.95  0.04 Pcr ¼ 0.09 0.52  0.05 Pcr ¼ 0.08 0.16  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

Statistics for Mirzapur Pcr ¼ 0.12 1.60  0.04 Pcr ¼ 0.11 1.51  0.04 Pcr ¼ 0.10 1.25  0.05 Pcr ¼ 0.09 1.17  0.04 Pcr ¼ 0.08 0.92  0.07 Pcr ¼ 0.07 0.83  0.12 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

2.07e0.88 1.97e0.54 1.93e0.42 1.77e0.81 1.48e0.24 1.26e0.36 0.00e0.00 0.00e0.00 0.00e0.00

100% 98% 92% 74% 52% 14% 0% 0% 0%

1.77e1.02 1.65e0.72 1.51e0.20 1.14e0.11 0.16e0.16 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 100% 58% 2% 0% 0% 0% 0%

Statistics for Ternifine Pcr ¼ 0.12 2.23  0.00 Pcr ¼ 0.11 2.17  0.01 Pcr ¼ 0.10 2.13  0.01 Pcr ¼ 0.09 2.03  0.02 Pcr ¼ 0.08 1.88  0.03 Pcr ¼ 0.07 1.67  0.04 Pcr ¼ 0.06 1.37  0.06 Pcr ¼ 0.05 1.26  0.06 Pcr ¼ 0.04 1.47  0.13

2.30e2.12 2.31e1.97 2.22e1.92 2.21e1.69 2.14e1.41 2.08e0.90 1.95e0.92 1.72e0.95 1.66e1.27

100% 100% 100% 100% 100% 94% 60% 28% 4%

Statistics for Pirro Nord Pcr ¼ 0.12 1.59  0.02 Pcr ¼ 0.11 1.45  0.02 Pcr ¼ 0.10 1.22  0.03 Pcr ¼ 0.09 0.95  0.06 Pcr ¼ 0.08 0.66  0.22 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.88e1.30 1.78e1.04 1.60e0.93 1.55e0.17 0.97e0.35 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

100% 98% 98% 56% 4% 0% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.33  0.03 1.10  0.04 Pcr ¼ 0.11 Pcr ¼ 0.10 0.84  0.04 Pcr ¼ 0.09 0.51  0.05 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.72e0.76 1.70e0.45 1.32e0.11 1.08e0.11 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 98% 42% 0% 0% 0% 0% 0%

Marginal survival in desert, 1.8 Ma transition Statistics for Mirzapur - India Pcr ¼ 0.12 1.99  0.02 Pcr ¼ 0.11 1.92  0.02 Pcr ¼ 0.10 1.82  0.02 Pcr ¼ 0.09 1.70  0.03 Pcr ¼ 0.08 1.40  0.04 Pcr ¼ 0.07 1.07  0.08 Pcr ¼ 0.06 0.82  0.09 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

2.19e1.66 2.16e1.52 2.11e1.46 2.13e1.21 1.89e0.76 1.76e0.10 1.26e0.13 0.00e0.00 0.00e0.00

100% 100% 100% 100% 98% 68% 24% 0% 0%

Statistics for Ternifine Pcr ¼ 0.12 2.26  0.01 Pcr ¼ 0.11 2.22  0.01 Pcr ¼ 0.10 2.18  0.01 Pcr ¼ 0.09 2.08  0.02 Pcr ¼ 0.08 1.87  0.03 Pcr ¼ 0.07 1.71  0.04 Pcr ¼ 0.06 1.46  0.06 Pcr ¼ 0.05 1.24  0.09 Pcr ¼ 0.04 0.83  0.00

2.33e2.13 2.31e2.00 2.28e1.96 2.26e1.67 2.20e1.21 2.10e1.06 2.01e0.86 1.85e0.80 0.83e0.83

100% 100% 100% 100% 100% 98% 72% 30% 2%

Statistics for Pirro Nord Pcr ¼ 0.12 1.65  0.02 Pcr ¼ 0.11 1.50  0.02 Pcr ¼ 0.10 1.32  0.03 Pcr ¼ 0.09 1.08  0.04 Pcr ¼ 0.08 0.83  0.15 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.90e1.16 1.80e1.07 1.66e0.92 1.59e0.16 1.02e0.18 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 94% 72% 10% 0% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.41  0.03 1.18  0.03 Pcr ¼ 0.11 Pcr ¼ 0.10 0.75  0.04 Pcr ¼ 0.09 0.46  0.03 Pcr ¼ 0.08 0.31  0.10 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00

1.82e0.91 1.53e0.57 1.23e0.17 0.84e0.11 0.55e0.16 0.00e0.00 0.00e0.00

100% 100% 98% 74% 6% 0% 0%

Appendix 2c. T. oswaldi mean, standard error, range of arrival dates (Ma), and probabilities of arrival for all scenarios and colonization rates (1.0 Ma climate transition) Ability equal in all habitats, 1.0 Ma transition Statistics for Mirzapur Pcr ¼ 0.12 2.15  0.01 Pcr ¼ 0.11 2.11  0.01 Pcr ¼ 0.10 2.05  0.01 Pcr ¼ 0.09 1.99  0.01 Pcr ¼ 0.08 1.91  0.01 Pcr ¼ 0.07 1.76  0.02 Pcr ¼ 0.06 1.58  0.03 Pcr ¼ 0.05 1.14  0.04 Pcr ¼ 0.04 0.63  0.04

2.24e2.07 2.20e2.02 2.19e1.87 2.10e1.81 2.05e1.75 1.98e1.28 1.81e0.95 1.67e0.29 1.18e0.23

100% 100% 100% 100% 100% 100% 100% 100% 46%

(continued on next page)

72

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

Appendix 2c (continued)

Pcr ¼ 0.04

0.00  0.00

0.00e0.00

0%

Statistics for Ternifine Pcr ¼ 0.12 2.33  0.00 Pcr ¼ 0.11 2.32  0.00 Pcr ¼ 0.10 2.29  0.00 Pcr ¼ 0.09 2.27  0.00 Pcr ¼ 0.08 2.23  0.01 Pcr ¼ 0.07 2.19  0.01 Pcr ¼ 0.06 2.12  0.01 Pcr ¼ 0.05 2.03  0.01 Pcr ¼ 0.04 1.85  0.01

2.36e2.30 2.34e2.28 2.37e2.23 2.32e2.22 2.31e2.13 2.27e2.06 2.23e1.97 2.13e1.92 2.01e1.54

