Study of Diversity and Estimation of Leaf Area in Different Mint Ecotypes Using Artificial Intelligence and Regression Models under Salinity Stress Conditions

     Leaf area is a key indicator for the growth and production of plant products and also determines the efficiency of light consumption. Therefore, the study of diversity and also the estimation of leaf area in different mint ecotypes is particular importance. One of the common methods for estimating leaf area is regression analysis, the leaf area as independent variable, and leaf length and width as dependent variable. In this study, leaf area of ​​18 mint ecotypes with different models of ANFIS, artificial neural network (MLP and RBF), linear and nonlinear regressions using two inputs of leaf length and width in four levels of salinity stress (control, 2.5, 5 and 7.5 dS/m) were estimated in two levels of harvesting. The results showed that there is a high correlation between length and width with leaf area, so that the width correlation with leaf area was greater than leaf length. Among the regression models in both harvest levels and at all levels of salinity stress, the NLR model is presented as the best and most accurate model. ANFIS model was harvested in both stages and at all levels of stress was more accurate than other models and less error rate was obtained. Also, the results of cluster analysis showed that there is a good variation between ecotypes. In addition, the mean comparison between different cluster analysis groups showed that the highest amount of length, width and leaf area was taken in both stages and at all levels of salinity stress was related to E18.