Morphology and environment in some fossil Hominoids and Pedetids (Mammalia)

Abstract

Linking the environment to functional anatomy is not an easy exercise, especially when dealing with fossils, which are often fragmentary and represent animals that are extinct. A holistic approach permits us to fill the gaps in reconstructing the evolutionary patterns in fossil groups. Identifying the environment where animals lived can help to interpret some anatomical structures and, vice versa, the functional morphological pattern can help to refine some fossil environments. Two examples focusing on locomotor behaviours in fossil mammals are considered in this paper: the hominoids and the origins of hominid bipedalism and the springing adaptations in fossil rodents (Pedetidae) in connection with different habitats. In the first case, the limits of the chimp‐based models and the necessity to take into account detailed environmental reconstructions will be addressed. The famous ‘savannah hypothesis’ is no longer tenable because the palaeontological data support a more vegetated environment for the origins of bipedal hominids. Data from the environment will be considered. The earliest putative hominid fossils which preserve skeletal remains of the locomotor apparatus show mixed adaptations to terrestrial bipedalism and arboreal activities. The second example focuses on the variation in springing adaptations in Pedetidae in the Lower Miocene of East Africa and Southern Africa. In the East, the sites where Pedetidae were preserved were mainly forested, whereas in the South the region was more open and drier, with extensive grassy patches. In the first case, pedetids were robust and heavy jumpers, whereas in the South they were smaller, their skeleton more gracile and their springing was lighter. During the desertification of the southern part of Africa, the large pedetid species became extinct, but a smaller species developed. In the case of primates, as in the case of rodents, the skeletal morphology was adapted to its environment.

Introduction

It is usually accepted that the morphology of animals is adapted to their habitats and linking functional morphology (related to diet and/or locomotion) to ecology has been the focus of numerous studies on vertebrates since the mid‐1970s. Morphology results from a compromise between size, substratum, environment and efficiency. Body size influences locomotor abilities, respiratory needs and other factors, but most of these data are not available from the fossil record. Reconstructing past morphology is thus particularly challenging for palaeontologists, who frequently deal with fragmentary bones or teeth, occasionally known from partial skeletons (which can be damaged or distorted), and sometimes ichnological traces. Comparative skeletal anatomy remains the basis of all studies, keeping in mind that there is no magic feature! A single bone can yield a lot of information or none, depending on its variability and/or the variability within the studied species. It is also important to study variability among primates and non‐primate mammals: for example, stabilization of a joint for terrestrial quadrupedalism can be similarly expressed in a baboon, an antelope or an elephant. Reconstructions are usually based on extant ‘models’ which may not have existed in the past, as indicated by early hominids such as Orrorin or the australopithecines, which exhibited their own locomotor pattern with a large component of bipedalism and climbing, a combination not known in modern hominoids. A pertinent case is the chimpanzee, which has been widely considered to exhibit the ‘primitive’ morphology, whereas modern humans are considered to exhibit a ‘derived’ one. This is not only an incorrect interpretation of the concept of primitive and/or derived, but also an erroneous interpretation of the genetic relationships between humans and African apes. Because chimpanzees are genetically closely related to us, their morphology was considered by some authors to be ancestral to hominids, which is an anti‐evolutionary statement! Humans and chimpanzees exhibit primitive and derived features in their own lineages. Considering their mode of locomotion, humans are derived in their permanent bipedalism just as chimpanzees are derived in their knuckle‐walking (Senut, 1989a,b, 1992). Bonobos have been considered for quite a while to be a good prototype of the common ancestor of African apes and hominids (Zihlman et al. 1978; Zihlman, 1984). Another factor which should be stressed is that anthropologists studying fossils tend to concentrate on the detailed anatomy of a structure but do not often reinsert it into a broader framework. An interesting aspect of the field is the search for the implications in terms of adaptability to a milieu or an environment. We can usually see the tree which hides the forest, but the forest remains forgotten. In this paper, two fossil mammalian examples are developed: (1) hominoid evolution and arboreality as a key factor in the emergence of hominids and (2) the evolution of locomotion in Pedetidae in different environments and in relation to the onset of desertification of the Namib. To address ecomorphology in fossils, studies must not be limited to morphology but must also include a broader dataset combining geological (including sedimentological and geochemical) contexts. By this means, the combination of different approaches may lead to a better understanding of the palaeoenvironment and of animal–environment interactions. In this article, I will focus on a few examples concerning arboreal locomotion in Lower and Middle Miocene hominoids, bipedality in Mio‐Pliocene hominids and springing locomotion in pedetid rodents which show variation through the Miocene in connection with desertification in Africa. Data yielded by the faunal and floral context of a fossil as well as results from isotopic analyses performed on associated large mammal remains will be provided to indicate the limits of the details we can obtain today.

There is no consensus about the definition of Hominidae, which creates a lot of confusion: for some authors it includes the large‐bodied apes from Asia, African Apes and Humans. For others, the large‐bodied apes from Asia belong to a distinct family, Pongidae, a position which is in agreement with the geological and palaeontological history of the group. Other authors prefer to keep only African Apes and Humans under Hominidae, whereas it is probably wiser to retain only Humans and their immediate precursors that show adaptations to bipedality in the family Hominidae, a position which is adopted here.

