The eyes have it

Article metrics

Females of many species prefer mates with extravagant traits. Studies of stalk-eyed flies show that, in this case at least, such preference is linked to suppression of a selfish gene that influences the sex ratio of offspring.

Peacocks would not have such large and elaborate tails were it not for the fact that peahens prefer their partners to be so ornate. But why are females so choosy? There are two standard answers. Fisher1 suggested that males have elaborate traits purely because females find them attractive. In this case females must be choosy to gain the advantage of passing on the attractive trait to their sons, and so improve the chances of passing on their genes to subsequent generations. Others have conjectured that showy traits indicate the quality of a male2. If only peacocks with ‘good genes’ can afford to bear bigger tails, peahens should mate with them to ensure that their offspring are viable and healthy.

Work by Gerald Wilkinson and colleagues3 on stalk-eyed flies (Diopsidae), described on page 276, provides the strongest evidence to date about the nature of some of the genes females prefer. As their name suggests, stalk-eyed flies have their eyes perched on the end of side-projecting stalks (see the cover of this issue). These can reach ridiculous proportions. In two of the species studied, males have considerably greater eye span than females, and male eye span even exceeds body length. In both of these species females show a strong preference for males with a large eye span. In a third species, male eye span was much shorter (indeed no different to female eye span) and there was no female choice.

Wilkinson's group noticed that, curiously, the two species in which females preferred a large eye span also tended to have populations with more females than males. Closer examination revealed that some, but not all, broods were almost exclusively female. In contrast, the one species without female preference had an even sex ratio and no evidence that individual broods were female biased.

How does the sex-ratio bias come about? Some of the males, it transpires, have an unusual X chromosome4. At the time when sperm are produced, by the process of meiosis, gene(s) on this X chromosome somehow destroy most of the Y-bearing sperm. This is an example of X-linked ‘meiotic drive’ and has been described in several other flies (see box)5,6,7. Most of the progeny of males with the driving X are female as they inherit an X chromosome from both parents (Fig. 1, overleaf). This accounts for the biased population sex ratios.

Figure 1: Meiotic drive in stalk-eyed flies.

A male with a non-driving version of the X chromosome produces as many X-bearing sperm as Y-bearing sperm. On mating with a female, whose eggs are all X-bearing, half the progeny are female (XX) and half male (XY). However, if (as shown here) a male has the driving X, then, by an unknown mechanism, this X kills most of the sperm from the same male that contain the Y chromosome. Typically, males produce many more sperm than are necessary to fertilize the eggs of a female. So the eggs are all fertilized and nearly all inherit the father's driving X chromosome. As a consequence, the progeny are mostly female. If a male has the X-driver and the Y-suppressor, then the action of the X-driver is inhibited and the sex ratio even ends up a little male biased (about 63% male). In the absence of the driving X, 50:50 sex ratios are found, regardless of the type of Y chromosome.

From the X chromosome's point of view, drive is a profitable strategy, as more eggs end up being fertilized by sperm bearing the driving X. But the death of Y-bearing sperm is clearly bad for the Y chromosome. It is no surprise to discover that some males have Y-linked genes that can inhibit the activity of the selfish X chromosome. The Y-suppressor can somehow reverse the direction of drive, leading to male-biased broods.

Could coexistence of female choice and the driving X chromosome be more than just a coincidence? To test this possibility, Wilkinson and colleagues subjected flies to 22 generations of artificial selection on eye span. Remarkably, they found that male eye span covaried with the ability to suppress the selfish X chromosome: males artificially selected for large eye span produced male-biased sex ratios; those selected for short eye span produced female-biased sex ratios. So females choosing males on the basis of their eye span will likewise be selecting a partner's ability to suppress the meiotic-drive gene. The covariance is consistent with the idea that at least some of the genes for long eye span are on the same Y chromosome as the suppressors of drive. Additional breeding experiments (G. Wilkinson, personal communication) reinforce this view.

Why should this linkage come about, and why do females prefer males with a large eye span? In the wild, the driving X chromosome coexists with the Y-suppressor, but not all X chromosomes drive and not all Y chromosomes suppress. Assuming that the frequencies of the four chromosomal types have reached equilibrium, there is by definition no selection in favour of the X-driver or Y-suppressor. However, at equilibrium the population sex-ratio is female biased and so, on average, each male has more offspring than each female. This means that selection favours mothers producing more sons than daughters as they will then have more grandchildren.

