In 1988, Trevor Price, Mark Kirkpatrick and Stevan Arnold published a paper in Science developing a model to explain why breeding date in birds, though heritable and under selection, does not evolve. Twenty-eight years after the paper was published, I spoke to Trevor Price about the making of this study, what we have learn since about this topic, and his recent attempts to test this model with breeding bird data from the Himalayas.
(Interview conducted on 20th July 2016 via Skype)
Paper citation: Price, T., Kirkpatrick, M. and Arnold, S.J., 1988. Directional selection and the evolution of breeding date in birds. Science, 240(4853), pp.798-799.
Hari Sridhar: This is the first time you worked on breeding dates in birds. What got you interested in this topic at this point in your career?
Trevor Price: Being an empiricist beforehand, I wanted to learn theory for my postdoc. When I went to Chicago, Steve Arnold was very interested in sexual selection. This idea about how sexual selection works in monogamous birds was really puzzling a lot of people. Darwin had come up with a solution, but there was this big question about why, if early breeding females have an advantage, doesn’t breeding keep evolving to an earlier time? Arnold had been puzzling about this and had read Fisher’s book in which Fisher puzzles about this. Fisher came up with his famous diamond, this explanation for how genetic values can’t be under directional selection even though breeding date can be. Mark Kirkpatrick had just begun this with Steve Arnold, and so I just jumped in really, and puzzled over Fisher’s diamond. We just puzzled over this for about a week, intensively actually, and then Mark said he understood it. That really freaked me out. I spent another day working really hard on this and then I understood it.
HS: Were you a postdoc. with Stevan Arnold?
TP: No, I was a postdoc with Russell Lande, the theoretician who put quantitative genetics into evolution. But in classic Lande style, he went off on a sabbatical, for a while, just when I arrived. So I ended up working with Mark Kirkpatrick, who was one of his disciples, and learnt a lot of the theory. Actually, I learnt a lot of the theory from both Russ and Mark. At that stage, Mark had been already collaborating with Steve on how sexual selection in monogamous birds might work. That’s the history of it.
HS: Were you and Mark Kirkpatrick in different universities at that time?
TP: Yes, he did his PhD in Seattle, Washington, and then did his postdoc. in Berkeley. He was a Miller fellow at Berkeley but he treated Chicago as his field site. It’s pretty amazing. They kept telling him he couldn’t be away from Berkeley for such extended periods, and he kept pointing out that if you were a field worker you were allowed to go off to Costa Rica, or Panama where Mike Ryan was, for six months at a time. So he came up to Chicago for two months at a time, which really was the hot bed of quantitative genetics. We met in Chicago when we were both postdocs.
HS: But the affiliations on the paper are University of California, San Diego, for you, and University of Texas, Austin, for Mark Kirkpatrick?
TP: Yeah, that’s right, because we had both finished our postdocs by the time this was published. And we were both just starting our jobs. Obviously, it’s a great thing to do, to have a Science paper in the first year of your job and put your university’s name on it!
HS: Can you give us a sense of how this collaboration worked? You say the work was done in a week or a couple of weeks…
TP: No, the work wasn’t done in a week – well, maybe the Science paper was. What was done in a week was very intensive scrutiny of Fisher’s diamond, to try and understand how you can have selection and inheritance but no evolution. Once we figured that out, it was quite straightforward for Mark to write out the model, Steve to come up with the path diagram, which really shows everything, and me to review the empirical literature. But what happened then was we basically put it aside and wrote this other paper for Evolution, which was on sexual selection, and included this little bit as part of it. So we got these two papers on the go, then we submitted the Science paper, and then we went back to the Evolution paper. So we worked on these two papers, while we were postdocs – the big one which had a female preference and the male traits in it, and then this one which was the foundation for that paper. We submitted the Science one first, but while we were still postdocs, Mark started to work on the Evolution one, which uses this and a lot more.
HS: Did the three of you meet in person to discuss the work that went into this paper?
