-A11 species of diurnal leaf-litter anurans in central Amazonia changed the type and size of their prey as they grew. Postmetamorphic individuals of all species ate principally mites and collembolans, and larger frogs ate larger prey of types. The shift in prey types was not a passive effect of selection for larger prey. There was a strong relationship between electivity for prey types and frog size, independent of electivities for prey size, in six of the seven species. This study showed that most species in the community had strong ontogenetic changes in diet composition and electivity for prey types, and these changes did not conform to simple models of the effects of predator size and diet quality. urnal of Herpetology, Vol. 32, No. 3, pp. 392-399, 1998 pyright 1998 Society for the Study of Amphibians and Reptiles e Ef ects of Size on the Di ts of Six Sympatric Spec es of stmetamorphic Litter Anurans in Ce tral Am zonia Change in diet with ontogeny has often been studied in fish (reviewed in Werner and Gilliam, 1984; Osenberg et al., 1992; Sheldon and Meffe, 1993). In anurans and reptiles, change in diet has been related to changes in selection for the type or size of prey (Schoener and Gorman, 1968; Pengilley, 1971; Labanick, 1976; Rose, 1976; Mushinsky et al., 1982; Christian, 1982; Dominguez and Salvador, 1990; Donnelly, 1991; Simon and Toft, 1991; Lima and Moreira, 1993). However, most of these studies have focused on one or two species (Lynch, 1985 is an exception). Consequently, little is know about change in diet with ontogeny in multi-species assemblages. However, the selection for prey size is not independent of selection for prey type because different types of arthropods have different mean sizes. The smallest arthropods available for anurans and lizards are principally mites and collembolans. The most abundant arthropods of intermediate sizes are ants, termites, and beetles, and the largest arthropods are orthopterans, butterflies, and spiders (Schoener and Janzen, 1968; Dominguez and Salvador, 1990; Lima and Moreira, 1993). Therefore, variiet ith ontogeny has often be n is (revie ed in Werner and Gilliam, et al., 1992; Sheldon and Mef e, a s and reptiles, change in diet late to changes in selection for the i prey (Schoener and Gorman, ille , 1971; Labanick, 1976; Rose, i s et al., 1982; Christian, 1982; a Salvador, 1990; Donnelly, 19 1; t, 1991; Li a and Moreira, 19 3). st of these studies have focused on s ecies (Lynch, 1985 is an exception). l , little is know about change in diet i ulti-species as emblages. e selection for prey size is not inselection for prey type because es of arthropods have dif erent . e s al est arthropods available s a lizards are principally mites ation in the taxonomic composition of the diet with the stage of growth of individuals is expected as a passive side effect of the capacity to ingest larger prey. Lima and Moreira (1993) d monstrated that the shift in prey types with growth was not just a passive effect of selection for larger prey in the leaf-litter frog, Colostethus stepheni. For that species, there was a strong relationship between electivity for prey types and frog size, independent of electivities for prey size. To test whether size-independent shifts in electivity are a general phenomenon in leaf-litter frogs or are restricted to C. stepheni, I studied the diet of postmetamorphic individuals of six species of diurnal leaf-litter frogs that occur syntopically with C. stepheni in Reserva Florestal Adolpho Ducke in Central Amaz6nia. MATERIALS AND METHODS Study Area and Distribution of the Species.-The study was conducted in a 10,000-ha tropical rainforest reserve, Reserva Florestal Adolfo Ducke (Reserva Ducke), located 25 km northeast of Manaus, Amazonas, Brazil (03?08'S; ion in the tax nomic composition of the diet ith the stage of growth of in ividuals is excted as a passive side ffect of the capacity to est larger prey. Lima and Moreira (1993) e onstrated that the sh ft in prey types with o th was not just passive effect of selection larger prey in the leaf-litter fr g, Colostethus pheni. For that sp cies, there was a st ong reionship between electivity for prey types and g size, ind pendent of electivities for prey 392 This content downloaded from 157.55.39.96 on Sun, 02 Oct 2016 05:58:51 UTC All use subject to http://about.jstor.org/terms DIETS OF SIX LITTER FROGS 60?04'W). The study was done at two sites separated by four km. The first site, described by Lima and Moreira (1993) lies along six trails (?7 km) in an undisturbed part of Reserva Ducke. The second site included a plateau and stream. I used an area of about 2 km by 200 m wide along the margins of the stream and about 2 km by 10 m on the plateau. The plateau is traversed by an unsealed forest road. Temporary pools form on the margins of the road in the rainy season. The community of diurnal leaf-litter frogs of Reserva Ducke is composed of