In a paper published in Science in 2001, Tim Coulson, Edward Catchpole, Steve Albon, Byron Morgan, Josephine Pemberton, Tim Clutton-Brock, Mick Crawley, and Bryan Grenfell examined the effect of density, climate and demography on population fluctuations in Soay sheep on the St. Kilda islands in the United Kingdom. Coulson and colleagues’ found that age and sex distributions of populations can fluctuate independent of population size, and different ages and sexes respond differently to density and weather conditions. These findings, together, suggested that populations of the same size could show different dynamics in identical weather conditions. Sixteen years after the paper was published, I spoke to Tim Coulson about his motivation to conduct this study, how this group of authors came together and what we have learnt since about population dynamics of Soay sheep.
Hari Sridhar: How did you get interested in Soay sheep?
Tim Coulson: I’d finished my PhD, which had been on interactions between plants and animals, and I saw a postdoc advertised to work on the dynamics of these two long-term individual-based datasets, on Red deer and Soay sheep. I applied, and it was very fortuitous, I got the job. Then I realized what amazingly rich datasets these were, and how I would be able to use them to address a range of questions that people hadn’t been able to address with other systems before.
HS: Where was this postdoc based?
TC: It was based in London at the Institute of Zoology, which is the research arm of London Zoo.
HS: Tell us a little about the connection, if any, between what you did during your PhD and this postdoc.
TC: For my PhD, I had done a certain amount of statistical modeling, but also theoretical modeling. The theoretical modeling had asked questions about how survival rates and death rates influenced plant communities. So, the skill set that I had, although I’d applied them to plants, was what the people who offered the postdoc were looking for.
HS: Who was your supervisor during your postdoc?
TC: A guy called Steve Albon, who’s now retired, and a woman called Josephine Pemberton, who is now based in Edinburgh.
HS: Tell us a little bit about the motivation for this particular piece of work described in the 2001 paper,and how it was connected to what you had done earlier on the same data set.
TC: I’d started on my postdoc by asking what factors influenced survival rates – what attributes of individuals, but also what aspects of ecological environment, influenced which individuals died and which individuals survived, at a particular point in time. I’d done that and had published some papers on that. I probably then thought, why am I just interested in survival? And so, I then started thinking, well, what about reproduction? So, I did some papers that looked to see, a little bit, about factors that influenced whether particular individuals succeeded in reproducing or not in a given year. And again, I would ask what attributes of the individual and what attributes of the environment determine whether individuals bred or died. And then, I thought, well, having done survival and having done reproduction, why don’t I tie them together, and then, having linked them together in a model, use that model to see if we could understand something about the dynamics of the whole population. So, the work that had gone before, on trying to understand factors that influence survival and reproduction, was, really, the catalyst that pushed me to linking those insights together, to try and come up with a population model that could explain why the population of sheep fluctuated in the way that it did.
HS: Before you started this work, had you already seen Soay sheep in the wild? Did you visit this island during the time when you’re doing this work?
TC: Yes. I started my postdoc in 1994, and I’d been up to the island most years, if not every year, since then, spending between two weeks and a month there. So, I’d been pretty familiar with the island. And, in fact, by the time that paper came out, I would have finished my postdoc, and I did the work when I was based, on a fellowship, at London Zoo.
HS: Did the data come from an earlier period, i.e. before your visits?
TC: The project was started by Steve Albon, Josephine Pemberton and Tim Clutton-Brock in 1985. I got involved in the project in 1994 and then helped collect the data that went into the paper. I think, we went up to 1999 or 2000 with the data in the paper. The project’s still going, so data is still being collected.
HS: When was the last time you visited this island?
TC: Sometime early in the 2000s. I can’t remember exactly which year, but maybe 2003 or 4, I should think. I got to a point when I’d answered the questions that I was interested in answering with the Soay sheep and the Red deer dataset. I mean, I still use them occasionally for questions, but I wanted to move on to other systems, and, particularly, systems that were a little bit more complicated, in the fact that it’s not just sheep on the island, but there might be other species within the ecosystem. And also, to start working on systems where we could do some experiments to test some hypotheses. One of the weaknesses of the sheep and the deer system is that it’s only one population, and if you bring sheep into the lab, it kind of causes a bit of chaos.
HS: Has this island and the sheep population changed, in any significant way, since the time you of this study?
TC: We have had some subsequent papers.The sheep had, on average, been getting a little bit smaller over time, and that population had been creeping up in size. We did subsequent papers – in fact, one in Science and one in Nature, if I remember correctly – where we examined why body size had been changing in the way that it had.
HS: If you were to redraw Figure 1 for the 10 years following the study, would you say it would look, more or less, the same?
