Patterns of postnatal ontogeny of the skull and lower jaw of snakes as revealed by micro‐CT scan data and three‐dimensional geometric morphometrics

Abstract

We compared the head skeleton (skull and lower jaw) of juvenile and adult specimens of five snake species [Anilios (=Ramphotyphlops) bicolor, Cylindrophis ruffus, Aspidites melanocephalus, Acrochordus arafurae, and Notechis scutatus] and two lizard outgroups (Ctenophorus decresii, Varanus gilleni). All major ontogenetic changes observed were documented both qualitatively and quantitatively. Qualitative comparisons were based on high‐resolution micro‐CT scanning of the specimens, and detailed quantitative analyses were performed using three‐dimensional geometric morphometrics. Two sets of landmarks were used, one for accurate representation of the intraspecific transformations of each skull and jaw configuration, and the other for comparison between taxa. Our results document the ontogenetic elaboration of crests and processes for muscle attachment (especially for cervical and adductor muscles); negative allometry in the braincase of all taxa; approximately isometric growth of the snout of all taxa except Varanus and Anilios (positively allometric); and positive allometry in the quadrates of the macrostomatan snakes Aspidites, Acrochordus and Notechis, but also, surprisingly, in the iguanian lizard Ctenophorus. Ontogenetic trajectories from principal component analysis provide evidence for paedomorphosis in Anilios and peramorphosis in Acrochordus. Some primitive (lizard‐like) features are described for the first time in the juvenile Cylindrophis. Two distinct developmental trajectories for the achievement of the macrostomatan (large‐gaped) condition in adult snakes are documented, driven either by positive allometry of supratemporal and quadrate (in pythons), or of quadrate alone (in sampled caenophidians); this is consistent with hypothesised homoplasy in this adaptive complex. Certain traits (e.g. shape of coronoid process, marginal tooth counts) are more stable throughout postnatal ontogeny than others (e.g. basisphenoid keel), with implications for their reliability as phylogenetic characters.

Introduction

Relatively few studies have focused on the postnatal development of the cranial skeleton of snakes (e.g. Rossman, 1980; Young, 1989; Monteiro, 1998; Scanferla & Bhullar, 2014), though these patterns are well‐documented in lizards (e.g. Maisano, 2001, 2002a,b,c; Bell et al. 2003; Torres‐Carvajal, 2003; Tarazona et al. 2008). The prenatal head skeleton development of squamate reptiles (lizards and snakes) has also been extensively studied (e.g. Brock, 1941; Kamal & Hammouda, 1965; Kamal & Abdeen, 1972; Bellairs & Kamal, 1981; Haluska & Alberch, 1983; Rieppel, 1994; Jackson, 2002; Boughner et al. 2007; Hugi et al. 2010; Boback et al. 2012; Hernández‐Jaimes et al. 2012; Roscito & Rodrigues, 2012a,b; Polachowski & Werneburg, 2013; Khannoon & Evans, 2015; Kovtum & Sheverdyukova, 2015).

Although these studies have generated observational data on skeletal ontogeny of selected squamates, few have aimed to discover major shared patterns of development across major clades. The aim of this study is to help identify major ontogenetic patterns across groups of squamate reptiles during postnatal growth, taking advantage of relatively new and precise quantitative analytical methods (i.e. digital tools for three‐dimensional geometric morphometrics; Zelditch et al. 2012).

The main focus of this work is to compare the patterns of morphological variation that occur between the earliest postnatal developmental stages and adult individuals in a series of snakes that are representative of the major living groups (e.g. Hsiang et al. 2015): Scolecophidia, basal Alethinophidia, basal Macrostomata, basal Caenophidia and Colubroidea. We have used an exemplar species from each of these groups to begin the process of documenting the major patterns of postembryonic ontogeny in the cranial skeleton of snakes, and to offer a starting point for future studies. To place variation within snakes in a broader evolutionary perspective, species from each of the two closest lizard outgroups to snakes are analysed for comparison: Iguania and Anguimorpha (Reeder et al. 2015; Zheng & Wiens, 2016).

The present study aims to provide: (i) a broad overview of the general growth patterns observed in snakes, in particular their ontogenetic allometry, or shape change correlated with growth (Gould, 1966); (ii) a detailed description of the morphological changes involved in ontogeny for all the taxa examined; a description that is not only qualitative, but also quantitative through the use of three‐dimensional (3D) geometric morphometrics (Bookstein, 1996; Zelditch et al. 2012); (iii) a comparison of the ontogenetic trajectories in shape space for the selected taxa (Zelditch et al. 2012); and (iv) a report of morphological differences that can be attributed to ontogeny rather than to evolutionary divergence, data that can be useful to discriminate between ontogenetic and taxonomic differences, in both living and fossil taxa.

Moreover, snake head skeletons are extremely specialised structures defined by a high degree of kinesis and enclosure of the braincase (Cundall & Irish, 2008); hence one of the main questions to be addressed in this study is whether lizards and snakes, despite their great morphological divergence, still share common growth patterns in their skulls and lower jaws, or if there is evidence for totally different ontogenetic trajectories. We are aware that our limited sampling cannot provide a definitive answer to this question, but we see this comparison as a vital first exploratory step.

The terminology used herein follows Romer (1956) and Hildebrand (1982), who use the term ‘skull’ to refer to the head skeleton minus the lower jaw and hyoid apparatus, and the term ‘cranium’ is used to refer to the whole head skeleton.

Material and methods

Micro‐CT scan data of a young juvenile (i.e. as close to the neonate stage as available material allowed) and an adult individual for each taxon studied were obtained from specimens in the herpetological collections of the South Australian Museum, Adelaide, South Australia (SAMA), the Queensland Museum, Brisbane, Queensland (QM), the Western Australia Museum, Perth, Western Australia (WAM), and the Field Museum of Natural History, Chicago, US (FMNH). A total of 14 specimens were analysed: Ctenophorus decresii (Agamidae) SAMA R53670 (juvenile), SAMA R28618 (adult male); Varanus gilleni (Varanidae) SAMA R18223 (juvenile), SAMA R32164 (adult); Anilios bicolor (Scolecophidia: formerly Ramphotyphlops; Hedges et al. 2014; not to be confused with the pipe snake, Anilius) SAMA R62252 (juvenile), SAMA R60626 (adult female); Cylindrophis ruffus (Cylindrophiidae) WAM R49553 (juvenile), WAM R121384 (adult); Aspidites ramsayi (Pythonidae) SAMA R40987 (juvenile), SAMA R49367 (adult); Acrochordus arafurae (Acrochordidae) QM J11033 (juvenile female), SAMA R4305 (adult female); and Notechis scutatus (Elapidae) SAMA R48107 (juvenile), SAMA R29514 (adult).

Most species studied do not show obvious sexual dimorphism, except for Ctenophorus, where adult males may have larger heads than females (Lebas, 2001), and Anilios bicolor and Acrochordus arafurae, where females are known to grow to a considerably larger body size than males (Shine, 1986; Rabovsky et al. 2016). Some dry skeletal specimens of various species of snakes were also examined for comparative purposes, in particular to verify the generality of some observations; these specimens belong to the collections of the South Australian Museum (SAMA), the American Museum of Natural History in New York (AMNH), the Museum of Comparative Zoology in Cambridge (MCZ), the Field Museum of Natural History in Chicago (FMNH), and the Zoologisches Forschungsmuseum Alexander Koenig in Bonn (ZFMK).

As the present study is mainly focused on identification and quantification of the major morphological changes that occur during the postnatal ontogeny of the cranium, the age gap was chosen to be as wide as possible (i.e. as permitted by the available specimens and the size limits of the micro‐CT scanner). Because postnatal ontogeny is typically a much greater source of intraspecific variation than individual or geographic variability across individuals of similar age (e.g. Barahona & Barbadillo, 1998), we did not sample multiple individuals at each ontogenetic stage. We investigated morphological variability across closely related species in Cylindrophis using the techniques below and found only minor differences between adults compared with juveniles and adults (C. ruffus WAM R121384, C. mirzae SAMA R12956, and C. cf. aruensis FMNH 60958).

All specimens except C. cf. aruensis FMNH 60958 (CT‐scan data courtesy of M. Kearney and O. Rieppel) were micro‐CT scanned using a Skyscan 1076 scanner (Bruker microCT) at Adelaide Microscopy (University of Adelaide, South Australia). Most specimens were scanned at a resolution of 17.30 μm, with the exceptions of A. bicolor and the juveniles of C. ruffus and V. gilleni, which were scanned at a higher resolution (8.65 μm). A 0.5‐mm aluminium filter was used to reduce scattering artefacts. Voltage and amperage were adjusted depending on the specimen, in the range 33–74 kv and 100–169 μA, respectively.

The CT‐scan data (raw X‐ray images) were processed using the software nrecon (Bruker microCT), which produced stacks of images (.bmp) that could then be visualised in avizo v.9.0 (Konrad‐Zuse‐Zentrum für Informationstechnik Berlin and Visualization Sciences Group) as a 3D digital reconstruction. Each dataset was then segmented in avizo v.9.0 to separate the skull from the lower jaw. This was done because of the extreme mobility of snakes jaws (which lack a bony symphysis), which prevents their study together with the skull. Other bones in snakes and lizards are also known to be somewhat mobile (Frazzetta, 1962, 1966), but to a much lesser degree, and in specimens at rest the differences are barely noticeable. Surface models of skulls and jaws were then exported from avizo v.9.0 as .ply files. The .ply files were then imported into the software landmark editor v.3.0 (Wiley et al. 2007), where a selection of landmarks was applied.

We used two main sets of landmarks, one to compare in great detail the geometric transformations between juveniles and adults of the same species (‘species‐specific landmarks’, SSL) and the other to compare the ontogenetic trajectories of all species in multivariate analyses of shape (‘overall comparison landmarking scheme’, OCL) (Zelditch et al. 2012).

With regard to the first type of analysis (SSL), 122 landmarks common to most snakes were selected. Because of their widely different cranial morphologies, Ctenophorus and Varanus each had their own set of landmarks (166 for Ctenophorus and 176 for Varanus). Graphical representations of these landmarks are provided in Figs 1 and 2; for a description of all landmarks and notes about taxon‐specific variation see Supporting Information Appendix S1.

Figure 1.

Diagrammatic representations of the skull and lower jaw of a snake showing bone abbreviations and selected landmarks. (A,F) dorsal view of the skull; (B,G) ventral view of the skull; (C,I) left lateral view of braincase; (D,J) right lower jaw in medial view; (E,K) right lower jaw in lateral view; (H), right lateral view of braincase; (L) right lower jaw in dorsal view. an, angular; bo, basioccipital; c, coronoid; co, compound bone; d, dentary; e, ectopterygoid; f, frontal; j, jugal; m, maxilla; n, nasal; ot, otoccipital; p, palatine; pa, parietal; pb, parabasisphenoid; pf, prefrontal; pfr, postfrontal; pm, premaxilla; pr, prootic; pt, pterygoid; q, quadrate; s, splenial; so, supraoccipital; st, supratemporal; v, vomer.

Figure 2.

