Ecological Barriers for an Amphibian Pathogen: A Narrow Ecological Niche for Batrachochytrium salamandrivorans in an Asian Chytrid Hotspot

The chytrid fungal pathogens Batrachochytrium salamandrivorans (Bsal) and B. dendrobatidis (Bd) are driving amphibian extinctions and population declines worldwide. As their origins are believed to be in East/Southeast Asia, this region is crucial for understanding their ecology. However, Bsal screening is relatively limited in this region, particularly in hotspots where Bd lineage diversity is high. To address this gap, we conducted an extensive Bsal screening involving 1101 individuals from 36 amphibian species, spanning 17 natural locations and four captive facilities in the biodiversity-rich Guangxi Zhuang Autonomous Region (GAR). Our PCR assays yielded unexpected results, revealing the complete absence of Bsal in all tested samples including 51 individuals with Bd presence. To understand the potential distribution of Bsal, we created niche models, utilizing existing occurrence records from both Asia and Europe. These models estimated potential suitable habitats for Bsal largely in the northern and southwestern parts of the GAR. Although Bsal was absent in our samples, the niche models identified 10 study sites as being potentially suitable for this pathogen. Interestingly, out of these 10 sites, Bd was detected at 8. This suggests that Bsal and Bd could possibly co-exist in these habitats, if Bsal were present. Several factors seem to influence the distribution of Bsal in Asia, including variations in temperature, local caudate species diversity, elevation, and human population density. However, it is climate-related factors that hold the greatest significance, accounting for a notable 60% contribution. The models propose that the specific climatic conditions of arid regions, primarily seen in the GAR, play a major role in the distribution of Bsal. Considering the increased pathogenicity of Bsal at stable and cooler temperatures (10–15 °C), species-dependent variations, and the potential for seasonal Bd-Bsal interactions, we emphasize the importance of periodic monitoring for Bsal within its projected range in the GAR. Our study provides deeper insights into Bsal’s ecological niche and the knowledge generated will facilitate conservation efforts in amphibian populations devastated by chytrid pathogens across other regions of the world.


Introduction
Amphibians worldwide face considerable threats, including habitat loss, climate change, and disease, which contribute to severe population declines and biodiversity loss [1].Chytridiomycosis, a disease caused by the chytrid fungi Batrachochytrium dendrobatidis (Bd) and B. salamandrivorans (Bsal), is a major driver of these declines [2][3][4].Both Bd and Bsal are thought to have originated in Asia, with endemic Bd lineages present in the region [5,6].Bd affects all three amphibian orders and has caused widespread population declines, particularly in anuran species across all amphibian-inhabited continents, and except in Asia [4,7,8].In contrast, Bsal mainly affects caudates and several anurans [9,10].It has led to severe declines in native salamanders only in Europe so far, although the pathogen is present in Asia [4][5][6]11]; however, the impending threat for the highly diverse North American and Neotropical salamanders is significant [12][13][14][15][16][17][18].Knowledge of Bsal ecology is critical as it provides proactive information for amphibian conservation, including identifying new susceptible species and analyzing their occurrences with the identification of the relevant ecological drivers.
Although Bsal primarily infects caudates, recent studies indicate that anurans can also act as reservoir hosts or as vectors for Bsal [9,[19][20][21][22].This role increases Bsal dispersal pathways as well as disease risk to sensitive salamander species and populations as some of these species tend co-occur with anurans in the same habitat.Therefore, identifying reservoir hosts is crucial for understanding infection dynamics and potential occurrences as Bsal continues to expand its range and threaten biodiversity [23][24][25].
Bsal tends to co-occur with Bd in several microhabitats [19,26].These Bd-Bsal cooccurrences perhaps facilitate hybridization, leading to the generation of new genotypes with heightened pathogenicity [11,27,28].In their overlapping natural habitats, three species of caudata-Salamandra salamandra, Triturus cristatus, and Ichthyosaura alpestrishave been found to carry concurrent infections of Bd and Bsal [19,29].It has also been demonstrated that these concurrent infections tend to escalate their severity under the specific conditions [30,31].Additionally, Bsal has been found to be more pathogenic than Bd within the same salamander host species, which results in more severe infection outcomes [32].
Although Bsal has its origin in Asia, our understanding of its infections within this region is still relatively limited.Significantly, Bsal has been observed cohabiting with the global pandemic lineage of Bd, also known as BdGPL, in Southeast Asia, specifically Vietnam [26].Certain areas within Asia, such as South China, harbor multiple Bd-Asian genotypes along with BdGPL [33 -35].Yet, we still do not know whether these genotypes coexist with Bsal.Therefore, conducting surveys on the co-occurrences of Bsal and Bd lineages in their natural habitats will yield important insights into the interactions of pathogenic chytrids.
In addition, in terms of the niche space and geographic distribution of Bsal, ecological niche models can be useful in predicting the potential suitable habitats of pathogens and evaluating the different influences of variables on pathogen occurrences [36].Prior work has predicted the potential distribution of Bsal in Asia, with diurnal temperature range identified as the most important factor [15,37], but they failed to consider biotic interactions that might alter the range estimate.For example, alterations in biodiversity significantly impact pathogen dynamics through mechanisms such as dilution and amplification effects [38,39].This suggests that biodiversity could play a crucial role in shaping the distribution patterns of Bsal.On the other hand, it is indicated that landscape structure and anthropogenic influence can further drive the dispersal risk of Bsal [40].Given that human-related factors and biotic interactions significantly influence pathogen/parasite occurrences [41][42][43][44], adding biotic factors and landscape features can improve the accuracy of estimated Bsal distributions and niche spaces.
To identify the potential reservoir hosts and test whether Bsal co-occurs together with Bd populations in microhabitats in South China, we screened for Bsal infection in a regional amphibian hotspot, the Guangxi Zhuang Autonomous Region (GAR), which is newly recognized for its diverse Bd genotypes spanning various natural habitats [35], including where Bsal was previously verified [45].To assess the influence of combined environmental factors and biotic interactions on Bsal occurrences in the region, we predicted the potential distribution of this pathogen through ecological niche models that included climate, landscape, and biotic characteristics.Finally, we examined the areas in the GAR that are suitable for Bsal.The knowledge generated that is relevant to Bsal ecology could advance our understanding of Bsal pathogenicity and disease dynamics which could have significant implications for amphibian conservation.

