| Summary: | Thesis (Ph.D.)--University of Washington, 2013 Trawls are used extensively during fisheries abundance surveys to derive estimates of fish density and, in the case of acoustic-based surveys, to identify acoustically sampled fish populations. However, trawls are selective in what fish they retain, resulting in biased estimates of density, species, and size compositions. Selectivity of the midwater trawl used in acoustic-based surveys of walleye pollock (Theragra chalcogramma) was evaluated using multiple methods. The effects of trawl selectivity on the acoustic-based survey abundance estimates and the stock assessment were evaluated for the Gulf of Alaska walleye pollock population. Selectivity was quantified using recapture, or pocket, nets attached to the outside of the trawl. Pocket net catches were modeled using a hierarchical Bayesian model to provide uncertainty in selectivity parameter estimates. Significant under-sampling of juvenile pollock by the midwater trawl was found, with lengths at 50% retention ranging from 14 - 26 cm over three experiments. Escapement was found to be light dependent, with more fish escaping in dark conditions. Highest escapement rates were observed in the aft of the trawl near to the codend though the bottom panel of the trawl. The behavioral mechanisms involved in the process of herding and escapement were evaluated using stereo-cameras, a DIDSON high frequency imaging sonar, and pocket nets. Fish maintained greater distances from the trawl panel during daylight, suggesting trawl modifications such as increased visibility of netting materials may evoke stronger herding responses and increased retention of fish. Selectivity and catchability of pollock by the midwater trawl was also investigated using acoustic density as an independent estimate of fish abundance to compare with trawl catches. A modeling framework was developed to evaluate potential explanatory factors for selectivity and catchability. Selectivity estimates were dependent on which vessel was used for the survey, ...
|
|---|