TC: More or less. There’s a bit more of a trend.There’s a few more animals on the island than there would have been back then, probably because of the winters being less harsh than they used to be. But in terms of the ups and downs and the fluctuations: yes. That population still fluctuates substantially.
HS: How did this group of authors come together and what did each bring to this study?
TC: I can’t, actually, remember, exactly, who the authors are. Tim Clutton-Brock, Steve Albon and Josephine Pemberton are on the paper for setting up the project on the island of St Kilda and collecting such detailed individual-based data on a large mammal. It was their motivation to, actually, dive in there and do it. I would have meetings with all three of them, and we would discuss the analysis that I’d done and the results.Bryan Grenfell had done some previous analyses of the population dynamics of Soay sheep, and Bryan and I worked quite closely together. In fact, Bryan provided quite a lot of inputs on testing how well some of the models worked. Mick Crawley had been my PhD supervisor and also worked on the system. He’d provided some insights into the likely drivers – which were the variables to use and what have you. I think Byron Morgan might have been on the paper. He’s a statistician. We’d worked together with him, quite a lot, to conduct mark-recapture analyses, which would provide insights into what influenced survival and reproduction. I think Ted Catchpole might have been on the paper as well. He was another statistician who provided input into the statistical analyses of the data, earlier on.
HS: Could we also, quickly, go over the names of the people you’ve acknowledged, to get a sense of how you knew them and how they helped?
TC: Yeah, you have to tell me who’s in it.
HS: First, you thank the Royal Artillery for logistical support. Can you tell us a little more about this?
TC: Yeah, that’s how we get to the island. In order to get to the island, and to get food to the island, in particular, and get equipment to the island, we relied on the army. The reason for that is St. Kilda was an army base. They have a big radar installation on the island. And, actually, the reason it was possible for us to do all of this work on the island was, to a large part, because we did have frequent help from the military, in terms of getting ourselves to the island, getting equipment to the island, getting food to the island, but also, it was the Army’s generator that provided electricity and the shower block that allowed us to wash and what have you. We were very grateful to them for providing the logistic support that allowed us to conduct research on the island.
HS: J. Pilkington
TC: Jill Pilkington was the woman who did most of the data collection for us on the island. She lived on the island for about six to eight months a year. She was instrumental in ensuring the data continued to come in.
HS: A. McColl
TC: Andrew McColl did the same job as Jill Pilkington, but he did it before she took over
HS: A. Robertson
TC: Same deal. He was involved in data collection at some point in the project.
HS: Were they all students at the Institute of Zoology?
TC: No. Jill Pilkington, at that point, I think, was employed out of Cambridge. She was a technician on the project. Andy McColl had done his PhD on the project. He’s now at Nottingham. I don’t know what’s happened to Robertson.
HS: You also thank other volunteers for collecting the data. Were they also other students from these universities?
TC: They were a mix of people. When we go up in the summer, to catch the sheep and take measurements, we’d put together an expedition of, maybe, a dozen people.They were people who volunteered to come and help on the project. Some of them had finished their undergraduate, some of them were graduate students, some of them were even faculty, who wanted to come to the island to do some field work and help out.
HS: And then there are some people you thank for comments on the manuscript. The first name is O. Bjornstad.
TC: Ottar Bjornstad. He’s a population dynamics modeler. He’s very good. He’s now based at Penn State in America. He had done various bits of population modeling. and I, probably, sat down in the pub at some point and discussed the work with him. I must have shared the manuscript and he would have provided some comments and feedback.
HS: Was he in the same institute as you, at that time?
TC: No. He was based in Oslo, but he came over to Cambridge quite a lot. If I was up in Cambridge, seeing Bryan Grenfell or Tim Clutton-Brock, I would often meet up with Ottar. He’s a good friend. He became a good friend around that time.
HS: S. Freeman
TC: Steve Freeman is a statistician who had been doing a postdoc with Byron Morgan and Ted Catchpole – or a PhD; I can’t remember which. He would have checked the statistical analyses that we’d done, just to cast her additional eye over them.
HS: M. Forchhammer
TC: Mads Forchhammer. He was a postdoc at Cambridge, who was also working on the sheep population at the time, and has some interest in the population dynamics. I suspect I showed the paper to him, out of a matter of courtesy, and he would have provided some comments that, obviously, helped improve the manuscript.
HS: A. Illius
TC: Andrew Illius. He was up in Edinburgh. He, again, had done some work on the Soay sheep population in the past. I must have showed him the paper, as well, and he would have provided some comments and suggestions.
