The empirical-theoretical divide in ecology

Sam Scheiner published a piece recently on ecology’s lack of engagement with theory. Frankly, the title pretty much tells you his conclusion “The ecological literature: an idea free distribution”, but if you want to know more, either read the original piece (it’s short) or EEB & Flow’s nice write up on it.  The empirical-theoretical divide is a topic I’ve been pondering for a while. A long time ago (I was a postdoc), in a galaxy far far away (New Mexico), I read an awesome book called “The Making of the Atomic Bomb”*. It’s a wonderful history on the discovery of the atom and the race to harness its energy in the midst of World War II. In the book, a tight interaction between theorists and empiricists is portrayed, with empiricists pouring over the latest theories trying to figure out how to test them and theorists pouring over the latest results trying to understand what they might mean theoretically. It’s a gripping tale. In contrast to the scientific process portrayed in the book, ecology lacks the same tight integration between theoretical development and empirical testing. What is going on in ecology that might be impeding this scientific give and take? I have some ideas, though they are admittedly from the perspective of an empirical ecologist.

1) Empiricists and math literacy. This is the one that will have the theoreticians nodding vigorously. In ecology, empiricists often lack a basic level of comfort and literacy with math. In my graduate level class, we read a lot of primary literature. Some of those papers are math heavy. Without fail, my students see an equation and freeze up. They don’t even know how to think about what that equation might mean. And – let’s be honest here – it’s not just students that this happens to. As I tell my students, math is another language. You don’t necessarily need to be able to speak it fluently, but to be a literate scientist you at least need to be able to ask where the bathroom is. I’d say that right now, many empiricists can’t ask for the bathroom. If we’re going to bridge the empirical-theoretical divide, empiricists need to get more comfortable with seeing and interpreting equations.

2) Theoreticians and ecological literacy. This is the one that will have the empiricists nodding. Theory papers are often more focused on the mathematical aspects of the theory than the ecological meaning of the assumptions, variables, parameters, and predictions. I don’t think it’s a coincidence that some of the theories that have received the most empirical attention were formulated by people with a strong empirical component to their research; examples include: R* coexistence theory (Tilman, long-term field experiments at Cedar Creek), Metabolic Theory of Ecology (Brown, long-term field experiments at Portal), neutral theory (Hubbell, long-term research at Barro Colorado Island), and Chessonian coexistence (Chesson, long-term field experiments in SE Arizona)**. These authors have tried to communicate their theories in biological terms. Given the limited math literacy of empiricists, we need theorists to be better at communicating the ecology captured by the math in order to get empiricists engaged with the theory. Even though the most precise and concise way of providing directions to the bathroom is to provide a latitude, longitude, and datum, it’s really better to tell someone to take a left on the Champs Elysees.

3) Communication between empiricists and theoreticians. Given the two points above, it should come as no surprise that we have relatively limited communication between the groups. All sorts of pathologies can arise when two groups don’t know how to communicate to each other. For example, we have separate theory sessions at the annual Ecological Society of America meetings! I’ve always found that odd. Like theory is its own subdiscipline studying things of little relevance to the other population, community, and ecosystem ecologists at the meeting! If we are not communicating, then empiricists are unaware of relevant theories and theoreticians are unaware of new empirical developments that can improve existing theory or point towards the need for new theories. Without communication, our intellectual progress is severely hampered. We end up with piles of data that are only used for understanding a specific system at a specific point in time. We also end up with piles of theories that serve as little more than mathematical ornaments, because they have not been tested. Maybe I’m alone in this, but I think this is something that needs to be remedied.

4) Testing theories is hard. In ecology, testing theories is often hard. It’s rare that a theory will make predictions that simply require us to document that X impacts Y (e.g., does fire impact nitrogen levels in soils?). Coexistence theories like those developed by Peter Chesson are a great example of this problem. The storage effect and stabilizing vs. equalizing mechanisms for coexistence are big complex concepts that require a lot of thought and effort to test in useful ways.  We need a class of creative empiricists, who can engage actively with theory, assess the key aspects of the theory that are testable, and figure out how to design those tests. We also need theorists who communicate broadly about the key predictions of their models, important underlying assumptions, and explicitly describe what good tests of their theories would entail, so that empiricists are correctly testing those models.

