One of the most contentious areas of paleoanthropological research has been determining whether the knuckle-walking behavior of Pan and Gorilla is a homologous or homoplastic trait. If it is a shared trait, this would imply the ancestor of the earliest bipedal hominins was also a knuckle-walker. If not a shared trait, then the positional behaviour from which hominin bipedalism evolved becomes even more nebulous. There are two places where we can look for clues to help resolve this debate: the fossil record, and our closest extant primate relatives. Evidence from the former is obscured by a fragmentary record, while evidence from the latter is obscured by the process of evolution and the passage of time. Although the homologous position is the most parsimonious, this does not necessarily mean it is correct. In this post I’ll examine the most recent iterations of this dispute in the primatological and paleontological literature. Focus is placed upon recent studies of extant great ape carpal and upper limb morphology, as well as an examination of the positional behaviours which preceded knuckle-walking in late Miocene hominoids. From this it is demonstrated that although Pan and Gorilla do exhibit different forms of knuckle-walking, this does not necessarily mean that this trait evolved independently. Instead, it suggests that knuckle-walking first evolved approximately 10 MA in the common ancestor of the hominid family, and, took unique forms in Pan and Gorilla as each species adapted to different ecological habitats.
In order to evaluate the evidence in the debate surrounding the evolution of hominoid knuckle-walking, this analysis begins by assuming the most parsimonious construction of the African great ape phylogeny as the null hypothesis. As knuckle-walking is a trait found in both Pan and Gorilla, the conclusion which makes the fewest new assumptions is that this trait was shared by a common ancestor of both these species, and, therefore, the lineage that leads to humans. Assuming this position a priori is consistent with a standard cladistic analysis (Begun, 2007; Lockwood and Fleagle, 1999; Hennig, 1975). The patterns of evolution are the result of historical contingencies, stochastic processes, and adaptation. Before one can begin to speculate as to how evolutionary change may have occurred, one must first recognize the pattern. Thus, the remainder of this post will examine recent challenges to the observed patterns of knuckle-walking evolution, beginning first with evidence from extant great apes and then moving into the hominoid fossil record.
Kivell and Schmitt (2009) provide a considerable obstacle to the idea that knuckle-walking in Pan and Gorilla is a homologous trait. They do this by examining the carpal bones of gorillas (Gorilla sp.) and chimpanzees (Pan troglodytes, Pan paniscus), morphologically and developmentally. Kivell and Schmitt (2009) propose that if the features of the African ape wrist did evolve from the same ancestral condition, they would expect to find three specific patterns. First, the features of the wrist which can be attributed to knuckle-walking should be found in both species. Secondly, as gorillas have been observed to spend more of their time knuckle-walking, and bear heavier loads than chimpanzees, these features should be more pronounced in gorillas. Finally, gorillas have a larger body mass, a faster growth rate, and they engage in more frequent knuckle-walking behaviour at an earlier age than chimpanzees; therefore, the features attributable to knuckle-walking should appear earlier in their ontological development (Kivell and Schmitt, 2009).
In regard to the first hypothesis tested by Kivell and Schmitt (2009), a statistically significant difference in the features of the carpal bones was detected. The specific features being examined included the dorsal concavity of the scaphoid, the beak of the scaphoid, the distal concavity of the capitates, the dorsal ridge of the capitates, the distal concavity of the hamate, and the dorsal ridge of the hamate. Kivell and Schmitt (2009) showed that only two of these six traits were shared by Pan and Gorilla. In the cases where these apes shared morphological features (distal concavity of the capitate and distal concavity of the hamate), their presence in both species was not absolute. For example, the dorsal concavity of the capitate was present in 100% of the all adult Pan specimens, but only 75% of the adult Gorilla specimens (Kivell and Schmitt, 2009). An example of a trait not commonly shared was the dorsal beak of scaphoid which appeared in 96% of the adult P. troglodytes specimens, but only 11% of the adult Gorilla specimens shared this trait. Their conclusion was that gorillas lacked any key features which had been previously assumed by multiple authors (Begun, 2004; Richmond et al., 2001; Richmond and Strait, 2000; Begun, 1992; Corruccini, 1978) to be crucial components that limited the extension of the wrist during knuckle-walking. This evidence was also used to falsify the second hypothesis: that these carpal features should be much more developed in gorillas. Many of the features were not only lacking in adult gorillas, but were almost always more highly developed among the adult chimpanzees. In order to falsify the third hypothesis, that the features of the carpals should be more highly developed in juvenile gorillas than in juvenile chimpanzees, Kivell and Schmitt (2009) examined the presence and absence of these traits in juveniles of Gorilla and P. troglodytes. Again, in almost all cases, the carpal features under examination were found much more frequently among the juvenile chimpanzees than in the juvenile gorillas.
