Are Hemipenial Spines Related to Limb Reduction? A Spiny Discussion Focused on Gymnophthalmid Lizards (Squamata: Gymnophthalmidae)
Pedro M. Sales Nunes,
Departamento de Zoologia, Instituto de Biociências, Universidade de São Paulo, São Paulo, SP, Brazil
Correspondence to: P. M. S. Nunes; Departamento de Zoologia, Instituto de Biociências, Universidade de São Paulo, Caixa Postal 11.461, CEP 05422-970, São Paulo, SP, Brazil. Fax: +55 11 3091 75 13. email@example.com
Limb reduction, defined as the partial or complete phylogenetic loss of skeletal elements of the limbs relative to a typical four-limbed, pentadactyl condition (Greer, 1991; Shapiro, 2002), is widespread in the evolution of Tetrapoda. Different degrees of limb reduction are particularly common in the order Squamata, with several independent events among its lineages (Greer, 1991; Wiens et al., 2006; Skinner, 2008).
The most pronounced instances of fore and/or hind limb reduction in squamates (complete limblessness or partial reduction affecting functionality) occur in the major clades Serpentes and Amphisbaenia, as well as in the lizard families Anguidae, Annielidae, Cordylidae, Dibamidae, Gymnophthalmidae, Pygopodidae, and Scincidae. Among lizards, the vast majority of reduction events occur in the Scincidae [16 to 20 events (Miralles et al., 2012) or 27 events only among Lerista (Skinner et al., 2008)]. Another group showing multiple events of limb reduction is the Gymnophthalmidae. According to Pellegrino et al. (2001), at least five independent events of limb reduction have occurred within the Gymnophthalmidae: one in the subfamily Rachisaurinae (Rachisaurus brachylepis); two in the subfamily Gymnophthalminae (tribes Gymnophthalmini and Heterodactylini); and two in the subfamily Cercosaurinae [genera Anotosaura (subfamily Epleopodinae, sensu Castoe et al., 2004) and Bachia (tribe Bachiini, sensu Castoe et al., 2004)] (see Pellegrino et al., 2001; Castoe et al., 2004; Rodrigues and Dos Santos, 2008; Rodrigues et al., 2009 for systematic accounts).
Limb reduction is usually associated with body elongation, resulting in a snake-like body form that is often related to fossorial, sand-dwelling, or grass-dwelling habits (Caputo et al., 1995; Lee, 1998; Wiens and Slingluff, 2001). In addition, such snake-like morphology implies in biomechanical and physiological adaptations of most body systems (Underwood, 1976; Caputo et al., 1995). However, there is only a handful of revisions about modifications of the genital system and/or reproductive behavior of squamates and its correlation with limb reduction (e.g., Presch, 1978; Olsson and Madsen, 1998; Sánchez-Martínez et al., 2007).
The use of genital morphology as an indication of kinship (Dowling, 1957; Arnold, 1986b; Nunes et al., 2012) is mostly supported by evidence from sexual selection (Eberhard, 1985, 2010). In contrast, most functional interpretations of hemipenial ornaments [e.g., mechanical stimulation (King et al., 2009), increase of copula duration (King et al., 2009; Olsson and Madsen, 1998), gaping of female cloaca (Pisani, 1976) and lifting of the anal plate (Pisani, 1976)] are widely speculative. Even though, there seems to be a widespread assumption that such structures must be directly involved in couple-anchoring during the sexual act to ensure efficient sperm transfer (Pope, 1941; Edgren, 1953; Pisani, 1976; Murphy and Baker, 1980).
