Morphological Patterns in the BSA of Jamaican Anoles
To ground our study we employed Iguana iguana and Cophosaurus texanus as outgroups to establish a comparative framework that could be explored qualitatively and quantitatively. Comparing the mounted skeleton of I. iguana with the three-dimensional renditions of C. texanus, it becomes apparent that in the scans only the extent of the suprascapula and epicoracoid exhibit noticeable variability in their rendition between different specimens. All other skeletal features of the BSA are readily observable to their full extent in the virtual reconstructions.
Among the species of Norops examined we were able to determine, through qualitative and quantitative comparison, that the components of the BSA of all species examined differ in their morphology from those of the other species, but that few of these differences are actually discriminatory (Fig. 5, Tables 8-11).
Our analysis also revealed that sexes differ in their morphospatial distribution, although a clear discrimination between sexes is not possible (Fig. 7, Tables 6 and 7). If a morphological difference between the BSA elements of male and female Norops exists, it is too subtle to recognize employing our methods. Because our focus is on the exploration of morphological differences between Jamaican anoles, we do not further explore sexual dimorphism at this time.
Since our data are based only on adult specimens, allometric effects should play only a minor role on the perceived size differences. However, it is still possible that one species merely represents an upscaled form of another, so the data needed to be tested for size-correlation. A regression of the PCs against logCS revealed only a very weak association between size and shape (Table 5). The strong size correlation that is evident in the vertebral column is not unexpected, since the landmarks that make up this moiety are aligned along the body long axis. This structural correlation with body length makes a size-based prediction of 12% for PC1 (Table 5) appear very low.
Powell and Russell (1991) examined external morphometric features of N. garmani, N. opalinus and N. grahami, and showed that the features that they investigated scale differently among these species. Our analysis results in a similar conclusion. Although they are similar in habitat preference, N. garmani and N. grahami differ morphologically from N. opalinus, and not merely in the scaling of their skeletal elements (Table 5, Fig. 8). Without further data relating to allometric growth of the species and skeletal elements examined, we are unable to argue conclusively whether the correlation between size and PCA is driven by phylogeny, ontogeny or absolute size. Furthermore, the covariation between size and shape that is evident in our analysis is relatively small (Table 5). Adopting a conservative definition of “significance” (Higgs, 2013), and recognizing the relatively small correlation between shape and absolute size, we decided not to “correct” for size in our data. Instead, we used the Procrustes data and principal components as represented by their respective analyses.
The PCA reveals that all species occupy distinctive regions of morphospace (Fig. 8). Although the analysis through principal components does not allow for clear-cut discrimination between these groups, the differential morphospatial distribution allows us to test whether distinctive areas of the morphospace are preferentially occupied by distinct species or ecomorphs. CVA reveals that the means of most species differ from each other (Tables 8-11. However, the data for the vertebrae have only very little discriminatory power, and only N. valencienni can be distinguished from the other species with some certainty (Table 8). The interclavicle-presternum moiety of N. grahami is barely distinguishable from that of any other species examined, but representatives of all other species differentiate relatively well (Table 9). N. grahami performs relatively well in the CVA of the clavicle data set, but all other species, including N. valencienni, appear to be less differentiable (Table 10). Relatively the best CVA scores are achieved for the scapulocoracoid, where at least 50% of every species group is correctly classified, and N. grahami can be distinguished from all species but N. opalinus (Table 11).
The ecomorph-based CVA revealed discriminatory power in the shape changes of the four moieties (Fig. 13, Tables 12-15 that are similar to those retrieved from the species-based CVA (Fig. 8, Tables 8-11. The strongest discrimination is achieved for the scapulocoracoid dataset, and the relatively weakest correlation values are found for the vertebral moiety, regardless of whether the data are classified by species or ecomorph designation (Tables 11 and 15). When N. grahami and N. opalinus are grouped together as a single trunk-crown ecomorph, the latter becomes comparable in sample size (Fig. 3) and disparity (Fig. 13) to the trunk-ground ecomorph (N. lineatopus), which makes these two groups more readily comparable as ecomorphs than they are as species. The morphometric differences revealed by the PCA and CVA are very similar (Figs. 8 and 13). However, since the morphometric disparity varies between ecomorph groups (Fig. 13), it is also easy to overinterpret the canonical variation.
