Date of Award


Document Type

Open Access Thesis


Marine Science


College of Arts and Sciences

First Advisor

Ryan Rykaczewski


The Pacific Decadal Oscillation (PDO) has been used to characterize the dominant basin-scale climate variability, in the North Pacific. The PDO index has been correlated to Pacific salmon (Oncorhynchus spp.) population dynamics in the Gulf of Alaska (GOA) during the late 1900s, but the stationarity of the statistical relationships between salmon production and the basin-scale index (and between the regional physical processes and the basin-scale index) has not been quantified. A change in the relationship between the PDO and salmon catches has been noted during the late 1980s, motivating further investigation of variability in climate and ecosystem properties during this period.

To test relationships between North Pacific basin-scale and GOA regional scale oceanographic variability, an empirical orthogonal function (EOF) analysis was used to explore changing temporal and spatial patterns in oceanographic properties during the mid 1900s to early 2000s. Physical properties and climate indices were compared across the winter of 1988/89 to characterize the variability in relationships between properties in different periods. EOF loading patterns were compared using rolling 15-year windows to estimate variability and consistency of patterns throughout the 1900s and 2000s. An EOF analysis was also used to compare temporal metrics of oceanic and atmospheric variability along the GOA continental shelf with residuals from Pacific salmon spawnerrecruitment curves.

In this thesis, I show that many physical properties in the GOA exhibit nonstationary relationships across the 1988/89 boundary. Relationships between physical

properties and salmon productivities also exhibit non-stationarities during the late 1980s. Only SSHa showed a shift in the dominant EOF spatial pattern in the late 1980s that was consistent with the proposed shift in salmon-climate relationships. SST did not show a sustained temporal or spatial change in the GOA, but rather a short-lived spatial change in the 1980s, that reverted back to the pre-1980s pattern. No other property showed a change in patterns of variability during the late 1980s, nor did any properties maintain a stationary relationship with salmon indices. Local and regional SSHa indices may better represent variability in the GOA and the properties that impact salmon survival than basin-scale indices.