In 2008, Nathan Kraft, Renato Valencia and David Ackerly published a paper in Science providing field evidence that suggested that niche-based processes structured the plant community in an Amazonian forest. Kraft et al.‘s study came at a time when there was a debate raging in the ecological community about the importance of niche vs. neutral processes in structuring biological communities. Eight years after the paper was published, I spoke to Nathan Kraft about how he got interested in this topic, memories of field work and what we have learnt since about the processes that structure tropical plant communities.
Citation: Kraft, N. J., Valencia, R., & Ackerly, D. D. (2008). Functional traits and niche-based tree community assembly in an Amazonian forest. Science, 322(5901), 580-582.
Date of interview: 8th August 2016 (on Skype)
Hari Sridhar: I would like to start by asking you what your motivation was to do the work presented in this paper. By looking at your publication profile, I came to know that this was work you did during your PhD. Could you tell us how you decided to work on this topic for your PhD?
Nathan Kraft: Sure. My interest in that study during my PhD came from two different threads. As an undergraduate, I had spent a month in the Amazon, right near where I ended up working, and that really inspired me to think about the diversity in those forests and to try to make sense of the hyperdiversity that is found in places like the Ecuadorian Amazon. In terms of the intellectual motivation, before beginning my PhD in 2003, I read Steve Hubbell‘s book that had just come out in 2001 on Neutral Theory. That really challenged, in many ways, how the field was thinking about the structuring of diversity in tropical forest communities. That was part of my interest, to really think about the challenge that Steve had laid out for everyone, basically asking: if we imagine a world without niches, how much diversity do we expect to see in these communities? That was a big inspiration. At the same time, there was also a thread of trait-based ecology that was developing, both in the lab where I was working on my PhD and in the field in general. David Ackerly was my PhD advisor and he was quite instrumental in pushing those approaches. Will Cornwell was a PhD student a few years ahead of me in the lab who was also working on similar kinds of analysis in coastal California. For me personally, I think I saw a question I was very interested in, in a place I was excited about, and the opportunity to use some new tools to think about how we might be able to use traits to understand communities in these hyperdiverse forests.
HS: Were you always interested in plants, or did you feel that plants were the ideal system to address these questions?
NK: I think I was interested in working in ecology before I necessarily had a taxonomic group in mind. I didn’t really start working on plants until my PhD, but plants are ideally suited to these high diversity community ecology questions because they don’t move and you can find them anytime you want!
HS: You say “Few large-scale tests of coexistence theories in tropical forests have explicitly examined the ecological strategy of co-occurring species, in part because of difficulties in identifying more than a few discrete plant strategies (such as shade tolerant, light-demanding pioneer, etc.) (6). Recent advances in functional ecology now permit a more precise quantification of woody plant strategy along a number of continuous, often orthogonal, axes of variation related to resource acquisition strategy, regeneration niche, environmental tolerance, and life history (7–10), opening the door for previously intractable analyses.”. Could you tell us a little more about these developments in functional ecology at that time?
NK: I am not sure quite how far back to go, but I would say, beginning in the mid-90s there was this interest in developing a suite of functional traits, which are components of the phenotype of an organism that are indicators of variation in ecological strategy across species within communities and across species globally. There was a lot of work around that time in developing our understanding of what these traits are, what they are correlated with ecologically, how best to sample them, and how plants and communities varied in terms of these traits globally. The promise of the approach at the time was that the traits are fairly straightforward to measure, but have more information than you might expect given how easy they are to sample. That hasn’t always been borne out in the years since, but that was certainly the motivation at that time.
HS: I want to ask you about the authors on this paper. You have already told us that David Ackerly was your PhD supervisor. Can you tell us a little bit about the role of Renato Valencia in this paper?
