Theoretical and Applied Genetics

, Volume 127, Issue 12, pp 2619–2633

Applying association mapping and genomic selection to the dissection of key traits in elite European wheat

  • Alison R. Bentley
  • Marco Scutari
  • Nicolas Gosman
  • Sebastien Faure
  • Felicity Bedford
  • Phil Howell
  • James Cockram
  • Gemma A. Rose
  • Tobias Barber
  • Jose Irigoyen
  • Richard Horsnell
  • Claire Pumfrey
  • Emma Winnie
  • Johannes Schacht
  • Katia Beauchêne
  • Sebastien Praud
  • Andy Greenland
  • David Balding
  • Ian J. Mackay
Original Paper

DOI: 10.1007/s00122-014-2403-y

Cite this article as:
Bentley, A.R., Scutari, M., Gosman, N. et al. Theor Appl Genet (2014) 127: 2619. doi:10.1007/s00122-014-2403-y

Abstract

Key message

We show the application of association mapping and genomic selection for key breeding targets using a large panel of elite winter wheat varieties and a large volume of agronomic data.

Abstract

The heightening urgency to increase wheat production in line with the needs of a growing population, and in the face of climatic uncertainty, mean new approaches, including association mapping (AM) and genomic selection (GS) need to be validated and applied in wheat breeding. Key adaptive responses are the cornerstone of regional breeding. There is evidence that new ideotypes for long-standing traits such as flowering time may be required. In order to detect targets for future marker-assisted improvement and validate the practical application of GS for wheat breeding we genotyped 376 elite wheat varieties with 3,046 DArT, single nucleotide polymorphism and gene markers and measured seven traits in replicated yield trials over 2 years in France, Germany and the UK. The scale of the phenotyping exceeds the breadth of previous AM and GS studies in these key economic wheat production regions of Northern Europe. Mixed-linear modelling (MLM) detected significant marker-trait associations across and within regions. Genomic prediction using elastic net gave low to high prediction accuracies depending on the trait, and could be experimentally increased by modifying the constituents of the training population (TP). We also tested the use of differentially penalised regression to integrate candidate gene and genome-wide markers to predict traits, demonstrating the validity and simplicity of this approach. Overall, our results suggest that whilst AM offers potential for application in both research and breeding, GS represents an exciting opportunity to select key traits, and that optimisation of the TP is crucial to its successful implementation.

Supplementary material

122_2014_2403_MOESM1_ESM.docx (807 kb)
Supplementary material 1 (DOCX 807 kb)

Copyright information

© Springer-Verlag Berlin Heidelberg 2014

Authors and Affiliations

  • Alison R. Bentley
    • 1
  • Marco Scutari
    • 2
  • Nicolas Gosman
    • 1
  • Sebastien Faure
    • 3
  • Felicity Bedford
    • 1
  • Phil Howell
    • 1
  • James Cockram
    • 1
  • Gemma A. Rose
    • 1
  • Tobias Barber
    • 1
  • Jose Irigoyen
    • 1
  • Richard Horsnell
    • 1
  • Claire Pumfrey
    • 1
  • Emma Winnie
    • 1
  • Johannes Schacht
    • 4
  • Katia Beauchêne
    • 5
  • Sebastien Praud
    • 3
  • Andy Greenland
    • 1
  • David Balding
    • 2
  • Ian J. Mackay
    • 1
  1. 1.The John Bingham Laboratory, NIABCambridgeUK
  2. 2.Genetics InstituteUniversity College LondonLondonUK
  3. 3.BiogemmaChappesFrance
  4. 4.Limagrain GmbHPeine-RosenthalGermany
  5. 5.ARVALIS-Institut du végétalOuzouer Le MarchéFrance