We have developed a dynamic epidemiological model to evaluate potential transmission routes maintaining a viral pathogen in anadromous fish populations of the Columbia River Basin (CRB), with consideration for how missing data influence inference. In the CRB, infectious hematopoietic necrosis virus (IHNV) causes severe disease, predominantly in juvenile steelhead trout (Oncorhynchus mykiss) and less frequently in Chinook salmon (O. tshawytscha). Mortality events following IHNV infection can be devastating for individual hatchery programs. Despite reports of high local mortality and extensive surveillance efforts, there are questions about how viral transmission is maintained. Modeling this system offers important insights into disease transmission in natural aquatic systems, as well as about the data requirements for generating accurate estimates about transmission routes and infection probabilities. We present a simulation analysis of the model in economically and culturally important salmonid populations and demonstrate model application using data from 14 cohorts (2000-2013) at 24 sites in a sub-region of the CRB, the lower Columbia River (LCR). We interpret model results from the LCR in light of conclusions from six simulated scenarios.
Our simulations demonstrated that the model accurately identified routes of transmission and inferred infection probabilities when there was testing of all cohort-sites. When testing records were incomplete, the model accurately inferred which transmission routes were exposing particular cohort-sites but infection probabilities were biased. In the LCR, exposure to IHNV via the return migration of adult fish is an important route for maintaining IHNV, and the probability of infection following this exposure was relatively high at 0.16. Although only 5% of cohort-sites experienced self-exposure by infected juvenile fish, this transmission route had the greatest probability of infection (0.22). Increased testing of adult fish may improve inferential ability. Increased use of secure water supplies and continued use of biosecurity protocols may reduce IHNV transmission to juvenile salmonids at hatcheries. Conclusions from this study can direct the use of the model to quantify the relative importance of transmission routes for other important aquatic pathogens, including the bacterial agents that cause bacterial kidney disease and coldwater disease in salmonids, or complex multicellular pathogens such as Ceratonova shasta.