TP: Mark and I became pretty good friends, and we discussed it quite a bit. But, in the end, I didn’t really talk much to Steve about it. Steve took a backseat and Mark and I did it. Steve was the motivator for the original idea, Mark and I wrote the manuscript, and then Steve came up with this path diagram – Figure 1. But basically he just let us work on it. The actual manuscripts were written once we had our jobs. I think I basically sat down and wrote a version of this manuscript that Mark then modified and sent back to me. We did it over whatever it was in those days; I guess it was email. And then we met couple of times and looked over each other’s shoulders when we were doing the writing. This is a pretty simple paper. It is a very simple model in this paper. But we carried on with the same motivation for the big Evolution one. Mark wrote all the equations, all the simulations, that we needed to use for that one. Then I went right through them all and checked them all and then wrote the first version. I wrote the discussion and the results, then Mark looked at it, and then we got together for the final draft.
HS: Do you remember how long it took you to write this?
TP: We put it aside for a while, when I came to work in the field in India. As you know, most papers take a couple of years from beginning to end. My postdoc was 1984-85, and this was three years after that. So, yes, it took 2-3 years to come out. That was partly because I had other things to do, of course. I was finishing up other papers. I don’t know about you, but I find the best way to write a paper is to sit on it for 6 months or so. And with three people, it takes a bit longer as well.
HS: Where did you do most of the writing?
TP: I think most of the writing for this particular paper was actually done at Austin. I visited Mark there a couple of times with a draft and we sat down at his computer and wrote it together.
HS: Do you have a writing routine – a particular time of the day and a place where you like to write?
TP: Yes, I usually write early in the morning and do my reviews in the afternoon. I can only be reactive by the time it gets to the afternoon.
HS: You said the motivation for this paper came from Arnold reading Fisher’s work…
TP: Yes, I don’t know how exactly Arnold came to that, but Arnold was extremely interested in sexual selection and in Russ’s model of sexual selection. You know, the famous runaway sexual selection model of Russ’s? He also, obviously, knew about Darwin’s model, and he must have comes across this diamond in Fisher’s book. Steve wanted Mark to do a model of it, that’s how it came about. But this paper itself has really got nothing to do with sexual selection and the Darwin-Fisher model. In retrospect, I feel it really deserved to be in Science, but most people at the time thought this was a crazy paper to have in Science. It didn’t seem that big. And still many people don’t get how you can have selection and inheritance but no evolution. It’s actually been quoted a lot since, but people still don’t accept that this is one way you can have persistent directional selection on lots of traits. It’s obvious to me. We are all told R = h2S – presumably you were at one point too – and that’s just not true.
HS: You say that what you discovered about breeding dates is probably also true of other traits. Subsequent to this paper, have other traits been discovered?
TP: Yes, I think so. Of course, seed germination was discovered pretty quickly after that. Body size is another classic trait actually, which seems to be under persistent directional selection. You know if you look at a lot of summaries of selection you find a lot of traits are under strong directional selection over their whole life history. And yet people still puzzle over that, they still say: ‘Well, they have got some unmeasured trade-off somewhere else in their life history’. But I’m pretty certain they haven’t. I’m pretty certain that it’s this phenomenon. So nearly all sexually selected traits, for example, we predict to be condition dependent. They should evolve to be condition dependent and as soon as they evolve to be condition dependent you get this phenomenon. Now what that does is it creates directional selection on the trait. But there’s a balance you see: you’ve got stabilising selection on the genetic component, directional selection on the non-heritable component, which sometimes can lead to stabilising selection when measured over the whole life history, but often it will lead to directional selection over the whole life history. So nearly all sexually selected traits are probably subjected to this kind of thing. In fact, most traits you can think about have got some kind of condition dependence in them.
HS: Was this the first piece of theoretical work you did?
