Using Cure Models for Analyzing the Influence of Pathogens on Salmon Survival

Abstract

Parasites and pathogens influence the size and stability of wildlife populations, yet many population models ignore the population-level effects of pathogens. Standard survival analysis methods (e.g., accelerated failure time models) are used to assess how survival rates are influenced by disease. However, they assume that each individual is equally susceptible and will eventually experience the event of interest; this assumption is not typically satisfied with regard to pathogens of wildlife populations. In contrast, mixture cure models, which comprise logistic regression and survival analysis components, allow for different covariates to be entered into each part of the model and provide better predictions of survival when a fraction of the population is expected to survive a disease outbreak. We fitted mixture cure models to the host–pathogen dynamics of Chinook Salmon Oncorhynchus tshawytscha and Coho Salmon O. kisutch and the myxozoan parasite Ceratomyxa shasta. Total parasite concentration, water temperature, and discharge were used as covariates to predict the observed parasite-induced mortality in juvenile salmonids collected as part of a long-term monitoring program in the Klamath River, California. The mixture cure models predicted the observed total mortality well, but some of the variability in observed mortality rates was not captured by the models. Parasite concentration and water temperature were positively associated with total mortality and the mortality rate of both Chinook Salmon and Coho Salmon. Discharge was positively associated with total mortality for both species but only affected the mortality rate for Coho Salmon. The mixture cure models provide insights into how daily survival rates change over time in Chinook Salmon and Coho Salmon after they become infected with C. shasta.

Received August 5, 2013; accepted October 26, 2013

ACKNOWLEDGMENTS

We thank Julie Alexander for providing feedback and support in the development and analysis of the models. The findings and conclusions in this article are those of the authors and do not necessarily represent the views of the USGS or the U.S. Fish and Wildlife Service. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government. This project was supported by funding from the U.S. Bureau of Reclamation through Cooperative Agreement R09AC20022, Cooperative Ecosystems Study Unit 3FC810873.