Summary: | Thesis (M.Sc.)--Memorial University of Newfoundland, 2001. Mathematics and Statistics Bibliography: leaves 99-101 In this dissertation, two model designs for estimating selection curves of trawl fishing gears are studied. -- The first, a probabilistic model design is introduced in this dissertation. It is a Three- Part Model Design use in Trouser Trawl Experiments. In this model design, the problem of Greenland halibut (turbot) fish meshed in the forepart component of the fishing net is addressed and retention probabilities for all length classes are derived from specific experimental data. Furthermore, a bell-shaped selection curve is adopted for the forepart component of the gear. This model design will enable us to make more accurate assessment of fishing efficiency of trawl gears and their ability to release undersized fish. -- The second probabilistic model design introduced in this dissertation is the Modified Alternate Haul Model Design use in Trawl Experiments. In this model design, the control gear (which measures the total population) involved in traditional alternate haul experiments is eliminated. By utilizing this model design, an arbitrary number of gear modifications (including standard gear) can be compared from limited experimental data. This model design will also enable the experimental data collected on fishing gears evaluation, to be used to produce their individual selection curves. Application of this model design has significant potential to reduce monetary costs and waste of undersized fish, and improve efficiency of future experimental work. -- In this dissertation, several examples are provided to illustrate applications of two proposed model designs. The results of applying the models to the experimental data sets collected in Phrase I study on Greenland halibut (turbot) in Canadian Waters are most encouraging.
|