NK: Renato is a professor at the Catholic University in Ecuador. He is the principal investigator for the long-term forest dynamics plot where that study was based. He had led the efforts, especially around the time when I was working there, to conduct the re-censuses of the 25-hectare forest where all the trees were mapped and tagged and identified. That spatial and taxonomic information was vital to our ability to go in and do that kind of study. We could not have done it without that foundation. Renato helped with access to the plot, and thinking about how to go about doing it, approving the protocols, commenting on the analysis, and interpreting what we found.
HS: Did you do most of the fieldwork for this project by yourself?
NK: Yes, I led all the leaf functional trait sampling that is in there over three summers – 2005, 2006 and 2007. In the summers of 2006 and 2007 I brought field assistants to help and they were fantastic. The seed mass trait data that is in there comes from a long-term project run by Joe Wright and Nancy Garwood. They declined to be co-authors on this paper, but they are acknowledged at the end.
HS: Could you give us a sense of what a typical day in the field was like during this project?
NK: Sure. The Estacion Cientifica Yasuní is just a few kilometers walk from the forest plot. This is a field station run by the Catholic university in Ecuador based in Quito. It’s a fantastic place to live and work. There is a herbarium, a dining hall, support staff, air conditioning – it is a very comfortable place to work. The typical day, when I was working there, would be to get up around six in the morning, have breakfast, and go out into the field till about one in the afternoon. During that time in the morning, we would go out with lists of randomly selected individuals of species I was trying to sample in that time. We would go out and try to find them in the plot, check their condition, see if they were in ok shape to sample, and if so, sample their traits. We would bring some samples back for nutrient analysis as well. Then we would come back around one for lunch and then work in the afternoon to process the samples from the morning, and to prepare a sampling plan for the next day if needed. That was how many researchers tended to work at that station, where you work a long morning and then you have time in the afternoon in the lab. That seemed to be the most sustainable pattern over a few months.
HS: Did the three authors ever meet as a group during the study?
NK: No, I don’t think that we all were ever in the same place. David had quite a lot of things going on at the time, so we spoke and emailed a lot about plans but he never actually came out to the site. I know that David and Renato met at some point, but I think it was after this paper was written.
HS: Did you do most of the writing when you are in Ecuador or after coming back to UC Berkeley?
NK: I wrote it when I was back in Berkeley. We completed the sampling in 2007 and the paper was submitted in 2008. There was some time to work on the analysis. I think I did most of the writing of the first draft in my house in San Francisco, where I was staying at the time. And then there were revisions and lots of help and input from my co-authors on the drafts as well.
HS: Was it all done over email?
NK: I am fairly certain that I would bring drafts of it into the Ackerly lab meeting. I think it was workshopped in the lab group at least once or twice. And I would meet with David, usually in person, to discuss drafts and he would also send me comments as tracked changes. Renato would send suggestions and comments as tracked changes over email.
HS: I would like to go over the list of people you have acknowledged to get a better idea of who they were and how they helped in this project? Can we do that?
NK: Sure. Let me pull the paper up.
HS: The first set of people – A. Martin, L. Williams, P. Alvia, and L. Dunn – who you thank for field assistance, were they all students from UC Berkeley?
NK: Adam Martin, at that time, was in between a master’s and PhD at University of Toronto. He volunteered as a field assistant for me in 2006. Laura Williams was an Australian ecology student. I think she had just finished her undergrad and was beginning to explore doing a PhD. She helped me in the summer of 2007. Pablo Alvia has been associated with the forest plot in Ecuador for decades now. He had worked a lot on the plot and also on identification. He helped me in a number of ways, but, in particular, he helped me to identify some larger individuals just adjacent to the plot, that we wanted to get some samples from. Laina Dunn was an undergraduate who came to volunteer with us for a few weeks in 2007.
HS: The next three names are of people with sample analysis…
NK: Andy Thompson helped us with preparing the samples for nutrient analysis. He ran the analyzer machine that we were working with at Berkeley. Tania and Matt had both helped us analyse some of the images that we took of leaves to calculate leaf areas. They were both associated with Berkeley as students. I think Matt was not enrolled at the time but he was living in Berkeley and interested in the research experience.
