Since the discovery of numerous Australopithecus afarensis postcranial remains at Hadar, Ethiopia, in the early 1970s (Taieb et al., 1974; Johanson and Taieb, 1976; Bush et al., 1982; Johanson et al., 1982), which built upon a collection of australopith fossils already uncovered in South Africa (e.g., Dart, 1925; Broom and Schepers, 1946; Straus, 1948; Broom and Robinson, 1949; Robinson, 1972), paleoanthropologists have increasingly focused on the inferences of posture and locomotion that can be drawn from fossil hominin (i.e., humans and their fossil relatives) skeletons. This focus on functional morphology led to contentious debates but also the development of new hypotheses, analytical approaches, and methods. Functional inferences have typically centered on understanding the form of bipedalism in which australopiths engaged and the relative importance of arboreality in their positional repertoire (e.g., Lovejoy, 1975; Stern and Susman, 1983; Susman et al., 1984; Senut and Tardieu, 1985; Latimer, 1991). Reconstructions of australopith locomotion and posture were made within the broader context of two foundational and longstanding evolutionary aims of paleoanthropology: (1) the identification of the primary locomotor behavior from which bipedalism evolved; and (2) understanding the evolutionary or selective process(es) leading to the establishment of hominin bipedal dependence. Multiple hypotheses about the primary locomotor behavior from which bipedalism evolved (e.g., Keith, 1923; Tuttle, 1969; Richmond and Strait, 2000) and the selective drivers of bipedalism have been proposed (e.g., Lovejoy, 1981; Darwin, 1987; Wheeler, 1991; Hunt, 1994), which remain contentious due to a paucity of Late Miocene and Early Pliocene hominin fossils and the virtual absence of an African ape fossil record (Crompton et al., 2008; Harcourt-Smith, 2010; Wood and Harrison, 2011; Andrews, 2020; Almécija et al., 2021; but see McBrearty and Jablonski, 2005; Pickford et al., 2008). In the absence of fossil evidence near the time of the hominin-panin divergence, these hypotheses have traditionally relied heavily on functional interpretation of australopith anatomy, observations of extant primate postural and locomotor behaviors, and the influence of phylogeny. One such hypothesis is the ‘postural feeding hypothesis’ proposed by Hunt (1994) in “The evolution of human bipedality: Ecology and functional morphology” (see also Hunt, 1996). In this review, we use Hunt's (1994) seminal paper on the origins of hominin bipedalism as inspiration for how the australopith postcranium has influenced our hypotheses about the selective pressures surrounding hominin bipedalism and the value of living primate models and ecological context in our interpretation of hominin functional morphology. We start with a brief review of the historical context in which Hunt's (1994) ‘postural feeding hypothesis’ was developed. We then further review some of the new australopith fossils discovered and new methods developed since 1994 that have helped paleoanthropologists to gain a better understanding of australopith diversity and make more informed functional inferences of the fossil morphology.

Hypotheses about the locomotor or postural behavior from which bipedalism emerged range from terrestrial knuckle-walking to arboreal vertical clinging or brachiation and from pronograde to orthograde postures (see reviews in Richmond et al., 2001; Harcourt-Smith and Aiello, 2004; Crompton et al., 2008; Senut et al., 2018). Keith (1903, 1923) was the first to explicitly propose hominoid-like, orthograde ancestor based on morphological features of the upper limb that are shared between humans and suspensory apes. Keith (1923) proposed three stages of hominoid evolution: a hylobatid-like (‘hylobatian’) ancestor that evolved into a larger-bodied African ape-like (‘troglodytian’) ancestor capable of orthograde climbing and terrestrial knuckle-walking, which in turn evolved into a bipedal (‘plantigrade’) hominin (see also Morton, 1926). Tuttle (1969, 1975, 1981; Tuttle et al., 1974) further developed Keith's ‘brachiating’ hypothesis through the study of great ape hand postures and anatomy, highlighting the lack of anatomical evidence for a knuckle-walking phase in hominin evolution. Tuttle (1969:p. 960) proposed “that the ancestors of [hu]man probably engaged in some form of suspensory posturing and that they assumed bipedal postures very soon after venturing to the ground.” It is important to note that neither Keith nor Tuttle at this time had a conception of hominoid phylogeny (although the seminal research on hominid proteins by Goodman, 1962, 1963, was published). Tuttle (1969) supported a closer evolutionary relationship between Pan and Gorilla (and, in fact, grouped gorillas within in the Pan genus) to the exclusion of humans. Thus, knuckle walking could be considered a synapomorphic behavior of African apes in his hypothesis that human bipedalism evolved from a suspensory ancestor.

