Georgi Raykov
Department of Crop Science, Agricultural University Plovdiv, 4000 Plovdiv, Bulgaria

Manuscript received 6 April 2026; accepted for publication 30 August 2026)
e-mail: georgi.raikov.dzi@gmail.com

Abstract. Grain yield formation in winter wheat is a complex process governed by the interaction of multiple yield components, whose relative contributions are difficult to quantify under multi-environment conditions due to strong interdependencies among traits. The present study aimed to identify the key determinants of grain yield and to clarify their relationships using an integrative analytical framework. A dataset comprising 36 winter wheat genotypes across four experimental locations in Bulgaria over four growing seasons was analyzed. Correlation analysis, linear mixed modelling, Random Forest, and Partial Least Squares regression were applied to assess trait contributions and address multicollinearity. Across all approaches, the number of grains per square meter consistently emerged as the dominant determinant of grain yield, showing the strongest association (r=0.92), the highest standardized effect (β=1.330), and the greatest importance in the Random Forest model (0.896). In contrast, the number of productive tillers and the number of grains per spike showed strong pairwise associations with yield but negligible or negative effects in multivariate and non-linear models, indicating predominantly indirect contributions mediated through grain number per unit area. Thousand grain weight exhibited a partially independent contribution, supported by its importance in the Random Forest model (0.090) and its strong loading on the second latent component in the Partial Least Squares analysis (0.956). These results indicate that grain yield formation is governed by a hierarchical structure dominated by grain number, while grain weight plays a complementary role. The integration of multiple analytical approaches provides a robust framework for understanding trait interactions and supports targeted trait prioritization in wheat breeding under variable environmental conditions.