Evaluation of the Use of Gill Nets for Monitoring Reservoir Striped Bass Fisheries
Brian J. McRae, James S. Bulak, Barbara E. Taylor, and Christian T. Waters
Abstract.—Gill nets are commonly used to evaluate recreational and commercial striped bass Morone saxatilis fisheries. Using historical gill-net data for striped bass from four southeastern reservoirs, we fitted selection curves to evaluate size selectivity of the nets, assessed the effect of net selectivity on stock assessment metrics, defined the variability associated with two commonly used survey metrics, catch per unit effort (CPUE) and length at age, and evaluated the likelihood of observing a true change in these metrics between two time periods. Selection curves fitted to data from an experimental gill net with bar mesh sizes of 25, 38, 51, 64, and 76 mm indicated that the net was most efficient at capturing striped bass around 600 mm in length; efficiency of the net declined sharply for fish less than 300 mm in length. Adjusting catch data for net selectivity did not have a significant effect on either mortality rates from age 1 to age 3 or mean length at age. Variability associated with length-at-age data was relatively small while variability in CPUE data was relatively large. We found that historical sample sizes of age-1 and age-2 striped bass would be adequate to detect changes in mean length at age as small as 15 mm between two 5-year periods with a power of 0.8. However, in most instances, the historical number of net sets per year needed to be substantially increased to detect either a 30% or 50% change in CPUE between the two periods with a power of 0.8. In three of the four reservoirs, the historical number of nets set per year was sufficient to detect a 100% change in CPUE between the two periods with a power greater than 0.8. Variance associated with abundance-based metrics will generally be more difficult to overcome than that associated with fish-specific data or cohort-specific data, such as mean length at age. We suggest that managers carefully evaluate the population-specific variability associated with the metrics of interest and consider the acceptable level of risk for failing to detect a change.