Environmental parameters affecting the structure of leaf-litter frog ( Amphibia : Anura ) communities in tropical forests : A case study from an Atlantic Rainforest area in southeastern Brazil

Despite a recent increase of information on leaf litter frog communities from Atlantic rainforests, few studies have analyzed the relationship between environmental parameters and community structure of these animals. We analyzed the effects of some environmental factors on a leaf litter frog community at an Atlantic Rainforest area in southeastern Brazil. Data collection lasted ten consecutive days in January 2010, at elevations ranging between 300 and 520 m above sea level. We established 50 quadrats of 5 x 5 m on the forest floor, totaling 1,250 m2 of sampled area, and recorded the mean leaf-litter depth and the number of trees within the plot, as well as altitude. We found 307 individuals belonging to ten frog species within the plots. The overall density of leaf-litter frogs estimated from the plots was 24.6 ind/100m2, with Euparkerella brasiliensis (Parker, 1926), Ischnocnema guentheri (Steindachner, 1864), Ischnocnema parva (Girard, 1853) and Haddadus binotatus (Spix, 1824) presenting the highest estimated densities. Among the environmental variables analyzed, only altitude influenced the parameters of anuran community. Our results indicate that the study area has a very high density of forest floor leaf litter frogs at altitudes of 300-500 m. Future estimates of litter frog density might benefit from taking the local altitudinal variation into consideration. Neglecting such variation might result in underestimated/overestimated values if they are extrapolated to the whole area.

ABSTRACT.Despite a recent increase of information on leaf litter frog communities from Atlantic rainforests, few studies have analyzed the relationship between environmental parameters and community structure of these animals.We analyzed the effects of some environmental factors on a leaf litter frog community at an Atlantic Rainforest area in southeastern Brazil.Data collection lasted ten consecutive days in January 2010, at elevations ranging between 300 and 520 m above sea level.We established 50 quadrats of 5 x 5 m on the forest floor, totaling 1,250 m 2 of sampled area, and recorded the mean leaf-litter depth and the number of trees within the plot, as well as altitude.We found 307 individuals belonging to ten frog species within the plots.The overall density of leaf-litter frogs estimated from the plots was 24.6 ind/100m 2 , with Euparkerella brasiliensis (Parker, 1926), Ischnocnema guentheri (Steindachner, 1864), Ischnocnema parva (Girard, 1853) and Haddadus binotatus (Spix, 1824) presenting the highest estimated densities.Among the environmental variables analyzed, only altitude influenced the parameters of anuran community.Our results indicate that the study area has a very high density of forest floor leaf litter frogs at altitudes of 300-500 m.Future estimates of litter frog density might benefit from taking the local altitudinal variation into consideration.Neglecting such variation might result in underestimated/overestimated values if they are extrapolated to the whole area.KEY WORDS.Frog richness; density estimates; environmental factors.
ZOOLOGIA 31 (2): 147-152, April, 2014 density and richness for the area.Our aim was to investigate the factors explaining the local ecological parameters of tropical frog communities.

MATERIAL AND METHODS
The study was carried out at the Reserva Ecológica de Guapiaçu (hereafter REGUA, 22°24'S, 42°44'W), municipality of Cachoeiras de Macacu, Rio de Janeiro State, southeastern Brazil.The area is inserted within one of the largest remnants of Atlantic forest in the state (over 60,000 ha), most of which is encompassed by the Três Picos State Park.Most of the area is covered with Atlantic Rainforest in different levels of conservation, with remnants of undisturbed forests occurring in the higher and less accessible areas of the reserve.The climate is wet and warm, with mean annual rainfall of about 2,600 mm and daily temperatures ranging from 14 to 37°C (BERNARDO et al. 2011).
