In a paper published in Animal Behaviour in 2002, Niels Dingemanse, Christiaan Both, Piet Drent, Kees van Oers and Arie van Noordwijk showed, using an open field test in the laboratory on wild caught great tits, that, 1. there is consistent individual variation in behaviour, and 2. this behaviour is heritable. Fourteen years after the paper was published, I spoke to Niels Dingemanse about his motivation to carry out this study, memories of field work and what we have learnt since about the repeatability and heritability of animal behaviour.
Citation: Dingemanse, N. J., Both, C., Drent, P. J., Van Oers, K., & Van Noordwijk, A. J. (2002). Repeatability and heritability of exploratory behaviour in great tits from the wild. Animal behaviour, 64(6), 929-938.
Date of interview: 24 and 29 November 2016 (via Skype)
Hari Sridhar: What was your motivation for doing this piece of work? By looking at your publication profile, I came to know that this work was part of your PhD. How did this fit in with the other things that you did in your PhD and what was the specific motivation for this piece of work?
Niels Dingemanse: This is some time ago! This was research done between 1998 and 2002. We were starting to become interested in the idea that behaviour is not just plastic. Individuals might change their behaviour from one moment to the next, but also differ between individuals. And that was a very new idea at that time. We were studying these differences in exploratory behaviour and aggressiveness, partly because there were colleagues at work at the University of Groningen, who had been working on aggressiveness and so-called “coping styles” in mice and rats. From the work on the mice and the rats they knew that one could, for example, select for aggressiveness and find an evolutionary response by means of artificial selection, implying that aggressiveness was a heritable trait. And they also found that individuals that were very aggressive were also more explorative, more active and so on. They mostly reached these conclusions based on very detailed experiments that these very aggressive animals, versus animals that are less aggressive, are mostly different in their ability to use environmental inputs in taking behavioural decisions, hence the term “coping” style, referring to how animals cope with environmental change. So the idea was, essentially, that these aggressive animals are internally organized, whereas these less aggressive animals are much more plastic; their behaviour is guided more by environmental stimuli. And this makes them, for example, often less aggressive because they would assess features of their opponents before working out whether they would respond aggressively or not. And so, we were one of the first research groups working on this topic as part of big research collaboration between the University of Groningen, the Netherlands Institute of Ecology, and the University of Utrecht. There were people that worked on, for example, behavioural physiology (Dr. Claudio Carere), asking questions about whether hormones in eggs would affect the phenotypes of the animals that hatched. There were people working on quantitative genetics (Prof. Kees van Oers), or focusing on population ecology (Prof. Christiaan Both and myself). So, the grant that I worked on really tried to take the next step, and not simply establish the fact that there were individual differences. We already knew that from this work on the mice and the rats. We wanted to ask the next question, which was why are these differences there? Is there selection acting on this variation? Can we understand why this variation is maintained?
The objective of my PhD study, was to, first, ask whether, in the wild, we would find that individuals birds also differed in their behaviour. This was a very difficult task, of course, because we first had to develop some sort of assay that allowed us to measure their behaviour in a standardized way. The assay had been developed by Monica Verbeek, a previous PhD student who had worked at the Netherlands Institute of Ecology with my PhD-advisors Prof. Arie van Noordwijk and Dr. Piet Drent. Monica had been assaying animals that came from the wild but were hand-raised in the laboratory. Monica was finding the same types of patterns that people were describing for the rats and mice. Basically, explorative birds are more active in a novel environment, they are more willing to explore more objects, they do that more quickly, as well as that they’re more aggressive. Our 2002 paper was quite a challenging and risky bit of research. We captured animals with mist nets, animals that were born and raised in the wilds, took them to the laboratory and kept them there overnight. So, they arrived typically in the afternoon, or even in the evening, stayed overnight in the lab, and the next morning we did this behavioural essay. When we started to do these essays, we weren’t sure at all that these individuals would show repeatable variation. So, this was really quite challenging and even a stressful enterprise. Now, 15 years later, no one would be surprised if you do such a study and find repeatable and heritable differences. Today, this is well established, but at that time, it was not. The way we proceeded was that we then returned these assayed individuals back to the wild. And then, a year later, we would catch them again. By doing so, we could be able to compare whether individuals that were, say, exploring the world very fast the first time, were also like that the second time around; in other words, was this a repeatable trait?
