Zombie ideas in ecology: the unimodal diversity-productivity relationship

Various people have been begging me to do a post on the humped diversity-productivity relationship as a zombie idea–a widely-believed idea that should be dead, but isn’t. And since my motto in blogging is “Give the people what they want!”*, here it is!

Grime (1973 Nature 242:344) may have been the first to note unimodal relationships between local community diversity and local resource productivity (or some surrogate such as total biomass). But it was the chapters by Rosenzweig & Abramsky, and Tilman & Pacala, in Ricklefs & Schluter’s influential 1993 book Species Diversity in Ecological Communities: Historical and Geographical Perspectives that first promoted the view that the unimodal pattern was both common (even “the true productivity pattern”), and to be expected on theoretical grounds. Neither chapter was (or was intended to be) a comprehensive review of the empirical literature. But the exemplary empirical examples cited, along with the theoretical explanations offered, were sufficiently compelling to sufficiently many people that the unimodal hypothesis quickly took hold.

Pushback arrived quickly, too. Abrams (1995) pointed out that the theoretical mechanisms widely thought to produce unimodal diversity-productivity relationships could produce non-unimodal relationships (and didn’t have much empirical support anyway), and that there were other, less familiar theoretical mechanisms that could produce unimodal relationships. Comprehensive and widely-cited empirical reviews by Waide et al. (1999) and Mittelbach et al. (2001) showed a variety of diversity-productivity relationships in empirical data, depending on study system, spatial grain and scale, and other unknown factors. But I think it’s fair to say that unimodal relationships have continued to attract the bulk of the theoretical and empirical attention (e.g., Kondoh 2001). It remains to be seen whether the recent massive study of Adler et al. (2011), which looked at grassland plant communities around the world using a standardized sampling regime, will finally kill off this zombie idea, at least for terrestrial plant communities.

In some ways, the unimodal diversity-productivity relationship is a weaker zombie than others I’ve discussed on this blog. The zombie idea took hold in part because it was predicted by some perfectly valid and testable mathematical models. It didn’t arise from logical fallacies like the IDH, or over-literal interpretation of mathematical models like r-K selection. For that reason, unimodal empirical relationships were never widely identified with, or worse, mistaken as evidence for, any one particular theoretical explanation. The other reason the zombie took hold was because of some clear-cut examples: there really are observational and experimental systems that show a unimodal diversity-productivity relationship (some of my favorites are Kassen et al. 2000, Dodson et al. 2000, and Chase & Leibold 2002).  These examples were probably given undue weight compared to examples showing other patterns. But thanks to the reviews of Waide et al. and Mittelbach et al., I don’t think that in recent years the unimodal empirical pattern was widely seen as universal. So if people saw it as somewhat more common than it really is, and so gave it more theoretical or empirical attention than it perhaps deserved, well, in the grand scheme of things that’s not so bad.

Indeed, I wonder a little if the unimodal diversity-productivity relationship is more of an “apparent zombie” (or better, a straw man). In the aftermath of Mittelbach et al., I haven’t seen too many people actually saying any more that the unimodal diversity-productivity relationship is universal, even at the local scale. Rather, people say that the unimodal diversity-productivity relationship is “widely accepted” (Adler et al. 2011), or “much discussed”, or etc. So maybe it’s not so much that everyone believes it, as that everyone thinks that everyone else believes it. Or maybe everyone just says that everyone else believes it, as a way of motivating further study of the diversity-productivity relationship! All of which is another reason to follow my advice to motivate your work by talking about ecology, not about what ecologists say about ecology. Because that straw man you’re setting up may be a straw zombie.

*footnote: This is not really my motto.

7 thoughts on “Zombie ideas in ecology: the unimodal diversity-productivity relationship

  1. Pingback: Trying to save a zombie idea « Oikos Blog

  2. Pingback: When, if ever, is it ok for a paper to gloss over or ignore criticisms of the authors’ approach? | Dynamic Ecology

  3. Pingback: WIWACS vs. zombie ideas | Dynamic Ecology

  4. When I started reading this blog (in 2011), this was one of the first posts I read. I think it took me until today to realize that (effect of) biodiversity -> productivity (e.g. Cardinale et al. 2011) was a different question from (effect of) productivity -> biodiversity (above). Seems like progress! 😉 For context, I am reading David Tilman’s 1999 MacArthur Lecture paper (Ecological Consequences of Changes in Biodiversity) and read “greater plant diversity leads to greater primary productivity” and I thought, wait, didn’t I read about that being a zombie idea on Dynamic Ecology way back when? FWIW, I think “Productivity is a poor predictor of plant species richness” is a clearer title than “Worldwide evidence of a unimodal relationship between productivity and plant species richness.” Relationship is an ambiguous term. Love this quote from the Grace et al. 2016 paper: “Our findings give reason for optimism about the future of ecology as a more precise and less ambiguous science.” I think that’s a good goal. 🙂

Leave a Comment

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 )

Facebook photo

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

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.