Using artificial neural network to evaluate salinity indices to identify rapeseed salinity tolerant cultivars

ABSTRACT Objective: This study aims to evaluate the salt tolerance of various canola genotypes in arid and semi-arid regions, particularly focusing on the identification of salt-tolerant varieties that can perform well under saline conditions. Methods: A total of 39 cultivars and lines of rapeseed were assessed for their tolerance to salinity using eight different indices. The evaluation was conducted through a randomized complete block experiment with four replications under two irrigation conditions: normal (0.831 dSm−1) and saline (8.7 dSm−1) in Kerman, Iran. Statistical analyses, including analysis of variance, regression, and artificial neural network (ANN) modeling, were employed to assess the performance of the canola varieties. Results: The results indicated significant differences between the cultivars and a notable interaction effect between environmental conditions and cultivar performance. Additionally, there was a non-significant correlation (0.021) between cultivar performance in the two irrigation conditions, suggesting genetic diversity among the genotypes for breeding purposes. The endurance indices—harmonic mean, stress tolerance index, mean product, and geometric mean product—demonstrated positive and significant correlations with seed performance in both irrigation scenarios, making them effective predictors of salinity tolerance. High-yielding varieties such as Talaye, Talaieh, T98007, Ahmadi, Modena, Option 500, and PP-4010 were identified as suitable for cultivation in saline soils. Conclusions:This research contributes to the understanding of salt tolerance in canola genotypes, providing valuable insights for breeding programs aimed at enhancing crop performance in saline environments. The identification of specific endurance indices as reliable predictors of tolerance offers a methodological advancement in the selection of salt-tolerant cultivars.