Feeding ecology of Liophis reginae semilineatus ( Serpentes : Colubridae : Xenodontinae ) in Eastern Amazon , Brazil

We studied the diet of Liophis reginae semilineatus (Wagler, 1824) from eastern Amazon, Brazil, based on the analysis of 182 preserved specimens. Thirty-six individuals had prey in their stomachs; 34 (95%) contained exclusively anurans and 2 (5%) contained both anurans and lizards. The most common prey items were small Leptodactylus sp. (33.3%), followed by Physalaemus ephippifer (Steindachner, 1864) (10.3%). Prey ingested head-first (78%; n = 25) were significantly larger than prey ingested tail-first (22%; n = 7). Females of L. reginae semilineatus have longer and wider heads than conspecific males with the same body length, which correspond to sexual divergences in the diet (size of the prey). No correlation was found between snake head length vs. prey size (SVL, width and mass). Liophis reginae semilineatus is an anurophagous snake that probably forages actively on the ground.

Liophis Wagler, 1830 belongs to Xenodontini and includes 35 species, occurring from the Antilles and continental Central America to southern South America (DIXON 1989).Even though most species are terrestrial, some have semi-aquatic habits.Species can be found in humid environments such as primary and secondary forests, flooded forests and floodplains, as well as in dry habitats such as semi-arid scrubland and open pastures (DIXON 1980, CUNHA & NASCIMENTO 1993).
The terrestrial L. reginae semilineatus (Wagler, 1824) is distributed in Brazil, Ecuador, part of Venezuela, Colombia, Bolivia, and Peru.In Brazil, this diurnal species can be found in the Amazon Forest and in the Atlantic Forest up to the northeast of the state of São Paulo (CUNHA & NASCIMENTO 1993), being common in anthropically disturbed areas (CUNHA & NASCIMENTO 1978) and occurrying in wet environments such as streams ("igarapés"), fresh-water springs, and ponds with emergent grasses in primary and secondary forests.The diet of L. reginae semilineatus consists primarily of anurans, fishes and lizards (CUNHA & NASCIMENTO 1993, MARTINS & OLIVEIRA 1998).
Despite the fact that L. reginae semilineatus is common in the Amazon region, particularly in disturbed environments, data on the natural history of this species in the region is available only in CUNHA & NASCIMENTO (1993) and MARTINS & OLIVEIRA (1998).In order to increase our knowledge of the natural his-tory of L. reginae semilineatus, this study performs a quali-quantitative analysis of the feeding biology of eastern Amazonian populations of this species.

MATERIAL AND METHODS
We examined 182 preserved specimens of L. reginae semilineatus from eastern Amazon, Pará, Brazil (96 males and 86 females).All specimens used for the analyses, collected between 1968 and 2005, are deposited in the Coleção Herpetológica of the Museu Paraense Emílio Goeldi (Appendix I).
The stomach contents of all dissected specimens were preserved in 70% ethanol and each was deposited in the same collection, under the same record number as its source.
The average temperature in eastern Amazon ranges between 24 and 27°C and annual average precipitation varies from approximately 1,500 mm to more than 3,000 mm (NHM 2007).A rainy season occurs from December to April with monthly precipitation of 300 mm, and a dry season occurs from June to November with monthly precipitation of 88 mm.The region has yellow and red-yellow latossoil that was originally covered by tropical rainforest.Presently, most of the area is disturbed (NHM 2007).
Data collection: we collected and recorded our data in the following manner: first, we recorded the snout-vent length (SVL, mm), head width (HW, mm, at widest part of head) and head length (HL, mm, from back of parietal to tip of snout) of each specimen.Next, we dissected each specimen in order to remove and sort out its stomach contents.Following preliminary sorting, we identified each intact or partially digested prey ZOOLOGIA 27 (1): 87-91, February, 2010 to the lowest possible taxonomic level, and determined the direction of ingestion (head-first or tail-first).Additionally, we measured the SVL, width and mass of each intact prey specimen.For digested prey, we estimated prey size based on anatomic parts using a regression analysis (ZAR 1999).
Statistical analysis: we performed all of our statistical tests using SYSTAT 10.In all cases, the Kolmogorov-Smirnov test was applied to test for a normal distribution prior to parametric tests.The level of significance adopted for all tests was 0.05.To compare the SVL of prey ingested in different directions, an ANCOVA was used (log10 snake SVL as co-variant, directions of ingestion as factor and log10 prey SVL as dependent variable).A reduced major axis regression was used to determine the relationships among prey dimensions (mass, SVL and width, log10 transformed) and snake head length (HL, log10 transformed), because there was no clear dependent-independent relationship between these variables (SOKAL & ROHLF 1981).To test for the presence of sexual dimorphism, the following tests were used: ANCOVA for the measures of HL and HW (with SVL as co-variant and sex as factor); the Chi-square test to compare differences in diet composition among the sexes; MANCOVA (snake SVL as co-variant, sex as factor and mass, width and SVL of the prey as dependent variables), together with an univariate F test to verify differences between prey dimensions among the sexes (ZAR 1999).
The most common anurans found in the stomach contents were members of the Leptodactylidae (six species), Bufonidae (two species), Hylidae (two species) and Microhylidae (one species).Besides anurans, two lizards of Gymnophthalmidae (two species) were also found.
Most snakes (88.2%; n = 30), had only one food item in the stomach.Three specimens had two and only one specimen had three items -1 Scinax nebulosus (Spix, 1824) and 2 Leptodactylus sp.The stomach of one snake had 128 leptodactylid eggs , each egg approximately 1.5 mm diameter (Tab.I).