HS: Tell us a little about the Centre for Tropical Forest Science (CTFS).
NK: CTFS is part of the Smithsonian Institution. The CTFS network of forest plots grew from of Steve Hubbell’s initial work in Panama to its current form as a network of forest research plots located all over the world, following a standard methodology. There is a lot of collaboration and interaction among the people who work in those plots. I believe CTFS helped in the establishment of the plot initially and also give me some funding for my trait sampling. They had a small grants programme for people working in the plots.
HS: You also thank a number of people for the interactions you had with them about the study. Could you tell us a little more about them?
NK: Will Cornwell was a PhD student in David Ackerly’s lab who helped a tremendous amount, especially in terms of thinking about the kinds of null model analysis to do. He had done very similar things in a community in California and his work was really the inspiration for the analyses that are there in this paper. Paul Fine was a professor at the time at Berkeley who was on my committee, who helped me with the revising of the manuscript. Steve Kembel was a postdoc at that time with David Ackerly who has worked a lot on phylogenetic community structure analysis and null models. He helped me to think about the analysis. Jeff Lake was one of Steve Hubbell’s PhD students at the time, who had a number of comments and thoughts about the analyses and their interpretation. We had spoken about some of the results before I had submitted the paper. Margaret Metz is a great friend. She was doing her PhD in Ecuador at the same time at the same site. She actually encouraged me to work there. She works on seedling Dynamics within the forest plot. We had overlapped quite a bit in the field and also at Berkeley and she had helped me to really get oriented there and to navigate all the challenges of working in that site. Nate Swenson was a PhD student at the time, finishing up with Brian Enquist at Arizona. He was interested in very similar kinds of analyses working in other neotropical forests, so we had begun to interact a bunch about these kinds of studies that they we were both beginning to do.
HS: What about Richard Condit?
NK: I don’t know if he is still listed as co-principle investigator, but at that time he was a co-principal investigator of the plot. He has really been instrumental in getting this network of plots up and going, helping everybody with how to do the sampling, how to handle the data, how to analyse it and so on. He was actually the first person to introduce me to the site in person. He happened to be at the site, the same time that I was out there, and he walked me around the plot and showed me everything. Both he and Renato were involved in giving me permission to work there.
HS: Which university were SJ Wright and N Garwood part of?
NK: Joe Wright is at the Smithsonian tropical Research Institute in Panama. He does a lot of work, especially in neotropical forests. He has worked throughout the network at this point, but he is especially known for his work on Barro Colorado Island (BCI) in Panama. Nancy Garwood is currently at Southern Illinois University in Carbondale. She has been working on seeds and seedlings in Yasuní for a long time.
HS: Could you tell us a little bit about the grants you acknowledge in this paper?
NK: All the grants that are listed there, in that last sentence, are basically the acknowledgement for the forest census work. So, they all come from Renato, essentially. None of those grants mentioned were specifically for my project, but they enabled the collection of the spatial and identification information that was essential to the analysis.
HS: Did this paper have a relatively easy ride through peer-review? Was Science the first place you submitted it to?
NK: Let me check. I believe we tried at Nature first and it was rejected without review. And then we sent it to Science. The review was pretty rigorous and there were a number of concerns that came up that we had to address. I don’t think it was the hardest paper that I had to get through the process, but it was not the easiest. It was about average I think, but the review process was quite constructive.
HS: Do you remember if it went through just one round of review?
NK: No, it went through at least two, if not three, rounds of review. It was a pretty extensive process.
HS: Do you remember if there were any major differences between the original manuscript and the final published version?
NK: Let me actually pull up the decision letter and look, if you don’t mind?