100% 100% 100% 100% 100% 100% 100% 100% 100%

Statistics for Cueva Victoria Pcr ¼ 0.12 0.00  0.00 Pcr ¼ 0.11 0.00  0.00 Pcr ¼ 0.10 0.00  0.00 Pcr ¼ 0.09 0.00  0.00 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

0% 0% 0% 0% 0% 0% 0% 0% 0%

Statistics for Pirro Nord Pcr ¼ 0.12 2.01  0.01 Pcr ¼ 0.11 1.96  0.01 Pcr ¼ 0.10 1.91  0.01 Pcr ¼ 0.09 1.79  0.01 Pcr ¼ 0.08 1.67  0.02 Pcr ¼ 0.07 1.49  0.03 Pcr ¼ 0.06 1.17  0.03 Pcr ¼ 0.05 0.72  0.04 Pcr ¼ 0.04 0.27  0.05

2.18e1.90 2.15e1.80 2.12e1.68 2.03e1.58 1.93e1.44 1.89e1.12 1.72e0.51 1.47e0.14 0.40e0.13

100% 100% 100% 100% 100% 100% 100% 94% 8%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.93  0.01 Pcr ¼ 0.11 1.87  0.01 Pcr ¼ 0.10 1.81  0.01 Pcr ¼ 0.09 1.70  0.01 Pcr ¼ 0.08 1.55  0.02 Pcr ¼ 0.07 1.33  0.03 Pcr ¼ 0.06 0.99  0.03 Pcr ¼ 0.05 0.52  0.04 Pcr ¼ 0.04 0.00  0.00

2.10e1.83 2.03e1.71 2.05e1.58 1.93e1.48 1.86e1.32 1.64e0.94 1.62e0.35 1.19e0.16 0.00e0.00

100% 100% 100% 100% 100% 100% 100% 88% 0%

Little chance of survival in non-grassland habitats, 1.0 Ma transition Statistics for Mirzapur Pcr ¼ 0.12 1.92  0.02 Pcr ¼ 0.11 1.82  0.02 Pcr ¼ 0.10 1.69  0.03 Pcr ¼ 0.09 1.48  0.04 Pcr ¼ 0.08 1.14  0.06 Pcr ¼ 0.07 0.78  0.06 Pcr ¼ 0.06 0.62  0.07 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

2.18e1.56 2.07e1.46 2.07e0.96 2.05e0.59 1.85e0.14 1.54e0.16 0.95e0.15 0.00e0.00 0.00e0.00

100% 100% 100% 100% 96% 78% 22% 0% 0%

Statistics for Ternifine Pcr ¼ 0.12 1.91  0.04 Pcr ¼ 0.11 1.78  0.04 Pcr ¼ 0.10 1.51  0.06 Pcr ¼ 0.09 1.40  0.07 Pcr ¼ 0.08 1.25  0.08 Pcr ¼ 0.07 1.11  0.08 Pcr ¼ 0.06 0.93  0.07 Pcr ¼ 0.05 0.81  0.21 Pcr ¼ 0.04 0.16  0.00

2.30e0.84 2.25e0.93 2.06e0.51 2.01e0.29 2.07e0.13 1.77e0.25 1.50e0.24 1.36e0.24 0.16e0.16

100% 100% 100% 92% 80% 60% 44% 10% 2%

Statistics for Pirro Nord Pcr ¼ 0.12 0.00  0.00 Pcr ¼ 0.11 0.00  0.00 Pcr ¼ 0.10 0.00  0.00 Pcr ¼ 0.09 0.00  0.00 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00

0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

0% 0% 0% 0% 0% 0% 0% 0%

Marginal survival in non-grassland habitats, 1.0 Ma transition Statistics for Mirzapur Pcr ¼ 0.12 1.98  0.01 Pcr ¼ 0.11 1.91  0.02 Pcr ¼ 0.10 1.80  0.02 Pcr ¼ 0.09 1.62  0.03 Pcr ¼ 0.08 1.39  0.03 Pcr ¼ 0.07 1.01  0.06 Pcr ¼ 0.06 0.69  0.08 Pcr ¼ 0.05 0.39  0.00 Pcr ¼ 0.04 0.00  0.00

2.16e1.78 2.18e1.58 2.00e1.49 1.96e0.78 1.82e0.85 1.78e0.12 1.25e0.29 0.39e0.39 0.00e0.00

100% 100% 100% 100% 98% 98% 30% 2% 0%

Statistics for Ternifine Pcr ¼ 0.12 2.08  0.02 Pcr ¼ 0.11 1.97  0.03 Pcr ¼ 0.10 1.87  0.03 Pcr ¼ 0.09 1.63  0.06 Pcr ¼ 0.08 1.38  0.06 Pcr ¼ 0.07 1.23  0.07 Pcr ¼ 0.06 0.90  0.07 Pcr ¼ 0.05 1.14  0.12 Pcr ¼ 0.04 0.30  0.04

2.29e1.65 2.30e1.34 2.23e1.12 2.12e0.53 2.12e0.40 2.02e0.18 1.57e0.37 1.65e0.47 0.35e0.25

100% 100% 100% 100% 100% 84% 54% 22% 4%

Statistics for Pirro Nord Pcr ¼ 0.12 0.00  0.00 Pcr ¼ 0.11 0.00  0.00 Pcr ¼ 0.10 0.00  0.00 Pcr ¼ 0.09 0.00  0.00 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

0% 0% 0% 0% 0% 0% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 0.00  0.00 0.00  0.00 Pcr ¼ 0.11 Pcr ¼ 0.10 0.00  0.00 Pcr ¼ 0.09 0.00  0.00 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

0% 0% 0% 0% 0% 0% 0% 0% 0%

Good survival in non-grassland habitats, 1.0 Ma transition Statistics for Mirzapur Pcr ¼ 0.12 2.10  0.01 Pcr ¼ 0.11 2.03  0.01 Pcr ¼ 0.10 1.97  0.02 Pcr ¼ 0.09 1.85  0.02 Pcr ¼ 0.08 1.68  0.02

2.25e1.98 2.16e1.87 2.16e1.70 2.05e1.61 1.92e1.16

100% 100% 100% 100% 100%

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77 Appendix 2c (continued )

