Yield-associated trait analysis and candidate high-yielding germplasm selection among 49 Castanea mollissima germplasm accessions

Abstract Chinese chestnut ( Castanea mollissima ) is an economically important tree crop in China, but its relatively low and variable yield remains a major constraint on production improvement. In this study, 20 yield-related traits encompassing inflorescence, stem, and fruit attributes, including spiny cupule and nut characteristics, were evaluated in 49 germplasm accessions to characterize phenotypic variation and diversity patterns. Three main findings emerged. First, all 20 traits exhibited substantial variation, with coefficients of variation (CVs) ranging from 19.00 to 70.06% and Shannon–Wiener phenotypic diversity indices (H′) ranging from 1.37 to 1.89. Spiny cupule number per tree (SCN-T) had the highest CV, whereas yield per tree (Y-T) had the highest H′, indicating that the two measures ranked the traits differently. Second, hierarchical clustering grouped the traits into three clusters. The composite score derived from Cluster I after excluding yield per tree showed the largest observed correlation with yield among the four trait sets ( r = 0.53, p < 0.01). Third, two-step cluster analysis based on Cluster I traits classified the 49 accessions into three yield groups: high-yielding (7 accessions, 14.3%), intermediate-yielding (19 accessions, 38.8%), and low-yielding (23 accessions, 46.9%). The seven accessions in the high-yield group were designated candidate high-yielding accessions. Their performance has not been independently validated across additional years, environments, or populations; the two-year observations do not establish long-term yield stability. Collectively, these results describe phenotypic associations among reproductive and fruit-related traits and yield, providing an empirical basis for germplasm evaluation and candidate selection for further breeding studies.