ORIGINAL PAPER Genome-wide association mapping of agronomic traits in sugar beet Tobias Wu ¨ rschum • Hans Peter Maurer • Thomas Kraft • Geert Janssen • Carolina Nilsson • Jochen Christoph Reif Received: 3 March 2011 / Accepted: 28 June 2011 / Published online: 15 July 2011 Ó Springer-Verlag 2011 Abstract Recent results indicate that association map- ping in populations from applied plant breeding is a pow- erful tool to detect QTL which are of direct relevance for breeding. The focus of this study was to unravel the genetic architecture of six agronomic traits in sugar beet. To this end, we employed an association mapping approach, based on a very large population of 924 elite sugar beet lines from applied plant breeding, fingerprinted with 677 single nucleotide polymorphism (SNP) markers covering the entire genome. We show that in this population linkage disequilibrium decays within a short genetic distance and is sufficient for the detection of QTL with a large effect size. To increase the QTL detection power and the mapping resolution a much higher number of SNPs is required. We found that for QTL detection, the mixed model including only the kinship matrix performed best, even in the pres- ence of a considerable population structure. In genome- wide scans, main effect QTL and epistatic QTL were detected for all six traits. Our full two-dimensional epis- tasis scan revealed that for complex traits there appear to be epistatic master regulators, loci which are involved in a large number of epistatic interactions throughout the genome. Introduction Association mapping is emerging as a novel tool in plant genomics (Myles et al. 2009) and has recently been shown to be suited for the analysis of populations from applied breeding programs (e.g. Reif et al. 2011a; Wu ¨rschum et al. 2011). The detection of QTL in these populations offers the advantage that QTL are identified which are of direct rel- evance for breeding. A potential problem of association mapping in plant populations is the inherent population structure. Any nonfunctional associations between the trait and the underlying population structure will also be detected as QTL (Zhao et al. 2007). Two methods have recently been suggested to correct for population stratifi- cation (P matrix) and familial relatedness (K matrix) (Price et al. 2006; Yu et al. 2006; Astle and Balding 2009; Cockram et al. 2010). As both, the K and the P matrix, are estimated based on marker data, their simultaneous use may result in overcorrection for population structure leading to a reduced power to detect QTL. Using experi- mental data in sugar beet, it has recently been shown that in breeding populations without a major population structure, the K matrix is sufficient to control the genetic background (Wu ¨rschum et al. 2011). Association mapping is based on linkage disequilibrium (LD) between the examined molecular markers and QTL associated with the trait. The mapping resolution of asso- ciation mapping is expected to be higher than in classical linkage mapping as it exploits all the recombination events that have occurred during the history of the population. LD is population specific and affected by many genetic factors (Flint-Garcia et al. 2003). Moreover, LD is highly variable across the genome. The power to detect QTL depends on the strength of the LD between the marker and the QTL. High r 2 values are required to detect medium and small Communicated by I. Mackay. Electronic supplementary material The online version of this article (doi:10.1007/s00122-011-1653-1) contains supplementary material, which is available to authorized users. T. Wu ¨rschum (&) Á H. P. Maurer Á J. C. Reif State Plant Breeding Institute, University of Hohenheim, 70593 Stuttgart, Germany e-mail: tobias.wuerschum@uni-hohenheim.de T. Kraft Á G. Janssen Á C. Nilsson Syngenta Seeds AB, Box 302, 261-23 Landskrona, Sweden 123 Theor Appl Genet (2011) 123:1121–1131 DOI 10.1007/s00122-011-1653-1