73

1.45  0.04 1.00  0.06 0.44  0.07 0.36  0.00

1.90e0.52 1.70e0.22 1.13e0.14 0.36e0.36

100% 94% 28% 2%

Pcr ¼ 0.08 Pcr ¼ 0.07 Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04

1.45  0.02 1.23  0.02 0.67  0.04 0.40  0.09 0.00  0.00

1.72e1.13 1.57e0.86 1.16e0.14 0.73e0.12 0.00e0.00

Statistics for Ternifine Pcr ¼ 0.12 2.31  0.00 Pcr ¼ 0.11 2.28  0.00 Pcr ¼ 0.10 2.25  0.01 Pcr ¼ 0.09 2.20  0.01 Pcr ¼ 0.08 2.12  0.01 Pcr ¼ 0.07 2.03  0.02 Pcr ¼ 0.06 1.66  0.05 Pcr ¼ 0.05 1.22  0.08 Pcr ¼ 0.04 1.00  0.17

2.36e2.22 2.34e2.21 2.32e2.13 2.29e1.97 2.26e1.69 2.20e1.55 2.10e0.91 1.98e0.26 1.72e0.18

100% 100% 100% 100% 100% 100% 100% 80% 24%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.86  0.01 Pcr ¼ 0.11 1.75  0.01 Pcr ¼ 0.10 1.68  0.01 Pcr ¼ 0.09 1.52  0.02 Pcr ¼ 0.08 1.25  0.02 Pcr ¼ 0.07 0.96  0.03 Pcr ¼ 0.06 0.45  0.04 Pcr ¼ 0.05 0.61  0.00 Pcr ¼ 0.04 0.00  0.00

2.01e1.69 1.93e1.60 1.87e1.46 1.85e1.16 1.49e0.96 1.32e0.40 0.86e0.10 0.61e0.61 0.00e0.00

Statistics for Pirro Nord Pcr ¼ 0.12 1.88  0.01 Pcr ¼ 0.11 1.76  0.01 Pcr ¼ 0.10 1.66  0.02 Pcr ¼ 0.09 1.46  0.02 Pcr ¼ 0.08 1.17  0.03 Pcr ¼ 0.07 0.76  0.04 Pcr ¼ 0.06 0.35  0.09 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.98e1.73 2.02e1.57 1.95e1.43 1.69e1.18 1.70e0.70 1.44e0.16 0.53e0.16 0.00e0.00 0.00e0.00

100% 100% 100% 100% 100% 96% 6% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.77  0.01 Pcr ¼ 0.11 1.60  0.02 Pcr ¼ 0.10 1.46  0.02 Pcr ¼ 0.09 1.13  0.03 Pcr ¼ 0.08 0.79  0.05 Pcr ¼ 0.07 0.32  0.04 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.88e1.57 1.89e1.36 1.83e1.10 1.54e0.66 1.56e0.19 0.68e0.13 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 100% 100% 86% 28% 0% 0% 0%

Pcr ¼ 0.07 Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04

Very good survival in non-grassland habitats, 1.0 Ma transition Statistics for Mirzapur Pcr ¼ 0.12 2.13  0.01 Pcr ¼ 0.11 2.07  0.01 Pcr ¼ 0.10 2.01  0.01 Pcr ¼ 0.09 1.93  0.01 Pcr ¼ 0.08 1.81  0.02 Pcr ¼ 0.07 1.60  0.02 Pcr ¼ 0.06 1.25  0.04 Pcr ¼ 0.05 0.70  0.05 Pcr ¼ 0.04 0.00  0.00 Statistics for Ternifine Pcr ¼ 0.12 2.32  0.00 Pcr ¼ 0.11 2.30  0.00 Pcr ¼ 0.10 2.27  0.00 Pcr ¼ 0.09 2.24  0.00 Pcr ¼ 0.08 2.20  0.00 Pcr ¼ 0.07 2.16  0.01 Pcr ¼ 0.06 2.04  0.01 Pcr ¼ 0.05 1.80  0.03 Pcr ¼ 0.04 1.19  0.08 Statistics for Pirro Nord Pcr ¼ 0.12 1.96  0.01 Pcr ¼ 0.11 1.86  0.01 Pcr ¼ 0.10 1.79  0.01 Pcr ¼ 0.09 1.66  0.02

2.22e2.01 2.16e1.95 2.15e1.79 2.10e1.68 1.99e1.42 1.81e1.30 1.80e0.45 1.25e0.12 0.00e0.00 2.36e2.25 2.35e2.25 2.32e2.23 2.31e2.20 2.25e2.13 2.25e2.07 2.16e1.80 2.03e0.91 1.97e0.12

100% 100% 100% 100% 100% 100% 100% 78% 0% 100% 100% 100% 100% 100% 100% 100% 100% 82%

100% 100% 92% 12% 0% 100% 100% 100% 100% 100% 100% 64% 2% 0%

Quite reasonable survival in desert, 1.0 Ma transition Statistics for Mirzapur Pcr ¼ 0.12 2.06  0.01 Pcr ¼ 0.11 1.97  0.01 Pcr ¼ 0.10 1.89  0.01 Pcr ¼ 0.09 1.79  0.02 Pcr ¼ 0.08 1.50  0.03 Pcr ¼ 0.07 1.19  0.05 Pcr ¼ 0.06 0.82  0.06 Pcr ¼ 0.05 0.79  0.06 Pcr ¼ 0.04 0.00  0.00

2.23e1.90 2.19e1.70 2.07e1.65 2.04e1.44 1.83e1.05 1.87e0.51 1.52e0.24 0.88e0.70 0.00e0.00

100% 100% 100% 100% 100% 100% 82% 4% 0%

Statistics for Ternifine Pcr ¼ 0.12 2.27  0.01 Pcr ¼ 0.11 2.23  0.01 Pcr ¼ 0.10 2.17  0.01 Pcr ¼ 0.09 2.07  0.02 Pcr ¼ 0.08 1.91  0.03 Pcr ¼ 0.07 1.64  0.06 Pcr ¼ 0.06 1.27  0.08 Pcr ¼ 0.05 0.94  0.10 Pcr ¼ 0.04 0.95  0.10

2.36e2.13 2.32e2.08 2.30e1.84 2.26e1.60 2.22e1.29 2.14e0.64 2.05e0.14 1.79e0.37 1.49e0.43