I can tell you a bit more about repeatability and heritability. We were doing this study in a natural population, in which we would monitor the performance of the adults when they were breeding. We would ring both the parents and the offspring. This way, we could assay the exploratory behaviour of both the parents and their offspring provided we would capture them in autumn/winter (when we conducted the behavioural tests). Similarly, we could compare uncles and nieces, for example. Basically, this approach allowed us to ask whether relatives resembled each other in their exploratory behaviour. At that time, we were using simple statistical approaches such as parent-offspring regressions. Basically, you do a regression between the parent’s behaviour and the offspring’s behaviour and the slope of that regression line tells you the level of heritability. We used these types of approaches to ask not just whether the behaviour was repeatable, but also whether it was heritable. And that was, basically, what the paper was about: we found that wild great tits are repeatable and heritable in this behaviour. And this really a relatively new finding at the time.
There were, also, some other interesting findings in that study. We discussed how extremely plastic the behaviour was. One of our findings was that there was a strong seasonal effect. The birds were more explorative closer to the breeding season, namely, in early spring, rather than in autumn or winter. Moreover, this was a within-individual effect: the same individuals were more explorative closer to the breeding season. What is interesting is that we have replicated this finding for four populations. We have a paper, exactly 10 years later, which was published in Journal of Animal Ecology, with a whole suite of other researchers from the University of Oxford, the University of Groningen, Netherlands Institute of Ecology and University of Antwerp. This was because a number of people started doing this particular behavioural essay, based on our studies, And so, we now know that the seasonal effects, for example, are in every population. This is, you could say, a ubiquitous feature of this population with about 40% percent of the variation, roughly, due to differences between individuals. Another finding that, I think, is intriguing is that the animals were clearly showing signs of learning: they were modifying their behaviour as a function of experience. What we were seeing, and this is also replicated for the other populations, is that when animals were assayed a second time, they were much more active in this room: they were less inhibited. We also found that this effect of learning or exposure is something that they forget – as you might expect! When the interval between the two tests was roughly more than a year, then they scored the same as the very first time. It’s only when the interval was a bit shorter that they behaved much more actively. So, on the one hand, there seems to be some habituation effect, which could be due to handling or could be a response to the novelty of the room; we don’t really know. And this is, of course, something that you would expect if the behaviour is representing a response to novelty. This is, in a nutshell, what the study showed.
HS: How did you get interested in animal personalities and studying great tits for your PhD?
ND: The PhD that I worked on was an advertised position at a very prestigious institute. The Netherlands Institute of Ecology is really at the forefront of ecological research. At the time, I had the opportunity to also work as PhD student on another project, continuing the work on the frugivorous butterflies that I had initiated with Prof. Paul Brakefield (University of Leiden) in Uganda as an MSc-student. I decided that work on the great tits would, probably, be more interesting and profitable, partly because some of my advisors, at University of Groningen (Prof. Franjo Weissing), had the impression that research on personality would become very hot. Also, I would not working alone but rather in a team. There were three PhD students and a postdoc that were all jointly working on different facets; different sub-questions. And that’s, of course a very nice environment for a young PhD student to work in. So, I hadn’t been working on these individual differences before. It became an interest during those studies and, at this very moment, it’s still the major line of research in our laboratory.
HS: Was your PhD was from the University of Groningen?
ND: No, it was at the University of Utrecht. This was because my promoter, Prof. Arie van Noordwijk- was working at the institute, who was affiliated with Utrecht as a professor. My daily supervisor was Dr. Piet Drent.
HS: Tell us about the authors on this paper – who were they and what they brought to this study?
ND: The second author, Christiaan Both, was the postdoc on this project (now a professor at the University of Groningen). We were collecting data together with some division of labour. I focused on captures during the autumn and the winter period, while he focused a bit more on the collection of breeding season data. Piet Drent was my main advisor. He had also supervised Monica Verbeek, who was the previous PhD student who had done the more exploratory work, based on which this grant was granted. Kees van Oers was the other PhD student (now a professor at the same institute where I did my PhD). His work, mostly, focused on quantitative genetics. He was also based at the institute, but he did laboratory work. He worked with selection lines, which is a very important method to estimate heritability. He helped, partly, in the collection of the data, as well as with the analysis. Finally, Arie van Noordwijk was my promoter and the formal advisor of the study.
