Morphological and biological variations of the glass snake Ophiodes striatus (Spix, 1825) (Squamata, Anguidae) from southeastern Brazil

: Lizards of the Ophiodes genus, popularly known as glass snakes, are exclusively Neotropical. Despite their wide geographic distribution, glass snakes are poorly studied and there are many knowledge gaps in their biology and ecology to be filled. In this context, the objective of this study was to analyze possible sexual and population morphological differences of Ophiodes striatus specimens collected from different municipalities in the state of Minas Gerais, Brazil. A total of 63 specimens belonging to the Herpetological Collection of the Federal University of Juiz de Fora - Reptiles (CHUFJF-Reptiles), collected between 1995 and 2015 were analyzed. All the specimens were necropsied and morphometric measures related to body and head were measured. The males and females analyzed showed similar morphometric measures, and there was no size-related sexual dimorphism. Principal component analysis did not group the analyzed specimens by sex or locality. A positive correlation was observed between the snout-vent length and the vestigial limb length. The present study brings new information related to sexual dimorphism and population morphometric variation of O . striatus , adding information that will help in better understanding the ecology, biology and evolution of the genus in Brazil.


MATERIALS AND METHODS
A total of 63 O. striatus specimens of the Coleção Herpetológica de Répteis da Universidade Federal De Juiz De Fora -Reptiles (CHUFJF-Répteis) (Appendix 1) were used in the present study. The lizards were collected in fields surveys between 1995 and 2015, in the cities of Juiz de Fora, Santa Bárbara do Monte Verde, Chácara, Lavras, Ingaí, Ijaci, Ribeirão Vermelho, Coronel Xavier Chaves, Ritápolis, São João del Rei and Tiradentes, in the state of Minas Gerais (Brazil).
The sexual determination for adults and sub-adults was performed through macroscopic observation of gonadal traits by necropsy. Specimens which had body length between 47.24 mm and 63.19 mm were considered juveniles and were not necropsied; therefore, sexual identification was not performed.
The external morphometric measurements calculated were: snout-vent length (SVL), tail length (TL), length of the vestigial hind limbs (LVL -measured from left side), distance between nostrils (DN), face commissure (FC), head height (HH) and head width (HW), as shown in Figure  2. Due to the curvature of the animal's body, linear measurements of the body (SVL) and tail (TL) were measured with the line of a string. It was placed on the animal ( Figure 1) and then extended to determine the size in millimeters using a digital caliper. Tail length was measured on all specimens, including those having regenerated or fragmented tails (the portions being measured and then summed to define the total value of the tail length).
Some descriptive statistics were investigated such as the mean, standard deviation, and amplitude (maximum = M and minimum = m) for each variable, which were calculated by analyzing the measurements of males, females and juveniles of the whole sample. The size variation of the specimens belonging to most geographically distant (Juiz de Fora and Ritápolis) was also analyzed, aiming to investigate the population variation since these populations.
Principal Component Analysis (PCA) was performed in order to analyze whether the sample would be organized into groups. PCA was done to find out if the morphometric data showed groupings of the sample by sex (male x female) and by location (Juiz de Fora x Ritápolis). Principal Component Analysis (PCA) was performed to determine if variables influenced the formation of groups based on sex and location. PCA was performed separately for body variables (SVL, TL and LVL; n = 50) and head (DN, FC, HH and HW; n = 45). Both analyzes were performed to determing whether these variables influenced grouping by sex or location. The Pearson Correlation test was performed to verify the correlation between the morphometric data. Both statistical tests were performed in Statistical Program R, version 3

