Fish dispersal in fragmented landscapes: a modeling framework for quantifying the permeability of structural barriers.

Dispersal is a key determinant of the spatial distribution and abundance of populations, but human-made fragmentation can create barriers that hinder dispersal and reduce population viability. This study presents a modeling framework based on dispersal kernels (modified Laplace distributions) that describe stream fish dispersal in the presence of obstacles to passage. We used mark-recapture trials to quantify summer dispersal of brook trout (Salvelinus fontinalis) in four streams crossed by a highway. The analysis identified population heterogeneity in dispersal behavior, as revealed by the presence of a dominant sedentary component (48-72% of all individuals) characterized by short mean dispersal distance (<10 m), and a secondary mobile component characterized by longer mean dispersal distance (56-1086 m). We did not detect evidence of barrier effects on dispersal through highway crossings. Simulation of various plausible scenarios indicated that detectability of barrier effects was strongly dependent on features of sampling design, such as spatial configuration of the sampling area, barrier extent, and sample size. The proposed modeling framework extends conventional dispersal kernels by incorporating structural barriers. A major strength of the approach is that ecological process (dispersal model) and sampling design (observation model) are incorporated simultaneously into the analysis. This feature can facilitate the use of prior knowledge to improve sampling efficiency of mark-recapture trials in movement studies. Model-based estimation of barrier permeability and its associated uncertainty provides a rigorous approach for quantifying the effect of barriers on stream fish dispersal and assessing population dynamics of stream fish in fragmented landscapes.