DETECTING CHANGES IN THE NEARSHORE FISH COMMUNITY

The response of a fish community to a change in habitat can be measure d only if an assessment of the community is done before and after the change. Decisions must be made about how many years of data will be co llected and how many samples per year. We show how these sampling requ irements are determined for index fishing surveys that use a spatially random sampling design. The methods are illustrated using trapnetting data of common species (e.g., rock bass (Ambloplites rupestris), pump kinseed (Lepomis gibbosus), smallmouth bass (Micropterus dolomieu), an d white sucker (Catostomus commersoni)) on five lakes (208-72500 ha). The results indicate that, for a sampling intensity of 45 sets per yr, 7 yr of sampling before and after the change would be needed to detec t a twofold change in abundance of any species (one-tailed t-test, alp ha=0.05 and beta=0.10). This sampling requirement cannot be greatly re duced by increasing the number of sets per year because the among-year variance component is large. Evaluating the effects of habitat change s within a shorter period would require improvements in sampling metho d that reduce the annual variation in catchability. Alternatively, if similar effects occur in several lakes, the required number of years c ould be reduced by analyzing lakes as a group rather than individually .