Three prediction models for the first generation of Dendrolimus punctatus (Lepidoptera: Lasiocampidae) larvae. (2nd October 2021)

• Record Type: Journal Article
• Authors: Song, Xue-Yu; Zhang, Lin; Zhang, Nan; Qian, Guangjing; Zhou, Xia-Zhi; Zhang, Guo-Qing; Zou, Yun-Ding; Fang, Guo-Fei; Zhang, Zhen; Bi, Shou-Dong
• Is Part Of: Oriental insects. Volume 55:Number 4(2021)
• Page Start: 513
• Page End: 528
• Publication Date: 2021-10-02
• Subjects: Dendrolimus punctatus larvae -- initial peak period -- Bayes discriminant method -- stationary time series method -- contingency table method; Insects -- Asia -- Periodicals; Entomology -- Periodicals; Insects -- Classification -- Periodicals; 595.7095
• DOI: 10.1080/00305316.2020.1841688
• Languages: English
• ISSNs: 0030-5316

Abstract

ABSTRACT: To improve the accuracy of forecasting the initial peak period of Dendrolimus punctatus (Walker) (Lepidoptera: Lasiocampidae) larvae, Bayes discriminant analysis, Stationary time series method and Weighted contingency table method were used to develop a prediction model for the initial peak period of the first generation of its larva based on occurrence data which spanned from 1983 to 2016 in Qianshan City, Anhui Province. Historical data were digitised, graded, programmed and used to build the forecast models. The predicted values were verified with data in 2017 and 2018. The historical coincidence rate from Bayes discriminant was 97.18%. In 2017, the forecasted value was level 1 and the actual level was level 2. In 2018, the forecasted and actual results were all level 1. The historical coincidence rate from the stationary time series was 82.76%, which indicated that the prediction result in 2017 was 1 level lower than the actual values, but was consistent in 2018. The historical coincidence rate from the weighted contingency table method was 93.94% and the forecasted results for 2017 and 2018 were consistent with the actual values. Among the three methods, Bayes discriminant method and weighted contingency table method gave better prediction results.