Compositional Optimization for Miticidal Activity, Ecotoxicity, and Phytotoxicity of <i>Rosmarinus officinalis</i> Essential Oils as Biorational Pesticides

Recognizing the challenges in using botanicals as sustainable pest control agents due to compositional variation, this study addresses the limitations of traditional component-based approaches such as Hewlett and Plackett or Wadley's models. Based on the assumption of noninteractivity among constituents, these models often fail to predict outcomes accurately due to dynamic intermolecular interactions. We introduce a whole mixture-based approach, employing a combination of experimental design and polynomial modeling. This technique accurately predicts miticidal activity on <i>Tetranychus urticae</i>, ecotoxicity on <i>Daphnia magna</i>, and phytotoxic activities on <i>Phaseolus vulgaris</i> of <i>Rosemarinus officinalis</i> essential oils with varying composition. The RMSE values from the polynomial model are 66.9 and 5.0 for miticidal activity and ecotoxicity, respectively, while they are much higher in component-based models, up to 1097.7 and 41.3, respectively. Additionally, we utilize multiobjective optimization algorithms to identify the optimal supplementary blending of oils and compounds. This strategy aims to maximize miticidal effectiveness while minimizing ecotoxicity and phytotoxicity. Our approach for predicting multicomponent mixture effects is likely to bridge the knowledge gap between research and commercialization.