Offspring sex ratios are stable across the life course in Drosophila simulans

Abstract

Within populations, adult sex ratios influence population growth and extinction risk, mating behaviours and parental care. Sex ratio adjustment can also have pronounced effects on individual fitness. Accordingly, it is important that we understand how often, and why, offspring sex ratios deviate from parity. In Drosophila melanogaster, females appear to improve their fitness by producing fewer sons when paired with older males. However, facultative sex ratio adjustment in D. melanogaster is controversial, and our understanding of how sex ratio skew affects fitness is hampered by pronounced sexual conflict in this species. Additionally, it is unclear whether maternal age or quality interacts with paternal age to influence offspring sex ratios. Here, we test whether offspring sex ratios vary as a function of maternal quality, and maternal and paternal age in Drosophila simulans, a sister species of D. melanogaster that lacks overt sexual conflict. We find that offspring sex ratios are slightly male‐biased overall, but constant across the female life course, and independent of female quality, or paternal age. To really understand if, how and when females skew offspring sex ratios, we need studies linking offspring sex ratios to paternal and maternal phenotypes that are predicted to shift optimal investment in sons and daughters.

Abstract

Drosophila simulans offspring sex ratios do not depend on paternal age (a), maternal age (b) or quality (c).

Open in new tabDownload slide

INTRODUCTION

The reproductive value of the rarer sex is greater than the reproductive value of the abundant sex (Fisher, 1930). That is, in a population with skewed sex ratios the rarer sex is more likely to secure mates and produce more per capita offspring than the abundant sex. This negative frequency‐dependent selection should mean that at a population level, offspring are generally produced at around a 1:1 sex ratio (Fisher, 1930). However, some population sex ratios deviate from equality. This includes large cooperative societies of spiders that are often female‐biased (Frank, 1987) and human populations where slightly more sons are produced than daughters (Dyson, 2012). Population‐level sex ratio skew can have pronounced and far‐reaching evolutionary and ecological consequences, affecting population dynamics (Hornett, Charlat, Wedell, Jiggins, & Hurst, 2009) and male and female mating behaviour (Jirotkul, 1999; Liker, Freckleton, & Székely, 2014). However, the relationship between population sex ratio and individual sex ratio adjustment is not straightforward (Frank & Swingland, 1988). Even when populations show gender parity, individual females can produce sex‐skewed broods or clutches, and this can increase dam fitness (e.g. Ellegren, Gustafsson, & Sheldon, 1996).

Enhanced fitness through skewing offspring sex ratios can occur when any factor predictably but differentially affects the fitness of sons versus daughters (Booksmythe, Mautz, Davis, Nakagawa, & Jennions, 2017; Cockburn, Legge, & Double, 2002; West, 2009). Maternal condition, where condition reflects how well individuals acquire environmental resources (Rowe & Houle, 1996), is one such factor that has been widely studied (i.e. the Trivers–Willard effect: Trivers & Willard, 1973). If maternal condition influences offspring quality, and if the fitness returns of producing high (or low)‐quality sons versus high (or low)‐quality daughters are unequal, then skewed offspring sex ratios are predicted (Trivers & Willard, 1973). Differential effects of maternal condition on offspring fitness are likely to be widespread. This is because sons tend to have more variance in reproductive success than daughters, and concurrently, the reproductive success of daughters is less dependent on their intrinsic quality (Bateman, 1948). Accordingly, high‐quality sons typically produce more offspring than high‐quality daughters, whereas low‐quality sons produce fewer offspring than low‐quality daughters. Therefore, females that produce high‐quality sons will have higher expected fitness than if they produced high‐quality daughters (all else being equal), whereas females that produce low‐quality sons will tend to have lower fitness than females that produce low‐quality daughters. Thus, when female condition affects offspring quality (e.g. high‐condition females produce larger, heavier offspring), females in good condition should produce excess sons and females in poor condition should produce excess daughters (Trivers & Willard, 1973). This prediction has received support from meta‐analyses of ungulates, although the relationship between maternal condition and son production is weak (Sheldon & West, 2004).

