Spatio-temporal patterns of foraging and feeding behavior of Elachistocleis bicolor ( Anura : Microhylidae )

We analyzed spatio-temporal patterns and behavior of the foraging activity of the myrmecophagous frog Elachistocleis bicolor. We delimit ten 1m2 plots and recorded the number of frogs, distance between individuals and foraging behavior at half-hour intervals, during nine days. Using a generalized linear mixed model, we evaluated the maximum, minimum, and average daily temperature, microhabitat temperature, relative humidity, atmospheric pressure, the number of days since the last rain and last 48 hours accumulated rainfall as explanatory variables of frogs’ activity. The variables that best explained frogs’ activity were: mean atmospheric pressure, number of days since last rain and accumulated rainfall. Frogs showed an aggregated distribution pattern when foraging (Morisita standardized index= 0.501). The average distance between frogs was 47.53±25.28 mm. Frogs displayed a combined (active-passive) foraging strategy, actively searching ant trails, digging with their heads on trails running under the mulch, and them passively preying on ants as they pass through. Results showed that the E. bicolor foraging activity followed a circadian cycle that is regulated by several climatic variables, and that these frogs aggregated when foraging for ants.


INTRODUCTION
Studies on amphibians' activity patterns classically refer to calling phenology, which is regulated by several abiotic and biotic environmental cues, such as climatic variables, intraspecific and interspecific social interactions, environmental physiology and reproductive behaviors (Feder and Burggren 1992, Oseen and Wassersug 2002, Saenz et al. 2006, Richter-Boix et al. 2006, Wells 2007, Canavero et al. 2008).Beyond reproduction, feeding is one of the activities that occupies most amphibians' time when they are not aestivating and, even though attention has been paid to some aspects of their feeding behavior, others remain poorly studied (Duellman and Trueb 1986).
There is a wide theoretical frame on feeding ecology of neotropical amphibians.Several topics have been studied such as diet composition of many anuran species, amphibians diet specializations, opportunism and ontogenetic changes, trophic plasticity, foraging modes, intraspecific trophic segregation (e.g.among sexes and age classes) and trophic resources partitioning among sympatric and syntopic species (Toft 1981, Maneyro and da Rosa 2004, Duré et al. 2009, López et al. 2007, 2015, Falico et al. 2012, Matsui 2016).But little is known about environmental variables such as microclimate, food availability, breeding season length, that regulate frogs spatio-temporal patterns of foraging (Feder and Burggren 1992, Solé and Pelz 2007, López et al. 2015).
In The Middle Paraná River floodplain (Argentina), this species inhabits the vicinity of permanent and temporary lentic water bodies, with an intensive reproduction phase during the summer (Sánchez et al. 2009, 2013, López et al. 2011).Its' diet is based on termites or ants or both, depending on prey environment availability (Solé et al. 2002, Berazategui et al. 2007, Cossovich et al. 2011).In environments associated to The Middle Paraná River floodplain, ants comprises about 90% of their diet, being the genus Solenopsis (Formicidae) their most important prey item (López et al. 2007).
It has been suggested that some microhylids actively search for ant trails using olfaction (Duellman and Trueb 1986).Once trails are located, frogs consume these hymenopterans as they pass through the trail.In this study, we aimed to analyze spatio-temporal patterns of foraging and feeding behavior of E. bicolor in a temperate wetland from the floodplain valley of The Middle Paraná River, and to evaluate the influence of abiotic environmental cues on frogs' activity.

MATERIALS AND METHODS
The study was conducted on nine days between December 2010 and February 2011, in the protected area "Reserva Ecológica de la Ciudad Universitaria UNL" (RECU) located in the floodplain of The Middle de forrajeo mixto (activo-pasivo), buscando activamente los caminos de hormigas y cavando con la cabeza en los sitios donde los caminos de hormigas discurrían bajo el mantillo, para luego depredar pasivamente sobre las hormigas que pasaban.Los resultados muestran que E. bicolor posee una actividad de forrajeo que sigue un ciclo circadiano regulado por varias variables climáticas y que estas ranas se agrupan durante la alimentación.
During December 2010 we conducted three exploratory searches to find the sites where frogs foraged and found shelter.Exploratory fieldwork was carried out between 19:00 hrs and 23:00 hrs.To analyze spatio-temporal patterns and foraging behavior, we delimited ten 1m 2 -plots on the perimeter of the lagoon, on an open space with grass (an area of ~4,000m 2 ).Plots were randomly allocated and the distances between them varied from 2-10 m.Frogs were observed foraging during 3 to 5 minutes per plot using a red light led flashlight.We recorded the number of individuals of E. bicolor observed per plot between 20:00 hrs and 00:30 hrs, at half an hour intervals.We estimated the distance between individuals (mm) when more than one frog was observed in one plot by carefully holding a carpentry meter a few centimeters (< 5 cm) above frogs.We analyzed the spatial distribution of frogs in the plots with standardized Morisita index (Ip) (Smith-Gill 1975).Observations were made by a single person (López J.A.) trying to keep the incidence of observer as low as possible.
We used a generalized linear mixed model (GLMM), binomial error distribution and logit link, to assess the relationship between the presence and absence of Elachistocleis bicolor in the plots with climate variables.
In the statistical analysis, plots were included as a random factor to account for repeated observations in each one (Zuur et al. 2009).Candidate explanatory variables were temperature of air in microhabitat (°C), obtained within the plots during each sampling time with a mercury glass thermometer placed five centimeters above ground level, and five climatic variables obtained from the weather station of the Centro de Informaciones Meteorológicas (CIM: FICH-UNL), located 200 m away from the study area: mean temperature (°C), relative humidity (HR%), atmospheric pressure recorded along the day (hPa), number of days since the last rain and accumulated rainfall (mm) in the past 48 hours.We used variance inflation factor (VIF) to assess multicollinearity among variables included in the models (Quinn and Keough 2002).VIF showed a high collinearity between temperature of microhabitat and mean temperature obtained in the weather station; thus, only the latter was maintained in the model.We used stepwise variable selection by backward elimination to build the model (Quinn and Keough 2002).We tested for individual effects by the likelihood ratio test for nested models; this statistic follows a X 2 distribution with one degree freedom (Zuur et al. 2009).GLMM analysis was performed through lme4 package (Bates et al. c2015) in R software (R Core Team c2015).

