Pre-planting Estimation of Damage Risk and Selection of Control Methods for the Soybean Pod Borer, Leguminivora glycinivorella（Lepidoptera: Tortricidae）

Abstract

To estimate the damage risk and select control methods for the soybean pod borer, Leguminivora glycinivorella(Matsumura)(Lepidoptera: Tortricidae), over 200 soybean fields cultivated by farmers and treated with insecticides 0–2 times for the borer were surveyed in Niigata and Nagano Prefectures from 2015 to 2018. A decision tree analysis was performed on the non-treated data. The tree-model had four terminal nodes and was split by three factors: soybean continuous cropping year, damage rate in the previous year, and distance to the nearest field where soybeans were cropped in the previous year. The four terminal nodes were ordered from low to high according to the damage level of the fields classified to each node. The order of nodes did not differ between the modeling data and test data, which were data from other uncontrolled fields. As the number of insecticide treatments increased, the ratio of low-damage-level fields increased to > 90% in all terminal nodes except the highest-damage-level node. These results indicate that the order of the four terminal nodes may represent the order of damage risk. The tree-model may enable rational selection of the number of insecticide treatments or rotation of soybean to rice before planting.