Published online 7 July 2009 | Nature | doi:10.1038/news.2009.641

News

Bigfoot study highlights habitat modelling flaws

Accurate prediction of climate change's effects is as elusive as the fabled apeman.

BigfootBigfoot has helped provide a reality-check on climate impact models.Alamy

Climate change, it turns out, is going to be a mixed blessing for the sasquatch. The legendary American apeman will lose some of its existing habitat in the coastal and lowland regions of the northwestern United States, but gain a lot of new land in the Rocky Mountains and Canada.

So say biologist Jeff Lozier of the University of Illinois at Urbana-Champaign and his colleagues, in an analysis just published in the Journal of Biogeography1. But they're not really worried about bigfoot. Instead, they're trying to warn their colleagues that ecological models are only as good as the data that go into them.

Lozier's team subjected bigfoot to a technique called ecological niche modelling. This involves taking records of where a species has been found, and then, by combining these with environmental data, deducing where it ought to live or has lived in the past, present or future.

Such models are among the main tools in efforts to predict and plan for the biological effects of climate change. And because their predictions can be displayed as intuitive and dramatic maps, they have a psychological power beyond most scientific graphics.

Mistaken identity

But researchers' enthusiasm for such analyses risks outpacing their understanding of them, says Lozier. "The method is really new, and it's not fully worked out. I think some people have been seduced by the pretty output."

“We were trying to do the same thing for the yeti.”

Carsten Rahbek
University of Copenhagen

One problem is misidentification. It's hard to judge whether someone really saw what they thought they saw where they saw it, particularly in less well-studied groups such as insects — or American apemen. Mistake one species for another, for example, and your model will mislead.

Such errors can be hard to spot, because even if all the data are all highly dubious, a model based on them can still give a plausible-looking result, as Lozier and his colleagues found when they analysed sightings recorded by the Bigfoot Field Researchers Organization.

The reported sightings imply that the wooded and mountainous areas of California, Oregon and Washington teem with sasquatch at present. Warm the climate, and, like many other species, it will probably move north and uphill.

"It's a perfect commentary on the potential problems of this approach," says Lozier. "Plus, it's a sasquatch paper."

No crystal ball

Unlikely as it sounds, Lozier's paper scooped work by another group. "We were trying to do the same thing for the yeti," says ecologist Carsten Rahbek of the University of Copenhagen. Like Lozier, he wanted to show that models could turn dubious data into plausible-looking predictions.

A few years ago, only a few labs had the expertise to do ecological niche models. But now they are accessible to just about everyone, thanks to online data sources and user-friendly modelling software.

Much of the resulting work is "very naive", says Rahbek. "I'm editor-in-chief of a journal (Ecography) that gets a lot of these studies, and we reject nine out of ten."

ADVERTISEMENT

Misidentification isn't even the biggest problem with these models, says ecologist Joaquin Hortal of Imperial College London. More important is bias: if researchers only collect along roads, for example, then models will suggest that the species lives only along roads. "Biodiversity data [are] usually environmentally and spatially biased," he says.

Even if accurate data go in, a model's predictions of where species will go, and which are most at risk of extinction, will be imprecise and uncertain. "We in the modelling community need to be a bit more humble about how precise our predictions are, and acknowledge the errors of estimates, which are huge, more than we do," says Rahbek. "It's just damn hard to predict the future."

So if you need to be cautious about ecological niche models' inputs and you can't be certain about their outputs, are they any use at all? Yes, says Rahbek, because their predictions show consistent trends, such as European wildlife moving north and east as the climate warms. If the data were all random noise, then the predictions would be, too. 

  • References

    1. Lozier, J. D., Aniello, P. & Hickerson, M. J. J. Biogeogr. published online. doi:10.1111/j.1365-2699.2009.02152.x (2009).

Comments

If you find something abusive or inappropriate or which does not otherwise comply with our Terms or Community Guidelines, please select the relevant 'Report this comment' link.

Comments on this thread are vetted after posting.

  • #60601

    There have been tales of hairs being sent off for DNA sequencing for years... and that's all you ever hear of it. There are accounts of results coming back as unknown ape or unknown canid, but they seem to be urban legends as far as I can tell. JustinI @webs.com

Commenting is now closed.