The attribution of fossils to extinct hominin taxa has relied almost exclusively on diagnostic morphological characteristics of the skull and dentition. This general, albeit not wholly universal, truism holds for fossils that range from the earliest recognizable members of the Hominini in the Late Miocene, such as Sahelanthropus tchadensis (Brunet et al., 2002), to later representatives, such as Homo rhodesiensis (Woodward, 1921) and Paranthropus boisei (Leakey, 1959). Indeed, there are some species (e.g., Homo rudolfensis) for which the postcranial skeleton is currently unknown.

This is not to say that postcranial bones have not featured in the diagnoses and definitions of extinct hominin taxa, as exemplified by Homo habilis (Leakey et al., 1964), Ardipithecus ramidus (White et al., 1994, White et al., 1995), Australopithecus sediba (Berger et al., 2010), and Homo naledi (Berger et al., 2015). Thus, the holotype of H. habilis (OH 7) consists of a dentate mandible, an isolated upper molar, two parietal bones, and a set of hand bones “of a single juvenile individual” (Leakey et al., 1964: 8). Aspects of manual morphology, such as the degree of dorsal curvature of the phalangeal shafts, the orientation of the trapezium, and the form of the scaphoid featured in the diagnosis of the species (Leakey et al., 1964: 8). However, the association of the manual and craniodental elements has been questioned by some workers (e.g., Moyà-Solà et al., 2008). Postcranial bones form part of the type series of A. ramidus, where the holotype (ARA-VP-7/2) is a set of associated permanent teeth, and two of the 16 paratypes comprise postcrania (i.e., the ARA-VP-1/4 humerus and the ARA-VP-7/2 humerus, radius, and ulna). Even though these elements cannot be directly associated with any of the diagnostic craniodental elements, they are very reasonably attributed to A. ramidus insofar as this is the only species recognized from the Aramis localities (White et al., 2009). Australopithecus sediba is an example where postcranial remains comprise part of the holotype (MH1) in undoubted association with a partial skull, and where they also comprise part of the paratype (MH2) in undoubted association with teeth and a partial mandible (Berger et al., 2010). In this instance, although the characters used in the diagnosis of the species were wholly craniodental, postcranial elements were observed to present “a unique combination of primitive and derived traits” relative to some species of Australopithecus and Homo (Berger et al., 2010: 196). Postcranial remains comprise part of the type series of H. naledi, where the holotype (DH1) is a partial, dentate skull, and four of the nine paratypes are postcranial elements (i.e., Dinaledi hand 1 [H1], Dinaledi foot 1 [F1], the U.W. 101-1391 proximal femur and the U.W. 101-484 tibial diaphysis). Although none of these postcranial bones can be definitely associated with any of the craniodental fossils, they (and another 700 or so elements) are very reasonably attributed to the species because only one taxon is represented by craniodental remains in the Dinaledi Chamber assemblage (Berger et al., 2015). In this instance, the postcranial paratypes also feature in the diagnosis of the species (Berger et al., 2015).

Because the attribution of postcranial skeletal elements to an extinct species obviously provides significantly enhanced appreciation of its paleobiology, it is important to recognize the level of confidence that can be ascribed to such attributions. There are clearly different levels of reliance that pertain to such attributions. High confidence in the taxonomic attribution of postcranial bones tends to be restricted to instances in which they derive from a single skeleton that also preserves diagnostic craniodental elements. Day (1976b: 507) recognized that in any attempt to assess postcranial fossils, “the greatest possible assistance is obtained from clear and unequivocal knowledge of anatomical association—that is, the certainty that a given group of bones belonged to one individual, one limb, or one functional complex such as a hand or a foot.” What constitutes the clear and unequivocal association of postcranial and craniodental remains to a single individual is, of course, an issue that must be addressed. Similarly, because the unequivocal association of postcranial elements with one another, whether they represent a single limb, a single complex, such as a hand or foot, or related bones of a single individual will provide enhanced information over that which can be gleaned from a solitary bone, the reason(s) behind such an association should be elucidated. Day (1976b) provided an Appendix to his article on the postcranial remains from the Koobi Fora Formation in which he set forth three norms each, of anatomical and geological evidence that should be met to satisfactorily conclude that fossils are associated with one another. Based on these criteria, he provided a set of guidelines that should be applied to reinforce claims with respect to various elements representing parts of a single individual.

