THE ANATOMICAL RECORD 297:410–432 (2014)

Geometric Morphometric Analysis of the Breast-Shoulder Apparatus of Lizards: A Test Case Using Jamaican Anoles (Squamata: Dactyloidae) ALEXANDER TINIUS* AND ANTHONY PATRICK RUSSELL Department of Biological Sciences, University of Calgary, Calgary Alberta, Canada

ABSTRACT The breast-shoulder apparatus (BSA) is a structurally and kinematically complex region of lizards. Compared with the pelvic region it has received little attention, even though its morphological variation is known to be extensive. This variability has seldom been the focus of functional explanation, possibly because the BSA has been difficult to explore as a composite entity. In this study we apply geometric morphometric techniques to the analysis of the BSA in an attempt to more fully understand its configuration in relation to differential use in locomotion. Our approach centers upon the Jamaican radiation of anoline lizards (genus Norops) as a tractable, small monophyletic assemblage consisting of species representing several ecomorphs. We hypothesized that the different species and ecomorphs would exhibit variation in the configuration of the BSA. Our findings indicate that this is so, and is expressed in the component parts of the BSA, although it is subtle except for Norops valencienni (twig ecomorph), which differs greatly in morphology (and behavior) from its island congeners. We further found similarities in the BSA of N. grahami, N. opalinus (both trunk-crown ecomorphs), and N. garmani (crown giant). These outcomes are promising for associating morphology with ecomorphological specialization and for furthering our understanding of the adaptive response of the BSA to demands on the locomotor system. C 2014 Wiley Periodicals, Inc. Anat Rec, 297:410–432, 2014. V

Key words: Norops; Anolis; ecomorphs; skeleton; pectoral girdle; shape analysis; locomotion

INTRODUCTION Locomotor forces of terrestrial tetrapods are transmitted from the limbs to the axial skeleton via the limb girdles. Evolutionary transformations that took place in the transition from the aquatic to the terrestrial environment (Clack, 2002) resulted in a segregation of the pectoral girdle from a firm association with the axial skeleton via dermal bones, to one which is less direct and mediated by the ribs, the newly-emergent sternum, and a system of ligaments (Russell and Bauer, 2008). The reverse happened in the pelvic region, whereby the once skeletally isolated pelvic girdle attained direct connection with the vertebral column via the sacroiliac articulation. Kinematically, therefore, the shoulder-region of tetrapods is more complex than the hip region. C 2014 WILEY PERIODICALS, INC. V

Most studies of the limb kinematics of tetrapods have focused on the hindlimb (salamanders—Ashley-Ross, 1992; Ijspeert et al., 2005; crocodylians—Gatesy, 1991; Reilly and Elias, 1998; Hutchinson and Gatesy, 2000;

Grant sponsor: Natural Sciences and Engineering Research Council of Canada; Grant number: A9745 (to A.P.R.). *Correspondence to: Alexander Tinius, Department of Biological Sciences, University of Calgary, 2500 University Drive NW, Calgary Alberta, Canada T2N 1N4. E-mail: [email protected] Received 28 February 2013; Accepted 11 September 2013. DOI 10.1002/ar.22869 Published online 31 January 2014 in Wiley Online Library (wileyonlinelibrary.com).

MORPHOMETRICS OF THE BSA OF JAMAICAN ANOLES

411

Kubo and Ozaki, 2009; lizards—Reilly, 1995, 1998; Reilly and Delancey, 1997a,b; Irschick and Jayne, 1999; Kubo and Ozaki, 2009; chelonians—Butcher and Blob, 2008; Wyneken et al., 2008; birds—Santi, 1990; Gatesy and Middleton, 1997; Hutchinson, 2004; mammals—Channon et al., 2010). Kinematic investigations of the forelimb, especially those incorporating a consideration of the shoulder region, are less numerous, except for the cases of birds (Goslow et al., 1989; Poore et al., 1997) and brachiating primates (Eaton, 1944; Schmidt et al., 2002; Wright-Fitzgerald et al., 2010), in which the pectoral limbs generate all, or the majority of, the propulsive forces. Analyses of pectoral kinematics have been conducted for some lizards (Renous and Gasc, 1977; Jenkins and Goslow, 1983; Peterson, 1984) and mammals (Jenkins, 1971; Klima et al., 1980; Hermanson and Altenbach, 1983; H€ogfors et al., 1987), but remain almost unexplored for salamanders and crocodylians (Jenkins, 1993). As long ago as 1888 Max F€ urbringer recognized the fundamental differences that exist in the anatomical arrangement of the pectoral and pelvic region (Fig. 1), and set forth the concept of the breast-shoulder apparatus (BSA) to accommodate the mobile relationships and indirect linkages between the composite (dermal and replacement bone origins) skeleton of the shoulder girdle

