Conceptual representations of the proposed model of inversion-associated balanced sexual antagonism. (A) Hypothetical skewed distribution of male mating success, yielding a greater variance in reproductive success for males than for females in our model. (B) Fitnesses of display-favoring alleles under sexually antagonistic balancing selection, illustrating a pleiotropic variant that should rise in frequency when rare, but decline if frequency exceeds the balanced equilibrium value. (C) Fitnesses of display-favoring alleles at two haploid loci under the same model, illustrating synergistic epistasis between displaying-favoring alleles. (D) A hypothetical trajectory of four such mutations, in which B outcompetes A, an inversion links B and C to create a more strongly display-favoring haplotype, and then the addition of D to that haplotype furthers the accumulation of antagonistic variants, reaching an equilibrium frequency while displacing less extreme display-favoring haplotypes.

The layout of a single simulated generation. Simulated individuals each reach the reproductive stage with probability in proportion to the product of the survival effects of their alleles. To generate each offspring in the next generation, a surviving female is first randomly sampled (with replacement) to be the mother. Then a specified number of surviving males are randomly sampled to generate the pool of males encountered by the sampled female - without replacement for each pool and with replacement between pools. For each encounter pool, each male is assigned an observed display quality as the sum of the display effects of its alleles, plus a normally distributed noise effect for each encounter. The highest quality male in the encounter pool is selected as the father, and the offspring genome is generated. The female gamete is generated with crossover and gene conversion events. Odd numbers of crossovers in the interior of an inversion are resampled from the same parent, to model the biology of an organism that eliminates aneuploid gametes during oogenesis, or exhibits reproductive compensation. D. melanogaster males do not cross-over, and so males do not recombine in the simulations presented.

Simulations show that a balanced equilibrium frequency exists for certain sexually antagonistic variants. The proportion of simulations retaining polymorphism (top) and the average final frequency (bottom) are plotted for a single locus at which one antagonistic allele with the indicated effects reducing survival proportion (of both sexes) and increasing male reproductive quality score is initiated at equal frequency with an alternative allele that has no such effects. Larger effect sizes are plotted in the left panels, with values incrementing by 0.025, and smaller effect sizes (a subset of the larger) are plotted in greater detail in the right panels, with values incrementing by 0.005. Most parameters result in either all simulations retaining polymorphism (purple), or else all simulations losing polymorphism (white). However, simulations retaining polymorphism have a broad range of equilibrium frequencies (shades of green). For each parameter combination, 500 replicates were simulated for a diploid population of 1,000 individuals evolving for 20,000 generations, with further details as indicated in the Materials and Methods.

Simulated inversions may persist as polymorphisms when linked to sexually antagonistic, pleiotropic variants. In turn, the presence of inversions facilitates the accumulation of antagonistic variation. Rates of long-term persistence are shown for simulations with two defined antagonistic variants alone (left panels) and with the additional presence of a defined inversion (right panels). To focus the potential addition of a second linked variant to an antagonistic haplotype, we begin with the stronger variant already at an intermediate frequency in the population (0.5). The weaker antagonistic variant and the inversion either both start at 0.5 frequency as well (upper row) or both start at a lower frequency of 0.05 (lower row). The 1 Morgan chromosomal segment is diagrammed above the plots, with the inversion breakpoints marked (0.21M, 0.55M), chosen to include space across and beyond the inverted region. The proportion of simulations in which the second, variably-positioned weaker antagonistic variant (survival and reproductive values 0.86, 0.12) maintains polymorphism is shown (light red line), dependent on the recombination distance from a stronger allele (0.75,0.3) at fixed position 0.225M, which always persists (dark red line at 1 across the top of each plot). In the right panels, the proportion of replicates in which the inversion is retained is also plotted (blue line). Values for each parameter combination were averaged over 1000 replicate simulations, for populations of 1000 diploid individuals run for 20N generations. Gene conversion occurs for each heterozygous variant at a probability of 10−2 per female meiosis with random direction. Crossover arrangements generating aneuploidy, where there are an odd number of crossovers within the inverted region of a heterokaryotypic parent, are resampled from the same parent to represent removal to polar bodies or reproductive compensation.