Building upon the work of previous researchers (Inouye and Shea, 2004; Inouye, 1994; Inouye, 1992; Inouye, 1989), Kivell and Schmitt (2009) proposed that Gorilla and Pan have two distinct postural positions, respectively labelled as “columnar” and “extended” (Kivell and Schmitt, 2009: 14244). In the Gorilla columnar posture, the carpal joints are in line with the hand and forearm. When in the support phase, the wrist is perpendicular to the substrate, and the weight of the gorilla is loaded upon it. Kivell and Schmitt (2009) further point out that this posture also gives gorillas increased wrist mobility, as they lack the bony adaptations seen in the carpals of Pan. It also gives them a more hyper-extended elbow joint (Inouye, 1989). This stands in contrast to the extended posture of Pan which is better equipped to deal with bending loads. According to Kivell and Schmitt (2009), the features associated with extended posture, such as the bony projections and the specific aspects of carpal morphology, have been previously identified by researchers as more generalized aspects of knuckle-walking. The authors contend that these are instead specific to Pan, and, thus, can be used to demonstrate the independent evolution of knuckle-walking in each lineage.
Although Pan and Gorilla may have two distinct forms of knuckle-walking, this does not necessitate independent evolution. There is an underlying assumption being made by Kivell and Schmitt (2009) that the differences in the carpal features are attributable to the unique origins of these traits in each species. It seems to imply that these features became frozen in time in each lineage since their initial evolution. It would be more parsimonious to suggest that knuckle-walking first arose in the shared common ancestor of Pan and Gorilla and then subsequently became different in each species. Gorilla and Pan are estimated to have diverged somewhere around 10 MA (Langergraber et al., 2012; Scally et al., 2012; Suwa et al., 2007; Kunimatsu et al., 2007). This allows enough time for each genus to evolve along its own evolutionary path, which consequently shaped the morphology of their carpals in different ways.
When one takes a broader look at the entire suite of morphological characteristics related to knuckle-walking in Pan and Gorilla, the similarities between these species outnumber a few morphological differences in the carpals. First and foremost is the similar manner in which the hand comes in contact with the ground during knuckle-walking in both Pan and Gorilla. The first bone to connect with the substrate is the fourth medial phalanx. Weight is forced down upon the hand and the radiocarpal and midcarpal joints of the wrist compress. The weight is then rolled onto the third and second medial phalanges (Begun, 2004). Sheer stress occurs between the metacarpals as the weight is shifted among these digits. The second metacarpal articulates with the capitate which in turn articulates with the scaphoid at the mid-carpal joint. The scaphoid then articulates with the upper arm where it meets the radius. The extra reinforcement and support provided by this configuration is unique among African apes and humans. This is because in African apes and humans the os centrale and the scaphoid are fused. This fusion happens early in ontogeny, sometime close to birth (Kivell and Begun, 2007).
In contrast, in Pongo and Hylobates the os centrale and scaphoid only fuse in rare instances, and when it does occur, it is always in adulthood (Kivell and Begun, 2007). The scaphoid does not articulate with the capitate in Asian apes, and, therefore, plays no role in the midcarpal joint (Begun, 2004). Despite what may appear to be rather clear homologous knuckle-walking traits in Pan and Gorilla, Kivell and Begun (2007) note that this is an area which needs to be explored further to fully understand the functional biomechanics. Although Kivell and Begun (2007) conducted an exhaustive study of os centrale fusion among extant hominoids (and some Malagasy strepsirrhines), their focus was primarily upon the timing of os centrale fusion. They were not able to either falsify or find evidence to support the sheer stress hypothesis. Still, as this trait is perhaps one of the most definitive homologous structures in the wrist of African apes, the sheer stress hypothesis cannot be easily discarded. As Begun (2004:20) suggests, “it appears as if sheer across the scaphoid-centrale joint due to loading of the second metacarpal-trapezoid joint in compression has selected for the fusion of these bones.” The fact that fusion occurs very early in African ape development supports the idea that it is genetically programmed and does not form as a result of use (Kivell and Begun, 2007).
