Experimental evidence for reduced male allocation under selfing in a simultaneously hermaphroditic animal

Abstract

Self-fertilization is widespread among simultaneously hermaphroditic animals and plants, but is often only facultatively deployed under circumstances that constrain outcrossing. A central prediction of sex allocation (SA) theory is that because exclusive selfing reduces sperm or pollen competition to zero, this should favour extreme economy in resources channelled to the male sex function. We can therefore expect that organisms switching from outcrossing to selfing should reduce their male allocation. However, to date this prediction has received relatively little support in animal taxa, especially compared to plants. Here we show that isolated individuals (under enforced selfing conditions) have a less male-biased SA than do grouped conspecifics (under outcrossing conditions) in the preferentially outcrossing flatworm Macrostomum hystrix. This shift arises from a reduced male allocation (testis area) in isolated individuals, although we did not find any evidence for a re-allocation of these resources to the female sex function (i.e. ovary area was unaffected by selfing/outcrossing conditions). Our results provide some of the clearest experimental evidence to date for reduced male allocation under selfing in simultaneously hermaphroditic animals, extending previous findings comparing SA between populations differing in selfing rates to the level of individual plasticity in gametogenesis.

1. Introduction

Many simultaneously hermaphroditic organisms reproduce via self-fertilization, either routinely, or facultatively under circumstances that prevent outcrossing [1,2]. In both cases, the male–male competition over fertilization typically found under outcrossing is absent. This in theory favours extreme economy in sperm/pollen production, at some minimal level to maintain full fertility, and the re-allocation of spared resources into own female function, resulting in a more female-biased sex allocation (SA) [3,4]. This can be seen as an extreme form of ‘local sperm (pollen) competition’—equivalent to local mate competition in gonochorists [5]—where all competition over fertilization (in this case of own eggs/ovules) occurs between sperm/pollen from the same individual, causing strongly diminishing returns on male investment [6].

In hermaphroditic animals, evidence to date bearing on the prediction of reduced male allocation under selfing is mixed. Johnston et al. [7] found correlative support in a freshwater mussel, Utterbackia imbecillis, reporting that mean SA compared across four populations closely tracked inferred levels of selfing, with male allocation decreasing with increasing selfing, as predicted [3]. This result mirrors the widespread empirical evidence of reduced pollen production under selfing in plants (e.g. [8,9]). By contrast, an experimental study of the marine bryozoan Celleporella hyalina, failed to support the prediction: experimentally imposed self- compared to outcross-matings led to a greater number of ovicells relative to male zooids in only one of eight replicate populations (with variable histories of selfing) [10]. Moreover, a recent experimental evolution study in the freshwater snail Physa acuta found that, although regular selfing purged inbreeding depression and resulted in a reduced waiting time to commence reproduction under selfing—in line with theoretical predictions [11]—there was no evidence for a concomitant reduction in male allocation [12].

Here we adopt an experimental approach to test the prediction of decreased male allocation under selfing in a free-living flatworm, Macrostomum hystrix. We reasoned that if individuals potentially experience conditions allowing or preventing outcrossing (e.g. owing to fluctuation in population density), there may be selection for phenotypic plasticity in SA to optimize male and female investment to fit the prevailing conditions. We already know that outcrossing is the preferred mating system in this species, owing to the fact that isolated individuals only switch to selfing behaviour (self-injecting sperm into their own body) after a significant delay compared to outcrossing, and at some apparent cost, producing fewer, less viable offspring [13,14]. Owing to their transparency, SA can easily be estimated in M. hystrix by measuring the relative size of the paired testes and ovaries, allowing us to test for the predicted shift in SA under selfing. Indeed, there is already a large body of evidence in an obligately outcrossing congener of our study species, M. lignano, that individual flatworms adjust their SA depending on social and mating group size (e.g. [15,16]), another major prediction of SA theory that similarly stems from the fact that as mating group size increases, sperm competition becomes increasingly important to gaining reproductive success through the male sex function [3]. Here we extend the observation of plasticity in SA to a mixed mating species.

