Microbial symbionts of vertebrate skin have an important function in defense of the host against pathogens. In particular, the emerging chytrid fungus Batrachochytrium dendrobatidis, causes widespread disease in amphibians but can be inhibited via secondary metabolites produced by many different skin‐associated bacteria. Similarly, the fungal pathogens of terrestrial salamander eggs Mariannaea elegans and Rhizomucor variabilis are also inhibited by a variety of skin‐associated bacteria. Indeed, probiotic therapy against fungal diseases is a recent approach in conservation medicine with growing experimental support. We present a comprehensive Antifungal Isolates Database of amphibian skin‐associated bacteria that have been cultured, isolated, and tested for antifungal properties. At the start, this database includes nearly 2000 cultured bacterial isolates from 37 amphibian host species across 18 studies on five continents: Africa, Oceania, Europe, and North and South America. As the research community gathers information on additional isolates, the database will be updated periodically. The resulting database can serve as a conservation tool for amphibians and other organisms, and provides empirical data for comparative and bioinformatic studies. The database consists of a FASTA file containing 16S rRNA gene sequences of the bacterial isolates, and a metadata file containing information on the host species, life‐stage, geographic region, and antifungal capacity and taxonomic identity of the isolate.

Comparisons of present and past occurrences suggest that populations of six frog species endemic to the tropical rainforests of northern Queensland have declined during the past ten years. Most declines have occurred at high altitudes in the southern portions of the tropical rainforest. An extensive survey conducted during the summer of 1991-1992 did not locate any individuals of two upland species, Litoria nyakalensis and Taudactylus rheophilus. Another upland species, T. acutirostris, which formerly was widely distributed, appears to have declined in rainforests south of the Daintree River. Three species (Litoria nannotis, L. rheocola and Nyctimystes dayi) were absent from most upland sites south of the Daintree River, but were common at lowland sites and at all sites north of the Daintree River. Aspects of water chemistry, including inorganic ions, heavy metals, and pesticide residues, were analysed for many sites. These analyses failed to identify any abnormalities that might have contributed to frog declines. Declines appear to be unrelated to the history of forestry or mining at sites, or to low rainfall in wet seasons. Levels of habitat disturbance by feral pigs appear to have increased at some sites in recent years and, either by this disturbance or through direct predation, feral pigs may have contributed to declines in some populations. However, pigs are unlikely to be the sole cause of frog population declines. Once declines have occurred, fragmentation of rainforest habitats may prevent recolonization from adjacent sites. Until causal agents associated with declines can be identified, management strategies to ensure the long-term survival of these species must involve protection of the riparian habitats in which they occur.

Contamination is a ubiquitous problem in microbiome research and can skew results, especially when small amounts of target DNA are available. Nevertheless, no clear solution has emerged for removing microbial contamination. To address this problem, we developed the R package microDecon (https://github.com/donaldtmcknight/microDecon), which uses the proportions of contaminant operational taxonomic units (OTUs) or amplicon sequence variants (ASVs) in blank samples to systematically identify and remove contaminant reads from metabarcoding data sets. We rigorously tested microDecon using a series of computer simulations and a sequencing experiment. We also compared it to the common practice of simply removing all contaminant OTUs/ASVs and other methods for removing contamination. Both the computer simulations and our sequencing data confirmed the utility of microDecon. In our largest simulation (100,000 samples), using microDecon improved the results in 98.1% of samples. Additionally, in the sequencing data and in simulations involving groups, it enabled accurate clustering of groups as well as the detection of previously obscured patterns. It also produced more accurate results than the existing methods for identifying and removing contamination. These results demonstrate that microDecon effectively removes contamination across a broad range of situations. It should, therefore, be widely applicable to microbiome studies, as well as to metabarcoding studies in general.

No abstract

No abstract