Synthetic seed production in Crataegus monogyna L. and prediction of regeneration of synthetic seeds with machine learning algorithms

Abstract Crataegus monogyna is a complex species that is essential in both ecological and therapeutic domains. Its versatility across several settings, along with its extensive phytochemical composition, renders it a significant focus of research in both botany and medicine. One of the research areas that could be focused on is the propagation of C. monogyna under in vitro conditions. In this study, we encapsulated nodal segments of sterile shoots from C. monogyna plants growing naturally in Van, Türkiye. Encapsulated propagules were cultured in hormone-free Murashige and Skoog medium or Murashige and Skoog medium supplemented with 2 mg/L indole-3-acetic acid (IAA) after being stored for 30, 60, or 90 days at −20, 4, or 24 °C. The regeneration of synthetic seeds under the effects of hormone (IAA), storage temperature, and storage period was predicted using five machine learning algorithms: Decision Tree (DT), Gaussian Process (GP), Multi-Layer Perceptron (MLP), Random Forest (RF), and XGBoost (Extreme Gradient Boosting). Feature importance analysis was conducted to identify the key factors influencing regeneration outcomes. The DT, MLP, RF, and XGBoost models achieved high prediction accuracy (97.3%). Furthermore, while the DT and XGBoost models identified temperature as the most influential factor, the MLP and RF models found hormone to be the most significant. Surface and contour plot analyses were also employed to assess the relationships visually between key features of the regeneration process.