Calcified hemipenial spines and spicules are apparently restricted to some gymnophthalmid lizards (e.g., Uzzel, 1965; Estes et al., 1988; Nunes, 2011; Nunes et al., 2012) and several representatives of a major lineage of snakes [Colubroides sensu Zaher et al. (2009)]. Calcified hemipenial structures have also been reported for other squamate groups, but the homology and general structure of such elements are debatable. For example, Arnold (1973, 1986a) mentions a possible seasonal presence of minute spines in the Lacertidae, but this observation has received no further attention. In addition, distinct calcified hemipenial structures have been reported in some Varanidae (McDowell and Bogert, 1954; Ziegler et al., 2007; Koch et al., 2009; Welton et al., 2010), Gekkonidae and Sphaerodactylidae (Kluge, 1982; Rösler and Böhme, 2006) representing internal skeletal structures referred to as a hemibaculum, as well as apical horns acting as extensions of the retractor muscles (Branch, 1982). Observations of hemipenial spines in Chamaeleonidae (Olsson and Madsen, 1998) are intriguing, considering that descriptions and illustrations of the hemipenis of several chameleons made by Klaver and Böhme (1986) provide no indication of the presence of such structures. Finally, Olsson and Madsen (1998), mention the presence of spines in the hemipenis of Anguis fragilis, an assumption likely based on the descriptions of Cope (1896) and on an illustration by Böhme (1988), since they provide no evidences that such structures are in fact calcified. The absence of any traces of calcified structures in the hemipenis of other Anguidae (Thomas and Hedges, 1998; personal observation), even after staining tests with Alizarin Red solution casts further doubts upon their work and leaves the question open for further investigation.
The remarkable degrees of limb-reduction seen in snakes and in some gymnophthalmids, combined with the presence of calcified hemipenial spines in both groups, may have led Presch (1978) to associate the complexity in hemipenial ornamentation with dramatic instances of limb reduction and/or total limblessness. This assumption is based on the premise that, in the absence of well-developed limbs, other anatomical structures (i.e., hemipenial spines) could take on the role of keeping male and female tightly attached while mating. However, with respect to gymnophthalmids, Presch's (1978) observations were not supportive of such a straightforward rationale, since he reported virtually nude hemipenes both in taxa with well-developed limbs (e.g., species of the genera Anadia, Euspondylus, Gymnophthalmus, Proctoporus, and Riama), as well as in two greatly limb-reduced species of the genus Bachia (B. intermedia and B. trisanale). Posterior studies (Myers and Donnelly, 2001; Nunes, 2011) revealed the presence of calcified spicules embedded in the hemipenial flounces of the fully limbed species Anadia ocellata, which were unnoticed by Presch (1978), most likely due to preparation artifacts. Thus, Presch's (1978) observations suggested an opposite pattern to the one predicted in his first assumption, with limb-reduced taxa showing poorly ornamented organs, whereas conspicuous ornaments were common among taxa with well-developed limbs.
Despite the inconclusiveness of Presch's (1978) results, the putative relationship of limblessness with complex hemipenial ornamentation deserves further attention for the following reasons: (i) Presch's (1978) sampling was far from comprehensive, including only one of the limb-reduced gymnophthalmid lineages (i.e., the tribe Bachiini, sensu Castoe et al., 2004); and (ii) the family Gymnophthalmidae is especially informative for studies concerning the “by-products” of limblessness because the several reduction events can be traced in the two recent phylogenies (Pellegrino et al., 2001; Castoe et al., 2004). Therefore, if the patterns of hemipenial ornamentation are confronted with well-supported phylogenetic trees, one can directly assess the possible hierarchic relationship between limb morphology and hemipenial ornamentation.
In this study, we dedicate special attention to hemipenial spines. A recent study on the hemipenial morphology of the Gymnophthalmidae allowed the unequivocal detection of distinct patterns of calcified spines and spicules in the hemipenes of several genera of the family (Nunes, 2011). Based on the considerable variation in size and organization of such structures among the gymnophthalmid taxa presenting variable degrees of limb reduction, we reassess the hypothesis of Presch (1978), regarding the phylogenetic congruence between major reductions in limb morphology, including complete loss of the limbs, and the presence of hemipenial spines, in light of the current phylogenetic hypothesis for the family.
