In a paper published in Nature in 2002, John Thompson and Bradley Cunningham showed, through careful study of a widespread plant-moth interaction across multiple habitats and years, that the same moth species functions as a pollinator in some places and as a parasite in other. Thompson & Cunningham’s results provided support for the idea that the geographic structure of species can create mosaics of selection across the overlapping ranges of the interacting partners and drive continuing coevolution. Fourteen years after the paper was published, I spoke to John Thompson about his motivation to do this study, memories of field work and what we have learnt since about geographic mosaics of coevolution.
Citation: Thompson, J. N., & Cunningham, B. M. (2002). Geographic structure and dynamics of coevolutionary selection. Nature, 417(6890), 735-738.
Date of interview: 14 December 2016 (via Skype)
Hari Sridhar: What was your motivation to do the work presented in this paper? At this point, you had already been working on this system for quite a while. The first paper you wrote on this plant species was in 1992, as was your first paper on the moth species. In the following years, you had some other papers on the interaction itself. What was the motivation for this particular paper, in relation to the prior work you had done on this system?
John Thompson: Actually, there’s a couple of different kinds of motivation that all come together in this paper. One, from a more theoretical standpoint, was a direct test of some of the components of Geographic Mosaic Theory. It was intended as an analysis of how the structure of reciprocal selection and the strength of reciprocal selection may vary across populations. From past work I had done on this system, I thought that this might be a good set of interactions to use to test those ideas. That, then, was the conceptual motivation. Another motivation, conceptually, for using these particular sets of interactions was many people done a lot of work on geographic variation and traits, from a whole variety of different perspectives. I mean, not just for coevolving, but in evolutionary biology in general. Working on these interactions suggested to me is that we may have a geographic mosaic here in the ecological outcomes of traits. One of my goals throughout my career has been to link ecological and evolutionary theory together, and have a more ecologically-based theory of evolution and theory of coevolution. And so, one of the things about this paper, or this analysis, that I thought might be interesting was to ask whether ecological outcomes varied across landscapes, which turned out to be so. That became both a motivation and an outcome of the paper. So, that was one line of argument and motivation for this study.
The second line of motivation was that in the late 1970s I made a decision to take on one of the classic stories of coevolution. I didn’t know which one yet, but I wanted to take one of the classic stories of coevolution and really understand where those interactions may have come from. How do you get these beautifully coevolved interactions seemingly out of thin air, which is the way we taught about them in courses. You had beautiful examples like figs and fig wasps, and yuccas and yucca moths. And so, what I did is I started looking at these different kinds of interactions. I spent time in Costa Rica looking at figs and fig wasps and I spend time looking at the yuccas and yucca moths. By sheer serendipity, while working on a different set of interactions between moths and plants, that is, between oecophorid moths on mbelliferous plants (plants in the carrot family), I saw something that looked like a yucca moth laying eggs into the seeds of one of these Umbellifers. I said to myself, either I’m a very bad natural historian and this isn’t at all related to yucca moths, or this could be really exciting and tell us something about the origins of coevolved interaction between plants, yucca moths, and the relatives of yucca moths. This yucca-moth relative was completely unexpected because it was laying eggs in a eudicot rather than a monocot, such as yuccas and grasses. These two lineages of plants separated very long ago in geological time, and single lineages of insects do not tend to use both of these plant lineages.Also, these moths on umbellifers were antagonists rather than mutualists. They were simply seed parasites, in which females lay egg within seeds that are already developing and each larva eats the seed in which it which it hatches In contrast, the pollinating yucca moths are pollinating floral parasites that pollinate the flowers in which they lay their eggs. By pollinating the flowering, a female moth guarantees that her offspring will have developing seeds to eat. The differences between these antagonistic and mutualistic moths motivated a whole series of studies during the 80s that let me follow up on those observations. In the process, we discovered the diversification of the whole set of undescribed prodoxid moths throughout western North America, of which the final extreme of these coevolved interactions turned out to be the yucca moths. So, that’s a very long-winded answer, but there were two very different kinds of motivations that came together in this particular paper.
HS: How did you become interested in this particular system?
JT: Well, that is sort of the answer that I was jumping ahead on in my previous answer. I had made that decision in the late 70s. I’d been studying insects and plants; I’d been studying geographic patterns of interactions between birds and fruits. And these were all questions on the spatial ecological and evolutionary diversification of interactions among species. This serendipitous study, done at a time when I was looking for ways to study diversification of co-evolving interactions, then led to studies of additional interactions and discoveries of how coevolving interactions diversify across landscapes and among species.