HS: How did you pick the two field sites in which you collected the birds?
ND: The different scientists in the group each had his or own study area. Piet Drent was working at the Westerheide and in Oosterhout. Since I was his PhD student, those areas became, naturally, our study areas too.
HS: You mention that there were 600 nest boxes put up in Westerheide from 1995 onwards. Were all these boxes put up by the research team?
ND: When I started my PhD in 1998, there was already an established population of great tits breeding in these 600 boxes. The habitat is mainly deciduous wood with some pine trees and some open landscape. These boxes were very loosely arranged into regular grids over a large area, which allowed us to work in the breeding season, and measure all kinds of breeding parameters of the birds in a standardized way, by checking the boxes and looking at a progression of all the nests. We also did field observations in the winter, for example, we determined dominance relationships based on social interactions on feeding stations baited with sunflower seeds.
HS: Was the laboratory where you did the experiments located in the Institute of Ecology?
ND: Yes, this was part of the laboratory where the birds were kept. This was where Kees van Oers conducted his PhD work on quantitative genetics. There were many aviaries where animals could be kept, rooms where animals could be raised by hand and two rooms that were empty except for artificial trees. The latter were used for these exploration tests. They are really just normal standard rooms that I used for behavioural research on birds. I think the nice thing about this setup was that that each room had multiple cages attached to it. We now have a setup of over 30 cages attached to one laboratory room, and this allows us to bring in many animals at the same time. We put them all in individual cages, and then, the next morning, we can screen them one after the other. That was the approach that we also used in this study. We would capture, maybe up to 20 birds, and then the next morning, we would test them one after the other. We did this for many days, in the autumn and the winter, and that allowed us to collect a very large sample size (hundreds of birds!), which is what is needed for these types of studies.
HS: Was this setup with the artificial trees and the attached cages already available, or was it something you designed for the study?
ND: It was available. I believe that it was designed by Monica Verbeek. She has an Animal Behaviour paper in 1994, where she describes this test for the first time. She shows that great tits differ in this exploratory tendency and that variation is correlated with aggressiveness and their response to novel objects.
HS: Would you know what has happened to the setup now?
ND: Kees van Oers did his PhD in Germany, returned to the institute, and is still working on these personality differences. He’s more interested in molecular genetics and genomics these days. But they’re still collecting the same data on this exploratory behaviour in the very same population in the Westerheide. This is, of course, fascinating because it’s rare that people can collect such long-term data. Some questions can only be answered with long-term data.
HS: Do you also continue to work in these field sites?
ND: No, I have my own field sites now. I moved in 2009 to Germany. I’m now a professor at the LMU – the Ludwig Maximilians University of Munich, where I work on crickets and great tits. The cricket work was, partially, initiated because one can keep them in the laboratory very easily and breed them in large numbers. This then allow doing quantitative genetics. We are studying great tits here in 12 nest box populations that I’ve put out, each with 50 nest boxes. This is something I always wished for because, when you have distinct populations, you can do treatments at the level of the population. For example, we’ve worked on perceived predation risk a lot, for the past few years, with replicates. We had six plots with one treatment – increased perceived predation risk – and six plots with a control treatment. If you have many populations, even if each is a bit small; only 50 nest boxes – it allows you to do research where you can manipulate factors at the level of the nest box population. This was not possible when I was doing my PhD because I was essentially working in one big population.
HS: I just noticed that paper where you compare city and forest, which sounds really interesting.
ND: That is partly based on these 12 populations, which are all forest populations, and one population in the city of Munich that was set up and monitored by Dr. Philipp Sprau. He worked as a postdoc in my group at the time.
HS: You said that you’ve also recreated the experimental setup with the artificial trees. Is that at LMU?