Maternal condition is not the only factor that affects optimal offspring sex ratios; the attractiveness and viability of fathers matter too. If females mate with high‐quality males (i.e. attractive or viable sires), and male quality positively influences offspring quality, females may do better to skew offspring sex ratios towards sons. For example, Burley (1981) showed that zebra finch (Poephila guttata) offspring sex ratios varied depending on the attractiveness of the coloured leg band worn by their fathers, with more attractive colours being associated with greater son production. More recently, meta‐analysis found that females that mated to more ornamented or larger males produced more sons than females that mated with less ornamented or smaller males (Booksmythe et al., 2017).

Given the potential impact of paternal quality on optimal offspring sex ratios, some research has explored how paternal age should affect sex ratio skew. Ageing can result in the accumulation of deleterious germ‐line mutations; increased human paternal age is associated with epigenetic changes, DNA mutations and chromosomal aneuploidies (Sharma et al., 2015). Accordingly, paternal age could affect siring quality and, in turn, optimal offspring sex ratios. In support of this idea, younger male aye‐ayes (Daubentonia madagascariensis) sire more sons than older males (Tanaka, Fukano, & Nakamura, 2019), and in humans, fathers aged < 25 are more likely to produce sons than fathers aged 25–34 (Khandwala et al., 2018). However, assessed across species paternal age does not generally affect offspring sex ratios (Booksmythe et al., 2017).

Some of the strongest evidence for sex ratio distortion with paternal age comes from Drosophila melanogaster, where females mated to old mates produce more daughters than sons (Long & Pischedda, 2005; Mange, 1970; Yanders, 1965). This shift may occur because ageing male D. melanogaster accumulate mutations and sons that inherit deleterious mutations suffer greater fitness reductions than daughters that inherit those mutations (Pischedda & Chippindale, 2005). In support of this idea, the sons of old D. melanogaster males are less fit than the sons of young males, suggesting that this sex ratio skew is adaptive (Long & Pischedda, 2005). However, these results are controversial—not all work manipulating paternal age find differences in sex ratios of D. melanogaster offspring (e.g. Mossman, Mabeza, Blake, Mehta, & Rand, 2019) or find declines in the quality of sons with increased paternal age (Han, 2014).

Evidently, there is mixed support across species that paternal age influences offspring sex ratios, and this is reflected in D. melanogaster studies. Some of this inconsistency might reflect that D. melanogaster experience pronounced intralocus sexual conflict, such that high‐quality mothers produce lower quality sons and higher quality daughters (Chippindale, Gibson, & Rice, 2001). When intralocus sexual conflict is present, we might expect that low‐quality mothers overproduce sons as these would be of higher fitness—a prediction that has received support in the broad‐horned flour beetle Gnatocerus cornutus (Katsuki, Harano, Miyatake, Okada, & Hosken, 2012). If genes improving female condition have sexually antagonistic effects, these effects complicate predictions about adaptive offspring sex ratio adjustment by dams (Katsuki et al., 2012). A more general complication is that even if mothers bias the sex ratio of their offspring, there may be only minor deviance from gender parity (Booksmythe et al., 2017), meaning that large sample sizes may be needed to detect sex ratio variation. There are also signs that negative results are not being published, leading to a skewed perspective of how common facultative sex ratio adjustment is (Booksmythe et al., 2017; Festa‐Bianchet, 1996). Given that variation in offspring sex ratios can affect maternal fitness, and population sex ratio has a pronounced impact on population dynamics (Lee et al., 2011) and mating behaviour (e.g. Liker et al., 2014), it is important that we better understand how, and when, dams bias the sex ratios of their offspring.

