Functional morphology of the douc langur (Pygathrix spp.) scapula

1 INTRODUCTION

Colobine monkeys are generally considered to be arboreal quadrupeds with most exhibiting leaping behaviors to some extent (e.g., Gebo & Chapman, 1995; McGraw, 1998a; Rose, 1973). Suspensory locomotion (i.e., below branch arm-swinging and hanging) is atypical for this group (e.g., Davies & Oates, 1994; Fleagle, 2013) but has been intermittently observed in captive douc langurs (Pygathrix), first in zoos (Hollihn, 1973) and, more recently, at the Endangered Primate Rescue Center (EPRC) in Cuc Phuong National Park, Vietnam, where high frequencies of forelimb suspension and arm-swinging have been reported (Figure 1) (Byron & Covert, 2004; Wright, Stevens, Covert, & Nadler, 2008). Douc langurs are capable of suspensory postures similar to those of gibbons (Nomascus leucogenys) (Wright et al., 2008) but these are better described as arm-swinging, rather than true brachiation. Outside of the captive setting, information about douc langur locomotion, especially arm-swinging frequency data, is more limited. However, one study (Rawson, 2009) has documented arm-swinging in nearly 10% of all travel bouts across all age and sex classes in wild black-shanked doucs (Pygathrix nigripes). While this is less frequent than what is reported for captive Douc langurs, it is likely safe to infer that arm-swinging is not simply an artifact of captivity.

Scapula shape differences between quadrupedal and brachiating primates lie along a spectrum; that is, there are no discrete differences between the groups (Roberts, 1974). However, there are general trends in scapular form for each group. Typical quadrupeds have more laterally situated scapulae while brachiators and suspensory primates have more dorsally placed scapulae as a result of the broader thorax found in most hominoids (Gebo, 1996). The broad thorax sets the scapulae further apart and allows greater shoulder mobility (Gebo, 1996). The glenoid fossa of brachiators is wider and more cranially oriented (Larson, 1993), which has been proposed to be an adaptation for distributing strain more evenly across the glenohumeral joints (particularly during unimanual arm-hanging) (Hunt, 1991). The orientation of the glenohumeral joint is relatively static during ontogeny and has thus been suggested to be more developmentally constrained (Green, 2013). The acromion process, which articulates with the clavicle, typically projects further past the glenoid fossa in brachiators relative to arboreal quadrupeds (Larson, 1993). Concomitantly, brachiator clavicles are more often elongated (Ashton & Oxnard, 1964; Jenkins, Dombrowski, & Gordon, 1978). In suspensory primates, the acromioclavicular joint is more robust to better transfer weight between the glenohumeral joint and the manubrium (via the clavicle) (Hunt, 1991). The deltoid muscle, which attaches to and overlays the clavicle, acromion process, and humerus (Moore, Agur, & Dalley, 2002) is a shoulder rotator, abductor, flexor, and extensor (Myatt et al., 2012). This muscle is an important component of the brachiating anatomy because it envelops and stabilizes the glenohumeral joint, and is the primary shoulder abductor (a motion central to brachiation) (Ashton, Healy, Oxnard, & Spence, 1965; Jungers & Stern, 1981). In addition to the deltoid, four other scapular muscles are critical for shoulder stabilization: infraspinatus, supraspinatus, teres minor, and subscapularis (also known as the rotator cuff muscles) (Dvir & Berme, 1978). These muscles are also involved in movement of the shoulder joint (Roberts, 1973). The supraspinatus muscle attaches on the supraspinous fossa of the scapula and inserts on the middle facet of the greater tubercle of the humerus. The supraspinatus assists the deltoid in arm abduction (Moore et al., 2002; Myatt et al., 2012). The origin of the supraspinatus muscle in the supraspinous fossa is quite broad in brachiators while it is comparatively long and narrow in quadrupeds (Ashton & Oxnard, 1963). Because the deltoid and supraspinatus muscles heavily contribute to the ability to brachiate, their attachment sites are typically more robust. The other three rotator cuff muscles (infraspinatus, subscapularis, and teres minor) show no considerable differences between brachiators and quadrupeds (Ashton & Oxnard, 1963).

