Markers associated with heading and aftermath heading in perennial ryegrass full-sib families

Abstract

Background: Heading and aftermath heading are important traits in perennial ryegrass because they impact forage quality. So far, genome-wide association analyses in this major forage species have only identified a small number of genetic variants associated with heading date that overall explained little of the variation. Some possible reasons include rare alleles with large phenotypic affects, allelic heterogeneity, or insufficient marker density. We established a genome-wide association panel with multiple genotypes from multiple full-sib families. This ensured alleles were present at the frequency needed to have sufficient statistical power to identify associations. We genotyped the panel via partial genome sequencing and performed genome-wide association analyses with multi-year phenotype data collected for heading date, and aftermath heading.

Results: Genome wide association using a mixed linear model failed to identify any variants significantly associated with heading date or aftermath heading. Our failure to identify associations for these traits is likely due to the extremely low linkage disequilibrium we observed in this population. However, using single marker analysis within each full-sib family we could identify markers and genomic regions associated with heading and aftermath heading. Using the ryegrass genome we identified putative orthologs of key heading genes, some of which were located in regions of marker-trait associations.

Conclusion: Given the very low levels of LD, genome wide association studies in perennial ryegrass populations are going to require very high SNP densities. Single marker analysis within full-sibs enabled us to identify significant marker-trait associations. One of these markers anchored proximal to a putative ortholog of TFL1, homologues of which have been shown to play a key role in continuous heading of some members of the rose family, Rosaceae.

Keywords: Aftermath heading; Flowering; Genome wide association; Heading; Lolium perenne; Perennial ryegrass; Single marker analysis.

Fig. 1

Phenotypic distribution of heading date in six full-sib families. Boxplots representing heading date in full-sib families with y-axis showing days to heading and families on x-axis

Fig. 2

QQ-plots for a heading and b aftermath heading

Fig. 3

Extent of linkage disequilibrium (LD) measured as the squared correlation of allele counts (y-axis), based on the maximum likelihood solution to the cubic equation. The x-axis shows inter marker distance in bp. LD estimates were sorted according to inter-marker distance, and divided into bins of 1000 estimates. Each point on the plot represents the mean R ² and mean inter-marker distance of 1000 measurements

Fig. 4

Heatmap illustrates regions associated with heading over six full-sib families on perennial ryegrass LG2. A Kruskal-Wallis test was performed on each marker to identify significant regions for heading. Using the perennial ryegrass genome zipper [26, 30] we identified a putative gene order for markers on LG2. These data were used to construct the heatmap for each family. A perennial ryegrass transcriptome-based genetic linkage map upon which GenomeZipper was based was used as reference to construct LG2 [26, 27] and placed above the heatmap. Each bar in the heatmap represents region between two genetic markers from the linkage map. The median Kruskal-Wallis test statistic was calculated for markers binned between markers on the genetic linkage map and used to construct the heatmap. Putative orthologs of LpPRR37 and LpTFL1, were identified in the phylogenetic analysis and placed onto LG2 using genetic positions from genome zipper. The genetic positions of these orthologs were extrapolated over the heatmap. Color of the heatmap illustrates the test-statistic of the Kruskal-wallis analysis from 0 to 23

Fig. 5

Heatmap illustrates regions associated with heading over six full-sib families in perennial ryegrass LG4. A Kruskal-Wallis test was done on each marker to identify significant regions for heading. Using perennial ryegrass genome zipper [26, 30] putative gene order for markers on LG4 was identified. These data was used to construct the heatmap for each family. Ryegrass transcriptome linkage map upon which GenomeZipper was based is used as reference to construct LG4 [26, 27] and placed above the heatmap. Each bar in the heatmap represents region between two genetic markers from the linkage map. The median Kruskal-Wallis test statistic was calculated for bins represented by gaps between markers on the genetic linkage map and used to construct the heatmap. Putative orthologs of LpVRN1, LpPHYA, LpPHYB and LpPHYC, were identified in the phylogenetic analysis and placed onto LG4 using genetic positions from genome zipper. The genetic positions of these orthologs were extrapolated over the heatmap as bars. Color of the heatmap illustrates the test-statistic of the Kruskal-wallis analysis from 0 to 21

Fig. 6

Heatmap illustrates regions associated with heading over six full-sib families in perennial ryegrass LG7. A Kruskal-Wallis test was done on each marker to identify significant regions for heading. Using perennial ryegrass genome zipper [26, 30] putative gene order for markers on LG7 was identified. These data was used to construct the heatmap for each family. Ryegrass transcriptome linkage map upon which GenomeZipper was based is used as reference to construct LG7 [26, 27] and placed above the heatmap. Each bar in the heatmap represent region between two genetic markers from linkage map. The median Kruskal-Wallis test statistic was calculated for bins represented by gaps between markers on the genetic linkage map and used to construct the heatmap. Putative orthologs of LpCO and LpFT, were identified in the phylogenetic analysis and placed onto LG7 using genetic positions from genome zipper. The genetic positions of these orthologs were extrapolated over the heatmap. Color of the heatmap illustrates the test-statistic of the Kruskal-wallis analysis from 0 to 12

Fig. 7

Principal component analysis (PCA) of 360 perennial ryegrass individuals, genotyped using 51,846 SNPs. The first two principal components explained 14.8 % of total variation. Components are colored according to family (color coding is listed in figure legend)

Fig. 8

Schematic view of genetic pathway controlling heading. Genes promoting heading were shown by arrows and genes acting as repressor shown as lines with bars. External factors like day light and extended cold periods were represented with respective symbols. Pathways were mentioned in grey boxes and genes shown in red were considered as key regulators in heading

Fig. 9

Phylogenetic analysis of FT/TFL1 gene family using Arabidopsis FT as query. The evolutionary history was inferred by using the Maximum Likelihood method based on the JTT matrix-based model [67]. Bootstrap values after 100 replicates were shown next to the branches. The tree is mid-point rooted, drawn to scale, with branch lengths proportional to the number of substitutions per site. The analysis involved 90 amino acid sequences. All positions containing gaps and missing data were eliminated. There were a total of 83 positions in the final dataset. Evolutionary analyses were conducted in MEGA 6.06 [66]. All the associated Lolium proteins are in red and Arabidopsis proteins were highlighted