Predicting seasonal density patterns of California cetaceans based on habitat models

Abstract

Temporal variability in species distribution remains a major source of uncertainty in managing protected marine species, particularly in ecosystems with significant seasonal or interannual variation, such as the California Current Ecosystem (CCE). Spatially explicit species- habitat models have become valuable tools for decision makers assisting in the development and implementation of measures to reduce adverse impacts (e.g. from fishery bycatch, ship strikes, anthropogenic sound), but such models are often not available for all seasons of interest. Broadscale migratory patterns of many of the large whale species are well known, while seasonal distribution shifts of small cetaceans are typically less well understood. Within the CCE, species- habitat models have been developed based on 6 summer-fall surveys conducted during 1991 to 2008. We evaluated whether the between-year oceanographic variability can inform species predictions during winter-spring periods. Generalized additive models were developed to predict abundance of 4 cetacean species/genera known to have year-round occurrence in the CCE: common dolphins Delphinus spp., Pacific white-sided dolphin Lagenorhynchus obliquidens, northern right whale dolphin Lissodelphis borealis, and Dall's porpoise Phocoenoides dalli. Predictor variables included a combination of temporally dynamic, remotely sensed environmental variables and geographically fixed variables. Across-season predictive ability was evaluated relative to aerial surveys conducted in winter-spring 1991 to 1992, using observed:predicted density ratios, nonparametric Spearman rank correlation tests, and visual inspection of predicted and observed distributions by species. Seasonal geographic patterns of species density were captured effectively for most species, although some model limitations were evident, particularly when the original summer-fall data did not adequately capture winter-spring habitat conditions.