Problems with palaeoenvironmental reconstructions

Studying past life necessitates taking into account several sets of data which, when combined, yield a general picture of the animal living in its environment, after which we can compare variations in the environments through time. The more complete the faunal and the floral records, the more accurate will be the representation of the milieu. However, in some cases, the usually accepted ideas jeopardize the acceptance of new results. Concerning the origins of bipedalism, there have been some fashionable scenarios such as the ‘savannah hypothesis’ since Lamarck (1809) (see reviews in Senut, 2006a; Bender et al. 2012; Domínguez‐Rodrigo, 2014) which influenced research on human evolution for at least a century, despite increasing evidence from the field showing that early hominids did not live in an open grassy savannah (Napier, 1967; Coppens, 1985; Haile‐Selassie et al. 2004; Senut, 2006b; Bonnefille, 2010; Cerling et al. 1997, 2011; Roche et al. 2013). In most scenarios explaining the evolution of man and/or the origin of bipedalism, a savannah‐like vegetation was considered to be the key environment in which hominids became bipedal, in which their brain enlarged and in which they manufactured tools. But what is a savannah? (Fig. 1). There is a major issue concerning the definition of some biomes and ‘savannah’ and/or ‘wooded savannah’ is commonly used without clear definition. It is mainly a grassland ecosystem, with can contain a few or many trees. There is a ‘world’ of savannahs under tropical climates: more temperate, more or less wooded. This is a highly variable type of vegetation following the hydrographic network, the type of soil and the humidity in the environment. Figure 1(A–D) shows examples of what we may call ‘savannah’; however, their structures are different and locomotor strategies used by early hominids could have varied according to each of them. In 1967, J. Napier shed new light on the type of environment in which hominids emerged and evolved – ‘An environment neglected by scholars but one far better for the origins of man is woodland savannah, which is neither high forest nor open grassland’ – and he is probably right in his statement. He was referring to what was called later a ‘mosaic’ environment, a concept also widely used, but which is vague and poorly defined. It refers to the heterogeneity of the habitat (Reynolds et al. 2015). But it is obvious that a mosaic can be formed by quite different entities and mosaic environments in the Plio‐Pleistocene of Kenya and of South Africa were very probably different. Moreover, seasonal changes in the vegetation may have had an impact on hominid feeding behaviours and migrations. It is clear that woodlands would be appropriate habitats for the Late Miocene divergence between the lineages that eventually became chimpanzees and humans (Review in Elton, 2008) as shown by Australopithecus anamensis (Schoeninger et al.2003). Miombo woodland (Fig. 1D) has been considered to be a possible environment for evolution of early hominids (Sikes et al. 1999; Elton, 2008). It may have been present in the Upper Miocene in the Western Rift in Uganda (Dechamps & Ergo, 1994; Bonnefille, 2010) as well as in Ethiopia during the Mio‐Pliocene, as suggested by Bonnefille (2010): ‘In tropical Africa, many intermediate vegetation types offer both wooded and open habitat in close geographical proximity. Mixed tree and grass cover are among the widely spread vegetation and conditions for them persisted throughout the last 10 Ma’.

Figure 1.

Variability in extant African ‘savannahs’. (A) Wooded savannah with fever trees (Acacia xanthophloea) in Kenya. (B) Wooded savannah in Murchison Falls Park, Uganda. (C) Borassus savannah in Murchison Falls Park, Uganda. (D) Miombo woodland at Gorongosa Park, Mozambique (photo by M. Pickford). (E) Dambo in Karamoja, Uganda. (F) Dry woodland in Kibwezi, Kenya (photo by M. Pickford). (G) Typical grassland and woodland in the Toro Game Reserve, Uganda. (H) Riparian forest in West Nile, Uganda.

