Fvets 08 685877

ORIGINAL RESEARCH
published: 16 July 2021

doi: 10.3389/fvets.2021.685877
Global Patterns of the Fungal

Pathogen Batrachochytrium
dendrobatidis Support Conservation
Urgency
Deanna H. Olson 1*, Kathryn L. Ronnenberg 1 , Caroline K. Glidden 2 , Kelly R. Christiansen 1
and Andrew R. Blaustein 3
1
Pacific Northwest Research Station, United States Department of Agriculture (USDA) Forest Service, Corvallis, OR,
United States, 2 Department of Biology, Stanford University, Stanford, CA, United States, 3 Department of Integrative Biology,
Oregon State University, Corvallis, OR, United States
The amphibian chytrid fungus Batrachochytrium dendrobatidis (Bd) is a skin pathogen

that can cause the emerging infectious disease chytridiomycosis in susceptible species.
It has been considered one of the most severe threats to amphibian biodiversity. We
aimed to provide an updated compilation of global Bd occurrences by host taxon and
geography, and with the larger global Bd dataset we reanalyzed Bd associations with
environmental metrics at the world and regional scales. We also compared our Bd data
compilation with a recent independent assessment to provide a more comprehensive
Edited by:
Rachel E. Marschang, count of species and countries with Bd occurrences. Bd has been detected in 1,375
Laboklin GmbH & Co. KG, Germany of 2,525 (55%) species sampled, more than doubling known species infections since
Reviewed by: 2013. Bd occurrence is known from 93 of 134 (69%) countries at this writing; this
Hudson Alves Pinto,
Minas Gerais State University, Brazil
compares to known occurrences in 56 of 82 (68%) countries in 2013. Climate-niche
Gonçalo Rosa, space is highly associated with Bd detection, with different climate metrics emerging
Zoological Society of London, as key predictors of Bd occurrence at regional scales; this warrants further assessment
United Kingdom
relative to climate-change projections. The accretion of Bd occurrence reports points to
*Correspondence:
Deanna H. Olson the common aims of worldwide investigators to understand the conservation concerns
[email protected] for amphibian biodiversity in the face of potential disease threat. Renewed calls for better
mitigation of amphibian disease threats resonate across continents with amphibians,
Specialty section:
This article was submitted to
especially outside Asia. As Bd appears to be able to infect about half of amphibian
Zoological Medicine, taxa and sites, there is considerable room for biosecurity actions to forestall its spread
a section of the journal
using both bottom-up community-run efforts and top-down national-to-international
Frontiers in Veterinary Science
policies. Conservation safeguards for sensitive species and biodiversity refugia are
Received: 26 March 2021
Accepted: 15 June 2021 continuing priorities.
Published: 16 July 2021
Keywords: amphibian chytrid, Bd, climate associations, emerging infectious disease, fungal pathogen
Citation:
Olson DH, Ronnenberg KL,
Glidden CK, Christiansen KR and
Blaustein AR (2021) Global Patterns of
INTRODUCTION
the Fungal Pathogen
Batrachochytrium dendrobatidis
The Earth is undergoing a “biodiversity crisis,” with population losses and species extinctions
Support Conservation Urgency. occurring at unprecedented rates (1–6). Contributing factors to biodiversity losses are multifaceted
Front. Vet. Sci. 8:685877. and are complicated by species, population- and site-specific differences. Anthropogenic stressors
doi: 10.3389/fvets.2021.685877 such as habitat loss and fragmentation, chemical contamination, introduced species, and climate
Frontiers in Veterinary Science | www.frontiersin.org 1 July 2021 | Volume 8 | Article 685877

Olson et al. Global Bd Occurrence Patterns
change are key factors influencing losses across taxonomic in 2008, hosted by Imperial College, UK. This exportable
groups. Furthermore, there is increasing recognition of health database and mapping application provided the first global-scale
concerns as species are exposed to emerging infectious diseases visualization of an amphibian panzootic (19, 28). The broad use
(EIDs) [e.g., coral disease outbreaks (7, 8); sea star wasting disease of Bd-Maps.net led to the development of an analogous but
(9); bat white-nose syndrome (10, 11); avian West Nile virus (12); more comprehensive online database for Ranavirus, the Global
>50 United States (US) wildlife disease factsheets (13)]. Ranavirus Reporting System [GRRS (31)]. The GRRS inspired
For the especially vulnerable vertebrate class Amphibia, the the development of the more sophisticated amphibian chytrid
Global Amphibian Assessment first reported 32.5% of species disease portal for both Bd and Bsal, AmphibianDisease.org,
as threatened with extinction (14) and that estimate has since hosted by AmphibiaWeb and the University of California at
risen to 40% of species (5, 6), with an increase in concern for Berkeley (41). The Bd-Maps.net dataset is currently in transition
disease impacts (2, 15–20). Amphibian diseases span both lethal to AmphibianDisease.org, providing continuity of and a means
and sublethal multiple-host species infections by microparasites to archive the Global Bd Mapping project from 2007 to present.
such as trematodes (21), bacteria (22), fungi (15, 23, 24), The importance of globally accessible online databases for
protists (25, 26), and viruses (16, 24). Information on amphibian discerning pathogen occurrence patterns is multifold. First, gaps
disease-causing pathogens has increased substantially in the in knowledge are readily apparent by species and location
last two decades, especially relative to field surveillance of and can guide subsequent inventory and monitoring efforts;
taxonomic and geographic patterns of pathogen occurrences at there is support that taxonomic and geographic gaps have
global scales (27–31) and experimental research that illustrates been filled over time (28, 42, 43). Second, world occurrence
species-specific vulnerabilities and interacting factors (24, 32). maps of amphibian pathogens [e.g., Bd (28, 42, 43)] have been
Keeping informed about rapid advances in research and widely used for education and outreach across disciplines, raising
monitoring of amphibian diseases is challenging. Further awareness of potential emerging threat factors and informing
complicating the challenge of tracking host-pathogen patterns conservation efforts. For example, global Bd maps have appeared
of disease threats, anthropogenic processes are linked with in textbooks (44), museum and zoo exhibits (e.g., Panama
amphibian disease dynamics. For example, human-mediated exhibit by Smithsonian Institution; US National Zoo exhibit,
translocation of amphibian EIDs is an increasing concern, Washington, DC; Fungi and Their Diversity exhibit, Hesse
especially for chytridiomycosis, the disease caused by the chytrid Museum, Wiesbaden, Germany), and other multimedia venues
fungal pathogens Batrachochytrium dendrobatidis (Bd) and [(45, 46); e.g., ArgoFilms 2009 film for the Public Broadcasting
B. salamandrivorans (Bsal). These two pathogens are associated System’s Nature TV show, Frogs: The Thin Green Line]. Third,
with amphibian infections across continents and disease-caused knowledge of pathogen occurrences can inform biosecurity
mortality resulting in population losses (14, 17–20, 33, 34). procedures to forestall human-mediated translocation (47–49).
Amphibian chytridiomycosis research has transitioned In addition, global datasets can enable novel metadata analyses
from initial pathogen identification associated with amphibian of specific hypotheses; the global Bd database has contributed to a
mortality [Bd (35); Bsal (23)] to understanding pathogen variety of analyses of host-pathogen and disease-threat dynamics
occurrences and patterns of amphibian losses as knowledge of [e.g., (50–53)]. However, the initial Bd online database had some
host-susceptibility, pathogen strain virulence, and transmission constraints. Occurrences of the disease chytridiomycosis were
scenarios has unfolded (15, 17, 19, 24, 33, 34). Several geographic not tracked, as Bd occurrence studies often do not report the
origins of Bd have been proposed, spanning Asia, Africa, and development of disease signs in sampled animals. Additionally,
North and South America (36–40). A recent genetic analysis we now understand that disease emergence varies with Bd strain
reported east Asia to be a Bd biodiversity hotspot, where the (34, 54), and as of this writing, the Bd database has not recorded
source of Bd was traced to the Korean Peninsula and one lineage Bd lineages with surveillance data, nor have most published
showed the signature of an ancestral population tied to global sources isolated or reported the strain(s) surveyed. Furthermore,
emergence in the early twentieth century (34). Scheele et al. the initial Bd database was not set up to report zoospore loads
(17) estimated that Bd chytridiomycosis has contributed to for samples—these were not being reported in 2007, and even
the declines of 6.5% of amphibian species, and categorized Bd today, not all publications report Bd zoospore loads. Owing to
as one of the most destructive invasive species. Lambert et al. continuing requests for the world Bd database, renewed calls for
(18) concurred that Bd chytridiomycosis irrefutably harmed comprehensive analyses of world-scale data of amphibian disease
amphibians but because their re-analysis could not reproduce threat patterns (18), and significantly increasing reports of Bd
the specific results of Scheele et al. (17), they called for a research and surveillance (24, 29, 30, 32), the global Bd database
more comprehensive approach to quantify the complexities of and web portal warrant maintenance and improved capacity.
interacting amphibian threat factors. The new web portal AmphibianDisease.org (41) is developing
The globalization of amphibian diseases and increasing need to enable broader chytrid data reporting (e.g., strains, captive
for both researchers and natural-resource stewards to understand hosts, eDNA, zoospore loads) and user-friendly data import and
EIDs and their incremental science advances has been aided export functions.
by the advent of amphibian pathogen databases at world- Our aim in this paper is to provide updated summaries of
accessible web portals. The Global Bd Mapping Project began taxonomic and geographic detections and non-detections from
in 2007, with its database of Bd occurrences by host taxon and newly compiled world Bd data through 2019. We use a format for
geographic location going online at the web portal Bd-Maps.net quick comparison to Bd occurrence patterns previously reported

