Landmark study demonstrates the potential of across-population genomic prediction in grapevine breeding

Crop breeding in grapevine (Vitis vinifera L.) typically relies on phenotypic selection, but this process is slow and labor-intensive. Genomic prediction (GP) offers a more efficient alternative, yet its application in grapevine remains limited to single-population studies. Current research primarily focuses on GP for predicting genotypes within crosses, but lacks comprehensive evaluation of its effectiveness across diverse populations. The key research gap is exploring the predictive ability of GP in more varied genetic contexts, essential for advancing genomic selection in grapevine breeding.

In February 2022, Horticulture Research published a research article entitled by “Across-population genomic prediction in grapevine opens up promising prospects for breeding”.  

In this study, researchers aimed to evaluate across-population genomic predictive ability (PA) in grapevine breeding, focusing on 15 traits related to yield, berry composition, phenology, and vigor. The study was conducted using phenotypic data from a diversity panel and a half-diallel of 10 bi-parental crosses. The genetic diversity within the half-diallel population was first examined through a principal component analysis (PCA), revealing that the half-diallel crosses were genetically close to the wine west (WW) subpopulation of the diversity panel. The broad-sense heritability (H²) values for the 15 traits were medium to high, and varied among half-diallel crosses. The cross effect on genetic value estimation also varied depending on the trait, and the observed phenotypic diversity and transgressive segregation within half-diallel progenies for multiple traits.

For predicting cross mean, the study compared methods (RR and LASSO) best suited to the genetic architecture of each trait. Predictive ability was measured by Pearson's correlation between observed and predicted genotypic values. In scenario 2, where allelic effects were estimated within the whole diversity panel, per-trait and per-cross PA was lower and more variable compared to the other two scenarios. The study found upward or downward bias in some traits and scenarios, with LASSO showing larger bias in scenario 1a. Regarding Mendelian sampling prediction, PA varied significantly among crosses and traits. In scenario 1a, the average PA per trait ranged from 0.18 to 0.57, with considerable variation among crosses. Scenario 1b, involving two PA values per cross, showed that PA was often lower than in scenario 1a. In scenario 2, the overall average PA was nearly half that of scenario 1a, with some traits showing a marked decrease in PA. The study also explored factors affecting PA, such as genetic relationship between parents, non-segregating markers, and the relative variance of cross effect. Using optimized training sets from the diversity panel did not improve PA, suggesting that genetic relationship did not significantly affect PA.

In conclusion, this study provided valuable insights into the genomic prediction in grapevine breeding, highlighting the effectiveness of GP for certain traits and crosses and identifying key variables influencing PA. The research demonstrated moderate to high PA values for some traits and crosses, showing the potential of GP in grapevine breeding and suggesting further areas for improvement and exploration.

###

References

Authors

Charlotte Brault^1,2,3, Vincent Segura^1,2, Patrice This^1,2, Loïc Le Cunff^1,2,3, Timothée Flutre⁴, Pierre François^1,2, Thierry Pons^1,2, Jean-Pierre Péros^1,2 and Agnès Doligez^1,2,*

Affiliations

¹UMT Geno-Vigne®, IFV-INRAE-Institut Agro, F-34398 Montpellier, France

²UMR AGAP Institut, Univ Montpellier, CIRAD, INRAE, Institut Agro, F-34398 Montpellier, France

³Institut Français de la Vigne et du Vin, F-34398 Montpellier, France

⁴Université Paris-Saclay, INRAE, CNRS, AgroParisTech, GQE- Le Moulon, 91190, Gif-sur-Yvette, France

About Agnès Doligez

Agnès Doligez currently works at the Amélioration Génétique et Adaptation des Plantes méditerranéennes et tropicales (AGAP), French National Institute for Agricultural Research (INRA). Agnès does research in Agricultural Plant Science and Genetics. Her current project is 'Development of genetic resources and tools to select new cultivars better coping with climate warming issues'.