MATERIAL AND METHODS
We examined the hemipenes of the vast majority of valid genera of Gymnophthalmidae, including all taxa with limb reduction. Additionally, we also analyzed the hemipenes of some representatives of Serpentes and Amphisbaenia, as well as selected taxa of the lizard families Anguidae, Anniellidae, and Scincidae (see Supporting Information Appendix I for a list of the specimens examined). Our observations were complemented by an extensive survey of the literature and, in some cases, such data represented the sole information regarding the hemipenial morphology of taxa to which we had no access to (e.g., Dibamidae; see Darevskyi, 1992). We could not obtain any information about the hemipenial morphology of limb-reduced species of the families Pygopodidae, Gerrhosauridae, and of the genus Chamaesaura (family Cordylidae).
Hemipenial preparations follow a combination of the techniques proposed in earlier studies (Manzani and Abe, 1988; Pesantes, 1994; Zaher, 1999; Zaher and Prudente, 1999), added by a staining step (immersion of the everted organ in a 70% alcoholic solution of Red Alizarin for 24 hr) used to enhance calcareous structures and improve their detection (see Uzzell, 1973; Harvey and Embert, 2008; Nunes et al., 2012). The retractor muscle was manually separated and the everted organ filled with a mixture of stained petroleum jelly and melted paraffin (liquefied). The hemipenes of Feylinia currori (Scincidae) and Anniella pulchra [Annielidae, sensu Uetz (2013)] could not be manually everted; alternatively, the inverted organs were longitudinally dissected to allow access to their external surface, and stained accordingly. The terminology of hemipenial structures follows Dowling and Savage (1960), Klaver and Böhme (1986), Savage (1997), and Myers and Donnelly (2001, 2008).
Information regarding the presence/absence of calcified structures in the hemipenes of Gymnophthalmidae were mapped onto the phylogenies of Pellegrino et al. (2001) and Castoe et al. (2004) with the aid the software Mesquite v. 2.75 (Maddison and Maddison, 2011). The analysis simply illustrates the presence or absence of calcified hemipenial spines in the distinct gymnophthalmid lineages. Nonetheless, one must emphasize that the considerable variation reported regarding size, shape, and number of ornaments, represents important taxonomic information and will be addressed in more appropriate contexts of taxonomic and/or systematic approaches.
Comments on the Hemipenial Morphology of the Gymnophthalmidae
The results presented below represent our own observations combined with a collection of literature data regarding calcified structures (e.g., spines and spicules) in gymnophthalmid lizards and in five other squamate groups with limb-reduced representatives (i.e., Serpentes, Amphisbaenia and the lizard families Anguidae, Anniellidae, Dibamidae, and, Scincidae).
We found no evidence of the presence of calcified hemipenial structures being exclusive to limb-reduced species among the 46 gymnophthalmid genera sampled (out of the 47 valid genera so far). Enlarged hemipenial body spines occur both in genera with well-developed limbs (e.g., Echinosaura) and in limb-reduced taxa (e.g., Calyptommatus); in contrast, poorly ornamented or even virtually nude hemipenis occur in genera, such as Amapasaurus and Ecpleopus, which have well-developed limbs, as well as in some representatives of the long-tailed genus Bachia, which displays an extreme degree of limb reduction.
Fully ornamented hemipenis, with large and hook-shaped body spines, as well as series of minute calcified spicules associated with hemipenial flounces, are present in the genus Echinosaura (three species examined herein) (Fig. 1; Uzzell, 1965) and in Teuchocercus keyi. Judging from their shape and general organization, it is possible that such structures are functionally associated with the reinforcement of couple-anchoring, independent of the role of limbs during the sexual act. Similar examples among gymnophthalmids are the genera Cercosaura, Placosoma, Potamites, Riama, and Proctoporus, as well as some species of Arthrosaura (A. reticulata, A. kocki) Colobosaura (C. modesta), and Leposoma (L. scincoides, L. southi and L. rugiceps).
