Trophic Plasticity, Environmental Gradients and Food-Web Structure of Tropical Pond Communities

Home , Freshwater biology, Phyllomedusa sauvagii

Freshwater Biology (2017) 62, 519–529 doi:10.1111/fwb.12882

Trophic plasticity, environmental gradients and food-web structure of tropical pond communities

~ , † CHRISTOPHER M. SCHALK* ,CARMENG.MONTANA* ,KIRKO.WINEMILLER*AND LEE A. FITZGERALD* *Department of Wildlife and Fisheries Sciences, and Biodiversity Research and Teaching Collections, Texas A&M University, College Station, TX, U.S.A. † Department of Biological Sciences, Sam Houston State University, Huntsville, TX, U.S.A.

SUMMARY

1. Freshwater pond communities exhibit strong patterns in species composition in response to environmental gradients such as ecosystem size, disturbance and productivity, serving as excellent systems for studies of food-web structure. 2. We surveyed 13 ponds that varied along environmental gradients of canopy cover, pond size and hydroperiod at the beginning and end of the rainy season in the semi-arid thorn forests of the Gran Chaco ecoregion of Bolivia. We collected basal resources and consumers (tadpoles, macroinvertebrates and ﬁshes) from these ponds and used stable isotope ratios of carbon (13C/12C) and nitrogen (15N/14N), to quantify the spatiotemporal dynamics of the food webs in these tropical ponds. 3. There were no relationships between vertical structure of the food webs and the environmental gradients associated with the ponds. Consumers within these ponds exhibited high trophic variability, with multiple taxa occupying more than one trophic position across space or time. 4. Consumers from larger ponds had a greater degree of trophic redundancy. More shaded ponds supported food webs that had lower trophic diversity. More permanent ponds had signiﬁcantly greater trophic diversity as well as a greater range of basal resource diversity. 5. The breeding ponds utilised by tadpoles and macroinvertebrates are patchily distributed across space and time. In these dynamic habitats, a feeding strategy of trophic generalism and plasticity enables consumers to exploit a broad range of resources and promote species coexistence. These results suggest that high diversity in tropical ponds does not necessarily translate into specialisation of trophic function.

Keywords: canopy cover, hydroperiod, Neotropics, pond size, stable isotopes

Freshwater pond communities exhibit strong patterns Introduction in species composition in response to environmental gra- A fundamental goal in ecology is to determine the mecha- dients such as area, hydroperiod and productivity, serv- nisms responsible for community organisation and species ing as excellent systems for studies of food-web coexistence (Chesson, 2000; Siepielski & McPeek, 2010). structure (Wellborn, Skelly & Werner, 1996; De Meester Studies of community structure can generate hypotheses et al., 2005; Williams, 2006; Werner et al., 2007). Most about how and why attributes of communities allow their studies of species inhabiting lentic waterbodies have assembly and persistence within certain environments. attributed patterns of community structure as a response Food webs are depictions of consumer–resource interac- to pond permanence, and trade-offs that are associated tions and can provide insights as to trophic relationships with the constraints of pond drying and predation and resource use among coexisting species. (Wellborn et al., 1996). Werner et al. (2007) documented

Correspondence: Christopher M. Schalk, Department of Wildlife and Fisheries Sciences, and Biodiversity Research and Teaching Collections, Texas A&M University, 210 Nagle Hall, MS 2258 College Station, TX 77843, U.S.A. E-mail: [email protected]

© 2016 John Wiley & Sons Ltd 519 520 C. M. Schalk et al. a decline in amphibian richness in ponds with longer food-web structure along gradients of habitat distur- hydroperiods, which was clearly attributed to the pres- bance (pond drying), pond area and productivity ence of fish in more permanent ponds. In addition to (canopy cover). We predicted that larger and more pond hydroperiod, canopy cover has also been shown to stable ponds (i.e. more permanent) would support food affect structure of pond communities (Werner & Glen- webs with lower trophic redundancy and greater trophic nemeier, 1999; Pazin et al., 2006; Binckley & Resetarits, diversity compared with more ephemeral ponds. We 2007; Werner et al., 2007). Patterns of amphibian distri- predicted that less productive (more shaded) ponds bution among ponds without fish were strongly linked would have lower trophic redundancy and diversity with abiotic variables, specifically canopy cover and because these have a lower availability and diversity of pond area (Werner et al., 2007). Some species appeared food resources for aquatic consumers. We further pre- to be intolerant to increased canopy cover, and changes dicted that most tadpoles in these tropical ponds would in distribution of species among ponds were associated be trophic specialists resulting in low trophic redun- with increasing forest encroachment around breeding dancy within local assemblages when compared to tad- ponds (Skelly, Werner & Cortwright, 1999; Werner & poles from temperate ponds. Glennemeier, 1999; Binckley & Resetarits, 2007, 2009). Canopy cover affects the resource type and quality by reducing productivity and this in turn affects consumer Methods developmental rates (Werner & Glennemeier, 1999; Study site Grether et al., 2001; Skelly, Freidenburg & Kiesecker, 2002; Halverson et al., 2003). Our study was conducted in the semi-arid thorn forests Theoretical studies have converged on the pattern that of the Gran Chaco ecoregion of south-eastern Bolivia. less predictable environments (e.g. those with less pre- The region has a warm, rainy season (November–March) dictable disturbance regimes or primary production) and a cool, dry season (April–October). Vegetation in tend to support food webs with shorter food chains this region is predominately thorn forest; common tree (Pimm & Lawton, 1978; Pimm, 1982). However, most species include Schinopsis lorentzii (Anacardiaceae) and empirical studies have been conducted in temperate Aspidosperma quebracho-blanco (Apocynaceae), with cacti regions (Post, 2002; McHugh, McIntosh & Jellyman, [e.g. Opuntia spp. (Cactaceae), Cleistocactus baumannii 2010; Schriever & Williams, 2013; Schriever, 2015). There (Cactaceae) and Eriocereus guelichii (Cactaceae)] and is much to learn from examination of the structure of bromeliads common in the understory (Navarro & Mal- tropical pond communities and how local assemblage donado, 2002). The study area is located in one of the structure of aquatic and semi-aquatic species may or most xeric regions of the Bolivian Chaco; annual rainfall may not follow the patterns and trade-offs that are and temperature average 513 mm and 24.6 °C respec- apparent in temperate systems. Tropical taxa tend to tively (Navarro & Maldonado, 2002). reveal greater specialisation and diversification in functional traits (Schemske et al., 2009). Consumers in Surveys and sample preparation temporary tropical ponds, such as tadpoles, have high eco-morphological diversity compared to their temperate We surveyed 13 ponds at two sites, Kuaridenda (19.17°S, counterparts (Altig & Johnston, 1989; Altig & McDiar- 62.53°W; n = 5 ponds) and Yapiroa (19.61°S, 62.57°W; mid, 1999). However, temporary ponds, by their very n = 8 ponds) in the indigenous territory of Isoso, nature, eventually dry, and species occupying these Cordillera Province, Santa Cruz Department, Bolivia. habitats experience changing environmental conditions, Both natural and constructed ponds were surveyed such as reduction in habitat size and/or water quality. along environmental gradients of pond permanency Harsh conditions such as these may act as a filter that (mean SD = 85.8 7.9 days; range = 76–93 days), restricts local assemblage composition to species that pond size (mean SD = 724.1 780.8 m2; range = 74.4– possess a similar suite of traits (Chase, 2007). In turn, 2441 m2) and canopy cover (mean SD = 26.9 23.6%; these coexisting species may be functionally redundant, range = 1.5–68.7%) (Appendix S1). All ponds sampled whereby distinct taxa fulfil similar ecological roles dry annually and lacked aquatic vegetation. Among frog (Rosenfeld, 2002). Thus, different patterns may emerge species that breed exclusively in ponds, reproduction is when comparing food-web structure in the tropics. most intense during the rainy season (Schalk & Saenz, Here, we analyse stable isotope ratios of major food- 2016). All frogs in the study area are pond breeders and web components in tropical ponds to reveal variation in have exotrophic tadpoles (Perotti, 1997).