RESULTS
During exploratory searches, we observed frogs feeding after sunset in an area of low herbaceous vegetation alongside the fluvial forest that borders the south margin of a permanent lagoon of the RECU.This area was the selected one to place the 1m 2 -plots.A few specimens were found sheltered under fallen logs in the fluvial forest and no frog was found out of their refuge before sunset.
Between checking periods of the plots we observed frogs in the surroundings.Specimens of E. bicolor were found hunting for ants on their trails.Frogs were observed digging with their heads and introducing it under leaf litter right where the ant trails run as shallow tunnels under the mulch, remaining in this position for at least 5 min (Fig. 1).In eight out of the nine sampled days frogs were recorded in plots.One of the days we recorded only one frog in plots.In the remaining seven days we recorded from five to 43 frogs foraging in the ten plots, showing a density of 0.4±0.8frogs/m 2 .In those seven days, between 21:00 hrs and 22:30 hrs, the time of the day with the highest activity, frogs' density reached 0.57±0.93frogs/m 2 .The maximum density recorded was 1.1 frogs/m 2 , on 26th January at 21:30 hrs.The average distance between frogs within the plots was 47.53±25.28mm; while the minimum distance was 15mm.An aggregated distribution pattern was observed (Ip = 0.501).

DISCUSSION
Until now, knowledge of Elachistocleis bicolor phenology was limited to its reproduction, which occurs explosively after heavy rains during spring and summer and that has also been related to variables such as temperature and humidity (Rodrigues et al. 2003, Martori et al. 2005, López et al. 2011, Sanchez et al. 2009, 2013).We found that foraging activity of E. bicolor was regulated by climatic variables such as atmospheric pressure, days since last rain and recently accumulated rainfall.Low barometric pressure has been associated with increase in calling activity in some amphibians (Henzi et al. 1995, Oseen andWassersug 2002); however, in E. bicolor foraging activity increases with higher barometric pressure; thus, the rise of atmospheric pressure after rainfalls may be triggering foraging activity.Calling activity of some amphibians from temperate climates is positively correlated to one and two day's prior rainfall (Saenz et al. 2006), indicating a similar relation to the one found here for E. bicolor, which peaks occurred the second or third day after rains.Also, there is evidence that explosive breeders practically do not feed during reproduction (Solé and Pelz 2007).One possible explanation of the delay between precipitation and foraging peak is that frogs reproduce during the first days after rain, and then, they need to feed extensively to recover the energy spent (Feder and Burggren 1992).
We also considered humidity and air temperature as candidate explanatory variables, but both failed to explain foraging activity of E. bicolor.Humidity mediates reproductive activity in several amphibian species (Oseen andWassersug 2002, Richter-Boix et al. 2006), and air temperature has been associated to activity of other species of genus Elachistocleis and is classically related to amphibian activity in temperate environments (Oseen and Wassersug 2002, Saenz et al. 2006, Richter-Boix et al. 2006, Canavero et al. 2008).Therefore, reproduction and foraging activities are regulated by different climatic variables.Indeed, we observed the presence of moisture at ground level on the days associated with frogs' foraging activity, and this variable should be tested on future studies (Dodd 2009).
Most anurans have adopted a sit-andwait strategy for preys' location, leading to a generalist diet (Duellman and Trueb 1986).However, Elachistocleis bicolor is a specialist forager (Berazategui et al. 2007, López et al. 2007, Cossovich et al. 2011), displaying a combined active-passive foraging strategy (Donnelly 1991, Maneyro andda Rosa 2004).The combined foraging strategy of E. bicolor consists of an 'active' searching of ant trails, which sometimes includes digging right where ant trails ran as shallow tunnels under the mulch (active foraging strategy), and then, the 'passive' predation of ants as they pass through following a sit-and-wait foraging strategy.It has been mentioned that the nuchal fold of E. bicolor serves as protection against ant bites, since it can be extended forward to cover the eyes when frogs are hunting, while the skin secretes a viscous substance (Langone 1994) (Fig. 1).We did not observe this behavior.
