In a paper published in PNAS in 2003, Jordi Bascompte, Pedro Jordano, Carlos Melián and Jens Olesen investigated the structural organisation of 52 mutualistic networks and found them to be highly nested, resulting in communities centred around a smaller core of interactions. Bascompte and colleagues also found that communities that contained a greater number of inteactions were more nested. Fourteen years after the paper was published, I asked Jordi Bascompte how he got interested in ecological networks, his memories of the collaboration and what we have learnt since about the nestedness of mutualistic networks.
Citation: Bascompte, J., Jordano, P., Melián, C. J., & Olesen, J. M. (2003). The nested assembly of plant–animal mutualistic networks. Proceedings of the National Academy of Sciences, 100(16), 9383-9387.
Date of interview: Questions sent by email on 19th December 2016; responses received by email on 5th January 2017.
Hari Sridhar: I would like to start by asking you about your motivation to do the work presented in this paper. Could you share with us how you got interested in the network approach at this point in your career, and, specifically, what motivated the work presented in this paper?
Jordi Bascompte: As with many other things in science, it was quite the consequence of chance and opportunity. I had just started my new Faculty job in Sevilla, Spain, after four years in the USA. This created the opportunity to start a long-term interaction with my new colleague, Pedro Jordano. Pedro and I immediately felt in love intellectually with each other and collaborating was just a matter of time. Timing was very good, as he and Jens Olesen – at the University of Aarhus in Denmark – had been compiling some data on plant-pollinator and plant-seed disperser interactions at the community level. The network framework was making a splash, and I was quite interested in networks from a more theoretical point of view. Again, timing was everything: we had a great biological problem, some really good data, and the right tool to address this question.
Our project was motivated by the realization that the main tenet was that mutualism has to lead towards either highly specialized one-on-one interactions or highly diffuse assemblages intractable to analysis. On the other hand, and from a theoretical point of view, it was surprising that while there was a sound community-wide theory on competition and predation, we were lacking such a conceptual framework for mutualism. The study of mutualism was full of natural history, but very much unrelated to major theories in community ecology. This is ironic, as we knew that interactions of mutual benefit had played a major role in generating biodiversity. And yet, they were studied depleted from such diversity.
HS: Stepping back a bit, could you tell us how you got interested in ecology, and specifically, in the analytical/theoretical/statistical aspects of ecology? The reason I ask this is because this seems to have been a major emphasis of your research right from your PhD days.
JB: My interest in ecology started very early. As a little boy, I was fascinated by the TV documentaries about wildlife by a great Spanish naturalist and communicator by name Félix Rodríguez de la Fuente. This initial fascination for nature matured across years into a more scientific dimension. At some point in this process, I become interested in theoretical ecology. Here my single biggest influence was the late Spanish ecologist Ramon Margalef. I happened to hear about him through a teacher in high school and through a relative who had worked with him in an insurance company before Margalef was formally appointed in academia (he had a late formal training). During an Eastern break, I read his book Perspectives in Theoretical Ecology, which was a transforming experience. Even when I could barely understand a few of those ideas, I was seduced by his elegant way to distil the relevant mechanisms in ecological systems and his way to relate them to other physical systems. It was like modern art, boiling complex ecosystems down to the single most relevant element while removing distracting details. I was very fortunate to get to know Margalef later on, during my college and PhD years, where he served in my PhD Committee. This was a major experience. My PhD supervisor, Ricard Solé, shared with me his enthusiasm about theory and provided an easy launching pad for me to start doing science. The fact that he was trained as both a physicist and a biologist certainly helped me to fully embrace theory.
HS: This paper has four authors. How did this group come together and what did each author bring to this study? Did you continue to work with your co-authors on this paper after this study?
