In a 1990 paper in Nature, David Reznick, Heather Bryga and John Endler, showed, through an 11-year experiment on a natural population of guppies in Trinidad, that predators can cause significant life-history evolution. Twenty-six years after the paper was published, I spoke to David Reznick about its making, the influence it had on its career and what we have learnt since about life-history evolution in guppies.
(Interview conducted via Skype on 28th July 2016)
Citation: Reznick, D. A., Bryga, H., & Endler, J. A. (1990). Experimentally induced life-history evolution in a natural population. Nature 346: 357-359.
Hari Sridhar: Before this study, you had already done a lot of work on life-history and evolution in this system. What was your motivation to do the work presented in this paper in relation to all that you had done before? At what stage in your career did this come?
David Reznick: My original motive for going into research was to test some aspect of the theory of evolution in nature. And to do experiments. Evolution had largely been dealt with as a historical discipline, or in work on model systems in the laboratory, like fruit flies. I felt that it should be possible to look at evolution in real time. I was originally inspired by work by Janis Antonovics on heavy metal tolerance in plants, which was coming out in the 1960s and 70s. Janis was seeing evidence for evolution on timescales of centuries, but I felt that something more might be possible, based upon what I’d learnt in population genetics. My earlier papers that led up to this were really to develop the potential to do this experiment. The first question was to consider the natural history of guppies in Trinidad. I had been to a seminar by John Endler where he had talked about localities where guppies lived either with or without predators, and how the predators had shaped the evolution of male colour patterns. What I took from that talk was that those localities differed in the risk of adults experiencing mortality. When there are predators, adult male guppies showing colour were at risk of being eaten, but when predators were absent, males that were mature and showed colour were at much less risk of being eaten. That’s the only way Endler could have gotten his result. For me that paired well with life history theory. There’s a body of theory developed in the 70s that predicted how life history should evolve in response to a change in the risk of adult mortality. This work on guppies gave me a link to a specific facet of evolutionary theory that made a prediction. It meant that I could make that prediction and test it. The first step was to actually see whether or not there is a pattern in nature that was consistent with the prediction – whether or not guppies that lived with predators were on average younger at maturity and more rapidly developing and devoting more resources to reproduction than ones that lived without predators. That’s what the earlier papers had accomplished.
HS: You said you were inspired to do this work after hearing a seminar by John Endler. Do you remember when and where this was?
DR: I do. The talk was given at the Academy of Natural Sciences in Philadelphia in October 1977.
HS: At this time, had you already finished your PhD?
DR: I was doing my PhD at the time and I was working on a very similar problem. I was working on mosquito fish from Illinois and North Carolina and New Jersey and looking at life histories and variation among populations in life histories. And it was approaching the point where I might have been able to do something like I did in guppies. I had found farm ponds in Illinois with introduced mosquito fish, of which some had and some did not have sunfish (family Centrarchidae) in them. So the question was whether or not there is evidence of life history evolution there. So when I went to John’s seminar, my mind was primed, and his system was much more attractive for a variety of reasons. I was beginning my fourth year of graduate school, but after I saw his seminar I wrote the sort of proposal that we were encouraged to write for our qualifiers as second year students. And I gave it to my committee in December that year and asked them to consider letting me change my PhD project. They liked the paper. They thought it was very good and worthwhile for me to, not really discontinue the work on mosquito fish, but to put it on a back burner, and switch to guppies. So the guppy work papers published in 1982 and 1983 were from my PhD thesis, which I submitted in 1980.
HS: And did you continue working on guppies for your postdoc?
DR: Well no. At the point when I finished my degree I had no papers and got nowhere on the job market. But I had this idea to transplant guppies and predators and study evolution in real time. So I wrote a proposal to the National Science Foundation (NSF), but I wasn’t allowed to be a principal investigator on the proposal. So I found somebody whose name I could put down as Principal Investigator and submitted it before I finished my degree. Then by the time I finished my degree, the proposal was funded and it included my salary. I was then, what I described as, an incipiently unemployed research scientist, meaning that I was paying my own salary with my grants.
HS: This paper has three authors – you, John Endler and Heather Bryga. How did this group come together?