(28, 42, 43). We summarize Bd detection and no-detection sources (newspapers, newsletters, online compilations); and (5)
data from wild and captive specimens, inclusive of wild-caught unpublished contributed datasets.
museum specimens that have been tested for Bd. Geographic Amphibian taxonomy and geographic locations of Bd
patterns of Bd detection are assessed for countries, sites, and sampling per report were examined for reporting consistency
US 5th -field hydrologic unit code (HUC) watersheds (42) which (Supplementary Appendix 1). Taxonomy used herein followed
have been used in some land-management decisions to forestall Frost [(72); Supplementary Appendix 1]. Geographic locations
inadvertent Bd translocation during water draws for fire-fighting with detectable errors (e.g., coordinates clearly outside the study
(55, 56). Furthermore, we examine environmental correlates area) were corrected by consultation with principal investigators,
with Bd site-level occurrence data compiled through 2019. or based on other location information provided in the report.
Several studies have investigated the importance of temperature Laboratory results of Bd analyses were not examined for scientific
and moisture regimes for Bd occurrence, growth, and host- integrity, including analytical sensitivity or accuracy (e.g., sample
infection dynamics using laboratory and local- to landscape- size analyzed; histological or PCR analyses). Hence, we caution
scale analyses (57–71). Here, we examine elevation and climate that “no detection” is not synonymous with Bd absence in a
parameters analyzed previously with the Bd-Maps.net dataset sample, as likelihood of detection can vary with population size,
compiled through 2013 (28) and June 2014 (43) to investigate sample size, Bd prevalence, and analysis method [e.g., (47, 73)].
whether there is stability in Bd predictors (e.g., temperature Data duplication was assessed for studies imported to the Bd-
range at a site). Owing to an abundance of new occurrence Maps database prior to publication that were later identified in
data and the potential climate-change implications for significant literature searches of published papers.
temperature and precipitation metrics with Bd occurrences, we
conduct downscaled analyses of environmental associations with
Bd occurrence for North America, South America, Europe, Bd Occurrence Categorization by Taxa and
Africa, eastern Asia, and Australia. Geography
Lastly, we compare our Bd database tallies by taxon and Bd data were compiled for species, records, sites, watersheds
countries through 2019 with the 2020 results reported by Castro (USA only), and at the region or country level when precise
Monzon et al. (30) who examined the peer-reviewed literature of coordinates or locations were not available. To assess whether Bd
Bd occurrences aggregated by a web-search engine. We combine had ever been detected in an amphibian taxon, the composite
unique taxonomic and country data from Castro Monzon et al. data records were compiled and the taxon was labeled as “Bd
(30) with our findings for an overarching summary of the detected” if there had ever been a single Bd-positive report.
taxonomic and geographic scope of Bd knowledge to date. “Bd not detected” was the usual alternative, however a few
reports have been challenged in the literature due to diagnostics
concerns, and as a precaution those were labeled as “uncertain,”
MATERIALS AND METHODS as were cases where the authors themselves reported an uncertain
result of a diagnostics test. A “record” was a database entry
Bd occurrence data management was based on methods reported for a species at a particular location for a study (28). There
previously, whereas analyses conducted here were intended were multiple records for a location if Bd sampling occurred for
to complement previous assessments (28, 43). To standardize multiple species, sampling occasions, or studies.
methods among years, Bd occurrence database oversight Site-level data compilations were composite records for a
including limited data quality assurance and quality control were common latitude/longitude coordinate, or a specific locality
conducted by the 2007–2019 Bd database manager (KLR). description (28). Site-level Bd occurrence was assigned to one
of three categories: Bd detected; Bd not detected; Bd detection
Data Compilation uncertain. Thus, even if multiple species were sampled for Bd at a
Bd occurrence data were compiled primarily by four methods. unique geographic coordinate, or the location was sampled over
First, an initial dataset was compiled by regional data multiple years, the site was designated “Bd detected” if Bd had
coordinators who submitted project data or reports for their ever been detected at that location for any species in any year.
regions for the 2007 Global Bd Mapping Project, presented in Site-level Bd detected and not-detected data were included in
the first Bd map at the International Bd Conference, Tempe, geospatial analyses described below. Countries were designated
Arizona, USA, in November 2007; this Global Bd Mapping as “Bd detected” based on field or museum specimens sampled or
Project database initiated the development of the Bd-Maps.net collected from the wild. If the only positive sample for a country
web portal (28). Second, Bd surveillance data were directly came from a captive sample, the country was not designated
uploaded to Bd-Maps.net by principal investigators, 2007–2014. as “Bd detected.” An analysis of continental USA watershed-
Third, web-based literature searches were conducted of the main scale Bd occurrence was conducted. As for sites, an individual
international and regional journals reporting on Bd studies watershed was designated as “Bd detected” if Bd had ever been
(Supplementary Appendix 1). Fourth, published or unpublished detected in samples of any species in any year. If a watershed had
reports were sent directly to us (DHO, KLR) for import to the been sampled but Bd had never been detected there, it was labeled
Bd-Maps database. We quantified the number of data sources as “Bd not detected.” For country-scale patterns, a country was
in our 2019 database by five types: (1) peer-reviewed journal labeled as Bd detected or not detected based on the composite
articles; (2) reports; (3) theses and dissertations; (4) online data in the database for that nation.

TABLE 1 | Environmental attributes analyzed for associations with with: 10-year average of lowest monthly precipitation, 10-year
Batrachochytrium dendrobatidis (Bd) occurrence (detection, no detection) across average of highest monthly precipitation, and 10-year average
world sites with Bd sampling compiled through 2019.
of average monthly precipitation. Thus, only 10-year mean
Attribute (units) Description annual precipitation was used in analyses. Similarly, 10-year
mean annual daily temperature was highly correlated (>0.7)
Elevation (m) Altitude above sea level with all other temperature variables (10-year lowest mean
Mean annual precipitation (mm) 10-year mean annual precipitation temperature, 10-year highest mean temperature, 10-year mean
Low average monthly precipitation (mm) 10-year average of lowest monthly low temperature, 10-year lowest mean low temperature, 10-year
precipitation highest mean low temperature, 10-year mean high temperature,
Mean average monthly precipitation (mm) 10-year mean of average monthly 10-year lowest mean high temperature, 10-year highest mean
precipitation
high temperature). Hence, only 10-year mean annual daily
High average monthly precipitation (mm) 10-year average of highest
temperature was used in analyses. The final six covariates used
monthly precipitation
in the models of environmental associations with Bd occurrence
Temperature range (◦ C) 10-year monthly average daily
maximum temperature (tmax)
were: 10-year mean annual precipitation, 10-year mean annual
minus 10-year monthly average daily temperature, 10-year average temperature range, mean
daily minimum temperature (tmin) elevation, minimum elevation within the cell, and maximum
Low average monthly temperature (◦ C) 10-year average of lowest monthly elevation within the cell.
temperature Both presence-only and presence-and-absence (i.e., absence
Mean average monthly temperature (◦ C) 10-year mean of average monthly = no detection) Species Distribution Models (SDMs) were
temperature
evaluated. Given the uncertain nature of true absences, presence-
High average monthly temperature (◦ C) 10-year average of highest only models have been considered more robust (75), whereas
monthly temperature
presence-absence data include the broader dataset assembled for
Low average monthly minimum temp. (◦ C) 10-year average of lowest monthly
minimum temperature
Bd and can be compared with previous models. With global and
Mean average monthly minimum temp. (◦ C) 10-year mean of average monthly
regional subsets of data, using presence-only (detections-only)
minimum temperature data, a maximum-entropy SDM was used to estimate the effect
High average monthly minimum temp. (◦ C) 10-year average of highest of environmental covariates on relative odds of Bd occurrence.
monthly minimum temperature With global and regional datasets, using both detection and no-
Low average monthly maximum temp. (◦ C) 10-year average of lowest monthly detection data, a logistic regression SDM was used to estimate
maximum temperature the effect of environmental covariates on odds of Bd occurrence.
Mean average monthly maximum temp. (◦ C) 10-year mean of average monthly As we expect non-linear relationships between environmental
maximum temperature covariates and probability of Bd occurrence, we transformed
High average monthly maximum temp. (◦ C) 10-year average of highest each covariate (linear, monotonous, deviation, forward hinge,
monthly maximum temperature
reverse hinge, threshold) and used forward selection to select the
transformations that best-explained variation in Bd occurrence.
After variable transformation, a subset selection procedure was
used to determine the best-fit model (i.e., select the final
Environmental Predictors of Bd environmental covariates). As we expected interactions among
Occurrence covariates (e.g., the effect of mean temperature depends on
Analyses of Bd occurrence associations with environmental annual precipitation), we allowed for interactions among all
attributes focused on elevation and 14 climate metrics (Table 1) covariates. To visualize the form of the relationship between
of world sites with Bd sampling compiled through 2019. final model covariates and probability of Bd occurrence (e.g.,
Elevation and climate data were derived from online global unimodal), we plotted model predictions for a range of the
geographic models (Supplementary Appendix 1). World environmental covariate while holding all other environmental
climate data were available for 0.5-degree latitude/longitude covariates at their mean. To visualize the form of interactions
grid cells, hence Bd site-level occurrences were consolidated among covariates, we plotted model predictions for a range of the
per grid cell for consistency with climate data (i.e., per grid environmental covariate while holding the interacting covariate
cell, Bd was either detected or not). This consolidation likely at the 0.25% percentile, mean, and 0.75%, and all other covariates
reduces potential spatial autocorrelation, data collection biases at the mean.
among sampling events, and geographic- and population-level Per SDM, we determined the fraction of total variation
redundancy considerations of the reported source data. accounted (FTVA) for by main parameters in the best-fit model
To avoid collinearity issues, we removed highly correlated [i.e., measure of the parameter contribution to explain variation
predictor variables (74). Consequently, we refined elevation and in Bd occurrence (76)]. We evaluated model performance by
climate data to six parameters for analyses. Three elevation calculating the area under the curve (AUC) which provides an
metrics were determined per 0.5-degree latitude and longitude aggregate measure of model sensitivity (i.e., ability to correctly
grid cell and used in analyses: mean elevation; minimum classify grid cells with Bd detection) and specificity (i.e., ability
elevation; and maximum elevation. For climate metrics, 10- to correctly classify grid cells with no Bd detection). At AUC
year mean annual precipitation was highly correlated (>0.7) = 1.0, the model can perfectly categorize true negatives and

positives, whereas if AUC = 0, it incorrectly categorizes all Western Ghats, India); one caudate family (Rhyacotritonidae, 4
true negatives and positives. If AUC = 0.5, the model makes spp.: Pacific Northwest USA); and one gymnophionan family
predictions equivalent to random guesses. For each SDM, we (Chikilidae, 4 spp.: Northeast India) (Tables 2, 3). However, we
trained the model using 75% of the data and tested the model are aware that in ongoing experiments of Bsal susceptibility
with the remaining 25%. The data were randomly split into a in USA salamanders, wild-caught members of Rhyacotritonidae
training and test set using the R package caTools (77). Final have been screened for Bd prior to use in laboratory trials, and
models were fit with the entire dataset. FTVA and AUC were Bd has not been detected (J. Piovia-Scott, Washington State
calculated using MIAMaxent. University, Vancouver, WA, USA; pers. commun.). Species-level
Finally, to visualize habitat suitability of Bd using analyzed Bd prevalence among families was highly variable (Table 3).
environmental parameters, we calculated model predictions Through 2019, 6 of 55 (11%) Anura families and 3 of 9 (33%)
for each grid location within a global or regional map and Gymnophiona families had no Bd detections among sampled
plotted using model predictions. Importantly, predictions for species (Tables 2, 3).
the presence-only models were scaled to the probability ratio
output [PRO (78)] and can be interpreted as relative habitat Geographic Patterns
suitability of Bd occurrence (79), whereas predictions from the Geographically, our compilation of studies detected Bd in the
presence-absence models represented absolute probability of Bd wild in 88 of 124 (71%) countries sampled through 2019
occurrence. The probability ratio output was log2 (log base 2) (Figures 1, 2 and Supplementary Table 3); these 124 countries
transformed to improve visualization. We excluded data from included 6 countries for which only Bd-negative (no-detection)
Madagascar when fitting the African regional models owing samples with a “country-centroid coordinate” were reported
to some uncertain results for the area in the literature (see (i.e., no location reported: Armenia, Barbados, Central African
Supplementary Table 1 footnote), but we projected the African Republic, Gambia, Iran, Latvia). We recognize that some country
regional models to Madagascar to show potential Bd occurrence. names and boundaries have been dynamic, our intention here is
We also excluded data from Papua New Guinea when fitting the to include recognizable principalities over time. For example, the
regional model because the amphibian fauna has similarities to record of Bd occurrence in North Korea is from the analysis of a
Australia, whereas habitat may be more reflective of Southeast museum specimen reported in 2015 (82) yet the animal had been
Asia; we projected the Asian model to Papua New Guinea collected in the year 1911 when the Korean Peninsula was a single
to predict potential Bd occurrence probability based on Asian political entity; Hong Kong is included separately here although
Bd environmental associations. SDM model predictions were it is now part of China. Our limited quality assurance and quality
calculated in the R package MIAMaxent (80) and global and control of reported data resulted in correction of a small minority
regional predictions were plotted in the R package ggplot2 (81). of location coordinates (Supplementary Appendix 1).
We examined our data by sites (i.e., locations having a
RESULTS common latitude/longitude coordinate for sampling of one
or more amphibian species for Bd infection), compiling Bd
Our Bd occurrence data compilation through 2019 included sampling across 14,647 discrete sites worldwide, inclusive of both
773 sources: 661 peer-reviewed journal articles; 16 reports or wild and captive animals but not including 67 reports from
proceedings; 13 theses and dissertations; 5 online sources; and regions lacking geographic specificity (Supplementary Table 1).
78 unpublished contributed datasets. Worldwide Bd surveillance We could not know if captive animals were infected with Bd
across amphibian taxa and geographies through 2019 showed in the wild or during captivity, so captive sites were not used
advancing knowledge of Bd occurrences, with geographic in subsequent analyses of environmental associations. Excluding
knowledge gaps filled compared to June 2014 (Figures 1–3). Bd captive animals and a small number of results with diagnostic
data were summarized across 33,753 overall sampling records uncertainties, Bd in wild-caught amphibians was detected at
(e.g., sampling effort for a species for a project location in a 5,550 of 14,413 (38.5%) sites. These 14,413 site-level Bd detection
year) with Bd detections and no-detections for wild (including and no-detection data were used in analyses of climatic and
museum specimens of wild-caught animals) and captive animals geographic correlates (below). We mapped Bd detections and
(Supplementary Table 1). no-detections for continental-USA 5th-field HUC watersheds:
Bd detections were reported for 916 of 1,874 (49%) sampled
Taxonomic Patterns watersheds (Figure 3).
Through 2019, our world Bd data compilation showed that Bd
had been detected in 1,294 of 2,412 (54%) amphibian species Environmental Associations With Bd
sampled, and that sampling had been conducted in 29% of Occurrence
all amphibian species (Table 2 and Supplementary Table 2). Consolidation of site-level Bd occurrence data compiled through
Anurans (frogs and toads) had the highest species-level December 2019 into 0.5-degree grid cells resulted in 3,777 grid
prevalence of infection (54.7%) compared to caudates (newts and cells used in global SDMs analyzed with both detection and no-
salamanders: 49.2%), and gymnophionans (caecilians: 29.2%). detection data. Using only the detection data in global SDMs
Through 2019, there were Bd detections in 86% of amphibian (presence-only models), 2,012 grid cells were analyzed.
families. Bd surveys have been reported for all amphibian Using only detection data, the best-fit global model included
families except one anuran family (Nasikabatrachidae, 2 spp.: four environmental parameters and their interactions per grid