ND: I’m currently working on a five-year grant at the Max Planck Institute for Ornithology. Between 2012 or 2017, I have a research group there, and as part of that group, I created one of these laboratory rooms. The funny story there is that we replicated, exactly, the room we used in the Netherlands. The specific dimensions used, originally, had no experimental basis; they were just dictated by how these rooms were built years ago in Netherlands And, for the birds, the exact dimensions don’t really matter. But now we have the same room in Germany, as well as in Groningen, which we built some years ago, when I was working there. Replicating them, exactly, makes no sense at all but it does allow us to say that it’s an exact replica! This is what people like, right?
HS: When was the last time you visited the field sites where you collected the birds for this study?
ND: I think in 2002, at the end of my PhD. That was the last time that I collected data there. And then, there was a period of write-up in 2002 and 2003.
HS: Is it likely that the same individuals or individuals from the same population, that you used are still being used for research?
ND: Yeah, they are. They still use the descendants of the birds I worked with. They are still being monitored. One thing that needs to be noted is that these great tit populations are very open. That means that when you monitor the new breeders, only a very small percentage of them, typically 3-5%, were actually born there. Most new breeders come from somewhere else, probably from somewhere in the neighbourhood, but maybe even further away. In that sense, most of this population is simply being renewed massively every generation and, therefore, there will be very few descendants of the birds that I had handled myself. That was so long ago.
HS: Give us a sense of what your daily routine was during this work.
ND: This differed tremendously between seasons. In the non-breeding season, between February and March, we would map territories. We would walk through the forest, typically between six and eight o’clock, listening for sounds and interactions between birds. We would then note down where we saw birds, based on which we made territory maps. Typically, that would end by about noon, when the birds become much more silent and difficult to observe. In the breeding season, between April and July, we would work the whole day. I don’t remember exactly what time we started there. In our current field sites, we start preparing for fieldwork at 6.30, leave at seven, and then get back by two. Each of us had different tasks in the field. For example, one of us would monitoring the nest boxes. Once or twice a week, we monitored every box. After some events, like the first egg being laid, the whole program was enrolled. We would work out when the female started incubating the eggs to predict when the eggs would hatch. When the eggs hatch we would take a blood sample of the nestlings, and figure out when we had to come back to the nest to ring the nestlings at a particular age. Breeding seasons are very, very tough; there’s a lot of hard work involved. At that time, I was doing the field work, mainly, with Piet de Goede, who was one of the technicians, but also helped by other people like Piet Drent, Christiaan Both and many students. It’s the type of work that’s impossible to do without many hands. In autumn and winter, we would do a lot of “mist netting”. We would set up the nets early in the day, with little feeders of sunflower seeds to attract the birds, and drive between the different feeders monitoring the animals caught, typically between eight and noon. We would do that a couple of days per week, targeting different sites in the area, to avoid that habituation by the birds to the capture. And then, the birds would have to be transported back to the lab, which involved a 20-30 minute ride. The following morning, we would do the exploration test assays and release the animals back in the wild. So, the autumn and winter season was also used very heavily for doing field work. The problem of that particular study was that I was basically doing field work year-round. Therefore, analysis and writing-up was done relatively late. The good thing was that I had four or five years, in total, to do this study, so this was not a problem. But in many other countries, like in the UK or Germany, there’s only funding for three years for a PhD project. So, these types of studies are actually difficult to do here, for sure.
HS: Can you tell me a little more about the people you acknowledge – how you knew them and how they helped?
ND: Piet de Goede was a technical assistant. He, basically, taught me how to catch birds, how to ring them and also assisted with the captures, both in the breeding season and in autumn and winter. Henri Bouwmeester was someone who was hired temporarily, I think, every year, to help with catching birds. He didn’t have a permanent job at the institute, but he was hired every now and then. The other people – Leontien Witjes, Anna den Held, Kathelijnede Meijer, Maarten Hageman, Michel Geven and Dirk Zoetebier – many of them were students, from the University of Utrecht. Kate Lessells was a senior research scientist who still works at that institute. Her technician, Christa Mateman, did the genetic sex determination. Bart, Mary-Lou and Tanja were people who did laboratory work in the animal facilities. Jan Visser was a technician who worked on the management of all the long-term databases, to make sure that the quality of the data was consistent. ‘Het Gelders Landschap’ is the environmental agency who gave us permissions to do the work. Mrs van Boetzelaer was, I think, the owner of the Oosterhout area. This was a personal property.