Conversely, in the monotypic taxon Rachisaurus brachylepis (Rachisaurinae) as well as in all the representatives of the limb-reduced genus Bachia (Cercosaurinae: Bachiini; 12 species examined) present very poorly ornamented hemipenes, lacking spines, spicules, and any vestiges of calcified structures (Figs. 2, 3), [see also Presch (1978), Rodrigues et al. (2008), and Teixeira Jr. et al. (2013)]. In Bachia, nude hemipenial flounces occur in B. cophias, B. flavescens, B. heteropa, and B. scolecoides, whereas the organs of B. bresslaui, B. dorbignyi, B. oxyrhina, and B. scaea are completely unornamented.
The tribe Gymnophthalmini (subfamily Gymnophthalminae; Pellegrino et al., 2001; Castoe et al., 2004; Rodrigues and Dos Santos, 2008) includes genera with variable degrees of limb reduction. The fossorial and remarkably limb-reduced taxa Calyptommatus, Nothobachia, and Scriptosaura, have enlarged and hook-shaped calcified hemipenial spines (Figs. 4, 5). Calcified spines are also present in the four-limbed genus Psilophthalmus, although not as developed as those of the previously mentioned Gymnophthalmini. The remaining gymnophthalmini genera, with limb morphology similar to that of Psilophthalmus, bear no hook-shaped hemipenial spines, but exhibit different ornamentation patterns such as finger-like and noncalcified papillae (e.g., Gymnophthalmus vanzoi and G. pleei), flounces modified into calcified laminae (e.g., Gymnophthalmus speciosus, Tretioscincus agilis, T. bifasciatus, and T. oriximinensis), or completely nude and roughly circular flounces (e.g., Micrablepharus spp., Procellosaurinus spp., and Vanzosaura rubricauda).
The genera Heterodactylus and Colobodactylus both allocated in the tribe Heterodactylini (following Pellegrino et al., 2001), together with the genus Caparaonia (Rodrigues et al., 2009), represent another limb-reduced radiation within Gymnophthalmidae. However, limb reduction in this group is not as remarkable as is in some Gymnophthalmini and in the tribe Bachini, and the Heterodactylini species show only minor reductions in phalanx number and/or size.
Among the Heterodactylini, the presence of calcified structures is subjected to intrageneric variation. Minute calcified spines are detected on the extremities of some flounces of the hemipenial body of Heterodactylus imbricatus (Fig. 6A,B), whereas the hemipenis of H. lundi (Fig. 6C,D) lack any traces of calcified structures. Similarly, Colobodactylus taunayi shows minute spines on the hemipenial flounces (Fig. 7A,B), while C. dalcyanus has a virtually nude hemipenis (Fig. 7C,D). We found no calcified structures in Caparaonia itaiquara.
Calcified spicules on the hemipenial body flounces were detected in the five Iphisini genera (sensu Rodrigues et al., 2009), which also present minor reductions of limbs, mainly represented by a decrease in phalanx size. In these genera, the hemipenial spicules are organized in rows along the flounces, in an arrangement referred to in the literature as “comb-like” flounces (Myers and Donnelly, 2001; Rodrigues et al., 2005, 2007).
Among the tribe Ecpleopodini, Anotosaura spp., Dryadosaura nordestina [sister taxa of Anotosaura vanzolinia, according to Rodrigues et al. (2005)], and Colobosauroides cearensis (sister taxon of the clade composed by the two former genera) (Pellegrino et al., 2001; Castoe et al., 2004; Rodrigues et al., 2005) exhibit stout and short limbs when compared to the remaining Ecpleopodini genera [i.e., Arthrosaura, Ecpleopus, Leposoma, and probably Amapasaurus, Kaieteurosaurus, Marinussaurus, and Pantepuisaurus (Pellegrino et al., 2001; Kok, 2005, 2009; Peloso et al., 2011)]. Despite differences in size and position of the hemipenial lobes, as well as in the general shape of the hemipenial body, the hemipenes of Anotosaura vanzolinia (Fig. 8), Dryadosaura nordestina (see Rodrigues et al., 2005) and Colobosauroides cearensis share the presence of small and isolated calcified spines at the margin of the transversal body flounces. Dryadosaura nordestina also presents series of mineralized comb-like spicules throughout the body flounces, an unusual characteristic among members of the tribe, which is only shared with three species of Arthrosaura (A. reticulata, A. synaptolepis, and A. montigena). The recently rediscovered species Anotosaura collaris differs from the congener A. vanzolinia by exhibiting the hemipenis ornamented with nude flounces lacking any vestige of calcified structures (see Rodrigues et al., 2013). Conversely, the Ecpleopodini taxa with well-developed limbs Adercosaurus vixadnexus (sensu Myers and Donnelly, 2001), Amapasaurus tetradactylus, and some species of the Leposoma parietale group (sensu Ruibal, 1952; Rodrigues, 1997) exhibit poorly ornamented hemipenes, with nude flounces sparcely distributed along the hemipenial body, whereas Ecpleopus gaudichaudii has completely nude hemipenes (Uzzell, 1969; Myers et al., 2009; Nunes, 2011). In contrast, species of the Leposoma scincoides group and some species of Arthrosaura (A. guianense and A. kocki) show isolated calcified spines on the border of each hemipenial body flounce.