Combining the data of N. opalinus and N. grahami into a trunk-crown ecomorph group results in a greater discriminatory power for the scapulocoracoid data set (Tables 11 and 15), but not for the vertebral moiety (Tables 8 and 12). This shows that ecomorph groups do not necessarily have greater explanatory power for the morphological diversity that is evident in our data than do species groups. There appear to be species-specific differences between N. opalinus and N. grahami that prevent a more effective discrimination of these two species from all others. The clavicle of N. grahami appears to be distinctly different from that of N. opalinus (Table 10), and their combination as trunk-crown ecomorph results in less discrimination from the other species (Table 14). The form of the clavicle is very variable among lizards, and characters related to its morphological diversity have featured repeatedly in phylogenetic analyses (Sukhanov, 1961; Lécuru, 1968b; Russell, 1988). However, little is known of the functional implications of these differences in clavicular form (Peterson, 1973; Russell, 1988; Russell and Bauer, 2008). We posit that the form of the clavicle is more variable within small radiations (such as the monophyletic radiation of Jamaican Norops) than is that of any of the other elements of the BSA, which makes it more susceptible to microecological adaptation than the other skeletal elements. This remains to be tested.
The major findings from the DF are that (i) the relative distance between the anterior extremity of the interclavicle and that of the presternum increases from N. valencienni, to N. opalinus, to N. grahami and N. garmani, and is greatest in N. lineatopus (Figs. 10 and 13b), and that this distance covaries with the angle between the lateral processes and the posterior process of this element. (ii) The relative length of the medioventral portion of the clavicle increases from N. valencienni, to N. lineatopus, to N. grahami and N. garmani, and is longest in N. opalinus (Fig. 11). (iii) The scapulocoracoid is lateromedially relatively the narrowest in N. valencienni and N. lineatopus, is slightly wider in N. opalinus, and is widest in N. grahami and N. garmani (Figs. 12 and 13d). (iv) The coracoid, scapulocoracoid and scapular rays are anteroposteriorly short in N. valencienni, relatively longer in N. garmani, N. grahami and N. lineatopus, and are longest in N. opalinus. No two patterns in any of these four major findings describe the same trajectory in terms of sequential change between the species, and the four patterns are distinct from one another, meaning that every element of the BSA conveys different information in relation to the differentiation between species and ecomorphs of Jamaican Norops.
In attempting to summarize our findings, and seek features that potentially discriminate N. valencienni, N. grahami, N. garmani and N. opalinus from N. lineatopus, and from each other, we note the following differences in comparison with N. lineatopus.
Norops valencienni (twig giant) is characterized by relatively long thoracic vertebrae; a lateromedially narrow and anteroposteriorly lengthened presternum; the dorsolateral process of the presternum is relatively long; its mesosternum articulates with the third sternal rib; the posterior process of the interclavicle is relatively short and there is a large angle between the lateral and posterior process; the apex of the primary curvature of the clavicle is ventrally shifted; the scapula and suprascapula are dorsoventrally tall; the scapula, coracoid and epicoracoid are anteroposteriorly short.
N. grahami (trunk-crown) exhibits anteroventrally shifted articulatory facets of the sternal ribs; a relatively lateromedially wide and anteroposteriorly short presternum; a slightly lengthened posterior process of the interclavicle; a relatively small dorsal shift of the apex of the clavicle; an anterolateral displacement of the anterior extremity of the scapular ray and of the scapula-suprascapular boundary; and a smaller angle between the ventral and dorsal portion of the coracoid.
N. garmani (crown giant) is characterized by a slightly lateromedially narrower and anteroposteriorly elongated presternum (relatively the second-longest presternum after that of N. valencienni); a slightly elongated posterior process of the interclavicle; a relatively small dorsal displacement of the apex of the clavicle (as also seen in N. grahami); an anterior and medial elongation of the epicoracoid; a dorsal displacement of the glenoid fossa; a relatively dorsoventrally taller scapula; and a lateromedially wider scapulocoracoid.
N. opalinus (trunk-crown) is characterized by a relatively dorsally displaced apex of the clavicle; a dorsoventrally short suprascapula; and torsion around the dorsoventral axis of the scapulocoracoid.
Placing this understanding of variation in structure of the BSA into the context of the ecology of these species, we note that differences between the taxa considered are subtle and relatively slight, except for the situation revealed for the twig ecomorph Norops valencienni. This twig anole exhibits some unusual traits within this radiation, such as the very narrow and elongated presternum (Figs. 10 and 13b), the elongated paired dorsolateral processes of the presternum (Fig. 6e), articulation of the third sternal rib with the mesosternum instead of the presternum (Figs. 5b and 6), the short posterior process of the interclavicle (Fig. 10), and the anteroposteriorly short scapulocoracoid (Figs. 12 and 13d).
