Flow- and substratum-mediated movement by a stream insect

1. In streams subject to frequent hydrologic disturbance, the ability of benthic invertebrates to disperse within the channel is key to understanding the mechanisms of flow refugium use and population persistence. This study focuses on crawling invertebrates, the effects on movement of abiotic factors (specifically, flow near the stream bed and bed micro‐topography) and the consequences for dispersal. 2. In a large flume, we observed individual cased caddisflies, Potamophylax latipennis, moving in fully turbulent flows over a precise replica of a water‐worked surface. From maps of movement paths, we quantified crawling behaviour and entrainment, and the influence of bed micro‐topography. We manipulated discharge and tested its effect on movement, linear displacement and areal dispersal. The highest discharge treatment was a disturbance to the caddis; the lowest discharge was not. Crawling behaviours were used to parameterise random walk models and estimate population dispersal, and to test the effects of abiotic factors on movement. 3. Bed micro‐topography influenced crawling in several ways. Caddis spent most of their time at the junctions between proud particles and the adjacent plane bed. The frequency distribution of turn angles was bimodal, with modal values approximating the angle required to travel around median‐sized particles. Larvae generally crawled downstream, but crawling direction relative to the flow was skewed by bed micro‐topography and was not directly downstream, unlike drift. 4. Caddis crawled for most of the time and discharge affected almost every aspect of their movement. As discharge increased, caddis crawled less often, more slowly and over shorter distances; they also became entrained more frequently and over greater distances. With increased discharge, caddis spent proportionately less time at the junctions between proud particles and the adjacent plane bed, and more time on the tops and sides of proud clasts. This is curious as most entrainment occurred from the tops and sides of clasts and entrainment is generally considered to be disadvantageous during disturbances. 5. Linear displacement (drift and entrainment combined) was downstream, but the relation between total displacement and discharge was complex. Total displacement decreased at intermediate discharge as crawling decreased, but increased at high discharge as entrainment and drift played a greater role in movement. 6. Within‐stream dispersal via crawling contained elements of both a correlated random walk (we observed directional persistence in turn angles) and a biased random walk (we observed downstream bias in move direction angles) and was best described as a biased correlated random walk. Dispersal was inversely related to discharge, suggesting that the ability of P. latipennis to crawl into flow refugia on the streambed is reduced at high flow.