A Meta-Analysis of North American Stream Fish Movement Studies

Riverscape theory envisions the population dynamics of stream fishes playing out at the watershed scale.  In contrast, mark-recapture studies of individual fish typically record more limited, reach-scale movement.  This contradiction might be reconciled if mark-recapture data were analyzed using techniques that focus on the "tails", rather than the centers, of movement distributions.  We conducted a meta-analysis of 39 published mark-recapture movement studies of North American stream fishes, which spanned a range of taxa but focused particularly on non-salmonids.  Bayesian hierarchical modeling was used to fit a generalized dispersal model to empirical movement-distance frequency distributions.  From best-fitting models, we estimated the distance threshold corresponding to the upper 10th percentile of movement probability and used this as an estimate of dispersal extent.  Estimated dispersal extent varied widely among populations, with much of the variation attributable to the spatial and temporal extent of a study.  After accounting for methodological factors, dispersal extent was not strongly related either to taxonomic affiliation or to life-history traits.  Although some fish populations seemed to disperse primarily at the reach scale, many others exhibited stream- to watershed-scale dispersal extents.  Such long-distance movements form the key, understudied link between individual movements and the population dynamics of stream fishes.