Temporal trends in portfolio effects and stability of a salmon population complex

Background/Question/Methods

Recent research has highlighted the importance of interpopulation diversity in fostering the stability of population complexes. Here we focus on California’s Central Valley fall-run Chinook salmon (Oncorhynchus tshawytscha) and use estimates of potential returns (escapement plus estimated harvest) to each major river within the Central Valley from 1957-2008 to quantify overall system stability as well as the correlation among returns to individual rivers, a crucial contributor to "portfolio effects" that can stabilize stock complexes.  It has been proposed that straying and/or constrained ocean entry dates of hatchery-sourced fish may have contributed to to increased homogenization and weakened portfolio effects in this system. To examine potential increasing homogenization of this system, we quantified temporal variation in the proportion of total escapement occurring to rivers with hatcheries and used moving-windows to quantify trends in stability of returns to the system as a whole, to rivers with and without hatcheries, and to examine trends in the correlations among rivers in their returns.  We also used wavelets to quantify changes in the intensity and period of variability.

Results/Conclusions

The proportion of total Central Valley escapement occurring to rivers with hatcheries increased through time.  Concurrently, variability in returns (escapement plus estimated harvest) for the entire Central Valley also increased through time with most of the increase in variability seemingly occurring since the mid-1980s.  Variability in escapement to rivers with hatcheries also increased during this period, but the increase was not statistically significant as variability was higher to begin with.  Until the mid-1980s, mean pairwise correlations in potential returns were consistently higher among rivers with hatcheries than those without.  Later, rolling 10-yr mean pairwise correlations increased rapidly for rivers without hatcheries until they nearly matched those for rivers with hatcheries.  There was also a slow but significant increase through time in correlations of returns among rivers with hatcheries over the entire period considered.  Wavelet analysis of basin-wide returns revealed that overall variability in returns increased, and there was a shift with reduced variation at short (i.e., approximately generation time) period and increased variation with longer period.  However, this pattern was not always displayed by individual rivers.