Here, we extend experimental assessment of adaptive sex ratio allocation to test whether females alter offspring sex ratios. Because sexual conflict can complicate predictions of sex ratio skew theory including the Trivers–Willard hypothesis, we used D. simulans, a close relative of D. melanogaster that does not display pronounced intralocus sexual conflict (Duffy et al., 2019). First, we mated young females to young and old males to test whether females adjusted offspring sex ratios to discriminate against the sons of old mates—effectively mirroring previous work on D. melanogaster (Long & Pischedda, 2005; Mange, 1970; Yanders, 1965). Then, we followed a cohort of females from adult eclosion until death and determined whether the sex ratio of offspring varied over the life course as dams were repeatedly mated to young males. Here, we tested whether maternal age influenced offspring sex ratios—as there is evidence in species including red‐winged blackbirds (Agelaius phoeniceus) (Blank & Nolan, 1983), mountain goats (Oreamnos americanus) (Côté & Festa‐Bianchet, 2001) and reindeer (Rangifer tarandus) (Weladji, Holand, Yoccoz, & Lenvik, 2003) that older mothers tend to produce more sons than daughters. Finally, we tested whether female residual reproductive value and lifetime reproductive success affected offspring sex ratios to test whether high‐quality mothers produce more sons—we phrase the discussion here in the light of maternal quality rather than condition per se, as the Trivers–Willard (1973) predictions also apply in this more general case. Additionally, quality and condition are linked—high‐quality females will be in better condition. To our knowledge, this is the first study to test how male age, female age and quality shape offspring sex ratios. We found a slight bias towards son production, but no evidence for facultative sex ratio adjustment as a function of parental age or maternal quality.

MATERIALS AND METHODS

Stocks

In April 2010, circa 100 wild‐type D. simulans females were collected from Athens, Greece, and used to establish isogenic female lines (hereafter isolines) at the University of Exeter, Cornwall, UK. These lines were maintained via full‐sib matings with nonoverlapping generations for 5 years prior to the beginning of this experiment. These isolines were used to establish a mixed population cage, which was maintained at ca. 500 individuals with overlapping generations for ca. 6 generations prior to establishing flies for experiments. Experimental flies were collected from this large stock population by providing laying vials (48 × 116 mm) containing an excess of Jazz‐Mix food (Fisher) in which females could lay eggs. These vials were incubated at 25°C under a 12:12‐hr light:dark cycle and monitored for eclosions. Focal experimental flies were collected from these vials under light CO₂ anaesthesia within 8 hr of hatching to ensure that they were virgins and immediately housed individually in 40‐ml vials with 8 ml of Jazz‐Mix food prior to experimentation.

Does male age affect offspring sex ratios?

We mated young females (average age 4 days post‐eclosion ± 0.08 standard error) to young (5 days post‐eclosion) and old (28 days post‐eclosion) males. These ages were chosen because work has shown that males aged 28 days are senescing—they exhibit pronounced functional performance declines (K.L Delaney, D.J. Hosken, L.M. Gill, A. Sutter, and C.R. Archer, unpublished data). To do this, virgin females and males were collected from the outbred experimental population and housed individually as described above until they reached the desired assay age. The day before a mating assay, females were moved to a clean 40‐ml vial with 8 ml of Jazz‐Mix food (vial A). At 9 a.m. the following day (when incubator lights came on), a single experimental male was added to these vials. Pairs were left in an incubator until 6 p.m., when both males and females were removed. Females were transferred into a new vial (vial B), containing 8 ml of Jazz‐Mix food, whereas males were transferred into an Eppendorf and frozen. Females were left in this second vial (B) for 72 hr, before being moved to a third vial (C) for a further 72 hr; finally, females were moved into a fourth vial (D) for 24 hr. This means that females were allowed seven days to lay eggs. Egg vials were left to incubate and monitored daily for eclosion. Seven days after the first eclosion was observed in each vial, vials were frozen and male and female offspring were counted at a later date. If the sex of offspring could not be determined (e.g. they had died prior to freezing and so were hard to accurately sex), their sex was recorded as ‘unknown’. Only flies of known sex were used to calculate sex ratios; however, unknown sex offspring were included to calculate lifetime reproductive success to use as a covariate in some analyses (please see section 2.4 below). Please note that of 11,478 offspring produced in this experiment, only 36 could not be identified by sex. After accounting for female deaths before all offspring could be collected, handling deaths and escapes, we assayed 182 females (99 mated to young males and 83 mated to old males) spread across three experimental blocks that were initiated within a week of one another. Note, 155 of 182 females produced offspring in this experiment (88 mated to young males and 67 mated to old males).

