Feeding Habits of the Leaf Litter Frog Haddadus Binotatus (anura, Craugastoridae) from Two Atlantic Forest Areas in Southeastern Brazil

Haddadus binotatus is an endemic anuran of the Brazilian Atlantic Forest and currently, there is no information about the diet of this species. We analyzed the diet of two populations of this anuran in two states in southeast Brazil. Samplings were carried out in 2004 in the state of Rio de Janeiro and in 2009 and 2010 in the state of Espírito Santo. Haddadus binotatus presented a rich diet composition, preying 19 prey types. Orthoptera, Coleoptera, and Blattodea were the most important preys in the Rio de Janeiro population, and Orthoptera, Araneae and Hemiptera were the most important in the Espírito Santo population. The diet composition differed numerically between the two localities, but not in terms of volume, which can reflect local differences in the prey availability in the two habitats. The jaw width limited the size of prey, which is expected for predators who swallow the preys without chewing. The proportion of individuals with empty stomachs was higher in the Rio de Janeiro population (39.2%) than in the Espírito Santo population (17.9%), suggesting that the former could be in a lower energy balance. The females of the species were larger than the males, which may result from the production of larger eggs.

Haddadus binotatus (Spix, 1824) (Craugastoridae) is endemic to the Brazilian Atlantic Forest biome, occurring from the states of Bahia to Rio Grande do Sul (Hedges et al. 2008).This species presents direct eggs development, independent of water (Haddad andSazima 1992, Canedo andRickli 2006), and is found mainly at dusk and at night (Rocha et al. 2007(Rocha et al. , 2011)).Haddadus binotatus is an abundant species in the leaf litter of some areas of Atlantic Rainforest in the state of Rio de Janeiro (e.g.Rocha et al. 2007, Almeida-Gomes et al. 2010), representing an important component of leaf litter communities at these localities.An abundant, and relatively large sized species as H. binotatus (Rocha et al. 2007(Rocha et al. , 2011) ) may prey on a variety of items of different sizes, representing locally an important source of death to prey, and potentially regulating some prey populations.
Currently, there is no information about feeding habits for any of the two living species in the genus Haddadus, including H. binotatus.Therefore, we aimed to characterize the diet composition of two populations of H. binotatus, one from an area of Atlantic Rainforest of Rio de Janeiro (hereafter RJP) and another from an area in the state of Espírito Santo (hereafter ESP), in southeastern Brazil.We specifically asked: 1) Which are the food items predominantly consumed by the two populations of H. binotatus?; 2) In which extent do H. binotatus of distinct populations differ in their diet composition?; 3) Does jaw width of the H. binotatus affect the volume of ingested preys?; 4) Does the body size of the H. binotatus affect the number of preys per stomach?; and 5) Does H. binotatus have sexual dimorphism in body size?