NK: You know, from what I remember, and certainly looking through the responses that we wrote, most of the discussion and improvements were in terms of the interpretation. There weren’t major changes to the analysis, but I think that there was a lot of discussion about whether the kinds of test that we were doing were actually testing predictions from Neutral Theory, and how to relate them to some of the larger discussions that people were having about Neutral Theory and species coexistence. I think there was a lot of thinking about how to frame what we found there. That was where most of the effort was happening, in the responses.
HS: Did this paper attract a lot of attention in academia when it was published?
NK: At the time, I did not have a great basis of comparison because that was only my second paper. It certainly got more attention than my first paper, but I don’t think I had a great sense of how much is normal for a paper like this. I certainly heard back from colleagues about it as soon as it was out – both, people who were excited about it and people who had concerns or questions about it.
NK: Sure. This came from Jeff Lake who had been a PhD student working with Steve Hubbell. At the time of this exchange, he was a postdoc. with Annette Ostling in Michigan. I think that that comment began as a conversation Jeff and I had at ESA [Ecological Society of America] where I gave a poster on this before it was published. There are a number of different issues that are in there, and during the revision process that comment and exchange went a number of different places. But I think that the issue that is most helpful from that exchange is the question about intra-specific trait variation. We had sampled over 1100 species, which, at that time, was a huge amount of work. But we didn’t do enough sampling to be able to quantify intra-specific variation for most species. We did for a subset of species that were relatively more abundant, but we couldn’t do it for everything, because many of the species were represented by only one or two trees in the entire plot. And I think Lake and Ostling were interested particularly in the ways in which including intraspecific variation in traits might change the conclusions we drew from the analysis. At that time, one could imagine a number of different ways in which we could make things look more neutral or more niche-structured. I think, in the time since that comment and reply, we have seen a number of studies that have quantified intraspecific variation and have shown both of those outcomes can occur. But at that time, I think they were concerned that if we didn’t include ITV [Intraspecific Trait Variation] that we might be missing something that is quite important. I knew that it was a huge amount of work to do that ITV sampling. I thought it was quite interesting but it was not possible at that point of time for a PhD student. Also, my intuition was that it was going to be a second-order effect. The main effects that we were going to see, were going to come from the species average differences. But we didn’t have the data to tell either way. That part of the exchange, on ITV, was an important one and one that, I don’t know if it inspired, it certainly raised questions that were answered by a whole bunch of other papers in the years to follow.
HS: Did this paper have a big impact on your career and the research you did after your PhD?
NK: I think a lot of what I’m doing now is still linked into this question. I think most of my career so far – it’s been short – but to date, has really been on thinking about how can we use traits to better understand the dynamics of communities, community assembly and coexistence. I think that there are a number of limitations with the kind of analyses that we did in that study using a null model approach, and really understanding that has, in part, inspired a lot of what we are doing now, which is much more experimental work to understand how traits lead to coexistence or exclusion type outcomes. The effort here, of trying to use traits to understand community assembly, is something that I’ve done in many ways throughout my career, so far, and it is still kind of a theme, in trying to unpack what the drivers are of the kinds of patterns that we see in a place like Yasuní.
In terms of my career, I’m certain that it helped. I think it was a high impact paper, it was read by a number of people and I’m sure that it helped in some ways in efforts to find jobs afterwards.
HS: Its now eight years since the paper was published. Would you say that the main conclusions from the paper still hold true more-or-less?
NK: I think so. The paper is basically reporting a pattern. As we have continued to work at the site, we see the same types of patterns showing up, in other ways in other kinds of analyses. So, I think that the patterns in there are real. At the time, we tried to suggest what we thought might by driving the patterns, and tried to make as strong an inference as we could, but also acknowledged some of the weaknesses that are there. But I also think there is a lot more uncertainty now in the field about how to link some of these trait dispersion patterns to particular community assembly processes. That’s something that myself and a number of my collaborators have been writing about a lot for the last 4-5 years now.
HS: If you were to redo this study today would you do anything differently?