DR: The connection with John Endler was his seminar in 1977. I went out to dinner with him that night and said: I have this idea based upon your work. He was very supportive. I first went to Trinidad in 1978 – in March and April of 1978 – and he was there then and guided me. He gave me all the kind of background I needed to be able to work there – he took me out to the field, showed me how to identify the fish, showed me where I could go to collect what I wanted, etc. And I continued to interact with him after that point. In fact, the experiment in the 1990 paper is one that he initiated in 1976, which he talked about in the seminar I saw, and based on which a publication came out in 1980. So there is a natural and continuing connection between us.
Heather Bryga was my lab technician. She was an undergraduate who started working with me when I began as a faculty member in UC Riverside in 1984, Later I was able to hire her and pay her as a full-time technician with my continuing support from the NSF. Heather was the one who oversaw the lab work. I was still in the lab a lot those days but not always. She is the one who sort of kept things on track.
David Reznick (background, peering into a dissecting scope) and colleagues engaged in experimental work on a tributary to the Quare River. Reznick is examining fish for individual marks as part of a mark-recapture study designed to quantify mortality and growth rates in natural populations (© Helen Rodd)
HS: In the paper you say you reared the descendants of these fishes through two generations. Where was this done? Did you bring the fish back to the US?
DR: Yes. It’s interesting, I was looking at the paper the other day and thinking – Nature papers had, I think, a 1200 word limit, because of which you can’t see what you need to see in that paper. So the key sentence, where it says that, refers you to earlier papers. You would have to go to the earlier papers to read the methods. What I did was I collected wild adult females from the two sites, brought them back to the laboratory at the University of California, where I had a lab in the Biology Department vivarium. The experimental side of the lab was really like a fish file system since I had hundreds of two-gallon aquaria and reared one or a few fish per aquarium. I isolated each female in her own aquarium. Guppies store sperm, so each female produces a series of litters of babies, and they became a numbered pedigree. I reared them to maturity and separated males and females before they matured so I had virgin stock. Then I did crosses among the different pedigrees. The cross design was such that each wild caught female was equally represented, and all crosses were unique. There was no sib-sib mating. So it meant that my first lab generation was mated to produce the second lab generation that evenly represented the genetic diversity of the original sample from the field. The goal in doing that was to separate out environmental effects and maternal effects from what I wanted to evaluate, which was genetic differences among populations. By working on grandchildren, the idea was that the environmental effects and the maternal effects had been separated out. They were no longer confounding feature of the design.
HS: In your email to me, you mentioned that you are going back to Trinidad soon. I’m assuming that means you are still working there. Do you continue to work in the same sites that you sampled in this study?
DR: No, our main work now is located in a neighbouring drainage. We have gone well into the headwaters of what’s called the Guanapo River. We still do, sometimes, work in those old sites. Those sites are a living resource and all the introduced and evolved populations are available for sampling. For example, we have a proposal that will be submitted this week, to study the genetics of adaptation in guppies. Part of that will be to compare guppies within a river that live either with or without predators, but another part will be to take advantage of the experiments, where we know which the ancestral and descendent populations were. So these sites continue to be a valuable resource for looking at the evolution of various traits. I and other people continue to use them. But I’m not doing specific work at those sites. I just occasionally use the fish for studies.
David Reznick hiking upstream to one of his study sites where he and colleagues are currently performing experimental studies of evolution in guppies and killifish (© Lynn Johnson)
HS: Have those sites changed a lot since the time you worked there for this study?
DR: The Aripo tributary is pretty much intact. It’s in good shape, and the main site downstream where we collected the guppies that lived with predators is okay. But other parts of it..you know, right now what I’ve heard is that in the last year, a land squatter – I mean somebody who didn’t own the land but has cleared a section of forest nearby –might have modified the lower channel of the tributary that was the basis of that 1990 paper. This has happened before; I’ve seen people come and try to garden there. It’s hard because it’s not very good soil and its very steep. In the past they have used it for a year or two and given up and left. Further back in the forest the site is still in very good shape; there is no serious modification there. But a lot of the sites I used to work on in Trinidad have changed considerably. In the other introduction that I started in 1981 on the El Cedro River, the downstream control – the ancestral site – is very severely modified by human activity.
