Study area

This study was conducted in the Estación de Biología Chamela of the Instituto de Biología, Universidad Nacional Autónoma de México (19°30’ - 19°32’ N & 105°03’ - 105° 05’ W) on the Jalisco coast, Mexico. The Chamela station is part of the Reserva de la Biosfera Chamela-Cuixmala, which protects 13 142 ha of tropical dry forest and associated ecosystems (^{Ceballos, Szekely, García, Rodríguez, & Noguera, 1999}). The climate is tropical sub-humid with a prolonged dry season of about eight months and a welldefined rainy season with 85 % of the annual precipitation occurring from July to October (^{Bullock, 1986}). Average annual precipitation and temperature is 748 mm and 25 °C, respectively. The landscape is comprised of small hills dominated by deciduous forest vegetation, with semi-deciduous forest vegetation restricted to small valleys (^{Lott, Bullock, & Solis-Magallanes, 1987}). Other vegetation types in the reserve include xerophilus scrub, wetlands, grassland and mangrove (Ceballos et al., 1999). This region comprehends considerable vertebrate species diversity, including 19 anuran species, 11 of which are endemic to Mexico, and 15 have indirect development (^{García & Ceballos, 1994}; ^{García & Cabrera-Reyes, 2008}).

Data collection

This study was carried out during two rainy seasons, from July 10 to October 4, 2013, and June 26 to September 26, 2014. We established five transects along the main streambeds (known locally as ‘arroyos’) and watersheds at the Chamela Station: 1 km long transects in each of three streams Colorado, Hornitos and Zarco, and two transects of 500 m long along the watershed drainages known as Cuenca 2 and Cuenca 3 (BC watersheds). Each line-transect was divided into 50 m sections, giving 80 survey sections in total, with 20 sections in the streams, and 10 in each of the two watersheds. After the first heavy rain (> 30 mm), we surveyed all transects to locate potential sites or pools for amphibian reproduction, and to record any signs of breeding activity (calling males, adults in amplexus, presence of eggs and/or larvae). Each pool located was georeferenced, and individually marked with flagging. We recorded the site, survey section, and chronological order in which the pools formed and were found.

Once we located all breeding sites along the transect survey sections, we defined a time schedule to ensure that all five line-transects were surveyed twice a week, for a period of one to two hours each. All surveys were conducted at night, from 21:00 - 02:00 h. During each survey, three samplers walked along the transect at a slow pace looking for adult anurans on the streambed and within a 2 m band along each bank, as well as those present in pools. Anurans were detected using simultaneous visual encounter surveys and auditory sampling (^{Crump & Scott, 1994}; ^{Zimmerman, 1994}). For each individual anuran found, we recorded the species, location, activity (calling, amplexus, resting), and microhabitat according to type of substrate (sand, leaf-litter, rock and water) or plant life-form (tree, grass, bromeliads, agaves, herbs). This survey effort consisted of a total of 25 weeks of continuous field work during two rainy seasons, comprising 120 nocturnal surveys in 2013, and 130 nocturnal surveys in 2014. This resulted in 1 800 and 1 950 personhours of visual encounter surveys in 2013 and 2014, respectively (Supplementary Material).

Using data collected from all survey sections along each transect, we calculated species richness and the number of individuals of each species per transect. We used nomenclature in accordance with ^{Frost (2016}) Amphibian Species of the World. Information on habitat use was collected to determine potential differences in the occurrence of anuran species across the different microhabitats. We also evaluated whether species conducted specific activities (calling, amplexus and resting) in different microhabitats at the pools, and whether there is habitat partitioning among species.

Climatic variables

We obtained information on precipitation, temperature and relative humidity from the Chamela meteorological station, to explore the relationship between these variables and the reproductive activity and occupancy of anuran species. We calculated total weekly and accumulated precipitation, as well as weekly averages of maximum and minimum temperature, and relative humidity for each sampling season. We then analyzed the relationship between weekly climate variables and weekly occurrence of each anuran species.

We determined species richness and the number of individuals of each species per stream and watershed. Species diversity was estimated using the three indices of Shannon-Wiener, Simpson, and diversity of order one. The Jaccard species similarity index was used to compare composition of anuran species between sites. To explore the relationship between abiotic variables and anuran reproduction activity we conducted a Spearman correlation analysis of number of individuals of each species per week with weekly accumulated precipitation. To describe the reproductive phenology of the anuran assemblage we used weekly records of adult activity related to weekly and cumulative precipitation for each sampling year.

Reproductive sites

We located 161 pools formed during the 2013 rainy season, and 147 pools in 2014. The number of pools formed per survey section was higher in Arroyo Zarco in both years (65 and 59, respectively), than in Arroyo Colorado (16 and 30), Arroyo Hornitos (26 and 24), and the two watershed drainages combined (Cuencas BC: 54 and 44, respectively). During 2013, about 45 % of sites (72 pools) were used by anurans for reproduction at some point (34 in Zarco, 9 in Colorado, 13 in Hornitos and 16 in Cuencas BC), whereas in 2014 only about 18% (27 pools) were used (12 in Zarco, 4 in Colorado, 4 in Hornitos, and 7 in Cuencas BC).

