Differential success in sampling of Atlantic Forest amphibians among different periods of the day

In general, anurans tend to be nocturnal, though diurnal activity is characteristic of some groups. Studies show that frog activity may be inferred based on the number of individuals collected at different periods of the day, during large-scale field surveys. We investigated the best period of the day to conduct amphibian sampling in nine Atlantic Rainforest areas in southeastern Brazil, based on intensive field surveys. At each locality we employed similar sampling effort during diurnal, crepuscular and nocturnal searches (totaling 704.5 sampling hours). We pooled data from all localities for each period and estimated the proportion of frogs of each species active at each period based on the total number of individuals and on the number of species found during all surveys for that period. We recorded a total of 817 individual frogs from 69 species. Species richness was highest at night (median = 12 species), intermediate at dusk (median = 8), and lowest during the day (median = 4). The percentage of the total number of individual frogs found (pooled species) was highest during the night (ca. 53%) and lowest during the day (ca. 14%). Analyzing each species separately, the number of individuals recorded was consistently higher at dusk and night for most species. Our study evidences a trend for nocturnal activity for most Atlantic Rainforest frogs, with few species having primarily diurnal habits. Those results may favor future studies and conservation efforts for amphibian species.


Introduction
Studies investigating amphibian communities commonly employ different sampling methods, such as 'plots' or 'quadrats' (Jaeger and Inger, 1994), pitfall traps (Corn, 1994), and visual encounter surveys -VES (Crump and Scott, 1994).In the Neotropics, VES and plots are the methods that have been the most successful during short-term anuran surveys, for both total abundance and species richness (e.g.Doan, 2003;Almeida-Gomes et al., 2008, 2010;Siqueira et al., 2009).However, studies conducted at different times of day in some areas show that capture success is not equal throughout the day, being the highest during the night (e.g.Rocha et al., 2000Rocha et al., , 2007;;Menin et al., 2008).Therefore, temporal differences in rates of frog activity (by "active" we mean individuals that are not at rest, i.e. that are performing tasks such as calling, mating, foraging, fighting, dispersing, etc) may be inferred based on the number of individuals collected at different periods of the day during field surveys.However, this type of inference has its limitations, since not all the frogs may actually be active at the moment of encounter.Nevertheless, this approach has been used in studies of anuran assemblages/guilds in Atlantic (Rocha et al., 2000(Rocha et al., , 2007;;Almeida-Gomes et al., 2008, 2010;Siqueira et al., 2009) and Amazonian rainforests (Menin et al., 2008) in Brazil.In two studies (Rocha et al., 2000;Menin et al., 2008) a similar proportion of individuals (ca.70%) was recorded during nocturnal sampling, indicating that most anurans in Neotropical rainforests tend to be found at night.In the present study, we evaluate the differences in the success in detecting amphibians among three periods of the day (diurnal, crepuscular and nocturnal) based on the results of intensive short-term surveys carried out in nine Atlantic Rainforest areas in southeastern Brazil.

Material and Methods
Anuran surveys were conducted from 2004 to 2006 in nine areas of Atlantic Rainforest in the state of Rio de Janeiro, southeastern Brazil (Table 1).We sampled frogs using time-constrained visual search surveys (Crump and Scott, 1994) during five to seven consecutive days in each area.Surveys were conducted during the day (09:00H-16:00H), at dusk (17:30H-18:30H) and at night (19:00H-22:00H) in each locality (usually one survey episode per period per day).During the surveys, each observer walked slowly for 30 minutes carefully looking for frogs on the ground, on tree trunks, branches and shrubs, under and on logs, twigs, roots and stones, and on arbustive and herbaceous vegetation.In addition, some water bodies such as temporary and permanent ponds, shallow rivulets and streams were also searched.A total of 1409 survey episodes were performed resulting in 704.5 hours of sampling in the nine areas studied (Table 1).
Individual species were considered as predominantly diurnal, crepuscular or nocturnal if 35% or more individuals were sampled during one of those periods; if approximately the same proportion of individuals was sampled in each of the three periods, we considered the species' activity trend as "extensive".To test for differences among sampling periods (day, dusk and night) in species richness per site and total abundance of frogs per site, we performed one-way analysis of variance (ANOVA) coupled with Tukey post-hoc test, using the software Systat 11.0 (for these analyses we did not consider the locality Estação Ecológica do Paraíso, as the sampling effort was unequal among periods).Because sampling effort was not equal among sampling periods, we used an individual-based rarefaction technique (Gotelli and Colwell, 2001), which provides a richness estimation that is comparable among sampling periods.This analysis was made using software EcoSim 7.71 (Gotelli and Entsminger, 2004), using 1000 iterations.Descriptive statistics are represented in the text as mean ± SD.

