As nicely conveyed by Stinchcombe & Hoekstra (2008), ‘population genomics is simply population genetics writ large — that is, population genetic analyses of a large number of loci, distributed throughout the genome’. This transition from ‘genetics’ to ‘genomics’ sprang from the recent breakthrough in genotyping throughputs and data analysis capacities, and it quickly found a particular echo in evolutionary studies of adaptation. Information about many independent loci indeed enables to statistically detect those displaying atypical patterns of diversity compared to the rest of the genome. These ‘outlier’ loci are of particular interest because they are expected to be influenced by locus-specific evolutionary forces like selection or, more likely, to be linked to a locus experiencing such forces (Luikart et al. 2003; Storz 2005).
As a result, the past few years have witnessed the multiplication of genome scans aiming at exploring adaptation or speciation in a variety of organisms, for example, walking-sticks (Nosil et al. 2008), beech (Jump et al. 2006), whitefish (Campbell & Bernatchez 2004) or oyster (Murray & Hare 2006). However, once outlier markers are revealed, population genomics alone is often helpless to discover the adaptive mutation(s) or gene(s), if any, driving outlier behaviour (but see Wood et al. 2008). A number of reasons can be blamed for this lack of success (see a recent review in Stinchcombe & Hoekstra 2008), but the most important one may be that most of the genome scans carried out so far look for a needle in a haystack. As a matter of fact, they generally rely on a priori neutral markers such as microsatellites, single nucleotide polymorphisms (SNPs) and amplified fragment length polymorphisms (AFLPs), which tend to fall in noncoding sequences of the genome (introns and intergenic regions). Yet genomes are big, and odds are low for a marker to land closely enough to an adaptive mutation or a gene to show a selection signature. This is especially true in natural populations where linkage disequilibrium tends to decay rapidly, therefore shortening the window of outlier behaviour around selected genes. Conversely, under the influence of particular mating systems, population histories or low recombination rates, linkage disequilibrium can sometimes span large genomic areas encompassing several genes. This in turn makes it difficult to link an outlier locus to a specific candidate gene.
One solution to boost the efficiency of population genomics would be to ‘filter out the hay’ from the outset, i.e. to directly target coding regions because they are more likely to be under selection. This is where expressed sequence tags (ESTs) promisingly get into the picture. Briefly, ESTs are short (~200–700 nucleotides) subsequences of transcribed and spliced DNA. They are generated by partially sequencing a pool of mRNA extracted from a specific tissue at a particular developmental stage of an organism (Bouck & Vision 2007). From a population genomics perspective, ESTs are a great genomic tool for several reasons. First and foremost perhaps, EST-based genome scans have the big advantage to focus on coding regions while avoiding wasting resource elsewhere in the genome. Second, EST sequences offer the raw material for population genomics in abundance: they are a great source of SNPs and microsatellites, and the primers used for their genotyping are usually highly transferable between closely related species. Third, EST sequences can be used as primer anchors to discover markers in flanking sequences, and thus allow exploring the vicinity of genes, i.e. regulatory motifs and intronic regions. Finally, even if EST-based markers still remain more expensive to develop than random markers like AFLPs, they are rather affordable and prices are likely to decrease in the future thanks to the new ultra-high throughput sequencing technologies (Schuster 2008).
In this issue of Molecular Ecology, Namroud et al. (2008) describe one of the first EST-based genome scans designed for tracking adaptive loci. More specifically, their aim was to identify candidate genes for local adaptation in six Canadian populations of white spruce (Picea glauca, Fig. 1) known to be moderately differentiated for traits pertaining to wood formation, phenology and growth. For this purpose, the authors examined 534 SNP loci associated with 345 unigenes from a white spruce EST library. They used two different statistical methods to reveal loci whose genetic differentiation was higher than expected under a neutral model of evolution. Overall, up to 9.2% of the markers could be seen as outliers; thus, 13.6% of the studied genes showed trends towards adaptation. Interestingly, patterns in trait differentiation as well as the genes’ putative functions brought independent and corroborative evidence of the adaptive status of these candidate genes.
Beyond the results specific to white spruce, Namroud et al.'s findings have broader implications that deserve to be emphasized here. First, they show the potential for EST-based genome scans to easily discover candidate genes for adaptation in a nonmodel species. Of course, it remains to be demonstrated that these genes are truly under selection. Yet, the consistency between genetic patterns, trait differentiation and gene functions allows some level of confidence in this assumption. Second, in this study, the authors were able to test SNPs situated in both translated and untranslated regions of genes. This illustrates the capacity of EST-based genome scans to provide some preliminary insight about the nature of adaptive mutations (e.g. exonic or regulatory). This is not trivial, as adaptation can be triggered by modifications in protein-coding sequences as well as gene expression changes (Whitehead & Crawford 2006). Third, it has to be noted that population genomics was here successfully applied to a model species a priori unbefitting to this type of approach. White spruce (Fig. 1) has indeed a large genome where linkage disequilibrium is usually limited.
Admittedly, there is still a long way to go to get a fine picture of the genetic basis of adaptation in white spruce. Further research will help establish, for example, if these genes display selection signatures at the nucleotide level, and/or if their expression patterns show deviations from neutral expectations in the examined populations. But this exciting article, as well as a few others (Vasemägi et al. 2005; Oetjen & Reusch 2007; Zayed & Whitfield 2008) should yet be credited for bridging the gap existing between population and functional genomics, and for paving the way for the next generation of genome scans investigating adaptation and speciation.