Other factors impact on the evolution of locomotor strategies such as the percentage of trees in the environment, the slope of the ground, the herbaceous cover which can help while climbing abrupt slopes, variable composition of the substratum (smooth, hard, wet, coarse, etc.) which is very rarely flat, and tree trunks which can be vertical or slanting. Primates exhibit a wide range of locomotor behaviours. In different wooded environment, the types of trees can also be different and this can induce variation in the locomotion and diets. Another more recent model has been proposed which considers that tectonics has been the driving evolutionary stimulus of the dispersal of hominids by providing a rough topography, and offering many infractuosities in the landscape in which to hide, rest, etc., the complex topography hypothesis, which highlights the fact that neither savannah nor arboreal environments were good models for explaining the origins of bipedalism in hominids (King & Bailey, 2006; Winder et al. 2012). However, the latter hypothesis does not take into account the pre‐hominid period. And it is difficult to accept that only the topography played a role in early hominid evolution. Moreover, palaeontological evidence concerning early hominids suggests an important component of tree‐climbing. The topography has played a role in the sense that it affected the vegetation in which our ancestors could live and survive, but probably more so after hominids became true bipeds. At the time when Orrorin lived, the major faulting of the Tugen Hills block had not yet occurred and the topography was probably much smoother than is the case today (Pickford, 1975a,b, 1978; Chapman et al. 1978; Hill et al. 1986; Williams & Chapman, 1986). Riparian forests or gallery‐forest suggested by several faunal and floral analyses can also be considered as good environments for early hominids (see Singer‐Polignac Symposium Coppens, 1985; Potts, 1998; Pickford & Senut, 2001; Vignaud et al. 2002; Haile‐Selassie et al. 2004) as they provide shade, food and water necessary to survive and it is a suitable environment for arboreal hominoids. Reconstructing past environments is not an easy task and an early hominid from Ethiopia, Kenya or South Africa may not exhibit the same morphologies in relation to these variations. There is probably not a clear, uniform environment in which early hominids lived as we can see today in gorillas, the mountain gorilla being less arboreal than the lowland gorilla, or in chimpanzees, which inhabit miombo woodland in Tanzania or more forested areas in Western Uganda or the Democratic Republic of Congo. They exhibit anatomical differences in their skeleton (Schultz, 1930, 1934; Tocheri et al. 2011; Dunn et al. 2014) but even if they live in a rough environment (altitude, slopes) (Taylor & Goldsmith, 2003), they still deal with vegetated environments.

The fossil hominoids

The reconstruction of hominid bipedalism has been biased for decades by the fact that researchers were comparing hominid fossil remains mainly with those of modern apes, the Miocene apes being marginalized most of the time. To some extent, this is still the case today, as most articles are based on a modern comparative typology. However, taking the Miocene data into account provides the time dimension, the evolutionary dimension and a completely new framework. There is a tendency to forget that modern locomotor patterns are different from the fossil ones and the best example comprises the early hominids which exhibit a mixed adaptation: terrestrial and arboreal, a system which does not exist anymore except in a weakened occasional form. Bipedalism is not restricted to humans, all primates can walk bipedally; however, humans are the only animals capable of walking on two legs for long distances and for a long time. This is reflected in our specific musculo‐skeletal modifications, but also in our nervous system and respiratory system. How did human bipedalism evolve through time, and what was the impact of the environment? To answer the question, we will consider some key fossils ranging from the Lower Miocene to Upper Pliocene.

In the Miocene, few skeletal remains permit the reconstruction of the locomotor patterns in hominoids.

