As recent studies have shown (see Highlights, November 2000), an important issue for those mapping human disease genes is how and why linkage disequilibrium (LD) levels vary across the genome. Often such studies have looked at only a few loci. Now a genome-wide study of LD has found that long-range LD is a feature of European genomes but not of some African ones. These results illustrate how a population's history can determine its usefulness for disease-mapping studies and indicate that northern European populations experienced a severe bottleneck — perhaps during the emergence of modern humans out of Africa.

In this study, Reich and colleagues documented the extent of LD around 19 high-frequency coding single nucleotide polymorphisms (cSNPs) selected from a multi-ethnic DNA panel. They chose high-frequency SNPs for several reasons but, most importantly, because they represent the 'worst-case' scenario — the extent of LD is usually less around common SNPs than around rare, younger SNPs, as haplotypes with common SNPs have often been broken down over time by recombination.

Reich and colleagues identified SNPs at various distances away from the core cSNP in 44 unrelated individuals of N. European descent and measured the distance at which LD drops away (the half length). By comparing the 19 regions, Reich et al. found that LD has an average half length of 60 kb, indicating the presence of blocks of LD, larger than previously estimated. However, they also found great variation in LD between different genomic regions. For example, LD extends for 155 kb around the WASL gene, but for only 6 kb around PCI . These regional variations emphasize the value of a global view.

So why do large blocks of LD exist? The simplest explanation is that an extreme population bottleneck occurred — according to the authors' simulations — ∼27,000–53,000 years ago. But could differing recombination rates also account for this pattern? If so, these regions should show long-range LD in all populations. To assess this, the authors analysed LD in 48 Swedes and 96 Yorubans from Nigeria — the Swedes had an identical pattern of long-range LD but in the Yorubans the half length was dramatically reduced to ∼5 kb. These results strongly indicate that N. European populations indeed underwent a bottleneck after their divergence from Africans — one of such enormous proportions that only a handful of chromosomes gave rise to most of the N. European gene pool. This population contraction might have occurred during the migration from Africa, or perhaps more recently during the colonization of Europe or the last Ice Age.

The authors advocate that geneticists should make good use of these differing LD rates between populations for finding disease genes — their long-range LD makes European populations ideal for gene-mapping studies but not for finer mapping, which could be done in populations with less extensive LD, such as the Yorubans.