Friday, January 14, 2011

How to infer importance

Quantifying the relative importance of ecological factors in determining the distribution and abundance of species is one of the most critical endeavors to understand the assembly of communities and ultimately the evolution of different species. Yet, how do we do this?

One approach is experiments. Another is gradient analyses. Each has its pluses and minuses. Yet, is there a way to infer importance without manipulations or gradients. If I go to a single place and determine who is rare and who is abundant, does that offer insight?

Seems like it should.

Let's take drought. Drought has long thought to structure grasslands. We know that it can from the Great Drought. But how important is drought in determining the assembly of extant communities?

One way is to compare a trait that should confer advantage during drought to the relative abundance of species currently. But what are the interpretations of the relationships. Let's say there is a positive relationship between the drought tolerance trait and abundance. Probably good evidence that drought structures the community since those species that do not have the trait are less abundant.

But what if there is no relationship? Does this mean drought is not important since it doesn't confer any net advantage? Or is it equally important as some other factor, since one might expect the drought tolerance trait to penalize plants in wet times? And what if physiologically drought-tolerant species are less abundant than intolerant ones? Is drought not important? Or does drought serve as a disturbance and structure the community in a different way.

In all, using an inferential approach to interpret patterns depends strongly on our mental model of how different factors structure communities. As much as being able to measure the right traits, linking these mental models to testable hypotheses is one of the most limiting steps.

In short, we need to spend more time clarifying our concepts before testing the importance of factors.

No comments:

Post a Comment