Batrachochytrium salamandrivorans is the predominant chytrid fungus in Vietnamese salamanders

The amphibian chytrid fungi, Batrachochytrium dendrobatidis (Bd) and B. salamandrivorans (Bsal), pose a major threat to amphibian biodiversity. Recent evidence suggests Southeast Asia as a potential cradle for both fungi, which likely resulted in widespread host-pathogen co-existence. We sampled 583 salamanders from 8 species across Vietnam in 55 locations for Bsal and Bd, determined scaled mass index as a proxy for fitness and collected environmental data. Bsal was found within 14 of the 55 habitats (2 of which it was detected in 2013), in 5 salamandrid species, with a prevalence of 2.92%. The globalized pandemic lineage of Bd was found within one pond on one species with a prevalence of 0.69%. Combined with a complete lack of correlation between infection and individual body condition and absence of indication of associated disease, this suggests low level pathogen endemism and Bsal and Bd co-existence with Vietnamese salamandrid populations. Bsal was more widespread than Bd, and occurs at temperatures higher than tolerated by the type strain, suggesting a wider thermal niche than currently known. Therefore, this study provides support for the hypothesis that these chytrid fungi may be endemic to Asia and that species within this region may act as a disease reservoir.


Results
Across 11 provinces located in northern Vietnam, 44 ponds and 11 streams were sampled and 583 individuals from 8 species, belonging to the genera Paramesotriton and Tylototriton, were sampled for the presence of Bsal and Bd and their body condition determined (Table 1). A total of 17 individuals belonging to 5 species (T. vietnamensis, T. ziegleri, T. asperrimus, P. deloustali and P. sp.) tested positive for B. salamandrivorans resulting in an overall prevalence of 2.92% (95% CI: 1.80%, 4.65%). Two of the three species which did not contain positive individuals had small sample sizes (T. anguliceps (8), T. sp. (17) and P. guanxiensis (54)) and as the prevalence is low within this area more samples need to be gathered to definitively determine if these populations contain infected individuals. The average prevalence of the 14 positive sites was 17.56% (SD ± 25.27, 95% CI: 3.06%, 32.24%). Four individuals of T. ziegleri, in one pond, tested positive for Bd resulting in an overall prevalence of 0.69% (95% CI: 0.20%, 1.83%) and a prevalence of 30.77% (95% CI: 12.35%, 57.96%) within the positive pond. The Bd samples amplified with the global pandemic lineage (BdGPL) probes indicating a BdGPL-like lineage.
Bsal infection prevalence did not differ between provinces or species (Fisher's exact tests, P = 0.34 and P = 0.25, respectively). There were no correlations between infection intensity of Bsal or Bd and scaled mass index (SMI) for any of the eight species (Spearman's correlations < ± 0.18). Elevation, province, species, scaled mass index, water temperature and ambient temperature were not significant predictors of infection intensity for Bsal (Generalized linear model, P > 0.05).

Discussion
Our study provides the first large scale screening for Batrachochytrium salamandrivorans (Bsal) within Southeast Asia, specifically Vietnam. We found that Bsal was much more widespread throughout salamander populations in Vietnam than its sister species B. dendrobatidis (Bd), occurring in 14 of the 55 locations compared to the one pond in which Bd was detected (Fig. 1), corroborating the predilection of Bsal for urodelan hosts 12 and providing further evidence for a Bd prevalence that is low throughout Asia [15][16][17][18][19]22 . We did not find any signs of Bsal associated disease such as decreased body condition, lesions, or death in the infected populations. The overall low prevalence for both Bsal (2.92%) and Bd (0.69%), despite presence of Bsal in Vietnamese salamander populations at least since 2013 12 , combined with the absence of disease signs indicates endemism of both species of chytrid fungi within Vietnamese salamander populations, adds support to the hypothesis that South East Asian salamanders constitute a Bsal reservoir. Similarly low prevalence but widespread occurrence has been demonstrated for Bd in regions where Bd is considered endemic in the absence of epidemic disease 13,23,24 . If compensated, co-existence does not exclude an actual cost of Bsal on the Vietnamese salamander populations, as has been demonstrated for Bd 23,25 , especially given the potential of both pathogens to cause juvenile mortality under experimental conditions 12,22 .
Through the use of lineage specific qPCR, we found that the Bd we detected amplified as global pandemic lineage (BdGPL) indicating it could be the true BdGPL, endemic or non-endemic BdGPL-like, or a hybrid of an endemic and BdGPL cross 26 . BdGPL is associated with all of the known epizootic events and has experimentally been shown to be more virulent than other lineages 27 . As Bd seems to occur in an endemic state within Asia and our study site of Vietnam, it is likely an endemic BdGPL-like lineage, further testing should be done to determine the strain as well as the virulence of the Bd within Vietnam.
The thermal optimum for the type strain of Bsal is 15-20 °C 28 , whereas the majority of infected individuals we found in Vietnam were in ponds or streams with water temperatures between 20-25 °C, even reaching 26 in one positive location (Fig. 2). This might have serious implications as differential thermal preferences of strains could significantly expand the invasive niche if the fungus is vectored into a naive region (Yap et al. 29 ).
In conclusion, due to the low prevalence of Bsal and BdGPL and lack of correlation between any of the environmental or physical parameters, we provide evidence supporting the hypothesis that these chytrid fungi may be endemic to Asia and that species within this region may act as a disease reservoir for naive regions. We found that Bsal is much more widespread within northern Vietnam, occurring at temperatures higher than those tolerated by the fungus' type strain, which suggests a wider thermal niche than currently known. Further screening should be conducted to determine the range of Bsal and which species carry the fungus. These results reinforce the notion that there should be strict biosecurity measures in place, in order to prevent the further spread of Bsal and avoid a global impact similar to that of Bd.  ). Nets were utilized to capture salamanders within the stream or pond. Each salamander was weighed, using a digital scale (MS-Series,G&G GmbH), and measured for snout to vent and tail length, using a 0-150 mm digital caliper (Fixpoint). A scaled mass index was calculated following the method developed by Peig and Green 30 .
Data collection. Since the Bsal type strain has a low tolerance for high environmental temperatures, these were predicted to drive Bsal infection. Therefore, at each pond or stream water temperature and ambient temperature were measured using an HHC201 thermocouple thermometer, with a hermetically sealed thermocouple (OMEGA ® ). Prevalence and infection loads of Bsal and Bd in the captured amphibians was determined by collecting non-invasive skin swabs 29  Sample processing. DNA was extracted from the swabs using 100 μ l of Prepman Ultra DNA extraction buffer. Extracted DNA samples were diluted 1/10 with HPLC water to reduce PCR inhibition 31 and then stored at −20 °C until processing. Samples were processed using the B. salamandrivorans and B. dendrobatidis specific duplex real-time PCR procedures described by Blooi et al. 32 on CFX96 real-time system (Bio-Rad Laboratories, Hercules, CA). Real-time PCR results (genomic equivalents (GE) of B. salamandrivorans and B. dendrobatidis zoospores) were corrected for the applied dilution factor. Lineage specific qPCR was conducted utilizing Taqman MGB probes that are able to discriminate single nucleotide polymorphisms in the Bd mitochondrial genome which are diagnostic for the globalized pandemic lineage (BdGPL), following the protocol of Bletz et al. 26 .
Data analysis. The modified Wald method used to calculate the 95% CI for proportion of infected individuals. Fisher exact tests were used to compare prevalence between species and provinces. We examined how GE load was affected by species, SMI, SVL, water temperature and elevation using generalized linear models (Gaussian functions) and spearman correlations in R (version 3.2.3), due to the non-normal distribution of the data.