Morphological Discrimination of Genetically Distinct Chinook Salmon Populations: an Example from California's Central Valley

Abstract

Listen

Translate article

AbstractWe compared a length‐at‐date growth model to truss network (morphometric) models for classifying emigrating juvenile (age 0 through age 1) California Central Valley Chinook Salmon Oncorhynchus tshawytscha to three genetically identified races at upstream (salmon generally younger and smaller) and downstream (salmon generally older and larger) monitoring stations. Morphometric models including capture date and head to standard length ratio performed better than the growth model at distinguishing genetically assigned fall‐run and late fall‐run juveniles; prediction accuracy increased from 54.6% to 63.8% for upstream salmon and from 17% to 73% for downstream salmon. The growth model may over inflate downstream estimates of federally listed Chinook Salmon by misidentifying fall and late‐fall runs (nonlisted) as winter and spring runs (both listed) as much as 83% of the time. Morphometric models did not improve run assignment for the listed runs; the growth model outperformed morphometric models at downstream stations. Morphometric models including head shape and sample date had similar accuracy measures to models, including multiple fish measurement ratios, suggesting head shape is the strongest predictor of the juvenile Central Valley Chinook Salmon race. Our results indicate morphometric modeling can improve identification of nonlisted juvenile Central Valley Chinook Salmon from federally listed runs, potentially benefitting monitoring, water management, and protection of sensitive species.Received May 5, 2014; accepted August 13, 2014