and the axial skeleton. In its entirety, the BSA also includes the muscles and ligaments that respectively induce and control the displacement of the skeletal components that include the ribs, sternum and vertebral column. F€ urbringer’s (1888, 1900) considerations were almost purely descriptive. Subsequently other authors explored the BSA in terms of its functional attributes (Skinner, 1959; Renous and Gasc, 1977; Russell and Bauer 2008), but its kinematics, other than in birds and mammals (Goslow et al., 1989; Poore et al., 1997; Jenkins, 1971; Klima et al., 1980; Hermanson and Altenbach, 1983; H€ogfors et al., 1987), have remained largely uninvestigated. Relative displacements in this region are more difficult to study than are those in the pelvic region, and morphological variation between taxa can be quite considerable, although the functional implications of that variation remain largely unknown (see Russell and Bauer, 2008, for a review). The variation in the configuration of the BSA of lizards (F€ urbringer, 1900; Russell and Bauer, 2008) invites further functional investigation, but qualitative descriptions of the region provide limited means of interpreting the differences that are evident. Assessment of the integrative locomotor morphology and kinematics of the BSA will benefit from a more synthetic understanding of the

Fig. 1. Left and anterior views of the generalized morphology of the breast-shoulder apparatus of lizards, and a similar conceptualization of the pelvic region. The number and type of joints that are found between the skeletal elements allow for different degrees of mobility between the axial skeleton and the stylopodium, depending on limb position. In the hindlimb, the acetabulum is solely associated with the mobility possible between the axial skeleton and the stylopodium

(femur). In the forelimb the frame carrying the stylopodium (humerus) is able to move on the body wall anteroposteriorly and dorsoventrally, and also potentially mediolaterally in cases in which the orientation and shape of the coracosternal articulation permits this, with the coracoid translating along an obliquely-oriented articulation. (After €rbringer, 1900; Vickaryous and Hall, 2006) Fu

412

TINIUS AND RUSSELL

variability of form and configuration within and between its contributory elements. Here we describe the variability in the morphology of the BSA in situ, employing computed tomography (CT) and geometric morphometric techniques to examine its structural components. We confine this exploratory study to a small radiation of lizards (Fig. 2, Jamaican members of the genus Norops, part of the anoline radiation, Alf€oldi et al., 2011; Nicholson et al., 2012) whose ecology is well understood. The species examined are the trunk-ground Norops lineatopus; the twig-giant N. valencienni; the trunkcrown N. grahami; the crown giant N. garmani; and the trunk-crown dwarf N. opalinus. Although the Jamaican anoles examined in this study are very closely related (Alf€oldi et al., 2011; Nicholson et al., 2012), they differ in their geographic distribution (Barbour, 1922; Underwood and Williams, 1959; Schoener and Schoener, 1971; Crombie et al., 1984), microhabitat preference and use (Lazell, 1966; Rand, 1967; Schoener and Schoener, 1971; Williams, 1983; Bundy et al., 1987; Landwer et al., 1995; Langerhans et al., 2006; Butler, 2007; Singhal et al., 2007), frequency and speed of locomotor movements (Hicks and Trivers, 1983; Losos, 1990a,b,c), jumping performance (Losos, 1990a; Toro et al., 2004), competition with congeners (Jenssen, 1973; Schoener, 1975), sexual interactions (Lynn and Grant, 1940; Jenssen, 1977; Lailvaux and Irschick, 2007), expression of sexual dimorphism (Jenssen and Nunez, 1994; Butler et al., 2000; Butler and Losos, 2002; Losos et al., 2003; Butler, 2007), feeding habits (Floyd and Jenssen, 1983), and external morphology (Lazell, 1966; Irschick et al., 1997; Langerhans et al., 2006; Kolbe et al., 2011). Much of the morphological variation that is correlated with sexual dimorphism and/or ecomorph designation is related to relative lengths of the pectoral limbs and elements thereof (Losos, 1990c; Powell and Russell, 1991; Irschick et al., 1997; Toro et al., 2004; Langerhans et al., 2006; Kolbe et al., 2011). Studies exploring the morphological differences of skeletal elements of anoline ecomorphs are not abundant (Butler and Losos, 2002; Herrel et al., 2007b; Sanger et al., 2011), meaning that relationships, if any, between ecomorph designation and skeletal anatomy remain relatively unknown. Here we raise the question whether species identity or ecomorph designation are associable with skeletal features, and choose the BSA complex as a vehicle to explore that question. We do so because of the great variability shown by the BSA in lizards (Lecuru, 1968a,b; Russell and Bauer, 2008), and because the BSA is likely to be influenced by locomotor and social signalling adaptations that characterize anole species within