Ton Groothuis was the professor who had written the overall grant proposal. Jaap Koolhaas did the work on the rodents, based on which this project was proposed. David Thomson was a postdoc working at the institute with whom I talked a lot about statistics. Joost Tinbergen, who was also at University of Groningen, supervised Christiaan Both. And finally, Marcel Visser was the head of our department. So, some of those people were directly involved in the project and some were colleagues who gave us advice and feedback.
HS: Were Bart, Mary Lou and Tanya students?
ND: No, they were animal caretakers. The students were Leontien Witjes, Anna den Held and Kathelijne de Meijer.
HS: Do you remember how long it took to write the paper, and where and when you did most of the writing?
ND: Yes, I did most of the writing at the institute, in-between the field work. It took me a long time to write this paper. Now, if I have solid time, I can really write a paper in a week without any problem; at that time, no way. I was very inexperienced. It took me many months, maybe three or four months, to write a paper. We submitted it to Animal Behaviour, they liked the paper, but we had to implement rather major revisions. As you know, this typically doesn’t change anything in the paper – the data are the data, the results are the results- it’s just written up in a slightly different way. So, I guess, all this might have taken me half a year or so. This is typical for a first paper from your PhD. It takes a lot of time because you’re not experienced as a writer.
HS: Were the other authors involved in the writing? If yes, how did you work together during the writing?
ND: The typical setup is that the first author writes the paper. In this case, I don’t remember exactly, but the paper was certainly co-written by Christiaan Both and Kees van Oers. Piet Drent and Arie van Noordwijk would, essentially, read our material and give feedback.
HS: Did all the authors ever meet together?
ND: Oh, sure. We were all in the same place. This was a local paper, you could say, based on research done in the lab.
HS: Did the paper attract a lot of attention when it was published? Was it considered controversial at that time?
ND: I don’t think so. This was a basic first paper. Only later on the paper became highly cited! The follow up papers were certainly noticed. For example, the one that talked about the links with dispersal was mentioned in the comments section in Nature.
HS: Would you say that it had any kind of direct impact on your career?
ND: I think the suite of papers did. They were the first papers that looked at whether personality differences, as we call them now, existed and whether that mattered. Whether there were links with metapopulation structure, dispersal and settlement, whether they were under selection, whether there was mate choice, and so on. I published four papers in my PhD, maybe five, and they’re all very highly cited. In about 2004, when I finished my PhD, when these papers were all published, there were a couple of other research groups, both in the US and elsewhere, that all started working on these individual differences. A couple of major perspectives and review papers were published, which led to the development of a whole new field that focused on the behaviour ecology of individual differences. As a consequence, these first papers are all very, very heavily cited. I certainly benefited from that. It’s very profitable to be known, especially, when applying for grants and so on. It enabled me to acquire some prestigious grants and allowed me to further explore this topic.
HS: Do you have a sense of what the paper gets cited for? Is it usually cited as being the first study in a particular area?
ND: I don’t think it’s often cited as the first study. I find this a very interesting notion. If, as a reviewer or editor, I read that, I say, come on, take that out, science is not really concerned with that. I think it is cited as evidence that individuals differ and this variation is heritable. Once, I looked up for what reasons papers are cited and it was very depressing, I stopped doing that! In 2003, I published a paper on dispersal, which showed that explorative animals disperse further, and that immigrants into a population are more explorative than locally-born animals. That paper is often cited for the idea that birds have personalities or that personality research is done on birds, which was not the point of the paper.
HS: Would you say that the main conclusions of this paper still hold true today, more or less?
ND: I think they completely stand. This is an interesting story, too. In 2012, we published a paper in Journal of Animal Ecology where we, essentially, repeated the same analysis for four populations, and also with more modern statistics. Our previous analysis was a bit inappropriately focused on using ANOVA approaches, but in 2012 we were using mixed models, for example. We were able to confirm our results – effects of learning, the seasonal effects and level of repeatability -through studies in other populations. We were comparing all these parameters across four different populations. One major problem in science is that studies are, often, not replicated. So, we don’t know whether a result is specific to the population studied, or to the species, or maybe even to the researcher who collected the data. This is why I’m pretty happy about this sort of confirmation, based on much more data from many different populations.