seven species: Adenomera andreae and Eleutherodactylus fenestratus (Leptodactylidae), Colostethus marchesianus, Colostethus stepheni, and Epipedobates femoralis (Dendrobatidae), and Bufo cf. typhonius and Dendrophryniscus minutus (Bufonidae). The juveniles of all species metamorphose at ca. 6.5 mm snout-urostyle length (SUL), except for E. femoralis and E. fenestratus, which metamorphose at >8.0 mm SUL. Individuals of Eleutherodactylus fenestratus >24 mm are nocturnal. All species are active during the day and can be found in the same place foraging for arthropods over the leaf-litter. At both study sites, five of the species (D. minutus, C. marchesianus, E. fenestratus, A. andreae, and B. cf. typhonius), are usually found near streams, especially in the dry season. In the rainy season, those species have wide distributions and may be found on valleys, slopes, and plateaus. Epipedobates femoralis occurred only on the plateau in the second site in both the rainy and dry seasons. Food Availability.-All species forage principally on the surface of the litter. However, litter arthropods migrate between the surface and the first few centimeters of litter depending on climatic conditions. The pool of invertebrates generally available was estimated from samples of litter and soil to a depth of 5 cm, collected with a 6.4 x 6.6 cm soil corer. A total of 144 samples were collected from February 1991 to March 1992 (except September and October 1991). The invertebrates were extracted in Berlese funnels for eight days and identified to order or family. The samples were collected at different points distributed throughout the length of each trail in each month so that all areas in which frogs were collected or observed were sampled. There was little variation in the proportion of prey types in samples of availability throughout the year. Therefore, arthropod availability used in analyses of electivity was based on the pooled sample for the whole year. The extraction of invertebrates from litter samples using Berlese funnels, when not done with care, could sample 10% or less of mites and collembolans in the sample (Toft and Levings, 1979). In this study, extractions were done slowly and carefully, and mites constituted 65% of the fauna and collembolans 28%, proportions similar to those encountered in detailed studies (Simon and Toft, 1991). Analyses of Diet.-A total of 1065 frogs were used for the diet analyses. Approximately 20% of these were collected during 2-3 d periods each month from April 1988 to October 1989 (except June and July 1989) and about 60% were collected from January 1991 to March 1992. I walked all trails in both areas, capturing every frog within one meter of the center of the trail between 0800h and 1700h. The remainder of the nimals wer collected from March 1993 to May 1995. During that period, I walked all trails capturing only fr gs in ize classes that were lacking from the origi al collections. The animals were killed immediately and fixed in 10% formalin. Their lengths, from the tip of the snout o the posterior end of the urostyle (SUL), were measured with Vernier callipers after preservation. For ana yses of di t, I used only items in the stomach. They were sorted into prey categories (generally orders or family). Prey categories that contributed less than 10% (mass) of the diet of all size ca egories were grouped as other invertebr tes or larvae, For analyses of electivity, I considered only seven prey categories (Collembola, Acarina, Formicidae, COA, Isoptera, insect larv e, and invertebrates). The COA category included Coleoptera, Orth ptera, and Ar neae. Those categories were combined based on the supposition that sit-andwait foragers will be more efficient in capture of mobile species of those prey types (Huey and Pianka, 1981; Toft, 1985; Pough and Taigen, 1990). I d d not group this category with other invert rates, because it may contain important independent information about the diet composition of the species. Sizes of prey were measured under a dissecting microscope with a gridded eyepiece. In the case of disarticulated items, the original size of the prey was estimated by comparison with a reference collection. Wet masses (excess water removed with blotting paper) of the preserved animals were determined on a digital balance accurate to 0.0001 g. I estimated the masses of animals that were lighter than the limit of reading of the balance by grouping 10-100 preserved arthropods of similar size from the same taxon, weighing them, and calculating the mean mass. Statistical Analyses.-Electivities were based on numbers of individuals rather than mass and calculated by the method of Jacobs (1974): D = (Rk Pk)/[(Rk + Pk)(2RkPk)], where Rk is the proportion of prey k in the stomach contents and Pk is the proportion of prey k 393 This content downloaded from 157.55.39.96 on Sun, 02 Oct 2016 05:58:51 UTC All use subject to http://about.jstor.org/terms