Hominoids discovered in the Lower Miocene of Uganda and Kenya traditionally attributed to Proconsul major were generally considered to be a scaled‐up version of Proconsul africanus, Proconsul heseloni and Proconsul nyanzae (Andrews, 1978; Bishop, 1964a,b; Martin, 1981; and see Pickford et al. 1999 for review and Harrison, 2010). However a detailed study of the cranio‐dental material shows that this taxon is different from Proconsul. Proconsul major was renamed in a new combination Ugandapithecus major (Senut et al. 2000; Pickford et al. 2009b). This is still a matter of debate today (see for example Harrison, 2010 and the more recent but misleading paper by McNulty et al. 2015). However, the differences between Proconsul and Ugandapithecus are evident not only in the dentognathic anatomy but also in the postcranial morphology. Proconsul is basically a generalized arboreal quadruped (based on the Rusinga fossils) as evidenced by its limbs and vertebral column (Napier & Davis, 1959; Walker & Pickford, 1983; Senut, 1989b; Rose, 1993, 1994, 1997; Walker, 1997; Nakatsukasa, 2006; Gommery, 2006 2004), whereas Ugandapithecus exhibits a more ape‐like anatomy, suggesting that it was a heavy climber as shown notably by the morphology of its femora (Gommery et al. 1998, 2002) (Fig. 2) but also of the tibia (Rafferty et al. 1995). MacLatchy & DeSilva (2009) noticed differences between humeral, radial and tibial large hominoid fragments from Napak (Uganda) and those of P. africanus, P. heseloni and P. nyanzae. These data suggest a different locomotor repertoire which supports the validity of the genus Ugandapithecus. The Ugandapithecus proximal femur from Napak IX and the femoral shaft from Napak I (Gommery et al. 1998, 2002) differs from those of Proconsul in several features such as the position of the greater trochanter which is lower, the intertrochanteric line which ends in the lesser trochanter but not in Proconsul from Mfwangano (KNM RU 13142 A), the tuberculum for the insertion of m. quadratus femoris is well expressed at Napak, but is absent in Proconsul, which is confluent with the internal border of the greater trochanter. From Songhor, a hominoid distal humerus (KNM SO 31232) has been described (Gebo et al. 2009) which has been attributed to Proconsuloids (either P. africanus or Rangwapithecus gordoni). It is said to be similar anatomically and functionally to Proconsul but was allocated to R. gordoni on the basis of size. However, the morphology is different from that of Proconsul: the capitulum is elongated in Proconsul from Rusinga and round in the specimen from Songhor, the morphology of the trochlea which is shorter and the zona conoidea narrower and the lateral flange for the m. brachioradialis more expanded in the specimen from Songhor (Fig. 3). Another large fragmentary distal humerus from Songhor, KNM SO 1007 (Senut, 1989b), exhibits a wide brachioradialis flange, as is the case in KNM SO 31232, but not in Proconsul (Fig. 3). KNM SO 1007 is larger than KNM SO 31232, but this could be due to sexual dimorphism. It may be considered that the Songhor and Napak large hominoids do not belong to Proconsul or to Rangwapithecus, but more probably to Ugandapithecus. Ugandapithecus major was not just a scaled‐up version of P. africanus, P. heseloni and P. nyanzae and the naming of a new genus is valid. It is confirmed by a new discovery made during the 1998 survey of the Uganda Palaeontology Expedition, which has been prospecting the Napak area since 1985 and which has greatly increased the collections started by Wayland in the 1920s and continued by Bishop and Wilson in the 1960s. In the Lower Miocene deposits at Napak I a left scapular specimen found in 1998 (NAP I 80′98) (Pickford et al. 1999; Senut et al. 2000) (Fig. 4) is attributed to Ugandapithecus major, for which we give here the first detailed analysis. Although the specimen is not complete, it is the most complete and best preserved scapular fragment of an African Miocene hominoid. It preserves the cavitas glenoidalis, the base of the spina scapulae and the proximal part of the axillary border (Fig. 4). The glenoid margin is beaded as is usual in hominoids (B. Senut, pers. obs.). The upper part of the glenoid cavity is broken away. as is the tuberculum infraglenoide. The ovoid cavitas glenoidalis (minimal estimated measurements, length 35.0 mm; breadth 27.1 mm) is morphologically close to that of modern apes (Roberts, 1974; Senut, 1981) (Fig. 5) and does not exhibit the piriform shape which can generally be seen in other Simiiformes. The axillo‐glenoid angle cannot be estimated as the axillary border is not well enough preserved. In lateral view, the notch between the spine and the glenoid is deep, the root part of the spine thick, and the dorsal projection of the spine well marked. Taking into account that the superior border of the glenoid fossa is broken, the base of the spine would be located at mid‐height of the fossa and resembles the morphology usually seen in Simiiformes, which suggests a more cranially oriented scapula (Senut, 1981) (Figs 4 and 5) linked with the upper limb usually held above the head (Larson, 1995). This width is interesting as it permits the distinction of locomotor groups in primates, the more arboreal species exhibiting greater widths. This position can be more accurately estimated by measuring the width of the infraspinatous fossa at the neck of the scapula following the method of Larson (1995). She measured the minimum width of the infraspinatous fossa at the neck of the scapula and calculated an index representing the ratio between the interval between this distance and the glenoid size. She confirmed that this distance relative to glenoid size is important for distinguishing locomotor groups and concluded that hominoids have the widest intervals. The distance between the base of the spine and the axillary border is estimated to be 17.2 mm in the Ugandapithecus scapula, which gives an index between 61 and 66 (related to maximum width and length of the glenoid) (Fig. 6). This value falls within the range of variation of modern hominoids (including Pongo) and supports its more ape‐like morphology in relation to arboreal abilities (probably with an erect trunk during climbing) in this early Miocene hominoid. Moreover, the base of the spine is thickened (8.7 mm), a feature which is also present in Proconsul from Rusinga (Walker & Pickford, 1983) and in Nacholapithecus (Senut et al. 2004). The buttressing is probably related to reinforcement of the spine, which would suggest bulky muscles and/or powerful stabilization of the joint (Senut et al. 2004). According to Lewis (1989), the tali and calcanei from Rusinga and Songhor are ‘devoid of the specializations distinguishing Pan and Gorilla’. However, the cuboid facet of the calcaneus appears to be morphologically more like Pan in the specimens of Proconsul from Rusinga, but is strikingly close to that of Pongo in the Songhor fossils. This reinforces the differences between Proconsul and Ugandapithecus. Ugandapithecus major was probably a heavy, large‐bodied tree‐climber, with erect trunk. The environment where Proconsul lived was reconstructed from the floral remains (Chesters, 1957; Collinson et al. 2009; Maxbauer et al. 2013) of riparian forest, composed of patches of forested areas and woodland in a strongly seasonal and warm climate. This is in agreement with the data obtained from the land gastropods (Pickford, 1995), which show that the environment was composed of savannah and bush, with gallery forest, with evergreen parts. At Napak, the levels where Ugandapithecus remains were found have yielded fossil mammalian remains and gastropods that indicate a well forested and humid environment (Pickford et al. 1986a; Pickford, 2004; Roche, 2012), which would fit with a large arboreal climber such as Ugandapithecus.

Figure 2.

Proximal femur of Ugandapithecus major (NAP IX 46′99, NAP IX B 64; NAP IX 65, P. 67). (A) Posterior view; (B) anterior view. Scale bars: 1 cm.

Figure 3.

