| Summary: | Traditional fisheries models are based upon simplistic Poisson assumptions concerning fleet behavior. Fishing vessels are assumed to operate independently of one another and to sample the fish population in a random fashion. This dissertation provides a review of the major component processes contributing to the operation of fishing fleets and uses field data on salmon purse seiners to present tests of the hypotheses contained in the historical assumptions. Data pertaining to interference competition and information were analyzed. Since fisheries management usually entails assumptions concerning the form of exploitation rate responses to effort, the consequences of non-random fisherman behavior were explored. Alternative models of the fishing process were proposed and examined.
In British Columbia, salmon purse seiners line up at fishing access points, forming well defined queues. These queues were measured over time in Johnstone Strait using a one dimensional recording scale.
The distribution of effort was fit to theoretical truncated Poisson and truncated negative binomial distributions. Most data fit the negative binomial rather than the Poisson. Movement patterns and time series of catches were also non-random. Analysis of variance methods indicated that line-up lengths reflected set catch rates.
Waiting times were quantified using functional and statistical models. Using the waiting times, the fleet set effort and number of sets per boat were calculated. Although the fleet set effort was a near linear function of the number of boats in the area, interference competition produced an initial decrease in sets per vessel.
Two models were presented for exploitation rates in relation to queuing patterns. The overflight model was based upon the line-up distributions and assumed that information was good. The model fit well and the parameter estimates reflected anecdotal and statistical information about fish behavior. The exploitation rates saturated at an effort level of 100 vessels.
As an alternative model, the negative binomial distribution was used to estimate exploitation rates from catch per vessel distributions. It was assumed that salmon abundance does not affect the shape of the distributions. As effort increased, the distribution of catch per vessel was predicted to become more skewed to the origin. The parameter describing the shape of the distribution, k, should have tracked the fishing power of the fleet (decreasing as the distribution became more skewed).
After fitting the weekly distributions, it was found that the relative exploitation rates from the sales slip model did not saturate like the parameters of the overflight model. An alternative derivation indicated that the shape of catch per unit effort distributions responds to the size and aggregative properties of the fleet and to the magnitude of the catch. As the mean catch per set increases, k will increase. Salmon abundance, fleet numerical responses and vessel aggregations affect the skewness of the catch per unit effort distributions.
In general, traditional model assumptions were rejected. Vessels did not operate independently. Boats were not distributed in a random fashion. The overflight model provided predicted exploitation rates. The exploitation response to effort was qualitatively distinct from the forms incorporated in traditional models. === Science, Faculty of === Zoology, Department of === Graduate
|
|---|
