AUSTRALIAN FRONTIERS OF SCIENCE, 2005
Walter and Eliza Hall Institute of Medical Research, Melbourne, 12-13 April
The evolution of attractiveness: Insights from fitness landscapes
Dr Rob Brooks, School of Biological, Earth and Environmental Science, University of New South Wales
I am going to talk about very similar issues to those that Mark Blows has outlined so patiently and steadily in his talk. Many of the same concepts, such as genetic variance/covariance, matrices and selection gradients, are going to come up.
(Click on image for a larger version)
Like Mark, I am going to start with a bird slide. I puzzle over how people who don’t work on birds tend to start sexual selection talks with a slide involving birds, but I guess it is in some ways a tribute to the fact that a lot of the very important early work has been done on birds, presumably because they are so colourful and they call so beautifully, getting you thinking about these types of things.
(Click on image for a larger version)
That is not necessarily what appeals to everybody. I think if you want to take some kind of a barometer of what people are concerned about and what they are worried about, you can do very well to read the literature. In my field the ‘literature’ is quite accessible and reasonably cheap – the subscriptions aren’t all that extortionate!
So what are people worried about? According to magazines such as this one, they are concerned about sex and interested in it – which is good if you happen to work on sex. They are also interested in questions of age and diet.
Now, whether these are separate issues or are all parts of the same issue is something that I hadn’t thought about until recently, but today I will start off with some discussion of issues around the lek paradox and genetic variation, and then make a transition about halfway through to teasing apart some of the issues involved in this relationship between age, sex and diet, seeing whether or not the magazines actually have a point.
There is a fourth concept that people seem to be very interested in. In my field we call it phenotypic plasticity. It is not quite the same thing as Sharon Stone, whose photograph appears on this magazine cover, apparently had done, but it is a good metaphor.
(Click on image for a larger version)
This is the last bird slide, I can promise. It demonstrates the types of evidence for sexual selection that you see in the field: a relationship between an obvious and quantifiable trait, and attractiveness measured in terms of some kind of behavioural response, or – even better if you can get it – real mating success data. There is a nice tight, linear relationship and it is quite simple. Looking at this little finch from Africa, the pin-tailed whydah, you can see how evolution would be driving the exaggeration of tail length. The males with the longest tails are going to be the most successful, and so over time you would expect a change in tail length, if there is genetic variation.
I guess that is another important point, because as Mark pointed out, selection should tend to deplete that type of genetic variation. As you get towards where you are going, the variation that is out there in the population should be depleted.
(Click on image for a larger version)
So that brings in the issue of the lek paradox, which is encapsulated by the question: why doesn’t variation disappear?
A related issue is: what keeps males honest? If males are signalling some underlying benefit, obviously females don’t choose just because they feel like it. There is some kind of an adaptive basis. And what keeps males from cheating, signalling that they are a better quality than they really are? That is a related issue to those to do with the lek paradox.
There is an additional complication, that it is very seldom as simple as one dimension – the length of a tail. Very often we are talking about a number of traits and a number of dimensions of those traits, and that adds certain complications on which Mark has made a very good case – complications that actually change things quite substantially.
(Click on image for a larger version)
This is my main study organism, guppies. I have been working on these guys for about 12 years. Guppies occur ferally just about anywhere that humans occur, if that is warm enough to support them. Anyone who has ever kept them would know that they are very hardy animals, and people tend to like to release them, either to control mosquitoes, which is what was done in North Queensland in the early part of last century, or just because they can’t face killing their pets.
They are proliferative breeders, which is a wonderful thing in a study organism. But, of the things that I find quite cool about them, just looking at this slide I would say that every male is different – except that, having used the slide about 10 times, I suddenly spotted that the two males at the bottom of the second column are not different after all. But I can certainly, after looking at a male for five or six seconds and drawing a very brief sketch, come back and pick that male out of a tank. Once you know which particular colour pattern elements to look for, it is fairly easy to recognise individuals.