JB: This paper was the starting point of a long-term collaboration on mutualistic networks with my two colleagues Pedro Jordano and Jens Olesen. As noted above, Pedro and Jens had started working together by the time I moved to Sevilla, and the three of us got together a few months after my arrival. That first meeting was like an explosion of ideas that we would develop in the next few years. The three of us complemented very well. Pedro and Jens were empiricists, one mainly working on seed dispersal, the other on pollination. I was the most analytically-oriented one. This blend of quantitative approaches and an exquisite understanding of the natural history of these systems is what made our work so unique. The fourth co-author, Carlos Melián, was my PhD student at the time. He was working on the structure and dynamics of food webs. Carlos provided some food webs that we used to compare the patterns in the mutualistic networks; they served as a sort of baseline comparison. In retrospect, I think that part of the reason for this long-term collaboration was the pleasant and enjoyable collaborative atmosphere. We had the feeling that we were after something really big, and enjoyed the process.
HS: The analysis in this paper is built on ’52 mutualistic networks’. Could you share with us how this dataset was put together, i.e. were you and the authors already aware of most of these papers from your earlier work, or did you specifically search for relevant papers for this study?
JB: This was the merit of both Jens and Pedro, who knew very well the literature and had started compiling a data set that has grown through years and is now publicly available in our repository www.web-of-life.es. Some of these networks were the fruit of the excellent fieldwork by Pedro and Jens. Other networks were provided by close colleagues or were already in the literature.
HS: How often did the authors meet during this study? How was this collaboration carried out?
JB: Pedro and I were constantly interchanging ideas and discussing about the paper almost on a daily basis. Pedro, Jens, and I had more temporally spaced meetings, either when he was visiting us in Sevilla or in conferences. After the original burst of ideas, the paper was pretty much unfolding on its own. I was leading it with constant input from Pedro and more periodic interactions with Jens.
HS: If you don’t mind, I would like to go over the names of people you acknowledge in the paper to find out a little more about who these people were, how you knew them and how they helped.
JB: This is a paper I cared about very much and for which we sought feedback. Several colleagues read the manuscript and provided important advice. Mainly, people I was acquainted with during my Californian days. I had only recently realized how critical writing properly was and how much I needed to improve. Many of the colleagues cited in the acknowledgments are really master writers and helped very much in those years to improve my writing skills. One of those names, Stuart Pimm, was one of the formal reviewers. He had a very important input as he made us realize that the expected hypothesis against which to compare our results was not a random network, but a compartmentalized one arising from tight, parallel specialization. This is reflected by a key phrase added to the abstract in the last minute. Other colleagues had provided insight into the analysis of data or theory of food webs. Particularly relevant was Thomas Lewinsohn, who was spending some time in our lab in Sevilla. He had been thinking in very similar terms and had a great understanding of these systems. His major contribution was to make us realize that the basic null model we were using to detect network patterns had a strong type I error (incorrectly rejecting the null hypothesis). This forced us to think carefully about the importance of null models and to develop a better one.
HS: Would you remember how long it took you to write this paper, and where and when you did most of the writing?
JB: As with all major papers, it was a long process. I tend to follow a research notebook, something I learnt from Ricard Solé. It is, therefore, easy for me now to go back to my notes and find it out. In doing so, I can determine that the first ideas on the topic of this manuscript came from an informal talk between Pedro and I in my office, towards the end of 2000. On 24 January 2001, Pedro, Jens, and I had this first brainstorm I was referring to in a previous question. That summer, while at NCEAS in Santa Barbara, I plotted the first preliminary figure and worked on the Table of the manuscript. The writing of the first draft took place a few months later in Lanzarote, one of the Canary Islands. I spent there a few weeks with Carlos, who is a native from the Islands, to write and discuss about science. We rented a small flat in front of the sea and enjoyed the beautiful surroundings of Caleta de Famara. With such a peace of mind, I could focus on drafting a first version of the paper. This was elaborated later on in Sevilla during several months. Around summer of 2002 we finally had a definitive manuscript that we submitted for publication.
HS: Did this paper have a relatively smooth ride through peer review? How did you decide to submit it to PNAS?
JB: It took us some time. The first draft was sent to Nature in 2002 and come back rejected shortly after that. In August that year, we submitted it to Science. It went out for peer review, but ended up rejected again. We then decided to repack it for PNAS, and submitted it to Editor Mary Arroyo on October 2002. The process was smoother at PNAS, albeit somehow slow. In due time, we received two enthusiastic reviews just suggesting minor changes. The paper ended up being accepted on June 24, 2003, almost three years after the time we started working on it.