© 2016 John Wiley & Sons Ltd, Freshwater Biology, 62, 519–529 Tropical pond food webs 521 We collected tadpoles, macroinvertebrates and fishes one facing each cardinal direction. Canopy cover was at the beginning (December 2010/January 2011) and at estimated as the average of four readings taken at each the end (April 2011) of the rainy season using an area- measurement site, then averaging those to get an overall based sampler and dip net. At the beginning of the mean canopy cover for a pond (Werner et al., 2007). rainy season, all ponds were sampled within 2 weeks Pond area was determined using a measuring tape to after filling with water. Only two ponds were in close measure the pond’s longest and widest axis and using proximity to an ephemeral stream (ponds 7 and 8 were the formula of an ellipse to calculate surface area. 150 and 100 m from the stream respectively), which can Hydroperiod (the number of days the ponds held water connect to these ponds during heavy rain events. We within the study period) was determined by checking used a 120-L plastic cylinder (trashcan) as a drop sam- the ponds for water approximately every 2 weeks. When pler (height = 0.43 m, sampling area = 1.3 m2). We a pond was found to be dry, we assumed the pond dropped the sampler every 3 m along the pond’s longest dried in the midway point between surveys. axis and cleared the sampler using a dip net (mesh size = 2 mm). To ensure that all organisms had been Data analyses removed from the sampler, we conducted at least 10 sweeps with the dip net and 10 additional sweeps after Trophic position from isotopic data (TP SIA) was esti- the last animal was collected (Werner et al., 2007). Dur- mated based on fractionation of 15N between the con- ing each survey period, we retained samples of tadpoles, sumer and basal production sources collected from its fishes and aquatic invertebrates per pond for isotopic locality (Vander Zanden & Rasmussen, 1999; Post, 2002) analyses. using the formula: Tissue samples for stable isotope analysis were taken Trophic position ¼½ðd15N d15N Þ=2:54 from three specimens of each species during each survey consumer reference þ 1 period. In addition, we collected potential basal sources d15 d15 (organic sediment, seston and C3 plants) from each pond where Nreference was the mean N of basal sources by hand. We collected seston by filtering 100 mL of water (C3 plants, seston, sediment) and 2.54& is the mean through pre-combusted Whatman GF/F filter (GE trophic fractionation value (Vanderklift & Ponsard, 2003). Healthcare, Little Chalfont). Samples for stable isotope To compare food-web structure between ponds and analysis were preserved in salt as described by Arrington periods, we used community-wide metrics based on iso- & Winemiller (2002). In the laboratory, tissues were topic data (Layman et al., 2007). We examined the degree soaked in distilled water for 4 h, and then rinsed again to of trophic redundancy [mean nearest neighbour distance remove salt. Samples were then dried at 60 °C for 48 h in (NND)], evenness of trophic niches (standard deviation of a drying oven. Dried samples were ground to a fine pow- NND), average degree of isotopic diversity [centroid dis- der with a mortar and pestle, and then stored in clean tance (CD)], magnitude of food-web trophic diversity glass vials. Subsamples of each ground samples were [convex hull of the total area (TA)], maximum vertical weighed (1.5–3 mg) and packaged into a Ultra-Pure tin structure [d15N range (NR)] and basal resource diversity capsules (Costech Analytical, Valencia) and sent to the [d13C range (CR)] (Layman et al., 2007). These metrics Analytical Chemistry Laboratory, Institute of Ecology, may be biased when isotopic signatures of basal sources University of Georgia, for analysis of stable isotope ratios are not considered, because d13C values might be influ- of carbon (13C/12C) and nitrogen (15N/14N). During mass enced by physicochemical and other environmental char- spectrometry, two different standards were processed acteristics across systems (Hoeinghaus & Zeug, 2008). In between every 12 samples, and precision was 0.22& for our study, however, these metrics are appropriate given d13C and 0.20& for d15N. Isotope ratios were reported that between-pond variation in ratios of basal sources in parts per thousand (&) standardised in relation to ref- was minimal within a given survey period (Fig. 1 and erence material (Pee Dee Belemnite for C, atmospheric Appendix S2). The SIAR (Stable Isotope Analysis in R) d nitrogen for N) and reported as X=[(Rsample/Rstan- package in the R software version 3.0.2 (R Core Team, 3 13 12 15 14 dard) 1)]910 , where R= C/ Cor N/ N. 2013) was used to calculate food-web metrics. We used We quantified environmental gradients at each pond generalised linear models (GLMs; using a normal distri- monthly from January to April 2011. Canopy cover was bution and link function) to explore relationships between estimated using a spherical densitometer with a mea- food-web structure and environmental gradients. Because surement taken at every 5 m along the pond’s longest there was no relationship between pond area, hydrope- axis. At each sampling site, four readings were taken, riod or canopy cover (Appendix S2), we analysed the

