Herding in groundfish and effective pathwidth of trawls

Abstract

The swept area method requires quantitative information on the effective pathwidth W eff of a trawl to estimate absolute densities of groundfish. Herding of some groundfish by the bridles, sweeps and doors of a trawl will extend W eff beyond that of the net proper and limits applicability of that method. Generally, the two most commonly used values for W eff (net width and door spread) will result in biased density estimates. We developed a model relating fish catch to net width and door spread that allows estimation of effective herding distance H, catch due to the net proper C W, catch due to herding C H, and W eff, thereby significantly improving the swept area method. We analysed trawl catch data using a special case of this model and found that herding occurred in at least 14 of 36 abundant northern Australian groundfish. For Lutjanus malabaricus, the target species in the Arafura Sea trawl fishery, H = 73.90 (asymptotic standard error (ASE) 18.11) m. Thus C W = 26.44 (ASE 6.06) kg h −1, C H = 43.48 (ASE 13.57) kg h −1 and W eff = 35.64 (ASE 5.97) m for a trawl with a door spread of 60 m and a net width of 15 m. Other species had H > 80 m or were not herded. Our findings stress the need to examine herding and review previous applications of the swept area method for estimating fish biomass.