In addition to specific scapular differences between brachiators and quadrupeds, there are also disparities in the morphological gestalt. Overall, the brachiator scapula is taller from the most superior to inferior points and more narrow across the scapular spine (Larson, 1993). This narrower scapula is thought to maximize the range of rotation of the shoulder joint, thus bringing the glenoid fossa closer to a position over the center of gravity during arm-hanging. In this position, the glenoid is aligned with the spinal column so that the two elements are in the same line of action. This functions to decrease bending of the spinal column, reduce shear stress on the structures between the glenoid and spine, and more evenly distribute compressive stress on the rib cage (Hunt, 1991). By contrast, a quadruped scapula is more laterally placed, shorter in overall height, and wider across the scapular spine. This positioning and shape function to limit the mobility of the shoulder joint and increase stability (Larson, 1993). The brachiator scapula is also generally more robust and this robusticity corresponds to the physiological cross-sectional area of the muscles of the forelimb. Greater cross-sectional area increases the maximum available force of a muscle, which is especially important in the brachiator forelimb muscles (Anapol & Gray, 2003; Fleagle, 1976; Wright et al., 2008). One of the ways gibbons have adapted an efficient brachiating pattern is by distributing forelimb muscle mass at the proximal ends of the limb, keeping the majority of that muscle's mass centered near the trunk (Michilsens, Vereecke, D'Août, & Aerts, 2009).

Observations of the douc langurs arm-swinging at the EPRC invites speculation that their scapular morphology reflects this derived—relative to other colobines—locomotor behavior. Given the differences in scapular anatomy between brachiators and quadrupeds, we predict that the douc langur scapula possesses features analogous to obligate brachiators (such as gibbons). Previous studies have used indices to measure scapular form but capture only a portion of the overall shape. For example, Su and Jablonski (2008) found that odd-nosed monkeys (Pygathrix, Simias, Nasalis, Rhinopithecus) have derived scapular morphology more similar to that of extant apes and Covert, Workman, and Byron (2004) found douc langurs to have a longer vertebral border of the scapula compared to the Delacour langur (Trachypithecus delacouri). Here we present a more comprehensive analysis of overall scapular form. We specifically aim to measure douc langur scapular morphology relative to other cercopithecoid quadrupeds and to hylobatid brachiators, and to assess their similarities and differences. Should the results indicate that Pygathrix scapular morphology is more similar to quadrupeds, this would suggest that the arm-swinging behaviors of the douc langurs are more facultative in nature and their scapular morphology is reflective of more habitual, quadrupedal locomotor behaviors. A more similar morphology to true brachiators (i.e., hylobatids) would suggest selective pressures have acted on the douc langur scapula to accommodate suspensory activity.

2 METHODS

This research adhered to the American Society of Primatologists Principles for the Ethical Treatment of Non Human Primates. Seventeen landmarks (Table 1, Figure 2) were collected following the methods of Young (2006, 2008) from scapulae of 100 adult individuals from 15 different species of catarrhines (Table 2) using a Microscribe G2X from Immersion Technologies (San Jose, CA). All landmarks were recorded from the dorsal view to avoid potential errors arising from repositioning the specimens during data collection. Juveniles were excluded due to exaggerated variability in their skeletal morphology (see: Workman & Covert, 2005; Young, 2006). Specimens showing disease, which deformed the scapulae, or significant amounts of post-mortem damage obscuring collection of three or more landmarks, were also excluded from the analyses. Only one scapula, either left or right, was sampled from each specimen. Most of the specimens measured were from the wild but the Pygathrix sample included both wild and captive individuals.

For the purposes of this study the three douc langur species, Pygathrix nemaeus, Pygathrix nigripes, and Pygathrix cinerea, were grouped into a single entry (Pygathrix) and assigned to the “unknown” locomotor category. This reflects our agnostic position regarding their status as brachiators or quadrupeds. Previous work has found no significant difference in scapular shape between the three species of douc langurs, or between wild and captive specimens (Bailey & Pampush, 2015), suggesting their condensation into a single group is appropriate for these analyses. The other cercopithecoids in the sample include Piliocolobus badius, Cercopithecus diana, Cercopithecus campbelli, Cercopithecus petaurista, and Trachypithecus germainii. These taxa were assigned to the “quadrupedal” locomotor category, consistent with previous research (e.g. Fleagle, 2013; Napier & Napier, 1967). The hylobatids (Hylobates spp. and Nomascus spp.) were assigned to the “brachiator” locomotor category also based on the literature (e.g., Fleagle, 2013; Napier & Napier, 1967).

