Responses of bluegill to habitat manipulations: power to detect effects
Power, the probability of detecting a particular effect of an experimental manipulation, has rarely been considered in whole-lake fishery experiments. We assessed the statistical power of experiments to detect effects of habitat manipulation on growth and population size of bluegills Lepomis macrochirus. In general, power was greater for detecting changes in growth increments than for population estimates. Power was greatest for first-year growth but decreased for growth increments in subsequent years. However, when length-specific growth was calculated, power remained high across all lengths. For 1.5× changes, considered biologically significant in Wisconsin lakes, more than 90\% of experiments with 3 manipulated and 3 control lakes detected changes in growth of 1- or 2-year-old bluegills and intercepts of growth curves. About 5 replicate lakes per treatment were needed to detect 1.5× changes in size-specific growth or growth increments in ages 2–3 with power greater than 90\%. About 10 replicate lakes per treatment would be needed to assess 1.5× changes in growth increments in ages 3–4 or population density with power greater than 90\%. Sampling an additional lake for bluegill population estimates, however, costs more than 50× more person-hours than sampling an additional lake for growth data. In the past, whole-lake studies of fish responses to habitat manipulations have generally not been conclusive. Our results quantify the replication needed to provide clear guidance to managers.