Some zombie ideas in ecology–ideas that should be dead, but aren’t–are about purported empirical generalities that actually aren’t generalities at all. Diversity mostly doesn’t peak at intermediate disturbance or intermediate productivity. Local-regional richness relationships are not predominantly linear. There’s no general trend toward stronger herbivory and plant defenses in the tropics. And there are probably other examples I don’t know about. It’s not a matter of general patterns having a few exceptions, or being somewhat noisy, or being a bit more complicated or limited in scope than widely believed. In most or all these cases, there’s not much there worth calling a “pattern” at all (at least not in my view or in the view of many others).
Clearly, once an idea (zombie or otherwise) is established, it’s hard to get rid of. Which isn’t necessarily a bad thing–indeed, often it’s a good thing. But like all good things, reluctance to change one’s mind can be taken too far. In the comments on a previous post, Angela Moles relates the experience of having had her own work testing for (and failing to find) latitudinal gradients in herbivory and plant defense rejected seven times, on the grounds that she “must” have done something wrong because we “know” those gradients exist. It was only after she published a meta-analysis showing that in fact most studies don’t find those gradients that reviewers stopped assuming her own empirical work “must” be wrong.
But how do zombie ideas about empirical generalities get established in the first place? I’ve been wondering about this because many ecologists are noted for their reluctance to generalize. WIWACS is the acronym Mick Crawley coined for the “world is infinitely wonderful and complex school” of ecological thought. Also known as the “but in my system things are different” school. Anecdotal evidence suggests that it’s a big school of thought (e.g., about 2/3 of the audience at Tony Ives’ recent MacArthur award lecture denied that ecology is about the study of “general laws”, although I freely admit that that anecdote is tough to interpret because different people mean different things by “general laws”). And there’s often good reason to be skeptical of empirical generalities in ecology. At some level, ecological systems really are complex–there are lots of species, interacting in lots of different ways, affected in various ways by lots of different abiotic variables, etc. And there’s a lot of variation among species, places, and times. So how come ecologists, who in many respects seem so reluctant to generalize (again, often with good reason), in several cases have been quick to believe purported empirical generalizations that turned out not to be general at all?
The only answer I can think of is publication bias. Studies finding a pattern, and finding it to be especially strong, are more likely to get published first, and get published in more widely-read venues. Thereby establishing the pattern. In other words, it’s precisely the atypical studies that get published first and most prominently, and so get taken as typical by readers.
Probably it also helps if the pattern was predicted by some prominent theory. If you expect to find something, and then happen to find it the first time you look for it, you’ll be quick to assume that you’ll find it (or “should” find it) everywhere you look.
But even that picture still assumes that ecologists are reasonably willing to generalize based on just a few empirical studies. Which kind of contrasts with the image of many ecologists as members of the WIWACS, doesn’t it? So I don’t know. Maybe, despite that show of hands at Tony Ives’ talk, most ecologists actually aren’t members of the WIWACS after all? Maybe most ecologists are actually quite prepared or even eager to believe that there are lots of empirical generalities out there? Maybe many ecologists are skeptical of general theories or models, but are happy to believe in general empirical patterns?
It’s also interesting to ask the reverse question. Are there cases where the evidence for a general empirical pattern is very strong, but many ecologists don’t accept it? I can’t think of any great examples of the top of my head. Maybe the closest would be the prevalence of interspecific competition and trophic cascades? Although those aren’t great examples because I think as good data came in people mostly did come around reasonably quickly (am I wrong about that?) Can anyone think of any better examples?
I have no answers here. I’m just asking the question. How do zombie ideas about purported empirical generalities get established in the first place? And are there any cases in which real empirical generalities haven’t been widely accepted despite strong evidence?
Just to be clear: there are plenty of real generalities that all ecologists, including me, believe in! Tropical regions have more species, polar regions have fewer. Larger areas have more species. Species-abundance distributions have a few common species and many rare or fairly rare ones. Etc. etc. And there are also cases in which many ecologists were skeptical of a purported empirical generality and were vindicated when the “generality” turned out not to be general. Think for instance of the reluctance of many ecologists to believe in purported patterns in early studies of food web topology, which were overturned as better data came in. I don’t think that ecologists always, or even usually, misidentify generalities. But even if cases of misidentified generalities are exceptional, I still think it’s interesting and useful to ask how those exceptions came to be. If there are common threads in cases where many ecologists have been too quick, or too slow, to believe in claimed empirical generalities, recognizing those common threads can help us avoid such mistakes in future.
Interesting post, thank you.
“Maybe many ecologists are skeptical of general theories or models, but are happy to believe in general empirical patterns?”
Maybe it’s a psychological effect? Humans love patterns and symbols, and general empirical patterns are ‘tangible results’ that are accessible to everyone…they can be used as foundations for new studies, or as examples to teach students, or to back up an argument etc. But general theories/models are seen as just that, theoretical, and they are more amenable to being tested, built on or even disproven.
Also, accepting generalities can depend on context and scale, which I think you hint at above – it can be useful, and sometimes even fairly accurate, to make generalities at some levels (e.g. all roses), but not others (e.g. all flowering plants).
+1 for psychology! Draw a point pattern with complete spatial randomness and most will see clustered distributions. And the idea that random processes represent approximately such a complex world in many cases could be more difficult to see from the perspective of one empirical system/area only.
Pingback: Posterior samples | Sam Clifford
Pingback: On progress in ecology | Dynamic Ecology