One of the big questions regarding terrestrial ecosystems was how would these ecosystems respond to the change in the climate system being driven by elevated atmospheric carbon dioxide? In particular, how would changes in the distribution and abundance of terrestrial ecosystems feed back to the elevated CO₂? Would the changes in the terrestrial ecosystem result in increased productivity, therefore increasing the rate of uptake of carbon dioxide from the atmosphere and functioning as a negative feedback, or would it result in a reduction of greenness, if you will, and actually pump carbon from the terrestrial environment into the atmosphere, thus functioning as a positive feedback? This was a very important question in the context of modeling possible climate changes.

At the time, my colleagues and I had been evaluating the consequence of a changed climate system using relatively simple correlative models that we called "bio-geographical models." We would produce maps of potential changes in the distribution and abundance of ecosystems under different climate-change scenarios. One of the advantages of this very broad-brush view of terrestrial ecosystems is that we have estimates of the amount of carbon in various ecosystem types—tropical rain forests, for example. So you could do a very simple exercise of classifying different ecosystems into these broad categories, and then do the bookkeeping of how much carbon is in the vegetation, in the soils, etc. And you could do that under current conditions and do it under the redistribution of ecosystems under climate change.

All the scenarios suggested that, overall, climate change would result in a greener planet, which suggested that there was negative feedback—the amount of carbon stored in the terrestrial ecosystems would increase under these climate-change scenarios. The problem with the analyses done up to that point was that they were all what are called "equilibrium solutions." In other words, climate modelers would double the concentration of carbon dioxide in the atmosphere and then look for a solution once everything is back in equilibrium. What they don't tell you is how the system got to equilibrium. What happened? What were the transient dynamics, as the system went from point A—ambient conditions—to point B—a doubling of the carbon dioxide concentration? So what we got interested in—the big question—was how did these ecosystems get from point A to point B, and what did that imply about the transient dynamics of carbon exchange between the terrestrial surface and the atmosphere?

  A naïve guess would suggest that climate modelers went for the equilibrium solutions because they were much simpler, and that calculating the transient dynamics would be very hard. Was that the case?

Yes. This turns out to be an incredibly difficult problem. In the fullest sense, it requires that you literally simulate the dynamics, spatially and temporally, of all these ecosystems. We still have not arrived at the point where we can do that. What we did was ask, how can we use our models to get a first estimate of what the transient dynamics would be? So we took a very simple approach: for each location on the Earth's surface, we did an analysis to calculate the implied shift from one ecosystem to another as the climate changed. We looked at all those transitions—going from forest to grasslands, for instance, or vice versa—and we assigned a time frame on which these different processes operate. Then, by classifying these transitions into categories based on the processes that would control these transitions, we were able to put, in effect, a transient face on this. We were able to give an estimate of what would happen as the CO₂ concentration in the atmosphere doubled, as we went from point A to point B.

  And what was the result?

It was interesting. Even though the doubling of carbon dioxide implied that terrestrial ecosystems would hold more carbon—that is, act as a negative feedback—the transient dynamics going from point A to point B implied the opposite and that a lot of carbon would move into the atmosphere. Any way you looked at it, the transient dynamics said the ecosystems would be pumping carbon into the atmosphere. And the beauty of this analysis and approach was that it didn't depend on the particular numbers used. It didn't matter if we said that to go from grassland to forest took 100 years or 200 years or 300 years, and to go from forest to grassland was 10 years or 20 years or 50 years. The result was robust. Simply put, the processes that result in the loss of carbon into the atmosphere operate on a much faster time-scale than those processes associated with the storage of carbon in the terrestrial ecosystems. I always jokingly say that now climate modelers have realized something every gambler eventually realizes: it takes a long time to build up your money but you can lose it all in one hand.

  Were you surprised with the impact of your Nature paper?

Not really, because I think it was answering a very important question with a very robust solution. I also think one reason it may have been so highly cited, however, is because there are very few papers that have been written on the topic. And that may be because going to the next step is infinitely more difficult. So I think it may have received more attention because we're sort of stalled in the process of moving forward. If it were just one more step on a continuous process of advances, its shelf life wouldn't be as great as it's been. The next step, however, requires a greater leap than we've been capable of making.

  What is it that's preventing researchers from taking that leap?