3.3 Genome wide association study (GWAS) for thermal plasticity of the circadian clock output
Next, we used the SensyPAM platform to investigate the genetic basis for the period and amplitude of circadian photosynthetic rhythms and their thermal plasticity (Table S3 ). To this end we selected 285 wild barley from the B1K collection (Hubner et al. , 2009) that originate from different geographic habitats. The rationale was to try to include single representative from each of the 255 micro-site of this hierarchical collection (51 sites including each 5 micro-sites with 4 representatives). We performed a GWAS using the traits values per se and by applying a QTL by Environment (Q x E) model for the OT and HT environments (Yamamoto & Matsunaga, 2021). Moreover, since the analysis of RH (this study) and the ASHER populations (Bdolach et al.,2019) indicated a significant effect of the plasmotype diversity on the photosynthetic rhythm plasticity, we included sequencing information from isolated chloroplast DNA for a portion of this panel (see Methods; Table S4 ).
We wanted to identify nuclear loci for which allelic diversity is associated with plasticity of photosynthetic rhythms. First, we increased the current coverage of B1K genomic analysis by harvesting genic and intergenic DNA variation from several sources (see Supporting Information and Methods). Prior to GWAS analysis, missing values were imputed with the ”missForest” algorithm and filtered for markers with missing data >0.2, minor allele frequency < 0.03, monomorphic and multiallelic markers. The selection process led to a final set of 13,786 informative markers for GWAS analysis (TableS5 ). We wished to detect QTLs with persistent effects across the two environments (OT and HT) but also with Q × E effects, i.e., loci with specific effects to a certain environment ( Malosetti et al.2013; Yamamoto & Matsunaga, 2021).
Briefly, the output of the analysis for each SNP included “Additive Main Effect” (p.ame ), “All SNP effects” (p.all ), “Interaction terms” (p.int ) and a Wald score for each environment indicated the environment with the most significant effect of the locus on the trait (Table S6 ; Figure 3 ). In a previous study, we identified several DOCs loci that modulated the circadian clock output in the HEB-25 interspecific mapping population (Prusty et al., 2021). Interestingly, some of theseDOC loci are overlapping with signals in this current genome scan of the B1K, including some that contain genes reported to be involved in the circadian clock. Of note is the gene GIGANTEA (GI ) that resides in the long arm of chromosome 3. Previously, we identified a large interval on chromosome 3, DOC3.2 , associated with significant pleiotropic effects on the clock period and growth in the field. Nevertheless, DOC3.2 , although harboring the barley ortholog of the GI gene (Fowler et al., 1999), stretched in the previous HEB population analysis to a distance of 45.98 Mbp (Chr3H_35066186 – Chr3H_81047480). Here, in the GWAS in the B1K collection, we identified SNPs around the GI gene that were significantly associated with the amplitude of the clock and its plasticity (Figure 3a ). In fact, the current SNP arrangement point to a causal variation in 150 kb downstream to the GI gene. Phenotypic analysis showed that the mild non-significant difference in the amplitude between two allele under OT became more pronounced and significant under HT condition (Figure S4 ).