Kivell and Schmitt (2009) do point out that the distal concavity of the scaphoid is noticeably absent or reduced in most adult gorillas, but it is only one part of larger mechanism used to create greater stability in the radiocarpal joint of African apes (Richmond and Strait, 2000; Richmond et al., 2001). One shared component of the radiocarpal joint in extant knuckle-walkers is the orientation, shape, and size of the scaphoid notch of the radius (Figure 1; Richmond et al., 2001; Richmond and Strait, 2000). The scaphoid notch is a feature common to all primates, but Pan and Gorilla are unique in that the radius has a prominent dorsal ridge which projects distally, causing the scaphoid notch to be more dorsally oriented (Richmond and Strait, 2000). The scaphoid notch and scaphoid-lunate angle are remarkably small in African apes compared to other anthropoid quadrupeds, but still larger than Pongo (Richmond and Strait, 2000). This allows the convex proximal articular surface of the scaphoid to rotate until the concave portion of the scaphoid locks into a close-packed position with the scaphoid notch during wrist extension. Essentially, the radiocarpal joint is stabilized and limits the range of wrist extension (Begun, 2004; Richmond et al., 2001; Richmond and Strait, 2000). Although this mechanism might not be exactly the same in Pan and Gorilla (Kivell and Schmitt, 2009), the mechanics of the scaphoid notch in extant Asian apes are remarkably different. Pongo and Hylobates are aided in clambering/brachiation by preserving a wrist that allows a greater range of wrist extension (Richmond and Strait, 2000). Additionally, as Richmond and Strait (2000) point out, the scaphoid notch reduces stress by distributing the weight of the ape over a larger area. If Kivell and Schmitt (2009) are correct, that Pan and Gorilla exhibit two unique and independently evolved postures, it would not make sense for both of these apes to have the same adaptation to weight distribution. It would, however, make sense for both taxa to have this feature of the radius if they shared an ancestral knuckle-walking condition, but then the Gorilla evolved a slightly different form of knuckle-walking (and hence weight distribution) secondarily.
Aside from limiting the range of extension to create greater wrist stability, there are other features shared by Pan and Gorilla which are not found in Pongo. One of these features is the jagged morphology of the carpo-metacarpal joint. The carpo-metacarpal joint are the five joints between the distal row of the carpals and the proximal ends of the metacarpals. In African apes and fossil humans, the bases of the metacarpals have keeled articular surfaces. These surfaces articulate with the facets of the carpals of the distal carpal row. The facets are much more divergent and pronounced in African apes, especially in the capitate, which articulates with the second, third, and fourth metacarpals (Begun, 2004; Richmond et al., 2001). The trapezoid, which shares the articulation of the second metacarpal with the capitate, has a strong keel (Richmond et al., 2001). This keel articulates with a deep notch in the second metacarpal. The primary function of the jagged morphology is to resist the twisting at the carpo-metacarpal joint during the stance phase, when body mass is transferred across this joint (Begun, 2004; McHenry, 1983). The jagged morphology of the African apes persists in Australopithecus sp. (McHenry, 1983; Ward et al., 1999; Kivell et al., 2011), but stands in stark contrast to smooth articulations found at the carpo-metacarpal joint in Pongo.
The capitate has a number of other unique features which can be used to demonstrate shared derived knuckle-walking traits among the African apes. In fact, Begun (2004) states that the capitate shows the most pronounced differences between African apes and Pongo. In Pan, Gorilla, and Homo the capitate is considerably larger compared to Pongo. The dorsal non-articulated portion of the capitate is expanded to prevent the lunate from riding up the dorsal side, as is observed in Asian apes and digitgrade cercopithecids (Begun, 2004). The capitates of Gorilla and Pan show a much more pronounced degree of mediolateral waisting compared to other anthropoids (Richmond et al., 2001). The waisting itself is not necessarily a condition that aids in knuckle-walking, but a by-product of the expansion of proximal and dorsal portions, somewhat akin to the spandrels of San Marco as famously described by Gould and Lewontin (1979). The proximal portion (capitate head) articulates with the scaphoid and the trapezoid laterally and is close-packed during extension in African apes and humans (Lewis, 1985; Begun, 2004). Again, the close-packing of the carpals is another feature which helps to limit the range of extension at the mid-carpal joint (Begun, 2004). Although mediolateral waisiting of the capitate is also present among fossil hominins, such as in some Australopithecus sp. (McHenry, 1983; Ward et al., 1999), it is reduced among Australopithecus sediba and members of Homo (Kivell et al., 2011). One possible explanation is that the expanded proximal and distal ends of the capitate were crucial components of the knuckle-walking suite which were kept in check by natural selection. Once free of the role of reinforcement during knuckle-walking, the degree of waisting in capitates became more variable in bipedal hominins and gradually reduced. Kivell and Schmitt (2009) show that the degree of capitate waisting is reduced in Gorilla compared to Pan; however, the morphology of the capitate in Pongo in relation to this feature is considerably different from the African apes.