FIGURE 1 | World maps of Batrachochytrium dendrobatidis (Bd) occurrences at unique sites from data compiled for the Global Bd Mapping project through: (A) June
2014 (42); and (B) December 2019. Sites shown with Bd detections also may have sampling results with no detection; records with only country-level coordinates are
not mapped.
cell: 10-year mean annual daily temperature (mean temp); (annual precipitation ∗ temp range) + (mean temp ∗ elevation
10-year mean annual precipitation (annual precipitation); max) + (mean temp ∗ temp range) + (mean temp ∗ annual
10-year average temperature range (temp range); and precipitation) + (elevation max ∗ temp range). Mean annual
maximum elevation within the grid cell (elevation max). daily temperature accounted for the highest fraction of total
The relative probability of Bd occurrence was a function variation in probability of Bd occurrence (0.713), with annual
of mean temp + annual precipitation + temp range + precipitation accounting for the second highest fraction
elevation max + (annual precipitation ∗ elevation max) + (0.218; Table 4).

FIGURE 2 | Regional maps of Batrachochytrium dendrobatidis (Bd) occurrences at unique sites from data compiled for the Global Bd Mapping project: (A) North
America; (B) South America; (C) Europe; (D) Africa; (E) Eastern Asia; (F) Australia; and (G) New Zealand. Sites shown with Bd detections also may have sampling
results with no detection; records with only country-level coordinates are not mapped.

FIGURE 3 | United States 5th -field hydrologic unit code watershed maps of Batrachochytrium dendrobatidis (Bd) occurrences from data compiled for the Global Bd
Mapping project through: (A) June 2014 (42); and (B) December 2019. Watersheds shown with Bd detections also may have sampling results with no detection.

TABLE 2 | Global Batrachochytrium dendrobatidis (Bd) detections in amphibians as compiled through December 2019.
Species Families
Bd detected Tested Prevalence (%) Total species Bd detected Tested Prevalence (%) Total families
Anura 1,153 2,106 54.7 7,311 49 55 89.1 56

Caudata 127 258 49.2 762 8 8 100 9
Gymnophiona 14 48 29.2 210 6 9 66.7 10
Total 1,294 2,412 53.6 8,283 63 72 85.9 75
In the SDM derived from both detection and no-detection model with less sensitivity and specificity than the presence-
data at the global scale, the best-fit model included two only model.
environmental parameters and their interactions: mean temp and The maps of Bd habitat suitability (presence-only model) and
temp range. Probability of Bd occurrence was a function of mean probability of Bd occurrence (presence-absence model) from
temp + temp range + mean temp ∗ temp range. Variation in our best-fit global models (Figure 6) were reflective of our dot
probability of Bd occurrence was primarily described by mean distribution of Bd occurrences (Figure 1). Areas of heightened
temperature (0.969; Table 4). likelihoods of Bd occurrence included mesic mid-latitude and
In the global SDMs, the relative probability (presence-only coastal influences, especially when considering the presence-only
model) or absolute probability (presence-absence model) of model. North-temperate, interior-continental, and arid zones
Bd occurrence responded non-linearly to each environmental had lowest Bd probabilities.
covariate (Figures 4, 5). A more detailed representation of In regional SDMs, relative probability (using detection-only
these figures showing frequency of observations for each data) and absolute probability of Bd (using detection and no-
environmental covariate and kernel-estimated data density detection data) also responded non-linearly to environmental
showing the sampling effort (no. grid cells) are depicted in covariates. For regional SDMs using detection-only data, 10-year
Supplementary Figures 1, 2. mean annual daily temperature (mean temp) was retained in
In the SDM derived from detection-only data, there was all final models (Table 5), with mean temperature capturing the
a unimodal relationship between 10-year mean annual daily most variation in relative probability of Bd occurrence in North
temperature and relative probability of Bd occurrence, peaking America, South America, and Europe (Table 5). In all three
at ∼12–13◦ C, when holding all other covariates at their regional models, shape of the response of Bd occurrence reflected
mean (Figure 4A). Relative probability of Bd occurrence had that of the regional model (unimodal with a peak around ∼12–
a more complex relationship with mean annual precipitation, 13◦ C). Notably, in North America and South America, the effect
showing an initial modal maximum at ∼1,200–1,400 mm, a of mean temperature changed with temperature range and max
dip at ∼3,000 mm, before increasing around 4,000 mm, when elevation, respectively. In Africa, max elevation in a ∼55-km
holding all other covariates at their mean (Figure 4B). Relative grid cell (max elevation) and annual precipitation contributed
probability of Bd also formed a unimodal relationship with the most variation to relative probability of Bd occurrence
maximum elevation, peaking at ∼4,000 m, when holding all other (0.413 and 0.339, respectively; Table 5). When all other covariates
covariates at their mean (Figure 4C). The relationship between were held at their mean, relative probability of Bd occurrence
temperature range and Bd occurrence plateaued between ∼8 increased with max elevation until a plateau around 2,000 m and
and 21◦ C, with a stark decrease in relative probability of Bd linearly increased with annual precipitation. However, the effect
occurrence at both cooler and warmer temperatures, when of precipitation was dependent upon mean temperature. In Asia,
holding all other covariates at their mean (Figure 4D). To annual precipitation contributed the most variation in relative
visualize interactions among covariates, model predictions were probability of Bd occurrence followed by mean temperature
plotted with interacting covariates held at the 0.25% percentile, (0.626 and 0.374, respectively; Table 5). The relative probability
mean, and 0.75% percentile (Supplementary Figures 3A,B). The of Bd occurrence increased with annual precipitation, whereas
AUC for the global presence-only model was 0.86, indicating a model response to mean temperature followed a unimodal
model with high sensitivity and specificity. pattern reflective of the global presence only model. In Australia,
In the presence-absence model, at the average temperature temperature range contributed the most variation to relative
range, probability of Bd detection increased once mean probability of Bd occurrence (0.675, Table 5). When all other
temperature increased past 0◦ C and decreased when mean covariates were held at their mean, relative probability of Bd
temperature exceeded ∼20◦ C (Figure 5A). At the mean temp, followed a unimodal response to temperature range, with relative
probability of Bd detection tended to increase as temperature probability of Bd occurrence peaking at a temperature range of
range increased, with a peak around 18◦ C (Figure 5B). ∼10◦ C. Model predictions from the presence-only models (PRO
Interactions between mean temperature and temperature of Bd occurrence) were projected in regional maps (Figure 7).
range are plotted in Supplementary Figure 4. The AUC for AUC for regional presence-only models ranged from 0.79 to 0.93
the presence-absence global model was 0.63, indicating a (Table 5).

TABLE 3 | Family-level summary of Batrachochytrium dendrobatidis (Bd) TABLE 3 | Continued

detections among species sampled for Bd.
Family No. spp. Bd No. spp. Spp. Total spp. in
Family No. spp. Bd No. spp. Spp. Total spp. in detected tested prevalence family
detected tested prevalence family
Ranixalidae 2 6 0.33 18
Anuraa Rhacophoridae 27 62 0.44 434
Allophrynidae 0 1 0.00 3 Rhinodermatidaeh 2 2 1.00 3
Alsodidae 8 17 0.47 26 Rhinophrynidae 0 1 0.00 1
Alytidae 7 9 0.78 12 Scaphiopodidae 4 7 0.57 7
Aromobatidae 17 25 0.68 128 Sooglossidae 0 3 0.00 4
Arthroleptidae 52 96 0.54 149 Telmatobiidae 21 25 0.84 63
Ascaphidae 1 2 0.50 2 Total Anura 1,153 2106 0.55 7,311
Batrachylidae 6 7 0.86 12 Caudatai
Bombinatoridae 4 5 0.80 8 Ambystomatidae 20 24 0.83 37
Brachycephalidae 9 12 0.75 74 Amphiumidae 2 3 0.67 3
Brevicipitidae 1 7 0.14 37 Cryptobranchidae 3 3j 1.00 4
Bufonidae 102 204 0.50 630 Hynobiidae 3 20 0.15 85
Calyptocephalellidae 2 2 1.00 5 Plethodontidae 70 150 0.47 491
Centrolenidae 18 33 0.54 156 Proteidae 4 4 1.00 9
Ceratobatrachidae 2 14 0.14 102 Salamandridae 22 50 0.44 128
Ceratophryidae 7 7 1.00 12 Sirenidae 3 4 0.75 5
Conrauidae 2 5 0.40 6 Total Caudata 127 258 0.49 762
Craugastoridaeb 97 185 0.52 874 Gymnophionak
Cycloramphidaec 13 19 0.68 36 Caeciliidae 0 7 0.00 43
Dendrobatidae 33 49 0.67 203 Dermophiidae 2 5 0.40 14
Dicroglossidae 17 46 0.37 215 Herpelidae 2 3 0.67 10
Eleutherodactylidae 36 68 0.53 230 Ichthyophiidae 0 5 0.00 57
Heleophrynidae 5 5 1.00 7 Indotyphlidae 1 7 0.14 24
Hemiphractidae 20 225 0.80 118 Rhinatrematidae 0 2 0.00 14
Hemisotidae 0 2 0.00 9 Scolecomorphidae 2 3 0.67 6
Hylidaed 190 255 0.74 734 Siphonopidae 3 7 0.43 26
Hylodidae 23 29 0.79 47 Typhlonectidae 4 9 0.44 14
Hyperoliidae 68 107 0.64 228 Total gymnophiona 14 48 0.29 210
Leiopelmatidae 1 4 0.25 4
Total species in family as of November 2020 based on Frost 2020 (72).
Leptodactylidae 49 87 0.56 231 a Family Nasikabatrachidae (Western Ghats of India, with 2 species) not yet sampled.
Limnodynastidae 14 23 0.61 43 b Includes former family Strabomantidae.
c Not including family Rhinodermatidae, listed separately below; genus Proceratophrys
Mantellidae 1e 77 0.01 233
moved to Odontophrynidae.
Megophryidae 5 31 0.16 280 d Not including species split off into new family Pelodryadidae.
Micrixalidae 1 2 0.50 24 e Twelve species with uncertain positive tests in Madagascar, 1 positive captive animal in
Microhylidae 26 96 0.27 703 the USA, see Supplementary Material and discussion of Madagascar results in the text.
f Genus Proceratophrys moved from Cycloramphidae to Odontophrynidae.
Myobatrachidae 20 37 0.54 89
g New family split from Hylidae.
Nyctibatrachidae 2 2 1.00 39 h New family split from Cyclorhamphidae.
Odontobatrachidae 0 1 0.00 5 i Family Rhyacotritonidae (Pacific Northwest United States, 4 spp.) may have been
Odontophrynidaef 7 11 0.64 50 sampled, but results have not yet been reported.
j Prior versions of this table treated Cryptobranchus alleganiensis alleganiensis and C. a.
Pelobatidae 3 4 0.75 5
bishopi as separate species, but the current taxonomy regards them as one species.
Pelodryadidaeg 35 69 0.51 219 k Family Chikilidae (Northeast India, 4 spp.) has not been sampled.
Pelodytidae 0 1 0.00 5
Petropedetidae 6 11 0.54 13
Phrynobatrachidae 20 40 0.50 95
Phyllomedusidae 18 25 0.72 67 Like regional models using detection-only data, for regional
Pipidae 21 26 0.81 41 SDMs using detection and no-detection data, mean temperature
Ptychadenidae 12 23 0.52 64 was retained in all final models (Table 6), with mean temperature
Pyxicephalidae 20 28 0.71 85 capturing the majority of variation in relative probability of
Ranidae 96 166 0.58 419 Bd detection in North America, South America, and Europe
(Table 6). Furthermore, mean temperature accounted for the
(Continued) majority (or all) of variation in Bd detection in Africa, Asia, and

