Reproduction and diet of Imantodes cenchoa ( Dipsadidae : Dipsadinae ) from the Brazilian Amazon

Imantodes cenchoa (Linnaeus, 1758) is distributed from the east coast of Mexico to Argentina. In Brazil, it occurs in the north, central-west and northeast regions. We present information on the reproductive biology and diet of I. cenchoa from analysis of 314 specimens deposited in the Herpetological Collection of the Museu Paraense Emílio Goeldi (MPEG). Imantodes cenchoa displays sexual dimorphism in the snout-vent length, where sexually mature females are larger than mature males (t = 4.02, p < 0.01; N males = 150, N females = 71), head length (f1.218 = 98.29, p < 0.01; N males = 150, N females = 71), and head width (f1.218 = 112.77, p < 0.01, N males = 150; N females = 71). Bi-sexual maturity is observed, with males becoming sexually mature earlier than females. Females with eggs were recorded from November to January (rainy season) and from April to July (dry season), suggesting two reproductive peaks throughout the year, with recruitment occurring mainly during the rainy season, when there is a greater supply of food. Imantodes cenchoa is a nocturnal active forager, capturing prey that are asleep on the vegetation. In 32.80% of the analyzed specimens, food contents were present, of which 84.11% were lizards of the genera Norops (69.16%, N = 74) and Gonatodes (14.95%, N = 16). The other 15.89% of the contents were made up of items in an advanced state of digestion, preventing their identification. Some specimens had more than one food item in their digestive tract, accounting for 107 prey items in total. There was no ontogenetic variation in the diet of I. cenchoa, and the predominant direction of prey ingestion was antero-posterior (71.96%). Larger snakes tended to feed on larger prey, although these did not exclude small prey from their diet.

Imantodes cenchoa is distributed from the tropical region of the island of Trinidad and Panama, through Mexico, Venezuela, Colombia, Guyana, Suriname, French Guiana, Bolivia and Brazil (north, central-west and northeast regions), reaching Paraguay and Argentina (CUNHA & NASCIMENTO 1978).It is an oviparous species and feeds on small frogs and lizards, mainly of Norops (Daudin, 1802) (ROBINSON 1977, ZUG et al. 1979, AVEIRO-LINS et al. 2006).Individuals can be seen foraging during the night on bromeliads, shrubs, in the underbrush, or on palm bracts, which during the day are primarily used for sleeping (HENDERSON & NICKERSON 1976, BARTLETT & BARTLETT 2003).Despite the arboreal habits of I. cenchoa, this snake has been observed on the ground of forests (HENDERSON & NICKERSON 1976, MARQUES & SAZIMA 2004) and dead on paved roads, killed by cars, indicating possible terrestrial foraging.
In this study, we present data to increase our understanding of the reproductive and dietary strategies of I. cenchoa from the Brazilian Amazon region.