We sampled during ten consecutive days (from 20 to 29 January 2010), at elevations ranging between 300 and 520 m above sea level.We established 50 plots (quadrats) of 5 x 5 m (25 m 2 ) (see JAEGER & INGER 1994) on the forest floor (five per day), totaling 1250 m 2 of sampled area.Each quadrat was an independent sampling unit and was at least 100 m apart from the nearest one.While setting the plots, we also maintained a minimum distance of 10 m from streams, in order to minimize the effect of the proximity of water bodies on the frogs (except in one case, where the plot was set ca. 6 m from a stream).During the day, we delimited quadrats by completely enclosing them with a 80 cm high soft mesh fence.The bottom of the fence was buried or attached to the ground with strings and sticks, to prevent frogs from escaping the plot (ROCHA et al. 2001).After sunset, each plot was carefully searched for frogs by a crew of four people wearing head lamps.During searches, each crew member walked the entire plot on hands and knees, side-by-side, using hand rakes.Inside each plot, crew members checked for frogs on and underneath leaves and pieces of dead wood, rock crevices and spaces between tree roots.Additionally, they overturned fallen branches and stones.Searches in each plot lasted about 20-30 minutes.All anurans encountered within a given plot were identified to species and released after the crew had finished searching.Although no individuals were collected in the present study (except for a caecilian), voucher specimens of all frog species recorded herein, collected at the study area during previous fieldwork, are deposited at the Museu Nacional, Rio de Janeiro (see Appendix).For each plot, we recorded the altitude, estimated the mean leaf-litter depth (by measuring the litter depth at each corner of the plot and calculating the mean value), and counted the number of trees with a trunk diameter (DBH) wider than 50 mm.
The evaluation of collection effectiveness was undertaken by a species accumulation curve (collector curve) and by estimators (Bootstrap and Chao 1), using the program EstimateS 8.2.0 (COLWELL 2005).To analyze the relative importance of some environmental parameters such as leaf litter depth, altitude and number of trees affecting the abundance and richness of leaf litter frogs, we performed multiple regression analyses with frog abundance/richness as the dependent variable and environmental parameters as the independent variables.
The influence of the environmental variables altitude, mean leaf-litter depth and number of trees on anuran species distribution (composition data) was assessed through a Canonical Correspondence Analysis (CCA; TER BRAAK 1986).Data on environmental factors were standardized by centering and normalizing.The statistical significance of the species-environment correlation was evaluated by Monte Carlo test (1000 randomized runs).
Those four species accounted for 83.7% of all frogs found in the plots.In addition to frogs, one individual of the caecilian Siphonops hardyi Boulenger, 1888 (Gymnophiona, Siphonopidae) was found inside a plot.The species accumulation curve analysis produced a clear asymptotic shape, and the predicted estimates of frog richness (Bootstrap = 10.8 ± 0.3 species; Chao 1 = 11.0 ± 2.3 species) were close to the richness value obtained by us with plot sampling (10 species) (Fig. 1).
The first three axes of the CCA explained only 9.9% of the variance in species data (6.0, 2.7, and 1.2%, respectively).The biplot scores of the variable "altitude" showed a higher correlation to axis 1 (= -0.345), while "number of trees" had a higher correlation with axis 2 (= 0.216) (Table II).Randomization (Monte Carlo) test did not show statistical significance for species-environment correlations (p = 0.07 for axis 1; P was not reported for axes 2 and 3 because using a simple randomization test for these axes could bias the P values).
Figure 1.Cumulative (black line) and rarefaction (solid gray line; with the standard deviation represented by dotted gray lines) curves of richness of leaf litter frogs recorded at the Atlantic rainforest of Reserva Ecológica de Guapiaçu, in southeastern Brazil, using 5 x 5 plot sampling method.
The overall abundance of frogs and the three environmental variables (altitude, leaf litter depth and number of trees) were not significantly correlated in the results of the multiple regression analysis (r = 0.39, F 3,46 = 2.68, p = 0.06).Partial correlations showed a positive and significant relationship between frog abundance and altitude (p < 0.01), but not for the other variables (leaf litter depth, p = 0.60; number of trees, p = 0.63).