Comments on the Hemipenial Morphology of Other Limb-Reduced Squamata
Among Serpentes, the presence of calcified spines and spicules are confirmed only in the advanced and highly diverse clade Colubroides (sensu Zaher et al., 2009). Hemipenial spines of snakes are scattered throughout the hemipenial body and vary greatly in size. Some authors suggest that such structures are functionally related to couple-anchoring, as well as to the positioning of the hemipenis for sperm transfer (Pope, 1941; Edgren, 1953; Pisani, 1976; Murphy and Baker, 1980).
The direct examination of several colubroid genera (e.g., Apostolepis, Boiruna, Erythrolamprus, Helicops, Hydrops, Philodryas, Pseudoeryx, Sordellina, Xenodon—see Supporting Information Appendix I for a list of the species analyzed), combined with literature data (e.g., Mao et al., 1984; Zaher, 1999; Harvey and Embert, 2008; Silva and Rodrigues, 2008; Guo et al., 2009; Klaczko et al., 2010; Montingelli et al., 2011; Zaher et al., 2012) corroborates the widespread occurrence of hemipenial spines in the lineage. In contrast, more basal radiations of the Serpentes, such as Cylindrophiidae (Stuebing, 1994), Leptotyphlopidae (Passos et al., 2005, 2006; Pinto and Curcio, 2011; Pinto and Fernandes, 2012), Boidae (Passos and Fernandes, 2008), and Tropidophiidae (Stull, 1928; Gibson, 1970; Curcio et al., 2012) seem to lack calcified ornaments. Hemipenial ornamentation in basal snake lineages is mostly represented by flounces, papillae, and/or calyces, as well as modified structures that may also aid in couple-anchoring (e.g., the “heart-shaped” papillate structures described in some tropidophiids; see Curcio et al., 2012). The report of seven to nine spines on the asulcate face of the hemipenes of Typhlops reticulatus (see Dixon and Hendricks, 1979) demonstrates that poorly studied snake taxa (e.g., Scolecophidia) deserve further attention regarding hemipenial morphology.
The general structure of amphisbaenian hemipenes is much simpler than that of the vast majority of other squamates. The direct examination of the hemipenis of Amphisbaena brasiliana (Fig. 9), A. cuiabana, A. mertensi, and A. microcephala, coupled with sparse literature information (Rosenberg, 1967; Broadley et al., 1976; Gans, 1978; Böhme, 1989; Rosenberg et al., 1991; Pinna et al., 2010; Pinna, 2012), brings no evidence of calcified structures. Ornamentation of amphisbaenian hemipenes are either entirely absent, resulting in completely nude organs [e.g., Amphisbaena carli (Pinna et al., 2010), A. bolivica (Pinna, 2012) and Bipes canaliculatus (Böhme, 1989)], or is restricted to the presence of flounces on the body and/or lobes [e.g., A. caudalis (Thomas and Hedges, 2006), A. microcephala, A. trachura (Pinna, 2012), Monopeltis galeata (Cope, 1896), Trogonophis wiegmanni (Böhme, 1989)]. Some reports in the literature point the presence of hemipenial lobes distinctly curved or perpendicular to the hemipenial body, such as that seen in A. brasiliana (Fig. 9; Pinna, 2012), A. cuiabana (Strüssmann and Carvalho, 2001), A. innocens (Rosenberg et al., 1991), A. schmidti (Rosenberg, 1967), and Blanus cinereus (Rosenberg et al., 1991). In other instances, there are reports of lateral papillae positioned as secondary lobes, as is the case for A. kingi (Rosenberg, 1967; Vanzolini, 1999). Lamellate tips are documented in some species of Amphisbaena (Rosenberg et al., 1991; Thomas and Hedges, 2006; Pinna, 2012), Dalophia pistillum (Broadley et al., 1976), and Monopeltis sphenorhynchus (Rosenberg et al., 1991).
