I interviewed awesome ecologists at the 2011 Ecological Society of America meeting in exchange for reader donations, which paid for my conference attendance. This is one in a series of posts about those interviews.

Scientists are often portrayed in popular culture as utterly focused on their work, absolutely excluding personal relationships and all other interests and passions from their lives. But that’s not at all true!

My next interviewee, Professor Nick Gotelli from the University of Vermont, is a good example of a successful scientist who has the time (and inclination) to pursue things-that-are-not-science: in addition to being quite a well known ecologist, he’s also a musician. I’ve seen him play at ESA in the evenings at the informal Musicians Central events, and he organizes a popular session at ESA meetings that’s sort of an open mic night for musician-ecologists. If you want to know more about his music, you should check out his music site. If you want to know more about his science, read on!

Pitcher plants and community assembly

A lot of Dr. Gotelli’s work is in a very particular (and clever) study system that I mentioned a few weeks ago: inside of pitcher plants’ pitchers. When I asked him how he got started working with pitcher plants he told me it was

Completely a random walk! I got into them because I was contacted by Aaron Ellison when he was a faculty member at Mount Holyoke college. I had done a lot of statistical work before that on community assembly and Aaron and I were both in the same position of having worked on other systems in the past for a long time and having moved to new universities.

So they were both looking for a new experimental system where they could try to answer questions about community assembly: (why species are are found with certain other species (or not!)). The systems they’d worked in before (Ellison: mangroves, Gotelli: doodlebugs!) weren’t too common in New England.

One of the things they looked at in pitcher plants were the food webs. See, inside the pitcher plant’s pitcher is a bunch of water. Some bugs fall in and drown, but there’s a whole little food web of little bugs living in there, too. But why do we care about studying food webs? After all, we all learned about food webs as kids in elementary school. Remember the pyramid with the plants at the bottom and the big predator at the top? But it turns out there are a lot of things we still need to get straight about food webs.

The first thing is they’re not a pyramid!! That’s the first thing we have to get over. There are huge omnivore and detritivore links so most food webs don’t really look at all like the drawings that we show.

The challenge with food webs is that the research is sort of split in two directions.

One path, the one that’s gained ascendancy so far is the network analysis path. We elucidate all the different linkages that we can possible find in a web and we run those through network motifs and other metrics to try and figure out structure. And that’s very much in vogue right now especially because many social networks show the same sorts of patterns that a lot of our food webs do. And they’re very visually appealing.

This looks very cool and impressive in food webs with lots and lots of species, but Gotelli has good reasons for working in the very simple food webs in pitcher plants.

I’m skeptical though that [the big network food web models] are actually going to tell us much about structure and function and that’s because it’s very hard for us to understand the function of truly complex systems. So the other direction that food web analysis has gone – that to me is much more satisfying – is to deliberately simplify and collapse all that complexity; to deliberately distort nature and pigeonhole it into a couple of key little trophic groups of predators or consumers or parasitoids and work out the critical linkages there.

Experimentally that’s given us our great advances in, for example, trophic cascades, and top-down and bottom-up effects. We learn about all those things by making those deliberate simplifications.

But it’s certainly not time to throw out what we have learned from more complex network analysis!

In some ways we maybe need kind of an intermediate path where we can add a little bit more complexity, find a few more linkages that we’re going to want to study. Adding and removing those links and asking what happens to systems is really important because, of course, (like it or not) that’s what’s happening to all our natural systems: at the top end we’re stripping out and (either deliberately or accidentally) removing all the top predators and and the top consumers from our webs. At the bottom end we are overly enriching the productive forces at the base. So we need to understand what those kinds of changes are going to do to communities.

The beautiful simplicity of the pitcher plant community was what initially attracted Ellison and Gotelli to it, but it had some surprises.

We went off in completely unexpected directions! When we started looking at the pitcher plant and the basics of their nitrogen budget, it became clear that they were getting far more nitrogen than just from what was was coming in from their prey. So this led us into some long term modeling studies of the consequences of air pollution and atmospheric deposition of nitrogen and potential extinction risk.

And most recently it’s gotten us to thinking about using the pitcher plant as sort of a model system for understanding eutrophication. So, we can experimentally push this system by loading it up with excess prey and we can drive it into a eutrophic state, just like other aquatic systems.

It’s possible we can take what we learn about eutrophication from the tiny little system in the pitcher and apply it to a lake! (I’ve written before about the problems extra nitrogen causes.)

So what exactly are Nick Gotelli and his collaborators trying to learn?

Our latest work is to start to look for sort of proteomic indicators of a system that might be heading for a tipping point, or a change. So we want to know if we can use proteomic profiles to see if there are early bioindicators.

We’re hoping to do some kind of real world intervention experiments, to say if the protein indicator gets to this level and we stop the nutrient treatment can we prevent it from going all the way to eutrophication?

