Life finds a way: the recovery of frog populations from a chytridiomycosis outbreak

McKnight, Donald T. (2019) Life finds a way: the recovery of frog populations from a chytridiomycosis outbreak. PhD thesis, James Cook University.

PDF (Thesis)
Download (7MB) | Preview
View at Publisher Website:


Emerging infectious diseases are a serious threat to wildlife, but not all populations or species have the same response to outbreaks. In some cases, diseases shift from being epizootic to enzootic, allowing populations to recover, but both the causes of recoveries and the long-term consequences of disease outbreaks remain poorly understood. My PhD aimed to further our knowledge of these important topics by using a frog assemblage in the Australian Wet Tropics as a model system for understanding recoveries from disease outbreaks.

This region was impacted by an outbreak of the fungal disease chytridiomycosis (caused by the pathogen Batrachochytrium dendrobatidis [Bd]) in the late 1980s and early 1990s, during which high elevation populations of several frog species declined or disappeared, while low elevation populations remained stable. Following the outbreak, some species recovered at upland sites, but the patterns of both declines and recoveries vary among species. Litoria dayi disappeared from upland sites and has never recovered. Litoria nannotis disappeared from upland sites and has largely recovered. Litoria serrata declined at upland sites and has recovered, and Litoria wilcoxii did not decline substantially at any elevation. These different histories with the disease presented a great opportunity for studying the factors that allowed some species to recover, while apparently precluding recovery in others, and my thesis examined both population genetics and microbiomes of frogs in this system. My primary goals were to examine the long-term consequences of the outbreak (e.g., fragmentation, inbreeding, loss of diversity) and test several hypotheses for the differences in the history of declines and recoveries among species (e.g., differences in dispersal abilities, a lack of adaptive potential due to lost diversity, differences in microbiomes).

I used single nucleotide polymorphisms to examine connectivity patterns, test for a loss of diversity, and test for Bd-driven selection. I examined low elevation populations of L. nannotis, L. serrata, and L. dayi that survived the outbreak, and compared them to recovered upland populations of L. nannotis and L. serrata. I sampled L. dayi at three national parks and L. nannotis and L. serrata at two national parks. All three species showed high levels of connectivity within a given park, and there was no structuring along streams, suggesting that all three species have good dispersal abilities. No inbreeding was present in any species, and all species showed high genetic diversity levels north of Paluma Range National Park. At Paluma, however, both L. nannotis and L. serrata had reduced genetic diversity, and diversity levels followed a west–east pattern, with higher diversity on the western half of the park (L. dayi does not occur at Paluma). These diversity patterns matched habitat patterns, with higher diversity in wetter areas with larger sections of rainforest, suggesting that the size and quality of refuge habitat may play an important role in the retention of genetic diversity during a disease outbreak. I did not find consistent evidence of selection in L. nannotis, but there was consistency among outlier testes for L. dayi. These tests could not conclusively demonstrate that L. dayi was undergoing diseaseinduced selection, but they were suggestive.

Prior to analysing the microbiomes of the frog species, it was necessary to test or develop several microbiome methodologies. First, microbiome data often need to be normalized prior to analysis, and many methods are available, but several of the most popular methods use variance standardizing techniques that can distort ecological data. Therefore, I compared six methods (rarefaction, proportions, upper quartile, CSS, edgeRTMM, and DESeq-VS) using both a published data set and simulations. My results showed that upper quartile, CSS, edgeR-TMM, and DESeq-VS failed to fully standardize reads, and inflated minor differences among rare micro-organisms while suppressing large differences among common micro-organisms, thus distorting community comparisons. In contrast, using proportions or rarefaction produced accurate results, with proportions outperforming rarefaction.

Another common issue with microbiome studies is the ubiquitous presence of bacterial contamination. This problem has been widely documented, but no method of accurately removing contaminate reads exists. Therefore, I developed an algorithm for identifying and removing contaminate reads, wrote an R package (microDecon) to implement it, and tested it using two large simulations, a published data set, and a sequencing experiment. All tests showed that microDecon was highly accurate and improved the results in 98.1% of cases.

Having tested and developed these methods, I was able to apply them to the microbiomes of frog populations. Multiple laboratory studies have documented beneficial effects of bacteria for amphibian hosts during Bd infections, and several field studies have suggested that microbiomes may play important roles in infection dynamics. Nearly all of this research has focused on bacteria, while the fungal microbiomes of amphibians remain largely unexplored. Therefore, I examined both the fungal and bacterial microbiomes of L. dayi, L. nannotis, L. serrata, and L. wilcoxii to make one of the first comparisons of bacteria and fungi in frog populations and test the hypothesis that differences in microbiomes could explain the differences in patterns of declines and recoveries in the Wet Tropics frog assemblage. I also used qPCR to examine Bd infection prevalence and intensity.

