Effects of vegetation structure and landscape complexity on insect parasitism across an agricultural frontier in Argentina

Abstract

Insect parasitism patterns are influenced by vegetation structure and landscape complexity. Our objective was to examine the effects of vegetation structure and landscape complexity on parasitism based on direct measurements of structure and diversity indices as well as on metrics based on remote sensing using Quickbird images. We collected 2266 lepidopteran larvae and pupae, including different families and habits, to estimate parasitism, and recorded vegetation characteristics in five 100-m2 transects and 18 1 ha-plots in the dry Chaco, Northwest Argentina. We calculated landscape metrics and semivariograms in the plots from the image. The plots represented four “complexity groups”: agricultural, riparian/hedgerow, bare ground, and forest plots. Mean parasitism in the study sites was 10.7% (min: 0%, max: 23%). Parasitism was highest in agricultural plots, lowest in forest plots, and intermediate in riparian/hedgerow and bare ground plots. The landscape model explained parasitism more than the vegetation model. The landscape final model included Normalized Difference Vegetation Index (NDVI) Range, a measure of landscape heterogeneity, and Mean Shape Index, a measure of patch shape irregularity, and their interaction. The vegetation model included basal area and the Coefficient of Variation of tree density among transects, a measure of tree spatial distribution within a plot. Our results agree with previous studies that found higher parasitism in agricultural vs. non-agricultural environments in the subtropics, while riparian/hedgerow plots were important for conserving parasitism, as reported for temperate environments. We showed that under-explored tools such as the semivariogram and satellite band combinations were useful for the assessment of parasitism and that studying vegetation and landscape complexity simultaneously can help us examine mechanisms in detail. The identified variables related to high parasitism should be used for image classifications with a functional approach.