HS: J. Lindstrom
TC: Similar deal as Mads Forchhammer. He was a postdoc who was working on the Soay sheep and Red deer projects, who had interested in population dynamics. I asked him for comments.
HS: I. Stevenson.
TC: Ian Stevenson. He did his PhD on the project and had been involved in a lot of data collection. I must have discussed the paper with him at some point too.
HS: K. Wilson
TC: Same as all the others. He was based in Cambridge, at the time, doing a postdoc.
HS: Did all the authors ever meet during this study?
TC: Yes. Every year we have an annual sheep meeting, where everyone who’s involved in the project would get together for a few days. People would give a talk about the research they were doing. I would have given a talk about that research as it was ongoing. And then, we would have had a discussion with those people and everyone who was involved in that paper.
HS: Would this happen at the Institute of Zoology?
TC: It would move around between different places. Sometimes we’d do it at the Institute of Zoology, sometimes we’d do it in Cambridge, sometimes in Edinburgh, and sometimes in Sheffield. This is a very big project with lots of people working on it. So, we’d take turns for different groups to organize it.
HS: Do you remember how long it took, from the time you started the analysis,to getting it published?
TC: I remember we submitted it first, we got, sort of, a reject and resubmit, we went away and did a bit more work, and then it went back to review again. I think we got minor revisions.We may have even submitted it to Nature first – I can’t remember exactly – but they rejected it with some comments. It was probably a year to 18 months. I was doing other things at the time as well, but in terms of putting that manuscript together, I would say, something like that.
HS: Did you take the lead on the analysis and the writing?
TC: And the modeling.
HS: Did you do most of this work at the Institute of Zoology?
HS: What do you remember about the peer-review of this paper at Science? Would you say it was a relatively smooth ride?
TC: It was a relatively smooth ride. We got some good comments on the manuscript after the first review, if I remember correctly. I think Nils Christian Stenseth was a reviewer, and I think he wrote a News and Views on the paper. Is that possible? Maybe, it was a Perspectives piece. I don’t know who the other reviewer was. The only reason I know that it was Nils is because he wrote a commentary on the paper. I think I emailed him and said I enjoyed his commentary, and he said he’d reviewed it; he didn’t sign it, at the time. The reviewing process, from memory – it was some time ago now – worked well. The comments were designed to be, and were, helpful. Sometimes you get reviewers who perhaps have an agenda that they don’t want to see a particular bit of work published in a particular journal, but we didn’t experience that with this paper at all.
HS: I was also curious about the striking title. Do you remember if there was a lot discussion on this, among the authors, or with the journal?
TC: We probably did go backwards and forwards. I think Science made us change the title a little bit. I have no idea what it was before.
HS: At the time it was published. do you remember how it was received? Did it attract a lot of attention?
TC: Yeah, I mean, it certainly received some attention. As I said, it had a commentary about it. I was asked to give various talks. It was before the time that blogging was widespread. Twitter hadn’t been invented. Facebook wasn’t around. It must have got some attention, because it’s sort of, been reasonably consistently cited.
HS: Do you have a sense of what it mostly gets cited for?
TC: Yeah. It’s many things. It’s about doing something first. And I think what we did first was to construct a realistic matrix Model, which had stochasticity and density dependence and population structure in it. I think it was one of the first realistic population dynamic models out there. We also showed that the model did okay in capturing the dynamics of the population. So, you know, I think it’s precedence. Lots of people have done similar models, subsequently, but because we got there first, they will often come along and they will cite that paper for the method side of it. But I also think it was one of the first papers to show that density and climate naturally interact. There’d been a lot of debate in the literature about the relative roles of climate and competition for food. What we demonstrated in that paper is that those two things interact. And it’s not the fact that one is better or more dominant than the other. They both can be important and can interact with one another. So, I think it was a nice timely, ecological message, as well as a sort of nice methodological demonstration that this sort of thing could be done.
HS: Would you say that this paper played an important role in your career?
TC: I’m sure it must have played a role; probably that, along with a number of other papers. It made me realize that demography had to be central to all ecological and evolutionary dynamics, that it’s at the center of biology, and that, in order to understand both ecological and evolutionary change, you need to focus on who lives and who dies. Those are the fundamental biological drivers. You need to identify the factors that influence who lives and who dies. My first postdoc would have been when I started doing this work. I then moved on to a fellowship at the Institute of Zoology. I left the Institute of Zoology in early 2001, I think, and moved up to Cambridge University, where I was on a fellowship, and then moved on to a lectureship. And then after that I moved on to a position at Imperial College. I’m sure, having reasonably well-cited papers in Nature and Science doesn’t do any harm when it comes to looking for jobs and what have you.
