Ecologists Peter Reich and Sarah Hobbie at the University of Minnesota in St. Paul, Minn., conducted a 13-year study of perennial grassland species, where they mimicked the projected atmospheric CO2 levels of the year 2070. They found that unfertilized grass plots grown under the increased CO2 levels grew only half as fast as plots fertilized with nitrogen.

Other studies have found that computer models could be overestimating land plants' carbon absorption capacity by as much as 23 percent, because they don't take into account these nitrogen and phosphorus limitations.

The Higgins and Harte study did not address the nutrient limitations on plant migration. But Higgins said that when those limits are factored in, the potential impacts of climate change would be narrowed and likely to be more extreme.

How we choose to deal with the range of possibilities these studies present is a risk-management question that shouldn't be ignored, Higgins said.

"There's not a clear answer," he said. "How deep the cuts we should make in emissions depends on how much risk society wants to run."