The fungal disease chytridiomycosis, caused by Batrachochytrium dendrobatidis, is enigmatic because it occurs globally in both declining and apparently healthy (non-declining) amphibian populations. This distribution has fueled debate concerning whether, in sites where it has recently been found, the pathogen was introduced or is endemic. In this study, we addressed the molecular population genetics of a global collection of fungal strains from both declining and healthy amphibian populations using DNA sequence variation from 17 nuclear loci and a large fragment from the mitochondrial genome. We found a low rate of DNA polymorphism, with only two sequence alleles detected at each locus, but a high diversity of diploid genotypes. Half of the loci displayed an excess of heterozygous genotypes, consistent with a primarily clonal mode of reproduction. Despite the absence of obvious sex, genotypic diversity was high (44 unique genotypes out of 59 strains). We provide evidence that the observed genotypic variation can be generated by loss of heterozygosity through mitotic recombination. One strain isolated from a bullfrog possessed as much allelic diversity as the entire global sample, suggesting the current epidemic can be traced back to the outbreak of a single clonal lineage. These data are consistent with the current chytridiomycosis epidemic resulting from a novel pathogen undergoing a rapid and recent range expansion. The widespread occurrence of the same lineage in both healthy and declining populations suggests that the outcome of the disease is contingent on environmental factors and host resistance.
Amphibian species are facing a current global extinction crisis of unprecedented magnitude. The major factors causing their decline are the emerging disease chytridiomycosis and habitat destruction. Chytridiomycosis is caused by the aquatic fungus Batrachochytrium dendrobatidis and has been linked to species extinctions and population declines in montane regions including Australia, Panama, North America, and Spain. Currently, it is debated whether the recent emergence of the pathogen is largely the result of environmental factors triggering an outbreak of an endemic pathogen or if the epidemic has been caused by widespread introduction of the pathogen into naïve host populations (“pathogen pollution”). We studied the population genetics of chytridiomycosis using DNA sequences from a global panel of strains. These data showed evidence of a strong genetic bottleneck in the history of the pathogen, and the epidemic appears traceable to the widespread dispersal of a single genotype. Populations were not structured by host-origin, and the same lineage was detected in populations of both resistant and highly sensitive species. The data suggest that the chytridiomycosis epidemic is caused by the emergence of a novel pathogen but that disease outcome is contingent on host resistance and environmental factors.
Citation: James TY, Litvintseva AP, Vilgalys R, Morgan JAT, Taylor JW, et al. (2009) Rapid Global Expansion of the Fungal Disease Chytridiomycosis into Declining and Healthy Amphibian Populations. PLoS Pathog 5(5): e1000458. doi:10.1371/journal.ppat.1000458
Editor: Robin Charles May, University of Birmingham, United Kingdom
Received: February 17, 2009; Accepted: April 29, 2009; Published: May 29, 2009
Copyright: © 2009 James et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: Funding for this project was provided by the National Science Foundation (GEO 0213851 and IOS 9977063). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Globally, amphibian species are threatened by the emergence of a novel pathogen, the chytrid fungus Batrachochytrium dendrobatidis (Bd) ,. The pathogen proliferates in epidermal cells of amphibians leading to hyperkeratosis, electrolyte loss ,, and ultimately death in susceptible species. The precise cause of osmotic imbalance and death is uncertain, but is hypothesized to be caused by physical skin damage or unidentified toxins produced by the pathogen ,. Some species, such as the North American bullfrog, show no evidence of morbidity and are likely to have acted as vectors by facilitating the widespread dispersal of the pathogen and by providing a reservoir of infection ,,. Amphibian declines caused by chytridiomycosis have not affected all areas of the planet homogeneously but are well documented in montane regions of Australia, western North America, Panama, and Spain ,,,,. In other regions such as eastern North America and South Africa, chytridiomycosis occurs but has not been linked to declines ,,.
The disease chytridiomycosis was only recently discovered ,,, and debate centers on whether Bd is an endemic pathogen whose emergence is due to recent changes in the environment versus the alternative that Bd is a novel pathogen introduced into naïve host populations, or a combination of the two ,,,,. Analyses of the pattern of declines in Australia ,, Panama , and California  provided convincing evidence of recent spread of the pathogen, consistent with the idea that the pathogen is novel in these areas. Histological examinations of museum specimens also support a recent origin of the disease, with the oldest known infections dating to 1938 on Xenopus laevis from Africa . In contrast, the endemic pathogen hypothesis has been invoked to explain the correlation between harlequin frog (Atelopus spp.) extinctions in Central America and global climate change measured as sea surface temperature .
Despite growing evidence that the disease is increasing in geographic range, no clear evidence points to a source population of chytridiomycosis. Two hypotheses explain how Bd may have been spread across the globe but remain to be tested. One postulates that the disease evolved in Africa, being thereafter transported intercontinentally through dissemination of Xenopus for aquaria, scientific research, and pregnancy assays . A second proposes that the disease spread through purposeful and unintentional range expansions of the North American Bullfrog Rana catesbeiana . R. catesbeiana is native to eastern North America but has been widely introduced to Asia, Europe, western North America and South America ; its current distribution cannot fully explain the arrival of Bd in regions such as Australia and South Africa.
The inability to disentangle the endemic from the novel pathogen hypotheses stems from the fact that little is known about the global population genetic structure of Bd. Previous studies of Bd have suggested low genetic variation in the pathogen and widespread dispersal of closely related pathogen genotypes ,. However, the only broadly sampled global study to date included only three polymorphic loci, making it difficult to infer statistically significant patterns in the data. Several questions must be addressed to resolve the ongoing debate: is there evidence for population structure within Bd, how is genetic diversity generated, is there evidence for a source population, and do Bd isolates from healthy amphibian populations (defined in this study as species or geographic regions having no evidence of population declines) have the same genetic structure as those from declining populations? The central prediction of the endemic hypothesis is that geographic population structure will be detected because populations are at equilibrium. In contrast, the novel pathogen hypothesis predicts geographically widespread genotypes and the occurrence of a genetic bottleneck resulting from the rapid range expansion of the disease. A key assumption is that source populations can be identified because they harbor greater genetic diversity ,,.
In this study we address these questions of genetic diversity in Bd with 59 strains isolated from five continents and 31 host species (Table 1) and 17 polymorphic nuclear loci. We also sequenced a non-coding region (>11 kbp) of the mitochondrial genome for a subset of 16 strains. These data were used to test the predictions of the endemic versus the novel pathogen hypotheses. Genetic diversity and heterozygosity of the diploid pathogen were compared between various subpopulations to test for geographic structure and to identify if any were a source of the current epidemic. Consistent with the novel pathogen hypothesis, we observed an extremely low allelic diversity in the pathogen, suggesting a severe bottleneck during the current epidemic. The data also demonstrate that the pathogen reproduces primarily asexually, but that genotypic diversity is high. Using information on the genomic location of marker loci we show how this paradox can be reconciled by invoking loss of heterozygosity as a product of mitotic recombination.
