Humpback whale (Megaptera novaeangliae) location in Southeast Alaska : modeling the influence of mesoscale krill (Euphausiacea) patch depth and size

Abstract: Humpback whales ([i]Megaptera novaeangliae[/i]) require dense prey aggregations due to the high energetic cost of their feeding methods. Modeling studies have shown that when whales are on their feeding grounds, they are closely associated with high prey concentrations. These studies collapse the prey field into a two dimensional measure of prey abundance in horizontal space. However, patchiness is one of the dominant characteristics of marine systems. The three-dimensional structure of prey is known to affect rorqual whale (family Balaenopteridae) feeding kinematics and prey patchiness affects the foraging decisions of central-based marine predators. This study aims to assess how spatial and temporal patterns in krill distribution affect the distribution of humpback whales across the feeding area of Frederick Sound and Lower Stephens Passage, Southeast Alaska, from 2006-2008. Data from hydroacoustic surveys were used to identify prey patches, and statistics related to their depth, size and backscatter intensity were calculated. These statistics, along with measures of prey hot spots for each survey, persistent prey hot spots, bathymetry and season were used in a spatially explicit Generalized Additive Model (GAM) to explain humpback whale distribution. The model showed that whales preferred large, dense prey patches that occurred persistently. Whales were rarely found at patches covering less than 22 000 square meters vertically. With regards to depth, shallower patches at 40 meters depth were preferred. However, patches located at depths of approximately 90 meters were larger, occurred frequently, and were favored by whales as well. There was some indication that whales were more likely to be located at persistent prey hot spots. A comparison of the prey patch explicit model and a model built using horizontal prey densities indicated that incorporating measures of prey depth and patch size greatly improved models of humpback whale distribution. The patch explicit model explained 59% of whale counts, while the horizontally averaged prey model explained 35%. Incorporating patch-explicit variables can be used to improve future modeling efforts for rorqual whale species abundance and distribution.