Now consider a rare autosomal gene (that is, one not on the X or Y chromosome) acting in females that codes for a preference for males with greater eye span. Will this gene become more common or will natural selection eliminate it? Spread of the gene depends on the linkage, in males, between genes influencing eye span and those for drive suppression. Preference will be favoured if genes for large eye span are on the suppressing form of the Y chromosome because males with the Y-suppressor produce male-biased broods. In contrast, preference for a large eye span is deleterious if genes for big eyes are on the non-suppressing Y chromosome, because males with this chromosome produce female-biased broods. Finally, preference will be neither favoured nor selected against if eye-span genes are located on autosomes, as these eye-span genes have no influence over the sex of offspring. So we can understand both when female preference will evolve and why at least some of the genes for the preferred trait are linked to Y-suppressors of meiotic drive. This model, however, cannot be the whole truth. For one thing, the daughters of males selected for short eyes themselves prefer short-eyed males8.

Wilkinson and colleagues suggest another reason why female preference is favoured. In choosing males resistant to drive, the female is ensuring her sons have high fertility, because males with the Y-suppressor never have half their sperm killed. But this is only half of the equation. The Y-suppressor is probably costly when with a non-driving X chromosome (which would explain why some of the Y chromosomes do not suppress the driving X, and why at equilibrium there is a female-biased sex ratio). So the sons of a choosy female may do well compared to the average male with the driving X, but they do badly when compared with the average male without the driver. These forces will tend to balance out, in which case the effects of male fertility will be relatively unimportant.

Most people will find it surprising that the ‘good genes’ females prefer are not for general viability but for resistance to a selfish gene (see box). Few of those working on sexual selection would have imagined this answer, and few working on meiotic drive would have thought that their work related to the problem of female choice. That said, there is some suggestive evidence from mice9 and flies10 showing that females prefer males without meiotic drive (although in neither case was this preference associated with an extravagant trait, and there are counter examples11).

The fact that the genes appear not to be ones for general viability also upsets our understanding of why females usually prefer seemingly unnecessarily largemale traits. Those espousing the ‘good genes’ argument suppose that such traits are costly and these costs ensure that trait size reflects male quality12. In stalk-eyed flies, however, females do not seem to be assessing general male quality. In principle, the Y-linked suppressor is equally likely to be in a poor-quality male as in a good-quality one. Nonetheless, the traits are large rather than small. So why is size important?

It is also unclear just how general an explanation of female choice this might be. Data5,6 showing that X-linked meiotic drive is more common than previously supposed support the idea that such preferences might be common, at least in flies. And in guppies there is one report of Y-linked variation for sex ratio13. (Is it a coincidence that guppies have numerous Y-linked sexually selected traits14?) But no good evidence of a sex-ratio distorter has been found in birds in which female choice is common. It would nonetheless be worth looking for a sex-ratio effect following artificial selection for trait size.


  1. 1

    Fisher, R. A. The Genetical Theory of Natural Selection (Clarendon, Oxford, 1930).

  2. 2

    Williams, G. C. Adaptation and Natural Selection: A Critique of Some Current Evolutionary Thought (Princeton Univ. Press, 1966).

  3. 3

    Wilkinson, G. S., Presgraves, D. C. & Crymes, L. Nature 391, 276–279 (1998).

  4. 4

    Presgraves, D. C., Severance, E. & Wilkinson, G. S. Genetics 147, 1169–1180 (1997).

  5. 5

    Jaenike, J. Am. Nat. 148, 237–254 (1996).

  6. 6

    Merçot, H., Atlan, A., Jacques, M. & Montchamp-Moreau, C. J. Evol. Biol. 8, 283–300 (1995).

  7. 7

    Hurst, L. D. Genetics 142, 641–643 (1996).

  8. 8

    Wilkinson, G. S. & Reillo, P. R. Proc. R. Soc. Lond. B 255, 1–6 (1994).

  9. 9

    Lenington, S. Adv. Study Behav. 20, 51–86 (1991).

  10. 10

    Wu, C.-I. Genetics 105, 651–679 (1983).

  11. 11

    James, A. C. & Jaenike, J. Anim. Behav. 44, 168–170 (1992).

  12. 12

    Iwasa, Y., Pomiankowski, A. & Nee, S. Evolution 45, 1431–1442 (1991).

  13. 13

    Farr, J. A. Heredity 47, 237–248 (1981).

  14. 14

    Winge, O. J. Genet. 18, 1–43 (1927).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.