(A) A population with polymorphism and heterozygote advantage. The population includes high-fitness heterozygotes and low-fitness homozygotes. (B) A haplotype bearing two alleles in tandem forms high-fitness homozygotes. A population monomorphic for such a haplotype consists of only high-fitness individuals.

MHC diversity over time in simulations with parameter values that lead to high diversity (red curve) and with slightly altered parameter values but no compensatory adjustment of cmax.

The bottom panel is a vertical zoom of the top panel.

Gene family expansion in the Gaussian model.

Top: diversity over time in twenty simulation runs. In each run, haplotypes bearing two alleles come to predominate, leading to a sudden loss of any diversity that has accumulated. Bottom: the distribution of time until such an event in 10,000 simulation runs.

Diversity over time in simulations of the bitstring model with gene family expansion.

Twenty simulation runs are shown. Diversity becomes and remains low when expanded haplotypes come to predominate.

Diversity over time in simulations with gene family expansion but only weak expression of additional alleles carried by a haplotype.

Top: Gaussian model. Bottom: bitstring model.

Simulation results for a model in which mutation can alter the breadth of peptides presented.

The initial allele specifies v=9, and cmax is set accordingly. Mutation can change v to 8.12 (top) or 8.9 (bottom). Results of one hundred simulation runs are shown in each plot.

Diversity over time in simulations of the bitstring model in which v can evolve between 9 and 8.12 and overly wide presentation breadth is lethal.

The initial allele has narrow presentation breadth (v=9) in the top panel and broader presentation breadth (v=8.12) in the bottom panel. One hundred simulation runs are shown for each. Note that many curves in both plots fall on top of each other, with effective number of alleles close to 1 throughout the simulations.

Simulation results for the bitstring model with various values of parameters m and v and identical values of other parameters, including cmax.

One hundred simulations runs are represented in each plot. Equilibrium diversity is high with m=100 and v=20 (A), but low if these parameters are changed slightly (B-E). The thick black curve in some of the plots (B, D, and E) represents the harmonic mean among runs.

Simulations like those in Fig. 6, top, except that diversity is allowed to accumulate for one million generations before mutations affecting the breadth of presentation are allowed.

In all 20 runs, diversity collapses quickly once such mutations are allowed. The bottom panel is a horizontal zoom of the top panel.

Simulations incorporating a 50% decrease in all peptide detection probabilities in individuals with certain genotypes. Other conditions are as in Fig. 6, top.

Despite the inclusion of this effect, alleles with higher presentation breadth quickly come to predominate, preventing the development of high diversity.