Adding weight to the hypothesis that knuckle-walking in African apes originated from a shared ancestral condition is the fact that many of these features discussed above can also still be found in humans, but are not found in the other hominoids outside the African ape and human clade. There are also many more of these features, such as the fact that African apes and humans have a reduced brachial index compared to Pongo and Hylobates (Begun, 2004). African apes and humans also have shared characteristics in the large size and shape of the ulnar head which forms a meniscus between the ulna and the proximal carpal row (Begun, 2004). This feature, which is absent in Pongo, serves to prevent direct articulation between the ulna and the carpals (triquetrum and pisiform), but still allows compressive stress to be transmitted from the ulna to the aforementioned proximal carpal row (Begun, 2004; Sarmiento, 1988). This appears to be a modified form of ulnar deviation, a trait common among suspensory hominoids (Lewis, 1985), which helps deal with the additional stresses placed on the wrist due to weight bearing.
African apes and humans also demonstrate “greater of torsion in the [humeral] head relative to the transverse plane of the distal end” (Begun, 2004:15). Interestingly, this was first demonstrated by Larson (1988, 1996), who suggested these traits evolved independently among African apes and humans. Larson (1996) examined postcranial material from Australopithecus afarensis, Australopithecus africanus, and Homo habilis and concluded that these transitional species lack the degree of humeral torsion seen in extant African apes and humans. This diagnosis has been shown to be rather problematic as two of the specimens examined by Larson are highly distorted and crushed (Begun, 2004). Furthermore, Begun (2004) notes that the most intact specimen, OMO 119-73-2718, actually falls within the range of humeral head torsion expected in Pan and Homo. Taking these facts into consideration, the least assumptive conclusion is that degree of humeral head torsion seen in African apes, Pliocene hominins, and modern humans is homologous.
There are additional components of early bipedal hominin anatomy which further suggest that knuckle-walking among Pan and Gorilla is a homologous trait. Richmond and Strait (2000) employed a data clustering method known as UPGMA (Unweighted Pair Group Method with Arithmetic Mean) to look at the similarities and differences between fossil hominins and a variety of extant catarrhine taxa (apes and old world monkeys). More specifically, they examined a combination of features of the arm and wrist that can be attributed to knuckle-walking, including features of the radius: a prominent distally projecting dorsal ridge, and a dorsally oriented scaphoid notch, as well as the scaphoid-lunate angle. Their analysis showed that the earlier hominins, such as Australopithecus anamensis and Australopithecus afarensis, clustered more closely with the extant African apes, whereas the later hominins, Australopithecus africanus and Paranthropus robustus clustered more closely with Homo sapiens. Thus, it suggests these features were more prominent in early bipedal hominins, as they were more closely related to their knuckle-walking ancestor. Then, these features reduced in prominence over time as the hominin skeleton adapted to bipedalism. If early bipedal hominins evolved from a knuckle-walking posture, as this study indicates, it becomes even more parsimonious to suggest that this was a trait shared by the last common ancestor of the African apes. Otherwise, knuckle-walking would have had to evolve on at least three separate occasions.
Taking all these lines of evidence into consideration, the conclusion that the last common ancestor of African ape and human clade was a knuckle-walking ancestor is not only more parsimonious, but also demonstratively stronger than the opposing position. In the examples explored above, the morphological features of extant taxa and Pliocene hominins are used to inform our picture of the past. In order to better understand the origins of knuckle-walking, we need to explore the evidence provided by the fossil record that preceeds its emergence. As noted previously, the common ancestor of Gorilla, Pan, and Homo is suspected to have lived approximately 10 MA (Langergraber et al., 2012; Scally et al., 2012; Suwa et al., 2007; Kunimatsu et al., 2007).