MATERIAL AND METHODS
We analyzed 314 specimens (200 males and 114 females) of I. cenchoa, from forested environments of the Brazilian Amazon.They are deposited in the Herpetological Collection of the Museu Paraense Emílio Goeldi (MPEG), state of Pará, Brazil (Appendix).
The Amazon region is formed mostly of constantly humid environments, with precipitation close to 3000 mm/year.The driest month never has less than 60 mm of rainfall and humidity is around 80%, with an average temperature of 25.9°C.The climate of this region is classified as type Af (according to Köppen's classification).The vegetation varies from a low-altitude forest (the Andean portion) to the Amazon rainforest, with predominance of the "terra firme" forests, flooded forests, "várzeas" and "igapós" (PNRH 2006).
For each specimen, the snout-vent length (SVL), tail length (TL), head width (HW), head length (HL), head height (HH) and distance between the eyes (DBE) were measured, as well as the mass.
Through a longitudinal incision in the abdominal region, the gonads of both males and females were analyzed macroscopically, in order to assess the state of the vas deferens of the males and the oviduct of the females, and to infer the number and length of ovarian follicles.Thus, females with SVL equal to or larger than females having follicles in secondary vitellogenesis (diameter 10.0 mm), and/or oviductal eggs, and/ or oviducts with evidence of eggs, were considered sexually mature.Males with SVL equal to or larger than the smallest male with turgid testes, convoluted and opaque efferent ducts, were considered sexually mature (SHINE 1977b, c, 1988, BALESTRIN & DI-BERNARDO 2005).Fecundity was inferred from the ratio of the number of follicles in secondary vitellogenesis and the number of oviductal eggs with snout-vent length of the female (ALDRIDGE 1979, DENARDO 1996, THOMPSON & SPEAKE 2002, SANTOS & LLORENTE 2004).
Analysis of the reproductive cycle was performed using adult (sexually mature) specimens, observing the temporal distribution of follicles in secondary vitellogenesis or the presence of oviductal eggs (SHINE 1977b(SHINE , 1988)).
Stomach contents were observed directly in the digestive tract of each specimen.The quantitative analysis of the diet was performed on the number of prey items observed in the stomach or intestine; in order to perform the qualitative analysis, we identified each prey item to the lowest possible taxonomic level, using the help of experts and the literature.Partially digested prey were measured (SVL and TL) and their mass was inferred from comparison with intact conspecific specimens of similar size (according to RODRIGUEZ-ROBLES & GREENE 1999) from nearby locations, and preserved in the Herpetological Collection of MPEG.
The direction of ingestion was classified according to the alignment of the head of the prey in relation to the body of the snake.Thus, the direction of items ingested head first was considered antero-posterior, and the direction of items ingested tail first was considered postero-anterior.
To verify the existence of ontogenetic variation in the diet, the types and size of prey found in the digestive tract of sexually immature and mature snakes were compared.
For data analyses, the program Statistica 7.1 was used.The significance level (␣) used for all tests was 0.05.A Student t-test was performed to test for the presence of sexual dimorphism in the SVL (for data with normality and homogeneity of variance).To compare the sexual dimorphism in relation to the TL (excluding specimens with broken tails), a one-way ANCOVA was used, with sex as a factor and SVL as a covariate.To test the sexual dimorphism of the HL, a one-way ANCOVA was used with sex as a factor and HL subtracted from SVL as a covariate.In the case of HW, a one-way ANCOVA was used with sex as a factor and SVL as a covariate, with log-transformed data.All data submitted to ANCOVA were tested for normality and homogeneity of variance.Pearson correlation analysis was performed to find the relationships between SVL and HL, and SVL and HW.In all tests of sexual dimorphism, only data from sexually mature females and males were used.Pearson correla-tion analysis was performed to find the ratio between the SVL of females and the number of eggs and vitellogenic follicles.Pearson correlation analysis was also used to find the relationship between predator-prey SVL, snake HL and prey SVL, and between mass of prey and mass of snake.
Females of I. cenchoa with eggs were found between November and January, during the rainiest season, and from April to July, during the dry season (Fig. 4).

Diet
Of the 314 specimens of I. cenchoa analyzed, 32.80% (N = 103) had some type of content in the stomach and/or intestines.Among the females with secondary follicles and/or eggs (N = 21), six (28.57%) presented food contents.
Among the food items, 84.11% were Norops (69.16%,N = 74) and Gonatodes (Fitzinger, 1843) (14.95%,N = 16) lizards, and the other 15.89% was made up of insect wings, eggs, scales, tails, lizard skulls, and other items in an advanced state of digestion, preventing their identification.Seven specimens had two food items in their digestive tract, yielding a total of 107 preys (Table II).
The predominant direction of ingestion was antero-posterior (71.96%,N = 77), and prey ingested postero-anteriorly accounted for 1.87% (N = 2) of the total.The direction of ingestion could not be determined for 26.17% (N = 28) of the items (Table II).A correlation was found between SVL and HL (r 2 = 0.68, p < 0.01) and between SVL and HW (r 2 = 0.44, p < 0.01), where individuals with larger snout-vent length had larger head length and width.
The SVL of sexually mature specimens ranged from 438 to 829 mm in males (Fig. 1) and between 560 and 900 mm in females (Fig. 2), emphasizing the sexual bi-maturity, with males reaching sexual maturity at a smaller size than females.
One to two follicles in secondary vitellogenesis (N = 8, mean = 1.87) and one to three eggs (N = 14, mean = 2.07) were found in mature females.The length of eggs ranged from 22.85 to 45.10 mm (N = 14, mean = 30.06mm).Larger females had a

