Evolution at the mouse t complex: why is the t haplotype preserved as an integral unit?

Segregation distorters are selfish genetic elements that bias Mendelian segregation in their favor. All well-known segregation distortion systems consist of one or more "distorter" loci that act upon a "responder" locus. At the t complex of the house mouse, segregation distortion is brought about by the harmful effect of t alleles at a number of distorter loci on the wild-type variant of the responder locus. The responder and distorter alleles are closely linked by a number of inversions, thus forming a coherent t haplotype. It has been conjectured that the close integration of the various components into a "complete" t haplotype has been crucial for the evolutionary success of these selfish genetic elements. By means of a population genetical metapopulation model, we show that this intuition may be unfounded. In fact, under most circumstances an "insensitive" t haplotype retaining only the responder did invade and reach a high frequency, despite the fact that this haplotype has a strong segregation disadvantage. For certain population structures, the complete t haplotype was even competitively excluded by partial t haplotypes with lower segregation ratios. Moreover, t haplotypes carrying one or more recessive lethals only prevailed over their nonlethal counterparts if the product of local population size and migration rate (Nm) was not much smaller or larger than one. These phenomena occurred for rather realistic fitness, segregation, and recombination values. It is therefore quite puzzling that partial t haplotypes are absent from natural house mousepopulations, and that t haplotypes carrying recessive lethals prevail over nonlethal t haplotypes.