UBC Theses and Dissertations
Search and decision in fishing systems Shotton, Ross
Methods of search in marine fisheries are reviewed, and aspects of variables necessary to evaluate exploitation strategies are considered. Estimates of fish school densities based on gas diffusion principles are made. Possible school patterns are examined using the Poisson, Poisson with added zeroes, Neyman type A and Negative Binomial distributions. The Negative Binomial distribution was found to give best fit to the data used although the Neyman type A distribution gave a better estimate of the data variance. School size was reasonably well described by a log distribution, Mean square dispersion rates, modal velocities and direction of movement for sets of observations on tuna schools are found. Tests on the significance of direction of movement by groups of tuna school are done and those groups showing significance were tested for homogeneity of direction. Confidence intervals on the movement of direction are also calculated. Decision methods so as to increase catch are considered for three situations: (1) Where the position of a school is known with some error, and it must be decided whether to attempt to locate it or remain searching on the present grounds. (2) When the Bayes estimate of catch rate on the present grounds is less than that expected on an alternate ground, and the decision of changing depends on minimizing an expected loss function. (3) For the Vancouver trawl fishery where fish occurrence may be considered of a present or absent nature, and it is desired to minimize the time spent steaming between grounds. The use of entropy as a criterion of search effectiveness as used by other workers is also discussed.
Item Citations and Data