Herbivore movement and spatial population dynamics in a heterogeneous landscape

Abstract

To date, few metapopulation studies have isolated the effects of the intervening matrix from other attributes of landscape structure (e.g., patch geography or quality) that might influence dispersal among patches. Furthermore, there has been little exploration of the mechanisms underlying effects of the matrix on the spatial ecology of species. In this dissertation, I examine how the movement and distribution of the planthopper Prokelisia crocea within and among host-plant patches (prairie cordgrass, Spartina pectinata) is affected by the composition of the matrix (the invasive grass smooth brome [Bromus inermis], native non-host grasses, or mudflat). First, using a mark-recapture study in networks of experimental cordgrass patches that were made identical in size, isolation, and plant quality, I found that the interpatch movement rate of the planthopper was highest in the brome matrix, intermediate in the native grass matrix, and lowest in mudflat. Second, field surveys revealed that individuals accumulated against patch edges in mudflat-bordered patches, but not in patches bordered by non-host grasses. Among patches, incidence and density increased with the proportion of the matrix composed of open mud. Third, I investigated the behavioral bases of these matrix effects using individual movement trials. Whereas movement through mudflat was highly linear, movement was much more tortuous through brome. Within patches, brome edges were three times more permeable to emigration than mudflat edges. I suggest that the effect of matrix composition on the rate of planthopper movement among patches is driven largely by differences in movement tortuosity within the matrix. Tortuous movement through brome likely increases the planthopper’s rate of encounter with spatially aggregated host-plant patches. Furthermore, the effect of the matrix on the planthopper’s within-patch distribution can be attributed to the differences in edge permeability between matrix types. Finally, a literature review revealed that matrix composition and patch quality often covary in plant-herbivore systems, and that most matrix studies have failed to experimentally or statistically isolate the effects of the matrix from potential patch quality effects on dispersal. These findings highlight the value of a mechanistic understanding of the links between landscape structure and dispersal in spatial ecological population studies.