Acoustic estimates of zooplankton biomass and distribution: application of canonical correlation to scaling of multifrequency acoustic data

A technique is presented that can be used to scale multifrequency acoustic data to estimates of wet weight biomass from concurrently collected zooplankton samples. The technique uses canonical correlation to identify those taxa that correlate with the acoustic data. Canoncial correlation is then used to compute scaling coefficients that can be used to convert transect data to biomass equivalents for the correlating taxa. Examples are offered to illustrate how the technique can identify euphausiid aggregates, pteropods, and larval fish. The technique can discriminate between taxa with gas bladders, taxa with shells, and fluid-filled taxa. It also has the advantage of quantifying the correlation between the target taxa and acoustic data, thus allowing the researcher to judge the usefulness of the acoustic data in characterizing the distribution and density of the zooplankton taxa.