The Miocene (~23 – 5.3 Ma) has been dubbed “the real planet of the apes” (Begun, 2015) and is therefore a critical period for exploring the deep roots of locomotion in hominoids (all apes), hominids (the great apes), and hominins (the human lineage) (Almécija et al., 2021; Urciuoli and Alba, 2023). Unfortunately, the African ape record is sparse between ~14 and 10 Ma but rare exceptions, such as the discovery of an infant ape cranium in Kenya, Nyanzapithecus alesi (~13 Ma), provide insights into potential stem hominoids. Nyanzapithecines share similarities with extant hylobatids (gibbons) with their relatively small bodies and arboreality, but the inner ear of the N. alesi cranium suggests that their movements were probably less acrobatic than modern-day gibbons (Nengo et al., 2017).

Ongoing debate about where and when the earliest hominids emerged will likely remain unresolved whilst this critical gap in the African fossil record persists. One possibility is an “out of Africa” and then “back-to-Africa” sequence of events, with aridification of the Sahara creating a biogeographic barrier between African and southern European faunas (Bibi, 2011; Schuster et al., 2006, 2009). The first fossil evidence of apes outside of Africa comes from Europe in the middle Miocene (~16.5–14 Ma) and many of the most recent Miocene apes for which we have postcranial remains have been found in Europe. These include Pierolapithecus catalaunicus (~11.9 Ma) (A. S. Hammond et al., 2013; McNutt et al., 2018; Moyà-Solà et al., 2004), Danuvius guggenmosi (~11.6 Ma) (Böhme et al., 2019a), and Rudapithecus hungaricus (~10 Ma) (Begun et al., 2012; Ward et al., 2019).

The foot, hand, and pelvic morphologies of Pierolapithecus suggest that it lived a fully arboreal lifestyle, did not engage in suspensory locomotion, and was more frequently engaged in orthograde (upright body posture) behaviours than earlier hominoids, but not as frequently as extant great apes (Hammond et al., 2013; Moyà-Solà et al., 2004). In contrast, Rudapithecus and Danuvius show adaptations to suspensory locomotion and there has even been an argument for evidence of above-branch bipedalism in the latter (Böhme et al., 2019a; Ward et al., 2019), although this is contested (Williams et al., 2020).

These Miocene apes from Europe provide a diverse set of possibilities for the roots of hominid locomotion, but the three genera described above share characteristics that could have been present in the LCA of African hominids. For example, at ~30kg in body mass, these apes were all smaller than extant great apes (perhaps with the exception of bonobos) [Pierolapithecus: (Moyà-Solà et al., 2004); Rudapithecus: (Smith et al., 2019); Danuvius: (Böhme et al., 2019b)]. They show adaptations to an arboreal lifestyle and indicate early origins of an orthograde body plan that could then have been co-opted for behaviours including below-branch suspension, knuckle-walking, and/or arboreal or terrestrial bipedalism in different hominid taxa. Emergence of this upright posture is documented alongside an increase in ape brain size – both examples of traits that are central to discussions about human evolution but are in fact rooted in evolutionary trends seen as far back as the Miocene (Alba, 2010; Almécija et al., 2021; Moyà-Solà et al., 2004).

As evidence about the diversity, evolutionary relationships, and geographic dispersals of Miocene apes continues to emerge, albeit mired in controversy (Grabowski and Jungers, 2017; McNulty, 2010; Urciuoli and Alba, 2023), more fossil discoveries will help piece together the connections between ape species from European fossil sites and those found in later Miocene African sites. One such connection has been suggested based on the resemblance between Nakalipithecus (9.9 – 9.8 Ma) fossils, found in Kenya, and those of Ouranopithecus (9.6–8.7 Ma), found in Greece and a proposed candidate close relative of the extant African great apes (Kunimatsu et al., 2007). Both of these genera, along with Samburupithecus (9.5 Ma) (Ishida and Pickford, 1997) and Chororapithecus (10 Ma) (Suwa et al., 2007), have been hypothesized as larger-bodied apes, perhaps comparable to extant gorillas. However, most analyses come only from craniodental evidence, except for two isolated phalanges from Ouranopithecus which have been used to propose that it might have been “a ground dweller closer to Papio and Macaca than to Hylobates or even to Pan” (de Bonis and Koufos, 2014). However, this suggestion is hard to substantiate without further postcranial evidence, and would benefit from an integrated analysis of other ecological drivers that might have driven this locomotor style.

By one analysis, there have been “a series of selective regime shifts” that have influenced ape body size (and locomotion) across the Miocene, shifting from a hominoid LCA that was gibbon-like to a Pan-Homo LCA that was chimpanzee-like in size (Grabowski and Jungers, 2017) and – by some arguments – in behaviour (Williams et al., 2023). However, it is difficult to plot out the diverging evolutionary trajectories of the different hominid taxa, due to the scarcity of fossils from the ancestors of extant African ape clades. Other than a few teeth and a possible proximal femur, we have no fossil evidence for the chimpanzee or gorilla lineages (Cote, 2004; DeSilva et al., 2006; Kunimatsu et al., 2007; McBrearty and Jablonski, 2005; McNulty, 2010; Pickford and Senut, 2004; Suwa et al., 2007). There is fossil evidence of three ape genera from northern and eastern Africa in the late Miocene, all of which have been putatively assigned to the hominin lineage: Sahelanthropus, Orrorin, and Ardipithecus. The traits of these genera are discussed below, with a focus on their inferred styles of locomotion.

