Due to the concerns over their environmental and health impacts, there have been attempts for shift towards biorational pesticides from synthetic pesticides. Among them, plant essential oils have emerged as promising active ingredients. Due to the complex interactions among their constituents, the bioactivities of essential oils can vary depending on the compositions, which often undermine their stability in efficacy. Here, we present a model-based optimization approach to develop reliable rosemary oil-based biorational pesticide, against two-spotted spider mites, Tetranychus urticae Koch. The ecotoxicity against Daphnia magna and foliar phytotoxicity against Phaseolus vulgaris were also evaluated. Our quadratic models accurately predicted miticidal activity, ecotoxicity, and phytotoxicity. We aimed to maximize, minimize, and minimize these parameters, respectively. We employed seven multi-objective evolutionary algorithms in Matlab. Among them, the nondominated sorting genetic algorithm II with adaptive rotation based simulated binary crossover (NSGA-II-ARSBX) performed best. We experimentally determined the thresholds for miticidal activity and phytotoxicity, based on the current approval process for agricultural pesticide products in Korea. After applying the thresholds, we validated the obtained viable solutions. Our study offers a novel framework to enhance the reliable and responsible use of essential oils as biorational pesticides.

• Jun-Hyung Tak(Department of Agricultural Biotechnology, Seoul National University, Research Institute of Agricultural and Life Sciences, Seoul National University)
• Junho Yoon(Department of Agricultural Biotechnology, Seoul National University) Corresponding author