Evaluation of agronomy and botanical traits using multivariate statistical methods in Soybean (Glycine max L.) genotypes

Soybean (Glycine max L.) is one of the main sources of vegetable oil and protein in the world, which contains 18-23 percent oil and 40-30 percent protein (Smith & Huyser, 1987). Various factors such as susceptibility to pests, various diseases and environmental stresses, etc. contribute to limited yield in soybeans. In order to overcome these limitations, it is necessary to identify different sources of genetic resources and desirable traits, which will be beneficial for soybean yield improvement (Moe & Girdthai, 2013). The current research was aimed at evaluating important agronomic and botanical characteristics of 354 soybean germplasm obtained from Karaj Seed and Plant Improvement Research Institute as well as classifying them based on some effective characteristics in seed-yield breeding.

In 2013, after the preparation of the experimental land by using conventional cultivation operations (plowing, two times disk, application of fertilizers in soil and forming 60-cm planting rows), 354 soybean genotypes were grown in a nested statistical design with four sets of 51 genotypes and 3 sets of 50 genotypes with two replications . in duration growth period important botanical characteristics such as: day to flowering, day to podding, day to start seed filling, day to end of seed filling, day to physiologic maturity, day to maturity, seed filling duration, long reproductive duration index, growth habit, evaluated based on Fehr and Caviness method (1977). In maturity stage agronomic traits such as: plant Height, no. of nods, no. of pods per plant, no. of branches, no. of seed per plant and no. of seed per pod evaluated and after harvesting seed yield per plant and 100 seed weight determined. Statistical indexes such as: Mean, minimum, maximum and variation coefficient of the traits calculated and to variance analysis of data based on the nested design was used, SAS software, Soybean genotypes were classified using cluster analysis based on hierarchical method (Ward criterion) by SPSS software. To verify the accuracy and the validity of the groups derived from cluster analysis, the principal component analysis was conducted by SPSS software Ver.15. Moreover, the two-dimensional graphical diagram of the first and second components was generated by Gen Stat 12th Edition.

The results of analysis of variance showed a significant difference between genotypic sets for all the studied traits. Significant variation was observed for the important agronomic traits in the studied genotypes. So that the highest variation coefficient were for growth habit (55%), grain yield (40%) and number of seeds per plant (38%). the variation range of the most important traits were for grain yield 2.1- 25.7 g,. Growth period 126-86 days, length of grain filling period 29 - 13 days, Ratio of reproductive duration to length of growth period 0.71-0.47, number of seeds per plant 21.2-192 and 100 seed weight 6.8 - 24.7 gr. Cluster analysis divided 354 genotypes into four clusters, in which cluster 1 Most of the genotypes were late-maturity, indeterminate type, with a higher height. in cluster 2, Most of the genotypes were determinate type , medium height, in cluster 3, the most genotypes had high yield and seed number per plant with longer growth period and in cluster 4, the most genotypes were semi-determinate, very early maturity and had short height and low yield. The analysis of the principal component analysis showed that the first five components explained 32.5%, 18.1%, 9.8%, 7.98% and 7.5% of the total variance of the data respectively. The biplot of the phenological traits (PC1) and the yield component (PC2) showed that distribution pattern of genotypes had a good agreement with the cluster analysis results and genotypes GN 1130, GN 1028, GN 1040, GN 2129 and GN 2122 can be genotypes due to the high potential yield and number of seeds per plant, and the indeterminate type as superior genotypes for use in breeding programs