Shea and Mangel, 2001

Populations of coho salmon (Oncorhynchus kisutch) in California are listed as threatened under the U.S. Endangered Species Act. Such listings refer to adult populations, but often, juvenile life history stages are censused, so it is important to understand what affects the relationship between true adult and observed juvenile numbers. We present models to address how observational uncertainty, census length, and autocorrelation in vital rates affect our ability to observe
trends. We ask two questions about our ability to detect declines in one life history stage from censuses of another. First, given an observed decline in parr numbers, what is the chance that this reflects a decline in adults? Second, given that adult numbers are declining, what is the chance that we see that decline in parr? Our results indicate that statistical power decreases with increasing observational uncertainty and decreasing census lengths and demonstrate how these two parameters interact. Power increases as the level of autocorrelation in mortality rates increases. Management recommendations include obtaining more accurate estimates of autocorrelation in mortality and of observational uncertainty.