The heterogeneous levels of linkage disequilibrium in white spruce genes and comparative analysis with other conifers

Abstract

In plants, knowledge about linkage disequilibrium (LD) is relevant for the design of efficient single-nucleotide polymorphism arrays in relation to their use in population and association genomics studies. Previous studies of conifer genes have shown LD to decay rapidly within gene limits, but exceptions have been reported. To evaluate the extent of heterogeneity of LD among conifer genes and its potential causes, we examined LD in 105 genes of white spruce (Picea glauca) by sequencing a panel of 48 haploid megagametophytes from natural populations and further compared it with LD in other conifer species. The average pairwise r(2) value was 0.19 (s.d.=0.19), and LD dropped quickly with a half-decay being reached at a distance of 65 nucleotides between sites. However, LD was significantly heterogeneous among genes. A first group of 29 genes had stronger LD (mean r(2)=0.28), and a second group of 38 genes had weaker LD (mean r(2)=0.12). While a strong relationship was found with the recombination rate, there was no obvious relationship between LD and functional classification. The level of nucleotide diversity, which was highly heterogeneous across genes, was also not significantly correlated with LD. A search for selection signatures highlighted significant deviations from the standard neutral model, which could be mostly attributed to recent demographic changes. Little evidence was seen for hitchhiking and clear relationships with LD. When compared among conifer species, on average, levels of LD were similar in genes from white spruce, Norway spruce and Scots pine, whereas loblolly pine and Douglas fir genes exhibited a significantly higher LD.

Figure 1

Linkage disequilibrium as a function of distance for white spruce genes. x axis is the distance in nucleotides; y axis is the correlation coefficient (r²) between nucleotide sites. Only informative sites were considered. In red, the curve obtained for the merged set of 105 genes. (a) Curves obtained for the 67 genes with at least 15 pairwise site comparisons; (b) in blue, curves obtained for members of the myb R2R3 gene family; (c) in magenta, curves obtained for members of the wrky gene family; (d) in green, curves obtained for members of the knox-I gene family.

Figure 2

Plot of the mean r² values obtained for gene data sets from five conifer species. Curves were estimated from the model of Hill and Weir (1988) from 105 genes of Picea glauca (this study), 22 genes of Picea abies (Heuertz et al., 2006), 19 genes of Pinus taeda (González-Martínez et al., 2006b), 16 genes of Pinus sylvestris (Pyhäjärvi et al., 2007) and 121 genes of Pseudotsuga menziesii var. menziesii (Eckert et al., 2009a).

Figure 3

Linkage disequilibrium and nucleotide diversity in genes sequenced in four conifer species. Mean r² values are plotted along the y axis and nucleotide diversity π values are plotted on the x axis. Data were calculated for genes exhibiting at least 15 pairwise site comparisons.

Figure 4

Plot of the values of Tajima's D and Fay and Wu's H for 52 white spruce genes with appropriate outgroup sequence. The plotted genes belonged to one of two groups of genes with a mean r² above (high LD series in blue) or under (low LD series in red) the overall mean.