Counting capelin: a comparison of acoustic density and trawl catchability

Abstract

Acoustic estimates of capelin, Mallotus villosus (Müller), density were compared with catches in 97 Campelen 1800 bottom and 113 IYGPT midwater fishing sets to assess catching efficiency (q _e ) relative to acoustic estimates, and the ability of trawl and acoustic surveys to index capelin abundance. Catches in experimental IYGPT sets targeted at capelin and towed at constant depth or undulated over 40 m did not differ from acoustic predictions over a range of densities from 0.00001 to 0.2 fish per m ² (q _e =1; p<0.05). Catch and q _e did not differ between fishing methods (p>0.05, paired t-test). Campelen catches were typically bigger than acoustic estimates in the trawl zone at low or medium densities, likely a consequence of dead-zone non-detectability. Campelen q _e s were strongly density-dependent (>1 at densities >0.05 fish per m ³ ; <1 at higher densities), and ranged over several orders of magnitude, making catches representative only of presence/absence. For the IYGPT trawl undulated at fixed survey stations, not targeted at capelin, mean q _e averaged 11.6, and high variability, in part related to vertical distribution (s.e.=3.76), limited usefulness as an abundance index. A recognition threshold caused acoustic integrations based on a priori allocation of backscatter to underestimate capelin distributed at low densities (<0.05 fish per m ³ ), as did dead-zone bias. A posteriori allocation reduced bias. It is concluded that acoustic integration supported by directed trawling is the most reliable method of counting capelin.