Limb proportions have been used to infer important aspects of early positional hominin behavior (i.e., locomotion and posture; Prost, 1965), but gaps in our knowledge presently hinder their effective use. Limb length proportions (the relative lengths of the limbs and their segments: e.g., intermembral, humerofemoral, brachial, and crural ratios) have been crucial in primate comparative studies since Tyson's (2018) recognition that differences in body proportions among apes, monkeys, and modern humans are linked to their behavior. Mollison (1910) and Schultz (1930, 1933, 1937, 1973) published detailed morphological data linking the diversity of primate limb length proportions to various aspects of their biology. Napier (1967) and Napier and Napier (1967, 1985) defined categories of extant primate locomotion and linked them to adaptive trends in body size and limb length proportions, with a focus on the intermembral index. Subsequently, Napier and Napier (1967, 1985; see also Napier and Walker (1967) became commonly cited to support hypotheses about locomotor mode in extinct hominins (Asfaw et al., 1999; Hunt et al., 1996; Jungers, 1985; Lordkipanidze et al., 2007; Lovejoy et al., 2009; Richmond et al., 2002; Wright et al., 2015). Some criticized the locomotor categories presented by Napier and colleagues as being broad and/or inaccurate (e.g., Cartmill, 1974; Fleagle, 1988; Mittermeier and Fleagle, 1976), but most researchers still support a link between predominant locomotor mode and limb length proportions, although that link has yet to be quantified in a comprehensive analysis of extant primates (e.g., Biegert and Maurer, 1972; Fleagle, 1988, 2013; Fleagle and Mittermeier, 1980; Jungers, 1985).

A lack of standardization in recording methods and behavioral categories has been an obstacle to collecting comprehensive behavioral data about wild primates, and thus an obstacle to quantifying the link to limb proportions. Hunt et al. (1996), using both their own and published observations, proposed using 32 positional modes subdivided into 74 locomotor and 52 postural sub-modes, to provide standardized categories of primate positional and locomotor behavior. In a more recent publication, Hunt (2016) provided the frequencies with which 33 extant hominoid and cercopithecoid taxa use the standardized behaviors in their positional repertoires, but as yet there has been no attempt to examine the relationship between limb length proportions and Hunt's classification in a diverse range of extant primates.

Researchers have investigated the effects of ontogeny, sex, phylogeny, environment, and scaling on limb length and cross-sectional proportions, locomotor behaviors, and functional morphology of the bones and joints of extant primates (Carlson and Pickering, 2004; Druelle and Berillon, 2013; Gordon et al., 2020; Hurov, 1987; Jungers et al., 1998; O'Neill and Dobson, 2008; Patel et al., 2013; Ramírez-Llorens et al., 2008; Ruff, 2002; Ruff et al., 2018, 2019, 2022; Ruff and Runestad, 1992; Takahashi, 1990; Turnquist et al., 1999; Vereecke et al., 2011). However, studies linking limb length proportions to the behaviors of primates in the wild are often restricted to comparisons among closely-related or sympatric species, or the methods used are not quantitative, or the comparisons are limited to one region of the postcranial skeleton (Strasser, 1992; Jungers, 1979; Nakatsukasa, 1996; Anapol et al., 2005; Su and Jablonski, 2009; Byron et al., 2017; Kimura, 2021; see also Polk, 2004). For example, although Rose (1996) and Wright et al. (2015; see also Wright, 2007) describe the limb proportions of extant platyrrhines and Miocene catarrhines, and capuchins, respectively, they avoid exploring quantitative links between morphology and behavior. Ford and Hobbs (1996) distinguish dietary, locomotor, and skeletal traits including limb bone lengths among Cebus species, but do not quantitatively support the observed differences, and the Rein et al. (2015) exploration of shape covariation to infer forelimb suspensory behavior in extinct primates was restricted to the ulna.