Comparisons of humeri of Proconsul heseloni (KNM RU 2036AH – reversed) from Rusinga and Ugandapithecus major from Songhor area (KNM SO 1007 from Senut, 1989b; KNM SO 31232, adapted from Gebo et al. 2009). (A) Posterior views; (B) anterior views. Note the distinctive morphology of the distal humerus in KNM RU 2036AH and KNM SO 31232.

Figure 4.

Scapula of Ugandapithecus major (NAP I 80′98). (A) Dorsal view; (B) ventral view; (C) lateral view; (D) axillary view. Scale bars:  1 cm.

Figure 5.

Comparisons of scapulae of extant primates in lateral view (from Senut, 1981).

Figure 6.

Means of relative width of the infraspinatous fossa/neck in some Simiiformes (adapted from Larson, 1995; Senut et al. 2004). As the glenoid is slightly damaged in the Ugandan specimen, the three red crosses correspond to the range between the biggest and smallest values. It is noticeable that the value falls within the large ape variation including Pongo.

Most of the confusion concerning the middle Miocene hominoids is due to the fact that there is probably a wide range of locomotor behaviours which are much more diverse than what can be seen in modern non‐human hominoids. However, it is also probably due to misinterpretations of palaeontological data. At Moroto in Uganda, fluviatile deposits have yielded large hominoid remains. There is debate about the dating of the strata. The most plausible date based on radiometric and biochronological data is around 17.6 Ma (Pickford et al. 1986b); the older estimation of more than 20 Ma (Gebo et al. 1997) is not supported by the fauna. Despite the usually accepted view that only one large hominoid is known at Moroto, Morotopithecus bishopi, the available data support the presence of two large hominoids (Pickford et al. 1999; Senut, 2006a). It has been shown that Morotopithecus is a synonym of Afropithecus (Pickford, 2002; Patel & Grossman, 2006). The most complete right femur from Moroto published by Gebo et al. (1997) was wrongly reconstructed due to the fact that some fragments in Bishop's collection housed in the Uganda Museum in Kampala (Bishop, 1963, 1964a,b) were not recognized by these authors (Fig. 7). Another reconstruction was proposed (Senut, 2012) in which the femur is shorter. The Moroto snout is clearly Afropithecine‐like, but we cannot make comparisons with an Afropithecus femur as there is none preserved in the Kenyan collection from Buluk and Kalodirr (Leakey & Walker, 1997) despite the mention of a specimen by Ward (1997, table p. 103). Most other authors have suggested that the postcranial elements of Afropithecus were close to Proconsul and, more probably, P. nyanzae. The Moroto specimen is not dramatically different from the specimen from Rusinga KNM RU 2557 (Bacon, 2001) or from Mfwangano KNM MW 13402 (Ward et al. 1993) (Fig. 7), but the fragmentary left proximal femoral fragments from Moroto resemble those of Ugandapithecus (Gommery et al. 1998, 2002; Pickford et al. 1999). The attribution of the Moroto glenoid cavity (MacLatchy et al. 2000) remains uncertain as it is fragmentary and it may belong to a non‐primate mammal (Pickford et al. 1999; Senut, 1998; Benefit, 1999; Johnson et al. 2000), probably Morotochoerus (Pickford, 2011). Some vertebrae from Moroto and especially a lumbar vertebra of a large hominoid add fuel to the debate (Walker & Rose, 1968). Leakey et al. (1988) and Ward (1997, 1998) showed that limb bones of Afropithecus resemble those of Proconsul, from which it can be proposed that the vertebral morphology would have been similar. However, the vertebra from Moroto, UMP 67.28, differs strongly from those of Proconsul in indicating a more dorsostable lumbar column (Walker & Rose, 1968; Gommery, 2006; Nakatsukasa, 2008; Nakatsukasa & Kunimatsu, 2009). Thus, it is quite probable that the Moroto vertebra belongs to Ugandapithecus, a genus that is also present at Moroto (Pickford et al. 2009a, b). This is the first evidence of modern hominoid morphology of a lumbar vertebral block. In Moroto, the environment, based on the faunal associations, was probably comparable to a modern dambo, a shallow wetland with wooded patches (most of the time miombo woodland) which dries up during the dry season and can be flooded during the rainy season. It occurs today in Uganda in the flat plains between Soroti and Katakwi in Karamoja. In this environment, apes would have found the water necessary for their survival as well the food they needed (Figs 1E and 8).

Figure 7.

Comparisons of the femora of Proconsul nyanzae from Rusinga (KNM RU 5227) (A,E) (from Bacon, 2001), Mfwangano (KNM MW 13142 A) (B,F) (cast) and the composite Moroto specimen (MUZM 80 + MOR II 23′05+ unnumbered fragments from Bishop's collection) (C,D). Scale bar:  1 cm. The orientation of the caput femoris in the Moroto specimen must be accepted with caution as it was not originally correctly glued.

Figure 8.

Modern dambo in Eastern Uganda. Top left: during the rainy season, Right: detail of the preceding image; Bottom view: same area during the dry season.

