SPATIAL AUTOCORRELATION, SPATIAL MODELING, AND IMPROVEMENTS IN GRASSHOPPER SURVEY METHODOLOGY

Abstract

AbstractAnalysis of 10 years of grasshopper survey data (1978–1987) indicated that grasshopper populations in fields can be reliably predicted from roadside survey counts. Direct estimation of grasshopper densities in crop fields is no longer required for summaries of infestation levels or forecasts. Spatial autocorrelation was significant and positive for both roadside and field counts. The coefficient of variation of the field counts was greater than that of the roadside counts in each of the last 10 years. Population density was summarized by crop type and sampling method for the last 10 years. Linear regressions fitted to the 1978–1984 grasshopper survey data were used to estimate field population density from crop type and roadside counts in 1985–1987. Maps of population density were generated from the predicted and observed field counts with SPANS, a microcomputer-based geographic information system. Large coefficients of association (73–79%) between the predicted and observed maps attested to the sufficiency of road-side counts as the basis for production of population density maps.