Integrating Genomic and Phenomic Data Improves Prediction of Key Traits in Potato Breeding

Abstract Potato breeding faces challenges due to clonal propagation and complex polyploid genetics, limiting the acceleration of genetic gains. Predictive breeding integrating genomic and high-throughput phenotypic data offers potential to enhance selection accuracy. Here we evaluated multiple regression models combining genomic single nucleotide polymorphism (SNP) data and UAV-derived image-based traits across diverse potato clones and cultivars to predict key agronomic traits. Combining genomic and phenomic data significantly improved prediction accuracy for total tuber yield compared to using either data source alone, with phenomic data contributing complementary environmental information. Prediction accuracy varied by trait, with genomic data better predicting starch content and phenomic data excelling for tuber yield-related traits. Our findings demonstrate that integrating genomic and phenomic information can enhance predictive breeding strategies, potentially accelerating genetic gains and optimizing resource use in potato breeding programs.