Landscape composition is more important than local management for crop virus-insect vector interactions

Managing crop viruses is difficult due to complex interactions among vectors, reservoirs, and mediating factors such as land cover. Identifying the appropriate ecological neighborhood, or the spatial area in which the most influential interactions occur affecting virus epidemiology, would therefore be beneficial in exposing which of the many explanatory variables to target in the plant-pathogen system. We constructed partial least squares path models to find the neighborhood size for vectors of stylet-borne nonpersistent viruses infecting pumpkins, and compare the relative influence of within-field and extra-field land cover. Two economically important aphid-vectored viruses in the U.S. Midwest are included in these analyses: papaya ringspot virus (PRSV) and watermelon mosaic virus (WMV). In 2010 and 2011, we used commercial pumpkin fields to quantify virus infection, vector (aphid) alightment (i.e., landing rates), and within-field weed communities, and subsequently analyzed extra-field cover from 1 to 5 km concentric buffer zones within the surrounding landscape. Alightment rates of total noncolonizing vectors and the top three numerically dominant species (Aphis craccivora Koch, Therioaphis trifolii (Monell), and Rhopalosiphum padi [L.]) were included in individual path models. Overall, we found that extra-field landscape composition had a far stronger influence on vector alightment than within-field weed cover; this pattern was consistent for seven of eight statistical models. In one exception, weed cover influenced alightment of the putative PRSV vector, A. craccivora, while surrounding landscape had no effect. In this case, weed coverage and vector alightment were inversely related, demonstrating that aphids were less likely to land in weed-infested fields. However, weed cover did not predict alightment of T. trifolii or total noncolonizers. The neighborhood size scales for total noncolonizers' tended to be larger than for individual species (4-5 km), suggesting future studies of dispersal by multi-species aphid groups may benefit from an extended gradient. On balance, our results indicate that while surrounding land cover interactions are complex, they exert greater influence over vector dispersal than within-field weed cover, calling into question whether management of local weeds is an effective method of crop virus prevention in some systems.