Exploring the role of oceanographic features in the spatial distribution of Pacific halibut and other longline-caught fishes off the west coast from southern Oregon to Queen Charlotte Sound, British Columbia

Knowing how groundfish distribute in relation to ocean conditions is of primary importance to fishery managers as they are tasked with estimating stock size and designing effective monitoring programs amidst a changing climate. This research examined near-bottom environmental data alongside Pacific halibut survey catch data for the years 2006-2009 on the continental shelf of Oregon, Washington, and southern British Columbia. The objectives of the research were: 1) characterize summer environmental and fish distribution conditions at depth observed during the setline survey throughout the study area; 2) explore ranges and possible tolerance thresholds for temperature, DO, salinity, and pH for Pacific halibut and six other species commonly caught during the survey including arrowtooth flounder, spiny dogfish, sablefish, longnose skate, lingcod, and yelloweye rockfish; 3) identify the primary environmental factors affecting distribution of these species and model the observed relationships; and 4) assess the ability of the longline survey to adequately account for target species across environmental gradients.;All of the organisms except sablefish were found in increasing numbers from south to north and occupied varying depth strata. Upwelling-induced seasonal hypoxia (i.e. DO less than 1.4 ml/L) is a feature of the study area and all of the organisms except sablefish exhibited apparent species-specific DO minimum thresholds between 0.75 ml/L to 1.0 ml/L, i.e. where animals were found above, but not below these thresholds. Ordinary least squares (OLS) multiple regression analysis indicated a variety of variables unique to each species as significant in predicting distribution, with temperature, DO, and pressure as the most common. Poor model performance led to the use of two other methods, geographically weighted regression (GWR) and tree regression, to examine regional variation and mitigate the effect of correlated explanatory variables. The tree regression model, which allows for predictor variable correlation, was a comparable or better fit than OLS and GWR for all species except Pacific halibut, where GWR, using DO as a predictor variable, was the best fit. GWR addresses regional variation of processes. Results here suggest that many of the animals, including halibut, use avoidance as a coping mechanism for below-threshold DO. However, given that GWR yielded the best fitting model for halibut, low (but above threshold) DO may also be contributing to catchability differences in the survey. Multivariate analysis was used to examine the relationship among the environmental variables and to examine species assemblages in relation to these variables. Depth and latitude were of primary importance in describing relationships.