Dynamics and regulation of the southern brook trout (Salvelinusfontinalis) population in an Appalachian stream

1. We used information theoretic statistics [Akaike’s Information Criterion (AIC)] and regression analysis in a multiple hypothesis testing approach to assess the processes capable of explaining long-term demographic variation in a lightly exploited brook trout population in Ball Creek, NC. We sampled a 100-m-long second-order site during both spring and autumn 1991–2004, using three-pass electrofishing. 2. Principle component analysis indicated that the site had lower average velocity, greater amounts of depositional substrata and lower amount of erosional substrata during the 1999–2002 drought than in non-drought years. In addition, drought years had lower flows, and lower variation in flows, than non-drought years. 3. Both young-of-the-year (YOY) and adult densities varied by an order of magnitude during the study. AIC analysis conducted on regressions of per capita rate of increase versus various population and habitat parameters for the population, adults and YOY, for both spring and autumn data sets, indicated that simple density dependence almost always was the only interpretable model with Akaike weights (wi) ranging from 0.262 to 0.836. 4. Growth analyses yielded more variable results, with simple density dependence being the only interpretable model for both adult spring data (wi = 0.999) and YOY autumn data (wi = 0.905), and positive density dependence (wi = 0.636) and simple density independence (wi = 0.241) representing interpretable models for spring YOY data. 5. We detected a significant stock–recruitment relationship between both spring and autumn densities of adults in year t and autumn YOY density in year t + 1. Finally, spring YOY density was positively correlated with both autumn YOY density and spring mean YOY standard length (SL), suggesting that processes affecting recruitment show residual effects at least in the first year of life. This population appears to be regulated primarily by densitydependent processes, although high flows also negatively affected mean SLs of YOY.