Humans and other hominins are characterized by a shared reliance on upright, hind limb-dominated positional behaviors. Although the obligate bipedal walking and running of humans is unique among primates, the hominoid lineage is itself characterized by an incredible diversity of locomotor and postural adaptations, ranging from the mixed quadrupedal and suspensory behaviors of chimpanzees and gorillas (Hunt, 1992; Doran, 1997), to the slow clambering of orangutans (Thorpe and Crompton, 2006), and the high-speed brachiation of hylobatids (Hollihn, 1984; Chang et al., 2000). A dependence on orthograde body postures, with the concomitant reorganization of the vertebral column, further characterizes all living hominoid taxa, and hominins in particular, and distinguishes them from other primates and from mammals generally (Schmitt, 2003). As a result, understanding the evolution of such behaviors within and across lineages is of primary importance in paleoanthropology, particularly as we seek to reconstruct the origins of bipedality.

Adaptations to specialized locomotor behaviors and orthograde posture encompass all regions of the body, most notably the axial skeleton. Early hominoids display differences associated with positional behavior in the position of the pedicle, orientation of the spinous process, and position of the transverse processes relative to non-hominoids, while extant hominoids also display differences to living cercopithecoids in vertebral body proportions, and the size and orientation of the zygapophyses and zygapophyseal processes (Sanders and Bodenbender, 1994; Nakatsukasa, 2008; Shapiro and Russo, 2019). Differences between humans and other living hominoid taxa specifically include a reorientation of the cranial base (Moore et al., 1973; Dean and Wood, 1981; Kimbel et al., 2014; but see Lieberman et al., 2000), a shift in the modal numbers of both thoracic and lumbar vertebrae (Williams et al., 2016), changes in the size and shape of the lumbar vertebral bodies (Shapiro, 1993; Lovejoy, 2005), and an increase in lumbar mobility in humans compared to great apes (Lovejoy, 2005; McCollum et al., 2010). These differences are typically attributed to orthograde bipedality, but the sparse fossil record as regards the vertebral column makes assessments of vertebral shape, number, and function among fossil hominins difficult, especially as these differences are difficult to assess from single elements. Partial vertebral columns from the same individual are rare, but they are available for Homo erectus (KNM-WT 15000, D2673, D2674, D2721, D2713, and D2672; Latimer and Ward, 1993; Lordkipanidze et al., 2007), Homo naledi (UW-101, UW-102; Hawks et al., 2017; Williams et al., 2017), Australopithecus afarensis (A.L. 288-1, DIK-1-1, KSD-VP-1/1; Johanson et al., 1982; Alemseged et al., 2006; Haile-Selassie et al., 2010; Ward et al., 2017), Australopithecus africanus (Sts 14, StW 431; Robinson, 1972; Haeusler et al., 2002; Toussaint et al., 2003), Australopithecus prometheus (StW 573; Crompton et al., 2022), and Australopithecus sediba (MH1, MH2; Williams et al., 2013, 2018). Other early hominin genera and the earlier hominoid fossil record are largely limited to isolated vertebral elements, including Paranthropus robustus (SK 854, SK 3981a–b, CD 5773, SKX 3342, SKX 41692, SKW 4776, KNM-ER 1825; Broom and Robinson, 1949; Robinson, 1970, 1972 but see Leakey and Walker, 1985; Susman, 1988, 1989; de Ruiter et al., 2009; Meyer and Williams, 2019), and Miocene hominoids Ekembo heseloni (KNM-RU 2036, KNM-KPS; Walker and Pickford, 1983; Ward, 1991; Ward et al., 1993), Ekembo nyanzae (KNM-MW 13142, KNM-RU 5944, KNM-RU 18385; Ward, 1993; Nakatsukasa et al., 2004), Morotopithecus bishopi (Moroto II; Nakatsukasa, 2008), Nacholapithecus keroi (KNM-BG 35250, KNM-BG 15527, KNM-BG-17826, KNM-BG 17822, KNM-BG 17822, KM-BG 427531, KNM-BG 47687A, KNM-BG 40949; Rose et al., 1996; Nakatsukasa et al., 2003; Nakatsukasa et al., 2007), Equatorius africanus (KNM-TH 28860; Sherwood et al., 2002) Otavipithecus namibiensis (BA 91–104; Conroy et al., 1996), Pierolapithecus catalaunicus (IPS-21350; Moyà-Solà et al., 2004), Hispanopithecus laietanus (IPS-18800; Moyà-Solà and Köhler, 1996), and Oreopithecus bambolii (IGF 11778 #50; Schultz, 1960). Together, specimens for which lumbar counts can be estimated suggest that early hominoids had six to seven lumbar elements (Nakatsukasa, 2019), more similar to cercopithecoids (Williams, 2012a) or hylobatids and hominins than to most extant apes, which display lumbar counts of three or four elements (Williams, 2012a; Williams et al., 2016). They also indicate that fossil hominins differ from both modern humans and extant great apes in the modal position of the thoracolumbar transitional vertebra (Williams et al., 2017). Finally, they demonstrate that human-like vertebral body and neural canal dimensions did not appear until early Homo (Meyer and Williams, 2019). Given the differences between fossil hominins and extant members of the Hominoidea and the lack of specimens from the period of divergence between panins and hominins, there is no clear way to assess the state of the last common ancestor (LCA) of these groups and the behaviors and vertebral anatomies that underlay hominin bipedalism.

