Spatial heterogeneity of the incidence of grape downy mildew

Heterogeneity of the incidence of downy mildew of grape, caused by Plasmopara viticola, was quantified in an experimental Ohio vineyard. The proportion of diseased leaves on each of 15 shoots (sampling units) per plot was determined for 18 separate plots at two times (generally August and September) during each of 3 yr. The binary data analogue of Taylor's power law, in which the logarithm of the observed variance is regressed on the logarithm of the theoretical variance for a binomial (random) distribution, provided strong and consistent evidence that diseased leaves were aggregated. Year and assessment time did not affect power law parameters. The estimate of the regression slope (b), an overall measure of heterogeneity, was 1.30 (SE = 0.04). Heterogeneity in individual plots was measured with variance ratio and C(alpha) tests and with the aggregation parameter of the beta-binomial distribution fitted to the data. Except for mean incidence less than 0.05, the majority of the plots had significant heterogeneity, and the data were better described by the beta-binomial than by the binomial distribution. Estimates of aggregation parameter were variable but were highest in the middle range of disease incidence. Using the modified power law, sampling curves were generated to precisely estimate disease incidence.