If they succeed, we could have a new way to monitor lakes and streams. We’d just go out, take a sample of the water and the critters in it, and see how much of certain proteins they’re making. It would be kind of like having a eutrophication thermometer – certain proteins are normal (98.6 degrees), others are worrisome (100 degrees), and certain levels of other proteins mean you’re never going back to normal and might die (105 degrees).

I wanted to know about some of the challenges Gotelli faces in answering his questions. Some of what makes his questions hard to answer are nearly universal in ecology:

• We’re small and the world isn’t

First the fact that we’re so limited in the temporal and spatial scales at which we can collect data. That’s probably the main limitation.

• Money

If we could increase our grant budgets in size by an order of magnitude we could start collecting all sorts of different kinds of data that we don’t currently collect. And that would make it easier to estimate parameters.

• The world is complicated

I think the other thing is that I think ecological experiments are really hard to do well because it’s really difficult to manipulate one and only one variable at a time and to come up with appropriate controls and sufficient replication to do a really good experiment.

He also pointed out a problem that I hadn’t thought about before – not only are ecological experiments hard to do, but we’re probably doing the wrong ones!

It’s also hard to look beyond the end of the experiment. We’re usually interested in what the qualitative result of an experiment is (‘do the treatments differ from one another?’). But the better experiments are used to actually estimate some sort of a model parameter. Sometimes those are going to require a very different sort of experimental design.

The standard ANOVA designs are oftentimes not very good for estimating model parameters. The reason is that for ANOVAs we typically chop up a variable into little discrete chunks and we set those as discrete treatment levels. That’s fine for estimating those particular values but what we really usually want is a continuous function. So a lot of times we should really be using a regression design rather than an ANOVA design in our experiments.

Climate change and conservation

Like most ecologists, Gotelli is very concerned about climate change and sees a lot of big ecological questions we need to answer:

So what’s going to happen to communities? What’s going to happen to ecosystem fluxes? What’s going to happen to the human population? It’s not too hard to imagine as time goes on we’re going to see more severe coastal storms. We’re going to start seeing more movement than we’ve seen in the past away from uninhabitable coastal areas.

One of the big questions we don’t know is how much extinction are we going to see versus how much evolutionary response and phenotypic plasticity that will actually allow things to stay in place. We don’t know that yet.

Despite all these big unknowns, we need to make decisions now. I asked Dr. Gotelli what our conservation priorities should be.

My immediate impulse is to say that the most important conservation decision we need to deal with is controlling human population size. If we don’t get that under control then all these other elements are short term and ultimately are going to fail.

But calling for controlling the human population is more than a little contentious and ecologists aren’t politicians.

I’m not even sure we need to say we need to control it, but what we do need to say is what are the consequences if we don’t? I’m skeptical about [ecologists’] ability to make direct changes in how things happen. I still think that science’s most important position is to provide knowledge and information. Society as a whole is going to have to act on that. We aren’t going to be the ones who will be able to implement these things, but we can certainly use our ecological knowledge to make some useful predictions about the changes we’ll see and then society is going to have to decide what to do with that information, even though it’s oftentimes very unpopular.

Dr. Gotelli spends a lot of time thinking about pitcher plants and their risk of extinction, but why should the average person care about these plants?

The first thing is: I don’t do my work in order to persuade people to conserve pitcher plants. As far as my work goes, I am ultimately a hedonist. The real reason I do my work is that I love it, and it’s intellectually satisfying to me. Everything else follows secondarily.

Now I would hope that my enthusiasm and love for my subjects will translate into other people recognizing their importance. Certainly in my grant proposals I will emphasize the importance of pitcher plants as a great model system. We do have some very good reasons for that because it’s so experimentally tractable. But I don’t hold great expectations I’m necessarily going to persuade people just on the basis of the writing that it’s an important thing to conserve. But I do think that the more we learn about the natural world, the more we’re going to want to conserve it. And so just the additional knowledge itself I consider worthwhile to have.

I loved this answer because I think it really captures the position of many ecologists. We didn’t start doing this to save the world, but to understand it. But now we’re invested in the systems we study and can’t just stand idly by why they’re destroyed.

Favorite papers

I asked Dr. Gotelli about his favorite papers – one he’d worked on and another he hadn’t. Like most of my interviewees, he chose a paper of his own that isn’t particularly well known.

I guess one of my favorites is a paper from 1992 on sexual selection in Acrothoracican barnacles, or burrowing barnacles. These are Darwin’s barnacles and they have dwarf parasitic males. The male is greatly reduced in body size – it’s really nothing more than a penis that attaches to the female’s body. We were able to show in this paper that these males are probably actually competing with one another for where they settle on the female’s body for access to her for fertilization.