Hemipenes of some limb-reduced anguids (e.g., Ophisaurus spp., Ophiodes intermedius, and Anguis fragilis) have been previously described or illustrated (Böhme, 1988). Those of A. fragilis present spine-like structures, but it is not possible to recognize the presence of calcified structures only by examining the illustrations provided in Böhme's work (1988).
The analyses of hemipenes of Ophiodes fragilis (a limb-reduced anguid; Fig. 10) and Diploglossus fasciatus (an anguid lizard with no reduction of limbs; Fig. 11) confirmed the presence of pleated hemipenial flounces (Cope, 1896; Thomas and Hedges, 1998; Balej and Jablonski, 2006) in both taxa, but no calcified structures were detected after Alizarin Red staining (not performed in any of the previous studies). Therefore, the presumed presence of spines [Thomas and Hedges (1998); “recurved osseous spines”, sensu Cope (1896); “spinous hemipenis”, sensu McDowell and Bogert (1954)] was not supported by our data.
In Diploglossus, pleated flounces cover the entire hemipenis [Fig. 11; Thomas and Hedges (1998 – Fig. 4)], whereas in Ophiodes fragilis they are restricted to lobes and to the distal end of the hemipenial body (Fig. 10). Regardless of the absence of spines or spicules, the general structure of the flounces is highly suggestive of a couple-anchoring role during mating.
Literature data on annielid hemipenes are scarce: Coe and Kunkel (1904) illustrated the partly everted organ of an embryo of Anniella pulchra, and Cope (1896) provided a brief description of the hemipenis of an unidentified Anniella specimen. We were unable to properly prepare the hemipenis of the single specimen of A. pulchra (MZUSP 2227) available for examination; alternatively, we performed a longitudinal dissection of the organ that allowed us to examine its external surface. The Alizarin Red staining failed to show any particular calcified structures in this specimen, suggesting that such elements must be absent. In addition, like Coe and Kunkel (1904), we were unable to detect vestiges of other ornaments, such as the wrinkled body flounces described by Cope (1896). However, we regard our data as inconclusive due to the absence of a well-prepared hemipenis.
The data on hemipenial morphology of dibamids rely strictly on a brief description of the hemipenis of the holotype of Dibamus greeri (Darevsky, 1992); in addition, the organ is not fully everted and no staining tests were performed for detecting calcareous structures. The illustrations of the anal region of D. greeri (Darevsky, 1992: Figs. 8, 11) suggest a very reduced organ (1.1 mm length and 0.4 mm width, according to the original description), and the absence of ornaments is inconclusive given that eversion is not complete. Hence, further analyses are needed in order to access the diversity of hemipenial morphology in dibamid lizards.
There are virtually no descriptions of hemipenes among the more than 15 remarkably limb-reduced skink genera (according to Miralles et al., 2012). The single record on hemipenial morphology of a limb-reduced skink relies on a brief characterization and a single illustration of the hemipenis of Lerista terdigitata (Greer, 1979: Fig. 15), as well as minor additional information on other Australian skinks given in the same paper. The illustration provided by Greer (1979) reveals a completely nude organ with no evident ornaments in the hemipenial body or lobes. We had access to a male specimen of the limbless Feylinia currori (MZUSP 97826), but the small size of its hemipenis prevented proper preparation. Nonetheless, the dissection of the organ and the staining test revealed no traces of calcified structures, following the condition seen in limb-developed skinks for which hemipenial morphology is known (Cope, 1896; Greer, 1979; Linkem et al., 2011; C. Linkem pers. comm.).
