Predicting Oblivion: Are Existing Models Up to the Task?

Science, August 17, 2007

The most authoritative guide to today’s extinction crisis is a database known as the Red List. Later this month, a group of scientists will gather in England to consider whether the Red List should be opened up to species that, for the moment, show no signs of trouble. Many scientists suspect that the next few decades of global warming could push some species toward oblivion. “The concern,” says the meeting’s organizer, H. Resit Akçakaya, an ecologist at ecological software company Applied Biomathematics in Setauket, New York, “is that maybe some species that are threatened by climate are not reflected on the Red List.”

But Akçakaya and others caution that the meeting is unlikely to come up with firm predictions of how many species will become extinct, let alone which ones will be particularly at risk.

The science of predicting extinctions from global warming is only a few years old, and the best models are rife with uncertainties. Experts generally agree that the models may be useful for giving a rough idea of the potential impact of global warming and may also offer guidance for planning preserves. But some scientists are concerned that policymakers will be expecting them to provide more precise estimates than they can deliver. “It’s worrying, says Miguel Araújo, an ecologist at the Spanish National Research Council in Madrid.

Much of the current debate over climate-triggered extinctions focuses on what are known as climate-envelope models. Scientists analyze all the places where a species has been recorded and look for features of the climate that those places share. The key factors may be rainfall, for example, or the temperature during the winter.

In the early 2000s, scientists began to look at what happened to these climate envelopes in the scenarios climate scientists have projected for the coming century. “A number of us were noticing that these envelopes seemed to be winking out entirely,” says Lee Hannah, chief climate change biologist at the Center for Applied Biodiversity Science at Conservation International, a nonprofit in Arlington, Virginia.

Concerned about the prospect of mass extinctions, an international team of scientists, including Hannah, combined their data into a global analysis. They estimated the size of future climate envelopes, assuming shrinking climate envelopes meant an increased risk of extinction. Their sobering conclusion, published in Nature in 2004: Based on a midrange climate-warming scenario for 2050, “15-37% of species in our sample of regions and taxa will be ‘committed to extinction.’”

The paper was enormously influential and figures prominently in the Intergovernmental Panel on Climate Change’s (IPCC’s) upcoming report on the impact of global warming. In a summary for policymakers, the IPCC authors warn that “approximately 20-30% of plant and animal species assessed so far are likely to be at increased risk of extinction if increases in global average temperature exceed 1.5-2.5°C.”

Some experts have criticized the IPCC for implying that climate-envelope models are more precise than they actually are. “Simply presenting those numbers as factual, saying this is how many species will go extinct, is misleading,” says Richard Pearson, a postdoctoral researcher at the American Museum of Natural History in New York City.

Pearson and other researchers have been testing climate-envelope models for their accuracy and consistency, and they’ve found some serious causes for concern. Araújo and his colleagues studied the ranges of 116 species of birds in England in the 1970s and 1990s. The red-backed shrike’s range shrank dramatically to southeast England in the 1990s, for example, but climate-envelope models based on the 1970s data predicted that the bird’s range would stretch all the way to the northern tip of Scotland. “We found that there were lots of uncertainties,” he says.

Araújo, Pearson, and other researchers published a study last year in which they compared the projections for a group of plant species in South Africa from several frequently used envelope models. “We found a huge difference between the models,” says Pearson. Their projections ranged from a 92% range reduction to a 322% expansion.

Scientists are currently debating how to make better predictions. Climate-envelope models are “simply mapping programs,” complains Daniel Botkin, a professor emeritus at the University of California, Santa Barbara. “There’s no biology in that.”

To improve the performance of these models, Botkin urges researchers to include biological details about species, such as how quickly they disperse and how they interact with other species. Pearson and other modelers have already had some success at doing so, Pearson says. In a paper in press at Global Ecology and Biogeography, Pearson and his colleagues report that they can do a much better job of predicting the ranges of owls in Finland if they also factor in where woodpeckers live, as owls make their nests in woodpecker cavities. “In my opinion,” says Pearson, the role of biological interactions “is the biggest question out there at the moment, but we’re just nibbling on the edges of that.”

Other researchers believe that a better strategy is to analyze existing climate-envelope models more effectively. “You have to find automated ways to extract information in intelligent ways from the data you have,” says Araújo. He and his colleagues have found that averaging the results of many climate-envelope models provides a more accurate prediction of where species can be found than any one model. “I think that’s a much more useful way to go,” says Araújo. “This is likely to be a closer match to the truth than anything else we can produce so far.”

One problem with these so-called ensemble forecasts, however, is that they are a huge undertaking. Running thousands of models of thousands of species across an entire continent is far beyond the capacity of any existing software. “In the next 2 to 3 years, we won’t be able to do it,” says Araújo, who is now developing a program he hopes will be up to the task.

Scientists are debating not just how to make these models better but also the best way they can be used to make conservation decisions. Some researchers are trying to estimate the percentage of species that global warming will put at risk of extinction across entire continents or even the entire planet. Walter Jetz, a biologist at the University of California, San Diego, and his colleagues recently published a study of the combined impact of climate change and land-use change on birds. They found that several hundred species may lose over half of their range by 2050. Land use will have the biggest impact on birds in the tropics, whereas climate change will be stronger at higher latitudes.

Araújo and his colleagues have a more modest goal: trying to predict patterns of change in different regions. They’ve been forecasting which parts of Europe will be particularly vulnerable to losing species through climate change, for instance. They’ve found that for plants, the mountainous regions in southern Europe will be hit hardest. For amphibians, the arid parts of southwestern Europe are most vulnerable. For now, he suggests, such estimates may be more useful for conservation than a misleadingly precise estimate of a rate of extinction for a particular species.

Identifying these sensitive regions may reveal how existing preserves may change and offer hints about how to design new ones. As the temperature warms, some preserves will no longer have a climate suitable to the species they are supposed to protect.

A number of researchers are using climate-envelope models to study how preserves may have to be altered as species shift their ranges. For instance, existing preserves could be linked by corridors to enable animals and plants to disperse from one habitat to another.

Hannah believes that scientists must move forward with this sort of planning now, even if the models have plenty of room for improvement. “The scary thing for me is that the stuff our models is showing happening decades from now, we’re already seeing,” he says. He points to the extinction of frogs in the Andes, where researchers suspect that a changing climate may have fostered the spread of a lethal fungus. “These models are the best we’ve got at the moment, and when we see how the complexity of the world operates, it seems that it may be worse than these models are indicating.”