Consistency of Published Results on the Pathogen Batrachochytrium dendrobatidis in Madagascar: Formal Comment on Kolby et al. Rapid Response to Evaluate the Presence of Amphibian Chytrid Fungus (Batrachochytrium dendrobatidis) and Ranavirus in Wild Amphibian Populations in Madagascar

A recent paper by Kolby et al. [1], surveying for Batrachochytrium dendrobatidis (Bd) and ranavirus in Madagascar, presents results for 508 amphibian specimens and 68 water bodies sampled during a 2-month period of the 2013–14 wet season. Kolby et al. [1] did not detect Bd in any of the samples, presenting evidence that add to our understanding of Bd dynamics in Madagascar. Earlier in 2015, we published “Widespread presence of the pathogenic fungus Batrachochytrium dendrobatidis in wild amphibian communities in Madagascar” in the journal Scientific Reports [2]. We presented rigorous spatial and temporal surveillance data for 4,155 amphibians sampled across a 10-year period, and used two independent molecular diagnostics to demonstrate the occurrence of a molecular signature of Bd infection at multiple locations across the island. We focus here on solely the Bd results, which directly relate to our published study. 
 
While the conclusions of Bd’s occurrence and prevalence in Madagascar may appear to conflict between these papers, upon closer investigation the data sets actually complement each other. Our evidence for Bd’s presence and its widespread incidence is based on multi-year monitoring data carried out through the National Monitoring Program [3] and allied survey efforts, occurring in both the wet and dry season. Our data collected during the same time as Kolby et al.’s sampling (2013–14 wet season), is consistent with their recently published results [1,2] (summarized in Table 1). Therefore, Kolby et al.’s conclusion that our data “highly contradict” those reported in their study is inaccurate. In the 2013–14 wet season, we sampled 569 frogs from 8 locations, of which only 3 samples showed a positive signal for Bd. The positive samples were collected from one individual at each of three sites: Antoetra, Ranomafana, and Ankaratra. While both datasets (the sampling reported in Kolby et al. 2015 and the wet season 2013–2014 sampling reported in Bletz et al. 2015) surveyed numerous individuals and locations across the island, there are some differences in sampling locations. More specifically, one of our positive occurrences came from Antoetra, which was not surveyed by Kolby et al [1]. Two of our positives do come from locations surveyed by both groups: Ranomafana and Ankaratra; if we combine the sub-sites within these locations, the prevalence is 0.0043 and 0.0062 respectively, which falls within the prevalence confidence intervals presented in Kolby et al. [1] (Table 1). This same logic is used by Kolby et al. [1] to show the complementarity of their field survey data and Kolby’s previous work showing Bd’s presence in amphibians imported into the US from Madagascar [4]. Both datasets are consistent with the conclusion that Bd had a very low prevalence during the 2013–2014 wet season. 
 
 
 
Table 1 
 
Summary of published Bd survey data for the 2013–2014 wet season. 
 
 
 
The low prevalence detected in the wet season may likely be explained by seasonality of Bd. Both papers discuss the possibility of seasonal patterns, where Bd prevalence decreases in the warmer, wetter season and increases in the cooler, dryer season due to climatic or so-far un-described environmental factors. This phenomenon is not unusual and several studies have noted a high degree of seasonal variation in the prevalence of Bd (e.g. [5, 6, 7, 8]). In our published study [2] we present preliminary evidence of a seasonal pattern of Bd, showing that prevalence and/or detection was greater in the dryer, cooler season (May-Oct) than the wetter, warmer season (Nov-April) [2]. This seasonal pattern we documented likely explains the lack of detection by Kolby et al. since they sampled only in the wet (and warmer) season. Kolby et al. [1] supplement their individual sampling with the analysis of filtered water from natural habitats. In this case, the lack of Bd detection might be associated with increased water flow due to increased rainfall during the wet season, which could lower the concentration of Bd zoospores to undetectable levels. Additionally, water-filters of natural water bodies have also been found to be less sensitive than direct sampling of amphibians [9,10]. To better understand seasonal variation as well as other factors such as geographic distribution and host species variation, additional sampling across wet and dry season in a uniform and standardize manner will be important. 
 