circumscribed radiations (Beuttell and Losos, 1999; Butler and Losos, 2002; Herrel et al., 2007a). Powell and Russell (1991) examined several external attributes of a subset of the Jamaican anole radiation, and found that the features measured (limb length, scansor area, and intergirdle distance, among others) differed among the ecomorphs investigated, and that these differences were evident throughout ontogeny. This indicated that dimensional differences among anoline ecomorphs are not simply a function of body size alone, but are characteristic of the species throughout their life spans (Losos, 1990c; Irschick et al., 1997; Losos, 2009; Kolbe et al., 2011). Here we examine skeletal features of five anole species representing four distinctive ecomorphs (trunk-ground, trunk-crown, crown giant, twig giant, Fig. 2) that exploit different parts of the locomotor resource space and that may be expected to exhibit distinctive, but subtle, differences in locomotor mechanics. We investigate whether these are reflected in the three-dimensional configuration of the BSA. Here we pose the following questions: 1. How much qualitatively and quantitatively observable variation exists in the morphology of the BSA of these Jamaican anoline species, and can that be expressed as observable differences between ecomorphs? 2. Can species-specific differences of this structural complex be correlated with habitat preferences?

MATERIALS AND METHODS We firstly provide qualitative descriptions of the BSA, and place these into a comparative framework by employing the detailed descriptions of the BSA of Iguana iguana (Iguanidae) provided by Russell and Bauer (2008). We compare this to the morphology of the BSA of Cophosaurus texanus (Phrynosomatidae). The form of the BSA in the latter is similar to that of I. iguana, whereas its absolute size is comparable to that of Norops (Fig. 3). It was, therefore, employed as a test for the resolution attainable via CT scanning for the skeletal features that have been established for the BSA of I. iguana. We then provide a comparative summary of the form of the elements of C. texanus and Norops lineatopus, the latter representing a relatively basal branch in the phylogeny of Jamaican anoles (Fig. 2), using scanned renditions of each. Employing N. lineatopus as a baseline, we then summarize qualitative variation in the morphology of the elements of the BSA of all Jamaican Norops examined. Finally, we present an integrative summary of the

Fig. 2. Phylogenetic relationships and ecomorph designations of the five Jamaican anole species. Phy€ ldi et al. (2011). logenetic hypothesis after Nicholson et al. (2012) and Alfo

MORPHOMETRICS OF THE BSA OF JAMAICAN ANOLES

Fig. 3. Boxplot showing the size range and number of specimens by sex, and ecomorph designation of each species examined in this study. Sexes could not be determined for all specimens of Cophosaurus texanus.

variation observed in the skeletal components of the BSA among Jamaican anoles. That summary then serves as the basis for comparison with the findings of our geometric morphometric analysis of the BSA of the same taxa (and the same specimens). The phylogeny of the Jamaican anoles is still disputed (Nicholson et al., 2012; Poe, 2013; Casta~ neda and de Queiroz, 2013), and we, therefore, assess our data in the light of two recent phylogenetic hypotheses (Fig. 2; Alf€oldi et al., 2011; Nicholson et al., 2012). Our qualitative and geometric morphometric comparison is based on the following taxa: Cophosaurus texanus (11 specimens), Norops lineatopus (31), N. valencienni (6), N. grahami (14), N. garmani (4), and N. opalinus (12). C. texanus was not included in the geometric morphometric analyses, and serves only as a qualitative baseline for the CT renditions. The description of I. iguana (Russell and Bauer, 2008) is used here in lieu of a generalized lizard to provide a qualitatively comparative framework. C. texanus is similar in size to Norops (Fig. 3), and the respective specimens can be subjected to the same radiographic scanning procedures as the latter. However, its BSA is structurally very similar to that of I. iguana, and we therefore use C. texanus to relate the thorough description of I. iguana to our analysis of Jamaican Norops. To obtain three-dimensional renditions of the BSA of each species of Norops examined, and of Cophosaurus texanus, each specimen was scanned using a Skyscan microCT. All skeletal elements of the BSA were voluR 6.3.1 (VSG, Visualization Scienmized using vsg AvizoV ces Group, Burlington, MA). The resulting renditions (Bruker, Billerica, MA) were used for the qualitative descriptions of all skeletal elements of the BSA, and the configured renditions were exported as obj-files. Since every scan was performed on an intact specimen, the renditions represent the in situ configuration of the skeletal components of the BSA. R All bone models were imported into Autodesk MayaV (Autodesk GmbH, Munich, Germany), within which each element can be moved independently of all others. To obtain a uniform position of elements of the BSA we

