Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Exotic Fish in Exotic Plantations: A Multi-Scale Approach to Understand Amphibian Occurrence in the Mediterranean Region

  • Joana Cruz ,

    terraetudo@gmail.com

    Affiliations Environment Department, University of York, Heslington, York, YO10 5DD, United Kingdom, CIBIO Research Centre in Biodiversity and Genetic Resources, InBIO, Universidade do Porto, Campus Agrário de Vairão, Rua Padre Armando Quintas, 7. 4485–661 Vairão, Vila do Conde, Portugal

  • Pedro Sarmento,

    Affiliation CESAM, Universidade de Aveiro, Campus Universitário de Santiago, 3810–193, Aveiro, Portugal

  • Miguel A. Carretero,

    Affiliation CIBIO Research Centre in Biodiversity and Genetic Resources, InBIO, Universidade do Porto, Campus Agrário de Vairão, Rua Padre Armando Quintas, 7. 4485–661 Vairão, Vila do Conde, Portugal

  • Piran C. L. White

    Affiliation Environment Department, University of York, Heslington, York, YO10 5DD, United Kingdom

Exotic Fish in Exotic Plantations: A Multi-Scale Approach to Understand Amphibian Occurrence in the Mediterranean Region

  • Joana Cruz, 
  • Pedro Sarmento, 
  • Miguel A. Carretero, 
  • Piran C. L. White
PLOS
x

Abstract

Globally, amphibian populations are threatened by a diverse range of factors including habitat destruction and alteration. Forestry practices have been linked with low diversity and abundance of amphibians. The effect of exotic Eucalyptus spp. plantations on amphibian communities has been studied in a number of biodiversity hotspots, but little is known of its impact in the Mediterranean region. Here, we identify the environmental factors influencing the presence of six species of amphibians (the Caudata Pleurodeles waltl, Salamandra salamandra, Lissotriton boscai, Triturus marmoratus and the anurans Pelobates cultripes and Hyla arborea/meridionalis) occupying 88 ponds. The study was conducted in a Mediterranean landscape dominated by eucalypt plantations alternated with traditional use (agricultural, montados and native forest) at three different scales: local (pond), intermediate (400 metres radius buffer) and broad (1000 metres radius buffer). Using the Akaike Information Criterion for small samples (AICc), we selected the top-ranked models for estimating the probability of occurrence of each species at each spatial scale separately and across all three spatial scales, using a combination of covariates from the different magnitudes. Models with a combination of covariates at the different spatial scales had a stronger support than those at individual scales. The presence of predatory fish in a pond had a strong effect on Caudata presence. Permanent ponds were selected by Hyla arborea/meridionalis over temporary ponds. Species occurrence was not increased by a higher density of streams, but the density of ponds impacted negatively on Lissotriton boscai. The proximity of ponds occupied by their conspecifics had a positive effect on the occurrence of Lissotriton boscai and Pleurodeles waltl. Eucalypt plantations had a negative effect on the occurrence of the newt Lissotriton boscai and anurans Hyla arborea/meridionalis, but had a positive effect on the presence of Salamandra salamandra, while no effect on any of the other species was detected. In conclusion, eucalypts had limited effects on the amphibian community at the intermediate and broad scales, but predatory fish had a major impact when considering all the scales combined. The over-riding importance of introduced fish as a negative impact suggests that forest managers should prevent new fish introductions and eradicate fish from already-occupied ponds whenever possible.

Introduction

Amphibians are one of the most threatened vertebrate groups, with nearly one third of the total number of species now at risk of extinction [1]. In the Mediterranean region, one of the global biodiversity hotspots [2], 29% of amphibian species are threatened with extinction, with habitat alteration and fragmentation cited as the primary reasons for past and future extinctions [1, 3]. One change in land use that has contributed to these problems in the Mediterranean has been the spread of forest plantations, which expanded by 5 million ha per year between 2000 and 2010 [4]. Forestry practices have been associated with low diversity and abundance of amphibians, due to land cover disturbance, alteration of microclimates, and exposure during terrestrial phase, mainly affecting forest specialist species [59]. Practices such as clear-cutting may lead to abrupt increase in surface temperature and loss of soil-litter moisture [10], decreasing survival and causing poor body condition [11], as well as altering migration behaviour [12] and connectivity. These processes may ultimately lead to species extirpation in the affected area [13, 14].

Exotic trees are used commonly in preference to native ones for forestry worldwide [15]. Eucalypt is one of the most commonly-planted trees in the world [4], and the negative impacts of forestry can expect to be exacerbated in exotic plantations [16]. Eucalypt has been associated with altered soil conditions, leading to lower pH in both soil [17] and water, disturbing the aquatic macroinvertebrate community, which is a key food source for amphibians [18], and also causing water depletion [19]. Changes in the soil and land characteristics may have a negative impact during aestivating and overwinter periods, especially on fossorial species. The effects of eucalypt plantations on amphibian communities have been documented in some biodiversity hotspots outside the eucalypt native range. In Madagascar [20], Brazil [21] and South Africa [16], where species richness is lower when compared to native forests; Costa Rica, where eucalypt plantations were a suitable habitat for the Eleutherodactylus coqui [22]; and USA, where the species richness was similar to native forests although differed in composition [23]. However, little equivalent research has been carried out in the Mediterranean basin hotspot [24], despite the widespread occurrence of eucalypt plantations in the region and the possible impact on habitat connectivity.

Connectivity is crucial for amphibians due to: (1) their distinct habitat requirements for feeding, breeding and overwintering; (2) an obligatory aquatic reproductive phase, for the Mediterranean species; (3) seasonal terrestrial adult migrations which make them susceptible to changes in landscape structure; (4) juvenile dispersal; and (5) their low vagilities and high risk of desiccation [3, 25]. Most estimates of migration and dispersal distance for the Mediterranean species (or related species) do not exceed 400 and 1000 metres, respectively [2629]. Nevertheless, there are accounts of individuals exceeding this distance [29]. Land cover, the network of ponds and streams (ephemeral, temporary and permanent), and the proximity of other ponds occupied by their conspecifics are covariates that can influence connectivity of amphibian populations [30, 31]. The proximity and high density of ponds and streams may provide a route for migration, facilitating the movements whilst maintaining moist conditions [25]. During migration or dispersal, the preference for occupied ponds by conspecifics is common [30]. However, migration and dispersal are species- and individual-specific and influenced by the ability to overcome predation, challenging microclimatic conditions and the resistance to movements of the substrates [3234]. For instance, clearcut areas may be more difficult to cross due to exposure to more extreme weather conditions and lack of refuge than habitats with vegetation cover [35]. Nevertheless, landscape processes are not the onlyones to affect amphibians’ population dynamics. At the local scale, pond characteristics also restrict occupation by certain species. The absence of exotic fish [3638], presence of temporary ponds [39] and the presence of aquatic vegetation [40, 41] may all favour a diverse amphibian community. Due to their different life stages and requirements (embryo, larvae, juvenile and adult), amphibians make use of resources at various scales, so a cross-scale analysis can be valuable to assess the presence of a certain species.

There has been previous research worldwide to investigate local- and landscape-scale variables influencing occurrence patterns in amphibians, e.g. [42, 43], but the results are highly variable [44, 45] and region- and context-specific [42, 46]. Given the significance of the Mediterranean region for native biodiversity [47], including amphibians, and the predominance of eucalypt forest cover in the region, there is an urgent need to evaluate the impact of these plantations on the amphibian community and assess local and landscape-scale covariates of species occurrence.

Here, we evaluate the impacts of different landscape and environmental factors on six species of amphibians (the Caudata Pleurodeles waltl, Salamandra salamandra, Lissotriton boscai and Triturus marmoratus and the anurans Pelobates cultripes and Hyla arborea/meridionalis). We evaluated the factors affecting pond occupancy by these species at three different scales, appropriate to the scale of individual ponds (local), migration distances (intermediate; 400 m) and dispersal distances (broad; 1000 m). Using different scales is useful as different variables may only become significant at a specific scale, improving the quality of the models [42, 46, 48, 49].

At local scale, we tested the hypothesis that all amphibian species would survive in ponds without fish, with a temporary hydroperiod and with high aquatic vegetation cover. At the intermediate and broad scales, we hypothesised that amphibian species occurrence in a pond would increase with the density of streams and ponds, and decline with increasing eucalypt cover and distance to streams. Finally, we hypothesised that amphibian species occurrence would be explained better by a combination of covariates across each scale than by covariates at any one scale.