Despite the plethora of bipedal hominin fossils that permeate the Pliocene fossil record, the fossil record of late Miocene hominoids in Africa is considerably sparse. Some fossils, such as Samburupithecus kiptalami from 9.5 MA in Kenya, Chororapithecus abyssinicus from 10.5-10 MA in Ethiopia and Nakalipithecus nakayamai 9.9-9.8 MA in Kenya, have been described as possible ancestors of, or sister taxa to, Gorilla (Kunimatsu et al., 2007; Suwa et al., 2007; Ishida and Pickford, 1997). Samburupithecus is known from a maxilla with post-canine teeth. Ishida and Pickford (1997) claim Samburupithecus shows affinities with the post-canine teeth of Gorilla. However, other researchers have pointed out there is not enough information provided by these teeth that enables us to know whether these belonged to a member of the great ape clade or a terminal proconsulid (Begun et al., 2012). Chororapithecus is described as a possible sister taxa to Gorilla on the basis that mirco-computed tomography scans showing dentine crests at the enamel junction similar to the sheering crests seen on the occlusal surface of Gorilla molars (Suwa et al., 2007). The problem with this connection, as Begun et al. (2012) have pointed out, is that these crests are not found on the enamel surface of the teeth. They are inaccessible to the process of natural selection and, therefore, any connection to the crested molars found in Gorilla is extremely tenuous. Finally, Nakaliopithecus suffers from the same fate as Samburupithecus in that the molars which have been used to describe this species are undecipherable from the teeth of late period proconsulids (Begun et al., 2012).
Comparatively, the fossil record of Late Miocene hominoids outside of Africa is much more plentiful. Even though Pan, Gorilla, and Homo are often considered to be African in origin, the fossil evidence suggests that their shared common ancestor was a member of the Dryopithecini tribe from Eurasia (Begun et al., 2012; Begun, 2010; Kordos and Begun, 2002; Kordos, 2002; Stewart and Disotell, 1998: Begun et al., 1997; Begun and Kordos, 1997; Begun, 1992). The evidence used to tie the Dryopithecines to the African ape and human clade is primarily found in the crania. These include a short premaxilla which has little overlap with the maxilla, an elevated root of the zygomatic, and in some later Dryopithecine taxa (Hispanopithecus and Rudapithecus), a large frontal sinus derived from the ethmoid (Begun et al., 2012; Begun, 2010).
More importantly to this analysis, the Dryopithecines have humeri which greatly resemble great apes in their general morphology (Begun, 1992; Morbek, 1983; Rose, 1983; Pilbeam and Simons, 1971). In an analysis of a humeral diaphysis from Saint Gaudens, France, attributed to Dryopithecus fontani, Pilbeam and Simons (1971) argued that the size and morphology of the humeri more closely resembled P. paniscus than any cercopithecoid. The delto-triceps were poorly developed and curvature of the shaft looked more similar to Pan than a cercopithecoid monkey. The biceptal groove faced more anteriorly than laterally, and this was used to suggest that the humeral head was posteriorly positioned. The humerus also exhibited a poorly developed epicondylar ridge, another feature which demonstrated the bone was more morphologically similar to other hominoids than it was cercopithecoids. Pilbeam and Simons (1971) used this and other features of the distal humerus to suggest D. fontani had a capitulum which resembles the strongly pronounced capitulums seen in later hominoids. This feature, if it were present, would have been a necessary for component below branch suspensory behaviour. This is not to say that D. fontani was necessarily a below branch suspensory hominoid. Pilbeam and Simons (1971) even speculated that D. fontani could be an early knuckle-walker. Although this latter claim is unlikely to be correct, the important aspect of Pilbeam and Simons (1971) study is that it shows there are a number of visible features which could demonstrate a link between Eurasian Miocene taxa and extant apes.