TABLE 4 | Fraction of total variation accounted for (FTVA) by each variable in

best-fit global species distribution models (SDMs) of Batrachochytrium
dendrobatidis (Bd) occurrence from data compiled through December 2019.
Model Variable FTVA
Presence-only global 10-year mean annual daily temperature 0.713

(AUC = 0.86) 10-year mean annual precipitation 0.218
maximum elevation in a 55-km grid cell 0.046
10-year average temperature range 0.023
Presence-absence global 10-year mean annual daily temperature 0.969
(AUC = 0.63) 10-year average temperature range 0.031
The presence-only model used Bd detection data and the presence-absence model
used Bd detection and no-detection data. AUC, area under the curve, measure of
model sensitivity (ability to correctly classify 0.5-degree latitude/longitude grid cells with
Bd detection).
FIGURE 5 | Absolute probability of Batrachochytrium dendrobatidis (Bd)

occurrence for the environmental attributes (A) mean temperature and (B)
temperature range from the global presence-absence best-fit species
distribution model.
presence-absence model, where probability of Bd increased until

∼0◦ C, plateaued, and then decreased after ∼20◦ C. In Europe
and Asia, probability of Bd detection to mean temperature
was more typically unimodal and similar in shape to the
global presence-only model: probability of occurrence increased
up to ∼12–13◦ C and then decreased. In North America,
temperature range contributed a substantial portion of variation
to probability of Bd detection (0.355; Table 6). In South America,
max elevation also contributed a small portion of variation to
probability of Bd detection (0.079; Table 6). In Australia, annual
precipitation accounted for almost half of variation in probability
of Bd detection (0.433; Table 6). Regional model predictions
(probability of Bd detection) from the Bd detection and no-
detection model were projected in maps (Figure 8). AUC for
regional presence-absence models ranged from 0.495 to 0.724
(Table 6).
DISCUSSION
Our results provide new insights into a pathogen that has
emerged as one of the most severe threats to amphibian
biodiversity (2, 34), representing a “paradigm shift in our
understanding of how emerging infectious diseases contribute
to global patterns of biodiversity loss” (19). Our compilation
FIGURE 4 | Relative habitat suitability of Batrachochytrium dendrobatidis (Bd) of Bd sampling shows increasing taxonomic scope of sampled
occurrence (probability ratio output) for each environmental attribute [(A) mean families and species, increased incidence of species infection,
temperature; (B) annual precipitation; (C) maximum elevation; (D) temperature and increased geographic occurrence. Below, we combine
range] from the global presence-only best-fit species distribution model. Each
our findings with those of Castro Monzon et al. (30) to
environmental attribute marginal-response plot is calculated while holding all
other covariates at the mean.
yield a more comprehensive tally of total Bd occurrence
patterns. In addition, our analyses support emergence of
new key climate predictors of Bd occurrence and geographic
variance in climate metrics associated with the occurrence of
Australia (Table 6). In North America, South America, Africa Bd. These support new hypotheses for downscaled analyses
and Australia, response of probability of Bd detection to mean of regional contexts associated with pathogen occurrence
temperature followed a hinge-type pattern similar to the global patterns and renewed efforts for species and microrefugia

FIGURE 6 | Global maps of predicted Batrachochytrium dendrobatidis (Bd): (A) habitat suitability derived from the best-fit presence-only species distribution model
(log2 -transformed [log base 2] probability ratio output, area under curve [AUC] = 0.86); and (B) probability of Bd occurrence from the best-fit presence-absence
species distribution model (AUC = 0.62). Both maps were derived using Bd data compiled through December 2019.
identification and management for more effective conservation. even together, the compilation is incomplete; past reports that
The large number of Bd occurrence reports accruing over have not yet been included in these composite summaries are
time (Table 2 and Supplementary Table 2) points to the continually brought to our attention. Below, we also compared
common aims of worldwide investigators to understand our compilation by taxonomy and geography to the earlier
the taxonomic and geographic scope of Bd infections and Bd-Maps database (28, 42) to estimate accretion of knowledge
the underlying global conservation concerns for amphibian over time.
biodiversity in the face of potential disease threat. A Castro Monzon et al. (30) reported Bd sampling across
better understanding of linkages between these pathogen 71 amphibian families; these families were included in our
occurrence patterns and amphibian disease threats are needed compilation, although in our data compilation we added
across continents. an unpublished captive report of a detection in Mantellidae
(Mantella sp.; Supplementary Table 2), bringing the total to
Global Bd Occurrence 72 families. Also, if the caudate family Rhyacotritonidae is
To gain a more comprehensive summary of Bd occurrence by included (Piovia-Scott, pers. commun.), then Bd is now known
taxonomy and geography, we compared our Bd data compilation to have been sampled in a total of 73 amphibian families.
through 2019 with the web search conducted by Castro Monzon Since 2014, 6 additional families (including Rhyacotritonidae,
et al. (30), who independently compiled Bd occurrences from plus Pelodryadidae which has since split off from Hylidae)
the published literature through mid-2020. We summed unique have been sampled for Bd (42). Compared to amphibian
reports from our analyses and Castro Monzon et al. (30) to family tallies of Bd occurrence from 2014 (42), family patterns
calculate the total numbers of families, species, and countries changed slightly over the ∼5 years. For example, for species-
with Bd detections from these combined datasets. Our two data level prevalence in families with over 100 species sampled,
compilation approaches differed in data sources. Castro Monzon in 2014, Bd occurrence was highest in hylids (60%), ranids
et al. (30) cited 554 papers produced by their web search. (59%), craugastorids (57%), and bufonids (44%), whereas in
In comparison, our data compilation through 2019 included 2019, Bd occurrence was higher in hylids + pelodryadids (67%)
773 total sources, including sources outside the peer-reviewed and hyperoliids (64%), followed by ranids (58%), craugastorids
literature. Together, a more comprehensive perspective is gained (52%), bufonids (50%), and plethodontids (47%). Species tested
by combining our two approaches, but we acknowledge that nearly doubled for four relatively under-sampled families in 2014

(Microhylidae, Rhacophoridae, Hynobiidae, Typhlonectidae). TABLE 5 | Final best-fit model covariates of regional presence-only species
Overall, knowledge of previously under-sampled species and distribution models (North America, South America, Europe, Africa, Asia,
Australia), including regional model area under the curve (AUC), and fraction of
families grew, suggesting there have been focal efforts targeting
total variation accounted for (FTVA) for each variable.
taxonomic knowledge gaps. Previously, amphibian family has
been reported to be a strong predictor of Bd infection Region and final best-fit model Variable FTVA
status, including severity of infection and development of
chytridiomycosis (51, 83). North America (AUC = 0.89)
Castro Monzon et al. (30) reported 1,062 of 1,966 (54%) Mean temp + annual precipitation + Mean temp 0.910
elevation max + mean temp*elevation max Annual precipitation 0.034
Bd-infected species from their web search. We report on Bd
occurrences in 1,286 of 2,389 (54%) amphibian species, an Elevation max 0.056
additional 423 species but a comparable rate of infection. South America (AUC = 0.91)
Upon closer comparison of these two datasets, we found Mean temp + elevation max + temp Mean temp 0.667
that Castro Monzon et al. (30) included Bd sampling range + mean temp*elevation max Elevation max 0.292
in 126 different, additional species (excluding hybrid Temp range 0.042
species and uncertain species designations, i.e., Genus sp.: Europe (AUC = 0.79)
Supplementary Appendix 2) which were not in our data Mean temp + annual precipitation Mean temp 0.833
compilation. Adding these species to our total, Bd has been Annual precipitation 0.167
detected in 1,375 of 2,525 (55%) species sampled and compiled Africa (AUC = 0.90)
from both datasets. Our knowledge of world Bd surveillance Elevation max + annual precipitation + Elevation max 0.413
across species has more than doubled since the 2013 paper by mean temp + temp range + annual
Annual precipitation 0.339
Olson et al. (28) where Bd detection was reported in 516 of 1,240 precipitation*mean temp
Mean temp 0.183
(42%) sampled species. The incidence of known species infection Temp range 0.064
has increased by 13% over this relatively short time period,
Asia (AUC = 0.91)
2013 to 2020.
Annual precipitation + mean temp Annual precipitation 0.626
Geographically, we compared our Bd-in-the-wild occurrence
Mean temp 0.374
results by country with the web search conducted by Castro
Australia (AUC = 0.93)
Monzon et al. (30) (Supplementary Appendix 2). In comparison
Temp range + elevation max + mean Temp range 0.675
to our reported Bd sampling in 124 countries through 2019 temp + annual precipitation Elevation max 0.211
(88 with detections), they reported sampling in 119 countries
(86 with detections) through early May 2020. Our country
Mean temp 0.048
lists differed (Supplementary Appendix 2) in that we reported
Bd sampling in 13 countries that they did not include, and
they reported Bd sampling from 9 countries that we did not
include; hence, our datasets compiled different reports for 22
countries. Adding their 9 additional countries with 4 additional Environmental Associations of Bd
Bd detections to our sample (124 countries) yields 133 countries Occurrences
with known Bd sampling, with Bd detected in a total of 92 Our analyses of environmental associations with Bd occurrence
countries. From a very recent publication, we became aware of Bd through 2019 further support the importance of climate-niche
sampling in one additional country that had not been included space for this pathogenic aquatic fungus [e.g., (28, 43, 57, 59–61,
in either compilation, the Kingdom of Bhutan [Bd not detected 63–66)]. In our global SDMs using the largest dataset to date with
(84)]. Adding this to the grand total, Bd has been detected in both detection and no-detection data, mean temperature was
wild samples from 93 of 134 (69%) countries to our knowledge the most important environmental correlate of Bd occurrence,
at this writing. This compares to Bd detections in 71 of 105 (68%) and accounted for 97% of the variation in Bd occurrence. In
sampled countries in 2014 (42) and in 56 of 82 (68%) countries the more predictive global model using Bd detection-only data,
in 2013 (28). mean temperature accounted for 71% of the variation in Bd
For comparisons of site-level knowledge gain over time, occurrence whereas annual precipitation accounted for 22%.
using similar methods, Olson et al. (28) reported compilation Although the relationship between probability of Bd occurrence
of Bd sampling data at 4,281 sites, Xie et al. (43) reported and mean temperature (Figure 4A) is consistent with our
5,166 site-level records through June 2014, and herein we report knowledge of temperature constraints on Bd growth [e.g., (57)],
14,647 total sites. Site-level Bd data more than tripled since the pattern of Bd occurrence with annual precipitation is not
our initial report in 2013. In data compiled through June 2014, easily reconciled (Figure 4B) and may result from: a sampling
Bd sampling had occurred in 923 total USA 5th -field HUC artifact of Bd occurrence patterns in our dataset—perhaps
watersheds, with 560 (60%) watershed having Bd detections relating to underlying, complex host-pathogen interactions
(42). In comparison, by December 2019, our knowledge of with temperature; an artifact of our 0.5-degree latitude and
Bd sampling had doubled across US watersheds, with 1,874 longitude grid cells being the unit of analysis, within which
watersheds sampled, and Bd detections were reported for 916 heterogeneous precipitation patterns are likely; regional diversity
(49%) watersheds. in Bd environmental associations; or Bd lineage effects. Sampling