TP: Yes, absolutely. This is how I learnt quantitative genetics theory. I started off working with Russ trying to do a theoretical model of reinforcement, but that had hybridisation in it, which rapidly creates non-normality. Non-normality is Russ’s biggest nightmare. In the end I did that with a student of mine for her thesis. So this was the first piece of theoretical work I did.
HS: Since then you have done a lot of theoretical work. Do you enjoy that as much as doing empirical work in the field?
TP: I think what I like about theoretical work is I do it very rarely. I mean I do theory, but the number of papers I have written on my own is tiny. I really enjoy working with very smart theoreticians and solving problems that have been an outstanding mystery to me. I do really enjoy the light forming in front of my eyes when something like this is solved. I mean, it was a puzzle for a long time and now we understand it. So I’m carrying on in that way even today, working with theoreticians on issues I can’t understand, which become clearer when they model them. So that I enjoy. But I wouldn’t say I enjoy it as much as fieldwork, because that I do on my own.
HS: Did this paper have a relatively smooth ride through peer-review? Was Science the first place you submitted it?
TP: No, we submitted it to Nature first. We got rejected, we appealed, and the appeal was accepted. But then it got rejected again. Then it also got rejected at Science, the first time. But Mark already had a name for himself, and so the editor at Science actually called him up and said: ‘Why should we publish this?’ I think Mark did a very good PR job on it. Despite all this, I think this paper probably had a smoother ride than many of my other papers.
Hs: Was the final published version very different from the first submitted draft?
TP: No, it didn’t change much. I mean, with three people working on it you know, it was pretty good.
HS: Did this paper attract a lot of attention when it was published?
TP: You know those were different days. A paper in Science tended to attract attention. There weren’t 10 other journals and only about 10% of the science being done now. So it attracted a fair bit of attention, but many people just couldn’t understand why this was in Science. I think people do now.
I should also say, one of the reasons I’m resurrecting the work at the moment is we have got a lot of interesting data on breeding dates of birds in Himalayas. Figure 2 in this paper shows that there should be a mismatch between breeding data and optimal time to breed. I am sure you are aware of this: people will talk about how a mismatch induced by climate change is causing bird reproductive stuff to go down. But it is not quite as simple as that because it turns out the optimal timing can be a mismatch. And this second part of the paper basically has not been played up at all. Figure 2, which shows this mismatch, has not been discussed in the literature very much. So I’m now trying to do models and use empirical data from the Himalayas to show mismatches are often optimal. You can’t actually interpret a mismatch as bad. What people do is they measure the average breeding date of the bird, but what natural selection is doing is working on the average breeding date of all the offspring that are produced. And birds that breed earlier in the season produce more offspring because they are in higher condition. So there’s stronger selection on those birds. Therefore, if you actually were to reformulate it as the distribution of surviving offspring they would lie right under the optimum. In this paper, we didn’t discuss why it should be that birds that breed earlier should be in higher condition. We just said we are leaving that bit aside. Now, I’ve worked out the full model which shows that birds in higher condition should breed earlier, but also calculates what the mismatch should be, under different selective conditions. That’s what we are working on right now. We have got 25 years of data from the Himalayas about how breeding date is moving forwards. There are a lot of ramifications of this whole thing. One of the things we don’t talk about is why high condition birds breed early, and we now know why that is. Another important point is that high condition birds should impart that condition to their offspring, so you expect maternal effects. Maternal effects are not in this paper either. Now, we have built the complete model where birds in different condition produce different numbers of offspring. And we have shown where the breeding date should evolve to and how much of mismatch that should produce. And the empirical data from the Himalayas is really showing that this condition effect is very important. But it’s still a long way from being published.
HS: What kind of impact did this paper have on your career? How did it influence the future course of your research?
TP: On my career it didn’t have any impact, in terms of things like tenure. People just said: “Oh, that was before you got here” and stuff. But in terms of my research it had a strong impact. I wrote another paper in American Naturalist talking about clutch size and how you could get an optimal clutch size in a similar way. Everyone was talking about optimal clutch size at that time. A lot of the theory I have done since was coming out of extending this model and building on it. I had one very influential paper with Dolph Schluter where we put condition effects into sexually-selected traits. That came out in ‘93 in Proc. Roy. Soc. and has been widely quoted. So, basically, this was foundational for my understanding of quantitative genetics, and subsequent theory I did, and which I have been doing ever since.
