bs_bs_banner Austral Ecology (2014) 39, 864–874 Amphibian richness patterns in Atlantic Forest areas invaded by American bullfrogs CAMILA BOTH,1,2* BRUNO MADALOZZO,3 RODRIGO LINGNAU4 AND TARAN GRANT1,5 1 Programa de Pós-Graudação em Zoologia, Pontifícia Universidade Católica do Rio Grande do Sul, Porto Alegre, 3Programa de Pós-Graduação em Biodiversidade Animal, Universidade Federal de Santa Maria, Santa Maria, 4Universidade Tecnológica Federal do Paraná, Francisco Beltrão, and 5 Departamento de Zoologia, Instituto de Biociências, Universidade de São Paulo, São Paulo, Brazil; and 2 School of Biological Sciences, University of Sydney, Sydney, New South Wales, Australia (Email: camilaboth@gmail.com) Abstract The relationship between invasion success and native biodiversity is central to biological invasion research. New theoretical and analytical approaches have revealed that spatial scale, land-use factors and community assemblages are important predictors of the relationship between community diversity and invasibility and the negative effects of invasive species on community diversity. In this study we assess if the abundance of Lithobates catesbeianus, the American bullfrog, negatively affects the richness of native amphibian species in Atlantic Forest waterbodies in Brazil. Although this species has been invading Atlantic Forest areas since the 1930s, studies that estimate the invasion effects upon native species diversity are lacking. We developed a model to understand the impact of environmental, spatial and species composition gradients on the relationships between bullfrogs and native species richness. We found a weak positive relationship between bullfrog abundance and species richness in invaded areas. The path model revealed that this is an indirect relationship mediated by community composition gradients. Our results indicate that bullfrogs are more abundant in certain amphibian communities, which can be species-rich. Local factors describing habitat heterogeneity were the main predictors of amphibian species richness and composition and bullfrog abundance. Our results reinforce the important role of habitats in determining both native species diversity and potential invasibility. Key words: Anura, Brazil, diversity, invasion, Lithobates catesbeianus, Ranidae. INTRODUCTION The immigration of novel species into communities or uncolonized patches is an important process for community development, structure and composition (MacArthur & Wilson 1967; Connell & Slatyer 1977; Loreau & Mouquet 1999).We are currently witnessing an age of biological invasions, with novel species being incorporated into local communities not only by dispersal, but also by intentional or accidental introductions. Such invasions are not restricted to taxa, regions, biomes, or continents (Soulé 1990), and they have attracted much attention because they can be economically expensive. For instance, exotic species can negatively affect crops, fisheries, public health systems etc., resulting in the expense of billions of dollars (Pimentel et al. 2001). Further, invasions can also be expensive in an ecological and evolutionary context, since they have been linked to recent species extinctions (Clavero & García-Berthou 2005). *Corresponding author. Accepted for publication April 2014. © 2014 The Authors Austral Ecology © 2014 Ecological Society of Australia Charles Elton predicted a scenario of ecological and economical losses linked with species invasions (Simberloff 2011), and since his seminal synthesis in ‘The Ecology of Invasions by Animals and Plants’ (Elton 1958) there is growing interest to identify species traits linked to invasion propensity and community properties facilitating or preventing invasions. In his book, Elton observed that ‘tropical systems are less invaded than species poor temperate and boreal systems’. Elton associated ‘tropical resistance’ with highly diverse communities, harbouring large numbers of enemies and parasites, which, in turn, can impede the establishment of novel species. This idea is currently recognized as the diversity–invasibility hypothesis (see Justus 2008). Elton himself recognized that this should only partially explain the low invasion rates in the tropics. Studies subsequent to Elton (1958) explained the diversity–invasibility hypothesis in the light of resource exploitation. Following this rationale, diverse tropical communities are more resistant because richer communities tend to be saturated with species that exploit resources in complementary ways (MacArthur & Levins 1967; MacArthur 1970). doi:10.1111/aec.12155 B U L L F R O G I N VA S I O N A N D A N U R A N D I V E R S I T Y The diversity–invasibility hypothesis is still a central issue in ecology (Fridley et al. 2007).There is evidence from observational studies that alien species diversity is low in richer systems, although some studies have found no relationship at all (e.g. Case & Bolger 1991). Experimental studies supplied evidence that supports both negative (e.g. Tilman 1997) and positive relationships (e.g. Robinson et al. 1995) between diversity and invasion. Nevertheless, there are also experimental studies highlighting the lack of a relationship between diversity and invasion (e.g. Robinson & Dickerson 1984). The presence of such contrasting results has been termed ‘the invasion paradox’ (Fridley et al. 2007) and seems to be related to spatial scale. In an extensive review, Fridley et al. (2007) showed that large-scale studies suggest a positive relationship between invasive species and native richness. Such positive correlation indicates that richer communities are able to accommodate introduced species, despite high numbers and abundance of native species, known as ‘biotic acceptance’ (Stohlgren et al. 2006). In contrast, at smaller scales relationships are often negative, especially in experimental studies (Fridley et al. 2007). Nonetheless, a positive relationship between nonnative and native species might be due to sampling artefacts (Fridley et al. 2004). Apparently positive or negative relationships might also result from analytical models that ignore important co-variables that might be related to invasive species abundance or richness and native species diversity (Taylor & Irwin 2004). Factors such as propagule pressure, disturbance and other environmental gradients often function synergistically to explain invasion patterns (Elton 1958; Von Holle & Simberloff 2005). Therefore, in situ studies that consider distinct spatial scales and address the relationship between native richness and invasive species are required for a thorough understanding of the diversity–invasibility hypothesis. This is especially true for approaches that consider