FIGURE 7 | Regional predictions of Batrachochytrium dendrobatidis (Bd) habitat suitability from our best-fit presence-only species distribution models
(log2 -transformed [log base 2] probability ration output, area under curve values in Table 5) using Bd data compiled through December 2019 for: (A) North America;
(B) South America; (C) Europe; (D) Africa; (E) eastern Asia; and (F) Australia.
intensity across each covariate (Supplementary Figure 1) and TABLE 6 | Final best-fit model covariates of regional presence-absence species
interactions among covariates are important considerations distribution models (North America, South America, Europe, Africa, Asia,
Australia), including regional model area under the curve (AUC), and fraction of
to fully understand the role of the different parameters
total variation accounted for (FTVA) for each variable.
across their range extents (Supplementary Figures 3A,B). Mean
temperature was not the top predictor in previous presence- Region and final best-fit model Variable FTVA
absence models of globally compiled data, as temperature range
North America (AUC = 0.545)
had previously emerged as a highly predictive covariate (28,
Mean temp + temp range Mean temp 0.645
43). Our approach of examining correlations among similar
Temp range 0.355
climate metrics and consolidating to fewer potential covariates
South America (AUC = 0.714)
prior to SDM analyses may have contributed to this difference.
Mean temp + temp range Mean temp 0.921
Additionally, the change in surveillance patterns geographically,
Elevation max 0.079
2014 to 2019 (Figures 1, 3), may have led to emergence of
Europe (AUC = 0.495)
different predictor covariates at the global level. Previously,
Mean temp Mean temp 1.00
sampling bias favoring species or locations in the United States,
Africa (AUC = 0.724)
for example, may have led to skewed environmental associations
during global assessments. With many former data gaps filled by
Asia (AUC = 0.664)
the time of our 2019 data snapshot, this single-region bias is a
lesser concern. However, the different covariates that emerged in
Australia (AUC = 0.664)
our regional SDMs support the unique role that different climate
Mean temp + annual precipitation Mean temp 0.57
metrics in each area may have on emerging Bd patterns.
Differences among regional SDMs with our more robust
2019 dataset support the importance of additional downscaled
analyses to understand potential geographic context-specific
patterns of Bd emergence. In the presence-only regional close-second predictor in Asia (0.37); (2) annual precipitation
models (Table 2), the models with highest sensitivity: (1) was a top predictor or close-second predictor in Asia (0.63)
mean temperature dominated Bd predictors in North America and Africa (0.34); (3) maximum elevation was a top predictor
(0.91), Europe (0.83), and South America (0.67), and was a or close-second predictor in Africa (0.41) and Australia (0.21);

FIGURE 8 | Regional predictions of Batrachochytrium dendrobatidis (Bd) absolute probability of Bd occurrence from our best-fit presence-absence species
distribution models (area under curve values in Table 6) using Bd data compiled through December 2019 for: (A) North America; (B) South America; (C) Europe; (D)
Africa; (E) eastern Asia; and (F) Australia.
and (4) temperature range was a top predictor in Australia DOI = https://n2t.net/ark:/21547/DsA2; updated public data
(0.67). Per continent, scenarios could be developed to add 2015 to 2019, DOI = https://gcc02.safelinks.protection.outlook
more specificity to these regional patterns and explore their .com/?url=https%3A%2F%2Ffanyv88.com%3A443%2Fhttps%2Fn2t.net%2Fark%3A%2F21547%2
complexities. For example, the high gradients in temperature FDsM2&data=04%7C01%7C%7Cd307ffc12a5e407dad1e08d92e
ranges across coastal-to-interior Australia were a significant a75cce%7Ced5b36e701ee4ebc867ee03cfa0d4697%7C0%7C0%7
contributing factor to the downscaled models already developed C637592118618600952%7CUnknown%7CTWFpbGZsb3d8eyJ
by Murray et al. (65). Regional patterns warrant additional WIjoiMC4wLjAwMDAiLCJQIjoi V2luMzIiLCJBTiI6Ik1haWwi
study at smaller spatial scales, and relative to additional LCJXVCI6Mn0%3D%7C1000&sdata=hBkSl%2FScrh83TNo1H
interactions among environmental covariates. At smaller scales, eQkG7QTovBwrI3OJ8E8RblAXFY%3D&reserved=0), showed
relationships between temperature and the pathogen biology, significant Bd-knowledge gaps have been filled across amphibian
host biology, and their interplay could be further explored families and countries, with updated world occurrence patterns
(85). Spatially downscaled approaches could have ramifications likely to inform novel research investigations and conservation
for the direction of regionally specific conservation actions to actions. In combination with additional more-recent data
forestall disease threat, such as site-specific efforts to manage including the independent Bd-data compilation by Castro
microclimate conditions (86). Monzon et al. (30), these composite surveillance efforts are
an unparalleled accomplishment by a vast global community
Research and Management Implications of natural-resource managers and amphibian scientists. With
Although Bd is globally distributed, occurring on every about half of sampled amphibian species being infected and
continent with amphibians, support is growing for Bd to have Bd occurring at <40% of sites sampled, the need for effective
expanded its global range relatively recently (34, 36–39). Our pathogen biosecurity-and-mitigation is paramount to reduce
analyses do not quantify recent spread but are an update further Bd transmission and losses of vulnerable hosts.
of knowledge of Bd occurrence. Despite its present broad Management actions to curtail Bd can take several paths.
occurrence, Bd is clearly not ubiquitous across amphibian Sampled taxa and geographies without Bd detections warrant
taxa or geographies, likely owing to a complex combination continued assessments for novel pathogen detection and disease
of transmission dynamics, host susceptibilities to infection, threat, and consideration for heightened biosecurity to forestall
and pathogen environmental associations. Our updated Bd- human-mediated pathogen translocation pathways [e.g., global
occurrence data (downloadable from the AmphibiaWeb portal, trade markets (19, 20, 34)]. Priorities for Bd monitoring and
AmphibianDisease.org: previously released public data to 2014, exclusion include sensitive species habitat strongholds, both

micro- or macro-scale refugia, and broader Bd-free geographies antifungal properties that afford protection to host amphibians
that are habitat for high- or unique-diversity communities from adverse effects of Bd infections (93–96). Despite advances
(49). Biodiverse locations of Africa, Asia, and South America in understanding amphibian immune responses, this is an active
with fragmented Bd occurrences warrant attention for elevated area of ongoing research that is likely to offer new insights to
biosecurity to protect rich endemic fauna from potential Bd control (97–100). Such ongoing research interleaves with
disease threats. In more highly sampled areas, such as the novel Bd-management opportunities (e.g., microbiota vaccines)
USA, similar patterns of patchy Bd occurrences are evident and could help forestall mass-mortality events in susceptible
at finer geographic scales (Figures 2, 3), supporting the value species. In situations where at-risk taxa appear threatened, the
of downscaled biosecurity efforts to reduce human-mediated ability to develop rescue measures and learn from their efficacy
spread into current apparent host macro- or micro-refugia can inform later efforts [e.g., (101, 102)]. Furthermore, relative
from the pathogen, such as Bd-free watersheds (Figure 3). efficacy of additional conceptual field mitigations for amphibian
Watersheds could be a practical spatial unit for aquatic-pathogen EIDs have been qualitatively evaluated but warrant field trials to
management (28), as a variety of water resources are often see if they can alter site-scale Bd infection dynamics [e.g., habitat
managed by watershed boundaries. attributes such as shading and water temperature management—
As supported by O’Hanlon et al. (34), listing Bd as a essentially microclimate manipulations—could alter site-scale Bd
notifiable disease by the World Organization for Animal infection dynamics (86)]. Adaptive management and learning
Health (OIE) in 2008 has had little effect on human-mediated from such field trials is needed to advance effective mitigations
translocation. Instead, reliance on local, regional, and national with knowledge of the risks and benefits they may entail. Each
jurisdictions with Bd-clean trade, transportation, and fieldwork intervention that might safeguard species from severe infection
procedures is a more effective biosecurity strategy. Pathogen merits study for efficacy and practicality as part of research
biosecurity approaches for wild amphibians include species- and conservation trials, while biosecurity measures could stall
and geographic-specific risk assessment and prioritization of inadvertent spread.
between-site measures that are applicable for public and natural-
resource manager implementation (49, 87, 88). Enforceable Next Steps
regulations protecting the national heritage of non-game wildlife Our newly updated dataset points to the broad human
in separate jurisdictions could be considered for development dimensions of Bd surveillance, and specifically the contributions
with a focus on wildlife health, clean trade, and management of to our current understanding of global Bd occurrences from
injurious invasive species [e.g., salamander import restrictions a broad world community. Our 769 data sources show
to forestall Bsal transmission to the USA and Canada (89, 90); international partnerships have been established between local
inclusion of Bd on aquatic invasive species lists of injurious faunal and land-management experts and personnel from
species, with hygiene measures promoted across geographic numerous universities and institutions to pursue Bd surveillance
boundaries]. Biosecurity guidance is available for some types (Supplementary Table 2). Such co-production underscores the
of field activities that could have broad implementation [e.g., local and global interest in Bd occurrence, spread, and threat.
large equipment use at field sites (48); water draws for wildfire Although published data from peer-reviewed journals dominate
management (55, 56)], as well as for amphibian research our compilation of data sources (86%), incentives are needed to
where within-site methods are applicable (47). These biosecurity improve this rate to ensure data quality assurance of sampling
measures transcend application to Bd and are relevant for and analytical procedures. Publishing could be promoted prior
cross-taxonomic pathogens, parasites, and invasive species. to graduate student defenses, or with permitting procedures.
These actions are consistent with One Health approaches, to Communication of the global Bd database move to
recognize the interconnections among people, species, and our AmphibianDisease.org and encouragement of ongoing project
shared environment, and to work collaboratively to optimize plans as well as data imports is important going forward, as
health of each component [e.g., CDC (91)]. Building upon the metadata analyses of global data compilations can be important
increasing public awareness of the linkages between human and for hypothesis testing and pattern revelations [e.g., (50–53)].
wildlife pathogens and their diseases resulting from the recent The consistency of the results presented here from joint web-
coronavirus pandemic could bolster biosecurity implementation portal data imports compilation and literature searches was
for broader One Health aims. comparable to a more technical web search of journals (30).
The value of forestalling human-mediated spread of Bd is Clearly, the combined approaches yield a more comprehensive
several-fold. First, we are still learning about the pathogen picture, and may be useful for a more complete Bd dataset at
and its complex context-dependent interactions with hosts, AmphibianDisease.org into the future.
environmental conditions, and other threats factors [e.g., (24, With initiation of the new web portal AmphibianDisease.org,
32)]. Bd-strain differences have emerged as a key element in host- we anticipate phases of Bd database updates over the near term.
pathogen dynamics [e.g., (34, 54, 92)]. Our current Bd dataset First, as novel Bd-data imports have been made already to the
does not record Bd strains, which sets a new challenge for the new portal from early users and local projects, cross-checking
global community conducting surveillance to contribute strain between the Bd-Maps database upload and these datasets will
data to the next phase of the Bd web portal (41). Also, there is be needed to reconcile redundancies. Data gaps discovered from
increasing information about the role of microbiotic community 2019 and earlier should be addressed as they are identified,
interactions on amphibian skin, with some bacteria having including data sources uncovered by Castro Monzon et al. (30)