Materials and Methods

Study Area

More than one-third of mainland Portugal is covered by forest (35%). Within this forested area, eucalypt (Eucalyptus spp.) is the dominant tree (26%), and both Maritime pine (Pinus pinaster) and native cork oak (Quercus suber) occupy 23% each [50]. We carried out the study in central-east Portugal, Castelo Branco district (39°40’– 40°10’N, 7°0’– 7°35’W). The area has a Mediterranean climate, with a mean temperature of 16.7°C (mean minimum: 11.0°C; mean maximum: 22.4°C) and an average precipitation of 758 mm [51].

In the study area, the forest land cover is dominated by eucalypt (Eucalyptus globulus) plantations (36%), with different age stands, natural forest of cork oak (Quercus suber) and Holm oak (Quercus ilex), Maritime pine plantations, scrubland areas dominated by Cytisus spp., Cistus spp. and Erica spp. (all comprising 23%), and montados (oak savannah-like woodland) (16%). In addition to forestry, the landscape is used patchily for livestock grazing, olive (Olea europaea) groves, wheat (Triticum spp.) production, and small-scale subsistence agriculture (24%) (Fig 1). Most of the eucalypt stands are on their third rotation, planted for the first time in the mid-1970s. Each rotation lasts between 12 to 16 years depending on site productivity and plantations are managed by coppicing. The montados in the study area are actively exploited, with cattle grazing and cork extraction.

thumbnail
Fig 1. The location of the study area.

Distribution of the 88 ponds surveyed monthly between February and June 2011 in central-east Portugal. There were three major study sites distributed in the region: Site A, with 67 ponds; Site B: 2 ponds; Site C: 19 ponds. Each site was divided into subsets (FC, SM, SRIN, CAT, CF. FR, MG, GAL), according to geographical, topographical or barrier features.

http://dx.doi.org/10.1371/journal.pone.0129891.g001

Field sampling

We selected the ponds in order to represent roughly the proportion of habitats in the landscape. We also wanted to have spatial replicates of the ponds in the different habitats, that is why we chose ponds in eucalypt plantations that geographically were similar and near agriculture land and montados. We tried to sample all the ponds that were in a particular area as long as we had permission from the Estate managers. We recorded the presence/absence of six species—the urodeles Iberian ribbed newt (Pleurodeles waltl) (PW), fire salamander (Salamandra salamandra) (SS), Bosca's newt (Lissotriton boscai) (LB) and marbled newt (Triturus marmoratus) (TM), and the anurans Western spadefoot (Pelobates cultripes) (PC) and tree frog (Hyla arborea/meridionalis) (HY). Sampling was carried out monthly from February to June 2011, in 88 ponds, distributed in three major areas (sites A, B and C; Fig 1). Considering the different detectability of the species studied and their life stages, we used a range of techniques to increase the detectability and to record their presence/absence: dipnetting, visual surveys and acoustic night surveys. [52] refer that to achieve 95% of detection probability for all amphibian species in the Mediterranean region, a minimum of three visits is required, spread out during the breeding season and including a combination of nighttime call count, nighttime visual encounter and daytime netting. We surveyed the ponds five times, and applied all the methods each time we did so. If one species was detected by one single method once, it was considered to be present in that pond. During each visit, surveys were conducted by two independent observers. Each observer began their surveys at opposite sides of the water body and walked around the perimeter of the pond in the opposite direction, separately recording detections of all life stages of encountered amphibian species. Sampling effort was proportional to the pond size. Dipnetting was complemented with visual surveys in and around each water point to detect eggs, larvae, juveniles and adults. Amphibians were identified to the species level, whenever possible, using identification keys [53, 54]. Tree frog Hyla arborea/meridionalis tadpoles were identified to genus, because they could not be reliably identified in the field [55].

The visual surveys took place during day and night-time, the latter with the aid of torchlight (Streamlight, model Fire Vulcan Led). For the night survey, we had a 1-minute pause after arrival and then conducted a 3-minute aural survey, in which we identified each species call, before the visual survey started [26].

Amphibian data were collected following all legal requirements and we had the needed permits by Instituto da Conservação da Natureza e da Biodiversidade (ICNB) (the national authority for nature conservation and wildlife protection) to sample protected species. The ICNB permit was issued under the EU Habitats Directive and considered safety measures to avoid spread of any pathology, namely the disinfection of equipment with 1% hypochlorite solution. Considering this safety measure and the fact that no animals were sacrificed there was no need for approval by any animal ethics committee. The sampling within private land was performed with the authorization of the land owners.

The authors confirm that all data underlying the findings are fully available without restriction. All data files are available from the corresponding author’s Research Gate website upon request.

Local scale

At each pond, we recorded the presence of predatory fish (FISH), the hydroperiod (HYDRO), soil type (muddy or shale) (SOIL) and the percentage of aquatic vegetation [floating (FLOAT), emergent (EMER) and submerged (SUBMER)]. The presence of predatory fish was assessed while doing the surveys by visual observation and interviewing the estate managers. We recorded the presence of pumpkinseed sunfish (Lepomis gibbosus), eastern mosquito fish (Gambusia holbrooki) and largemouth bass (Micropterus salmoides); these species are non-native to the region and classified as invasive under national law. We divided the hydroperiod into two levels: temporary (retains water between 3 to 6 months) and permanent (maintains water all year around), based on previous knowledge by the surveyors and, when necessary, confirmed by the estate managers. The soil type was assessed by visual observation, and it was classified as muddy or shale depending if it was an earthy or rock substrate, respectively.

The aquatic vegetation estimated visually by the same surveyor and classified into three classes: floating—plants rooted or free that float on the water surface, like Ranunculus spp.; emergent—rooted plants that grow above the water surface, such as Typha spp.; submerged—rooted plants that grow up to the water surface but not above it.

Intermediate scale

The variables at this scale were measured within a 400 m radius buffer of each pond, using ESRI ArcGIS 10 and the land cover map of Portugal, from 2007 (COS2007) [56] and the Ordnance Survey maps.

The proportion of each land cover level was classified according to the main classes [agriculture (AGRIC), eucalypt plantations (EUC), montados (MONT) and native forest (NATFOR)]. We applied the data exploration guidelines described by [57] to the datasets for each of the six individual species to assess collinearity between the predictors and possible interactions. We carried out interaction plots and according to their results (non-consistent effect across all levels) we chose those interactions. Hence, we added to the models the interaction AGRIC400:MONT400 to the models. The distance to the nearest streams was estimated for both ephemeral streams (hydrologically dependent on rainfall, and are dry for most of the year, retaining water for less than 4 weeks after the last rainfall event) (NEPH) and temporary streams (retaining water for more than 6 months in a year) (NTEMP). Distance to ponds (NPOND) was also estimated. Following initial data exploration, we added the interaction term NEPH:NPOND. The density of ephemeral and temporary streams (DEPH and DTEMP) and ponds (DPOND) was also calculated. For each amphibian species we also measured the distance to the nearest other pond occupied by their conspecifics during the current survey (NPW, NSS, NLB, NTM, NPC, NHY).

Broad scale

To investigate habitat associations at a broad scale, we assessed the same variables as in the intermediate scale—land cover, the distance to the nearest ephemeral and temporary streams, distance to ponds and density of ephemeral and temporary streams and ponds, and the distance to the nearest other pond occupied by their conspecifics—but we applied a 1000 m buffer around each pond. Following data exploration and analysis of the interaction plots, we added the interaction AGRIC1000:NATFOR1000.

Model building and model selection

As mentioned before, we followed the data exploration guidelines of [57]. To assess collinearity, we used the Spearman rank correlation coefficient |r| because it makes no assumption about function of the relationships between two variables [58]; |r| > 0.6 was chosen to indicate high collinearity between variables, and where this was found, the variables were not used together in the same model. This value was chosen as a compromise, since the threshold for high collinearity is defined by some authors as |r| > 0.5 [58], whereas other authors propose a value of |r| > 0.7 [59].