At approximately 7 million years old (Lebatard et al., 2008), Sahelanthropus tchadensis fossils from Chad represent the oldest putative hominin, initially classified as such based on its dentition and the position of its foramen magnum, perhaps indicative of bipedality (Brunet et al., 2002; Neves et al., 2024; Zollikofer et al., 2005). However, this ape retains many primitive features similar to earlier apes like Pierolapithecus, and the foramen magnum positioning may simply indicate orthograde posture rather than necessitating bipedality (Andrews, 2020). Recent discovery and analyses of postcranial material have resulted in varying interpretations of Sahelanthropus ranging from habitually bipedal but still engaged in “substantial arboreal behaviour” (Daver et al., 2022), to “not habitually bipedal” (Macchiarelli et al., 2020), and “not an obligate biped, but [ … ] a Miocene hominid with knuckle-walking adaptations” (Meyer et al., 2023). New fossils, further analyses, and clearer definitions of bipedalism will all be needed to clarify the locomotor style and status of this taxon.

Orrorin tugenensis fossils discovered in Kenya and dated to ~6 Ma have also been given hominin status by some, based largely on an initial proposal that its femur morphology was more similar to modern humans than to other extant apes or Australopithecus species, indicating that it was adapted to terrestrial bipedalism of some sort, whilst retaining good climbing adaptations (Senut et al., 2001). More recent interpretations of femoral evidence highlight that Orrorin shares many features with earlier Miocene apes and later hominins, and likely represents an “intermediate” morphology between them (Almécija et al., 2013; Richmond and Jungers, 2008).

At the very end of the Miocene (~5.8 – 5.2 Ma) evidence of Ardipithecus kadabba from the Middle Awash, Ethiopia, has also been dubbed “the first hominin”, due to the possible absence of a functional canine honing complex, and some indicators of “an early form of terrestrial bipedality” (Haile-Selassie, 2001). However, depending on how morphological comparisons are made, different conclusions can and have been reached about Ardipithecus. These range from assigning it the status of the first hominin, to the Homo-Pan LCA, to a genus more related to chimpanzees than to the hominin lineage (Haile-Selassie, 2001; Sayers et al., 2012; Senut et al., 2001).

Interpretations relying on assumptions that the LCA was essentially chimpanzee-like use chimpanzee morphology as a reference point to assess how “hominin-like” a fossil is, rather than acknowledging that both later hominins and chimpanzees are likely to display very different, derived features from their LCA. Researchers trying to move away from an over-reliance on this “Chimpanzee Referential Doctrine” have highlighted that because Ardipithecus does not exhibit clear adaptations to below-branch suspension, knuckle-walking, or vertical climbing, it should not be assumed that these styles of locomotion existed in the LCA of Homo and Pan (Sayers et al., 2012).

The Plio-Pleistocene represents a key period for understanding hominin evolution. It encompasses the existence, and even co-existence of at least three hominin genera – Australopithecus, Paranthropus, and Homo – but this diverse clade was pruned down by the disappearance of Australopithecus and Paranthropus by 1 Ma. Meanwhile, Homo appears to have flourished within Africa and in its dispersal out of Africa and across Eurasia ~1.8 Ma (Behrensmeyer, 2006; Potts, 2013; Stringer, 2002; Wood and Boyle, 2016), likely signalling the successful locomotor strategy of a committed terrestrial biped and initiating the subsequent ecological hegemony of H. sapiens. The Plio-Pleistocene also marks a period of climatic and ecological change that resulted in the expansion of grasslands across Africa (Cerling et al., 2011; Levin, 2015; Negash et al., 2024). It is this shift towards more open environments that laid the foundations for the “Savannah Hypothesis” – the idea that human bipedality is an adaptation to receding forests (Dart, 1925). This hypothesis gained and retained attention in popular understandings of human evolution and has been closely linked with human tool-use, hunting, and expansion of brain size (Brain, 1981; Dart, 1925, 1949; Darwin, 1871; Harcourt-Smith, 2010; Senut et al., 2018; Vrba et al., 1989; Washburn, 1960). However, several lines of evidence – discussed below – suggest that bipedality was not driven solely by expanding savannahs, and that the relationships between these characteristics and open environments are nuanced.

After appearing in the Miocene, Ardipithecus is represented in the Pliocene by the species Ardipithecus ramidus, a “primitive” bipedal ape that ranged in wooded habitats in the area now known as the Afar of Ethiopia, ~4.4 Ma (Semaw et al., 2005; White et al., 2009). Initial interpretations of Ardipithecus postcranial material posit that this possible hominin combined some form of terrestrial bipedality with arboreal clambering, which involved weight-bearing in the palms and was unlikely to include knuckle-walking. However, this style of locomotion was not well-adapted to long bouts of terrestrial bipedalism, nor to the levels of suspensory arboreal locomotion seen in the African apes of today (White et al., 2009, 2015). Recent analyses of Ardipithecus fossils have prompted some researchers to say that it did in fact engage in suspensory locomotion and maybe even knuckle-walking, because its hand morphology is closer to extant apes than more generalised quadruped primates (Prang et al., 2021). A similar argument has been used to propose that its feet were also more similar to those of extant African apes than previously thought (Prang, 2019, 2022). Other researchers do not believe that these similarities warrant the conclusion that Ardipithecus’ locomotion was anything like that of modern African apes, and highlight that comparisons using extant primate data continue to be limited whilst we have almost no fossil record for Pan or Gorilla (Chaney et al., 2022).

Separating homologous and homoplastic traits amongst extant apes will certainly be made easier through the discovery of more fossils, but it can also be aided by considerations of selection pressures that shaped the behavioural ecology of these lineages. For example, the LCA of all extant African apes is often assumed to have been at least semiterrestrial, because terrestrial behaviour is seen in gorillas, chimpanzees, bonobos, and humans. However, evidence from several Miocene ape fossils (discussed above) suggests they were arboreal creatures, and some interpretations of the Ardipithecus evidence indicate that the LCA of Homo, Pan, and Gorilla was probably an above-branch clambering quadruped from whom the extant apes independently acquired their specialisations for suspension, vertical climbing, knuckle-walking, and terrestrial bipedality (Dainton and Macho, 1999; Kivell, 2019; Lovejoy et al., 2009a). We must therefore consider ecological factors that might have selected for increased terrestrial behaviour across all these lineages (convergent evolution), rather than assuming that the shared characteristic necessitates its existence in their LCA.