Survey Region
We performed Bsal screening throughout South China's Guangxi Zhuang Autonomous Region (GAR) (Figure 1), which shares its southwestern border with Vietnam, a country known to harbor Bsal [26] from where one location was previously reported for the presence of Bsal [45].The presence of Bsal also has been confirmed in the adjacent Guangzhou province [45].The closest record of Bsal presence is located at a distance of ca. 100 km of one studied site in the GAR.In the survey region, Bsal has so far been detected in an individual of a salamander species, Pachytriton wuguanfui [45], and it is also now known that basal Asian and global lineages of Bd exist [35].
advance our understanding of Bsal pathogenicity and disease dynamics which could have significant implications for amphibian conservation.

Survey Region
We performed Bsal screening throughout South China's Guangxi Zhuang Autonomous Region (GAR) (Figure 1), which shares its southwestern border with Vietnam, a country known to harbor Bsal [26] from where one location was previously reported for the presence of Bsal [45].The presence of Bsal also has been confirmed in the adjacent Guangzhou province [45].The closest record of Bsal presence is located at a distance of ca. 100 km of one studied site in the GAR.In the survey region, Bsal has so far been detected in an individual of a salamander species, Pachytriton wuguanfui [45], and it is also now known that basal Asian and global lineages of Bd exist [35].

Sample Collection
To test whether Bsal co-occurs with Bd in the same habitats and Bsal co-infects with Bd in the same hosts in the surveyed region, we used previously extracted DNA from skin swabs (N = 1006) from amphibian adults collected between 2019 and 2021 across seventeen natural sites including nine sites confirmed for Bd presence [35].These samples included 372 individuals collected in spring (April), 769 in summer (May, June, and July), 40 in autumn (October), and 12 in winter (November).Given that Bsal can occasionally spill over from captive populations into wildlife communities [46][47][48], we also tested the collected 95 skin swabs from a pet market and three frog farms between 2019 and 2020 and then extracted the genomic DNA from the skin swabs using PrepMan Ultra (Life Technologies, Warrington, United Kingdom) [49].

