Multi-locus genome-wide association studies reveal novel alleles for flowering time under vernalisation and extended photoperiod in a barley MAGIC population

Abstract

Key genes controlling flowering and interactions of different photoperiod alleles with various environments were identified in a barley MAGIC population. A new candidate gene for vernalisation requirements was also detected. Optimal flowering time has a major impact on grain yield in crop species, including the globally important temperate cereal crop barley (Hordeum vulgare L.). Understanding the genetics of flowering is a key avenue to enhancing yield potential. Although bi-parental populations were used intensively to map genes controlling flowering, their lack of genetic diversity requires additional work to obtain desired gene combinations in the selected lines, especially when the two parental cultivars did not carry the genes. Multi-parent mapping populations, which use a combination of four or eight parental cultivars, have higher genetic and phenotypic diversity and can provide novel genetic combinations that cannot be achieved using bi-parental populations. This study uses a Multi-parent advanced generation intercross (MAGIC) population from four commercial barley cultivars to identify genes controlling flowering time in different environmental conditions. Genome-wide association studies (GWAS) were performed using 5,112 high-quality markers from Diversity Arrays Technology sequencing (DArT-seq), and Kompetitive allele-specific polymerase chain reaction (KASP) genetic markers were developed. Phenotypic data were collected from fifteen different field trials for three consecutive years. Planting was conducted at various sowing times, and plants were grown with/without additional vernalisation and extended photoperiod treatments. This study detected fourteen stable regions associated with flowering time across multiple environments. GWAS combined with pangenome data highlighted the role of CEN gene in flowering and enabled the prediction of different CEN alleles from parental lines. As the founder lines of the multi-parental population are elite germplasm, the favourable alleles identified in this study are directly relevant to breeding, increasing the efficiency of subsequent breeding strategies and offering better grain yield and adaptation to growing conditions.

Fig. 1

Scatter plot comparing the flowering time between vernalised, extended photoperiod and normal conditions. Each box represents a different trial conducted in Perth in 2019 with sowing time in a April, b May, c June, d July and e August; and comparing f extended and regular photoperiod. Dots depicted the date to awn appearance (Z49) of individual plants

Fig. 2

Population structure analysis result. a Ancestry plot of the population structure data from Admixture 1.3.0 12 clusters (K = 12) and b average cross-validation error of different K values from 1 to 20 from population structure analysis using Admixture 1.3.0

Fig. 3

Genome-wide distribution of significant QTNs detected by GWAS. a The outermost ring with the scale represents the 7 barley chromosomes. b The colour lines in the inside ring represent the marker positions. The blue scatters represent the positions of significant QTNs in normal conditions, with each ring representing different trials conducted in c Corrigin 2018, d Esperance 2018, e Perth 2017, f Perth 2018, different sowing times in Perth in g April, h May, i June, j July, k August in 2019, respectively. The yellow scatters represent the positions of significant QTNs in vernalised trials, with each ring represent vernalised trials with different sowing times in Perth in l April, m May, n June, o July, p August in 2019, respectively. q The red scatters and yellow ring represent the positions of significant QTNs in extended photoperiod trial. The green scatters represent the positions of significant QTNs in vernalised/non-vernalised comparison, with each ring representing the vernalised/non-vernalised comparison for different sowing times in r April, s May, t June, u July, v August in 2019, respectively

Fig. 4

Boxplots for flowering time of different D2H519658782_AC QTN in different trials. a–e Non-vernalised trials with different sowing times from April to August in Perth 2019. f–j) Trials with vernalisation were sown from April to August in Perth 2019. k–m Trials at different locations in Corrigin, Esperance and Perth 2018. n Extended photoperiod trial. The population was divided into two groups according to allele types in each trial. The X-axis represents the two alleles, while the Y-axis corresponds to flowering time (Z49). Student’s t-test results shown for comparison: **p-value < 0.01, ***p-value < 0.001

Fig. 5

Sequence analysis of four parental cultivars Compass, GrangeR, La Trobe and Lockyer with 20 barley accessions from the pan-genome dataset. a Phylogenetic tree based on 13 genetic variants in the CEN gene region. b A C-to-G SNP caused an amino acid change from Proline to Alanine in cvs. Barke, RGT Planet, and Golden Promise

Fig. 6

Boxplots for flowering time of different D2H29454480_AC QTN in different trials. a–e Non-vernalisation trials with different sowing times from April to August in Perth 2019. f–j Trials with vernalisation were sown from April to August in Perth 2019. k–m Trials at different locations in Corrigin, Esperance and Perth 2018. n Extended photoperiod trial. The population was divided into two groups according to allele types in each trial. The X-axis represents the two alleles, while the Y-axis corresponds to flowering time (Z49). Student’s t-test results shown for comparison: ns: not significant at p-value = 0.05, *** p-value < 0.001