Diet of the South American frog Leptodactylus ocellatus ( Anura , Leptodactylidae ) in Uruguay

The diet of Leptodactylus ocellatus (L., 1758) was studied in specimens collected at Maldonado Department (Uruguay), where monthly surveys were made between August 1998 and March 2000. A total of 143 frogs were collected, measured, sexed, and dissected, to assess stomach contents. The anurans were grouped into three age classes and four categories (considering sex and capture date). The trophic amplitude for each age class and sexseason category was quantified using Shannon-Weaver Index, and the trophic niche overlap between ages and categories by using the Pianka Index. Comparisons among treatments were made by G-tests. The most important prey items were coleopterans (IRI = 2547), and significant correlations were found between predator and prey sizes. The major differences on diet composition were found between extreme age classes (froglets and adults). Significant differences were also detected between sexes in the cold season but not in the warm season. Any of these changes in diet may be related with the availability of prey.


INTRODUCTION
The knowledge and understanding of the paths and processes related to amphibian trophic ecology is of interest not only for herpetologists, but also for ecologists (HIRAI & MATSUI, 1999).The pivotal role of amphibians in aquatic communities was pointed out by several authors (DUELLMAN & TRUEB, 1994;STEBBINS & COHEN, 1995).The global phenomenon of declining amphibian populations (BLAUSTEIN, 1994;MCCOY, 1994;ALFORD & RICHARDS, 1999) elevated these vertebrates to a zoological group of special interest.
The leptodactylids are one of the more widely distributed anuran families in South America, and specially Leptodactylus ocellatus (L., 1758) occupies a wide geographical area, including Brazil, Argentina, Paraguay and Uruguay.An extensive literature has been published about this species mainly because of its complex reproductive behaviour.VAZ-FERREIRA & GERHAU (1975, 1986) described parental care of foam nest and tadpole schools by females.GALLARDO (1964) and CEI (1980) gave some data about the adult diet for several Argentinian populations.BASSO (1990) provided quali-quantitative information about the diet.The feeding ecology of tadpoles and juveniles was described by LAJMANOVICH (1994LAJMANOVICH ( , 1996) ) for Argentinian populations, but knowledge on the diet of this species in Uruguay is very scarce.LANGONE (1994) reported some data about systematics, behaviour, reproduction and, based on the diet, he stated that this frog may feed on land or in water.The name of the species, L. ocellatus, is here adopted in the sense of LANGONE (1994).Some references on the trophic niche of adults and on cannibalistic events in an Uruguayan population, were given by ACHAVAL & OLMOS (1997).
The objective is to describe the diet of an Uruguayan population of Leptodactylus ocellatus in the wild, analyzing the ontogenetic, intersexual, and seasonal changes that may occur.

MATERIAL AND METHODS
The field work was conducted at Arroyo Espinas (34º47 , S, 55º22 , W), Maldonado, Uruguay, between August 1998 and March 2000.The frogs were obtained with pitfall traps and supplementary specimens were hand collected.These animals were anesthetized in situ with 2-phenoxy-ethanol and immediately fixed in 10% formalin as suggested by CALDWELL (1996).All specimens were deposited in the Vertebrate Zoology Collection, Facultad de Ciencias, Universidad de la República, Uruguay (ZVCB).In the laboratory, snout-vent length (SVL) and mandibular width (MW) were measured to the nearest 0.1 mm.Specimens were ventrally dissected, and stomach and first portion of the intestine were removed (SCHOENER, 1989).The prey contents were analyzed under a binocular microscope and items were determined to order level.The total length (L) and maximum width (W) of all prey items were measured with an ocular micrometric grid (in mm), and the contents were kept in cans with 10% formalin.The individual prey volume (in mm 3 ) was estimated after DUNHAM (1983) criteria, under the expression V = 4/3Π (L/ 2)(W/2) 2 .The total volume of each stomach (TVS) was calculated as the sum of the volume of all prey items present in a given stomach.The minimum sample size was estimated from the cumulative diet diversity curve using Shannon-Weaver index (SHANNON & WEAVER, 1949) as suggested by KOVÁCS & TÖRÖK (1997).
Based on the SVL, frogs were grouped into three age classes: froglets (SVL < 50 mm), juveniles (50 < SVL < 75 mm) and adults (SVL > 75 mm).The TVS data allowed to differentiate two feeding periods: a warm season with high foraging rate (November to February) and a cold season with low foraging activity (March to October).Sex of each individual was determined by examination of gonads.From the season (warm and cold) and sex (males and females) factors, were determined four comparison MANEYRO et al. groups.
The trophic diversity was analyzed by standardized Shannon-Weaver index: J = H / log(n), where H = -Σ [p i x log(p i )], p i is the proportion of the i resource in the sample, and n is the total number of different resources.
To describe the importance of each item consumed, the index of relative importance (PINKAS et al., 1971) was calculated as: IRIt = (POt) x (PIt + PVt), where POt is the percentage of occurrence (100 x number of stomachs contained t item / total number of stomachs), PIt is the percentage of individuals (100 x total number of individuals of t in all stomachs / total number of individuals of all taxa in all stomachs), and PVt is the percentage of volume (100 x total volume of individuals of t in all stomachs / total volume of all taxa in all stomachs).
The dietary overlap among groups (age classes, sex-seasons) was evaluated using the Pianka index (PIANKA, 1973) ) 0.5 , where p ij and p ik are the proportion of the i resource used by the j and k paired categories in each treatment.Changes on diets among groups were evaluated by G-tests using IRIs.Regressions between prey length (mean, minimum, and maximum of each prey) and predator size (SVL and MW) were made.