causal links connecting environmental and community gradients. In many systems, invasions are a concrete reality, despite several decades of research and efforts for their prevention (Richardson 2011). In previously invaded ecosystems, we are challenged to estimate the invasive species effects on diversity rather than to understand whether diversity will prevent invasions.The same predictions about the relationship between diversity and invasibility described above also hold for the relationship between invasive species effects on diversity. Accordingly, at fine spatial scales negative relationships are expected to occur, because it is at smaller spatial scales that species interact. The problem faced when studying an invasion that has reached the final stages is that, without data collected prior to the invasion, we cannot know if invasive species promoted changes to communities in the past. Nevertheless, we can study the relationship between current invasive © 2014 The Authors Austral Ecology © 2014 Ecological Society of Australia 865 species and community diversity patterns, asking whether invasions are related to less diverse communities or differences in community composition, while controlling for environmental and spatial variation. Classical examples of invasion effects on ecosystem diversity come from freshwater habitats that have suffered historically from structural modifications. These are among the most endangered ecosystems, and biotic modifications imposed by the addition of invasive species could lead to local extinction of freshwater species (Witte et al. 1992; Dodds & Whiles 2010). Waterbodies such as lakes, ponds, pools and temporary streams can be viewed as small islands, which seem to be very sensitive to invasions (Ricciardi & MacIsaac 2011). The bullfrog (or American bullfrog) Lithobates catesbeianus (Shaw 1802) is a member of the ‘hall of fame’ of freshwater invasive species. It is cited as one of the 100 worst alien species in the world (Lever 2003) and is one of the causes of amphibian declines in North America. Furthermore, the species appears to be an important vector of fatal amphibian diseases such as chytridiomycosis (Daszak et al. 2004; Schloegel et al. 2010). Bullfrogs benefit from humanmodified landscapes (Adams 1999; Zampella & Bunnell 2000; D’Amore et al. 2010; Fuller et al. 2011), the presence of non-native fish in fresh waters (Adams & Wasson 2000; Adams et al. 2003), and simple aquaculture enclosures that enable a continuous source of propagules (Liu & Li 2009). Bullfrogs originate from eastern North America and are distributed from Mexico to south Canada; however, invasive populations occur in western North America, Europe, Asia, and Central and South American countries where they were introduced for aquaculture practices (Santos-Barrera et al. 2009). In invaded sites, they can be negatively correlated with native frog abundance and with the occurrence of certain native frog species; they can also reduce tadpole survival (Kupferberg 1997; Kiesecker & Blaustein 1998; Kats & Ferrer 2003; Wang & Li 2009) and are often implicated as a factor in the decline of native amphibians and potential species loss (Kraus 2009). A negative relationship between bullfrogs and native anuran richness has been reported in China (Li et al. 2011). Bullfrogs have been present in Brazil since the 1930s, and invasions have continued to occur and expand since then, probably growing exponentially from the 1970s to the 1990s when bullfrog farms received governmental incentives (Lima & Agostinho 1988). Brazil is highly suitable for bullfrog establishment, especially in the Atlantic Forest, and climate change might make protected forest areas even more prone to invasion by bullfrog populations (Nori et al. 2011; Loyola et al. 2012). Although the relationship between invasive bullfrogs and the diversity of native communities in the Atlantic Forest is currently unknown, we do know that bullfrogs are widespread in doi:10.1111/aec.12155 866 C . B OT H ET AL. Brazil and that the number of known localities is increasing (Both et al. 2011b). In this study, we tested relationships between the abundance of the non-native American bullfrog and native diversity in areas where the invasion had reached a fully invasive degree (sensu Blackburn et al. 2011). Our primary goal was to test whether bullfrog abundance can predict richness patterns by assessing how amphibians in Atlantic Forest communities vary in richness in invaded areas. We expected a negative relationship because we were investigating the relationship between invasive species abundance and native community richness at a fine spatial scale. Furthermore, we developed a model to understand the interplay of environmental, spatial and species composition gradients with the relationship between bullfrogs and native species richness. METHODS Study areas The study was conducted in three different areas of the Atlantic Forest domain in southern Brazil: the central region of Rio Grande do Sul State and the western and eastern portions of Santa Catarina State (see Appendix S1 for coordinates). These areas are highly susceptible to bullfrog invasions and are in regions where records of L. catesbeianus in Brazil are concentrated (Both et al. 2011b). Invasions date to at least 10 years ago in all cases. There were no active bullfrog farms near the study areas. In study area A1 (central Rio Grande do Sul), the vegetation is characterized by seasonal deciduous forest (IBGE (Instituto Brasileiro de Geografia e Estatística 2004). In western Santa Catarina, study area A2, two phytogeographic forms of Atlantic Forest occur, including seasonal deciduous forest and mixed ombrophilous forest (IBGE (Instituto Brasileiro de Geografia e Estatística 2004; Lucas & Fortes 2008). In eastern Santa Catarina, study area A3, the vegetation is formed by dense ombrophilous forest (IBGE (Instituto Brasileiro de Geografia e Estatística 2004). The Atlantic Forest in the three study areas is highly fragmented. The general land use is farming and cattle grazing, with larger forest fragments found in protected areas. In A1, Parque Estadual da Quarta Colonia protects 1847.90 ha along the border of the Dona Francisca hydroelectric dam. A2 includes the Floresta Nacional de Chapecó, with 1606.3 ha divided in two areas. Parque Nacional da Serra do Itajaí, with 57 374 ha, is located within A3. breeding (calling males, eggs, tadpoles), and 15 waterbodies without, alternating spatially.This site selection allowed us to collect data for invaded and non-invaded sites in each study area.We considered sites that presented evidence of breeding to be invaded, given that eggs, tadpoles or adults were present. Water bodies with no evidence