Results
We recorded a total of 817 individual frogs from 69 species (Tables 2 and 3).Specimens of Flectonotus recorded during fieldwork may represent more than one species (E.Izecksohn, pers.comm.), but due to taxonomic uncertainties were treated as a single species for the purposes of this study.
The number of species recorded per site varied from nine at the Parque Estadual do Desengano to 21 at the Estação Ecológica do Paraíso (Table 2).The highest species richness was recorded at night (median = 12, range = 6 to 19 frog species), followed by the crepuscular period (median = 8, range = 3 to 14 species) (Table 2).During the diurnal surveys we recorded the lowest number of frog species (median = 4, range = 1 to 6) (Table 2).Species richness differed significantly among sampling periods (ANOVA: F 2,21 = 7.395; p < 0.005), with differences being significant between diurnal and nocturnal periods (p < 0.005), and non-significant between diurnal and crepuscular (p = 0.087), and between crepuscular and nocturnal periods (p = 0.275).
In most areas, except Serra da Concórdia, P. E. do Desengano and P. E. dos Três Picos, more frogs were found at night than during the day or at dusk (Table 2).Overall, most frogs were found during nocturnal (N = 437 individuals or 53.5%) and crepuscular surveys (N = 267 individuals or 32.7%), whereas only 13.8% of individual frogs (N = 113) were found during diurnal samplings (Table 2).The total number of individuals per site (values log-transformed) differed significantly among sampling periods (ANOVA: F 2,21 = 10.982;p = 0.001), with values for the diurnal period being significantly lower than those of the crepuscular (p < 0.01) and nocturnal (p = 0.001) periods, but the latter two not differing between themselves (p = 0.531).
Analyzing each species separately, the number of individuals was consistently higher at dusk and night for most species (Table 3).Fifteen species (21.7%) were found during both the crepuscular and nocturnal surveys, 24 (34.8%) were found only at night, and 15 (21.7%) were found during all three periods (Table 3).Three species (0.4%) were found only during the day, but two of them are represented by a single individual (Table 3).
The individual-based rarefaction (based on N = 113 from diurnal period) gave estimates of higher species richness for the nocturnal period (37.8 spp.) than for the crepuscular (27.6 spp.) and diurnal (23 spp.) periods.

Discussion
In our study, approximately 53% of the frogs were recorded during the nocturnal period.If we consider the comparatively shorter crepuscular period together with the nocturnal one, we have about 86% of the individuals recorded during the dusk and night.Rocha et al. (2000) surveyed another Atlantic forest leaf litter frog assemblage in southeast Brazil using the "quadrat" methodology (see Jaeger and Inger, 1994) and recorded 29% of the individuals (belonging to six species) during diurnal sampling and 71% (nine species) during nocturnal sampling (even though the total sampling effort for diurnal quadrats was twice that of nocturnal ones).Similarly, Menin et al. (2008), studying an anuran community in Central Amazonia (total of 30 species recorded) found 27% of the individuals (eight species) during diurnal sampling and 73% (25 species) at night.In contrast, Summers (2002) found similar values of relative abundance and richness for diurnal and nocturnal surveys of a leaf litter frog assemblage in a Panamanian forest.However, his sampling effort was much greater during the day than at night, and the values of relative abundance for his daylight samples were partly influenced by one species of diurnal dendrobatid that was highly abundant in the area.
Assuming that the number of individuals recorded at each time of the day may reflect (at least in part) amphibian activity, our data suggest some patterns for specific taxonomic groups, even though the sample size for some species was reduced.Hylids, for instance, are predominantly nocturnal, and the terrestrial direct-developers [comprising    Parmelee (1999) and Menin et al. (2008).In these studies, most species in the Leptodactylidae (sensu Fouquet et al., 2013) and Hylidae were characteristically nocturnal.Among the species sampled in our study, only those of the family Hylodidae can be considered as diurnal.Individuals of Hylodes fredi (= H. phyllodes; see Canedo and Pombal, 2007) and Crossodactylus gaudichaudii at the Atlantic forest of Ilha Grande, in the state of Rio de Janeiro, begin activity at sunrise and remain active all day, normally ceasing activity at sunset (Hatano et al., 2002;Almeida-Gomes et al., 2007).In the present study, a few individuals of Hylodes and Crossodactylus were found during crepuscular or nocturnal searches.Some of those were probably close to ceasing their activity when found (one individual of Hylodes pipilans collected at dusk was active and calling).However, all Hylodes spp.found during nocturnal samplings were apparently inactive, most of them resting on vegetation, as reported for other species in the genus (Heyer et al., 1990;Hatano et al., 2002;Narvaes and Rodrigues, 2005).Stream-dwelling frog species such as those in the genera Hylodes and Crossodactylus live in moisture-saturated environments, and are thus less subjected to dehydration, which allows them to be active during the day (Duellman and Trueb, 1994;Haddad and Giaretta, 1999).
Our study shows a trend for nocturnal activity for most frogs of the Atlantic Rainforest, with few species having primarily diurnal habits.Those results may favor future studies, optimizing sampling effort for individual frog species and maximizing the gathering of information about them.

Table 1 .
Number of 30-minutes survey bouts (total = 1409) and sampling time (total = 704.5 h) for three periods (diurnal, crepuscular and nocturnal) of searching for frogs in nine Atlantic rainforest areas of Rio de Janeiro State in Brazil.
SPP = Surveys performed in the period; THSP = total of hours of searching in the period.

Table 2 .
Richness and abundance of frogs found in three sampling periods (diurnal, crepuscular and nocturnal) in nine Atlantic rainforest areas of Rio de Janeiro State in Brazil.
SR = Species richness; SRP = Species richness in the period.The column "Total" indicates the total of individuals found at each locality.

Table 3 .
Sampling frequencies (in absolute numbers and percentages, in parentheses) during diurnal (D), crepuscular (C) and nocturnal (N) periods for 69 frog species sampled in nine Atlantic Rainforest areas in Rio de Janeiro State, southeastern Brazil, with inferred activity trends for each.
IDI = Insufficient data (N < 4) to infer on activity of the species.IDI = Insufficient data (N < 4) to infer on activity of the species.

Table 3 .
Continued... the Terrarana sensuHedges et al. (2008)and represented here by the families Brachycephalidae and Craugastoridae] tend to have crepuscular-nocturnal activity.A tendency for nocturnal or crepuscular-nocturnal activity can also be visualized for the Leptodactylidae, the Bufonidae, and the Cycloramphidae (except for Proceratophrys melanopogon), but less clearly due to insufficient data for many species.Similar trends have been observed for Amazonian rainforest frog communities studied by