HS: Are these sites protected?
DR: Well [laughs] technically they are National Forest and should be protected, but they are not.
HS: Could you tell us a little about how the writing of this paper happened. Did the three authors ever get together in one place, or did all the discussion happen over the phone?
DR: I wrote the first complete draft on my own. At that point, John Endler was in Santa Barbara, which is, about, a 2.5-3 hour drive from here [Riverside]. So I just drove to Santa Barbara and we sat down at John’s computer and reworked the paper together. And then I submitted it. I can’t remember if Heather Bryga read or commented or participated in the writing at all.
HS: Do you remember how long it took you to write the first draft?
DR: That’s interesting. I, actually, got the results in 1988. But the inconvenient event at the time was that I was getting a divorce, so I wasn’t able to work on it as quickly as I would have otherwise. I think, I wrote the first draft fairly quickly. I can write reasonably quickly and I don’t think I spent more than, maybe, four sessions of a couple of hours each to write that paper. Then, when I met with John, I think we spent 2-3 hours working on it together on his computer.
HS: That is really quick! At that time, did you have a specific writing routine, i.e. with regard to when and where you wrote?
DR: Yes, I write in the morning. I do things like outlining and working on figures and tables and things like that in the night. In terms of writer’s block, I always tell myself that outlining isn’t writing. Outlining is just putting your ideas down and I can do that any time of the day. So I tend to work on outlines of papers and sketch out the order of ideas in the evenings, and then I get up and do the writing, the actually sentence construction, in the morning.
HS: Where do you write, usually?
DR: At that time, I was living in Riverside, in a small town house that I had just gotten, and did all the writing there, at the dining table.
HS: Did this paper have a relatively-smooth ride through peer-review?
DR: Yeah, it did. I got word of it when I was in Trinidad. They just asked for some small changes. It was not controversial, in that regard.
HS: How did you make the figures for this paper? Did you hand-draw them?
DR: 1990.. No, I think I just used some kind of computer program we used for drawing figures then. I think Heather Bryga did those figures, or certainly helped with them.
HS: How was the paper received when it came out? Did it get a lot of attention?
DR: Oh my, it changed my life. It appeared on the front page of the New York Times and the World Herald Tribune and in many other newspapers. I was called to do a TV interview, but I couldn’t make it. They woke me up at 8 am and asked if I can be in downtown LA at 9 am! I, also, got many invitations for seminars. It did a lot to help establish my career.
HS: Why do you think this study attracted so much attention?
DR: Because it showed evolution in real-time. You take it for granted now. It’s like no big deal now to say – ‘Oh, I study contemporary evolution’, since so many others have done it. But what people don’t realise is that this change in how people would respond to the study today versus in 1990 is what you could think of as a paradigm shift. When I proposed the experiment in 1980, in the original grant I wrote for the NSF to do this experiment, people used to smile at me and say – “it’s a good idea and we hope you live long enough to see something happen.” Even Jerry Coyne, who is a very prominent evolutionary biologist, saw me present this in a seminar in 1984 and said – “I hope this works. It’s a good idea.” It seemed like I was expecting a lot, to be able to see any kind of change over such a short time-scale. The Grants‘ work was coming out just then. Their work was 1980-81 and it was getting a lot of attention. Initially, however, they were looking at selection, not evolution. They didn’t nail it down as evolution until a couple of years later. But it wasn’t experimental work, and it wasn’t linked to a body of theory that made specific predictions. So I think the thing that was attractive about my work was that I had a prediction based on a body of evolutionary theory and I went and found a plausible situation in nature where we could test it. By the way, this is the second paper. A paper by me and Heather Bryga appeared in 1987, on the earlier experiments showing the evolution of male life history traits, and it got no attention at all. This was from the the El Cedro river experiment, but in that we looked for evolution after only four years. That seems kind of bold now, but I was an assistant professor looking for tenure, so I figured I would take a chance and see if it worked. And it did work, in the sense that the male traits had evolved, but there was no evolution of female traits. The 1990 paper was based on an eleven year experiment, and there the female traits showed up. But if you look at Table 1 in this paper, you will see the 1987 experiment is also in the left hand column. That’s from the earlier paper.