Another hominoid taxon from the Middle Miocene is Nacholapithecus of the Aka‐Aiteputh Formation (Nachola, Kenya). It is well represented, notably by several partial skeletons, of which a rather complete one is known. The individuals seem to have been buried alive in a mud flow (i.e. lahar deposit) (Sawada et al. 2006). The material has been described (Ishida et al. 1999, 2004; Nakatsukasa et al. 1998, 2003a,b, 2007; Nakatsukasa & Kunimatsu, 2009) and I will not give details of the anatomy. What is interesting is that this hominoid exhibits a mixture of Proconsul‐like features in the vertebral column (except for the morphology of the lumbar spinous process), the narrow trunk, and more ape‐like ones with predominance of the upper limb compared with the hindlimb. As suggested by its body proportions, the shoulder was powerful in the reinforced base of the spine (Senut et al. 2004) and generally the animal was adapted to orthograde climbing and bridging. Some of the shoulder features recall modern colobines and the ratio estimating the position of the spine has a value of 50.9, which falls within the variation of the Colobine and Platyrrhine monkeys. Moreover, the base of the spine is thickened as in Ugandapithecus and to some extant Proconsul, which suggests a reinforcement of the pectoral girdle. Even if Nacholapithecus exhibits some ape‐like features in its skeleton, it was not specialized for suspensory activities (Nakatsukasa et al. 1998, 2003a; Ishida et al. 2004; Senut et al. 2004, 2004). It lived in a wooded environment (Pickford et al. 1987; Nakaya & Tsujikawa, 2006) and the anatomy of Nacholapithecus is compatible with this kind of environment.

Other hominoids are known from the Miocene of Eurasia and Africa. The Spanish ones are better preserved. They had an orthograde trunk and were tree‐dwellers, as evidenced by Hispanopithecus (Moyá‐Solá & Köhler, 1996) and Pierolapithecus (Moyá‐Solá et al. 2004), but none of them matches the patterns of modern apes. Miocene locomotor adaptations were different from those of extant hominoids (Alba, 2012; Moyà‐Solà et al. 2009). They were all living in arboreal and more or less humid environments, and they display a variety of locomotor patterns: Hispanopithecus in a closed forest, marshy environment (Casanovas‐Vilar & Agustí, 2007; Marmi et al. 2012) and Pierolapithecus in very humid and warm forests (Casanovas‐Vilar et al. 2008). These environments are compatible with their arboreal life inferred from their skeletons.

Considering the earliest hominids in Africa, the only material which may yield some information for the late Miocene are the bones (even fragmentary) of Orrorin tugenensis. The femur (Figs 9 and 10), as previously described (Senut et al. 2001; Pickford et al. 2002; Galik et al. 2004; Richmond & Jungers, 2008), exhibits some features linked with bipedalism. Compared with modern chimpanzees of similar size, the femoral shaft is longer, the neck is long and narrow, the trochanteric fossa is poorly depressed and there is a precursor of the linea aspera. The distribution of the cortical bone in the neck is like that of humans and australopithecines. However, although a biped, it was also a climber, as shown by its humeral fragment, which exhibits an important brachioradialis flange and the manual phalanx which is markedly curved. In addition, a distal thumb phalanx of Orrorin (Gommery & Senut, 2006) presents some interesting features usually thought to be related to the capacity to manipulate and/or to manufacture tools (Susman, 1988). However these traits are present in Orrorin, which did not have any tools (as far as we know). A detailed comparative analysis of the specimen (BAR 1901′01) (Gommery & Senut, 2006) suggests that the phalanx indeed belongs to a hominid (deep depression for the m. flexor pollicis longus with a marked asymmetry of its radial edge, distally a transversal bulge, a horseshoe‐shaped asymmetrical apical tuft with an ungual spine and the presence of an ungual fossa). Having a human‐like morphology is not a synonym for having the capacity to manipulate and/or manufacture tools (Gommery & Senut, 2006; Senut, 2012). As shown by the rest of the skeleton, Orrorin was still spending some time in the trees and the human aspect of its distal thumb phalanx could be related to the developed precision grip essential for climbing and balancing for an animal which differs from the modern apes in proportions and morphology. This precision grip would be an advantage for manipulating and/or manufacturing tools in later hominids. This could be considered as an exaptation (Gould & Vrba, 1982).

Figure 9.

Comparisons of femora of an extant chimpanzee (left) and Orrorin tugenensis (right). (A) Posterior views; (B) anterior views. Scale bar: 1 cm.

Figure 10.

Comparison of anterior views of femora of Orrorin tugenensis (BAR 1002′00) (left) and Australopithecus afarensis (AL 288. 1) (right). Scale bar: 1 cm.