Association is most commonly made based on spatial proximity among elements at the time of discovery. In rare, but fortunate circumstances, cranial and postcranial elements may be discovered in articulation (e.g., DIK-1-1 and StW 573). Skeletal and craniodental remains are usually expected to be recovered within a few meters of one another if they are held to have derived from a single individual. However, this criterion will vary depending on the circumstances surrounding the taphonomic and diagenetic histories of the remains. Certainly, elements in purported association should derive from the same stratigraphic context or horizon. Moreover, bones that are held to represent a single individual should not be incompatible in size or morphology, and there should be no duplication of elements (e.g., an individual cannot have two right humeri). In some instances, such as the case with the FLK NN site in Bed I of Olduvai Gorge, bones that were recovered within meters of one another have been prudently assigned separate specimen numbers if there is no direct evidence to support their association (e.g., Leakey et al., 1964; see also Leakey, 1973a; Coppens and Sakka, 1983; Hammond et al., 2021).

Moderate to sometimes high confidence in the attribution of postcrania to a particular species may also pertain in instances in which the bones derive from temporal horizons and/or geographic locales from which only a single taxon is known. Such attributions, however, may be transient as future paleontological discoveries might reveal the presence of a previously unknown species in the temporal and/or geographic framework from which only a single taxon had been formerly identified. Thus, for example, before 2015, all hominin fossils from the Pliocene sediments of the Afar Triangle, Ethiopia, were confidently assigned to Australopithecus afarensis (e.g., Kimbel and Delezene, 2009). However, the recognition of Australopithecus deyiremeda by Haile-Selassie et al. (2015) in 3.5- to 3.3-Ma-old deposits in the Woranso-Mille area has meant that more caution must be exercised in the specific ascription of postcranial specimens from the Middle Pliocene of the Afar.

When postcranial elements can be securely attributed to a taxon they may add significant information relating to functional aspects of the paleobiology of a species, such as its locomotor habits and manipulative capabilities, and the presence of primitive vs. derived features that may be useful for phylogenetic postulates (e.g., McHenry and Brown, 2008; von Cramon-Taubadel and Lycett, 2014). However, to date, hominin postcranial skeletal morphologies have been interpreted almost solely in relation to body size, locomotor, manipulative, and obstetric considerations; they have yet to be fully explored in phylogenetic reconstructions. Indeed, almost all of the major analyses of hominin phylogeny that have been conducted to date rely solely on craniodental characters (e.g., Strait et al., 1997; Collard and Wood, 2000; Kimbel et al., 2004; Strait and Grine, 2004; Dembo et al., 2015; Mongle et al., 2019; Parins-Fukuchi et al., 2019), whereas those that incorporate postcranial features are comparatively rare (e.g., Argue et al., 2017).

It has been suggested that postcranial characters have been eschewed from phylogenetic analyses owing to the notion that, because they are susceptible to functional convergence, they are more prone to homoplasy than craniodental traits (e.g., Pilbeam, 1996; Lockwood, 1999; Lockwood and Fleagle, 1999). However, numerous studies across primates and other mammals have shown that this is not necessarily the case (e.g., Sánchez-Villagra and Williams, 1998; Fleagle and McGraw, 1999, 2002; Rae, 1999; Young, 2003, 2005; Williams, 2007; von Cramon-Taubadel and Lycett, 2014; McCarthy et al., 2017). We suspect, rather, that the failure to incorporate postcranial characters in hominin phylogenetics is owing primarily to the nature of the fossil record itself. The addition of morphological characters to phylogenetic analyses of fossil taxa, even when they are associated with missing data for some species, may serve to improve the accuracy of postulated evolutionary relationships, while excluding them to produce a matrix that contains fewer missing data cells may result in reduced accuracy (Wiens and Morrill, 2011; Koch et al., 2021).