In order to determine whether there was a difference in the studied response variables between eucalypt plantation and other land covers (agriculture, montados and native forest), we applied a chi-square test for multiple independent samples followed by post-hoc pairwise comparisons using False Discovery Rate method (R Package “fifer”) [60]. To assess which predictors better explained the behaviour of the response variables, we used a generalised linear mixed model (GLMM) fit to Laplace approximation, with a binomial error distribution (to model species occurrence) with the local subsets as the random variables, using package glmmADMB [61]. The subsets were defined according to geographical, topographical or barrier features (e.g. roads) which created eight local subsets (Fig 1), to account for spatial intercorrelation. The steps taken to determine the best top-ranked models are explained in Fig 2. Firstly, we combined all covariates, avoiding multicollinearity, in (1) one single full-model at a local scale, (2) 10 models at an intermediate scale, and (3) 10 models at a broad scale (Table A in S1 File). At the local scale the variables were not collinear unlike at the intermediate and broad scale and that is the reason for the different number of models. There was collinearity between agriculture cover and eucalypt cover at both scales (intermediate rS = -0.79, p < 0.001; broad rS = -0.79, p < 0.001), and between eucalypt cover and montados (intermediate rs = -0.63, p < 0.001; broader scale rS = -0.64, p < 0.001). Secondly, we used data dredge statistics (dredge—MuMIn R package) [62] to run GLMM on those models. Dredge statistics are a valid method, used when multiple variables may contribute to a response behaviour and it is important not to exclude any combinations of these variables. All the variables that we used have been described as significant determinants of amphibian occurrence in the literature. We did not want to omit any of these potentially important variables, especially since the effect of eucalypt stands on amphibians is poorly studied in the Mediterranean area. In addition, when considering a guild of species, it is difficult to choose an assemblage of variables that is relevant for all the species. In such cases, dredging is a recommended procedure. Recent papers covering genetics [63, 64], carnivore monitoring [65], amphibian studies [66] and bird movements and fragmentation [67] have all used dredge statistics.

thumbnail
Fig 2. Flowchart describing the statistical analysis applied to the data.

http://dx.doi.org/10.1371/journal.pone.0129891.g002

Thirdly, we approximated model parsimony, given our data and model set by species, using Akaike Information Criterion for small samples (AICc) [68] and weighted the support of each model using AICc weights [69]. We used model averaging to determine the direction and magnitude of the effect of each predictor variable and we report significantly positive or negative effects when 95% confidence intervals of parameter estimates did not contain zero [6972]. The variables chosen to build the combined scale models were selected using the results from the previous steps, i.e., variables from the models which Akaike weights was more or equal to 0.9 and which confidence intervals did not overlap zero, in case of multiple models, of each scale (local, intermediate and broad). The number of full models varied according to the species (Table A), and model selection was applied using their ΔAICc and Akaike weights. Finally, we considered all together the top ranked models from the four scales and ranked them according to their ΔAICc and Akaike weights and averaged the models coefficients which cumulative Akaike weights was ≥0.9. The relative importance of each predictor variable [69] was then calculated based on AICc weights (‘importance’ function in ‘MuMIn’).

In order to evaluate the effect size of each covariate, we calculated the odds ratio (OR) using the model average parameter estimates for each response variable [73]. Odds ratios >1 indicate a positive effect; ratios <1 indicate a negative effect. Only variables for which the confidence interval of the coefficient did not overlap zero and cumulatively had an OR≥1.1 or OR≤0.90 were plotted.

Results

All of the studied species were detected at least once in each dominant land cover (Table 1). The most common species were Hyla arborea/meridionalis, L. boscai, P. waltl and T. marmoratus, all of which were present in more than 60% of the surveyed ponds.

Pleurodeles waltl

P. waltl presence was significantly different between eucalypt and native forest and agriculture and native forests at the intermediate scale (χ2 = 10.28, df = 3, P<0.05) and the broad scale ((χ2 = 32.72, df = 3, P<0.001) (Table 1;Table B in S1 File). At broad scale only, P. waltl presence was significantly different between agriculture and montados, and between eucalypt stands and montados (Table B in S1 File) P. waltl presence was negatively associated strongly with the presence of predatory fish (OR = 0.12, IC95% [0.03–0.55]) and distance to the nearest other pond occupied by conspecifics (OR = 0.36, IC95% [0.16–0.82]) (Table 2 and Fig 3). The model selection results provided strong support for a positive relationship between the probability of presence of P. waltl and the proportion of agricultural land at the broad scale (OR = 1393.29, IC95% [16.59–1.17e05]). The remaining variables whose confidence intervals did not overlap zero—distance to the nearest ephemeral stream, distance to the nearest pond, and their interaction—did not influence substantially the occurrence of P. waltl (OR≈1) (Table 2).

thumbnail
Table 2. Model averaged parameter estimates (β) (top-ranked models) for each of the Caudata, odds ratio (OR) and respective 95% confidence intervals (IC95%).

http://dx.doi.org/10.1371/journal.pone.0129891.t002

thumbnail
Fig 3. Fitted values predicted by the top ranked model for the Caudata for each of the response variables with strong effect according to the odds ratio results.

The dashed line are the confidence intervals at 95%.

http://dx.doi.org/10.1371/journal.pone.0129891.g003

The two top ranked models combined local and broad scale variables and accounted for 0.95 of the model’s Akaike weights (Table C in S1 File). Models at the different scales considered individually had little support (ΔAICc>6: ΔAICc Broad scale<ΔAICc Local scale<<ΔAICc Intermediate scale) (Table C in S1 File). The intermediate scale models had the weakest support, with the confidence intervals of the covariates density of ephemeral streams, density of temporary streams and proportion of agricultural land all overlapping zero (Table C in S1 File).

Salamandra salamandra

S. salamandra presence was significantly different between eucalypt stands and agriculture, between eucalypt stands and montados, and between eucalypt stands and native forest at the intermediate (χ2 = 42.99, df = 3, P<0.001) and broad scales (χ2 = 50.15, df = 3, P<0.001) (Table 1; Table B in S1 File). At the broad scale only, there was also significant difference regarding S. salamandra occurrence between agriculture and native forests (Table B in S1 File). The model averaged parameter estimates provided strong support for a negative influence of fish (OR = 0.14, IC95% [0.03–0.68]) and a positive influence of the proportion of eucalypt on S. salamandra presence—intermediate (OR = 18.85, IC95% [3.39–104.65]) and broad scale (OR = 37.56, IC95% [4.66–302.76]) (Table 2 and Fig 3). The distance to ephemeral streams did not have a strong impact on S. salamandra occurrence (OR = 0.97, IC95% [0.94–0.99]) (Table 2 and Fig 3). The three top-ranked models accounted for 0.94 of the model’s Akaike weights and incorporated local, intermediate and broad scale covariates together (Table C in S1 File). When considered separately, models at each scale had little support, with ΔAICc>5.5 (ΔAICc Intermediate scale ≈ ΔAICc Broad scale << ΔAICc Local scale) (Table C in S1 File). Some of the covariates measured across the three scales had little support, overlapping zero in their confidence intervals (proportion of native forest at broad scale, distance to pond and to temporary streams, at both temporary and intermediate scale, and hydroperiod and submerged vegetation, at the local scale) (Table C in S1 File).

Lissotriton boscai

L. boscai presence was significantly different between agriculture and native forest, eucalypt stands and montados, and eucalypt stands and native forest and montados and native forest, at intermediate (χ2 = 23.63, df = 3, P<0.001) and broad scales (χ2 = 32.47, df = 3, P<0.001); and between agriculture and eucalypt stands, at the intermediate scale alone (Table 1; Table B in S1 File). A single top-ranked model with a combination of covariates from the local and broad scales best explained L. boscai presence, accounting for 0.99 of the model’s weight selection (Table C in S1 File). The single scale models had a ΔAICc>10 (ΔAICc Intermediate scale ≈ ΔAICc Broad scale << ΔAICc Local scale). There was strong evidence of a negative relationship between occurrence of L. boscai and the presence of fish (OR = 0.18, IC95% [0.05–0.70]), distance to the nearest other pond occupied by their conspecifics (OR = 0.39, IC95% [0.17–0.89]), density of ponds at the broad scale (OR = 0.76, IC95% [0.61–0.96]) and proportion of eucalypt at the broad scale (OR = 0.06, IC95% [0.005–0.75]) (Table 2 and Fig 3). The proportion of submerged aquatic vegetation had a positive and strong influence on L. boscai presence (OR = 29.10, IC95% [3.78–223.90]) (Table 2 and Fig 3). Distance to temporary streams did not affect strongly the occurrence of L. boscai (OR = 1.00, IC95% [1.00–1.00]).

