Comparison of yield stability of potato genotypes under water deficit stress via parametric and non-parametric stability analysis

The interaction of genotype with the environment provides the possibility of selecting stable genotypes for a wide range of environments. Evaluation of the interaction of genotype with the environment is necessary to increase the efficiency of selecting varieties with high and stable performance in a wide range of different environments. The objectives of this paper are: (1) evaluating the stability of 60 potato genotypes for tuber yield in two years using parametric and non-parametric stability methods, (2) identifying genotypes with good and stable performance when evaluated in variable environments and (3) investigating the relationship and correlation between stability statistics of tuber performance under water deficit conditions in Iran.

To evaluate the performance stability and adaptability of the 60 potato genotypes, two cultivars and 58 advanced clones, 17 parametric and non-parametric statistics were evaluated for tuber yield across eight environments during the 2018-2019 growing seasons. The genotypes were evaluated under normal and water deficit conditions in Karaj and Ardabil. In this study, the parametric analysis for yield was determined by such parameters as regression coefficient (bi), environmental variance (Si2), coefficient of variation (CVi), deviation from regression (sdi2), and Wricke’s ecovalence (Wi2), The non-parametric analysis included Nassar and Huhn's statistics (S(1) and S(2)), Huhn's equation (S(3) and S(6)), Shukla’s stability variance (σ2i), Plaisted and Peterson’s (θi), Thennarasu’s non-parametric (NP(1), NP(2), NP(3), and NP(4)), and Kang’s rank-sum (KR) parameter. Parametric and non-parametric statistics are used by agronomists and plant breeders. Currently, researchers are interested in applying several stability statistics to obtain the desired results, crucial for the selection of stable varieties

Composite variance analysis showed that the effect of place and year as well as the effect of genotype are significant. The interaction effect of year × location × genotype was significant at the probability level of 1%. The genotype effect was also significant at the 1% probability level. The interaction effect of genotype x location and genotype x year was not significant, which indicates that the average performance of genotypes is not different in different locations and years. Grouping of genotypes based on average performance and parametric and non-parametric stability statistics showed that genotypes are divided into four main groups. In general, based on the average rank of parametric and non-parametric stability parameters, genotypes G31, G21 and G36 had the least changes and were recognized as the most stable genotypes, and therefore they can be introduced as stable genotypes. The results of stability statistics and cluster analysis showed that G31, G21 and G36 genotypes can be introduced as stable and compatible genotypes.

Our results showed that G21, G31 and G36 genotypes contributed the least to the genetic × environment interaction (G*E) and were considered as stable genotypes under water deficit conditions. The different parametric and non-parametric stability procedures can be proposed to select drought tolerant genotypes under different environments conditions; these procedures could be useable for recognition of the best genotypes under drought stress conditions. Therefore, the analysis of yield stability could be utilized in combination with parametric and non-parametric methods to evaluate and identify drought tolerance genotypes. Dendrogram results confirmed each other with the results of parametric and non-parametric statistics. While G49, G51, G53 and G56 genotypes with the highest values were the most unstable genotypes.