Does maternal age and quality affect offspring sex ratios?

Female offspring sex ratios were monitored over each female's lifetime, but the age of sire was consistent (2–6 days old). Males and females were initially collected and housed as described above. After 4 days, female flies were moved into a new vial of clean food. The following morning, two stock males were added and kept with the females for 3 hr. Females were subsequently exposed to males for 3 hr every 5 days. This treatment schedule appears to allow females to reproduce without being sperm‐limited but does not reduce female survival (Taylor, Wigmore, Hodgson, Wedell, & Hosken, 2008). The only difference between our treatment and the procedure in Taylor et al., (2008) is that our males were not age‐matched with the females, but rather, we aimed to collect tester males such that they were around 5 days old when provided to a female. Once females were removed from vials, vials were treated as above (i.e. incubated and offspring counted). The same females that were assayed for age‐dependent reproductive success were also monitored for survival daily. After accounting for escapes and handling deaths, we had 84 females monitored across the life course in two experimental blocks, established on two sequential days. Three of these females did not produce any offspring. Of the 20,452 offspring produced during this experiment, only 27 could not be identified by sex.

Statistics

All analyses were conducted in R version 3.6.1 (R Core Development Team, 2017). To test whether male age affects female offspring sex ratios (Experiment 1), a proportion test was used. Next, a binomial generalized linear model was used with offspring sex as the response variable (cbind(number_of_male_offspring,number_of_female_offspring)) with male age (factor, two levels) and block (factor, three levels) and interactions between them as explanatory variables. There was no over‐dispersion in this model.

A proportion test was used to determine whether the total proportion of daughters (i.e. all daughters produced across each female's lifetime, pooled across females) was equal to 0.5. To see whether this ratio was stable across the life course or not, a binomial generalized mixed model was used with offspring sex as the response variable (as above) with female residual reproductive value (lifetime reproductive success minus cumulative reproductive success at each age), age, block and interactions between them as explanatory variables. Because explanatory values were on different scales, the R ‘scale’ function was used to scale and centre ‘age’ and ‘residual reproductive value’. Centring was achieved by subtracting column means from each value of the corresponding explanatory variable, values were then scaled by dividing the centred explanatory variable by their corresponding standard deviation. Female ID was included as a random effect, and the model was not over‐dispersed.

An additional model to test how maternal quality (here measured as lifetime reproductive success) affected offspring sex ratios had the same basic structure as the female age‐dependent model, but here, sex ratios over the entire life course were pooled and lifetime reproductive success (scaled as above) was included as an explanatory variable, together with block, and an interaction between the two. Here, over‐dispersion was observed, and so, an observation‐level random effect was added to the model (Harrison, 2014). In all cases, significance was assessed using the ‘ANOVA’ package in ‘car’ (Fox & Weisberg, 2018) to avoid the ‘winner's curse’ (elevated risk of false positives due to overestimation of effect sizes in backward model simplification) (Forstmeier & Schielzeth, 2011).

RESULTS

Does male age affect offspring sex ratios?

Females mated to young males produced an average of 65.04 offspring (±4.07 SE), whereas females mated to old males produced 60.71 offspring (±4.62 SE). A proportion test that compared the total number of sons and daughters produced by all females showed that the proportion of daughters produced by young females did not deviate from 0.5, irrespective of whether fathers were young (3,217 daughters, 3,218 sons, X₂ = 0, df = 1, p = 1.0) or old (2476 daughters, 2531 sons, X₂ = 0.582, df = 1, p = 0.45) (Figure 1). Next, a binomial model showed that the proportion of offspring produced by each female was not affected by age—averaging across females, 49.2% (± 5% SE) of offspring sired by young fathers were daughters, whereas 49.8% (± 5% SE) of offspring sired by old fathers were daughters (offspring sex ratio effects: male age—X₂ = 0.206, df = 1; p = 0.650: block—X₂ = 0.160, df = 2; p = 0.924: block‐by‐age interaction—X₂ = 1.031, df = 1; p = 0.31—all nonsignificant). Please note that the numbers reported for the proportion test are the sums of all offspring produced in each treatment group, whereas the percentages are son and daughter production averaged across females in each treatment group.

