2006. Current velocity and habitat patchiness shape stream herbivore movement. Á Oikos 115: 358 Á368.

Understanding not just where organisms move but how they move is an important step towards integrating animal behaviour into landscape ecology. The three-dimensional landscape of a streambed provides an ideal setting for forging this integration because of the persuasive effects of flowing water. In this study, we experimentally examine the larval movement of the case-building caddisfly Agapetus boulderensis Milne, 1936 in response to two current velocities in each of five levels of contrasting habitat types (i.e., smooth patches that facilitate movement and thick algal patches that constrain movement). Detailed behavioural observations showed that larvae employed two distinctly different strategies of movement in different current velocities: faster crawling and slower pivoting. Our results suggest that individual decision-making between crawling and pivoting is related to the magnitude of current velocity across the streambed, and the frequency at which larvae employ these behaviours translates into differential movement rates and directions. Strong concordance between a conceptual model and our results supports the notion that the presence of structural "nonhabitat" patches at high current velocities may create areas of local flow interruption and refugia. This, in turn, plays an important role in eliciting either crawling or pivoting and in shaping patterns and directions of larval movement, and by extension resource acquisition.

Ecological patterns and processes often change with scale, and determining how they do so is a necessary first step in understanding these relationships. We documented how the spatial patterns of beetle species richness, environmental variables, and correlations between the two change with spatial scale. Tenebrionid beetles and vegetation coverage were sampled on the shortgrass steppe of northeastern Colorado during June and August 1998. We used triplet quadrat variance (tQV) and triplet quadrat covariance (tQC) to evaluate the spatial patterns of beetle distribution and environmental variables. As revealed by tQV, spatial patterns of beetle species richness and environmental measures were scale dependent but differed among transects. Bare ground exhibited significant patchiness at spatial scales in the 460–500‐m range, grass at 460–540 m, and shrubs at 420–620 m. Tenebrionid species richness in June showed significant patchiness in the 400–420‐m and 560–640‐m range, while August patch dimensions were in the 200–360‐m and 400–520‐m ranges. As revealed by tQC, correlations between tenebrionid species richness and environmental measures were not scale dependent, nor was there much variation between transects. Univariate, multiscaled analyses revealed potentially significant scales; however, covariate, multiscaled analyses revealed no scale dependence of interaction. These analyses showed (1) that while spatial scale is important, it is often unpredictable and (2) that multiscaled analysis in realistic settings may not capture species' relevant heterogeneity.