of breeding are likely to be non-invaded. Juveniles are known to disperse from their natal waterbody and can occasionally be found in ephemeral pools or when crossing roads at night (Both, pers. obs., 2010); therefore, the presence of juveniles in ponds is likely to be temporary and we did not consider it to be evidence that a site has been invaded. We sampled a total of 90 waterbodies: 32 in A1, 30 in A2 and 28 in A3 (see Appendix S1). We surveyed each site twice for post-metamorphic individuals, tadpoles and egg clutches. Collections were made over two 25-day periods in February/ March (late summer) and October/November (spring) of 2010. Sampling months were chosen to represent periods of longer photoperiod, which is a predictor of high native amphibian breeding activity in subtropical regions (Canavero & Arim 2009), and also coincides with the breeding period of bullfrogs in southern Brazil (Kaefer et al. 2007). We sampled 4–6 waterbodies per day for eggs and tadpoles in the daytime and post-metamorphic individuals at night. We sampled tadpoles in distinct microhabitats with dip net sweeps (40 × 30 cm, mesh 0.02 mm), always using the same collector. At least five sweeps were undertaken in each microhabitat (covering approx. 1 m2).The effort was proportional to pond size and heterogeneity (Shaffer et al. 1994). Microhabitats were classified into marginal/shallow vegetated, marginal/shallow non-vegetated, centre/deep vegetated and centre/deep non-vegetated types. Marginal/ shallow microhabitas are generally <30–50 cm deep, while centre/deep habitats are generally >50 cm deep. Heterogeneity is positively correlated with waterbody depth and size; therefore, fewer microhabitats were sampled in small, shallow, non-vegetated waterbodies (approx. 4–6) than in large, permanent, vegetated waterbodies with all four types of microhabitats (approx. 20 microhabitats sampled). In the largest ponds we also sampled microhabitats midway between the margin and the centre. The total number of microhabitats sampled in each waterbody in a single survey ranged from 4–30. Samples were spaced at least 2 m apart. Samples were taken between 09:00–19:00 h, always during daylight. Post-metamorphic surveys were begun 30 min after sunset, and we searched for individuals along the perimeters of breeding sites.We counted all post-metamorphic individuals, separating calling and non-calling individuals. Sampling effort was proportional to size and heterogeneity of the waterbody (Scott & Woodward 1994). Waterbody and landscape descriptors Waterbody selection and sampling In spring 2009, we carried out initial surveys to select waterbodies for subsequent sampling, including natural marshes, ponds, stream sections with low water flow (streamside pools), and artificial ponds. In each area we identified 15 waterbodies with evidence of L. catesbeianus doi:10.1111/aec.12155 We measured the distance from the waterbody to the nearest forest fragment and to the nearest road using GPS coordinates. We expanded these distances in second- and third-order monomials to test for potential non-linear relationships with richness. Area and depth of each waterbody were measured at each collection event. For depth, we took the measurements in each of the microhabitats sampled for © 2014 The Authors Austral Ecology © 2014 Ecological Society of Australia B U L L F R O G I N VA S I O N A N D A N U R A N D I V E R S I T Y tadpoles and recorded the mean, maximum, and minimum depth. Water surface area ratio obtained from the two sampling events (summer area/spring area) was used as a measure of pond permanence. We recorded hydrophyte morphotype (floating and emergent macrophytes) richness in the pond. We visually estimated the hydrophyte coverage of waterbodies and classified it into three classes: <30%, 30–60% and >60%. Fish presence/absence was determined by visual surveys or dip net captures or was based on information provided by local rural owners. We classified the vegetation of waterbody banks into low grass, tall grass, shrubs, and trees and used the sum of types present in a site as a descriptor of the bank vegetation structure. Spatial descriptors This study was performed at two major scales: between areas (A1, A2 and A3) and within areas. To account for this nested spatial arrangement, we considered two kinds of spatial descriptors: the three large study areas represented by dummy variables, and Moran’s eigenvector maps (MEMs). MEMs provide orthogonal spatial variables ranging from broad, inter-area scales to the finest scales derived from the geographic coordinates of the sites (Dray et al. 2006). The spatial arrangement in this study shows a high truncation value (minimum distance connecting all points = 334 km) between study areas. We used a nested MEM model (Declerck et al. 2011), where MEM variables are blocked within study areas to describe spatial relationships at this scale. We used the R function create.MEM.model (Declerck et al. 2011) to build the MEM model using a matrix of dummy variables that described the three regions as the connectivity criteria. The nested analysis resulted in seven spatial vectors describing the spatial arrangement of sites within areas. These seven MEM vectors and the dummy variables of the three large study areas were used as spatial descriptors. 867 and spatial variation. This analytical approach allowed us to depict the co-variation structure between predictors (Shipley 2000). First, we tested whether bullfrog abundance affects native richness. Using path analysis, we then tested whether species richness is related to (i) waterbody, (ii) landscape, and (iii) spatial descriptors, jointly with (iv) community composition and (v) bullfrog abundance. In order to assess the relative importance of all five sets of predictors, we built a hierarchical theoretical model of potential causal relationships explaining native species richness (Fig. 1). This model assumes a hierarchical causal order between predictors, where spatial descriptors have the highest causal order and all other variables are endogenous (see causal links in Fig. 1). We tested this model using the analytical steps proposed by Brum et al. (2012). First, we performed a model selection procedure using each set of descriptors as predictors of native richness to identify the descriptors that are directly related with native richness. We then regressed all selected descriptors from all sets against richness to test for direct causal links. Obeying the causal hierarchy, we regressed all variables directly linked with richness on their respective potential predictors (immediate higher hierarchy factors, Fig. 1), and so on, until we reached the exogenous descriptors, or spatial descriptors in this