Multilocus sequence typing (MLST) of Bd strains confirmed that the fungus is diploid because all strains were heterozygous at multiple loci. Despite the geographic breadth of sampling, only two alleles (sequence variants) were detected at each of the 17 loci. The limited allelic variation is not restricted to the nuclear genome. Zero nucleotide polymorphisms were found in the >11 kbp portion of the mitochondrial genome surveyed in the subsample of global strains, though one large deletion of 3,904 bp occurred in 4 of the 16 strains. This low genetic variation is consistent with a previous study describing minimal global genetic variation using fewer genetic markers and sparser sampling . The previous study found that 7 of 10 loci were monomorphic; in this study only polymorphic loci were included.
Host and geographic patterns of population substructure
We analyzed the genetic similarity of isolates with MLST data to determine the association between genotype and geographic origin (Figure 1). Strains from the same geographic region showed some tendency to cluster together, including several strains of identical multilocus genotype (Figure 1). Larger clusters of strains with identical genotypes are one Panamanian group of six strains and a group of four strains from South Africa. An association between strains from Spain and the United States was also detected. However, broad geographic structure is limited; for example, strains from Australia and Africa are found in several unrelated portions of the dendrogram. Few branches in the dendrogram are supported by bootstrap resampling, presumably due to the limited resolving power of two alleles per sequenced locus. Because of the likelihood of homoplasy among multilocus genotypes due to recombination, a phenetic method (neighbor joining) for tree reconstruction was used rather than a cladistic method. Network approaches can account for this homoplasy by introducing loops where recombination could have occurred in the history of the genealogy. A network of the multilocus genotypes derived using statistical parsimony was constructed and revealed a very similar pattern of genotypic relatedness as the neighbor-joining tree (Figure S1), including the close association of Spanish and the United States strains.
The tree was computed using neighbor-joining in PAUP v4.0b10  with “hetequal” coding, and the thickened branches indicate bootstrap values of 50% or greater.
The absence of an overarching geographic structure to the dendrogram could result from low resolving power of the markers, recombination among loci, or gene flow between regions. Although recombination has occurred in the history of the Bd epidemic , the marker loci do possess a phylogenetic signal as estimated by the permutation tail probability test. The shortest trees estimated by parsimony for the non-permuted dataset were 118 (“hetequal” coding), whereas the mean tree length of 100 permuted datasets was 217.1 (P<0.01). This test remained highly significant after removing repeated genotypes (clone-correction). These data suggest that recombination among loci has not obliterated the phylogenetic signal among genotypes, and that the absence of a broad geographic pattern (Figure 1) lends support to the novel pathogen hypothesis during which long distance dispersal has occurred.
To test whether significant geographic structuring exists between regions, strains were grouped by geographic origin and were tested for significant differences in allele frequency. These tests identified significant differences (P<0.05) between strains from tropical America and all other populations. Temperate American strains were also statistically differentiated from African strains. However, following clone-correction of the dataset by eliminating strains with repeated genotypes, only the difference between temperate American versus tropical American strains remained significant (P<0.05). The power to detect additional genetic structure within the global sample is likely to be limited by low numbers of isolates from poorly sampled regions such as Europe. Nonetheless, some alleles (BdC24, 9893X2, and R6046) were unique to temperate America and Europe and were never detected in African, Australian, or tropical American strains. However, until additional samples are obtained from outside North America, the absence of these alleles may be due to sampling error. Altogether, these data indicate substructure within the global Bd population, implying that contemporary intercontinental gene flow is not prevalent. Instead, the MLST population structure of our isolates is consistent with a single, or a few, introductions between regions followed by population differentiation.
Visual inspection of the dendrogram (Figure 1) shows little effect of host species on pathogen relatedness, suggesting no host specificity of pathogen genotypes. The six strains from Panama of identical genotype were each isolated from a different host species. The stronger geographic than host pattern confirms results from field surveys showing that Bd is a pathogen that undergoes rapid local proliferation with low host specificity ,,. Neither isolates from Xenopus nor from R. catesbeiana formed a cluster and instead were scattered throughout the dendrogram. Although host species is not a good indicator of genetic relatedness or virulence, recent evidence suggests that strains may differ in their virulence ,,. In laboratory trials, strain Melbourne-Llesueuri-00-LB-1 (Me00LB) caused more rapid mortality in Litoria caerulea than strain Rockhampton-Lcaerulea-99-LB-1 (Ro99LB) . These strains are in separate parts of the dendrogram and differ in genotype at 10 loci. Furthermore, a recent study demonstrated that virulence towards Bufo bufo and other phenotypic traits of the pathogen, such as sporangia size, were correlated with genetic relatedness between strains as assessed using the same MLST markers employed in this study . Future studies that address the relative virulence of a large number of Bd strains on a single common model host species (e.g., Bufo bufo) will be required to determine the evolution of virulence in Bd and to reconcile differences in virulence observed in various host by strain inoculation studies ,. These data may help explain the observation of widespread and closely related genotypes found in regions such as Panama and Australia (Figure 1).
Host and geographic patterns of genetic diversity
If Bd has been recently dispersed from a geographic-restricted or host-restricted source population, that source population is expected to harbor greater genetic diversity than populations where the pathogen is newly established. Furthermore, if Bd has been stochastically spread from region to region via long distance dispersal, then each introduction is likely to have caused a concurrent bottleneck in genetic diversity. Genetic diversity can be expressed as “allelic diversity” (number of alleles per locus), “genotypic diversity” (number of unique genotypes in a sample), or as “gene diversity” (expected heterozygosity under Hardy Weinberg equilibrium [HWE] or HE). The difference between genotypic diversity and gene diversity relates to the manner of reproduction; for example, clonal reproduction leads to a proliferation of strains of the same genotype. In the present study, we found a striking lack of allelic diversity. “Global” allele diversity was limited to two alleles per locus, and was even lower in Africa, tropical America, and Australia where the second alleles at BdC24, 9893X2, and R6046 have never been detected (Table 2). In contrast, genotypic and gene diversity were high; the 59 strains were of 44 multilocus genotypes, with a global HE of 0.468 (Table 2). Geographic regions differed little in the number of genotypes recovered per strain sequenced but did differ in HE, with strains from Africa, Australia, and tropical America having lower gene diversity (Table 2). The lower heterozygosity (HE and HO) in tropical American (primarily Panamanian) and Australian populations is consistent with a bottleneck and epidemic spread in these regions that have undergone some of the most severe population declines and pathogen expansions ,. An alternative explanation is that the geographic breadth of these populations is significantly smaller than that sampled in temperate North America.
Measures of allelic and genotypic diversity of Bd strains were not different between healthy (aclinical or subclinical) and sick or dying hosts presenting clinical symptoms of chytridiomycosis (Table 2). Over all host species, HE of strains isolated from aclinical hosts was similar to that of strains from clinical or dead hosts (0.461 vs. 0.453); these differences were not significantly different when tested by permuting samples between groups (P = 0.502). It should be noted, however, that the incubation period of chytridiomycosis may be as long as two months , meaning that classification of animals into sick and aclinical groups can be confounded. Nonetheless, these data provide evidence that there are not highly virulent strains that cause mortality that are a separate genetic pool from strains isolated from highly resistant hosts (e.g., Xenopus).