2
Both immature and mature females and males frequently preyed on Norops fuscoauratus and no evidence of ontogenetic variation was observed (Figs 5 and 6).
The results show that larger snakes tend to eat larger prey items with larger mass, although smaller prey with smaller mass were not eliminated from their diet, (r 2 = 0.13, p = 0.01; r 2 = 0.22; p < 0.01, respectively) (Figs 7 and 8).

DISCUSSION
Sexual dimorphism in snakes can occur in characters such as size and/or body shape, color, position and/or size of organs, and behavior (KING 1989a, SHINE 1993, BONNET et al. 1998, KEOGH & WALLACH 1999, PIZZATTO & MARQUES 2006), as well as the relative size of the head (CAMILLERI & SHINE 1990).Most frequently sexually mature females have greater snout-vent length than their sexually mature male counterparts, as observed in this study, is (e.g., MARQUES & PUORTO 1998).This is related to the fact that there is a positive relationship between body size and the ability to produce and maintain a larger amount of eggs/embryos (SHINE 1994, PIZZATTO et al. 2006).
The absence of sexual dimorphism in the length of the tail of Imantodes cenchoa, observed in this study, had been previously reported by MYERS (1982), as was the exponential increase in length and width of the head of females relative to that of males.Adult males of I. cenchoa in Honduras and Costa Rica had a slightly longer tail than females.However, there was an absence of sexual dimorphism in tail length in specimens from Panama, Ecuador and Peru (ZUG et al. 1979).Sexual dimorphism in tail length occurs in most colubrids, in which males have larger tails than females.The larger tail of males is necessary to hold the copulatory organ and associated muscles.Conversely, females need a larger body in order to produce more offspring (KING 1989).The absence of sexual dimorphism in the tail of I. cenchoa can be related to its arboreal habit, since selection favors longer tails, which provide better balance and movement in trees (PIZZATTO et al. 2007).
Sexual dimorphism in head size, as found in I. cenchoa, has evolved, according to SHINE (1986), to enable males and females to feed on prey of different sizes, maximizing foraging efficiency and minimizing competition between the sexes.Simi-Figures 3-4.(3) Fecundity of Imantodes cenchoa, from the Brazilian Amazon, indicating the relationship between the number of follicles in secondary vitellogenesis ( ) and eggs ( ) with female SVL.(4) Seasonal distribution of the largest follicles ( ) and eggs ( ) of Imantodes cenchoa, from the Brazilian Amazon.
The fact that males reach sexual maturity earlier than females is probably related to energy cost, which is higher for females than for males of different sizes (PARKER & PLUMMER 1987, SANTOS-COSTA et al. 2006), but similar between large females and large males (MADSEN & SHINE 1993).Natural selection would favor of a more delayed sexual maturity in females.This results in larger females that are consequently more fertile (FITCH 1970, 1982, VITT & VANGILDER 1983, SHINE 1994) and could explain the bi-maturity in some species of snakes, such as Erythrolamprus aesculapii (Linnaeus, 1766) (MARQUES 1996) The results on the fecundity of Imantodes cenchoa, presented in this study, are consistent with a trend found for most species of snakes, where larger females produce more eggs (FITCH 1970, SEIGEL & FORD 1987).The abdominal space may limit the reproductive investment relative to the size or the total mass of offspring, resulting, in many species, a strong relationship

6
Figures 5-6.Relationship between the number and type of prey consumed by sexually immature and mature females ( 5) and male (6) of Imantodes cenchoa from the Brazilian Amazon.( ) Immature, ( ) mature.
Figures 7-8.( 7) Relationship between snout-vent length (SVL, in mm) of Imantodes cenchoa, and snout-vent length (SVL, in mm) of their prey, from the Brazilian Amazon.( 8) Relationship between the mass (g) of Imantodes cenchoa and mass (g) of their prey, from the Brazilian Amazon.