Species richness, number of individuals of each species, composition and diversity

During nocturnal surveys, we obtained a total of 7 313 observations of individuals belonging to 15 anuran species, representing 12 genera and five families (Table 1). The most diverse family was Hylidae with 8 species, which also included the most abundant species of Tlalocohyla smithii and Exerodonta smaragdina (Table 1). The species of Leptodactylus melanonotus (Leptodactylidae), Incilius mazatlanensis and Rhinella horribilis (Bufonidae) had the least number of individuals (Table 1). The occurrence and the number of individuals of each species was not uniform in all at the sampling sites (Table 1). Ten of fifteen anuran species were recorded in all sites, whereas five species were found in only three sites (Table 1). The number of individuals of each species was on average three to five times higher in Arroyo Zarco than in other sites (Table 1). Species richness was highest in Arroyo Zarco, although diversity indices were similar among sites, but were slightly higher in Arroyo Hornitos (Table 1). The Jaccard similarity indices also indicated a high similarity of anuran species among sites, particularly between Zarco and Colorado (Table 2). When we compared the relative number of individuals reported for each species in a particular site with respect to the total number recorded for that species in all sites, it was clear that the majority of individuals of most species were observed in Arroyo Zarco (average 53.8 % of individuals per species; Fig. 1). The exceptions were R. horribilis and Trachycephalus typhonius found mainly in the Cuencas BC and Arroyo Colorado, respectively.

Fig. 1 Percentile distribution of individuals of each species by sampling sites at Estación de Biología Chamela, Jalisco, Mexico. 

Correlations between abiotic factors and anuran occupancy of reproductive sites

Multiple Spearman correlation tests found that reproductive activity of species such as Agalychnis dacnicolor, Hypopachus ustus and Smilisca baudinii, were positively correlated with maximum temperature (Spearman r = 0.553, 0.464, 0.41 respectively), but negatively correlated with an increment in accumulated precipitation (Spearman r = -0.841, -0.589, -0.479, respectively). This group of species was found active from June to August, with a subsequent decline in activity when streambeds had running water. Even though the number of active S. baudinii individuals increased after heavy rains (> 25 mm), there was no significant correlation between number of individuals of this species and weekly precipitation. In contrast, abundance of E. smaragdina and T. smithii were positively correlated with accumulated precipitation (Spearman= 0.672, and 0.687, respectively) and relative humidity (Spearman= 0.585, and 0.623, respectively), and negatively correlated with the maximum temperature (Spearman= -0.776, and -0.774, respectively). For these two species, the number of individuals increased during September to October, when arroyo streams filled with water after accumulated rainfall exceeded 300 mm. These species were considered prolonged breeding species that reproduce several times throughout the rainy season (Fig. 2). Activity of six species, Diaglena spatulata, H. variolosus, I. marmoreus, I. mazatlanensis, S. fodiens and T. typhonius were positively correlated with weekly precipitation (Spearman=0.489, 0.586, 0.540, 0.576, 0.460, 0.418, respectively). These species were active exclusively during or after heavy rains (generally >30 mm), and were considered explosive breeders.

Fig. 2 Activity and use of microhabitat of more abundant species of anurans at Estación de Biología de Chamela, Jalisco, Mexico. 

Based on these results, we grouped species according to two main reproductive strategies of prolonged or explosive breeders. Species with prolonged breeding could be further subdivided into two groups based on the correlations of reproductive activity with climate variables: 1) species that are positively correlated with accumulated precipitation and humidity, and negatively correlated with temperature; and 2) species that are negatively correlated with the accumulated precipitation and positively correlated with temperature. We could not determine the temporal reproductive strategy of Lithobates forreri, Dendrosophus sartori, and R. horribilis as their activity was not correlated with any of the climate variables studied.

Temporal segregation

Weekly records of males of each species calling at all sites showed that there was a tendency towards species replacement at breeding sites throughout the rainy season. The species that were reproductively active during the first weeks of the rainy season included one of the most abundant species recorded, H. ustus, and large-bodied hylids such as A. dacnicolor and S. baudinii. The small-bodied hylids with greatest overall abundance, such as E. smaragdina and T. smithii, were more active during the late rainy season, when cumulative rainfall exceeded 300 mm, and the arroyos formed into strong flowing streams (Table 3). However, there was generally an overlap in the occurrence of most anuran species that was associated with the occurrence of strong weekly rainfall (Table 3).

Microhabitat use

A. dacnicolor showed a strong association with logs, branches, and leaves of trees near the pools, where it was found calling, resting, or in amplexus, although the species also occurred in smaller proportion in bushes, on rocks, and other substrates (Fig. 2). Clutches were observed on leaves, trunks, shrubs and rocks next to the pools. D. spatulata was more frequently observed in the water of pools or on nearby rocks, but was also found calling and resting on tree trunks. Amplexing pairs were also observed in pools and on rocks (Fig. 2). Calling males and amplexing pairs of E. smaragdina were found generally on rocks emerging above the water flow or directly in water in shallow areas of streams (< 2 cm), but resting individuals were observed more frequently on small rocks, tree branches, leaves and shrubs near streams (Fig. 2). Individuals of S. baudinii were found mainly in the pools and nearby rocks when calling, resting or in amplexus, although sometimes individuals occurred on other substrates (Fig. 2). Clutches of this species were always found at the water surface of pools. The species T. smithii was associated almost exclusively with tree leaves, shrubs, and grass near streams, when singing, resting, and during amplexus (Fig. 2). No clutches were observed for the three species of D. spatulata, E. smaragdina, and T. smithii.

When calling, resting or in amplexus, H. ustus was always found at ground level, either on litter (most individuals), on the sand in the dry stream bed, or in cavities under rocks (Fig. 2). Clutches were found on the surface of the pools. By comparison, individuals of H. variolosus were found calling and amplexing almost exclusively in the pools, while resting individuals were observed in water, leaf-litter, and under rocks, or at the base of grassy vegetation in the dry sandy streambeds (Fig. 2). Clutches of this species were observed on the water surface of pools. Resting I. marmoreus were observed frequently on leaves, rocks, and in the water of pools, while those calling or amplexing were observed mainly in the water of pools and rocks close to the pools (Fig. 2).

Clutches of this species were observed on the surface of pools.