Are some general ecological concepts TOO general?: revisiting an Oikos classic

Submitted by drupaladmin on 12 May 2011.

As I've discussed elsewhere, the level of generality with which we conduct ecological research is up to us (well, usually). We can choose to focus on the forest, or the trees. And studies of different levels of generality complement one another--you don't fully understand the forest unless you understand the trees, and vice-versa.

But is there such a thing as too much generality, or the wrong kind of generality? It's a good thing to step back and see the forest for the trees, but what if you step back too far (into deep space, say)? Don't you lose sight of the forest, and maybe even end up mistaking the forest for something else?

I think so. In seminars, I use the hypothetical example of a 'general theory of growth'. Try to imagine a general theory of growth that would apply to everything that can be said to 'grow'. Not just individual organisms, but populations, economies, egos, the entire universe...The point is that just because two things can both be said to 'grow' doesn't mean they're comparable in any interesting or useful sense. Note that one could statistically compare the growth rates of, say, organisms and egos. But that doesn't make the comparison scientifically meaningful. I use this example because it's obviously silly--at least, I thought it was obviously silly. Then I attended a philosophy seminar at which the speaker argued, in all seriousness, that major disciplinary boundaries (e.g., between astronomy and biology) inhibit scientific progress because they inhibit the development of general theories--such as a general theory of growth that might apply to everything from growing organisms to the expanding universe. So at least to some, there's no such thing as too much generality!

Moving away from hypothetical examples and the sort of philosophers who focus exclusively on them, there are real-world examples of overly-general theories or concepts in ecology and evolution. The obvious examples are theories which were empirically false due to their over-generality. An example is the pre-Darwinian idea that development of individual organisms and evolution of species are closely analogous. Darwin showed that that's just false--development is directed towards a pre-determined goal, evolution isn't. So the notion of 'development' doesn't apply nearly as generally as some pre-Darwinians thought.

But more interesting cases are when theories or concepts are over-general without necessarily being empirically false. Over-generality in this sense is trickier to pin down, but can involve such features as:

  • vaguely-defined terms
  • loose analogies
  • lack of a general mathematical version of the theory or explication of the concept, thereby forcing reliance either on purely verbal models, or else on less-general mathematical models of specific cases, from which over-general conclusions are drawn
  • reliance on purely statistical tools to compare and contrast different case studies in a phenomenological way (as in statistical comparison of the growth rates of organisms and egos)

All these features have the effect of facilitating comparisons of 'apples to oranges'--comparisons that highlight comparatively superficial commonalities among different cases, while obscuring deeper distinctions that need to be drawn in order for explanatory progress to be made.

I've argued that the idea of 'biodiversity affecting ecosystem function' (BDEF) is over-general in this way (Fox 2006, Fox and Harpole 2008). The terms 'biodiversity' and 'ecosystem function' are like the term 'growth'--they embrace many phenomena which really need to be kept separate in order for explanatory progress to be made. I have used the Price Equation to suggest what distinctions ought to be drawn between different classes of BDEF problems (a 'divide and conquer' approach, if you like). This formal mathematical approach forces precise definition of terms, and highlights not only important distinctions but also important but unrecognized commonalities. For instance, 'community variability' is not generally regarded as an 'ecosystem function', but effects of biodiversity on community variability can be analyzed within the Price Equation framework (Fox 2010 Oikos).

One very prominent ecological idea which has been criticized as over-general is from the most highly-cited Oikos paper of all time: Clive Jones, John Lawton, and and Moshe Shachak's 'Organisms as ecosystem engineers' (Oikos 69:373 [1994]). Jones et al. 1994 is the Oikos classic referred to in the title of this post, and what a classic it is--cited over 1200 times (!) By way of comparison, that's substantially more citations than even older Oikos classics such as Connell 1980 (ghost of competition past), Hamilton 1980 (parasites explain the evolution of sex), and Hanski 1982 (core-satellite hypothesis). Jones et al. define ecosystem engineers as "Organisms that directly or indirectly modulate the availability of resources [other than themselves] to other species, by causing physical state changes in biotic or abiotic materials. In so doing, they modify, maintain and create habitats." The concept has subsequently been broadened to include essentially any effect of organisms on their physical environment (e.g., Harmon et al. 2009). Ecologists now have a huge range of case studies of ecosystem engineering (reviewed in Wright and Jones 2006), mathematical models of a few specific examples (e.g., Wright et al. 2004 Oikos), and an admirable recent effort to unify these disparate examples within a general (but unfortunately, not fully specified) mathematical framework (Jones et al. 2010 Oikos). Despite all this, even ecosystem engineering's staunchest advocates recognize that they have yet to present a fully-convincing argument that they're not engaging in apples-to-oranges comparisons (Jones et al. 2010). That some specific models of ecosystem engineering (e.g., Wright et al. 2004), and some very prominent case studies (e.g., Harmon et al. 2009) appear not to fit within the general mathematical framework of Jones et al. 2010 is a little worrisome. I honestly admire the imagination and ambition of the idea of ecosystem engineering--these are the kind of ideas that drive real progress in ecology. But I remain to be convinced that there is a general, non-trivial theory that covers everything from the creation of aquatic habitat by beavers (Wright et al. 2004) to the effects of stickleback fish on the chemical composition of DOC (Harmon et al. 2009).

In fairness, I would be remiss if I didn't acknowledge that the ideas of BDEF and ecosystem engineering have spurred huge amounts of useful work, which will retain much value even if the original ideas which inspired that work ultimately prove to have been over-expansive. Indeed, I'm sure that these ideas would've been much less inspirational and influential if they had initially been more narrowly and precisely tailored. So if over-generalizing is at some level a mistake, well, sometimes ambitious mistakes can be more productive than cautious correctness. Probably, if we're not overgeneralizing sometimes, we're not trying hard enough to generalize in the first place.

New ideas