HS: You said that one of the things that you changed, in the new paper, was to use mixed models instead of an ANOVA approach. If you were to redo the experiment, today, would you change anything else about it?
ND: One of the concerns that we have with this behaviour we were measuring in this novel room is that we were not entirely sure what it represents. What we were assaying is how active the bird is in a novel environment. If I were to start studying this problem now, I would not do behavioural assays with birds that we capture and bring to the laboratory, because that’s a very stressful experience for birds. We have now developed assays that we can do directly in the wild. This has a number of advantages. The first one is that we are able to assay many more individuals. Also, all our behaviour essays were done in the winter. Only in winter, we could take animals out of the natural habitat and return them without affecting them too severely. If you do than in the summer, in the breeding season, they will, most likely, desert their nests. This meant that, because not all animals that we capture in winter become breeders, and not all breeders are captured in winter, our assays in the novel room represent a biased sample of the population. This was another reason we were interested in developing more field-based assays. These days, we carry a little cage with us to the field, to do field-based assays of this activity in a novel environment. This allows us to target every single breeder.
HS: Let me see if I’ve understood this correctly – you take the cage to the field, and do the assay of a wild-caught bird in the cage in the field?
ND: Yes, that’s what we do. In the breeding season, when the nestlings are between seven and 10 days old, we capture both parents. We ring the parents, measure their morphology, for example, wing length, the length of their tarsus, the length of their legs, etc. Before we do that, we put them in this little cage. This allows us then to get information about how active they are. We’ve done this now since 2010. So, we have now collected rather substantial datasets and we have various publications where we are using that information. We know, for example, that this type of behaviour also has very long-term repeatability: how animals respond in one year is actually predictive of how they respond in other years.
HS: In the paper you say, “Exploration tests in the field would provide an interesting new step to evaluate whether the behaviour in wild-caught great tits in the laboratory also extends to the wild.” From what you just said, it looks like that is the case, in terms of repeatability.
ND: Yes, certainly for repeatability. There are also other assays that other people have been doing, for example, measurements of how much ground animals cover in the field. And, generally, they’ve found that exploratory tests in the laboratory are predictive of movement patterns in the wild. This is something that we didn’t know at the time. But now, there are quite a few studies that show, for example, that dispersal distances, which is the distance animals move from the place where they’re born to the place where they first breed, is correlated with these types of assays, as well as other types of behaviours, such as home ranges and so on.
HS: You said that you realized the problem with novel environment tests in the lab and therefore decided to do it differently. Is this approach still used by other people?
ND: Yes. It’s very heavily used because it is a very quick and standardized test, but there is also criticism because it’s not so clear what the assay means. One way to overcome that is to make sure you collect a lot of additional behavioural data that allows you to interpret what this test means. To give you an example, we know from some recent work we have published in Behavioural Ecology that, if we increase perceived predation risk by broadcasting sounds of predators in the forest, this exploratory behaviour in the lab is affected. Animals decreased exploratory tendency when subjected to broadcasts of predator calls in the wild. That tells us that this behaviour has something to do with perceived predation risk. We are thinking that explorative animals are under increased risk of predation in the wild, and therefore these individuals down-regulate their behaviour when they perceive increased predation risk. We’re also finding, for example, that these fast explorers respond to increased perceived predation risk by lowering their body weights, whereas slow explorers do not do that. That, too, tells us that this behaviour is related to the probability of predation, that these fast explorers probably have an increased risk of predation. Now, we have some more supporting evidence. When we installed spy cameras in nest boxes, the slow explorers responded very strongly – they leave the nest boxes – whereas the fast explorers do not. This is completely in line with the notion that fast explorers might be more bold and willing to stay in situations that are a bit more dangerous. This is how we could reap the benefits of using these very quick standardized tests, but, at the same time, gain information about what they mean. It is important not to forget that there has been a lot of criticism on the specific usage of these standardized assays. We are using them because the questions that we ask, about selection and evolution on individual types, require massive amounts of data. They are “data hungry” questions. The statistical analyses (mixed models) are really data hungry. So, the only way to address them is to use an assay that can be executed extremely quickly on large numbers of individuals and lots of repeated measures, But then, of course, it is necessary to also cross-validate the meaning of that test by assaying other types of behaviours on the sites.
