First records in Brazil for Arthrosaura versteegii van Lidth de Jeude , 1904 ( Reptilia , Gymnophthalmidae ) , a possible species complex , and distribution extension for Arthrosaura montigena Myers & Donnelly , 2008 , in Venezuela

Arthrosaura versteegii van Lidth de Jeude, 1904 is known to occur on the Guiana Shield from eastern Venezuela to western French Guiana, between 100 m and 1400 m a.s.l. No records for Brazil are known. Herein, we report specimens morphologically similar to A. versteegii from 2 areas in Brazilian Amazonia, south of the Amazon river. However, the Brazilian specimens exhibit 2 distinct types of hemipenes that also differ from that of Guianan males of A. versteegii. Moreover, the reexamination of 2 specimens of A. versteegii collected in the Venezuela highlands show that they are A. montigena Myers & Donnelly, 2008, leaving A. versteegii restricted to the eastern Amazonian lowlands.


Introduction
Van Lidth de Jeude (1904) described Arthrosaura ver steegii based on a single specimen (RMNH 4469) from the Cottica Mountains ("Suriname") in extreme western French Guiana, just east of the Marowijne river that forms the border between Suriname and French Guiana.The specimen had been collected by a Dutch topographical expedition and Van Lidth de Jeude apparently had not realized that the Cottica Mountains were outside Suriname.Van Lidth de Jeude (1904) and Burt and Burt (1931) noted the similarities between A. versteegii and A. reticulata (O'Shaugnessy, 1881) from Ecuador.Bron-gersma (1932) reviewed all species which at that time were considered to belong to Arthrosaura and suggested that A. versteegii might be a subspecies of A. reticulata.However, he had no specimens of A. reticulata for direct comparison.Brongersma (1935) compared the holotypes of A. reticulata and A. versteegii (the only specimens of both taxa known at that time) and came to the conclusion that the differences between the 2 species were rather on a subspecific level, so that they should be considered subspecies of 1 species.Thus, the taxon described by Van Lidth de Jeude (1904) became A. reticulata versteegii.Cunha (1961) followed Brongersma's (1935) conclusion, but later, Cunha (1967) described A. amapaense and con-

