UBC Theses and Dissertations
Growth and survival responses to experimental fishing : models, experiments and lessons from the Northern pikeminnow of South Central British Columbia Taylor, Nathan Gabriel
The objective of this was study was to determine if species specific fishing could produce "cultivation-depensation effects" in an aquatic ecosystem with two predatory, competing fish species. I identified a unique stock of northern pikeminnow living in series of connected lakes that has obligatory rearing in specific nursery lakes; developed two novel likelihoods to measure the growth, movement and mortality responses; developed an ecosystem model to predict how the system would respond to fishing, and finally, compared the ecosystem modeling predictions to observed responses. My research showed northern pikeminnow in South Central B.C. have obligatory rearing in specific nursery lakes then disperse to other lakes as adults. I argue that this large scale spatial ontogeny can be solely explained by temperature cues to spawn and that the distribution of adults is determined by density-dependent dispersal that equalize very large productivity and effective density differences between lakes. I showed that in spite of being included in many stock assessments, and being used as proxies for natural mortality estimates and for exploitation rate targets, von Bertalanffy growth parameters are not generally estimated correctly. The data used to do so are virtually always biased due to: size-selective gears, populations subjected to fishing and natural mortality and in some cases, size-dependent movement. I developed two new likelihoods to simultaneously estimate growth and mortality parameters: one for length-age data, and another for mark-recapture data. The first performs well across a range of recruitment anomalies and steady state fishing mortalities but fails when fishing rates have been variable (especially increasing) and when gear selectivity is dome-shaped. The second likelihood works well with simulated data but is not robust to assumptions of constant recruitment and measurement error being violated. I combined length-age and mark-recapture data to show using simulated sampling that it is possible to simultaneously estimate growth, mortality, and movement parameters where sufficient numbers of fish are observed moving. The assumptions required for these models to perform well are very restrictive. I used a simple ecosystem model and compared the predictions to observed responses following depletion fishing in two-fish lake systems with rainbow trout and northern pikeminnow. Consistent with model predictions, growth was slower and mortality of juvenile rainbow trout higher relative to the control in lakes where northern pikeminnow were removed, while adult rainbow trout survival remained unchanged. Visual survey indices of northern pikeminnow fry indicated survival of 1+ fish worsened and 2+ improved following fishing. Consistent with model predictions, no obvious mortality or growth responses were observed in adult fish in either rainbow trout or northern pikeminnow removals. While the agreement between the model and observations was encouraging, field testing such complex predictions was fraught with difficulty. The probability distributions of the parameters of interest were very broad. Also, the model predicts that survival and behavioral dynamics producing the greatest differences in direction and magnitude of ecosystem response occur in size classes of fish and groups of zooplankton that are difficult to observe. It was not possible to conclude whether an alternate state was produced through cultivation-depensation effects. To do so would require longer term data on recruitment responses, vulnerability exchange processes and survival data of young age classes of fish. This study identifies several shortcomings in our ability to predict and detect how ecosystems will respond to fishing. First, our ability to measure even simple response variables such as growth and mortality is not good. Secondly, even if we could, the direction and magnitude of these responses can be highly counter-intuitive. Finally, those processes with the most violent effects on our predictions are those for which we have very little information, namely dynamics determining the spatial distribution of the stock, the dynamics of young fish and behaviorally mediated predation rates.
Item Citations and Data