Identifying interactions among salmon populations from observed dynamics

A simple direct correlation analysis of individual counts between different populations often fails to characterize the true nature of population interactions; however, the most common data type available for population studies is count data, and one of the most important objectives in population and community ecology is to identify interactions among populations. Here, I examine the dynamics of the spawning abundance of fall-run chinook salmon spawning within the California Central Valley and the Klamath Basin, California, and the Columbia River Basin, Oregon. I analyzed multiple time series from each watershed using a multivariate time-series technique called maximum autocorrelation factor analysis. This technique was used for finding common underlying trends in escapement abundance within each watershed. These trends were further investigated to identify potential resource-mediated interactions among the three groups of salmon. Each group is affected by multiple trends that are likely to be affected by environmental factors. In addition, some of the trends are coherent with each other, and the differences in population dynamics originate from variations in the relative importance of these trends among the three watershed groups.