Pedetidae from the Sperrgebiet

Another mammalian example is examined here: some aspects of pedetids. Today, pedetids (spring hares) constitute an exclusively African group of saltatorial rodents in Eastern Africa (Pedetes surdaster) and Southern Africa (Pedetes capensis) of about the same size. They are known for their kangaroo‐like springing adaptations evidenced in their gracile skeleton with long lower limbs and reduced upper limbs. The earliest representatives of the family are recorded in the Lower Miocene of Southern Africa and Eastern Africa dated around 20 Ma. Initially described by MacInnes (1957), a giant form of Lower Miocene pedetid from Songhor and Rusinga Island (Kenya) was described as a new rodent genus and species, Megapedetes pentadactylus. One of its distinctive features is the presence of five digits in the foot, contrasting with the four‐toed modern Pedetes. Remains of Megapedetes have been recorded in the Lower Miocene (20–21 Ma) of East Africa and Middle Miocene (ca. 17.4 Ma) of the Sperrgebiet, Namibia (Senut, 1997; Mein & Senut, 2003; Mein & Pickford, 2008). However, from the older strata, only a single tooth of Megapedetes is known at Langental, the richest deposits being the Lower Middle Miocene ones at Arrisdrift along the Oranje River, which have yielded several cranio‐dental as well as well‐preserved postcranial specimens, belonging probably to two species on the basis of size, Megapedetes gariepensis the large one and Megapedetes pickfordi the smaller one. Most of the postcranial bones (femur, tibia, calcaneum and metatarsals) belong to the latter species and will be considered here. Detailed morphological comparisons were made with modern Pedetes (Senut, 1997; Mein & Senut, 2003) showing that the postcranial bones of the Kenyan Megapedetes are generally very robust (MacInnes, 1957). The femur (Fig. 11) has a long neck in modern pedetids but a short to very short one in Megapedetes; the greater tuberosity is medio‐laterally expanded in Megapedetes, but antero‐posteriorly flattened in the modern specimens; in distal view, the trochlea femoris is wide and not deep in the fossils, whereas it is long and narrow in the modern bone; the femoral shaft is cylindrical, robust, but in Pedetes it is antero‐posteriorly flattened and slender. In the Southern African fossils, these characteristics are present, but to less of an extreme than in the Kenyan fossil. The tibia (Fig. 12) in Megapedetes has a rounded shaft, a distal articular surface which is wide medio‐laterally, deep and oblique (in distal view); in Pedetes the shaft is antero‐posteriorly flattened, and the distal articular surface is narrow medio‐laterally, very deep and very oblique (in distal view), the processus posterior medialis is salient in Pedetes, but non salient in the fossils. In Megapedetes, the calcaneum (Fig. 13) is long but not as elongated as in Pedetes; the two talar facets are separated and the lateral one salient (reflecting the development of the sustentaculum tali) whereas in Pedetes, the facets are confluent and not salient. The shape of the cuboid facet is also distinct in the two genera: in Megapedetes it is wide, short and thick, and in Pedetes it is elongated and triangular. The metatarsals V also exhibit some differences: a robust, cylindrical and straight shaft in Megapedetes vs. a gracile, flattened and oblique shaft in Pedetes. Moreover, the presence of a robust metatarsal I (absent in Pedetes) indicates good stabilization and reinforcement of the foot on the ground during springing activities. The combination of these features suggests that Megapedetes was a heavy springer (Senut, 1997). The Middle Miocene species from Namibia mainly resembles the older East African one but it is not as large and robust as shown in Figs 11, 12, 13. However, compared with the extant genus, it is much more robust. What about the environments in which these animals lived? In Eastern Africa (Songhor in Kenya) and Southern Africa (Sperrgebiet in Namibia) the environments were different: during the Lower Miocene, Western Kenya was well forested; many hominoids lived there and a diverse fauna included forest‐dwelling animals (such as Anomalurids, Tragulids). In this closed environment, pedetids could not jump as easily as in a more open environment. They were big but not as agile as the modern species adapted to desert conditions. In the middle Miocene of Namibia, the environment was not as forested as the Kenyan sites but was a warm and humid wooded area, and in some places, there were more open grassy patches where hypsodont elephant‐shrews were common (Pickford & Senut, 2003; Senut, 2003, 2008). This environment was better suited for a more slender type of Megapedetes which was smaller and less robust than the East African form. The proportions of the fossils are difficult to compare as the bones are not complete, but it seems that the proportions their lower limbs were not as developed as those of the extant animals.

Figure 11.

Comparisons of pedetid femora. (a) Megapedetes pentadactylus (from the Lower Miocene of Songhor, Kenya (from MacInnes, 1957); (b) Megapedetes pickfordi (AD 215′95) from the Middle Miocene of Arrisdrift, Namibia; (c) extant Pedetes capensis (reversed for comparisons). (1) femoral head; (2) femoral neck; (3) greater trochanter; (4) lesser trochanter; (5) insertion for lig. teres; (6) femoral trochlea; (7) intertrochanteric fossa. Scale bar: 1 cm (adapted from Senut, 1997). (A) Anterior views; (B) distal views (modified from Senut, 1997).

Figure 12.

Comparisons of pedetid tibiae. (A) Megapedetes pentadactylus from the Lower Miocene of Songhor, Kenya (from MacInnes, 1957); (B) Megapedetes pickfordi (AD 216′95) from the Middle Miocene of Arrisdrift, Namibia (reversed for comparisons); (C) extant Pedetes capensis. (1) talar articular surface; (2) malleoli. Anterior views. Scale bar:  1 cm (modified from Senut, 1997).

