Examining the utility of fishery and survey data to detect prey removal effects on Steller sea lions (Eumetopias jubatus)

Abstract

One focus of mitigation for Steller sea lion (Eumetopias jubatus) declines in Alaska has been to restrict commercial fishery activity around sea lion rookeries and haul-outs. However, a variety of statistical hypothesis tests have failed to relate sea lion population metrics to fish and fishing variables, prompting speculation that regulations may be unwarranted. In this study, we use simulation to show that standard hypothesis tests often have overstated power to detect a relationship between Steller sea lion vital rates and fish or fishing variables. The power and utility of hypothesis tests largely depend on choosing appropriate dependent and independent variables. In particular, pup counts were the most effective for diagnosing fecundity effects, and successive ratios of adult counts were the most effective for diagnosing survival effects. Fish relative abundance was the most effective independent variable, with other choices (e.g., fishery catch) often resulting in misleading inferences. We argue that Bayes factors are best suited for characterizing the relationship between fish abundance and Steller sea lion vital rates and that existing evidence does not preclude a strong relationship between sea lion fecundity and the availability of commercially harvested fish stocks.