Elachistocleis bicolor preference for small, hard prey like ants or mites (Solé et al. 2002, Berazategui et al. 2007, López et al. 2007, Cossovich et al. 2011), coincides with the feeding preference of other specialist anurans (Toft 1981, Lajmanovich 1995, Duré et al. 2009).Along its distribution area, E. bicolor is sympatric with diurnal and nocturnal mirmecophagous anurans (e.g.Bonansea and Vaira 2007, Duré et al. 2009, Da Rosa et al. 2002, Bortolini et al. 2013).An spatiotemporal segregation in foraging activity may reduce food competition with other ant predators.Day time activity could contribute to trophic segregation with sympatric diurnal toads of genus Melanophryniscus (Bonansea andVaira 2007, Bortolini et al. 2013).While microhabitat segregation could diminish trophic niche overlap with sympatric nocturnal toads of genus Rhinella (Duré and Kher 2004, Duré et al. 2009, Matsui 2016).In this study area, E. bicolor is syntopic with Rhinella fernanadezae (Gallardo, 1957), which also specializes in ant predation in open terrestrial habitats (Duré et al. 2009); but we did not observe E. bicolor overlapping foraging microhabitats with R. fernanadezae in the RECU.Indeed, seasonal activity of R. fernanadezae concentrates in spring, while E. bicolor activity concentrates in summer (López et al. 2011); thus, temporal, or seasonal segregation should be favoring coexistence of these species.Species of the genus Elachistocleis are usually associated with open environments (Lavilla et al. 2003, Rodrigues et al. 2003).And, although E. bicolor foraging habitat was an open area of low herbaceous vegetation, the few frogs that found shelter were found in the adjacent fluvial forest besides the RECU lagoon (E.bicolor refugees in fossae or under fallen logs: Maneyro and Carreira 2012).Therefore, studied frogs should have move after sunset from their day shelters in the fluvial forest (forest strip of 10 to 40 m wide) to contiguous foraging sites.In the foraging area, where observation plots were delimited, E. bicolor displayed an aggregate distribution.This pattern probably reflected the patchy distribution of ants (Theunis et al. 2005, Andersen 2008), their principal food resource (Berazategui et al. 2007, López et al. 2007).
Foraging activity of Elachistocleis bicolor is nocturnal, with a peak activity around 22:00 hours.The highest foraging activity of the frogs coincided with the times of the day where ant workers of the genus Solenopsis explore the environment searching for food (Claborn andPhillips 1986, Norasmah et al. 2006), suggesting synchronization with the activity of their prey.Studies on diet of E. bicolor showed an average of about thirty ants per digestive tract (Berazategui et al. 2007, López et al. 2007).With the high densities that E. bicolor can reach while foraging (more than one frog/m 2 ), this species should be considered an important ant predator.Taking into account that is a frequent species in agroecosystems (Sánchez et al. 2013), its role as biological control organism should be investigated.
Finally, when analyzing the role of abiotic environmental cues in amphibian activity, we should consider not only interspecific differences, but also differences among activity types within each species (e.g.reproduction or feeding).Since the same variable could relate in opposite directions to calling or foraging activity such is the case of atmospheric pressure.Also, variables classically related to reproductive activity not necessary explain foraging activity (e.g.temperature or humidity).Additionally, biological variables such as prey availability certainly are likewise influencing E. bicolor foraging activity (Falico et al. 2012, López et al. 2015).Moreover, environmental conditions could directly influence prey activity and then be indirectly correlated to frogs foraging activity (Donnelly 1991).Fully understanding of foraging activity implies the knowledge of a diversity of factors such as spatio-temporal patterns of activity, predator diet, prey-predator relationships and trophic niche segregation among predators, feeding strategy and related behaviors.And many of these issues are still unknown for most of amphibian species.

Figure 1 .
Figure 1.Frogs during foraging activity: a. frog digging with its head, arrow indicates nuchal fold, b. frog submerging its head under leaf litter right where ant trails run under the mulch.

Figure 2 .
Figure 2. Foraging circadian cycle of Elachistocleis bicolor.Daily changes every half hour in frogs' frequency of occurrence in plots.

Figure 3 .
Figure 3. Frequency of occurrence of Elachistocleis bicolor frogs registered in relation to atmospheric pressure (hPa), days since the last rain and with accumulated rainfall in the past 48 hours (mm).