Comparisons of diversity among host species revealed that the HE and HO of strains from clawed frogs (Xenopus) was lower than the global mean, whereas that of bullfrog strains was greater than the global mean. Despite the limited sampling of isolates from both of these hosts, permutation tests suggested that the increased diversity (HE) among bullfrog strains relative to other hosts was significantly greater than random expectations (P = 0.041), whereas Xenopus diversity was not significantly lower (P = 0.073). Xenopus spp. and the related genus Silurana are found throughout sub-Saharan Africa and appear to have originated in central or eastern equatorial Africa . However, in the present study few African strains and only ones from South Africa and Ghana were included, implying that additional samples encompassing more geographic and species diversity of Xenopus will be required to fully test the “out of Africa” hypothesis for the origin of the Bd pandemic. The absence of increased genetic diversity in African clawed frog samples suggests that X. laevis and S. tropicalis may not be a source population, but these data do not effect the hypothesis that clawed frogs were a vector of the disease from Africa through amphibian exportation.
Evidence for both clonality and recombination
Asexual reproduction is the only form of Bd reproduction documented in the lab, and there is no morphological evidence of sexual reproduction in nature. Examination of molecular genotype data has produced multiple hypotheses on the role of recombination in shaping the population genetics of Bd. Recovery of multiple strains of identical genotype and an excess of heterozygosity at some loci in a global sample of strains suggested outcrossing to be rare or absent , whereas genotype distributions in the Sierra Nevada of California were consistent with both clonality within a site, e.g., lake, and recombination occurring in the history of divergence between lakes . Previous results of fixed heterozygosity for some Bd loci were also coupled with the observation that other loci show HWE proportions of genotypes, a contradiction that was reconciled by invoking a recombination process that is not homogenous throughout the genome . By now including additional MLST loci from a broader sampling of strains and by mapping the loci onto their genomic location, we have determined that genotypic diversity of many strains can be explained as the result of loss of heterozygosity (LOH) that occurs during asexual reproduction, a mechanism that does not involve outcrossing.
Although none of the marker loci showed completely fixed heterozygosity, 7 of the 17 markers had heterozygote excess (negative inbreeding coefficients or FIS) significantly greater than HWE expectations (Table 3; P<0.01). Three of the loci with positive inbreeding coefficients (FIS) showed significant heterozygote deficiency (P<0.05). After applying a clone-correction (elimination of repeated genotypes) all of the loci displaying heterozygote excess were still significantly different from HWE, whereas only one of the loci showing heterozygote deficit (9893X2) was significant. This test for heterozygote excess is highly conservative because the test was computed for the entire global dataset. When data from multiple subpopulations are analyzed together, the Wahlund effect  increases homozygosity, even for randomly mating populations. The Wahlund effect probably explains the significant heterozygote deficit of 9893X2, where the second allele has been detected only in North America and Europe.
The recent sequencing of the genomes of two strains of Bd by the Broad Institute and the Joint Genome Institute now allows these marker loci to be placed into a genomic context. The genome of JEL423 has been assembled into 69 supercontigs: 10 are greater than 1 Mbp, and the largest is 4.44 Mbp (http://www.broad.mit.edu/annotation/genome/batrachochytrium_dendrobatidis). Chromosome sizes in Bd were shown by pulsed-field gel electrophoresis to range from ~0.7–6.0 Mbp  and thus the largest supercontig probably represents a majority of the largest chromosome. Mapping our MLST markers onto their genomic location revealed that all markers on the smaller supercontigs except R6046 had an excess of heterozygotes (>70% observed heterozygosity; significantly different from HWE expectations [P<0.05]), whereas 7 of 9 marker loci on the largest supercontig had higher numbers of homozygous genotypes (positive FIS), and none of the loci displayed significant heterozygote excess (Table 3).
These data suggest that heterozygosity is not uniform across the genome but is reduced on the largest chromosome. Such a pattern is not consistent with frequent sexual reproduction and independent assortment of chromosomes  but could be explained by chromosome-specific variation in mitotic recombination or aneuploidy causing loss of heterozygosity (LOH) within diploid individuals. Selective loss of specific chromosomes, such as chromosome 5 in Candida albicans strains adapting to sorbose containing media , as well as a general relationship between chromosome size and number of sister chromatid exchanges , should both create heterogeneity in rates of LOH among chromosomes. At the population level, chromosome-specific levels of heterozygosity have also been observed in alloploid hybrids of the fungal pathogen Cryptococcus neoformans between varieties neoformans and grubii, “serotype AD hybrids.” These hybrids are unable to complete normal meiotic reduction, and these naturally occurring diploid or aneuploid strains appear to have retained heterozygosity on specific chromosomes while other chromosomes show higher degrees of homozygosity . Although mitotic recombination can only eliminate variation by reducing heterozygosity at certain loci, it has great power to generate genotypic diversity, and this genotypic diversity could facilitate adaptation by exposing beneficial recessive alleles and by increasing the rate of fixation of beneficial mutations ,,. LOH in diploid pathogens has also been implicated in attenuation of virulence and acquisition of drug resistance ,,. Specific evidence for a role of LOH in generating genotypic differences among Bd strains is found in several strain clusters (Figure 2A) that differ only by heterozygosity at a single locus or linked loci (e.g., the two related genotypes PM-05+PM-07 and Me00LB+To05LB differ only in heterozygosity of the marker 8702X2 on supercontig 1.9; Figure 2A). LOH of all of the markers on supercontig 1.1 was found when the six strains of identical multilocus genotype from Panama (JEL408, JEL409, JEL415, JEL423, JEL424, JEL425; Figure 2B) were compared with the two strains from Panama collected a few years earlier (PM-05, PM-07, Figure 2A).
Genotypes for each locus are purple for heterozygous genotypes, red for the higher frequency homozygous genotype, blue for the minority homozygous genotype, and white for missing data. Locus names are shown above the genotypes and are ordered into linkage groups of descending supercontig size based on comparison to genome supercontigs (number shown above the locus names) from the assembly of strain JEL423 (http://www.broad.mit.edu/annotation/genome/batrachochytrium_dendrobatidis). For precise genomic locations, see Table 3. (A) Pattern of LOH showing closely related strains from Panama (green highlighting) and Australia (red highlighting) differing in genotype at a single locus (8702X2). (B) Prevalent genotype from Panama differing from Panamanian genotype shown in (A) by LOH of all markers on supercontig 1.1. (C) Genotypes of strains from the Sierra Nevada showing both LOH (compare JAM083 and JAM084) and highly recombined genotypes (JAM033 and LRJ089).
