Accelerating perennial crop improvement via multi-omics-based predictive breeding

Perennial crops are positioned at a critical juncture, facing intensifying environmental challenges that threaten productivity. Despite the high value of these crops, breeding gains in perennials are notably slow due to prolonged breeding cycles, often exceeding several decades, and thereby limiting their capacity to adapt to increasing climatic stressors. In contrast, annual crops have begun to leverage predictive breeding methods to incorporate multi-omics data, paving the way for a new era of accelerated genetic improvement. Multi-omics approaches integrate diverse datasets, ranging from genomic to proteomic layers, and likely more comprehensively capturing system features of regulatory networks that link the genome and phenotype. In this review, we assess the current landscape of predictive breeding in perennials by examining single-omic approaches alongside emerging omics resources, and we compare these trends with established multi-omics-based prediction frameworks in annual crops that have yielded enhanced predictive ability and novel biological insights. Building on these comparisons, we outline key considerations for implementing multi-omics-based genetic improvement frameworks in perennials, emphasizing the need for an end-to-end, reproducible, and scalable system that integrates multidimensional datasets and models both additive and nonadditive genetic effects across genotype-by-environment-by-management interactions. We also address significant challenges, including high data dimensionality, complex genotype-by-environment interactions, and limited training population sizes, and propose cross-institutional collaborations to pool resources, as well as the use of breeding program simulation tools to optimize multi-omics integration into practical breeding strategies. Despite current limitations, multi-omics-based predictive breeding holds great promise as a powerful tool for rapid genetic improvement in perennial crops.