Bacterial microbiomes generally had higher operational taxonomic unit (OTU) richness but lower evenness than the fungal microbiomes. Bacterial microbiomes also tended to be less variable within groups of samples (e.g., frog species), resulting in stronger clustering in ordination plots. Nevertheless, fungal and bacterial Bray-Curtis dissimilarities were positively correlated within frog species (i.e., two individuals with similar fungal microbiomes tended to also have similar bacterial microbiomes). Fungal and bacterial richness were also correlated. This is a somewhat novel result that suggests that either one microbiome is driving the other, or both are being affected similarly by environmental variables.

Results for associations with Bd were mixed. I did not find associations between Bd and beta-diversity for fungi or bacteria. Also, the relative abundance of bacteria that are inhibitory to Bd (based on previous culturing studies) did not follow the expected patterns of association with Bd. Litoria dayi had the highest relative abundance of inhibitory bacteria despite having never recovered from the outbreak, while L. wilcoxii (which never declined) had a low relative abundance of inhibitory bacteria. Additionally, for L. dayi and L. wilcoxii there were significant positive associations between the relative abundance of inhibitory bacteria and Bd infection intensity. In contrast, OTU richness showed negative associations with Bd infection intensity for both fungi and bacteria. Additionally, for both fungi and bacteria, L. dayi had the lowest OTU richness of any frog species. These results are consistent with a protective effect of OTU richness and suggest that a lack of richness in L. dayi has played a role in its inability to recover from the outbreak.

In summary, I found that having large areas of high-quality lowland habitat is likely important for allowing populations to retain genetic diversity during an outbreak, and they should be a focus of conservation efforts. Additionally, neither differences in genetic diversity nor differences in dispersal abilities could explain why L. dayi has been unable to recover from population declines. There was some evidence that L. dayi is in the process of adapting, but this was not conclusive. The microbiome data did not show significant associations between Bd and either total community composition or the relative abundance of inhibitory bacteria, but there were associations with the OTU richness of both fungal and bacterial microbiomes, suggesting that richness may be an important factor in infection dynamics.

Item ID: 64195
Item Type: Thesis (PhD)
Keywords: 16S, adaptation, bacteria, bioinformatics, Bray-Curtis, chytridiomycosis, community comparisons, controls, decontaminate, diversity, emerging infectious diseases, evenness, fragmentation, gene flow, genetic drift, microbiome, ordination, population genetics, principal coordinates analysis, simulation
Related URLs:
Copyright Information: Copyright © 2019 Donald T. McKnight.
Additional Information:

For this thesis, Donald McKnight received the Graduate Research School Medal of Excellence.

Five publications arising from this thesis are stored in ResearchOnline@JCU, at the time of processing. Please see the Related URLs. The publications are:

Chapter 2: McKnight, Donald T., Schwarzkopf, Lin, Alford, Ross A., Bower, Deborah S., and Zenger, Kyall R. (2017) Effects of emerging infectious diseases on host population genetics: a review. Conservation Genetics, 18. pp. 1235-1245.

Chapter 3: McKnight, Donald T., Lal, Monal M., Bower, Deborah S., Schwarzkopf, Lin, Alford, Ross A., and Zenger, Kyall R. (2019) The return of the frogs: the importance of habitat refugia in maintaining diversity during a disease outbreak. Molecular Ecology, 28 (11). pp. 2731-2745.

Chapter 5: McKnight, Donald T., Huerlimann, Roger, Bower, Deborah S., Schwarzkopf, Lin, Alford, Ross A., and Zenger, Kyall R. (2019) Methods for normalizing microbiome data: an ecological perspective. Methods in Ecology and Evolution, 10 (3). pp. 389-400.

Chapter 6: McKnight, Donald T., Huerlimann, Roger, Bower, Deborah S., Schwarzkopf, Lin, Alford, Ross A., and Zenger, Kyall R. (2019) microDecon: a highly accurate read‐subtraction tool for the post‐sequencing removal of contamination in metabarcoding studies. Environmental DNA, 1 (1). pp. 14-25.

Appendix 1: McKnight, Donald T., Alford, Ross A., Hoskin, Conrad J., Schwarzkopf, Lin, Greenspan, Sasha E., Zenger, Kyall R., and Bower, Deborah S. (2017) Fighting an uphill battle: the recovery of frogs in Australia’s Wet Tropics. Ecology, 98 (12). pp. 3221-3223.

Date Deposited: 28 Aug 2020 00:26
FoR Codes: 05 ENVIRONMENTAL SCIENCES > 0502 Environmental Science and Management > 050202 Conservation and Biodiversity @ 30%
06 BIOLOGICAL SCIENCES > 0605 Microbiology > 060502 Infectious Agents @ 35%
06 BIOLOGICAL SCIENCES > 0604 Genetics > 060411 Population, Ecological and Evolutionary Genetics @ 35%
SEO Codes: 96 ENVIRONMENT > 9604 Control of Pests, Diseases and Exotic Species > 960405 Control of Pests, Diseases and Exotic Species at Regional or Larger Scales @ 50%
96 ENVIRONMENT > 9608 Flora, Fauna and Biodiversity > 960806 Forest and Woodlands Flora, Fauna and Biodiversity @ 50%
Downloads: Total: 23
Last 12 Months: 23
More Statistics

Actions (Repository Staff Only)

Item Control Page Item Control Page