RESULTS
The minimum sample size was estimated in 115 individuals (fig.1).From a total of 143 stomachs analyzed, 30 (21%) were empty and 113 (79,0%) presented at least one of the 19 taxa identified (tab.I).The overall mean richness of taxa by stomach was x = 2.32 (SD = ± 2.22), reaching a value of x = 2.96 (SD = ± 2.10), when empty stomachs were not considered.The maximum richness found in a single stomach was 11 items.
The overall diversity (standardized Shannon-Weaver index) was H = 0.67.The total number of prey individuals retrieved was 1159.The maximum number of prey individuals found in a single stomach was 118.The mean number of individuals per stomach was x = 8.11 (SD = ± 18.24), reaching a value of x = 10.35 (SD = ± 20.06), when stomachs without content were not considered.
Regressions between prey size (mean, minimum, and maximum prey length in each stomach) and predator length (SVL and MW) were made using all stomachs containing at least one prey individual (tab.II).In five of the six regressions (fig.2), a positive and statistically significant (p < 0.05) slope was found, except for the relationship between minimum prey length and predator SVL, where the slope was also positive but not statistically significant (p = 0.12).
The trophic diversity and the index of relative importance change between age classes for some items 0,0   0,2   0,4   0,6   0,8   1,0   1,2   1,4   1,6   1,8   2,   overlap between juveniles and adults was 0.834 (G = 45.7,df = 7, p < 0.05).On taxa level, the isopods were the only group not consumed by froglets.Some changes have been observed between the four sex-season categories in the trophic diversity and the index of relative importance for each item (tab.III) and the niche overlap (tab.IV).The maximum overlap value was reached by the males in warm season -females in warm season pair (Pianka index = 0.963).In this season (warm), the trophic diversity of each sex is the highest.The diet shows significant differences among the four categories (G = 59.8, df = 21, p < 0.05), and between sexseason classes in paired comparisons.The exception was the males in warm season -females in warm season pair, where no significant differences were found (G = 4.41, df = 7, p = 0.73).

DISCUSSION
Correlations between the diet of tropical anuran species and environmental variables allowed the classification of these species as generalist or specialist predators (TOFT, 1981).This information is scarce for amphibians from temperate regions (HIRAI & MATSUI, 1999), particularly in South America.
The trophic niche of L. ocellatus in the studied Uruguayan locality would indicate that this frog is nearer to the wide niche predator than to the narrow one, in the continuous specialist-generalist.Characteristics of this kind of frogs (TOFT, 1981) include lower number of prey items and a "sit and wait" foraging mode.The present results agree with those obtained for other leptodactylid frogs, such as Ceratophrys cornuta L., 1758 (DUELLMAN & LIZANA, 1994), Leptodactylus latinasus Jiménez de la Espada, 1875 (BASSO, 1990) and a species in the L. ocellatus group (now L. macrosternum Miranda-Ribeiro, 1926) from Amazonas River basin (STRÜSSMAN et al., 1984).GALLARDO (1964) described the diet of L. ocellatus based on 88 specimens, demonstrating that the majority had eaten coleopterans.This prey item has been also found frequently in the diet of a wild population in Argentina (BASSO, 1990).
Regressions between frog size (SVL, MW) and prey length showed significant positive correlations.This fact may be interpreted as an ontogenetic change on diet.In the present study, ontogenetic change on diet has been found among three age classes.This kind of change has also been described in some Amazonic leptodactylids by LIMA (1998) and LIMA & MAGNUSSON (2000).These authors suggested that ontogenetic changes could be explained by prey size and type.LAJMANOVICH (1996) quantified the trophic spectrum of juveniles of L. ocellatus in a wild population at Paraná River, and he found that diet changes with increasing body size.He hypothesized that these changes may be correlated with dispersion mechanisms of the species, since adult frogs move to terrestrial environments with higher prey availability.Our results may partially confirm this hypothesis, because the trophic diversity increases with age classes and the niche overlap (similarity) is greater between neighbour age classes.Some terrestrial prey items, like isopods and ants, increased their value of IRI in a positive way with age class, whereas coleopterans (that include many aquatic taxa) decreased in the value of this index.Spatial changes in prey availability might be discarded because specimens of all age classes were collected in the same localities (traps).A daily segregation may be considered (temporal changes), and may not be excluded as a factor that would explain differences in diet as a response to prey availability.
Sex-seasonal changes have been also detected in the studied population.The occurrence of the high value of overlap index (and not significant differences) between both sexes in the warm season means that both sexes shared a great portion of trophic resources.Significant differences between sexes in the cold season may be interpreted as an "adaptative specialization" related to decreasing of prey availability.These seasonal and intersexual changes may support this hypothesis, because significant fluctuations on dietary composition among classes were detected, with the cold season being the period where the trophic spectrum was lowest.Some authors have suggested this kind of predator responses to prey availability diminishing (JACKSIC, 2001).The dietary plasticity of L. ocellatus may explain, in part, the success of the species in fluctuant environments of temperate regions.
Fig. 1.Cumulative diversity for the overall sample of Leptodactylus ocellatus.Arrow indicates the minimum sample size (115 individuals).