Whilst debate about the positional behaviour of Ardipithecus continues, evidence suggests that it lived in forested habitats and had a C₃-heavy diet. Although not absolute, C₃ and C₄ isotopic signatures have been associated with woodland and grassland vegetation respectively (Cerling et al., 2015). Together, this paleoenvironmental and isotopic evidence indicates that locomotor adaptations seen in Ardipithecus fossils were not driven by a move onto the savannah (White et al., 2009). Indeed, the influence of expanding grasslands has more often been linked to the emergence of the genus Australopithecus – the first apes considered to be indisputably hominin and referred to as “committed”, “obligate”, or at least “habitual” bipeds (Anaya et al., 2021; Bobe et al., 2020; Harcourt-Smith, 2010; Stamos and Alemseged, 2023; White et al., 2009). However, as this range of descriptors suggests, there is still much debate around the exact nature of locomotion in Australopithecus species, and there was probably locomotor diversity within the genus (Senut et al., 2018). At first, Australopithecus was contrasted with the “more primitive” Ardipithecus to suggest that the derived postcranial features of Australopithecus showed that they “had largely abandoned locomotion in the arboreal canopy” (White et al., 2009). However, there is increasing evidence that the morphology, paleoecology, and diet of Australopithecus was more adapted to woodland-living than initially thought, especially when the genus first emerged (Bobe et al., 2020; Cerling et al., 2013; Sponheimer et al., 2013).

Au. anamensis first appears in the Kenyan and Ethiopian fossil records from ~4.2 Ma (Ward et al., 1999, 2001; White et al., 2006). Similarities in the paleoecology and faunal records at Kanapoi in the Turkana Basin, Kenya, and Asa Issie in the Afar, Ethiopia, highlight the biogeographic connections that must have existed between these sites, and suggest that Au. anamensis thrived in habitats with a mix of C₃ and C₄ vegetation, although they still showed a preference for C₃ foods at this time (Bobe et al., 2020). Their probable descendants, Au. afarensis, were the first hominins to incorporate C₄ foods into their diets ~3.7 Ma, whilst still living in relatively wooded environments (Sponheimer et al., 2013). Au. afarensis is probably also the hominin species most associated with early bipedality, famous for both the “Lucy” skeleton and the Laetoli footprints that have been attributed to the species (Kimbel and Delezene, 2009; Leakey and Hay, 1979; Raichlen et al., 2008). These footprints are strong evidence that hominins did travel on the ground with a bipedal gait, but the tracks could have been made by a stride very different to modern human locomotion (Raichlen et al., 2008). Recent analyses of additional trackways indicate that there may even have been more than one species of “small, cross-stepping bipedal hominin” traversing the Laetoli landscape over 3.5 Ma (McNutt et al., 2021). If so, this evidence would suggest that the emergence of terrestrial bipedality was not necessarily rare or unique to one species, but an adaptive response by several ape species to shared selection pressures at the time. Indeed, analyses of hominin foot morphologies indicate a diversity at the base of the lineage suggestive of different styles of locomotion and “experimentation in bipedalism” (DeSilva et al., 2019).

Whilst Australopithecus is the first hominin genus in which bipedal locomotion is undisputed (Potts, 2013), it does not necessarily mark a transition to exclusive terrestriality, as will be outlined in the sections below. Species of Australopithecus disappear from both the eastern and southern African fossil records during the Early Pleistocene; Au. sediba from South Africa less than 2 Ma (Dirks et al., 2010), and other members of the genus from eastern and southern Africa ~2.5 Ma (Wood and Boyle, 2016). Evidence shows that Australopithecus overlapped both temporally and geographically with two other hominin genera; the megadont, Paranthropus, and early species of our own genus, Homo (Herries et al., 2020; Jablonski et al., 2008; Potts, 2013; Wood and Boyle, 2016).

Paranthropus appears in the eastern and southern African fossil records ~2.7 Ma and 2 Ma respectively, and disappears from both ~1.2 Ma (Potts, 2013; Wood and Boyle, 2016). Carbon isotope analyses of teeth from P. boisei, present in eastern Africa from 2.3 Ma until 1.2 Ma, suggest that the species was a C₄ specialist and indicate that it’s style of locomotion must have allowed it to exploit open grasslands (Sponheimer et al., 2013). The earliest evidence of the genus Homo is a little older, just pre-dating the start of the Pleistocene; one specimen identified as a Homo mandible is thought to be ~2.8 million years old (Villmoare et al., 2015). H. habilis and H. rudolfensis, appear in the fossil record between 2.5 and 2 Ma, and disappear between 1.8 and 1.6 Ma, whilst the first clearly “premodern human”, H. erectus, originated at least 1.89 Ma, dispersed across and out of Africa, and only went extinct less than 200 Ka (Hammond et al., 2021; Herries et al., 2020; Rizal et al., 2019; Wood and Collard, 1999).

Compared to the C₄ specialisation indicated by P. boisei fossils, Homo teeth indicate a mixed diet, composed of both C₃ and C₄ resources (Sponheimer et al., 2013). This suggests that, although Paranthropus and Homo species were at times sharing relatively small habitats (Bobe and Carvalho, 2019; Hatala et al., 2024) and dealing with the same climatic shifts – increasing aridity, seasonality, and mosaic habitats – they were utilising the environment differently. It seems that perhaps Paranthropus lived and foraged predominantly in the more open C4 areas, whilst Homo inhabited a wider range of environments (Cerling et al., 2013). Both genera appear to have been bipedal, but morphological evidence suggests that this mode of locomotion was less specialised in Paranthropus than in Homo species (Harcourt-Smith and Aiello, 2004; Richmond et al., 2020).

