multilocus balancing selection
Navarro A & Barton NH 2002 The effects of multilocus balancing selection on neutral viability. Genetics 161:849-863.
- forward simulations are probably the most straightforward method (Barton & Navarro 2002)
- we aim to focus on qualitative results that will not depend on the exact nature of selection or on the details of the model
- the main factors influencing the degree of polymorphism and linkage disequilibrium at the selected loci are, together with recombination and population size, the strength and nature of selection
- the amount of linkage disequilibrium in the system is an important issue
- because it determine the actual degree of subdivision in the population
- the results ... (with low recombination) become clear if one considers that the population depicted there consists essentially of two complementary background, independently of the number of selected loci
- one needs only to recalculate the coalescent predictions, taking into account the real background equilibrium frequencies
- when recombination is low, nondominant backgrounds have such low frequencies that most of the time they are absent from the population and, thus, can be ignored
- with intermediate or high recombination one needs to take into account the exact equilibrium frequencies in the coalescent
- the helpful assumption made by Barton and Navarro (2002) of no linkage disequilibrium between selected loci cannot be used
- independently of the number of loci and the selection regime, neutral variability does not increase for markers at at extreme of the set of selected loci when Nr > 1
- as expected, simulated and predicted results diverge when the number of loci is large
- independently of this discrepancy, the effect of the set of selected loci on neutral variability dissipates when recombination between them and the neutral locus is > 1 / N
- chromosomal regions between selected loci
- negative epistasis and intermediate or low recombination (r ≥ 10− 3, Figure 7a)
- maximum neutral variability (which is always found in regions closely linked to the selected loci) is increased beyond the one-locus (two backgrounds) limit
- the population is more subdivided with two loci (four backgrounds) than with a single one
- variability is increased in a wider region of the chromosome than expected for a single locus alone
- still, high recombination rates (r > 10− 3) preclude any relevant multilocus effect and the predicted values are almost identical for one as for two selected loci
- Figure 7b
- multiplicative fitnesses
- results for positive epistasis (not shown) are qualitatively equivalent
- variability is not increased beyond the one-locus limit because the population is dominated by only two backgrounds
- in segments between selected loci, however, a second kind of multilocus effect is detected
- crossing over between the two selected loci, which would allow neutral alleles to recombine away, breaks linkage disequilibrium and generates gametes that are eliminated by selection
- the effective recombination rate is reduced in regions between selected loci
- if selection is strong enough, variability can be increased even when recombination is high
- Figure 8b with r = 10− 2
- the extended coalescent needs only information about the frequencies of the selectively relevant haplotypes and not about the kind of selection producing them
- diversity is enhanced at neutral markers located in regions among the set
- even if they are a long way from the selected loci themselves
- Figure 8 with r = 10− 2, Nr = 10
- this mechanism has been already described by Kelly and Wade (2000)
- who analyzed a system formed by two diallelic, epistatically interacting loci and obtained predictions for the patterns of neutral sequence variation linked to them
- their model is an instance of the positive epistasis scenario presented here