MATERIALS AND METHODS
We carried out the study at three areas of the Atlantic Rainforest in two states of Southeastern Brazil (Rio de Janeiro and Espírito Santo).The two areas, the Estação Ecológica Estadual Paraíso (EEEP) (22°29'S, 42°55'W; elevation 150 -300 m) and the Reserva Ecológica de Guapiaçu (REGUA) (22°24'S,42°44'W; elevation 40 -400 m), were located in the municipality of Guapimirim and Cachoeiras de Macacu respectively, both in the central region of the state of Rio de Janeiro.The climate of EEEP and REGUA is wet and warm, total annual rainfall ranges between 2000 mm and 3000 mm, and daily temperature vary from 14°C to 37°C (Bernardo et al. 2011).The third area was in the region of Serra das Torres (21°00'S, 41°14'W; elevation1000 m) in the municipality of Atílio Vivacqua, south of the state of Espírito Santo.The climate of the region is wet and warm, with mean annual rainfall of about 1300 mm, and mean annual temperature of about 24.5°C (INCAPER 2010).The two areas from the state of Rio de Janeiro (RJP), EEEP and REGUA, are about nine km apart from each other, whereas the area sampled in Serra das Torres (in the state of Espírito Santo) (ESP) was nearly 230 km away from the localities in Rio de Janeiro.
We did Haddadus binotatus samplings during September of 2004 in EEEP and October of 2004 FEEDING HABITS OF Haddadus binotatus in REGUA, in the state of Rio de Janeiro, and during August, September and December of 2009 and March of 2010 in Serra das Torres, in the state of Espírito Santo.In EEEP and REGUA, we used two sampling methods: large-plot (quadrat) (Jaeger and Inger 1994) and time-constrained visual search (Crump and Scott Jr 1994) for searching the frogs.In Serra das Torres, we only used the large-plot sampling method.For the large-plot method, we established both in REGUA and in EEEP 28 quadrats of 5 x 5 m (25 m²), totaling 56 plots (1400 m² of forest floor sampled), and 180 quadrats of 4 x 4 m (16 m 2 ) in Serra das Torres (2880 m²).We searched each plot for about half an hour and only at night [for a complete description of large-plot method see Rocha et al. (2007)].For time-constrained visual search, we performed transects of 30 minutes distributed along three periods of the day, of which 79.5 h were of diurnal, 60 h were of crepuscular and 71 h were of nocturnal search, totaling 212.5 hours of sampling effort in each area.During each transect, the observer moved at a slow walking pace, carefully searching the surroundings for the presence of H. binotatus.All individuals of H. binotatus found during samplings were collected, euthanized with lidocaine ointment (anesthetic) or 10% ethanol, about 3 -4 hours after the capture, and fixed in 10% formalin.All procedures used to capture and euthanasia followed the guidelines of the Herpetological Animal Care and Use Committee (HACC) of the American Society of Ichthyologists and Herpetologists (available at: http://www.asih.org/publications),aiming to reduce animal suffering, and were authorized by the Instituto Chico Mendes de Conservação da Biodiversidade (ICMBio) and the Instituto Estadual do Ambiente (INEA).We deposited specimens at the herpetological collection of the Museu Nacional, Rio de Janeiro.
We measured specimens in their snout-vent length (SVL) and in their jaw width (JW) with a 100 mm Digital Vernier Calliper (Maplin) (precision of 0.01 mm).We dissected the individuals, removed the stomachs and identified the sex through gonadal analyses.We analyzed their stomach contents qualitatively and quantitatively under a K-700L Stereo Microscope (Motic).We identified all prey items to the taxonomic level of Order and Family, in the case of Formicidae.Arthropod remains that could not be identified were grouped in the category "Arthropod remains".We analyzed the diet of H. binotatus in terms of number, volume (in mm 3 ) and frequency of occurrence of each type of prey.We counted prey items and measured each one in their width (W) and length (L) with a 100 mm Digital Vernier Calliper (Maplin) (precision of 0.01 mm).We estimated the volume (in mm 3 ) for each prey category by using the formula for an ovoid-spheroid: V = 4/3π (L/2) (W/2) 2 , where "L" is the length and "W" the width of a given prey item (Dunham 1983).We obtained the frequency of occurrence of each prey category in the diet dividing the number of stomachs which contained that category by the total number of stomachs analyzed, with the exception of the empty ones.For each item, we also obtained an importance index (I x ) through the mean of number, volume, and frequency of occurrence proportions (Howard et al. 1999).
For data analysis, we considered the samples from the state of Rio de Janeiro (EEEP and REGUA) as one population (RJP), given the proximity of these two localities, which belong to the same mountain range named Serra dos Órgãos.To evaluate the feeding strategy (specialist or generalist) and width niche of H. binotatus populations from Espírito Santo and Rio de Janeiro, we did a "specialization tendency diagram" (Amundsen et al. 1996).In the diagram, we considered the proportion of the prey item only for the individuals that consumed it [P i =(ΣS i /ΣS ti ) x 100] where S i is the stomach content composed of the prey i, and S ti represents the total of stomach LÍVIA COCO et al. content in only those predators with the prey i in their stomachs (Y axis) and the frequency of occurrence of the prey items in the diet [F i =(N i /N) x 100] where N i represents the number of predators that had the prey i in their stomach and N represents the total number of predators with stomach contents (X axis).To evaluate if diet composition differed between populations we performed a Kolmogorov-Smirnov test for two independent samples, based on numeric and volumetric proportions of prey items (Zar 1999).We also estimated the trophic similarity between the numerical and volumetric proportions of prey categories of the two populations using the "MacArthur and Levins Similarity Equation" modified by Pianka (1986) , where p ij and p ik are the volumetric/numerical proportions of prey item category i present in the diet of populations j and k, respectively.We estimated the correlations between the mean volume of the prey ingested and the JW of the individuals, and between the number of prey per stomach and the SVL of frogs using the Spearman Rank Correlation (Zar 1999).For all diet analyses we excluded plant pieces and arthropod remains.Finally, to test if there was sexual dimorphism in body size (SVL) for each population, we used the one-way analysis of variance (ANOVA) (Zar 1999).We presented the results for the descriptive statistics throughout the text as mean ± standard deviation.We logtransformed the variable whenever the criteria for normality and homoscedasticity were not achieved (Zar 1999).We performed all statistical analyses using statistic software (SYSTAT 11 ® , SYSTAT Software Inc.).

