REMOTE PEST MONITORING: STATE OF THE ART AND CASE STUDY USING ISCOUT TRAPS IN DOURO REGION, PORTUGAL

Insect populations in agriculture exhibit high spatial, temporal, and species-specific diversity, posing significant challenges for effective pest management. Traditional pest monitoring relies on manual checks of sticky traps, which are labor-intensive and provide limited spatial and temporal coverage. Automated systems using image sensors and machine learning algorithms to detect and identify pest species offer significant advantages, enabling real-time, remote monitoring and improving data resolution across larger vineyard areas. This approach aligns with the principles of Integrated Pest Management (IPM), promoting more efficient and sustainable pest control strategies. This thesis consists of both a review and experimental work. The review provides an overview of state-of-the-art remote insect monitoring technologies in agriculture, highlighting the evolution of systems based on image sensors and machine learning. It discusses the benefits and limitations of these technologies, offering a comprehensive understanding of their role in sustainable pest management. The experimental study was conducted as part of the Biod'Agro project, which aims to support agrobiodiversity management and water availability monitoring in viticulture by integrating innovative technologies with sustainable agricultural practices. The research was carried out in the Quinta da Extrema vineyard, located within Portugal’s International Douro Natural Park. The focus of this experimental study is to evaluate the performance of iScout traps equipped with automatic pest identification algorithms for monitoring Lobesia botrana populations, comparing their effectiveness with traditional Delta Sticky Traps. The study was conducted during 2022 and 2023 across two vineyard plots of the Touriga Nacional grape variety. It assessed the accuracy of the iScout traps, the types of errors they might generate, and their side effects on non-target insect captures. Manual image analysis revealed discrepancies between the algorithm’s pest identification results and laboratory observations, highlighting opportunities to improve the system's accuracy. Despite these limitations, the iScout traps demonstrated potential in providing a more sustainable and efficient approach to pest monitoring by reducing non-target insect captures and offering enhanced spatial and temporal resolution. In conclusion, while further refinements to the algorithm's accuracy are necessary, ongoing use and additional research are expected to enhance its performance. The findings highlight the potential of iScout traps as a valuable tool for remote and sustainable pest management in viticulture, contributing to the broader goals of the Biod'Agro project.