We tested the fit of a completely non-outcrossing model for the Bd genotype data by inferring ancestral recombination events that occurred during the spread of the epidemic. The genealogy of the strains was assumed to be the same as the dendrogram shown in Figure 1, and the ancestral genotypic states for each locus were inferred using parsimony . Under a completely asexual mode of reproduction (with recombination occurring solely by LOH), a single genealogy should be compatible with all of the loci because genomes are inherited without admixture . Alternatively, where ancestral state transitions occur from one homozygous genotype to the other homozygous genotype or from a homozygous genotype to a heterozygous genotype, genotypic changes must have occurred as the result of outcrossing or mutation. The number of outcrossing/mutation events needed to reconcile the neighbor-joining tree shown in Figure 1 range from 0–6 for each locus. Over half (9/17) of the loci were completely compatible with a LOH-only mode of recombination (zero inferred outcrossing/mutation events). However, results for the remaining 8 loci suggest that sex may have occurred in the history of the epidemic. The possibility of sex in Bd has important epidemiological consequences as sex in the chytrid fungi always leads to the production of a resistant sporangium stage that could facilitate long distance dispersal and survival during hostile seasons. In contrast, the inferred outcrossing events could merely be due to the imprecise estimation of the genealogy or the failure to accommodate linkage into our LOH model. Distinguishing among true outcrossing and a mitotic-only recombination system will require typing a much more extended set of marker loci. Currently, the nearly-fixed heterozygosity of multiple loci in the global population favors the hypothesis that meiosis has not occurred during the current pandemic, perhaps due to genome-wide heterosis.
The population structure and basic genetics of Bd shares a number of similarities with the fungus Candida albicans, a species that is normally a commensal of the gastrointestinal and genitourinary tract biota but also an important opportunistic pathogen. Both are diploid and the population genetic structures of both show a high level of heterozygosity and a primarily asexual mode of reproduction ,,,,. Yet, populations of both species harbor extensive genotypic diversity ,, and in C. albicans it has been suggested that a large proportion of genotypic diversity results from LOH during mitotic division ,. In a recent population genetic analysis of C. albicans strains, a negative correlation was observed between distance of the locus to the centromere and number of sites per locus displaying heterozygote excess . Heterozygosity is expected to decline with increasing distance to the centromere because exchange between sister chromatids during mitosis causes LOH of all loci distal to the breakpoint ,. Such a pattern was observed in the genome sequence of chromosome 7 of C. albicans strain SC5314 , and commensal C. albicans isolates also undergo a process of LOH by sister chromatid exchange in which all markers distal to the breakpoint became homozygous . Centromeres have not been located for Bd, but the data indicate that, at least for the largest chromosome (supercontig1.1), HO is highest in the middle (Table 3), possibly where the centromere would be located.
A model explaining all genotypic diversity for Bd by LOH may not be parsimonious for all portions of the globe. Sierra Nevada populations are at the edge of a current epidemic, yet Bd strains from this region encompass a wide range of HO . California strains (LRJ089, JAM033, JAM083, JAM084; Figure 2C) show evidence of genotypic divergence by LOH (compare JAM083 to JAM084) as well as highly recombined genotypes that would require short gene conversion tracts or double crossovers (JAM033 and LRJ089; Figure 2C). A comparison of commensal strains of C. albicans obtained from the same patient or patient family has revealed three mechanisms for LOH between strain pairs: sister chromatid exchange, chromosomal loss and reduplication, and short tracks of gene conversion . Interestingly, the mechanism depended on both the strain pair and the chromosome investigated, suggesting LOH mechanisms may vary depending on the physical structure and chromosomal gene/locus constituency. In C. albicans mating can occur between diploid strains homozygous for alternate MAT alleles followed by random chromosome loss including homologous recombination to revert back to a diploid genome ,. In tetraploid products of the same fungus, Forche et al.  found that homologous chromosomes commonly experienced multiple recombination/gene conversion events during the completion of the parasexual cycle. The possibility for a parasexual cycle involving diploid-diploid cell fusions in Bd is an interesting possibility, but has been unexplored. One possibility that may explain differences in mode of recombination among populations of Bd is that some populations, such as those from Panama and Australia, lack sexual reproduction, whereas other populations, such as California, may be capable of outcrossing. These differences in mating system among populations could be from the loss of complementary mating types and outcrossing ability through genetic bottlenecks in regions where Bd has been recently introduced; unfortunately knowledge of mating types in Chytridiomycota is essentially nonexistent. Alternatively, the differences in pattern may be due to different routes of LOH in these areas, e.g., sister chromatid exchange in Panama and short gene conversion tracts or outcrossing between very closely related strains in California.
The ancestral genotype of the pandemic
These findings of an excess of heterozygosity throughout the genome but only two alleles per locus suggests that a single diploid lineage of Bd has recently expanded throughout global amphibian populations; this pattern is consistent with the novel pathogen hypothesis. The question remains as to why the fungus has so rapidly spread through apparently healthy populations. Did the pathogen emerge because of a chance dispersal event from a still uncertain source population that allowed a single lineage to exploit novel habitats? Under this hypothesis Bd would be ancestrally diploid and asexual and the pandemic is merely due to dispersal into a novel niche. An alternative scenario is that emergence occurred after an initial diploidization/hybridization event followed by loss of sex and clonal proliferation. In this scenario the ancestral Bd population was haploid and may or may not have been sexual. This hypothesized scenario is realized in Cryptococcus neoformans serotype AD hybrids which display hybrid vigor but may have attenuated virulence ,. The C. neoformans hybrids are nonetheless commonly found in the clinic and environment but are hybrid-sterile and unable to complete meiosis ,. The low amount of variation observed across the Bd genome does not support an allopolyploid hybridization event . However, if the ancestral population had zygotic meiosis and a normally haploid life cycle like the majority of fungi, a diploidization event could be related to the origin of increased infectivity. Diploidization could be the result of an autopolyploidy due to spontaneous genome doubling. A final possibility is that Bd was ancestrally diploid and capable of both sexual and mitotic reproduction (as in most fungi) and that the current clonal epidemic is due to very strong selection on a single genotype. The fungus may be capable of outcrossing but this may be very rare and largely restricted to self-fusions or prevented from LOH by genome-wide heterosis. The absence of mating type control over self fusion of clonal gametes in Chytridiomycota and Blastocladiomycota has been discussed . Currently we are exploring the potential for sex by investigating the polymorphism of genes that encode high mobility group-domain proteins that may be part of putative mating-type loci, as observed in another basal fungal lineage, Mucoromycotina . Distinguishing among these competing hypotheses on the ancestral genome of the pandemic would be greatly facilitated by determining the ancestral ploidy of Bd through investigations of related chytrid species.