that we had not included in our data compilation for this manuscript development. All authors provided critical feedback
paper. Ongoing efforts to educate researchers on new procedures and contributed to the final manuscript.
to archive their data will be critical. The Amphibian Disease
database may include records of captive animals and museum
specimens, allowing unpublished data to be added. Additionally, FUNDING
Bd occurrence patterns derived from eDNA or fomite samples
This project was supported by the US Department of Agriculture,
provide additional data sources requiring special attention for
Forest Service, Pacific Northwest Research Station.
compilation, especially as new multi-taxonomic community-
based analyses are conducted [e.g., (103)]. Analyses of Bd
samples for genetic variants is a needed technological bridge, ACKNOWLEDGMENTS
expanding basic Bd surveillance and monitoring objectives to
another level of specificity. Innovative collaborations among We thank Matt Gregory (Oregon State University) for his
diagnostics laboratories and amphibian researchers in a variety of assistance with geographic analytical procedures for the
subdisciplines is likely needed to meet this objective. In addition compilation of environmental metrics for world Bd sampling
to pathogen strains, virulence parameters could be included sites, Michelle Koo for providing comments on an earlier
in the updated database, addressing new research priorities to draft of this paper and creating the two Bd database DOIs in
understand pathogen demographics, pathogenicity, and disease amphibiandisease.org, and two reviewers for their comments
dynamics (19, 34, 92). Most importantly, expanding the new on our submitted manuscript. We thank Mat Fisher and
portal to identify chytridiomycosis occurrences, rather than David Aanensen for their pivotal roles in the initiation of
simply the pathogen Bd, is a key goal. the Bd-Maps.net web portal for world Bd data, and Imperial
The Amphibian Disease database has new web applications College, UK for hosting that initial website. We thank the
compared with Bd-Maps.net (41). Importantly, user-friendly world amphibian science and management community for
import and export capacities have been a priority in portal contributing data to this project. The use of trade or firm names
development. Data can be uploaded by “projects” for both Bd and is for reader information and does not imply endorsement
Bsal studies. Data for a project can be assigned a DOI. Koo and by the U.S. Department of Agriculture of any product
Olson (41) explain further how AmphibianDisease.org can tap or service.
into additional online databases via their network partnerships,
such as genetic and genomic public databases.
As Bd data accumulation accelerates, a corresponding increase SUPPLEMENTARY MATERIAL
in the depth of knowledge of species status and threat occurs,
The Supplementary Material for this article can be found
echoing calls for conservation urgency. As Bd chytridiomycosis
online at: https://www.frontiersin.org/articles/10.3389/fvets.
appears to be about a half-Earth pandemic across amphibian
2021.685877/full#supplementary-material
taxa and sites, there is considerable room for action from both
Supplementary Appendix 1 | Detailed methods.
bottom-up community-run efforts and top-down national-to-
international policies having importance. Supplementary Appendix 2 | Comparing the current study with results from
Castro Monzon et al. (30).
Supplementary Figure 1 | Relative habitat suitability of Batrachochytrium

DATA AVAILABILITY STATEMENT dendrobatidis (Bd) occurrence (probability ratio output) for each environmental
attribute from the global presence-only best-fit species distribution model. Each
The raw data supporting the conclusions of this environmental attribute [(A) mean temperature; (B) annual precipitation; (C)
temperature range; (D) maximum elevation] marginal-response plot is calculated
article will be made available by the authors, without
while holding all other covariates at the mean. The frequency of observed
undue reservation. presence (FOP) plots (black dots, orange line estimating trend) show the number
of sites with Bd occurrence across the range of the explanatory variable (e.g., the
frequency of Bd occurrence points increases as max elevation increases, until
ETHICS STATEMENT about 4,000 m). The kernel estimated data density (light gray background) shows
the sampling effort.
Ethical review and approval was not required for the animal study Supplementary Figure 2 | Absolute probability of Batrachochytrium
because no live animals were used for this research, we only dendrobatidis (Bd) occurrence for the one environmental attribute (mean
summarized existing datasets. temperature, ◦ C) from the global presence-absence best-fit species distribution
model. The frequency of observed presence (FOP) plot (black dots, orange line
estimating trend) shows the number of sites with Bd occurrence across the range
of the variable. The kernel estimated data density (light gray background) shows
AUTHOR CONTRIBUTIONS the sampling effort.
DO designed research and led manuscript development. KR Supplementary Figure 3 | Marginal-response plots [N = 24; 18 shown in (A), 12
compiled and summarized data, drafted figures and tables, and shown in (B)] depicting interactions of covariates from the best-fit presence-only
global species distribution model of Batrachochytrium dendrobatidis (Bd)
developed Supplementary Material. CG conducted statistical occurrence using detection-only data for 0.5-degree latitude/longitude grid cells.
analyses. KC developed geographic information for analyses The probability ratio output (PRO, model predictions) is calculated for a range of
and drafted figures. AB assisted with logistical support and the environmental covariate (named below plot) while holding the interacting

covariate (named above plot) at the 0.25 percentile (left), mean (center), and 0.75 Supplementary Table 1 | Summary of Batrachochytrium dendrobatidis (Bd)
percentile (right). Non-interacting covariates per plot are held at their mean. detection (+) and no-detection (–) records and sites in the Global Bd Mapping
Supplementary Figure 4 | Marginal-response plots (N = 6) depicting interactions Project (Bd-Maps) database through December 2019.
of covariates from the best-fit presence-absence global species distribution model Supplementary Table 2 | Batrachochytrium dendrobatidis detections by
of Batrachochytrium dendrobatidis (Bd) occurrence, using detection and species, with references and countries of detection.
no-detection data for 0.5-degree latitude/longitude grid cells. The predicted
probability of Bd occurrence is calculated for a range of the environmental Supplementary Table 3 | Batrachochytrium dendrobatidis (Bd) detections by
covariate (named below plot) while holding the interacting covariate (named above country, with source citations (see Data References following
plot) at the 0.25 percentile (Left), mean (Center), and 0.75 percentile (Right). Supplementary Table 2).
REFERENCES 18. Lambert MR, Womack MC, Byrne AQ, Hernandez-Gomez O, Noss
CF, Rothstein AP, et al. Comment on “amphibian fungal panzootic
1. Dirzo R, Raven PH. Global state of biodiversity and loss. Annu Rev Env causes catastrophic and ongoing loss of biodiversity”. Science. (2020)
Resour. (2003) 28:137–67. doi: 10.1146/annurev.energy.28.050302.105532 367:eaay1838. doi: 10.1126/science.aay1838
2. Wake DB, Vredenburg VT. Are we in the midst of the sixth mass extinction? 19. Fisher MC, Garner TWJ. Chytrid fungi and global amphibian declines. Nat
A view from the world of amphibians. Proc Natl Acad Sci USA. (2008) Rev Microbiol. (2020) 18:332–43. doi: 10.1038/s41579-020-0335-x
105(Suppl. 1):11466–73. doi: 10.1073/pnas.0801921105 20. Kerby J, Berger L. Global trade in frogs has led to catastrophic amphibian
3. Barnosky AD, Matzke N, Tomiya S, Wogan GOU, Swartz B, Quental TB, declines. In: Tsing A, Deger J, Saxena AK, Zhou F, editors. Feral Atlas: The
et al. Has the earth’s sixth mass extinction already arrived? Nature. (2011) More-than-Human Anthropocene. Bd chytrid fungus. Redwood City, CA:
471:51–7. doi: 10.1038/nature09678 Stanford University Press (2020). Available online at: feralatlas.org (accessed
4. Ceballos G, Ehrlich PR, Barnosky AD, Garcia A, Pringle RM, Palmer October 22, 2020).
TM. Accelerated modern human-induced species losses: entering the 21. Calhoun DM, Leslie KL, Riepe TB, Achatz TJ, McDevitt-Galles T, Tkach
sixth mass extinction. Sci Adv. (2015) 1:e1400253. doi: 10.1126/sciadv. VV, et al. Patterns of Clinostomum marginatum infection in fishes and
1400253 amphibians: integration of field genetic, and experimental approaches. J
5. International Union for the Conservation of Nature. IUCN Red List of Helminthol. (2020) 94:e44. doi: 10.1017/S0022149X18001244
Threatened Species. Version 2020-3 (2021). Available online at: https://www. 22. Pessier AP. Hopping over red leg: the metamorphosis of amphibian
iucnredlist.org (accessed February 1, 2021). pathology. Vet Pathol. (2017) 54:355–7. doi: 10.1177/0300985817699861
6. International Union for the Conservation of Nature. IUCN Species Survival 23. Martel A, Spitzen-van der Sluijs A, Blooi M, Bert W, Ducatelle R,
Commission (SSC) Amphibian Specialist Group. (2021). Available online at: Fisher MC, et al. Batrachochytrium salamandrivorans sp. nov. causes lethal
https://www.iucn-amphibians.org/ (accessed December 1, 2020). chytridiomycosis in amphibians. Proc Natl Acad Sci USA. (2013) 110:15325–
7. Kuta KG, Richardson LL. Ecological aspects of black band disease of corals: 29. doi: 10.1073/pnas.1307356110
relationships between disease incidence and environmental factors. Coral 24. Blaustein AR, Urbina J, Snyder PW, Reynolds E, Dang T, Hoverman JT,
Reefs. (2002) 21:393–398. doi: 10.1007/s00338-002-0261-6 et al. Effects of emerging infectious diseases on amphibians: a review of
8. Howells E, Vaughan G, Work TM, Burt J, Abrego D. Annual outbreaks of experimental studies. Diversity. (2018) 10:1–49. doi: 10.3390/d10030081
coral disease coincide with extreme seasonal warming. Coral Reefs. (2020) 25. Isidoro-Ayza M, Grear DA, Chambouvet A. Pathology and case definition
29:771–81. doi: 10.1007/s00338-020-01946-2 of severe perkinsea infection of frogs. Vet Pathol. (2019) 56:133–
9. Aquino CA, Besemer RM, DeRito CM, Kocian J, Porter IR, 42. doi: 10.1177/0300985818798132
Raimondi PT, et al. Evidence that microorganisms at the animal- 26. Chambouvet A, Smilansky V, Jirku M, Isidoro-Ayza M, Itoïz S, Derelle E,
water interface drive sea star wasting disease. Front Microbiol. (2021) et al. Diverse alveolate infections of tadpoles, a new threat to frogs? PLoS
11:610009. doi: 10.3389/fmicb.2020.610009 Pathogens. (2020) 16:e1008107. doi: 10.1371/journal.ppat.1008107
10. Frick WF, Pollock JF, Hicks AC, Langwig KE, Reynolds DS, Turner 27. Duffus A, Olson D. The Establishment of a Global Ranavirus Reporting
GG, et al. An emerging disease causes regional population collapse System. Vol. 96. FrogLog (2011). p. 37. Available online at: https://
of a common North American bat species. Science. (2010) 329:679– www.amphibians.org/wp-content/uploads/2018/12/Froglog96.pdf (accessed
82. doi: 10.1126/science.1188594 March 22, 2021).
11. Verant ML, Boyles JG Waldrep W Jr., Wibbelt G, Blehert DS. 28. Olson DH, Aanensen DM, Ronnenberg KL, Powell CI, Walker SF,
Temperature-dependent growth of Geomyces destructans, the Bielby J, et al. Mapping the global emergence of Batrachochytrium
fungus that causes bat white-nose syndrome. PLoS ONE. (2012) dendrobatidis, the amphibian chytrid fungus. PLoS ONE. (2013)
7:e46280. doi: 10.1371/journal.pone.0046280 8:e56802. doi: 10.1371/journal.pone.0056802
12. LaDeau SL, Kilpatrick AM, Marra PP. West Nile virus emergence and 29. Olson DH. A decade of herpetological disease papers: puzzle pieces of a
large-scale declines of North American bird populations. Nature. (2007) bigger picture. Herpetol Rev. (2019) 50:37–40. Available online at: https://
447:710. doi: 10.1038/nature05829 www.fs.usda.gov/treesearch/pubs/57951 (accessed June 24, 2021).
13. Northeast Wildlife Disease Cooperative. Disease Fact Sheets. Northeast 30. Castro Monzon F, Rödel MO, Jeschke JM. Tracking Batrachochytrium
USA Wildlife Disease Cooperative (2020). Available online at: https://www. dendrobatidis. infection across the globe. EcoHealth. (2020) 17:270–
northeastwildlife.org/disease-fact-sheets (accessed May 19, 2021). 9. doi: 10.1007/s10393-020-01504-w
14. Stuart SN, Chanson JS, Cox NA, Young BE, Rodrigues AS, Fischman DL, 31. Brunner JL, Olson DH, Gray MJ, Miller DL, Duffus ALJ. Global patterns
et al. Status and trends of amphibian declines and extinctions worldwide. of Ranavirus detections. Collection: Ranavirus research: 10 years of global
Science. (2004) 306:1783–6. doi: 10.1126/science.1103538 collaboration. FACETS. (2021) 6:912–24. doi: 10.1139/facets-2020-0013
15. Martel A, Blooi M, Adriaensen C, Van Rooij P, Beukema W, Fisher MC, 32. Blaustein AR, Han BA, Relyea RA, Johnson TJ, Buck JC, Gervasi SS,
et al. Recent introduction of a chytrid fungus endangers Western Palearctic et al. The complexity of amphibian declines: understanding the role of
salamanders. Science. (2014) 346:630–1. doi: 10.1126/science.1258268 cofactors in driving amphibian losses. Ann NY Acad Sci. (2011) 1223:108–
16. Gray MJ, Chinchar VG, editors. Ranaviruses: Lethal Pathogens of Ectothermic 19. doi: 10.1111/j.1749-6632.2010.05909.x
Vertebrates. Switzerland: Springer International Publishing (2015). p. 246. 33. Laking A, Ngo HN, Pasmans F, Martel A, Nguyen TT.
17. Scheele BC, Pasmans F, Skerratt LF, Berger L, Martel A, Beukema W, Batrachochytrium salamandrivorans is the predominant chytrid fungus
et al. Amphibian fungal panzootic causes catastrophic and ongoing loss of in Vietnamese salamanders. Sci Rep. (2017) 7:44443. doi: 10.1038/srep
biodiversity. Science. (2019) 363:1459–63. doi: 10.1126/science.aav0379 44443

