Predictive modeling of Satureja rechingeri essential oil yield and composition under water deficit and soil amendment conditions using artificial neural networks (ANNs)

Abstract

The cultivation of medicinal plants is a complex process that involves the processing of large data across various factors. With the introduction of artificial networks in plant sciences, it has become possible to decipher complex relationships in nature. The composition and yield of Satureja rechingeri essential oil is a multifactor trait. Therefore, it is essential to estimate the performance of essential oils using fast and low-cost methods with acceptable accuracy. An intelligent technique was developed using an artificial neural network (ANN) to predict the content and main components of savory essential oil across diverse conditions. Specifically, an ANN was developed to model the relationship between soil amendments (biochar and superabsorbent) in different drought stress conditions as independent variables, and oil content and main essential oil compounds as dependable variables. Results of GC–MS analyses of savory revealed that carvacrol (78.6%), γ-terpinene (3.0%), p-cymene (2.0%), terpinene-4-ol (2.5%), 1, 8-cineole (2.0%) and linalool (2.3%) were the major constituents of essential oil. The ANN model with a 3–12-9 topology was applied as the ideal scheme to predict carvacrol, γ-terpinene, p-cymene, terpinene-4-ol, 1,8-cineole, and linalool levels in S. rechingeri. The efficiency of the model was confirmed using the r value as the statistical parameter. As a result, the highest (0.9687) and lowest (0.6642) correlation coefficients between predicted and experimental values were obtained for terpinene-4-ol and p-cymene, respectively. The current work demonstrates the potential of the ANN model in predicting essential oil content and main oil components in savory plants accurately as a low-cost non-invasive method.