Wednesday, August 22, 2012

Justifying assumptions: tests of seed size/mass tradeoffs

When ecologists develop theory and models, we generally need to make assumptions. The nicest definition of an assumption is that they are the framework we use to capture our beliefs about a system. All future analyses will treat these assumptions as true, and so ultimately the validity of a model is tied to the validity of its assumptions. As Joseph Connell said: “Ecological theory does not establish or show anything about nature. It simply lays out the consequences of certain assumptions. Only a study of nature itself can tell us whether these assumptions and consequences are true.” Often times the most interesting advances in ecology come when we questions popular assumptions, such as that species are ecologically different, that interspecific differences are more important than intraspecific differences, or that ecological interactions occur much more rapidly then evolutionary changes. 

Assumptions in models and theory can often serve as a sort of shorthand for ideas that there is some general evidence for, but for which comprehensive data may be lacking. Community ecology is full of assumptions about functional tradeoffs that mediate coexistence between species. Various assumptions about plant species coexistence include that species experience tradeoffs between competition and colonization, growth versus reproduction, or seed size versus seed number. A simplistic explanation for such tradeoffs is that you can’t do everything well: a strong competitor can’t be a good colonizer too, which creates opportunities for strong colonizers but poor competitors, etc.

Tests of these functional tradeoffs are lacking, or lag behind the theory that relies on them. For example, the idea that there should be a tradeoff between seed size and seed number has long been proposed to explain why plants have highly variable seed sizes. Plants with small seeds should produce more offspring, and these seeds should be more successful at reaching empty sites. Large seeded species should be more competitive in the seedling stage or more tolerant of difficult conditions, and so have higher survival. Theoretical models that rely on such a tradeoff suggest that many species could co-exist and that the resulting community would exhibit a wide variety of seed sizes. 

But though many studies and theories depend on this assumed tradeoff, a comprehensive experimental test was lacking. Ben-Hur et al. have finally provided such an experiment, testing the basic prediction that a negative correlation between seed size and seed number should increase species richness. They also tested whether small-seeded species were more likely to remain in the community when this tradeoff existed, increasing the amount of among-species variation in seed size. To do so, the authors created 3 ‘community treatments’ of 15 plant species. The abundance of each species in the starting seed mix was manipulated to create either (1) positive correlation between seed mass and seed number; (2) negative correlation between seed mass and seed number or (3) random allocation of the 15 species regardless of seed size.

From Ben-Hur et al. 2012. Ecology Letters. a) Final number of species in the community, when the correlation between seed size and seed number is negative, random, or positive. b) Seed mass distribution in community under positive correlation between seed size and seed number (left), and negative correlation (right). 

Ben-Hur et al.’s results strongly suggest that a seed size/seed mass functional tradeoff can increase species richness (figure, a). Further, when there is such a tradeoff, the variation in seed size represented in a community increases, again in agreement with predictions (figure, b). The results are particularly convincing because the authors used experimental manipulation of the strength of the correlation (i.e. from negative to positive) to test its importance. The authors suggest that the tradeoff they simulated did not involve competitive differences (i.e. was not a competition-colonisation tradeoff), and more likely reflects a trade-off in establishment probability and colonisation (Dalling and Hubbell 2002; Muller-Landau 2010).

Of course, these results represent relatively short-term coexistence, and community richness may have changed had the experiment been allowed to continue for longer. But as a starting point, this suggests that theories that rely on functional tradeoffs in seed characteristics to explain coexistence are capturing a mechanism that has some experimental support. 


Anonymous said...

Great post...interesting paper.

Sorry to get a little pet peevey here.

"Tests of these functional tradeoffs are lacking, or lag behind the theory that relies on them."

Hmmm. I admit I'm no expert on functional tradeoffs, but I'm always skeptical of statements about how data collection lags behind theory. It seems to me that statements like this are very popular amongst ecologists, but to me it's not clear that either one lags behind the other.

For example, I myself am more interested in theory and observation, than experiment. Not saying that experiments are worthless, they're just not my bag. But I often feel like everyone else is doing experiments and that theory and observation are 'lagging behind'! Could there be some kind of perceptual bias that we tend to feel that what we are interested in is always 'lagging behind' what we're not interested in, because people naturally feel that what they're interested in should be worked on more?

In general the 'approach X is lagging behind approach Y' statements assume that the optimal mix of X and Y is obvious...but I don't think it usually is obvious.

Oh other thing...Connell's wrong...ecological theory *does* teach us about nature. ;)

Caroline Tucker said...

Hey Steve - first, of course Connell is wrong - any pat statement about the weaknesses of theory is limited. But there's a kernel of truth in most of these stereotypical statements.

Is there some obvious mix of theory and experimentation, and does our perspective of what that mix is depend on our own biases? I don't know. But I'll give an anecdote that illustrates my thoughts - the concept of competition/colonization tradeoffs was really popular in theory and models developed around 2000-2010. All sorts of coexistence models depend on this sort of tradeoff, but there was actually relatively little evidence that such a tradeoff was frequent and important in natural systems. It was only after a time lag that experimental evidence began to roll in which suggested that this kind of tradeoff wasn't very commonly observed. So suddenly the earlier theory, though technically correct, was less applicable to natural systems because its assumptions were not very general.

Jeremy Fox said...

There are cases where theory runs ahead and experiments catch up, and vice-versa. Early empirical work on biodiversity and ecosystem function--e.g., Naeem et al. 1994, the original Ecotron paper--was based on "hypotheses" that were just arm-waving, verbal metaphors, and guesses. Theory came later.

Unknown said...

Great post.

Thank you.

Anonymous said...

Makes sense. I think I was just giving an overly general involuntary irrational response. My apologies. I've got these buttons and one of them is 'experiment is lagging behind theory'.

Unknown said...

Thanks for reviewing the article, you wrote a great post.
One little typo mistake, in "(1) positive correlation between seed mass and seed" the word 'number' is missing.

Caroline Tucker said...

Sorry about that Eyal, I've fixed it.