Simple mechanical principles described in Napier and Napier (1967) predict that relatively longer fore- or hindlimbs permit, for instance, relatively longer reaches or strides and greater acceleration during swinging or leaping and bipedalism, respectively (but posture, limb excursion, and body mass are other important, mediating factors; Preuschoft, 2002; Polk, 2004). Because of their clear links with limb dominance in primates, the intermembral index and its close approximate, the humerofemoral index, are frequently used to study locomotor differences, and the brachial, and crural indices are also informative about primate behavioral repertoires involving suspension and bipedalism due to similar rules (Anapol et al., 2005; Byron et al., 2017; Fleagle and Meldrum, 1988). However each index captures only part of the overall picture and, on its own, fails to capture body mass information and other limb segment information influencing a group's positional repertoire, thus they have been criticized as sole predictors of positional behaviors (Jungers, 1979, 1984, 1985; Tallman, 2013). Oxnard et al. (1981) combined multiple postcranial indices in discriminant analyses to reveal previously unrecognized group structures among prosimians, and Young (2003) used multivariate cladistic, principal component analyses, and cluster analyses on a large set of postcranial variables to demonstrate the same in the three anthropoid superfamilies. However, none of the traditional limb length proportions directly account for body size, which is a major selective factor affecting behavior, limb scaling, and function that is likely to affect results. Analyses of relative limb lengths that account for body size using proxy measurements (e.g., geometric means) could help improve efforts to use limb length data to retrodict the locomotor and postural behaviors of extinct primates including early hominins.

Here, we use a large, phylogenetically, behaviorally, and morphologically diverse anthropoid dataset to investigate how behavior and phylogeny influence limb proportions. We used a multivariate approach to understand overall appendicular shape by analyzing combinations of traditional limb bone length indices and size-adjusted ratios as shape variables. We build upon previous studies (e.g., Napier and Napier, 1967; Napier and Walker, 1967; Fleagle, 1988; see also Jungers, 1985; Tallman, 2013; Almécija et al., 2015; Hunt, 2016) by utilizing standardized locomotor categories based on published field observations to quantify multivariate correlations between overall limb length proportions and behavioral repertoires in a comprehensive comparative sample that includes multiple genera of all of the extant anthropoid superfamilies. We address three main research questions:

How well can extant anthropoid taxa be differentiated using limb bone length proportions alone, and which limb length proportions best separate them? If individual, or combined, limb proportions reliably separate large, phylogenetically diverse, samples of anthropoid taxa, then it is parsimonious to expect that the same is true of their fossil close relatives at equivalent taxonomic levels. We will test the hypothesis that limb length proportions differentiate extant anthropoid taxa using individual ratios and major axes of variation, with a focus on genera. We predict that major axes of multivariate variation differentiate taxa within our sample at least as well as individual limb proportions. Consistent with existing literature (e.g., Jungers, 1984, 1985), we also predict that of all individual ratios, intermembral and humerofemoral indices are most effective for distinguishing taxa.

To what extent do limb length proportions correlate with behavioral repertoires in extant anthropoids? Phenotypic convergence (e.g., Hylobates-like postcranial dimensions and forelimb-dominated locomotion in atelids; see Larson, 1998; Young, 2003), and unexpected behavior based on phenotype and phylogeny (e.g., modern human-like limb proportions and rare suspensory behaviors in colobines; Gebo and Chapman, 1995; Hunt, 2016; Byron et al., 2017) indicate a complex relationship between morphology and behavior. These relationships have yet to be quantified across extant anthropoids, which calls into question inferences about their fossil relatives. We will assess associations between limb length proportions and positional behaviors in extant anthropoids. Based on the extensive published literature, we hypothesize that taxa using a relatively greater frequency of forelimb-dominated behaviors have relatively elongated forelimb elements, and taxa using a relatively greater frequency of hindlimb-dominated behaviors have relatively elongated hindlimb elements.

How much of the observed variations in limb length proportions and positional behaviors are explained by phylogeny? Limb lengths are recognized to be under strong genetic control (Young et al., 2010; Rolian, 2014), but the extent to which phylogeny explains differences in limb length proportions and positional behaviors in extant hominoids and other anthropoids remains incompletely explored (but see O'Neill and Dobson, 2008 regarding phylogenetic signals in limb lengths of non-hominoids; Ruff et al., 2019 regarding behavioral correlates of articular and cross-sectional properties in cercopithecoid monkeys). This information is crucial for accurately interpreting fossil limb proportions in terms of ancestry, behavioral specializations, locomotor transitions, or convergence. We will assess hypotheses regarding the relationships among limb proportions, locomotor behaviors, and phylogenetic structure, and determine the strength of the resulting signals.

Our aim is to better understand the links between positional behavior, appendicular morphology, and phylogeny in extant anthropoids to improve future predictions regarding extinct hominins.