Observational and experimental research from extant nonhuman primates and modern humans can be applied to the study of early hominin variation and adaptation in body form. Limb proportions—including limb length, joint size, and limb robusticity—can provide information about arboreal behavioral activity (Stern and Susman, 1983; Susman et al., 1984), bipedal efficiency (Steudel-Numbers and Tilkens, 2004; Steudel-Numbers et al., 2007; Hora et al., 2014), thermoregulation (Wheeler, 1984; Ruff, 1991, 1993, 1994; Tilkens et al., 2007; Holliday and Hilton, 2010; Wall-Scheffler, 2014), and phylogenetic relationships (Holliday, 1997, 1999, 2012; McHenry and Berger, 1998a,b; Richmond et al., 2002; Green et al., 2007; Prabhat et al., 2021).

Selection pressures acting on limb proportions played an important role in the mosaic transition from Australopithecus to Homo. Most hominin taxa fall between extant apes and modern humans (Homo sapiens) with regard to limb length and limb size proportions (Aiello and Dean, 1990; Richmond et al., 2002; Haeusler and McHenry, 2004; Holliday, 2012; Heaton et al., 2019; Prabhat et al., 2021). Australopithecus and early Homo taxa possessed more apelike, forelimb-dominant proportions relative to the more humanlike, hindlimb-dominant proportions of later Homo. The taxa with more humanlike proportions are generally thought to have a reduced arboreal repertoire and an increased reliance on bipedal locomotion (McHenry and Berger, 1998a, b; Haeusler and McHenry, 2004; Green et al., 2007; Prabhat et al., 2021). However, most extinct hominins are thought to have incorporated both arboreal and terrestrial components into their positional behaviors (Susman et al., 1984; McHenry and Berger, 1998a,b; Haeusler and McHenry, 2004; Green et al., 2007; Heaton et al., 2019; Gordon et al., 2020).

The earliest members of the genus Homo, such as Homo habilis and Homo floresiensis, demonstrate limb proportions that are dissimilar to modern humans, suggesting a more mosaic transition in ‘bauplan’ from Australopithecus to Homo (Haeusler and McHenry, 2004; Holliday, 2012). Wood and Collard (1999) argued for reassigning H. habilis to the genus Australopithecus, citing the australopith-like limb proportions, among other skeletal evidence. Regrettably, the postcranial material attributed to H. habilis is highly fragmentary and less certain in attribution, so the limb estimates remain uncertain (Richmond et al., 2002). Because of its retention of many primitive apelike and australopith-like postcranial and cranial features, even though H. floresiensis is geologically much younger, this species is thought to have separated early from the common ancestors of other members of the genus Homo, with its lineage originating near the branching point at which the H. habilis lineage began (Morwood and Jungers, 2009; Dembo et al., 2015).

Significant limb proportion variation occurs among individuals attributed to Homo erectus individuals as well. For example, the earlier H. erectus Dmanisi individual, D4507, exhibits humerofemoral indices that are significantly different from those of a large sample of Pleistocene and Holocene H. sapiens. In contrast, these indices for KNM-WT 15000 are similar to those of H. sapiens (Holliday, 2012). To further blur the distinction between early Homo and Australopithecus, Australopithecus afarensis and Australopithecus sediba more closely approximate a humanlike pattern of upper to lower limb joint size proportions (McHenry and Berger, 1998a; Green et al., 2007; Haile-Selassie et al., 2010; Holliday et al., 2018) though both species demonstrate the typical Australopithecus interlimb length proportion that falls between those of Pan troglodytes and H. sapiens (Holliday et al., 2018; Heaton et al., 2019). This overlap among different representatives of Homo and Australopithecus suggests that limb proportions did not set all branches of the genus Homo apart from other hominins.

An evaluation of the limb proportions of Homo naledi is critical for understanding the transition from Australopithecus to Homo. Many aspects of H. naledi morphology suggest that it is phylogenetically rooted deep in the genus Homo (Dembo et al., 2016; Hawks and Berger, 2016; Berger et al., 2017; Argue et al., 2017). Although present at ∼300 ka (Dirks et al., 2017), H. naledi demonstrates many traits similar to A. afarensis or Australopithecus africanus (e.g., small cranial capacity, scapular and pelvic morphology, phalangeal curvature). Accordingly, a secure estimate of relative upper to lower limb size in H. naledi would shed more light on variation in the genus Homo and help us evaluate the range of locomotor strategies present in the common ancestor of H. erectus and other anatomically primitive forms of Homo.