With greater evidence and acceptance of the phylogenetic relationship between Pan and humans (Goodman, 1963; Sarich and Wilson, 1967; Miyamoto et al., 1987), Washburn (1967:p. 23) proposed, albeit with little morphological evidence, that hominin bipedalism evolved from a terrestrial knuckle-walking ancestor because it offers a logical ‘intermediate condition’ in which, as bipedalism is positively selected, the long upper limb can be used less and less. The terrestrial knuckle-walking hypothesis continued to gain support through comparative anatomy, notably synapomorphic features of the upper limb, wrist, and hand shared between humans and African apes (Marzke, 1971; Corruccini, 1978; Gebo, 1992, 1996; Shea and Inouye, 1993; Begun, 1993, 1994; Richmond and Strait, 2000; Richmond et al., 2001) and the identification of ‘knuckle-walking features’ in australopiths (McHenry, 1983; Richmond and Strait, 2000). Parsimony also played a key supporting role with the sentiment that knuckle-walking locomotion is too unusual to have evolved independently in Pan and Gorilla (Gebo, 1992, 1996; Begun, 1994; Richmond and Strait, 2000; Richmond et al., 2001). Although other hypotheses on the locomotor or postural origin of hominin bipedalism emphasize arboreality, most notably a vertical climbing origin supported by biomechanical similarities between primate vertical climbing and human bipedalism (Prost, 1980; Fleagle et al., 1981; Stern and Susman, 1981; Ishida et al., 1985; Senut, 1988), the terrestrial knuckle-walking hypothesis envisions a locomotor repertoire that also includes arboreal climbing and some suspension (Richmond et al., 2001).

The role of arboreality has been central in the debate over early hominin functional morphology, and particularly that of australopiths, since the early 1980s, following the publication of the Au. afarensis fossils from Hadar (Taieb et al., 1974; Johanson and Taieb, 1976; Bush et al., 1982; Johanson et al., 1982). This debate stems from the duality of the Au. afarensis, and that of australopiths more generally, postcranium that can be broadly summarized as possessing human-like lower limbs and ape-like upper limbs. The functional interpretation of Au. afarensis morphology is typically divided into two camps that differ in the significance they grant to the ape-like morphological features, as aptly summarized by Daegling (2022; see also Ward, 2002). In the ‘efficiency camp,’ researchers focused on human-like (i.e., derived) morphological features and biomechanical data to support the hypothesis that Au. afarensis had a bipedal gait similar to that of humans, for example, striding gait with extended hip and knee (Lovejoy et al., 1973; Lovejoy, 1975, 1978, 1988; Day and Wickens, 1980; White, 1980; Latimer, 1983, 1991; Ohman, 1986; Latimer et al., 1987; Latimer and Lovejoy, 1989, 1990; Crompton et al., 1998; Kramer, 1999). In this camp, Au. afarensis was reconstructed as fully committed to bipedalism because energetically costly bipedalism would not be positively selected, thus compromising any ability to climb competently. Arboreal behaviors were viewed as a trivial component of australopith positional repertoire.

In contrast, researchers in the ‘compromise camp’ incorporated both basal and derived morphology as well as biomechanical data to support the hypothesis that Au. afarensis engaged in a bipedal gait that was unlike that of modern humans and, instead, was compliant (i.e., bent hip and bent knee) and energetically costly (Senut, 1980; Stern and Susman, 1981, 1983, 1991; Feldesman, 1982; Jungers, 1982, 1991; Jungers and Stern, 1983; Schmid, 1983; Rose, 1984, 1991; Susman et al., 1984; Deloison, 1985, 1991, 1992; Tardieu, 1986a, Tardieu, 1986b; Susman and Stern, 1991; Duncan et al., 1994; Stern, 2000). This energetically costly bipedalism allowed Au. afarensis to be arboreally competent because trees were essential to, for example, foraging and avoiding predation (Susman et al., 1984; Preuschoft and Witte, 1991; Rak, 1991; Cartmill and Schmitt, 1996; MacLatchy, 1996; Schmitt et al., 1996, 1999; Ruff, 1998; Stern, 1999; also see below).

Numerous hypotheses have also been offered regarding the selective drivers underlying the origin of bipedalism (Rose, 1991, and references therein). These hypotheses are dependent not only on the functional interpretation of australopith morphology and on the form of positional behavior from which bipedalism evolved, but also on the paleoecological context in which early hominins may have been living. Many of the initial hypotheses linked the origin of bipedalism to a savanna landscape, either the traditional view of an open, grassland savanna or a savanna mosaic that includes areas of woodland (see review in Domínguez-Rodrigo, 2014). In an open landscape, bipedalism may have been an effective means of long-distance travel (Sinclair et al., 1986) or moving through patches of open terrain to reach woodland habitats (Rodman and McHenry, 1980; Isbell and Young, 1996; Potts, 1998). Others focused on the use of bipedalism for feeding in terrestrial, savanna-dwelling primates as a potential model for the selective driver of hominin bipedalism, all of which were grouped under the umbrella of the ‘terrestrial feeding hypothesis’ (Jolly, 1970; Rose, 1976, 1984; Wrangham, 1980). Jolly's (1970) ‘seed-eater’ model is particularly relevant to Hunt's (1994) ‘postural feeding hypothesis’ as it was the first time bipedal posture, rather than locomotion, was proposed as the selective target (see also Du Brul, 1962; Prost, 1980; Wrangham, 1980).