Climbing Mount Improbable

by Richard Dawkins

  Winged males, then, can prosper only if there are some figs that happen to have female wasps but no wingless males. How likely is this? It depends upon how densely populated the wasps are, relative to the figs. And it also depends upon the fraction of males that have wings. If the wasps in general are rare compared to figs, wasp eggs will be few and far between, and you can hope to find at least some figs that have females only. Winged males will do relatively well under these conditions. Now look what happens if the wasp population is high. Most of the figs will have several wasps in them, of both sexes. Most of the females will be mated by wingless males before they leave their figs, and the winged males will do badly.

  Hamilton did these calculations more precisely. He concluded that if the average number of male eggs per fig is greater than about three, winged males will almost never reproduce. At all densities higher than this, selection will favour winglessness among the males. If the average number of male eggs per fig is one or fewer, wingless males will do badly because they'll almost never find themselves sharing a fig with another wasp, let alone a female. Natural selection under these conditions will favour wings among males. At intermediate population densities, the stable balance theory comes into play and natural selection favours a mixture of winged and wingless males in the population.

  Once we have the stable balance theory in action, natural selection favours whichever male type is in the minority, or rather, whichever male type exists at less than the critical frequency, whatever the critical frequency is. We can then say as a shorthand that natural selection is favouring the critical frequency. As for the critical frequency itself, it will vary from species to species, depending upon the absolute density of wasps compared to figs. We can think of different species of wasp, with their various wasp/fig densities, as like different thermostatically controlled rooms. Each room has its thermostat set to a different {316} temperature. For instance in a species where the average number of male eggs per fig is three, natural selection favours a mixture of males in the population with about 90 per cent of them wingless. In a species where the average number of males per fig is two, natural selection favours a mixture of males with about 80 per cent of them wingless. Remember that the condition we are talking about is an average of two males per fig. This does not mean that every fig has exactly two males. This average of two is made up of some figs with no males, some with one, some with two and some with more than two. The 20 per cent winged males make their genetic living not from the figs with two males (whose females are likely to be snapped up before they leave the fig) but from the figs with no males at all.

  What is the evidence, in practice, that the stable balance theory is working among these wasps, rather than the best of a bad job theory? The key difference between the two theories is that on the first theory, but not the second, both kinds of male wasp should do equally well. The Hamiltons found evidence suggesting that the two kinds of male really do succeed equally well in mating with females. They looked at ten different species. They found that, in all species, the proportion of winged to wingless males was approximately equal to the proportion of females that left their natal figs unmated. Thus a species in which 80 per cent of the females flew out of their natal figs unmated, was also a species in which 80 per cent of the males were winged. A species in which 70 per cent of the females were mated before leaving their natal figs was also a species in which 70 per cent of the males were wingless. It really does look as though the proportions of the two kinds of males are exactly what they should be to ensure an equal share of females all round. This is evidence for the stable balance theory and against the best of a bad job theory. Sorry it was so complicated, but that is par for the course in the world of figs.

  Let's leave the freeloader wasps and return to the true fig wasps, the bona fide, specialist fig pollinators. If you think the freeloader saga was complicated, get ready for this, my final story. I take some pride in trying my hardest to explain difficult and complicated matters but the following may defeat me. Let me do my best and, if I fail, {317} blame the figs and their partners the wasps. Or, don't blame them or me but credit evolution for the subtle wonder of this complex dance through evolutionary time. It takes some work to follow these final pages of the book but I like to hope that it is worth the effort.

  The figs in which the following story is set are ‘dioecious’. This means that, instead of having one kind of fig containing both male and female flowers like the ‘monoecious’ trees we've so far talked about, there are male trees and female trees. Female trees produce figs containing only female flowers. Male trees make figs containing male flowers, but that isn't all. These so-called male figs also contain pseudo-female flowers and this is very important for the wasps. Unlike the real female flowers in the female figs, the pseudo-female flowers in the male figs cannot set seed, even if they are pollinated. What they are good at is providing food for baby wasps, and — thereby will hang a tale — they need to be pollinated before they will do this. The fertile female flowers in the all-female figs are a genetic graveyard for wasps though they are vital for fig reproduction. Female wasps enter them and pollinate them, but their eggs cannot grow in them.

  Here we have the elements of a rich strategic game which we can describe in terms of the ‘wants’ (in the special Darwinian sense) of the various players. Both male and female figs ‘want’ wasps to enter them, but the wasps want to enter only male figs — and then only because of the nutritious pseudo-female flowers within them. The male tree wants eggs to be laid in its pseudo-female flowers so that the female young that hatch out will then load themselves up with the male fig's pollen and fly off with it. The tree has no direct interest in male wasps hatching inside its figs because male wasps don't transport pollen. This may seem surprising since male wasps are, after all, necessary for the continuation of the race of fig wasps. We people, with our penchant for looking ahead and thinking of the larger consequences of our actions, find it hard to purge our minds of the idea that natural selection too looks ahead. I have already made this point in another connection. If natural selection were capable of taking the long view, animals and plants would take steps to preserve the race — their own and those that they depend on like their prey and their pollinators. But nature, unlike humans with {318} brains, has no foresight. ‘Selfish genes’ and short-term benefit are always favoured in a world where others are coping with the long-term needs of the race. If an individual fig tree could get away with fostering nothing but female wasps it would do so, relying on other fig trees to produce the males needed to preserve the race of wasps. The point is that, as long as the other trees are producing male wasps, a single selfish rebel tree that discovered a way to increase its production of female wasps, and hence increase the amount of pollen that it could export, would have an advantage. As the generations went by, more and more trees would become selfish, relying on fewer and fewer trees to produce the needed male wasps. Finally, the last tree with a penchant for nurturing male wasps would die because it would still be doing less well than its rivals producing only female wasps.