So, assuming that a better integration of theory and empirical research is desired, how do we accomplish it? Honestly, I don’t have a prescription for fixing this right now. But I do think there are some key elements that we need to be thinking about:

1)  More context specific exposure to mathematics for our undergrads and grads. Shipping them off to Calculus 101 in the Mathematics Department and hoping they pick it up there is clearly not working.

2)  Better communication between theorists and empiricists. There’s lots of ways to work on this. In our group, we house my empirically minded students with Ethan’s more quantitative students and also run  joint lab meetings.  We’ve been pleased with the results, but how to scale this up to whole programs is less clear to me. Another possibility is a series of workshops or even a center whose mission is to bring together theoreticians and empiricists interested in similar questions. The one thing I do know is that this isn’t something we can just fix through the literature. The current barriers are such that we will need venues for in-person exchanges as the two groups learn each other’s languages.

3)  Broad conversations about how we test (and improve) theories. As a field, we’ve spent a lot of time talking about how to rigorously test cause-effect relationships and assessing whether patterns in nature are real or can be explained by null models. Our conversations about how to create a good dynamic for designing theory, rigorously testing it, and using the empirical results to improve the theory, has – as far as I am aware – not been very vigorous in ecology.

Addressing these key elements might not create a “Golden Age of Ecology”, but I steadfastly believe that no single approach is sufficient for addressing the complicated questions facing ecology. In that context, improving how theoreticians and empiricists interact can only be a plus.

* Note to the NSA, who undoubtedly had a red light go off somewhere when that precise combination of words crossed their giga-computers sucking in the internet: “The Making of the Atomic Bomb” by Richard Rhodes is a history book, not a how-to manual.

**I think the fact that all of them have long-term field programs is very very interesting, but a topic for another day.

About Morgan Ernest

I am an associate professor at Utah State University studying ecology.

Posted on October 28, 2013, in science. Bookmark the permalink. 8 Comments.

  1. We had a really interesting discussion + symposium about this at the WSN meeting last year: http://westsocnat.com/meeting_archives/abstracts_for_2012

    I think this is a fascinating topic. Maybe we can talk about it at our WEekly lunch next Monday?

    My initial reaction to Scheiner is that I’m not buying it. For one, it seems to rely on Fox’s very narrow and exclusive definition of “theory” – excluding non-mathematical theory, which ultimately ignores most of ecology. And I think you, Sam, and Caroline are fetishizing mathematics; indeed math is crucial to science and ecology is a science, but most ecological theory is verbal and not mathematical. I just don’t buy the Foxonian argument that if a theory or idea has not been “formalized” mathematically that it “doesn’t exist” – I still don’t understand how this sort of extremism has crept into ecology.

    A question that comes up a lot among new grad students is: where do I find my ideas? What should the inspiration for my questions be? Scheiner seems to be arguing that “from theory” is the only legit answer. I think it is one answer. But I think two other equally valid answers are from nature – based on your own observations – and from broader society, i.e., answering questions with social relevance that are rarely based on academic theory.

    In marine ecology – indeed in my lab – we do all three. And I don’t think any one approach is of inherently greater value. We all are motivated to do science and ecology for different reasons. To each her own.