413

employed the standardized pectoral limb configuration set out by Russell and Bauer (2008). In order to obtain that position the scapulocoracoid was shifted in the scapulocoracoid groove of the presternum in thirty-three of the 67 Norops specimens. Forty-three three-dimensional landmarks were then placed on the rendition of the BSA of each specimen, as listed in Table 1. However, a realignment as outlined above cannot guarantee unambiguous standardization of all skeletal elements, especially those related to the highly deformable rib cage. The landmark data were, thus, split into four subsets for further analysis: the vertebrae; presternum-interclavicle moiety; clavicle; and scapulocoracoid (Table 2). R ExcelV R Landmarks were assembled in a MicrosoftV worksheet, exported as txt-files, and analyzed using MorphoJ 1.02e (Klingenberg, 2011). All landmark sets were standardized using Procrustes superimposition (aligned by principal axis) and the covariance matrix was calculated directly from the Procrustes coordinates. Our principal component analyses (PCA) employ the covariance matrix. Using PAST 2.17b, a broken-stick model was applied to the PCs of each data set to determine the number of PCs to be analyzed. The relevant PC scores for all data subsets were pooled, together with centroid size, logarithmic centroid size and snout-vent length (SVL). In order to test how well log-transformed centroid size (logCS) approximates body size, we employed “SPSS” 9.0.1 (IBM Corporation, Armonk, NY) to calculate Pearson’s correlation coefficient with two-sided significance between logCS and logSVL. In order to test whether the data covary with size, we then computed a regression of the Procrustes raw coordinates against log-transformed centroid size using MorphoJ. Using MorphoJ we performed a discriminant function analysis (DF) of the Procrustes coordinates to visualize the shape changes (in all four data sets) between (i) male and female individuals of N. lineatopus, and (ii) between N. lineatopus and every other species of Norops examined. Leave-one-out cross validation was performed via MorphoJ to test for discrimination between male and female representatives of N. lineatopus. A canonical variate analysis (CVA) of the principal components was performed using “SPSS” 9.0.1, and was used to assess the distinctiveness of groups as they are defined by (i) species and (ii) ecomorph. The sample sizes per species and ecomorph correspond to the numbers provided in Fig. 3. PAST 2.17b was used to visualize the shape changes from N. lineatopus to the target Norops species as thin plate spline transformation grids (Hammer et al., 2001). Employing the phylogenetic hypotheses of Alf€oldi et al. (2011) and Nicholson et al. (2012), we performed a permutation test against the null hypothesis of no phylogenetic signal in the principal component data via MorphoJ. This calculates an average shape for every species, and using 10,000 permutations reassigns each species to a different terminal node of the phylogeny, thus testing the hypothesis that the distribution of these average shapes in morphospace arose in correlation with the phylogenetic relationship between these species (Klingenberg and Gidaszewski, 2010).

RESULTS Qualitative Comparison Vertebral column and ribcage. Iguana iguana and Cophosaurus texanus. Both species exhibit the

414

TINIUS AND RUSSELL

TABLE 1. Location and type of landmarks used in this study Landmark # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43