8
between the body size of the mother and the size of the litter (FITCH 1981), emphasizing why larger females are favored.LILLYWHITE and HENDERSON (1993) speculated that the slender body shape of arboreal snakes may constrain the female's capacity to carry eggs.This hypothesis was strongly supported by the results of PIZZATTO et al. (2007), and would explain the small amount of eggs produced by I. cenchoa.Some previous studies (e.g., JAMES & JOHNSTON 1998, PLAUT 2002, GHALAMBOR et al. 2004) have demonstrated that the movement and escape ability of females is related to the habitat type and the offspring/egg weight.These results suggest that the weight of eggs has a significant cost.Likewise, the use of arboreal habitats by snakes could have provided selective advantages, such as a better ability to cradle and camouflage in between tree branches (POUGH et al. 1988, LILLYWHITE & HENDERSON 1993), along with the laterally compressed body shape, as in I. cenchoa.
Snakes have different reproductive cycles, and males and females of the same species may also have distinct cycles (SEIGEL & FORD 1987).PIZZATTO et al. (2008a) suggested that the cycle of I. cenchoa is continuous for about eight months, differing from that observed in this study, where females containing eggs were found in the beginning of the rainy season (November to January) and in the first months of the dry season (April to July), suggesting two reproductive peaks throughout the year, but with recruitment occurring mainly in the period (rainy season) when there seems to be more food available.A third pattern was observed by ZUG et al. (1979).In his data, females of I. cenchoa with eggs were found throughout the year, indicating a prolonged breeding season, which is correlated with the rainy season in the region.
In temperate regions, the period when the temperatures are higher influence the timing of the reproductive cycle of snakes.Higher temperatures are important not only for egg development, but are also associated with greater availability of food during recruitment (SHINE 1977a). In Iquitos, Peru, FITCH (1982) found no evidence that snake species reproduce only in certain seasons.DUELLMAN (1978), based on work carried out in Santa Cecilia, Ecuador, concluded that non-seasonal reproduction is a trend among Amazonian snakes.However, according to data obtained by MARTINS & OLIVEIRA (1999), recruitment for most species occurs during the rainy season, when there is a greater supply of food than in the dry season.
In the present study, only 32.8% of the specimens analyzed had food contents in their stomachs.These results were expected, since the frequency of specimens containing food items is usually low, ranging between 14 and 30% (MASCHIO et al. 2010).According to GREGORY & ISAAC (2004), this low frequency may be related to the period in which the specimen was collected.SHINE (1987) suggested that female snakes generally tend to reduce food consumption during their gestation period.This was observed in females of Natrix natrix (Linnaeus, 1758) and Anilius scytale (Linnaeus, 1758) (READING & DAVIES 1996, MASCHIO et al. 2010, respectively), which began to feed again after the reproductive period.The small number of females of I. cenchoa containing eggs and/or follicles in this study does not allow us to draw conclusions about the influence of gestation on stomach contents.
Analyses of stomach contents of I. cenchoa showed that Norops lizards are the most frequent items in their diet, demonstrating that these are their main prey (see also STUART 1948, 1958, WEHEKIND 1955, LANDY et al. 1966, HENDERSON & NICKERSON 1976).This result is in agreement with the findings of ZUG et al. (1979), who stated that I. cenchoa forages actively during the night, feeding on small diurnal arboreal lizards (mainly Norops) that sleep on the vegetation (see ÁVILA-PIRES 1995).