HS: In general, would you say that what you propose in this model holds true even today? If you were to rewrite this paper today would you change anything?
TP: That’s a very difficult question. At that time we just said we do not discuss why birds in high condition should breed early. I think if we write the paper today, people will say you have got to discuss that. But we just didn’t know why then. Now we do. So it will be a longer, more complicated paper. That’s the trouble. But there’s nothing in the paper that’s wrong.
HS: You say “It is not known whether the average breeding date actually is later than the ecologically optimal time for raising young in natural populations”. I’m guessing you have a better idea about this now, at least from the Himalayas?
TP: It’s always advantageous to an adult’s survival to breed as early as possible because you can go through your moult and come back. So that may push breeding date before the optimal time for raising young. Birds that breed very early are going to have higher survival because they can just simply get through the season quicker and get ready for the winter. So we’ve got two forces acting here: the condition effect which pushes breeding date later, and a trade-off, a genuine trade-off between high fecundity and high survival. What we haven’t worked out in the model yet is how these two conflicting things, this condition effect and the breeding date effect, resolve themselves. So putting everything together you get these two effects, one is pushing you before the optimal and one is pushing you after. But in the Himalayas, it looks like it’s resolved to be before the optimal. For example, the yellow-browed leaf warbler, which is a bird we study, breeds early. In contrast, the greenish warbler comes back later, is in the same habitat, but breeds 2-3 weeks later. It seems to us that the greenish warbler is breeding when food is optimal. So I don’t know if its competition or if it’s just simply that yellow-browed leaf warblers are smaller, so it can do with smaller food items and less food. But it breeds before the greenish warbler and the food abundance still seems to be increasing when the greenish warbler breeds.
I don’t really understand the European tit studies, in the light of what goes on in the Himalayas. European studies have measured caterpillar frass and shown a mismatch. But tits breed even earlier in the Himalayas than warblers do. There’s relatively a small amount of food out by the time they are raising their young. So I don’t really understand the European studies that talk about mismatches.
HS: Apart from the Himalayas, is there data on this from other parts of the world?
TP: There’s lots of data but it is nearly all based on nest box studies. Nest box studies have a number of issues, but it’s hard to criticize them because the people who do them review your papers. But there are two big issues. One is that they are relatively predation free. The other, probably the more important one, is that the densities are artificially increased. So you’ve actually got all sorts of weird things going on. There are studies in tits all over Europe, studies on parrots in South America and studies on swallows in North America. They have all used nest boxes, changing the densities of the bird populations they are working on and cutting out predation. So I just don’t know how relevant they are.
HS: You say that earlier studies “confound the effects of the environment  and the effects of nutrition  on female fecundity” and “It will be critical to distinguish these two effects” when investigating selection on breeding date. Has this happened?
TP: Yes, people have tried. There was, actually, stimulated by our paper, for about 10 years, massive attempts to do that. So there was really an attempt to test the model, by doing things like taking eggs and putting them in the fridge so they would then hatch later. And the idea was that you would not be changing condition but you would be changing breeding date. Other cases where they tried to change condition by plucking feathers from adults. So there was a lot of work on that, but I think it’s all been very inconclusive. It’s been very hard to do, especially when you didn’t have the full model. We didn’t have the model to explain why high condition birds should breed early. So I think now, with the full model, that could be revisited. It’s also been very difficult to separate the two effects. It might be now, now that we have got the full model, but I think it still probably would be very hard to do. The full model has this trade-off, where if you breed early you get an advantage, and I don’t think that trade-off was in the thinking of these early papers. But there were a flurry of empirical tests that, you know, were quite ingenious, but they have either been equivocal or they have done what you might predict, which is show that both condition and direct effects of breeding date are important.
