Relating marine mammal distribution to water column prey structure derived from echosounding

Abstract

Managing human impacts on marine mammal populations depends on understanding their distributions in space and time. Knowledge of these distributions is becoming increasingly important due to offshore energy development and climate-induced shifts in oceanic habitat. Most models assessing pelagic marine mammal abundance and distribution primarily use ocean surface or ocean floor variables as proxies for water column habitat. Here, we used a ship-based marine mammal sighting survey to test the utility of echosounder-based predictive variables for modeling marine mammal distribution and abundance. We assessed the distribution of 7 marine mammal taxa and 3 feeding guilds along the shelf break off the northeast USA relative to prey structure derived from acoustic data. We classified prey into 4 categories: (1) fish with swimbladders; (2) small resonant bubbles as phytoplankton, fish larvae, and gelatinous zooplankton; (3) fluid-like scatterers such as krill and copepods; and (4) fish with no swimbladder. We quantified the spatial structure of prey by calculating backscattering strength, location, dispersion, occupied areas, evenness, and aggregation. Spatial resolution along the survey track line was set to bins 1000 m in distance and either 50 or 200 m in depth. We then built generalized additive models (GAMs) using these acoustically derived variables to explain marine mammal distribution. The resulting GAMs explained between 9 and 38% of deviance, with model fit often reflecting aspects of foraging depth and prey preference. This approach could contribute to improved management through more accurate species distribution models that employ direct measurements of prey.