Do maternal age and quality affect offspring sex ratio?

Median female lifespan was 49.36 days, and females produced 243.5 ± 12.38 (mean ± SE) offspring over their lives. The proportion of daughters produced by all females deviated from 0.5, as slightly more sons were produced than daughters overall (10026 daughters, 10399 sons, X₂ = 6.78, df = 1, p = 0.01). Averaging across females, 49.41% ± 0.01% (mean ± SE) of all offspring were daughters and offspring sex ratios did not depend on female quality (measured as residual reproductive value) (X₂ = 0.010, df = 1, p = 0.921), age (X₂ = 0.150, df = 1, p = 0.699) or the interaction between the two (X₂ = 0.018, df = 1, p = 0.893). Experimental block did not affect offspring sex ratios singly or via any interaction terms (p > 0.1) (Figure 2). In the model, where sex ratios were summed over the life course and female lifetime reproductive success was included as a proxy for female quality rather than female residual reproductive value, neither female quality (X₂ = 0.358, df = 1, p = 0.551), block (X₂ = 0.667, df = 1, p = 0.414) or an interaction between the two (X₂ = 2.458, df = 1, p = 0.117) affected sex ratios (Figure 3).

DISCUSSION

Offspring sex ratios often deviate from gender parity, but the extent of facultative sex ratio adjustment and its drivers remain controversial (Booksmythe et al., 2017; Festa‐Bianchet, 1996). Although there are many examples of skewed offspring sex ratios as a function of maternal condition (Sheldon & West, 2004) or paternal age (Long & Pischedda, 2005; Mange, 1970; Yanders, 1965), there appears to be a pronounced ‘file‐drawer’ effect where data showing the lack of skewed sex ratios are under‐represented in the literature (Booksmythe et al., 2017; Festa‐Bianchet, 1996). This makes it hard to understand whether sex ratio adjustment is common or not. Here, at the population level there was an excess of sons in our lifelong female dataset but individual mothers did not adjust the sex ratio of offspring depending on their own age or the age of their mate. Further, offspring sex ratios were independent of female residual reproductive value or lifetime reproductive success—our measures of female quality.

Interestingly, our results do not mirror findings from D. melanogaster (a close relative of D. simulans) where females mated to old sires produced more daughters than sons (Long & Pischedda, 2005; Mange, 1970; Yanders, 1965). Instead, we found that young D. simulans females mated to young males (5 days old) produced 49.2% daughters and 50.8% sons, whereas young females mated to old males (28 days old) produced 49.8% females and 50.2% males. Sex ratios were equal and independent of paternal age. There are several possible reasons for the disparity between studies. Foremost, we used D. simulans rather than D. melanogaster, and there are many differences between these species (Taylor, Sharma, & Hosken, 2009). Although this is likely not the whole explanation as Mossman et al. (2019) also failed to detect shifts in D. melanogaster offspring sex ratios as a function of paternal age, and Han (2014) did not detect a reduction in the fitness of D. melanogaster sons as their fathers aged—suggesting that a shift towards producing more daughters with aged males might not be adaptive. Our failure to detect an effect of paternal age could be a power issue—sample sizes were higher in previous work and any effect sizes of sex ratio skews are likely to be small (Booksmythe et al., 2017). Although this is a limitation of the current work, Mange (1970) found that females mated to 13‐day‐old males produced 44.7% male offspring—this evidently represents a sizeable skew that we quite clearly do not detect in the current work even though our males were much older and D. simulans males are senescing by 28 days (K.L Delaney, D.J. Hosken, L.M. Gill, A. Sutter, and C.R. Archer, unpublished data). Furthermore, both Mange (1970) and Long and Pischedda (2005) found that female sex ratios became female‐biased when females were mated to old males—a result we do not see here where offspring remain slightly male‐biased in both male age categories. Interestingly, Mange (1970) only found that females produced excess daughters until their mates were approximately 13 days old, after which point sex ratios deviated in an unpredictable manner as paternal age rose to ca. 26 days. If sex ratio adjustment is adaptive because harmful mutations accumulate in ageing male Drosophila that have a disproportionately large impact on sons (Pischedda & Chippindale, 2005), it is hard to understand why skewed sex ratios cease with 13‐day‐old sires given that mutation loads continue to increase with age (Garcia et al., 2010).

