Evaluation of visual survey methods for monitoring Pacific salmon (Oncorhynchus spp.) escapement in relation to conservation guidelines

Abstract

Canada’s Wild Salmon Policy requires that biological status of conservation units of Pacific salmon (Oncorhynchus spp.) be assessed regularly in relation to abundance-based benchmarks. Visual survey methods, in which periodic counts of spawning fish are made throughout a season, will likely be used for this purpose because they provide a cost-effective means of monitoring interannual trends in escapement. Trend detection performance for visual survey methods depends mainly upon consistency in (i) the ability of observers to detect fish and (ii) the annual timing of fish presence in the survey area. We developed a Monte Carlo simulation procedure to evaluate the ability of four visual survey methods (peak count, mean count, trapezoidal area-under-the-curve (AUC), and likelihood AUC) to detect 30% declines in coho salmon (Oncorhynchus kisutch) escapement over 10 years (i.e., the magnitude of trend that would warrant listing a coho population as threatened using the listing criteria of the Committee on the Status of Endangered Wildlife in Canada (COSEWIC)) given realistic levels of variability in these two factors. The mean count outperformed all other approaches across a wide range of scenarios about true population dynamics and survey designs, suggesting that a simple mean count method is suitable for monitoring coho escapements in relation to COSEWIC guidelines.