Genomic determinants, architecture, and constraints in drought-related traits in Corymbia calophylla

Abstract

Background: Drought adaptation is critical to many tree species persisting under climate change, however our knowledge of the genetic basis for trees to adapt to drought is limited. This knowledge gap impedes our fundamental understanding of drought response and application to forest production and conservation. To improve our understanding of the genomic determinants, architecture, and trait constraints, we assembled a reference genome and detected ~ 6.5 M variants in 432 phenotyped individuals for the foundational tree Corymbia calophylla.

Results: We found 273 genomic variants determining traits with moderate heritability (h²_SNP = 0.26-0.64). Significant variants were predominantly in gene regulatory elements distributed among several haplotype blocks across all chromosomes. Furthermore, traits were constrained by frequent epistatic and pleiotropic interactions.

Conclusions: Our results on the genetic basis for drought traits in Corymbia calophylla have several implications for the ability to adapt to climate change: (1) drought related traits are controlled by complex genomic architectures with large haplotypes, epistatic, and pleiotropic interactions; (2) the most significant variants determining drought related traits occurred in regulatory regions; and (3) models incorporating epistatic interactions increase trait predictions. Our findings indicate that despite moderate heritability drought traits are likely constrained by complex genomic architecture potentially limiting trees response to climate change.

Keywords: Epistasis; Eucalyptus; Genome wide association study (GWAS); Heritability; Pleiotropy; Water use efficiency.

Fig. 1

Phenotypic traits were studied in a common garden of populations sampled from across the range of Corymbia calophylla, denoted here by the precipitation layer. a Location of populations sampled and the experimental site mapped with precipitation of the driest month (P_DM; mm; BIO14); b-d trait values (grey) with their best linear unbiased predictions (BLUPs; yellow) for (b) δ¹³C, (c) SLA, and (d) NDVI. Population colours are coded red for northern populations and blue for southern populations and ordered from wettest (left) to driest (right). Star represents the location of the experimental site. Inset shows location of study area within Australia. NDVI was scaled (y2-axis; yellow) to meet assumptions of normality before estimating BLUPs

Fig. 2

Genome sampling and GWAS outputs for Corymbia calophylla. a Manhattan plots for three traits and SNP density. Points represent SNPs significantly associated with the trait (red = δ¹³C; yellow = NDVI; blue = SLA; grey = not significant) at an FDR value < 0.00001. The density plot (below Manhattan plot) shows the number of SNPs in 1 million base pair segments across the genome with colour (white to green). b Magnified view of the significant peaks within the three ‘hotspot’ regions of adaptive variation. Underneath the magnified views are haploview plots that detect significant blocks, bounded by black lines. The linkage disequilibrium (r.²) across the haploview plot is denoted from low (white) to high (red)

Fig. 3

Patterns of significant epistatic and pleiotropic interactions in Corymbia calophylla. a Epistatic interactions are shown with coloured arrows and pleiotropic effects between traits are shown in the circular bands. Chromosomes are in black; chromosome six is not present because no SNPs were significant in the analysis. The direction of influence is shown by colour, where orange indicates that the SNP affects a different SNP in a positive way and blue is indicative of a negative effect. Interactions between a SNP and multiple traits indicate pleiotropy, while the same colours are indicative of the same effects. Antagonistic pleiotropy is inferred if the colours are different among SNPs in the same chromosomal location. Points on chromosome 3, 8, and 10 have been manually separated due to severe overlapping to visualise the antagonistic effects. b Genotypes for negatively influenced epistatic interaction between two SNP variants (grey points in (b) & blue arrows in (a)) and positively influenced epistatic interactions between two SNP variants (orange points in (b) & orange arrows in (a)). c Visualisation of one significant epistatic interaction where the main effect of a SNP (grey in (c)) and trait is conditioned on a second SNP (yellow in (c)), black dashed line is the interaction effect between the two variants. P-value – *** < 0.001; * < 0.05