In addition to the overall lack of early specimens, adaptations to locomotion and posture are convergent among primate and non-primate taxa that engage in similar behaviors. These convergences include a reconfiguration of at least some aspects of the cranial base in orthograde taxa (Russo and Kirk, 2013; Villamil, 2017), differences in spinal cord white matter according to forelimb-versus hindlimb-dominated locomotion (MacLarnon, 1995), greater variation in the number of presacral vertebral elements in suspensory taxa (Williams et al., 2019), and specific changes to the morphology of particular vertebral elements, such as more dorsally located lumbar transverse processes in suspensory taxa (Shapiro, 1993; Nakatsukasa, 2008). Anatomical convergence is thought to reflect adaptation to specific biomechanical constraints, such as the magnitude of forces acting on muscles at particular orientations (Shapiro, 1993) or, in some cases, to reflect a loss of biomechanical constraints associated with a particular kind of positional behavior (Williams et al., 2019).

Convergence among taxa has complicated our ability to interpret the fossil record as regards hominoid evolution generally, and hominin evolution specifically. New fossil finds call into question basic assumptions that similarity among traits in closely related species is a result of shared ancestry (Wood and Harrison, 2011). For example, analyses of the Orrorin tugenensis femur, a putative hominin dated to six million years ago, suggest that many shared aspects of hominoid femoral morphology have been independently derived in each taxon, and that both hominins and panins derive their particular morphologies from a more intermediate-looking ancestor (Almécija et al., 2013). Even orthogrady itself may have evolved multiple times within the hominoid lineage (Nakatsukasa, 2019). These kinds of findings have led to debates about the traits of the last common ancestor of hominini and panini, and about the presence and frequency of homoplasy in hominin evolution, with many studies devoted to assessing homoplasy and homology of specific morphologies and behaviors or their ontogenetic basis (MacLatchy et al., 2000; Nakatsukasa, 2004; Hall, 2007; Crompton et al., 2008; Begun and Kivell, 2011; Williams, 2012a; Reno, 2014; Prabhat et al., 2021). Such studies have often addressed both indirect indicators of posture (e.g., MacLatchy et al., 2000; Crompton et al., 2008), as well as axial morphologies directly linked to posture and locomotion, including the number of vertebral elements (Williams, 2012a) and aspects of thoracolumbar functional morphology (Nakatsukasa, 2008). Because of these difficulties, we need new ways to understand the evolution of hominoid locomotor and postural adaptations, which have implications for the origins of orthograde bipedalism and its associated anatomies in hominins.

One way to clarify evolutionary processes is by understanding ontogenetic processes (Reno, 2014). The process of adaptation shapes the pleiotropic pathways affecting the development and function of particular structures, resulting in evolutionary constraints (Jones et al., 2003; Melo and Marroig, 2015). Understanding the ways in which adaptations to positional behavior are linked to evolutionary constraints or the loss of constraints can provide insight into the regions that are under selection in particular taxa as well as clarify whether some adaptations will facilitate or limit subsequent evolution (Sansalone et al., 2022) or whether certain morphological features are likely to evolve multiple times (Goswami et al., 2014), for example, if apparently divergent suites of traits represent various points along a continuum of integrated options. One way to measure constraints is by examining phenotypic integration, which refers to covariation among traits resulting from shared genetic, ontogenetic, and functional pathways (Olson and Miller, 1958; Rolian, 2014). Covariation among traits is thought to facilitate or constrain evolutionary change by limiting the response to selection along particular paths of least resistance (Schluter, 1996; Hansen and Houle, 2004; Goswami and Polly, 2010; Goswami et al., 2014). Among taxa with similar covariation patterns, the same paths of least resistance can be exploited to generate a variety of forms and can also result in convergence and homoplasy as particular morphologies are repeatedly favored (Goswami et al., 2014). Importantly, however, phenotypic integration itself evolves in response to selection (Hansen and Houle, 2004; Melo and Marroig, 2015), including selection for locomotor traits, which has been demonstrated in the appendicular skeleton of hominoids, including humans, who diverge significantly from other taxa in the strength and sometimes pattern of integration (Young and Hallgrìmsson, 2005; Young et al., 2010; Grabowski et al., 2011; Lewton, 2012; Rolian, 2020). Generally, traits that function together are thought to evolve greater covariation (Cheverud, 1996).

