Given the increasing use of sympatric species to investigate niche differentiation and resource partition in biological communities, our study analyzes diet composition differences and trophic niche overlap between the sympatric species Physalaemus cuvieri and P. atim in the municipalities of Silvânia and Leopoldo Bulhões, Brazil. We used stomach flushing to obtain stomach contents from each individual and identified 11 prey categories for P. cuvieri and 17 for P. atim. Isoptera had the highest proportional volume in both species. No difference was found in the mean prey volume per stomach. Detrended correspondence analysis distribution scores indicated a greater diet breadth for P. atim than P. cuvieri, which is consistent with a higher trophic niche breadth in P. atim than P. cuvieri. According to a multi-response permutation procedure analysis, P. cuvieri and P. atim have different diet compositions. Our findings suggest that these congeneric species occupy sufficiently different dietary niches to enable them to persist in sympatry.
INTRODUCTION
Ecologists have applied sympatric species models seeking to understand species coexistence and community structure by investigating niche differentiation and resource partitioning in biological communities. Coexistence of sympatric species derives from coevolution through their process of adapting to a habitat by avoiding competition with reduced overlap of resource use regarding at least one niche dimension (Leibold and McPeek, 2006). Thus, spatial and temporal dimensions as well as food availability are regarded as the most important factors in niche differentiation (Pianka, 1974; Schoener, 1974).
Many studies have found differences in the trophic niches of sympatric species of tropical anurans, including findings on food resource overlap (e.g., de Paula-Lima et al., 2010; Sabagh et al., 2012; Oliveira et al., 2015) and identification of food resource partitioning in prey size, feeding time period, and predator size classes (e.g., Lima and Magnusson, 1998; Menin et al., 2005). These studies are useful to determine ecological differences of closely related species.
Physalaemus Fitzinger, 1826 is one of the most speciose genera of Leptodactylidae Werner, 1896 (1838), with 48 recognized species widely distributed throughout South America, including from the Guianas, the lowlands of southern Venezuela, the llanos of southeastern Colombia, western Ecuador, Bolivia, Paraguay, Uruguay, northern and central Argentina, and Brazil (Nascimento et al., 2005; Lourenço et al., 2015; Frost, 2020). Brazil harbors 44 species of Physalaemus (Frost, 2020). Physalaemus cu– vieri Fitzinger, 1826 is widely distributed throughout South America, while P. atim Brasileiro and Haddad, 2015 has been recorded only in the municipalities of Campo Limpo de Goiás (type locality) and Terezópolis de Goiás (Brasileiro and Haddad, 2015; Frost, 2020). Distributions of both species overlap in central Brazil, with sympatry in open Cerrado areas of Goiás state. Detailed studies on the diet of P. cuvieri have been carried out in some localities (Moreira and Barreto, 1996; Becker et al., 2007; Araújo et al., 2009; Menin et al., 2015; Leivas et al., 2018), but no studies on the diet of P. atim have been conducted.
Our study provides information on the diet and trophic niche overlap of Physalaemus cuvieri and P. atim in aquatic habitats of open Cerrado areas in central Brazil and report on (1) the diet composition of the species at the study site and (2) the food niche overlap between them. Physalae– mus species are morphologically, behaviorally, and phylogenetically similar, which suggests that sympatric species of the genus might show similar dietary compositions, as has been reported for other sympatric anuran species (Sabagh et al., 2012; Oliveira et al., 2015). However, we expect to find differences in dietary composition, as this would enable their sympatry by way of low trophic niche overlap.
MATERIALS AND METHODS
Study site
Our study encompassed nine cattle ranches (Table 1) located in the municipalities of Silvânia and Leopoldo Bulhões, central Goiás state, central Brazil (Fig. 1). The ranches are characterized by Cerrado sensu lato physiognomies (campo limpo, areas without shrubs or trees; campo sujo, areas with scattered small trees and shrubs; Oliveira-Filho and Ratter, 2002) surrounded by extensive pasture areas. We surveyed one water body per ranch, totalling nine water bodies (Fig. 2), which were selected due to the sympatric occurrence of Physalaemus cuvieri and P. atim in those locations.
Figure 1.
