Estimation of Contemporary Dispersal in River Networks Using Genetic Time Series Data

Quantification of migration and recolonization following disturbance is a significant problem for large river fishery management.  We explored the utility of genetic time series data as a means to estimate contemporary dispersal in riverine fishes.  A time series is defined as set of three or more temporally adjacent genetic samples taken from two or more distinct localities arrayed in a river network.  Our approach treated observed alleles as ‘individuals’ in an occupancy modeling context, where transition probabilities of occurrence, extirpation, recolonization and migration within and between localities across time steps were of interest. We used forward-time simulation modeling implemented in the program simuPOP to evaluate statistical power to detect movement of alleles (and fish) from one distinct locality across time steps.  Models were parameterized using a genetic data set collected for longfin dace in the upper Gila River Basin in southwestern New Mexico. With a microsatellite data at 10 loci over three generations, we could identify events that were consistent with movement of fishes between localities from one generation to the next.  Detection probabilities were low due to small sample sizes and a preponderance of low-frequency alleles.  Large scale parent-based genetic tagging could overcome these constraints.