case. For instance, if a waterbody descriptor was directly related to local richness, we further investigated which landscape and spatial descriptors could explain this descriptor. If a landscape descriptor was selected, we inspected its relationships with high-order spatial descriptors. Fish presence was considered together with other waterbody descriptors. However, we investigated further causal links associated with this variable (see Fig. 1) because it is a well-known filter for lentic communities (Wellborn et al. 1996). Beta regression coefficients of linear models were taken as path coefficients. Model selection was based on corrected Akaike information criterion (Anderson 2008). All regression models were performed in R (R Development Core Team 2012). RESULTS Data analyses To remove the effects of sampling effort on richness values, we regressed richness on the abundance of postmetamorphic individuals, which showed a 96% concordance with total richness, and used the residuals, henceforth called richness, in subsequent analyses. To describe species composition across waterbodies, we calculated Jaccard’s index of similarity for pairs of communities and performed a principal coordinate analysis (PCoA) (Legendre & Legendre 1998) using only occurrence (presence/absence) data. We chose occurrences instead of species abundances because they allowed us to combine post-metamorphic and tadpole surveys, thereby dispensing with further weighting assumptions about the importance of distinct life history stages to the characterization of community composition.The stability of ordination axes was evaluated by bootstrap resampling (Pillar 1999). The PCoA analysis was performed in Multiv (Pillar 2006). We used successive regression models to analyse the effects of bullfrogs on richness, taking into account environmental © 2014 The Authors Austral Ecology © 2014 Ecological Society of Australia Across the three study areas we recorded 40 native amphibian species: 19 species in A1, 19 in A2 and 23 in A3 (Table 1). Native amphibian richness ranged from 0–11 species, with an average of 3.9 species per waterbody (SD = 2.8). The ordination of species composition resulted in 37 PCoA axes. The first three axes represented 41% of the species composition gradients and were used in the regression models. The stability of the first three ordination axes was validated by bootstrap resampling. The first two axes represented the composition gradient from ombrophilous vegetation to mixed ombrophilous vegetation to seasonal deciduous forest amphibian communities (Fig. 2A). The first principal coordinate (PC1) axis clearly varied from amphibian compositions restricted to dense ombrophilous forest (e.g. Sphaenorhynchus sp. and Bokermanohyla hylax) to those occurring in high frequencies across the study regions (e.g. Dendropsophus minutus and Physalaemus cuvieri) (Fig. 2B). doi:10.1111/aec.12155 868 C . B OT H ET AL. Fig. 1. Theoretical path model explaining relationships of native amphibian richness with community gradients, space, landscape, waterbody descriptors and American bullfrog abundance in invaded areas. We detected bullfrogs at 62 of the 90 sampled waterbodies, 44 of which showed evidence of breeding (16 in A1, 16 in A2 and 12 in A3), indicating the presence of established populations. At least eight sites in each study area lacked evidence of bullfrogs at any life stage. The linear model relating native amphibian richness to bullfrog abundance showed a weak positive relationship (Table 2). Richness was also positively related to the first two principal coordinates of community composition, PC1 and PC2 (Fig. 2). Three MEMs were selected in the spatial model: MEM-2, which described spatial structure within A2; MEM-3, which described spatial structure within A3; and MEM-7 describing spatial structure within A1. The model relating richness with local waterbody descriptors selected maximum depth, hydrophyte richness, bank vegetation structure, fish presence and streamside pools as predictors (Table 2). The path model showed that native amphibian richness was directly determined by the first two principal coordinates of community composition (PC1 and PC2), maximum depth, and the binary descriptors for fish presence, as well as by one spatial model, MEM-2 (Fig. 3). The composition gradients described by PC1 and PC2 showed the highest path coefficients, and therefore are the main predictors of native amphibian richness. Stream side waterbodies and two spatial models (MEM-3 and MEM-4), did not show significant causal links with amphibian richness. Bullfrog abundance was only indirectly related to amphibian species richness through its relationship with the community composition gradient of PC1. Hydrophyte doi:10.1111/aec.12155 richness was also only indirectly linked to richness through the composition gradients. Hydrophyte richness directly determined bullfrog abundance and community composition gradients PC1 and PC2 (Fig. 3). The final model accounted for 63% of native richness variation. DISCUSSION Traditionally, the diversity–invasibility relationship has been explored in order to investigate whether diversity will prevent invasions. In the face of increasing invasion rates and successful establishment of invaders, it has become important to explore the effect of invasions on species diversity at a local scale. In this study, we assessed whether invasion affects diversity in already successfully occupied areas, expecting a negative relationship between bullfrog abundance and native amphibian richness. Contrary to our initial expectation, we found a weak positive relationship, which was revealed to be indirectly linked to species composition gradients. Local environmental gradients describing waterbody features directly explained all biotic components of the model. They also showed an additional and important influence of species composition gradients on native richness. Bullfrog abundance also responded to the same waterbody gradients. Across the three Atlantic Forest areas, the influence of space upon richness patterns was small.We found only weak relationships between micro-regions and lower richness in A1 and A2, and higher richness in A3. © 2014 The Authors Austral Ecology © 2014 Ecological Society of Australia B U L L F R O G I N VA S I O N A N D A N U R A N D I V E R S I T Y Table 1. 