NK: No. I can understand now that it did fit at a certain moment in time. I think that Steve Hubbell’s work had, in some ways, set the expectation fairly low for the contribution of niche differences to community structure. Going to, at this point, one of the most diverse communities that we had information for, and exploring and finding patterns that were consistent with that, was exciting, in part, because of what Steve had written. I don’t know if that result had been, for instance, submitted in 1998, two years before Steve’s book was out, it would have been as exciting to the field. It really only worked in that format because of the dialogue with the ideas that Steve was putting out at that time.
At the same time, I think that the framework we used to interpret the patterns was far too simple. That is something that we really have come to understand in the years after this paper had come out. If I wrote this paper now, I would write a different discussion of the potential drivers of the pattern that we found. It would be more nuanced.
HS: This paper has been cited over 450 times. Do you know what it mostly gets cited for?
NK: My sense is that it is often cited just for the general idea that traits can be used to look at community structure in plant communities. Most of the times I have seen it cited, it is in that context. It has been a while since I have checked the recent citation patterns.
HS: You say, “Logistical concerns related to the extremely high diversity of the system limited us to these practical traits that are established proxies for plant strategy, although additional traits such as rooting depth, leaf secondary chemistry, and seedling relative growth rate would be of great interest, if and when data become available.” Has such data become available after this study?
NK: We continue to work at the site and I have a PhD student Ian McFadden who is just returning from the site right now, who has been sampling leaf osmotic potential, which is a predictor of turgor loss point in leaves, which is, in turn, an indicator of drought tolerance. I’ve had a cursory look at his data and I think we are seeing similar patterns – species are more drought tolerant in the ridge tops as opposed to the valley bottoms. So, we do continue to see the same kind of patterns in new traits. We also have a lot more data on wood densities. Joe Wright, who has been acknowledged in the paper, has been working with Renato to sample wood densities on individuals at the site. They had a nice paper a few years back, and we’ve used that data in a new Ecology Letters paper that just came out on Online Early a month or so ago. I don’t know if it’s reported in that paper, but I know from using that new extensive set of trait measurements – wood density and more data on seed size from Joe Wright and Nancy Garwood – that we see the same kinds of patterns even with more extensive trait data.
HS: What is it that allows you to measure these new traits today that wasn’t available then? Is it just the realization that these traits are important or the availability of some new technology that allows measurement?
NK: I would say it is mainly changes in the techniques available for easily sampling traits that we perhaps already had some sense would be important but are impractical to sample across a species-rich community. There is a PhD student, Megan Bartlett, here at UCLA [University of California, Los Angeles], where I’m right now, working with Dr. Lawren Sack who developed this approach to look at leaf osmotic potential around 2012. It’s a new trait that’s out there and that was not available at that time.
HS: You say “The evenly distributed trait patterns that we observed may be produced by direct competition (14) or by other density-dependent processes. For example, previous studies have shown that density dependent attack by specialist herbivores or pathogens may be pervasive in tropical forests (23) and that the probability of attack by natural enemies for plants has a strong phylogenetic component (24). In general, studies have shown that closely related plants have similar ecological strategies (19, 25), including qualitative defenses (26). Thus, our results may reflect non-random mortality inflicted by natural enemies.” Do we know more about this now, i.e. what the density-dependent driver is?
NK: Unfortunately, we don’t necessarily know that even now. Claire Fortunel, who is a postdoc in the lab here, has just published a paper in Ecology Letters, that’s Online Early, that is looking at the effect of crowding and neighbor identity on variation in tree growth in the forest plot. We do see that some species are impacted by the traits of neighbours. They grow more slowly when they are surrounded by species that are similar to them. So, we do have information for those same traits, that at least some of the species in the plot have reduced growth when they are around functionally similar species. But as for the pathogen end of it, that’s something we are very interested in, but I think it’s quite challenging to work on that at the community scale. But I’m always trying to see when there might be opportunities to expand in that direction.