Methods
The MPEG collection, which has a good representation of specimens from Brazilian Amazonia, was thoroughly searched for additional A. versteegii-like specimens.We reexamined the specimens of A. versteegii from the Guianas, including the Venezuelan specimens of A. versteegii (RMNH 25261-62) reported by Hoogmoed and Avila-Pires (1992).Following the suggestion of Myers and Donnelly (2008), we compared these specimens with the paratype of A. montigena (AMNH R-140230).Material (including types) of A. versteegii and A. montigena from the following musea (museum acronyms follow Sabaj 2016) were studied and are listed in Table 1: AMNH, EBRG, MHNLS, MNHN (Paris), MPEG and RMNH.Specimens from EBRG and MHNLS from Venezuela were examined based on photos, all others directly.Coordinates and altitude may be approximations, secondarily obtained based on available locality data.We recovered data from field notes, in order to obtain an idea of the habitat of the species.
In addition, we examined the hemipenes of some of these specimens from both north and south of the Amazon River.We prepared hemipenes of A. versteegii (MNHN 1999.4910,from French Guiana, previously depicted and described by Gasc, 1981 while still on the specimen) and of the morphologically similar specimens from south of the Amazon (MPEG 24950, from Vitória do Xingu, Pará, Brazil;MPEG 28440 and MPEG 28443, from Novo Progresso, Pará, Brazil).For comparative purposes, we also prepared hemipenes of A. reticulata (RMNH 25265 from the Lucie river, Suriname, north of the Amazon river; MPEG 25967 from Parauapebas and MPEG 32008 from Vitória do Xingu, both in Pará, south of the Amazon river), and we used data from Myers and Donnelly (2008) and Nunes (2011) on the hemipenis of A. montigena.Hemipenes were removed and prepared following the procedures described by Pesantes (1994), Manzani and Abe (1988) and, for eversion and filling, Zaher and Prudente (2003).For coloration of mineralized structures, we used an alcohol solution of Alizarin Red (Uzzell 1973).Hemipenial terminology follows Myers and Donnelly (2008).
The Brazilian localities, from where A. versteegii was not previously known, are all well south of the Amazon River, and the southernmost locality (W of Castelo dos Sonhos) is about 1200 km due south of the 2 southernmost localities (Sipaliwini, Suriname, and Koulimapopane, French Guiana) in the Guiana region (Table 1, Fig. 1).Thus, the presently known distribution covers part of the eastern lowlands of the Guiana Shield and reaches far south of the Amazon between the rivers Tapajós and Xingu, nearly reaching the northern border of the Brazilian Shield.In the Belo Monte area, A. versteegii was only collected on the west bank of the Xingu river, apparently being absent from the east bank of that river.
In Venezuela, 2 specimens (RMNH 25261-62) previously identified as A. versteegii by Hoogmoed and Avila-Pires (1992) are here re-identified as A. monti gena (see below).This species was previously known only from Auyántepui, approximately 140 km to the northwest.In addition, MHNLS 11168, from a locality about 60 km north of that of the RMNH specimens and reported by Gorzula and Señaris (1998) as A. versteegii, seems also to be A. montigena.These data indicate that A. versteegii occurs only in eastern Amazonian lowlands, while A. montigena occupies the uplands and highlands of southeastern Venezuela.
Arthrosaura versteegii was found in Suriname both in savanna forest and in high primary forest (Hoogmoed and Avila-Pires 1992).The Brazilian material was collected in forested areas, mostly in pitfalls.In Castelo dos Sonhos and the locality near the airport of Novo Progresso, W of the road BR-163, specimens were caught in high, original to lightly logged rainforest.In the Novo Progresso locality 13 km E of BR-163, specimens were collected in pitfalls in a heavily disturbed open forest on a hillside between a road and pasture.Three of the specimens collected in Vitória do Xingu were said to have been collected in "mata" (i.e.any type of forest), but most likely rainforest (personal observation MSH).The remaining 2 specimens from Vitória do Xingu have no data on habitat, but the coordinates seem to fall in (degraded) forested areas.One hand-collected specimen was found in leaf litter, during daytime.One female (MPEG 28442, SVL 46 mm), collected 27 November 2005, had 2 well-developed eggs in the oviducts.Preanal and femoral pores were small in males with SVL 35-39 mm, and well developed in males with SVL 45-49 mm.