All of these theories on the origin of the ancestral genome of a pathogenic Bd predict that samples taken earlier in space or time during the spread of the clone should have the highest heterozygosity (HO). By such measures, North America, Europe, and bullfrog populations are potentially ancestral or sites of earlier introductions, with greater heterozygosity than other populations (Table 2). Our sampling regime was biased towards North America, the native range of the bullfrog, and the effects of this bias are that the population genetic parameters within non-North American populations are difficult to estimate and large pockets of genetic diversity may have been overlooked. Nonetheless, one strain (JEL404) isolated from a bullfrog tadpole from Maine, U.S.A., was heterozygous at all 17 of the loci investigated in this study. Thus, this single strain possesses all of the allelic diversity in the entire global sample, and each strain in the data set could, in theory, be derived from JEL404 by LOH. Two of the loci in this study were microsatellite markers, which are normally among the most polymorphic regions of the genome; this makes the finding of only two alleles per locus surprising given the current large population size predicted by the widespread distribution of the disease. If the global sample represents the expansion of a single widespread diploid lineage, this presents a conundrum because the same pandemic has spread through both declining and apparently healthy host populations.
We hypothesize that a mutation or genomic rearrangement occurred that increased the virulence or transmissibility of Bd from its ancestral state. It is possible that the Bd ancestor may have even been saprobic or on a different host than amphibians. The ability to grow as a cutaneous pathogen could have been the adaptation that spawned the chytridiomycosis epidemic. However, Bd has a number of specialized adaptations that enable it to grow intracellularly within the dynamic tissue of the epidermis, suggesting a long cutaneous history . These include a life cycle length that matches the maturation of the epidermal cells and a discharge tube that opens at the skin surface. Cutaneous diseases may be more rapidly spread than sexually transmitted or blood-borne pathogens because they can be transmitted by incidental direct contact. Cutaneous fungal infections may be directly transferred from host to host without exposure to harsh environmental conditions; because of this effective means of spread and rising human population sizes, skin mycoses are globally on the rise . Frog-to-frog transmission may be important because Bd zoospores do not have a wall and appear to lack sexual reproduction and a concomitant thick-walled resting spore , but see  for possible evidence. We hypothesize that the human mediated movement of frogs for the pet trade, scientific research, or food production followed by direct frog-to-frog transmission may have facilitated the international spread of chytridiomycosis. In addition, climate change may play a role in the transmission of the disease as, for example, longer dry seasons could have led to an increased density of frogs in ponds , and Bd is killed by high temperatures and desiccation ,. While the current pandemic appears to have occurred after the introduction of a novel pathogen into naïve host populations, the outcome of the disease is highly contingent on host as well as environmental factors.
In conclusion, our data show that the current outbreak of Bd is due to the recent emergence of a single successful diploid lineage. We have sequenced 17 loci from 59 strains isolated from across the globe and have found no new alleles beyond the two that had already been detected in a study focused on the Sierra Nevada in California . Thus these data indicate a recent and severe genetic bottleneck, apparently down to a single diploid ancestral strain. Both the extremely low genetic variation and the widespread dispersal of closely related genotypes provide strong support for the novel pathogen over the endemic pathogen hypothesis. None of the sampled regions included in this study provided strong evidence for a source population. However, increased heterozygosity of strains from North American bullfrogs may suggest that this invasive species vectored Bd from more diverse and healthy amphibian populations in eastern North America to ones where declines from chytridiomycosis have occurred ,.
Despite the apparent absence of meiosis, the fungus harbors extensive genotypic diversity, which we hypothesize is largely caused by LOH during mitosis. No more than the global dispersal of a single diploid strain followed by LOH is needed to explain all of the diversity of existing strains. The occurrence of the same lineage in both declining populations and species and apparently stable populations supports previous results suggesting that context dependent effects such as innate host immunity and environmental factors, such as season and temperature, are critical for understanding the outcome of disease for a particular species and region ,. Following the introduction of this lineage of Bd into amphibian populations, some host populations have collapsed, others have crashed and then recovered, and many carry the infection with no obvious population-level effects ,,. The low pathogen genetic diversity and absence of frequent outcrossing imply that Bd may have reduced potential to coevolve with an adapting host population and could suffer from mutational meltdown . Further sampling of polymorphic loci, geographic regions, and hosts will be required to test whether the current global sample is representative of true genetic diversity of the pathogen or is only the first drop of a much larger unsampled pool.
Materials and Methods
All experiments involving animals were performed following protocols approved by the University of Maine Institutional Animal Care and Use Committee (Protocols A98-09-03 and A2001-09-01; JEL), Imperial College London and the British Home Office (MCF), and the Office of Laboratory Animal Care at University of Berkeley (JATM & JWT).
Strains and collection strategy
Molecular analysis of Batrachochytrium dendrobatidis (Bd) was performed on DNA extracted from axenic cultures. Cultures were obtained from living and dead amphibians by isolation onto 1% tryptone agar using routine techniques . The global sample of strains encapsulated all of the known geographic and host diversity of Bd represented by living cultures at the time the study was conducted. Isolation of Bd requires both sterile conditions and technical skill; because of this, few cultures exist in comparison with the known distribution of the disease based on histology and molecular analysis (http://www.parcplace.org/bdmap2008update.html). The sampling strategy used was therefore opportunistic and is biased towards North America for practical reasons. Most of the strains from the Morehouse et al.  study were included as were a subsample encompassing the genotypic diversity of the California strains from the Morgan et al.  study. Strains that were isolated from animals housed or shipped together were specifically eliminated in order to avoid non-independent isolates of the same clonal genotype.
DNA was extracted from cultures grown on 1% tryptone medium with 1% agar. DNA procedures and genotyping used the protocols previously published with the following marker loci: APRT13, CTSYN1, R6046, 6164Y2, 9893X2, mb-13-8b, b7-10c, 6677X2, 6873X2, 8009X2, 8329X2, 8392X2, 8702X2 ,. An additional three loci (BdC5, BdC18, BdC24) were amplified with the following primer pairs: BdC5-F (5′-TAATAGCGCCGACCGAACTA-3′), BdC5-R (5′-ATGCCAAACCATGAGCAAAT-3′), BdC18-F (5′-GCGAATACGACTGCAAATGA-3′) BdC18-R (5′-TGAGCTCTAGCCGACATTGA-3′), BdC24-F (5′-GACAATGTGCTCACGGCTTA-3′), BdC24-R (5′-CTCTCCAAGGCTGAATCTGG-3′). Only polymorphic loci were included in this study. Complete details on the coding of sequences into diploid genotypes and the data set used to conduct genetic analyses are given in Dataset S1. This file displays the coding scheme for each of the sequence types.
An 11,026 bp fragment of the mitochondrial genome was sequenced from 16 strains: JEL203, JEL213, JEL225, JEL230, JEL239, JEL240, JEL253, JEL254, JEL258, JEL262, JEL270, JEL289, JEL404, PM-01, PM-05, PM-07. This fragment spans the intergenic region between the mitochondrial large ribosomal RNA subunit and cytochrome c oxidase subunit 1 and was amplified using long PCR with the enzyme ExTaq (Takara). Sequencing across the fragment was accomplished by primer walking with synthesized oligonucleotides. A deletion spanning 3,904 bp of DNA with no homology to any known sequences in GenBank was observed in the following four strains: JEL225, JEL262, PM-05, and PM-07.