HS: At the time it was published, did this paper attract a lot of attention and discussion?
JB: I remember a few close colleagues seeing the importance of the paper and already discussing its implications. The single most important example was John Thompson, who sent us an e-mail on 8 August 2003 (three days after the paper was published), which carried a strong significance to me. He wrote us that reading our paper left him quite excited and spurred him to write what he saw as a few implications for his soon-to-be published book The Geographic Mosaic of Coevolution. John made the point that our paper would go a long way towards dispelling the main view that mutualisms between free-living species should co-evolve as either pair-wise mutualisms or diffuse assemblages intractable for co-evolutionary analysis.
HS: This paper has been cited over 1000 times. When you were doing the study did you anticipate at all the kind of impact this study would have? Do you have a sense of what it mostly gets cited for?
JB: Indeed, I had the feeling this paper would become one of the most important contributions in my career. I think its impact comes from being one of the first manuscripts providing a conceptual framework to understand mutualistic interactions at the community level. It may also reflect the idea of finding of a simple pattern in complex communities, and its blend of quantitative and natural history approaches. One thing I did not anticipate, however, is its attention beyond ecology, namely, by people working in socio-economic systems. This has been a pleasant surprise.
HS: In what way did it influence the future course of your research? Does the analysis and interpretation of mutualistic networks continue to be a major area of research for you?
JB: It was a foundational paper in the sense of opening the field of mutualistic networks. It influenced my future work as the paper ends up by suggesting potential implications of the patterns we found. But to seriously advance into the dynamical implications of these patterns we had to work hard in the next few years trying to add an analytical approach. It has been a main line of research until now.
HS: Today, 13 years after it was published, would you say that the main findings still hold true, more-or-less?
JB: We have improved our measures of network patterns and the use of null models. We have also a more complete data set. More importantly, there is now abundant theory trying to make sense of these network patterns. But overall, I think it is fair to say the paper remains as valid today as it was back in 2003 in the sense of advocating the existence of simple patterns in the architecture of species-rich mutualisms.
HS: You found that nestedness in resource-consumer networks was lower than mutualistic networks and say “It is not clear whether this difference reflects a different biological organization or differences in sampling resolution”.
Today, do we have a better understanding of the differences between resource-consumer and mutualistic networks?
JB: Yes, our knowledge has improved. I would say that after considering more resolved antagonistic networks, the tendency continues for those to be less nested and more compartmentalized. I think it is safe to say that this may reflect some underlying biological difference. For example, as elegantly put by John Thompson, one would expect mutualisms to lead to nested networks through the combination of coevolutionary complementarity and coevolutionary convergence, while in antagonistic interactions the tendency is towards smaller groups of species being arranged in compartments through coevolutionary alternation, as for example when herbivores are selected to attack less defended plants.
HS: At the end of your paper you say “A nontrivial question that deserves further study is how the assembly pattern described in this paper affects the coevolutionary process in species-rich networks.”
Subsequent to this paper, did this form a focus of your research? How good is our understanding of how network assembly affects coevolutionary processes today?
JB: Interestingly enough, this is the point at which we are right now in our research, which implies we have been moving slowly! Indeed, up to now we have focused on the implications of these network patterns at the level of community ecology (e.g., How robust are these networks? How many species can they support?). We are now at a point where we are interested in focusing on the coevolutionary implications. A first preliminary step in this direction has been a paper published by Scott Nuismer, Pedro, and myself in Evolution in 2013. In this paper, we were using theory to address whether coevolution matters for large mutualistic networks. The rational for that was that although coevolution is acknowledged as an important evolutionary process for pairs of reciprocally specialized species, its importance within species-rich communities has been questioned. This preliminary result, together with a few others, suggests that indeed coevolution matters for species-rich communities.
HS: Have you ever read this paper after it was published? If yes, in what context? When you read it today, what strikes you the most about it?
JB: I have revisited the paper a few times and am surprised by how many ideas were already latent in the manuscript, albeit as suggestions. I have the feeling that our research in the few years after the paper was published sort of elaborated on some of these ideas.
HS: Would you count this paper as a favorite, among all the papers you have written?
JB: Certainly. This paper is my number one favorite.
Reflections on Papers Past 