Biomass traits and candidate genes for bioenergy revealed through association genetics in coppiced European Populus nigra (L.)

Abstract

Background: Second generation (2G) bioenergy from lignocellulosic feedstocks has the potential to develop as a sustainable source of renewable energy; however, significant hurdles still remain for large-scale commercialisation. Populus is considered as a promising 2G feedstock and understanding the genetic basis of biomass yield and feedstock quality are a research priority in this model tree species.

Results: We report the first coppiced biomass study for 714 members of a wide population of European black poplar (Populus nigra L.), a native European tree, selected from 20 river populations ranging in latitude and longitude between 40.5 and 52.1°N and 1.0 and 16.4°E, respectively. When grown at a single site in southern UK, significant Site of Origin (SO) effects were seen for 14 of the 15 directly measured or derived traits including biomass yield, leaf area and stomatal index. There was significant correlation (p < 0.001) between biomass yield traits over 3 years of harvest which identified leaf size and cell production as strong predictors of biomass yield. A 12 K Illumina genotyping array (constructed from 10,331 SNPs in 14 QTL regions and 4648 genes) highlighted significant population genetic structure with pairwise FST showing strong differentiation (p < 0.001) between the Spanish and Italian subpopulations. Robust associations reaching genome-wide significance are reported for main stem height and cell number per leaf; two traits tightly linked to biomass yield. These genotyping and phenotypic data were also used to show the presence of significant isolation by distance (IBD) and isolation by adaption (IBA) within this population.

Conclusions: The three associations identified reaching genome-wide significance at p < 0.05 include a transcription factor; a putative stress response gene and a gene of unknown function. None of them have been previously linked to bioenergy yield; were shown to be differentially expressed in a panel of three selected genotypes from the collection and represent exciting, novel candidates for further study in a bioenergy tree native to Europe and Euro-Asia. A further 26 markers (22 genes) were found to reach putative significance and are also of interest for biomass yield, leaf area, epidermal cell expansion and stomatal patterning. This research on European P. nigra provides an important foundation for the development of commercial native trees for bioenergy and for advanced, molecular breeding in these species.

Keywords: Genetics; Leaf area; Lignocellulosic; Salicaceae; Short rotation coppice (SRC); Yield.

Fig. 1

Map illustrating the nations and major river locations from which the P. nigra association population is drawn and the colours employed to illustrate these nations in subsequent figures

Fig. 2

Histograms illustrate trait frequency distribution following Box-Cox transformation for a Estimated biomass yield 2013 (EB-13); b Epidermal cell number per leaf (CNPL-13) and c Saccharification potential (glucose yield) 2012 (SP-12)

Fig. 3

Box plots depict range, interquartile range, median and mean (cross) for a Epidermal cell number per leaf 2013 (CNPL-13); b Estimated oven-dry biomass yield 2013 (EB-13); c Saccharification potential (glucose release) 2012 (SP-12); d Leaf area 2013 (LA-13); e Epidermal cell area 2013 (CA-13); f Stomatal index 2013 (SI-13)

Fig. 4

Pairwise trait correlations are visualised with line colours and widths conferred according to the strength and direction of Pearson’s correlation coefficient (r) between trait pairs. Non-significant correlations are depicted with grey, point 1 lines. Significant positive and negative correlations (p < 0.05) are depicted with point 2 lines coloured light green or light red, respectively. Strong positive and negative correlations (r > 0.5) are depicted with point 3 lines coloured dark green or dark red, respectively. Very strong positive correlations (r > 0.8) are also shown in dark green with point 4 lines

Fig. 5

Trait heritabilities show significant positive regression with their correlation coefficients (r) for a longitude of origin (r ² = 0.643) but not b latitude of origin (r ² = 0.112)

Fig. 6

Satellite map of P. nigra association population subpopulation locations and their mean proportional cluster allocations from STRUCTURE for a K = 2 and b K = 7

Fig. 7

Genetic distance matrix (pairwise F_ST) between 20 subpopulations of P. nigra association population. F_ST values are shaded according to magnitude (white to dark grey) with Italian subpopulations in purple; French in orange; Spanish in red; German in blue and Netherlands in green

Fig. 8

QQ and Manhattan plots for the Q + K (optimal) models for the 3 traits with SNPs reaching genome-wide significance. Red and blue lines on Manhattan plots illustrate genome wide (α < 8.79 × 10⁻⁶) and putative (α < 1.76 × 10⁻⁴) significance levels, respectively. a QQ plot for Height-11 associated SNP on chromosome 7; b Manhattan plot for Height-11 association; c QQ plot for Height-13 associated SNP on chromosome 4; d Manhattan plot for Height-13 associated SNP; e QQ plot for CNPL-13 associated SNP on chromosome 13; f Manhattan plot for CNPL-13 associated SNP

Fig. 9

Bar plots of raw effects sizes (with standard error bars) for each trait-associated SNP with genome-wide significance from trait-specific optimal model for a Height-11 associated SNP; b Height-13 associated SNP and c CNPL-13 associated SNP. The x-axis of each plot gives the identity of each allelic variant (MM, MN or NN) with its sample size (n) within the population given in adjacent brackets