Triturus marmoratus

T. marmoratus presence was significantly different between agriculture and montados, agriculture and native forest, eucalypt stands and montados, and eucalypt stands and native forests, at the intermediate (χ2 = 25.61, df = 3, P<0.001) and broad scales (χ2 = 23.51, df = 3, P<0.001) (Table 1; Table B in S1 File). Of the four top-ranked models, three included a combination of covariates at the different scales, and one included covariates exclusively from the local scale, accounting for 0.92 of the Akaike weights. There was little support for models at the intermediate and broad scale, with ΔAICc>9.9 (ΔAICc Local scale < ΔAICc Broad scale ≈ ΔAICc Intermediate scale). The presence of predatory fish was the only variable whose confidence intervals did not overlap zero and had a negative effect on T. marmoratus occurrence (OR = 0.15, IC95% [0.05–0.47]) (Table 2 and Fig 3).

Pelobates cultripes

P. cultripes presence was significantly different between agriculture and native forest, at both the intermediate (χ2 = 11.51, df = 3, P<0.05) and the broad scale (χ2 = 29.11, P<0.001) (Table 1; Table B in S1 File). At the broad scale, P. cultripes occurrence was also significantly different between agriculture and eucalypt stands, agriculture and montados and eucalypt and native forest (Table B in S1 File). A combination of local and broad-scale spatial covariates were included in the models that supported P. cultripes presence, namely the proportion of submerged aquatic vegetation, which had a negative effect on the presence of this anuran (OR = 0.06, IC95% [7.85e-03-0.47]), and agricultural cover (positive effect) (OR = 182.12, IC95% [3.25–1.02e04]) and native forest cover (negative effect) (OR = 0.001, IC95% [3.94e-06-0.40]) at broad scale (Table 3 and Fig 4). The remaining scale models had a ΔAICc>2 (ΔAICc Local scale ≈ ΔAICc Broad scale < ΔAICc Intermediate scale).

thumbnail
Table 3. Model averaged parameter estimates (β)(top-ranked models) for each of the anurans, odds ratio (OR) and respective 95% confidence intervals (IC95%).

http://dx.doi.org/10.1371/journal.pone.0129891.t003

thumbnail
Fig 4. Fitted values predicted by the top ranked model for the anurans for each of the response variables with strong effect according to the odds ratio results.

The dashed line are the confidence intervals at 95%.

http://dx.doi.org/10.1371/journal.pone.0129891.g004

Hyla arborea/meridionalis

H. arborea/meridionalis presence was significantly different between agriculture and native forest, and between eucalypt stands and native forest, at the intermediate (χ2 = 18.96, df = 3, P<0.001) and broad scales (χ2 = 34.02, df = 3, P<0.001). At the intermediate scale only, there were differences in presence between montados and native forest, and at the broad scale only, there were differences between agriculture and montados and eucalypt and montados (Table 1; Table B in S1 File).

A high number of models (21) accounted for 0.9 of Akaike weights, comprising models of combined scales and models with just local covariates (Table C in S1 File). There was a strong negative influence of temporary ponds (OR = 0.10, IC95% [0.02–0.53]) as well as of eucalypt cover at the broad scale (OR = 0.10, IC95% [0.001–0.93]) and a positive effect of submerged vegetation (OR = 9.57, IC95% [1.09–83.90]) (Table 3 and Fig 4).

Discussion

In summary, and considering the effect size of the variables, at the local scale, the presence of predatory fish had a consistently strong negative effect on the presence of Caudata. At this local scale, submerged vegetation had opposing influences on L. boscai and H. arborea/meridionalis (positive) and P. cultripes (negative) and Hyla arborea/meridionalis avoided temporary ponds. At the intermediate scale, distance to the nearest occupied pond by conspecifics had a strong negative effect on the presence of both P. waltl and L. boscai. Eucalypt cover had an intense positive effect on S. salamandra occurrence at both intermediate and broad scales, but had the opposite effect on L. boscai occurrence. L. boscai avoided also areas with high density of ponds, at the broad scale. At this scale also, the proportion of agriculture cover resulted in a strong positive influence on P. waltl and P. cultripes occurrences. Finally, still at the broad scale, P. cultripes strongly avoided native forests while H. arborea/meridionalis avoided eucalypt stands.

Some of our initial hypotheses were supported by our results. At a local scale, predatory fish presence was the single variable present in all top-ranked occurrence models for Caudata, having a negative influence on species occurrence. In addition, models with covariates across the three spatial scales had stronger support than models taken individually. However, some of our results were contrary to expectations. At the local scale, H. arborea/meridionalis preferred permanent ponds over temporary ponds and P. cultripes was not favoured by the increase of submerged vegetation. At the intermediate and broad scales, density of streams and/or ponds did not increase species occurrence and the decrease of distance of the connectivity variables did not show a consistent positive relationship with species occurrence. At the broad scale, eucalypt plantations had a negative effect on L. boscai and H. arborea/meridionalis occurrence, and had no negative effect on the other species, even favouring the presence of S. salamandra.

The negative impact of introduced predatory fish on amphibians has been reported in several studies worldwide [7479] and in the Mediterranean region [36, 8082]. These impacts occur through direct predation, competition or pathogen transfer [81, 8387]. Certain amphibians may be more susceptible to these threats because they are not usually exposed to predatory fish, either because they are associated with temporary ponds, where fish are absent, and therefore lack appropriate defences [88] or because they do not recognise these fish as threats since they are all non-native species and they had little or no evolutionary history with these predators [87]. Nevertheless, there are amphibians that co-exist with alien fish, and have defence mechanisms either because they may have developed them during the course of evolution and conserve them in the absence of predators or because those mechanisms work against both native and non-native predators [87]. In this situation, tadpoles may show changes in morphological traits (e.g. increased tail area) and in behaviour (lower activity rate, aggregation, higher use of complex aquatic vegetation for refuge) to adapt to the novel situation besides tadpole unpalatability and/or chemically mediated predator avoidance [8992]. Presence or absence of fish was irrelevant for the occurrence models for anurans, although other studies have detected a negative relationship (P. cultripes: [55]; H. arborea: [78, 84]). It has been reported in previous studies that these species are resilient against predatory fish due to their morphological traits or plasticity. Specifically, P. cultripes larvae attain a large body size (on average around 80 mm, [93]) and H. arborea larvae are able to develop deeper tail fins and deeper tail muscles in presence of fish [89]. On the other hand, the swimming behaviour of both these larvae, which are nektonic, may increase the chance of being preyed upon by making them more visible to visually oriented predatory fish like L. gibbosus [84, 9496].

Avoidance of temporary ponds by H. arborea/meridionalis has also been reported previously, e.g. [97]. Both species have a long larval stage, on average 3 months [98], and temporary ponds can dry out before metamorphosis is complete [99]. Although temporary ponds cannot support predatory fish, the desiccation risk in the Mediterranean region is high, so the preference for ponds with a long and stable hydroperiod may still improve the recruitment success of these amphibian species [39].

Aquatic vegetation can provide refuge, food [94] and protection against UV-B radiation, which can affect some species during early developmental stages [100]. Three of the studied species occurrences were affected differently by aquatic vegetation. L. boscai and H. arborea/meridionalis were found more in ponds with a high percentage of vegetation, which is likely to be related to the oviposition behaviour shown by newts of wrapping each egg individually in leaves to protect them from UV-B radiation and predators [40]. In contrast, the anuran P. cultripes avoided ponds with a high percentage of submerged vegetation. P. cultripes is a good swimmer, feeds within the water column, and it is possible that too much aquatic vegetation interferes with its foraging.

Connectivity covariates—distance to streams or ponds and density of streams and ponds—showed different trends amongst the studied species, and only P. cultripes and T. marmoratus occurrence were not affected by these covariates. The increased probability of occurrence of a certain species in a pond was often related to a decrease in the distance to the nearest other pond occupied by that same species (P. waltl and L. boscai). Most juvenile Caudata do not disperse more than 500 metres from the breeding pond and adults show a high level of site fidelity to the pond they first reproduced in [31]. In fact, juveniles are not well adapted to dispersal. They are smaller, more prone to desiccation and have less locomotor capacity than adults to travel long distances, and they sustain high mortality rates when they leave the pond [25]. Due to all these constraints it is most likely that juvenile Caudata disperse to nearby ponds, depending on close “networks” of ponds where the species is already present [101]. Chemical cues, both aquatic and terrestrial, may lead these juveniles to non-natal ponds [30, 31]. In a laboratory setting, L. boscai preferred water that contained chemical cues of themselves or conspecifics [30]. Heterospecific auditory cues may also attract some species and help with pond orientation. T. marmoratus showed positive phonotactic orientation when exposed to Bufo calamita advertisement calls [102], and Lissotriton helveticus showed the same behaviour when exposed to P. perezi calls [103]. Triturus alpestris is capable of long distance homing using only magnetic compass [104]. However, a high density of ponds had a negative impact on L. boscai occurrence. Our results are partially coincident with those by [105], with species being more abundant at intermediate pond density. Both local (aquatic; within-pond) and landscape (terrestrial) features are expected to influence species occurrence. The contribution of each feature may depend on their spatial configuration and quality [106]. Water is a scarce resource in Mediterranean regions, so aquatic habitats are expected to act as a constraint to population occurrence and dynamics, with the distribution and characteristics of terrestrial habitats only having a major role when ponds are plentiful [105].