The brains of Homo and Paranthropus also display significant differences; the skulls and endocasts of the latter are not distinct from those of Australopithecus, and are possibly even less developed, whilst the big size and shape of the Homo brain are two of the most distinguishing features of the genus (Falk et al., 2000). This difference suggests that the larger brains observed in Homo species cannot be attributed only to environmental shifts, as evidenced by the fact that smaller-brained Paranthropus hominins successfully existed through at least a million years of climatically variable Pleistocene, as well as at least half a million years of co-existence with Homo. To disentangle assumptions that connect modern human tool-use, hunting, and brain size with a pre-historic move into grasslands, it is thus useful to compare Homo records to those of Paranthropus and grassland-dwelling papionins, like Theropithecus, who were subject to the same selection pressures. Furthermore, as discussed below, the behavioural ecology of the extant Theropithecus gelada can complement fossil record interpretations to hypothesise about foraging strategies, substrate-use, and exposure to competition or predation pressure in grasslands.

Despite strong evidence of some form of terrestrial bipedality from as far back as the footprints and derived foot morphology of Au. afarensis (Ward et al., 2011), adaptations for and evidence of arboreal locomotion are seen in that species (Kappelman et al., 2016; Senut and Tardieu, 1985; Stern and Susman, 1983; Ward, 2002) as well as in several hominin taxa that both pre and postdate Au. afarensis. There has been heated debate about whether such morphological traits are simply a retention of “primitive” ancestral adaptations, no longer in use nor adaptive in hominins, or if they are proof of continued arboreality in multiple hominin genera (Anaya et al., 2021; Kimbel and Delezene, 2009; Ward, 2002). Several lines of evidence – including hominin fossils themselves, and the integration of paleoecological data with applied insights from primate behavioural ecology – suggest that these signals should not be dismissed as primitive relics “leftover” from pre-hominin ancestors.

From hominin fossils themselves, we see indicators of arboreality in a diversity of taxa. Described above is the postcranial evidence from the eastern African Pliocene species, Au. anamensis and Au. afarensis (Kimbel and Delezene, 2009; Ward, 2002; Ward et al., 2001). Arboreal adaptations are also seen in southern African species of the genus. Au. africanus appears in the southern African fossil record ~3 Ma, which is more recent than the eastern African Au. afarensis. Although there is some evidence of reduced arboreality in this species (at least in comparison to extant great apes) (Georgiou et al., 2020), its morphology is generally interpreted as better suited to climbing than the eastern African Australopithecus (Green et al., 2007). This interpretation prompts questions about whether climbing adaptations in Au. africanus were “secondarily derived” after arboreality declined in the ancestral Au. afarensis. Alternatively, both Au. afarensis and Au. africanus might have inherited their arboreal adaptations from a primitive ancestor. However, if Homo is a descendant of Au. africanus (as has been proposed), this scenario would suggest that derived terrestrial bipedality evolved independently in the Au. afarensis and Homo lineages at different times and in different forms (Green et al., 2007; Prabhat et al., 2021). Again, such a scenario suggests that mosaic styles of ape locomotion represent sensitive adaptations to ecological pressures, and remind us that modern human bipedality did not evolve along a steady linear path.

In Au. sediba, a southern African species that lived more recently than 2 Ma, internal finger bone structure demonstrates in vivo power grasping – evidence of more than just a primitive relic of arboreality. This proven climbing ability existed alongside knee, ankle, and lower back morphology suggestive of bipedalism (Dunmore et al., 2020; Williams et al., 2021). Recent dating of South African fossil sites indicates that this species existed contemporaneously and within a 250 km² area inhabited by both Paranthropus robustus and Homo aff. erectus (Herries et al., 2020). Not only does this highlight the diversity of the Pleistocene hominin family tree, but it once again brings up questions about the role of homology and homoplasy in shaping “unique” features of human morphology. Whilst Au. sediba has often been considered to sit between Au. africanus and Homo in the ancestral tree, the discovery of their contemporaneity suggests it may not be ancestral to Homo, in which case their shared features could be homoplastic rather than homologous (Du and Alemseged, 2019; Herries et al., 2020).

Evidence of arboreality persists even beyond the genus Australopithecus. There are now analyses of the upper limb of Paranthropus boisei suggesting that this species was regularly climbing trees in eastern Africa ~1.5 Ma (Richmond et al., 2020), despite its C₄ diet indicating regular exploitation of terrestrial resources. And a study of the trabecular structure of southern African hominin femurs indicates regular climbing behaviours in a specimen attributed to either Paranthropus robustus or Homo (Georgiou et al., 2020). Within the genus Homo, it appears that 1.8-million-year-old H. habilis had forelimbs that were conducive to climbing (Ruff, 2009), a feature that also seems to have been present in H. naledi, a southern African hominin that lived as recently as 200 to 400 thousand years ago (Dirks et al., 2017; Feuerriegel et al., 2017). Even from within the past hundred thousand years, we have postcranial evidence from H. floresiensis that suggests the small-bodied hominin had a very different gait to its contemporaneous Homo species (Harcourt-Smith and Aiello, 2004; Jungers et al., 2009; Larson et al., 2009). Whilst an interpretation of the locomotor style of H. floresiensis remains elusive, one analysis of hominin limb joint proportions place H. floresiensis in a group with Au. africanus, Au. sediba, P. robustus, P. boisei, and H. habilis, all of which the study describes as having more “ape-like” proportions than those of Au. afarensis, H. erectus, and H. naledi which more closely resemble those of modern humans (Prabhat et al., 2021).

