Insightful Analysis: Hodgson Calls Out Naeem and His Box-Ecosystems

This week’s insightful analysis: from 1997, Michael Huston critiquing Naeem et al’s ecosystems-in-boxes experiments (insightfully analyzed here). Huston claims that the Ecotron experimenters dropped the ball by building in an inevitable bias towards a wide range of plant sizing in their high-diversity boxes, and further claims that debunks the diversity-ecosystem function link. He brings some big-gun databases of plant survey data into the act, and proposes the way forward. I think he’s got some good fundamental points, but then goes too far. Mostly I think he’s got a pretty trivial idea of what “function” means, which makes his claims that Naeem et al‘s results are trivial sound a bit whiny. But I don’t say that in my analysis, because I’m nice. Is there anything more heart-pumping than scientific debate?

The paper:

Hidden treatments in ecological experiments: re-evaluating the ecosystem function of biodiversity. Huston MA. 1997 Oecologia 110:449-460. (Don’t issue and volume citations look weird on the internet?)

Text of the analysis after the fold.

The central assertion of the critique is that, although the original Naeem et al Ecotron experiment claimed that “The only experimentally manipulated factor was the number of plant and animal species”, in fact “in the experiment…species richness varied in parallel with variation in diversity, making it impossible to isolate the effect of either”.

The authors of the critique frame this objection as a repudiation of the Naeem et al experiment. I agree with them that the Ecotron experiment was not designed in such a way as to resolve what mechanism linked the manipulated species richness with the measured productivity (and by inference, with ecosystem function). This is a strong objection, and a good one. However, the authors of the critique seem to imply that for a diversity-function experiment to be valid, it would have to vary species richness while holding constant all correlated physical attributes in order to directly measure the effect of diversity on function. If this is what the authors are suggesting, it is also a fallacy. Species richness is just a number. There cannot be a “direct” effect of a number on ecosystem function. In order for diversity to have some such effect, there must be one or more physical mechanisms correlated with that diversity. Requiring that all such mechanisms be controlled (for instance, by “including more species in the low-diversity treatment” to reduce the range of functional type between treatments), will inevitably eliminate effects on ecosystem function. Would that prove that diversity does or doesn’t have an effect on function?

I would reframe the author’s criticism as being that the Naeem paper failed to allow us to distinguish the particular nature of the mechanisms linking diversity to function (function in this case measured purely as productivity). After that the authors of the critique, while still framing their statements as a repudiation, rather suggest a possible mechanism linking diversity to productivity. They use empirical data on plant body structure to suggest that the higher diversity treatments included more physically dominant species. This doesn’t seem to repudiate that diversity and function were linked in the Ecotron experiment, just that they were linked in a trivial way, which was a “foreseeable consequence of the experimental design”. Perhaps this boring, superficial mechanism is the actual mechanism linking diversity and function in the real world. In order for the critique to be a true invalidation, the authors would have to prove this not to be so. As it stands, it is rather a useful addition to the original experiment.

The authors also question where on the axis of diversity of the model graphs of diversity vs. function (figures 1 and 3 in the original and critique papers) the Ecotron samples were located. This is an important question. I found their conclusions to be unconvincing. The link between the Ecotron assemblages and the data they used to calibrate the axis was weak (they used the “closest approach to a community which often contains many of them” and modified the range upwards because “a substantial fraction of the potential arable weed flora is already extinct”). They also ignore that the Ecotron experiments took place in isolated growth chambers, while their analog ecosystems exist in the much more complicated real world. Consequently I feel that any measure of the “maximum possible” species richness in not transferable between the two settings.

leave a comment