Although N. valencienni is absolutely larger than most of the other examined species, the specimens of N. garmani are of about the same size, or larger (Fig. 3), without showing a morphology as distinctive as that of N. valencienni. Thus, the unique shape of the skeletal elements of the BSA of N. valencienni cannot be attributed to its size alone. According to recent phylogenetic studies, N. valencienni is a relatively basal member of the Jamaican anole radiation (Fig. 2, Alföldi et al., 2011; Nicholson et al., 2012), and according to our analysis there exists little phylogenetic signal in the data (Table 16). It is rather unlikely, although not impossible, that its distinctive morphology evolved at or near the phylogenetic node that it shares with the other species of Norops examined.
An alternate explanation for the discrete morphology of the BSA of N. valencienni is that it has deviated from its Jamaican congeners in relation to its adaptations relating to locomotor and habitat occupancy characteristics that differ markedly from those of the other examined species. Indeed, N. valencienni is unusual in that it forages widely (contrasting with the sit-and-wait foraging mode of the other Norops species examined; Losos, 1990b). In comparison to other Jamaican Norops, it moves for relatively long periods of time, but moves relatively slowly (Hicks and Trivers, 1983). Like other twig anoles it exhibits a great range of motion at the shoulder joint (Peterson 1974). Peterson's observations (1974) described twig anoles as being able to markedly extend their stride length via excursion at the coracosternal joint. This contradicts the findings of Higham et al. (2001), who reported a relatively short stride length for N. valencienni as compared with N. lineatopus and N. garmani. However, the latter authors studied anoles running at their maximum speed, and thus their results are not directly comparable with those reported by Peterson (1973, 1974), who studied the locomotion of these anoles in their natural environment. Since twig anoles manoeuvre on relatively the narrowest branches, in comparison to other ecomorphs (Butler and Losos, 2002), they profit from increased limb mobility in exchange for a relative reduction in locomotor speed.
Our data agree with the findings of Mahler et al. (2013) who showed that many island species of anoles are represented on other islands by almost perfect morphological copies, whereas other species exhibit distinct character sets that set them apart. Such gross morphological deviation between closely related lizards is not limited to anoles, but also occurs, for example, in geckos (Higham and Russell, 2010). The external morphology of the desert-dwelling Rhoptropus afer differs greatly from that of its sister species R. bradfieldi, whereas the latter is morphologically very similar to its other congeners (Bauer et al., 1996). Higham and Russell (2010) associated this great deviation in morphology, that occurred over a relatively short evolutionary time span, with a suite of locomotor adaptations peculiar to R. afer. We posit that morphometric changes along the branch leading to N. valencienni do not represent changes at the ancestral node, but instead bear evidence that the morphology of N. valencienni is uniquely highly divergent. We thus infer that the remainder of the Jamaican anole radiation has remained more conservative in the structure of the BSA, and that they exhibit more subtle changes with respect to each other.
The degree of mobility of the scapulocoracoid varies with the magnitude of the angle between the sagittal plane and the coracosternal groove, it being greatest when the angle is low (Peterson, 1973; Russell and Bauer, 2008). Peterson (1973) reported a relatively narrow presternum in the crown-giant Anolis richardi, and a decrease in said coracosternal angle with species occupying perches of smaller diameter. We encountered the greatest angle in the crown-giant N. garmani (Figs. 5b, 10, and 13b), which contrasts with Peterson's findings of a low angle in the crown-giant A. richardi. However, in the Jamaican radiation N. garmani occupies the widest perches, and the other four species occupy successively narrower perches (the perch diameter decreases from N. opalinus to N. grahami, to N. lineatopus, to N. valencienni; Butler and Losos, 2002). Thus, the increase in the coracosternal angle is directly reflected in an increase of the average perch diameter from N. valencienni to N. garmani. The greater mobility of the scapulocoracoid in the sagittal plane allows for a greater range of motion of the arms, thus increasing stability of the gait, especially on narrow branches (Peterson, 1973). However, the great excursion of the shoulder also leads to a decrease in relative locomotor speed (Peterson, 1973; Higham et al., 2001).
The length of the posterior process of the interclavicle is relatively the shortest in the twig anole N. valencienni (Figs. 10 and 13b), and the anteroposteriorly longest interclavicle is found in N. grahami, N. opalinus and N. garmani. The relative length of the interclavicle correlates with the distance between the anterior extremity of the interclavicle and the presternum (Figs. 10 and 13b). The length of the interclavicle has played a prominent role in anoline systematics (Guyer and Savage, 1986, 1992), but little functional interpretation has been attempted. Peterson (1973) found a relatively short posterior process of the interclavicle in the trunk-ground form Anolis cybotes and the trunk anole A. distichus. This contrasts with our findings, in which the trunk-ground form (N. lineatopus) shows an interclavicle of intermediate length between the other species. Additional data are required to determine whether there are any ecomorphological patterns that are reflective of variations in the shape of the interclavicle.
