Simulation and empirical analysis of novel sibship-based genetic determination of fish passage

We develop and test a new analytic approach, termed sib-split, to detect fish passage through road crossings. This new approach is based on the genetic analysis of full-siblings on opposite sides of potential barriers. We used simulations and data from two empirical case studies involving brook trout (Salvelinus fontinalis) movement with respect to barriers that varied in strength of effect on fish passage. Simulations revealed that both sib-split and the population assignment-based method (STRUCTURE) were highly accurate (mean accuracy > 99%) under easy-to-detect conditions (moderate to strong genetic differentiation and no movement). However, under difficult-to-detect simulated conditions (no genetic differentiation, 10% movement each generation), sib-split had higher accuracy (mean accuracy = 98%) than STRUCTURE (mean accuracy = 84%). Sib-split also outperformed STRUCTURE (mean accuracy 98% versus 89%) under a more difficult-to-detect simulated management scenario (simulated construction of a new complete barrier to movement). Sib-split provided more reliable and easily interpretable movement detection in both easy-and difficult-to-detect empirical case studies. With the empirical case studies, sensitivity to the prior on migration rate precluded use of STRUCTURE by itself, but a two-step approach where sib-split results were used to provide an informed migration prior for STRUCTURE provided additional information for both case studies.