DISCUSSION
Liophis reginae semilineatus is a predominantly anurophagous snake, like most snakes of this genus (VITT 1983, VITT & VANGILDER 1983, MICHAUD & DIXON 1989, PINTO & FERNANDES 2004), that occasionally preys upon other animals such as lizards.Our findings are consistent with previous reports that confirm the presence of anurans and lizards in the diet of this snake (CUNHA & NASCIMENTO 1993, MARTINS & OLIVEIRA 1998).Additionally, we have found anuran eggs in the diet of L. reginae semilineatus.Anuran egg predation had been previously reported for Liophis jaegeri (Günther, 1858) (SOLÉ & KWET 2003).Feeding on a large quantity of eggs and tadpoles may be advantageous, because it may result in the comsumption of a large quantities of prey in a short period of time, with little exposure to predation.Even though we did not find any fish in our samples, this item had been previously reported as prey by CUNHA & NASCIMENTO (1993).
The use of anurans (Leptodactylus andreae Müller, 1923, Leptodactylus wagneri (Peters, 1862), Allobates sp.) and lizards (Cercosaura ocellata Wagler, 1830) as prey had been previously reported for different areas of the Amazon (DUELLMAN 1978, ÁVILA-PIRES 1995, MARTINS & OLIVEIRA 1998).However, variations in diet composition among populations from these areas may be a result of sampling biases caused by the geographical distribution patterns of the prey, for example.
Liophis reginae semilineatus is active during the day (CUNHA & NASCIMENTO 1978), while most of the prey consumed is nocturnal (with the exception of lizards) and can be found by foraging the ground (DUELLMAN 1978, ÁVILA-PIRES 1995, ÁVILA-PIRES & HOOGMOED 1997, ESTUPIÑAN et al. 2002, LIMA et al. 2006).Based on this information, we conclude that our subject forages the ground predominantly during the day seeking for inactive (frogs) or active prey (lizards).
Most prey were consumed head-first, a trend observed in most snakes (MUSHINSKY 1987, GREENE 1997, RUFFATO et al. 2003, AGUIAR & DI-BERNARDO 2004, ALBUQUERQUE et al. 2007, LEITE et al. 2007).This behavior reduces time and effort spent swallowing, reducing the resistance imposed by the limbs.It also mini-mizes the amount of time the predator is vulnerable during predation, and the likelyhood of injuries inflicted by the prey (SAZIMA 1989, PINTO & LEMA 2002).Prey ingested head-first were significantly larger than items ingested tail-first.Two hypotheses can explain these results: 1) the snake manipulates large prey in a way that results in head-first ingestion (GREENE 1997, AGUIAR & DI-BERNARDO 2004, ALBUQUERQUE et al. 2007), or 2) bitting the prey on the anterior portion of the body is morel likely to result in successful ingestion by the snake (see VINCENT & MORI 2008).
No correlation was found between snake head length and prey dimensions (mass, SVL and width), an unusual finding for snakes (SHINE 1991, ARNOLD 1993).One possible explanation for this result is the opportunistic predatory behavior of L. r. semilineatus which results in the capture of prey of differ-  SHINE (1977) suggested that the capture of small prey is a strategy for reducing the risk of being injured, minimizing energy spending.Consistent with findings for most snakes (SHINE 1991(SHINE , 1994)), mature females of L. reginae semilineatus have larger and wider heads (HL and HW) with respect to males.In snakes, differences in HL and HW are usually correlated with ecological divergences (SHINE 1989(SHINE , 1993)).Our results confirm this hypothesis, since sexual differences were observed in the size of the prey consumed (females eat larger prey than males).Such dimorphism could represent an advantage for females, which have higer reprodutive costs than males.Consumption of larger prey could help avoid intersexual competition and increase the likelihood of energetic gain (SHINE 1991(SHINE , 1993)).

Table I .
Food items and their frequency in the diet of L. reginae semilineatus from eastern Amazon, Pará.(*) The Leptodactylidae eggs (n = 128) were considered as a single food item.