869 Amphibian composition from southern Atlantic Forest areas invaded by Lithobates catesbeianus Species Bufonidae Rhinella abei Rhinella fernandezae Rhinella icterica Cycloramphidae Limnomedusa macroglossa Hylidae Bokermannohyla hylax Dendropsophus minutus Dendropsophus microps Dendropsophus nahdereri Dendropsophus nanus Dendropsophus sanborni Dendropsophus werneri Hypsiboas albomarginatus Hypsiboas albopunctatus Hypsiboas bishoffi Hypsiboas faber Hypsiboas pulchellus Hypsiboas semilineatus Hypsiboas cf. semiguttatus 1 Hypsiboas cf. semiguttatus 2 Phyllomedusa tetraploidea Phyllomedusa distincta Scinax alter Scinax fuscovarius Scinax granulatus Scinax perereca Sphaenorhynchus sp. Hylodidae Crossodactylus sp. Leiuperidae Physalaemus gracilis Physalaemus cuvieri Physalaemus nanus Pseudopaludicola falcipes Leptodactylidae Leptodactylus fuscus Leptodactylus gracilis Leptodactylus labyrynthicus Leptodactylus joly Leptodactylus latinasus Leptodactylus latrans Leptodactylus mystacinus Leptodactylus plaumani Microhylidae Elachistocleis bicolor Total Abbreviation Post-metamorphics Tadpoles Rab Rfe Ric 1 1 1 1 Lma 1 Bhy Dmin Dmic Dnah Dnan Dsa Dwe Halb Halbp Hbi Hfa Hpu Hse Hsm1 Hsm2 Pte Pdi Sal Sfu Sgr Spe Sphe 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Cro A1 1 1 1 1 Lfu Lgr Llab Ljo Llat Llat Lmy Lpl 1 1 Ebi A3 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 34 1 30 1 19 1 1 1 1 1 1 1 Pgr Pcu Pna Pfa A2 1 1 1 1 1 1 1 1 1 1 1 19 1 23 A1 = central region of Rio Grande do Sul, deciduous forest; A2 = western region of Santa Catarina, deciduous and mixed ombrophilous forest; A3 = east of Santa Catarina, ombrophilous forest. However, only one spatial model was causally significant when considering the entire co-variation structure between all sets of predictors. At fine spatial scales, negative relationships are expected because it is at these scales that individuals actually interact. At this local scale, biotic resistance of the native community or impact of invasive species are © 2014 The Authors Austral Ecology © 2014 Ecological Society of Australia expected to operate, and both would lead to negative relationships (Elton 1958). However, it has been shown that native species differ greatly in the way they are affected by invasive species. For instance, Thomaz and Michelan (2011) studied the effects of an invasive macrophyte on several native macrophytes at distinct spatial scales and observed that only some of the native doi:10.1111/aec.12155 870 C . B OT H ET AL. Fig. 2. Ordination of amphibian species composition across the three study areas. (A) First two axes of principal coordinate analysis (PCoA) revealing composition gradients across the study areas (PC1 and PC2). Triangles indicate species composition from A1 (seasonal deciduous forest), circles indicate compositions from A2 (mixed ombrophilous forest), and squares from A3 (ombrophilous forest). (B) Correlation scatter plot for amphibian species with the first two PCoA axes. Amphibian species are identified by abbreviations (see Table 2). macrophytes were affected, and, further, that these negative associations are more common at smaller scales. A frequently neglected issue is that at fine scales species facilitation can occur, which would lead to positive relationships (Bruno et al. 2003). In our case, a structurally complex community might provide greater resources for bullfrogs, which are generalist predators (Bury & Whelan 1984). In turn, bullfrogs might be able to explore diverse resources in waterbodies, including anuran prey, without affecting native diversity. Here, we tested whether bullfrog abundance can predict anuran richness and found that this relationdoi:10.1111/aec.12155 ship is mediated by community composition. Bullfrogs were more abundant in certain species-rich communities. Bullfrog abundance was positively related to the PC1 community gradient, varying in a similar fashion in this gradient to common species like Physalaemus cuvieri and Dendropsophus minutus that are broadly distributed in South America (IUCN 2011). These species have been recognized as anthropogenically adapted (Santos et al. 2007), like bullfrogs. These results are in accordance with those of Bunnell and Zampella (2008), who found that bullfrogs are associated with distinct species compositions, being more closely associated with species that have a widespread distribution. However, bullfrogs do co-occur with other communities in the Atlantic Forest, albeit at lower abundance, and might affect the communities by other pathways. For instance, Both and Grant (2012) found evidence that bullfrogs have the potential to affect the acoustic niche of native amphibians: tree-frogs naïve to bullfrog vocalizations significantly altered advertisement call frequencies in response to bullfrog advertisement calls. In this case, bullfrogs might not cause an immediate decrease in richness but could affect female mate selection and result in important but less immediate fitness consequences (Both & Grant 2012). Our results highlight how all biotic components – species richness, composition, fish presence, and bullfrog abundance – respond directly or indirectly to waterbody descriptors. These waterbody descriptors act as local scale filters, structuring both native species diversity and bullfrog abundance. Hydrophyte richness and maximum depth were the main predictors describing waterbodies selected in our final model. These predictors could be viewed as a measure of habitat heterogeneity, which undoubtedly plays an important role for both faunal biodiversity and abundance. Heterogeneity has the potential to decouple trophic interactions, promoting greater diversity (Kovalenko et al. 2012). Despite using a simplistic measure of heterogeneity of the water surface (richness of hydrophyte morphotypes), we found that it was an important predictor of community gradients. Hydrophytes could promote diversity by providing shelter for tadpoles and postmetamorphic individuals and calling sites or for adult males. Maximum depth is a gradient that contains variation related to pond area, permanence, availability of other lower depths, and is also an important resource for communities in small waterbodies. It governs the coexistence of nektonic and benthic tadpoles in ponds and streams (Eterovick & Fernandes 2001; Both et al. 2011a). Fish and bullfrog presence are also regulated by this gradient, because only waterbodies with a certain depth can support these organisms (Kushlan 1976; Bury & Whelan 1984). © 2014 The Authors Austral Ecology © 2014 Ecological Society of Australia B U L L F R O G I N VA S I O N A N D A N U R A N D I V E R S I T Y 871 Table 2. Linear model relating native amphibian richness and bullfrog abundance (1) and best models relating native richness with three different sets of predictors: (2) community gradients, (3) waterbody descriptors and (4) spatial models (1) Bullfrog abundance (2) Community gradients PC1 + PC2 (3) Waterbody descriptors Depthmax + bank vegetation structure* + hydrophyte richness + stream* + fish presence* (4) Spatial descriptors MEM.2 + MEM.3 + MEM.7 R2 AICc AICc wi 0.05 0.51 405.17 348.24 – 0.75 0.31 384.94 0.27 0.14 401.22 0.10 The selection of best models was based on corrected Akaike information criterion (AICc) and Akaike weight (AICc wi). The selected models have the highest AICc wi, which describes the relative likelihood of the model, normalized across the set of all possible models to sum 1. Fixed factors are identified by (*). Principal coordinates of community composition (PC) were used as predictors in model 2 and Moran’s eigenvector maps (MEM) were used as predictors in model 4. Abbreviations in model (3): Depthmax = maximum depth recorded for a given waterbody; bank vegetation structure = number of types of vegetation present in waterbody banks; stream = stream side pool (dummy variable describing type of waterbody). Fig. 3. Causal relationships between selected spatial and environmental predictors, amphibian species composition gradients, bullfrog abundance and native amphibian richness. Standardized path correlation coefficient values follow each path. Solid arrows (—) represent positive relationships and dashed arrows (- - -) represent negative relationships. Only paths with P ≤ 0.05 are shown. U = non-determination coefficient (U = 1 − R2). Richness = native amphibian richness; Hydrophyte R = hydrophyte richness; Veg. types = number of structural vegetation types around the ponds; Depth = maximum depth recorded for a waterbody; bullfrogs = Lithobates catesbeianus abundance; Fish = fish presence/absence; PC1 = first principal coordinate of community composition; PC2 = second principal coordinate of community composition; MEM.2 = Moran’s eigenvector two, describing spatial structure within study area A1. Habitat heterogeneity has been recognized as a key factor determining the persistence of native frog populations in areas invaded by bullfrogs in North America. Adams et al. (2011) studied extinction probabilities of Rana aurora populations, testing whether they could be related to bullfrog invasion and habitat modification. They did not find supporting evidence for an effect of bullfrogs on local extinction probability for the native species. Their results showed that vegetation cover and riparian vegetation are the best predictors of low extinction risk on local scales for the native frog populations. The relationship between species composition and bullfrog © 2014 The Authors Austral Ecology © 2014 Ecological Society of Australia abundance with local waterbody descriptors observed here are encouraging, because they suggest a means for habitat management to protect native frogs. In this study, we found that waterbody depth was causally linked with bullfrog abundance and species richness through positive and negative paths, respectively. In addition, previous studies showed that waterbody modification favours bullfrogs and promotes changes in community composition (e.g. Bunnell & Zampella 2008; Fuller et al. 2011). Waterbodies in areas of special interest for conservation could be managed to favour native amphibians, or to prevent the establishment of bullfrogs. doi:10.1111/aec.12155 872 C . B OT H ET AL. Our study is not the first to note a positive relationship between amphibian community richness and invasion.This pattern has been reported for native and invasive amphibian and reptile communities on broader spatial scales, suggesting the occurrence of biotic acceptance (Stohlgren et al. 2006). Poessel et al. (2013) studied how native richness and alien amphibian species establishment are related in Europe and North America and found positive relationships in both regions. They used invasive species richness as a measure of invasibility, and not single species abundance, as we used in the present study to access bullfrog effects. Nevertheless, they also concluded that environmental gradients favouring higher native species richness also favour invasions. Additionally, it is worth noting that only recent studies dealing with bullfrog invasion patterns highlight neutral relationships between bullfrogs and richness or emphasize the role of habitat filtering to promote or facilitate invasions (e.g. D’Amore et al. 2010; Adams et al. 2011). Such results might be due to study spatial scale and/or refinement of analytical approaches. Another possibility is that over time, some communities might reach a new dynamic equilibrium state in which bullfrogs are not a major threat. An important consideration when comparing the results of the present study with those of previous studies in other regions is that the phylogenetic relationship between bullfrogs and native species is much closer in those regions than in the Atlantic Forest localities we studied. In Asia, Europe and North America there are more ranid species. In China, where bullfrogs are known to negatively affect richness, native species composition mostly consists of ranids (Li et al. 2011). In contrast, only a single native ranid species occurs in the Atlantic Forest (Lithobates palmipes; Hillis & de Sá 1988), and it is restricted to the northern portion of the biome. No ranids occurred at any of the localities we studied, and the closest relative we found in sympatry with bullfrogs is the microhylid Elachistocleis bicolor. The potential influence of relatedness upon invasions was recognized by Darwin (1859) in ‘The Origin of Species’. He suspected that a low degree of relatedness should give some advantage to novel species in new habitats. This idea deserves further investigation, considering the global scale of invasions by bullfrogs and many other alien species. In conclusion, we found only a weak and indirect, positive relationship between native anuran richness and bullfrog abundance, which is mediated by species composition at the invaded sites. Our results indicate biotic acceptance of bullfrogs in the Atlantic Forest.We observed that bullfrogs were more abundant in some amphibian communities that include anthropogenicadapted species. The main result is that local waterbody descriptors predict native richness and composition and bullfrog abundance. These results doi:10.1111/aec.12155 agree with recent studies suggesting that environmental gradients directly affect both native and invasive species. Although we did not detect negative effects of bullfrogs on native richness, we do not conclude that such effects are absent in general. Disease transmission, differential predation, and competition by interference are mechanisms that could plausibly operate in invaded areas, even when bullfrogs have low abundance. ACKNOWLEDGEMENTS This study was supported by IGRE Associação Sócio Ambientalista and received grants from Fundação o Boticário de Proteção a Natureza (0835_20092) and the Rufford Small Grants Foundation. The Conselho Nacional de Desenvolvimento Científico provided a student fellowship to CB and a grant (Proc. 476789/ 2009-5) and fellowship (Procs. 