S. salamandra is often associated with temporary streams for breeding, avoiding ephemeral streams [107], contrary to our results. This forest species has a long terrestrial phase, using ponds or streams only to deposit their larvae, spending the rest of their time on land. Our results showed also the positive effect of eucalypt plantation at both intermediate and broad scale on the occurrence of this species. Eucalypt plantations place large demands on soil water. In some cases water depletion caused by eucalypts can reach 8 metres depth [19], leading to low levels of moisture at the surface. Eucalypt plantations in the Mediterranean region are also characterised by a lower macro-arthropod abundance when compared to native habitats such as cork oak woodlands, olive groves or riparian vegetation [108]. Despite this, eucalypt plantations apparently favoured S. salamandra occurrence, a species that has a strong association with high woodland cover [109], and these stands are the only forest-cover type that cover extensive areas in the region. We tentatively interpret this to be a consequence of the proximity of ephemeral streams which may supply sufficient humidity to reduce the risk of desiccation and also serve as a source of prey, especially if the original riparian vegetation is maintained [108], which was verified in most of our study area.

Nevertheless, the conditions that favoured S. salamandra had the opposite effect on L. boscai. This species, although being one of the most aquatic European newts, makes terrestrial incursions throughout the year and goes into summer dormancy in hot and dry regions [110]. L. boscai has a low ecological plasticity [111] and low dispersal ability [40]. Thus, any additional barrier, like chemical fertilisers, soil disturbances and low soil depth can add costs to the distance travelled [33, 112]. In addition to the impact of eucalypt plantations in causing soil water depletion, the smaller size of this newt compared with S. salamandra, may make it more susceptible to predation and desiccation when crossing extensive areas of exotic stands.

Agricultural land is often associated negatively with amphibian presence due to multiple interventions throughout the year, altering the soil humidity and jeopardizing refuges during aestivation, as well as potentially causing direct mortality due to injuries [113]. The use of fertilisers may also affect the body condition of amphibians and their ability to disperse depending upon concentrations, time of the year and species sensitivity [114, 115]. In addition, cattle grazing may have a negative impact on water quality through nitrogenous deposits, increasing eutrophication, degrading water quality but also by grazing on the shoreline vegetation, that acts as refuge and source of food and oviposition sites [116, 117]. However, the agricultural use in the study area is extensive, and comprises olive groves, wheat areas, and small-scale farming for personal use, with low use of fertilisers and grazing at low densities, mainly by cattle. Hence, the practice of small-scale agriculture, which represents an anthropogenic disturbance of only intermediate impact, favoured the occurrence of P. waltl and P. cultripes, especially at the broad scale. Nonetheless, that was not the case for native forests, which combined pine, oak and mixed forests, and P. cultripes avoided these areas. Adult P. cultripes need soft soils to dig their burrows and they might have some difficulties digging in forested areas when compared with agricultural land.

Overall, our results showed that eucalypt stands had a negative impact on the occurrence of L. boscai and H. arborea/meridionalis, whilst having a positive effect on S. salamandra occurrence. For the remaining species, eucalypt cover was an unimportant factor compared with other landscape and local variables, especially the presence of predatory fish, which had a strong negative impact on the occurrence of Caudata The eucalypt stands of the study area were embedded in a traditional agro-forestry matrix, with intermediate disturbance, and a conservative approach must be taken to extrapolate these results to larger extensions of eucalypt plantations surrounded by a degraded matrix, with a high level of disturbance (e.g. intensive agriculture, barriers such as roads). Therefore, sensitive management of these plantations in terms of conservation is advised. As further research, we suggest the evaluation of functional connectivity. This was not possible in our study because there are only a few studies that relate costs of travelling with habitat structure, and to our knowledge, for some species, there is an absolute lack of information, like for P. cultripes or L. boscai. In conclusion, eucalypts had limited effects on the amphibian community at the migration and dispersal scales, but fish presence had a major impact at all scales. Our results highlight the importance of context-dependency in predicting impacts of landscape composition and structure on amphibian populations. However, the over-riding importance of fish as a negative impact suggests that forest managers should prevent new introductions fish and other exotic predators and eradicate fish from already-occupied ponds whenever possible. When fish eradication is not possible, creation of new permanent fish-free ponds nearby fish-occupied ponds may be an alternative strategy.

Supporting Information

S1 File. Full models built for the six species studied.

For the local, intermediate and broad scales the models were the same for each of the six species. For the combined scale they differ as explained by the procedure in Fig 2 (Table A). Chi-square results, between pairs of land cover and species occurrence at the intermediate and broad scale Significant difference obtained if p<0.05 (in bold) (Table B). Model selection results for analysis of the species occurrence. Models which AICc weight sums up to 0.9 or more are shown, as well as the two highest ranked models at each spatial scale and the reference if they were used in the model average (✓). For each response variable is presented the model description, the number of estimable parameters (K), the sample-size adjusted AIC (AICc), Akaike differences (ΔAICc), Akaike weights and the log-likelihood (logLik). In bold are the covariates which confidence intervals do not overlap zero. All models have the covariate subset added as a random variable (Table C).

doi:10.1371/journal.pone.0129891.s001

(DOCX)

Author Contributions

Conceived and designed the experiments: JC PS MAC PCLW. Performed the experiments: JC PS. Analyzed the data: JC. Contributed reagents/materials/analysis tools: JC PS MAC PCLW. Wrote the paper: JC PS MAC PCLW.