Location

Type

Posterodorsal edge of neural canal of the first thoracic vertebra Posterodorsal edge of neural canal of the third thoracic vertebra Posterodorsal edge of neural canal of the fifth thoracic vertebra Posterodorsal edge of neural canal of the sixth cervical vertebra Posterodorsal edge of neural canal of the fourth cervical vertebra Articulatory point between the vertebral column and the dorsal extremity of the first sternal rib Juncture between the dorsal and ventral portion of the first sternal rib Articulatory point between the first sternal rib and the sternum Articulatory point between the vertebral column and the dorsal extremity of the second sternal rib Juncture between the dorsal and ventral portion of the second sternal rib Articulatory point between the second sternal rib and the sternum Articulatory point between the vertebral column and the dorsal extremity of the third sternal rib Juncture between the dorsal and ventral portion of the third sternal rib Articulatory point between the third sternal rib and the sternum Posteromedial extremity of the presternum Anteromedial extremity of the presternum Lateral extremity of the presternum at the posterior edge of the dorsal lip of the coracosternal groove Posterior-most contact point between the sternum and the interclavicle Posterior-most extremity of the epicoracoid Medial extremity of the epicoracoid at the medial contact with the dorsal lip of the coracosternal groove Anteromedial extremity of the epicoracoid Dorsal anterior extremity of the first coracoid ray at its point of contact with the epicoracoid Dorsal anterior extremity of the second coracoid ray at its point of contact with the epicoracoid Medial extremity of the coracoid at its point of contact with the epicoracoid Posterior extremity of the coracoid at its point of contact with the epicoracoid Posterior extremity of the coracoid foramen Ventral extremity of the inferior glenoid buttress Anterior extremity of the glenoid fossa at the point of contact between the coracoid and scapula Dorsal extremity of the superior glenoid buttress Ventroanterior extremity of the scapular ray at its point of contact with the epicoracoid Anterior-most point of contact between the suprascapula and the scapula Posterior-most point of contact between the suprascapula and the scapula Dorsoposterior-most contact between the clavicle and the scapulocoracoid of the acromion region Anterior-most extremity of the suprascapula Dorsomedial extremity of the suprascapula Most posterior extremity of the suprascapula Lateral extremity of the lateral process of the interclavicle Anteroventral-most extremity of the interclavicle Posterior-most extremity of the articulation between the interclavicle and clavicle Position of the lateral apex of the primary curvature of the clavicle Posterior extremity of the primary coracoid fenestra Ventroposterior extremity of the scapulocoracoid fenestra Dorsal extremity of the dorsolateral process of the presternum

II II II II II II II II II II II II II II II II II I II II II I I II II II II II II II I I II II II II II II I III III III II

All landmarks of bilaterally symmetrical elements were placed on the left side.

TABLE 2. Contribution of the various landmarks examined in this study to specific data subsets analyzed Landmarks 1–6, 9, 12 8, 11, 14–18, 37, 38, 43 33, 39, 40 19–36, 41, 42

Data set Vetebrae Presternum-interclavicle moiety Clavicle Scapulocoracoid

Total no. landmarks 8 10 3 19

same gross vertebral structure. Posterior to the atlas and axis are five additional cervical vertebrae. Vertebrae 6 and 7 carry cervical ribs that wrap around the body

cavity, but terminate freely some distance from the presternal plate. The eighth vertebra in the complete series is the first of the thoracics (Fig. 5a), and there are five thoracic vertebrae in total. The articulatory facet for the first sternal rib is borne laterally on the anterior third of the vertebral centrum. The first three thoracic ribs connect with the presternum, whereas the next two ribs in the anteroposterior series connect with the mesosternum. Cophosaurus texanus and Norops lineatopus. The vertebral column of N. lineatopus is qualitatively indistinguishable from that of C. texanus (Fig. 5a). Variation among Jamaican Norops. The vertebral column of N. lineatopus is qualitatively indistinguishable from that of all other species of Norops examined (Fig. 5a).

MORPHOMETRICS OF THE BSA OF JAMAICAN ANOLES

415

Fig. 4. CT reconstruction of the breast-shoulder apparatus of Norops lineatopus. (a) Ventrolateral view with the scapulocoracoid in unadjusted (unrotated) position. (b–d) Configuration following rotation of the scapulocoracoid into the standardized position (see text for details) and showing the position of the 43 landmarks (Table 1) used in the geometric morphometric analyses. (b) Ventrolateral view, (c) anterior view, (d) dorsolateral view.

