Regional Mapping of Perennial Weeds in Cotton with the Use of Geostatistics

Abstract

Perennial weeds constitute a serious problem in Greek cotton-growing areas, as they strongly competing against the crop and downgrade the final product. Monitoring weeds at a regional scale and relating their occurrence with abiotic factors will assist in the control of these species. Purple nutsedge, field bindweed, bermudagrass, and johnsongrass were studied in cotton crops for three consecutive growing seasons (2007 through 2009) in a large area of central Greece. Weed densities and uniformities per sampling site were assessed in relation to soil and climatic data. Abundance index (AI), which is highly dependent on abiotic factors, was also estimated, and revealed purple nutsedge to the most persistent and damaging species among the recorded weeds. Field bindweed showed the highest correlation with soil properties and especially with clay content. Furthermore, correlation analysis was used over the sampling years in order to assess the stability of weed occurrence in the sampling sites. Purple nutsedge, field bindweed, and bermudagrass proved to be stable in location and intensity. The weed density spatial distribution was evaluated by using local indicators of spatial autocorrelation (LISA) statistics, and was mapped by ordinary kriging and co-kriging interpolation methods. Only 1 to 3 spatial outliers were identified in each 1 of the 3 yr. Between the two interpolation methods co-kriging delivered better results for field bindweed and purple nutsedge, indicating that soil data could improve the estimation of weed occurrence. These co-kriging interpolated weed maps would be a very useful tool for decision makers in taking appropriate weed control measures.