For each locus except BdC18, all of the heterozygous genotypes could be explained by the combination of the two homozygous genotypes, making assignment to alleles trivial. Only for locus BdC18 was there any ambiguity in the phasing of diploid genotypes into haplotypes, apparently due to intralocus recombination. For the purposes of analysis, locus BdC18 was divided into two loci (BdC18.1 and BdC18.2) to reflect the two regions in this marker that did not display complete linkage disequilibrium. The six genotypes observed at the locus could be completely explained by loss of heterozygosity at one or both of two polymorphic sites (BdC18.1 and BdC18.2).
The dendrogram in Figure 1 was constructed using the neighbor-joining algorithm on a distance matrix computed in PAUP v4.0b10  and by statistical parsimony using TCS 1.21 (, Figure S1). Genetic distances between strains were calculated by considering for each locus the two different types of homozygotes two steps apart and the heterozygote 1 step from each homozygous type (hetequal coding). The data were resampled using 1,000 bootstrap pseudo-replicates to test for branch support and tested for phylogenetic signal using the permutation tail probability test in PAUP. The minimum number of outcrossing/mutation events for each locus was mapped onto the neighbor-joining phylogeny using parsimony. Ancestral state reconstruction of genotypes was inferred using MacClade 4 , and the number of instances incompatible with asexual reproduction recorded. For each tree, if the inferred genotype of the ancestor (root) was not heterozygous, then one mutation occurring on an internal node was allowed and not counted as an incompatibility for these trees. Calculation of Hardy-Weinberg disequilibrium, linkage disequilibrium, and tests for population differentiation were conducted using Genepop v3.4 . Basic population descriptors (HE and HO) were estimated in Genetic Data Analysis 1.0d16c , and permutation tests were computed using custom Perl scripts. Estimates of allele richness using rarefaction were calculated using FSTAT v188.8.131.52 .
Genealogical network depicting relatedness among multilocus genotypes of Bd. The network was estimated using the software TCS 1.21 on a distance matrix calculated using “hetequal” coding of alleles (see Materials and Methods section). Each branch represents a single mutational or recombinational step. Ancestral intermediate genotypes not observed in the data set are indicated with open circles, and dashed lines indicate stretched connecting mutations.
(0.04 MB PDF)
Coding scheme for the multilocus sequence typing markers and genotype matrix of global population.
(0.02 MB TXT)
We thank the Broad Institute and the Joint Genome Institute for making publically available draft annotations of the genomes of two Bd strains, and Christina Cuomo and Jason Stajich for helpful insights into the evolution of Bd genomes. We express our gratitude to Peter Murphy who isolated three of the Bd strains. Thanks also to Rebecca Webb for preparations of Bd zoospores.
Conceived and designed the experiments: TJ APL RV JATM JWT MF LB CW LdP JEL. Performed the experiments: TJ APL JATM MF LB CW JEL. Analyzed the data: TJ APL JATM JWT. Contributed reagents/materials/analysis tools: RV JWT LdP JEL. Wrote the paper: TJ MF LB JEL.
- 1. Berger L, Speare R, Daszak P, Green DE, Cunningham AA, et al. (1998) Chytridiomycosis causes amphibian mortality associated with population declines in the rain forests of Australia and Central America. Proc Natl Acad Sci U S A 95: 9031–9036.
- 2. Daszak P, Cunningham AA, Hyatt AD (2003) Infectious disease and amphibian population declines. Divers Distrib 9: 141–150.
- 3. Voyles J, Berger L, Young S, Speare R, Webb R, et al. (2007) Electrolyte depletion and osmotic imbalance in amphibians with chytridiomycosis. Dis Aquat Organ 77: 113–118.
- 4. Blaustein AR, Romansic JM, Scheessele EA, Han BA, Pessier AP, et al. (2005) Interspecific variation in susceptibility of frog tadpoles to the pathogenic fungus Batracbochytrium dendrobatidis. Conserv Biol 19: 1460–1468.
- 5. Daszak P, Strieby A, Cunningham AA, Longcore JE, Brown CC, et al. (2002) Experimental evidence that the bullfrog (Rana catesbeiana) is a potential carrier of chytridiomycosis, an emerging fungal disease of amphibians. Herpetol J 14: 201–207.
- 6. Fisher MC, Garner TWJ (2007) The relationship between the emergence of Batrachochytrium dendrobatidis, the international trade in amphibians and introduced amphibian species. Fungal Biol Rev 21: 2–9.
- 7. Walker SF, Bosch J, James TY, Litvintseva AP, Valls JAO, et al. (2008) Introduced pathogens threaten species recovery programs. Curr Biol 18: R853–R854.
- 8. Bosch J, Martinez-Solano I, Garcia-Paris M (2001) Evidence of a chytrid fungus infection involved in the decline of the common midwife toad (Alytes obstetricans) in protected areas of central Spain. Biol Conserv 97: 331–337.
- 9. Skerratt LF, Berger L, Speare R, Cashins S, McDonald KR, et al. (2007) Spread of chytridiomycosis has caused the rapid global decline and extinction of frogs. EcoHealth 4: 125–134.
- 10. Morgan JAT, Vredenburg VT, Rachowicz LJ, Knapp RA, Stice MJ, et al. (2007) Population genetics of the frog-killing fungus Batrachochytrium dendrobatidis. Proc Natl Acad Sci U S A 104: 13845–13850.
- 11. Rachowicz LJ, Knapp RA, Morgan JAT, Stice MJ, Vredenburg VT, et al. (2006) Emerging infectious disease as a proximate cause of amphibian mass mortality. Ecology 87: 1671–1683.
- 12. Ouellet M, Mikaelian I, Pauli BD, Rodrigue J, Green DM (2005) Historical evidence of widespread chytrid infection in North American amphibian populations. Conserv Biol 19: 1431–1440.
- 13. Longcore JR, Longcore JE, Pessier AP, Halteman WA (2007) Chytridiomycosis widespread in anurans of northeastern United States. J Wildlife Manage 71: 435–444.
- 14. Rothermel BB, Walls SC, Mitchell JC, Dodd CK, Irwin LK, et al. (2008) Widespread occurrence of the amphibian chytrid fungus Batrachochytrium dendrobatidis in the southeastern USA. Dis Aquat Organ 82: 3–18.
- 15. Longcore JE, Pessier AP, Nichols DK (1999) Batrachochytrium dendrobatidis gen et sp nov, a chytrid pathogenic to amphibians. Mycologia 91: 219–227.
- 16. Pessier AP, Nichols DK, Longcore JE, Fuller MS (1999) Cutaneous chytridiomycosis in poison dart frogs (Dendrobates spp.) and White's tree frogs (Litoria caerulea). J Vet Diagn Invest 11: 194–199.
- 17. Lips KR, Diffendorfer J, Mendelson JR, Sears MW (2008) Riding the wave: Reconciling the roles of disease and climate change in amphibian declines. PLoS Biol 6: 441–454. doi:10.1371/journal.pbio.0060072.
- 18. Pounds JA, Bustamante MR, Coloma LA, Consuegra JA, Fogden MPL, et al. (2006) Widespread amphibian extinctions from epidemic disease driven by global warming. Nature 439: 161–167.
