Prospects of applying genomic selection in natural populations of Populus trichocarpa.

An emerging pattern from phenotype-genotype association studies in forest trees is that the genomic architectures of most traits of interest for breeding and conservation are highly complex. The prediction of genetic or breeding values from molecular marker data is therefore likely to require high-density genome coverage. Until recently, this approach, known as genomic selection, was only feasible in synthetic populations characterized by very extensive linkage disequilibrium (i.e., allowing prediction based on up to several thousand markers). However, advances in sequencing and genotyping technology are rapidly changing this, potentially making genomic selection much more widely applicable. We illustrate this concept using whole-genome re-sequencing data for 289 Populus trichocarpa trees, which have also been extensively phenotyped as part of ongoing association studies, and discuss the implications of our findings.