100% 100% 100% 100% 100% 100% 90% 46% 18%

Statistics for Pirro Nord Pcr ¼ 0.12 1.80  0.01 Pcr ¼ 0.11 1.60  0.02 Pcr ¼ 0.10 1.42  0.02 Pcr ¼ 0.09 1.01  0.04 Pcr ¼ 0.08 0.52  0.04 Pcr ¼ 0.07 0.40  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.97e1.60 1.93e1.35 1.84e1.12 1.70e0.17 0.89e0.18 0.40e0.40 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 100% 100% 60% 2% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.58  0.02 1.23  0.04 Pcr ¼ 0.11 Pcr ¼ 0.10 0.90  0.04 Pcr ¼ 0.09 0.53  0.09 Pcr ¼ 0.08 0.11  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.86e1.16 1.64e0.24 1.49e0.24 1.05e0.15 0.11e0.11 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 96% 28% 2% 0% 0% 0% 0%

Marginal survival in desert, 1.0 Ma transition 2.12e1.80 2.04e1.68 2.04e1.56 1.96e1.25

100% 100% 100% 100%

Statistics for Mirzapur Pcr ¼ 0.12 2.02  0.02 Pcr ¼ 0.11 1.93  0.02

2.22e1.58 100% 2.15e1.64 100% (continued on next page)

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

74 Appendix 2c (continued) Pcr ¼ 0.10 Pcr ¼ 0.09 Pcr ¼ 0.08 Pcr ¼ 0.07 Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04

1.83  0.02 1.64  0.03 1.37  0.04 0.92  0.06 0.68  0.10 0.90  0.11 0.00  0.00

2.06e1.33 2.06e0.94 1.92e0.73 1.80e0.20 1.57e0.10 1.06e0.73 0.00e0.00

100% 100% 100% 88% 38% 4% 0%

Statistics for Ternifine Pcr ¼ 0.12 2.26  0.01 Pcr ¼ 0.11 2.21  0.01 Pcr ¼ 0.10 2.16  0.01 Pcr ¼ 0.09 2.08  0.02 Pcr ¼ 0.08 1.92  0.03 Pcr ¼ 0.07 1.69  0.05 Pcr ¼ 0.06 1.29  0.07 Pcr ¼ 0.05 0.98  0.11 Pcr ¼ 0.04 0.51  0.04

2.35e2.08 2.34e1.93 2.29e1.87 2.29e1.77 2.17e1.32 2.10e0.76 1.98e0.20 1.67e0.24 0.60e0.42

100% 100% 100% 100% 100% 100% 82% 44% 6%

Statistics for Pirro Nord Pcr ¼ 0.12 1.70  0.02 Pcr ¼ 0.11 1.52  0.02 Pcr ¼ 0.10 1.32  0.03 Pcr ¼ 0.09 0.92  0.04 Pcr ¼ 0.08 0.53  0.06 Pcr ¼ 0.07 0.13  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.96e1.47 1.71e1.18 1.74e0.90 1.49e0.34 0.97e0.15 0.13e0.13 0.00e0.00 0.00e0.00 0.00e0.00

100% 100% 100% 96% 40% 2% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.49  0.02 Pcr ¼ 0.11 1.19  0.04 Pcr ¼ 0.10 0.85  0.05 Pcr ¼ 0.09 0.41  0.06 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.86e1.12 1.60e0.59 1.40e0.10 0.80e0.11 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

100% 98% 78% 26% 0% 0% 0% 0% 0%

Extinction in desert, 1.0 Ma transition Statistics for Mirzapur Pcr ¼ 0.12 1.50  0.05 Pcr ¼ 0.11 1.48  0.04 Pcr ¼ 0.10 1.21  0.05 Pcr ¼ 0.09 1.04  0.06 Pcr ¼ 0.08 0.80  0.07 Pcr ¼ 0.07 0.51  0.10 Pcr ¼ 0.06 0.51  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.99e0.30 1.93e0.46 1.92e0.32 1.73e0.14 1.69e0.12 1.03e0.10 0.51e0.51 0.00e0.00 0.00e0.00

100% 100% 100% 82% 50% 20% 2% 0% 0%

Statistics for Ternifine Pcr ¼ 0.12 2.21  0.00 Pcr ¼ 0.11 2.17  0.01 Pcr ¼ 0.10 2.14  0.01 Pcr ¼ 0.09 2.02  0.02 Pcr ¼ 0.08 1.87  0.03 Pcr ¼ 0.07 1.57  0.06 Pcr ¼ 0.06 1.17  0.07 Pcr ¼ 0.05 0.93  0.10 Pcr ¼ 0.04 1.17  0.19

2.26e2.13 2.27e1.93 2.24e1.74 2.16e1.64 2.12e1.37 2.03e0.34 1.93e0.36 1.66e0.13 1.55e0.73

100% 100% 100% 100% 100% 96% 90% 54% 6%

Statistics for Pirro Nord Pcr ¼ 0.12 1.58  0.02 Pcr ¼ 0.11 1.42  0.02 Pcr ¼ 0.10 1.14  0.04

1.89e1.27 1.71e0.96 1.77e0.46

100% 100% 100%

Pcr ¼ 0.09 Pcr ¼ 0.08 Pcr ¼ 0.07 Pcr ¼ 0.06 Pcr ¼ 0.05 Pcr ¼ 0.04

0.84  0.04 0.43  0.05 0.00  0.00 0.00  0.00 0.00  0.00 0.00  0.00

1.42e0.22 1.11e0.15 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

98% 42% 0% 0% 0% 0%

Statistics for Cueva Victoria Pcr ¼ 0.12 1.36  0.03 Pcr ¼ 0.11 1.08  0.04 Pcr ¼ 0.10 0.79  0.06 Pcr ¼ 0.09 0.60  0.07 Pcr ¼ 0.08 0.00  0.00 Pcr ¼ 0.07 0.00  0.00 Pcr ¼ 0.06 0.00  0.00 Pcr ¼ 0.05 0.00  0.00 Pcr ¼ 0.04 0.00  0.00

1.73e0.89 1.52e0.21 1.56e0.10 0.99e0.20 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00 0.00e0.00

