The evaluation of elite lines obtained from red common bean (Phaseolus vulgaris L.) local populations

Phaseolous vulgaris is highly important among bean species. China, Iran and Japan are the most important countries producing bean in Asia. According to reported information, cultivated area of bean was 94000 ha in Iran which 17841 ha located in Lorestan province. Improvement of new cultivar with high genetic potential for grain yield is the final goal in many breeding programs. To achieve this goal, many characteristics should be considered. Positive and negative significant correlation has been reported between traits in common bean. Various, traits had different direct and indirect effects on grain yield which should be considered. In addition to path analysis, evaluation and relationship of traits by other multivariate statistical methods such as cluster and factor analysis have been studied for better understanding of these relationships. This study aimed to investigate superior lines, study the relationship between important traits with seed yield by some univariate and multivariate statistical methods, and provide functional recommendation to breeding programs.
 

Elite lines obtained from red common bean (Phaseolus vulgaris L.) local populations of Lorestan province and suburbs, Iran, including 14 lines belonged to Azna (2 lines), Aligudarz (4 lines), Aleshtar (1 line), Borujerd (1 lines), Durood (2 lines), Shazand (2 line) and Nourabad (2 lines), with two control cultivars Goli and Sayyad were evaluated in randomized complete block design (RCBD) with 4 replications in Borujerd station, Lorestan, Iran, in 2011. Plant height, node number, stem number, pod length, pod number per plant, seed number per plant, 100 seed weight and seed yield characteristics were measured. Data were normalized by Kolmogorov-Smirnov test at 0.01 probability level. Error, genetic and phenotypic variances, and broad sense heritability pertreatment mean were estimated. Pearson correlation coefficient was used to determine the relationship between traits. Cluster analysis based on UPGMA was performed and Wilks' lambda statistic was used for cutting dendrogram. Then, a code was allocated to each group and these groups were compared. However, in order to fix type I error, Hotelling’s T2 test was used at 0.05 probability level. Before that, the assumption of homogeneous variance-covariance matrix was tested by BOX test at 0.01 probability level. Then, traits were compared. Seed yield was considered as dependent variable (y) and regression analysis was performed by stepwise method. Traits, that had significant standardized coefficient, were used in path analysis. For statistical analysis, Excel, SPSS and MSTAT-C software were used.
 

The results of ANOVA showed that the lines had significant difference for all studied traits. Mean comparison by Duncan test showed the highest and lowest seed yield belonged to Aligodarz-3 line (6058 Kg ha-1) and Goli cultivar (3175 Kg ha-1), respectively. All of the genotypes that were located in high seed yield group by cluster analysis belonged to elite lines that indicated the local red common bean populations, as a heterozygote germplasm source that been mixed of homozygous genotypes, had high potential to selection of superior lines in the breeding program. Comparison between groups by Hotelling-T2 statistics indicated seed number per plant and had significant difference at 0.01 probability level. The highest and lowest significant correlation coefficient was observed for seed number per plant with seed yield (0.91**) and node number with stem number (0.27*), respectively. The results of multiple regression for seed yield as dependent variable showed that seed number per plant and 100 seed weight had significant standardized regression coefficient. While, 100 seed weight had no significant correlation coefficient with seed yield that breeders should be considered different aspect of trait relationships. Path analysis results showed seed number per plant with direct effect 1.03 had more effect on seed yield than 100 seed weight with direct effect 0.43. While, both seed number per plant and 100 seed weight had negative indirect effect on seed yield by each other. The results of factor analysis showed four factors explained 93% of total variation. The first and second factor were called yield and yield component, and phenological type explained 60% of total variation. The lines that located in high yield group had the highest value for yield and yield components factor. While, the lowest score for mentioned factor belonged to Goli cultivar that showed factor analysis can be used both to summarize many dependent variables (traits) into little independent variables (factors) and to selective genotypes based on factor value. According to reports, the genetic base of bean cultivars has been limited. One of the useful results of multivariate analysis is the investigation of genotypes that locate at the end of distribution for several traits and these genotypes can be used as parents for better utilizing of probably heterosis. Heterosis and crossing one of the sustainable agricultural goals would improve germplasm, increase human food, promotion of farmer’s livelihoods and provide food.
 

Local populations of red common bean should be considered by breeders to select superior lines because of its potential and adaptation. Some studied traits had high diversity that could be exploited in breeding programs. While, other traits need to increase diversity by breeding strategies. Considering that each of studied analysis showed the different aspects of traits relationships and genotypes potential. So, the results should be considered simultaneously by breeders to better interpretation. Among the studied traits seed number per plant, according to high heritability, the identification of groups with different grain yield, high correlation and positive direct effect on grain yield was appropriate to increase seed yield. Multivariate analysis can be used to evaluate desirable genotypes and accumulate favorable alleles in breeding programs. The results of this study can be useful for breeders to investigate and utilize both traits and genotypes in breeding programs.