HS: What kind of impact did this paper have on your career?
DR: I think it made me known in my university. And it got me a lot of seminar invitations. I had linked it to my earlier papers, so it attracted more attention to the whole sequence of the work. But I will tell you what the biggest impact was, which I find hard to talk about. My choosing to become a scientist didn’t go over well in my family. For them, it was okay if that’s what I wanted to do, but my father wanted me to go into the family business or to veterinary school, which I had gotten admission into. Becoming a scientist seemed like an eccentric and not so meaningful thing to do. But you know, even though they didn’t understand my work, seeing me in the newspapers gave it legitimacy in their eyes. You don’t usually hear about this, but that really was, more than anything, the most important consequence of that paper.
HS: How did this paper influence the future course of your research?
DR: The other thing I was trying to work on at the time, which I was having some success with, but which I didn’t publish till much later, was the demography of natural populations through mark-recapture. Up until this point in time, the assumption was that if you live with a predator your probability of dying is higher. That seemed reasonable, but I wanted to actually prove it. So between 1986 and 1990 I got NSF money for detailed mark-recapture work. I was marking guppies using acrylic latex paint that I made less viscous by diluting it with Teleost Ringer’s solution. I found that, in a lot of these streams, the probability of catching a fish if it was alive was exceedingly high. Said differently, the odds of not catching the fish were so low that I could interpret the number re-caught as being very close to the number still alive. So it became a way of estimating mortality rate and I could show that, in fact, mortality rates were higher in streams with predators.
There was a major revolution in how people perform and analyse mark-recapture data that came between when | began and finished the work. The new statistics let you discriminate between causes for not seeing an individual in a given census. It may be because it died, emigrated or was present but not caught. If you sample properly you can get independent estimates of all of these things. I had collected data in a way that would allow me to know if emigration or escaping capture were important issues because they could bias the data, but I had not collected data in a form amenable to the new sorts of statistics. This means there was a special burden to address these potential sources of bias. I delayed publishing so I could add some extra experiments to convince people the data were okay.
One inspiration for the delay and extra work is that I presented my results at a conference in France with one of the founders of the new theory in the audience (J.D. Lebreton). He raised his hand and announced that he did not believe I could catch all of the fish. The extra work I did showed that I can, indeed, come so close to catching them all, and that the data were okay.
The weird thing though was – and this paper didn’t come out until 1996 – that the shape of the mortality curves for the high- and low-predation guppies was not what it needed to be, to be consistent with the original theory that predicted how life history should evolve. The original theory was based upon differences among age-classes in risk of mortality. And what I found was that in high-predation localities, all the age-classes had a higher risk of mortality to an approximately equal degree. And the theory I was using said they shouldn’t evolve, and that you couldn’t get evolution unless there was heterogeneity among age-classes in risk. At first, the result was upsetting, but then I realised – ‘Well no, I had seen them evolve!’ I had already done it. And so it wasn’t a question of whether or not they evolve, or whether or not predation had anything to do with it, but it still said that there is something more going on. And that has led directly, in two ways, to the kind of work that I’m doing now. First, the reason the low-predation life history evolves is not because of the lower risk of mortality, but because of the indirect effect of population densities being much higher and depleting the environment of resources. It’s a version of what people now call eco-evo. interactions. In 2006, I got the biggest grant I’ve ever got – a multi-investigator grant – to use guppies to evaluate the importance of eco-evo. interactions in a natural ecosystem. That was a partnership that included theoreticians, geneticists and ecosystem ecologists. That work still goes on. The second thing was that, once we perfected the mark-recapture, it became possible to repeat the kind of introduction experiment we had done earlier, but to begin with marked individuals. We could then census them monthly, mark all new recruits, save scales from each of them, and get DNA from the scales. Using the DNA, we can construct pedigrees, quantify individual reproductive success and look at evolution in a very different way, i.e. as variation in individual reproductive success. Then we could associate this variation in reproduction with individual-based traits we had measured. One of the nice things about this system is that, at each stage, as I learnt more, I was able to use the learning to answer a greater diversity of questions. I continue to work on guppies in Trinidad till today and am, in fact, just finishing a manuscript on them.