The overall Dinaledi Chamber sample of H. naledi includes elements from at least 15 individuals. To date, few of these have been matched to the same skeleton (Berger et al., 2015), although this work continues (Bolter et al., 2020). The Lesedi Chamber sample consists of at least three individuals, including a partial adult skeleton (LES 1; Hawks et al., 2017). Garvin and colleagues (Garvin et al., 2017) estimated the maximum length of the most complete adult H. naledi humerus and tibia from the Dinaledi Chamber to derive a humerotibial length index. Given the possibility that these bones belong to different individuals, they compared this value to resampled distributions of humerotibial length indices from P. troglodytes, Gorilla spp., and single population of smaller-bodied Kulubnarti individuals from medieval Nubia. The H. naledi value fell within the 95% confidence interval of the Kulubnarti human distribution and well below the Pan and Gorilla distributions, suggesting that H. naledi had a humanlike relative limb length (Garvin et al., 2017). Lower limb length may be very informative about the adaptation of H. naledi to striding bipedalism (Garvin et al., 2017; Marchi et al., 2017; Walker et al., 2019), with additional evidence provided by its derived foot morphology (Harcourt-Smith et al., 2015). However, the relative joint sizes and shaft robusticity of H. naledi are also informative with regard to this adaptation. In addition, although some features of the hands and shoulders provide evidence that H. naledi was adapted to climbing (Kivell et al., 2015; Feuerriegel et al., 2017, 2019), establishing the relative limb size of H. naledi with a large sample of H. naledi elements will provide useful context for evaluating hypotheses about the locomotor habits of this group in light of other morphological analyses. Prabhat et al. (2021) estimated the relative limb size of the LES 1 partial skeleton using four upper and lower limb elements. They determined the relative limb size of LES 1 to be humanlike as the relative limb size value of LES 1 fell within the modern human distribution and well outside of the hylobatids, Gorilla spp., Pan spp., and Pongo spp. distributions (Prabhat et al., 2021). Here we use a wider range of unassociated elements from the Dinaledi and Lesedi Chambers and resampling techniques to estimate the relative limb size of H. naledi.

Relative limb size is a size-based limb proportion measure and is calculated using joint size and long bone robusticity. It is different from length-based proportions such as the intermembral index. Joint surfaces and long bone diaphyses are dynamic structures that respond to joint forces, muscle activity, and other mechanical demands associated with postural and locomotor behaviors. Joint surfaces and diaphyses are more robust, or larger in size, when the magnitude of mechanical loads (forces) on the bone is greater (Jungers, 1988; Aiello and Dean, 1990; Ruff and Runestad, 1992; Godfrey et al., 1995). Any change in the size of the joint, measured by the size of the articular surface or the diaphysis, will influence the relative limb size index of a species. Modern human limb proportions are driven by larger lower limb joints, which are a functional adaptation to habitual terrestrial bipedalism. This results in a derived, smaller upper to lower limb joint ratio in humans relative to great apes (Jungers, 1988). Green et al. (2007) showed a qualitative association between the higher relative limb size values of forelimb-dominant groups with higher levels of arboreality in extant apes. In addition, Gordon et al. (2020) applied dimensions of articular surfaces and diaphyses to develop a limb size index similar to relative limb size and demonstrated that this index corresponds with the extent of arboreality in extant great apes and humans.

Following these studies, we consider both articular surface and diaphyseal dimensions and broad categories of behavior such as arboreality compared to terrestriality. We apply a Monte Carlo resampling approach to estimate relative limb size in H. naledi, similar to the approach used by Green et al. (2007) in their consideration of A. afarensis and A. africanus limb elements. The findings of Green et al. (2007) supported the findings of McHenry and Berger (1998a), who suggested that A. africanus had relatively larger upper limbs that were more similar to those of apes, whereas A. afarensis had upper to lower limb size proportions more like modern humans. The higher, more apelike relative limb size in A. africanus suggests a morphological response to greater mechanical loading of the upper limbs (Green et al., 2007). Behaviorally, the contrast in relative limb size between these two species suggests that in addition to bipedalism, A. africanus performed activities that required this greater loading more than A. afarensis, including arboreal behaviors (Green et al., 2007).

Most phylogenetic analyses align A. africanus more closely with modern humans relative to A. afarensis based on shared, derived craniodental elements (White et al., 1983; Strait et al., 1997; Dembo et al., 2015). The greater similarity in relative limb size between A. afarensis and modern humans implies homoplasy in the evolution of limb proportions in these lineages (McHenry and Berger, 1998a; Prabhat et al., 2021). We adapt the Green et al. (2007) resampling method to generate a distribution of relative limb size values for H. naledi and compare this to a modern human distribution to better understand the locomotor repertoire and phylogenetic history of H. naledi. Based on previous research on the humanlike relative length proportion of H. naledi (Garvin et al., 2017) and the humanlike relative limb size proportion of LES 1 (Prahbat et al., 2021), we predict that the relative limb size of this larger sample of H. naledi is also humanlike. If this hypothesis is supported, it would suggest that H. naledi engaged in fewer positional or locomotor behaviors that required significant loading of the upper limb. Alternatively, if relative limb size is found to differ from modern humans, then this would suggest that H. naledi engaged in more positional or locomotor behaviors requiring significant loading of the upper limb than is seen in modern humans.