- 19. Rachowicz LJ, Hero JM, Alford RA, Taylor JW, Morgan JAT, et al. (2005) The novel and endemic pathogen hypotheses: Competing explanations for the origin of emerging infectious diseases of wildlife. Conserv Biol 19: 1441–1448.
- 20. Daszak P, Berger L, Cunningham AA, Hyatt AD, Green DE, et al. (1999) Emerging infectious diseases and amphibian population declines. Emerg Infect Dis 5: 735–748.
- 21. Laurance WF, McDonald KR, Speare R (1996) Epidemic disease and the catastrophic decline of Australian rain forest frogs. Conserv Biol 10: 406–413.
- 22. Berger L, Speare R, Hyatt A (1999) Chytrid fungi and amphibian declines: overview, implications and future directions. In: Campbell A, editor. Declines and Disappearances of Australian Frogs. Canberra: Environment Australia. pp. 23–33.
- 23. Lips KR, Brem F, Brenes R, Reeve JD, Alford RA, et al. (2006) Emerging infectious disease and the loss of biodiversity in a Neotropical amphibian community. Proc Natl Acad Sci U S A 103: 3165–3170.
- 24. Padgett-Flohr GE, Hopkins RL (2009) Batrachochytrium dendrobatidis, a novel pathogen approaching endemism in central California. Dis Aquat Organ 83: 1–9.
- 25. Weldon C, du Preez LH, Hyatt AD, Muller R, Speare R (2004) Origin of the amphibian chytrid fungus. Emerg Infect Dis 10: 2100–2105.
- 26. Lever C (2003) Naturalized Amphibians and Reptiles of the World. New York: Oxford University Press.
- 27. Morehouse EA, James TY, Ganley ARD, Vilgalys R, Berger L, et al. (2003) Multilocus sequence typing suggests the chytrid pathogen of amphibians is a recently emerged clone. Mol Ecol 12: 395–403.
- 28. Litvintseva AP, Thakur R, Vilgalys R, Mitchell TG (2006) Multilocus sequence typing reveals three genetic subpopulations of Cryptococcus neoformans var. grubii (Serotype A), including a unique population in Botswana. Genetics 172: 2223–2238.
- 29. Ciosi M, Miller NJ, Kim KS, Giordano R, Estoup A, et al. (2008) Invasion of Europe by the western corn rootworm, Diabrotica virgifera virgifera: multiple transatlantic introductions with various reductions of genetic diversity. Mol Ecol 17: 3614–3627.
- 30. Besnard G, Henry P, Wille L, Cooke D, Chapuis E (2007) On the origin of the invasive olives (Olea europaea L., Oleaceae). Heredity 99: 608–619.
- 31. Bosch J, Rincón PA (2008) Chytridiomycosis-mediated expansion of Bufo bufo in a montane area of Central Spain: an indirect effect of the disease. Divers Distrib 14: 637–643.
- 32. Berger L, Marantelli G, Skerratt LL, Speare R (2005) Virulence of the amphibian chytrid fungus Batrachochytrium dendrobatidis varies with the strain. Dis Aquat Organ 68: 47–50.
- 33. Retallick RWR, Miera V (2007) Strain differences in the amphibian chytrid Batrachochytrium dendrobatidis and non-permanent, sub-lethal effects of infection. Dis Aquat Organ 75: 201–207.
- 34. Fisher MC, Bosch J, Yin Z, Stead DA, Walker J, et al. (2009) Proteomic and phenotypic profiling of the amphibian pathogen Batrachochytrium dendrobatidis shows that genotype is linked to virulence. Mol Ecol 18: 415–429.
- 35. Davidson EW, Parris M, Collins JP, Longcore JE, Pessier AP, et al. (2003) Pathogenicity and transmission of chytridiomycosis in tiger salamanders (Ambystoma tigrinum). Copeia 3: 601–607.
- 36. Young S, Berger L, Speare R (2007) Amphibian chytridiomycosis: strategies for captive management and conservation. Int Zoo Yb 41: 85–95.
- 37. Evans BJ, Kelley DB, Tinsley RC, Melnick DJ, Cannatella DC (2004) A mitochondrial DNA phylogeny of African clawed frogs: phylogeography and implications for polyploid evolution. Mol Phylogenet Evol 33: 197–213.
- 38. Hartl DL, Clark AG (1997) Principles of Population Genetics. Sunderland, MA: Sinauer.
- 39. Tibayrenc M, Kjellberg F, Ayala FJ (1990) A clonal theory of parasitic protozoa: the population structures of Entamoeba, Giardia, Leishmania, Naegleria, Plasmodium, Trichomonas, and Trypanosoma and their medical and taxonomical consequences. Proc Natl Acad Sci U S A 87: 2414–2418.
- 40. Janbon G, Sherman F, Rustchenko E (1998) Monosomy of a specific chromosome determines L-sorbose utilization: A novel regulatory mechanism in Candida albicans. Proc Natl Acad Sci U S A 95: 5150–5155.
- 41. Latt SA (1974) Localization of sister chromatid exchanges in human chromosomes. Science 185: 74–76.
- 42. Sun S, Xu J (2009) Chromosomal rearrangements between serotype A and D strains in Cryptococcus neoformans. PLoS ONE 4: e5524. doi:10.1371/journal.pone.0005524.
- 43. Mandegar MA, Otto SP (2007) Mitotic recombination counteracts the benefits of genetic segregation. Proc Roy Soc B-Biol Sci 274: 1301–1307.
- 44. Schoustra SE, Debets AJM, Slakhorst M, Hoekstra RF (2007) Mitotic recombination accelerates adaptation in the fungus Aspergillus nidulans. PLoS Genet 3: e68. doi:10.1371/journal.pgen.0030068.
- 45. Bougnoux ME, Pujol C, Diogo D, Bouchier C, Soll DR, et al. (2008) Mating is rare within as well as between clades of the human pathogen Candida albicans. Fungal Genet Biol 45: 221–231.
- 46. Chen X, Magee BB, Dawson D, Magee PT, Kumamoto CA (2004) Chromosome 1 trisomy compromises the virulence of Candida albicans. Mol Microbiol 51: 551–565.
- 47. Selmecki A, Forche A, Berman J (2006) Aneuploidy and isochromosome formation in drug-resistant Candida albicans. Science 313: 367–370.
- 48. Wu W, Lockhart SR, Pujol C, Srikantha T, Soll DR (2007) Heterozygosity of genes on the sex chromosome regulates Candida albicans virulence. Mol Microbiol 64: 1587–1604.
- 49. Maddison D, Maddison W (2002) MacClade version 4.05: analysis of phylogeny and character evolution. Sunderland, Massachusetts, USA: Sinauer Associates.
- 50. Smith JM, Smith NH, O'Rourke M, Spratt BG (1993) How clonal are bacteria? Proc Natl Acad Sci U S A 90: 4384–4388.
- 51. Miller M, Johnson A (2006) Mating and parasexual genetics in Candida albicans. In: Heitman J, Filler SG, Edwards JE Jr, Mitchell AP, editors. Molecular Principles of Fungal Pathogenesis: Amer Soc Microbiology. pp. 71–88.