HS: I’m sorry, maybe I misunderstood. Are you saying that the species you discovered was the Greya species or something closely related to that?
JT: I wasn’t clear. The one I found on the umbellifer (a species of Lomatium) turned out to be what we eventually named Greya, but Greya didn’t exist as a genus then. I had some of these moths in my collection and I sent them to Don Davis, who was a systematist at the Smithsonian Institution working on yucca moth and many other small moths. He wrote back and he said, well, John, these are interesting. What yucca does it feed on? And I said, No, no, no, no, it feeds on a carrot – something of the carrot family character – and he said, of course it doesn’t. This comes from a group of moths that feeds on monocots. That’s when I knew it was interesting. I then studied that species in some detail, published some papers on it, and in the process, and found additional Greya moth species in the same habitat, feeding on plants in saxifragaceous plants rather than umbelllifers. That is when I realized there was a whole radiation of species that varied from antagonists to mutualists on dicots and monocots. And this paper, the first on the saxifrage-feeding species, became one component of all the follow-up studies that have used these interactions to study how coevolving interactions diversifications among ecosystems.
HS: Stepping back a bit, how did you get interested in coevolution?
JT: So, the interest in understanding ecology and evolution of species interactions was there right from the beginning. When I worked on my doctorate, I worked on two completely different kinds of interactions of which, I turned in for my thesis, the work on insects and plants. The other part of it, which I’ll start with, was the work that I did with Mary Willson on interactions between birds and fruits, specifically asking how phenological patterns might evolve in plants to take advantage of migratory birds. And so, it’s really very much a geographic question at a very big scale, about how plants and birds might coevolve through changes in phenology, through changes in plant traits, and changes in foraging patterns of birds. But what we could get out of it, in that short timescale, was really ecological kinds of results, rather than evolutionary results. We were able to identify certain kinds of fruiting strategies, which we published in journals like Evolution, but we could only go so far. These were hard interactions to manipulate.
Meanwhile, I was working on interactions between insects and plants with Peter Price, and that became the thesis work. The interactions I studied were those between introduced wild parsnips and specialized moths called parsnip webworms. Those papers ended up being ecological but they started out being evolutionary – and this is an interesting story, which I’ll keep brief. I noticed that there was a lot of variation in the patterns of attack on the plants – some plants were attacked, some plants were unattacked – and I was trying to understand why that might be so. That led to a series of studies on how plant position within populations and plant sizes could influence patterns of attack. But I knew I was missing something. And the papers I ended up publishing from that work were ecological papers, because I couldn’t solve the evolutionary problem of why there is variation in patterns of attack. As I was finishing my doctorate – and this is where the work became evolutionary after I left it – May Berenbaum was leaving Cornell after her doctorate and moving to Illinois. She said, “I’m very interested in these interactions you’ve been studying, John. Are you going to continue with them?” I said, “No, I’m moving west, and it would be great if you picked up on that story”. She took over my field sites. She knew plant chemistry, which I didn’t know. Her work, and that of her long-time collaborator Art Zangerl, eventually showed that there’s a beautiful coevolutionary interaction going between the moths and the plants that explains the patterns of variation. Her studies demonstrated to me that, if I would have known my chemistry, I would have, right from the beginning, had one of the best coevolutionary stories that anybody’s ever unraveled.
HS: When did you first start using the phrase “Geographic Mosaic”?
JT: The term first appeared in some of my papers in the early 90s. It followed an Annual Review paper that I wrote in 1988 called “Variation in interspecific interactions”, at a time when I was starting to think about how to take studies of coevolution and turn them into much more of a science of coevolutionary biology. We had some examples of coevolution that were starting to develop, but the way that we presented them in evolution courses would be as just stories. We would discuss rigorous theory of evolution, and then, if coevolution was mentioned at all, it was just a story. My motivation was to understand how we could link ecology and evolution into a much more ecologically-based theory of coevolution. I started to work toward that specific goal in the 1980s, in papers such as the Annual Review paper and some subsequent papers, and then articulated it in full form in the 1994 book, The Coevolutionary Process. In the decades since then, it has undergone much refinement, through mathematical and empirical studies by a various labs worldwide using a wide range of approaches.
HS: How did the collaboration with Bradley Cunningham come about?
JT: Brad was a technician in my lab for some time and eventually became a master’s student in my lab. He and I worked together on this paper, and we did a lot of field work together for the paper. Brad spent a lot of his time with me and separately in the field as we collected the data. Multiple other people were involved in this field work too.It was sort of a group effort within the lab. Eventually Brad did his Master’s work on the way that these moths have shifted onto yet another genus of plants in some habitats near to where we conducted the studies for this paper. After he finished his Master’s, Brad became a technician in the lab of one of my former doctoral students and then he went on to other work in environmental consultancy.