In light of the potential utility of postcranial characters for phylogenetic analysis, we review African hominin postcranial remains that have been assigned to various Late Miocene to Middle Pleistocene taxa, with a view to assessing the level of confidence that can be assigned to such attributions. In so doing, we revisit and scrutinize the ways in which these associations and attributions have been made in particular historical contexts over the past century.

We recognize three general categories into which postcranial fossils can be grouped in relation to their associations (or lack thereof) and the level of confidence that can be assigned to their taxonomic attribution (Table 1).

The highest level of confidence in the specific attribution of postcranial bones pertains to those instances in which they are clearly associated with taxonomically diagnostic craniodental remains. This first general category (A–very high confidence taxonomic attribution) can be assigned to only 17 specimens that span over four million years from the Early Pliocene (i.e., A. ramidus at ca. 4.51 Ma) to the Middle Pleistocene (e.g., H. naledi from the Lesedi Chamber at ca. 335–236 ka and Omo I from the Kibish Formation at ca. >233 ka). These 17 variably incomplete skeletons have been attributed to 10 or possibly 11 species.

A moderate to high level of confidence in the specific attribution of postcranial elements pertains to those instances in which they are found in a taxon-specific temporal and/or geographic context. This second general category (B–moderate to high confidence taxonomic attribution) can be assigned to several variably associated and isolated bones.

Within this second general category, in some instances, the bones are possibly (but not assuredly) associated with diagnostic craniodental remains (category B-1). In other instances, several bones may be clearly associated with one another, providing more information on the postcranial skeletal morphology and interelement relationships of an individual that is likely a representative of a given taxon (Category B-2). A sizable number of isolated postcranial elements fall into this category (Category B-3). These include, for example, a good number of isolated bones from Makapansgat and Sterkfontein Member 4 that are reasonably referable to Australopithecus africanus (see Grine, 2019 and Ward and Zipfel, 2020 for reviews), although, as discussed below, some workers envision the presence of postcrania of two species of australopith from these deposits (e.g., Macho et al., 2020; Fornai et al., 2021). They also include a good number of elements from the Hadar Formation, with the caveat that some, such as the BRT-VP-2/73 foot bones from Burtele (Haile-Selassie et al., 2012) may not represent Au. afarensis.

A much lower level of confidence in the taxonomic attribution of postcranial elements is realized in those instances in which they derive from temporal spans and geographic regions in which more than one hominin taxon has been identified based on craniodental remains. Specimens that constitute this third general category (C–low confidence taxonomic attributions) are, unfortunately, much more common than would seem to be generally recognized or acknowledged.

Within this third category, in some instances, the bones are possibly (but not assuredly) associated with craniodental elements (Category C-1). These craniodental elements may be diagnostic. One such example, discussed at length below, pertains to the holotype (OH 7) of H. habilis. In other cases, the diagnostic value of the craniodental remains is more questionable. Within this general category, there are instances in which several bones may be clearly associated with one another, providing more information on the postcranial skeletal morphology and interelement relationships of an individual, but whose taxonomic identity is in doubt (Category C-2). The lack of taxonomic specificity limits the amount of information that they can provide. An example of a fossil in this subcategory is the OH 8 foot. Finally, within this general category, there are hundreds of isolated postcranial bones from sites in eastern and southern Africa (Category C-3). The taxonomic ascriptions of bones in Category C have commonly relied upon assumptions about postcranial morphotypes and/or the ‘statistical probability’ of species attribution based on the comparative abundance of taxonomically identifiable craniodental fossils at a site.