Plant height, which is an estimator of vegetative yield, and crown rust tolerance are major criteria for perennial ryegrass breeding. Genetic improvement has been achieved through phenotypic selection but it should be speeded up using marker-assisted selection, especially in this heterozygous species suffering from inbreeding depression. Using connected multiparental populations should increase the diversity studied and could substantially increase the power of quantitative trait loci (QTL) detection. The objective of this study was to detect the best alleles for plant height and rust tolerance among three connected populations derived from elite material by comparing an analysis per parent and a multipopulation connected analysis. For the studied traits, 17 QTL were detected with the analysis per parent while the additive and dominance models of the multipopulation connected analysis made it possible to detect 33 and 21 QTL, respectively. Favorable alleles have been detected in all parents. Only a few dominance effects were detected and they generally had lower values than the additive effects. The additive model of the multipopulation connected analysis was the most powerful as it made it possible to detect most of the QTL identified in the other analyses and 11 additional QTL. Using this model, plant growth QTL and rust tolerance QTL explained up to 19 and 38.6% of phenotypic variance, respectively. This example involving three connected populations is promising for an application on polycross progenies, traditionally used in breeding programs. Indeed, polycross progenies actually are a set of several connected populations.

Connected multiparental crosses are valuable for detecting quantitative trait loci (QTL) with multiple alleles. The objective of this study was to show that the progeny of a polycross can be considered as connected mutiparental crosses and used for QTL identification. This is particularly relevant in outbreeding species showing strong inbreeding depression and for which synthetic varieties are created. A total of 191 genotypes from a polycross with six parents were phenotyped for plant height (PH) and plant growth rate (PGR) and genotyped with 82 codominant markers. Markers allowed the identification of the male parent for each sibling and so the 191 genotypes were divided into 15 full-sib families. The number of genotypes per full-sib family varied from 2 to 28. A consensus map of 491 cM was built and QTL were detected with MCQTL-software dedicated to QTL detection in connected mapping populations. Two major QTL for PH and PGR in spring were identified on linkage groups 3 and 4. These explained from 12% to 22% of phenotypic variance. The additive effects reached 12.4 mm for PH and 0.11 mm/C • d for PGR. This study shows that the progenies of polycrosses can be used to detect QTL.

No abstract