34. O’Hanlon SJ, Rieux A, Farrer RA, Rosa GM, Waldman B, Bataille A, et al. to an emerging pathogen, the amphibian chytrid fungus. PLoS ONE. (2017)
Recent Asian origin of chytrid fungi causing global amphibian declines. 12:e0167882. doi: 10.1371/journal.pone.0167882
Science. (2018) 360:621–7. doi: 10.1126/science.aar1965 52. Yap TA, Koo MS, Ambrose RF, Vredenburg VT. Introduced bullfrog
35. Longcore JE, Pessier AP, Nichols DK. Batrachochytrium dendrobatidis gen. facilitates pathogen invasion in the Western United States. PLoS ONE. (2018)
et sp. nov. a chytrid pathogenic to amphibians. Mycologia. (1999) 91:219– 13:e0188384. doi: 10.1371/journal.pone.0188384
27. doi: 10.1080/00275514.1999.12061011 53. Jairam R. A historical overview of Batrachochytrium dendrobatidis infection
36. Weldon C, Du Preez L, Hyatt A, Muller R, Speare R. Origin of from specimens at the national zoological collection Suriname. PLoS ONE.
the amphibian chytrid fungus. Emerg Infect Dis. (2004) 10:2100– (2020) 15:e0239220. doi: 10.1371/journal.pone.0239220
5. doi: 10.3201/eid1012.030804 54. Dang TD, Searle CL, Blaustein AR. Virulence variation among strains
37. Goka K, Yokoyama JUN, Une Y, Kuroki T, Suzuki K, Nakahara M, of the emerging infectious fungus Batrachochytrium dendrobatidis (Bd)
et al. Amphibian chytridiomycosis in Japan: distribution, haplotypes in multiple amphibian host species. Dis Aquat Organ. (2017) 124:233–
and possible route of entry into Japan. Mol Ecol. (2009) 18:4757– 9. doi: 10.3354/dao03125
74. doi: 10.1111/j.1365-294X.2009.04384.x 55. National Wildfire Coordinating Group. Guide to Preventing Aquatic
38. Bataille A, Fong JJ, Cha M, Wogan GOU, Baek HJ, Lee H, et al. Genetic Invasive Species Transport by Wildland Fire Operations. Invasive species
evidence for a high diversity and wide distribution of endemic strains of subcommittee, Equipment Technology Committee, National Wildfire
the pathogenic chytrid fungus Batrachochytrium dendrobatidis in Wild Asian Coordinating Group, United States. PMS 444. (2017). p. 64. Available
amphibians. Mol Ecol. (2013) 22:4196–209. doi: 10.1111/mec.12385 online at: https://www.nwcg.gov/sites/default/files/publications/pms444.pdf
39. Rodriguez D, Becker CG, Pupin NC, Haddad CFB, Zamudio KR. Long- (accessed August 26, 2020).
term endemism of two highly divergent lineages of the amphibian- 56. National Wildfire Coordinating Group. Invasive Species Mitigation for
killing fungus in the Atlantic forest of Brazil. Mol Ecol. (2014) 23:774– Ground Resources. Invasive species subcommittee, Equipment Technology
87. doi: 10.1111/mec.12615 Committee, National Wildfire Coordinating Group, United States.
40. Talley BL, Muletz CR, Vredenburg VT, Fleischer RC, Lips KR. A century Operations video (2020). Available online at: https://www.nwcg.gov/
of Batrachochytrium dendrobatidis in Illinois amphibians (1888–1989). Biol publications/training-courses/rt-130/operations/op819 (accessed August
Conserv. (2015) 182:254–61. doi: 10.1016/j.biocon.2014.12.007 26, 2020).
41. Koo MS, Olson DH. Data management working group. In: North American 57. Piotrowski JS, Annis SL, Longcore JE. Physiology of Batrachochytrium
Bsal Task Force. A North American Strategic Plan to Control Invasions of the dendrobatidis, a chytrid pathogen of amphibians. Mycologia. (2004) 96:9–
Lethal Salamander Pathogen Batrachochytrium salamandrivorans. (2020). p. 15. doi: 10.1080/15572536.2005.11832990
47–52. Available online at: https://www.salamanderfungus.org/wp-content/ 58. Berger L, Speare R, Hines HB, Marantelli G, Hyatt AD, McDonald
uploads/2020/04/Bsal-Strategic-Plan-2020-Posted-OM-edits.pdf (accessed KR, et al. Effect of season and temperature on mortality
May 19, 2021). in amphibians due to chytridiomycosis. Austr Vet J. (2004)
42. Olson DH, Ronnenberg KL. Global Bd Mapping Project: 2014 Update. 82:434–9. doi: 10.1111/j.1751-0813.2004.tb11137.x
Vol. 111. FrogLog (2014). p. 17–21. Available online at: https://www. 59. Ron SR. Predicting the distribution of the amphibian pathogen
iucn-amphibians.org/wp-content/uploads/2018/10/froglog110_lowres.pdf Batrachochytrium dendrobatidis in the new world. Biotropica. (2005)
(accessed March 22, 2021). 37:209–21. doi: 10.1111/j.1744-7429.2005.00028.x
43. Xie GY, Olson DH, Blaustein AR. Projecting the global distribution 60. Pounds JA, Bustamante MR, Coloma LA, Consuegra JA, Fogden MPL, Foster
of the emerging amphibian fungal pathogen, Batrachochytrium PN, et al. Widespread amphibian extinctions from epidemic disease driven
dendrobatidis, based on IPCC climate futures. PLoS ONE. (2016) by global warming. Nature. (2006) 439:161–7. doi: 10.1038/nature04246
11:e0160746. doi: 10.1371/journal.pone.0160746 61. Bosch J, Carrascal LM, Duran L, Walker S, Fisher MC. Climate change
44. Wessner D, Dupont C, Charles T. Microbiology. New York, NY: John Wiley and outbreaks of amphibian chytridiomycosis in a montane area of
& Sons (2013). Central Spain; is there a link? P R Soc Lond B Bio. (2007) 274:253–
45. Holland JS. Bd map, In: The Vanishing. National Geographic Magazine. 60. doi: 10.1098/rspb.2006.3713
Washington, DC: National Geographic Society (2009). p. 145. 62. Kriger KM, Hero J-M. Large-scale seasonal variation in the
46. Tsing A, Deger J, Saxena AK, Zhou F, editors. Bd Chytrid Fungus. In: Feral prevalence and severity of chytridiomycosis. J Zool. (2007)
Atlas: The More-than-Human Anthropocene. Redwood City, CA: Stanford 271:352–9. doi: 10.1111/j.1469-7998.2006.00220.x
University Press (2020). Available online at: feralatlas.org (accessed October 63. Rödder D, Kielgast J, Lötters S. Future potential distribution of the emerging
22, 2020). doi: 10.21627/2020fa amphibian chytrid fungus under anthropogenic climate change. Dis Aquat
47. Gray MJ, Duffus ALJ, Haman KH, Harris RN, Allender MC, Thompson Organ. (2010) 92:201–7. doi: 10.3354/dao02197
TA, et al. Pathogen surveillance in herpetofaunal populations: guidance 64. Rohr JR, Raffel TR. Linking global climate and temperature variability to
on study design, sample collection, biosecurity, and intervention strategies. widespread amphibian declines putatively caused by disease. Proc Natl Acad
Herpetol Rev. (2017) 48:334–51. Available online at: https://parcplace.org/ Sci USA. (2010) 107:8269–74. doi: 10.1073/pnas.0912883107
wp-content/uploads/2017/08/Grayetal2017.pdf (accessed June 24, 2021). 65. Murray KA, Retallick RWR, Puschendorf R, Skerratt LF, Rosauer
48. Julian JT, Henry PFP, Drasher JM, Jewell SD, Michell K, Oxenrider KJ, et al. D, McCallum HI, et al. Assessing spatial patterns of disease risk
Minimizing the spread of herpetofaunal pathogens in aquatic habitats by to biodiversity: implications for the management of the amphibian
decontaminating construction equipment. Herpetol Rev. (2020) 51:472– pathogen, Batrachochytrium dendrobatidis. J Appl Ecol. (2011) 48:163–
83. Available online at: https://parcplace.org/wp-content/uploads/2020/ 73. doi: 10.1111/j.1365-2664.2010.01890.x
11/Julian-2020-Decontamination-for-Herps-for-large-equipment.pdf 66. Puschendorf R, Carnaval AC, VanDerWal J, Zumbado-Ulate H,
(accessed June 24, 2021). Chaves G, Bolanos F, et al. Distribution models for the amphibian
49. Olson DH, Haman KH, Gray MJ, Harris R, Thompson TA, Iredale M, chytrid Batrachochytrium dendrobatidis in Costa Rica: proposing
et al. Enhanced between-site biosecurity to minimize herpetofaunal disease- climatic refuges as a conservation tool. Divers Distrib. (2009)
causing pathogen transmission. Herpetol Rev. (2021) 52:36–46. Available 15:401–8. doi: 10.1111/j.1472-4642.2008.00548.x
online at: https://www.fs.usda.gov/treesearch/pubs/62316 (accessed June 24, 67. Voyles J, Johnson LR, Briggs CJ, Cashins SD, Alford RA, Berger L, et al.
2021). Temperature alters reproductive life history patterns in Batrachochytrium
50. Grant ECH, Miller DAW, Schmidt BR, Adams MJ, Amburgey SM, dendrobatidis, a lethal pathogen associated with the global loss of
Chambert T, et al. Quantitative evidence for the effects of multiple amphibians. Ecol Evol. (2012) 2:2241–9. doi: 10.1002/ece3.334
drivers on continental-scale amphibian declines. Nature. (2016) 68. Voyles J, Johnson LR, Rohr J, Kelly R, Barron C, Miller D, et al.
6:25625. doi: 10.1038/srep25625 Diversity in growth patterns among strains of the lethal fungal pathogen
51. Gervasi SS, Stephens PR, Hua J, Searle CL, Urbina J, Olson DH, et al. Linking Batrachochytrium dendrobatidis across extended thermal optima. Oecologia.
ecology and epidemiology to understand predictors of multi-host responses (2017) 184:363–373. doi: 10.1007/s00442-017-3866-8