100% 96% 80% 28% 0% 0% 0% 0% 0%

References Agusti, J., Anto´n, M., 2002. Mammoths, Sabretooths and Hominids. Columbia University Press, New York. Agusti, J., Moya`-Sola`, S., Pons-Moya`, J., 1986. Venta Micena (Guadix-Baza basin, south-eastern Spain): its place in the Plio-Pleistocene mammal succession in Europe. Geol. Rom. 25, 33e62. Alemseged, Z., Geraads, D., 1998. Theropithecus atlanticus (Thomas, 1884) (Primates: Cercopithecidae) from the late Pliocene of Ahl al Oughlam, Casablanca, Morocco. J. Hum. Evol. 34, 609e621. Anto´n, S.C., Leonard, W.R., Robertson, M.L., 2002. An ecomorphological model of the initial hominid dispersal from Africa. J. Hum. Evol. 43, 773e785. Ascenzi, A., Mallegni, F., Manzi, G., Segre, A.G., Segre Naldini, E., 2000. A re-appraisal of Ceprano calvaria affinities with Homo erectus, after the new reconstruction. J. Hum. Evol. 39, 443e450. Azzaroli, A., Napoleone, G., 1982. Magnetostratigraphic investigation in the Upper Siwaliks near Pinjor, India. Riv. It. Palaeontol. Strat. 87, 739e762. Barry, J.C., 1987. The history and chronology of Siwalik Cercopithecids. Hum. Evol. 2, 47e58. Belmaker, M., 2002. First evidence of Theropithecus sp. in the Southern Levant. Israel. J. Zool. 48, 165. Benefit, B.R., 1999a. Victoriapithecus: the key to Old World monkey and catarrhine origins. Evol. Anthropol. 7, 155e174. Benefit, B., 1999b. Biogeography, dietary specialization, and the diversification of African Plio-Pleistocene monkeys. In: Bromage, T.G., Schrenk, F. (Eds.), African Biogeography, Climate Change and Evolution. Oxford University Press, Oxford, pp. 172e188. Bobe´, R., Behrensmeyer, A.K., 2004. The expansion of grassland ecosystems in Africa in relation to mammalian evolution and the origin of the genus Homo. Palaeogeogr. Palaeoclimatol. Palaeoecol. 207, 399e420. Brain, C.K., 1993. Swartkrans: A Cave’s Chronicle of Early Man. In: Transvaal Museum Monograph, No. 8. Transvaal Museum, Pretoria. Cadman, A., Rayner, R.J., 1989. Climatic change and the appearance of Australopithecus africanus in the Makapansgat sediments. J. Hum. Evol. 18, 107e113. Cerling, T.E., Hay, R.L., 1988. An isotopic study of paleosol carbonates from Olduvai Gorge. Quatern. Res. 25, 63e78. Codron, D., Luyt, J., Lee-Thorp, J.A., Sponheimer, M., de Ruiter, D., Codron, J., 2005. Utilizations of savanna-based resources by Plio-Pleistocene baboons. S. Afr. J. Sci. 101, 245e248. Delson, E., 1984. Cercopithecid biochronology of the African Plio-Pleistocene: correlation among eastern and southern hominid-bearing localities. Cour. Forsch.-Inst. Senckenberg 69, 199e218. Delson, E., 1993. Theropithecus fossils from Africa and India and the taxonomy of the genus. In: Jablonski, N.G. (Ed.), Theropithecus: The Rise and Fall of a Primate Genus. Cambridge University Press, Cambridge, pp. 157e189.

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77 Delson, E., 1994. Evolutionary history of the colobine monkeys in palaeoenvironmental perspective. In: Davies, A.G., Oates, J.F. (Eds.), Colobine Monkeys. Cambridge University Press, Cambridge, pp. 11e43. Delson, E., Dean, D., 1993. Are Papio baringensis (R. Leakey, 1969), and Papio quadratirostris (Iwamoto, 1982), species of Papio or Theropithecus? In: Jablonski, N.G. (Ed.), Theropithecus: The Rise and Fall of a Primate Genus. Cambridge University Press, Cambridge, pp. 125e156. Delson, E., Eck, G.G., Leakey, M.G., Jablonski, N.G., 1993. A partial catalogue of fossil remains of Theropithecus. In: Jablonski, N.G. (Ed.), Theropithecus: The Rise and Fall of a Primate Genus. Cambridge University Press, Cambridge, pp. 499e525. Delson, E., Terranova, C.J., Jungers, W.L., Sargis, E.J., Jablonski, N.G., Dechow, P.C., 2000. Body mass in Cercopithecidae (Primates, Mammalia): estimation and scaling in extinct and extant taxa. Am. Museum Nat. Hist. Anthropol. Papers 83, 1e159. DeMenocal, P.B., 2004. African climate change and faunal evolution during the Pliocene-Pleistocene. Earth Planet. Sci. Lett. 220, 3e24. Dennell, R., 2003. Dispersal and colonisation, long and short chronologies: how continuous is the Early Pleistocene record for hominids outside East Africa? J. Hum. Evol. 45, 421e440. Dennell, R., Roebroeks, W., 2005. An Asian perspective on early human dispersal from Africa. Nature 438, 1099e1104. Dodonov, A.E., Baiguzina, L.L., 1995. Loess stratigraphy of Central Asia: Palaeoclimatic and palaeoenvironmental aspects. Quatern. Sci. Rev. 14, 707e720. Dowsett, H.J., Barron, J.A., Poore, R.Z., Thompson, R.S., Cronin, T.M., Ishman, S.E., Willard, D.A., 1999. Middle Pliocene paleoenvironmental reconstruction: PRISM2. U.S. Geol. Surv., Reston, Va., Open File Rep. 99e535. Eck, G.G., 1993. Theropithecus darti from Hadar. In: Jablonski, N.G. (Ed.), Theropithecus: The Rise and Fall of a Primate Genus. Cambridge University Press, Cambridge, pp. 15e83. Eck, G.G., Jablonski, N.G., 1984. A reassessment of the taxonomic status and phyletic relationships of Papio baringensis and Papio quadratirostris (Primates: Cercopithecidae). Am. J. Phys. Anthropol. 65, 109e134. Eck, G.G., Jablonski, N.G., 1987. The skull of Theropithecus brumpti compared with those of other species of the genus Theropithecus. In: Coppens, Y., Howell, F.C. (Eds.), Les Faunes Plio-Pleistocenes de la Bas Vallee de l’Omo (Ethiopie), Cercopithecidae de la Formation de Shungura. Cahiers de Paleontologie, Tome 3. Editions du Centre National de la Recherche Scientifique, Paris, pp. 11e122. Elton, S. 2000. Ecomorphology and evolutionary biology of African cercopithecoids: providing an ecological context for hominin evolution. Unpublished Ph.D. Dissertation, University of Cambridge, UK. Elton, S., 2001. Locomotor and habitat classification of cercopithecoid postcranial material from Sterkfontein Member 4, Bolt’s Farm and Swartkrans Members 1 and 2, S. Afr. Palaeontol. Africana 37, 115e126. Elton, S., 2002. A reappraisal of the locomotion and habitat preference of Theropithecus oswaldi. Folia Primatol. 73, 252e280. Elton, S., 2006. Forty years on and still going strong: the use of the hominincercopithecid comparison in palaeoanthropology. J.R. Anthropol. Inst. 12, 19e38. Elton, S., 2007. Environmental correlates of the cercopithecoid radiations. Folia Primatol 78, 344e364. Elton, S., Barham, L., Andrews, P., Smith, G.H.S., 2003. Pliocene femur of Theropithecus from the Luangwa Valley, Zambia. J. Hum. Evol. 44, 133e139. Falgue`res, C., Bahain, J.-J., Yokoyama, Y., Arsuaga, J.L., Bermudex de Castro, J.M., Carbonell, E., Bischoff, J.L., Dolo, J.-M., 1999. Earliest humans in Europe: the age of TD6 Gran Dolina, Atapuerca, Spain. J. Hum. Evol. 37, 343e352. Foley, R.A., 1993. Comparative evolutionary biology of Theropithecus and the Hominidae. In: Jablonski, N.G. (Ed.), Theropithecus: The Rise and Fall of a Primate Genus. Cambridge University Press, Cambridge, pp. 254e270. Frost, S., Delson, E., 2002. Fossil Cercopithecidae from the Hadar Formation and surrounding areas, Pliocene of Ethiopia. J. Hum. Evol. 43, 687e748. Gabunia, L., Vekua, A., Lordkipanidze, D., Swisher III, C.C., Ferring, R., Justus, A., Nioradze, M., Tvalchrelidze, M., Anto´n, S.C., Bosinski, G., Jo¨ris, O., de Lumley, M.-A., Majsuradze, G., Mouskhelishvili, A., 2000.

