Scaling and the effects of plant, soil, and landscape characteristics on sap-feeding herbivores in cotton

Soil physical and chemical properties can affect plant growth and nutrition, which in turn can affect a plant's attractiveness and susceptibility to insect herbivores. A further source of variation in these relationships is the spatial scale at which patterns are measured. Both the size of the area being sampled, or scale, and the distance between measurements, or grain, are parameters that affect interpretation of insect abundance patterns. Our objectives in this study were to determine both the relationship of various landscape, plant, and soil characteristics to densities of sap-feeding insect herbivores in cotton and to determine the effects of sampling scale and sampling grain on these relationships. We included three sap-feeding herbivores in our study: Aphis gossyphii Glover, Frankliniella occidentalis (Pergande), and Bemisia tabaci Gennadius. We found that abundance of each insect species was related to several single factors within the cotton field and that these relationships were always dependent on the scale and grain of the measurements. No one variable or set of variables was related to a particular insect density for each scale and grain examined. However, some variables were significantly correlated with insect densities at the larger scale (622 by 31 m), although none were significantly correlated for all plots Lit the smaller scale (154 by 31 m). In comparing the separate effects of each variable using partial correlations, elevation was negatively correlated with A. gossypii density at both grains (samples taken 25 m apart and samples taken 50 m apart), whereas in multiple regression analyses including all variables, plant moisture and soil nitrates were positively correlated and plant height and clay negatively correlated with A. gossypii density. In examining the separate effect of each variable on E occidentalis density, plant moisture was negatively associated with E occidentalis density at each grain. In multiple regression analyses, no variable was associated with E occidentalis density at each grain. For B. tabaci, soil salinity was positively associated when variables were examined separately or in multiple regression. We discuss the possible reasons for why particular variables are related to the densities of each species.