REDESCRIPTION OF ATRACTUS ALBUQUERQUEI (SERPENTES: COLUBRIDAE: DIPSADINAE), WITH COMMENTS ON GEOGRAPHICAL DISTRIBUTION AND INTRASPECIFIC VARIATION

Atractus albuquerquei Cunha and Nascimento, 1983 was previously known from a holotype from eastern Pará, and 15 specimens from Rondônia and Acre, all in Brazil. We report on 23 additional specimens from the Brazilian states of Rondônia, Goiás, Mato Grosso, and Mato Grosso do Sul. These specimens extend the known range of A. albuquerquei substantially, and beyond the southern limits of the Amazon basin. The holotype of A. albuquerquei is redescribed and intraspecific variation in external morphology, hemipenes, and colour is documented. Sexual dimorphism exists in total length, and number of ventral (significantly greater in females) and subcaudal scales (greater in males). There is a significant correlation between number of subcaudal scales and longitude (decreasing from East to West) for both males and females.


INTRODUCTION
Species of the neotropical colubrid snake genus Atractus are known to be poorly represented in museum collections (Myers, 2003), although there are exceptions (e.g.Atractus reticulatus).As a result, knowledge of the natural history, morphological variation, and geographical distribution of most species of Atractus is far from satisfactory.Atractus albuquerquei is one of the many species of this genus that has been recorded only sporadically after its formal description by Cunha and Nascimento (1983).These authors based their description on a single specimen from Vila Nova, a locality near the river Timboteua, in the vicinity of Paragominas, in the eastern part of the state of Pará, Brazil.
Since its original description, A. albuquerquei has been mentioned only three times in the literature (Vanzolini, 1986;Nascimento et al., 1988;Jorge da Silva, 1993).Vanzolini (1986:25) reported the occurrence of the species in the Brazilian state of Rondônia, based on eight specimens collected in the localities of Nova Brasília, Santa Bárbara, Jaru, and Nova Colina (only seven specimens are deposited in the Museu de Zoologia da USP).Nascimento et al. (1988) also reported on a herpetological collection from the state of Rondônia, where two other specimens of A. albuquerquei were collected in the vicinities of Ji-Paraná and Ouro Preto d'Oeste.Additionally, these authors mentioned the existence of another specimen collected near Rio Branco, state of Acre, and deposited in the collection of the Universidade Federal do Ceará (Nascimento et al., 1988:44).Jorge da Silva (1993) reported four specimens from the area of the Samuel hydroelectric power plant, near the city of Porto Velho, state of Rondônia.Except for the holotype, which is from the easternmost part of the state of Pará, in the Amazonian basin, all 15 specimens mentioned above are from the states of Rondônia and Acre, on the western border of Brazil.This results in a disjunct distribution along the southern border of the Amazonian basin, with a wide gap between Rondônia and eastern Pará.
Here we report 23 additional specimens of Atractus albuquerquei from the states of Rondônia (eight specimens), Goiás (nine), Mato Grosso (five) and Mato Grosso do Sul (one) (see list of specimens examined for details).These new records extend significantly the distribution of the species beyond the southernmost limits of the Amazon basin.We also redescribe the holotype of the species, and discuss patterns of intraspecific variation.