References

  1. 1. Vié J-C, Hilton-Taylor C, Stuart SN, editors. Wildlife in a changing world—an analysis of the 2008 IUCN Red List of Threatened Species. Gland, Switzerland: IUCN; 2009.
  2. 2. Myers N, Mittermeier RA, Mittermeier CG, da Fonseca GAB, Kent J. Biodiversity hotspots for conservation priorities. Nature. 2000;403:853–8. 0450. pmid:10706275
  3. 3. Dodd CK Jr, editor. Amphibian ecology and conservation: a handbook of techniques. New York: Oxford University Press; 2010.
  4. 4. FAO. Global forest resources assessment 2010—Main report. Rome: Food and Agriculture Organization of the United Nations, 2010. doi: 10.1073/pnas.0910467107. pmid:20921413
  5. 5. Popescu VD, Patrick DA, Hunter ML Jr, Calhoun AJK. The role of forest harvesting and subsequent vegetative regrowth in determining patterns of amphibian habitat use. Forest Ecology and Management. 2012;270:163–74. doi: 10.1016/j.foreco.2012.01.027 4777.
  6. 6. Adum GB, Eichhorn MP, Oduro W, Ofori-Boateng C, Rodel M- O. Two-stage recovery of amphibian assemblages following selective logging of tropical forests. Conservation Biology. 2012;27(2):354–63. doi: 10.1111/cobi.12006 5152.
  7. 7. Popescu VD, Hunter ML. Clear-cutting affects habitat connectivity for a forest amphibian by decreasing permeability to juvenile movements. Ecological Applications. 2011;21(4):1283–95. doi: 10.1890/10-0658.1. pmid:21774430
  8. 8. Freidenfelds NA, Purrenhage JL, Babbitt KJ. The effects of clearcuts and forest buffer size on post-breeding emigration of adult wood frogs (Lithobates sylvaticus). Forest Ecology and Management. 2011;261(11):2115–22. doi: 10.1016/j.foreco.2011.03.005 3923.
  9. 9. Karraker NE, Welsh HH Jr. Long-term impacts of even-aged timber management on abundance and body condition of terrestrial amphibians in Northwestern California. Biological Conservation. 2006;131(1):132–40. 0997.
  10. 10. Semlitsch RD, Todd BD, Blomquist SM, Calhoun AJK, Gibbons JW, Gibbs JP, et al. Effects of timber harvest on amphibian populations: Understanding mechanisms from forest experiments. BioScience. 2009;59(10):853–62. 1103.
  11. 11. Todd BD, Rothermel BB. Assessing quality of clearcut habitats for amphibians: Effects on abundances versus vital rates in the southern toad (Bufo terrestris). Biological Conservation. 2006;133(2):178–85. 0989.
  12. 12. Todd BD, Luhring TM, Rothermel BB, Gibbons JW. Effects of forest removal on amphibian migrations: implications for habitat and landscape connectivity. Journal of Applied Ecology. 2009;46(3):554–61. doi: 10.1111/j.1365-2664.2009.01645.x 3338.
  13. 13. Petranka JW, Brannon MP, Hopey ME, Smith CK. Effects of timber harvesting on low elevation populations of southern Appalachian salamanders. Forest Ecology and Management. 1994;67(1–3):135–47. 3325.
  14. 14. Dupuis LA, Smith JNM, Bunnell F. Relation of terrestrial-breeding amphibian abundance to tree-stand age. Conservation Biology. 1995;9(3):645–53. 3682.
  15. 15. Evans J, editor. Planted forests: Uses, impacts and sustainability Preston, UK: Food and Agriculture Organisation of the United Nations and CABI; 2009.
  16. 16. Russell C, Downs CT. Effect of land use on anuran species composition in north-eastern KwaZulu-Natal, South Africa. Applied Geography. 2012;35(1–2):247–56. doi: 10.1016/j.apgeog.2012.07.003. pmid:22736875
  17. 17. Jobbágy E, Jackson R. Patterns and mechanisms of soil acidification in the conversion of grasslands to forests. Biogeochemistry. 2003;64(2):205–29. doi: 10.1023/a:1024985629259 5775.
  18. 18. Canhoto C, Laranjeira C. Leachates of Eucalyptus globulus in intermittent streams affect water parameters and invertebrates. International Review of Hydrobiology. 2007;92(2):173–82. doi: 10.1002/iroh.200510956 4765.
  19. 19. Mendham DS, White DA, Battaglia M, McGrath JF, Short TM, Ogden GN, et al. Soil water depletion and replenishment during first- and early second-rotation Eucalyptus globulus plantations with deep soil profiles. Agricultural and Forest Meteorology. 2011;151(12):1568–79. doi: 10.1016/j.agrformet.2011.06.014 4240.
  20. 20. Vallan D. Effects of anthropogenic environmental changes on amphibian diversity in the rain forests of eastern Madagascar. Journal of Tropical Ecology. 2002;18(05):725–42. doi: 10.1017/S026646740200247X 0089.
  21. 21. Gardner TA, Ribeiro-Júnior MA, Barlow J, Ávila-Pires TCS, Hoogmoed MS, Peres CA. The value of primary, secondary, and plantation forests for a neotropical herpetofauna. Conservation Biology. 2007;21(3):775–87. 0090. pmid:17531055
  22. 22. Fogarty JH, Vilella FJ. Use of native forest and Eucalyptus plantations by Eleutherodactylus frogs. Journal of Wildlife Management. 2003;67(1):186–95. 4536.
  23. 23. Sax DF. Equal diversity in disparate species assemblages: a comparison of native and exotic woodlands in California. Global Ecology and Biogeography. 2002;11(1):49–57. doi: 10.1046/j.1466-822X.2001.00262.x 3308.
  24. 24. Goiti U, Garin I, Almenar D, Salsamendi E, Aihartza J. Foraging by Mediterranean horseshoe bats (Rhinolophus euryale) in relation to prey distribution and edge habitat. Journal of Mammalogy. 2008;89(2):493–502. 0519.
  25. 25. Semlitsch RD. Differentiating migration and dispersal processes for pond-breeding amphibians. Journal of Wildlife Management. 2008;72(1):260–7. doi: 10.2193/2007-082.
  26. 26. Guerry AD, Hunter ML. Amphibian distributions in a landscape of forests and agriculture: an examination of landscape composition and configuration. Conservation Biology. 2002;16(3):745–54. doi: 10.1046/j.1523-1739.2002.00557.x 3295.
  27. 27. Lemckert FL. Variations in anuran movements and habitat use: Implications for conservation. Applied Herpetology. 2004;1:165–81. 1623.
  28. 28. Kovar R, Brabec M, Vita R, Bocek R. Spring migration distances of some Central European amphibian species. Amphibia-Reptilia. 2009;30:367–78. 1172.
  29. 29. Trochet A, Moulherat S, Calvez O, Stevens V, Clobert J, Schmeller D. A database of life-history traits of European amphibians. Biodiversity Data Journal. 2014;2:p.e4123. [Online]. doi: 10.3897/BDJ.2.e4123. 7308. pmid:25425939
  30. 30. Aragón P, López P, Martín J. Conspecific chemical cues influence pond selection by male newts Triturus boscai. Copeia. 2000;2000(3):874–8. 1373.
  31. 31. Russell AP, Bauer AM, Johnson MK. Migration in amphibians and reptiles: an overview of patterns and orientation mechanisms in relation to life history strategies. In: Elewa AMT, editor. Migration of Organisms: Climate Geography Ecology. Berlin, Heidelberg, New York: Springer; 2005. p. 151–203.
  32. 32. Fahrig L. How much habitat is enough? Biological Conservation. 2001;100(1):65–74. 3379.
  33. 33. Mazerolle MJ, Desrochers A. Landscape resistance to frog movements. Canadian Journal of Zoology. 2005;83(3):455–64. 5032.
  34. 34. Janin A, Léna J-P, Deblois S, Joly P. Use of stress-hormone levels and habitat selection to assess functional connectivity of a landscape for an amphibian. Conservation Biology. 2012;26(5):923–31. doi: 10.1111/j.1523-1739.2012.01910.x. pmid:22891816
  35. 35. Chan-McLeod ACA. Factors affecting the permeability of clearcuts to red-legged frogs. Journal of Wildlife Management. 2003;67(4):663–71. 1095.
  36. 36. Ferreira M, Beja P. Mediterranean amphibians and the loss of temporary ponds: Are there alternative breeding habitats? Biological Conservation. 2013;165:179–86. doi: 10.1016/j.biocon.2013.05.029.
  37. 37. Reshetnikov AN. The introduced fish, rotan (Perccottus glenii), depresses populations of aquatic animals (macroinvertebrates, amphibians, and a fish). Hydrobiologia. 2003;510(1):83–90. doi: 10.1023/B:HYDR.0000008634.92659.b4 3314.
  38. 38. Adams MJ. Pond permanence and the effects of exotic vertebrates on anurans. Ecological Applications. 2000;10(2):559–68. 2688.
  39. 39. Jakob C, Poizat G, Veith M, Seitz A, Crivelli AJ. Breeding phenology and larval distribution of amphibians in a Mediterranean pond network with unpredictable hydrology. Hydrobiologia. 2003;499(1):51–61. 1387.