So, as with humans, we are dealing with something where there is an enormous amount of diversity, and that diversity is very visually available. Why is there so much genetic variation when we know, from previous studies, that these colour patterns are important in making males attractive? Why don’t we converge on a single most attractive colour pattern?
(Click on image for a larger version)
Just to demonstrate that that variation is genetic, here are a few pairs of full siblings, and you can see the similarities between their colour patterns. The males in the top right-hand panel are actually half-siblings of the males in the panel to their left – they share the same father – and you can see that there is tremendous resemblance across half-siblings. That is because many of the genes involved are on the Y chromosome. This is one of the few systems where genes that have some kind of an adaptive benefit are on the Y chromosome to such an extent. (That’s a different story.)
(Click on image for a larger version)
We have a special kind of algebra in selection analysis where the symbol above the column of numbers in this table means β. We are working on getting adoption of that across the field. This is a vector of selection gradients exactly like the β vector that Mark showed you in his slides. This is just to demonstrate that there is significant selection on these traits – the area of various colour pattern elements, the size of the tail, and how intense the orange colouration is, are all under significant selection. And those are quite big numbers as far as selection gradients go.
So there is selection and there is genetic variation. Why aren’t we converging on a single most attractive phenotype?
(Click on image for a larger version)
The first general type of solution is that potentially colour patterns that become rare tend to be favoured, because of their rarity. And so they are prevented from being eliminated from the population. We know that preference for rare colour patterns or for unfamiliar males will have this type of an effect. There also may be variation in what females like, and there may be more than one way for males to be attractive. So we might not be looking at this the right way, in expecting that selection is going to drive towards a single most attractive type. There being more than one way to attract is not, in itself, negative frequency dependence, but if it is going to be involved, it needs to be paired with one of the other mechanisms.
The alternative is that net selection on the colour patterns is in fact rather weak. The multivariate constraint argument, which is exactly what Mark spoke about, is one possibility. And trade-offs between the traits that make males attractive, and other components of fitness, like survival or female fecundity, might be implicated. So, across an organism’s entire life history, there may not be very strong net selection. I am going to just run through a few of those possibilities.
(Click on image for a larger version)
I am not actually going to consider any evidence for preferences for rare or unfamiliar males; we have some experiments in our lab but certainly it has been known for 15 or 20 years that females do prefer males with rare colour patterns and they do prefer males that are unfamiliar. If you think about the ecology of living in a stream that occasionally dries up, I guess, or dries up into little pools, you would realise that you don’t really want to be mating with your brothers or your cousins. There is very strong inbreeding depression in this species, and I think what you want to be able to do is recognise males that come from upstream or downstream – newcomers. And those newcomers do tend to have a big advantage.
(Click on image for a larger version)
I am now going to discuss some of the evidence for the other possibilities. This table simply shows the linear selection gradients and a matrix of nonlinear selection gradients describing the curvature of the relationship between these traits and fitness. (I don’t expect you to really scrutinise that table.)
(Click on image for a larger version)
This is a very bad animation metaphor for a canonical rotation. Basically, what we have done is to take that big table and crunch it down into six major axes of nonlinear selection.
(Click on image for a larger version)
I am going to consider the first two here in a thin plate spline. This is work that Mark and I did together. What this figure says, quite surprisingly, is that there are three different peaks to this surface, which is quite an unusual situation. I don’t know of any other surface that has three or more peaks. But what is really nice about it is that they correspond quite nicely to attractive elements of the male colour pattern.
So you can be attractive by having high levels of orange and black colouration, or you can be attractive by having a large tail, or you can be attractive by having lots of iridescent colouration, but there doesn’t seem to be a combination that optimises all three of those. There is more than one way to be attractive – which is quite a liberating message, I think!