Once again, the analyses above can be criticised for their dichotomous comparison of “ape-like” and “human-like” traits and an over-reliance on extant ape locomotion as a model for extinct species. Furthermore, different methods and interpretations of evidence, as well as the use of different reference skeletons, lead researchers to different and even contradictory conclusions (Chaney et al., 2022; Haeusler and McHenry, 2004; Kramer, 2012; Prang, 2019). However, there is still much to learn from this accumulation of fossil evidence. Not only does it highlight that arboreality was likely present in a diverse range of hominin taxa and across multiple geographic regions and time periods, but it also demonstrates that terrestrial bipedality emerged in different forms, at different times, and in varied combinations with other traits such as larger brains – all contributing to the mosaic nature of hominin evolution (Foley, 2016). Au. afarensis, for example, demonstrates a combination of some form of terrestrial bipedalism with a small endocranial volume and ape-like brain organisation (Gunz et al., 2020), and P. boisei was clearly exploiting open grasslands but also retains a smaller brain and arboreal adaptations (Falk et al., 2000; Richmond et al., 2020; Sponheimer et al., 2013). Meanwhile H. habilis demonstrates a combination of the “typical” larger brain of Homo with signals of continued arboreality (Ruff, 2009; Spoor et al., 2015), and H. naledi is a relatively recent species of Homo but demonstrates a combination of small brain and climbing adaptations (Feuerriegel et al., 2017). Evidence of this locomotor diversity calls attention to the many selective pressures that act and interact to shape behaviour and morphology. It raises questions such as: What were the biotic and abiotic variables that released some primate species from the ancestral state of arboreality? Which of these variables (and the interactions between them) were the strongest drivers of opportunistic terrestriality? And what were the subsequent benefits of opportunistic terrestriality that selected for longer term behavioural and morphological adaptations? Additionally, what dictated the style of locomotion (e.g., bipedalism vs. quadrupedalism, or knuckle-walking vs. palmigrade locomotion) in which a transition to the ground was made by various taxa? These are all questions that can be answered by integrating insights from paleoecology and modern ecosystems to better understand the interactions between primate morphology, behaviour, substrate-use, and ecology.

As discussed, a degree of bipedality has often been attributed to hominin fossils, with species of Australopithecus described variously as “committed”, “obligate”, or “habitual” bipeds, and with Homo fossils as different as H. habilis and H. erectus classified as “obligate” bipeds, albeit with the latter distinguished as a “fully committed” biped (Harcourt-Smith, 2016; Ruff, 2009). The difference between these terms has traditionally not been clearly defined and is often inferred from fossil features, rather than a definition of each behaviour in vivo (Stamos and Alemseged, 2023). This issue is compounded when interpretations of hominin morphology draw from comparisons with extant species classified as “arboreal”, “semi-terrestrial”, or “terrestrial” – another set of terms without clear definitions (Williams et al., 2023). These terms also overlook the fact that even the most terrestrial of non-human primates engage in relatively frequent arboreal activity, at the very least ascending into trees or cliffs to sleep at night. And conversely, even the most arboreal of non-human primates engage in opportunistic terrestriality (a recent finding only facilitated by remote monitoring because human presence had previously prevented them from coming to the ground) (Estrada and Marshall, 2024). To effectively interpret the primate fossil record, it is thus essential to understand how form relates to function, for example how extant primate morphology relates to proportional substrate-use, and how that function is driven by ecological factors.

In the case of hominins, we must first disentangle bipedality and terrestriality to understand their independent ecological drivers. Overreliance on reference to extant African apes has created assumptions that their LCA was at least semi-terrestrial, and questions have therefore asked how and when hominins “stood up” from terrestrial knuckle-walking to bipedal striding (Richmond and Strait, 2000; Sayers et al., 2012). However, there is evidence that knuckle-walking in non-human apes might be an example of convergent evolution (Dainton and Macho, 1999; Kivell and Schmitt, 2009), and that hominin bipedality might have an arboreal origin. Proponents of this argument highlight that arboreality provides a relatively secure food supply and security from predators, and fits better with the palaeoecological evidence that hominins emerged in wooded environments (Crompton et al., 2010b; Senut et al., 2018; Thorpe et al., 2007). Furthermore, by broadening the scope of comparative analyses beyond African apes to also incorporate extant orangutan behaviour in interpretations of the fossil record, an argument can be made that arboreal bipedalism might have given the LCA of extant apes an advantageous way of moving about on flexible end-branches of trees. This might explain apes’ shared adaptations to orthogrady, as well as the fossil evidence that many early and possible hominins display hindlimb adaptations to bipedalism whilst retaining forelimb adaptations for arboreality (Thorpe et al., 2007). An arboreal origin of ape bipedalism might have been driven by growing competition from monkey species during the middle Miocene. Monkeys can digest less-ripe fruits in the centre of trees, potentially forcing apes to compete by using more suspensory and clambering behaviours to reach riper fruits at the end of branches (Hunt, 2016). Once again, this highlights dynamic aspects of paleoenvironments – competition and niche partitioning – that might have shaped hominin locomotion but cannot be inferred directly from fossils alone.

So, if the evidence increasingly points towards arboreal origins for the hominid LCA, as well as arboreal origins of hominin bipedality, then what and when were the changes that eventually led to a predominantly ground-dwelling Homo species? The combined evidence indicates that terrestriality is more common across all extant African great apes than it was in their LCA. Until we have more extensive fossil records for Gorilla and Pan, we can only hypothesise about the locomotion and postcranial morphology of their ancestors, but they were likely more arboreal than their descendants. For example, we know from extant apes that there is a trade-off between body size and time spent on arboreal travel. Gorillas are both the largest of the non-human apes and the most terrestrial. Male chimpanzees and orangutans are more likely to travel terrestrially than smaller females, and there is some evidence that chimpanzees are more terrestrial than smaller-bodied bonobos (Doran, 1993). Although it is certainly still a matter of debate (Almécija et al., 2021; Grabowski and Jungers, 2017; Williams et al., 2023), if extant African apes are larger bodied than their LCA (and living in more fragmented environments), body size might have been a driver of more terrestrial activity across the clade, but with different locomotor styles that have evolved to fit the different three-dimensional features of their respective environments.