Presternum. Iguana iguana and Cophosaurus texanus. The presternum is a depressed, rhomboidal structure. The posterior half of the presternal plate bears articulatory facets for three sternal ribs. The demarcation between the presternum and mesosternum is clear cut. The coracoid articulates with the coracosternal groove that spans the anterolateral face of the presternum (Figs. 4 and 6). In C. texanus the dorsal lip of this sulcus does not extend to the anterior extremity of the presternal plate, although it does in I. iguana. The presternum of C. texanus bears a short but prominent lateral process that extends from the posterior extremity of the dorsal lip of the sulcus. I. iguana lacks such a process. In C. texanus there is a single large fontanelle in the posterior half of the presternal plate (that in some specimens is paired and divided in the midline, Fig. 5b). I. iguana lacks this fontanelle. Siebenrock (1895) and Skinner (1959) both noted that the extent of the fontanelle can vary between different developmental stages. Both the existence and the extent of this fontanelle probably reflect limitations of the imaging technique rather than actual skeletal features, and are, therefore not examined here. Cophosaurus texanus and Norops lineatopus. In Norops lineatopus the dorsal lip of the sulcus spans the entire anterior half of the presternal plate, as is also seen I. iguana, but not in C. texanus (Fig. 5b). The presternum of N. lineatopus is narrower and relatively longer than that of C. texanus (Fig. 5b). The posterior

extremity of the presternal plate carries a shallow, anteriorly-directed indentation, to either side of which attach the mesosternal rods. The latter, in many instances, are fused together in their anterior portions, carrying the above-mentioned indentation further posteriorly. This results in a gradual, rather than an abrupt, transition from presternum to mesosternum (Fig. 5b). Variation among Jamaican Norops. In comparison to that of the other species of Norops investigated, the presternum of N. lineatopus is most similar to that of N. grahami in overall form and configuration. In contrast, the presternum of the other three species appears more attenuated (Fig. 5b). The paired dorsolateral processes of the dorsal sulcus are, in a relative sense, about three times as long in N. valencienni as they are in the other species (Fig. 6). In N. valencienni the third sternal rib uniquely articulates with the mesosternum (Fig. 5b).

Interclavicle. Iguana iguana and Cophosaurus texanus. The interclavicle is T-shaped (Fig. 5c). In C. texanus it is strongly depressed, whereas in I. iguana it is cylindrical, and only its posterior process is depressed. The interclavicle makes contact with the presternum via a wide longitudinal groove in the ventral face of the latter, which occupies the anterior half of the presternum. The lateral processes are very short (about one sixth of the total width of the BSA) in I. iguana, whereas in C. texanus they span almost one half of the total width of the BSA. The anteroventral edge of the interclavicle

416

TINIUS AND RUSSELL

Fig. 5. CT reconstructions of the components of the breast-shoulder apparatus of Cophosaurus texanus (far left column) and all five species of Norops examined in this study. Each species of Norops examined is represented by a column of reconstructions, in the sequence N. lineatopus, N. valencienni, N. grahami, N. garmani, and N. opalinus. Ele-

ments are represented as follows: (a) vertebral column in lateral view, (b) presternum in dorsal view, (c) interclavicle in dorsal view, (d) clavicle in anterolateral view, (e) scapulocoracoid in lateral view, (f) scapulocoracoid in posterior view.

forms a short, shovel-like process (Fig. 6), which is relatively longer in C. texanus than it is in I. iguana. The lateral processes meet the posterior process at an angle of about 85 in both species.

directly with this groove. The very short anterior-medial process is cylindrical and forms a proboscis-like extension. The equivalent process in I. iguana and C. texanus is strongly depressed (Fig. 6). Variation among Jamaican Norops. In its general form the interclavicle is very similar in all examined Norops species (Fig. 5c). The angle between the lateral processes and the posterior process varies from 45 to 65 in N. opalinus, 55 to 60 in N. grahami and N. garmani, to 55 to 65 in N. valencienni (Fig. 5c).

Cophosaurus

texanus

and

Norops

lineatopus.

Although generally cruciform, the very short anterior process of the interclavicle gives this element a T-shaped appearance in N. lineatopus (Fig. 5c). The body of the interclavicle is depressed. The lateral processes form an angle of about 55 to 60 with the posterior process, which is considerably less than that in C. texanus. A longitudinal groove in the lateral process divides its medial half into dorsally and posteriorly oriented components, a separation that is absent from I. iguana and C. texanus (Fig. 5c). The medial third of the clavicle articulates

Clavicle. Iguana iguana and Cophosaurus texanus. In both species the shaft of the clavicle is cylindrical and the primary curvature is very smooth, the