The first Dryopithecine which can probably be considered a below branch suspensory hominid is Pierolapithecus catalunicus from the middle Miocene of Spain. This is despite the fact that the researchers who initially described Pierolapithecus suggested it was a strong climber which used palmigrade locomotion to maneuver above the branches of trees (Moyà-Solà et al., 2004; Almécija et al., 2009). Moyà-Solà et al. (2004) were willing to claim that Pierolapithecus was close to the last common ancestor of all the extant great apes based upon the typical suite of Dryopithecine cranial features. They were, however, unwilling to say that it engaged in below branch suspensory behaviour, even though Pierolapithecus had a broad, shallow thorax which they contrasted with the narrow, deep thoraxes of earlier monkey-like hominoids like Proconsul heseloni and Equatorius africanus. The carpals showed even greater signs of below branch suspensory features. The triquetrum is elongated proximodistally and showed no facet for articulation with the ulnar styloid (Moyà-Solà et al., 2004). This is an extremely important feature, as it is the earliest known example of ulnar deviation in the hominoid fossil record. Ulnar deviation creates a more versatile wrist which allows adduction of the hand (Lewis, 1985). This is different from the stiff, locking wrists which are necessary components of above branch palmigrade primates (Lewis, 1985). Moyà-Solà et al. (2004) claimed Pierolapithecus was an above branch palmigrade primate primarily because of features of the phalanges. These included the proximodistal tilt of the proximal articular facets, the large and widely separated plantar tubercles, and the fact that the phalanges were generally shorter and less curved than those of extant suspensory apes (Moyà-Solà et al., 2004).
A subsequent analysis by Deane and Begun (2008) demonstrated the curvature of the phalanges exhibited by Pierolapithecus actually put it outside the range of orthograde climbers. Coming to the opposite conclusion of Moyà-Solà et al. (2004), they argued that Pierolapithecus had phalangeal curvature much more similar to that which is seen among suspensory primates. It showed slightly less curvature than Hylobates and came close to the curvature exhibited by Pan. Phalangeal curvature has been consistently shown to be an excellent indicator of locomotive capabilities (Ward et al., 2012; Rein, 2011; Richmond, 2007; Richmond, 2003; Rose, 1986; Stern and Susman, 1983;). Combined with the fact that Pierolapithecus is estimated to have had a mass of approximately 30 kg (Moyà-Solà et al.,2004), which is comparable to a female chimpanzee (Boesch and Boesch-Achermann, 2000), it is extremely likely that Pierolapithecus was a suspensory ape.
Also included in Deane and Begun’s (2008) study were phalanges from the Dryopithecines Hispanopithecus laietanus (formerly Dryopithecus laietanus) and Rudapithecus hungaricus (formerly Dryopithecus brancoi). Both of these taxa come from the late Miocene (10 MA), the former from Spain and the latter from Hungary. Like all the aforementioned Dryopithecines, Hispanopithecus and Rudapithecus can be tied to extant great apes craniodentally (Begun et al., 2012). Additionally, their post-cranial remains exhibit a number of transitional steps which preceded the evolution of knuckle-walking in African apes.
In Hispanopithecus these include lumbar vertebrae in which the transverse processes are more posteriorly positioned (Begun et al., 2012; Almécija et al., 2007). Hispanopithecus seemingly had a short, stiff, lower back that permitted orthograde behaviour (Begun et al., 2012; Almécija et al., 2007). Interestingly, Hispanopithecus also has some characteristics which more closely resemble Pongo than the African great apes. These include the curvature of the phalanges, as well as a Pongo-like femur. It is likely these are primitive retentions, as evidence shows that the lineage which led to the Ponginae clade (Ankarapithecus, Sivapithecus, Lufengpithecus, Khoratpithecus, and Pongo) diverged much earlier from the hominins, sometime between 16 and 12 MA (Begun, 2002; Kelley, 2002). The Pongo-like femur is quite possibly a homoplastic character. Sivapithecus, which is thought by many to be a sister taxa to Pongo (Begun, 2002; Kelley, 2002; Ward and Kimble, 1983), has a femur which does not have the same degree of curvature as modern orangutans (Pilbeam et al., 1990).
Rudapithecus has characteristics which bring it closer to the base of African ape and human clade than Hispanopithecus (Begun et al., 2012; Begun, 2010). Although many of these characteristics are found in the cranium, there are also more postcranial features which Rudapithecus shares with modern African apes. A detailed comparative analysis of the post-cranial elements of Rudapithecus performed by Morbeck (1983) describes a distal humeral fragment (RUD 53 – Figure 2), a proximal ulnar fragment (RUD 22), and a fragmentary radial head (RUD 66) recovered from the Rudabánya site. At the time of Morbeck’s analysis, the affinity of these fragments was unclear. They were suspected of belonging to either Rudapithecus hungaricus or Bodvapithecus altipalatus. However, subsequent work at Rudabánya has shown that these taxa are actually the same species: Rudapithecus (Begun, 1988; Kordos and Begun, 1997). Morbek (1983) came to the conclusion that the radius and ulna, although obviously from different sized individuals, had structural similarities to below branch suspensory primates. The humeral fragment also shared many of these hominid similarities, and Morbeck noted it was most similar in size to Pan and Homo.