HS: Towards the end of the paper you say “Using a functional trait approach, we have found evidence for niche-based processes known to have stabilizing effects on diversity (27) in one of the most species-rich tropical forest assemblages on the planet. Although the magnitude of these processes still needs to be quantified (28).” Has that happened?
NK: I think that we know that the effects, by my best guess, are small in the plot, although we haven’t really quantified them directly. But the patterns that we found are quite subtle. I don’t think that they are overwhelmingly strong. When we wrote that phrase that you are referring to, I think we were interested in the question of how many more species are able to persist in this community because of these effects that we might be seeing. Is it all of them? Is it 10% of the species? That was the question that we were trying to ask there. We have found patterns that suggest what these processes are going on, but we still don’t know how important they are to the overall dynamic of the forest. But there have been a number of advances, around the same time as this paper but especially afterwards, that have posed frameworks for thinking about quantifying how important niche differences are in structuring communities. That’s based in a large part on the work of Peter Chesson. Now there are approaches out there for doing that, but the data that you need to actually apply those approaches aren’t necessarily available from these observational long-term forest communities where many species are, by definition, very rare.
HS: You say you continue to work in this plot. Has it changed from the time you worked there for this study, in 2005-2007?
NK: It hasn’t changed that much. I think that this part of Yasuní National Park has had an evolving and changing human landscape nearby. At the time I was doing this work there was I believe, a lot more hunting and poaching going on, than there is now. I think that the park is a lot better regulated right now in terms of illegal activities of that sort. In the years that I’ve been there since my PhD I’ve seen more wildlife around than I’d seen at that time. But in terms of the trees themselves, I don’t think there has been that much change.
HS: Is the plot protected?
NK: The plot is in a National Park, but there are people who live inside the park. So that’s a complicated dynamic. And there is ongoing oil extraction in the park.
HS: Have you ever read the paper after it was published?
NK: I’ve read it on and off. Sometimes, when people ask me specific questions about it. At one point, I got a paper to review that had plagiarized a paragraph of it, so I went back to check that! Now that I have students and postdocs. in my lab that are continuing to work at the site, I go back to it when we frame new papers and projects that they are interested in.
HS: Do you notice any differences in the way you wrote then as a PhD student and now?
NK: I think my approach is similar- trying to present interesting results with as much transparency and nuance as we can, in the space that’s allowed. It is always a challenge, particularly in short-format journals.
HS: Would you count this paper as a favorite among the papers you have published?
NK: I think that I’m quite proud of the work that it took to pull this paper off. It was a tremendous amount of work in the field, and I feel like we were trying to present a message that was not particularly easy at the time, given some of the confusions in my own learning at the time about how these results might relate to these larger discussions that were going on. In that sense, I’m quite proud of the work that went into it. But as I mentioned before, I do think that it fits into a timeline in some ways, of a response to what Steve Hubbell had then been thinking about in the early 2000s. It fits into a kind of sequence, and I think that the field and my own work has certainly moved into more nuanced ways of thinking about these kinds of questions. For these reasons, I don’t think it’s perhaps as timeless a paper as some other ones that are my favorites from other authors.
HS: What would you say to a student who is about to read this paper today? What should he or she take away from this paper published eight years ago? Would you add any caveats?
NK: As I’ve begun to teach graduate students now, I notice something. When I began this project as a PhD student, everybody I knew, knew about Neutral Theory. It was front and centre to how most of the incoming students were thinking about communities, even if it wasn’t the focus of their work. It was certainly there as a thing to consider. I don’t necessarily see that when I teach graduate courses now. So, I think, in some ways, that this is a paper that’s a bit of that moment, especially in tropical ecology.
I think the other thing that is important for readers to keep in mind is the inferences that we drew from the patterns. I think that the framework we used was much simpler than we now understand it to be. I think there’s more uncertainty today in linking the kinds of patterns that we see to particular ecological processes. That’s something that has emerged in the field, especially in the last seven or eight years.