Frog richness was not significantly influenced by the combined effect of the three environmental variables tested, according to the multiple regression analysis (r = 0.36, F 3,46 = 2.22, p = 0.10).Partial correlations showed a positive and significant relationship between frog richness and altitude (p < 0.05), but not for the remaining variables (leaf litter depth, p = 0. 80; number of trees, p = 0.46).(Girard, 1853), are added to the list of leaf litter frogs of the REGUA, previously compiled by ROCHA et al. (2007), and which featured 12 species.However, these previous records include the hylid Scinax aff.xsignatus, a scansorial species that is considered accidental in the leaf litter, and Euparkerella brasiliensis, which was erroneously identified as E. cochranae Izeksohn, 1988 in that work (L.A. Fusinatto, pers. comm.).Thus, there are currently 15 species known to comprise the leaf litter frog community of the REGUA (excluding Scinax aff.x-signatus which we do not consider as a leaf litter inhabitant).Most of those species are typical constituents of anuran communities in the leaf litter floor of Atlantic Forest areas in southeastern Brazil (e.g., ROCHA et al. 2000, 2007, 2011, ALMEIDA-GOMES et al. 2008, SIQUEIRA et al. 2009, ARAÚJO et al. 2009).However, the hylodid C. aeneus is typically found in association with streams (JORDÃO-NOGUEIRA et al. 2006), and the only individual recorded during our study was found in a plot set ca. 6 m from a small stream.
Despite the fact that our focus was on leaf litter anurans, it is worth mentioning that we also recorded the caecilian S. hardyi during our fieldwork.This finding represents another new record of a forest floor amphibian for the REGUA.
In the present study, the estimated density of forest floor leaf litter frogs at the REGUA (24.6 ind/100m 2 ) was about three times higher than that of a previous plot study carried out in the same locality (8.4 ind/100 m 2 , ROCHA et al. 2007).It is possible that the different densities found in the two studies are, at least in part, related to the altitude: from 300-520 m a.s.l. in our study versus 100-400 m a.s.l.(but predominantly at the lower sites) in the study of ROCHA et al. (2007).Another plot study carried out between 500 and 800 m a.s.l., in a site only ca. 15 km distant from the REGUA, also yielded relatively high litter frog density estimates (17.1 ind/100 m 2 ) (SIQUEIRA et al. 2009).It is also possible that the months of sampling may have been responsible for part of the differences between our study (January -peak of the rainy season) and that of ROCHA et al. (2007) (October -beginning of the rainy season), since Atlantic Forest leaf litter frog composition and abundances may vary seasonally (e.g., SANTOS-PEREIRA et al. 2011).
The litter frog density estimated for the REGUA in our study was not only higher than previously reported for the same region, but also higher than in all other Atlantic Forest areas for which similar data are available (see ROCHA et al. 2013).Comparing our data with other studies in different tropical forest regions in South America, the estimated density for the REGUA was higher than in some Amazon rainforest areas in Brazil (3.0-6.0 ind/100 m 2 ; ALLMON 1991) and in Peru (4.4-15.5 ind/100 m 2 ; TOFT 1980a).Compared to rainforest sites in Central America, our estimate was higher (11.5 ind/100 m 2 -HEINEN 1992; 15.7 ind/100 m 2 -LIEBERMAN 1986; 7.5-19.4ind/100 m 2 -TOFT 1980b), lower (30.2 ind/100 m 2 -HEATWOLE & SEXTON 1966), or higher/lower (13.4-62.3ind/100 m 2 -SCOTT JR 1976), depending on the locality and/or time of the year when sampling was performed.On the other hand, the estimated leaf-litter frog density obtained in our study was higher than reported for most Old World tropical forests surveyed so far (e.g., 0.5-2.6 ind/100 m 2 in Thailand -INGER & COLWELL 1977; 1.2 ind/100 m 2 in Borneo -LLOYD et al. 1968;1.5-2.2 ind/100 m 2 in Uganda -VONESH 2001; 1.5 ind/100 m 2 in the southern Western Ghats, India -VASUDEVAN et al. 2008;9.4ind/100 m 2 in Cameroon -SCOTT JR 1982; 3.5-10.2ind/100 m 2 in Taiwan -HUANG & HOU 2004), with one exception (41.8 ind/100 m 2 in Iriomote island, Japan -WATANABE et al. 2005).Hence, our data generally support the idea that higher densities of leaf-litter frogs tend to occur in the Neotropical region compared to the Old World tropics.