Body mass is very closely linked to terrestriality in primates and has been proposed as a driver of substrate-use. Indeed, in an analysis of 515 extant primate taxa, body mass was the most reliable predictor of ground use (Estrada and Marshall, 2024). However, when studying extant taxa, it is hard to pinpoint the evolutionary “first movers” in pairs or groups of traits that likely co-evolved. Whilst it might be true that extant primates with bigger bodies spend more time on the ground in modern environments, it is difficult to explain why they would have evolved significantly bigger body sizes in the first place if arboreality and substrate structure constrain body mass. The correlation we see today might thus be a result of increased terrestriality in certain taxa having facilitated the evolution of bigger bodies in those lineages.

Ancestral state reconstructions (ASR) used to estimate how often terrestriality evolved in primates suggest it emerged on at least two separate occasions during the Miocene (within the Cercopithecinae lineage ~22–15 Ma and the Homininae lineage ~15–9 Ma) (Estrada and Marshall, 2024). The earliest fossil evidence of terrestriality in Old World primates comes from the Kenyan fossil record at Maboko Island 15 Ma, in both a cercopithecoid, Victoriapithecus, and a hominoid, Equatorius (formerly known as Kenyapithecus). There is no evidence that this emergence of terrestriality was preceded by an opening up of habitats in the region, nor by an increase in the body sizes of either genus. But the fact that adaptations to terrestriality emerged in two separate clades in the early to middle Miocene of Africa does suggest that something about their ecological context was driving substrate-use (McCrossin et al., 1998). This evidence not only highlights that primate terrestriality emerged as early as the Miocene and was independent from bipedality, but it also provides a period and site ripe for further examination of the biotic and abiotic variables that might have driven primates to spend more time on the ground.

Body mass is just one of several factors associated with primate terrestriality. Others include bigger home ranges and social groups, higher male: female ratios within troops, the use of open habitats, warmer and/or drier habitats, lower predation pressure, dietary shifts, and the use of tools (Bandini et al., 2022; Clutton‐Brock and Harvey, 1977; Estrada and Marshall, 2024; Fleagle, 2013; Heldstab et al., 2016; Kim et al., 2019; McCrossin et al., 1998; Meulman and van Schaik, 2013; Willems and Van Schaik, 2017; Wright et al., 2019). Many of these characteristics are also central to the study of human “uniqueness”. However, it is difficult to disentangle which, if any, of the factors cause an increase in primate terrestriality, and which are consequences of more time spent on the ground (McCrossin et al., 1998). Evidence, both from the fossil record and living primates, suggests that many of these traits fall in the latter category.

Data from the “Ecological traits of the world’s primates” database – representative of 504 primate species (Galán-Acedo et al., 2019) – show that more terrestrial species have larger home ranges and inhabit a greater number of habitat types than arboreal species, with body mass in terrestrial species being an additional predictor of home range size (see Supplementary Material for analyses). These extant relationships suggest that terrestriality, particularly in combination with an increase in body size, allowed species to disperse further and exploit novel and diverse environments.

Another likely consequence of spending more time on the ground appears to be tool-use (Falótico and Ottoni, 2023; Meulman et al., 2012; Ottoni and Izar, 2008). Once thought to be unique to humans, and specifically to the genus Homo (Kivell, 2015; le Gros Clark, 1966), tool-use is now studied in several species across the animal kingdom, and several lines of evidence link it with terrestriality. Tool-use is most prolific amongst primate and bird species that spend a significant amount of time foraging on the ground (Heldstab et al., 2016) and is more common and complex in terrestrial than arboreal settings (Meulman et al., 2012; Visalberghi et al., 2015). Terrestriality is also positively associated with technological diversity, as reflected in variation amongst capuchin populations (Falótico and Ottoni, 2023) and perhaps also in the higher rates and types of tool-use in chimpanzees compared to more arboreal bonobos (Doran, 1993; Samuni et al., 2022). It is thought that terrestriality facilitates more frequent and complex use of technology by increasing individuals’ exposure to a greater range of food and raw materials, as well as providing better opportunities for social learning. The stability of the ground frees up primates’ hands, acts as a more reliable substrate than tree branches for activities like nut cracking, and provides space for juveniles to observe and scrounge from more proficient tool users. Furthermore, the accumulation of materials at specific terrestrial sites increases encounter rates with the materials and settings needed for primates to engage in tool-use, learning, and innovation (Meulman et al., 2012; Ottoni and Izar, 2008; Visalberghi et al., 2015).

Orangutans, the most arboreal of the extant great apes, very rarely make complex tools in the wild, but are able to do so in captivity where they lead more terrestrial lives. This suggests that they are not limited by cognitive factors but by their ecological setting (Meulman and van Schaik, 2013). Baboons are another primate species that use tools in captivity, but not in the wild (Laidre, 2008). This lack of tool-use in the wild is somewhat surprising given their degree of terrestriality, but it serves as a reminder about the complex web of relationships that link ecology with behaviour. Whilst harder to investigate than differences in tool-use across chimpanzee groups, there is some indication that the type of object manipulation exhibited by different baboon groups might vary according to environmental factors (Hamilton et al., 1978). Similarly, whilst highly terrestrial geladas are predominantly gramnivorous, they also engage in a range of complex foraging behaviours (including digging, cleaning, and peeling various plant foods as well as catching and removing appendages of flying invertebrates), particularly in more intact ecosystems (Fashing et al., 2014). Comparative research could explore this further to understand how terrestriality influences the complexity or frequency of object manipulation and extractive foraging. If terrestriality conferred advantages to certain taxa, it likely drove further adaptation to ground-use as well as co-evolution of cognitive and technological abilities.