18 Early season 18 Late season

16 16 Seston 14 14

12 12 N 15 10 Seston 10 8 8 C plant C3 plant 3 6 Sediment 6 Sediment

4 4 –35 –30 –25 –20 –15 –35 –30 –25 –20 –15 d13 d15 13 Fig. 1 Bi-plots of C and N for tad- 13C C poles (triangles), aquatic invertebrates Aquatic Invertebrates Tadpoles (diamonds) and three basal sources Belostomatidae Anostraca Ceratophrys cranwelli (Ceratophryidae) (circles) for one of the ponds (Pond 6) in the Gran Chaco ecoregion of Bolivia. Notostraca Hydrophilidae Phyllomedusa sauvagii (Hylidae) Isotope ratios were compared for taxa Corixidae Dysticidae Leptodactylus bufonius (Leptodactylidae) Dermatonotus muelleri (Microhylidae) that occurred near the early season and Notonectidae Aeshnidae late season of the annual rainy season. influence of these gradients on food-web structure sepa- estimates of trophic position for several taxa exceeded rately. Data were log transformed prior to analyses that an entire level (Fig. 2). Consumers tended to occupy were conducted using the statistical software PAST higher trophic positions near the end of the rainy season (Hammer, Ryan & Harper, 2001). (Fig. 3). Near the beginning of the rainy season, mean trophic position of consumer taxa ranged from 1.3 [Der- matonotus muelleri (Microhylidae) tadpoles] to 3.2 [Neo- Results fundulus ornatipinnis (Rivulidae) killifish]. Near the end Samples from 318 consumers (early season = 152, late of the rainy season, mean trophic position of consumer season = 165) and 98 basal sources (early season = 54, taxa ranged from 1.1 (Hydrophilidae beetles) to 5.7 [Sci- late season = 44) were analysed for stable isotope ratios. nax nasicus (Hylidae) tadpoles] (Fig. 2). During the early Consumer samples included 11 tadpole species, 5 fish rainy season, tadpoles tended to have lower trophic species and 16 macroinvertebrate taxa. Consumers d13C positions compared to tadpoles sampled near the end of ranged from 34.9 to 16.9& at the beginning of the the rainy season (Fig. 3a, b) and mean trophic position rainy season, and from 32.7 to 19.7& at the end of of tadpoles was similar to mean values of other con- rainy season. Consumer d15N ranged from 4.3 to 15.9& at sumer groups during this period [tadpole mean the beginning of the rainy season, and from 5.8 to 19.2& (SD) = 2.1 (0.4), macroinvertebrate = 2.1 (0.3), at the end of rainy season. Producer d13C ranged fish = 2.3 (0.6); Fig. 3a, b]. A few macroinvertebrate from 31.3 to 19.4& at the beginning of the rainy sea- taxa had high trophic positions near the end of the dry son, and from 31.3 to 20.5& at the end of the rainy season (Fig. 3a). Near the end of the rainy season, season. Producer d15N ranged from 4.5 to 13.4& at the trophic positions of tadpoles tended to be higher with beginning of the rainy season, and from 4.2 to 13.7& at distributions more left skewed compared to other con- the end of rainy season. Stable isotope bi-plots indicated sumer groups (tadpole mean SD = 3.2 1.2, macroin- consumers became more enriched in d15N near the end of vertebrate mean SD = 2.5 1.0, fish mean SD = the rainy season (Fig. 1). Most consumer d15N values 2.3 0.6; Fig. 3b). were consistently higher than those of basal production sources, but there were some exceptions (e.g. in ponds 5 Food-web structure and 7 in the early rainy season; Appendix S2). Seston d15N was highly variable and mean values sometimes Food-web structure varied along environmental gradi- were greater than those of consumer taxa (Appendix S2). ents across seasons. Near the beginning of the rainy season, consumers were significantly more packed within the isotopic space occupied by consumers with increas- Trophic positions ing pond size [GLM, R2 = 0.17, P = 0.02; mean Macroinvertebrates and tadpoles revealed large isotopic NND = 0.40285 0.16142log 10 (area)] (Fig. 4). There variation across space and time, and the range for was a statistically non-significant trend showing that

6 (a) Early season Late season (a) 50 Early season Macroinvertebrates Tadpoles 5 Fish 40 4

3 30 2 Trophic position 1 20

0 10

0 6 (b) 1.0 – 1.4 1.5 – 1.9 2.0 – 2.4 2.5 – 2.9 3.0 – 3.4 3.5 – 3.9 5 4.0 – 4.4 4.5 – 5.0> 4 (b) 50 Late season 3

2 40 Trophic position 1 30 0

20 % consumer taxa % consumer taxa 6 (c) 10

5 0 4

3 2.0 – 2.4 2.5 – 2.9 1.0 – 1.4 1.5 – 1.9 3.0 – 3.4 3.5 – 3.9 4.0 – 4.4 4.5 – 5.0> 2 Trophic position Fig. 3 Frequency histograms of trophic positions of macroinverte- 1 brates (black bars), tadpoles (white bars) and ﬁsh (grey bars) collected across 13 ponds at the (a) early season and (b) late season of 0 Bryconamericus sp. Characidium sp. Leporinus sp. N. ornatopninis S. marmoratus the annual rainy season.