305473/2008-5, 307001/2011-3) to TG. The Fundação de Amparo à Pesquisa do Estado de São Paulo provided a grant to TG (Proc. 2012/10000-5). ICMBio granted a collection permit for this study (Proc. 23009-1).This manuscript benefited from comments by Sandra Hartz, Célio Haddad, Márcio Borges-Martins, Adriano Sanches Melo, and three anonymous reviewers.We are thankful to all rural owners who granted access to study sites and the Parque Nacional da Serra do Itajaí for logistical support. We also thank S. Declerck for sharing the nested MEM function, V.F. Caorsi for her help with laboratory procedures, and D. BorgesProvette for help with the literature for tadpole identification. REFERENCES Adams M. J. (1999) Correlated factors in amphibian decline: exotic species and habitat change in Western Washington. J. Wildl. Manage. 63, 1162–71. Adams M. J., Pearl C. A. & Bury R. B. (2003) Indirect facilitation of an anuran invasion by non-native fishes. Ecol. Lett. 6, 343–51. Adams M. J., Pearl C. A., Galvan S. & Mccreary B. (2011) Non-native species impacts on pond occupancy by an anuran. J.Wildl. Manage. 75, 30–5. Adams M. J. & Wasson K. (2000) Pond permanence and the effects of exotic vertebrates on anurans. Ecol. Appl. 10, 559– 68. Anderson D. R. (2008) Model based Inference in the Life Sciences. Springer, New York. Blackburn T. M., Pyšek P., Bacher S. et al. (2011) A proposed unified framework for biological invasions. Trends Ecol. Evol. 26, 333–9. Both C., Cechin S. Z., Melo A. S. & Hartz S. M. (2011a) What controls tadpole richness and guild composition in ponds in subtropical grasslands? Austral Ecol. 36, 530–6. © 2014 The Authors Austral Ecology © 2014 Ecological Society of Australia B U L L F R O G I N VA S I O N A N D A N U R A N D I V E R S I T Y Both C. & Grant T. (2012) Biological invasions and the acoustic niche: the effect of bullfrog calls on the acoustic signals of white-banded tree frog. Biol. Lett. 8, 714–16. Both C., Lingnau R., Santos A. Jr. et al. (2011b) Widespread occurrence of the American bullfrog, Lithobates catesbeianus (Shaw, 1802) (Anura: Ranidae), in Brazil. S. Am. J. Herpetol. 6, 127–34. Brum F. T., Kindel A., Hartz S. M. & Duarte L. D. S. (2012) Spatial and phylogenetic structure drive frugivory in Tyrannidae birds across the range of Brazilian Araucaria forests. Oikos 121, 899–906. Bruno J. F., Stachowicz J. J. & Bertness M. D. (2003) Inclusion of facilitation into ecological theory. Trends Ecol. Evol. 18, 119–25. Bunnell J. F. & Zampella R. A. (2008) Native fish and anuran assemblages differ between impoundments with and without non-native centrarchids and bullfrogs. Copeia 2008, 931–9. Bury R. B. & Whelan J. A. (1984) Ecology and Management of the Bullfrog. US Department of the Interior, Fish and Wildlife Service, Washington, DC. Canavero A. & Arim M. (2009) Clues supporting photoperiod as the main determinant of seasonal variation in amphibian activity. J. Nat. Hist. 43, 2975–84. Case T. J. & Bolger D. T. (1991) The role of introduced species in shaping the distribution and abundance of island reptiles. Evol. Ecol. 5, 272–90. Clavero M. & García-Berthou E. (2005) Invasive species are a leading cause of animal extinctions. Trends Ecol. Evol. 20, 110. Connell J. H. & Slatyer R. O. (1977) Mechanisms of succession in natural communities and their role in community stability and organisation. Am. Nat. 111, 1119–44. Darwin C. (1859) On the Origin of Species by Means of Natural Selection. Murray, London. Daszak P., Strieby A., Cunningham A. A. et al. (2004) Experimental evidence that the bullfrog (Rana catesbeiana) is a potential carrier of chytridiomycosis, an emerging fungal disease of amphibians. Herpetol. J. 14, 201–7. D’Amore A., Hemingway V. & Wasson K. (2010) Do a threatened native amphibian and its invasive congener differ in response to human alteration of the landscape? Biol. Invasions 12, 145–54. Declerck S. A. J., Coronel J. S., Legendre P. et al. (2011) Scale dependency of processes structuring metacommunities of cladocerans in temporary pools of High-Andes wetlands. Ecography 34, 296–305. Dodds W. & Whiles M. (2010) Freshwater Ecology: Concepts and Environmental Applications of Limnology, 2nd edn. Academic Press, San Diego. Dray S., Legendre P. & Peres-Neto P. R. (2006) Spatial modelling: a comprehensive framework for principal coordinate analysis of neighbour matrices (PCNM). Ecol. Model. 196, 483–93. Elton C. (1958) The ecology of Invasions by Animals and Plants, 2000th edn. The University of Chicago Press, Chicago, London. Eterovick P. C. & Fernandes G. W. (2001) Tadpole distribution within montane meadow streams at the Serra do Cipó, southeastern Brazil: ecological or phylogenetic constraints? J. Trop. Ecol. 17, 683–93. Fridley J. D., Brown R. L. & Bruno J. F. (2004) Null models of exotic invasion and scale-dependent patterns of native and exotic species richness. Ecology 85, 3215–22. © 2014 The Authors Austral Ecology © 2014 Ecological Society of Australia 873 Fridley J. D., Stachowicz J. J., Naeem S. et al. (2007) The invasion paradox: reconciling pattern and process in species invasions. Ecology 88, 3–17. Fuller T. E., Pope K. L., Ashton D. T. & Welsh H. H. J. (2011) Linking the distribution of an invasive amphibian (Rana catesbeiana) to habitat conditions in a managed river system in northern California. Restor. Ecol. 19, 204–13. Hillis D. M. & de Sá R. O. (1988) Phylogeny and taxonomy of the Rana palmipes group (Salientia: Ranidae). Herpetol. Monogr. 2, 1–26. IBGE (Instituto Brasileiro de Geografia e Estatística) (2004) Mapa de vegetação do Brasil. [Cited 12 Apr 2012.] Available from URL: http://www.ibge.gov.br/home/geociencias/ default_prod.shtm IUCN (2011) The IUCN Red list of threatened species. Version 2011.2. [Cited 10 Nov 2011.] Available from URL: http:// www.iucnredlist.org/ Justus J. (2008) Complexity, diversity, stability. In: A companion to the Philosophy of Biology (eds S. Sarkar & A. Plutynski) pp. 321–50. Blackwell, Oxford. Kaefer Í. L., Boelter R. A. & Cechin S. Z. (2007) Reproductive biology of the invasive bullfrog. Ann. Zool. Fenn. 44, 435–44. Kats L. B. & Ferrer R. P. (2003) Alien predators and amphibian declines: review of two decades of science and the transition to conservation. Divers. Distrib. 