MORPHOMETRICS OF THE BSA OF JAMAICAN ANOLES

417

Fig. 6. CT reconstruction of the left scapulocoracoid of Cophosaurus texanus in (a) lateral and (b) ventral view, showing the anatomical features of this element (nomenclature after Russell and Bauer, 2008). (c) Interclavicle of Cophosaurus texanus and (d) Norops lineatopus in

anterolateral view, indicating the morphological difference in the form of the anterior process. (e) Presternum of Norops valencienni in anterolateral view showing the coracosternal groove and the long dorsolateral processes that are characteristic of N. valencienni.

element forming an almost halfmoon-shape in anterior view (Fig. 5d). In comparison to C. texanus, the clavicular shaft is relatively thicker in I. iguana. In C. texanus a short portion (10 to 20% of the total length of the clavicle) of the dorsal quarter of the shaft is anteroposteriorly compressed (Fig. 5d), a feature that is barely noticeable in I. iguana. The dorsal articulatory surface of the clavicle tapers into a mediolaterally compressed process that forms a shallow articulatory groove. Cophosaurus texanus and Norops lineatopus. In N. lineatopus the shaft of the clavicle is lateromedially compressed, and the apex of the primary curvature segregates it into dorsal and medial portions (Fig. 5d). Both of these are almost straight, and at the midpoint of the shaft both meet at an angle of about 120 . The shaft is asymmetrically flattened (it is thickest dorsomedially and strongly compressed ventrolaterally), contrasting with the smooth curvature of the cylindrical shaft of C. texanus and I. iguana (Fig. 5d). The surfaces of the acromio-clavicular joint are shaped similarly in N. lineatopus, C. texanus and I. iguana. Variation among Jamaican Norops. In N. lineatopus the medioventral and dorsal portions of the shaft constitute about three fifths and two fifths of the total length of the clavicle, respectively (Fig. 5d). The same is so for N. garmani. In N. opalinus the ventral aspect of the clavicle is relatively longer, and the apex of the primary curvature is displaced farther towards the dorsal extremity of the clavicle. The form of the clavicle in N. grahami is intermediate between that of N. lineatopus and N. opalinus. In N. valencienni the dorsal and medio-

ventral portions each constitute about half the length of the clavicle (Fig. 5d).

Scapulocoracoid. Iguana iguana and Cophosaurus texanus. The scapulocoracoid of lizards forms a single functional and structural unit. However, it is divisible into four distinct topographical elements: the suprascapula, scapula, coracoid and epicoracoid (Fig. 6a,b). With its four fenestrae the scapulocoracoid of I. iguana is assignable to type 6 of Lecuru’s (1968a) scheme, whereas C. texanus lacks the secondary coracoid fenestra (Fig. 6b), and is assignable to type 5. In I. iguana the secondary coracoid ray lies almost in the frontal plane, and is only slightly angled in an anteroventral direction. Medioposterior to that ray lies the circular secondary coracoid fenestra, which is absent from C. texanus. In I. iguana the epicoracoid borders about one quarter to one half of the total circumference of each scapulocoracoid fenestra, whereas in C. texanus it borders only about 10% of their total circumference (Fig. 6b), resulting in the epicoracoid being relatively more extensive in I. iguana. Cophosaurus texanus and Norops lineatopus. The free edges of the suprascapula and epicoracoid are not precisely discernible in CT images, because the density contrast between these two elements and the surrounding soft tissue varies greatly in C. texanus (and also in Norops). Thus, the anterior border of the scapulocoracoid fenestra is unresolved in most of our reconstructions. The scapulocoracoid of N. lineatopus is assignable to

418

TINIUS AND RUSSELL

type 3 of Lecuru (1968a), with only the scapulocoracoid and primary coracoid fenestrae present (Fig. 5e). The position and orientation of the fenestrae that are present are almost identical to said openings in I. iguana and C. texanus (Fig. 5e). However, the scapular ray is oriented directly anteriorly in N. lineatopus, although it sometimes has a small anteroventral inclination. The rays are relatively narrower and the primary coracoid fenestra and the scapulocoracoid fenestra are relatively wider and longer in N. lineatopus in comparison to C. texanus (Fig. 5e). Medially the epicoracoid constitutes a bandlike continuation of the coracoid in N. lineatopus. It articulates with the presternum for about three fifths of its anteroposterior extent, whereas its anterior portion lies dorsal to the interclavicle (Fig. 4). Variation among Jamaican Norops. In N. valencienni the anteroposterior extent of the scapula, and, to a lesser degree, that of the suprascapula, is markedly shorter than it is in N. lineatopus (Fig. 5e). The scapulocoracoid of N. grahami and N. garmani are qualitatively indistinguishable from that of N. lineatopus (Fig. 5e,f). The form of the curvature between the lateral and ventromedial portion of the coracoid, as well as the extent of the scapula and suprascapula, vary equally both within and between these species (Fig. 5f). In comparison with N. lineatopus the reconstructed dorsal edge of the suprascapula is strongly emarginated in both C. texanus and N. opalinus (Fig. 5e). In the latter, the true extent of the suprascapula reaches further dorsally than is evident in the reconstruction based on CT images.