(Click on image for a larger version)
I will now consider the multivariate constraint. This slide shows the basic equation of evolutionary change. (I thought I would put an equation in there as a compensation thing, really.) Selection and variation are the two essential elements for evolutionary change to take place. So one may have lots of genetic variation, one may have lots of selection, but as Mark demonstrated in his talk, if you put the two things together you may have very little net selection as a consequence.
(Click on image for a larger version)
What I did in this table was to put the gradients of selection into that equation, together with the genetic variance/covariance matrix. And even though there’s lots of selection, there is very little expected change.
These are to some extent the same units. The right-hand column represents how much each trait should change in units of standard deviation, in one generation of selection. So even though there is quite a lot of selection out there, you don’t predict very much change. It is a similar situation where most of the genetic variation that is out there is in fact not in the same direction as selection wants to go.
(Click on image for a larger version)
We come now to the last idea, trade-offs. There seems to be a bit of evidence for this as well. These two graphs show the genetic relationship between attractiveness and survival. This was quite a surprise to us. These are breeding values, so each data point represents the mean of the offspring of a given male. If you look at the relationship between attractiveness and offspring survival to maturity, you see a very strong negative relationship. Genetic correlation should never be less than minus one (although statistically it is possible) but this is actually a very strong genetic correlation in terms of that type of statistic, and it is negative.
When I scrutinised the data a little bit more, it was quite clear that this was driven by male mortality, not by the mortality of daughters. What is happening is that attractive males are having attractive sons – there is lots of genetic variation in attractiveness – but they are having fewer of them at maturity. And, in fact, beyond maturity male survival is also negatively genetically correlated with attractiveness.
(Click on image for a larger version)
That brings to a close the consideration of questions of variation. I am not one who likes ‘I’m okay, you’re okay’ types of solutions to problems, but it does seem that all four processes are important in this particular system, and that might be why there is so much variation in guppies compared with a number of other systems. It does seem as though it is not simply one process driving things.
Now I am going to talk about trade-offs a little bit more. The principle revealed in that previous graph, the negative genetic correlation between attractiveness and survival, is that being sexy comes at a cost. That is one of the very basic tenets of evolutionary theory: if you have resources, you can invest them into something that is going to make you attractive or you can save them for later – for survival, et cetera.
(Click on image for a larger version)
So let’s play a little game called ‘Two Ordinary Australians’. As you can see, they have got some money. (This is not a linear scale, I must add.) The crucial thing is that if you spend money you can’t spend it again on something else. I think the guy on the left here might have found a way to do that, but certainly I find that once I have spent my money it is gone. Let’s assume that some kind of success – it doesn’t have to be attractiveness – depends on advertising.
(Click on image for a larger version)
If these two people spend a similar proportion of their income on advertising, although the fellow on the left advertises a lot more, he succeeds far more than the other guy, but he also has a lot more remaining. And as a consequence he is able to spend that on health care, personal trainers et cetera, and he is therefore able to live a lot longer.
So this idea basically shows that even though being attractive is costly, you expect generally to see a positive relationship between survival (and other fitness components) and attractiveness. And that is in fact by far the most common finding out there. The predominant finding is that high quality males not only are more attractive but survive for longer. In fact, I think that has reached the status of dogma in the field, to some extent. So the guppy example, this negative genetic correlation, was quite unexpected and surprising.
(Click on image for a larger version)
If, however, the benefits of advertising are so immense, it is conceivable that the game can be played a different way. The guy on the left might actually derive a net benefit. He might win the game by spending so much on advertising that he in fact is left with less money in the end than the more prudent investor.
(Click on image for a larger version)
This is exactly what is encapsulated in the model that a collaborator of mine, Hanna Kokko and I did – well, I looked over her shoulder while she did all the really cunning maths. What she has modelled here is the strength of choice, on a negative scale. Where there is no choice, males don’t advertise at all. It doesn’t pay them to. But as the strength of choice in the population increases, so males increase their investment. High quality males are increasing their investment at a much higher rate than low quality males are.