HS: His affiliation on this paper is Washington State University. Had he moved there by the time this paper came out?
JT: Brad and I were both at Washington State during the years we conducted the field work. I moved from there to University of California, Santa Cruz in the year 2000. He had also left for his new position about the same time. The writing was done right after I arrived in California, but all the work itself and all the analysis were done when I was still at Washington State.
HS: You say this work is based on surveys done over five years in many different sites. From the beginning, was the motivation to understand the variation in this interaction or was there a different motivation to do surveys in multiple sites?
JT: The motivation was explicitly to test various components of Geographic Mosaic Theory, to ask whether the strength of selection and structure of selection varied among environments. And it was also to evaluate whether there was geographic variation in the ecological outcomes of those interactions, and not just in traits. So, the work was designed from the very beginning to take the same interaction and look at it in very different environments, from deep river canyons to essentially the tops of relatively tall mountains. We knew that potential for co-pollinators, beyond Greya moths, differed among those habitats. We knew that in the deep river canyons, for example, there were many bombyliid flies (bee flies) and solitary bees that visited plants that flower in the early spring. And so, part of the motivation was to determine whether the interactions between plants and Greya moths varied among habitats in whether they were mutualistic, commensalistic or antagonistic. That is, does the interaction vary in ecological outcomes and natural selection among environments.
HS: Who were the people involved in doing the surveys?
JT: We had a group of us in the lab that were doing them together, including me. I have not missed a field season in over 40 years. I was not playing lab director. I was directly out there with postdocs, graduate students, undergraduate students, and technicians collecting these data, because those are the moments in life that are among my most enjoyable for me.
HS: Could we go over the names of the people you acknowledge, to get a sense of how you knew them and how they helped?
JT: Yes, please.
HS: D. Althoff
JT: David Althoff, was PhD student in my lab. He worked on different Greya species and their parasitoids on different plant species. He worked mostly on the species of Greya moth that’s most closely related to the one that’s in this paper. The question we wanted to ask was whether the parasitoids vary geographically and how they searched for Greya moths on their host plants. In some habitats, there were two Greya moths on a plant species. One species placed eggs in the flowers and the other placed eggs in the stems. Dave’s dissertation showed showed phylogeographic divergence these parasitoids. These are Agathis parasitoids, which are braconid wasps that attack moth larvae. The parasitoids that Dave studied search the host plants of Greya moths in different ways in different habitats, depending upon which Greya species are in those plants. Some populations of the parasitoid search only on the reproductive parts of the plants, that is the seed capsules, and in others they search both on the seed capsules and the stems. That corresponded to whether those habitats had one or two Greya species. While he was working on his dissertation, we worked together in the field, simultaneously on this paper and on collecting for his PhD work.
HS: E. Bartlett
JT: She was an undergraduate helper in our laboratory that helped do many of the dissections that we had to do, to count the number of developing seeds and determine whether flowers had eggs in them or not.
HS: R. Beck
JT: Also the same. There’s a whole group of people that helped us dissect the seed capsules., what I think eventually turned out to be, about three quarters of a million seed capsules to look for eggs.
HS: N. Bonilla
JT: Same thing.
HS: S. Bradford
JT: Same thing.
HS: C. Clark
JT: Same thing.
HS: C. Cody
JT: Same thing.
HS: M. Jacoby
JT: Same thing.
HS: S Lambert
JT: Same thing.
HS: Y. Martinez
JT: Same thing. All the above undergraduate students worked very hard both in the field and in the lab. We had a great time together.
HS: K Merg
JT: Kurt Merg was a PhD student of mine who worked on pollination of a closely related species of plant (Heuchera grossulariifolia), asking whether the Greya moths on those plants were mutualists or antagonists. Kurt worked on pollination by a Greya moth that is either the same species or a closely related Greya species to the one we studied for this paper. We already knew that, on the plant species that Kurt was studying, the Greya moths were probably not the primary pollinator and that co-pollinating bees and flies differed among habitats. But there was another component to it, in that, in that genus of plants, some of the plants are diploid and some of the plants are autotetraploid, and we wanted to ask whether polyploidy can shape the evolution of the interactions between moths and the plants, and also the evolution of interaction between the plants and other co-pollinators.