The major difference between these papers is that we draw from data collected in multiple years and seasons and from a much larger sample of amphibians with further validation using chytrid lineage-based PCR amplification, making our study more comprehensive in nature. Kolby et al. [1] surveyed for a 2-month period, which makes it difficult to make general assumptions about pathogen occurrence from such a small snap shot in time. Our data set thus allows for the conclusion of the “widespread presence of Bd” as we document repeated detections of Bd at geographically distant locations in Madagascar albeit with varying degree of prevalence among sites and seasons. We also have secondary confirmation and validation of Bd’s presence from an independent non-nuclear lineage specific qPCR designed to the Bd mtDNA locus, which is unique to our study. Kolby et al. [1] suggest that Bd in Madagascar cannot yet be described with certainty in part due to the variability of sampling and detection methods. We acknowledge ourselves that our use of various methods may confound some of our findings, such as the seasonal pattern of Bd; however, and importantly, this does not negate or question the evidence for Bd-positive samples collected from Madagascar. 
 
While the data presented by both studies indicate a low prevalence of Bd in Madagascar in 2014, we argue that the additional multi-year data we have collected strongly supports the occurrence of one or more Bd lineage(s) in the samples collected from wild Madagascar amphibians. A similar conclusion was also made by Kolby et al. [4] based on their observations of Bd in wild-caught frogs from Madagascar that were imported into the USA. Definitive and final confirmation of Bd in Madagascar awaits histopathology, isolation of a Bd culture, and/or genome sequencing. These additional analyses can clarify whether Madagascar is facing the panzootic, hypervirulent Bd-GPL or a different (possibly endemic) Bd-lineage. 
 
More importantly, our results may have serious conservation implications. We presented strong evidence that at least one lineage of Bd exists in Madagascar, with increased prevalence at some locations during the dry season. It remains to be understood if this genotype is virulent with respect to the resident anuran fauna and capable of causing population declines. Using the ‘precautionary principle’ in reacting to suspected introductions of novel emerging infectious diseases infecting wildlife [11] it is essential to initiate conservation actions. Continuing ongoing population monitoring of Madagascar's amphibians and pathogen surveillance through the NMP are therefore essential and are a priority of the national amphibian conservation strategy for the country known as ‘A Conservation Strategy for the Amphibians of Madagascar (ACSAM) [12,13]. If it is relatively hypovirulent, it gives conservationists time to engage in mitigation strategies and to plan for the possible (and likely inevitable) arrival of a virulent genotype, which could threaten the diverse, endemic frog communities.

A recent paper by Kolby et al. [1], surveying for Batrachochytrium dendrobatidis (Bd) and ranavirus in Madagascar, presents results for 508 amphibian specimens and 68 water bodies sampled during a 2-month period of the 2013-14 wet season. Kolby et al. [1] did not detect Bd in any of the samples, presenting evidence that add to our understanding of Bd dynamics in Madagascar. Earlier in 2015, we published "Widespread presence of the pathogenic fungus Batrachochytrium dendrobatidis in wild amphibian communities in Madagascar" in the journal PLOS  Scientific Reports [2]. We presented rigorous spatial and temporal surveillance data for 4,155 amphibians sampled across a 10-year period, and used two independent molecular diagnostics to demonstrate the occurrence of a molecular signature of Bd infection at multiple locations across the island. We focus here on solely the Bd results, which directly relate to our published study. While the conclusions of Bd's occurrence and prevalence in Madagascar may appear to conflict between these papers, upon closer investigation the data sets actually complement each other. Our evidence for Bd's presence and its widespread incidence is based on multi-year monitoring data carried out through the National Monitoring Program [3] and allied survey efforts, occurring in both the wet and dry season. Our data collected during the same time as Kolby et al.'s sampling (2013-14 wet season), is consistent with their recently published results [1,2] (summarized in Table 1). Therefore, Kolby et al.'s conclusion that our data "highly contradict" those reported in their study is inaccurate. In the 2013-14 wet season, we sampled 569 frogs from 8  More specifically, one of our positive occurrences came from Antoetra, which was not surveyed by Kolby et al [1]. Two of our positives do come from locations surveyed by both groups: Ranomafana and Ankaratra; if we combine the sub-sites within these locations, the prevalence is 0.0043 and 0.0062 respectively, which falls within the prevalence confidence intervals presented in Kolby et al. [1] (Table 1). This same logic is used by Kolby et al. [1] to show the complementarity of their field survey data and Kolby's previous work showing Bd's presence in amphibians imported into the US from Madagascar [4]. Both datasets are consistent with the conclusion that Bd had a very low prevalence during the 2013-2014 wet season.
The low prevalence detected in the wet season may likely be explained by seasonality of Bd. Both papers discuss the possibility of seasonal patterns, where Bd prevalence decreases in the warmer, wetter season and increases in the cooler, dryer season due to climatic or so-far undescribed environmental factors. This phenomenon is not unusual and several studies have noted a high degree of seasonal variation in the prevalence of Bd (e.g. [5,6,7,8]). In our published study [2] we present preliminary evidence of a seasonal pattern of Bd, showing that prevalence and/or detection was greater in the dryer, cooler season (May-Oct) than the wetter, warmer season (Nov-April) [2]. This seasonal pattern we documented likely explains the lack of detection by Kolby et al. since they sampled only in the wet (and warmer) season. Kolby et al. [1] supplement their individual sampling with the analysis of filtered water from natural habitats. In this case, the lack of Bd detection might be associated with increased water flow due to increased rainfall during the wet season, which could lower the concentration of Bd zoospores to undetectable levels. Additionally, water-filters of natural water bodies have also been found to be less sensitive than direct sampling of amphibians [9,10]. To better understand seasonal variation as well as other factors such as geographic distribution and host species variation, additional sampling across wet and dry season in a uniform and standardize manner will be important.
The major difference between these papers is that we draw from data collected in multiple years and seasons and from a much larger sample of amphibians with further validation using chytrid lineage-based PCR amplification, making our study more comprehensive in nature. Kolby et al. [1] surveyed for a 2-month period, which makes it difficult to make general assumptions about pathogen occurrence from such a small snap shot in time. Our data set thus allows for the conclusion of the "widespread presence of Bd" as we document repeated detections of Bd at geographically distant locations in Madagascar albeit with varying degree of prevalence among sites and seasons. We also have secondary confirmation and validation of Bd's presence from an independent non-nuclear lineage specific qPCR designed to the Bd mtDNA locus, which is unique to our study. Kolby et al. [1] suggest that Bd in Madagascar cannot yet be described with certainty in part due to the variability of sampling and detection methods. We acknowledge ourselves that our use of various methods may confound some of our findings, such as the seasonal pattern of Bd; however, and importantly, this does not negate or question the evidence for Bd-positive samples collected from Madagascar.
While the data presented by both studies indicate a low prevalence of Bd in Madagascar in 2014, we argue that the additional multi-year data we have collected strongly supports the occurrence of one or more Bd lineage(s) in the samples collected from wild Madagascar amphibians. A similar conclusion was also made by Kolby et al. [4] based on their observations of Bd in wild-caught frogs from Madagascar that were imported into the USA. Definitive and final confirmation of Bd in Madagascar awaits histopathology, isolation of a Bd culture, and/or genome sequencing. These additional analyses can clarify whether Madagascar is facing the panzootic, hypervirulent Bd-GPL or a different (possibly endemic) Bd-lineage.
More importantly, our results may have serious conservation implications. We presented strong evidence that at least one lineage of Bd exists in Madagascar, with increased prevalence at some locations during the dry season. It remains to be understood if this genotype is virulent with respect to the resident anuran fauna and capable of causing population declines. Using the 'precautionary principle' in reacting to suspected introductions of novel emerging infectious diseases infecting wildlife [11] it is essential to initiate conservation actions. Continuing ongoing population monitoring of Madagascar's amphibians and pathogen surveillance through the NMP are therefore essential and are a priority of the national amphibian conservation strategy for the country known as 'A Conservation Strategy for the Amphibians of Madagascar (ACSAM) [12,13]. If it is relatively hypovirulent, it gives conservationists time to engage   in mitigation strategies and to plan for the possible (and likely inevitable) arrival of a virulent genotype, which could threaten the diverse, endemic frog communities.