Despite scattered evidence on anguid lizards (Böhme, 1988) and scolecophidians (Dixon and Hendricks, 1979), our results reinforce the assumption that calcified ornaments are circumscribed to advance snakes (Colubroides) and gymnophthalmids. Moreover, the significant degree of limb reduction seen in both lineages is tempting to associating this feature with the presence of calcified spines as an apomorphic trait related to couple-anchoring, following Presch's hypothesis (Presch, 1978). However, our comprehensive sampling of the Gymnophthalmidae, combined with data for other squamate groups, suggests that such an association may be exceedingly simplistic.
Even in the absence of specific morpho-functional approaches, the general structure of calcified hemipenial ornaments is, in itself, highly suggestive of a couple-anchoring role. Nonetheless, if the presence of calcified structures was in any way phylogenetically associated with limb reduction, one should expect their predominant occurrence in the limb-reduced squamate taxa.
Observations from the present study, together with most literature data, do not support such a relationship. Basal snake lineages, all known amphisbaenians, and several limb-reduced lizards lack any traces of calcified spines or spicules. In contrast, extremely developed, hook-shaped and calcified spines are present in gymnophthalmid genera with well-developed limbs, such as Teuchocercus and Echinosaura (Fig. 1). In addition, despite lacking spines, the pleated flounces of anguid lizards are also suggestive of couple-anchoring and are present both in the limbed genus Diploglossus (Fig. 11) and in the limb-reduced genus Ophiodes (Fig. 10), suggesting that the development of couple-anchoring systems are by no means phylogenetically associated with particular events of reduction/loss of limbs.
The presence of hemipenial spines was already considered synapomorphic for the Gymnophthalmidae (Estes et al., 1988), thus rendering the absence of calcified ornaments in the several taxa analyzed here as isolated instances of secondary losses. However, distinct and more parsimonious optimizations are possible in light of the two most recent and comprehensive phylogenies of the family (Pellegrino et al., 2001; Castoe et al., 2004). The topologies of Pellegrino et al. (2001) and Castoe et al. (2004) are congruent regarding the placement of the subfamily Alopoglossinae as the sister-group to all other gymnophthalmids. Since alopoglossine species have well-developed limbs and no vestiges of calcified hemipenial structures, Este's hypothesis of the spines being present in the most recent common ancestor of the Gymnophthalmidae should be set aside.
Considering the topology obtained by Pellegrino et al. (2001), the presence of calcified structures may represent either a synapomorphy for a less inclusive clade within the family (i.e., Gymnophthalmidae except Alopoglossinae) with independent reversions to the plesiomorphic condition, or a homoplastic character acquired at least twice within the family (Fig. 12A). Alternatively, considering the topology presented by Castoe et al. (2004) implies in assuming the homoplastic appearance of calcified hemipenial structures in at least three clades, each containing one single reversal event (Fig. 12B). Regardless of these scenarios, none of the topologies support a direct phylogenetic association between the presence of calcified hemipenial structures and limb reduction, because limb-reduced taxa (i) are nested within terminals represented by species with spiny hemipenes and well-developed limbs (e.g., Heterodactylus imbricatus, and Colobodactylus taunayi); and (ii) may present no calcified hemipenial structures at all (e.g., Bachia spp, Rachisaurus brachylepis, Heterodactylus lundii, and Colobodactylus dalcyanus).
