Integrated genomic analysis of a diverse maize population reveals novel genes and superior predictive models for ear diameter.

BACKGROUND: Ear diameter (ED) is a key agronomic trait that significantly influences maize yield and is regulated by both genetic and environmental factors. RESULTS: We systematically dissected the genetic basis of ED using a diverse multi-parent population of 858 recombinant inbred lines (RILs). Through an integrated approach combining quantitative trait locus (QTL) mapping and genome-wide association study (GWAS), we identified multiple stable loci associated with ED. Notably, we pinpointed two novel and significant loci on chromosomes 5 and 7, which harbor key candidate genes: Zm00001d020000, encoding a phosphatidate cytidylyltransferase involved in membrane biosynthesis, and Zm00001d016356, a putative Kinesin-like protein potentially regulating cell division. Functional annotation, haplotype, and protein structural analyses support their roles in regulating ear development. A comprehensive comparison of 13 genomic selection (GS) models, which demonstrated that the non-linear support vector regression (SVR) model achieved superior predictive accuracy. This highlights the substantial contribution of non-additive genetic effects, such as epistasis, to ED. Furthermore, we identified 45 elite RILs with stable performance across environments, providing valuable breeding materials. CONCLUSIONS: Our study unveils novel SNPs and candidate genes, elucidated the complex genetic architecture underlying maize ear diameter. We demonstrate that integrating GWAS and linkage mapping with advanced GS models provides a powerful strategy to dissect complex traits and deliver practical resources for accelerating yield improvement in maize breeding programs. The candidate genes identified represent promising targets for future functional validation.