Novel insights into joint estimations of demography, mutation rate, and selection using UV sex chromosomes

A central goal in evolutionary genomics is to understand the processes that shape genetic variation in natural populations. In anisogamous species, these processes may generate asymmetries between genes transmitted through sperm or eggs. The unique inheritance of sex chromosomes facilitates studying such asymmetries, but in many systems sex-biased mutation, demography, and selection are confounded with suppressed recombination in only one sex (the W in females, or the Y in males). However, in a UV sex-determination system, both sex chromosomes are sex-specific and experience suppressed recombination. Here we built a spatially-structured simulation to examine the effects of population density and sex-ratio on female and male effective population size in haploids and compare the results to polymorphism data from whole-genome resequencing of the moss Ceratodon purpureus. In the parameter space we simulated, males nearly always had a lower effective population size than females. Using the C. purpureus resequencing data, we found the U and V have lower nucleotide diversity than the autosomal mean, and the V is much lower than the U, however, we found no parameter set in the model that explained both the U/V and U/autosome ratios we observed. We next used standard molecular evolutionary analyses to test for sex-biased mutation and selection. We found that males had a higher mutation rate but that natural selection shapes variation on the UV sex chromosomes. All together the moss system highlights how anisogamy alone can exert a profound influence on genome-wide patterns of molecular evolution.


Introduction
Anisogamy, the condition in which genetic information is transmitted from one generation 44 to the next through two different sized gametes, is widely shared among eukaryotes. The 45 smaller gametes, typically called sperm, are abundant and motile, while the larger gametes, typically called eggs, are less abundant, better provisioned, and often sessile or retained on the parent. In species with two separate sexes, males produce sperm and females produce eggs, 7 thousands of viable spores (Norrell, Jones, Payton, & McDaniel, 2014;A. J. Shaw & Gaughan, 147 1993;Shortlidge et al., 2020), most of which fall near the parent sporophyte, but some are 148 captured by air currents and travel great distances (Biersma et al., 2020;McDaniel & Shaw, Demographically-informed expectations that specifically incorporate anisogamy are 151 necessary to fully understand the role of sex-specific evolutionary forces shaping patterns of 152 polymorphism in dioecious species. In principle, a single male could fertilize many nearby 153 females, an inference supported by field observations and experiments (Johnson & Shaw, 2016; 154 Shortlidge et al., 2020), which increases variance in male reproductive success. Many 155 bryophyte populations, including mosses, have an apparent female-biased sex ratio, due to sex-156 biased differences in clonal growth rates, differences in mortality, or differences in the number of For each run of the simulation (equivalent to one generation of mating), individuals were 198 randomly placed onto a 100 X 100 cell grid. Females mated in a random order. Each female 199 searched the eight adjacent cells to it for males to mate with. If there was more than one 200 adjacent male, the female randomly selected one of them to mate with. If no males were in the 201 cells adjacent to the female, that female did not mate. An example of the population following 202 one run of the simulation can be seen in Figure S1. All simulations were run in R (3.5.1; (R Core 203 Team, 2013) and plotted using the packages reshape2 (v1.4.3;(Wickham, 2007(Wickham, , 2012 and

214
To evaluate a variety of demographic scenarios, we ran simulations for a range of 219 Generating resequence data to test the model. We generated U-linked, V-linked, and 220 autosomal polymorphism data from 23 C. purpureus isolates collected from nine locations 221 ( Figure 1; Table S1). To start these lines, sporophytes were surface sterilized, and a single 222 germinated spore was isolated following (Norrell et al., 2014). DNA was extracted using a 223 modified CTAB protocol following (Norrell et al., 2014 (v0.11.4;(Andrews, 2010)).

12
We called variants on all BAMs together using BCFtools (v1.9; (H. Li, 2011)) mpileup 270 and call using a ploidy of one. The resulting VCF file was filtered using BCFtools filter by 271 excluding variants with a Phred-based quality score of the alternate base (QUAL) <30, 272 combined depth across samples (DP) <10, and mapping quality (MQ) <30, where these filters 273 had to be met in at least one sample (&&). We subset the VCFs using view to have females for 274 the U, males for the V, and both sexes for the chloroplast and autosomal analyses, excluding 275 isolates from localities where both sexes were not present ( Figure 1; Table S1). The VCFs were 276 finally filtered to remove variants with >20% missing data.  Table S1). For each of these metrics we 290 did sliding-window analyses using a window size of 100,000 and jump of 10,000 and plotted 291 these with a loess correction span of 0.03 using karyoploteR (v1.8.8; (Gel & Serra, 2017)). We 292 excluded the chloroplast, however, because the contig we analyzed was barely larger than the 293 windows (105,555 bp). We generated 95% confidence intervals (CI) by bootstrapping 1,000 of 294 the sliding windows with replacement and tested for differences between the autosomes and sex chromosomes using the Mann-Whitney U test with a Benjamini and Hochberg correction for 296 multiple tests (Benjamini & Hochberg, 1995;McKnight & Najab, 2010).