Skull and lower jaw of the lizards Ctenophorus decresii and Varanus gilleni showing bone abbreviations and selected landmarks. Images not to scale, anterior to the left. (A) Skull of Ctenophorus in dorsal view; (B) skull of Ctenophorus in ventral view; (C) skull of Ctenophorus in left lateral view; (D) close‐up of otic region of Ctenophorus in left posterolateral view; (E) lower jaw of Ctenophorus in lateral view; (F) lower jaw of Ctenophorus in medial view; (G) skull of Varanus in dorsal view; (H) skull of Varanus in ventral view; (I) skull of Varanus in left lateral view; (J) close‐up of otic region of Varanus in left posterolateral view; (K) lower jaw of Varanus in lateral view; (L) lower jaw of Varanus in medial view. The slash (/) symbol separates landmarks from the right and left side, respectively (e.g. ‘97/8’ stands for landmarks 97, from the right side of the body, and 98, from the left side of the body). Abbreviations as in Fig. 1, plus: a, articular‐prearticular; ep, epipterygoid; os, orbitosphenoid; pl, palpebral; po, postorbital; pof, postorbitofrontal; sa, surangular; sq, squamosal.

The second set of landmarks (OCL), common to all the taxa included in this study, consisted of a selection of 50 landmarks in the skull and nine in the lower jaw (mostly a subset of the landmarks used for snakes in the first scheme). The list and description of these landmarks are provided in the Supporting Information Appendix S2, and a graphic representation of their location on the average skull and jaw configurations of the selected taxa is provided in Fig. 3.

Figure 3.

Wireframe diagrams showing landmark distribution on the average configuration of the symmetric component of the skull and lower jaw of the sampled taxa, anterior to the left. (A) Representation of the skull as if in dorsal view; (B) representation of the skull as if in ventral view; (C) representation of the skull as if in left lateral view; (D) representation of the lower jaw as if in lateral view; (E) representation of the lower jaw as if in dorsoventral view. Slash symbol (/) separates right and left side landmarks, respectively (e.g. ‘35/6’ stands for landmarks 35, from the right side of the body, and 36, from the left side of the body).

The size of skulls (head length, HL), lower jaws (jaw length, JL; average between left and right side), and quadrate bones (quadrate length, QL; average between left and right side) was measured in avizo v.9.0 using the 3D length tool to quantify the magnitude of size change between the two age categories. The measurement for the skull was taken between the anterior tip of the snout (premaxilla) and the occipital condyle; the jaws were measured between the anterior tip of the dentary and the posterior end of the retroarticular process; and the quadrates were measured from the lateral end of the ventral condyle to the farthest point along the main axis of the bone. All measurements are provided in Supporting Information Table S1.

We wish to point out that here we use Gould's (1966: p. 43) definition of allometry, intended as ‘the study of proportion changes correlated with variation in size of either the total organism or the part under consideration […], the variates may arise in ontogeny, phylogeny, or static comparison of related forms differing in size; the term is not confined to any one form of mathematical expression, such as the power function.’ We regard as ‘positive ontogenetic allometry’ any relative increase in size (including centroid size) of a part of an organism correlated with age (i.e. from juvenile to adult). Similarly, ‘negative ontogenetic allometry’ refers to any relative decrease in size of a part of an organism with age, and ‘isometry’ refers to lack of appreciable differential growth (i.e. overall shape is retained between juvenile and adult, or equivalently, there is no relative change in size of the anatomical part under consideration) (Zelditch et al. 2012).

The equation of simple allometry is logY = α(logX) + β, where Y is a measurement of a part of the body (e.g. jaw length) that is considered to vary in relative size with respect to another part of the body X (e.g. skull length), α is the slope of the linear function (allometric coefficient), and β is a constant (Huxley, 1924, 1932; Gould, 1966; Sprent, 1972; Klingenberg, 1998). As the function represents a line, the value of α can be obtained from (logY_A − logY_j)/(logX_a − logX_j), where A and J refer to measurements in the adult and juvenile specimen, respectively. A value of α > 1 indicates positive allometry, α < 1 indicates negative allometry, and α = 1 indicates isometry (Gould, 1966). Values of α were obtained for the snout, braincase, quadrate, and lower jaws of all taxa based on the comparison between their log(CS) and the log(CS) of the whole skull, where CS stands for centroid size of the OCL landmark configurations (OCL was preferred over SSL because the former produces values that can be compared among different taxa). The snout was defined by landmarks 1–11, 15–16, 22–29 (i.e. anterior to the posterodorsal end of the prefrontals); the braincase was defined by landmarks 8–14, 17–19, 34–44; and the quadrate was defined by landmarks 45, 47 and 49 (Fig. 3).

The landmark configurations were exported from landmark editor v.3.0 (Wiley et al. 2007) and aligned with a generalised Procrustes superimposition in morphoj v.1.06d (Klingenberg, 2011). Object symmetry was enforced for the skulls (Klingenberg, 2002) but not for the jaws, of which only the right counterpart was landmarked (because of extensive mobility between the two jaw rami). morphoj v.1.06d (Klingenberg, 2011) was then used to calculate centroid sizes (CS) of all specimen jaws, skulls, and parts of the latter (i.e. snout, braincase, and quadrate), and to produce ‘wireframe’ diagrams to visualise the amount of shape change (Klingenberg, 2013).

To provide a measure of overall shape change, the values of full Procrustes distance between skulls and jaws of juvenile and adult of each species (OCL) were calculated using r v.3.1.3 (R Core Team, 2016) and the R package geomorph v.3.0.0 (Adams & Otarola‐Castillo, 2013).

morphoj v.1.06d (Klingenberg, 2011) was used to perform multivariate analyses of shape. In particular, principal components analyses (PCA) of all the selected taxa were carried out using covariance matrices of the OCL symmetric component of skulls and of the Procrustes coordinates of the jaws.

Results

Below we describe the ontogenetic changes observed in the individual snake taxa examined, followed by a description of patterns common across snakes. We then describe the ontogenetic changes that were observed in the two lizard outgroups, and compare them with those typical of snakes.

Distinctive aspects of each snake taxon

Anilios (=Ramphotyphlops) bicolor (Scolecophidia, Typhlopidae): the skull of the juvenile (Fig. 4) shows poor ossification in its braincase (i.e. unfused elements and fontanelles), like lizards but unlike most other snakes. However, the unossified fissure that completely separates the parietals extends posteriorly to divide also the supraoccipital, a bone that is typically undivided in squamates (Gauthier et al. 2012). Large gaps are also present between frontals and parietals, between parietals and supraoccipitals, and between supraoccipitals and otoocipitals. In the adult the gaps are all closed by bone, and the parietals are completely fused, leaving no trace of their originally paired nature, with the exception of a very small mid‐sagittal notch posteriorly. In the adult the supraoccipitals remain separated by a suture into right and left counterparts. Interestingly, despite a general lack of ossification on the skull roof in the juvenile Anilios, the otic capsules are well ossified, and the footplate of the stapes is already fully enclosed by bone, with only its shaft emerging from the skull. No major difference can be noticed when the otic capsule is compared with that in the adult, with the exception of an incomplete sutural contact between prootic and otoccipital dorsal to the stapes.

Figure 4.

Comparison between skull and lower jaw of juvenile (SAMA R62252) and adult (SAMA R60626) Anilios (=Ramphotyphlops) bicolor. Anterior is to the left. (A) Juvenile skull in dorsal view; (B) juvenile skull in ventral view; (C) juvenile skull in left lateral view; (D) close‐up of otic region of juvenile in left ventrolateral view (quadrate digitally removed); (E) lower jaw of juvenile in lateral view; (F) lower jaw of juvenile in medial view; (G) adult skull in dorsal view; (H) adult skull in ventral view; (I) adult skull in left lateral view; (J) close‐up of otic region of adult in left ventrolateral view (quadrate digitally removed); (K) lower jaw of adult in lateral view; (L) lower jaw of adult in medial view. Skulls and mandibles are not to scale (juvenile HL = 6.55 mm; juvenile JL = 4.07 mm; adult HL = 12.36 mm; adult JL = 8.59 mm). bt, basioccipital tuber; jf, jugular foramen; pdf, posteroventral end of the descending flange of the frontal; so, supraoccipital; sta, stapes; V2 + V3, exit for the maxillary and mandibular branches of the trigeminal nerve.

Another difference between juvenile and adult is that the jugular foramen (foramen located posteroventral to the opening for the stapes) becomes more deeply recessed inside a funnel‐shaped depression. Moreover, whereas in the juvenile the parietal forms the whole anterior margin of the trigeminal foramen, in the adult this foramen is almost completely enclosed by the prootic alone, with only a very small contribution from the parietal (Fig. 4D,J). The adult also has more strongly developed basioccipital tubera. Juvenile and adult share the same tooth count of five maxillary teeth.

In terms of changes in geometric proportions, as highlighted by comparison of the wireframe diagrams (Klingenberg, 2013), the braincase is relatively smaller in the adult compared with the rest of the skull, and the relative compression due to ontogeny is purely anteroposterior (i.e. the frontoparietal suture is shifted posteriorly and the back of the skull is shifted anteriorly), with no evident mediolateral or dorsoventral components (Fig. 5A–C). Moreover, the snout (the region anterior to the posterior margin of the prefrontals) does not expand isometrically but exhibits distinct positive allometry, showing a relative expansion both anteriorly and laterally (Fig. 5A–C, Table 2). No landmarks were placed on the maxillae of Anilios because of their highly mobile nature (in life they can be rotated vertically about 90° relative to the rest of the mandible; Kley, 2001), but from Fig. 4A,G it is clear that the maxilla is displaced laterally as a consequence of the lateral expansion of the snout. This shift is registered also by the lateral extension of the maxillary process of the palatines (landmarks 71/72). Another clear difference between juvenile and adult is in the relative position of the posteroventral end of the descending flanges of the frontal, which in the adult are shifted further back, to the point that they make extensive contact with the prootic, a contact prevented by the interposition of parietal and basisphenoid in the juvenile (Figs 4C,I and 5C). This is clearly correlated with the anteroposterior compression of the braincase relative to the rest of the skull. The quadrate bone also appears to follow this same contraction, is relatively much shorter and stouter in the adult (Table 2), and leaves the large opening of the trigeminal nerve exposed in lateral view (Fig. 4I). Interestingly, the lower jaw of Anilios displays some degree of positive allometry relative to the skull (Table 2), possibly to compensate for the shortening of the quadrate. There is also a noticeable difference between juvenile and adult in the shape of the coronoid bone, which becomes broader in lateral view (Figs 4E,F,K,L and 5D,E).

Figure 5.

Superimposition of the landmark configurations and wireframe diagrams for the juvenile (orange) and adult (blue) of Anilios (=Ramphotyphlops) bicolor and Cylindrophis ruffus (symmetric component only, scaling factor 1.0; anterior to the left). (A) Skull of Anilios as if in in dorsal view; (B) skull of Anilios as if in ventral view; (C) skull of Anilios as if in left lateral view; (D) lower jaw of Anilios as if in medial view; (E) lower jaw of Anilios as if in lateral view; (F) skull of Cylindrophis as if in dorsal view; (G) skull of Cylindrophis as if in ventral view; (H) skull of Cylindrophis as if in left lateral view; (I) lower jaw of Cylindrophis as if in medial view; (J) lower jaw of Cylindrophis as if in lateral view.

Cylindrophis ruffus (Cylindrophiidae): the ventral parasagittal depressions on the basioccipital become visibly deeper with age, and on the skull roof a tall longitudinal crest develops in the adult (Fig. 6). The prootic, supratemporal, and otoccipital jointly develop a cylindrical paroccipital process posterolaterally (Fig. 6A,G,D,J). The openings for the exit of the maxillary (V2) and mandibular (V3) branches of the trigeminal are still confluent in the juvenile but are separated by the broad ‘laterosphenoid’ ossification in the adult (Fig. 6D,J). In the juvenile a gap separates parietal and basisphenoid, and otoccipital and prootic ventral to the stapes, but in the adult these gaps are sealed off and these bones are in sutural contact. Interestingly, the juvenile Cylindrophis also retains a sub‐circular fontanelle on the posterior roof of the skull, between supraoccipital and parietal, reminiscent of the gap for the processus ascendens tecti synotici of lizards (Fig. 6A; see below for a comparison with lizard skulls). Such a discrete fontanelle is absent in the adult and as far as we know has never been reported before in any snake (but see skull roof fissure discussed above for Anilios).

Figure 6.

Comparison between skull and lower jaw of juvenile (WAM R49553) and adult (WAM R121384) Cylindrophis ruffus. Anterior is to the left. (A) Juvenile skull in dorsal view; (B) juvenile skull in ventral view; (C) juvenile skull in left lateral view; (D) close‐up of otic region of juvenile in left lateral view (quadrate and pterygoid digitally removed); (E) lower jaw of juvenile in lateral view; (F) lower jaw of juvenile in medial view; (G) adult skull in dorsal view; (H) adult skull in ventral view; (I) adult skull in left lateral view; (J) close‐up of otic region of adult in left lateral view (quadrate and pterygoid digitally removed); (K) lower jaw of adult in lateral view; (L) lower jaw of adult in medial view. Skulls and mandibles are not to scale (juvenile HL = 8.92 mm; juvenile JL = 7.79 mm; adult HL = 24.02 mm; adult JL = 23.27 mm). VII, exit for facial nerve; V2, exit for maxillary branch of trigeminal nerve; V3, exit for mandibular branch of trigeminal nerve.

With regard to geometric changes (wireframe diagrams), the braincase of the adult Cylindrophis is relatively much narrower than in the juvenile (Fig. 5F,G; very noticeable also in Fig. 6A,G), and the posterior region of the skull (supraoccipital, otoccipitals, and basioccipital) is somewhat displaced more anteriorly. In ventral view, the tooth‐bearing anterior portions of the pterygoids are relatively larger, situated further medially, and parallel to each other (L81/82), resulting in a marked angle with the posterior portions of the bones that continue obliquely towards the quadrates. The ectopterygoids become relatively much larger in the adult (positive allometry), the anterolateral processes of the parietal (L21/22) are displaced anteromedially, and the maxillae grow somewhat laterally and posteriorly. With the exclusion of the maxillae, the snout grows approximately isometrically (Fig. 5F–H, Table 2).

The lower jaw displays moderate positive allometry relative to the skull (Table 2), and the anterior end of the dentary is relatively more upturned in the adult, whereas the compound bone becomes relatively shorter (Fig. 5I,J). The posterior margin of the coronoid extends further posteriorly in the adult, and the dentary becomes deeper. The deepening of the dentary affects also the position of the intramandibular joint, between splenial and angular, which is shifted more ventrally as well (Fig. 5I). The quadrate bone displays negative allometry and, as in Anilios, this is associated with some degree of positive allometry in the lower jaw (Table 2).

With regard to marginal tooth counts the juvenile has 11 maxillary and 12 dentary teeth, whereas the adult has 12 maxillary and 12 dentary teeth. In the palate, the juvenile has seven palatine teeth, like the adult, but only five pterygoid teeth, whereas the adult has eight. The minor variation in the dentary is not necessarily linked to ontogeny because tooth number can vary slightly (+/− one tooth) from side to side even in the same individual (e.g. seven to eight palatine and seven to eight pterygoid teeth in SAMA R36778). However, the large discrepancy in the pterygoid (three teeth) is likely due to ontogeny. There is variation in tooth shape between juvenile and adult, with the latter having teeth that are more strongly recurved posteriorly.

Aspidites ramsayi (Pythonidae): this genus, like Cylindrophis, shows the appearance of a distinct mid‐sagittal crest on parietal and supraoccipital in adults that was completely absent in the juvenile (Fig. 7). The adult Aspidites also has a thin short mid‐sagittal crest on the ventral side of the basisphenoid, just posterior to the basipterygoid processes. As in Cylindrophis, a gap separates parietal and basisphenoid in the juvenile, but this gap is sealed off by a sutural contact in the adult. The para‐sagittal depressions on the basioccipital are much better defined in the adult Aspidites. The otic region of the juvenile Aspidites is not yet completely ossified, the sutural contact between prootic and otoccipital is not yet fully formed, and a distinct gap persists ventral to the juxtastapedial recess (as in Cylindrophis), where the crista tuberalis does not meet the crista interfenestralis or the crista prootica anteroventrally; on the other hand, the juxtastapedial recess is fully enclosed by bone in the adult, where the prootic and otoccipital have also produced a large oblique crest that overhangs the stapes (Fig. 7D,J). Other small qualitative differences that can be highlighted are in the shape of the ectopterygoids, which in the adult acquire a posterolateral process just posterior to the articulation with the maxilla, and in the premaxilla, which in the adult has a trapezoidal outline in dorsal view, whereas in the juvenile it is gently rounded.

Figure 7.

Comparison between skull and lower jaw of juvenile (SAMA R40987) and adult (SAMA R49367) Aspidites ramsayi. Anterior is to the left. (A) Juvenile skull in dorsal view; (B) juvenile skull in ventral view; (C) juvenile skull in left lateral view; (D) close‐up of otic region of juvenile in left posterolateral view (quadrate and quadrate ramus of pterygoid digitally removed); (E) lower jaw of juvenile in lateral view; (F) lower jaw of juvenile in medial view; (G) adult skull in dorsal view; (H) adult skull in ventral view; (I) adult skull in left lateral view; (J) close‐up of otic region of adult in left posterolateral view (quadrate digitally removed); (K) lower jaw of adult in lateral view; (L) lower jaw of adult in medial view. Skulls and mandibles are not to scale (juvenile HL = 18.78 mm; juvenile JL = 17.79 mm; adult HL = 44.94 mm; adult JL = 44.99 mm). ci, crista interfenestralis; ct, crista tuberalis, jf, jugular foramen; larst, lateral aperture of recessus scalae tympani.

The adult Aspidites has 16 teeth on the dentary, 18 on the maxillary, six on the palatine, nine on the pterygoid, and none on the premaxilla. The juvenile seems to possess approximately the same counts for the marginal dentition (posterior‐most teeth on the dentary and maxilla are very small and set in a groove rather than in distinct alveoli, making exact counts difficult); however, its premaxilla has two teeth (plus a large, median, anteriorly directed egg‐tooth) and the teeth on its pterygoid are very small and possibly only starting to erupt (no clear count possible). The teeth on the palatine are already fairly well developed in the juvenile and the count is the same as in the adult (6).

Geometrically, the wireframe diagrams of the skull of Aspidites show a distinct narrowing of the braincase both mediolaterally and dorsoventrally (Fig. 8A–C). This narrowing also affects the posterior half of the frontals (evident also in Fig. 7A,G). The braincase is also relatively shorter in the adult (the landmarks defining the back of the skull are shifted anteriorly). This foreshortening of the braincase is accompanied by a lengthening of the supratemporals, so that the quadrate articulation is not carried anteriorly as well. The lengthening of the supratemporals more than compensates for the foreshortening of the braincase, and as a result the quadrate is shifted posterolaterally in the adult compared with the juvenile. This shift of the quadrate is also accompanied by posterolateral flaring of the palatomaxillary arch (palatine, pterygoid, ectopterygoid, and maxilla). This posterolateral divergence of the palatomaxillary arches could potentially be explained by the skull kinesis, as pythonids are well known to possess a palatomaxillary arch that can rotate on the horizontal plane using the articulation between maxilla and prefrontal as a pivot point (Frazzetta, 1966). However, the fact that the supratemporal is extended posterolaterally as well, indicates that at least part of this posterolateral flaring of the palatomaxillary arches and quadrates is not an artefact of skull kinesis but an actual ontogenetic transformation. This is also supported by the fact that both the lower jaw and quadrate are relatively longer in the adult (Table S1) in order to adjust to this transformation. The orbit of the adult is relatively smaller, mostly because of a posterior extension of the prefrontal and dorsoventral compression of the braincase.

Figure 8.

Superimposition of the landmark configurations and wireframe diagrams for the juvenile (orange) and adult (blue) of Aspidites ramsayi, Acrochordus arafurae, and Notechis scutatus (symmetric component only, scaling factor 1.0; anterior is to the left). (A) Skull of Aspidites as if in dorsal view; (B) skull of Aspidites as if in ventral view; (C) skull of Aspidites as if in left lateral view; (D) lower jaw of Aspidites as if in medial view; (E) lower jaw of Aspidites as if in lateral view; (F) skull of Acrochordus as if in dorsal view; (G) skull of Acrochordus as if in ventral view; (H) skull of Acrochordus as if in left lateral view; (I) lower jaw of Acrochordus as if in medial view; (J) lower jaw of Acrochordus as if in lateral view; (K) skull of Notechis as if in dorsal view; (L) skull of Notechis as if in ventral view; (M) skull of Notechis as if in left lateral view; (N) lower jaw of Notechis as if in medial view; (O) lower jaw of Notechis as if in lateral view.

In Fig. 7(A,G) the premaxilla of the adult Aspidites may appear to have shifted further anteriorly relative to the rest of the skull, but Fig. 8 reveals that (when all landmarks are considered) it is actually a shift in the maxillae that is responsible for the increased gap that makes the premaxilla so prominent, the anterior shift of the premaxilla being very small. The maxillae of the adult have a more pronounced inward curvature anteriorly, so that the anteriormost maxillary tooth locus is placed almost posterior to the posterior corner of the premaxilla, rather than posterolateral to it as in the juvenile.

With regard to the lower jaw of Aspidites, the element appears somewhat deeper in the adult, and a mild dorsoventral bowing is accompanied by a relative shortening of the dentary (Fig. 8D,E). Because of the kinesis in the lower jaw of pythonids (Frazzetta, 1966) the bowing could potentially be an artefact of movement between dentary and postdentary bones, but because not all dentary landmarks are similarly affected (L4 and L15 are almost stationary, and although L15 could be at the centre of rotation, L4 is not) the displacement of the landmarks must reflect a real ontogenetic transformation. The lower jaw as a whole displays an allometric coefficient (α) that is consistent with either isometry or weak positive allometry (Table 2); a larger statistical sample would be necessary to discriminate between the two because the value of α is very close to 1, so we consider values of α between 0.95 and 1.05 approximatively isometric, values of α < 0.90 negatively allometric, values of α > 1.10 positively allometric, and values in the ranges 0.95–0.90 and 1.05–1.10 as ambiguous (i.e. isometric or weakly allometric).

Acrochordus arafurae (Caenophidia, Acrochordidae): in the adult a low and robust mid‐sagittal keel develops ventrally, along the basisphenoid and basioccipital (Fig. 9). The vomeronasal capsule (i.e. vomers plus septomaxillae) is still not fully ossified in the juvenile, whereas the otic capsules are. No major difference can be observed in the ossification patterns of the otic capsule of juvenile and adult Acrochordus, and they both show the unusual condition in snakes of having the lateral opening of the recessus scalae tympani confluent with the jugular foramen (fissura metotica; see also Rieppel & Zaher, 2001). A small dorsal mid‐sagittal crest develops in the adult Acrochordus, but does not extend to the supraoccipital (Fig. 9G).

Figure 9.

Comparison between skull and lower jaw of juvenile (QM J11033) and adult (SAMA R4305) Acrochordus arafurae. Anterior is to the left. (A) Juvenile skull in dorsal view; (B) juvenile skull in ventral view; (C) juvenile skull in left lateral view; (D) close‐up of otic region of juvenile in left posterolateral view (quadrate digitally removed); (E) lower jaw of juvenile in lateral view; (F) lower jaw of juvenile in medial view; (G) adult skull in dorsal view; (H) adult skull in ventral view; (I) adult skull in left lateral view; (J) close‐up of otic region of adult in left posterolateral view (quadrate digitally removed); (K) lower jaw of adult in lateral view; (L) lower jaw of adult in medial view. Skulls and mandibles are not to scale (juvenile HL = 16.58 mm; juvenile JL = 24.12 mm; adult HL = 30.95 mm; adult JL = 52.30 mm). cr, common recess for fissura metotica and two small foramina for the exit of the hypoglossal nerve.

The juvenile Acrochordus has 17 teeth on the maxillae, nine on the palatines, 14 on the pterygoids, and 15–16 on the dentaries; the adult has 16 teeth on the maxillae, 9–10 on the palatines, 13 on the pterygoids, and 15 on the dentaries. As noted above, such minor variability in the tooth counts might be due to individual rather than ontogenetic variability (examples of variable tooth counts in Acrochordus and other snakes that are not obviously correlated with age are provided by McDowell, 1975, 1979).

A comparison of the landmark configurations of the juvenile and adult specimens of Acrochordus (Fig. 8F–J) highlights a relative decrease in size of the braincase in the latter, which is compressed anteroposteriorly, mediolaterally, and dorsoventrally relative to the rest of the skull. Moreover, in the adult the region adjacent to the frontoparietal suture is somewhat sunk ventrally, which gives the lateral outline of the skull a dorsally concave shape (also evident in Fig. 9C,I). The shaft of the quadrate is visibly more elongated in the adult (positive allometry; Table 2) and this causes the articular condyle for the lower jaw to be displaced posteriorly and somewhat laterally. Unlike Aspidites there is no clear increase in the length of the supratemporal posteriorly, and this bone appears to be mostly expanded anteromedially relative to the rest of the skull. The posterior end of the pterygoid follows the posterior displacement of the quadrate mandibular condyle and associated lengthening of the lower jaw (Tables 2 and S1); however, due to the relative mediolateral constriction of the braincase, both the dorsal end of the quadrate and the palatomaxillary arch are located more medially in the adult. This change in configuration cannot be explained as an artefact of skull kinesis because if the palatomaxillary arch of the adult had simply rotated posteromedially using the prefrontal as a pivot point, then the anterior end of the maxilla would have been displaced anterolaterally, which is the opposite of what is observed. Alternatively, if the whole palatomaxillary arch had simply shifted medially due to skull kinesis, then the anterolateral corners of the maxillae should have followed, but again this is not the case (landmarks 67/68). Instead, the relative medial displacement of the palatomaxillary arch is more extreme at the level of the ectopterygoid, and follows the same type of displacement observed for the dorsal end of the quadrate and the supratemporal (located just dorsal to the ectopterygoid). This indicates that the shift in relative position of the suspensorium was followed by the posterior portion of the palatomaxillary arch, most likely constrained by its ligamentous connection to the quadrate (pterygo‐quadrate ligament; Frazzetta, 1966). Moreover, due to the elongation of the quadrate and the consequent shift of the mandibular articulation posteroventrally, the posterior end of the pterygoid and that of the maxilla are also shifted ventrally in a similar fashion, probably because of the constraints imposed by ligamentous connections (pterygo‐quadrate and quadrato‐maxillary ligaments; Frazzetta, 1966). The ectopterygoid bone is relatively longer in the adult (Fig. 8F,G).

Table 2.

Change in snout, braincase, quadrate and jaw centroid sizes (CS) expressed as a percentage of skull centroid size at the same developmental stage, and allometric coefficients (α) for snout, braincase, quadrate and jaw (see text for details)

Taxon	%SJ	%SA	%BJ	%BA	%QJ	%QA	%JJ	%JA	αS	αB	αQ	αJ
Snakes
Anilios bicolor	32	37	63	61	10	7	27	30	1.24	0.92	0.54	1.17
Cylindrophis ruffus	23	24	58	55	5	4	31	35	1.02	0.94	0.80	1.13
Aspidites ramsayi	26	26	51	46	5	7	33	35	1.01	0.90	1.24	1.08
Acrochordus arafurae	28	28	47	40	19	23	51	54	1.03	0.80	1.27	1.08
Notechis scutatus	25	26	53	48	7	9	36	40	1.07	0.86	1.38	1.11
Lizards
Ctenophorus decresii	27	28	53	46	6	7	31	37	1.04	0.79	1.23	1.27
Varanus gilleni	29	32	48	45	7	7	36	38	1.23	0.84	1.00	1.14

B_A, braincase of adult; B_J, braincase of juvenile; J_A, jaw of adult; J_J, jaw of juvenile; Q_A, quadrate of adult; Q_J, quadrate of juvenile; S_A, snout of adult; S_J, snout of juvenile; αB, allometric coefficient of braincase; αJ, allometric coefficient of jaw; αS, allometric coefficient of snout; αQ, allometric coefficient of quadrate.

The adult Acrochordus has more strongly developed crests for the insertion of the adductor mandibulae complex in the lower jaw, especially the crest medial to the adductor fossa (Fig. 9E,F,K,L). The slightly more upturned dentary of the adult Acrochordus is likely an artefact of kinesis (all landmarks marking the position of the dentary in Fig. 8I,J appear to be slightly rotated clockwise). As in Aspidites, the lower jaw has an α value that is indicative of either isometry or weak positive allometry (Table 2). The dentary is relatively shorter in the adult, and this is especially noticeable in the length of the posterior dentigerous process (Fig. 8I,J).

Notechis scutatus (Colubroidea, Elapidae): the skull of the juvenile is relatively well‐ossified from the vomeronasal capsule to the otic region of the skull, where the fenestra ovalis (the opening for the stapes), fenestra rotunda and jugular foramen are all well‐defined, individual openings (Fig. 10), and the crista tuberalis has already merged with the crista interfenestralis to close the ventral margin of the juxtastapedial recess. The low transverse supraoccipital crest of the juvenile has developed into a sharp semicircular crest in the adult (Fig. 10A,G). As the skull of this snake grows, the bony crest that develops anterolaterally on the parietal to join the posterior orbital element (jugal of Palci & Caldwell, 2013) becomes progressively detached from the latter element ventrally, and a large triangular notch appears between the two bones in the adult (Fig. 10C,I). On the ventral side of the braincase, two transversely directed crests with three emarginations each (one midsagittal and two para‐sagittal emarginations) appear in the skull of the adult, one at the level of the suture between basioccipital and basisphenoid and the other just posterior to it and extending to the lateral corners of the basioccipital (Fig. 10B,H). In the adult Notechis, the dorsal musculature does not expand onto the anterodorsal surface of the parietal (unlike Cylindrophis or Aspidites), so that there is a triangular parietal table instead of a sharp mid‐sagittal crest.

Figure 10.

Comparison between skull and lower jaw of juvenile (SAMA R48107) and adult (SAMA R29514) Notechis scutatus. Anterior is to the left. (A) Juvenile skull in dorsal view; (B) juvenile skull in ventral view; (C) juvenile skull in left lateral view; (D) close‐up of otic region of juvenile in left lateral view (quadrate digitally removed); (E) lower jaw of juvenile in lateral view; (F) lower jaw of juvenile in medial view; (G) adult skull in dorsal view; (H) adult skull in ventral view; (I) adult skull in left lateral view; (J) close‐up of otic region of adult in left posterolateral view (quadrate digitally removed); (K) lower jaw of adult in lateral view; (L) lower jaw of adult in medial view. Skulls and mandibles are not to scale (juvenile HL = 11.89 mm; juvenile JL = 12.55 mm; adult HL = 23.10 mm; adult JL = 27.26 mm). fr, fenestra rotunda; jf, jugular foramen.

The fangs of the adult Notechis appear to be much more strongly developed than in the juvenile (Fig. 10C,I). The number of tooth positions in the maxilla of the juvenile is six (two fangs and four smaller posterior teeth), nine on the palatine, 17–18 on the pterygoid, and 19 dentary teeth. In the adult there are six maxillary teeth (two fangs and four smaller teeth), 8–10 palatine, 18 pterygoid, and 19 dentary teeth.

In terms of landmark displacements, the adult configuration shows a distinct relative contraction of the braincase, which is mediolaterally, anteroposteriorly, and dorsoventrally compressed relative to the juvenile (Fig. 8K–M). This is very similar to what can be observed in the basal caenophidian Acrochordus, but with the difference that in Notechis there is no posterior shift of the fronto‐parietal contact. As in Acrochordus, a relative lengthening of the quadrate is here observed (Tables 2 and S1), but in Notechis this results in a displacement of the mandibular condyle that is posterolateral rather than mostly posterior. The displacement of the condyle is similar to that observed for Aspidites but, unlike Aspidites, in the adult Notechis the posterior end of the supratemporal does not grow further posteriorly relative to the rest of the braincase (most of the relative expansion of the supratemporal occurs anteromedially). The palatine and pterygoid, which were parallel in the juvenile, bow outwards in the adult (Fig. 8L). Most of this bowing takes place at the pterygo‐palatine joint, but some degree of bending is also observed in the shape of the palatine (Fig. 10B,H). The posterolateral displacement (rotation in the horizontal plane) of the palatines is known to occur during the swallowing cycle (Albright & Nelson, 1959; Kardong, 1986), but in this case the movement would be unilateral (i.e. would affect only one side of the skull at a time).

Another difference between juvenile and adult configurations resides in the relative position of the space between the dorsal lamina of the nasal and the anterior edge of the frontal (part of the prokinetic joint), which is displaced further posteriorly in the adult Notechis (landmarks 11/12, 19/20). As already described above, the contact between parietal and posterior orbital element extends further ventrally in the juvenile, and this is highlighted by the displacement of landmarks 26/27 (Fig. 8M).

The lower jaw of Notechis displays some degree of positive allometry (Table 2), and the anterior end of the dentary is slightly more upturned in the adult (Fig. 8N,O). Based on the static position of all other landmarks in the dentary, this does not appear to be an artefact of kinesis. The more recurved shape of the dentary in the adult compared with the juvenile can also be observed in Fig. 10(E,K).

Comparison across snakes

The species examined in this study all show a rather similar relative size increase of the skull from early juvenile to adult, approximately doubling in centroid size (CS) and linear dimensions during growth (mean increase in CS across all taxa: 2.09; mean increase in length: 2.02 times; Tables 1 and S1). The greatest relative increase was observed in Cylindrophis ruffus (adult skull CS 2.76 times juvenile; adult skull length 2.69 times juvenile) and in the pythonid Aspidites (adult skull CS 2.47 times juvenile; adult skull length 2.34 times juvenile). This is likely due to the fact that both Cylindrophis and Aspidites are species where the selected juvenile specimens are closest to the start of postembryonic ontogeny, as evidenced by retention of the egg tooth in the former and a remnant of the umbilical cord in the latter.

Table 1.

Values of centroid size (CS) for the skulls and jaws of the juveniles (Juv) and adults (Ad) of the sampled taxa, and values of full Procrustes distance (PD) between juveniles and adults of the same species (values of CS are in mm)

Taxon	Juv Skull CS	Ad Skull CS	Juv‐Adu Skull PD	Juv Jaw CS	Adu Jaw CS	Juv‐Adu Jaw PD
Anilios	15.87	29.73	0.0205	4.24	8.88	0.0213
Cylindrophis	23.06	63.60	0.0088	7.09	22.58	0.0246
Aspidites	48.52	119.84	0.0191	16.00	42.07	0.0240
Acrochordus	48.29	97.26	0.0065	24.75	53.03	0.0116
Notechis	32.31	65.65	0.0123	11.82	26.10	0.0310
Ctenophorus	27.94	53.99	0.0205	8.72	20.06	0.0090
Varanus	37.78	58.69	0.0062	13.56	22.37	0.0088

In these earliest stages some features related to incomplete ossification of the braincase can be observed; in particular, the lack of sutural contact between parietal and basisphenoid, and the lack of sutural contact between prootic and otoccipital ventrolateral to the stapes. Unlike other examined snakes, where the otic capsule is typically well ossified in the juvenile as in the adult, in the juveniles of Cylindrophis and Aspidites a distinct gap persists between the crista tuberalis (otoccipital) and the prootic, so that the lateral aperture of the recessus scalae tympani (LARST) is exposed in lateral view. Moreover, in the juvenile Aspidites the subdivision of the fissura metotica into jugular foramen and LARST is still in progress, as it is incomplete on one side (the thin bridge of bone separating the two openings is still incomplete in the middle).

In the snakes examined, with the only exception of Anilios, growth is nearly isometric in the snout (the region anterior to the posterior margin of the prefrontals) (Table 2). The braincase becomes relatively smaller in all species, generally along all three axes, although there are exceptions where relative compression is only in one plane or along the longitudinal axis.

The skulls of juveniles and adults generally differ in that the adults show well‐developed muscle crests, such as a distinct mid‐sagittal crest on the parietal and supraoccipital, and crests on the ventral side of the basisphenoid and basioccipital. In some taxa, (Cylindrophis, Aspidites) para‐sagittal depressions on basisphenoid and basioccipital, barely visible in the juveniles, are much better defined in the adults.

Both Cylindrophis and Acrochordus show an increase in the relative size of the ectopterygoid with age (Figs 5F,G and 8F,G). All snakes examined except Anilios and Cylindrophis show distinct positive allometry in their quadrate bones relative to their skull (Table 2). Interestingly, Anilios and Cylindrophis both show negative allometry of the quadrate, and this is associated with some degree of positive allometry in the lower jaw; the quadrate is oriented ventrally or anteroventrally in these taxa, so positive allometry of the quadrate would not increase relative gape size.

There is no correlation between amount of size change during growth and overall change in skull shape (Table 1); despite the fact that Cylindrophis and Aspidites show the largest difference in size between sampled juvenile and adult, the greatest shape change, as indicated by the value of full Procrustes distance between age categories, is observed in the skulls of Anilios and Aspidites.

In both Cylindrophis and Aspidites there is ontogenetic variation in tooth shape, where the juvenile has more uniformly recurved crowns, whereas the adult shows a distinct angle (~120°) between the base and the tip of its teeth. Cylindrophis and Aspidites suggest that teeth on the pterygoid of snakes are the last to appear in ontogeny, and that, at least in some cases, they may increase in number with age.

Postnatal changes in lizards

Ctenophorus decresii (Iguania, Agamidae): the skull of the juvenile Ctenophorus decresii (Fig. 11A–D) shows weak ossification in the braincase (both endochondral and dermal elements). The roof of the skull presents an enormous parietal fontanelle, which decreases considerably in size in the adult but does not completely disappear, persisting as an unossified area surrounding the parietal eye (a common feature within Iguania, where the relative size of the parietal fontanelle at sexual maturity is similar to that of neonates in other lizards; Maisano, 2001). Lack of ossification between elements of the braincase in the juvenile also results in the confluence of the fenestra rotunda with a large space (presumably filled by cartilage in life) between basisphenoid, basioccipital, and prootic. In the adult this space is completely sealed by sutural contacts between the above‐mentioned bones, and a distinct finger‐like basioccipital process has developed ventrolaterally, just ventral to the fenestra rotunda. On the ventral side of the braincase, along the midline and between basisphenoid and basioccipital, a small unossified portion (basicranial fenestra, commonly reported in neonates of other lizards; Maisano, 2002b,c; Nance, 2007) can be observed in the juvenile but is completely sealed off by bone in the adult.

Figure 11.

Comparison between skull and lower jaw of juvenile (SAMA R53670) and adult (SAMA R28618) Ctenophorus decresii. Anterior is to the left. (A) Juvenile skull in dorsal view; (B) juvenile skull in ventral view; (C) juvenile skull in left lateral view; (D) close‐up of otic region of juvenile in left ventrolateral view (quadrate digitally removed); (E) lower jaw of juvenile in lateral view; (F) lower jaw of juvenile in medial view; (G) adult skull in dorsal view; (H) adult skull in ventral view; (I) adult skull in left lateral view; (J) close‐up of otic region of adult in left posterolateral view (quadrate and quadrate ramus of pterygoid digitally removed); (K) lower jaw of adult in lateral view; (L) lower jaw of adult in medial view. Skulls and mandibles are not to scale (juvenile HL = 10.45 mm; juvenile JL = 10.75 mm; adult HL = 19.67 mm; adult JL = 24.74 mm). bp, basioccipital process; fr. fenestra rotunda.

Another evident morphological difference between the juvenile and adult skulls is the appearance of longer, recurved teeth at the front of the mouth in the adult, a change that occurs also in the lower jaw (Fig. 11). In particular, in the adult, two fang‐like teeth appear at the front of each maxilla, whereas on the premaxilla all four teeth become relatively larger, especially the central pair, which in the juvenile are very small. There is also a substantial increase in the tooth count, as the juvenile has nine teeth both in the maxilla and dentary, whereas the adult has 14 on each (ontogenetic increase in tooth number on maxilla and dentary is a common feature within agamids; Cooper et al. 1970). The locus for the newest tooth can be seen at the back of the tooth row (Fig. 11F), with the new tooth developing ventrally from the crown.

Comparison of the wireframe diagrams indicates that the braincase of the adult Ctenophorus decreases in relative size (negative ontogenetic allometry), and appears to be shifted anteriorly, compressed dorsoventrally and mediolaterally (Fig. 12A,C, Table 2). In the adult there is a small relative posterolateral shift of the posterior rim of the upper temporal fenestra, a shift that is followed also by the quadrate. The ectopterygoid processes are clearly more strongly developed in the adult, and extend further ventrally (landmarks 71/72). Increased ossification of the palatal elements (especially palatines and pterygoids) with age results in a relative expansion of the bony roof of the mouth and narrowing of the suborbital fenestrae (visible also in Fig. 11). The orbits become relatively smaller with age (Fig. 12C). The snout of Ctenophorus grows almost isometrically (Table 2) but its tip extends further forward and upward in the adult, which results in a more convex outline of the margin of the maxillary tooth row in lateral view. A complementary, concave curvature develops in the lower jaw, where both the posterior and the anterior end appear to be upturned in the adult relative to the rest of the mandible. This shape change of the jaw is associated with an anterior shift of the ventral process of the coronoid (landmark 21) and by a relative shortening of its dorsal process (landmark 3). The dentary of Ctenophorus is relatively shorter in the adult, whereas the retroarticular process extends further posteriorly as well as dorsally. Both the lower jaw and the quadrate display distinct positive allometry relative to the skull (Table 2).

Figure 12.

Superimposition of the landmark configurations and wireframe diagrams for the juvenile (orange) and adult (blue) of Ctenophorus decresii and Varanus gilleni (symmetric component only, scaling factor 1.0; anterior is to the left). (A) Skull of Ctenophorus as if in dorsal view; (B) skull of Ctenophorus as if in ventral view; (C) skull of Ctenophorus as if in left lateral view; (D) lower jaw of Ctenophorus as if in medial view; (E) lower jaw of Ctenophorus as if in lateral view; (F) skull of Varanus as if in dorsal view; (G) skull of Varanus as if in ventral view; (H) skull of Varanus as if in left lateral view; (I) lower jaw of Varanus as if in medial view; (J) lower jaw of Varanus as if in lateral view.

Varanus gilleni (Anguimorpha, Varanidae): the ossification pattern in the braincase of Varanus gilleni (Fig. 13) is strikingly different from that observed in Ctenophorus. A large unossified area of the parietal is not observed in the juvenile Varanus, although it retains a slit‐like fissure between the parietal foramen and the frontals. This fissure is completely sealed in the adult. The braincase of the juvenile Varanus is also better ossified than that of the juvenile Ctenophorus, and no distinct porosity can be observed. The fenestra rotunda is almost completely enclosed by bone but still communicates with a small gap that persists between prootic, otoccipital and basioccipital (as in the juvenile Ctenophorus). On the ventral side of the braincase, between basisphenoid and basioccipital, the juvenile has a distinct basicranial fenestra, which is absent in the adult. The crista tuberalis (the thin oblique crest located posterior to the fenestrae ovalis and rotunda) is more strongly developed in the skull of the adult, as are the basioccipital processes. The pterygoid processes of the basisphenoid do not reach the pterygoids in the juvenile, but these bones meet in what appears to be a sliding contact in the adult. The alar process of the prootic that juts out from the rounded anterodorsal margin of the otic capsule is relatively small and trapezoidal in the juvenile, but has become markedly enlarged relative to the otic capsule in the adult, and has acquired a distinctly square outline. The tooth count in Varanus is conserved between the juvenile and the adult stage (10 maxillary teeth, 10 dentary teeth, and nine premaxillary teeth). The nasals are paired in both juvenile and adult.

Figure 13.

Comparison between skull and lower jaw of juvenile (SAMA R18223) and adult (SAMA R32164) Varanus gilleni. Anterior is to the left. (A) Juvenile skull in dorsal view; (B) juvenile skull in ventral view; (C) juvenile skull in left lateral view; (D) close‐up of otic region of juvenile in left ventrolateral view (quadrate digitally removed); (E) lower jaw of juvenile in lateral view; (F) lower jaw of juvenile in medial view; (G) adult skull in dorsal view; (H) adult skull in ventral view; (I) adult skull in left lateral view; (J) close‐up of otic region of adult in left ventrolateral view (quadrate digitally removed); (K) lower jaw of adult in lateral view; (L) lower jaw of adult in medial view. Skulls and mandibles are not to scale (juvenile HL = 15.94 mm; juvenile JL = 14.95 mm; adult HL = 26.25 mm; adult JL = 25.48 mm). alp, alar process of the prootic; ct, crista tuberalis; fr, fenestra rotunda.

