Introduction: Grass pea (Lathyrus sativus) is an annual legume valued for its high yield, protein-rich seeds, rapid spring growth, and low irrigation requirements. Its tolerance to drought, salinity, and flooding, along with its nitrogen fixation ability, makes it important in crop rotation, soil improvement, and reducing weeds and diseases. In breeding programs, yield is a complex trait with low heritability, making indirect selection through associated traits more effective. Because breeders often deal with many correlated traits, multivariate statistical methods are valuable for simplifying data and identifying useful variation. Principal component analysis, cluster analysis, and discriminant analysis enable simultaneous evaluation of genotypes based on morphological, biochemical, or molecular traits. Cluster analysis, in particular, helps group genotypes by genetic similarity and select representative parents, increasing the chance of heterosis and superior offspring. This study applied multivariate methods to evaluate and classify grass pea genotypes for breeding advancement.
In plant breeding programs, selection is based on a large number of agronomic traits with positive and negative correlations, so statistical analysis methods that reduce the number of traits affecting yield are valuable for plant breeders. Analysis of genetic diversity in germplasm, especially with increasing number of variables and sample size, requires the use of multivariate statistical methods. The use of such tools allows for accurate classification of the samples under evaluation and helps the plant breeder in identifying the genetic material needed for subsequent programs and advancing breeding goals more quickly. Multivariate statistical methods that simultaneously evaluate genotypes in terms of several traits are widely used in the evaluation of genetic diversity regardless of the type of data, i.e. morphological, biochemical and molecular. Among the most important of these methods are principal component analysis, cluster analysis, and discriminant analysis.
Given that the foundation of plant breeding research is based on the evaluation, description, and introduction of suitable plant parents, this research was conducted with the aim of using multivariate analysis methods in grass pea genotypes.
Materials and Methods: To investigate sixteen promising lines of grass pea along with a control cultivar using multivariate statistical methods, a study was conducted at four research stations: Gachsaran, Kohdasht, Mehran, and Shirvan. The experiment was arranged in a randomized complete block design with three replications over three consecutive years (2017–2019). Nine traits were evaluated: 100-grain weight, grain yield, forage dry weight, forage fresh weight, grains per pod, pods per plant, plant height, days to seed maturity, and days to flowering.
Results: Cluster analysis using Ward's method and the Euclidean distance measure classified sixteen grass pea genotypes into three groups. The first cluster included eight genotypes (5, 6, 10, 11, 13, 14, 15, and 16). The second cluster consisted of genotypes 1, 2, 3, 4, 8, 9, and 12, while the third cluster contained a single genotype (7). In the principal component analysis, the first four components explained 77.54% of the total variation. The first component accounted for 23.18%, the second for 20.70%, the third for 17.46%, and the fourth for 16.19%. Days to flowering (0.893) had the highest positive loading in the first component; forage fresh yield (0.829) and 100-grain weight (0.733) in the second; days to grain maturity (0.642) in the third; and plant height (0.734) in the fourth. The second component, which can be regarded as the "forage yield component," may be used for selecting superior grass pea genotypes. Genotypes with the highest yield also had the highest factor scores in the principal component analysis, and these genotypes were grouped together in the cluster analysis.
Conclusion: The results of principal component and cluster analyses were generally consistent. These findings indicate significant variation among grass pea genotypes, which can be exploited in breeding programs to improve yield and other agronomic traits.