Figure 13.

Comparisons of pedetid calcanea. (a) Megapedetes pentadactylus from the Lower Miocene of Songhor, Kenya (from MacInnes, 1957); (b) Megapedetes pickfordi (PQAD 2019) from the Middle Miocene of Arrisdrift, Namibia; (c) extant Pedetes capensis. (from MacInnes, 1957). (1) Anterior talar facet; (2) posterior talar facet; (3) cuboid facet; (4) sustentaculum tali; (5) tuber calcanei. Scale bar:  1 cm (modified from Senut, 1997).

Discussion

Considering the available data on the Miocene hominoid (including hominids) locomotor behaviours, it is obvious that these hominoids were adapted to more or less wooded environments, an aspect which is confirmed by the data from palaeontology and geochemistry (Pickford et al. 1987; Wang & Cerling, 1994; Cerling et al. 1997, 2011; Lee‐Thorp et al. 2003; Nakaya & Tsujikawa, 2006; Uno et al. 2011; Roche, 2012; Domingo et al. 2013; Roche et al. 2013, 2013). We do not know all the locomotor repertoires for the numerous Miocene hominoid species, but the more complete skeletons indicate clear arboreal adaptations such as arboreal quadrupedalism, climbing and orthograde postures. Some of these species might have been terrestrial for some of the time, but as far as we know there is no good anatomical, functional and biomechanical evidence of knuckle‐walking adaptations in the Miocene palaeontological record, nor any modern ape type of locomotion (see Crompton et al. 2008). At 6 million years, early hominids show a combination of adaptation to arboreal life and bipedality. The same is true of australopithecines such as A. afarensis, A. prometheus and A. sebida from the Pliocene. In the Pliocene Ardipithecus ramidus this duality of behaviours has been reported (White et al. 2009), but the evidence for bipedalism has been challenged by several authors (Sarmiento, 2010; Wood & Harrison, 2011; Senut, 2012, 2014). However, it seems that a group of Miocene hominoids developed some abilities to walk on two legs and to climb trees but up to now it remains difficult to evaluate the percentage of time devoted to each type of behaviour and how exactly to reconstruct them. Several hominoids could have developed at different periods and/or in different regions the capacity to move on two legs, but the direct link with our own ancestors and our own bipedalism is still debated. These results also indicate that bipedalism originated in a wooded environment as several authors have suggested for a long time (see Thorpe et al. 2007, 2014) and contra the established view that these behaviours were different from the ones seen in modern apes, an idea which has been marginalized for a long time but seems to gain more support. They were probably several hominoid genera in the upper Miocene which lived in a more or less wooded environment and which developed different arboreal and terrestrial behaviours. In this sense, studies of modern apes will help us to understand these possibilities which need to be developed to better understand the variability (Thorpe & Crompton, 2006; Crompton et al. 2010 among others).

Concerning pedetids, it is interesting to evidence variability in the springing adaptations in relation to the environment. In the Lower Miocene of Namibia, other pedetids are represented by Parapedetes (Stomer, 1926) and Propedetes (Pickford, Mein, 2011) which are quite small and gracile. For Parapedetes, a complete skeleton is known from Elisabethfeld which was described by Stomer (1926). It is basically a smaller version of the modern pedetids and it might have inhabited grassy patches present in a wooded and humid environment (Pickford & Senut, 2003). Their main food competitors were the Macroscelididae (elephant‐shrews), which were hypsodont to very hyposodont, and they may have kept a low profile when facing this well‐developed group.

Concluding remarks

The samples from two different mammalian groups show not only that locomotor reconstructions can be used to understand the environment, but also (in some respects) that knowledge of the past environment can help better to reconstruct the fossil behaviours. This is why functional locomotor studies of the animals which lived in the same environments as our ancestors, can help to reconstruct the environment of upper Miocene hominoids. Another question is raised: we understand quite well the locomotor patterns of the earliest hominids, but what about those of the ancestors of gorillas and chimpanzees? We are beginning to fill the gaps in the histories of the African great apes (Pickford & Senut, 2005; Kunimatsu et al. 2007; Suwa et al. 2007; Pickford et al. 2009a; Senut, 2015). However, no postcranial bones are yet known for Samburupithecus, Nakalipithecus, Chororapithecus, the Lukeino and Ngorora non‐hominid hominoids, and the proto‐chimpanzee from Niger. Interestingly, Ardipithecus , originally claimed to be the earliest bipedal hominin (White et al. 2009), is now considered by the same authors to be a common ancestor of chimpanzees and humans (White et al. 2015). However, in the absence of a good postcranial record of Late Miocene African apes, the debate remains open. This change of interpretation reveals that we have to be extremely cautious when dealing with crushed skeletons that have to be reconstructed, because many reconstructions are possible. Concerning pedetid evolution, new surveys in the fossil dunes of the Namib Desert have yielded postcranial elements of Propedetes and Parapedetes; it seems that we can evidence changes in the bones tracking the processes of desertification. This is under study at the moment.