- 52. Xu JP, Vilgalys R, Mitchell TG (1999) Lack of genetic differentiation between two geographically diverse samples of Candida albicans isolated from patients infected with human immunodeficiency virus. J Bacteriol 181: 1369–1373.
- 53. Chibana H, Beckerman JL, Magee PT (2000) Fine-resolution physical mapping of genomic diversity in Candida albicans. Genome Res 10: 1865–1877.
- 54. Odds FC, Bougnoux ME, Shaw DJ, Bain JM, Davidson AD, et al. (2007) Molecular phylogenetics of Candida albicans. Eukaryot Cell 6: 1041–1052.
- 55. Xu JP, Boyd CM, Livingston E, Meyer W, Madden JF, et al. (1999) Species and genotypic diversities and similarities of pathogenic yeasts colonizing women. J Clin Microbiol 37: 3835–3843.
- 56. Forche A, May G, Magee PT (2005) Demonstration of loss of heterozygosity by single-nucleotide polymorphism microarray analysis and alterations in strain morphology in Candida albicans strains during infection. Eukaryot Cell 4: 156–165.
- 57. Welker DL, Williams KL (1982) A genetic map of Dictyostelium discoideum based on mitotic recombination. Genetics 102: 691–710.
- 58. Jones T, Federspiel NA, Chibana H, Dungan J, Kalman S, et al. (2004) The diploid genome sequence of Candida albicans. Proc Natl Acad Sci U S A 101: 7329–7334.
- 59. Diogo D, Bouchier C, d'Enfert C, Bougnoux M-E (2009) Loss of heterozygosity in commensal isolates of the asexual diploid yeast Candida albicans. Fungal Genet Biol 46: 159–168.
- 60. Bennett RJ, Johnson AD (2005) Mating in Candida albicans and the search for a sexual cycle. Annu Rev Microbiol 59: 233–255.
- 61. Forche A, Alby K, Schaefer D, Johnson AD, Berman J, et al. (2008) The parasexual cycle in Candida albicans provides an alternative pathway to meiosis for the formation of recombinant strains. PLoS Biol 6: 1084–1097. doi:10.1371/journal.pbio.0060110.
- 62. Lin XR, Nielsen K, Patel S, Heitman J (2008) Impact of mating type, serotype, and ploidy on the virulence of Cryptococcus neoformans. Infect Immun 76: 2923–2938.
- 63. Litvintseva AP, Lin X, Templeton I, Heitman J, Mitchell TG (2007) Many globally isolated AD hybrid strains of Cryptococcus neoformans originated in Africa. PLoS Pathog 3: 1109–1117. doi:10.1371/journal/ppat.1000283.
- 64. Boekhout T, Theelen B, Diaz M, Fell JW, Hop WCJ, et al. (2001) Hybrid genotypes in the pathogenic yeast Cryptococcus neoformans. Microbiology-UK 147: 891–907.
- 65. Litvintseva AP, Kestenbaum L, Vilgalys R, Mitchell TG (2005) Comparative analysis of environmental and clinical populations of Cryptococcus neoformans. J Clin Microbiol 43: 556–564.
- 66. Idnurm A, James TY, Vilgalys R (2007) Sex in the rest: mysterious mating in the Chytridiomycota and Zygomycota. In: J. Heitman JK, Taylor JW, Casselton LA, editors. Sex in fungi: molecular determination and evolutionary implications. Washington, D. C.: ASM Press. pp. 407–418.
- 67. Idnurm A, Walton FJ, Floyd A, Heitman J (2008) Identification of the sex genes in an early diverged fungus. Nature 451: 193–U198.
- 68. Berger L, Hyatt AD, Speare R, Longcore JE (2005) Life cycle stages of the amphibian chytrid Batrachochytrium dendrobatidis. Dis Aquat Organ 68: 51–63.
- 69. Havlickova B, Czaika VA, Friedrich M (2008) Epidemiological trends in skin mycoses worldwide. Mycoses 51: 2–15.
- 70. Di Rosa I, Simoncelli F, Fagotti A, Pascolini R (2007) Ecology - The proximate cause of frog declines? Nature 447: E4–E5.
- 71. Lampo M, Rodriguez-Contreras A, La Marca E, Daszak P (2006) A chytridiomycosis epidemic and a severe dry season precede the disappearance of Atelopus species from the Venezuelan Andes. Herpetol J 16: 395–402.
- 72. Berger L, Speare R, Hines HB, Marantelli G, Hyatt AD, et al. (2004) Effect of season and temperature on mortality in amphibians due to chytridiomycosis. Aust Vet J 82: 434–439.
- 73. Johnson ML, Berger L, Philips L, Speare R (2003) Fungicidal effects of chemical disinfectants, UV light, desiccation and heat on the amphibian chytrid Batrachochytrium dendrobatidis. Dis Aquat Organ 57: 255–260.
- 74. Garner TWJ, Perkins MW, Govindarajulu P, Seglie D, Walker S, et al. (2006) The emerging amphibian pathogen Batrachochytrium dendrobatidis globally infects introduced populations of the North American bullfrog, Rana catesbeiana. Biol Letters 2: 455–459.
- 75. Mazzoni R, Cunningham AA, Daszak P, Apolo A, Perdomo E, et al. (2003) Emerging pathogen of wild amphibians in frogs (Rana catesbeiana) farmed for international trade. Emerg Infect Dis 9: 995–998.
- 76. Woodhams DC, Voyles J, Lips KR, Carey C, Rollins-Smith LA (2006) Predicted disease susceptibility in a Panamanian amphibian assemblage based on skin peptide defenses. J Wildlife Dis 42: 207–218.
- 77. Retallick RWR, McCallum H, Speare R (2004) Endemic infection of the amphibian chytrid fungus in a frog community post-decline. PLoS Biol 2: 1965–1971. 10.1371/journal.pbio.0020351.
- 78. Lynch M, Burger R, Butcher D, Gabriel W (1993) The mutational meltdown in asexual populations. J Hered 84: 339–344.
- 79. Swofford DL (2002) PAUP*: phylogenetic analysis using parsimony (*and other methods). Version 4. Sunderland, Massachusetts: Sinauer Associates.
- 80. Clement M, Posada D, Crandall KA (2000) TCS: a computer program to estimate gene genealogies. Mol Ecol 9: 1657–1660.
- 81. Raymond M, Rousset F (1995) GENEPOP (version 1.2): population genetics software for exact tests and ecumenicism. J Hered 86: 248–249.
- 82. Lewis PO, Zaykin D (2001) Genetic Data Analysis: Computer program for the analysis of allelic data. Version 1.0 (d16c). Available at http://lewis.eeb.uconn.edu/lewishome/software.html.
- 83. Goudet J (1995) FSTAT (Version 1.2): A computer program to calculate F-statistics. J Hered 86: 485–486.
- 84. Haldane JBS (1954) An exact test for randomness of mating. J Genet 52: 631–635.