We detected a significant effect of altitude on the richness and abundance of the leaf litter anuran community at REGUA, despite the relatively small altitudinal range included in our study (ca. 200 m).The influence of altitude on anuran community parameters could be due to variation in environmental factors such as temperature and humidity, for instance (e.g., GIARETTA et al. 1999).Moreover, it is possible that habitat quality varies with altitude within the area, with more disturbances at lower sites resulting in comparatively less favorable conditions for leaf litter frogs (e.g., ROCHA et al. 2013).A longer-term study monitoring environmental parameters and quantifying habitat quality (or disturbance), including a wider altitudinal gradient, may clarify the relationship between altitude and leaf litter frog density.
Altitude, in our analysis, was the only variable that significantly affected the parameters of the leaf litter frog community.GIARETTA et al. (1999) found that altitude, depth of soil cover, leaf litter mass and fallen trunk area all influence the density of leaf litter frogs at an Atlantic Forest area in the state of São Paulo, Brazil.VAN SLUYS et al. (2007), in a study conducted at the island of Ilha Grande (Rio de Janeiro State), found that mean litter depth and the proportion of leaves in the leaf litter were significantly correlated with abundance and richness of litter frogs ( VAN SLUYS et al. 2007).In a "restinga" habitat (a relatively xeric ecosystem within the Atlantic Forest biome) in northeastern Brazil, the amount of leaves, percentage of leaf litter, percentage of terrestrial bromeliads exposed to direct sunlight, number of terrestrial bromeliads and soil moisture were the environmental variables that better explained the composition of the local anuran community (BASTAZINI et al. 2007).At a site in the Brazilian Amazon, ALLMON (1991) did not obtain any significant results regarding the influence of leaf litter moisture, depth or average dry mass on the abundance of frogs, but plots with frogs had higher average dry litter mass than plots without frogs.FAUTH et al. (1989) found a negative effect of elevation and a positive effect of leaf litter depth on the density and richness of amphibians in Costa Rica.Data from the aforementioned studies indicate that the effect of different environmental variables on the richness and densities of tropical forest floor herpetofauna varies among sites, but the amount/depth of leaf litter seems to be a consistently important variable overall.In the present study, however, litter depth was not significantly correlated with frog abundance or richness, contrary to what would be expected based on previous studies in other areas.
The results of the present study indicate that the area of the REGUA has a very high density of forest floor leaf litter frogs at altitudes of 300-500 m.This, together with the significant effect of altitude on litter frog abundance and richness in our analyses (and comparisons with an earlier study done mostly at lower altitudes in the same area), shows that estimates of litter frog density within a given area should consider the local altitudinal variation, and that data obtained at a limited altitudinal range may lead to underestimated/overestimated values if extrapolated to the whole area.

Table I .
Total number of individuals sampled and estimated density (individuals/100 m 2 , in parenthesis) of each frog species found in the leaf-litter frog community of the Atlantic rainforest of Reserva Ecológica de Guapiaçu, southeastern Brazil, using 5 x 5 plot sampling method.(*) Nine individuals of unidentified species evaded capture.

Table II .
Summary of the CCA performed for three features used to characterize plots and anuran species distribution among them at the Reserva Ecológica de Guapiaçu, southeastern Brazil.The variable scores with the highest influence in the first two axes are shown in boldface.