Map of the study sites. Distribution of the surveyed water bodies (survey points) in nine cattle ranches in the municipalities of Silvânia and Leopoldo Bulhões, Goiás state, central Brazil. See general views of the surveyed water bodies in Figure 2.

Table 1.
Surveyed water bodies in nine cattle ranches in the municipalities of Silvânia and Leopoldo Bulhões, Goiás state, central Brazil.

DATA COLLECTION
We conducted 29 expeditions between December 2013 and February 2014 during the breeding season of most anuran species of the region, using randomized order to avoid temporal pseudoreplication. Nocturnal surveys (18:00–00:00) were performed through visual and auditory searching (Scott and Woodward, 1994) to detect Physalaemus cuvieri and P. atim individuals (Fig. 3).
The stomach-flushing method (Solé et al., 2005) was employed to obtain each individual's stomach contents, and all retrieved food items were preserved in 70% ethanol (Luna, 2005). To avoid flushing the same frog more than once, all individuals were marked using fluorescent alphanumeric labels inserted into the subcutaneous inter-femoral lymph sacs (Clemas et al., 2009). Voucher specimens were deposited in the Coleção Zoológica da Universidade Federal de Goiás (Physalaemus atim: ZUFG 8497–8500; P. cuvieri: ZUFG 8504–8512).
We performed the following laboratory procedures: stomach content analysis with the aid of a stereomicroscope (Resh and Cardé, 2003); diet analysis based on occurrence, number, and volume per prey category identified to order or family; prey category measurement, expressed as the total volume percentage through the volumetric method (Hyslop, 1980) using a glass counting plate (Hellawell and Abel, 1971); incidence calculation according to the frequency of occurrence of each prey catagory per individual of each species. In addition, we obtained the volume of each item in mm3 using a milimiter plate for subsequent transformation into mL (Hellawell and Abel, 1971). Preys, especially insects, were highly digestible, because of this we didn't weigh or count the items.
Figure 2.
General views of the surveyed water bodies (survey points) in nine cattle ranches in the municipalities of Silvânia and Leopoldo Bulhões, Goiás state, central Brazil. See geographic coordinates of the survey points in Table 1.

Data analysis
We calculated the feeding index (IAi; Kawakami and Vazzoler, 1980) to assess each species' diet on the basis of the total volume and occurrence frequency per food item according to the following equation:
where IAi is the feeding index, FO is occurrence frequency expressed as a percentage, VO is volumetric frequency expressed as a proportion, and i is the number of food items. We performed a detrended correspondence analysis (DCA; Gauch Jr., 1982) to identify gradients in food item composition as well as a multi-response permutation procedure (MRPP) to compare the diet compositions of Physalaemus cuvieri and P. atim. All analyses were carried out using the PC-ORD software for Windows, Version 5 (MacCune and Mefford, 1999).
We calculated Levins' Index (Levins, 1968) to determine trophic niche amplitude according to the following equation:
where B is the trophic niche amplitude value and Pi is the proportion of item i volume for a species. At least two variables with values above 2.36 indicates generalist species (Rossa-Feres and Jim, 2001). We calculated the Trophic Niche Overlap index of Pianka (Ojk; Pianka, 1973) to determine trophic niche overlaps between two species based on degree of similarity, according to the following equation:
where Ojk is the niche overlap index between species j and k; pij and pik are the proportions of prey categories consumed by the species; and j, k, and n are the total number of resource categories consumed by species j and k. Index ranged from 0 to 1, in which values equal to zero indicate absence of overlap, and values equal to 1 point out to a complete overlap between species' diets (Krebs, 1999). This analysis was performed using EcoSim Version 7.71 (Gotelli and Entsminger, 2012).
RESULTS
We examined 67 Physalaemus cuvieri specimens, among which 53 individuals (79%) had stomach contents, as well as 58 P. atim specimens (Table 2), including 45 individuals (78%) with stomach contents (Table 2).