Both the posterior process of the interclavicle and the mediolateral portion of the clavicle are relatively long in the trunk-crown species N. grahami and N. opalinus, and in the crown-giant N. garmani (Figs. 10 and 13). The relatively great distance between the anterior tips of the presternum and the interclavicle relates to a more tapered appearance of the anterior portion of the BSA, reflecting a more acute angle between the lateral processes and the posterior process of the interclavicle, and a lengthened ventromedial portion of the clavicle (Fig. 5). Higham et al. (2001) studied sprinting performance in Jamaican anoles, and found that N. grahami runs faster on a strongly inclined surface than does N. lineatopus, which, in turn, is faster than N. valencienni. This makes it plausible, although only suggestive at present, that the anteriorly lengthened and more tapered BSA of crown anoles directly relates to their ability to climb on vertical surfaces. One possible reason for such a relationship is a greater muscle mass of the M. clavodeltoideus and M. pectoralis. Both of these muscles originate from the lateral and posterior processes of the interclavicle, and play important roles during the swing phase of locomotion (Fürbringer, 1900; Jenkins and Goslow, 1983). Wataru et al. (2013) compared the extent of the appendicular muscles of various anoles, and found differences that relate to variation in habitat occupation and signalling behaviour.
The relative width of the ventromedial portion of the coracoid and epicoracoid increases from N. lineatopus and N. valencienni, to N. garmani and N. grahami, and is greatest in N. opalinus. The ventromedial portion of the clavicle lengthens following a similar sequence, except that N. valencienni exhibits relatively the shortest ventromedial portion of the five species of Norops examined. This is probably not reflective of a wider chest in N. garmani, N. grahami and N. opalinus, because the lateral processes of the interclavicle and the width of the presternum are both relatively narrow in N. garmani, N. opalinus and N. valencienni (Figs. 10 and 13b). Peterson (1973, 1974) hypothesized that a correlative relationship exists between the form of the clavicle and that of other elements of the shoulder girdle. We did not observe such a relationship in our data, and examination of additional anoline radiations is necessary to determine whether more consistent patterns of variation in the form of the elements of the BSA are evident among anoles.
The shape variation in the vertebral column and presternum-interclavicle appear to carry no phylogenetic signal (Table 16). Only the distribution of the mean shapes of the scapulocoracoid appears to be related to phylogenetic history, and the topology of Alföldi et al. (2011) yields a P value of 0.048. However, the phylogenetic hypothesis of Nicholson et al. (2012) yields a P value of 0.164 (Table 16). The relatively small difference in the P values indicates that the topology of the phylogenetic tree does not have much influence on the outcome of the analysis.
Previous authors have found morphological characters of the clavicle that are useful indicators of phylogenetic relationships at the genus or family level (Sukhanov, 1961; Lécuru, 1968b; Russell, 1988). We argue that the form of the clavicle is more malleable in its adaptability to microecological differences relating to habitat occupation and exploitation than is the form of the vertebrae, presternum-interclavicle, and scapulocoracoid moieties.
From our results it is difficult to determine whether the differentiation in shape of the components of the BSA is reflective of ecomorph designation. More investigations are needed to determine whether similar ecomorphs to those comprising the Jamaican radiation, but occupying different Caribbean islands, exhibit similar patterns of configuration of the BSA. By so doing it will be possible to further investigate whether ecomorphs are structurally recognizable according to skeletal patterns that reflect their occupancy of their particular sector of the physical resource space.
This exploratory study, and the literature sources that relate to anole ecology, provide a solid foundation for designing additional studies that may further our understanding of covarying morphological and ecological patterns, and the ability to place these into a functional-morphological context that assists in explaining processes of adaptive radiation of anoline island populations. Renous and Gasc (1977) outlined an integrated approach that would yield a holistic picture of the morphology and locomotor behaviour of a focal lizard taxon, and this was reiterated by Russell and Bels (2001) for designing a research program that aims at a comprehensive understanding of locomotor kinematics among lizards. Similar studies have been conducted on the forelimb of Tupinambis (Renous and Gasc, 1977), and on the limbs and girdles of Chameleo (Fischer et al., 2010), and we designate Anolis as a next logical step in this series of kinematic and locomotor studies.