Identification.We found specimens similar to Arthro saura versteegii in the herpetological collection of MPEG identified as A. reticulata and Leposoma percarinatum.Some of those had been reported earlier (Hoogmoed et al. 2007(Hoogmoed et al. , 2008) ) as Ptychoglossus brevifrontalis, others as A. reticulata (Galatti et al. 2012, Leme Engenharia 2012, Vaz-Silva et al. 2015) and Leposoma (= Loxopholis) percarinatum (Müller 1923) (Vaz-Silva et al. 2015).Our comparisons of specimens of A. versteegii from Suriname and French Guiana showed no differences in external morphology with the specimens from the Tapajós-Xingu interfluve.Considering the similarity with A. montigena and the doubts raised by Myers and Donnelly (2008) about the identity of the Venezuelan specimens identified as A. versteegii, we also included A. montigena in our  . versteegii).For locality numbers, see Table 1.Rectangle in the inset shows the area in South America represented in the larger figure .comparisons (Table 2).The 2 species differ in the number of transverse rows of dorsals (at least in the few specimens of A. montigena known), in the shape and contacts of some of the head scales, and in details of the dorsolateral light lines that are present in both species (Fig. 2).We call attention to the fact that differences in head scales are all related: as the prefrontals tend to be longer in A. versteegii, they present a longer medial suture; they contact the loreal, precluding the contact between the frontonasal and first supraocular, giving the prefrontals a pentagonal (not a quadrangular) shape; and they extend further posteriorly, in some cases even reaching the second supraocular, so that the frontal has only a short contact, or no contact at all, with the first supraocular.Specimens from south of the Amazon river agree with A. versteegii from Suriname and French Guiana in all these characters.
On the other hand, RMNH 25261-62, from Venezuela, differ from all specimens of A. versteegii we examined and agree in most characters with A. montigena (Table 2).MHNLS 11168, reported by Gorzula and Señaris (1998) as A. versteegii, also has head scales similar to those of A. montigena.We therefore consider these records as referring to A. montigena.
The hemipenes (Fig. 3) of A. versteegii from French Guiana and A. versteegii-like specimens from Vitória do Xingu and Novo Progresso are similar in the presence of lobes with small protuberances around each apex; sulcate face with a wide nude area; sulcus spermaticus deep, straight, distally divided; plicae in 2 longitudinal rows separated by a narrow space, variably extending into the asulcate face, lined with small spicules.No orificium was observed between the lobes.Besides these common traits, however, we observed several differences between localities, shown in Table 3.The hemipenes from French Guiana and Vitória do Xingu are more similar to each other and to those of A. reticulata and A. montigena, all presenting a globose shape with distinct lobes, and a nude area on the asulcate face that separates a row of plicae on each side (Fig. 3A, B).The hemipenis of specimens from Novo Progresso is roughly cylindrical, with hardly divided lobes and asulcate face covered with continuous plicae (Fig. 3C).They also differ from each other in the number of plicae and some other characters.
In summary, at the moment specimens of A. versteegii are known from 5 localities in Suriname between 100 and 630 m above sea level (a.s.l.), the type locality in French Guiana (Cottica Mountains, maximally 700 m a.s.l.; Hoogmoed and Avila-Pires 1992), and 2 other localities in French Guiana, lower than 500 m a.s.l.(Gasc 1990).South of the Amazon, in Pará, we found Arthrosaura versteegii-like specimens in 3 localities in the municipality of Novo Progresso and 4 localities in the municipality of Vitória do Xingu, at altitudes between 70 and 340 m a.s.l.Differences in hemipenes suggest that they may form a complex of cryptic species, but until more data is available (e.g., hemipenes from other localities, molecular data), we prefer to consider all the specimens as A. versteegii.Below we present a chresonym list for A. ver steegii and A. montigena:

Discussion
Myers and Donnelly ( 2008) presented an identification key for all the species of Arthrosaura, except A. hoog moedi.This latter species differs from both A. versteegii and A. montigena by presenting 4, instead of 3, supraoculars (Kok 2008).Myers and Donnelly (2008), in their description of A. montigena, emphasized its distinctiveness from A. versteegii, the most similar species, but this was done based on the description by Hoogmoed and Avila-Pires (1992); they did not compare specimens directly.By comparing specimens, we did not observe 2 of the differences they pointed out.The first one, concerning the guttural fold, was due to confusion, since we talked about the gular fold, which borders the collar; a guttural fold is indistinct in both species.The second one regards the ventrolaterals of Myers and Donnelly (2008) or lateralmost ventrals of Hoogmoed and Avila-Pires  1992), and their delimitation with the lateral scales.Even though the former authors considered that they are "not or but indistinctly demarcated" and the latter authors that they are "sharply demarcated by a zone of small scales", the 2 species hardly differ in these scales and the contact between them.Two other differences indicated by Myers and Donnelly (2008) seem to vary within the species.In A. versteegii, the interparietal is usually narrower than the parietals, but they are subequal in some specimens, and it is also possible to suppose that A. montigena shows the same variation.The nasal scale is undivided in all specimens of A. versteegii and semidivided in the 2 types of A. montigena.In RMNH 25262, it is also semidivided (division running from naris to supralabial), but in RMNH 25261, a division was not detected; therefore, this character may be variable in A. montigena.Nunes (2011) examined the hemipenis of 5 of the 9 known species of Arthrosaura.In all these species, the hemipenis is globose, bilobed, with a nude area on the asulcate face, as we observed in A. versteegii (not examined by Nunes 2011) from French Guiana and Vitória do Xingu, Brazil, and quite distinct from the hemipenis found in the specimens from Novo Progresso, Brazil.Intraspecific variation in hemipenis has already been observed in lizards and snakes.Myers and McDowell (2014) discuss several cases of hemipenial polymorphism, especially in snakes, demonstrating that it is not possible to assume any explanation a priori as granted.Despite frequently being considered conservative, in some cases hemipenes seem to evolve rapidly, showing large differences within a species or in closely related species.Examples in lizards are the cases of Norops altae (Dunn, 1930) and N. monteverde (Köhler, 2009), presented by Köhler (2009); Norops osa (Köhler, Dehling & Köhler, 2010) and N. polylepis (Peters, 1874), discussed by Köhler et al. (2012); and Iphisa elegans Gray, 1851 (Nunes et al. 2012).Köhler et al. (2012) found that, at least considering sequences of mitochondrial cytochrome b, N. osa forms a clade within N. polylepis, even if their hemipenes differ in several characters, demonstrating the close affinity of these 2 taxa.In the case of the specimens studied here, it is interesting to note that the largest differences were not found between the population north of the Amazon and those south of it, but between the 2 populations in Pará, both in the Xingu-Tapajós interfluve.Irrespective of whether they represent the same or distinct species, considering that the hemipenis of the specimens from Novo Progresso differs from those of most Arthrosaura species, we may suppose that this population may have undergone some strong selective pressure that led to hemipenial divergence, as proposed in other cases by Myers and McDowell (2014).However, without additional studies to give a better picture of the variation in the whole group and how these populations relate to each other, any conclusion would be too speculative.
At least in part of its distribution area, A. versteegii occurs syntopically with A. reticulata.Hoogmoed and Avila-Pires (1992) mentioned both species from the proximities of the airstrip Sipaliwini (1 and 4 specimens, respectively).On the left (west) margin of Xingu river, in the surroundings of UHE Belo Monte, Pará, 8 specimens of A. versteegii and 51 of A. reticulata were collected (while 32 A. reticulata and not a single A. versteegii were collected on the right [east] margin).On the other hand, in Novo Progresso, Pará, the 9 Arthrosaura collected were all A. versteegii, and during recent work in the Calha Norte area of Pará, north of the Amazon, specimens of only A. reticulata were found (Avila-Pires et al. 2010).In French Guiana, all records (see chresonym list) refer to A. versteegii, while the presence of A. reticulata needs to be confirmed.
The recent appearance of A. versteegii in collections from Pará south of the Amazon may be a consequence of a relatively restricted area of occupancy due to microhabitat restrictions (together with an increase in collecting effort), but it can also be related to the use of pitfall traps, which capture species that rarely emerge from the leaf litter and therefore are difficult to detect during active collecting.In this case, the species' apparent rarity could be a result of seclusive habits.A similar case happened with Ptychoglossus brevifrontalis Boulenger 1912, which, for a long time was thought to have a peripheral Amazonian distribution, but more recently was found to occur all over Amazonia (Peloso and Avila-Pires 2010).Lehr (2002: 202), in his list of amphibians and reptiles from Peru, reports A. versteegii from Peru, but it is unclear on which basis he does so.The species is not mentioned in the main text of his book, and he does not provide any literature reference for this supposed occurrence.

Figure 1 .
Figure 1.Map showing known records of Arthrosaura versteegii (circles) and A. montigena (squares).New records are represented by open symbols (in the case of A. montigena, it indicates the locality of the specimens previously identified as A. versteegii).For locality numbers, see Table1.Rectangle in the inset shows the area in South America represented in the larger figure.

Table 1 .
Data on known specimens of Arthrosaura versteegii and A. montigena.

Table 2 .
Comparison between specimens of Arthrosaura dealt with in this paper. A.

versteegii: Suriname & French Guiana A. versteegii: Pará, Brazil A. montigena: Auyántepui, Venezuela † A. montigena: Kavanayen, Venezuela
*Dorsolateral light lineFrom rostral along canthus rostralis and lateral margin of supraoculars to forelimbs, and on base of tail Mainly visible from nape to shortly posterior to forelimbs, and from sacrum to base of tail; fainter between posterior corner of eye and nape, and on middle part of body † Data based onMyers and Donnelly (2008)and examination of the paratype (AMNH R-140230).* Characters that differ between A. versteegii and A. montigena.

Table 3 .
Comparison of the hemipenes of Arthrosaura versteegii-like specimens.