  Fortunately, it appears that fig trees have no control over the sex ratio of the wasps reared in their figs. If they could control it, it is likely that male figs would disappear and the race of fig wasps would die out. The fact that the race of fig trees would then also die out would be just too bad. Natural selection cannot look that far ahead. The reason fig trees have no control over wasp sex ratios is probably that the wasps, who also have an interest in controlling their own sex ratio, have overriding power.

  A female fig tree also wants (again in the special Darwinian sense) female wasps to enter its figs, otherwise its female flowers won't be pollinated. A female wasp wants to enter male figs, because only there will she find the pseudo-female flowers in which her larvae can grow. She wants to avoid female figs like the plague because once she has entered one she is genetically dead. She will have no descendants. Expressed more strictly, genes that make wasps enter femal
e figs will not be passed on to future generations. If natural selection were working on wasps alone, the world would become full of wasps that discriminate against female figs and in favour of male figs with their lovely pseudo-female egg-cosies.

  Once again, we humans want to interrupt and say: ‘But surely the fig wasps should want some of their number to enter female figs because, though these may be graveyards for individual wasp genes, they {319} are vitally important for the continuation of the race of fig trees. If the race of fig trees goes extinct, the race of fig wasps would soon follow them.’ This is the exact mirror image of our previous argument. Given that some fig wasps are foolish enough, or altruistic enough, to enter female figs, natural selection will favour a selfish individual wasp that discovers how to avoid female figs and enter only male figs. Selfishness among wasps is bound to be favoured over any public-spirited tendency to work for the continuation of the race. So why don't fig trees and fig wasps go extinct? Not because of altruism or foresight, but because selfishness on each side of the wasp/fig divide is prevented by selfish counter-measures on the other side. What prevents female wasps from doing the selfish thing and avoiding female figs is direct action taken by the fig trees themselves to thwart would-be selfish wasps. Natural selection has favoured deceptive tactics by female figs, making them become so like male figs that the wasps can't tell the difference.

  So, our game between wasps and figs has a fascinating symmetry. There are opportunities on both sides for individuals to be selfish. If either of these two selfish impulses succeeded, both the wasps and the fig trees might go extinct. What stops this happening is not altruistic restraint, nor ecologically aware foresight. What stops it happening is direct police action by individual players on the other side in each case, acting in their own selfish interest. Fig trees would, if they could, abolish male wasps, incidentally thereby ensuring the wasps’, and their own, extinction. They are prevented from doing so by the wasps, who have an interest in rearing both male and female wasps. Fig wasps would, if they could, avoid entering female figs and incidentally thereby ensure the trees’ and their own extinction. They are prevented from doing so by the trees, who make it difficult to tell the difference between male and female figs.

  To summarize so far, we can expect both male and female fig trees to do all in their power to lure wasps into their own kind of fig. And we can expect the wasps to struggle to tell the difference between male and female figs, to enter the first and shun the second. Remember that ‘struggle’ means that, over evolutionary time, they will come {320} to possess genes that confer a predilection for male figs. More contortedly, we shall also find that both male and female fig trees should have an interest in fostering wasps that enter figs of the other sex. In this difficult argument, I am following a brilliant paper by two British biologists: Alan Grafen, one of modern Darwinism's foremost mathematical theorists, and Charles Godfray, a leading ecologist and entomologist.

  What weapons do fig trees have in playing their game of strategy? Female trees can make their figs look, and smell, as much like male figs as possible. Mimicry, as we saw in earlier chapters, is a common phenomenon in the living kingdoms. Stick insects resemble inedible sticks and are therefore ignored by birds. Many palatable butterflies resemble distasteful butterflies of a completely different species that birds have learned to avoid. Orchids of various species mimic bees, flies or wasps. Mimicry of this kind has delighted naturalists since the nineteenth century and has often fooled collectors just as effectively as it presumably fools other animals. Although the object of awe and incomprehension in the past, it is now clear that mimicry, of almost limitless perfection, readily evolves by natural selection. Mimicry of (desirable to wasps) male figs by (undesirable to wasps) female figs is certainly to be expected, but the sequel is — to put it mildly — less obvious and needs a lot more thinking out. We also expect male figs to go out of their way to look, and smell, like female figs. Here's why.