MATERIALS AND METHODS
We examined a total of 34 specimens of Atractus albuquerquei, including the holotype.We also examined and compared our sample with other species of Atractus with 15 dorsal scale rows for which we had access to specimens.
Methods for hemipenial preparation and terminology follow Zaher (1999) and Zaher and Prudente (2003).Head, snout, and head plate lengths of the holotype were measured to the nearest 0.01 mm with a digital caliper.Total length and tail length were measured to the nearest 1.00 mm by stretching carefully the specimens along a ruler.Images of specimens were taken with a digital camera and mounted in plates with the aid of Adobe Photoshop.Ventral scales were counted using Dowling's method (Dowling, 1951).
The specimen is an adult male of 418 mm total length, 58 mm tail length (13.8% of total length); head length 13.8 mm (3.3% of total length); head width 7 mm at broadest point.Head is not distinct from neck.Dorsal scales smooth, in 15-15-15 rows, without apical pits.There are 172 ventrals, an undivided anal plate, and 38 paired subcaudals.Six maxillary teeth.
Rostral plate 1.5 times wider than high, and visible from above.Paired internasals 1.6 times wider than long.Paired prefrontals 1.4 times wider than long, in contact with each other and with frontal, internasal, posterior nasal, loreal, supraocular, and the orbit.Frontal pentagonal, 1.2 times wider than long.Supraocular as long as wide.Parietals 1.9 times longer than wide.
Nasal plate in contact with supralabials 1-2, and divided above and below the naris.Loreal plate 2.3 times longer than wide.Preocular absent.Two small postoculars of same size (fused on the right side of the head).One anterior temporal and two posterior temporals.On the right side of the head, the anterior temporal and the upper posterior temporal are of equal size whereas the lower posterior temporal is 1.3 times shorter than the former two.On the left side of the head, the anterior and lower posterior temporals are about 1.3 times shorter than the upper posterior temporal.Six supralabials, 2-3 contact the loreal and 3-4 border the orbit.Mental 2.7 times wider than long, separated from genials by the first pair of infralabials.Six infralabials, 1-3 are in contact with the anterior genials.Anterior genials 2.3 times longer than wide.Posterior genials absent.
Hemipenis: According to Cunha and Nascimento (1983), the hemipenis of the holotype is bilobed, with a bifurcated sulcus spermaticus, and has the undifferentiated type B pattern described by Savage (1960).However, our preparation of the left hemipenis of the holotype revealed a different pattern.Our preparation rendered a fully everted and almost maximally expanded hemipenis (sensu Myers and Cadle, 2003; see also Zaher and Prudente, 2003).The hemipenis is bilobed, semicapitate and semicaliculate (Figure 3).The capitular groove is well marked on the asulcate side, whereas on the sulcate and lateral sides it is only slightly developed.The lobes are restricted to the distal half of the capitulum, which is marked by the capitular groove, the latter being positioned just above the sulcus bifurcation, well below the lobular crotch.The lobes are covered by large and shallow papillate calyces.The left lobe is slightly longer than the right.The papillae are progressively replaced by spinules toward the base of the capitulum, where the calyces tend to lose their vertical walls and form poorly defined spinulate flounces.The distal two thirds of the hemipenial body is covered by medium-sized, hooked spines which tend to be slightly longer on the lateral surface and proximal region of the organ.The base of the organ is mostly nude, except for some dispersed spinules on the sulcate side.The sulcus spermaticus is bordered by spinules from the base of the organ to its bifurcation at the distal region of the hemipenial body, where both branches diverge to a centrifugal position ending at the tip of the lobes.A small naked pocket extends from the base of the organ  to the proximal half of the hemipenial body on the left lateral surface.
Coloration: The coloration of the holotype is poorly preserved and faded, precluding a detailed description.The dorsal surface of the head and body is light brown from the rostral scale to the end of the tail.The infralabials, mental region, and ventral surface of the body are uniformly light cream, except for the dark brown lower edges of the first, second, and third infralabials.The ventral surface of the tail retains tiny light brown spots on the sagital axis, between the two rows of scales, giving rise to a light cream longitudinal stripe.Supralabials 1-2 are light cream, 4 and 6 are light brown, and 3 and 5 are light brown dorsally, light cream ventrally.