HS: In this paper you quantified exploratory behaviour through something called the ‘exploratory score’, which is the total number of flights and hops within the first two minutes. Do you use a similar kind of measure even now?
ND: Yes. We use an activity score. We also state, very explicitly, that this is just a label for their personality or their risk-averse behaviour, rather than that it represents a measure of it. It’s a proxy for the type of behaviour in which we are interested. People have also been criticizing the usage of this test and our metric, because it just reflects activity. Our point is that, as I said before, we’re interested in working out what an individual’s behavioural type is, and this is one way to do that.
HS: In one place in the paper, you mention that, “Heritabilities were calculated for Westerheide only, because sample sizes to do so were insufficient for Oosterhout.” Later on, did you also calculate heritabilities for Oosterhout, and is that published?
ND: No, we haven’t done that. I don’t even know whether this data collection was continued. We have published estimates of heritability for other Dutch populations, in later years – for example, the Lauwersmeer population, which is mentioned in the 2012 paper, and in another paper that’s published by Nicolaus et al. in 2012 in Proceedings B – it’s information in the supplementary material – and we showed that, in those populations too, behaviour is repeatable, and also heritable; heritability is about 10%. Also, other researchers at the University of Oxford and the University of Antwerp have published papers where they’ve looked at how heritable this behaviour is. And, if I remember correctly, heritability is between five and 10%.
HS: In the ethical note in the Methods section you say, “male great tits usually do not lose their territory when removed for less than 48 h (Krebs 1982; N. J. Dingemanse, unpublished data).” Was this published subsequently in another paper?
ND: No, we haven’t published that. The way this worked is that we made all kinds of direct observations. We captured these animals, released them, and then made some direct observations to find out what the animals are doing once they are released. From those observations, we got the impression that the birds simply regain their territory directly. It’s one of those issues that are important to mention in a paper – a methodological issue – but it’s not interesting to publish on its own merits. No one is interested in, you know, if there an effect of catching a great tit, bringing it to the laboratory and releasing it. Keep in mind, also, that we collect so much data that we have to prioritize what to publish, given that there’s only so much time in a single day!
HS: Similarly, in another place, you described the calculation of the F statistic based on unpublished data by D. C. Thomson, N. J. Dingemanse &A. J. van Noordwijk. Was this published later?
ND: In this case, we did actually write this paper. It was also, I think, reviewed by a journal and rejected. By the time we tried to publish this, there were other methods, particularly the one using the mixed model, which, really easily, allowed you to compare these repeatabilities. So, in subsequent papers, we have used the mixed model approach to do such comparisons. We have, for example, published a paper in 2013, a “How to” paper, in the Journal of Animal Ecology, which is a cooking book recipe that tells you how to measure individuality or calculate repeatability or compare repeatabilities. That earlier paper by Thomson is also one such paper where we showed how to do something. There was nothing wrong with the paper. It was not incorrect. It worked. But the method, basically, had become obsolete by the time we tried to publish it and, whatever we did, we didn’t move forward.
HS: In the beginning of the discussion, you say, “This is one of the first studies showing heritable variation in a behavioural reaction towards a novel situation in nonhuman animals from the wild.” Have other examples been discovered subsequently?
ND: Yes, there’s been an explosion of studies. There’s a really massive amount – hundreds of studies – that have been done using these novel environment assays, both for wild animals and captive. This is well known now. Today, we would not be able to publish that paper very easily in Animal Behaviour. At that time, in 2002, the simple demonstration of repeatability and heritability of these types of behaviours was very novel. But nowadays, this is something that we know; there have been done hundreds and hundreds of studies on that particular question. Now, we are moving very rapidly towards studies that don’t just ask whether these differences are there – these differences are typically demonstrated as part of the study and might be described in two sentences in the results – but what are the ecological or the evolutionary conditions that generate this variation. The simple demonstration of these differences is no longer an area of major research.
