Experimental Harvest Policies for a Mixed-Stock Fishery: Fraser River Sockeye Salmon, Oncorhynchus nerka

Abstract

We investigated harvest strategies aimed at rebuilding the less abundant stocks of Fraser River sockeye salmon, Oncorhynchus nerka. Monte Carlo simulations were run to estimate catch under four different harvest policy designs and three alternative parameterizations of the Ricker model. A pooled regression model was fit to 34 yr of spawner–recruit data from the 10 major stocks of Fraser sockeye. Compared with estimating separate parameters for each stock, the pooled regression model resulted in a more precise estimate of the Ricker a parameter (productivity at low stock sizes). Exploitation rate for maximum sustainable yield depends only on the a parameter and is thus well defined by the pooled regression even though the corresponding optimal escapement levels remain uncertain. A reduction in harvest rates to 70% from the current average 80% increased the simulated 40-yr catch by 31%. More extreme harvest-rate reductions, involving 50% exploitation rates on some stocks for four generations, allowed faster rebuilding and further increases in catch. Extreme harvest-rate reductions were necessary to obtain a 0.5 or greater probability of correctly detecting increased escapements, should the less-abundant runs increase as expected.