MEPS 336:249-265 (2007)  -  doi:10.3354/meps336249

Modeling the pelagic habitat of salmon off the Pacific Northwest (USA) coast using logistic regression

Hongsheng Bi1,*, Rachel E. Ruppel1, William T. Peterson2

1Cooperative Institute for Marine Resources Studies, Oregon State University, and 2National Marine Fisheries Service, Northwest Fisheries Science Center, Hatfield Marine Science Center, 2030 Marine Science Drive, Newport, Oregon 97365, USA

ABSTRACT: Defining marine habitat use for Pacific salmon Oncorhynchus spp. is important for effective resource management because salmon production has been linked to ocean conditions in the Northeast Pacific. Towards that goal, Chinook O. tshawytscha and coho salmon O. kisutch populations were sampled off Washington and Oregon, USA, in June 1998 to 2005 along with habitat variables including temperature, salinity, water depth, and chlorophyll a concentration. Correlation analysis and stepwise logistic regressions were run to identify the physical and biological factors that predict the presence of Chinook and coho salmon. Low zero-catch probability was used to indicate used habitat. For all life history stages, zero-catch probability decreased with increased chlorophyll concentration and decreased depth. Temperature was a significant predictor variable for subyearling Chinook and yearling coho presence based on stepwise logistic regression. The size of used habitat showed large spatial and temporal variations, where more used habitat occurred off Washington and the Columbia River mouth than off Oregon. This pattern may relate to a wider shelf and greater primary production to the north. The largest amount of used habitat occurred in 2000 and 2003 for all 5 life history stages examined. Sea-viewing Wide Field-of-Sensor (SeaWiFS) satellite images indicated high chlorophyll concentration in that period.

KEY WORDS: Salmon · Habitat modeling · Logistic regression · Chlorophyll a · Depth · Temperature · Salinity · Fisheries

Full text in pdf format