Cytonuclear analysis of barley spike traits using a cytoplasm-aware population

The interplay between nuclear and cytoplasmic genomes—collectively known as cytonuclear interactions (CNIs)—is increasingly recognized as a key driver of phenotypic variation and adaptive potential across diverse organisms. Yet, leveraging cytoplasmic diversity and fully understanding CNIs’ contributions to agriculturally important traits remain major challenges in crop improvement, largely due to the scarcity of tailored genetic resources. In cultivated barley ( Hordeum vulgare ), limited genetic diversity relative to its wild relatives constrains adaptability to changing environments. While wild germplasm offers a reservoir of valuable alleles, the role of cytoplasmic variation and CNIs in shaping complex traits is still poorly understood. To address this gap, we present the Cytonuclear Multi-Parent Population (CMPP)—a novel interspecific resource comprising 951 BC ₂ DH lines, generated from crosses between ten genetically diverse wild barley accessions ( H. vulgare ssp. spontaneum ) used as female founders and the elite cultivar Noga. This design facilitates the concurrent segregation and analysis of nuclear introgressions involving distinct wild versus cultivated cytoplasmic backgrounds from ten subfamilies. Phenotyping across multiple environments revealed that up to 5% of variation in key spike and grain trait BLUPs are explained by cytoplasm (η² = 0.05), including Thousand Grain Weight (TGW), Grain Width (GW), and Fruiting Efficiency at Maturity (FEm). Notably, wild cytoplasms influenced trait stability, with the B1K-50-04 cytoplasm increasing TGW stability based on Shukla’s measure. Genome-wide association studies (GWAS) employing Nested Association Mapping (NAM), FASTmrMLM, and MatrixEpistasis (ME) identified 76 marker-trait associations (MTAs). The ME approach specifically uncovered 16 cytonuclear QTL (cnQTL) exhibiting cytoplasm-dependent effects. Furthermore, we developed a genomic prediction (GP) strategy incorporating interactions between significant MTAs and population structure variables (subfamily and cytoplasm). This targeted interaction model (“Peaks + I”) achieved cross-validation accuracies comparable to, or even exceeding, models using the full set of 6,679 SNPs, despite utilizing substantially fewer predictors, enabling quicker and more efficient validation runs. The CMPP provides a unique platform for dissecting cytoplasmic effects and CNIs, highlighting the importance of incorporating cytonuclear context in genetic mapping and prediction to effectively harness both nuclear and cytoplasmic diversity for crop improvement.