RESULTS
We analyzed 79 individuals of Haddadus binotatus, of which 51 were from RJP and 28 were from ESP.Twenty frogs (39.2%) from RJP and five (17.9%) from ESP had empty stomachs.Diets of both populations were composed by arthropods.We identified 19 food items consumed by H. binotatus from RJP.
The data showed that all categories of prey were consumed by few individuals in both populations.In both graphics, we observed that the two populations of H. binotatus had a high betweenphenotype contribution to the niche width, both of them with wide trophic niche (Figs. 1, 2).Diet composition differed between populations in terms of numerical proportion (D max = 0.450; P = 0.023), but not volumetrically (D max = 0.200; P = 0.749).The trophic similarity between the two populations (O jk ) in terms of number was 0.68 and 0.58 in terms of volume.

DISCUSSION
Haddadus binotatus of RJP and ESP fed on a wide array of arthropods, with 20 prey types identified to the level of Order.Leaf litter frogs of the Atlantic Forest fed upon a variety of prey from different Classes as Arthropoda, Gastropoda, and even vertebrates as anurans (e.g.Van Sluys et al. 2001, Boquimpani-Freitas et al. 2002, Marra et al. 2004, Almeida-Gomes et al. 2007, Dietl et al. 2009, Martins et al. 2010, Almeida-Santos et al. 2011, Klaion et al. 2011, Sugai et al. 2012).The number of prey types registered to the taxonomic level of Order can also differ considerably among leaf litter frog species, varying from six (e.g.Boquimpani-Freitas et al. 2002) to 22 items consumed (e.g.Siqueira et al. 2006, Teixeira et al. 2006).In other biomes, such as the Cerrado and the Amazonia, the number of preys identified to this taxonomic level and consumed by leaf litter frogs, also seemed to be around 22 (e.g.Biavati et al. 2004, Ortega et al. 2005).The most important preys were Orthoptera, Blattodea and Coleoptera in RJP and Orthoptera and Araneae in ESP.It has been suggested that Blattodea, Coleoptera, and Orthoptera are important prey for large leaf litter frogs, as they present large body sizes and mass, which provide an appropriate amount of food and energy (Boquimpani-Freitas et al. 2002, Klaion et al. 2011).Other arthropods such as Isopoda, Chilopoda and Araneae are also recognized as being an important volumetric contribution to the diet of medium-sized frogs (Van Sluys et al. 2001, Dietl et al. 2009).Although arthropods are the main prey of leaf litter frogs (e.g.Lima and Moreira 1993, Marra et al. 2004, Almeida-Gomes et al. 2007, Almeida-Santos et al. 2011), the diet composition of frog species may be composed by similar prey types but with different relative importance.This fact can be observed in some species that coexist with H. binotatus in the forest floor.For example, at Ilha Grande (RJ), Orthoptera was the most important prey of Proceratophrys tumpinamba (Cycloramphidae) (Boquimpani-Freitas et al. 2002), whereas Araneae, Isopoda and Formicidae were the main preys of the sympatric and syntopic Ischnocnema parva (Brachycephalidae) (Marra et al. 2004).Furthermore, the most important preys of Stereocyclops incrassatus (Microhylidae) at a locality in Espírito Santo were Formicidae, Coleoptera, and Isopoda (Teixeira et al. 2006).Similarly, Rhinella ornata (Bufonidae) consumed mainly Formicidae, Coleoptera and Orthoptera in EEEP and REGUA (Maia-Carneiro et al. 2013), the same areas of the present study in the state of Rio de Janeiro.However, Proceratophrys boiei, a syntopic species with similar size to that of H. binotatus, had not only a similar diet, composed mainly by Coleoptera, Blattodea, and Orthoptera, but the prey shared similar importance in their diets (Giaretta et al. 1998, Klaion et al. 2011).The preference for different prey types, in leaf litter frog assemblages, may be evolved as a mechanism to allow the coexistence of species with similar food and habitat requirements, or simply may reflect the more frequent encountered prey.
In both populations, some individuals of H. binotatus had vegetal remains in stomachs but in different proportions.Plant remains were usually represented only by pieces of dead leaves.The occurrence of plant remains has been reported in diet studies of some anuran species (e.g.Marra et al. 2004, Siqueira et al. 2006, Martins et al. 2010, Almeida-Santos et al. 2011), and its ingestion has been usually considered non-intentional and supposed to be ingested accidentally when the anuran swallows the prey among dead leaves on the forest floor (e.g.Van Sluys et al. 2001, Martínez-Coronel andPérez-Gutiérrez 2011), which seems to be the case of H. binotatus.Only for the hylid Xenohyla truncata has the ingestion of plant matter been considered intentional (Silva et al. 1989, Silva andBrito-Pereira 2006).Our results indicated that H. bonotatus feed mainly on arthropods and has a relatively generalized diet, amongst the richest of the Atlantic Forest in terms of taxonomic level of Order.Furthermore, Haddadus binotatus was among the frog species that had the higher density and mass per hectare in some studied sites at the Atlantic rainforest (EEEP) (Rocha et al. 2007(Rocha et al. , 2011)), which suggests that the species can play an important role on its prey populations at least at these sites.