75

Earliest Pleistocene hominid cranial remains from Dmanisi, Republic of Georgia: taxonomy, geological setting and age. Science 288, 1019e1025. Geraads, D., 1997. Carnivores du Plioce`ne terminal de Ahl al Oughlam (Casablanca, Maroc). Ge´obios 30, 127e164. Gibert, J., Ribot, F., Gibert, L., Leakey, M., Arribas, A., Martinez, B., 1995. Presence of the Cercopithecid genus Theropithecus in Cueva Victoria (Murcia, Spain). J. Hum. Evol. 28, 487e493. Gordon, C., Cooper, C., Senior, C.A., Banks, H., Gregory, J.M., Johns, T.C., Mitchell, J.F.B., Wood, R.A., 2000. The simulation of SST, sea ice extents and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments. Clim. Dyn. 16, 147e168. Goudie, A.S., 2005. The drainage of Africa since the cretaceous. Geomorphology 67, 437e456. Gupta, V.I., Sahni, A., 1981. Theropithecus delsoni, a new cercopithecine species from the Upper Siwaliks of India. Bull. Indian Geol. Assoc. 14, 69e71. Harris, E.E., 2000. Molecular systematics of the Old World monkey tribe Papionini: analysis of the total available genetic sequences. J. Hum. Evol. 38, 235e256. Haywood, A.M., Valdes, P.J., Francis, J.E., Sellwood, B.W., 2002. Global middle Pliocene biome reconstruction: a data/model synthesis. Geochem. Geophys., Geosyst. 3 (12), 1072. Horowitz, A., 1989. Continuous pollen diagrams for the last 3.5 My from Israel- vegetation, climate and correlation with the oxygen isotope record. Palaeogeogr. Palaeoclimatol. Palaeoecol. 72, 63e78. Hughes, J.K., Haywood, A., Mithen, S., Sellwood, B., Valdes, P.J., 2007. Investigating early hominin dispersal patterns: developing a framework for climate data integration. J. Hum. Evol. 53, 465e474. Iwamoto, M., 1982. A fossil baboon skull from the lower Omo Basin, southwestern Ethiopia. Primates 23, 533e541. Iwamoto, T., 1993. Food digestion and energetic conditions in Theropithecus gelada. In: Jablonski, N.G. (Ed.), Theropithecus: The Rise and Fall of a Primate Genus. Cambridge University Press, Cambridge, pp. 453e462. Jablonski, N.G., 1986. The hand of Theropithecus brumpti. In: Else, J.G., Lee, P.C. (Eds.), Primate Evolution, Proc. 10th Congr. Int. Primatol. Soc. Cambridge University Press, Cambridge, Volume 1, pp. 173e182. Jablonski, N.G., 1994. New fossil cercopithecid remains from the Humpata Plateau, southern Angola. Am. J. Phys. Anthropol. 94, 435e464. Jablonski, N.G., 1998. The evolution of the doucs and snub-nosed monkey and the question of the phyletic unity of the odd-nosed colobines. In: Jablonski, N.G. (Ed.), The Natural History of the Doucs and Snub-nosed Monkeys. World Scientific, Singapore, pp. 13e52. Jablonski, N.G., 2002. Fossil Old World monkeys: the late Neogene radiation. In: Hartwig, W.C. (Ed.), The Primate Fossil Record. Cambridge University Press, Cambridge, pp. 255e299. Jin, C.-Z., Kawamura, Y., Taruno, H., 1999. Pliocene and Early Pleistocene insectivore and rodent faunas from Dajushan, Qipanshan and Haimao in North China and the reconstruction of the faunal succession from the Late Miocene to Middle Pleistocene. J. Geosci., Osaka City University 41, 1e19. Jolly, C.J., 1972. The classification and natural history of Theropithecus (Simopithecus) (Andrews, 1916), baboons of the African Plio-Pleistocene. Bull. Br. Mus. (Nat. Hist.) Geol. 22, 1e123. Jolly, D., Prentice, I.C., Bonnefille, R., Ballouche, A., Bengo, M., Brenac, P., Buchet, G., Burney, D., Cazet, J.-P., Cheddadi, R., Edorh, T., Elenga, H., Elmoutaki, S., Guiot, J., Laarif, F., Lamb, H., Lezine, A.-M., Maley, J., Mbenza, M., Peyron, O., Reille, M., Reynaud-Farrera, I., Riollet, G., Ritchie, J., Roche, E., Scott, L., Semmanda, I., Straka, H., Umer, M., Van Campo, E., Vilimumbalo, S., Vincens, A., Waller, M., 1998. Biome reconstruction from pollen and plant macrofossil data for Africa and the Arabian peninsula at 0 and 6000 years. J. Biogeogr. 25, 1007e1027. Joos, F., Gerber, S., Prentice, I.C., Otto-Bliesner, B.L., Valdes, P.J., 2004. Transient simulations of Holocene atmospheric carbon dioxide and terrestrial carbon since the Last Glacial Maximum. Global Biogeochem. Cycles 18, doi:10.1029/2003GB002156. Joussaume, S., Taylor, K.E., Braconnot, P., Mitchell, J.F.B., Kutzbach, J.E., Harrison, S.P., Prentice, I.C., Broccoli, A.J., Abe-Ouchi, A., Bartlein, P.J., Bonfils, C., Dong, B., Guiot, J., Herterich, K.,