Simulated populations with stochastic mutation of new inversions and sexually antagonistic variants generate persistent inversion polymorphisms and accumulate linked sexually antagonistic variants. (A) In simulated populations with stochastic mutation of new variants having pleiotropic effects on survival and male display, levels of antagonism increased with time before stabilizing at an intermediate value (left panel). Whereas, simulations which also allowed the mutation of new inversions continued accumulating stronger antagonism throughout the simulated time interval (right panel). (B) In simulations which allowed the mutation of new inversions, a low crossover rate (left panel, 43.6 cM in females after scaling) allows populations to accumulate larger differences between the most common karyotype and all others than when the rate of crossover is particularly high (right panel, 43.6 M in females after scaling). Gene conversion occurs at each heterozygous variant with a scaled probability of 0.1295 per heterozygous variant in both scenarios. Values given are the mean of 1000 replicate simulations of 1000 diploid individuals run for 20N generations.

Antagonism-associated karyotypes reach predictable equilibrium frequencies in simulations with antagonistic and inversion mutations, when survival costs are shared between sexes. (A) Histograms of the number of arrangements at a specified frequency across all simulation replicates in females (left) and males (right), colored by how the average survival effect of that arrangement ranks compared to the others in its population. Ranks are normalized to be relative to the median arrangement value, to better account for rare arrangements and ties. These simulations reach an equilibrium with two predominant arrangements, with the more survival-focused arrangement around an overall frequency of 0.25 and more common in females, and the more male-competition-focused arrangement around a frequency of 0.75 and more common in males. (B) Because successful males tend to be homozygotes and successful females heterozygotes, the less frequent of the two major arrangements (which tends to favor survival) has homokaryotype frequencies far below the Hardy-Weinberg expectation, and there is an excess of heterokaryotypes. These genotype frequencies shift somewhat between zygote (left panel) and adult (right panel) populations as the genotype affects survival.

In simulations with female-limited survival costs, three distinct outcomes are observed involving either two or three balanced haplotypes. From simulations with randomly occurring antagonistic and inversion mutations, histograms of inversion frequencies across simulations for surviving females (left) and males (right) are partitioned into three categories based on sex-specific haplotype frequency outcomes. The “Favors Female” category (top) includes simulations in which there are exactly two arrangements with frequencies between 0.1 and 0.9 in both sexes (20.0% of replicates). The “Male-Specific” category (middle) includes simulations in which there are two arrangements with whole-population frequencies between 0.1 and 0.9, but one arrangement is absent in adult females (37.0% of replicates). The “Three Arrangement” category (bottom) includes simulations in which there are exactly three arrangements with whole-population frequencies between 0.1 and 0.9 frequency (40.6% of replicates). 2.3% of simulations did not fit in these categories and had only one predominant arrangement. As in Figure 6, inversions are colored by how the average survival effect of that arrangement ranks compared to the others in its population, and ranks are normalized to be relative to the median arrangement value, to better account for rare arrangements and ties.

The layout and potential expectations of the laboratory evolution experiment. We cross outbred males from a high-inversion population to inbred, non-inverted females from a specific inbred line, and collect DNA samples from the fathers, embryo offspring, and aged adult offspring (left). We hypothesize from our antagonistic pleiotropy model that one or more of the four inversions present on separate chromosome arms in our experimental populations may experience opposing selection between fathers and embryos versus between embryos and aged adults (middle). To test the significance of observed frequency changes, we designed a resampling model to represent the sources of neutral variation expected in the experiment (right).

Some Drosophila inversions show evidence for antagonistic tradeoffs and sex-specific survival. (A) Testing for antagonistic fitness effects on reproduction and survival. Each inversion’s frequency in parents, embryo offspring, and adult offspring are shown for each maternal inbred line. Parental frequency is taken as the paternal frequency divided by two to account for the inversion-free inbred mothers. Adult offspring frequencies are taken from the sum of estimated allele counts across all adult offspring cohorts. * indicates p-value < 0.05 for evidence of significant reversed frequency change from parental to embryo and from embryo to offspring, combined across all four maternal line crosses and corrected for multiple tests, when compared to neutral simulated experiments (details in Materials and Methods). (B) Testing for evidence of sex-specific survival effects. The changes in inversion frequencies between embryos and both female and male adult offspring are illustrated. * indicates p-value < 0.05 for significant differences in female versus male inversion frequency among adult offspring, combined across all four maternal crosses and corrected for multiple tests.