HS: S. Nuismer
JT: That’s Scott Nuismer. He was a PhD student of mine, working at the time on the same species of plant species (Heuchera grossulariifolia) that Kurt Merg was working on, and he was asking good questions about whether Greya moths of two different species were more likely to attack diploid or tetraploid plants. But again, we were all helping each other with the work at the field sites, because the plants and insects we were all studying overlapped in their habitats. Much of the work was in and near wilderness areas, and we spent many days and nights camping in tents at those sites.
HS: J. Olmsted
JT: Another one of the helpers for dissection.
HS: S. Ringo
HS: K. Segraves
JT: Kari Segraves was a Master’s student in my lab working, at the time, working on phylogeographic divergence in Heuchera grossulariifolia species in which some populations were diploid and others were tetraploid. In addition to her molecular analysis, she analyses differences in vegetative and reproductive traits of diploid and tetraploid plants in a common garden set up on the campus of Washington State University. As with David Althoff, Scott Nuismer, and Kurt Merg, she spent part of her time at field sites working on the research for her thesis and part of the time working with us on collecting data for this paper. She then went on to do a PhD with Olle Pellmyr, who had been a former postdoctoral associate in my lab.
HS: D. Shepard
JT: Also one of the people who helped with dissecting seed capsules.
HS: Z. Stanley
JT: Same thing.
HS: T. Steury
JT: Tim was a university science reporter who came with us in the field for some days during one field season to see our research first-hand and then write about it for university publications. He had a marvelous ability to quickly grasp the questions we were asking and then write about it in way that was accessible to non-biologists. He was also willing to pitch in to help us with the field work.
HS: Were all the people who helped with dissections undergraduate students?
JT: They were undergraduate students. I am worried that I’m underselling them in my earlier brief comments, because the lab work demanded care and accuracy. Some of them also worked in the field with us. We would have big groups of us, sometimes as many as seven, that would go out for a week or longer at a time.We were set up for expedition camping out in the wilderness areas of Idaho. We would set up at trailheads and camp there for days at a time and work from those sites. So it was a pretty big operation.
HS: Why did the fieldwork take five years?
JT: First, we needed several years of results. Any single year was not going to be a convincing result. And so we really needed to do it multiple times. In the first year of any project, you set up protocols. Some aspects of the protocols work and some aspects of the protocol aren’t as good as you think they can be, so you refine them. So there’s a little bit of that. And part of it is that these are time-limited data each year – you can’t repeat them within a year; you have to repeat them among years. It also takes some time to move among all these sites. These sites are not right next to one another. They’re in wilderness areas or near wilderness areas, and you move among them by driving, sometimes long distances, and hiking up trails. You have to be there at exactly the right time. If you are too early in a season, you have to go back. If you miss it in one year, you go back the next year. This work took a lot of time because logistically it was a very complicated project that required multiple years of results and groups of people working together as teams.
HS: This might be an unfair and a difficult question, but which was your favorite among these sites?
JT: That is an unfair and difficult question. Oh my goodness. It wouldn’t be the deepest wilderness site, which surprises me, because those are the places where I like to be. It would probably be the site at Turnbull National Wildlife Refuge in Washington State, which is in a beautiful ponderosa pine forest. It has some of the richest diversity of floral displays in the spring that I have seen anywhere. And I think one of the other reasons it is my favorite is that it has some of the greatest numbers of the woodland star plants and moths anywhere. It is easier to work there than anywhere else, just for sheer numbers of plants and moths. , and therefore there’s more satisfaction because you get better data.
HS: I wanted to ask you specifically about that site, because that’s a site where you worked again in 99, looking at the other pollinators. Why did you pick this one, among all the sites, to continue your work?
JT: It was mostly because of logistics. It’s flat, unlike some of these other sites, which are straight up the sides of mountains or deep into canyons. It’s much more accessible than some of our other sites from where I was living at the time. It didn’t take many hours or half a day or more to get to the site. I didn’t have to camp out there for long periods of time. We also had more laboratory access from that site. But, I think the biggest advantage is it was quite clear that that site was very strongly mutualistic, and we could ask how the interactions between the plants and the moths evolve when there is strong mutualism. Finally, there were so many moths and so many plants, we could work on those populations without any concern at all about damaging the populations.
HS: Was the laboratory where you did all the dissections in Washington State University?
JT: Yes, that was my lab.
HS: What has happened to that laboratory now? Is it being used by someone else?
JT: I’m sure it is, but I have no idea who it is. I can tell you that when I left Washington State University, which I enjoyed greatly for 22 years before moving to University of California, Santa, the university was very kind. I was still doing work in Washington, Idaho, and Oregon, and WSU let me keep that laboratory for almost two years after I left for UCSC. They eventually gave it to someone else, and I hope that whoever it is has enjoyed working there.