2. Methods

To test for differential SA, we performed a group size experiment similar to those conducted previously in Macrostomum. Briefly, subjects were from an inbred line of M. hystrix (designated SR1) that was produced through a total of eight generations of selfing followed by several more of sib–sib breeding from 10/2010–11/2011 (estimated inbreeding coefficient, F = 0.998) [13]. The line is maintained in small laboratory populations, and we began (in April 2016) by pooling around 50 adult individuals to lay eggs. After hatching, a total of 1344 experimental worms were (within 24 h) randomly distributed in 24-well plates containing 6‰ artificial seawater and ad libitum food (Nitzischia curvilineata diatoms), in groups of either one (isolated), two (paired) or eight (octet) individuals. Isolated worms can only reproduce by selfing, whereas paired and octet worms could also outcross. The isolated treatment therefore guarantees no sperm competition, whereas the octet treatment likely represents a high sperm competition intensity (worms generally have high numbers of received sperm when in groups, and the mode of fertilization precludes significant displacement). We would argue the paired treatment likely represents a low (rather than no) perceived sperm competition risk, for two reasons: (i) there is the possibility of competition between own and the partner's sperm and (ii) although we do not know the proximate cues involved, evidence from a congener suggests that worms may not recognize individual mating partners, and might instead rely on simpler cues of sperm competition such as mating rate [17].

Experimental subjects remained in their treatment groups for around four weeks until they were adult, but were transferred to fresh wells after 14 and 21 days to avoid the accumulation of hatchlings of the next generation. Owing to worm mortality, some replicates were lost during the course of the experiment. Wells with missing worms in the isolated and paired treatments were excluded, but for the octets the loss of one or two worms was tolerated (we reasoned that a group size of six or seven worms was still substantially higher than the paired treatment).

When worms were aged 29–31 days (worms mature at ca 18 days, and selfing typically commences in this line at ca 21 days), we then performed standardized measurements of their SA, estimated from the relative size of the paired testes and ovaries. To do so, worms were gradually relaxed using 21 µl 7,14% magnesium chloride and then squeezed to a consistent depth between a microscope slide and coverslip. Images were captured at less than or equal to 400× magnification (Olympus BX50 microscope coupled to a Canon EOS 600D digital camera) and analysed blindly with respect to treatment group, using the software ImageJ (https://imagej.nih.gov/ij/) to estimate body area, testes area and ovaries area [15]. From the latter two measures, SA was calculated as the proportion of reproductive investment devoted to the male sex function, i.e. as (testes area/(testes area + ovaries area)). This measuring protocol results in repeatable estimates of these traits (intra-class correlation coefficients, r_testis = 0.85, r_ovary = 0.6, A. Giannakara 2017, personal communication).

Because it sometimes proved difficult to ‘turn’ the mounted worms so as to obtain suitable images of all four gonads, we produced two datasets: (i) a smaller dataset (n = 105 worms, electronic supplementary material, table S1) for which all four gonad measurements (left and right testis, left and right ovary) could be obtained and (ii) a larger dataset (n = 205, electronic supplementary material, table S2) for which at least one testis and ovary was measured for each subject, with missing measurements extrapolated from the opposite gonad (assuming symmetry). For brevity, we report only results for the smaller dataset, but these are qualitatively unchanged if we instead use the larger one (see electronic supplementary material, figure S1). Statistical analyses were performed in R (v. 3.1.1) using the package multcomp 1.4–4 to calculate Tukey contrasts.

