Tag Archives: Macroecology

Continuing the Conversation: The Role of Theory in Conservation, a follow-up with Justin Kitzes

justin_kitzes_crop

In my conversation with John Harte, he mentions work by his post-doc, Justin Kitzes, who is interested in how ecological theory can be used in conservation. Two weeks after the Gordon Conference, where I interviewed John, I found myself at ESA, where Justin was coincidentally hosting a symposium on “Advancing Ecological Theory for Conservation Biology. I snagged a quick 10-minute conversation with Justin after what turned out to be a fascinating symposium. Here is, in his own words, why theory is important for conservation:

I’ll start with a more general answer to your question, why theory is relevant for conservation. In some sense, without theory ecology is a collection of stories. We can go to individual systems and we can study them very deeply. We can understand a lot about how they work and what makes them unique, but what fundamentally makes ecology a science is that we believe that there is some deeper order and deeper pattern underneath all of these individual observations of species and systems. In conservation, we’re often in a situation where we don’t have all of that deep information. We need to take what we’ve learned somewhere else and apply it in a context where we don’t have a lot of data, where we need to make a decision rapidly, and in those situations it’s often the case that theory offers some of the best information that we’re going to get in practice in order to be able to make decisions. In a broad sense, I think that the role of theory in conservation is filling in the gaps. When we don’t have time or money to study everything to the extent that we want to, we use theory to do the best we can.

The particular type of theory that I work with and that John works with is macroecology, which is, generally speaking, the focus on statistical patterns. If we have evidence that there are some sort of general universal underlying patterns that govern how communities structure themselves, we can use that information to make decisions where we don’t have a lot of time or a lot of money. Probably the canonical example of this is the species-area relationship, which tells you as area grows and shrinks, how the number of species goes up and down. That pattern has been used in conservation for probably 40 years or more by now. It’s a good way at providing a first pass estimate of something like extinction risk when you really don’t have much else to go on.

The species area relationship is a really interesting case. Arrhenius in the 1920’s was probably the first one to put a number on it, and pretty early on it was thought that it was a power law. So on a log-log plot it comes out to a straight line and the slope of that line was about 0.25. There was some early work, for example Jared Diamond’s paper on land bridge islands, that showed empirical fits close to 0.25. Frank Preston and Robert May followed that up in the 60s and 70s with some great work showing that a particular form of the species abundance distribution, the canonical log-normal, would lead precisely to a power law of a slope of 0.25. So for a while there everyone was happy. Of course there was always some scatter, but maybe that was just noise. Rosensweig comes along in 1995 with his book and really hits home the fact that, no, it’s not just scatter, there are patterns in how systems deviate from that traditional model. Over time, people like my PI, John Harte, start to look and see, there’s not just scatter. There’s curvature on a log-log plot. It’s concave downward. It bends over. And that seems to be pretty consistent. So it’s not just that the slopes are bouncing around. There’s something systematic going on here. One of the most interesting outcomes of MaxEnt, which was really not realized until after the original theory came out, is that it makes a prediction for the slope of the species-area relationship. What we normally take to be constant at 0.25, is actually decreasing as the number of individuals per species increases. So basically, as plots get large, the slope of the relationship is predicted to decrease. It turns out that really works well so long as you’re not crossing major habitat boundaries. It does an amazing job of collapsing what looked like an enormous shotgun blast of scatter down to something that follows a predicted curve, pretty darn closely, for what we consider close in ecology.

You never know what the right answer is until the future happens. When we’re talking about global change we’re in the business of predicting the future. So you have a couple choices. You could do nothing, that’s always one choice, right? You could look at the data qualitatively as best you can and try to decide what to do, and that can be a very reasonable option. Or you can rely on what, in the best case scenario, we’ve been spending 100 years trying to do, which is figure out how ecosystems work and to try to apply some of that knowledge to try to figure out what might happen in future. The embodied knowledge of how ecosystems work has shown up in the body of theory that underlies ecology. It’s also important to recognize that theory is not just mathematical theory. A lot of theorists work with mathematical theory, but things like the intermediate disturbance hypothesis or trophic cascades, those can be qualitative theories and those can play a role in trying to make predictions.

I do think that theorists and conservation biologists don’t talk to each other enough. I think there’s a cultural divide there in addition to differences in the backgrounds of the different communities. But if there was one thing I could say, it is that I really believe that theory is underutilized in conservation. I think across the board, not just type of theory I do, we’re not taking advantage of that body of knowledge that we put into our equations when it comes time to actually make decisions on the ground. And that’s not to say that theory is going to answer our questions and we’re going to have a technocratic world where we can predict exactly what to do. But there’s information there that is missing from the applied conversations that we should try to do a better job of bringing in.

August 30, 2014