  A male tree ‘wants’ female wasps to enter its figs and lay eggs in the pseudo-female flowers there. But the fig gains from this only insofar as the young female wasps that subsequently hatch out go on to play their allotted role. The new females must load themselves up with pollen, leave the natal fig, and then at least some of them must enter the genetic graveyard of a female fig and pollinate it (thereby propagating the fig's genes, though not the wasps’ own). A male fig that looks very unlike female figs may be very successful in helping female wasps to achieve their goal of entering only male flowers and laying their eggs. But the daughters of those wasps will tend to inherit their mothers’ taste in figs. The daughter wasps will inherit a tendency to {321} go only for male figs and they will be useless at propagating the genes of the fig in which they hatched (though good at propagating their own genes).

  Now consider a rival male tree whose figs resemble female figs. It may be harder for it to lure female wasps, who will be put off since they are trying to avoid female figs. But those female wasps that it does manage to lure will be a specially selected subset of female wasps: they will be female wasps who have been foolish enough (from their own point of view) to enter a fig that looks like a female fig. These wasps will lay their eggs in the pseudo-female flowers, as before. As before, their daughters will inherit their mothers’ taste in figs. And now, think about what that taste will be. These young wasps came from mothers who willingly and eagerly entered a male fig that looked like a female fig, and their daughters will inherit their (foolish from their point of view) proclivity. Their daughters will go out into the world, looking for figs that look like female figs. And a healthy proportion will find them — thereby killing their own genes but placing the male fig's pollen exactly where it wants it. These duped females throw away their own genes but they carry successful fig genes in their pollen baskets, including genes for making male figs mimic female figs. Fig genes from rival trees, genes that make male figs very different from female figs, will also be carried in wasp pollen baskets. But those bas-ketfuls of pollen are more likely to be thrown away — from the male fig's point of view — in the genetic graveyard of other male figs. Hence male figs will ‘conspire’ with female figs to make it difficult for wasps to discriminate between them and avoid their genetic graveyard. Male figs and female figs will ‘agree’ in ‘wanting’ to be indistinguishable.

  As Einstein once exulted, subtle is the Lord! But, if you can bear it, the plot thickens. The pseudo-female flowers, inside male figs, require to be pollinated if they are to provide the food that a wasp larva needs. So there is no difficulty in understanding, from a female wasp's point of view, why she actively loads herself up with pollen; no difficulty in understanding why the females have special pollen-carrying baskets rather than just becoming accidentally dusted with pollen. Female wasps have everything to gain from carrying pollen. They need {322} pollen to provoke pseudo-female flowers into making food for their grubs. But Grafen and Godfray point out that we do still have a problem on the other side of this remarkable relationship. We have a problem when we turn back to the figs. Why do the pseudo-female flowers in a male fig need to be pollinated before they will nourish wasp larvae? Wouldn't it be simpler just to provide food for wasp larvae whether pollinated or not? Male figs need to feed wasp larvae so that they will take pollen away to female figs. But why do the pseudo-female flowers insist on being pollinated before they will yield food?

  Imagine a mutant male tree that became less fussy: a sport that relaxed this requirement and allowed wasp larvae to develop even if laid in unpollinated flowers. This mutant tree would seem to have an advantage over its more fussy rivals because it would produce a larger crop of young wasps. Think about it. Any fig will be entered by some females who, for one reason or another, don't have any pollen in their baskets. In the fussy fig, these females may lay eggs but the resulting larvae will starve and no young pollinators will result. But now look at the rival, mutant, unfussy fig. If it is entered by a pollen-less female, no matter. Her larvae will grow rega
rdless and will hatch into healthy young wasps. The unfussy fig will produce a larger crop of young wasps because it will rear the progeny not only of pollen-carrying wasps but also of wasps that failed to carry pollen. So the unfussy male fig will have a clear advantage over the fussy male fig, because it will produce a larger army of young female wasps to carry its pollen off and into the genetic future. Won't it?

  No it won't, and here is the almost too convoluted subtlety that Grafen and Godfray discerned. This great little army of young female wasps, swarming out of the unfussy fig, will indeed be numerous. But — the argument is like the previous one — they will inherit the proclivities of their mothers. Their mothers — specifically the mothers of the surplus wasps, the wasps that the unfussy fig produces over and above its fussy rivals — had a flaw. They failed to pick up pollen, or for some other reason failed to pollinate the flower in which their larvae grew. This is why the extra larvae are extra at all. And the extra larvae will tend to inherit the flaw. They will tend not to pick up pollen, or otherwise will tend to be bad pollinators. It is almost as {323} though the fussy male fig deliberately imposes a hurdle for the wasps that enter it. It tests them to see if they will do to the pseudo-female flowers everything that they would have to do to a real female flower. If they do not, their larvae are not allowed to develop. By imposing this test, the male fig is selecting those wasp genes that tend to make wasps good at passing on fig genes. Grafen and Godfray call it ‘vicarious selection. It is a little bit like artificial selection, such as we met in Chapter I, and yet not completely like it. Pseudo-female flowers are like simulators used to weed out pilots unqualified to fly real planes.