HS: Could you tell us a little more about the people you acknowledge – how you knew them and how they helped?
TP: They all read and commented on a draft of the paper. Sara Via and Susan Kalisz were both postdocs in Chicago and good buddies of mine. Peter Grant was my PhD supervisor, Jack Bradbury was a colleague in San Diego, and Dolph Schluter and I were graduate students together. In fact, at that time, Dolph and I were sharing every paper we wrote.
HS: Do you remember if the grants you mention were specifically for this paper?
TP: Let me look. Ah, they weren’t grants to me. That’s a familiar story! Two grants to Steve and they were not for this work at all; they were for salamander work. And Mark’s grant was to do quantitative genetics in sexual selection, which is what he ended up doing. He had a whole bunch of papers on handicap and things like that, at that time.
HS: Is this analysis with the Himalayan data the first time you are testing this model?
TP: Yeah, that’s right. After this paper, I basically went off and did stuff for the Richman and Price paper. We were trying to do quantitative genetics but it was really hard. But we did have a paper, a few years ago now, in Current Biology in 2012, which built on this paper. It was looking at the wing bar. So, no, we didn’t ever try and test it earlier, although it could have been a focus of my work. I was just caught up in the phylogeny stuff.
HS: Have you ever read this paper after it was published?
TP: I think I probably have, for one reason or the other. When I have been teaching classes and so on. You know, I like this paper. It’s very short of course, extremely short. Some of my papers I read and think they are horrible, but some papers you read and think: “God, I was really thinking very deeply about that. About issues I have completely forgotten about.” This one’s like that. It’s got stuff in it that I have completely forgotten. And I think it was a pretty good paper at the time. But that’s going to happen when you work with two really smart people.
HS: Do you think your writing has changed since the time of this paper?
TP: I have no idea. I mean we would all like to think we are getting better at these things. I probably think my creativity has gone down. That paper was much more creative than the ones I’m doing now. But that’s why I work with students. They are the creative side, and I try and write the paper so it’s accessible to people. My writing has probably gotten better, but not much. It’s always hard, man. Writing papers is hard. But one thing that does happen is that there’s not the same pressure to work fast. It’s getting a bit ridiculous, but the two papers I’m working on with theoreticians at the moment, we started over 5 years ago. Today, I am able to put it aside, do other things, come back to it and make it better. Which is also true about writing a book, by the way. By the time you get to the last chapter, its two years since you wrote the first chapter, so you’ve had that time to revise and get things right.
HS: What does this paper mostly get cited for?
TP: I know what happened. It sort of didn’t get cited for a while, and then the Oxford group that studies tits, and the Finnish group that studies tits and flycatchers, got very interested in this question about lack of evolution of breeding dates. So I would say it’s mostly cited for what it says. But people still don’t believe it. You will often see a list of six reasons why heritable traits won’t evolve and ours will be number six, and they are still trying to sort out which one it is. I would say it’s cited mainly because the Oxford and Finnish groups got interested in it. Also now, there is interest in it in the context of breeding date and climate change.
HS: What would you say to a student who is about to read this paper today? What should he or she take away from it?
TP: It’s still really relevant today. But of course, there is this big big chunk missing, which is: why birds in high conditions should breed earlier or be bigger or have longer tails. A lot of unfinished work. So if you are reading this paper today, make sure you understand why heritable traits don’t evolve, and then go back to the literature of the last 30 years to understand why condition correlates with these traits.
HS: Among all the papers you have written, is this one of your favourites?
TP: No, not really. My favourite papers are probably the empirical ones like Richman and Price. And usually, the one I have just written! I started out doing correlative field work and then I had theoretician envy, but now I appreciate that its experimental work, really, where the skill lies. I still haven’t written a paper that’s got an experiment in it, so I would like to do that. But I think my favourite papers are basically the ones where it involved some field work and I learnt something directly about nature.