The primate shoulder girdle exhibits the greatest interspecific morphological variation of all placental mammal orders (Oxnard, 1968; Preuschoft et al., 2010). The observed morphological variation among primates matches the diversity of upper limb functional demands, and corresponds to the size, shape, and orientation of the muscles attaching to the bony elements of the shoulder girdle, especially the scapula (Ashton and Oxnard, 1964a, Ashton and Oxnard, 1964b; Oxnard, 1967, 1968; Badoux, 1974; Roberts, 1974; Ashton et al., 1976; Larson and Stern, 1986, 1992; Larson, 1993, 1995, 2015; Irwin and Larson, 2000; Young, 2008; Preuschoft et al., 2010). Morphological differences that correspond with functional variation in the scapula are present from birth and are maintained through ontogeny (Young, 2006). This indicates that variation in scapular shape is not solely a plastic response to biomechanical demands encountered during the lifetime of individuals. Nevertheless, whether this variation arose in response to direct selection brought about through physical demands on the shoulder related to its function, by correlated responses to directional selection on integrated traits (e.g., Rolian et al., 2010; Rolian, 2014; Grabowski and Roseman, 2015), or through neutral processes remains undetermined.

Studies of morphological variation in the primate shoulder girdle are often focused solely on the scapula (Schultz, 1930,1950; Ashton and Oxnard, 1961, 1964a,b; Oxnard, 1963,1967, 1968, 1969; Ashton et al., 1965; Roberts, 1974; Shea, 1986; Inouye and Shea, 1997; Taylor, 1997; MacLatchy et al., 2000; Taylor and Slice, 2005; Larson et al., 2007; Green, 2013; Green et al., 2106; Feuerriegel et al., 2017; Selby and Lovejoy, 2017), as it serves as a main attachment point for many of the muscles associated with upper limb and shoulder movement. This muscle–bone relationship informs a form-function model where the form of the scapula reflects movements of the shoulder region. Using this model, researchers have drawn evolutionary conclusions in their studies that implicitly make two assumptions: 1) trait differences among taxa result from directional selection acting on those morphologies; and 2) variation within individual traits between primate species reflect adaptations for specific motor functions or environments. While these conclusions may be merited, most studies have not used evolutionary models to quantify how evolution could have occurred, or might occur in the future, to explain interspecific variation (cf. Roseman and Auerbach, 2015). Rather they use a framework based on comparative anatomy. Moreover, the individual assessment of traits implicitly assumes their independent evolution, an assumption that is not supported by the evolutionary biology literature for most traits (Lande, 1979; Lande and Arnold, 1983; Arnold, 1983; Ackermann and Cheverud, 2000; Hansen and Houle, 2008; Walsh and Blows, 2009; Rolian et al., 2010; Roseman and Auerbach, 2015; Katz et al., 2016; Savell et al., 2016). Traits within an organism do not exist in isolation. Rather, they covary with whole-organismal traits (such as body size) and with traits from other anatomical regions that share mechanical and developmental variation.

In a recent study, Agosto and Auerbach (2021) used a quantitative genetics approach to assess the potential evolutionary covariance between the shoulder girdle and two anatomical regions with which it shares mechanical or developmental covariance: the basicranium and pelvic girdle. A novel result of this study indicates that traits of the basicranium and pelvic girdle can exert similar degrees of evolutionary constraint on traits of the shoulder girdle. Furthermore, the results suggest the basicranium imposes evolutionary constraint on both the shoulder and pelvic girdles but is not equally constrained by either the shoulder girdle or pelvic girdle. Additional analyses by Agosto and Auerbach (2021), which include traits of the proximal humerus, also demonstrate greater potential evolutionary independence of these traits from the basicranium and pelvis compared with the scapula. This result aligns with known developmental and functional relationships among these anatomical regions. Overall, Agosto and Auerbach (2021) demonstrated that the evolution of the primate shoulder is more complex than previously perceived, uncovering evolutionary covariance with anatomical regions once assumed to be independent of the shoulder (i.e., the basicranium). These relationships, however, are not evident when evaluating the scapula or its traits alone.

While the study by Agosto and Auerbach (2021) established a model where the evolution of the shoulder girdle is influenced by its underlying covariances with anatomical regions in which it shares functional and developmental relationships, a non-comparative approach was taken by investigating these relationships within a single genus: Colobus. How these relationships relate to the larger breadth of primate shoulder girdle morphological variation remains unanswered. Researchers cannot assume that the results indicated for the monkey taxa investigated by Agosto and Auerbach (2021) are representative of the potential evolutionary relationships among the three anatomical regions for apes, let alone all primates. For example, other studies (Young, 2004; Rolian, 2009; Young et al., 2010; Lewton, 2012) have demonstrated that integration, and thus the evolvability, of postcranial traits vary considerably among primates, and that apes generally are less integrated than monkeys. The present study therefore expands upon the findings of Agosto and Auerbach (2021) by comparing the patterns of evolutionary potential among anatomical regions between a broad sample of primate genera.

By investigating morphological integration and estimates of evolvability, we seek to understand how trait covariance that has resulted from developmental and mechanical relationships between the shoulder, basicranium, and pelvis varies among primate genera, and whether these could have influenced the observed diversity in shoulder girdle morphology among primates. We use a model where shared development and function influence evolutionary potential among these anatomical regions. Additionally, we include measures of the distal humerus as a more independent anatomical region under our model, as the distal humerus has no direct muscle attachments or known shared developmental processes or factors with either the basicranium or pelvis. We asked the following two questions:

Is there evidence that evolutionary constraint among the basicranium, shoulder girdle, and pelvic girdle is present among a range of primate taxa?

Do patterns of evolutionary constraint among primates correspond to functional variation of the upper limb as broadly reflected by locomotion (e.g., Roberts, 1974; Ashton et al., 1976; Larson, 1993, 1995; Irwin and Larson 2000)?

To address these questions, the magnitude of integration and measures of evolvability, namely conditioned covariance and evolutionary flexibility, were compared across anatomical groups within and among eight primate genera. As discussed in Section 2.4, these statistics provide estimates of the constraint, and thus potential of traits to respond to directional selection. We expect that there will be evidence that, especially for the basicranium, shoulder, and pelvis, relationships with each other will lead to constraints in the evolutionary potential of the traits within each anatomical region, following the results of Agosto and Auerbach (2021). Furthermore, it is expected that anatomical regions with direct functional anatomy and/or developmental relationships, such as the basicranium and shoulder girdle (see Matsuoka et al., 2005; Diogo and Wood, 2012), will have greater magnitudes of morphological integration than functionally unrelated traits within each taxon (see Cheverud, 1996; Wagner, 1996; Rolian, 2009). Among primates, we expect monkeys and apes to show a pattern of integration and evolvability where monkeys have greater integration and lower evolvability than apes for the same pairs of traits, following the broad patterns reported previously (Young, 2004; Rolian, 2009; Young et al., 2010; Lewton, 2012).