3 Populations and Pathways
85
Nevertheless, when grouping populations by type of habitats, significant differences
were found regarding genetic and allelic diversities. They also found that tandem
repeat markers in ESTs did not deviate more frequently from neutral expectations
than anonymous genomic microsatellite loci. This could be due to the unbalanced
number of markers of each type in this particular study, but similar results have
been reported in other organisms (e.g. Woodhead et al. 2005). This result can be
explained by the tiny fraction of EST-SSR markers (more generally gene-linked)
that are influenced by selection. Indeed, Vasemägi et al. (2005) finally identified
only nine putative EST-SSRs (12%) that were potentially under selective pressure.
Moen et al. (2008) carried out a similar study on cod using SNPs markers.
However, the cod study did not include other types of markers to survey gene or
allele diversities and levels of population differentiation. In cod, forty-eight SNPs
out of a total of 318 (≈15%) had levels of genetic differentiation significantly different from zero, and twenty-nine were found to be potentially selected (≈9%). Of
the latter, seventeen were associated with genes of known function.
Together these two studies indicate that the percentage of markers under selection is >9% for studies based mainly on EST-SSRs or SNPs. This figure can be
compared to estimates obtained in the few genome scans that have been carried out
using AFLP markers. These reported a range of ≈1.5–12.5% of loci that could have
been or were influenced by selective constraints, and subsequently conferred or are
actually prone to confer a potential fitness gain within a given environment (Wilding
et al. 2001, Campbell and Bernatchez 2004, Bonin et al. 2006, Gruenthal and Burton
2008). The highest value reported for AFLPs (12.5%, 49 loci of a total of 392) is
debatable as indicated by Bonin et al. (2006), as it depends on the statistical method
used to detect loci under selection, and on the data sampling design (i.e. the geographical scale and/or the main ecological factor considered in the hierarchical data
analysis). For AFLPs, this number is probably overestimated and the correct value in
this example is more likely ≈2% (Bonin et al. 2006). More studies are needed so that
rigorous comparisons of marker types can be carried out, but current data indicates
that coding sequences involved in EST-SSRs or gene-based SNPs are (somewhat
logically) better sources of adaptive polymorphisms. This number probably does
not exceed 15%, indicating that a huge effort is required when establishing ESTlibraries to recover a “significant” number of potentially selected genes. However,
this does not negate EST libraries as a substantial resource for mining adaptive traits,
particularly because they are usually in the public domain and available for anyone
to exploit.
To date, very few population studies have been carried out to look at genetic
variation in aquatic organisms based solely on EST-SSR markers, but the study of
Vasemägi et al. (2005) does reveal some pitfalls to such an approach in non-model
organisms. For example, in this study, a large number of ESTs did not match any
known gene in the databases. The percentage of “unknown” genes has been estimated to be as high as 70% in the European sea bass (Dicentrarchus labrax (Boutet
et al. 2006, Chini et al. 2006) or the polychaete Alvinella pompejana (Alvinella
Consortium) and as low as ≈13% in halibut (Douglas et al. 2007). Hence, obtaining EST-SSRs by random EST database mining for any non-model species could
85
Nevertheless, when grouping populations by type of habitats, significant differences
were found regarding genetic and allelic diversities. They also found that tandem
repeat markers in ESTs did not deviate more frequently from neutral expectations
than anonymous genomic microsatellite loci. This could be due to the unbalanced
number of markers of each type in this particular study, but similar results have
been reported in other organisms (e.g. Woodhead et al. 2005). This result can be
explained by the tiny fraction of EST-SSR markers (more generally gene-linked)
that are influenced by selection. Indeed, Vasemägi et al. (2005) finally identified
only nine putative EST-SSRs (12%) that were potentially under selective pressure.
Moen et al. (2008) carried out a similar study on cod using SNPs markers.
However, the cod study did not include other types of markers to survey gene or
allele diversities and levels of population differentiation. In cod, forty-eight SNPs
out of a total of 318 (≈15%) had levels of genetic differentiation significantly different from zero, and twenty-nine were found to be potentially selected (≈9%). Of
the latter, seventeen were associated with genes of known function.
Together these two studies indicate that the percentage of markers under selection is >9% for studies based mainly on EST-SSRs or SNPs. This figure can be
compared to estimates obtained in the few genome scans that have been carried out
using AFLP markers. These reported a range of ≈1.5–12.5% of loci that could have
been or were influenced by selective constraints, and subsequently conferred or are
actually prone to confer a potential fitness gain within a given environment (Wilding
et al. 2001, Campbell and Bernatchez 2004, Bonin et al. 2006, Gruenthal and Burton
2008). The highest value reported for AFLPs (12.5%, 49 loci of a total of 392) is
debatable as indicated by Bonin et al. (2006), as it depends on the statistical method
used to detect loci under selection, and on the data sampling design (i.e. the geographical scale and/or the main ecological factor considered in the hierarchical data
analysis). For AFLPs, this number is probably overestimated and the correct value in
this example is more likely ≈2% (Bonin et al. 2006). More studies are needed so that
rigorous comparisons of marker types can be carried out, but current data indicates
that coding sequences involved in EST-SSRs or gene-based SNPs are (somewhat
logically) better sources of adaptive polymorphisms. This number probably does
not exceed 15%, indicating that a huge effort is required when establishing ESTlibraries to recover a “significant” number of potentially selected genes. However,
this does not negate EST libraries as a substantial resource for mining adaptive traits,
particularly because they are usually in the public domain and available for anyone
to exploit.
To date, very few population studies have been carried out to look at genetic
variation in aquatic organisms based solely on EST-SSR markers, but the study of
Vasemägi et al. (2005) does reveal some pitfalls to such an approach in non-model
organisms. For example, in this study, a large number of ESTs did not match any
known gene in the databases. The percentage of “unknown” genes has been estimated to be as high as 70% in the European sea bass (Dicentrarchus labrax (Boutet
et al. 2006, Chini et al. 2006) or the polychaete Alvinella pompejana (Alvinella
Consortium) and as low as ≈13% in halibut (Douglas et al. 2007). Hence, obtaining EST-SSRs by random EST database mining for any non-model species could