Fig. 2 Mean trophic position (1 standard deviation) of (a) macroinvertebrates, (b) tadpoles and (c) fish collected in 13 ponds (canopy cover)] (Fig. 4). Near the end of the rainy sea- at the early season (white bars) and late season (grey bars) of the son, ponds with longer hydroperiods had significantly annual rainy season. R. schneideri = Rhinella schneideri, R. ma- greater trophic isotopic diversity [GLM, R2 = 0.29, = = jor Rhinella major, C. cranwelli Ceratophrys cranwelli, P = 0.047; CD = 1.532 + 1.088log 10 (hydroperiod)] as P. sauvagii = Phyllomedusa sauvagii, S. nasicus = Scinax nasicus, T. venulosus = Trachycephalus venulosus, L. bufonius = Leptodactylus well as a greater basal resource diversity as indicated by d13 2 = = bufonius, P. albonotatus = Physalaemus albonotatus, P. biligo- a greater range of C [GLM, R 0.0073, P 0.01; nigerus = Physalaemus biligonigerus, D. muelleri = Dermatonotus muel- range d13C = 1.741 2.3367log 10 (hydroperiod)] = leri, N. ornatipinnis Neofundulus ornatipinnis, (Fig. 4). There was no relationship between CR for basal S. marmoratus = Synbranchus marmoratus. sources or taxa richness and the three environmental gradients (Appendices S4 and S5 respectively). The dis- consumers were more packed within isotopic space with tribution of taxa within assemblage isotopic space increasing species richness (Appendix S3). Isotopic became less even with increasing canopy cover near the diversity was significantly lower with increasing canopy end of rainy season [GLM, R2 = 0.19, P = 0.036; cover near the beginning of the rainy season [GLM, CD = 0.13948 0.18091log 10 (canopy cover)] (Fig. 4). R2 = 0.47, P < 0.001; CD = 0.68649 0.16587log 10 Neither the range of isotopic diversity (range N) nor the

© 2016 John Wiley & Sons Ltd, Freshwater Biology, 62, 519–529 524 C. M. Schalk et al. extent of isotopic diversity (TA) revealed significant pat- Dayton & Fitzgerald, 2001). Whether this trade-off drives terns in relation to any of the three environmental gradi- variation in trophic positions of macroinvertebrates and ents (Fig. 4). tadpoles in tropical ponds needs to be tested experimen- tally (sensu Caut et al., 2013; Arribas et al., 2015). Another challenge is to understand the functional roles of diverse Discussion tadpoles in tropical assemblages (Altig et al., 2007). For Trophic positions of macroinvertebrates and tadpoles example, it is not known whether Chacoan tadpoles with varied considerably between ponds and within ponds high trophic positions function as predators (Schalk et al., between periods. Many consumer taxa would be classi- 2014b) or scavengers, or both depending on context. fied as omnivores, and in some cases, spatiotemporal Certain aspects of food-web structure were non-ran- variation in species trophic level ranged from herbivo- dom in relation to environmental gradients associated rous to carnivorous. Tadpoles were traditionally consid- with pond habitats. Overall, the range of d13C was posi- ered to be herbivores or detritivores (Altig, Whiles & tively correlated with pond hydroperiod. In temperate Taylor, 2007); however, our results indicate many tad- regions, ponds with long hydroperiods have seasonally poles fed at higher trophic levels, which corroborates variable seston production and inputs from leaf fall results from other studies of temperate tadpoles (Pet- (Schriever, 2015). We did not find a relationship between ranka & Kennedy, 1999; Schiesari, Werner & Kling, the range of d13C of the basal sources and hydroperiod 2009). With the exception N. ornatipinnis, an annual killi- of Chacoan ponds, however, it cannot be ruled out that fish with a life cycle adapted for ephemeral aquatic habi- samples of one or more important sources were not tats (Schalk, Montana~ & Libson, 2014a), the fish species included in our analysis. collected from these Chacoan ponds would have colo- Food chain length was shown to be positively corre- nised them during intervals when they had surface lated with hydroperiod in studies of temperate-zone water connections with the nearby stream. On average, ponds (Schriever & Williams, 2013; Schriever, 2015). In N. ornatipinnis was a tertiary consumer, but like tadpoles contrast, we found no relationship between vertical and macroinvertebrates, had large within-taxon variation structure (range of d15N) and any of the three environ- in trophic level during the early rainy season. While the mental gradients tested. There are important distinctions other four fish species could be classified as herbivores for the way hydrology affects ecological dynamics of or omnivores, our small sample sizes render this gener- ponds. Temporary ponds in temperate regions dry in a alisation speculative. Although tropical tadpoles tend to more predictable seasonal manner compared to ponds reveal greater ecological specialisation in terms of func- in many tropical and subtropical regions (Perotti, Jara & tional traits (Altig & Johnston, 1989; Schemske et al., Ubeda, 2011; Schriever & Williams, 2013; Schriever, 2009), our results suggest that greater eco-morphological 2015). In the Chaco, ponds are very dynamic and may specialisation does not necessarily translate into trophic dry and refill multiple times within a single rainy season specialisation for many tadpoles inhabiting Chacoan (Schalk, 2016a,b; Schalk & Saenz, 2016). Selection ponds. In these dynamic aquatic habitats, tadpoles dis- imposed by this unpredictable variation in pond drying play considerable trophic plasticity, consuming diverse is reflected in the diverse oviposition strategies of the resources opportunistically dependent upon availability. regional frog fauna. Several species lay their eggs in The flexible trophic strategy of tadpoles may explain the foam nests or terrestrial nest chambers (Crump, 1974; lack of a relationship between food-web vertical struc- Duellman & Trueb, 1994; Perotti, 1997; Schalk & Saenz, ture and environmental gradients in Chacoan ponds. 2016), which likely protects developing larvae from des- As for many animals, tadpoles have diets that are con- iccation associated with unpredictable pond drying text-dependent and often mediated by the density of con- (Crump, 1974, 2015; Duellman & Trueb, 1994). Temper- specifics, heterospecific competitors or non-consumptive ate ponds with intermediate hydroperiods that dry in a effects of predators (Caut et al., 2013; Greig, Wissinger & predictable fashion tend to harbour the greatest diversity McIntosh, 2013; Arribas et al., 2015). In a temperate of amphibians and macroinvertebrates (Werner et al., amphibian community, the presence of a crayfish com- 2007; Schriever & Williams, 2013; Semlitsch et al., 2015). petitor caused tadpoles to shift to a more detritus-based Colonisation of these habitats by species with broad diet and consume resources with lower d13C (Arribas trophic niches naturally leads to an increase in overall et al., 2015). Trade-offs between foraging activity and pre- trophic diversity by the end of the rainy season. dation risk are well established in tadpoles inhabiting Among-pond variation in food-web structure in the ponds in temperate regions (Werner & Anholt, 1993; Chaco apparently is influenced by pond size as well as

