Deleterious variation shapes the genomic landscape of introgression

Abstract

While it is appreciated that population size changes can impact patterns of deleterious variation in natural populations, less attention has been paid to how gene flow affects and is affected by the dynamics of deleterious variation. Here we use population genetic simulations to examine how gene flow impacts deleterious variation under a variety of demographic scenarios, mating systems, dominance coefficients, and recombination rates. Our results show that admixture between populations can temporarily reduce the genetic load of smaller populations and cause increases in the frequency of introgressed ancestry, especially if deleterious mutations are recessive. Additionally, when fitness effects of new mutations are recessive, between-population differences in the sites at which deleterious variants exist creates heterosis in hybrid individuals. Together, these factors lead to an increase in introgressed ancestry, particularly when recombination rates are low. Under certain scenarios, introgressed ancestry can increase from an initial frequency of 5% to 30-75% and fix at many loci, even in the absence of beneficial mutations. Further, deleterious variation and admixture can generate correlations between the frequency of introgressed ancestry and recombination rate or exon density, even in the absence of other types of selection. The direction of these correlations is determined by the specific demography and whether mutations are additive or recessive. Therefore, it is essential that null models of admixture include both demography and deleterious variation before invoking other mechanisms to explain unusual patterns of genetic variation.

Fig 1. The demographic models used for the simulations.

After a burn-in period of 10N_A (100,000) generations, a single population diverged into two subpopulations. The demography of the subpopulations was modified in ways that changed the distribution of deleterious variation. 2N_A (20,000) generations after the split, a single pulse of admixture occurred such that 5% of the ancestry of the recipient population came from the donor population. Arrows in each panel denote the direction of gene flow. The simulation was run for N_A (10,000) additional generations after admixture. Population sizes were changed as shown for each model. See S1 Table for specific parameter values for each model.

Fig 2. The change in the ratio of fitness over time due to demography.

Each individual plot depicts the ratio of the mean fitness of the recipient population (w_R) to the donor population (w_D) for the demographic models shown in Fig 1. The mean (dotted line) and the 25^th to 75^th percent quantiles are shown for 200 simulation replicates. The vertical gray line depicts the time of gene flow, and the horizontal dashed black line depicts w_R/w_S = 1. Different colors denote distinct recombination rates used in the simulations. Left panel denotes additive mutations (h = 0.5) while the right panel shows recessive mutations (h = 0).

Fig 3. The frequency of introgression-derived ancestry (p_I) in each model.

Earlier generations are not shown since p_I = 0 prior to admixture. The mean (dotted line) and the 25^th to 75^th percent quantiles are shown for 200 simulation replicates. The vertical gray line depicts the time of gene flow, and the horizontal dashed black line depicts the initial admixture proportion of 0.05. Different colors denote distinct recombination rates used in the simulations. Left panel denotes additive mutations (h = 0.5) while the right panel shows recessive mutations (h = 0).

Fig 4. The effect of divergence and population size on introgression-derived ancestry when mutations are recessive.

The proportion of ancestry that is introgression-derived, p_I, at the time of N_A (10,000) generations after admixture, is shown for 200 simulation replicates and two demographic models (Model 0 and Model 4, refer to Fig 1) for a range of times between subpopulation divergence and the admixture event. The recombination rate in all simulations is r = 10⁻⁹ per base pair. Violin plots represent the density while dot and whiskers represent the mean and one standard deviation to either side. The horizontal dashed black line represents the initial admixture proportion of 0.05. Note that as the split time increases, p_I also increases. Values of F_ST reflect the amount of population differentiation at the split time, that is, immediately before admixture has occurs in each of these models.

Fig 5. The average genomic landscape of introgression in simulations with human genomic structure.

The frequency of ancestry that is introgression-derived is shown for non-overlapping 100 kb windows in a simulated 100 Mb region of chromosome 1. The model numbers refer to the models shown in Fig 1. Points represent a single value for each 100 kb window and lines are loess curves fitted to the data. The horizontal black dashed line represents the initial frequency of introgression-derived ancestry, p_I = 0.05. Vertical blue bars represent genes in which deleterious mutations can occur. Red curves denote the results for recessive mutations, orange curves show the results for additive mutations, and blue curves show the results for simulations with a h(s) relationship.

Fig 6. The relationship between recombination rate and introgression for different models of demography and selection.

The frequency of introgression-derived ancestry (p_I) is plotted against the ranked average recombination rate of non-overlapping 100 kb windows in each window at time N_A (10,000) generations after admixture. Gray dots represent the average p_I of a single window in 100 simulation replicates, while red dots represent the average p_I of 5% of windows as ordered by rank of recombination rate. Rank was randomly assigned for ties. The horizontal black line represents the initial p_I of 5%. Spearman’s ρ is computed for the relationship between recombination rate and p_I in each window and p-values indicate the significance of H₁: ρ≠0. The model numbers refer to the models shown in Fig 1.

Fig 7. The relationship between exon density and introgression for different models of demography and selection.

The frequency of introgression-derived ancestry (p_I) is plotted against the average exon density of non-overlapping 100 kb windows in each window at time N_A (10,000) generations after admixture. Gray dots represent the average p_I of a single window in 100 simulation replicates, while red dots represent the average p_I of 5% of windows as ordered by rank of exon density. Rank was randomly assigned for ties. The horizontal black line represents the initial p_I of 5%. Spearman’s ρ is computed for the relationship between recombination rate and p_I in each window and p-values indicate the significance of H₁: ρ≠0. The model numbers refer to the models shown in Fig 1.

Fig 8. Differences in introgression between the X chromosome and autosomes.

The average frequency of introgression-derived ancestry across the entire simulated chromosome (p_I) at time N_A (10,000) generations after admixture is shown for three demographic models and three models of fitness. Model numbers refer to the models shown in Fig 1. Bars represent the mean p_I of 100 simulation replicates and error bars represent standard errors of the means. The horizontal dashed black line represents the initial p_I of 5%.

Fig 9. The average genomic landscape of introgression in simulations with Arabidopsis genomic structure.

The frequency of ancestry that is introgression-derived is shown for non-overlapping 100 kb windows in a simulated 29.1 Mb region of chromosome 1. The model numbers refer to the models shown in Fig 1. Points represent a single value for each 100 kb window and lines are loess curves fitted to the data. The horizontal black dashed line represents the initial frequency of introgression-derived ancestry, p_I = 0.05. Vertical blue bars represent genes in which deleterious mutations can occur. Blue curves show the results for simulations with a h(s) relationship.

Fig 10. The impact of partial selfing on the frequency of introgression-derived ancestry.

The simulated demographic model is Model 0 (Fig 1) with Arabidopsis genomic structure. The frequency of introgression-derived ancestry (p_I) at time N_A (10,000) generations after admixture is plotted for seven different scenarios of admixture between a partially selfing population and an outcrossing population. Bar plots denote the average p_I of 100 simulation replicates and error bars represent standard errors of the averages. The horizontal dashed black line represents the initial p_I of 5%, and the horizontal dashed blue line represents the p_I that is expected when both subpopulations are outcrossers. Labels on the x-axis denote the probability of selfing in the population that is partially selfing.