Morphological Variation Within Norops The vertebral column of N. lineatopus is qualitatively indistinguishable from that of all other species of Norops examined (Fig. 5a). The first three sternal ribs articulate with the presternal plate, and the following two ribs contact the mesosternum. This pattern is deviated from only by N. valencienni, in which the articulatory facet for the third sternal rib is borne by the mesosternum (Fig. 5b). The rhomboidal presternal plate is lateromedially narrow in N. valencienni, relatively wider in N. garmani and N. opalinus, and widest in N. lineatopus and N. grahami (Fig. 5b). The presternum bears a pair of dorsolateral processes that are relatively elongated in N. valencienni when compared with the other Norops species (Fig. 6).

The interclavicle is very similar in all examined Norops species. The angle between the lateral processes and the posterior process increases from N. opalinus, to N. lineatopus, N. grahami, and N. garmani, and then to N. valencienni (Fig. 5c). The clavicle is divisible into a ventromedial and a dorsal portion at the apex of its primary curvature. The length of the ventromedial portion increases, proportionally to the dorsal portion, from N. valencienni to N. garmani and N. lineatopus, to N. grahami, and is greatest in N. opalinus (Fig. 5d). The scapulocoracoid is a complex three-dimensional structure (Fig. 5e,f), which makes it difficult to qualitatively assess its morphological variation within and between species. In N. valencienni the anteroposterior extent of the scapula, and, to a lesser degree that of the suprascapula, is markedly shorter than that of the other species of Norops examined (Fig. 5e). The dorsal extent of the suprascapula of N. opalinus is relatively short in comparison to that of the other species of Norops, which is probably reflective of low density contrast between the material of the suprascapula and the surrounding soft tissue in this, the smallest species examined.

Geometric Morphometric Quantitative Comparison We divided the BSA into four moieties (Table 2), the components of which are generally stably and immovably linked to each other. The vertebral column poses the only exception in that regard, since we could not correct for displacement at the intervertebral joints. When examining landmarks 1 through 5, we thus focus on anteroposterior disposition. The eigenvalues and variance explained by the PCs of the dataset for each of the moieties are provided in Table 3. Only the PCs that are deemed informative after application of the broken-stick model are shown and discussed.

Testing for size-dependence of the principal components The Pearson-correlation between log-transformed snout-vent length and centroid size (logCS) indicates that the latter closely models the former (Table 4). We, therefore, used only logCS to test for size-dependence of the shape data. All PCs that are accounted for in Table 3 were regressed against logCS. Doing so revealed only a very

TABLE 3. Eigenvalues and variance of the explored PC scores for each data set Data set Complete BSA Vertebrae Presternuminterclavicle Clavicle Scapulocoracoid

PC1

PC2

PC3

PC4

PC5

PC6

0.00181233 25.118 0.00118661 44.407 0.00183076 27.820 0.00136259 82.725 0.00439470 30.116

0.00091543 12.687 0.00047770 17.877 0.00148110 22.506 0.00063359 19.025 0.00204665 14.026

0.00059948 8.308 0.00039550 14.801 0.00069534 10.566

0.00053199 7.373 0.00019832 7.422 0.00053999 8.205

0.00044647 6.188

0.00036288 5.029

0.00137751 9.440

0.00104479 7.160

0.00095248 6.527

0.00073888 5.063

Eigenvalue Variance [%] Eigenvalue Variance [%] Eigenvalue Variance [%] Eigenvalue Variance [%] Eigenvalue Variance [%]

We show only PCs that are informative according to the broken-stick model (broken stick calculated using PAST, V2.17, Hammer, 2012).

419

MORPHOMETRICS OF THE BSA OF JAMAICAN ANOLES

TABLE 4. Correlation coefficients of a two-sided Pearson correlation of log-transformed SVL with logtransformed centroid size Test group

Vertebrae

r P

0.874