Defining Reporting Groups: Essential Framework for Mixed Stock Analysis

Recent laboratory advances have increased the potential resolution available with mixed stock analysis of fishery mixtures.  Genetic data are now available from increasing numbers of individuals and populations across many geographic scales.  In addition, more genetic markers are available, including markers specifically developed to differentiate among populations. With increased resolution comes the benefit of the ability to identify groups of populations on a finer scale when reporting stock contribution estimates.  However, there is a cost to increasing the number of groups, especially when most groups contribute small proportions (0-5%) to the mixture.  These costs are generally associated with error due to bias and decreased precision of the estimates.  In light of these costs and benefits, the definition of reporting groups for mixed stock analysis provides the essential framework affecting accuracy and precision when reporting fishery stock composition estimates.  As end-users of the information, stakeholder interests provide a starting point and influence the final determination of reporting groups while biology and statistics provide the constraints to the definition.  We present methods to 1) use stakeholder input to establish desired reporting groups, 2) test our ability to identify the desired reporting groups, and 3) arrive at a final set of reporting groups.  We also propose a dynamic reporting group strategy for tabulating stock composition information that takes into account the geographic location of the fishery.  The method is demonstrated in the context of the Western Alaska Salmon Stock Identification Project.