HS: Do you continue to work in these field sites?
JT: I have continued to work at some of those sites over time, but mostly have moved on now to other sites, especially in California, where the interactions between woodland stars and Greya moths have radiated into multiple species and have diversified even more in how the plants and moths interact with each other. Turnbull National Wildlife Refuge is the one site that we continued to go back to almost every single year for many years, because we use that, really, as the base site to compare with everything else.
HS: Is this a system you still continue to work on.
JT: Yes, we continue to work on it. The focus of the work has shifted directions now, but I continue to ask questions about how these interactions coevolve. We are trying to amplify our understanding of the full range of traits that are coevolving in these interactions. And what that’s meant in recent years is moving from analyses of ecological outcomes to analyses of the morphological traits and, more recently, to analysis of chemical traits. And so we’ve been looking a great deal in the last few years at evolution of floral volatiles, because it was clear, even to my nose, many years ago that these populations of plants smelled very different,and that had to be important to these interactions. And so, by first working with Rob Raguso, at Cornell University, we developed some techniques and then when Magne Friberg came to work with me from Uppsala University as a postdoc, and now an ongoing collaborator more recently, when Florian Schiestl came on sabbatical to work with me from Zurich, we’ve been able to really start getting some deep understandings of the tremendous diversification in floral chemistry. We have just a bewildering number of compounds that have diversified among populations of these plants. And we’re looking at the chemical diversification, and we’re also looking at the insect response to that diversification – how much local adaptation is there in the insects to respond to that diversification? What compounds are they responding to? It’s turned out to be a fairly complex story, because we went into it with the prediction that it would be like the yuccas and yucca moths, which were the close relatives here of these moths, in which the plants have a very simple chemical profile that varies very little. Here we got like the opposite extreme and we’re trying to understand why that’s so.
HS: When and where you did most of the writing and, roughly, how long it took you to write a draft of this paper?
JT: That was too long ago.What I can tell you is it was a slow process, partly because, although the results are simple and clear, there are a lot of data. And it’s a fairly complex problem to try to collapse down into something for a paper that is going to be short. If I would have tried to publish it in Ecological Monographs, or even an Evolution paper, there would have been a lot more room to explain it. But to try to hone it down to something that you could put into a journal like Nature took a lot of time. Plus, there’s another component that is part of the process when I do statistical analyses and get statistically significant outcomes. I get the results and then I often spend months trying to search for biases in those outcomes. That is, may I have inadvertently biased my analyses to get to these statistically significant ones ? What do I need to do now to see how biased I was? This is an ongoing process with me for every paper. I do the analyses, put them aside for awhile, bring it back out later to think about what I did, and if I am satisfied that I have analyzed them in a fair and reasonable way. So that’s a long process for me, for each paper, for that reason.
HS: Would you say this paper had a relatively smooth ride to peer review and was Nature the first place is submitted to?
JT: I submitted it to Nature and it was rejected. I looked at the comments and realized that what I had done was not explain exactly what I was trying to do. The reviewers were very good reviewers, but I think what they expected was a paper that was a little bit more traditional. They expected me to ask how morphological or other traits vary among populations. But what I was trying to do here was something completely different, which was ask how ecological outcomes vary among populations. I didn’t articulate that as well in that original draft. I just made the assumption that they would get that point, and they ended up focusing on other points. So, I wrote the editor and said that I realized the fault was mine, but with just a small amount of rewriting, I think I could articulate the goal and novelty in a way the reviewers would understand. I asked him to let me do that and then have them take another look at the manuscript. If they still do not think it is acceptable, then the design is my fault. The editor, who was Rory Howlett, who was a remarkable editor for Nature, looked at my comments, looked back to the original manuscript and said, I see what you’ve done and what you mean. He sent it back to the reviewers and the reviewers said, ah, I see; now I get it. This is interesting. They re-reviewed the manuscript, I made some additional changes to incorporate their suggestions, and Rory Howlett then accepted it. I was grateful for the way in which the editor and the reviewers kept an open mind throughout the process.
HS: That’s an interesting story. Do you remember how the paper was received when it was published?
JT: It certainly generated comments from colleagues and mail and email. I mean, we weren’t emailing as much back then as we do now. But it certainly generated responses from people saying, now you’ve given me ideas on how to analyze my own data. Another set of comments, which I found very interesting, were follow-ups on comments I received after publication of my 1994 book on Geographic Mosaic Theory. When the book was published, some colleagues said, well, this is interesting, it might even be right, but how would you go about testing any of this?. After they read this paper, some said, oh, now I can understand a bit how I could go out and do something like this myself. This paper, then, helped generate a series of subsequent studies on the geographic structure of evolving and coevolving interactions among species.