All analyses were completed using the R package geomorph (Adams, Rohlf, & Slice, 2004; R Core Team, 2013). The sample includes specimens from both the left and right sides, therefore, mirroring was required. A Procrustes analysis was used to uniformly scale and align the point clouds, which allowed for sexes to be pooled together (as size is accounted for with the Procrustes transform). Seven specimens (two P. nigripes specimens and five P. badius specimens) missing three or fewer landmarks were imputated (i.e., the missing landmarks were statistically reconstructed based on species averages and the other landmarks of the particular individual). A Principal Component Analysis was performed to examine variance in scapular shape and Procrustes ANOVAs were conducted to test for shape differences among the scapulae. A pairwise analysis was conducted to compare the means for the three locomotor groups to determine which groups (if any) were significantly different from each other. To be sure phylogeny was not having an outsized effect on the measure of scapular shape, a phylogenetically controlled Procrustes ANOVA was performed. Performing such an analysis required two additional sets of data. First, a phylogenetic tree was downloaded from 10kTrees (Arnold, Matthews, & Nunn, 2010). Second, a set of mean coordinates was produced for each taxon, this was done using the “mean shape” tool contained within geomorph. The raw data used in this study has been provided in a supplementary excel file, which can be found online.

3 RESULTS

The three locomotor classifications grouped into distinct morphological clusters in the Principal Component Analysis (Figure 3). Results of the PCA are presented in Table 3 and Figure 3. The wireframes in Figure 3 illustrate the scapula shape of each of the three locomotor categories. The quadrupeds (other cercopithecoids) clustered negatively on the first axis and distributed positively and negatively across the second axis (but mainly positively). The brachiators (hylobatids) are clustered positively on the first axis and are predominantly positively on the second axis. Pygathrix falls between the cercopithecoids and the hylobatids with a positive–negative spread on the first axis and a largely negative distribution across the second axis. PC1 accounts for 39.66% of the observed variation and PC2 accounts for 20.12% of the variation (59.78% cumulatively). PC1 discriminates all three locomotor groups with the greatest variation while PC2 contains more overlap. Positive PC1 scores are characterized by a more inferiorly placed scapular spine along the vertebral border, a relatively longer acromion process (and by extension, scapular spine), and a superiorly oriented glenoid fossa. Pygathrix shares a superiorly placed scapular spine along the vertebral border with the cercopithecoids and a long acromion process and scapular spine with the hylobatids. The more acute angle between the coracoid prominence and the superior and inferior tips of coracoid process also links Pygathrix to other cercopithecoids on PC1. This angle on the hylobatids is not nearly as acute. While Pygathrix is more similar to the cercopithecoids on PC1, they are separated on PC2. Negative PC2 scores are characterized by a more medially placed superior angle thereby shortening the superior border and lengthening the superior aspect of the vertebral border. Additionally, the lower PC2 scores are distinguished by a mediolaterally compressed and craniocaudally elongated scapular body.

An ANOVA comparing scapula shape for all locomotor types indicated significant differences among the groups (p < 0.01) (Table 4). Further, pairwise analysis showed Pygathrix scapular morphology to be significantly different from both quadrupeds (p < 0.01) as well as brachiators (p < 0.01) (Tables 5 and 6). Results of the phylogenetically controlled Procrustes ANOVA also showed a significant difference (p = 0.008) (Table 7).

4 DISCUSSION

The results presented here show that the scapular morphology of Pygathrix is distinct from that of both quadrupedal cercopithecids and brachiating hylobatids. The morphology of the douc langur scapula is intermediate between typical brachiators and quadrupeds and may indicate the uncommon locomotor repertoire of this genus. On PC1 Pygathrix is similar to other ceropithecoids with its long vertebral border but the laterally projecting acromion process, long scapular spine, and more cranially oriented glenoid fossa resemble the hylobatids. On PC2 Pygathrix is distinct from both the other cercopithecoids and the hylobatids with its mediolaterally compressed scapula and more medially located superior angle.

The difference in η² values between the phylogenetically controlled and non-phylogenetically controlled procrustes ANOVAs prompted an unplanned test for phylogenetic signal. This was performed in geomorph using the function “physignal.” This returned a Blomberg's K value of 0.0073 (Table 8), which is very small yet significantly different from zero. So, while there is some effect of shared descent on the distribution of these data, the effect is rather small and is unlikely to be the cause of the η² differences. Rather, the difference in η² values is likely the product of digesting the point cloud data into species means to perform the phylogenetically corrected analysis. This data aggregation dramatically lowered the number of data points and limited the power of the analysis.