76

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77

Hewitt, C.D., Jolly, D., Kim, J.W., Kislov, A., Kitoh, A., Loutre, M.F., Masson, V., McAvaney, B., McFarlane, N., de Noblet, N., Peltier, W.R., Peterschmitt, J.Y., Pollard, D., Rind, D., Royer, J.F., Schlesinger, M.E., Syktus, J., Thompson, S., Valdes, P., Vettoretti, G., Webb, R.S., Wyputta, U., 1999. Monsoon changes for 6000 years ago: results of 18 simulations from the Paleoclimate Modeling Intercomparison Project (PMIP). Geophys. Res. Lett. 26, 859e862. Kaplan, J.O., Bigelow, N.H., Prentice, I.C., Harrison, S.P., Bartlein, P.J., Christensen, T.R., Cramer, W., Matveyeva, N.V., McGuire, A.D., Murray, D.F., Razzhivin, V.Y., Smith, B., Walker, D.A., Anderson, P.M., Andreev, A.A., Brubaker, L.B., Edwards, M.E., Lozhkin, A.V., 2003. Climate change and Arctic ecosystems: 2. Modeling, paleodata-model comparisons, and future projections. J. Geophys. Res. -Atmospheres 108 (D19). , art. no. 8171. Krentz, H.B., 1993. Postcranial anatomy of extant and extinct species of Theropithecus. In: Jablonski, N.G. (Ed.), Theropithecus: The Rise and Fall of a Primate Genus. Cambridge University Press, Cambridge, pp. 383e422. Kuman, K., Clarke, R.J., 2000. Stratigraphy, artefact industries and hominid associations for Sterkfontein Member 5. J. Hum. Evol. 38, 827e847. Larick, R., Ciochon, R.L., Zaim, Y., Suminto, S., Izal, Y., Aziz, F., Reagan, M., Heizler, M., 2001. Early Pleistocene Ar-40/Ar-39 ages for Bapang Formation hominins, Central Jawa, Indonesia. Proc. Natl. Acad. Sci. U. S. A. 98, 4866e4871. Larrasoa~na, J.C., Roberts, A.P., Rohling, E.J., Winklhofer, M., Wehausen, R., 2003. Three million years of monsoon variability over the northern Sahara. Clim. Dyn. 21, 689e698. Leakey, M.G., 1993. Evolution of Theropithecus in the Turkana Basin. In: Jablonski, N.G. (Ed.), Theropithecus: The Rise and Fall of a Primate Genus. Cambridge University Press, Cambridge, pp. 85e124. Leakey, M.G., Feibel, C.S., Bernor, R.L., Harris, J.M., Cerling, T.E., Stewart, K.M., Storrs, G.W., Walker, A., Werdelin, L., Winkler, A.J., 1996. Lothagam: a record of faunal change in the Late Miocene of East Africa. J. Vertebr. Paleontol. 16, 556e570. Leakey, R.E.F., 1969. New Cercopithecidae from the Chemeron Beds of Lake Baringo, Kenya. Fossil Vertebr. Afr. 1, 53e69. Lee-Thorp, J.A., van der Merwe, N.J., Brain, C.K., 1989. Isotopic evidence for dietary differences between two extinct baboon species from Swartkrans. J. Hum. Evol. 18, 183e190. Leroy, S., Dupont, L., 1994. Development of vegetation and continental aridity in Northwestern Africa during the Late Pliocene - the pollen record of Odp Site-658. Palaeogeogr. Palaeoclimatol. Palaeoecol. 109, 295e316. Marlow, J.R., Lange, C.B., Wefer, G., Rosell-Mele´, A., 2000. Upwelling intensification as part of the Pliocene-Pleistocene climate transition. Science 290, 2288e2291. Martinez-Navarro, B., Claret, A., Shabel, A.B., Perez-Claros, J.A., Lorenzo, C., Palmqvist, P., 2005. Early Pleistocene ‘hominid remains’ from southern Spain and the taxonomic assignment of the Cueva Victoria phalanx. J. Hum. Evol. 48, 517e523. Maschenko, E.N., 1994. Papio (Paradolichopithecus) suschkini a revision ofsystematics, morphofunctional peculiarities of the skull and mandible. Paleotheriologiya, Moscow, pp. 15e57 (in Russian). Mithen, S., Reed, M., 2002. Stepping out: a computer simulation of hominid dispersal from Africa. J. Hum. Evol. 43, 433e462. Nanda, A.C., 2002. Upper Siwalik mammalian faunas of India and associated events. J. Asian Earth Sci. 21, 47e58. Napier, J.R., Napier, P.H., 1967. A Handbook of Living Primates. Academic Press, London. O’Regan, H.J., Bishop, L.C., Elton, S., Lamb, A., Turner, A., 2006. Afro-Eurasian mammalian dispersions of the late Pliocene and early Pleistocene. Cour. Forsch.-Inst. Senckenberg 256, 305e314. Pasini, G., Colalongo, M.L., 1997. The Pliocene-Pleistocene boundary-stratotype at Vrica, Italy. In: van Couvering, J.A. (Ed.), The Pleistocene Boundary and the Beginning of the Quaternary. Cambridge University Press, Cambridge, pp. 15e45. Patel, B.A., Gilbert, C.C., Ericson, K.E., 2007. Cercopithecoid cervical vertebral morphology and implications for the presence of Theropithecus in early Pleistocene Europe. J. Hum. Evol. 52, 113e129.

