Habitat requirements and landscape features can exert strong influences on the population structure of organisms. For aquatic organisms in particular, hydrologic requirements can dictate the extent of available habitat, and thus the degree of genetic connectivity among populations. We used a landscape genetics approach to evaluate hypotheses regarding the influence of landscape features on connectivity among populations of the giant water bug Abedus herberti (Hemiptera: Belostomatidae). Abedus herberti is restricted to naturally-fragmented, perennial stream habitats in arid regions of North America. This species is exceptional because it is flightless at all life stages. Thus, we hypothesized a high degree of population genetic structure in A. herberti due to hydrologic constraints on habitat and low dispersal ability of the organism. A total of 617 individuals were sampled from 20 populations across southeastern Arizona, USA and genotyped at 10 microsatellite loci. We used a Bayesian clustering method to delineate genetic groups among populations. To determine which of six landscape variables (representing hypotheses of landscape-level connectivity) has the strongest association with genetic connectivity in A. herberti, we used information-theoretic model selection. Strong population structure was evident among A. herberti populations, even at small spatial scales. At a larger scale, A. herberti populations were hierarchically structured across the study region, with groups of related populations generally occurring in the same mountain range, rather than in the same major watershed. Surprisingly, stream network connectivity was not important for explaining among-population patterns. Only the Curvature landscape variable was identified as having an association with genetic connectivity in A. herberti. The Curvature variable hypothesizes that gene flow tends to occur where local topography is concave, such as within stream drainages and dry gullies. Thus, our results suggest that population connectivity may depend on the shape of local overland topography rather than direct connectivity within stream drainage networks.