We identified 11 prey categories for Physalaemus cuvieri and 17 for P. atim (Table 2). Except for Vespoidea Latreille 1802 all prey categories found in P. cuvieri were also recorded in P. atim. Preys consumed exclusively by P. atim included larvae of Coleoptera Linnaeus, 1758, Diptera Linnaeus, 1758, and Lepidoptera Linnaeus, 1758 as well as adults of Blattaria Burmeister, 1829, Collembola Lubbock, 1873, and Gastropoda Cuvier, 1795. Addicionally, we found an aquatic invertebrate (Bryozoa Ehrenber, 1831) as prey of P. atim (Table 2). Isoptera Brullé, 1832, Formicidae Latreille, 1809, and Coleoptera were the most numerous and frequent prey in P. cuvieri, as were Formicidae, Isoptera, and Aranae Clerck, 1757 for P. atim (Table 2). Plant material (i.e., fragments of leaves, twigs, and seeds) were found in the stomachs of 36 P. cuvieri specimens (68%) and 26 P. atim individuals (58%).
Isoptera had the highest proportional volume in the stomachs of individuals of both frog species. The mean volume of preys ingested per Physalaemus cuvieri individual was 12.73 mL, whereas that for P. atim specimens was 9.38 mL. No difference in the mean volume of preys per stomach was found between P. cuvieri and P. atim (F(1.96) = 1.03; P = 0.31). Isoptera was the most important prey category for both frog species (P. cuvieri, IAi = 86.74; P. atim, IAi = 56.47), followed by plant material (P. cuvi– eri, IAi = 8.48; P. atim, IAi = 8.71). Formicidae (IAi = 3.38) and Homoptera (IAi = 8.60; Table 2) were the third most important prey category for P. cuvieri and P. atim, respectively.
The DCA score distribution showed that Physalae– mus atim has greater diet breadth than P. cuvieri, which is consistent with a higher trophic niche breadth in P. atim (B = 4.63) than P. cuvieri (B = 1.30). According to the MRPP analysis, P. cuvieri and P. atim have different diet compositions. Nevertheless, the Trophic Niche Overlap index of Pianka revealed high niche overlap between both species (Ojk = 88.28%).
DISCUSSION
The number of prey categories consumed by Phy– salaemus cuvieri and P. atim was similar to that reported for congenerics, ranging from 12–20 prey categories (Araújo et al., 2009; Rodrigues and Santos-Costa, 2014; Oliveira et al., 2015; Farina et al., 2018). We found P. cu– vieri individuals to consume fewer prey categories than individuals collected in a remnant of original Cerrado vegetation (shrubby grassland areas) surrounded by wet areas in the municipality of Uberlândia, Minas Gerais state (Araújo et al., 2009). We also found P. cuvieri individuals to consume more prey categories than individuals collected in three agroecosystems and forest remnants in the municipalities of Jaboticabal and Guaíra, São Paulo state (Menin et al., 2015) and in disturbed areas within the Araucaria Forest in the municipality of Campina Grande do Sul, Paraná state (Leivas et al., 2018). These differences are due to variation in prey availability in different habitats (Born et al., 2010; Caldart et al., 2012), which regulates the dietary plasticity in P. cuvieri populations at different localities.
Table 2.
Prey categories consumed by sympatric specimens of Physalaemus cuvieri and Physalaemus atim surveyed in water bodies in nine cattle ranches in the municipalities of Silvânia and Leopoldo Bulhões, Goiás state, central Brazil. FO, percentage of frequency of occurrence; V, volume percentage of each prey category; IAi, feeding index.

It is worth emphasizing that even though most amphibians are predators, plant material is commonly found in the stomach contents of anurans (Kovács et al., 2007; Solé and Pelz, 2007). We also found the stomach content of both Physalaemus cuvieri and P. atim to contain plant material, probably ingested accidentally, as reported in previous studies on anurans (Protázio et al., 2015; Vogt et al., 2017). Anderson et al. (1999) concluded that the main functions of plant ingestion in anurans include facilitating insect exoskeleton digestion and parasite elimination and providing water to prevent dehydration. We suggest further behavioral studies to understand how Physalaemus species feed on termites and ants in the leaf litter and determine if ingestion of leaves is intencional or accidental, and analysis of stable isotopes (e.g., Molina-Burgos et al., 2018) would help determine if ingested plant material is assimilated.