Detection of Bsal
We used nested PCR to detect Bsal on DNA extracts from skin swabs [50,51].For the first amplification, the primers ITS1f and ITS4 were used, which specifically combined the 18S and 28S rRNA genes.The PCR amplification conditions, 4 min at 94 °C, followed by 30 cycles of 30 s at 94 °C, 30 s at 55 °C, 1 min at 72 °C, and a final 10 min at 72 °C were implemented.For the second amplification, the specific primers (STerF and STerR) for Bsal were used to amplify a fragment gene of the ITS-5.8SrRNA region [3].The conditions of PCR amplification included 4 min at 94 °C, followed by 30 cycles of 30 s at 94 °C, 60 °C for 30s, 1 min at 72 °C, and a final 10 min at 72 °C.PCR amplification products were visualized using 1.5% agarose gel electrophoresis.We used the synthetic DNA sequences of Bsal as positive controls and two negative controls in each plate.If PCR products yielded positive results, Sanger sequencing was conducted to verify if the amplified DNA fragments were

Sample Collection
To test whether Bsal co-occurs with Bd in the same habitats and Bsal co-infects with Bd in the same hosts in the surveyed region, we used previously extracted DNA from skin swabs (N = 1006) from amphibian adults collected between 2019 and 2021 across seventeen natural sites including nine sites confirmed for Bd presence [35].These samples included 372 individuals collected in spring (April), 769 in summer (May, June, and July), 40 in autumn (October), and 12 in winter (November).Given that Bsal can occasionally spill over from captive populations into wildlife communities [46][47][48], we also tested the collected 95 skin swabs from a pet market and three frog farms between 2019 and 2020 and then extracted the genomic DNA from the skin swabs using PrepMan Ultra (Life Technologies, Warrington, United Kingdom) [49].

Detection of Bsal
We used nested PCR to detect Bsal on DNA extracts from skin swabs [50,51].For the first amplification, the primers ITS1f and ITS4 were used, which specifically combined the 18S and 28S rRNA genes.The PCR amplification conditions, 4 min at 94 • C, followed by 30 cycles of 30 s at 94 • C, 30 s at 55 • C, 1 min at 72 • C, and a final 10 min at 72 • C were implemented.For the second amplification, the specific primers (STerF and STerR) for Bsal were used to amplify a fragment gene of the ITS-5.8SrRNA region [3].The conditions of PCR amplification included 4 min at 94 • C, followed by 30 cycles of 30 s at 94 • C, 60 • C for 30 s, 1 min at 72 • C, and a final 10 min at 72 • C. PCR amplification products were visualized using 1.5% agarose gel electrophoresis.We used the synthetic DNA sequences of Bsal as positive controls and two negative controls in each plate.If PCR products yielded positive results, Sanger sequencing was conducted to verify if the amplified DNA fragments were indeed associated with Bsal.The methods for detection of Bd and its results have been described in [35].
For the individual samples detected with Bd infection based on nested PCR assays, we also used the simplex real-time PCR method described by [52] to test whether they were infected with Bsal.The real-time PCR assays were performed with the ViiA TM 7 Real-Time PCR System (Life Technologies, Woodlands, Singapore) using amplification conditions consisting of 3 min at 94 • C followed by 40 cycles of denaturation at 94 • C for 5 s and annealing/extension at 60 • C for 90 s.We used duplicate samples, two negative controls, and a series of plasmid dilution standards ranging from 1.68 × 10 0 to 1.68 × 10 6 Molecules/µL in the real-time PCR run (Pisces Molecular, Boulder, CO, USA).

