Correlation and path analysis of morphological traits in a bread wheat collection using Structural Equation Modelling

The aim of this study was to investigate the structural determinants of grain yield per plant in bread wheat (Triticum aestivum L.) and to assess the contributions of morphological and yield-related traits. Descriptive statistics, Pearson’s correlation analysis, path coefficient analysis, and Structural Equation Modelling (SEM) were employed. A total of 53 genotypes from the bread wheat collection of the Genetic Resources Institute (Baku, Azerbaijan) were evaluated for grain weight per spike, thousand-kernel weight (TKW), grains per spike (GPS), plant height (PH), and main spike length (SL). Analysis revealed that grain yield per plant (GWPP) was primarily influenced by numerical (GPS, r = 0.47) and mass-related (TKW, r = 0.29) traits, whereas architectural traits (PH, SL) contributed mainly indirectly. SEM showed that grain weight per spike (GWPS) acted as a central mediating variable, channelling the effects of GPS and TKW to overall yield. The mass/sink axis (TKW → GWPS) exhibited a stronger direct effect (standardised coefficient = 2.18) than the architectural axis (0.41), while architectural traits primarily supported sink capacity indirectly. The final model demonstrated excellent fit (χ²(3) = 1.26, p = 0.74; Comparative Fit Index = 1.00; Root Mean Square Error of Approximation = 0.00; Standardised Root Mean Square Residual = 0.036), supporting the biological interpretability of the observed trait relationships. These results provide practical insights for plant breeders and geneticists seeking to select high-yielding wheat genotypes under controlled field conditions and can be utilised in wheat breeding programmes to optimise strategies for improving bread wheat production