Maximum Likelihood Estimation of Kokanee and Sockeye Salmon Spawners in a Stream Using Visual Survey Data, 2003-2017

Abstract

A Maximum Likelihood estimation procedure was developed for the joint assessment of resident kokanee and anadromous \n sockeye salmon in the Okanagan River, British Columbia. The model uses visual survey counts and is an extension of other areaunder- \n the type models used for estimation of salmon escapement. Alternative hypotheses were tried concerning observation error \n structure, arrival times, and survival patterns for a common observer efficiency estimate. The final model is described here, with \n information on setting bounds and constraints for parameter estimation, comparisons made and the results obtained. Estimates \n of abundance and approximate confidence intervals are comparable to those obtained from other investigations. Since 2001, \n Sockeye accounted from 6%-38% of the spawners, with no evidence from surveys and estimates to indicate kokanee are negatively \n impacted Sockeye, perhaps because of the estimated spatio-temporal segregation pattern. Recommendations are made concerning \n future adjustments and surveys for systems with two Nerkids that cannot always be readily distinguished in visual surveys. The \n results obtained are important to justify continued efforts to re-introduce Sockeye into Canada, maintain hatchery production \n to supplement Columbia River Sockeye fisheries, and protect existing biodiversity in this multi-species ecosystem. Management \n implications for the Columbia River fisheries in the US and Canada are discussed.