Begun and Kivell (2009) described three carpal bones from Rudabánya, two of which were successfully attributed to Rudapithecus, a scaphoid (RUD 202 – Figure 2) and a capitate (RUD 167 – Figure 2). Morphologically and functionally, these carpals seem to more closely resemble Pongo than the extant African apes. The scaphoid had tubercle that projected distally as in Pongo, and the os centrale was unfused. The capitate is not waisted as it is in Pan or Gorilla and it lacked many of the stabilizing adaptations for limiting wrist extension as described in the previous section. If Rudapithecus, or some similar Dryopithecine, is ancestral to the hominin clade, it appears as though the truly unique adaptations of the carpals found in extant knuckle-walkers must have evolved at a later period in time.
In a more recent study, Begun and Kivell (2011) examined two carpals, a capitate and hamate, from Sivapithecus. Although previous researchers had seen some similarities to Pan (Rose, 1984) and Gorilla (Spoor et al., 1991), Kivell and Begun (2011) suggested that some of the features of these carpals were best explained by a form of terrestrial knuckle-walking. They described the overall morphology of the capitate as “that of a primitive hominid” (Begun and Kivell, 2012: 160), but also as having features that were unlike Asian apes. One such feature was the keeled capitate facet which articulates with the third metacarpal. This feature is also found in African apes and evolved as an adaptation to deal with compressive loading and limiting wrist extension. The triquetral facate of the hamate is short proximodistally, closely resembling that of Gorilla. Again, this is a feature which is commonly seen as an adaptation to the stresses that arise from terrestrial. Sivapithecus is generally considered a member of the Ponginae subfamily (Begun, 2010; Begun, 2002; Kelley, 2002; Ward, 1997; Ward and Kimbel, 1983), so the presence of these features was unexpected. As Begun and Kivell disagree about the evolution of knuckle-walking in African apes, three possible hypotheses were put forth to explain the traits seen in the carpals of Sivapithecus. The first is that knuckle-walking evolved even earlier than previously believed in the common ancestor of the Ponginae and the hominins. This is the most unlikely of the scenarios given that Pierolapithecus, Hispanopithecus, or Rudapithecus do not exhibit any distinct knuckle-walking attributes. The second hypothesis is that knuckle-walking is extremely homoplastic and evolved on at least three separate occasions. The third hypothesis is that knuckle-walking must have evolved a least twice, once in each of distinct hominoid subfamilies (Ponginae and Homininae). Given that the number of homologous knuckle-walking features shared by Pan and Gorilla outweigh the differences, the third scenario is the most plausible.
Thus, some degree of homoplasy has to be accepted in relation to the evolution of knuckle-walking. However, allowing for homoplasy in this situation does not by any means strengthen (or weaken) the case for the independent evolution of knuckle-walking in African apes. Instead, Sivapithecus presents the precise type of scenario in which homoplastic evolution becomes the most parsimonious explanation, as the alternative hypotheses presented by Begun and Kivell (2011) would require even more assumptions. In this manner, the evolution of knuckle-walking Sivapithecus is perhaps better thought of as an example of convergent evolution rather than parallel evolution.
This knuckle-walking convergence in primates as distantly related as Sivapithecus and the common ancestor of the African apes suggests a strong environmental factor. Despite sharing a number of facial similarities with Pongo (Ward and Kimbel, 1983), the post-cranial morphology of Sivapithecus is noticeably different, showing no specialized adaptations to arboreal clambering (Pilbeam et al., 1990). Begun and Kivell (2011) propose that Sivapithecus went through a terrestrial knuckle-walking phase before evolving into a species that is more Pongo-like. This explanation not only ties together the work of previous researchers that appeared to be oppositional (Ward and Kimbel, 1983; Pilbeam et al., 1990), but it also helps explain biogeographical differences between Sivapithecus and Pongo. Sivapithecus was native to Pakistan and India, whereas the earliest orangutans are native to China (Bacon and Long, 2001), and modern orangutans are found only on the islands of Borneo and Sumatra (Wich, 2009). Allowing Sivapithecus the freedom of terrestrial knuckle-walking makes it possible for this species to disperse across a wider range of areas, and as Kivell and Begun (2011) suggest, this would include areas in which there was likely to be no continuous closed forests.