3. Results

There was a significant effect of treatment (mating regime) on SA (F_2,102 = 7.82; p = 0.0007), with worms becoming increasingly male-biased in their SA in larger groups (figure 1a). The difference between worms from pairs and octets was not significant (t = 0.90; p = 0.64), but isolated worms exhibit significantly less male-biased SA compared to both of the other two groups (versus paired: t = 3.24; p = 0.004; versus octet: t = 3.66; p = 0.001). This result stems from a significant reduction in combined (log-transformed) testis area in isolated worms (versus paired: t = 3.02; p = 0.009, versus octets: t = 5.98; p < 0.001; figure 1b), as well as a significant difference between pairs and octets (t = −3.67; p = 0.001; figure 1b); this latter result implies that sperm competition level under outcrossing is also important in determining allocation to sperm production. By contrast, there were no significant differences between treatment groups in combined (log-transformed) ovary area (F_2,102 = 2.114; p = 0.13, figure 1c). Despite some evidence for a marginally significant increase in body area in octets (electronic supplementary material, table S3), all conclusions for testis area, ovary area and SA remain unchanged if body area is included as a covariate (electronic supplementary material, tables S4 and S5).

Figure 1.

SA plasticity under selfing versus outcrossing conditions in Macrostomum hystrix. The boxplots depict (a) SA, (b) testis size and (c) ovary size of M. hystrix measured under three different social environments differing in the opportunity for outcrossing and sperm competition level. See main text for details and test statistics. (Online version in colour.)

4. Discussion

Previous observations in M. hystrix have suggested that this species is a preferential outcrosser capable of switching to self-fertilization in the extended absence of mating opportunities [13]. Our results extend these findings, by indicating that this switch to self-fertilization is accompanied by a reduced allocation to the male sex function, in both absolute and relative terms. This represents, to our knowledge, some of the clearest experimental support to date for this aspect of SA theory in animals [3,6].

The ability to adjust SA—as either a phenotypically plastic or evolutionary response—is likely adaptive under various scenarios, and various flatworm species have been found to respond to such factors as resource availability (e.g. [18]), mating group size (e.g. [16]) and shifts in mating system from outcrossing to both parthenogenesis (e.g. [19]) and selfing ([20], this study). A reduced allocation to male function was also previously observed in isolated individuals of the congener M. lignano, but because this species is an obligate outcrosser, this was interpreted with respect to the reduced perceived sperm competition level in isolated worms [21], and it was suggested that worms might simply conserve investment in sperm production when they cannot mate; this explanation alone seems less likely in M. hystrix, because even isolated worms can reproduce by selfing. Nevertheless, the responses of these two species are in some sense comparable, because selfing is ultimately a special case leading to zero sperm competition. We also note that plasticity was observed despite the fact that selection to flexibly increase male allocation had likely been relaxed during the several generations of selfing that generated the inbred line we used in our experiment. Plasticity in SA was similarly maintained after many generations of relaxed selection due to monogamy in the congener M. lignano [22].

In principle, we might have expected that any resources conserved through not investing into the male function would immediately have been channelled into the female function, with worms continuously producing eggs as adults. Somewhat surprisingly, however, the reduced male allocation we observed was not accompanied by any obvious increase in female allocation, at least as captured by ovary area. Although this contradicts the central assumption of an SA trade-off in hermaphrodites [3], two important caveats need to be borne in mind. First, ovary area is a morphological proxy for female reproductive allocation that does not necessarily capture well the dynamic process of egg production itself, and probably also underestimates this parameter because much of egg development actually occurs outside the ovary. This last point likely also explains why our mean SA estimates exceed 0.5, which, theoretically, is not normally expected [3]. Second, we performed our experiment under ad libitum feeding conditions, which may mask any underlying trade-offs (cf. [23]).

Data accessibility

Raw data are provided in the electronic supplementary material.

Authors' contributions

L.W. and S.A.R. designed the experiment and drafted the manuscript; L.W. conducted the experiment and analysed the data. Both authors approved the final version of the manuscript, and both authors agree to be held accountable for the content therein.

Competing interests

We have no competing interests.

Funding

This study was funded by research grant RA 2468/1-1 from the Deutsche Forschungsgemeinschaft (to S.A.R.).