Fringe effects: detecting bull trout (Salvelinus confluentus) at distributional boundaries in a montane watershed

Robust assessment and monitoring programs are critical for effective conservation, yet for many taxa we fail to understand how trade-offs in sampling design affect power to detect population trends and describe spatial patterns. We tested an occupancy-based sampling approach to evaluate design considerations for detecting watershed-scale population trends associated with juvenile bull trout (Salvelinus confluentus) distributions. Electrofishing surveys were conducted across 275 stream sites from the Prairie Creek watershed, Northwest Territories, Canada. Site-level detectability of juvenile bull trout was not uniform, and imperfect detection affected modelled occupancy probabilities most in fringe habitats near distributional boundaries in steep reaches and large streams. We show that detecting a 30% change in watershed-level occupancy >= 78% of the time, as conservation guidelines suggest, may require three repeat surveys (i.e., temporal replicates) and increased spatial sampling intensity of fringe habitats. Additional sampling effort in fringe sites could be offset by sampling fewer sites in core habitats to optimize designs for detecting demographic shifts in bull trout, while still minimizing risk of nondetection for this cryptic species.