Setting up of efficient sampling schemes to detect Xylella fastidiosa in Apulian olive groves

In the absence of an accurate sampling scheme for monitoring Xylella fastidiosa in Apulian olive groves, a "parcel scale" (single olive grove) sampling approach was developed in this study based on official 2016-2019 monitoring data. A total of 34 olive groves of different shapes were extracted within 13 selected outbreak areas with true incidence ranging from 1.96% to 38.29%. A SADIE analysis revealed for most outbreaks an aggregate spatial distribution of infection. Three types of sampling schemes were chosen and applied to the selected olive groves: Random (RS), Systematic ("Simple" SS and "in Groups" GS) and Perimetric (PS). For SS and GS, different inter-plant distances (10, 25, 35, 50, 70, 90, 120 and 150 meters) were evaluated and compared with the corresponding RS scheme. The process of local spread of infection within the olive groves was established using the "Random walks" mechanism: generating 100 scenarios of infection patterns for each olive grove, ranging between 0 and 50 % of the total number of trees. In simulations conducted in MATLAB between the different sampling schemes, total errors (RMSE) and introduced biases (MBE) were evaluated and compared. The SS, GS, and RS schemes introduced increasing RMSE values from 1.72 to 81.295 as a function of the percentage of trees sampled (from 83% to 0.65% of total trees at inter-distances between 10 and 150 meters). RS schemes performed best at 10-meter inter-distances, while SS schemes for inter-distances between 35 and 70 meters. The SS scheme at 35 meters was the best for 68.75% of the analysed olive groves, introducing a low MBE. The PS scheme (corresponding to two rows of trees along the perimeter of the grove) detected the presence of at least one positive tree in all olive groves studied.IT; PDF; fsantoro@iamb.i