Worldwide, infectious diseases represent a major source of mortality in humans and livestock. For wildlife populations, disease-induced mortality is likely even greater, but remains notoriously difficult to estimate -- especially for endemic infections. Approaches for quantifying wildlife mortality due to endemic infections have historically been limited by an inability to directly observe wildlife mortality in nature.

Here, we address a question that can rarely be answered for endemic pathogens of wildlife: what are the population- and landscape-level effects of infection on host mortality? We combined laboratory experiments, extensive field data, and novel mathematical models to indirectly estimate the magnitude of mortality induced by an endemic, virulent trematode parasite (Ribeiroia ondatrae) on hundreds of amphibian populations spanning four native species.

We developed a flexible statistical model that uses patterns of aggregation in parasite abundance to infer host mortality.  Our model improves on previous approaches for inferring host mortality from parasite abundance data by i.) relaxing restrictive assumptions on the timing of host mortality and sampling, ii.) placing all mortality inference within a Bayesian framework to better quantify uncertainty, and iii.) accommodating data from laboratory experiments and field sampling to allow for estimates and comparisons of mortality within and among host populations.

Applying our approach to 301 amphibian populations, we found that trematode infection was associated with an average of between 13 and 40% population-level mortality. For three of the four amphibian species, our models predicted that some populations experienced >90% mortality due to infection, leading to mortality of thousands of amphibian larvae within a pond.  At the landscape scale, the total number of amphibians predicted to succumb to infection was driven by a few high mortality sites, with fewer than 20% of sites contributing to greater than 80% of amphibian mortality on the landscape.

The mortality estimates in this study provide a rare glimpse into the magnitude of effects that endemic parasites can have on wildlife populations and our theoretical framework for indirectly inferring parasite-induced mortality can be applied to other host-parasite systems to help reveal the hidden death toll of pathogens on wildlife hosts.

Lab data: We used controlled laboratory experiments to estimate the dose-response curves between R. ondatrae exposure and survival for larvae of the four amphibian species. We collected recently deposited amphibian egg masses (P. regilla, A. boreas, and T. torosa) from field sites or reproductive adults (T. granulosa) and allowed them to lay eggs in the laboratory. Hatching larvae were maintained in carbon-filtered, UV-sterilized tap water at 22 deg. C until being assigned randomly to one of five exposure dosages (0 [control], 20, 40, 100, or 200 cercariae). Snails (Helisoma trivolvis) naturally infected with R. ondatrae were collected from field sites, isolated into 50 ml vials, and allowed to release free-swimming cercariae that were harvested within 4 hr of emergence. Pooled cercariae from multiple snails were administered to an individual amphibian larva within 500 ml of water. Larval amphibians were exposed to cercariae in a single pulse event timed to correspond to early limb development (anurans stage 28, caudates stage 2T). After exposure, we monitored amphibian larvae until death or 20 days following exposure, at which point we dissected hosts to quantify infection.

Field data: To characterize patterns of infection in wild-caught amphibians among species, across populations, and through time, we sampled amphibian hosts from unique ponds (sites) in the East Bay region of California (Alameda, Contra Costa and Santa Clara counties) between 2009 and 2014. This included sites from publicly accessible parks, open space preserves, municipal watershed districts, and private ranches.  Parasitological sampling was timed to overlap with amphibian metamorphosis. We focused on recently metamorphosed amphibians, as these provide a reliable and standardized indicator of R. ondatrae infections acquired during aquatic development.  To measure Ribeiroia abundance per host, we performed a systematic examination of all major tissues and organs in sampled hosts. In the analysis of parasite-induced mortality, we focused on the 301 site-by-year-by-host species combinations with at least five sampled hosts and five detected R. ondatrae metacercariae.

See accompanying README file for metadata.