25 25 25 R2 = 0.31 20 20 20

15 15 15

Crange 10

13 10 10

5 5 5

0 0 0 0 500 1000 1500 2000 2500 3000 75 80 85 90 95 0 20406080 16 16 16 14 14 14 12 12 12 10 10 10

N range 8 8 8 15 6 6 6 4 4 4 2 2 2 0 0 0 0 500 1000 1500 2000 2500 3000 75 80 85 90 95 0 20406080 2.5 2.5 2.5 R2 = 0.17 2 2 2

D 1.5 1.5 1.5

1 1 1 Mean NN 0.5 0.5 0.5

0 0 0 0 500 1000 1500 2000 2500 3000 75 80 85 90 95 020406080 2 2 2 2 R = 0.20 1.8 1.8 1.6 1.6 1.5 1.4 1.4 1.2 1.2 1 1 1 0.8 0.8 SD-NND 0.6 0.6 0.5 0.4 0.4 0.2 0.2 0 0 0 0 500 1000 1500 2000 2500 3000 75 80 85 90 95 020406080 6 6 6 R2 = 0.47 R2 = 0.30 5 5 5 4 4 4 3 CD 3 3

2 2 2

1 1 1

0 0 0 0 500 1000 1500 2000 2500 3000 75 80 85 90 95 0 20406080 180 180 180 160 160 160 140 140 140 120 120 120 100 100 100

TA 80 80 80 60 60 60 40 40 40 20 20 20 0 0 0 0 500 1000 1500 2000 2500 3000 75 80 85 90 95 0 20406080 Pond area (m2) Hydroperiod (days) Canopy cover (%)

Fig. 4 Metrics of food-web structure in relation to the environmental gradients defined by pond area (first column), hydroperiod (second column) and canopy cover (third column) in the early season (open circles) and late season (filled circles) of the annual rainy season. Signifi- cant relationships between food-web structure and an environmental gradient are depicted by the presence of a solid line (early season) and dashed line (late season). NND, nearest neighbour distance; SD-NND, standard deviation of nearest neighbour distance; CD, distance to centroid; TA, total area.

