Integration of satellite remote sensing and MaxEnt modeling for improved detection and management of forest pests

Forest pests pose a major threat to ecosystem services worldwide, requiring effective monitoring and management strategies. Recently, satellite remote sensing has emerged as a valuable tool to detect defoliation caused by these pests. Lymantria dispar, a major forest pest native to Japan, Siberia, and Europe, as well as introduced regions in North America, is of particular concern. In this study, we used Sentinel-2 satellite imagery to estimate the defoliation area and predict the distribution of L. dispar in Toyama Prefecture, central Japan. The primary aim was to understand the spatial distribution of L. dispar. The normalized difference vegetation index (NDVI) difference analysis estimated a defoliation area of 7.89 km² in Toyama Prefecture for the year 2022. MaxEnt modeling, using defoliation map as occurrence data, identified the deciduous forests between approximately 35° and 50° at elevations of 400 m and 700 m as highly suitable for L. dispar. This predicted suitability was also high for larval locations but low for egg mass locations, likely due to differences in larval habitats and ovipositing sites. This study is the first attempt to utilize NDVI-based estimates as a proxy for MaxEnt. Our results showed higher prediction accuracy than a previous study based on the occurrence records including larvae, adults, and egg masses, indicating better discrimination of the distribution of L. dispar defoliation. Therefore, our approach to integrating satellite data and species distribution models can potentially enhance the assessment of areas affected by pests for effective forest management.