HS: It is now 26 years since this paper was published. Would you say that the main conclusions still hold true, more-or-less?
DR: Yes, and the result has proved repeatable. In the new experiments we have four replicates, and we show that it’s happening, but we also have a much better idea about why it’s happening, than we did then. The aspect of that paper that didn’t hold up was that it was written around the idea that it was differences in age-specific mortality that caused the life-history patterns that we saw. As I said earlier, we now know that that’s not the explanation. Indirect effects of predators and density regulation are playing a very important role.
HS: If you were to redo these experiments today, would you change anything, given the advances in technology, theory, statistical techniques etc.?
DR: Well yes. In the new wave of experiments we did, we did change things. In the earlier experiment, I just collected a bunch of fish and introduced them. Actually John Endler set that experiment up, the one in 1990. I had done the one in 1987. But in both we just collected a mixture of fish and put them in. What that meant was that we didn’t have complete knowledge of who went in or what their genetic makeups were. In the new experiments, what I did was to collect juveniles from the source site, rear them to maturity in single-sex groups, mark them, mate them, collect scales from all of them, photograph them and then introduce them. Then we continued with the same mark-recapture for every new recruit. Through this we know we know about individual movements and the community in which they are growing. And we have their DNA, using which we can work out their pedigree. So the recent experiments are yielding a much richer body of information. None of this was conceivable when this work was done in the late 70s or early 80s.
HS: Do you continue to collaborate with John Endler on this work?
DR: Yeah, actually we do. For a long time we didn’t, but he was the co-Principal Investigator on the grant I got from 2012 through 2016. He is looking at the evolution of colour patterns again, but using all of his new methods that weren’t available in 1980. I could tell you a little bit about that if you want to know.
DR: So after that work in the 80s-early 90s, John almost became like a neurophysiologist. He was interested in, sort of, the neurobiology of how organisms perceive colour, and in the innate structure of colour in the environment and how light changes through the course of the day. He developed these predictive models that integrated the perceptual sensitivity of females with the colour spectra of reflectance of the males, to ask whether or not the pattern of evolution would affect the way females perceived males, or how predators saw males. He was able to show that, in some circumstances, you could become more brightly coloured and attractive to a female but yet not more conspicuous to the predator, because of differences in their visual sensitivities. In the current experiments, we have two localities where we have thinned the canopy to increase primary productivity which also changes light falling on the stream. The question is whether or not the change in light would affect the colour patterns of males and how they were perceived by females. John used that as an experimental treatment to look at male colour pattern evolution, and we are now at the point where we can write the papers. We know that the canopy has a significant influence on the evolution of structural components of male colour, the structural green and blue. Part of the guppy coloration is structural and part of it is pigments, and we now have clear evidence for the evolution of the structural colouration. John will now be plugging in his models to see whether or not this is predictable based upon the nature of the light and nature of the visual sensitivity of the females. Darrell Kemp is also playing a big role in this work.
HS: In one place in the paper you compare the results of the four- year study and this 11-year study, based on which you argue for the need for long-term field experiments. In the years since this paper was published, to what extent do you think that has happened?
DR: There’s been some. I’m actually giving a Skype talk in October to Florida International University where they have a Long-Term Research in Environmental Biology (LTREB) grant. That’s an NSF programme that I think came into existence in the 70s. There is a lot of talk about the importance of long-term work. Experiments like we are doing, there are some out there, but there aren’t many. The thing that has happened instead is long-term mark recapture on model systems. That’s mostly a British type thing – Soay sheep, red deer of Rhum, the meerkats in Africa, and various great tit populations. Long-term mark recapture on individual populations of birds has really blossomed. There’s a lot of that kind of work.
HS: Towards the end of the paper you say your results demonstrate “the importance of predation in moulding life history evolution in guppies, though of course other factors may be important.” Subsequent to this paper, were other such factors discovered?