HS: At the time when you did the work, did you anticipate at all that it would have such an impact on the field? And do you have a sense of what it mostly gets cited for?
JT: I don’t think any of us ever have a sense of whether a paper is going to have much of an impact, partly because the literature is so vast nowadays. But I certainly had a personal feeling that this was one of the more satisfying studies that I had done. I was hoping it would have some sort of impact on helping us understand how to study coevolution as an ongoing process. As to why it was cited – I think there’s several reasons, some of which have to do with science and some of which have to do more with sociology. Scientifically, it was a direct test of some components of Geographic Mosaic Theory, and it showed how that can be tested in nature. It also demonstrated geographic variation in ecological outcomes, not just traits. I think that helped to link ecological and evolutionary approaches to coevolution. The third reason is it was a fairly simple clear set of results. It is not the kind of analysis where you have to show, in detail, all your multivariate statistics and have a long discussion of residuals and statistical nuances. It was pretty straightforward as a paper once it was done. The final reason is it got published in a place like Nature, which means that it was immediately visible to a wide group of scientists worldwide. This last part is just the reality, part of the sociology, of life in science.
HS: I missed some names in the Acknowledgements, of some people with whom you discussed the manuscript. The first is R. Calsbeek.
JT: Ryan was a postdoc in my lab for a short period of time at UC Santa Cruz. He then went off to become a professor at Dartmouth University. Most of his work with me was on common patterns in the phylogeographic species across the landscapes of western North America. We were hoping to place the phylogenetic and geographic results for these saxifrages and moths into the broader context of geographic divergence in many other taxa. For this paper, he provided some helpful insights as the group of us in the lab talked about how best to analyze and present these results.
HS: D. Hembry
JT: David Hembry was an undergraduate student at Harvard who was working in my lab in summer and eventually went off to study different kinds of interesting pollinator-plant interactions in Berkeley, then in Japan, and then back in the U.S.
HS: K. Horjus
JT: She was a PhD student of mine who worked on phylogeographic divergence in Greya moths in western North America.
HS: R. Hufft
JT: She was, initially, a graduate student of mine at UC Santa Cruz, but her interests changed and she eventually moved to another laboratory.
HS: N. Janz
JT: He was a postdoc of mine at Washington State University, who worked on interactions between Greya moths and Heuchera plants. He was another of those in my lab who worked on the problem of how evolution of polyploidy in plants may shape plant-insect interactions. His work focused on how the relative colonization of diploid and tetraploid tents by Greya population in a region in Idaho where the moths seemed to speciating through specialization on Heuchera rather than Lithophragma. It was difficult work. He spend long periods camped out at the trailhead of a wilderness area, where bighorn sheep would sometime come into his camp and knock over his experimental cages, unless he had them protected.
HS: J. Richardson.
JT: He was a postdoc in my lab at the time working on phylogeographic differentiation in woodland star plants.
HS: Was Bradley Cunningham also involved in the writing and, if he was, how did you work together on the writing?
JT: I did the writing in this paper and Brad would write comments on the writing. That was the agreement right from the beginning. He was involved in the technical aspects, the field aspects and the analyses.
HS: Did this paper have any kind of direct impact on your career and the future course of your research?
JT: Nothing that I could directly identify. You would have to do a partial correlation analysis, right? I don’t think that I could point to any one outcome that came from this particular paper. Rather, it was part of a series of studies and analyses that helped build a broad phylogenetic, phylogeographic, geographic, and ecological understanding of the evolutionary and ecological process shaping adaptation and diversification in these coevolving interactions.
HS: Would you say that the main conclusions of this paper still hold true,more or less? I want to read out a sentence from the paper – you say, “These overall results confirm a major prediction from recent coevolutionary theory: that geographically structured species will tend to coevolve toward a complex spatial mosaic of coevolutionary hotspots and coldspots.” Would you say it the same way today, or would you modify it in some way?
JT: Having just had it read it to me, I think I would say it exactly the same way. I don’t think it’s because I’ve gotten stifled in my thinking. I think it’s more a matter of the fact that we have more evidence for this set of interactions and for interactions that others have studied. We have published follow ups on this, going back to some of the same populations again, in addition years, and reinforced those results, which is something I like to do when possible. So, I can even say it even more strongly now than we could have 15 years ago.
HS: If you redo these experiments today, and I’m guessing you have redone some of these experiments, what would you do differently now?