One factor that is more clearly a cause of variation in primate substrate-use is the risk of predation – or even just the perceived risk of predation. “Landscapes of fear” – the spatial distribution of perceived risk in environments (Gaynor et al., 2019; Palmer et al., 2022) – have been widely studied in modern ecological settings, and are increasingly recognised as having both immediate and long-term effects on the amount of time primates spend on the ground (Campbell et al., 2005; Hammond et al., 2022; Isbell and Young, 1993; Monteza-Moreno et al., 2020; Mourthé et al., 2007; Nowak et al., 2014). Fear of predators is also thought to be the primary driver of one of the most conserved behavioural traits across the primate order: ascending from the ground to sleep at night (Anderson, 1998; Bidner et al., 2018). We must therefore explore the relationships between predators and primate terrestriality throughout evolutionary history. It is also worth considering when and how hominins overcame the fear of predators to the extent that they no longer needed to climb to safety at night. This would surely mark the true transition to a “fully committed” terrestrial biped. The section below discusses methods and issues around studying primate landscapes of fear in the fossil record.

One avenue for further exploration is to examine primate fossils for tooth or claw marks. The Taung assemblage in South Africa, which includes the skull of an Australopithecus africanus infant, is likely to have been accumulated by a large bird of prey ~2.5 Ma (Berger and Clarke, 1995; McGraw et al., 2006), which serves as a reminder that hominins faced threats beyond large terrestrial carnivores, including raptors and crocodiles (Njau and Blumenschine, 2012). Many other assemblages that contain primate fossils – particularly in South Africa – show strong signals of carnivore activity, although it is difficult to determine which carnivores were responsible for marks observed on bones (DeSilva et al., 2013; Fourvel et al., 2018; Pickering et al., 2004; Val et al., 2014). For example, the Cooper’s D assemblage in South Africa contains a large proportion of large-bodied primate fossils, accumulated in a cave ~1.5 Ma. Fossilised primate genera include Papio, Theropithecus, Gorgopithecus, and Paranthropus, with 6.2% of the primate bones showing carnivore damage, including a mandibular fragment from P. robustus (Val et al., 2014). This has been attributed to leopards or hyenas based on comparisons with modern examples of bone damage, but the site was also inhabited by sabertooth cats, Megantereon and Dinofelis, whose signatures might be harder to detect due to a lack of a modern comparison (Val et al., 2014). At Swartkrans, a tooth-marked P. robustus cranium has also been attributed to leopard predation (Brain, 1981). However, isotopic analyses from carnivores at the site indicate that Megantereon and Crocuta species, as well as leopards, are all possible candidates for predators of the hominins and baboons at that site (Lee-Thorp et al., 2000).

Predators appear to have been less actively involved in the accumulation of Plio-Pleistocene fossils at eastern African hominin sites than southern African ones. However, there is still some evidence of likely carnivore damage to several hominin fossils from eastern Africa. These include signs of possible leopard predation on Orrorin tugenensis 6 Ma (Gommery et al., 2007), carnivore damage to Ardipithecus ramidus bones 4.4 Ma (Woldegabriel et al., 1994), and both crocodilian and mammalian carnivore damage to Homo habilis fossils 1.8 Ma (Njau and Blumenschine, 2012). Signs of fossil damage indicate that hominins continued to face threats of predation as they dispersed out of Africa (Margvelashvili et al., 2022) and through the recent past to the modern day (Camarós et al., 2016; Hart and Sussman, 2011). Whilst more research is needed to improve identification of specific predators from marks that they leave on bones, the presence of the marks themselves highlights the importance of considering the landscape of fear as an influential driver of behavioural evolution in hominins and other primate lineages.

Beyond identifying key predators of different primate species, advancing methods provide insights into other aspects of landscapes of fear throughout evolutionary history. Carnivoran fossil morphology reveals a lot about hunting styles and niche partitioning amongst guild members through time. For example, sabertooth predators have evolved multiple times throughout evolutionary history, in different families (another example of homoplasy). Compared to the more versatile diets of extant carnivores, sabertooths are characterised as hypercarnivores with dentition so specialised for cutting meat that they could not consume bone or other food items. Although often characterised as targeting megaherbivores, there is stronger evidence to suggest that they targeted medium-sized herbivores but had adaptations to perform quicker kills and more efficient stripping of meat from carcasses than conical-toothed cats (Mauricio, 2013). Amongst sabertooth taxa, there appear to have evolved two general “ecomorphs”, or strategies, for hypercarnivory. “Dirktooth” predators had very long sabers and a muscular build suited to a more solitary, ambush hunting strategy, likely conducted in closed environments. “Scimitar-tooths” on the other hand were more gracile with longer limbs and greater speed to pursue prey, likely hunting in more open environments and perhaps in groups (Mauricio, 2013). Building on this, isotopic analyses provide evidence that scimitar cats, such as the early Pleistocene Homotherium, were coursing predators that hunted predominantly large-bodied herbivores in open habitats. Meanwhile, dirktooth cats such as Megantereon were ambush hunters of browsers in closed habitats (Palmqvist et al., 2008). The inferred features of Megantereon indicate that it might once have filled a similar niche to the modern-day leopard, perhaps making it a key predator of medium to large-bodied primates – including early hominins.

Whilst informative, the examples above serve as a reminder that there are many features of these ancient landscapes of fear that do not have modern analogues, making it difficult to do more than hypothesise about the prey preferences and hunting techniques of animals like sabertooth cats, very large hyenas, and short-faced bears (Hart and Sussman, 2011; Van Valkenburgh et al., 2016). Additionally, whilst Africa has the most intact and diverse predator guild of the modern day (Cozzi et al., 2012; Dalerum et al., 2009), evidence from the eastern African fossil record suggests that the functional richness within the guild has declined by almost 99% since 3.5 Ma (Werdelin and Lewis, 2013b). This makes it difficult to imagine how prey species might have been affected by a far greater diversity of carnivores and very different intraguild dynamics. For example, Pleistocene ecosystems were home to many large (>100kg) hypercarnivores, several of which lived sympatrically. Today, only lions and tigers fill this niche and they do not overlap geographically. These differences in the make-up of extant and extinct carnivore guilds suggest that prey species, including primates, probably inhabited much more salient and complex landscapes of fear than exist today. On the other hand, predation pressure might actually be heightened in some modern ecosystems compared to paleo environments. This is because anthropogenically-induced habitat fragmentation and restricted-size reserves (protected areas that are smaller than naturally functioning ecosystems) might create more concentrated or “artificial” spatial overlap amongst carnivores and prey species compared to what would exist in an environment unmarred by modern human activity (Hayward and Slotow, 2009; Searle et al., 2021). This highlights the need to study community-level dynamics like predation and competition and how they interact with the diversity and intactness of an ecosystem. A better understanding of ecosystem regulation and trophic cascades is not only crucial for conservation of modern environments, but also to learn about significant ecological shifts in evolutionary history (Estes et al., 2011; Terborgh et al., 2001).