9, 99–110. Kiesecker J. M. & Blaustein A. R. (1998) Effects of introduced bullfrogs and smallmouth bass on microhabitat use, growth, and survival of native red-legged frogs (Rana aurora). Conserv. Biol. 12, 776–87. Kovalenko K. E., Thomaz S. M. & Warfe D. M. (2012) Habitat complexity: approaches and future directions. Hydrobiologia 685, 1–17. Kraus F. (2009) Alien Reptiles and Amphibians: A Scientific Compendium and Analysis. Springer, Dordrecht. Kupferberg S. J. (1997) Bullfrog (Rana catesbeiana) invasion of a California river: the role of larval competition. Ecology 78, 1736–51. Kushlan J. A. (1976) Environmental stability and fish community diversity. Ecology 57, 821–5. Legendre P. & Legendre L. (1998) Numerical Ecology, 2nd edn. Elsevier, Amsterdam, Boston, London, New York, Oxford, Paris, San Diego, San Francisco, Singapore, Sydney,Toquio. Lever C. (2003) Naturalized Reptiles and Amphibians of theWorld. Oxford University Press, Oxford. Li Y., Ke Z., Wang Y. & Blackburn T. M. (2011) Frog community responses to recent American bullfrog invasions. Can. J. Zool. 57, 83–92. Lima S. L. & Agostinho C. A. (1988) A Criação de Rãs, 2nd edn. Editora Globo, São Paulo. Liu X. & Li Y. (2009) Aquaculture enclosures relate to the establishment of feral populations of introduced species. PLoS ONE 4, e6199. Loreau M. & Mouquet N. (1999) Immigration and the maintenance of local species diversity. Am. Nat. 154, 427–40. Loyola R. D., Nabout J. C., Trindade-Filho J. et al. (2012) Climate change might drive species into reserves: a case study of the American bullfrog in the Atlantic Forest Biodiversity Hotspot. Alytes 29, 61–74. Lucas E. M. & Fortes V. B. (2008) Frog diversity in the Floresta Nacional de Chapecó, Atlantic Forest of southern Brazil. Biota Neotrop. 8, 51–61. MacArthur R. (1970) Species packing and competitive equilibrium for many species. Theor. Popul. Biol. 1, 1–11. MacArthur R. H. & Wilson E. O. (1967) The Theory of Island Biogeography. Princeton University Press, Princeton. doi:10.1111/aec.12155 874 C . B OT H ET AL. MacArthur R. & Levins R. (1967) The limiting similarity, convergence, and divergence of coexisting species. Am. Nat. 101, 377–85. Nori J., Urbina-Cardona J. N., Loyola R. D., Lescano J. N. & Leynaud G. C. (2011) Climate change and American bullfrog invasion: what could we expect in South America? PLoS ONE 6, e25718. Pillar V. D. (2006) Multiv: multivariate exploratory analysis, randomization testing and bootstrap resampling. [Cited 22 Nov 2011.] Available from URL: http://ecoqua.ecologia. ufrgs.br/arquivos/Software/MULTIV/ Pillar V. P. (1999) How sharp are classifications? Ecology 80, 2508. Pimentel D., McNair S., Janecka J. et al. (2001) Economic and environmental threats of alien plant, animal, and microbe invasions. Agric. Ecosyst. Environ. 84, 1–20. Poessel S. A., Beard K. H., Callahan C. M., Ferreira R. B. & Stevenson E. T. (2013) Biotic acceptance in introduced amphibians and reptiles in Europe and North America (ed M. Olalla-Tárraga). Glob. Ecol. Biogeogr. 22, 192–201. R Development Core Team (2012) R: a language and environment for statistical computing. [Cited 15 March 2012.] Available from URL: www.R-project.org Ricciardi A. & MacIsaac H. J. (2011) Impacts of biological invasions on freshwater ecosystems. In: FiftyYears of Invasion Ecology: The Legacy of Charles Elton (ed. D. M. Richardson) pp. 211–24. Wiley-Blackwell, Oxford. Richardson D. M. (2011) Fifty Years of Invasion Ecology. WileyBlackwell, Oxford. Robinson G. R., Quinn J. F. & Stanton M. L. (1995) Invasibility of experimental habitat islands in a California winter annual grassland. Ecology 76, 786–94. Robinson J.V. & Dickerson J. E. (1984) Testing the invulnerability of laboratory island communities to invasion. Oecologia 61, 169–74. Santos T. G., Rossa-feres D. C. & Casatti L. (2007) Diversidade e distribuição espaço-temporal de anuros em região com pronunciada estação seca no sudeste do Brasil. Iheringia Ser. Zool. 97, 37–49. Santos-Barrera G., Hammerson G., Hedges B. et al. (2009) Lithobates catesbeianus. : IUCN Red List of Threatened Species IUCN. [Cited 23 Nov 2011.] Available from URL: http://www.iucnredlist.org Schloegel L. M., Ferreira C. M., James T. Y. et al. (2010) The North American bullfrog as a reservoir for the spread of Batrachochytrium dendrobatidis in Brazil. Anim. Conserv. 13, 53–61. Scott J. & Woodward B. D. (1994) Standard techniques for inventory and monitoring: surveys at breeding sites. In: Measuring and Monitoring Biological Diversity, Standard Methods for Amphibians (eds W. R. Heyer, M. A. Donnelly, R. W. McDiarmid, L. C. Hayek & M. S. Foster) pp. 92–6. Smithsonian Institution Press, Washington, DC. doi:10.1111/aec.12155 Shaffer H. B., Alford R. A., Woodward B. D. et al. (1994) Quantitative sampling of amphibian larvae. In: Measuring and Monitoring Biological Diversity, Standard Methods for Amphibians (eds W. R. Heyer, M. A. Donnelly, R.W. McDiarmid, L. A. C. Hayek & M. S. Foster) pp. 130–41. Smithsonian Institution Press, Washington, DC. Shipley B. (2000) Cause and Correlation in Biology:A User’s Guide To path Analysis, Structural Equations and Causal Inference. Cambridge University Press, Cambridge. Simberloff D. (2011) Charles Elton: neither founder nor siren, but prophet. In: Fifty Years of Invasion Ecology: The legacy of Charles Elton (ed. D. M. Richardson) pp. 11–24. WileyBlackwell, Oxford. Soulé M. E. (1990) The onslaught of alien species, and other challenges in the coming decades. Conserv. Biol. 4, 233–40. Stohlgren T. J., Jarnevich C., Chong G. W. & Evangelista P. H. (2006) Scale and plant invasions: a theory of biotic acceptance. Preslia 78, 405–26. Taylor B. W. & Irwin R. E. (2004) Linking economic activities to the distribution of exotic plants. Proc. Natl Acad. Sci. USA 101, 17725–30. Thomaz S. M. & Michelan T. S. (2011) Associations between a highly invasive species and native macrophytes differ across spatial scales. Biol. Invasions 13, 1881–91. Tilman D. (1997) Community invasibility, recruitment limitation, and grassland biodiversity. Ecology 78, 81–92. Von Holle B. & Simberloff D. (2005) Ecological resistance to biological invasion overwhelmed by propagule pressure. Ecology 86, 3212–18. Wang Y. & Li Y. (2009) Habitat selection by the introduced American bullfrog (Lithobates catesbeianus) on Daishan Island, China. J. Herpetol. 43, 205–11. Wellborn G. A., Skelly D. K. & Werner E. E. (1996) Mechanisms creating community structure across a freshwater habitat gradient. Annu. Rev. Ecol. Syst. 27, 337–63. Witte F., Goldschmidt T., Wanink J. et al. (1992) The destruction of an endemic species flock: quantitative data on the decline of the haplochromine cichlids of Lake Victoria. Environ. Biol. Fishes 34, 1–28. Zampella R. A. & Bunnell J. F. (2000) The distribution of anurans in two river systems of a coastal plain watershed. J. Herpetol. 34, 210–21. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher’s web-site: Appendix S1. Geographic coordinates of the 90 sampled waterbodies. © 2014 The Authors Austral Ecology © 2014 Ecological Society of Australia