Accuracy of catch-effort methods for estimating fish density and biomass in streams

Abstract

At each of 11 localities a section of stream was closed off with nets and an electrofisher used to estimate the abundance of fishes in the section. Each section was fished from 5-7 times with each fishing equalling one unit of effort. Using the catch-effort methods of Leslie, DeLury and Ricker, separate estimates were made for each species. In several cases species were split into size groups and estimates made for each group. The fish remaining in each section after the fishings were collected using rotenone. Thus the estimates could be compared to the actual number of fish present. Estimates were considered to be either 'good', if the regressions used in the above methods were statistically significant or 'bad' if they were not significant. Lower limits for the number of fish and mean weight of a fish for 'good' estimates were identified. The Leslie and Ricker estimates, which did not differ significantly, were least in error. They tended to underestimate (-21.6% on the average for the Leslie method). Direct estimates of biomass did not differ significantly from those made using the estimates for numbers and the mean weight of fish caught. The interrelationships among variables such as mean weight, numbers, catchability, density, biomass, number of catches used, proportion of fish taken during the estimate, number of fish in the last catch and their relationships with the error of the estimates were examined using correlation and principal components analysis. Error was most closely related to the proportion of fish collected. The effects of other variables such as mean weight affected error through catchability and subsequently the proportion of fishes caught. It was not possible to predict a significant proportion of the error using variables which could be measured without a complete collection. The effects of locality, electrofisher, and species on error were examined. Each accounted for a significant proportion of the variability in error but primarily by affecting the proportion of fish caught. These results suggest that the most appropriate way of decreasing error would be to increase the total effort and consequently the proportion of fish collected. This would be best done by increasing the number of fishings used in the estimate. Catchability tended to decrease in successive fishings. The observed trends in changing catchability accounted for most of the error. Size-selectivity, which was evident as a change in mean weight in successive catches, was not significantly associated with changing catchability. © 1980 Dr. W. Junk b.v., Publishers.