607 American Fisheries Society Symposium 77:607–620, 2011 © 2011 by the American Fisheries Society Effect of Electrofishing Sampling Design on Bias of Size-Related Metrics for Blue Catfish in Reservoirs KRISTOPHER A. BODINE*, DAVID L. BUCKMEIER, AND J. WARREN SCHLECHTE Texas Parks and Wildlife Department, Heart of the Hills Fisheries Science Center 5103 Junction Highway, Mountain Home, Texas 78058, USA DANIEL E. SHOUP Department of Natural Resource Ecology and Management, Oklahoma State University 008C Ag Hall, Stillwater, Oklahoma 74078, USA Abstract.—We used electrofishing data from Oklahoma and Texas reservoirs to evaluate potential temporal (spring, summer, and fall) and spatial (reservoir section and habitat) biases associated with different sampling strategies for estimating size-related metrics (mean total length, proportional size distribution [PSD], and proportion of stock length fish [PSL]) of blue catfish Ictalurus furcatus. Regardless of how many individual fish were sampled, site-specific estimates of mean total length often deviated from the population mean, suggesting that fish within a site were of similar size. Bias across seasons was not consistent for any of the length metrics tested. Only one population had length-related differences between reservoir sections. Blue catfish collected from channel habitats were consistently larger than those collected from point or flat habitats. To identify the number of sites required to reduce deviations in estimates of size-related metrics, we used a Monte Carlo simulation technique to evaluate 3, 5, 10, 20, or 50 randomly selected sites. Regardless of how many individual fish were sampled, when too few sites were sampled, size-related metrics deviated farthest from the population mean. No- tably, the population with a truncated length distribution had the least deviation. Simulations indicated that randomly sampling 10–20 sites resulted in estimates with consistent deviations 50 mm from the population mean. More effort may be required to routinely estimate PSD and PSL within 10 units. In some situations, biologists may consider stratifying the sample by habitat; however, gains in accuracy and precision may not compensate for increased effort needed to precisely quantify the habitat. * Corresponding author: kris.bodine@tpwd.state.tx.us Introduction Low-frequency electrofishing (LFE) efficiently col- lects blue catfish Ictalurus furcatus, with samples often producing more than 100 fish. Recent studies indicate that individual LFE samples accurately rep- resent size structure of blue catfish within a given sampling location (Buckmeier and Schlechte 2009; Bodine and Shoup 2010). While this could suggest that just a few LFE samples could produce enough fish to provide precise length-frequency data, it is not clear how well single, large samples represent the overall population. Information regarding size- specific distributional patterns of this species is needed to develop sampling strategies that accu- rately represent total population size structure. For many fishes, variables such as season, reservoir sec- tion, habitat, water depth, and time of day affect their distribution (Hubbard and Miranda 1986; Post et al. 1995; Schael et al. 1995). Size-related spatial and temporal migrations, both vertically and horizontal- ly, may also affect the portion of the population that is vulnerable to electrofishing at a given place and time. In addition, similarly sized fish often school to- gether (Krause et al. 1996; Hoare et al. 2000; Barber 2003). Consequently, individual samples likely only represent a portion of the entire size structure and oversampling any one location or habitat type could bias metrics typically used to assess populations. By understanding these issues, sampling designs can be developed to reduce most sources of bias by ac- counting for nonrandom distributions of fish. Proportional size distribution (PSD), proportion