Upon casual observation, the Pygathrix scapulae resembles that of Piliocolobus badius (Figure 4). However, upon closer inspection there are several characteristics that set them apart. The acromion process is more robust and projects further laterally in Pygathrix than the other cercopithecoids, in addition to a more robust scapular spine. The relationship of the acromioclavicular joint to arm elevation is an obvious explanation for the more robust acromion found in brachiators (Hunt, 1991). A major difference between the hylobatids and the colobines is the teres major attachment site (the boney portion between the teres major fossa and inferior angle). In the colobines, the teres major attachment site projects further laterally compared to the hylobatids. In P. badius, this has been proposed to be a result of adduction of the forelimb after superior retrieval of food during foraging (Dunham, Kane, & McGraw, 2015). Pygathrix has a teres major attachment site that is similar to P. badius but it is not quite as large. The superior angle of the superior border is distributed more medially in the douc langurs compared to both the hylobatids and cercopithecids. A plausible explanation for this is the abundant use of the supraspinatus muscle, which attaches along the superior aspect of the supraspinatus fossa (superior border). The supraspinatus is critical for stabilization in the glenohumeral joint in both brachiators and quadrupeds, and for the elevation of the arm in brachiators (Potau et al., 2011). The glenoid fossa is less cranially oriented in Pygathrix than in the hylobatids. These differences are to be expected given how muscle insertions and skeletal morphology of the shoulder girdle of arm-swinging colobines are described as intermediate between quadrupeds and true brachiators (Ashton & Oxnard, 1964).

Another factor related to below-branch locomotion is body size. A general trend observed within and between primate clades is that as body size increases, the frequency of below-branch locomotion also increases, presumably because it is easier for large-bodied primates to hang below a branch than to balance on top of it (Cartmill, 1985; Fleagle & Mittermeier, 1980; Napier, 1967; Napier & Napier, 1967; Ward, 2007). Among the colobines, Pygathrix lies towards the larger end of the body mass spectrum (Delson et al., 2000; Smith & Jungers, 1997). Colobines that rarely engage in arm-swinging, such as P. badius and Trachypithecus francoisi, (McGraw, 1998b; Zhou, Luo, Wei, & Huang, 2013) weigh less than the more suspensory Pygathrix nemaeus. A parallel trend appears in New World monkeys as well (Fleagle & Mittermeier, 1980). Two groups of large-bodied New World monkeys Ateles spp. (spider monkey) and Lagothrix spp. (woolly monkey) are well known for their suspensory behaviors (Napier & Napier, 1967; Smith & Jungers, 1997), with suspensory locomotion comprising 23.3% and 11.7% of travel time, respectively (Cant, 1986; Defler, 2000). Smaller new world monkeys, such as Saimiri boliviensis (Fleagle, 2013) use suspensory postures <1% of the time (Arms, Voges, Fischer, & Preuschoft, 2002). The brachiation frequencies of Ateles and Lagothrix are less than the arm-swinging frequencies reported for Pygathrix, but this corresponds to their slightly smaller body size.

The suspensory locomotion used by the atelines differs from that of Old World brachiators by use of a prehensile tail in the former. Use of the prehensile tail results in a more pronograde suspension in the atelines, compared to the more orthograde suspension observed in Pygathrix and the hylobatids (Hunt et al., 1996). Despite this difference, the scapular morphology of Ateles and Lagothrix resembles that of the hylobatids (Kagaya, 2007); Ateles has a wide glenoid fossa while both Ateles and Lagotrhix have a wide supraspinatus fossa (Campbell, 1937), which is consistent with typical brachiator morphology (Ashton & Oxnard, 1963). Overall, the scapula morphology of Lagothrix is described as intermediate between that of Ateles and Alouatta, with Ateles being the most similar to the hylobatids and Alouatta representing a more typical quadruped (Gebo, 1996). Further, both Ateles and Lagothrix have a cranially oriented glenoid fossa (Gebo, 1996) a feature they share with Pygathrix and the hylobatids. Features shared by these taxa may be helpful for identifying suspensory behaviors in the fossil record. If such features appear in concordance with body size increases during hominoid evolution, particularly in the hylobatid lineage, this could be a useful line of evidence illuminating the origins of below-branch locomotion in hominoids.

This study lays the foundation for future research. A wild-based study focusing exclusively on locomotion and positional behaviors is the next logical step in identifying factors influencing arm-swinging frequency in douc langurs' natural habitat. Coupled with the present scapular morphology investigation, a locomotor study could explain why arm-swinging is observed in this genus at frequencies unique among colobines. Furthermore, red and black-shanked douc langurs are classified as Endangered and grey-shanked douc langurs are classified as Critically Endangered by the International Union for Conservation of Nature (IUCN) (Rawson, Lippold, Timmins, Ngoc Thanh, & Manh Ha, 2008; Ngoc Thanh, Lippold, & Nadler, 2008; Ngoc Thanh, Lippold, & Timmins, 2008). Behavioral and ecological data are imperative for ensuring their future conservation.