Within the Gymnophthalminae, interspecific variation regarding the presence of calcified hemipenial spines is seen in the genera Heterodactylus and Colobodactlylus. Calcified spines are present in the hemipenial body flounces of H. imbricatus and C. taunayi, whereas H. lundii and C. dalcyanus lack such structures, presenting no vestiges of calcium in the organs. Despite being considered representatives of a limb-reduced lineage among the Gymnophthalmidae (Pellegrino et al., 2001), these species show minor degrees of limb reduction (reductions in size or loss of individual phalanges). These minor changes are unlikely to affect limb functionality, in contrast to the snake-like species, which exhibit more drastic reductions. Thus, intrageneric variation regarding the presence or absence of calcified spines in species with similar body architecture, like that observed in the genera Heterodactylus, Colobodactylus, and Leposoma, reinforces the lack of a relationship between hemipenis ornamentation and limb development, and suggests that other historical factors might be more strongly associated with the evolution of hemipenial structure.
The subgroup of gymnophthalmini lizards formed by the limb-reduced genera Calyptommatus, Nothobachia, and possibly Scriptosaura (Rodrigues and Dos Santos, 2008), and by the four-limbed Psilophthalmus (included in this group in the topology of Castoe et al., 2004), is unique within the subfamily for presenting distinctly developed hook-shaped and calcified hemipenial spines. In contrast, the other genera included in Gymnophthalminae exhibit reduced comb-like spicules in the hemipenes or organs completely lacking calcified structures. The enlarged spines of the limb-reduced gymnophthalminis might represent an adaptation to limb-reduction and/or to psammophilic environments, considering that all of these genera have sand-dwelling habits. Similar hook-shaped spines on the basal region of the hemipenial body are also found in other psammophilic squamates such as the snake species Apostolepis longicaudata (Curcio et al., 2011).
Reports of the seasonal presence of micro-ornamentations such as minute spines, hooks or multi-spined tubercles in hemipenial ridges of sexually active lacertids (Arnold, 1986b), a family with no limb-reduced taxa, support the assumption that these anchoring structures must have arisen independently multiple times within Squamata. However, the presumed seasonal development of calcified hemipenial structures deserves further investigation.
The results of this study do not refute the putative functional role of calcified hemipenial ornaments for couple-anchoring. Nonetheless, the wide occurrence of these structures in several gymnophthalmids with well-developed limbs, combined with their virtual absence in amphisbaenians, several snake lineages and limb-reduced gymnophthalmid genera demonstrates that such structures are not necessarily associated with particular events of limb-reduction.
Hemipenial ornaments seem to be related to couple-anchoring in some limbless squamate lineages, but certainly do not represent the single mechanism that keeps male and female attached during mating. For example, in many snakes and lizards the tail can aid in this process (e.g., Tryron, 1979; Crews and Fitzgerald, 1980). However, the poor knowledge of reproductive behavior of squamates deter further functional inferences about the role of limbs or hemipenial ornamentation during mating.
We are grateful to D. Frost and D. Kizirian (AMNH); K. de Queiroz, R. McDiarmid, R. Heyer and G. Zug (USNM); W. Duellman and L. Trueb (KU); M. Altamirano (DHMECN), P. Kok (IRSNB); J. H. C. Santa Gadea (MHNSM); G. Rivas and T. Barros (MBLUZ); L. Coloma and O. Torres-Carvajal (QCAZ); A. L. Prudente and T. C. Ávila-Pires (MPEG); J. Hanken and J. Rosado (MCZ); E. La Marca (ULABG); D. Borges-Nojosa (UFC); A. Resetar and K. Kelly (FMNH); T. Mott (UFAL); and H. Zaher and C. Castro-Mello (MZUSP) for providing access to specimens under their care. We are also grateful to M. Sena, M. Teixeira-Jr, R. Gaiga, T. Mott, C. Strüssmann, S. Ruane and J. Cassimiro for making available some of the pictures of live specimens that illustrate this work, to Mary Andriani for revising the English writing and to P. Pinna for providing unpublished information on amphisbaenian hemipenial morphology. We would like to thank to S. J. Sánchez-Pacheco and to an anonymous reviewer for their important comments, which helped in improving the manuscript. Finally, we are specially grateful to Juan D. Daza and Scott Miller for the invitation to take part in this volume and for the contributions during the revision process of the manuscript. The work was funded by Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq).