Files
Abstract
Quantifying changes in any ecological assemblage is a difficult, but reservoir fish assemblages are particularly challenging. Reservoir sampling data are highly variable temporally and spatially because of variation in fish distribution and behavior (especially schooling), differential sampling efficiency across different habitats, and gear selectivity. The affect these different sources of variation have on the strength of inferences regarding total fish abundance is poorly understood, especially for passive gears such as gill-netting. Consequently, I used a series of novel approaches evaluate reservoir fish assemblages and the utility of those approaches to make strong inferences. In the first approach, I model the relationship between gill-net catch and acoustic-derived densities and a variety of covariates that could reasonably be expected to affect encounter rates with gill-nets. The modelling approach did not elicit specific time periods or sampling conditions that would result in more reliable estimates of relative abundance. However, additional avenues of pursuit for reconciling discrepancies were provided. In the second approach, I evaluated whether reduced sampling effort could adequately characterize the reservoir fish assemblage as compared to results obtained from monthly sampling. Results showed that characterizing the assemblage with gill-nets can be performed with 3 or 4 months of sampling per year, provided the sampling is carried out across seasons. In the third approach, I evaluated Mean Rank Shift, a new species rank approach, that has been proposed as offering promise for detecting changes in assemblages derived from long-term monitoring studies. MRS has been used in a fisheries context in a few published studies, but the approach has not been critically evaluated. Through simulation, I evaluated the sensitivity of MRS (as proposed) and several formulations I developed, for detecting directional changes in a reservoir fish assemblage. Results indicate that two novel formulations of MRS show promise for assessing changes in a community over time, especially in the face of perturbations that result in directional changes. Though challenges still exist, my results show that reservoir fish managers can characterize fish assemblages with a reduced amount of effort and that MRS shows promise for assessing changes in a community over time.