JT: Well, I would add components to it that I couldn’t have included in that paper because we were evaluating so many populations at once. The results in the paper are population-level results, providing a mean for each population. The ideal set of results would have evaluated the distribution of individual plant fitnesses within those populations. That would have involved recording, for example, all the seeds produced per plant relative to the number of visits by Greya moths. Because moths lay eggs in the flowers, the simplest form of that analysis would include all the flowers with and without Greya eggs. Some plants would have no flowers with eggs, some would have all flowers with eggs, and other plants would fall between those two extremes. Having the distribution of fitnesses relative to visit Greya moths visits for each population would provide a distribution of outcomes from antagonism to commensalism to mutualism. Those distributions are the raw material for the evolution of interactions, because natural selection would act to shape the distribution of outcomes toward mutualism, antagonism, or commensalism depending on the ecological circumstances such as presence of co-pollinators in some populations but not in other populations. That’s a different kind of paper from this one, but that’s what I’d add into that analysis, if I could redo it now with a large enough of crew of helpers.
HS: You say that the sites you sampled covered only a part of the geographic range of this interspecific interaction. Subsequently, have you sampled other parts of the range?
JT: We have sampled from the very northern edge of it up into Washington State and very southern British Columbia down into very southern California. There are a few known populations in Baja California, which we have not sampled, but otherwise we have sampled, over the years, the entire latitudinal range of this particular interaction. And right now (2016), we’re in the midst of writing papers that analyze up to 90 populations for traits – not ecological outcomes – but morphological traits and chemical traits, asking questions of how the interactions have diversified in places where you have the pairwise interaction between one Greya species and one plant species, to places where we now know you have networks, small networks of interactions, where you have, for example, two Greya moth species with one plant,you know, together. And so, it’s the beginning of understanding how small webs of interaction form. So that’s where it’s become a little bit more complex, in that way. But the idea has been to understand the overall diversification from the plant side and the moth side.And then what happens when you get secondary contact, is you get different combinations of plants and the moths. So it’s continuing and there’s more papers to come. (NOTE added in 2020: Some of the papers based on analyses of plant and moth morphological traits and plant chemical traits for more than 90 populations have now been published.)
HS: You say, “Although coevolutionary hotspots and coldspots have now been demonstrated in a few interactions on the basis of the geographic distribution of coevolved trait, the distribution of current coevolutionary selection is unknown for any interaction.The absence of such data precludes the development of moreecologically grounded models of coevolutionary dynamics. Apart from your work on this system, do we have similar examples from other systems, from other people’s work? And do we have better models of evolutionary dynamics that are grounded in ecology?
JT: Yes, we have very good studies from other labs, and I can give you a few examples. Craig Benkman and his colleagues have done on crossbills and conifers some of the most elegant work on coevolution that anybody has done. It’s probably work that you know because of your work on birds. In that case, we know who are the selective agents, and we know how and why they work – this is the ‘We’ of “We are all in this together”. Of course, it’s actually ‘he’ that knows and I’m a great admirer of that work. His studies have shown a geographic mosaic in the coevolution of lodgepole pines, squirrels, and crossbills. In regions in the North America Rocky Mountains where squirrels are found, lodgepole pine coevolve primarily with squirrels, and crossbills, which are often nomadic, impose only weak selection at most on the pines. In regions in the Rockies, where squirrels have long been absent since the retreat of the glaciers over 10,000 years ago, lodgepole pines coevolve with crossbills, In those regions, the crossbills tend to be sedentary rather than nomadic, and therefore become adapted to the local lodgepole pine populations. Craig and his colleague have repeated these studies on conifers, squirrels, and crossbills in other parts of the work and have found similar geographic mosaic in these other regions. Another excellent set of studies is the work on chemical coevolution between wild parsnips and parsnip webworms that May Berenbaum and Art Zangerl have done over the past several decades. I studied that interaction early in my career, and saw the variation in interactions within and among populations, but I did not have the chemical tools to understand what May and Art eventually learned about geographic mosaics of chemical defenses in these plants and chemical counter-defenses in the insects. Yet another beautiful set of results and ongoing studies is the work that the Brodies – father and son Brodie -and their students and colleagues have been doing with Taricha newts, which have tetrodotoxins, and garter snakes that vary geographically and their ability to detoxify these toxins. There is a growing number of other examples, but these are three of the best in which I think we really understand the selective drivers that shape the geographic mosaics of coevolving traits and ecological outcomes.
HS: What about the models? Do we have better models now, which are grounded in ecology?