Termites and ants have been considered major food items in several anuran species (Forti et al., 2011; Caldart et al., 2012; Oliveira et al., 2015), and also in the diet of Physalaemus species in other regions of Brazil (Santos et al., 2004; Oliveira et al., 2015). Their great importance in both Physalaemus cuvieri and P. atim suggests that these species forage actively and opportunistically by consuming small preys largely available in the environment (Toft, 1981; Solé and Rödder, 2009). Here, Physalaemus species showed preferences for ants, as reported previously for species of this genus, including P. ephippfer (Steindachner, 1864) (Rodrigues and Santos-Costa, 2014), Physalae– mus gracilis (Boulenger, 1883), Physalaemus biligonigerus (Cope, 1861) (Oliveira et al., 2015), and P. cuvieri (Santos et al., 2004).
According to Clarke (1974), the prevalence of these food items is due to their abundants in the environment and exploitation by few predators. Although the majority of Physalaemus food items were terrestrial invertebrates, we identified Bryozoa (aquatic invertebrate) as a prey of Physalaemus atim. Cogălniceanu et al. (2000) found aquatic organisms as prey items in the Pelophylax esculenta complex. Probably this occurs because the frogs inhabit aquatic environments during reproductive periods.
Aposematic coloration in some frogs is associated with alkaloid toxins accumulated in their skins, derived from the consumption of leaf-litter arthropods, especially ants (Saporito et al., 2004). For Physalaemus cuvieri, aposematic coloration appears in the inguinal region, displaying a reddish or orange color pattern. However, there is no evidence that the alkaloid sequestration process occurs in Physalaemus species (Moser et al., 2017). Also, aposematic coloration is associated especially to predators-preys interactions.
High diet overlap also has been reported in Physalae– mus species in sympatry (e.g., Sabagh et al., 2012; Moser et al., 2017). In this study, both Physalaemus atim and P. cuvieri had high diet overlap since termites are highly important in their diet compositions. However, we found variation in food item composition expressed by MRPP analyses, which might be related to their different body sizes and food resource availability, considering that food resource availability determines the diet of opportunistic species (Sabagh et al., 2012). Moser et al. (2017) performed their study in a subtemperate forest and Sabagh et al. (2012) carried out theirs in a forested area in central Amazonia, both of which have different resource food availability than Cerrado.
Variation of food item compositon is reinforced by niche differentiation, indicating that the sympatric occurrence of P. atim and P. cuvieri is driven by limiting environmental conditions required for their coexistence in the surveyed habitat (MacArthur and Levins, 1967). Futhermore, the surveyed locations indicate that P. atim is a more generalist predator than P. cuvieri, which suggests food as a non-limiting factor.
Prey category exclusiveness in sympatric species might be related to differences in foraging behavior (Oliveira et al., 2015) because of trade-offs that can force species to segregate throughout niche dimensions (McPeek, 1996) or food-avoidance to avoid competition (Durbin and Durbin, 1975). Therefore, despite their high overlap, the two species share only a few food items, which differentiates them in the habitat. Shared consumption of other items probably occurs because these preys are abundant in the environment (Pianka, 1974).
In conclusion, Physalaemus atim has higher niche amplitude than P. cuvieri and, consequently, a more generalist diet. Despite their different sizes, high niche overlap occurred between the species; however, they differed in composition, which ensures a degree of segregation in their sympatric environment. These differences can occur due to different microhabitats being used by each species (Moser et al., 2017). Moreover, other niche dimensions could also be explored to explain the coexistence of these two species, not mediated by their differentiation, but by their similarity, which prevents interspecific competition (Mandai, 2014). Further studies on anuran diet are essential to delimit behavior patterns, such as foraging, and achieve a better understanding of the importance of different food items for each species, as well as to test pertinent ecological hypotheses.
ACKNOWLEDGMENTS
The Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) provided financial support during the fieldwork. Alessandro R. Morais and Vinicius Guerra provided critical reading of the manuscript, and Mirco Solé solved all our doubts about the stomach flushing procedure. Priscilla G. Gambale received fellowships from CAPES. Fabrício H. Oda received a fellowship from Fundação Cearense de Apoio ao Desenvolvim– ento Científico e Tecnológico – FUNCAP/CAPES (process # 88887.162751/2018-00). Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) provided a research fellowship (process # 304363/2010-3) to Rogério P. Bastos. Instituto Chico Mendes de Conservação da Biodiversidade/Centro Nacional de Pesquisa e Conservação de Répteis e Anfíbios (ICMBio/RAN process #41168-3) provided a collection permit.