If fed into a more integrated and cyclical feedback loop, advances in both paleontology and behavioural ecology will contribute to increasingly accurate models of past landscapes of fear. From these we can then generate insights and hypotheses about the role of risk in behavioural evolution. The collection of more fossil evidence will provide us with information on which prey species were vulnerable to certain predators, and will increase the resolution of data about presence and abundance of carnivore species through time and space. Meanwhile, advances in remote sensing and community-level ecology will expand our understanding of extant intraguild dynamics and their effects on entire ecosystems, which could perhaps then be modelled in hypothetical environments without the artificial constraints of national borders or park boundaries.

Whilst it will take time to build increasingly accurate models of these broad paleo environments, there are particular sites at which localised patterns of carnivore speciation, extinction, and primate substrate-use might be studied. As discussed, primate terrestriality emerges in two primate clades at Maboko Island, Kenya, ~15 Ma. The simultaneous emergence of the characteristic in both the Victoriapithecus and Equatorius lineages precedes the expansion of grasslands in the region and appears to have happened when both taxa were relatively small-bodied (McCrossin et al., 1998). In the case of Equatorius, morphological analyses indicate that it likely engaged in palmigrade terrestrial locomotion which evolved independently to adaptations like the knuckle-walking seen in living hominoids (Patel et al., 2009). Researchers have hypothesised that a shift in diet might have caused the terrestriality seen in the lineages at Maboko Island (McCrossin et al., 1998). This hypothesis should be explored alongside considerations of pressures – including predation – that would have deterred them from descending to the ground. Perhaps the emergence of primate terrestriality was preceded by a drop in either the abundance or diversity of predators at the site? If temporal resolution is too low to explore sequential patterns at the site, the inferred landscape of fear at Maboko Island could be compared to those of other sites from a similar time period to investigate whether there were unusually low numbers of carnivores in the Maboko region ~15 Ma. It would also be a good site to study primate fossils more closely for carnivore damage, both as a proxy for predation rate amongst the sample and to identify key predators of primates at the site.

A similar example exists at Koobi Fora and surrounding deposits in the Omo-Turkana Basin, where the fossil record captures increasing terrestriality in both hominins and the cercopithecid genus, Theropithecus, over millions of years (Bobe et al., 2022; Cerling et al., 2013; Jablonski, 1993). Koobi Fora has yielded vast records of paleoclimatic, fossil, and archaeological materials representative of millions of years of ecological and technological transitions in the area (Bobe et al., 2022; Feibel, 2011). It is an excellent site for integrating evidence over broad geographic and temporal scales, with sediment layers ranging from 4.3 to 0.6 Ma, and with discrete subsections that can be geochemically matched across formations, allowing researchers to analyse temporal correlations of materials found across the Omo-Turkana region (Gathogo and Brown, 2006; Jablonski et al., 2008; Rogers et al., 1994).

The Koobi Fora record captures several longitudinal trends. It documents the presence of Au. anamensis ~4 Ma and the high diversity when H. habilis, H. rudolfensis, H. erectus, and P. boisei occupied the region between 2 and 1.4 Ma (Bobe et al., 2022). Its archaeological record demonstrates a shift from a few Oldowan sites before 1.9 Ma (Braun et al., 2010), through the transition to Acheulean technology 1.76 Ma (Lepre et al., 2011), to an abundance of many sites with large numbers of artefacts spread across the landscape by 1.5 Ma (Presnyakova et al., 2018; Rogers et al., 1994). This increase in production of lithic technology is proof that hominins were spending a significant amount of time on the ground by 1.5 Ma. It is worth exploring whether or when their levels of terrestriality decoupled from trends seen in other primate lineages at the site. For example, Koobi Fora also documents a rise in the numbers of Theropithecus, a primate genus that became increasingly terrestrial and widespread until it was the most common primate in the landscape by 1.5 Ma (Bobe et al., 2022; Jablonski, 1993). Broadly, the increasing terrestriality in both Theropithecus and hominins indicates that they were facing similar selection pressures. Indeed, Koobi Fora provides evidence of a shift from wooded environments to more open grasslands, particularly after 2 Ma (Bobe et al., 2022). It also provides evidence of the decline in both the functional and taxonomic diversity of the large carnivore guild in the area, between 2 and 1.5 Ma (Werdelin and Lewis, 2013b, 2013a) which might have released restrictions on primate terrestriality. Future work will need to examine the relationships between carnivoran and primate fossils from the site at higher temporal and spatial resolution, aided by new fossil discoveries, improved methodologies, and integrated insights.

As evidenced by studies of extant primate terrestriality, a behavioural shift can arise in a relatively short period of time within a small population of animals and a localised ecological setting. Over the course of primate evolutionary history, there have likely been various spatiotemporal pockets in which the “right” ecological factors have come together to favour terrestrial behaviour in primates, sometimes lasting long enough to shift species’ behaviour or even morphology. Using this behavioural ecology lens, we can better explore how locomotor diversity arose in Miocene and Plio-Pleistocene primates, and we might be able to pinpoint the conditions or place where terrestrial bipedalism emerged in our own lineage.