Morphological and phenotypical traits and their relationship with soybean grain yield: a multivariate analysis approach

An alternative to the conventional variance analysis to extract more adequate answers involving the relationship of morphological and phenological variables with grain yield is using multivariate statistical analysis. The objectives of the work were: (i) to group soybean cultivars, planted in two planting seasons, based on their similarity for morphological and phenological traits; (ii) evaluate the relationship between grain yield and morphological and phenological traits within and between cultivar groups. Principal component analysis, cluster, discriminant and canonical analysis were applied to data. For cultivars planted in the first planting season, the phenological cycle and the number of nodes emitted can help to select cultivars with higher yield potential only for a group of cultivars, in this case, cultivars classified in G1 and G5. For these groups, the lowest phenological cycle and node emission are traits related to high yield cultivars, reaching up to 95 bags ha⁻¹. Cultivar groups with a shorter cycle and a larger plant stand are favored when planted in the second soybean planting season, showing a high potential for grain yield. For cultivars with yield potential ranging from 55 to 85 bags ha⁻¹, a shorter cycle, lower height of insertion of the first pod and number of nodes should be chosen. Based on the results, it can be concluded that the application of principal components analysis, cluster and discriminant analysis allow a better understanding of the existing relationships between the morphological, phenological components, and yield of soybean.