297
To test for adaptive evolution in autosomal and sex-linked genes we first calculated the non-synonymous polymorphisms (Pn) to synonymous polymorphisms (Ps) to the ratio of non-

305
We evaluated the significance of deviations from neutrality using Fisher's exact test (Fisher,

312
To determine if the mutation rate differed between the U and V chromosomes, using  where θA is the Wu and Watterson estimator for an autosome, μA is the autosomal mutation 341 rate, and N is the census population size assuming that the sex ratio is equal (a different 342 equation is needed for unequal sex ratios).

344
From our resequence data, we estimated the ratio between θ for the V and U 345 chromosomes, θV/θU, which we used to solve for the ratio ⍺ = Vm/Vf to determine how different 346 the variances in reproductive success would need to be to explain the results, such that (7) 347 . (7) Similarly, given the ratio θV /θA, we can solve for the following ratio 348 .
This solution for does not hold for unequal sex ratios, however. Therefore, for unequal sex 349 ratios we evaluated the ratio 350 . (9)

351
For calculations from our empirical data, we assume that N = 400,000, which is increased approximately exponentially, and at high densities the effective population of the U 367 chromosome can exceed that of autosomes and even the census population size (Figure 2A).

368
This seemingly counterintuitive finding arises from the fact that, at high population densities, the 369 autosomal diversity is passed through relatively few males, while the U-linked variation is 370 passed exclusively through the females which have very low variation in reproductive success.

371
We also simulated the effect of variation in sex ratio on Ne of the U, V, and autosomes.

372
We explored the effects of sex-ratio variation at multiple population densities, but because the 373 trends were homogeneous we present only the results at a density of 20%. At this population 374 density, with an even sex ratio, the results most closely match the infinite-sites expectations. At 375 even modest male-biased sex ratios, the Ne of the U, V, and autosomes were very low ( Figure   376 2B). Even when males outnumbered females, relatively few males contributed to reproduction.

377
As the sex ratio became more female biased, in contrast, NeU increased dramatically ( Figure   378 2B). The NeA increased slightly with a modest female bias, but at more dramatic female biases 379 the NeA decreased slightly.

380
These models demonstrate that under reasonable demographic conditions, the infinite-381 sites expectations that the U and V each should have half the Ne of an autosome are met only 382 at low population densities. Moreover, the V can have lower Ne than the U and autosomes due 383 to a greater variance in reproductive success due to sex differences in life history alone (i.e., with UV sex chromosomes, we generated whole-genome resequence data for 23 isolates of C. 389 purpureus (Figure 1; Table S1). We found across isolates on average 80.87% of reads mapped 390 with BWA and 81.64% with NGM and our average coverage is ~28.5x (Table S1)

395
for simplicity we discuss the remaining results from using the BWA mapper, although the 396 summary statistics were similar with NGM and we report these in Table S2.

479
It is important to note that our simulation results relied on specific assumptions about the 480 direct links among sex chromosomes, anisogamy, and life history. For example, we assumed 481 that all individuals with a U-chromosome produced eggs and all individuals with a V-482 chromosome produced sperm, although it is well-known that these assumptions are violated in 483 many systems (Ming, Bendahmane, & Renner, 2011). In C. purpureus many individuals do not assumed that the egg-producing sex (females) and sperm-producing sex (males) each have specific, invariant life histories: females can only mate once and males can mate many times species, we caution that these results cannot be applied uncritically to other anisogamous 490 species (see (Sarah Blaffer Hrdy, 1986;S. B. Hrdy, 1981;Tang-Martínez, 2016)

546
We also found no evidence that the differences in θ between sex chromosomes was the 547 result of an elevated mutation rate in females. In fact, both dS and dN were higher on V-linked 548 genes (Mann Whitney U, dN p=0.044; dS p=0.005; Figure 4B-C), the opposite of the pattern 549 that would explain the higher θ on the U chromosome compared to the V. We did recover the 550 expected lower nucleotide diversity on the chloroplast (Table 1), consistent with other plants (D.

551
R. Smith, 2015;Wolfe et al., 1987) suggesting that the elevated male mutation rate is unlikely to 552 be an artifact of our sampling scheme.

553
Male-biased mutation rates are widely observed in animals and some seed plants Together these results support the long-standing notion that positive selection can dramatically decrease nucleotide diversity in non-recombining regions (Begun & Aquadro, 1992; illustration of this effect. Our simulations also suggest that density, mating system, and factors 641 that influence the variance in male reproductive success may also be confounded with selection 642 in analyses of Y chromosome polymorphism.

643
The geographic distribution of U and V-linked variants may provide insight into other  and sex-linked genes that were significant in the MK test at p<0.1. B) nonsynonymous mutation 1089 rate (dN) and C) synonymous mutation rate (dS) of one-to-one orthologous U and V-linked