Wheat seed germination prediction in response to temperature, water potential, and salinity using an artificial neural network

Abstract

Abstract Multi-layer perceptron (MLP) neural networks can be used to develop accurate models for quantifying plant responses to environmental factors. This study aimed to quantify wheat seed germination in response to temperature, water potential, and salinity using an ANN. Results indicated that the MLP model could predict total germination percentage with high model accuracy, including R 2 (0.99), MSE (0.342), RMSE (0.585), and MAE (2.166) for the test data. Time to 50% germination (T50) was also accurately estimated using the MLP model (R 2  = 0.97, MSE = 26.2, RMSE = 5.11, MAE = 5.70). Water potential was identified as the most significant variable affecting total seed germination and T50, followed by salinity and temperature. Seed germination was maximum at 20.5 °C and decreased at higher and lower temperatures. The optimal temperature for T50 was 25.3 °C. Higher salinity and more negative water potential led to lower total seed germination. The results of this study can be used to develop process-based models of crop growth and development and predict total seed germination and germination time under different conditions of temperature, water potential, and salinity.