Connectivity over a disease risk gradient enables recovery of rainforest frogs.

Chytridiomycosis has been a key driver of global frog declines and extinctions, particularly for high-altitude populations across Australia and the Americas. While recent evidence shows some species are recovering, the extent of such recoveries and the mechanisms underpinning them remain poorly resolved. We surveyed the historical latitudinal and elevational range of four Australian rainforest frogs that disappeared from upland sites between 1989 and 1994 to establish their contemporary distribution and elevational limits, and investigate factors affecting population recovery. Five rainforest streams were surveyed from mountain-base to summit (30 sites in total), with swabs collected from the target species (Litoria dayi, L. nannotis, L. rheocola, and L. serrata) to determine their infection status, and data loggers deployed to measure microclimatic variation across the elevational gradient. Infection probability increased with elevation and canopy cover as it was tightly and inversely correlated with stream-side air temperature. Occupancy patterns suggest varying responses to this disease threat gradient. Two species, L. rheocola and L. serrata, were found over their full historical elevational range (≥1,000 m above sea level [asl]), while L. dayi was not detected above 400 m (formerly known up to 900 m asl) and L. nannotis was not detected above 800 m (formerly known up to 1,200 m asl). Site occupancy probability was negatively related to predicted infection prevalence for L. dayi, L. nannotis, and L. rheocola, but not L. serrata, which appears to now tolerate high fungal burdens. This study highlights the importance of environmental refuges and connectivity across disease risk gradients for the persistence and natural recovery of frogs susceptible to chytridiomycosis. Likewise, in documenting both interspecific variation in recovery rates and intraspecific differences between sites, this study suggests interactions between disease threats and host selection exist that could be manipulated. For example, translocations may be warranted where connectivity is poor or the increase in disease risk is too steep to allow recolonization, combined with assisted selection or use of founders from populations that have already undergone natural selection.