Monday, 19 March 2012

Distribution of Ethiopian Bush-crow and the nature of explanations

Yesterday I was sent a link to a press release from the excellent BirdLife International (read it here). It's talking about some research by an international team to try and explain the remarkably restricted range of the Ethiopian Bush-crow (cute picture here, since I've never actually been there to take my own), and in it, Paul Donald the lead author makes some interesting comments:

“The mystery surrounding this bird and its odd behaviour has stumped scientists for decades – many have looked and failed to find an answer.  But the reason they failed, we now believe, is that they were looking for a barrier invisible to the human eye, like a glass wall. Inside the ‘climate bubble’, where the average temperature is less than 20°C, the bush-crow is almost everywhere.  Outside, where the average temperature hits 20°C or more, there are no bush-crows at all.  A cool bird, that appears to like staying that way.”

The reason this species is so completely trapped inside its little bubble is as yet unknown, but it seems likely that it is physically limited by temperature – either the adults, or more likely its chicks, simply cannot survive outside the bubble, even though there are thousands of square miles of identical habitat all around.

BirdLife International’s Dr Nigel Collar is co-author of the study. He added “Whatever the reason this bird is confined to a bubble, alarm bells are now ringing loudly.  The storm of climate change threatens to swamp the bush-crow’s little climatic lifeboat – and once it’s gone, it’s gone for good.”

Two things jump out from this to me (1) the nature and validity of the explanation itself: we're told the authors think there's probably a direct physical limitation, not an indirect effect on, say, a essential food source. (2) The concern about climate change removing the suitable bubble of cool in the highlands. (I should point out that the paper is available here and I'm going to base the rest of the discussion on the actual paper, not the press release - both these points are made in the discussion of the paper, however.)

The first thing that interests me is the statement about other scientists failing to find an explanation for this distribution, and I wonder whether actually this point comes down to a change in what is considered a sufficient explanation for ecologists. There's also been another recent paper on the way biologists have changed how they think about explanations over the last few decades. This paper, by two philosophers of science, charts the first discussions of different levels of explanation by Ernst Mayr back in 1961 who defined proximate explanations (e.g. it got too hot and the bird's died of sheatstroke) and ultimate causes (people burnt fossil fuels, releasing carbon dioxide into the atmosphere leading to global climate change, and that was the end of the species). They're both valid descriptions, but they answer slightly different questions. Not long after then Niko Tinbergen (who my little brother did his PhD about, by the way - rather strange having a brother who studied biologists..) also refined these further saying that before something could really be explained you needed to answer four types of question what's the physiological mechanism? What's the development pathway that gets you there? What's the ecological function? And what's the evolutionary history? Now, the latest paper suggests the whole cause and effect idea might be more complex than we originally thought, with evolution acting in feedback sometimes - but that's a whole different set of ideas. Still, ecologists looking for an explanation of distribution over the last few decades may have been looking to explain distribution in these ways. Recently, I'm wondering if we settle too fast on a rather superficial explanation that doesn't really meet these criteria.

Certainly, no-one could call the work on the Bush-crow an explanation for the distribution according to any of the four criteria Tinbergen was looking for. And of Mayr's original two it might, perhaps, be an ultimate explanation. But let's look at that - in this study, as many others of the type, the authors started by looking at five different climatic variables (we don't know why they chose these from the 19 they had available): mean annual temperature, mean annual precipitation, temperature seasonality, mean diurnal temperature range and annual temperature range. Now, let's ignore the results for a moment and ask what sort of explanation we could possibly derive from these chosen variables. Could any come close to identifying a physiological limit? As they're all means (IE averages) I really don't think so: by definition the actual temperatue on any one day is pretty much certain to be above or below the mean. Similarly, it seems likely that a physiological limit to mean diurnal termperature variation is unlikely - though if the variation is too high on any one day I could see a real limit. Unfortunately, there's no way to tell from an average alone whether the sort of extremes that could potentially cause direct physiological problems are present as well. So instead of a direct (proximate) limit, using means is more likely to identify an indirect effect: one mediated and perhaps integrated over time and space by something else, or maybe the mean is simple pointing to some associated measure that is correlated with it. That correlated variable could be the extremes that might ahve direct effects, or it could be the demographic of some slow-growing plant that's essential for nesting, for example. In fact, the author's rule out any indirect effect by noting that the species is a generalist feeder that likes degraded habitats - unlikely to suffer any shortage of food or nesting spaces in any habitat. So we're left with the explanation that mean temperature (which the authors show has the highest association with the observed range) might be acting as a proxy for some as yet unknown temperature-related variable that could have a direct effect.

Now, even without bringing up my general concerns with this sort of analysis (e.g. that the statistics used are inappropriate for spatial data and will 'explain' any distribution you care to throw at them), I have to wonder if this 'explanation' for the bird's distribution has never before been suggested for the simple reason that it's never before been considered adequate? Indeed, is it, as a biological explanation, any more satisfactory that simply saying these birds live on hills in southern Ethiopia? Looking at one of the main figures (right) from the paper I can see two things that would seems surprising to me: firstly both preferred temperature and preferred rainfall at incredibly narrow. If this is true then, as the authors rightly point out, it would be of extreme concern given the likelihood of future change in these areas. But it seems implausible to me: annual rainfall in the 600mm range of eastern Africa is so variable year on year that the average itself is essentially meaningless to anything except a long-lived plant like a tree or slow-growing bush. I can't imagine a plausible ecological mechanism that would allow such a restricted range on a map of mean rainfall, when the annual rainfall is such that it rarely falls within this optimal range anyway! Secondly, I see in the seasonality plot two peaks - the birds apparently like moderate seasonality and quite high seasonality, but don't 'moderate-to-high' so much. Such a dislike of intermediate seasonality immediately rings alarm-bells for me: it's physiologically not plausible and, together with the implausibly narrow limits identified for rainfall suggests the model is 'over-fitted' - in other words, it's not going to be reliable for prediction of distribution under changed climate conditions.

Unfortunately, therefore, I'm not convinced we've really learnt much here: as a description of the range, this works nicely. But as an explanation, it lacks something. Certainly, we have no explanation of distribution that could meet any of the older definitions of biological explanation (nor the newly suggested one), but we do have a fairly typical example of what seems to be a lowering of the barrier in ecological explanations: to me, a suspect correlation with little critical ecological discussion cannot be considered an explanation. Come on ecologists, we're better than this!

Main Reference:

ResearchBlogging.orgDonald, P., Gedeon, K., Collar, N., Spottiswoode, C., Wondafrash, M., & Buchanan, G. (2012). The restricted range of the Ethiopian Bush-crow Zavattariornis stresemanni is a consequence of high reliance on modified habitats within narrow climatic limits Journal of Ornithology DOI: 10.1007/s10336-012-0832-4

No comments:

Post a Comment