Peters, J., Po¨llath, N., 2004. Holocene faunas of the Eastern Sahara: zoogeographical and palaeoecological aspects. In: Lauwerier, R.C.G.M., Plug, I. (Eds.), The Future From the Past: Archaeozoology in Wildlife Conservation and Heritage Management. Oxbow Books, Oxford, pp. 34e50. Pickford, M., 1993. Climatic change, biogeography and Theropithecus. In: Jablonski, N.G. (Ed.), Theropithecus: The Rise and Fall of a Primate Genus. Cambridge University Press, Cambridge, pp. 227e243. Prentice, I.C., Webb III, T., 1998. BIOME 6000: reconstructing global midHolocene vegetation patterns from palaeoecological records. J. Biogeogr. 25, 997e1005. Quade, J., Cerling, T.E., 1995. Expansion of C4 grasses in the Late Miocene of northern Pakistan: evidence from stable isotopes in paleosols. Paleogeog. Paleoclimatol. Paleoecol. 115, 91e116. Reed, K.E., 1997. Early hominid evolution and ecological change through the African Plio-Pleistocene. J. Hum. Evol. 32, 289e322. Rook, L., Martinez-Navarro, B., Howell, F.C., 2004. Occurrence of Theropithecus sp. in the late Villafranchian of southern Italy and implication for Early Pleistocene ‘‘out of Africa’’ dispersals. J. Hum. Evol. 47, 267e277. Sardella, R., Caloi, L., Di Stefano, G., Palombo, M.R., Petronio, C., Abbazzi, L., Azzarli, A., Ficcarelli, G., Mazza, P., Mezzabotta, C., Rook, L., Torre, D., Argenti, P., Capasso Barbato, L., Kotsakis, T., Gliozzi, E., Masini, F., Sala, B., 1998. Mammal faunal turnover in Italy from the Middle Pliocene to the Holocene. Mededelingen Nederlands Instituut voor Toegepaste Geowetenschappen TNO 60, 499e511. Sato, K., Iwasa, Y., 2000. Pair approximations for lattice-based ecology models. In: Dieckmann, U., Law, R., Metz, J.A.J. (Eds.), The Geometry of Ecological Interactions. Cambridge University Press, Cambridge, pp. 341e358. Shackleton, N.J., Hall, M.A., 1989. Stable isotope history of the Pleistocene at ODP site 677. Proc. ODP Sci. Results 111. 295e316. Shipman, P., Bosler, W., Davis, K.L., 1981. Butchering of giant geladas at an Acheulean site. Curr. Anthropol. 22, 257e268. Sikes, N.E., Potts, R., Behrensmeyer, A.K., 1999. Early Pleistocene habitat in Member 1 Olorgesailie based on palaeosol stable isotopes. J. Hum. Evol. 37, 721e746. Spassov, N., 2003. The Plio-Pleistocene vertebrate fauna in South-Eastern Europe and the megafaunal migratory waves from the east to Europe. Revue Pale´obiol., Gene`ve 22, 197e229. Sponheimer, M., Reed, K.E., Lee-Thorp, J.A., 1999. Combining isotopic and ecomorphological data to refine bovid palaeodietary reconstruction: a case study from the Makapansgat Limeworks hominin locality. J. Hum. Evol. 36, 705e718. Strasser, E., 1988. Pedal evidence for the origin and diversification of cercopithecid clades. J. Hum. Evol. 17, 225e245. Szalay, F.S., Delson, E., 1979. Evolutionary History of the Primates. Academic Press, New York. Teaford, M.F., 1993. Dental microwear and diet in extant and extinct Theropithecus: preliminary analyses. In: Jablonski, N.G. (Ed.), Theropithecus: The Rise and Fall of a Primate Genus. Cambridge University Press, Cambridge, pp. 331e349. Thompson, R.S., Fleming, R.F., 1996. Middle Pliocene vegetation: reconstructions, paleoclimatic inferences, and boundary conditions for climatic modelling. Marine Micropaleontol. 27, 27e49. Toffoli, T., Margolus, N., 1987. Cellular Automata Machines: A New Environment for Modelling. MIT Press, Cambridge, Massachusetts. Trauth, M.H., Maslin, M.A., Deino, A., Strecker, M.R., 2005. Late Cenozoic moisture history of East Africa. Science 309, 2051e2053. Trofimov, B.A., 1977. Primate Paradolichopithecus sushkini sp. nov. from upper Pliocene of the Pamirs piedmont. J. Palaeontol. Soc. India 20, 26e32. Turner, A., 1999. Assessing earliest human settlement of Eurasia: Late Pliocene dispersions from Africa. Antiquity 73, 563e570. Vrba, E.S., 1995. The fossil record of African antelopes (Mammalia, Bovidae) in relation to human evolution and paleoclimate. In: Vrba, E.S., Denton, G.H., Partridge, T.C., Burckle, L.H. (Eds.), Paleoclimate and Evolution with Emphasis on Hominid Origins. Yale University Press, New Haven, pp. 385e424. van der Made, J., 2004. Megaloceros giganteus from the Middle Pleistocene of Neumark-Nord. Vero¨ffentlichungen des Landesamtes fu¨r Archa¨ologie 57, 373e378.

J.K. Hughes et al. / Journal of Human Evolution 54 (2008) 43e77 Wildman, D.E., Bergman, T.J., al-Aghbari, A., Sterner, K.N., Newman, T.K., Phillips-Conroy, J.E., Jolly, C.J., Disotell, T.R., 2004. Mitochondrial evidence for the origin of hamadryas baboons. Molec. Phyl. Evol. 32, 287e296. Winney, B.J., Hammond, R.L., Macasero, W., Flores, B., Boug, A., Biquand, V., Biquand, S., Bruford, M.W., 2004. Crossing the Red Sea: phylogeography of the hamadryas baboon, Papio hamadryas hamadryas. Mol. Ecol. 13, 2819e2827.

77

Yang, S., Ding, Z., 2006. Winter-spring precipitation as the principle control on predominance of C3 plants in Central Asia over the past 1.77 Myr: evidence from delta13C of loess organic matter in Tajikistan. Palaeogeogr. Palaeoclimatol. Palaeoecol. 235, 330e339. Zagwijn, W.H., 1998. Borders and boundaries: a century of stratigraphical research in the Tegelen-Reuver area of Limburg (the Netherlands). Mededelingen Nederlands Instituut voor Toegepaste Geowetenschappen TNO 60, 19e34.