© 2016 John Wiley & Sons Ltd, Freshwater Biology, 62, 519–529 526 C. M. Schalk et al. trophic guild structure of local assemblages. During the standing crops of primary producers in these ponds or early rainy season, there was evidence of higher trophic compare developmental/growth rates of taxa in shaded redundancy (i.e. consumers were more packed within versus unshaded ponds, our results nonetheless are con- isotopic space) with increasing pond size. Aquatic habi- sistent with previous studies that found an inverse rela- tat area is recognised as a major factor affecting popula- tionship between trophic diversity and canopy cover. tion abundance, assemblage composition and, Isotopic analysis revealed that trophic plasticity and consequently, food-web structure (McHugh et al., 2015; redundancy are high among tadpoles and macroinverte- Montana,~ Layman & Winemiller, 2015; Semlitsch et al., brate taxa inhabiting Chacoan ponds. Some tadpole spe- 2015). Larger habitats have higher rates of species cies apparently occupy multiple trophic positions, colonisation, further contributing to a positive species– depending on time and location. Similar patterns have area relationship (Rosenzweig, 1995). Greater species been observed for invertebrate communities in desert diversity often is associated with greater functional streams wherein coexisting species revealed high over- diversity that facilitates species coexistence. We did not lap among functional traits (Boersma et al., 2013). In observe a relationship between pond area and taxon regions with variable and often harsh environmental richness in Chacoan ponds. In these systems, stochastic conditions, such as the Gran Chaco, the regional species factors may play a strong role in community assembly pool may contain taxa with high functional similarity during the early rainy season (Chase, 2007; Chase & owing to environmental filtering at a relatively large Myers, 2011), especially for smaller ponds that are likely spatial scale (Chase, 2007; Schalk, Montana~ & Springer, to be colonised less frequently or predictably. Also, 2015; Schalk, 2016b). Because tadpoles and other aquatic many basal resources are probably rare during initial consumers can affect food resource availability and qual- stages of pond filling during the early wet season. Based ity, they can have a strong influence on aquatic food- on samples taken throughout the rainy season, there web dynamics (Wallace & Webster, 1996; Whiles et al., was a non-significant trend of increasing trophic redun- 2006, 2010). Moreover, pond-dwelling frogs and aquatic dancy in relation to taxa richness. If many consumers insects create trophic links between aquatic and terres- exploited similar resources due to low resource diver- trial habitats (Whiles et al., 2006; Earl et al., 2011). Func- sity, this also could explain high trophic redundancy. tional redundancy may provide a buffer that protects We also observed that trophic niches were more even ecosystem function against species loss, however, this with increasing pond size at the end of the rainy season. inference depends on the traits selected for analysis Since ponds were drying during this period, food (Petchey & Gaston, 2006; Boersma et al., 2013). Although resources may have been increasingly limited, thus pro- we conclude that tadpoles and aquatic macroinverte- moting resource partitioning among consumers. brates of Chacoan ponds have high trophic plasticity Trophic diversity was inversely correlated with canopy and redundancy, one also must consider that species cover during the early rainy season. Canopy cover can be could be dissimilar in other important functional an important environmental gradient affecting aquatic aspects, a topic that merits further study. ecosystems (Grether et al., 2001; Mosisch, Bunn & Davies, 2001; Schiesari, 2006; Earl et al., 2011). In temperate Acknowledgments regions, pond canopy cover affects the distribution and abundance of aquatic species (Werner & Glennemeier, We thank R.L. Cuellar and K. Rivero for providing logis- 1999; Skelly et al., 2002; Binckley & Resetarits, 2007, 2009; tical and permit support while in Bolivia. T.E. Lacher, A. Skelly, Bolden & Freidenburg, 2014). Field experiments Stronza and two anonymous reviewers provided con- that increased the canopy cover of tropical stream reaches structive comments that greatly improved the manu- reduced the abundance of primary producers, which in script. We also thank N. Angeli, D.E. Dittmer, T.J. turn reduced the diversity and abundance of aquatic con- Hibbitts, D.J. Leavitt, W.A. Ryberg, N.L. Smolensky, sumers (Ceneviva-Bastos & Casatti, 2014). Reduction in M.L. Treglia and M. Young for their support and primary producer abundance can reduce growth rates insightful discussions during the course of this project. and developmental times of consumers (Grether et al., Support was provided by the National Science Founda- 2001). Given that most aquatic organisms in ephemeral tion’s Graduate Research Fellowship Program, the ponds are larval stages of species with terrestrial adults, Applied Biodiversity Science NSF-IGERT Program at resource availability affecting growth also influences body Texas A&M University (NSF-IGERT Award #0654377) size and timing of metamorphosis as well as mortality and Carolyn Wierichs Kelso and George Kelso via the due to pond drying. Although we did not measure International Sportfish Fund. This is publication number

