Genetic diversity is important for ecological and evolutionary processes from individual fitness to ecosystem function (Hughes et al. 2008; Reed 2008). Allelic richness and heterozygosity serve as indicators of evolutionary potential (Willi et al. 2006) and are important in determining population dynamics and population viability (Reed et al. 2007). However, populations often face at least moderate inbreeding due to habitat reduction/fragmentation, naturally induced bottlenecks, habitat specialization combined with poor dispersal capabilities, or during artificial selection for traits valuable to humans. Whether genetic variation is lost in accordance with neutral expectations, or whether selection can maintain genetic diversity in the face of random genetic drift, is a very important question. The amount of genetic diversity lost impacts, for example, the ability of a small number of propagules to colonize (or invade) a new habitat (Roman & Darling 2007), for small populations to recover after a natural or artificial bottleneck, and for plants and animals to continue to respond to selection despite the inbreeding imposed by strong selection (Kristensen & Sørensen 2005).
The question that Demontis et al. (2009) set out to answer is this: if inbreeding is inevitable, does the speed at which inbreeding occurs matter? This is tantamount to asking how important is selection in slowing the loss of genetic diversity, because selection will have very little opportunity to act in populations that are inbred extremely rapidly (e.g. full-sib mating). Thus, it has been theorized that the amount of genetic diversity lost during an inbreeding event should be influenced by the rapidity with which the inbreeding proceeds, because selection can act more efficiently when inbreeding occurs slowly. It is important to note that Demontis et al. (2009) researched this question over ecologically relevant levels of inbreeding, rather than the extreme inbreeding sometimes imposed.
Selection can act in two important ways to slow (or even prevent) the loss of genetic variation in inbred populations. Selection can act directly to maintain genetic variation when individuals that are heterozygous at specific loci have higher fitness than either of the homozygotes. Genetic variation can also be maintained indirectly via associative overdominance. Associative overdominance is traditionally defined as higher mean fitness of heterozygotes at a neutral locus because the marker is in gametic disequilibrium with a locus that is under selection (i.e. via linkage).
Demontis et al. (2009) found large differences in the amount of genetic variation maintained in fast vs. slow inbreeding treatments (e.g. 25% more heterozygosity in lines more slowly inbred). The fast inbred lines also experienced 20% extinction compared to no extinction in the slow inbred lines. Such differences provide a very clear and powerful recommendation for conservation biology and selective breeding programmes. When possible, slow inbreeding is to be much preferred over fast inbreeding. For example, the difference between removing 5% of a species’ habitat each generation over the next 13 generations may have a very different outcome than removing 49% of the habitat all at once.
What is truly unique about this study is that the 40 SNPs used in the study were known to be contained within genes that are differentially expressed between inbred and outbred individuals. Thus, they are quite likely the targets of natural selection during inbreeding. Because these genes are differentially expressed in inbred vs. outbred populations and because many of them have a known function, this study allows us to begin to make powerful inferences about what class of genes and how many loci might be important to the inbreeding process. Most importantly, this selective suite of SNPs makes it obvious that selection was stronger in the slow inbred lines compared with the fast inbred lines and that selection is responsible for the differences in the amount of molecular genetic variation maintained in the two treatments. This study is an excellent companion piece to previous work showing that slower inbreeding maintains more genetic variation in quantitative traits as well (Kristensen et al. 2005). However, it goes much further than the previous study because of its ability to shed light on mechanisms.
Most intriguing is the authors’ finding that the increased heterozygosity is at least partially due to balancing selection. There is evidence for this, in their data set, at two loci. There is also strong evidence that one locus is under purging selection. This finding is so interesting because the bulk of the evidence thus far, despite a rapidly growing number of recent studies showing balancing selection to be much more common than generally thought (e.g. Ferreira & Amos 2006; Mäkinen et al. 2008), still suggests that the variation at most loci consists of a dominant allele with the highest fitness and one or more recessive deleterious alleles (e.g. Ayroles et al. 2009). In all likelihood, both forms of selection are acting on most genomes, but their proportional contribution is still being investigated and this study adds another piece of evidence.
With regards to the maintenance of genetic diversity during inbreeding, it is often less appreciated that associative overdominance can result more easily through variation in inbreeding levels among members of the population than through linkage. This can lead to a general and genome-wide heterozygosity–fitness correlation. This correlation can result from both selection against individuals homozygous for deleterious recessive alleles or through heterozygote superiority; as long as less inbred individuals have higher average fitness, their increased reproduction compared to more homozygous individuals will help prevent erosion of genetic variation in the face of random genetic drift. Thus, the results of this experiment suggest that genetic diversity has been maintained, in these populations during slower inbreeding, by associative overdominance from weak selection on large number of loci and strong selection on a much smaller number of loci, with some of those loci being subject to purging selection and some to balancing selection. This is similar to what was found in a genome-wide scan of Drosophila (Ayroles et al. 2009).
Demontis et al. (2009) suggest correctly that future research would benefit from looking for negative correlations between the amount of genetic variation retained and the amount of inbreeding depression suffered, but using traits more directly connected to fitness than those used in this study. I agree with this. I would also add that, possibly even more importantly, the slower inbred lines need to be tested in multiple novel environments to confirm that the smaller reduction in molecular genetic variation does indeed translate into higher levels of evolutionary potential.
The most immediate need may be for this type of research to be carried out in organisms other than Drosophila melanogaster. The advances in high throughput sequencing may make these types of experiments possible in nonmodel organisms in the near future.