To better understand how sex ratios might shift over the life course, we followed a cohort of females and tracked their reproductive success and offspring sex ratios, predicting that as females grew older (but continued mating with young males) we might see a progressive skew towards offspring of one sex. In theory, old females may skew towards either sex—outcomes depend on whether daughters bear a disproportionate cost of having an old mother, in which case we might see older females producing more sons. Alternatively, if both sons and daughters accrue equal costs, then we might see a shift towards old mothers producing more daughters whose reproductive success is less dependent on maternal quality. At the population level, slightly more sons were produced than daughters in this experiment. Within individual females, there were weak indications that older females might skew offspring sex ratios as once females were over 50 days old, there was a slight shift towards son production. However, the effect of maternal age was not statistically significant. The power of our sample, which sacrificed large sample sizes (Figure S1) in favour of high‐resolution data, might hinder our ability to detect a signal here. But if shifts in sex ratio only occur at these advanced ages, likely to exceed the maximum life expectancy of Drosophila in the field (Powell, 1997), it is hard to imagine that these effects have particularly pronounced demographic impacts on populations.

We expect parental age to influence maternal investment in offspring because older parents may produce lower quality young and age‐dependent reductions in offspring quality could have differential fitness effects on sons versus daughters. Declines in offspring quality as a function of parental age could be due to declines in maternal condition (i.e. resources available to invest after accounting for general maintenance) or declines in paternal viability, noting that somatic mutations accumulate over the lifetime of D. melanogaster (Garcia et al., 2010) and some mutations have costly effects on offspring fitness that differ in magnitude across the sexes (Pischedda & Chippindale, 2005). We also generally expect offspring sex ratios to be male‐biased for high‐quality mothers if they produce higher quality sons (Trivers & Willard, 1973). However, we found no evidence that residual reproductive value influenced offspring sex ratios over female age, and there was no relationship between female lifetime reproductive success and offspring sex ratio, suggesting that the sex ratio of offspring is unrelated to female quality.

In summary, although it appears that D. simulans females may produce slightly more sons than daughters at the population level, we found no support for individual adaptive sex ratio allocation in D. simulans as a function of parental age or maternal quality. Although we are happy to concede that power might be a limiting factor in our experiments, we do not see trends in the direction predicted by theory. There were some signs that at very advanced ages, females might produce more sons but more work is needed to really test this idea and effects were only seen at ages far greater than D. simulans are likely to reach in the field. This means that their ecological significance is likely to be minor, particularly relative to other factors influencing sex ratios in offspring (e.g. the widespread sex ratio distorting parasite Wolbachia: Hornett et al., 2009). Our results highlight the need for additional studies in Drosophila to resolve uncertainty with regard to sex ratio biasing. Furthermore, some theoretical clarification would be helpful. Crucially, we need more studies that link parental traits, such as age and condition, to offspring fitness—without these data, we cannot make strong a priori predictions about when sex ratio adjustment may be adaptive. Moreover, although it is clear that age could influence optimal sex ratios, it is unclear why existing data only show linear impacts of age in females mated to relatively young sires (Mange, 1970). The effects of intralocus sexual conflict also need to be more deeply incorporated into thinking about adaptive on sex ratio adjustment. Finally, the publication bias in the field of adaptive sex ratio adjustment does not appear to have been wholly resolved in the 20 years (Booksmythe et al., 2017) after it was first flagged (Festa‐Bianchet, 1996), highlighting the need for more even reporting of experimental outcomes. Without this, our understanding of sex ratio distortion will remain skewed.

ACKNOWLEDGMENTS

We thank all members of the Hosken and Wedell laboratory for useful discussion and assistance in the laboratory.

Peer Review

The peer review history for this article is available at https://publons.com/publon/10.1111/jeb.13696.

REFERENCES