Researchers have integrated genomic data into diploid banana breeding programs to improve predictions for yield and growth. Published on April 14, 2026, in Horticulture Research (DOI: 10.1093/hr/uhag139), a study reveals that a hybrid model combining pedigree records and DNA markers reduces bias and sharpens selection accuracy for vital staple crops.
Bananas and plantains rank among the most crucial fresh fruits and staple food crops worldwide, serving as a primary staple food for a weighty population in developing countries. They seize the fourth spot among the mainstay food crops in underdeveloped countries—as classified by the United Nations—based on overall production and food consumption. Producing roughly 179.3 million tons annually, the global sector generates an export value of US$13.8 billion across a volume of 26.48 million tons. Most of their production stems from small-scale farmers who either consume them locally or engage in domestic trade. Yet traditional breeding for these crops remains exceptionally slow and labor-intensive. A standard breeding cycle demands two to three years of field evaluation per generation, during which plants frequently suffer field losses that leave incomplete data records.
Genomic Architecture and Breeding Challenges in Musa Cultivars
Cultivated bananas and plantains stem from ancient hybridization events involving wild species including Musa acuminata (A), M. balbisiana (B), M. schizocarpa (S), and M. troglodytarum (T). These wild progenitors contributed distinct genetic building blocks, creating complex genomic mosaics characterized by polyploidy and structural heterozygosity. While these genetic traits confer seedlessness and stress resilience, they also cause reduced fertility. This low fertility severely restricts the number of viable hybrid offspring available for evaluation.
Conventional pedigree methods rely heavily on expected relationships calculated from family lineage records. Unfortunately, these records are frequently incomplete, particularly when breeding programs utilize landrace varieties with unknown ancestry. To bypass these hurdles, researchers are turning to high-throughput sequencing, molecular markers, and cytogenetics to characterize chromosome structures and biochemical signatures. These advanced genomic tools help capture allelic diversity, trace chromosomal translocations, and monitor clonal fidelity across commercial diploid and triploid cultivars.
Comparative Performance of Pedigree and Genomic Models
To establish more reliable genetic predictions, a collaborative team of scientists—representing the International Institute of Tropical Agriculture (IITA) in Uganda and Tanzania, the Swedish University of Agricultural Sciences, KU Leuven in Belgium, and Gnomixx B.V. in Belgium—evaluated 74 diploid Mchare banana hybrids derived from a factorial cross of four female and five male parents, utilizing 2,792 high-quality single nucleotide polymorphism (SNP) markers to construct genomic relationship matrices.
The researchers systematically compared three distinct approaches:
- Pedigree-based best linear unbiased prediction (P-BLUP)
- Genomic best linear unbiased prediction (G-BLUP)
- A hybrid model (H-BLUP) integrating both pedigree and DNA marker data
For yield-related traits—including bunch weight, number of hands, fruit length, and fruit circumference—additive genetic variance proved to be the dominant component across all three models. However, the P-BLUP model consistently overestimated these additive effects compared to G-BLUP and H-BLUP. Pedigree-based models fail to adequately separate additive from non-additive genetic effects because they rely on expected relationships rather than actual genetic sharing, which artificially inflates the additive component and misleads breeders about expected genetic gains.
Precision Breeding Gains Through the H-BLUP Approach
The study demonstrated that integrating genomic data with traditional pedigree records provides tangible advantages for crop improvement. The H-BLUP model produced the smallest standard errors for variance component estimates, signaling superior precision. While the P-BLUP model showed superior performance for predicting agronomic traits like plant height and girth, the H-BLUP model outperformed the others for yield-related and cycling traits.
Crucially, the hybrid model can generate predictions for individuals lacking DNA data—such as plants that fail genotyping or are lost during long evaluation cycles—because it incorporates information from both genotyped and non-genotyped individuals through combined pedigree and genomic matrices.
Authors
The analysis also identified four top-performing hybrids with consistently high breeding values across all three models, with three of those standout plants originating from crosses involving the female parent Huti-White. Funding for the research was provided by the Gates Foundation under grant numbers INV-002979 and OPP1213871, supporting efforts to secure food supplies and accelerate genetic gains in regions heavily reliant on the crop.