Ejaculate sperm number compensation in stalk-eyed flies carrying a selfish meiotic drive element

Meiotic drive genes cause the degeneration of non-carrier sperm to bias transmission in their favour. Males carrying meiotic drive are expected to suffer reduced fertility due to the loss of sperm and associated harmful side-effects of the mechanisms causing segregation distortion. However, sexual selection should promote adaptive compensation to overcome these deleterious effects. We investigate this using SR, an X-linked meiotic drive system in the stalk-eyed fly, Teleopsis dalmanni. Despite sperm destruction caused by drive, we find no evidence that SR males transfer fewer sperm to the female’s spermathecae (long-term storage organs). Likewise, migration from the spermathecae to the ventral receptacle for fertilisation is similar for SR and wildtype male sperm, both over short and long time-frames. In addition, sperm number in storage is similar even after males have mated multiple times. Our study challenges conventional assumptions about the deleterious effects of drive on male fertility. This suggests that SR male ejaculate investment per ejaculate has been adjusted to match sperm delivery by wildtype males. We interpret these results in the light of recent theoretical models that predict how ejaculate strategies evolve when males vary in the resources allocated to reproduction or in sperm fertility. Adaptive compensation is likely in species where meiotic drive has persisted over many generations and predicts a higher stable frequency of drive maintained in wild populations. Future research must determine exactly how drive males compensate for failed spermatogenesis, and how such compensation may trade-off with investment in other fitness traits.


SI-A1 Sperm storage in small and large females
In the main text we report on sperm storage in the female spermathecae after mating to an SR or an ST male. Females were either small (eyespan 4.1 -5.2 mm) or large (eyespan ≥6 mm).
Spermathecae size scales with female size, but female size did not influence the number of sperm stored in the spermathecae (Fig. SA1). Figure SA1: Number of sperm stored in large (L; eyespan ≥6 mm) or small female's spermathecae after mating with an SR (pair of plots on the left) or ST (pair of plots on the right) male. Boxplots (first to third quartile) with median line and whiskers (1.5 IQR), and mean ± s.e. (red points and lines). Female size did not influence number of sperm stored (P = 0.493), and this did not depend on SR (male type x female size P = 0.789).

SI-A2 Stock source and maintenance SI-A2.1 Standard stock population
The standard wildtype stock (ST-stock) population was created from collections in 2005, as described in the main manuscript. This population has been regularly monitored and does not contain meiotic drive and carries non-distorting standard X chromosomes (X ST ).

SI-A2.2 Sex-ratio meiotic drive stock population
Flies were collected in 2012 (by A. Cotton and S. Cotton) from the Ulu Gombak valley to create a sex ratio meiotic drive stock (SR-stock) population. To establish and maintain a stock with meiotic drive, a standard protocol was followed (Presgraves et al. 1997). Briefly: wild males (of unknown genotype) were mated to ST-stock females and their offspring (F1) were collected.
When an F1 brood was female biased (80% female, > 10 offspring), it was assumed that the father was a carrier of the sex-ratio distorting X SR chromosome, so that the F1 female offspring had genotype X SR /X ST .
When sexually mature (> 4 weeks, Baker et al. 2003), F1 X SR /X ST females were mated with ST-stock males and their offspring (F2) were collected. X SR /X ST females and ST-stock males were housed in cage populations of ∼100 individuals at 1:1 sex-ratio. Male F2 offspring are expected to be 50:50 X SR /Y:X ST /Y as they inherit either an X SR or X ST chromosome from their mother. F2 males were subsequently mated to ST-stock females to identify those males carrying X SR , and the process repeated.
Even though there was error in the assignment of individuals as carriers of X SR , the process maintains the X SR chromosome in this stock. Over generations the SR phenotype has become more distinct as the stock maintenance procedure selected for female biased broods, so most SR-stock males now produce only female offspring, or at least > 95% female biased broods. Note that because the SR-stock maintenance involves back-crossing to ST-stock males and females, the autosomes, Y-chromosome and mitochondrial backgrounds are homogenised across the two stocks. For brevity, we hereafter refer to X SR /Y and X ST /Y males as X SR and X ST males respectively.
The stock populations were kept at 25 • C, with a 12:12 h dark:light cycle and fed puréed sweetcorn twice weekly. Fifteen-minute artificial dawn and dusk periods were created by illumination from a single 60-W bulb at the start and end of the light phase.

SI-A3.1 Introduction
To conduct efficient experiments on drive in the stalk-eyed fly Teleopsis dalmanni, it is vital to have a convenient and accurate method of distinguishing drive from wildtype males. A predominant conventional method is through offspring counts and performing χ 2 tests of significance on deviations from a 1:1 sex ratio. However, this labour-intensive and time consuming method is limited in multiple ways. Firstly, a sample of at least 10 offspring is needed perform a χ 2 test (Cochran 1952), so it is not possible to assign a phenotype to less prolific males. Furthermore, various factors may influence brood sex-ratio independent of drive, causing false identification of drive and wildtype males. For example, selection may operate on larval survival and thus alter sex ratios. This may lead to variation in wildtype male brood sex-ratio, sufficient to emulate drive. Brood sex-ratios are also of limited use in identifying females that are heterozygous or homozygous for meiotic drive. Finally, and importantly, there are many situations in which there is no opportunity for males to sire offspring, for example, when collecting flies in the field, or when there is a need to use laboratory males in experiments as soon as they are sexually mature or as virgins. These factors severely limit the utility of offspring counts as a means of identifying individuals that carry meiotic drive.
An alternative approach is to use genetic markers that can reliably predict the phenotype of individuals. The task of finding useful markers is feasible because the sex ratio distortion X chromosome (X SR ) shows widespread divergence from the standard X chromosome (X ST ) (Christianson et al. 2011;Cotton et al. 2014;Reinhardt et al. 2014;Paczolt et al. 2017). Furthermore, recombination between X SR and X ST is rare or absent (Johns et al. 2005;Paczolt et al. 2017). A number of X-linked microsatellite markers were identified for T. dalmanni which showed association with meiotic drive (Johns et al. 2005). A further investigation using wild collected flies found that only one of the four microsatellites, ms395, was predictive of the drive phenotype in T. dalmanni, albeit with a rather high error (Cotton et al. 2014). Here we further investigate locus ms395, along with three additional X-linked INDEL markers, comp162710, cnv395 and cnv125. Using data from laboratory reared flies, we assess these markers for their predictive power of male phenotype (ST or SR). When samples carrying particular allele all tend to exhibit a single phenotype, this demonstrates high predictive power. The consistency of each marker can be additionally informative about its usefulness in detecting X SR . A marker is consistent when phenotypes are represented by a single allele size.

