Related Experiment Video
Updated: Sep 10, 2025
Assessment of Chemical Toxicity in Adult Drosophila Melanogaster
Published on: March 24, 2023
Prediction of mixture toxicity of essential oil constituents using nonparametric and parametric models against Musca
Junho Yoon1, Jun-Hyung Tak1,2
1Research Institute of Agricultural and Life Sciences, Seoul National University, Seoul, South Korea.
Background:
Growing concerns over the environmental and human health impacts of conventional insecticides have intensified interest in botanical alternatives, including essential oils. However, their compositional complexity poses challenges for quality control and consistent efficacy. This study evaluated and compared the predictability of nonparametric and parametric models for predicting the mixture toxicity of essential oil constituents against Musca domestica.
Results:
Conventional nonparametric null models, including Loewe, Bliss, Schindler and Highest Single Agent, were applied to 210 binary mixtures showed limited predictive accuracies of 30%, 46%, 42% and 41% of combinations, respectively. These limitations are due to their assumption of noninteractivity, which is often violated by essential oil constituents. By contrast, parametric models were evaluated using six combinations. Zimmer provided the best predictions for trans-anethole + eugenol, trans-anethole + thymol and borneol + thymol; MuSyC achieved the best fit for camphor + α-terpineol, carvacrol + carvone and d-limonene + α-terpineol, and had the lowest average RMSE overall (8.22%), followed by Zimmer (8.79%) and BRAID (9.10%).
Conclusion:
Parametric models provide a more accurate prediction than nonparametric ones by accounting for interactions, despite their extensive experimental data and challenges in parameter interpretation. These findings highlight that the selection of an appropriate modeling approach should be guided by the specific application: nonparametric models may be suitable for high-throughput screening of interacting combinations, whereas parametric models are more appropriate when accurate prediction is critical. Further development of modeling approaches grounded in mechanistic understanding will enable clearer integration between biological data and modeling frameworks. © 2025 The Author(s). Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.
