Geographic distribution ecology

The Austral Grass Tree - what happens to it with climate
change won't be just about climate space. Grasstrees
tend to be found on impoverished soils, so
predicting response is not just about projecting
future suitability of the climate.
(Photo: Michele Kohout)

I've just come back from the annual meeting of the Ecological Society of Australia (held jointly with the New Zealand Ecological Society this year). I saw lots of talks with climate change as their key focus. Many of the talks were excellent, and included projecting/forecasting future species distributions using modelling approaches (SDMs). However, I also saw a glaring absence of field experiments in the general area of 'climate change biology'. I want to raise this apparent research gap here (and perhaps start a discussion amongst the scientific community about how (or if) we can change this).

Field transplant experiments are the most important missing information for understanding species movements under climate change. There is lots of modelling suggesting species ranges will shift, i.e.  observed climate envelopes (realized niches in environmental space) are being projected into the future on a large scale using SDMs but the problem is, we don't know what stops species from establishing populations beyond their range boundaries, i.e.  is climate the direct or indirect driver of species distributions?  I'd argue that field transplant experiments across geographical range boundaries are urgent to test this basic question. They are the reality-test for what sets range boundaries, and especially for when it is climate directly versus when it is a competitive milieu mediated by climate.

Why didn't I see much on this?

Well, such field experiments are difficult and time-consuming. It is crucial that such experiments involve transplanting species beyond the range, with and without amelioration of suspected limiting factors, both physical and biotic. Hence, they can get pretty complicated and pretty quickly! 

But such field experiments are also the essential reality test in relation to SDMs. The general point here is that we need to know not how plants grow where they do occur (this is ecophysiology), nor how much two species compete with each other where both occur (this is experimental community ecology) but rather, what happens when species move into new areas where they experience new species compliments, soils and pathogens (this is geographical distribution ecology).

Unfortunately, the experiments conducted in the 1970s-90s were the era of field experiments on competition and predation, with the overwhelming majority directed at measuring intensity of interactions within a geographical range, with very few at determining why range boundaries are where they are. Hence, we've lost a couple of decades already in trying to understand this crucial question.

I suspect that if we rely on chance decisions by individual research groups to undertake geographical distribution experiments, it's likely that several more decades will pass before a useful number will accumulate to inform the general question: what sets range boundaries? We really need to arrive at a situation within a decade or two where we have a moderately well-founded overview so we can far better understand the response of biota to massive shifts in climate.  Suppose optimistically that, say, 30 experiments a decade could be achieved worldwide. Can the research community collaborate to target them in such a way as to obtain generalization across species and boundary-types as efficiently as possible?

One way might be to assemble a global network with standardised protocols - much like has been done in the Nutrient Network (which tests top-down versus bottom-up controls on diversity in herbaceous ecosystems) or the International Tundra Experiment  (which tests the effect of warming on tundra and alpine vegetation). Both networks have simple protocols and clear questions and they encourage collaborative research across a large range of field sites. Hence, the power of the experiments lies in their spatial distribution with uniform methods.

I think this approach is one that could work to test range boundaries (once it is clear what sort of factors would need to be manipulated), and am in the process of writing a review paper on this idea with (hopefully) a rallying call for such experiments. Stay tuned....