SI-A3.2.1 Phenotype assignment
To produce offspring to determine male brood sex-ratio, males were kept with three non-focal females for up to 4 weeks and egg-lays, consisting of damp cotton-wool and excess puréed sweetcorn contained in a Petri dish, were collected twice weekly. Eggs were allowed to develop into pupae and offspring were collected and sexed until no more offspring emerged from the egglay. Males were subsequently stored in 100% ethanol at -20 • C. Significant (P < 0.05) deviation from a 1:1 sex ratio was tested for using χ 2 tests on offspring counts with a minimum of 10 offspring. Males that had a significantly female biased brood sexratio of ≥ 80% were categorised as SR. Males were otherwise classed as ST.

SI-A3.2.2 Genotyping
Three X-linked INDEL markers (comp162710, cnv395 and cnv125) were developed from se- . The ms395 locus has previously been shown to have an association with the drive phenotype in wild males (Cotton et al. 2014), where large ms395 alleles (>218 bp) are associated with female-biased broods. Primer sequences can be found in Table SA1.
A standard protocol was followed to extract DNA (Bruford et al. 1998). For each sample,

SI-A3.2.3 Statistical analysis
Analyses were carried out in R version 3.31 (R Core Team 2016). Only males that produced at least 10 offspring were included in the analyses. The relationship between allele size and brood sex-ratio for each X-linked locus was examined using generalised linear models (GLMs).
Offspring counts were analysed as proportion data (total female off-spring, total male offspring) in binomial GLMs. These models assess sex ratio bias, while accounting for brood size. The data was over-dispersed, so models were fitted with a quasi-binomial error distribution and a logit link function. ms395 allele size was included as a nominal variable, split into groups of 10 base pairs, as in Cotton et al. (2014). The allele sizes of the three INDEL markers segregate into two distinct size groups (Table SA2), and so allele size for comp162710, cnv395 and cnv125 were split into two groups of small and large alleles. We subsequently split ms395 alleles into large and small depending in whether they were > 218 or not (Cotton et al. 2014), and for each locus we examined the frequency distribution of allele size groups between brood sex ratio phenotype categories using Fisher's exact test.
Families with significant sex-ratio distortion were mostly female biased (222), but a smaller number were significantly male biased (7). We have no reason to believe that a male biased sexratio is actually genetically distinct from an ST phenotype. 66 males with female biased broods, 7 males with male biased and 132 males with unbiased broods had allele size information for at least one marker (ms395, comp162710, cnv395, cnv125). After applying the criteria for defining male phenotype category, 174 males were classed as ST and 31 as SR (Fig. SA2).

SI-A3.3.2 Allele size consistency
Allele size consistency ranges from 0 (phenotype represented by either allele) to 1 (phenotype represented by single allele). Allele sizes were highly consistent for ST males for all markers except cnv125 (Table SA3,  .v Amplification success varied across the four loci. For the 211 males examined, 85% for ms395, 93% for comp162710, 89% for cnv395 and 55% for cnv125. Where samples failed to amplify for comp162710 (N = 15), all samples also failed for ms395, indicating minor technical issues because these loci were amplified in a multiplex. In contrast, all 15 amplified for cnv395. Figure SA2: Plot of brood sex-ratios, given as proportion of females, against total offspring. Males are categorised SR (red and pink; P < 0.05, and brood sex-ratio > 0.8) or as ST (dark and light blue), using χ 2 tests on offspring counts greater than 10 testing for significant (P < 0.05) deviations from a 1:1 sex ratio. Males carry either a large (light colour) or small (dark colour) allele for each marker (a, ms395 N = 179; b, comp162710 N = 196; c, cnv395 N = 186; d, cnv125 N = 115).

SI-A3.4 Conclusion
To be of value for identifying individuals which carry X SR , markers must associate with brood sex-ratio and be a reliable predictor of a phenotype category. Here we evaluated four X-linked markers (one microsatellite and three INDEL markers) and found that three of four markers associate with brood sex-ratio, and all have some predictive value (Table SA3). All loci are good at predicting ST (96 -98%), while comp162710, ms395 and cnv395 also reliably predict an SR phenotype (83 -92%). cnv125 is uninformative because whilst almost all SR males have the same sized allele, most ST males also have this allele. We can conclude that cnv125 is not useful for assigning SR and is not a worthwhile marker to be used for laboratory analyses.
Furthermore, markers ms395, comp162710 and cnv395 amplified well, while amplification rates for cnv125 were poor in comparison.