It was a paper done with a wonderful barnacle taxonomist while I was still in graduate school, and it’s just the most interesting natural history system that i’ve ever worked on. It’s also one of my most obscure papers – it’s very rarely cited, but it’s still kind of one of my favorites.

He couldn’t choose just one paper that he hadn’t worked on, instead referring me to a series of papers by Brad Efron.

He’s a statistician at Stanford, and he writes about Bayesian statistics. He’s done a lot of the pioneering work on statistical methods for microarrays and what to do in these situations where you’re doing like 5000 t-tests. (How do you possibly decide which of those significant values you’re going to use?) Those methods are actually applicable to many ecological problems – like if you’re interested in finding out which species pairs in the community show positive or negative associations; there are thousands of species pairs even in a moderately sized community! All those Efron papers are incredibly stimulating, and he’s an amazing writer. So I think those are really wonderfual and great reading for ecologists.

I was a little surprised by Dr. Gotelli’s choice, because it wasn’t even an ecology paper, but he recommends reading widely to get more inspiration for your own work.

How to do ecology

One of the tools Dr. Gotelli relies on when he designs experiments and analyzes his data are null models. He’s spent a fair amount of time and effort developing good null models. Why would he have worked so hard to come up with a model that shows that nothing is happening?

Aha! Because that’s the fundamental way we find things out in science!

Oh really? What exactly does he mean by that?

We always have to ask ‘patterns compared to what?’ In the absence of an experimental data set (and most data sets in ecology are not based on experiments), we have to have some base level against which we compare patterns. So the whole rationale behind using null models is to be explicit about what that baseline is. [A null model] forces a much more thorough discussion of exactly what our expectations are for a system, and it really focuses the question on exactly what kind of evidence is sufficient to claim a mechanism is going on.

Basically, if we want to say that competition or climate change or some other process is affecting a system, we have to know what the system would look like if that process isn’t affecting the system. It’s a little bit like having a control in an experiment. It’s important enough that Gotelli thought it was worthwhile to write a program other scientists could use for null model analysis in community ecology. He had this to say about ecologists wearing a programming hat:

I do think it’s critical that students in ecology to learn how to program on your own. If you don’t want to be stuck using the standard statistical software package (or even my software package!) you should be able to program your own software.

The downside is that programming is like learning a foreign language. It’s going to require a time investment. There’s a steep learning curve, and it’s sometimes not obvious to students that it’s worth the investment of time initially. So you have to be willing to wait for a payoff in the longer run in your research, but I really think it’s a huge one.

That’s a big piece of advice for new students – really take the time to learn a programming language.

He thinks we should all be learning R, which made me feel good about all that time I spent struggling with it during my undergrad. Now that I can swear in R instead of at R, I appreciate it much more.

One of the pillars of science as we know it is anonymous peer review. But reviewing is time consuming work and I don’t see any clear rewards for it. But Dr. Gotelli disagrees.

First off, there must be some reward because your editorial boards are filled with really good, productive scientists who are very selfish about their time. So you have to ask why are all these really good scientists on these editorial boards if there’s not a payoff in it? There’s actually a huge payoff! A couple of things. First, as you sit on the editorial boards of these journals, you’re actually getting to view the cutting edge of the science. You’re seeing it as much as a year or two in advance before it actually comes out. You’re also seeing the things that are failing and the things that are sources of controversy. And again those are not going to show up in the literature necessarily right away.

I actually think that as journals move more to non-print and open access, science will become even more interactive (commenting on papers, blogging the controversies) and publishing will happen much, much faster. So I can see those particular benefits disappearing, or diminishing. But there are other benefits that aren’t going anywhere that are especially important for young scientists.

Just for me myself, I learn so much more about how to effectively write and present my own science by getting examples of what others are doing.

What’s the most important thing for being successful in science? It’s being able to write effectively. You have to not only write and get published, you have to publish work that is going to be read and cited and appreciated by others. And the only way to do that is to immerse yourself in this world of writing. And you can do that by working as an editor or reviewer.

He went on to hammer home how important writing is for scientists:

Writing is a skill, but it’s also a craft. It’s like programming. It’s this essential thing that you have to learn in your sciences. I’ve seen many examples of scientists who are incredible modelers, incredible programmers, incredible experimentalists, but they’re often not successful in the field because they’ve not learned how to write or they’ve not been willing to put the time and energy into their writing that they’ve put into the other parts of their science.

And so the writing is a huge part of it and you need to expect to devote hours and hours and multiple drafts, multiple rewrites, intensive work with your collaborators to really get those manuscripts in a perfect way. I mean, some of our papers, you know, the email exchanges will go back and forth over a single paragraph or a single sentence. It’s got to be that good to really get your message out.

He’s right, but sometimes I just want to stomp my feet and complain about how unfair it is that scientists have to be writers, fundraisers, programmers, muddy field workers, etc. And we need to be good at it all. And then I remember that the variety and challenge is a big part of why I’m here.