RESULTS
Fifty-five (55) of the 63 O. striatus specimens belonging to CHUFJF-Reptiles were young/adult, while eight were juveniles. Among the young/ adult specimens, it was possible to identify the sex in 50 specimens (31 males and 19 females).
From the 19 necropsied females, four were pregnant with embryos at some gestational stage, and the specimens were collected between June and September. The morphometric measures were only verified in the individuals that were in an advanced gestational stage, with complete formation ( Table I). The mean litter size was 4.50 (± 1.26, n = 4). The specimen that had the largest litter (n = 6) had SVL of 196.18 mm, while the smallest litter (n = 3) had SVL of 192.03 mm.
The mean SVL of males and females was similar, and no apparent differences were observed between their sizes (Table I). A similar result was observed when the analysis was performed with the specimens belonging to the two most geographically distant (Juiz de Fora and Ritápolis) (Table II).
There was no sex grouping in either PCA (Figure 2a, b). PC1 explained 51.1% of the variation observed, and PC2 explained 31.8%, with the accumulated proportion of 82.82% when using body measures (Figure 2a). When using head measures, PC1 explained 57.1% and PC2 17.5% of the variation observed, with a cumulative proportion of 89.35% (Figure 2b).
According to the PCA data, it was also not possible to separate the populations from Juiz de Fora (n = 10) and Ritápolis (n = 17), indicating that both populations have similar size variation (Figure 3). PC1 explained 92.3% of the variation found in the sample, while PC2 only explained 5.5% (cumulative ratio = 97.8%) ( Figure 3).
As there was no grouping of males and females in PCA, Pearson's correlation test was performed with the whole sample without distinction by sex. The correlation indicated a positive and moderate relationship (r = 0.46) between SVL and LVL (Figure 4), meaning that the higher the SVL, the higher the LVL.

DISCUSSION
Ophiodes striatus morphometric data analysis indicated the non-existence of sexual dimorphism associated with body size. However, the studies by Barros & Teixeira (2007) and Montechiaro et al. (2011) with with different populations of O. striatus from Brazil (Espírito Santo and Rio Grande do Sul, respectively) recorded significant differences between females and males, indicating a possible dimorphism related to body size (SVL) and of the tail (TL). The females were significantly larger than males in both studies, which is common in many lizard species belonging to different families (Pianka & Vitt 2003). Our results on the existence of sexual size dimorphism in O. striatus stand in contrast to that of previous studies and suggest that sexual size dimorphism cannot be used to distinguish all populations of O. striatus. Analyzes using other parameters, besides the variables used in this study, may be an alternative to clarify the determination of external sexual dimorphism in this species.
No difference was observed in the tail length between the sexes, differing from that found by Montechiaro et al. (2011). We point out that this could be due to many males having undergone autotomy. Males with longer tails were expected according to the Morphological Restriction hypothesis of King (1989), as longer tails in males are usually related to the accommodation of hemipenis and its retractor muscles (Montechiaro et al. 2011). No difference was found in measures related to the head (labial commissure and head width). However, males with larger heads than females may be related to the aggressive interactions between males (Blackburn & Vitt 1992); moreover, males and females have different eating habits, preferentially feeding on different prey items (Montechiaro et al. 2011).
The study of vestigial limb size variation is not common, despite its great evolutionary and adaptive meaning for the genus. Montechiaro et al. (2011) also observed that males had larger vestigial members than females, which was not observed in the analyzes of this study. However, body length (SVL) and vestigial hind limbs (LVL) are positively correlated; meaning that the limbs grow simultaneously along with the body, being this fact an indication of functionality of this structure. This characteristic may not only be related to the use of these structures for sexual stimulation during mating, but also to the use in fighting between males (Carpenter et al. 1978, Montechiaro et al. 2011 (2007) observed a mean clutch size of 6.0 (ranging from three to eleven embryo/fetuses). In O. striatus specimens studied by Montechiaro et al. (2011) belonging to different museums and collections of Rio Grande do Sul (Rio Grande do Sul, Brazil), the mean clutch size was 7.96 (n = 51), mean SVL was 51.64 mm (n = 22). The mean litter size calculated for O. striatus of CHUFJF-Reptiles was 4.50 (n = 18), lower than that observed in the previously described studies.
Although a separation between the Juiz de Fora and Ritápolis populations was expected, this was not found (Figure 4). Morphological variations in natural populations are driven by evolutionary forces (Oliveira et al. 2016), and given that these populations are geographically distant, similar sizes of measured traits of individuals within these populations indicates their consistency as distinct populations of the same species. However, this fact will only be confirmed by complementary studies, such as genetic studies.
The present study presents new information on the biological characteristics of O. striatus. Despite the diversity of analyzed locations, no significant variation was observed. The results showed that there is no sexual dimorphism related to size, although it has been observed in previous studies. Our study indicated that the use of morphometric analysis of external morphology may not be applicable to sexual dimorphism in some species of reptiles such as the present study, since the external morphology may be influenced by different biotic and abiotic factors.