Morphological evolution of Ceratophryinae (Anura, Neobatrachia)

Ceratophryinae relationships

Cladistic analyses with implied weighting method show high support for the monophyly of pipids, scaphiopodids and some neobatrachian taxa (Ceratophryinae, Bufo spp., Leptodactylus spp., Phrynobatrachus spp., Ptychadena spp., Hoplobatrachus plus Pyxicephalinae, and Arthroleptidae) (Jacknifing absolute values >50 in Fig. 2). Ceratophryines, always monophyletic, appear as the sister taxon of all neobatrachians although in the strict consensus obtained with criterion of equally weighted characters, no relationships among pipids, pelobatids and neobatrachians can be proposed.

Heuristic searches with 1000 random-addition sequences replicates with the criterion of equally weighted characters found minimal-length topologies of 468 steps (CI 0.2711; RI 0.693). The branch swapping algorithm used was tree bisection and reconnection (TBR) and minimal-length trees replicates led to 2362 trees. Their strict consensus tree recovered Ceratophryinae's monophyly. For this analysis, the jacknifing value for ceratophryines is 99. Relationships within ceratophryines are less well supported; although the clade formed by Lepidobatrachus spp. reaches a jackknifing value of 89.

Using the implied weighting method, statistical coefficients of CI and RI exhibit light variation compared with the equally weighted characters method [K = 1 (CI = 0.25, RI = 0.666); K = 2 (CI = 0.256, RI = 0.669); and K = 3 (CI = 0.258, RI = 0.672)] and their results are more congruent to previous hypothesis (Ford and Cannatella 1993). All analyses found ceratophryines as a monophyletic group. Heuristic searches with 1000 random-addition sequences with implied weighting K = 2 led to three trees with 389 steps with a best score of 40.03388. The trees display differences in the position of some taxa (Astylosternus diadematus, Scinax fuscovarius) but most taxa do not vary. Relationships represented in tree 0 (Fig. 2) were considered enough to focus derived characters in the morphology of ceratophryines. Jacknifing support (with an independent character removal = 25) has values of 90 for ceratophryines, 68 for Chacophrys plus Lepidobatrachus, and 60 for Lepidobatrachus (Fig. 2). Monophyly of pipids, scaphiopodids, and neobatrachians are in agreement with current hypotheses on the phylogeny of frogs (Ford and Cannatella 1993), and unsolved relationships within suprageneric bufonoid and ranoid taxa are still the subject of discussion and study.

Synapomorphies defining the clade Ceratophryinae (Fig. 2) correspond to characters 2, 8, 11, 12, 14, 23, 24, 29, and 30; characters 23, 29, and 30 being autoapomorphies. Characters 2 (skull exostosis), 8 (presence of arch parieto-squamosal), 11 (zygomatic ramus of squamosal reaching the maxilla), and 12 (monocuspid teeth) are shared with some scaphiopodids, pipids, ranids, bufonids and hylids; this set of characters involves extensive development of skull bones. Character 14 (absence of pars palatina of premaxilla) occurs in some pipids and ceratophryines. Character 24 (strong fusion of dentary and mentomeckelian) appears in some ranids and ceratophryines. Character 23 (spur-like fang formed by dentary and mentomeckelian bones) is one of the autoapomorphies of the group. Ceratophrys is the sister group of Chacophrys plus Lepidobatrachus. Synapomorphies of Ceratophrys are characters 44 (dorsal shields, see Fig. 4b) and 45 (transverse processes of eighth presacral vertebra oriented forward). This last character exhibits 12 steps in the tree. The synapomorphy of the clade containing Chacophrys plus Lepidobatrachus (character 32) is the presence of a well developed parahyoid bone (Fig. 3c–f). Synapomorphies for the Lepidobatrachus node are represented by characters 25 (discontinuous ceratohyalia, Fig. 3d–f), also present in scaphiopodids; 75 (presence of double spiracles in tadpoles), as in pipid tadpoles; and 77 (absence of keratodonts), as in pipid and microhylid tadpoles. Synapomorphic characters of Chacophrys (52, presence of ileal ridge) and L. laevis (60 and 61; terminal phalanx of toe IV curved and pointed) appear many times (52, 9 steps; 60, 12 steps; and 61, 16 steps). Lepidobatrachus llanensis has a synapomorphy represented by character 44 (anterior and posterior dorsal shields, Fig. 4d).

Although this cladistic analysis could be improved by increasing the size of the data set of characters or by more extensive taxon sampling, the obtained hypothesis of phylogenetic relationships is sufficient to discuss intergeneric variation of selected traits of ceratophryines.

Skull shape variation in ceratophrynines

In the analysis of dorsal skull landmarks, RW1, RW2, and RW3 accounted for 72.13% of the variation (RW1 = 40.26%, RW2 = 19.62%, and RW3 = 12.25%). Relative warp values are given in Table 1. RW1 versus RW2 scatterplot reflects trajectories where lower values of RW1 are exhibited by Chacophrys, Ceratophrys, and smaller specimens of Lepidobatrachus spp. (Fig. 5a). RW1 explains shape changes involving a decrease in skull length and an increase in skull width at positive values (Fig. 5b). RW2 describes variation in the upper jaw length and posterior skull width; negative values represent smaller specimens of Lepidobatrachus spp. (Fig. 5b).

The lower jaw landmark data show that lower jaw shapes vary among ceratophryines, with clear differences between Lepidobatrachus and Chacophrys-Ceratophrys. Relative warp values are listed in Table 1. RW1, RW2, and RW3 accounted for 89.76% of the variation (RW1 = 56.98%, RW2 = 21.38%, and RW3 = 11.40%). RW1 versus RW2 scatterplot reflects trajectories where negative values of RW1 are exhibited by Chacophrys, Ceratophrys, and smaller specimens of Lepidobatrachus spp. (Fig. 5c). RW1 explains shape changes involving an increase of the length of the dentary and angulosplenial concomitant with a decrease in the length of the skull, from negative to positive values (Fig. 5d). RW2 displays variation in the relative position of lower jaw articulation with respect to the neurocranial articulation, from an anterior to a posterior position (Fig. 5d).

Skull shape variation among ceratophryines and other anurans

In dorsal skull, RW1, RW2, and RW3 comprised 59.80% of the variation (RW1 34.21%, RW2 14.54%, and RW3 11.05%). RW values for dorsal skull are listed in Table 2. RW1 versus RW2 scatterplot shows high values of RW1 for ceratophryines (Fig. 6a). Changes involve maxillary and squamosal length, skull width and length, and relative extension of nasal region relative to the consensus configuration (Fig. 6b). RW2 changes are reflected in squamosal and maxillary orientation, and nasal, frontoparietal, and maxillary lengths (Fig. 6b).

In lower jaw, RW1, RW2, and RW3 accumulated 87.16% of the variation (RW1 47.62%, RW2 23.39%, and RW3 16.15%). RW values for dorsal skull are listed in Table 2. RW1 versus RW2 scatterplot exhibits the extreme position of the ceratophryines with high positive values of RW1 (Fig. 6c). Changes of shape resulting in positive values of RW1 imply elongation of lower jaw bones and skull length shortening (Fig. 6d). As you precede from negative to positive values for RW2 the jaw morphology changes from acuminated to a rounded lower jaw, and ceratophryines exhibit intermediate position in this axis (Fig. 6d).

The Ceratophryinae clade

The Ceratophryinae was recognized by many studies (Reig and Limeses 1963; Lynch 1971; Maxson and Ruibal 1988; among others). Some authors proposed a familial status for Chacophrys, Ceratophrys, and Lepidobatrachus (Cei 1981) but at present, the group is considered part of the Ceratophryinae, with Macrogenioglottus, Odontophrynus and Proceratophrys, within the paraphyletic Leptodactylidae (Frost 2004).

Some features of ceratophryines seem to be primitive –e.g. tadpoles with 20 or 27 pairs of spinal nerves (Nishikawa and Wassersug 1989) or tadpoles having separate trigeminal and facial ganglia (Fabrezi and Chalabe 1997). Other ceratophryine's traits appear to be derived, or, at least, not present in those taxa considered basal – Ascaphidae, Leiopelmatidae, Bombinatoridae, and Discoglossidae – such as sculptured skulls, fangs (Fabrezi and Emerson 2003), monocuspid and non-pedicellate teeth (Smirnov and Vasil'eva 1995; Fabrezi 2001), etc.

Larval development in ceratophryines displays curious variation. Tadpoles of Chacophrys are typical pond larvae (Carrizo and Faivovich 1992); while Ceratophrys and Lepidobatrachus have carnivorous tadpoles with simplified internal oral structures (Wassersug and Heyer 1988), and a short alimentary tract with an adult-like stomach (Ruibal and Thomas 1988; Carroll et al. 1991; Ulloa Kriesler 2000). Further, Lepidobatrachus tadpoles have a symmetrical pair of large branchial openings (no siphons), derived from asymmetrical development resembling the sinistral spiracle of type 4 larvae (Ruibal and Thomas 1988), enlarged mouths without keratodonts and reduced beaks (Haas 2003), and external forelimb development. Thus, developmental larval variation in the ceratophryines may be one of the most important characteristics of their morphological evolution.

Peramorphosis is a developmental pattern of heterochrony that may be produced by an increase in rate (acceleration), a later offset time (hypermophosis), or an earlier onset time (pre-displacement) of development of some structure (Reilly et al. 1997). Some osteological features, sometimes shared with other taxa, occur altogether in the ceratophryines and are excellent models for studies of peramorphosis and its influence on morphological evolution. Some features supporting the monophyly of ceratophryines in this paper were described as peramorphic traits among anurans. For example, character 12 (monocuspid teeth, Fabrezi 2001) and fangs (character 23, Fabrezi and Emerson 2003) seem to be integrated and correlated to dietary specialization (these fanged frogs eat large preys) and aggressive biting behaviour (Fabrezi and Emerson 2003). The following discussion points out other peramorphic features of ceratophryines.

Ossification of the hyoid is rare among anurans (Fig. 3). Usually, older specimens of some anuran taxa may have mineralization on the hyoid cartilage. The Bombinatoridae possess a pair of flat, plate-like endochondral bones in the hyoid (Trewavas 1933; Cannatella 1985). Lepidobatrachus spp. and C. pierottii, however, display an extended (hyper-) ossification of the posteromedial processes reaching the ceratohyalia. These ossifications progress during post-metamorphic growth and are more extensive in larger adults, showing a pattern of hypermorphosis. Among ceratophrynines, an increase in the degree of ossification of the hyoid could be interpreted as concomitant to an increase in size, or functionally as a need for a more robust lower jaw-cranium relation given the feeding ecology of ceratophryines.

Absence of the parahyoid bone is a neobatrachian synapomorphy (Ford and Cannatella 1993). A dermal V-shaped parahyoid is a synapomorphy for Discoglossidae (and convergent with Pelodytes) (Ford and Cannatella 1993), and a medial parahyoid appears in several taxa, such as Ascaphus, Leiopelma, Barbourula, Bombina, and Rhinophrynus (Cannatella 1985). Except for Rhinophrynus, all these taxa bear a discrete and small parahyoid bone, which is especially large in Rhinophrynus and also in the Ceratophryinae (Fig. 3c–f). de Beer (1937) described the Y-shaped parahyoid of Polypterus as a tendon ossification, probably of paired origin. This element does not properly belong to the hyobranchial skeleton, and may be homologous with the ‘urohyal’ of osteichtyan fishes and the parahyoid of Anura (de Beer 1937). In ceratophryines, the parahyoid can be considered a derived feature, but not a morphological novelty, because it is present in other anuran taxa (Müller and Wagner 1991).

Most anurans have continuous ceratohyalia, but Lepidobatrachus is the only neobatrachian taxon that has discontinuous ceratohyalia (Fig. 3d–f). Incomplete ceratohyalia – ceratohyalia lacking the intermediate pars – are present in Rhinophrynus, Pelobates, Scaphiopus, and Pelodytes, while the complete or almost complete absence of ceratohyalia in adults is a synapomorphy of Megophryidae (Cannatella 1985; Ford and Cannatella 1993).