Morphological Variation in Atractus Albuquerquei
Proportions and scutellation: For the purpose of the following description, we used data from 33 specimens (17 males and 16 females; CHUNB 30245 discarded).
The largest specimen is a female with 772 mm of total length, 72 mm of tail length; largest male is 492 mm total, 63 mm tail.Color pattern in preservative: The dorsal surface of the head and body is nearly uniform light to dark brown.The dark brown head cap generally extends to the dorsal surface of the supralabial scales, and more extensively on the third and fourth (below the eye) which tend to be almost completely brown.The first and second supralabials are generally almost light cream, giving the impression of a light cream spot on the lateral surface of the snout.The light cream coloration of the supralabial scales extends slightly dorsally, above the last two supralabials and lateral surface of the neck.More extensively pigmented individuals (such as MZUSP 13368 and 13369) have completely dark brown supralabials, but retain a light cream spot on the first and second supralabials.In these more melanic specimens, the dark brown coloration extends to the lower edge of the infralabial scales and distal tips of the genials which are mostly light cream, as with the rest of the gular region.Some less melanic specimens have only a dark brown spot on the genials.All other specimens retain a uniformly light cream gular region.In adult specimens the dorsum is generally uniform brown or, in some cases, the vertebral region is of a darker brown than the rest of the dorsum, giving the impression that it retains a slightly darker longitudinal stripe.A striped condition with the presence of slightly darker longitudinal bands was found in some of the youngest specimens, and can be divided into three distinct dorsal patterns: 1) CEPB 1676 and MZUSP 4586 retain a darker vertebral stripe, one scale width, running from the neck to the tip of the tail; 2) MPEG 16870 and 17170 also have a vertebral stripe as well a thin paravertebral stripe on each side of the body; 3) MZUSP 8734 has one large vertebral stripe and two paravertebral stripes running on each side of the body, from the neck to the tip of the tail.Specimens from Cana Brava are darker than the remaining specimens, which tend to a uniform brown pattern.In all specimens, the flanks are conspicuously lighter than the dorsum due to the appearance of light cream spots on the last two rows of dorsal scales.In all specimens, the last dorsal row (in contact with the ventrals) has bicolored scales with their lower half to two-thirds being light cream (as the ventral scales) whereas the upper half or one-third of the scale remains brown.
The light cream coloration is variably immaculate or diffuse with dark brown marks dispersed on the light cream area of the scales.In six specimens (all from Cana Brava), the penultimate scale row also retains a bicolored pattern, with mostly brown scales with light cream spots on their distal tips or halves.In three specimens, the last dorsal scale row retains scales bordered by a brown edge with a light cream center forming a spot or tending to a longitudinal bar, and ventrals with brown lateral edges, resulting in a thin light cream stripe along the flanks.MZUSP 13366 has only inconspicuous light cream tips on both dorsal scale rows, with a dark brown pattern invading the lateral edge of the ventral scales.
The ventral surface of the body is an immaculate light cream in most specimens.Only three specimens (MZUSP 8679, 8733, CEPB 1674) have a few dispersed brown spots on the venter.In one specimen (MZUSP 8733) the spots are mostly condensed along the midline.Eleven specimens have a totally immaculate light cream ventral surface of the tail.Eighteen specimens have a dark brown spotted ventral surface of the tail, in which the spots tend to condense along the midline, between the two rows of subcaudals, forming a conspicuous brown stripe.A few specimens, including the holotype, have only a few dispersed spots on an almost light cream surface, whereas the majority of the sample has a distinctly spotted ventral surface of the tail.

Color pattern in live specimens:
The following description is based exclusively on the color pattern observed in specimens collected from Cana Brava and Parque Nacional das Emas, state of Goiás (Figure 2).Basically, specimens are uniform dark brown or black dorsally, and light yellowish ventrally.The dorsal surface of the head and body are a nearly uniform dark brown.The two juveniles (291 mm and 307 mm) are darker than the four adult specimens (387 mm, 420 mm, 508 mm, 664 mm) which tend to a uniform dark brown pattern.The last dorsal scale row (touching the ventrals) has scales with their lower half to two-thirds light yellow and the upper half or one-third dark brown.The penultimate dorsal scale row may be either completely dark brown or with light yellowish spots.Ventral scales are a uniform light yellow.Subcaudal scales are also light yellow, but often with dispersed dark spots.
Hemipenial variation in Atractus albuquerquei: We prepared 12 hemipenes of which eight are fully everted and maximally expanded, and four are fully everted but only partially expanded (sensu Myers and Cadle, 2003; see also Zaher and Prudente, 2003).The pattern present in all examined organs is similar to that described for the holotype.The organs are bilobed, semicapitate and semicaliculate.One lobe is always slightly longer than the other (except in MZUSP 13367 and MZUSP 13024).The longer lobe and basal naked pocket are always on the same side of the organ.In CEPB 1675, MZUSP 13367, and CHUNB 30344 the lobes are wider than in the other specimens.In all specimens the lobes are restricted to the distal half of the capitulum, which is clearly marked by a capitular groove disposed transversely, just above or at the level of the sulcus bifurcation.The capitular groove is always present, being only slightly pronounced in MZUSP 13367 (well developed in the other organs examined).In all organs, the capitulum is ornamented by distal papillate calyces that tend to calcify and change to spinules proximally.In two specimens (MZUSP 13367, 11242), the calyces tend to lose their walls and form spinulate flounces on the proximal region of the capitulum.The hemipenial body is always covered with medium-sized and hooked spines, which tend to be slightly more developed on the proximal region of the hemipenial body.In MZUSP 11157, the basalmost spines are clearly longer than the ones covering the rest of the hemipenial body.The sulcus spermaticus is bordered by spinules from the base of the organ to its bifurcation at the level of the distal region of the hemipenial body (except for MZUSP 8778 and CEPB 1675, in which the border of the sulcus is nude).The sulcus spermaticus always bifurcates centrifugally, ending at the tip of the lobes.The base of the organ is always nude and bears a naked pocket that extends vertically on its lateral surface.
Geographic variation: None of the counts or measurements presents any variation significantly correlated with latitude.However, there is a significant decrease in the number of subcaudal scales from East to West, both for males and females (Figure 5), d. f. = 1,15, F = 8.08, p = 0.01 for males; d. f. = 1,14, F = 17.71, p < 0.001 for females).Variation in numbers of ventral scales, total body length, snout-vent length, and caudal length did not correlate significantly with longitude.

