Sampling for detection: Leek rust as an example

Abstract

Abstract Computer simulation was used to compare four sampling patterns in their efficiency of detecting diseased plants under a variety of spatial distributions of infected plants. The sampling patterns compared were: diagonal sampling, double diagonal sampling, row sampling over two rows and row sampling over four rows. Row sampling over four rows was the most effective sampling pattern. The detection level, the lowest disease level at which the probability to detect the disease is sufficiently large, was studied in relation to temporal and spatial characteristics of the disease, as found in two epidemics in 1992 and 1993. For both epidemics, a weekly sample of 50 plants resulted in detection levels that were low enough to ensure adequate chemical control, indicating the feasibility of sampling for detection of leek rust. Weekly sampling is more efficient than fortnightly sampling.