DR: Yeah, the indirect effects that I told you about is one factor. Another is resource availability. When you talk about density dependence what it means is that guppies are adapting to themselves. People don’t usually think about density-dependence that way. The eco-evo. interaction idea was alive dating to about 1961 but it wasn’t mainstream at all. It was silent in the background, but now this idea is very much in the fore front. That’s the other main factor that emerged with guppies. I wasn’t thinking of that precisely then but I was wondering about the importance of resource availability, because it seemed to be a feature of the head-water streams versus the downstream localities. The head-water streams are much darker because they have completely closed canopies, which affects light and productivity.
HS: In the final sentence of your paper you say “The widespread evidence for size-specific predation in other species suggests that this could be a common factor in life-history evolution”. Has work after this paper found support for this suggestion?
DR: That’s a good question. I’m trying to think. In the bird literature, that wasn’t a new idea, conflict between predation and the ability of parents to provision their young. Certainly within the field of bird life history evolution, which is the biggest area where that kind of work is being done, predation emerged as a big deal. Tom Martin was reading the guppy papers at the same time as he was independently developing the idea that predation was the primary factor driving evolution of life histories of birds. It wasn’t original to him but he did more than anyone to develop the idea and show that it was important in bird evolution. I’m trying to think of other organisms. I know that it has turned up. Now, life history evolution has kind of faded in to the background. It’s not a premier topic as it was then, but I think it is fair to say that predation already was and it’s grown since, as an important factor in shaping how life histories evolved.
HS: This paper has been cited over 800 times. Do you have a sense of what it mostly gets cited for?
DR: I think for the idea of contemporary evolution and experimental studies of evolution in nature. Like I said you may have always grown up with the idea that evolution is contemporary and you can see it happening in real time. But that was a new idea then, and I think this paper has maintained some interest for helping pioneer that.
HS: In the 26 years since it was published have you ever read the paper again?
DR: I actually read it again yesterday because I was meeting with my undergraduates. I don’t always assign my own papers, but I figured it was a reasonable one to introduce them to the system. Maybe it was because you had made me think of it with your email. But at other times too, I remember looking back at it, every once in a while. I’m always impressed with how brief and simple it was. I don’t know if you noticed, but it doesn’t report any sample sizes, which is very embarrassing.
HS: What strikes you most about it when you read it now?
DR: The thing that strikes me was that it was simple. I guess I try to write papers like that today as well, you know, have a short simple sentence structure and not clutter it with extraneous ideas. I guess I didn’t realise I had figured that out at a pretty early stage in my career. I remember, one of the books I read about writing was “The Elements of Style” by Strunk and White. I had read it in 1988-89, sitting at my dining table. So that was fresh in my mind at the time I wrote this paper. So when I read it now, I’m glad to see did an okay job with it, in terms of simple sentences and being clear.
HS: Would you consider this one of your favourites, among all the papers you have published?
DR: Oh yes, it still is. It still serves as a standard I try to live up to, with the work that I do now.
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?
DR: They should take away the idea that evolution is a contemporary process happening in real time, and it’s one that can be studied empirically with experiments in nature. It’s a question of finding the appropriate setting to do it. So the contemporary nature of evolution is probably the most important message. Also, that it’s possible to extract, from evolutionary theory, specific predictions that can be tested in a natural setting. Finally, in terms of how the system has developed over time, it’s also a statement for why it’s good to do things in nature and not just do them in the lab. The lab version of reality is highly abstracted and it cannot capture the range of interacting factors that you see in nature. People will say you do it in the lab because you can control it and be clear about what’s going on. That’s true, but if you want to know why animals or plants are the way they are in the real world, then you need to work in the real world, because the full scope of factors that interact in shaping evolution can’t be anticipated or replicated in the lab. I made a similar argument one time, way back when people asked me why I wasn’t doing these experiments in the lab. I said it’s because I want to know why things are the way they are in the real world, and I don’t have faith in the lab being able to reproduce that. The way this project has developed has shown that there’s no way that any lab work could have led me to an understanding of ongoing interactions between ecology and evolution. You just can’t capture that in the lab. I couldn’t have anticipated that when designing the lab study. Therefore, I think it’s important to work in nature whenever you can.