HS What influence did this paper have on your research trajectory, both in the short- and long-term?
TC: In the immediate aftermath, I used a similar approach, with some collaborators, to look at the dynamics of a Red deer population. We used those papers to look at the effect of hunting on Red deer populations. We published a couple of papers in Nature from that. Then, I sort of thought, well, okay, as well as using age, what about using phenotypes, like body size or behaviors, within these models? In 2000 – actually, just before this paper came out – there’d been a very nice paper in Ecology – the first author was Mike Easterling – which showed how you could develop these things called Integral Projection Models,and I realized that those models offered great power to understand how phenotypes influence survival rates and birth rates. And we did a series of papers on a number of species, including marmots and wolves, and, again, we published those in Nature and Science. So, I suspect that the approach, and the success I had with that early paper in 2001, made me realize that this is a very, very flexible structured modeling approach that can be used to address a huge number of questions in ecology and evolution. I think, sort of subsequent to that, I’ve now shown how these models can be used to link the dynamics of genes, right the way through to the dynamics of ecosystems.
HS: Would you say that the main conclusions of this paper still hold true,more or less?
TC: Yeah, absolutely. The dynamics of populations are caused by a number of factors and those factors interact. So, that’s absolutely right.
HS: Has a similar kind of analysis been done with data collected after 2001?
TC: Not that I’m aware of. Unless we saw a substantial change in the population dynamics, I’m not sure anyone would dive back in to see. I think we’ve identified what the key drivers are. You may be able to refine things now.You may come up with a slightly different age structure, you may come up with different climatic drivers and what have you. But I don’t think the logic of that paper changes. I don’t see any reason why anyone would necessarily go back and do the same thing over and over again.
HS: If you were to redo this today, would you change anything about the way you did the analysis?
TC: If you were to do it again now, you could probably parameterize the whole model in one go, rather than doing piecemeal statistical analyses for each function that goes into the model. So, I suspect that if you did it now, given the advances in statistical methods, you would probably use a different statistical approach. My hunch, though, is it would probably give you very similar insight there.
HS: In the paper you say, “These results underline the importance of incorporating the age and sex structure of the population into models.” And later, you say,“Any attempt to predict future population size or the effects of climate change in this or similar systems must therefore take account of the age and sex structure.” Subsequent to this study, has it become the norm to incorporate age and sex structure into such models?
TC: Yes. If we go back to, sort of, the early 2000s, when this paper came out, what was very popular was to fit a, type of, time series model, called an autoregressive model. The structured model that I used in this paper was starting to gain interest, more from a theoretical perspective. And, in fact, in the same year 2001, Hal Caswell published a very influential book on this class of models, i.e. matrix population models. And, I think, because of his book, which demonstrated and how flexible these were, and, possibly, because of this paper, because it described how you’re able to gain novel biological insight by using these models, these models have now become extremely widely used. So, yes, I think that things have moved on, and people have understood the power and the strength of this structured modeling approach. And, of course, it’s not all down to this one paper. As I said, Hal’s book was certainly much more influential paper, but the paper probably did play a small role.
HS: In the last sentence of your paper, you say, “the processes we describe are likely to exist across a wide range of large herbivore populations.”What kind of evidence do we have for this today?
TC: There have been similar models for several herbivore populations that have been studied in detail, including Roe deer populations in France and other parts of Europe, Bighorn sheep populations in Canada, Kudu populations in Africa, etc. So, yes, this type of model has been used quite frequently for herbivores, and they’ve come to the same sorts of conclusions that we arrived at.
HS: Have you ever read this paper after it was published?
TC: No. I’ve used the figures in some undergraduate lectures that I gave.
HS: Would you count this paper and this work as one of your favorites?
TC: I guess, I like the biological story. Like most people, I’m probably proudest of the paper that I’m working on at the moment.The one that I find most exciting is the one that I’m writing at the moment. But, I mean, yeah, I like this paper. If I ever go and look at my citation records on Google Scholar, it’s the one that comes out top, so, yeah, I must quite like it!
HS: What would you say to a student who is about to read this paper today?Would you guide his or her reading in any way, maybe, point them to other papers they should read alongside this?Would you add any caveats to their reading?
TC: It depends on why they’re reading it.If they’re interested in learning about population dynamics and structured models, yes, I think I would suggest they read this. I would probably give them some theoretical papers by a guy called Shripad Tuljapurkar. I would probably give them Hal Caswell’s book. And I would probably give them some synthetic reviews of animal and bird populations that have come out, more recently around the same time, by Bernt-Erik Saether and colleagues, and by Jean-Michel Gaillard and colleagues.