Species identification of small pelagic fish schools by means of hydroacoustics in the Eastern Mediterranean Sea


Published: Jan 9, 2015
Keywords:
Schooling behaviour hydroacoustics school descriptors discriminant function analysis MEDIAS anchovy sardine
K. TSAGARAKIS
M. GIANNOULAKI
M. M. PYROUNAKI
A. MACHIAS
Abstract

 

Reliable biomass estimates by means of hydroacoustics largely depend on the correct identification of acoustic targets. Data collected during five summer acoustic surveys (2004-2008) in the North Aegean Sea (Greece) were analyzed to explore effective discrimination of small pelagic fish schools according to the species they belong. Discriminant Function Analyses (DFA) using bathymetric, energetic and morphometric school descriptors as explanatory variables were applied per research cruise as well as to pooled data from all surveys. Results revealed that the schools can be successfully classified into the five species considered (anchovy Engraulis encrasicolus, sardine Sardina pilchardus, round sardinella Sardinella aurita, blue whiting Micromessistius poutassou, and Atlantic mackerel Scomber scombrus). The percentage of correct classifications in terms of number of schools was higher in the analyses of the annual cruises (75.6%-95.36%) than in the analysis of pooled data (≈72%). This is because of (i) the lower number of species, as well as (ii) the reduced intraspecific variability, occurring in each separate cruise. Significant differences were detected among school descriptors for the different species, revealing discrete aspects of schooling behaviour for each species. The benefit of the specific approach is that the classification functions of the DFAs can be used to classify a larger set of schools, which has not been possible to assign to specific species. Overall the approach constitutes an objective, more automated and less time consuming procedure for the analysis of acoustic data and can contribute to the improvement of biomass estimates in the area.

Article Details
  • Section
  • Research Article
Downloads
Download data is not yet available.
Most read articles by the same author(s)