Predicting the distribution of native stream fishes is fundamental to the management and conservation of many species. Modeling species distributions often consists of quantifying relationships between species occurrence and abundance data at known locations with environmental data at those locations. However, it is well documented that native stream fish distributions can be altered as a result of asymmetric interactions between dominant exotic and subordinate native species. For example, the naturalized exotic Brown Trout Salmo trutta has been identified as a threat to native Brook Trout Salvelinus fontinalis in the eastern United States. To evaluate large‐scale patterns of co‐occurrence and to quantify the potential effects of Brown Trout presence on Brook Trout occupancy, we used data from 624 stream sites to fit two‐species occupancy models. These models assumed that asymmetric interactions occurred between the two species. In addition, we examined natural and anthropogenic landscape characteristics we hypothesized would be important predictors of occurrence of both species. Estimated occupancy for Brook Trout, from a co‐occurrence model with no landscape covariates, at sites with Brown Trout present was substantially lower than sites where Brown Trout were absent. We also observed opposing patterns for Brook and Brown Trout occurrence in relation to percentage forest, impervious surface, and agriculture within the network catchment. Our results are consistent with other studies and suggest that alterations to the landscape, and specifically the transition from a forested catchment to one that contains impervious surface or agriculture, reduces the occurrence probability of wild Brook Trout. Our results, however, also suggest that the presence of Brown Trout results in lower occurrence probability of Brook Trout over a range of anthropogenic landscape characteristics, compared with streams where Brown Trout were absent.

Many potential stressors to aquatic environments operate over large spatial scales, prompting the need to assess and monitor both site‐specific and regional dynamics of fish populations. We used hierarchical Bayesian models to evaluate the spatial and temporal variability in density and capture probability of age‐1 and older Brook Trout Salvelinus fontinalis from three‐pass removal data collected at 291 sites over a 37‐year time period (1975–2011) in Pennsylvania streams. There was high between‐year variability in density, with annual posterior means ranging from 2.1 to 10.2 fish/100 m2; however, there was no significant long‐term linear trend. Brook Trout density was positively correlated with elevation and negatively correlated with percent developed land use in the network catchment. Probability of capture did not vary substantially across sites or years but was negatively correlated with mean stream width. Because of the low spatiotemporal variation in capture probability and a strong correlation between first‐pass CPUE (catch/min) and three‐pass removal density estimates, the use of an abundance index based on first‐pass CPUE could represent a cost‐effective alternative to conducting multiple‐pass removal sampling for some Brook Trout monitoring and assessment objectives. Single‐pass indices may be particularly relevant for monitoring objectives that do not require precise site‐specific estimates, such as regional monitoring programs that are designed to detect long‐term linear trends in density. Received April 22, 2013; accepted September 18, 2013

In 2004, the Pennsylvania Fish and Boat Commission implemented catch‐and‐release (CR) regulations on headwater stream systems to determine if eliminating angler harvest would result in an increase in the number of adult (≥100 mm) or large (≥175 mm) Brook Trout Salvelinus fontinalis. Under the CR regulations, angling was permitted on a year‐round basis, no Brook Trout could be harvested at any time, and there were no tackle restrictions. A before‐after–control‐impact design was used to evaluate the experimental regulations. Brook Trout populations were monitored in 16 treatment (CR regulations) and 7 control streams (statewide regulations) using backpack electrofishing gear periodically for up to 15 years (from 1990 to 2003 or 2004) before the implementation of the CR regulations and over a 7–8‐year period (from 2004 or 2005 to 2011) after implementation. We used Poisson mixed models to evaluate whether electrofishing catch per effort (CPE; catch/100 m2) of adult (≥100 mm) or large (≥175 mm) Brook Trout increased in treatment streams as a result of implementing CR regulations. Brook Trout CPE varied among sites and among years, and there was no significant effect (increase or decrease) of CR regulations on the CPE of adult or large Brook Trout. Results of our evaluation suggest that CR regulations were not effective at improving the CPE of adult or large Brook Trout in Pennsylvania streams. Low angler use, high voluntary catch and release, and slow growth rates in infertile headwater streams are likely the primary reasons for the lack of response. Received June 14, 2013; accepted August 28, 2013 Published online January 24, 2014

Fisheries biologists often use backpack electrofishing to sample stream fish. A common goal of sampling is to estimate density and/or biomass to make inferences about the status and trends of fish populations. One challenge when estimating population size is determining an appropriate site or reach length to sample. In this study, we empirically determined the required length of stream that needs to be sampled, assuming the study design is one site per stream, in order to achieve a desired level of accuracy for brook trout density and biomass estimates in Pennsylvania headwater streams. Long sample reaches (600 m) were chosen on seven first to third order streams and these sites were broken into twelve 50-m subreaches. Each subreach was sampled by removal electrofishing techniques until either five electrofishing passes were completed or no brook trout were captured. The total density and biomass of brook trout over all 50-m subreaches was considered the “true” density and biomass for the entire reach. We then performed computer simulations in which various numbers of 50-m subreaches were randomly selected and catches from each subreach were summed within the first three electrofishing passes to simulate removal sampling of site lengths ranging from 50 to 550 m. Population estimates were made using a removal estimator and density and biomass were calculated using various stratification schemes based on fish age and size. Estimates of density and biomass were then compared to the true values to assess the possible range in bias of estimates for a given reach length. Results from our simulations suggest a 200- to 250-m-long or a 400- to 450-m-long stream reach or site is needed to estimate brook trout density and biomass within 50% and 25%, respectively, of the true density and biomass. This information and our methodology will be valuable to fisheries managers in developing standardized protocols for assessing trout populations in small streams.