Ecological Niche Modelling
We used verified presence records for Bsal in both Europe and Asia to build the ecological niche model.Previous presence records of Bsal were attained from [19,26,45,53,54].A single coordinate for each grid cell was retained to trim duplicate observation records.We used the ecoregions of Bsal occurrences as background to improve model calibration [55].
A total of 29 abiotic and biotic variable layers were obtained and resampled to the spatial resolution of 30 s (Table 1), and 19 bioclimatic variables represented the current climatic conditions.The human footprint that imposed pressure on the environment [56] and the human population density indicated the potential distribution of Bsal mediated by humans.The normalized difference vegetation index (NDVI), the enhanced vegetation index (EVI), and the net primary productivity (NPP) represented the natural state of the habitats.We used the maximum of monthly NDVIs and EVIs to synthesize the yearly NDVIs and EVIs.The average yearly NDVIs, EVIs, and NPP from 2019, 2020, and 2021 were used as the final predictor variables to reduce the potential inter-annual variation.The amphibian species' richness and caudate species' richness were used as biotic information to improve model fit and better understand the relationships between the species community and Bsal occurrences.To exclude the effects of high collinearity between predictors, we calculated the correlations between variables in the ENMTools [57], then retained eleven variables with Pearson's r < 0.7 to construct ecological niche models (Tables S1 and S2).
We applied two model classes: generalized liner model (GLM) and Maximum Entropy Modeling (MaxEnt) to evaluate the relative importance influencing Bsal occurrences and to estimate the suitable habitats of Bsal [58,59].The model performance was evaluated using the cross-validated area under the receiver (AUC) and true skill statistic (TSS) [60,61], which was calculated by splitting the training (70%) and testing (30%) observations.AUC values below 0.7 were poor, 0.7-0.9 were good, and 0.9-1.0 were excellent [62,63].TSS values below 0.4 were poor, 0.4-0.8 were useful, and >0.8 were good to excellent [64].GLM and Maxent models were performed in "sdm" packages [65], with 10 replications for each model.The ensemble model was generated using weighted averaging based on TSS statistics and the threshold of maximizing the sum of sensitivity and specificity [66].We finally extracted the studied region from generated Bsal distribution maps.All analyses in this study were performed in ArcGIS 10.4.1 (Environmental Systems Research Institute, 16th June 2021) and R 4.2.2 (R Development Core Team, 31th October 2022).
Table 1.All initial predictor variables, their sources, and the related biological hypotheses.Bold fonts indicate final predictor variables used in our ecological niche models, with Pearson's r < 0.7.These predictor variables represent three main categories including climate, landscape (non-climate), and biotic factors, and the corresponding hypotheses for the associations of these factors to Bsal presence.

Amphibian species' richness
Dilution or amplification effects of biodiversity have profound influence on pathogen transmission [38,71].Higher amphibian species richness might pose dilution effect since it possibly includes resistant species, whereas higher caudate species richness may pose an amplification effect since it may include more susceptible species.[72] Caudate species' richness [72] 3. Results

Bsal Absence in Wild and Captive Amphibians
We did not detect Bsal infection based on nested PCR assays in any of the individuals (N = 1101, representing 36 amphibian species), of which 51 individuals from 16 species were infected with Bd (Table 2).The individuals infected with Bd consisted of 48 wild individuals of 15 species and 3 captive individuals belonging to the Chinese giant salamander (Andrias davidianus).Also, Bsal was not detected with the simplex real-time PCR assay in these Bd-infected individuals.

Table 2. Identification of species and individual for
Bsal presence in the current study.Species highlighted in bold were previously recorded as Bd-positive according to [35], with the exception of A. davidianus, which was detected herein.Species denoted with an asterisk (*) represent samples obtained from captive amphibians.

Bsal Models
The average AUC and TSS values for the Bsal distribution model were 0.90 (SD ± 0.04) and 0.74 (SD ± 0.08), respectively, signifying good model performance.The top two variables for model predictions included mean diurnal temperature range and caudate species richness (Figure 2).The overall contribution of the climate-related five variables formed the largest importance (60%) in explaining Bsal distribution.Predictor variables associated with landscape factors (i.e., net primary productivity, human population density, and enhanced vegetation index) jointly contributed towards explaining about 12% of the Bsal distribution, although a single landscape-related variable did not show high relative importance.Factors associated with biotic interactions significantly affected the predicted Bsal distribution, with caudate species occurrence being positively related to the presence of this pathogen (Figure S1).
The niche models of Bsal estimated that the suitable habitats for Bsal are in the northern and southwestern parts of the GAR (Figure 3; Bsal suitability range: 0.49-0.94).The challenging ecological conditions prevalent in much of the GAR, which include high fluctuations in temperature and precipitation, generally exceed the environmental preferences of Bsal (Table S3).
Among the 21 surveyed wild sites, 10 were deemed suitable for Bsal, with suitability indexes ranging from low (0.54) to high (0.89).These suitable sites were confirmed for Bd presence except for one frog farm and one wild site (Figure 1).The niche models of Bsal estimated that the suitable habitats for Bsal are in the northern and southwestern parts of the GAR (Figure 3; Bsal suitability range: 0.49-0.94).The challenging ecological conditions prevalent in much of the GAR, which include high fluctuations in temperature and precipitation, generally exceed the environmental preferences of Bsal (Table S3).
Among the 21 surveyed wild sites, 10 were deemed suitable for Bsal, with suitability indexes ranging from low (0.54) to high (0.89).These suitable sites were confirmed for Bd presence except for one frog farm and one wild site (Figure 1).The niche models of Bsal estimated that the suitable habitats for Bsal are in the northern and southwestern parts of the GAR (Figure 3; Bsal suitability range: 0.49-0.94).The challenging ecological conditions prevalent in much of the GAR, which include high fluctuations in temperature and precipitation, generally exceed the environmental preferences of Bsal (Table S3).
Among the 21 surveyed wild sites, 10 were deemed suitable for Bsal, with suitability indexes ranging from low (0.54) to high (0.89).These suitable sites were confirmed for Bd presence except for one frog farm and one wild site (Figure 1).

Discussion
This study provides insights into the ecological niches of the deadly amphibian pathogen Batrachochytrium salamandrivorans (Bsal) within an Asian chytrid hotspot.Despite our extensive sampling of 1101 individuals across 36 species, Bsal was not detected.This absence contrasts with the detection of Bd, which was present in 51 individuals from 16 amphibian species, including the Chinese giant salamander A. davidianus, known to be susceptible to both pathogens [10,45,73].This differential distribution may indicate that these pathogens have distinct ecological niches, potentially driven by differences in their biotic or abiotic preferences.
Drawing upon presence data from extensive studies across Asia and Europe, our ecological niche models underscore the influence of certain biotic and abiotic factors in shaping the potential distribution of Bsal.As indicated by robust AUC and TSS values, the models provide a strong performance, pointing to the mean diurnal temperature range and caudate species richness as notable predictors.These insights suggest a unique thermal niche preference for Bsal, which may not be met within the habitats of the amphibians surveyed in our study.The observed absence of Bsal could be attributed to the scarcity of caudates in the area and the temperature-dependent pathogenicity of the organism.Bsal prefers lower thermal conditions for its survival; temperatures above 22 • C not only inhibit its growth but also reduce its ability to infect hosts, even those that are highly susceptible [67].Previous research supports this observation, suggesting that temperature is a key factor behind low prevalence of Bsal in Asian amphibians [45].Both the environmental and water temperatures in these regions are generally above 15 • C, which is considered suboptimal for survival and growth of Bsal [3,26,45,74].Furthermore, our models reinforce earlier findings regarding the significant influence of diurnal temperature range on Bsal distribution [15], and highlight the contribution of biotic interactions, particularly with caudate species, in shaping the niche space of this pathogen.
The absence of confirmed cases echoes previous studies [51,75], underscoring that Bsal may display a transient persistence with reduced transmission rates among anurans [22].This finding warrants a deeper investigation into host-specific factors, such as skin factors, that could be influencing the presence or detection of Bsal [76].Furthermore, the influence of environmental factors such as plant materials in the water on Bsal growth [77] raises the possibility of Bsal being present in specific microhabitats [78] or reservoir hosts that were not covered in our survey.It is important to note that the sample size for tested caudates in the study is relatively small-only 30 individuals from 3 species out of a total of 1101 samples representing 36 amphibian species.This is particularly noteworthy given that salamanders are documented to be the most susceptible to Bsal [10,18,22].To this end, extending the range of sampling efforts in environmental reservoirs and salamanders could enrich our understanding of the existence of Bsal.
Interestingly, the niche models predicted that some areas within the northern and southwestern regions of the GAR would provide suitable habitats for Bsal.Notably, 10 out of the 21 natural sites surveyed were deemed suitable for Bsal.This, coupled with the presence of Bd at eight of these sites, raises the possibility of a potential co-occurrence of these pathogens in these habitats, despite the current absence of Bsal at present.Hence, future monitoring of these habitats, especially across seasons, will be important to understand the infection dynamics of pathogens, pathogenicity, and potential resource partitioning between Bsal and Bd.Additionally, investigating how both pathogens affect the same amphibian host species (c.a.28, [10,79], Table S4) could offer valuable insights into their potential interactions and mutual influence on infections.For instance, in an initial Bsal inoculation experiment involving the North American eastern newt (Notophthalmus viridescens), it was observed that a Bd infection negatively impacted the subsequent Bsal infection [80].
The absence of Bsal in habitats suitable for its presence, as suggested by the niche models, opens up further possibilities as well.While it is possible that Bsal has not yet been introduced to these regions, it could also be the case that the pathogen has been introduced but is not able to establish itself due to unknown ecological or biological barriers like suboptimal habitat connectivity [78,81,82].For instance, previous studies highlight the significant role of seasonality, driven by variations in temperature and precipitation, in determining Bsal distribution and host susceptibility [67,83].Considering that our sampling primarily targeted amphibians during their reproductive period in spring and summer, future surveys focusing on cooler seasons could potentially yield different results, given the known preference of Bsal for consecutive colder thermal conditions [3,74].
Finally, our study suggests a complex interplay of ecological and biological factors in dictating the distribution of amphibian chytrid fungal pathogens.As climate change continues to alter key ecological factors, the risk of Bsal is expected to persist in coastal areas and higher elevations, even though habitats with medium-to-high climatic suitability for Bsal are likely to decrease [37,84].The international and regional trade in amphibians has been identified as a significant driver for the translocation of both Bd and Bsal pathogens [5,29].
Our discovery of Bd in A. davidianus-a species susceptible to Bsal [10]-sourced from a pet market suggests that the emergence of Bd in natural amphibian populations in the GAR could be linked to a spillover from captive animals, facilitated by human activities.As such, regular monitoring and surveillance programs are essential for the early detection of these pathogens, enabling timely interventions to mitigate disease risks [85].Further research is urgently needed to understand the infection dynamics of these deadly pathogens, as this knowledge will be pivotal for predicting and managing future outbreaks among amphibian populations, both in Asia and suitable habitats across the world.

Conclusions
Our study sheds light on the ecological niches of the amphibian pathogen Bsal within an Asian chytrid hotspot.Despite rigorous sampling, Bsal was not found, but Bd was detected.This disparity could signal distinct ecological niches for these pathogens, possibly influenced by different biotic or abiotic factors.Our niche models suggest the importance of temperature in explaining the absence of Bsal in the study sites.This research reinforces the significance of diurnal temperature range on Bsal distribution and reveals the role of biotic interactions in shaping its niche space.The lack of Bsal may be due to transient persistence with reduced transmission rates; this warrants further study into host-specific factors.Interestingly, some regions of the GAR are predicted to be suitable for Bsal, despite its current absence, which is suggestive of potential co-occurrence with Bd in the future.This suggests that unknown ecological or biological barriers maybe responsible for Bsal establishment at present.Our research underlines the complex interplay of ecological and biological factors in determining amphibian chytrid fungal distribution and the importance of understanding these dynamics for future disease management in the GAR, and other regions of the world that Bsal has not yet penetrated.

Figure 1 .
Figure 1.Study region and site distribution.Green points represent natural populations; orange points are for sampled captive individuals; red points indicate sites where there is a known presence of Bsal.

Figure 1 .
Figure 1.Study region and site distribution.Green points represent natural populations; orange points are for sampled captive individuals; red points indicate sites where there is a known presence of Bsal.

Figure 2 .
Figure 2. Importance of average relative variables in the final ecological niche models for Bsal.Importance (permutation importance) of predictor variable for GLM models, Maxent models, and ensemble models were evaluated and scaled based on the AUC metric.

Figure 3 .
Figure 3. Predicted suitable habitats of Bsal in the studied region: (A) habitat suitability; (B) presence-absence distribution.Blue points represent the studied sites suitable for Bsal with confirmed Bd presence.Orange represents the studied sites suitable for Bsal, where Bd was not detected.Grey represents the studied sites that are unsuitable for Bsal occurrence.

Figure 2 .
Figure 2. Importance of average relative variables in the final ecological niche models for Bsal.Importance (permutation importance) of predictor variable for GLM models, Maxent models, and ensemble models were evaluated and scaled based on the AUC metric.

Figure 2 .
Figure 2. Importance of average relative variables in the final ecological niche models for Bsal.Importance (permutation importance) of predictor variable for GLM models, Maxent models, and ensemble models were evaluated and scaled based on the AUC metric.

Figure 3 .
Figure 3. Predicted suitable habitats of Bsal in the studied region: (A) habitat suitability; (B) presence-absence distribution.Blue points represent the studied sites suitable for Bsal with confirmed Bd presence.Orange represents the studied sites suitable for Bsal, where Bd was not detected.Grey represents the studied sites that are unsuitable for Bsal occurrence.

Figure 3 .
Figure 3. Predicted suitable habitats of Bsal in the studied region: (A) habitat suitability; (B) presenceabsence distribution.Blue points represent the studied sites suitable for Bsal with confirmed Bd presence.Orange represents the studied sites suitable for Bsal, where Bd was not detected.Grey represents the studied sites that are unsuitable for Bsal occurrence.