Diet and feeding behaviour of the Neotropical parrot snake (Leptophis ahaetulla) in northern Brazil

Specimens (289) of Leptophis ahaetulla from northern Brazil and western Maranhão were examined for the composition of stomach contents. Most prey items were tree frogs, especially those of the family Hylidae (90%). Most of the anurans identified belong to the Scinax ruber species group (27%) and Scinax sp. (25%). Prey size was significantly related to snake length but not to head length. There was no significant difference related to sex in either case. Both sexes preyed on small to medium‐sized items (3–10% of snake snout–vent length). Leptophis ahaetulla seems to manipulate captured prey before ingestion since most of the prey items (83.6%) were swallowed head‐first. Leptophis ahaetulla is primarily diurnal and semi‐arboreal, inhabits disturbed and undisturbed forest, and forages primarily on the ground and in fallen vegetation, where its prey items are likely to be found at rest. †In memoriam.

The goal of this paper is to report the feeding habits of L. ahaetulla from northern Brazil and western Maranhão, addressing also the following questions: (1) what are the macrohabitat niches utilized by L. ahaetulla? (2) Are there differences in the size of prey items ingested by males and females of L. ahaetulla?

Material and methods
We analysed a total of 289 adult specimens of L. ahaetulla for the composition of stomach contents, of which 53 had identifiable prey items (Appendix I). The snakes are deposited in the Instituto Butantan (IBSP), São Paulo; Museu de Histó ria Natural Capão da Imbuia (MHNCI), Curitiba; Museu Nacional do Rio de Janeiro (MNRJ), Rio de Janeiro; Museu Paraense Emílio Goeldi (MPEG), Belém; and Museu de Zoologia da Universidade de São Paulo (MZUSP), São Paulo. Most of the stomach contents analysed were obtained from snakes collected in eastern Amazonia, mainly in Pará State, and western Maranhão, which have similar flora and rainfall distribution (Ab'Saber 1977;Cunha and Nascimento 1982).
We made a small incision in the stomach of each snake and removed all intact or partially digested prey items for further examination. Direction of ingestion (i.e. head-first or tailfirst) was recorded. Prey items that were transversely orientated in the gut were not included in the analyses. The sex of all snakes was determined by dissecting the base of the tail, and by inspection of gonads. Prey size was measured to the nearest 0.1 mm using a digital caliper.
Analyses of covariance (ANCOVA) were used to test for the correlation between prey length versus snake length and between prey length versus snake head length (covariates), using sex as factor. Assumptions of normality and homoscedasticity were evaluated using Kolmogorov-Smirnov's test and the Levene's test, respectively (Zar 1999).

Results
Fifty-three of the 289 stomachs of L. ahaetulla examined contained prey items. Anurans were the most frequent prey category (90% of the total prey items), including 10 species representing five genera (Dendropsophus, Hypsiboas, Osteocephalus, Scinax, and Sphaenorhynchus) of Hylidae. The most frequent prey items were frogs of the Scinax ruber species group (27%; Table I), which were ingested throughout the year, except September and December.
Three males contained lizard tails and remains of Gonatodes humeralis (Gekkonidae). One male contained remains of Hemidactylus mabouia and two females contained remains of H. mabouia (Gekkonidae) and Bolitoglossa paraensis (Amphibia, Plethodontidae). The lizards and salamander were not measured in our analysis.
Forty-seven of the 53 specimens (88.7%) contained a single prey item. Six had two prey items in their stomachs. Forty-six prey items (83.6%) were consumed head-first and nine were consumed tail-first. The direction of ingestion could not be determined for four prey items.

Discussion
Our results agree with published reports on the diet of L. ahaetulla (Oliver 1948;Lopez et al. 2003). Most prey items were treefrogs of the genus Scinax, which are among the most frequently encountered hylid frogs in the Amazon Basin (Duellman and Wiens 1993). They are nocturnal and inhabit primarily cleared areas in the rainforest (Lutz 1973;Duellman and Wiens 1993) on the ground and in bushes, and trees (Duelmann 1990), and include a perianthropic species (Hoogmoed and Á vila-Pires 1991;Á vila-Pires and Hoogmoed 1997). The species of Scinax have also been recorded as the major prey for other semi-arboreal snakes such as Chironius exoletus (Dixon et al. 1993) and Thamnodynastes strigatus (Bernarde et al. 2000).
The presence of the salamander Bolitoglossa paraensis was unexpected. This species is nocturnal (Crump 1977) and was probably taken at rest during the day. The anurans of the genus Hypsiboas and Dendropsophus are also nocturnal (Á vila-Pires and Hoogmoed 1997) and were probably also taken from their daytime resting sites.
The presence of two tails of Thecadactylus rapicauda may corroborate the efficiency of autotomy as one of the defences utilized by lizards for escaping predation attempts (Vitt and Vangilder 1983; see also Hero and Magnusson 1987). Thecadactylus rapicauda and Hemidactylus mabouia are primarily nocturnal (Á vila-Pires 1995) although the former may be active during the day (Hoogmoed and Á vila-Pires 1989). On the other hand, Gonatodes humeralis, the other gekkonid lizard consumed by Leptophis ahaetulla, is a diurnal species (Á vila-Pires 1995;Á vila-Pires and Hoogmoed 1997). Nocturnal lizards were probably taken in their daytime resting sites (see also Hero and Magnusson 1987). The presence of salamanders and lizards in low frequencies corroborates the results of other studies (Oliver 1948;Lopez et al. 2003).
Leptophis ahaetulla seems to manipulate captured prey before ingestion because 87% of prey items were consumed head-first. Such prey manipulation is probably facilitated by toxins which are lethal to small prey (Mertens 1971;Boos 2001).
The correlation between the body size of a snake and the size of its prey (Figure 1) corroborates findings by Shine (1991), but see Sazima and Martins (1990).
Based on information about habitat use by prey (Hoogmoed 1979;Duellman 1990;Zimmerman and Rodrigues 1990;Hoogmoed and Á vila-Pires 1991;Duellman and Wiens 1993;Á vila-Pires and Hoogmoed 1997;Van Sluys and Rocha 1998;Bernarde et al. 1999), we infer that Leptophis ahaetulla from northern Brazil and western Maranhão is one of the most wide-ranging habitat species of colubrine snakes, occurring in forested areas, along forest edges, open areas, lake borders, savanna and pasture areas, being also abundant in disturbed forest areas. Leptophis ahaetulla is primarily diurnal and semi-arboreal, forages primarily on the ground and in fallen vegetation, as an active forager. Our results also corroborate the hypothesis Nickerson 1976, 1977) that L. mexicanus and other arboreal snakes such as Imantodes and Leptodeira partition food resources, through the occupation of different temporal and dietary niches. Where snakes of the genus Imantodes and Leptodeira from forests of the Manaus region prey primarily upon lizards and several species of leptodactylid frogs, respectively (Vitt 1996;Martins and Oliveira 1999), L. ahaetulla eat primarily hylid frogs.
Zaher (MZUSP). Alessandra Travassos, Clarissa Canedo, Daniel Fernandes, José Pombal Jr, Marinus Hoogmoed, Roberta Pinto, and Teresa Á vila-Pires kindly identified some prey items for us. We are particularly grateful to Ana Prudente for her assistance in making material available. David Kizirian, Luis Felipe S. Aguiar, and two anonymous reviewers read the manuscript and offered valuable suggestions. Alessandra Travassos and Daniel Fernandes were especially helpful with identification of prey items. CAPES provided financial support to N.A.