We found a slight difference in number and volume of prey types consumed by the two populations of H. binotatus, which can reflect, at least partially, local differences in the prey availability of the two habitats of the Atlantic Rainforest, since invertebrate communities can change among areas (Lacerda et al. 1998, Santos et al. 1998) or as a result of seasonal changes (Toft 1980b, Anderson et al. 1999, Wells 2007) which in the case of our study, can also be a result of the sampling periods.
Mouth size in the H. binotatus of ESP explained approximately 36% of the variation in prey volume LÍVIA COCO et al. consumed, suggesting that prey size can be adjusted and constrained by mouth size of frogs (although this trend did not appear for the RJP).It is expected that animals that cannot handle their prey and need to swallow them whole, tend to concentrate their diet on relatively large prey (Siqueira et al. 2006) as observed for H. binotatus in the present study and for other anurans (e.g.Ortega et al. 2005, Martínez-Coronel andPérez-Gutiérrez 2011), most of them from Atlantic Forest areas (e.g.Boquimpani-Freitas et al. 2002, Dietl et al. 2009, Marra et al. 2004).As a consequence, anurans tend to show an adjustment between prey size and mouth size (Van Sluys et al. 2001, Jordão-Nogueira et al. 2006, Martins et al. 2010, Klaion et al. 2011).In some cases in where this relationship between prey size and mouth size is not clear or does not occur (as may be the case of RJP), usually a high consumption of small prey (low mean volume) or similar-sized prey, as Formicidae (Almeida-Gomes et al. 2007), Collembola, Isopoda and Acari, or the small size of frog (Van Sluys et al. 2006, Almeida-Santos et al. 2011) are involved.In fact a comparatively lower mean volume of prey in frog stomachs was registered for the RJP which could have prevented such relationship to be found.Additionally, we did not find correlation between the number of prey ingested and the frog body size indicating that a large size did not imply in an increase in the number of prey ingested.This result may be expected for frogs that select the prey size, according to their mouth size, as we found at Espírito Santo, and as it is expected for frogs that prefer to ingest larger preys to maximize the cost of capture.
The proportion of individuals having empty stomachs in the population of RJP (39.2%) was more than twice that of the ESP (17.9%) which suggests that the two H. binotatus populations could have had differences in their energy balance at the moment of the collection of the individuals.The observed proportion of individuals of a species having empty stomachs has been used as an index of instantaneous energy balance for lizards (Huey et al. 2001) and fishes (Arrington et al. 2002) in which individuals having empty stomachs are assumed to be in negative energy balance and dependent on reserve stores (Boivin and Power 1990, Huey et al. 2001, Arrington et al. 2002) whereas individuals having food in their stomachs would be gaining energy and thus, would be in positive energy balance (Huey et al. 2001).We still do not know to which extent this energy trends work among frogs but our data is suggestive that, comparatively, the RJP was in lower energy balance when compared with ESP.
Our data showed that there is sexual dimorphism in body size of H. binotatus, with females being larger than males.Sexual size dimorphism has been reported to leaf litter anurans (e.g.Woolbright 1989, Boquimpani-Freitas et al. 2002, Lee 2001, Van Sluys et al. 2006), and a positive relationship between the number of eggs and female size (Prado and Haddad 2005, Martins et al. 2010) is usually associated with advantages of females producing more eggs or larger eggs, although anurans with direct development have, in general, smaller clutches with larger eggs (Wake 1978, Zug et al. 2001, Almeida-Santos et al. 2011).As clutch size and female size is not positively related in H. binotatus (Canedo and Rickli 2006), the larger sizes attained by females could result from evolutionary forces favoring the production of larger eggs by females.
We conclude that the leaf litter frog Haddadus binotatus has a relatively generalized diet (20 prey types) which is composed mainly of arthropods being Orthoptera, Coleoptera, and Blattodea the most important preys among Rio de Janeiro population, and Orthoptera and Araneae the most important among the Espírito Santo population.Populations of this frog may differ somewhat in diet, potentially attributable to differences of the environments they live in.The size of H. binotatus mouth affected the size of prey consumed, as it is expected for frogs that swallow preys, according to their mouth size.The sexual dimorphism of FEEDING HABITS OF Haddadus binotatus the H. binotatus in its size, with females being comparatively larger than the males, could result from advantages of larger females being favored by the production of larger eggs.

