Characterization and classification of acoustically detected fish spatial distributions

High-resolution, two-dimensional measurements of aquatic-organism density are collected routinely during echo integration trawl surveys. School-detection algorithms are commonly used to describe and analyse spatial distributions of pelagic and semi-pelagic organisms observed in echograms. This approach is appropriate for species that form well-defined schools, but is limited when used for species that form demersal layers or diffuse pelagic shoals. As an alternative to metrics obtained from school-detection algorithms, we used landscape indices to quantify and characterize spatial heterogeneity in density distributions of walleye pollock (Theragra chalcogramma). Survey transects were divided into segments of equal length and echo integrated at a resolution of 20 m (horizontal) and 1 m (vertical). A series of 20 landscape metrics was calculated in each segment to measure occupancy, patchiness, size distribution of patches, distances among patches, acoustic density, and vertical location and dispersion. Factor analysis indicated that the metric set could be reduced to four factors: spatial occupancy, aggregation, packing density, and vertical distribution. Cluster analysis was used to develop a 12-category classification typology for distribution patterns. Visual inspection revealed that spatial patterns of segments assigned to each type were consistent, but that there was considerable overlap among types.