If the descendants of Rudapithecus, or a Dryopithecine similar to Rudapithecus dispersed back into Africa, as has been suggested (Begun et al., 2012), it is possible that subsequent adaptations to knuckle-walking could have aided in this process in overcoming geographical barriers in a way similar to Sivapithecus. Nearing the end of hominoid radiation in Eurasia, there was shift from tropical and subtropical forests to more deciduous forests (Begun et al., 2012). It is likely these changes might have created narrow ecological corridors which would have made hominin dispersal more problematic. Any morphological adaptation that could have helped a species confront and conquer these challenges is likely to have been favourably promoted by the process of natural selection.
The importance of ecology and ecological changes in the evolution of knuckle-walking is a subject which needs to be further explored. As Kivell and Schmitt (2009) clearly demonstrated, there are substantial differences in Pan and Gorilla wrist morphology and posture. Although this does not necessitate independent evolution of knuckle-walking, it does seem to indicate that clumping knuckle-walking into one over-arching category is perhaps too simplistic. The differences in the morphology of the carpals of Pan and Gorilla are better explained as adaptations to different ecological settings than as independent parallel evolution. This interpretation would be more parsimonious. It also implies that researchers need to pay special attention to which traits are specific to each genus and which traits are shared amongst genera. The shared traits will provide additional information about the ancestral form of knuckle-walking, and the unique traits can inform us about the changes that have taken place since divergence.
In this regard, it is interesting to note that P. paniscus is much more arboreal, yet it engages in less arboreal knuckle-walking than P. troglodytes. Doran (1993) reports that 85% of P. paniscus arboreal quadrupedalism is palmigrade; by contrast, only 30-40% of P. troglodytes arboreal quadrupedalsim is palmigrade. This difference is not reflected in the carpal wrist morphology of P. paniscus and P. troglodytes (Kivell and Schmitt, 2009). It is, however, reflected in the fact that P. paniscus has a longer, narrower scapula compared to P. troglodytes, a trait which Doran (1993) argues is an adaptation to increased suspensory behaviour.
Furthermore, although Kivell and Schmitt (2009) separate P. troglodytes and P. paniscus in their analysis, they treat Gorilla as a single entity. Kivell and Schmitt (2009) pool carpal measurements from both G. gorilla and G. beringei. G. gorilla and G. beringei are actually each composed of two subspecies, making four subspecies in total: the western lowland gorilla (G. gorilla gorilla), the cross river gorilla (G. gorilla diehli), the mountain gorilla (G. beringei beringei), and the eastern lowland gorilla (G. beringei graueri). Recent studies have shown that the Gorilla gorilla and Gorilla beringei diverged from each other approximately 1.75 MA, yet gene flow continued between these two groups (Scally et al., 2012; Langergraber et al., 2012; Ackermann and Bishop, 2010). Despite their relatively recent separation many researchers consider the degree of differences between these subspecies to be more equivalent to the differences commonly seen between species (Fleagle, 1999; Ruvolo et al., 1994). In terms of locomotion G. beringei beringei engage in considerably less arboreal activities and more terrestrial knuckle-walking than either G. gorilla gorilla or G. beringei graueri (Remis, 1998). These differences cannot be overlooked and need to be incorporated in further studies to in order to determine how they are represented morphologically.
Knuckle-walking did not evolve independently in Pan and Gorilla. The evidence in support of this statement clearly outweighs the evidence against it. This hypothesis is the most parsimonious explanation, but it cannot be taken for granted. Parsimony is only a null hypothesis, whose primary function is to be continually challenged. There are substantial differences in the forms of knuckle-walking exhibited by Pan and Gorilla, but these are likely to be modifications upon an ancestral knuckle-walking morphology. As each species diverged upon its unique evolutionary trajectory, some of the morphological characteristics changed as they adapted to different ecological niches. This would also suggest there are even more morphological differences within the various species and subspecies of extant knuckle-walkers yet to be discovered. Detecting these differences is likely to help further develop our understanding of the evolution of knuckle-walking, as well as the behaviour of our Miocene relatives.
References can be found in the comment section