The vertebral column additionally presents a useful system in which to explore the role of evolutionary constraints in evolution. Besides its functional importance in body support and as a muscle attachment site, and its particular importance in the transition to orthogrady among hominoids, the development of the vertebral column from somites and the determination of vertebral identity via specific patterns of Hox gene expression are generally well-understood (McGinnis and Krumlauf, 1992; Johnson and O'Higgins, 1996; Müller and Wagner, 1996; Carapuço et al., 2005; Mallo et al., 2010). Previous work has shown limited variability in both strength and pattern of integration among cervical vertebral elements in hominoids (Villamil, 2018; but see Arlegi et al., 2020, 2022), and mammals generally (Arnold et al., 2017; Arnold, 2021). Even relatively extreme mammals, such as giraffes, are exploiting existing patterns of integration to achieve novel shapes, rather than modifying the underlying relationships themselves (Arnold et al., 2017). However, the thoracolumbar region may be more responsive to locomotor-related selective pressures (Figueirido et al., 2021), and previous work in hominoids suggests variation between taxa in integration of individual thoracolumbar elements (Arlegi et al., 2020; Jung et al., 2021). What remains unknown, however, is the role of evolutionary constraints in the evolution of the hominoid vertebral column as a whole or in the reconfiguration of the hominin vertebral column.

The purpose of this paper is to assess phenotypic covariation among subaxial vertebral elements from three hominoid taxa and a macaque outgroup to determine whether adaptation to postural and locomotor behaviors has modified evolutionary constraints in this region and thus to clarify how the vertebral column is evolving in response to selection, particularly as it pertains to the obligate orthograde postures of modern humans.

We hypothesize the following:

Hominoids (gibbons, chimpanzees, and humans) will display shared patterns of high integration and low evolvability in the lumbar region, while macaques will not display such a pattern. Hominoids display a series of adaptations in lumbar vertebral form and column length, such as craniocaudally shorter vertebral bodies and pedicles and fewer numbers of vertebrae, that are typically linked to the stability and stiffness necessary for suspension and anti-pronograde postures (Williams, 2012b; Williams and Russo, 2015; Shapiro and Russo, 2019), and some authors have suggested that a similar suite of features evolved independently in different hominoid lineages (Nakatsukasa, 2019). These adaptations suggest the lumbar vertebrae are under selection to fulfill a specific, unified function in hominoids, a condition which is thought to lead to increased integration (Olson and Miller, 1958; Cheverud, 1984). Different authors have alternately found high (Arlegi et al., 2020) or low (Jung et al., 2021) integration for humans in this region, with variable results for other hominoid taxa (Jung et al., 2021).

Humans will display reduced integration and increased evolvability among pairs of elements in the cervical and thoracic regions. Humans not only lost the locomotor function of their upper limb, they also underwent anatomical and functional separation between the neck and shoulders when compared to other hominoids (Larson, 2009). Muscles of the shoulder typically attach to the cervical and thoracic vertebrae, so a loss of locomotor function likely represents a loss of evolutionary constraints on this region, leading to both reduced integration and increased evolvability.

Gibbons will display increased overall integration and decreased evolvability among pairs of elements in the cervical and thoracic regions. Gibbons engage in high velocity suspensory locomotion, which results in a ricocheting motion that is led by the upper limb, and which has led to the evolution of unique anatomical adaptations and powerful upper limb musculature (Schmitt, 2003; Jenkins, 1981; Michilsens et al., 2009). These specialized functions are expected to lead to strong selective pressures constraining trunk anatomy, including the anatomy of the vertebrae, leading to increased integration and reduced evolvability.

Chimpanzees and macaques will display similar patterns of integration and evolvability among pairs of elements in the cervical and thoracic regions. Although chimpanzees engage in a wider range of locomotor postures throughout ontogeny and display an orthograde body plan, both taxa are primarily quadrupedal in adulthood, using both upper and lower limbs to move and support their bodies (Hunt, 1992; Wells and Turnquist, 2001).