  40. 40. Alarcos G, Ortiz ME, Lizana M, Aragón A, Benéitez MJF. The colonization by amphibians of newly created aquatic environments as a conservation tool: the case of the Arribes del Duero. 2003.
  41. 41. Pavignano I, Giacoma C, Castellano S. A multivariate analysis of amphibian habitat determinants in north western Italy. Amphibia-Reptilia. 1990;11(4):311–24. 5671.
  42. 42. Scherer R, Muths E, Noon B. The importance of local and landscape-scale processes to the occupancy of wetlands by pond-breeding amphibians. Population Ecology. 2012:1–12. doi: 10.1007/s10144-012-0324-7 4884.
  43. 43. Trumbo D, Burgett A, Hopkins R, Biro E, Chase J, Knouft J. Integrating local breeding pond, landcover, and climate factors in predicting amphibian distributions. Landscape Ecology. 2012;27(8):1183–96. doi: 10.1007/s10980-012-9770-z 5045.
  44. 44. Mazerolle MJ, Desrochers A, Rochefort L. Landscape characteristics influence pond occupancy by frogs after accounting for detectability. Ecological Applications. 2005;15(3):824–34. 1277.
  45. 45. Van Buskirk J. Local and landscape influence on amphibian occurrence and abundance. Ecology. 2005;86(7):1936–47. 5861.
  46. 46. Zanini F, Pellet J, Schmidt BR. The transferability of distribution models across regions: an amphibian case study. Diversity and Distributions. 2009;15(3):469–80. 2713.
  47. 47. Brooks TM, Mittermeier RA, Mittermeier CG, Da Fonseca GAB, Rylands AB, Konstant WR, et al. Habitat loss and extinction in the hotspots of biodiversity. Conservation Biology. 2002;16(4):909–23. doi: 10.1046/j.1523-1739.2002.00530.x 5384.
  48. 48. Veysey J, Mattfeldt S, Babbitt K. Comparative influence of isolation, landscape, and wetland characteristics on egg-mass abundance of two pool-breeding amphibian species. Landscape Ecology. 2011;26(5):661–72. doi: 10.1007/s10980-011-9590-6 3932.
  49. 49. Altmoos M, Henle K. Relevance of multiple spatial scales in habitat models: A case study with amphibians and grasshoppers. Acta Oecologica. 2010;36(6):548–60. 3539. doi: 10.1016/j.envint.2010.04.008. pmid:20483456
  50. 50. ICNF. Áreas dos usos do solo e das espécies florestais de Portugal continental. Resultados preliminares. Lisbon, Portugal: Instituto da Conservação da Natureza e das Florestas, 2013.
  51. 51. IPMA/MAMAOT P. Temperatura média do ar (média anual) em Portugal Downloaded on 16 September 2013: Fundação Francisco Manuel dos Santos; 2013 [cited 2013 16-09-2013]. Available: http://www.pordata.pt/Portugal/Temperatura+media+do+ar+(media+anual)-1067.
  52. 52. Petitot M, Manceau N, Geniez P, Besnard A. Optimizing occupancy surveys by maximizing detection probability: application to amphibian monitoring in the Mediterranean region. Ecology and evolution. 2014;4(18):3538–49. doi: 10.1002/ece3.1207. 7347. pmid:25478146
  53. 53. Clave para la identificación de anfibios presentes en Aragón: I.E.S. PABLO GARGALLO; n.a. [cited 2011 01-04-2011]. Available: http://iespgaza.educa.aragon.es/departamentos/ciencias/claves_dicotomicas/claves_anfibios.pdf.
  54. 54. Villero JMB. Los anfibios y reptiles de la provincia de Cádiz. Seville, Spain: Consejería de Medio Ambiente; 1995.
  55. 55. Beja P, Alcazar R. Conservation of Mediterranean temporary ponds under agricultural intensification: an evaluation using amphibians. Biological Conservation. 2003;114(3):317–26. 0916.
  56. 56. Caetano M, Nunes V, Araújo A. Manual da nova COS (Carta de Ocupação/Uso do Solo para Portugal Continental)-2ª versão preliminar de 26 de Janeiro de 2007. 2007.
  57. 57. Zuur AF, Ieno EN, Elphick CS. A protocol for data exploration to avoid common statistical problems. Methods in Ecology and Evolution. 2010;1(1):3–14. 1280.
  58. 58. Zuur AF, Ieno EN, Walker NJ, Saveliev AA, Smith GM. Mixed effects models and extensions in ecology with R. Gail M, Krickeberg K, Samet J, Tsiatis A, Wong W, editors. New York: Springer; 2009. 574 p.
  59. 59. Dormann CF, Elith J, Bacher S, Buchmann C, Carl G, Carré G, et al. Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. Ecography. 2013;36(1):027–46. doi: 10.1111/j.1600-0587.2012.07348.x 5343.
  60. 60. Fife D. Package "fifer". 1.0 ed. CRAN: CRAN; 2014.
  61. 61. Skaug H, Fournier D, Nielsen A, Magnusson A, Bolker B. glmmADMB: generalized linear mixed models using AD Model Builder. R package version 06. 2011;5:r143.
  62. 62. Barton K. MuMIn: Multi-model inference. R package version 1.9.13. 1.7.11 ed2013.
  63. 63. van Strien MJ, Keller D, Holderegger R, Ghazoul J, Kienast F, Bolliger J. Landscape genetics as a tool for conservation planning: predicting the effects of landscape change on gene flow. Ecological Applications. 2014;24(2):327–39. 6234. pmid:24689144
  64. 64. Keller D, Van Strien MJ, Holderegger R. Do landscape barriers affect functional connectivity of populations of an endangered damselfly? Freshwater Biology. 2012;57(7):1373–84. doi: 10.1111/j.1365-2427.2012.02797.x 4839.
  65. 65. Güthlin D, Kröschel M, Küchenhoff H, Storch I. Faecal sampling along trails: a questionable standard for estimating red fox Vulpes vulpes abundance. Wildlife Biology. 2012;18(4):374–82. doi: 10.2981/11-065. 5155
  66. 66. da Silva F, Candeira C, de Cerqueira Rossa-Feres D. Dependence of anuran diversity on environmental descriptors in farmland ponds. Biodiversity and Conservation. 2012;21(6):1411–24. doi: 10.1007/s10531-012-0252-z 4984.
  67. 67. Aben J, Adriaensen F, Thijs KW, Pellikka P, Siljander M, Lens L, et al. Effects of matrix composition and configuration on forest bird movements in a fragmented Afromontane biodiversity hot spot. Animal Conservation. 2012;15(6):658–68. doi: 10.1111/j.1469-1795.2012.00562.x 4958.
  68. 68. Zuur AF, Hilbe J, Ieno EN. A Beginner's Guide to GLM and GLMM with R: A Frequentist and Bayesian Perspective for Ecologists: Highland Statistics; 2013.
  69. 69. Burnham KP, Anderson DR. Model selection and multimodel inference: a practical information-theoretic approach. 2nd ed. New York: Springer; 1998. 488 p.
  70. 70. Klug P, Wolfenbarger LL, McCarty JP. The nest predator community of grassland birds responds to agroecosystem habitat at multiple scales. Ecography. 2009;32(6):973–82. 7355.
  71. 71. Abad-Franch F, Valença-Barbosa C, Sarquis O, Lima MM. All that glisters is not gold: sampling-process uncertainty in disease-vector surveys with false-negative and false-positive detections. PLoS Negl Trop Dis. 2014;8(9):e3187. doi: 10.1371/journal.pntd.0003187. pmid:25233352
  72. 72. Payne CJ, Ritchie EG, Kelly LT, Nimmo DG. Does fire influence the landscape-scale distribution of an invasive mesopredator? PLoS ONE. 2014;9(10):e107862. doi: 10.1371/journal.pone.0107862. pmid:25291186
  73. 73. Nakagawa S, Cuthill IC. Effect size, confidence interval and statistical significance: a practical guide for biologists. Biological Reviews. 2007;82(4):591–605. doi: 10.1111/j.1469-185X.2007.00027.x. pmid:17944619
  74. 74. Pilliod DS, Hossack BR, Bahls PF, Bull EL, Corn PS, Hokit G, et al. Non-native salmonids affect amphibian occupancy at multiple spatial scales. Diversity and Distributions. 2010;16(6):959–74. 2923.
  75. 75. Johnson PTJ, McKenzie VJ, Peterson AC, Kerby JL, Brown J, Blaustein AR, et al. Regional decline of an iconic amphibian associated with elevation, land-use change, and invasive species. Conservation Biology. 2011;25(3):556–66. 3746. doi: 10.1111/j.1523-1739.2010.01645.x. pmid:21342266
  76. 76. Hamer AJ, Parris KM. Local and landscape determinants of amphibian communities in urban ponds. Ecological Applications. 2011;21(2):378–90. doi: 10.1890/10-0390.1. pmid:21563570
  77. 77. Adams MJ, Pearl CA, Galvan S, McCreary B. Non-native species impacts on pond occupancy by an anuran. The Journal of Wildlife Management. 2011;75(1):30–5. doi: 10.1002/jwmg.29. 4006.
  78. 78. Bronmark C, Edenhamn P. Does the presence of fish affect the distribution of tree frogs (Hyla arborea)? Conservation Biology. 1994;8(3):841–5. 1695.
  79. 79. Knapp RA, Matthews KR. Non-native fish introductions and the decline of the mountain yellow-legged frog from within protected areas. Conservation Biology. 2000;14(2):428–38. 2544.
  80. 80. Bosch J, Rincón PA, Boyero L, Martínez-Solano I. Effects of introduced salmonids on a montane population of Iberian frogs. Conservation Biology. 2006;20(1):180–9. doi: 10.1111/j.1523-1739.2005.00296.x. pmid:16909671
  81. 81. Denoel M, Dzukic G, Kalezic ML. Effects of widespread fish introductions on paedomorphic newts in Europe. Conservation Biology. 2005;19(1):162–70. 2712.
  82. 82. Montori A. Trophic segregation between the brown trout (Salmo trutta) and the Pyrenean newt (Euproctus asper). In: Bohme W, Bischoff W, Ziegler T, editors. Herpetologia Bonnensis; 1997; Bonn: Societas Europaea Herpetologica; 1997. p. 273–8.
  83. 83. Yonekura R, Kita M, Yuma M. Species diversity in native fish community in Japan: comparison between non-invaded and invaded ponds by exotic fish. Ichthyological Research. 2004;51(2):176–9. doi: 10.1007/s10228-003-0200-8 3276.
  84. 84. Hartel T, Nemes S, Cogălniceanu D, Öllerer K, Schweiger O, Moga C-I, et al. The effect of fish and aquatic habitat complexity on amphibians. Hydrobiologia. 2007;583(1):173–82. 0161.
  85. 85. Kiesecker JM, Blaustein AR, Miller CL. Transfer of a pathogen from fish to amphibians. Conservation Biology. 2001;15(4):1064–70. doi: 10.1046/j.1523-1739.2001.0150041064.x 5708.
  86. 86. Orizaola G, Braña F. Effect of salmonid introduction and other environmental characteristics on amphibian distribution and abundance in mountain lakes of northern Spain. Animal Conservation. 2006;9(2):171–8. 1403.
  87. 87. Kats LB, Ferrer RP. Alien predators and amphibian declines: review of two decades of science and the transition to conservation. Diversity and Distributions. 2003;9(2):99–110. 2708.
  88. 88. Kats LB, Petranka JW, Sih A. Antipredator defenses and the persistence of amphibian larvae with fishes. Ecology. 1988;69(6):1865–70. 3345.
  89. 89. Teplitsky C, Plénet S, Joly P. Tadpoles' responses to risk of fish introduction. Oecologia. 2003;134(2):270–7. 1707. pmid:12647168
  90. 90. Pujol-Buxó E, San Sebastián O, Garriga N, Llorente GA. How does the invasive/native nature of species influence tadpoles’ plastic responses to predators? Oikos. 2013;122(1):19–29. doi: 10.1111/j.1600-0706.2012.20617.x 4966.
  91. 91. Hecnar SJ, M'Closkey RT. The effects of predatory fish on amphibian species richness and distribution. Biological Conservation. 1997;79(2–3):123–31. 1069.
  92. 92. Watt PJ, Nottingham SF, Young S. Toad tadpole aggregation behaviour: evidence for a predator avoidance function. Animal Behaviour. 1997;54(4):865–72. 1616. pmid:9344439
  93. 93. Tejedo M. Size-dependent vulnerability and behavioral responses of tadpoles of two anuran species to beetle larvae predators. Herpetologica. 1993;49(3):287–94. 1508.
  94. 94. Waringer L, schenkohl A. An experimental study of microhabitat selection and microhabitat shifts in European tadpoles. Amphibia-Reptilia. 1988;9:219–36. 1723.
  95. 95. Escoriza D, Boix D. Assessing the potential impact of an invasive species on a Mediterranean amphibian assemblage: a morphological and ecological approach. Hydrobiologia. 2012;680(1):233–45. doi: 10.1007/s10750-011-0936-5 4981.
  96. 96. Diaz-Paniagua C. Larval diets related to morphological characters of five anuran species in the Biological Reserve of Donana (Huelva, Spain). Amphibia-Reptilia. 1985;6:307–21. 1680.
  97. 97. Gómez-Rodríguez C, Bustamante J, Díaz-Paniagua C, Guisan A. Integrating detection probabilities in species distribution models of amphibians breeding in Mediterranean temporary ponds. Diversity and Distributions. 2011;18(3):260–72. doi: 10.1111/j.1472-4642.2011.00837.x 4306.
  98. 98. Diaz-Paniagua C. Variability in timing of larval season in an amphibian community in SW Spain. Ecography. 1992;15(3):267–72. 0917.
  99. 99. Diaz-Paniagua C. Temporal segregation in larval amphibian communities in temporary ponds at a locality in SW Spain. Amphibia-Reptilia. 1988;9:15–26. 1368.
  100. 100. Häkkinen J, Pasanen S, Kukkonen JVK. The effects of solar UV-B radiation on embryonic mortality and development in three boreal anurans (Rana temporaria, Rana arvalis and Bufo bufo). Chemosphere. 2001;44(3):441–6. 1565. pmid:11459149
  101. 101. Ribeiro R, Carretero M, Sillero N, Alarcos G, Ortiz-Santaliestra M, Lizana M, et al. The pond network: can structural connectivity reflect on (amphibian) biodiversity patterns? Landscape Ecology. 2011;26(5):673–82. doi: 10.1007/s10980-011-9592-4 3933.
  102. 102. Diego-Rasilla FJ, Luengo RM. Heterospecific call recognition and phonotaxis in the orientation behavior of the marbled newt, Triturus marmoratus. Behavioral Ecology and Sociobiology. 2004;55(6):556–60. doi: 10.1007/s00265-003-0740-y.
  103. 103. Diego-Rasilla FJ, Luengo RM. Acoustic orientation in the palmate newt, Lissotriton helveticus. Behav Ecol Sociobiol. 2007;61(9):1329–35. doi: 10.1007/s00265-007-0363-9 5738.
  104. 104. Diego-Rasilla FJ, Luengo RM, Phillips JB. Use of a magnetic compass for nocturnal homing orientation in the palmate newt, Lissotriton helveticus. Ethology. 2008;114(8):808–15. doi: 10.1111/j.1439-0310.2008.01532.x 5777.
  105. 105. Joly P, Miaud C, Lehmann A, Grolet O. Habitat matrix effects on pond occupancy in newts. Conservation Biology. 2001;15(1):239–48. 2710.
  106. 106. Semlitsch RD. Biological delineation of terrestrial buffer zones for pond-breeding salamanders. Conservation Biology. 1998;12(5):1113–9. 1260.
  107. 107. Ficetola GF, Manenti R, De Bernardi F, Padoa-Schioppa E. Can patterns of spatial autocorrelation reveal population processes? An analysis with the fire salamander. Ecography. 2011;35(8):693–703. doi: 10.1111/j.1600-0587.2011.06483.x 4593.
  108. 108. Zahn A, Rainho A, Rodrigues L, Palmeirim JM. Low macro-arthropod abundance in exotic Eucalyptus plantations in the Mediterranean. Applied Ecology and Environmental Research. 2009;7(4):297–301. 2380.
  109. 109. Manenti R, Ficetola GF, De Bernardi F. Water, stream morphology and landscape: complex habitat determinants for the fire salamander Salamandra salamandra. Amphibia-Reptilia. 2009;30:7–15. 1817.
  110. 110. Caetano MH, Leclair R Jr. Comparative phenology and demography of Triturus boscai from Portugal. Journal of Herpetology. 1999;33(2):192–202. 1372.
  111. 111. García-Muñoz E, Ceacero F, Carretero MA, Pedrajas-Pulido L, Parra G, Guerrero F. Optimizing protection efforts for amphibian conservation in Mediterranean landscapes. Acta Oecologica. 2013;49:45–52. doi: 10.1016/j.actao.2013.02.013.
  112. 112. Lenhardt PP, Schäfer RB, Theissinger K, Brühl CA. An expert-based landscape permeability model for assessing the impact of agricultural management on amphibian migration. Basic and Applied Ecology. 2013;14(5):442–51. doi: 10.1016/j.baae.2013.05.004.
  113. 113. Humbert J- Y, Ghazoul J, Walter T. Meadow harvesting techniques and their impacts on field fauna. Agriculture, Ecosystems & Environment. 2009;130(1–2):1–8. 0950.
  114. 114. Hamer AJ, Makings JA, Lane SJ, Mahony MJ. Amphibian decline and fertilizers used on agricultural land in south-eastern Australia. Agriculture, Ecosystems & Environment. 2004;102(3):299–305. 1626.
  115. 115. Mann RM, Hyne RV, Choung CB, Wilson SP. Amphibians and agricultural chemicals: Review of the risks in a complex environment. Environmental Pollution. 2009;157(11):2903–27. 0942. doi: 10.1016/j.envpol.2009.05.015. pmid:19500891
  116. 116. Burton EC, Gray MJ, Schmutzer AC, Miller DL. Differential responses of postmetamorphic amphibians to cattle grazing in wetlands. Journal of Wildlife Management. 2009;73(2):269–77. 2727.
  117. 117. Schmutzer AC, Gray MJ, Burton EC, Miller DL. Impacts of cattle on amphibian larvae and the aquatic environment. Freshwater Biology. 2008;53(12):2613–25. 1790.