Figure 1 -
Figure 1 -Specialization diagram of Haddadus binotatus population from an Atlantic Rainforest area in the state of Espirito Santo, southeastern Brazil.The Y axis (Pi) shows the proportion of the prey item only for the individuals that consume it.The X axis shows the frequency of occurrence of the prey items in the diet.Each letter inside the diagram represents one prey type (A = Acari; B = Araneae; C = Blattodea; D = Chilopoda; E = Coleoptera; J = Hemiptera; K = Homoptera; L = Isopoda; N = Lepidoptera; O = Mantodea; P = Opiliones; Q = Orthoptera; S = Pseudoscorpiones; T = Thysanoptera).

Figure 2 -
Figure 2 -Specialization diagram of Haddadus binotatus population from an Atlantic Rainforest area in the state of Rio de Janeiro, southeastern Brazil.The Y axis (Pi) shows the proportion of the prey item only for the individuals that consume it.The X axis shows the frequency of occurrence of the prey items in the diet.Each letter inside the diagram represents one prey type (A = Acari; B = Araneae; C = Blattodea; D = Chilopoda; E = Coleoptera; F = Dermaptera; G = Diplopoda; H = Diptera; I = Hymenoptera (Formicidae); J = Hemiptera; K = Homoptera; L = Isopoda; M = Isoptera; N = Lepidoptera; O = Mantodea; P = Opiliones; Q = Orthoptera; R = Scorpiones).

Figure 3 -
Figure 3 -Relationship between the mean volume of prey (mm 3 ) and the jaw width (mm) of individuals of Haddadus binotatus population from an Atlantic Rainforest area in the state of Espírito Santo, southeastern Brazil (r s = 0.6025; P = 0.0063, n = 19).

TABLE I Number (N), volume (V, in mm 3 ), frequency of occurrence (F) and Index of Importance (I x , in %) of each food item in the diet of the anuran Haddadus binotatus in Atlantic Rainforest of the states of Rio de Janeiro (RJ) (n = 31) and Espírito Santo (ES) (n = 23), southeastern Brazil. Percentage values
(%) in brackets.LÍVIA COCO et al.