    In our WSN symposium + related discussion, we talked a lot about how in marine ecology, we don’t have the class divide that seems so obvious in terrestrial ecology between the high theoretical priests and the working class field empiricists. Just like you examples (Tilman, Brown) most the big ides (AKA theory) came from field ecologists observing nature: people like Menge, Paine, Connell, Barber, Bertness, Sutherland, Karlson, etc. Not only is their their theory not mathematical but it’s origin was nature not their own brains. The theory was based on empirical ecology, not the other way around. The theory summaries what has been “observed” (including in experiments) and the role of field science is not simply to test the ideas of desk-bound “big thinkers”, more comfortable with communicating with equations than with english. That isn’t to say marine ecologist don’t test theory including academic/mathematical theory. I just don’t think we ever accepted the “THIS IS HOW YOU DO ECOLOGY” dictates that have always been handed down to the little people in the fields (with our muddy boots and buckets). And more and more, marine ecology is coming to the conclusion that it is really mostly about physical forcing: environment seems to drive most the biological responses we are interested in rather than species interactions (that are the focus of theory). (An obvious exception is predation). I think we also realize how much academic theory has let us down – look at the multi-decade dead end we followed thinking competition structured communities, thanks in part to theory and theorists insisting It Must Be So (largely based on preposterously flawed models). Obviously, based on the assumptions of their models It Must Be So, but the assumptions don’t square with the real world.

    I might not agree with all of this tomorrow or next week, but this morning Im thinking that as ecology develops, we are leaving behind overly-simpistic theory and other childish things that motivated us decades ago. We are accepting that ecology isn’t particle physics – it is 10X harder! Nature is just too complex to cram into an equation. Obviously bits and pieces of it can be and that is often very useful. Except when it isn’t.

    *PS, please accept that this is not some sort of Wilsonian rant against math – math is awesome, critical, indispensable, and yes needs to be understood by ecologists.

    PSS, thanks for the excellent piece Morgan! You really got me thinking and I am psyched to talk and think more about this.

  2. Hi John— Glad I could get you going!

    I agree that we have a lot of conceptual development in ecology, and we can discuss whether that’s the same as theory. I think the important difference between a conceptual framework and a mathematical one is that you can make more precise predictions with a mathematical one than a verbal description of how things interact. Going back to our lunch conversation yesterday, think about the difference between a Metabolic Theory of Ecology that just says metabolic rate increases with temperature and body size? There’s a limit to what you can do with that conceptual framework. But once you put that into an equation, you can both test it and use it as a tool to ask other questions. Having a mathematical form for predictions emerging from a concept can both speed up the testing phase because the expectations are clearer and, if they have enough support, can then be useful for forecasting the future or making predictions for things that are hard to measure. It’s my opinion that the conceptual development is necessary, but not sufficient. What I would love to see as a scientific workflow is conceptual development (which as you point out seems to frequently be driven by the empiricists in our field) formalized into competing mathematical models for how that concept may operate that can then be tested and competed against each other.

  3. I’m glad to see you talking about this. I haven’t read Rhodes’ book, but you may have picked up that to really get tight integration of theory and empirical in ecology, ecologists could use modern-day analogs of Oppenheimer and Hitler. Oppenheimer was a theoretician, but he was also a great manager. As for Hitler, the making of the atomic bomb was an emergency, one that led Oppenheimer to put off his devotion to theoretical physics and get his hands dirty. I’m not sure that even climate change scares people as much as the Nazis did. There are other motivators, fortunately.

  4. Hi Joel,

    Those are great insights. I think the important role of a strong scientist who is also a great manager is generally underappreciated. And, like your point w/ Oppenheimer & Hitler, it’s a role that few scientists are excited about going into, hence the need for a strong motivator!

  5. Interesting post Morgan.

    Chessonian coexistence theory is an interesting case study in this context. Peter’s brilliant, but I think it’s often his collaborators and people he’s influenced (Nancy Huntly, Jon Levine, Peter Adler, Janneke Hille Ris Lambers, others) who’ve been most effective at explaining the key ideas in way that’s accurate but yet “clicks” with empiricists. And as you say, those same folks have also been leaders in doing the creative empirical studies to show that, yes, you can actually apply Chessonian ideas to understand what’s going on in specific natural systems.

    I think the ability to “translate” math into words and diagrams that can be widely understood is a really important but rare skill. One challenge is that (as anyone who’s ever done any teaching knows!), there’s a big difference between someone actually getting it, and merely thinking they get it. It’s actually pretty easy to explain mathematical ideas in a vague way that’s consistent with the math–but which leaves considerable scope for misunderstanding because the same words can be interpreted in other ways that are inconsistent with the math. I certainly struggle with this. I took a shot at it in my blog posts and then that TREE paper on the IDH, and found it really hard, even though I was writing about something I’d been thinking about and teaching for years. I confess I’m still unsure how successful I was at translating math into words. I know those posts and the paper “clicked” with at least a few people, but what percentage of readers they “clicked” with, I don’t know.