69. Raffel TR, Romansic JM, Halstead NT, McMahon TA, Venesky MD, Rohr Batrachochytrium salamandrivorans (Bsal) in the EU. EFSA J. (2018)
JR. Disease and thermal acclimation in a more variable and unpredictable 16: e05259. doi: 10.2903/j.efsa.2018.5259
climate. Nat Clim Change. (2013) 3:146–51. doi: 10.1038/nclimate1659 89. United States Fish and Wildlife Service. Listing Salamanders as Injurious
70. Chestnut T, Anderson C, Popa R, Blaustein AR, Voytek M, Olson DH, et al. Due to Risk of Salamander Chytrid Fungus. (2016). Available online at:
Heterogeneous occupancy and density estimates of the pathogenic fungus https://www.fws.gov/injuriouswildlife/salamanders.html (accessed October
Batrachochytrium dendrobatidis in waters of North America. PLoS ONE. 12, 2020).
(2014) 9:e106790. doi: 10.1371/journal.pone.0106790 90. Canada Border Services Agency. Environment and Climate Change Canada
71. Bradley PW, Brawner MD, Raffell TR, Rohr JR, Olson DH, (ECCC)’s Import Restrictions on Salamanders. Customs Notice 17-17. (2018).
Blaustein AR. Shifts in temperature influence how Batrachochytrium Available online at: https://www.cbsa-asfc.gc.ca/publications/cn-ad/cn17-
dendrobatidis Infects amphibian larvae. PLoS ONE. (2019) 17-eng.html (accessed October 12, 2020).
14:e0222237. doi: 10.1371/journal.pone.0222237 91. Center for Disease Control and Prevention. One Health. United States
72. Frost DR. Amphibian Species of the World: An Online Reference. Version 6.1. Department of Health and Human Services (2020). Available online at:
New York, NY: American Museum of Natural History (2020). https://www.cdc.gov/onehealth/index.html (accessed December 29, 2020).
73. Skerratt LF, Berger L, Hines HB, McDonald KR, Mendez D, Speare R. Survey 92. Byrne AQ, Vredenburg VT, Martel A, Pasmans F, Bell RC, Blackburn
protocol for detecting chytridiomycosis in all Australian frog populations. DC, et al. Cryptic diversity of a widespread global pathogen reveals
Dis Aquat Organ. (2008) 80: 85–94. doi: 10.3354/dao01923 expanded threats to amphibian conservation. Proc Natl Acad Sci USA. (2019)
74. Merow C, Smith MJ, Silander JA Jr. A Practical guide to 116:20382–7. doi: 10.1073/pnas.1908289116
MaxEnt for modelling species’ distributions: what it does, and 93. Harris RN, Brucker RM, Walke JB, Becker MH, Schwantes CR, Flaherty
why inputs and settings matter. Ecography. (2013) 36:1058– DC, et al. Skin microbes on frogs prevent morbidity and mortality caused
69. doi: 10.1111/j.1600-0587.2013.07872.x by a lethal skin fungus. ISME J. (2009) 3:818–24. doi: 10.1038/ismej.
75. Pearce JL, Boyce MS. Modelling distribution and abundance 2009.27
with presence-only data. J Appl Ecol. (2006) 43:405– 94. Harris RN, Lauer A, Simon MA, Banning JL, Alford RA. Addition
12. doi: 10.1111/j.1365-2664.2005.01112.x of antifungal skin bacteria to salamanders ameliorates the effects of
76. Halvorsen R, Mazzoni S, Bryn A, Bakkestuen V. Opportunities chytridiomycosis. Dis Aquat Organ. (2009) 83:11–16. doi: 10.3354/dao02004
for improved distribution modelling practice via a strict 95. Bletz MC, Loudon AH, Becker MH, Bell SC, Woodhams DC, Minbiole
maximum likelihood interpretation of MaxEnt. Ecography. (2015) KPC, et al. Mitigating amphibian chytridiomycosis with bioaugmentation:
38:172–83. doi: 10.1111/ecog.00565 characteristics of effective probiotics and strategies for their selection and
77. Tuszynski J. caTools: Tools: Moving Window Statistics, GIF, Base64, ROC use. Ecol Lett. (2013) 16:807–20. doi: 10.1111/ele.12099
AUC, etc. R Package, Version 1.18.0. (2020). Available online at: https:// 96. Bletz MC, Kelly M, Sabino-Pinto J, Bales E, Van Praet S, Bert W, et al.
CRAN.R-project.org/package=caTools (accessed December 1, 2020). Disruption of skin microbiota contributes to salamander disease. Proc R Soc
78. Halvorsen R. A strict maximum likelihood explanation of MaxEnt, and Lond B Bio. (2018) 285:20180758. doi: 10.1098/rspb.2018.0758
some implications for distribution modelling. Sommerfeltia. (2013) 36:1– 97. Knapp RA, Fellers GM, Kleeman PM, Miller DAW, Vredenburg VT,
32. doi: 10.2478/v10208-011-0016-2 Rosenblum EB, et al. Large-scale recovery of an endangered amphibian
79. Elith J, Phillips SJ, Hastie T, Dudík M, Chee YE, Yates CJ. A statistical despite ongoing exposure to multiple stressors. Proc Natl Acad Sci USA.
explanation of MaxEnt for ecologists. Divers Distrib. (2011) 17:43– (2016) 113:11889–94. doi: 10.1073/pnas.1600983113
57. doi: 10.1111/j.1472-4642.2010.00725.x 98. Wilbur MQ, Knapp RA, Toothman M, Briggs CJ. Resistance, tolerance
80. Vollering J, Halvorsen R, Mazzoni S. The MIAmaxent R package: variable and environmental transmission dynamics determine host extinction
transformation and model selection for species distribution models. Ecol risk in a load-dependent amphibian disease. Ecol Lett. (2017) 20:1169–
Evol. (2019) 9:12051–68. doi: 10.1002/ece3.5654 81. doi: 10.1111/ele.12814
81. Wickham H. ggplot2: Elegant Graphics for Data Analysis. New York, NY: 99. Rodriguez KM, Voyles J. The amphibian complement system and
Springer-Verlag (2016). doi: 10.1007/978-3-319-24277-4_9 chytridiomycosis. J Exper Zool. (2020) 333:706–19. doi: 10.1002/jez.2419
82. Fong JJ, Cheng TJ, Battaile A, Pessier AP, Waldman B, Vredenburg VT. Early 100. Rollins-Smith LA. Global amphibian declines, disease, and the ongoing battle
1900s detection of Batrachochytrium dendrobatidis in Korean amphibians. between Batrachochytrium fungi and the immune system. Herpetologica.
PLoS ONE. (2015) 10: e0115656. doi: 10.1371/journal.pone.0115656 (2020) 76:178–88. doi: 10.1655/0018-0831-76.2.178
83. Bancroft B, Han B, Searle C, Michael L, Olson DH, Lawler J, et al. 101. Bosch J, Sanchez-Tomé E, Fernández-Loras A, Oliver JA, Fisher MC, Garner
Species-level correlates of susceptibility to the pathogenic amphibian fungus TWJ. Successful elimination of a lethal wildlife infectious disease in nature.
Batrachochytrium dendrobatidis in the United States. Biodivers Conserv. Biol Lett. (2015) 11:20150874. doi: 10.1098/rsbl.2015.0874
(2011) 20:1911–20. doi: 10.1007/s10531-011-0066-4 102. Martel A, Vila-Escale M, Fernandez-Giberteau D, Martinez-Silvestre A,
84. Streicher JW, Mahony S, Kamei RG, Nidup T, Jervis P, Fisher MC. Canessa S, Van Praet S, et al. Integral chain management of wildlife diseases.
Preliminary survey reveals no evidence of Batrachochytrium dendrobatidis in Conserv Lett. (2020) 13:212707. doi: 10.1111/conl.12707
the Kingdom of Bhutan. Herpetol Rev. (2020) 51:494–7. Available online at: 103. Hauck LL, Weitemier KA, Penaluna BE, Garcia TS, Cronn R. Casting
https://ssarherps.org/herpetological-review-pdfs/ (accessed June 24, 2021). a broader net: using microfluidic metagenomics to capture aquatic
85. Kirk D, O’Connor MI, Mordecai EA. Temperature effects on individual- biodiversity data from diverse taxonomic targets. Environ DNA. (2019)
level parasitism translate into predictable effects on parasitism in 1:251–67. doi: 10.1002/edn3.26
populations. bioRxiv. (2020). doi: 10.1101/2020.12.01.406884. Available
online at: https://www.biorxiv.org/content/10.1101/2020.12.01.406884v2. Conflict of Interest: The authors declare that the research was conducted in the
full.pdf (accessed June 24, 2021). absence of any commercial or financial relationships that could be construed as a
86. Bernard RF, Grant EHC. Rapid assessment indicates context-dependent potential conflict of interest.
mitigation for amphibian disease risk. Wildlife Soc B. (2021). Available online
at: https://wildlife.onlinelibrary.wiley.com/journal/23285540 (accessed Copyright © 2021 Olson, Ronnenberg, Glidden, Christiansen and Blaustein. This
June 24, 2021). is an open-access article distributed under the terms of the Creative Commons
87. Phillott AD, Speare R, Hines HB, Skerratt LF, Meyer E, McDonald KR, et al. Attribution License (CC BY). The use, distribution or reproduction in other forums
Minimising exposure of amphibians to pathogens during field studies. Dis is permitted, provided the original author(s) and the copyright owner(s) are credited
Aquat Organ. (2010) 92:175–85. doi: 10.3354/dao02162 and that the original publication in this journal is cited, in accordance with accepted
88. More S, Miranda MA, Bicout D, Bøtner A, Butterworth A, academic practice. No use, distribution or reproduction is permitted which does not
Calistri P, et al. Risk of survival, establishment and spread of comply with these terms.

Fvets 08 685877

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

Fvets 08 685877

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fvets 08 685877

Uploaded by

Copyright:

Available Formats

ORIGINAL RESEARCH

published: 16 July 2021

Global Patterns of the Fungal

The amphibian chytrid fungus Batrachochytrium dendrobatidis (Bd) is a skin pathogen

Frontiers in Veterinary Science | www.frontiersin.org 1 July 2021 | Volume 8 | Article 685877

Frontiers in Veterinary Science | www.frontiersin.org 2 July 2021 | Volume 8 | Article 685877

Frontiers in Veterinary Science | www.frontiersin.org 3 July 2021 | Volume 8 | Article 685877

Frontiers in Veterinary Science | www.frontiersin.org 4 July 2021 | Volume 8 | Article 685877

Frontiers in Veterinary Science | www.frontiersin.org 5 July 2021 | Volume 8 | Article 685877

Frontiers in Veterinary Science | www.frontiersin.org 6 July 2021 | Volume 8 | Article 685877

Frontiers in Veterinary Science | www.frontiersin.org 7 July 2021 | Volume 8 | Article 685877

Frontiers in Veterinary Science | www.frontiersin.org 8 July 2021 | Volume 8 | Article 685877

Anura 1,153 2,106 54.7 7,311 49 55 89.1 56

Total 1,294 2,412 53.6 8,283 63 72 85.9 75

Frontiers in Veterinary Science | www.frontiersin.org 9 July 2021 | Volume 8 | Article 685877

TABLE 3 | Family-level summary of Batrachochytrium dendrobatidis (Bd) TABLE 3 | Continued

Frontiers in Veterinary Science | www.frontiersin.org 10 July 2021 | Volume 8 | Article 685877

TABLE 4 | Fraction of total variation accounted for (FTVA) by each variable in

Model Variable FTVA

Presence-only global 10-year mean annual daily temperature 0.713

FIGURE 5 | Absolute probability of Batrachochytrium dendrobatidis (Bd)

presence-absence model, where probability of Bd increased until

Frontiers in Veterinary Science | www.frontiersin.org 11 July 2021 | Volume 8 | Article 685877

Frontiers in Veterinary Science | www.frontiersin.org 12 July 2021 | Volume 8 | Article 685877

Frontiers in Veterinary Science | www.frontiersin.org 13 July 2021 | Volume 8 | Article 685877

Frontiers in Veterinary Science | www.frontiersin.org 14 July 2021 | Volume 8 | Article 685877

Frontiers in Veterinary Science | www.frontiersin.org 15 July 2021 | Volume 8 | Article 685877

Frontiers in Veterinary Science | www.frontiersin.org 16 July 2021 | Volume 8 | Article 685877

Supplementary Figure 1 | Relative habitat suitability of Batrachochytrium

Frontiers in Veterinary Science | www.frontiersin.org 17 July 2021 | Volume 8 | Article 685877

Frontiers in Veterinary Science | www.frontiersin.org 18 July 2021 | Volume 8 | Article 685877

Frontiers in Veterinary Science | www.frontiersin.org 19 July 2021 | Volume 8 | Article 685877

Frontiers in Veterinary Science | www.frontiersin.org 20 July 2021 | Volume 8 | Article 685877

You might also like