© 2016 John Wiley & Sons Ltd, Freshwater Biology, 62, 519–529 Tropical pond food webs 527 1533 of the Biodiversity Research and Teaching Collec- Crump M.L. (1974) Reproductive strategies in a tropical tions at Texas A&M University. anuran community. Miscellaneous Publication – University of Kansas, Museum of Natural History, 61,1–68. Crump M.L. (2015) Anuran reproductive modes: evolving References perspectives. Journal of Herpetology, 49,1–16. Dayton G.H. & Fitzgerald L.A. (2001) Competition, preda- Altig R. & Johnston G.F. (1989) Guilds of anuran larvae: tion, and the distributions of four desert anurans. Oecolo- relationships among developmental modes, morpholo- gia, 129, 430–435. gies, and habitats. Herpetological Monographs, 3,81–109. De Meester L., Declerck S., Stoks R., Louette G., Van de Altig R. & McDiarmid R.W. (1999) Diversity: familial and Meutter F., De Bie T. et al. (2005) Ponds and pools as generic characterizations. In: Tadpoles: The Biology of Anu- model systems in conservation biology, ecology and evo- ran Larvae (Eds R. Altig & R.W. McDiarmid), pp. 295–337. lutionary biology. Aquatic Conservation: Marine and Fresh- The University of Chicago Press, Chicago. water Ecosystems, 15, 715–725. Altig R., Whiles M.R. & Taylor C.L. (2007) What do tad- Duellman W.E. & Trueb L. (1994) Biology of Amphibians. poles really eat? Assessing the trophic status of an under- Johns Hopkins University Press, Baltimore. studied and imperiled group of consumers in freshwater Earl J.E., Luhring T.M., Williams B.K. & Semlitsch R.D. habitats. Freshwater Biology, 52, 386–395. (2011) Biomass export of salamanders and anurans from Arribas R., Dıaz-Paniagua C., Caut S. & Gomez-Mestre I. ponds is affected differentially by changes in canopy (2015) Stable isotopes reveal trophic partitioning and cover. Freshwater Biology, 56, 2473–2482. trophic plasticity of a larval amphibian guild. PLoS ONE, Greig H.S., Wissinger S.A. & McIntosh A.R. (2013) Top- 10, e0130897. down control of prey increases with drying disturbance Arrington D.A. & Winemiller K.O. (2002) Preservation in ponds: a consequence of non-consumptive interac- effects on stable isotope analysis of fish muscle. Transac- tions? Journal of Animal Ecology, 82, 598–607. tions of the American Fisheries Society, 131, 337–342. Grether G.F., Millie D.F., Bryant M.J., Reznick D.N. & Binckley C.A. & Resetarits W.J. Jr (2007) Effects of forest Mayea W. (2001) Rain forest canopy cover, resource avail- canopy on habitat selection in treefrogs and aquatic ability, and life history evolution in guppies. Ecology, 82, insects: implications for communities and metacommuni- 1546–1559. ties. Oecologia, 153, 951–958. Halverson M.A., Skelly D.K., Kiesecker J.M. & Freidenburg Binckley C.A. & Resetarits W.J. Jr (2009) Spatial and tem- L.K. (2003) Forest mediated light regime linked to poral dynamics of habitat selection across canopy gradi- amphibian distribution and performance. Oecologia, 134, ents generates patterns of species richness and 360–364. composition in aquatic beetles. Ecological Entomology, 34, Hammer Ø., Ryan P. & Harper D. (2001) PAST: paleontologi- 457–465. cal statistics software package for education and data anal- Boersma K.S., Bogan M.T., Henrichs B.A. & Lytle D.A. ysis. Palaeontologia Electronica, 4, 9. Available at: http:// (2013) Invertebrate assemblages of pools in arid - land palaeo-electronica.org/2001_1/past/issue1_01.htm. streams have high functional redundancy and are resis- Hoeinghaus D.J. & Zeug S.C. (2008) Can stable isotope tant to severe drying. Freshwater Biology, 59, 491–501. ratios provide for community-wide measures of trophic Caut S., Angulo E., Dıaz-Paniagua C. & Gomez-Mestre I. structure? Comment. Ecology, 89, 2353–2357. (2013) Plastic changes in tadpole trophic ecology revealed Layman C.A., Arrington D.A., Montana~ C.G. & Post D.M. by stable isotope analysis. Oecologia, 173,95–105. (2007) Can stable isotope ratios provide for community- Ceneviva-Bastos M. & Casatti L. (2014) Shading effects on wide measures of trophic structure? Ecology, 88,42–48. community composition and food web structure of a McHugh P.A., McIntosh A.R. & Jellyman P.G. (2010) Dual deforested pasture stream: evidences from a field experi- influences of ecosystem size and disturbance on food ment in Brazil. Limnologica, 46,9–21. chain length in streams. Ecology Letters, 13, 881–890. Chase J.M. (2007) Drought mediates the importance of McHugh P.A., Thompson R.M., Greig H.S., Warburton H.J. stochastic community assembly. Proceedings of the National & McIntosh A.R. (2015) Habitat size influences food web Academy of Sciences of the United States of America, 104, structure in drying streams. Ecography, 38, 700–712. 17430–17434. Montana~ C.G., Layman C.A. & Winemiller K.O. (2015) Chase J.M. & Myers J.A. (2011) Disentangling the impor- Species–area relationship within benthic habitat patches tance of ecological niches from stochastic processes across of a tropical floodplain river: an experimental test. Austral scales. Philosophical Transactions of the Royal Society of Lon- Ecology, 40, 331–336. don B: Biological Sciences, 366, 2351–2363. Mosisch T.D., Bunn S.E. & Davies P.M. (2001) The relative Chesson P.L. (2000) Mechanisms of maintenance of species importance of shading and nutrients on algal production diversity. Annual Review of Ecology and Systematics, 31, in subtropical streams. Freshwater Biology, 46, 1269–1278. 343–366.

