VARIABLE-INTENSITY SAMPLING FOR PROPORTION OF PLANTS INFESTED WITH PESTS

Construction of a variable-intensity sampling procedure for binomial data is described. Performance of an algorithm to calculate sampling intensity, which can be programmed into a portable computer for use in the field, was compared with that of a chart for determining sampling intensity, and was found to differ little. In computer simulations, performance of the procedure on the binomial distribution, using the algorithm described, was similar to that for previously described performance on the negative binomial distribution. Simulated between-site variation in the proportion of plants infested had little effect on the sample size or estimates of the average proportion infested. Simulated edge effects, higher proportion of infested plants at the edge of the field, increased overestimation; simulated patterns of infestation resulting from a clogged sprayer nozzle increased underestimation, but this bias would not affect treatment decisions. Testing the procedure on field data for five pests of cabbage resulted in performance similar to the simulations. Variable-intensity sampling for proportion of plants infested with pests could be the most useful procedure to use when samples must be taken along the entire length of a transect, yet total sample size must be chosen efficiently.