Spatial Pattern Analysis of Citrus Canker-Infected Plantings in São Paulo, Brazil, and Augmentation of Infection Elicited by the Asian Leafminer

Abstract

ABSTRACT Eradication of Asiatic citrus canker (ACC) has become increasingly difficult over the last decade, following the introduction of the Asian leafminer into Brazil and Florida, which has led to changes in the eradication protocols. The present study, undertaken in Brazil, was aimed at characterizing the spatial patterns of ACC in commercial citrus plantings to gain better understanding of the dynamics of the disease subsequent to introduction of the leafminer. The spatial patterns of ACC were mapped in 326 commercial citrus plantings and statistically assessed at various spatial dimensions. The presence of "within-group" aggregation in each plot was examined via beta-binomial analysis for groups of trees parsed into three-by-three-tree quadrats. The relative intensity of aggregation was expressed as a binomial index of dispersion (D) and heterogeneity among plots expressed as the intracluster correlation coefficient, rho. The population of data sets was found to fall into three D categories, D < 1.3, 1.3 </= D = 3.5, and D > 3.5. These categories then were related to other spatial characteristics. The binary form of Taylor's power law was used to assess the overdispersion of disease across plots and was highly significant. When the overall population of plots was parsed into D categories, the Taylor's R (2) improved in all cases. Although these methods assessed aggregation well, they do not give information on the number of foci or aggregations within each plot. Therefore, the number of foci per 1,000 trees was quantified and found to relate directly to the D categories. The lowest D category could be explained by a linear relationship of number of foci versus disease incidence, whereas the higher two categories were most easily explained by a generalized beta function for the same relationship. Spatial autocorrelation then was used to examine the spatial relationships "among groups" composed of three-by-three-tree quadrats and determine common distances between these groups of ACC-infected trees. Aggregation was found in >84% of cases at this spatial level and there was a direct relationship between increasing D category and increasing core cluster size, and aggregation at the among-group spatial hierarchy was generally stronger for the within-row than for the across-row orientation. Clusters of disease were estimated to average between 18 and 33 tree spaces apart, and the presence of multiple foci of infection was commonplace. The effectiveness of the eradication protocol of removing all "exposed" trees within 30 m surrounding each "ACC-infected tree" was examined, and the distance of subsequent infected trees beyond this 30-m zone from the original focal infected tree was measured for each plot. A frequency distribution was compiled over all plots to describe the distance that would have been needed to circumscribe all of these outliers as a theoretical alternative protocol to the 30-m eradication protocol. The frequency distribution was well described by a monomolecular model (R(2) = 0.98) and used to determine that 90, 95, and 99% of all newly infected trees occurred within 296, 396, and 623 m of prior-infected trees in commercial citrus plantings, respectively. These distances are very similar to previously reported distances determined for ACC in residential settings in Florida.