Navarro G. & Maldonado M. (2002) Geografıa ecologica de Boli- Schemske D.W., Mittelbach G.G., Cornell H.V., Sobel J.M. & via: Vegetacion y ambientes acuaticos. Centro de Ecologıa Roy K. (2009) Is there a latitudinal gradient in the impor- Simon I. Patino,~ Departamento de Difusion, Santa Cruz. tance of biotic interactions? Annual Review of Ecology, Evo- Pazin V.F.V., Magnusson W.E., Zuanon J. & Mendoncßa F.P. lution, and Systematics, 40, 245–269. (2006) Fish assemblages in temporary ponds adjacent to Schiesari L. (2006) Pond canopy cover: a resource gradient ‘terra-firme’ streams in Central Amazonia. Freshwater Biol- for anuran larvae. Freshwater Biology, 51, 412–423. ogy, 51, 1025–1037. Schiesari L., Werner E.E. & Kling G.W. (2009) Carnivory Perotti M.G. (1997) Modos reproductivos y variables repro- and resource-based niche differentiation in anuran larvae: ductivas cuantitativas de un ensamble de anuros del implications for food web and experimental ecology. Chaco semiarido, Salta, Argentina. Revista Chilena de His- Freshwater Biology, 54, 572–586. toria Natural, 70, 277–288. Schriever T.A. (2015) Food webs in relation to variation in Perotti M.G., Jara F.G. & Ubeda C.A. (2011) Adaptive plas- the environment and species assemblage: a multivariate ticity of life-history traits to pond drying in three species approach. PLoS ONE, 10, e0122719. of Patagonian anurans. Evolutionary Ecology Research, 13, Schriever T.A. & Williams D.D. (2013) Influence of pond 415–429. hydroperiod, size, and community richness on food-chain Petchey O.L. & Gaston K.J. (2006) Functional diversity: back length. Freshwater Science, 32, 964–975. to basics and looking forward. Ecology Letters, 9, 741–758. Semlitsch R.D., Peterman W.E., Anderson T.L., Drake D.L. Petranka J.W. & Kennedy C.A. (1999) Pond tadpoles with & Ousterhout B.H. (2015) Intermediate pond sizes contain generalized morphology: is it time to reconsider their the highest density, richness, and diversity of pond- functional roles in aquatic communities? Oecologia, 120, breeding amphibians. PLoS ONE, 10, e0123055. 621–631. Siepielski A.M. & McPeek M.A. (2010) On the evidence for Pimm S.L. (1982) Food Webs. Chapman and Hall, London. species coexistence: a critique of the coexistence program. Pimm S.L. & Lawton J.H. (1978) On feeding on more than Ecology, 91, 3153–3164. one trophic level. Nature, 275, 542–544. Skelly D.K., Bolden S.R. & Freidenburg L.K. (2014) Experi- Post D.M. (2002) Using stable isotopes to estimate trophic mental canopy removal enhances diversity of vernal position: models, methods, and assumptions. Ecology, 83, pond amphibians. Ecological Applications, 24, 340–345. 703–718. Skelly D.K., Freidenburg L.K. & Kiesecker J.M. (2002) Forest R Core Team (2013) R: A Language and Environment for Sta- canopy and the performance of larval amphibians. Ecol- tistical Computing. R Foundation for Statistical Comput- ogy, 83, 983–992. ing, Vienna. Available at: http://www.R-project.org. Skelly D.K., Werner E.E. & Cortwright S.A. (1999) Long- Rosenfeld J.S. (2002) Functional redundancy in ecology and term distributional dynamics of a Michigan amphibian conservation. Oikos, 98, 156–162. assemblage. Ecology, 80, 2326–2337. Rosenzweig M.L. (1995) Species Diversity in Space and Time. Vander Zanden M.J. & Rasmussen J.B. (1999) Primary con- Cambridge University Press, Cambridge. sumer d13C and d15N and the trophic position of aquatic Schalk C.M. (2016a) Predator-induced phenotypic plasticity consumers. Ecology, 80, 1395–1404. in an arid-adapted tropical tadpole. Austral Ecology, 41, Vanderklift M.A. & Ponsard S. (2003) Sources of variation 415–422. in consumer-diet d15N enrichment: a meta-analysis. Schalk C.M. (2016b) Community Assembly of Neotropical Frogs Oecologia, 136, 169–182. Across Ecological Scales. PhD Thesis, Texas A&M Univer- Wallace J.B. & Webster J.R. (1996) The role of macroinverte- sity, College Station. brates in stream ecosystem function. Annual Review of Schalk C.M., Montana~ C.G. & Libson M. (2014a) Reproduc- Entomology, 41, 115–139. tive strategies of two Neotropical killifish, Austrolebias van- Wellborn G.A., Skelly D.K. & Werner E.E. (1996) Mecha- denbergi and Neofundulus ornatipinnis (Cyprinodontiformes: nisms creating community structure across a freshwater Rivulidae) in the Bolivian Gran Chaco. Revista de Biologia habitat gradient. Annual Review of Ecology and Systematics, Tropical, 62, 109–117. 27, 337–363. Schalk C.M., Montana~ C.G., Klemish J.L. & Wild E.R. Werner E.E. & Anholt B.R. (1993) Ecological consequences (2014b) On the diet of the frogs of the Ceratophryidae: of the trade-off between growth and mortality rates medi- synopsis and new contributions. South American Journal of ated by foraging activity. The American Naturalist, 142, Herpetology, 9,90–105. 242–272. Schalk C.M., Montana~ C.G. & Springer L. (2015) Morpho- Werner E.E. & Glennemeier K.S. (1999) Influence of forest logical diversity and community organization of desert canopy cover on the breeding pond distributions of sev- anurans. Journal of Arid Environments, 122, 132–140. eral amphibian species. Copeia, 1999,1–12. Schalk C.M. & Saenz D. (2016) Environmental drivers of Werner E.E., Skelly D.K., Relyea R.A. & Yurewicz K.L. anuran calling phenology in a seasonal Neotropical (2007) Amphibian species richness across environmental ecosystem. Austral Ecology, 41,16–27. gradients. Oikos, 116, 1697–1712.

© 2016 John Wiley & Sons Ltd, Freshwater Biology, 62, 519–529 Tropical pond food webs 529 Whiles M.R., Gladyshev M.I., Sushchik N.N., Makhutova taxa that occurred in the early season and late season of O.N., Kalachova G.S., Peterson S.D. et al. (2010) Fatty acid the annual rainy season. The bi-plots for pond 6 are analyses reveal high degrees of omnivory and dietary depicted in Fig. 1 of the main text. plasticity in pond dwelling tadpoles. Freshwater Biology, Appendix S3. Bi-plots of the relationship between taxa – 55, 1533 1547. richness and mean nearest neighbour distance (NND) a Whiles M.R., Lips K.R., Pringle C.M., Kilham S.S., Bixby proxy for trophic redundancy for consumers collected R.J., Brenes R. et al. (2006) The effects of amphibian popu- during the early season (open circles) and late season lation declines on the structure and function of Neotropi- (filled circles) of the annual rainy season. cal stream ecosystems. Frontiers in Ecology and the d13 Environment, 4,27–34. Appendix S4. Bi-plots of the relationship between C Williams D.D. (2006) The Biology of Temporary Waters. range of the basal sources and the environmental gradi- Oxford University Press, New York. ents of (A) pond area, (B) pond hydroperiod and (C) canopy cover for consumers collected during the early season (open circles; solid trendline) and late season Supporting Information (filled circles; dashed trendline) of the annual rainy Additional Supporting Information may be found in the season. online version of this article: Appendix S5. Bi-plots of the relationship between taxa Appendix S1. Mean values of the environmental gradi- richness and the environmental gradients of (A) pond ents (canopy cover, pond area and pond hydroperiod) area, (B) pond hydroperiod and (C) canopy cover for of 13 ponds sampled in the Gran Chaco ecoregion, Boli- consumers collected during the early season (open cir- via. Variables were collected monthly from December cles; solid trendline) and late season (filled circles; 2010 to April 2011. dashed trendline) of the annual rainy season. Appendix S2. Bi-plots of d13C and d15N for consumers and basal sources for the remaining ponds surveyed in (Manuscript accepted 16 November 2016) the study area. Isotope ratios were compared for certain