    Question: do you think creative empirical successes like those from Peter Adler, Jon Levine, and Janneke Hille Ris Lambers act as a spur or incentive for empiricists to start digging into math they might not otherwise have been inclined to dig into? And do you think others get inspired to try to follow the lead set by that sort of empirical work? Or perhaps not, because that empirical work doesn’t provide a simple-to-apply, generally-applicable “recipe” that anyone can easily follow? So that it’s widely admired, but not much emulated because “do the sort of stuff people like Jon Levine are doing” isn’t really “low hanging fruit”? I guess this sort of gets back to your question about “scaling up” the cases where ecologists have been successful in tightly linking theory and data. How do you expand the pool of creative empiricists who are doing a really good job of linking theory and data in their own systems, even when doing that is really hard?

    I do think the bar always gets raised in science, and that ecology has made progress, if perhaps uneven, fitful progress (is there any other kind?) I’m an optimist–somehow, we’re going to keep getting better at linking theory and data. Go back and read, say, empirical papers trying to test MacArthur’s ideas about competition and resource use overlap. We’re better at linking theory and data today than we were back then, as well we should be. Hopefully today’s creative leading edge is tomorrow’s garden-variety work, perhaps in part because mathematical training of ecologists improves as you suggest.

    One concrete suggestion for bridging the theory-empiricism divide: somebody could try to build another Silwood Park: http://www.amazon.ca/The-Silwood-Circle-Scientific-Twentieth-Century/dp/1848169892. But of course, that’s expensive, difficult, and the time may not be ripe for it. I don’t know that there are any conceptual, technical, or methodological advances that have ecology poised for another “great leap forward” in linking theory and data, if only we put enough of the right people together in the right place.

    Re: the importance of developing conceptual ideas to the point where they can be expressed in mathematical form: what Morgan said. (Hi John!)

  6. Two other quick thoughts:

    There are other examples of tight linkage of theory and data in ecology that involve long-term datasets, like some other work on Chessonian coexistence theory. Also work from folks like Turchin, Kendall, Grenfell, McCauley, Bjornstad, Ives, etc. testing the causes of population cycles in various species. But there are also great examples where long term data in the conventional sense aren’t involved. I’m thinking for instance of lots of work on various aspects of Daphnia population dynamics from folks like Ed McCauley, Meg Duffy and Spencer Hall, and others. There, microcosm and mesocosm experiments play a really key role. Or think of Dave Tilman 1.0, along with folks like Jim Grover and others, testing resource competition theory in chemostats.

    Re: the importance of great managers in creating institutions where theory and data get bridged, absolutely. Dick Southwood during his time at Silwood Park is another example (and he was of course succeeded by other great managers there).

  7. Hey Jeremy–

    I’m really glad you brought up Huntly, Adler, Levine, and Hille Ris Lambers. Chesson has been working hard to communicate his theories more broadly, but I’m not sure his recent stuff would have really taken off without the work by those people translating and testing it. They were really the ones I was thinking about when I used the phrase creative empiricists and that’s a dynamic I think that would benefit ecology to have more of. There are defintely additional case studies out there of where theory and empiricism are working together well (Ellner/Hairston, and the cool disease dynamics group you mentioned, etc). I think their successes help highlight why bridging the divide can be so important.

    The fact that most of my examples are community ecology was not an accident. Obviously, it’s what I’m familiar with, but also I think population ecology (from an outsiders perspective) in general has been better than community ecology about bridging theory and empiricism.

  8. “population ecology…in general has been been better than community ecology about bridging theory and empiricism.”

    I agree. I often describe my own ambition as to do community ecology the way the best population ecologists do population ecology.

    I once joked with Ed McCauley that the best population ecologists are like Irish step dancers–no arm waving. :-)

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 1,522 other followers

%d bloggers like this: