Integrating models to predict distribution of the invertebrate host of myxosporean parasites

Abstract

Manayunkia speciosa, a freshwater polychaete, is the invertebrate host of myxosporean parasites that negatively affect salmonid populations in the Pacific Northwest of the USA. Factors that drive the distribution of M. speciosa are not well understood, which constrains our understanding of disease dynamics and the development of management solutions. We described the distribution of M. speciosa at 3 sites on the Klamath River, California, based on 2-dimensional hydraulic models (2DHMs) and a generalized linear mixed model (GLMM). 2DHMs were built to explain hydraulic variation at each site and used to stratify biological sampling effort along depth–velocity gradients and by substrate class. We assessed the presence/absence of M. speciosa at 362 georeferenced locations in July 2012 and built GLMMs to describe relationships between hydraulic and substrate variables and the distribution of M. speciosa. The best-fitting GLMMs demonstrated that M. speciosa distributions were associated with depth–velocity conditions and substrate size during base discharge (area under the receiver operating characteristic curve [AUC] = 0.88) and at peak discharge (AUC = 0.86). We evaluated the GLMMs with an independent data set collected in July 2013 (n = 280) and found that the top models predicted the distribution of M. speciosa with a high degree of accuracy (AUC = 0.90). These results support the conclusion that the summer distribution of M. speciosa is related to observed hydraulic and substrate conditions during base discharge (summer) and modeled hydraulic and substrate conditions during peak discharge (late winter to early spring). These results may have implications for the use of flow manipulation as a disease management tool. These results also illustrate the importance of examining species distribution data in the context of temporally disconnected environmental factors and demonstrate how models can fulfill this need.