In this study, no food item of I. cenchoa was identified as an anuran.Despite this result, it is possible that these amphibians may occasionally serve as food for this species, as noted by different authors, as follows: MARTINS & OLIVEIRA (1999) found Pristimantis fenestratus (Steindachner, 1864) in the digestive tract of an adult male and a gravid female; TEST et al. (1966), observed I. cenchoa under captivity feeding on Pristimantis sp.(Duméril & Bibron, 1841);and BEEBE (1946) observed an individual of I. cenchoa stalking an individual of Ololygon rubra (Daudin, 1803).
According to MARTINS & GORDO (1993) the occurrence of arthropods in the digestive tract of snakes is the result of their anuran-based diets.In our study, arthropods were characterized as secondary stomach contents because they were always associated with lizard vestiges.This may be explained by the fact that Norops sp.and/or Gonatodes sp. are insectivorous (ARAÚJO 1991).
There is no record of I. cenchoa eating reptile eggs.Even though eggs have been found in the stomachs of these snakes by some studies (e.g.LANDY et al. 1966), these findings are associated with the ingestion of a gravid female lizard, where most of the prey was digested, except the eggs (HENDERSON & NICKERSON 1976).
The anteroposterior direction of ingestion, prevalent in this study, follows the pattern found in most species of snakes (PALMUTI et al. 2009), for example Helicops infrataeniatus (Jan, 1865) (AGUIAR & DI-BERNARDO 2004) and Thamnodynastes strigatus (Günther, 1858) (RUFFATO et al. 2003).According to GREENE (1976), this direction prevents the limbs of the prey and the disposition of their scales from offering resistance to swallowing, and minimizes the potential risk of injuries caused by the claws or teeth of the prey.Consequently, intake by the head results in decreased time, effort and energy expenditure during the process of prey capture and ingestion (PINTO & LEMA 2002).Furthermore, the chances of prey escaping are minimized by ingesting the lizards head-first, since some species have caudal autotomy.Caudal autotomy is a successful defensive strategy used by many lizards (CHAPPLE & SWAIN 2002), including those of Norops (SCHOENER & SCHOENER 1980) and geckos (CONGDON et. al. 1974), for instance species of Gonatodes, the main prey of I. cenchoa.
The specimens containing N. fuscoauratus and G. humeralis were recorded in almost every month of the year except May, September and November for N. fuscoauratus and April, August and December for G. humeralis.Norops fuscoauratus, the most consumed item by I. cenchoa in this study, can produce at least three clutches per year, laying one egg per clutch (VITT et al. 2008), indicating that this species has a continuous reproductive cycle (ÁVILA-PIRES 1995), which would justify the occurrence of N. fuscoauratus in the digestive tract of I. cenchoa throughout the year (see FITCH 1970, HOOGMOED 1973, DIXON & SOINI 1975, 1986, DUELLMAN 1978).Females of Gonatodes humeralis lay one egg at a time and may return a few days later to the same place to lay another egg (HOOGMOED 1973).Its reproduction also seems to occur throughout the year (DIXON & SOINI 1975, 1986) and it can lay up to twelve eggs (ÁVILA-PIRES 1995).
The selection of prey with specific sizes can be influenced by prey density and availability (PLUMMER & GOY 1984, SHINE 1987, 1991, MASCHIO et al. 2010).Although species normally have continuous reproductive cycles in the tropics, reproduction of some species may be seasonal (SHINE 2003. ALVES et al. 2005), with recruitment occurring mainly during the rainy season, when prey availability appears to be higher (MARTINS & OLIVEIRA 1999).This does not seem to be the case of I. cenchoa, for which both immature and mature males and females feed on common lizard species, as seen previously, in all seasons, in the studied area.

Figures
Figures 1-2.Percentage of sexually immature and mature males (1) and females (2) of Imantodes cenchoa from the Brazilian Amazon for the different size classes.( ) Immature, ( ) mature.

Table I .
Morphometry of sexually mature females and males of Imantodes cenchoa from the Brazilian Amazon, showing the number of specimens examined (N), mean, standard deviation and range.(SVL) Snout-vent length (mm), (TL) tail length (mm), (HL) head length (mm), (HW) head width (mm).