SPATIAL AND TEMPORAL ANALYSES OF CITRUS TRISTEZA VIRUS IN EASTERN SPAIN

Citrus tristeza virus (CTV) was monitored for up to 14 years by monocl onal antibody probes via enzyme-linked immunosorbent assay in five ora nge and grapefruit orchards with symptomless trees in Valencia and Ali cante provinces, Spain. Linear, exponential, logistic, or Gompertz non linear temporal models were selected as the most appropriate, dependin g on the phase of the epidemic during the assessment period and based on correlation of observed versus predicted values and examination of the patterns of residual error. Ordinary runs analysis for within- or across-row association of CTV-positive trees indicated that the diseas e status of immediately adjacent trees was unpredictable. The beta-bin omial index of dispersion for different quadrat sizes indicated a tend ency for aggregations of infected trees at quadrat sizes of 2 x 2 or 4 x 4 CTV-positive trees in some plots, which when viewed with ordinary runs suggested the possible presence of longer distance spatial relat ionships. Two-dimensional distance class (2DCLASS) analysis indicated a random spatial pattern of CTV incidence and general lack of associat ion of infection among adjacent trees. Spatio-temporal distance class (STCLASS) analysis further indicated a lack of spatial dependency amon g adjacent CTV-positive trees over time. Significant edge effects dete cted by both 2DCLASS and STCLASS analyses suggested possible spread of CTV from inoculum originating outside the area of the individual plot s. Semi-variograms from spatio-temporal geostatistical analyses of fou r directions of orientation confirmed a lack of spatial dependency of infection among adjacent or nearby trees over time. These combined spa tial and temporal analyses gave some insight into possible underlying processes of CTV spread and suggested CTV spread must be predominantly to trees farther away rather than to immediately adjacent trees. If a nonrandom spatial structure of CTV incidence does exist, it may well be of a complexity beyond the detection ability of the spatial analysi s methods employed or perhaps on a scale that is larger than the dimen sions of the plots studied.