Case Study Applications of LRP Estimation Methods to Pacific Salmon Stock Management Units.

The revised Fisheries Act requires that Limit Reference Points (LRPs) be identifed for all major fsh stocks. For Pacifc salmon, major fsh stocks are represented by stock management units (SMUs). An SMU is composed of one or more salmon conservation units (CUs), which are the assessment units under the Wild Salmon Policy, WSP. We introduce methods to estimate LRPs at the SMU level that integrate statuses derived under the WSP at the CU level. We demonstrate and evaluate the LRPs for three case study SMUs: Interior Fraser Coho (Oncorhynchus kisutch), West Coast Vancouver Island (WCVI) Chinook (O. tshawytscha), and Inside South Coast Chum (O. keta) - excluding Fraser River. Methods are divided into two categories: CU status-based LRPs and aggregate abundance LRPs. CU status-based LRPs are recommended as the default method, and are based on the proportion of CUs above levels associated with increased risk of extinction (above 'Red' status) under the WSP. Aggregate abundance methods may be used supplementally to meet specifc fsheries management requirements. Aggregate abundance LRPs are subdivided into logistic regression LRPs and projection LRPs. Both types of aggregate abundance LRPs are defned at the SMU-level abundances associated with a desired probabilty of all component CUs being above Red status, but they differ in that logistic regression LRPs are determined directly from historical data while projection LRPs are determined from projections of CU-level population dynamics. We discuss suitability and requirements for the application of the various LRP estimation methods, drawing from the range of data and information availability among the case studies. In general, the application of aggregate abundance LRPs may be limited to SMUs where the CU-level populations covary, as demonstrated for the Interior Fraser Coho case study, and where covariance has not changed over time or, for projection LRPs, those changes can be parameterized. [ABSTRACT FROM AUTHOR]