HS: Towards the end of the paper, you say, “The great tit is one of the few wild-living species for which we now have substantial evidence for a heritable basis of a behavioural trait both from a laboratory (Verbeek et al. 1994; Drent & van Noordwijk 1997) and a wild population (this study). Because individual life histories are relatively easily obtained in wild populations of great tits, these findings provide a unique opportunity to study the consequences of individual behavioural strategies for major life history traits, including survival, dispersal and reproduction.” Have questions such as these formed a major part of your subsequent research?
ND: Yes, they have. This was really the first paper that asked whether these differences were observable in the wild. Once we knew that individuals were very different, and that this variation was heritable, we could ask a whole new range of questions. For example, whether there was selection on this trait, whether selection varied between different environments, where it was fluctuating, whether this could help us understand why this variation was maintained rather than eroded by selection, and so on. The papers that followed were party measuring proxies for fitness, such as the reproductive performance of individuals, how well they survived, how long the different types lived and so on. And, really, those studies were one of the first ones to ask these types of questions. There was also work by Denis Réale, who is a professor at UQAM, one of universities in Montreal. He had been asking similar questions on docility and trapability in bighorn sheep, and he published his findings at the same time. Those two study species – bighorn sheep and great tits – together, really stimulated research on selection on individual behaviour in the wild. This is also why Denis Réale and I wrote quite a few papers in the years to come. We were some of the pioneers, really, in this field.
HS: Have you ever read this paper after it was published?
ND: No, I haven’t. I have, for sure, looked at the estimates in the table, partly because we have continued doing similar analyses with the same types of traits for different populations. We quickly look back, to see if the estimates are similar and so on. But I haven’t really read the paper. Sometimes, when you do read your old papers, you’re surprised by the way it was written. You think that you should have written the paper in a very different way. But this is always the case, with hindsight. We know much more about the perspectives that we might apply to the paper. Therefore, rereading my own papers is something that doesn’t make me too happy!
HS: Do you think your writing has changed from the time of your PhD?
ND: Yes, my writing has certainly changed. This is difficult to say objectively. I think, with experience, one really knows to clarify, much better, why the study was conducted. One is better able to make an argument in a way that people understand. You also want to get these people on board that maybe thinking in a very different way. I think that people that stay in the game, are the ones who are able to improve their writing. As you may know, your ability to publish papers is very much also a function of the quality of the writing. It’s not just about the data that you’re trying to publish.
HS: Would you count this paper as one of your favourites?
ND: In some ways. It was one of my first papers, so there is some nostalgic argument to be made. I look back on it and I see how we did things in the past, which was very different from how we do things now. That is not, necessarily, meant to criticize the paper, but rather to illustrate that we are learning as we go. I, sometimes, do that with students. I point out how we do things very differently now and how, maybe, in the past, we did some things that we now consider “wrong” – particularly in terms of statistics. But then, I emphasize the point that this was the best that we could do then. It’s good to look at papers very critically, but you should also always look at them from this more historical perspective. This is, of course, difficult if you’re a very young and naive reader; you don’t have this perspective. And that might, sometimes, make our students very critical, maybe more critical than what they should be. But, certainly, I sometimes look back at these papers, partly nostalgically, but also partly to understand how we did things in the past and how we have improved the way we do our research.
HS: You have already, partly, answered my final question – what would you say to a student who’s about to read this particular paper today? What should he or she take away from it? Would you add any caveats to that reading?
ND: When one is reading a paper, one should really look at from a historical perspective. This particular paper was one of the first papers, in a field that is now very well developed but did not exist at that time. Without that knowledge, one would probably get a very different impression of the paper. The caveats to the paper would be to do with issues related to doing laboratory assays on wild caught birds and bringing them back into the wild to study their fitness, which we have been addressing over the next few years after the paper was published. We, and others, have published various papers on how these concerns may be alleviated or addressed. We also, now, have changed, substantially, how we do the statistical analysis of the papers, both, in terms of whether we describe patterns versus a priori work out what sort of effects are muddled in our analysis, and in switching from frequentist approaches, where we were focusing, maybe obsessively, on p-values, to Bayesian approaches, where we are much more interested in estimating effect sizes. So, when I look back on this paper, I really see that the field has moved, mostly, forwards. That, I think, is a positive way to conclude a self-criticism of the papers that one has published.
0 Comments