JT: Yes, we do. Before I answer that, let me add one other study that I think is also one of the best for empirical studies of geographic mosaic, and that is Hirokazu Toju’s work in Japan on camellias and camellia weevils. Camellia fruits vary geographically in size, and the weevils, which have to bore, with their long snouts, all the way to the seeds, before turning around and laying their eggs on the seeds, vary geographically in snout length. Toju has a set of studies showing not only the co-variation of fruit sizes and snout lengths, but also some of the underlying bases for the geographic variation. That’s another beautiful example that’s been worked out very well, from mechanisms upwards.
Now, you asked me about theory. One of the things that I was criticized for in the 1994 ‘Coevolutionary Process’ book is I didn’t have good mathematical theory in that book. At the time, I was working on it with some colleagues, but we simply weren’t there yet, because it turned out to be complicated, initially, to think about what kind of models we should really be doing. We eventually solved that, and we started with a set of models in the year 2000, with papers I worked on with Richard Gomulkiewicz and Scott Nuismer. Richard was a population genetics faculty member at WSU, and Scott was a doctoral student of mine, co-advised by Richard. Scott has gone on to become one of the best coevolutionary modelers. Since then we and others have produced an ever more refined set of models on the geographic mosaic of coevolution. In recent years, I have extensively and intensively with Paulo Guimarães at the University of Sao Paulo, his postdocs and students, and several other colleagues on a series of papers that have begun to explore how coevolution proceeds when multi-species networks of species coevolve with each other both locally and across ecosystems. So I think we now have very good mathematical theory that generates predictions for future work, asking questions such as, what’s the effect of varying this structure of selection across landscapes. What’s the effect of varying the strain? What’s the effect of trait remixing so we can formally partition these G*G*E interactions, in a mathematical sense? The additional kind of theory that we are working on coevolution in networks of interacting species, but it is a wonderfully interesting challenge. That work goes beyond my mathematical skills, and it is joy to work with colleagues such as Paulo Guimarães, Jordi Bascompte, and others whose mathematical skills are well beyond mine.
HS: Have you ever read the paper after it was published?
JT: I have read the paper, but primarily in order to ask what components of it I wanted to put into the books that I was writing. But other than that, life goes on, and there’s more to do.
HS: Would you count this paper as one of your favorite pieces of work?
JT: Yes, I would. I started it with a specific goal of asking whether we could test some of the components of Geographic Mosaic Theory. The study turned out to be useful for asking that. I also wanted to understand whether ecological outcomes could vary, not just traits. The study also turned out to be useful for asking that. At yet another level, it was just tremendously fun to work in these beautiful environments for long periods of time, and to do so with interesting colleagues and students. For me, that’s part of doing this kind of science – if it’s worth doing, it’s worth doing in beautiful places.
HS: I missed asking you this earlier – could you give us a sense of a typical day in field was like during this study?
JT: It was a major logistical operation to do something like this. We had to be prepared with all the camping supplies food supplies, field supplies, and scientific supplies that we needed. We had to make decisions beforehand on which sites to go to next, based on our guesses on when plants would be flowering at each site this year. We drove in multiple vehicles, with different subgroups sometimes going to different sites and then all of meeting up at a campsite in the evening. Sometimes we would get to a site, and immediately realize that we could not sample the site, then, because the plants were not yet at the right stage for collection of floral capsules. We always had to have a contingency plan in place on where to go next if the site we initially planned to visit was not ready for collections. These sites were regions where we could call up someone and simply ask about the flowering state of plants. Once we were done working at a site all day, we would either return to last night’s campsite or find a new one. Everybody had a different job in setting up the camp – it was a semi-military-like operation. Some people would be responsible for setting up the tents, others setting up the kitchen and preparing dinner for that evening, and others for getting the samples ready for processing after dinner. We would do this day after day after day, and then eventually take everything back to the lab back in Pullman. There we would sort and store the samples, spend a day or two getting things done at home and organizing supplies, and then go back again into the field for another set of days. We did this week after week after week during field season, sometimes eventually losing track of what day of the week it is.
HS: What would you say to a student who is about to read this paper today? Would you guide their reading in some way? Would you point them to what’s happened subsequent to this paper? Would you add any caveats?
JT: I think I would tell the student to read this as one of the early studies trying to understand how coevolving interactions may diversify ecologically across landscapes, knowing that it doesn’t have everything that you would want in a full study of this sort. To really understand that, you need to read the subsequent studies from our lab and from other labs for what you need for true tests of geographic mosaic. This study, though, helped to show that coevolving interactions vary ecologically and evolutionarily among ecosystems, and it helped to show that, with hard work, it is possible to study these mosaics in nature, which are the fuel for the relentlessness of evolutionary and coevolutionary change.