Disentangling risks to an endangered fish: using a state-space life cycle model to separate natural mortality from anthropogenic losses

Abstract

State-space population models are becoming a common tool to guide natural resource management, because they address the statistical challenges arising from high observation error and process variation while improving inference by integrating multiple, disparate datasets. A hierarchical state-space life cycle model was developed, motivated by delta smelt (Hypomesus transpacificus), an estuarine fish experiencing simultaneous risks of entrainment mortality from out-of-basin water export and natural mortality. Notable model features included a covariate-dependent instantaneous rates formulation of survival, allowing estimation of multiple sources of mortality, and inclusion of relative observation bias parameters, allowing integration of differently scaled abundance indices and entrainment estimates. Simulation testing confirmed that two sources of mortality, process variation, and data integration parameters could be estimated. Delta smelt entrainment mortality was associated with environmental conditions used to manage entrainment, and recruitment and natural mortality were related to temperature, freshwater flow, food, and predators. Although entrainment mortality was reduced in recent years, ecosystem conditions did not appear to support robust spawning or over-summer survival of new recruits, manifesting as a 98% reduction of adults during 1995–2015.