AUTHOR=Kim Steven , Heo Seong 

TITLE=Spatio-temporal time series forecasting with trap catch data of oriental fruit moth (Grapholita molesta) in peach (Prunus persica) orchards in South Korea

JOURNAL=Frontiers in Plant Science

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2025.1698144

DOI=10.3389/fpls.2025.1698144

ISSN=1664-462X

ABSTRACT=The oriental fruit moth (Grapholita molesta, Lepidoptera: Tortricidae, OFM) is a major pest causing significant economic damage to peach (Prunus persica, Rosales: Rosaceae) and stone fruits in South Korea. This study aimed to describe spatio-temperal patterns of the OFM population in South Korea, forecast the OFM population using time series models, evaluate their predictive performance, and provide data-driven guidance for region-specific targeted pest management strategies. This study presents the first spatio-temporal time series analysis for predicting OFM population dynamics in peach orchards using sex pheromone trap data (Z8-dodecenyl acetate, E8-dodecenyl acetate, and Z8-dodecenol in a ratio of 88.5:5.7:1.0) collected bimonthly between May and September for ten years (2016-2025). We compared the predictive performance of Seasonal Autoregressive Integrated Moving Average (SARIMA) and Prophet models across three major peach-producing provinces in South Korea: Gyeonggi (GG), Gyeongsangbuk (GB), and Chungcheongbuk (CB). The SARIMA and Prophet models were considered because the OFM generations follow temporal trend and seasonal patterns, and these time series models are flexible to describe and predict them. The Prophet model consistently outperformed the SARIMA model in all three provinces according to multiple evaluation metrics. The time series decomposition revealed a shift from traditional multi-peak (W-shaped) patterns to single-peak patterns of the OFM occurrence, where the mass emergence usually occurs in early May. This phenological shift appears to be driven by climate changes (warmer winters and rising temperatures in the spring) coupled with varying pesticide application strategies. Spatio-temporal analysis demonstrated regional-specific variations. The province of GB maintained low OFM populations through aggressive chemical control following a major outbreak in 2016, and the province of GG showed the highest predicted occurrence in 2026. These findings highlight the importance of region-specific pest management strategies, particularly for controlling the first-generation OFM population. The predictive time series models are valuable tools for establishing smart integrated pest management systems, enabling proactive control measures tailored to regional characteristics.