Postcranial dorsal shields rarely occur among anurans (Fig. 4b,d). As the species having these ossifications are terrestrial, it was assumed they have been independently derived on at least three occasions as protective modifications (Trueb 1973). Dorsal shields were mentioned in Lepidobatrachus asper, Ceratophrys aurita, Brachycephalus, and dendrobatids (Trueb 1973). Dorsal shields are formed from the calcified layer of dermis, and could be vestigial structures corresponding to a dermal skeleton developed in ancestral amphibians, and may be associated with physiological functions (Toledo and Jared 1993). The calcified layer is better developed in dorsal skin and the concentration of material deposited increases with the age of the animal (Toledo and Jared 1993). An armor formed by dermal plates firmly attached by ligament to expanded neural spines was described among Permian dissorophids, and DeMar (1966) proposed a scenario of terrestrial conditions where thicken dermis allow less transfer of moisture though the skin. This functional significance seems to be a reasonable explanation for the presacral complex present in species of Ceratophrys typical of arid and semi-arid environments, such as C. cranwelli and C. ornata. Interestingly, the aquatic L. llanensis has poorly developed dorsal shields suggesting that they are being lost, due perhaps to lack of necessity for them.

Traits that represent reappearances of lost structures, imply a morphological integration that seems to arise from an underlying/quiescent developmental program that was conserved in the historical patterns shared by the ceratophryines. Reappearances in neobatrachians suggest that the loss of these structures did not mean the loss of the ability to form them. This ability was preserved in the developmental program of anurans and has re-emerged in the peramorphic ontogeny typical of ceratophryines.

Adult ceratophryines share a set of features that are formed during or immediately after metamorphosis – earlier than in other anuran taxa – such as appearance of tooth anlagen, paired fangs, and fusion of lower jaw bones (Table 3). Developmental pathways may be characterized as peramorphic with respect to other anurans (Table 3). Some of them involve reappearances of structures considered lost in the phylogeny of neobatrachians, suggesting that a peramorphic ontogeny may play a role in creating new designs without producing true morphological novelties.

Patterns of peramorphosis in skull shapes of ceratophrynines

The skull shape of Lepidobatrachus could be represented by a near circle, where the medial axis (ratio) is the skull length. In Chacophrys and Ceratophrys, skull lengths are longer than in Lepidobatrachus. These shape differences represent spatial and temporal repatterning involving major changes in the elongation of lower and upper jaws bones concomitant with posterior wideness of skull. Chacophrys and Lepidobatrachus skull shapes represent the extremes within the Ceratophryinae, while Ceratophrys may be considered an intermediate. As froglets and juveniles, the skull shapes of Lepidobatrachus are similar to Ceratophrys. The most remarkable change in the adult skull shape of Lepidobatrachus is the displacement of the lower jaw articulation posterior to the occipital joint.

Among anurans, the distinctiveness of the skull shape of ceratophryines is evident and reflects peramorphosis in the ontogeny of the group. TPS analyses of shape among a sample of 50 anuran taxa emphasized differences between ceratophryines and other frogs. Yeh (2002) described changes in skull shapes among anurans where the increase of growth of upper and lower jaws reflects more ontogenetic changes. As described here, ceratophryine characteristics of elongated jaws with a decrease in skull length result in their unique skull morphology. In dorsal view, RW1 versus RW2 scatterplots placed skull shape of Pyxicephalus adspersus near the ceratophrynines, but not in the extreme position of Lepidobatrachus (Fig. 7a). This is expected because P. adspersus has developed cranial exostosis via an extension of the squamosals, maxillaries, and frontoparietals that has resulted in an overall skull shape similar to the ceratophryines. Skull shapes in Ceratophrys and Lepidobatrachus are consequences of a strong divergence from the developmental patterns of the typical anuran.

Despite the fact that individual traits in jaws and hyoid apparatus are shared by ceratophryines and other taxa of anurans, the way in which they are integrated in the skull of the Ceratophryinae defines unique morphologies where developmental patterns of peramorphosis drive changes toward unusual directions.

The comparative data support the hypothesis of the monophyly of the Ceratophryinae. They share a particular set of morphological features resulting from patterns of peramorphosis. Some of the features represent ancient characters that have been considered lost in the neobatrachian evolution. Diversification in the group is limited to three genera, with no more than 15 species, suggesting that as an evolutionary process, peramorphosis is not particularly significant in terms of species diversification, but is able to produce morphological evolution. The genus Lepidobatrachus could be considered the extreme of ceratophrynine's morphology. Comprehension of peramorphic development is necessary to elucidate the association between development, growth, and evolution. The Ceratophryinae represent an appropriate subject for such studies.