As a consequence, under weak selection, weak choice, one would normally expect high quality males to survive longer than low quality males – we have modelled that as mortality rate – but in fact it is conceivable that, under very strong choice, you would have that relationship flipping over.
So it is one and the same process. Our guppy findings of a negative relationship between survival and attractiveness don’t require a whole new body of theory; it is the same process. We just think that selection is probably very strong.
(Click on image for a larger version)
Let’s look at an independent test of that. This slide shows a field cricket, Teleogryllus commodus, a native animal which was calling last night in the city. It is probably the most common cricket around here.
(Click on image for a larger version)
The first thing about these calls is that there are a number of different elements of timing and of dominant frequency – acoustic elements – to the calls. And in a big lab study we demonstrated quite nicely that there is a single peak on that fitness surface, and that the population mean sits on that peak. This is not under very strong directional selection; we don’t expect mate choice to be pushing the population in a particular direction.
(Click on image for a larger version)
That is until we start considering just how much they call, whether they stay silent or whether they arc up and start calling on a given night.
(Click on image for a larger version)
And when you start looking at the calling effort that males put in, you get a very different story.
Here is something I call the crop-circle design. Being a biologist you do get to go to nice places: this is Swiss Lakes field station, up near Forster. We put out little units involving a CD player and a couple of speakers, with some nasty stuff called Tangle-Trap, which is a sticky trap, and we just played loops of different lengths over the night and saw how many animals came.
(Click on image for a larger version)
What we get with that fitness surface – once again these are just major axes of selection – is the structure stuff falling out quite nicely in terms of a single peak. But the dimension that has lots of call effort on it indicates that it doesn’t pay to just call at an average level (in fact you might as well be quiet) but when you get above average there are increasing benefits of calling more and more. And, of course, females aren’t going to find you if you don’t call as a male cricket. So we think the males that are silent are waiting – potentially, I guess, for something better to come along when they feel a little bit brighter.
(Click on image for a larger version)
Let’s go back now to our story about investment and look at male investment strategies. This is a whiz-bang piece of equipment, about as high-tech as it gets in animal behaviour research. It records 10 times a second, whether or not a male is making a noise, and you can do 64 males a night. I’m very proud of that machine.
(Click on image for a larger version)
If you feed males on different diets – if you give males lots of good food or you give them poor food – they change their strategy, they change their approach to attracting females. On a low protein diet, males take a long time to start calling. They’re sitting around and they’re waiting – possibly to run into a pot of high protein food or something like that; we’re not too sure. We were told when we started this experiment, ‘Males never call before day 10. Day 10 they might start calling.’
This slide shows how much noise the males make in a given night. And, as you can see, on high and medium diets all the males are calling when we start looking – and they are calling a lot more, as well. So what is happening is that males are changing the timing of their investment, as well as the amount of that investment, and they are doing it in such a way that it has quite an interesting result for survival.
(Click on image for a larger version)
The first graph here is for nymph survival. If you are on a high protein diet you are much more likely to survive than if you are on a low protein diet. It is the same for females: females live much longer on high protein diets than on low protein diets. But for males it is exactly the opposite: these males on high protein diets are dying much sooner than males on low protein diets. I don’t have the figure here, but we have the relationships demonstrating that because these males are calling earlier and more, they are essentially running themselves down, they are burning out. I think the press call it ‘living fast and dying young’.
(Click on image for a larger version)
So what should we take home from this? First is the liberating message that there is not always a single most attractive type, that beauty may be in the eye of the beholder. Secondly, attractiveness always comes at a cost, and that is also quite a consolation. And, thirdly, the gossip mags are right: there is a relationship between diet, sexual attractiveness and ageing. That relationship is, essentially, that if you invest a tremendous amount in being attractive you are going to age faster and you are probably going to die sooner – another comforting thought to leave you with.