The wireframe diagrams show that the braincase of Varanus, like that of Ctenophorus, decreases in relative size with age, is shifted anteriorly and compressed mediolaterally (Fig. 12F–H). Unlike Ctenophorus, however, there is no relative dorsoventral compression. The ventral condyle of the quadrate appears to be shifted posteromedially in the adult Varanus, but this may be an artefact of some degree of streptostyly (i.e. movable quadrate) present in this taxon (Frazzetta, 1962). The pterygoid and ectopterygoid of Varanus are shifted ventrally in the adult compared with the juvenile (see landmarks 93/94, 125/126). In Varanus the snout region displays positive allometry, mostly because of the tip, which in the adult extends a bit further anteriorly relative to the rest of the skull (Fig. 12F–H, Table 2). No major changes can be observed in the configuration of the lower jaw, which grows almost isometrically (Fig. 12I,J, Table 2). The quadrate of Varanus also grows isometrically (Table 2).

Comparison between lizards and snakes

No distinct difference could be observed in the general way the skull of lizards grows compared with snakes (this is quantified using PCA below). Despite major changes in skull kinesis and braincase enclosure, both groups of squamates seem to follow the same broad pattern, where the braincase decreases in relative size, the snout grows generally isometrically, and the jaws display isometry to positive allometry (Table 2). In the taxa examined the relative decrease in size of the braincase typically happens via compression along all three main body axes (anteroposterior, mediolateral, and dorsoventral). Exceptions, where the relative compression is limited to one or two axes, are provided by Varanus (anteroposterior and mediolateral only), Anilios (anteroposterior only), and Cylindrophis (anteroposterior and mediolateral only).

In Aspidites and Ctenophorus the orbit shows a distinct decrease in relative size with age, very much like in mammals (Segura & Prevosti, 2012; Flores et al. 2015), whereas a relative change in size is not so evident in Acrochordus, Notechis and Varanus (change in relative size of the orbit cannot be evaluated in Anilios and Cylindrophis because of their lack of bones framing the eye posteriorly). In both Aspidites and Ctenophorus the relative decrease in size of the orbit is mostly due to a posterior shift of the antorbital wall of the prefrontal combined with a relative compression of the skull posterior to the nasals.

The lower jaws of all snakes examined show high values of full Procrustes distance between juvenile and adult, which is indicative of a consistently greater degree of morphological remodelling compared with lizards (Table 1). On the other hand, the values of Procrustes distance of the sampled skulls do not highlight any consistent dichotomy between lizards and snakes.

Ctenophorus shares with Anilios and Aspidites a high value of full Procrustes distance between juvenile and adult skulls, whereas a similarly low degree of remodelling is shared by Varanus and Acrochordus (Table 1).

All lower jaws examined display allometry coefficients (α) relative to the skull that range between 1.08 and 1.27 (Table 2), where the highest value (1.27) was registered for Ctenophorus, but is likely due mostly to the extreme elongation of the retroarticular process in the adult (Fig. 11K,L). In some of the taxa examined (Aspidites, Acrochordus, Notechis) the anterior extent of the adductor fossa (marked by landmark 17 in snakes) retains its exact relative position with age.

The quadrate bone of most taxa shows positive allometry, but in Varanus this bone grows isometrically, whereas in Anilios and Cylindrophis the bone displays negative allometry (Table 2). Interestingly, the iguanian Ctenophorus shares with the macrostomatan snakes (i.e. Aspidites, Acrochordus, and Notechis) the strong positive allometry of the quadrate.

Results of principal components analyses

The PCA of skull shape produced 13 components, the first three of which account for about 80% of the total variance (Supporting Information Table S2). Graphic representations of the skulls with respect to the first three principal components are provided in Fig. 14 (PCA coordinates for the first seven eigenvalues, which account for over 95% of the total variance, are provided in Supporting Information Table S3).

Figure 14.

Visualization of ontogenetic trajectories of the skulls by means of PCA plots and wireframe diagrams. Arrows indicate ontogenetic trajectories (the arrowhead marks the position of the adult). Wireframe diagrams show morphological variation at the extremes (negative and positive) of each principal component for dorsal (top), ventral (middle), and lateral (bottom) representations of the skull (anterior to the left). Ac, Acrochordus arafurae; An, Anilios bicolor; As, Aspidites ramsayi; Ct, Ctenophorus decresii; Cy, Cylindrophis ruffus; No, Notechis scutatus; Va, Varanus gilleni.

Ontogenetic shifts along the first 2 PCs are very similar across all taxa except for Anilios (Fig. 14). The first principal component appears to be the best descriptor of ontogenetic shape change in most taxa, which in all but Anilios shift in the direction of negative values as they reach maturity. Interestingly, this is true for both lizards and snakes. Juvenile and adult Anilios both display very high, and very similar, values of both PC1 and PC2, and this sets this snake apart from all other squamates investigated here (Fig. 14). The main shape changes associated to a shift in the negative direction of PC1 (i.e. in the direction of growth for most taxa) consist of: (i) more upturned snout (premaxilla, nasals, vomers); (ii) mediolateral compression of the nasals; (iii) reduction in size of the frontals; (iv) anteroposterior compression of the prefrontals; (v) more elongate palatines; (vi) slimmer braincase, compressed mediolaterally and somewhat dorsoventrally (vii) anteroposteriorly elongate parietal and basisphenoid; (viii) more elongate quadrates; (ix) quadrate shaft shifting from an anteroventrally to a posteroventrally tilted position.

All taxa except Anilios show decreasing values of PC2 as they age. This transformation corresponds to the following changes: (i) premaxilla and nasals become narrower mediolaterally; (ii) prefrontals and vomers become wider mediolaterally; (iii) vomers extend further anteriorly; (iv) the posteroventral end of the prefrontals extends further anterodorsally; (v) palatine‐pterygoid arches bow inward rather than outward; (vi) the frontals taper forward; (vii) the braincase is compressed mediolaterally and dorsoventrally; (viii) the dorsal end of the quadrate becomes narrower anteroposteriorly.

The plot of PC1 vs. PC2 (Fig. 14) shows that whereas Anilios occupies the upper right corner of the plot (high values of PC1 and PC2), the lizards are placed close together in the lower right corner, and most snakes are located in the middle of the plot. Acrochordus, however, is located to the far left of the plot (lowest value of PC1).

PC3 does not appear to be as effective as PC1 or PC2 at capturing directions of transformation that are associated with ontogenetic growth. However, PC3 is the most useful component to discriminate shape changes occurring in Anilios, where juvenile and adult have very similar values of both PC1 and PC2. Shape change in the negative direction of PC3 involves: (i) a broader premaxilla; (ii) more elongate nasals; (iii) prefrontals that are compressed anteroposteriorly but elongate dorsoventrally; (iv) larger frontals; (v) broader and shorter braincase; (vi) pterygoids that flare out posteriorly; (vii) larger quadrate; (viii) quadrate shaft tilted posteroventrally rather than anteroventrally.

Interestingly, the lizards Ctenophorus and Varanus, unlike snakes, are characterised by relatively high values of PC1 and low values of PC2. The only exception among snakes is the aberrant‐looking scolecophidian Anilios, which also shows high values of PC1 but has high values of PC2 as well, unlike any other snake or lizard examined. On the other hand, Acrochordus has very low values of PC1, which sets it apart from all other taxa.

The PCA of lower jaw shape produced 13 components as well, the first three of which account for over 90% of the total variance (Supporting Information Table S4). Graphic representations of the jaws with respect to the first three principal components are provided in Fig. 15 (PCA coordinates for the first seven eigenvalues, which account for over 99% of the total variance, are provided in Supporting Information Table S5).

Figure 15.

Visualization of ontogenetic trajectories of the jaws by means of PCA plots and wireframe diagrams. Arrows indicate ontogenetic trajectories (the arrowhead marks the position of the adult). Wireframe diagrams show morphological variation at the extremes (negative and positive) of each principal component for lateral (top), and dorsoventral (bottom) representations of the jaw (anterior to the left). Ac, Acrochordus arafurae; An, Anilios bicolor; As, Aspidites ramsayi; Ct, Ctenophorus decresii; Cy, Cylindrophis ruffus; No, Notechis scutatus; Va, Varanus gilleni.

Ontogenetic shape changes across PC1 and PC2 were less concordant across taxa (Fig. 15). With the exception of Aspidites, Cylindrophis, and Ctenophorus, the jaws of most taxa do not appear to move far in shape space during ontogeny, and there is no common direction for all trajectories, although most taxa appear to develop towards positive values of PC1.

The major morphological changes associated with the positive direction of PC1 consist of a shortening of the dentary associated with a concomitant lengthening of the postdentary portion of the mandible (Fig. 15).

Morphological transformations associated with a shift along the positive direction of PC2 include: (i) a tapering of the posterior end of the lower jaw; (ii) a deeper notch between dorsal and ventral posterior rami of the dentary; and (iii) a jaw that bows further laterally.

PC3 mostly describes a dorsoventral expansion of the jaw occurring in the direction of positive values.

Interestingly, the shape of the jaw of the two snakes Aspidites and Cylindrophis tends to converge toward the same position in shape space in all three main principal components, whereas other taxa are distributed far and apart. Very little morphological change is registered for the jaws of Acrochordus, Anilios, Notechis, and Varanus, whose ontogenetic trajectories are consistently very short in all three plots (Fig. 15).

Discussion

The snakes and lizards examined here show certain similarities in their patterns of ontogenetic development. All show negative allometry in the size of the braincase, and except for Anilios (=Ramphotyphlops) and Varanus, isometry of the snout. At least for the regions and landmarks examined, there was no distinct allometric pattern that uniquely characterised snakes to the exclusion of the two lizard outgroups (Table 2). However, PCA analysis highlighted how the skulls of lizards differ from those of snakes (at least in our sample) in having relatively high values of PC1 and low values of PC2 (Fig. 14), although the ontogenetic shifts were largely parallel. Principal component analysis also showed that the most striking morphological changes shared by lizards and snakes during ontogeny involve the braincase, and the shape and orientation of the quadrate bone. These changes are well synthesised by the transformations occurring along the negative direction of PC1 (most ontogenetic trajectories, with the only exception of Anilios, form vectors oriented in approximately parallel directions) (Fig. 14).

Unlike the skulls, the lower jaws of most sampled taxa do not share a common direction of ontogenetic shape change in the first two PCs (Fig. 15). According to the first three principal components, shape change in the lower jaws appears to be smaller compared to the skulls, and major transformations were recorded only for Aspidites, Cylindrophis and Ctenophorus (Fig. 15). However, the values of full Procrustes distance suggest that the full extent of the morphological transformation of the jaw is generally higher in snakes than in lizards, which implies that likely such transformation is, at least for some taxa, best captured by some of the following principal components (PC) (i.e. beyond the 3rd PC). For example, this seems to be the case in Anilios, where the shift in shape space is very limited in the first three PC compared with other taxa, but is the largest in PC7 (Table S5). This is clearly an example of the fact that generally no PC should be expected to provide a complete representation of the morphological change correlated with size (Klingenberg, 2016).

One of the most consistent ontogenetic changes seen across all snake and lizard species examined consists of the elaboration of various crests and processes that serve for attachment of muscles and tendons, typically for the cervical (e.g. basioccipital tubera) and adductor musculature (e.g. mid‐sagittal crest on parietal). This ontogenetic transformation is commonly observed in other vertebrates as well (e.g. Segura & Prevosti, 2012; Stucchi, 2013; Flores et al. 2015; Jasinoski et al. 2015).

Cylindrophis, Aspidites and Acrochordus show the ontogenetic appearance of a distinct sagittal crest on the roof of the skull in the adult. If we consider that in all of these snakes the braincase becomes relatively much smaller in the adult, then this mid‐sagittal crest likely helps compensate for the relative decrease in surface area for attachment of the adductor musculature inserting on the parietal and/or supraoccipital. In Notechis we do not see the appearance of a mid‐sagittal crest in the adult, but the skull roof remains largely unaffected by relative compression, as indicated by the appearance of a triangular parietal table later in ontogeny (the braincase is compressed only around this area), which helps partially preserve the original relative position of the juvenile muscle attachments.

Among snakes, the most divergent taxon is the scolecophidian Anilios (=Ramphotyphlops), which not only shows a mostly one‐dimensional (anteroposterior) compression of the adult braincase relative to the rest of the skull, but also distinct positive allometry in its snout region. Scolecophidians in general differ markedly from all other snakes in the expanded and strongly sutured bones of the snout and absence of a prokinetic (fronto‐nasal) joint (List, 1966). This arrangement creates a very spacious nasal cavity and a strongly reinforced leading edge of the snout, suggesting potential importance in relation to chemoreception and/or fossoriality. Other burrowing snakes, including two in our study (Cylindrophis and Aspidites) are like most other non‐scolecophidian snakes in anterior skull bone shape, having a small premaxilla only weakly contacting adjacent skull bones, plus retention of a prokinetic joint (some other burrowing snakes, uropeltids, have more rigid, less kinetic snouts, but these are typically tapering anteriorly rather than expanded as in scolecophidians; Olori & Bell, 2012). The mechanical significance of the anatomical differences of burrowing scolecophidians compared with other burrowing snakes has been briefly discussed by Cundall & Rossman (1993), but as the burrowing biomechanics of scolecophidian snakes remain poorly known, we cannot yet provide an adaptive explanation for this ontogenetic pattern in Anilios. For now we are limited to stating that this pattern contributes to the distinctive adult shape of scolecophidian skulls, is potentially related to fossoriality and/or chemoreception, and could prove to be a strong developmental constraint.

The juvenile Anilios, unlike alethinophidian snakes and most lizards (Bellairs & Kamal, 1981; Maisano, 2001, 2002b; Nance, 2007; Werneburg et al. 2015; but see Maisano, 2002c for Xantusiids), shows a paired parietal (mid‐sagittal fontanelle) which does not ossify following a centripetal pattern; rather, completely ossified left and right counterparts later co‐ossifiy in adults. The complete separation of left and right parietals and the weakly ossified roof of the skull in the juvenile Anilios appears counterintuitive in a species that relies on a consolidated skull for burrowing. Oddly enough, a lack of sutural contact is often observed in scolecophidians between left and right parietal, and also between prootic, parietal and basisphenoid (List, 1966). The poor ossification of the braincase is difficult to explain adaptively. This lack of a sutural contact between the above‐mentioned skull bones may be indicative either of young age in the specimens examined or of paedomorphosis, a heterochronic process through which the adult skull configuration resembles that of the juvenile of more plesiomorphic taxa (McNamara, 2012) (lack of sutural contact between parietal and basisphenoid was also observed in the juveniles of both Cylindrophis and Aspidites). We found indications of paedomorphosis in Anilios in the form of retention of some juvenile skull proportions compared with other squamates. This was consistent with the high (and ontogenetically constant) values of PC1 and PC2, showing virtually no change between juvenile and adult along these principal components, in contrast to all other taxa (Fig. 14).

The paired parietals of scolecophidians may not be indicative of the primitive snake condition, as suggested by List (1966), but rather they may simply represent a by‐product of heterochrony. The lack of ossification in the braincase of juvenile Anilios is puzzling if the burrowing behaviour of these snakes remains unchanged between juvenile and adult stages. However, the head of Anilios is very small relative to the rest of the body (skull miniaturisation; Rieppel, 2012), and if the small skull of this snake develops as the result of heterochronic processes, then the apparent lack of adaptive value in the weakly constructed skull of the juvenile may simply represent a necessary, but temporary, non‐adaptive (or even maladaptive) ontogenetic constraint along the trajectory leading to the small‐headed adult proportions (McKinney & McNamara, 1991).

The juvenile Cylindrophis shows two very interesting features: (i) lack of separation between maxillary and mandibular branches of the trigeminal nerve (V2 and V3, respectively); and (ii) presence of a distinct gap between supraoccipital and parietal on the skull roof. Lack of subdivision of the exits for the trigeminal branches is typical of lizards and of the most basal snakes, like scolecophidians, Dinilysia, and madtsoiids (List, 1966; Scanlon, 2005, 2006; Zaher & Scanferla, 2012). All more derived snakes (alethinophidians), including Cylindrophis, are generally assumed to have the V2 and V3 exits separated by a laterosphenoid ossification. The second feature, presence of a fontanelle between supraoccipital and parietal, is similar to the condition observed in most lizards, where a gap between these two bones (Figs 11A,G and 13A,G) lodges a cartilagineous outgrowth of the supraoccipital (processus ascendens tecti synotici; Evans, 2008). Both of these features are interesting because they show that Cylindrophis, widely considered to be a relatively primitive alethinophidian, exhibits transitory primitive/lizard‐like conditions previously unreported in alethinophidians.

Another primitive and previously unreported feature is observed in the juvenile Aspidites, which retains premaxillary teeth (completely lost in the adult). Premaxillary teeth are typically present in adult Pythonidae (the family which includes Aspidites), including basal forms, so the absence of these teeth in the adult is a relatively recent loss, perhaps related to the semifossorial habits of this snake (Bruton, 2013). The premaxilla in this snake is relatively large and anteriorly projecting in the adult, evidently functioning as a reinforcement for the front of the snout to improve its effectiveness when scooping soil from burrows (Bruton, 2013). During growth there is an increase in the anteromedial curvature of the maxillae, directing the anteromedial tips of these bones towards one another, on an arc that runs posterior to the premaxilla. This results in a premaxilla that lies largely beyond the mouth cavity in the adult, which may be functionally correlated with the loss of premaxillary teeth.

The posterolateral flaring of the palatomaxillary arches is yet another striking feature observed in the adult Aspidites. It is tempting to think that this ontogenetic change could be related to a switch in dietary preference, but there is no evidence for this in A. ramsayi, which tends to feed mostly on reptiles (lizards and snakes) throughout its life (Shine & Slip, 1990).

Examination of the PCA plots reveals that the skulls of all taxa (except Anilios) shift towards negative values of PC1 as they mature. However, the entire ontogenetic trajectory of Acrochordus is greatly displaced towards negative values of PC1 compared with all other taxa (Fig. 14), suggesting overall peramorphosis.

Aspidites, Acrochordus, and Notechis all exhibit adaptations for large‐gape, termed the ‘macrostomatan’ condition. This was once thought to characterise a clade of advanced snakes (Rieppel, 1988) but genomic data has recently shown that these snakes do not form a monophyletic assemblage (e.g. Wilcox et al. 2002; Reeder et al. 2015; Streicher & Wiens, 2016; Zheng & Wiens, 2016; but see Hsiang et al. 2015). In particular, the phylogenetic distribution of macrostomy raises the possibility that the condition in Aspidites (pythonid) is not homologous to that in Acrochordus and Notechis (caenophidians). In Aspidites the jaw articulation is shifted posterior to the braincase largely as a result of the posterior expansion of the supratemporal (Fig. 8A–C); while although the quadrate does show positive allometry, its vertical orientation means that this allometry does not contribute to the posterior displacement of the lower jaw articulation but does contribute to a wider mouth via the lateral displacement of its ventral condyle. A similar configuration, where the supratemporal projects posteriorly far behind the braincase, whereas the quadrate shaft slopes ventrolaterally, can be observed in many other booids, suggesting that they share the same developmental pattern (e.g. Acrantophis madagascariensis ZFMK 21670, ZFMK 86469; Boa constrictor ZFMK 21661; Corallus caninus AMNH R57788, R73347; Epicrates cenchria ZFMK86470; Eunectes murinus AMNH R29349, AMNH R57474; Liasis albertisi ZFMK 5165; Morelia spilota FMNH 22380, ZFMK 84282; Python molurus ZFMK 5161, ZFMK 83431; Python reticulatus FMNH 51631, ZFMK 70207; Python sebae ZFMK 5200; A. Palci, pers. obs.). In Acrochordus and Notechis the jaw articulation is similarly displaced posteriorly (in Acrochordus) and posterolaterally (in Notechis), but the supratemporal does not expand further posteriorly with age; rather, the posterior shift of the jaw articulation in these snakes is achieved entirely by a relative lengthening of the quadrate bone, which is large (extraordinarily so in Acrochordus) and posteroventrally oriented (Figs 8F–H,K–M, Table 2). A quadrate bone that is strongly tilted posteroventrally can be observed in many other caenophidian snakes (e.g. Colubridae: Coluber caspius ZFMK 5221; Dasypeltis scabra MCZ 54849; Heterodon platyrhinos AMNH R69647, AMNH R155313; Lampropeltis getulus AMNH R70097, AMNH R128202; Malpolon monspessulanus ZFMK 5197; Thamnophis sirtalis AMNH R74849, AMNH R148084; Elapidae: Austrelaps superbus SAMA R13997, SAMA R35577; Dendroaspis angusticeps MCZ 49556; Hydrophis platurus FMNH 171632, FMNH 216510; Pseudonaja textilis SAMA R14065, SAMA R26961; Naja naja AMNH R74833, ZFMK 21704; Homalopsidae: Bitia hydroides NHML 1930.5.8.640; Cerberus rhynchops NHML 58.9.21.3, NHML 1964.10.20; Viperidae: Agkistrodon piscivorous ZFMK 21724; Bitis gabonica AMNH R57799, AMNH R137177; Causus rhombeatus FMNH 51692, FMNH 51693; Daboia russelii AMNH R74818, AMNH R75739; Vipera xanthina ZFMK 86077; AP per.obs.).

The existence of at least two different ontogenetic routes to obtain a ‘macrostomatan’ adult cranium is consistent with the phylogenetic evidence that this condition evolved convergently in pythons and in caenophidians (Streicher & Wiens, 2016). Interestingly, some snakes (e.g. Homalopsis buccata NHML 1964.11.25) have supratemporals that extend very far posterior to the braincase (as in pythons and boas) but also have long posterolaterally oriented quadrates (as in most caenophidians); this configuration could result from a third developmental pathway that may or may not be derived from those described above (posteriorly projecting supratemporals may result from their positive allometric growth, but also from a relative shortening of the rest of the skull); a denser sampling of ontogenetic data in a phylogenetic context, especially one including the Tropidophiidae (another potential instance of independent evolution of the macrostomatan condition; Rieppel, 2012; Streicher & Wiens, 2016), would be required to test whether macrostomy has evolved in multiple instances and via distinct developmental trajectories.

An additional pattern in Notechis worth discussing is the bilateral outward bowing of the palatine‐pterygoid arches. Such bowing appears to be a product of ontogeny in this snake, rather than an artefact of skull kinesis. A similar bowing was observed in some other skulls of adult Notechis (e.g. SAMA R2327, SAMA R543a, SAMA R26968) but was less apparent in others (e.g. SAMA R26959; SAMA R22574) (A. Palci, pers. obs.). Furthermore, it does not seem to be a universal condition among elapids (e.g. absent in: Furina diadema SAMA R6703, SAMA R6075; Hydrophis platurus FMNH 171632; FMNH 216510; Pseudonaja textilis, SAMA R14065, SAMA R26961, SAMA R44184; Dendroaspis angusticeps MCZ 49556; A. Palci, pers. obs.). Some degree of bowing of the palatine‐pterygoid bar can also be observed in Naja naja (AMNH R74833) and in Austrelaps superbus (SAMA R13997) (A. Palci, pers. obs.). Interestingly, in A. superbus the bilateral outward bowing is observed in a fairly large individual (HL = 27 mm), whereas a smaller individual (SAMA R35577; HL = 21 mm), despite having a clearly arched palatine, shows only a very weak lateral bowing of the palatine‐pterygoid bar, which parallels what we observed in Notechis. This modification of the palate could occur in order to achieve a wider mouth, which in turn may be correlated to an ontogenetic dietary shift. However, there appears to be some degree of intraspecific variability in the extent of the outward flexion of the palatine‐pterygoid bar in adult Notechis, which raises the possibility that this feature may be influenced by dietary preferences (or constraints) in certain populations. Shine (1987) and Aubret et al. (2004) documented a correlation between head morphology (jaw length) and diet in mainland and island populations of tiger‐snakes, where snakes feeding on larger prey have relatively longer jaws. The structure of the palate in adult tiger snakes may be a further plastic developmental feature correlated with habitat and/or diet. A systematic study of the palatal morphology of mainland and insular populations may shed light on this possibility.

The most interesting ontogenetic changes observed in the lizard Ctenophorus pertain to the jaws, including the differentiation of the dentition as the animal matures, the arching of the ventral edge of the maxilla, which becomes more convex, and the increase in relative size and change in orientation of the retroarticular process, which projects horizontally in the juvenile but angles posterodorsally in the adult. Most adult agamids are characterised by having cylindrical pointed pleurodont teeth on the premaxilla and the first few loci on the maxilla, followed by a series of triangular acrodont teeth that form a cutting edge that slices against a matching series of dentary teeth (Cooper et al. 1970; Evans, 2008). As shown by Cooper et al. (1970) in agamids the dentition of hatchlings is much less strongly differentiated than in adults, and the juvenile C. decresii shows this in that all its teeth are similar in shape, with the anteriormost teeth (on both dentary and premaxilla) somewhat smaller and narrower than the rest. In the adult C. decresii, the cylindrical teeth at the front of the maxilla and dentary are markedly projecting and have been described as caniniform or fang‐like. The reason why agamids retain some slender pleurodont teeth anteriorly has generally been canvassed in terms of adaptive reasons, such as weapons during social interactions (Lebas, 2001) or teeth that can penetrate and hold prey as it first enters the mouth (Cooper et al. 1970), but as this dentition is a virtually fixed characteristic of all agamids (regardless of diet or social system) it may simply stem from a developmental constraint established at the time the clade originated.

The more deeply arched profile of the maxilla probably improves the crushing and slicing capabilities of the adult jaws, providing a better distribution of the stresses during biting (i.e. increased critical load, which is inversely proportional to the radius of the arch; Karnovsky, 2012) as well as more efficient cutting (i.e. oblique cutting; Atkins, 2009). A relative increase in the volume of the adductor chamber in the adult, as indicated by the posterolateral expansion of the upper temporal fenestra, suggests a non‐linear increase in bite force that may be correlated with the morphological changes described above for the jaws.

The changing orientation and relative enlargement of the retroarticular process argue for a functional change in the mechanics involved in jaw opening. A longer lever‐arm at an optimal angle might enable more rapid opening of the mouth to improve prey capture or may have an involvement in disengaging the tightly meshed acrodont teeth at the start of a mouth‐opening cycle. The fact that chameleons, with similar interlocking acrodont teeth, essentially lack a retroarticular process altogether (Evans, 2008) would suggest that there are important differences in the mechanics of the jaws of the two major acrodont clades.

Interestingly, the iguanian Ctenophorus shares with ‘macrostomatan’ snakes a strong positive allometric growth of the quadrate. In the light of recent molecular phylogenies (e.g. Reeder et al. 2015; Zheng & Wiens, 2016) this could be viewed as a potential synapomorphy between iguanians and snakes; however, basal snakes like Anilios and Cylindrophis do not show positive allometry of the quadrate (Table 2). Further taxon sampling of both lizards and snakes will help resolve this issue.

The skull and lower jaw of Varanus gilleni display limited shape change during ontogeny. This is not likely true of all Varanus species, some of which show distinctive changes in diet and tooth morphology as they mature (Mertens, 1942; Estes & Williams, 1984; D'Amore, 2015). A possible explanation is the small size of V. gilleni (pygmy mulga monitor), with adults reaching a maximum total length of about 35 cm (Cogger, 2014). This small size may be correlated with paedomorphic processes such as neoteny (i.e. slower growth rate), postdisplacement (i.e. delayed onset of growth), and/or progenesis (i.e. early offset of growth), which may in turn attenuate the amount of ontogenetic change between juvenile and adult specimens (McNamara, 2012). The ontogenetic trajectories of both skull and lower jaw of Varanus gilleni are in fact shorter than those of the other lizard examined, C. decresii, in all of the first three principal components (Figs 14 and 15).

In V. gilleni the most striking difference between juvenile and adult morphology is in the contact between pterygoids and basisphenoid, which is absent in the juvenile and present as a sliding joint in the adult. This is probably due to the fact that the distal end of the basipterygoid processes is still cartilaginous in the juvenile, and ossifies only later in ontogeny, as happens in other lizards (Barahona & Barbadillo, 1998; Nance, 2007). The contact between basipterygoid processes and pterygoids also appears to be correlated with the overall mediolateral compression of the skull in Varanus, which pushes them closer together in the adult.

Another interesting feature in the juvenile V. gilleni is the small fontanelle between left and right sides of the parietal anterior to the pineal foramen. A study by Werneburg et al. (2015) has shown that this fontanelle is also present in the late embryo of V. panoptes. Therefore, the ossification pattern in Varanus is somewhat different from that of Ctenophorus and other iguanians, where closure of the parietal fontanelle is much slower (i.e. the fontanelle still persists in sexually mature individuals), and centripetal (Maisano, 2001, 2002b).

Conclusions

The recent advances in technology and the greater affordability of laser scanners, micro‐CT scanners, and digital microscribes, combined with computational tools for 3D geometric morphometric analysis, have resulted in a proliferation of 3D morphological studies focused on the osteology, ontogeny, allometry, sexual dimorphism, evolution, and speciation of vertebrates (e.g. Lieberman et al. 2007; Hedrick & Dodson, 2013; del Castillo et al. 2014; Owen et al. 2014; Sherratt et al. 2014). We have applied some of these techniques to produce a detailed ontogenetic study of the skull and lower jaw of a selection of snakes and lizards. Among the most interesting results are: (i) relatively conserved developmental patterns shared by lizard and snakes, most of which show isometric (or near‐isometric) growth in the snout and negative allometry in their braincase; (ii) divergent ontogenetic routes that generate the relative size reduction of the braincase (i.e. various combinations of anteroposterior, mediolateral, and dorsoventral compression); (iii) that among snakes a posterior shift of the jaw articulation, and consequently a larger gape, can be obtained via at least two distinct allometric growth pathways (due to posterior extension of either supratemporal or quadrate), which are then reflected on the relative positioning of the palatomaxillary arches; (iv) that the braincase (especially the skull roof) of the juvenile Anilios, and possibly also that of other scolecophidians, is very weakly ossified, despite fossorial habits being generally associated with robust skulls; (v) that at least some juveniles of Cylindrophis still retain a single opening for the exit of the maxillary and mandibular branches of the trigeminal nerve (V2 and V3), and a fontanelle between supraoccipital and parietal that is reminiscent of the condition in lizards; (vi) that the palatal dentition of Cylindrophis and Aspidites, and possibly other snakes, develops first on the palatine and later on the pterygoid, and the tooth count on the pterygoid may increase with age; (vii) that premaxillary teeth are present in juveniles of Aspidites and are subsequently lost through ontogeny; (viii) that the iguanian Ctenophorus shares with ‘macrostomatan’ snakes a strongly positive allometric growth of the quadrate; (ix) that snakes, compared with lizards, consistently show a greater degree of ontogenetic remodelling of the lower jaw (as suggested by comparison of full Procrustes distances between juveniles and adults).

An improved knowledge of the type and extent of morphological change that can be expected as a result of ontogeny can also help in the identification and systematics of fossil remains, as has been shown previously for both snakes and lizards (Barahona et al. 2000; Scanferla & Bhullar, 2014). It is also important to recognise that morphological features that change greatly during postnatal ontogeny may be misleading in phylogenetic analyses, when juvenile or even subadult specimens are used as representatives of terminal taxa. Other traits are relatively stable throughout postnatal ontogeny, and might be more reliable. For example, certain traits used in a comprehensive phylogenetic analysis of squamate morphology (Gauthier et al. 2012) are highly variable during postnatal ontogeny, such as relative width of the parietal, and presence/absence of a parietal crest or of the basisphenoid keel [Gauthier et al.'s (2012) characters 92, 93, 326]. Others are generally more stable, such as characters related to the shape of the coronoid and retroarticular process [e.g. Gauthier et al.'s (2012) characters 386, 387, 404–410; however, orientation of the retroarticular process, as described in character 407, may change ontogenetically, as shown in Ctenophorus]. Marginal tooth counts [Gauthier et al.'s (2012) characters 419–421], perhaps surprisingly, were very stable in all snake taxa examined (except on the premaxilla of Aspidites), and in Varanus. Although it is general practice to score only adult specimens in phylogenetic analyses of vertebrates, there are cases where only juveniles or specimens of uncertain ontogenetic stage are available (e.g. fossil taxa, or very rare extant taxa). In these situations morphological traits that are known to be susceptible to drastic postnatal ontogenetic change should be scored as unknown or at least treated with scepticism.

In the future it would be interesting to broaden our understanding of squamate ontogenetic patterns by sampling representatives of all remaining major groups of lizards (dibamids, amphisbaenians, gekkos, pygopodis, teiids, gymnophthalmids, lacertids, scincids, cordylids, anguids, chamaeleonids, and iguanids), and within snakes it would be interesting to explore in more detail the diversity within Colubroidea, the most successful and diverse living snakes (Greene, 1997). For example, it would be interesting to see whether highly fossorial species share specific ontogenetic patterns in the development of their skull with other species (e.g. positive allometry of the snout), how much ontogenetic variability is present among different burrowing lineages, and how this variability may relate to their specific fossorial ecologies. Last but not least, sampling more basal alethinophidians would help shed light on whether tropidophiid snakes represent an additional instance of convergent evolution of the macrostomatan skull condition.

Supporting information

Appendix S1. Lists of all the landmarks used in the ‘species‐specific’ landmarking schemes (SSL), their descriptions, and relevant notes.

Appendix S2. List of all the landmarks used in the ‘overall comparison’ landmarking scheme (OCL), their descriptions, and relevant notes.

Table S1. Skull (HL), mandible (JL), and quadrate (QL) lengths of the specimens examined, juveniles (J) and adults (A).

Table S2. Eigenvalues and percentage of variance for each principal component in the PCA of skull shape.

Table S3. Values of the first seven principal components for the skulls of all taxa examined (juv. = juvenile; ad. = adult).

Table S4. Eigenvalues and percentage of variance for each principal component in the PCA of jaw shape.

Table S5. Values of the first seven principal components for the jaws of all taxa examined (juv. = juvenile; ad. = adult).