DISCUSSION
According to Cunha and Nascimento (1983:9), A. albuquerquei might be a junior synonym of A. boettgeri.
However, this species differs from A. albuquerquei in its fewer subcaudals (21-22 instead of 27-44), a smaller mentum, a larger number of infralabials (7 or 8 instead of 6), infralabials contacting the genials (4 or 5 instead of 3), presence of a light brown band on the head that is reduced to the two posterior thirds of the parietal scales on the top of the head, and by its ventral color pattern with dark blotches in the venter and an entirely dark ventral surface of the tail.The holotype of A. boettgeri is shown in Figure 6.
Two other species of Atractus with 15 dorsal scales and six supralabials, not mentioned by Cunha and Nascimento (1983), are A. poeppigi and A. elaps.They differ markedly from A. albuquerquei in their coloration, having a banded pattern along the body (A.elaps) or only on the venter (A.poeppigi) instead of a uniformly brownish dorsum and immaculate light yellowish venter.These two species also differ from A. albuquerquei in their short loreals, fewer ventrals (135-161 in A. elaps, 139-148 in A. poeppigi) and subcaudals (16-37 in A. elaps, 25-29 in A. poeppigi) and, in the case of A. poeppigi, in having four infralabials in contact with the genial scales (see Table 1), and in the absence of the anterior temporal.There are several other species of Atractus with 15 dorsal scale rows that differ from A. albuquerquei in several aspects of their scutelation.These species and their distinctive features are listed in Table 1.
In his revision of the Ecuadorian species of Atractus, Savage (1960) describes a strong sexual dimorphism in the segmental scale counts, which is also present in the species studied here.We found in Atractus albuquerquei the same pattern of variation described by Savage (1960:21): the number of ventral scales is higher for females than for males, while the counts for subcaudal scales are higher for males.Similarly, total length and snout-vent length are significantly greater in females.Savage (1960:21) reported relatively longer tails in male Atractus, and explained this as a result of their possession of hemipenes.However, a simple interpretation of greater relative tail length in male A. albuquerquei is confounded by females having significantly greater snout-vent lengths.Additionally, absolute tail length is not significantly different between males and females, although male tails have significantly more subcaudal scales.
Geographically, the intraspecific variation in both male and female subcaudal scales (though not in tail length) in A. albuquerquei is correlated with longitude.There are no obvious adaptive or morphological constraint explanations for this, but it may be associated with some developmental process common to both FIGURE 6. Holotype of Atractus boettgeri (BMNH 1946.1.6.29).sexes.The absence of any similar pattern of correlation with latitude suggests that variation in subcaudal scales is not associated with variation in environmental temperature.It might be that the probably largely fossorial A. albuquerquei is subject to a less variable subterranean temperature regime than above-ground snakes, but this requires much more study.

FIGURE 3 .
FIGURE 3. Fully everted and almost maximally expanded left hemipenis of the holotype of Atractus albuquerquei in sulcate (left) and asulcate (right) views.

FIGURE 5 .
FIGURE 5. Regression of number of subcaudal scales against longitude in minutes, for males (closed circles) and females (open circles) of Atractus albuquerquei.Correlation is significant for both sexes (see text).

TABLE 1 .
List of species of Atractus with 15 dorsal scale rows showing relevant scale counts and bibliographic references.Numbers in parentheses represent supralabials and infralabials in contact with orbit and with genials, respectively.