Seeing the Rainforest for the Trees

While geographer Alan Grainger has upset conventional wisdom by suggesting the world's tropical forests are not shrinking, he sees his research as a clarion call.
Publish date:
Social count:
While geographer Alan Grainger has upset conventional wisdom by suggesting the world's tropical forests are not shrinking, he sees his research as a clarion call.

Dr. Alan Grainger, a senior lecturer in geography at the University of Leeds, is an internationally renowned expert on tropical deforestation. He earned his doctorate from Oxford in 1987 for producing the world's first global computer simulation model of the tropical forests, and wrote a seminal book on the subject in 1993 called Controlling Tropical Deforestation.

But his most recent publication, in the latest edition of the Proceedings of the U.S. National Academy of Sciences, challenges everything we thought we knew about the subject. Grainger spent more than three years combing through all available United Nations statistics, stretching back 30 years, and discovered that the data, in defiance of conventional wisdom, didn’t show that the world’s forests were shrinking.

Grainger first examined data published every 10 years by the United Nations Food and Agriculture Organization since 1980. FAO's Global Forest Resources Assessment in 2000, for example, showed that all tropical forest area fell from 1,926 million hectares to 1,799 million hectares between 1990 and 2000 — one hectare equals 2.47 acres. Ten years earlier, however, FAO’s report had said that tropical forest area plunged from 1,910 million ha to 1,756 million ha for the same 90 countries between 1980 and 1990. Grainger used data from FAO’s latest report, published in 2006, to show that in a few countries, such as Gambia and Vietnam, forest area has actually expanded since 1990, as the reforestation rate has exceeded deforestation.

To gather the kind of accurate, consistent data scientists need to chart the actual transformation of the tropical forests, Grainger calls for the establishment of a World Forest Observatory to conduct annual surveys of the forests. Grainger discussed his vision for such a group — and the surprising findings in his most recent study — with Miller-McCune’s staff writer Matt Palmquist.

Miller-McCune: It would seem logical that with better satellite imagery over the years, we might be able to see “more forest” than in the past. What did you expect to find before you actually began delving into the data? Did the levels of discrepancy surprise even you?

Grainger: I had no expectations before I began my study. The disparities between forest area trends in FAO's Forest Resources Assessments

For the benefit of your readers it may help to point out that there are two time series in my paper. The first deals with all tropical forest and comes solely from FRA statistics.

The second deals only with tropical moist forest and uses estimates from various sources. For the period between 1990 and 2000, these include estimates based on pan-tropical satellite surveys. These are slightly higher than those based on expert assessments up to 1990. It is indeed logical to expect, as I say in my paper, that some of this "rise" is due to the use of more accurate remote-sensing techniques.

Miller-McCune: What’s your explanation for the lack of a net long-term decline? Is reforestation happening more quickly than we thought?

Grainger: Before I answer this question, I should point out that tropical rain forest is just one of the two main types of tropical moist forest which occurs in the humid tropics. All tropical forest, meanwhile, includes both tropical moist forest and tropical dry forest, which is found in the dry tropics. Most tropical moist forest has a closed canopy. There is some “closed forest” in the dry tropics, but much of it is open forest, with a relatively open canopy. Such forests are often referred to as savannah woodlands.

The only inference I can reliably make from the first time series, for all tropical forest, is that inconsistencies between the different trends raise questions about the presence of long-term decline. The size of the difference between the FRA 1990 and FRA 2000 trends, however, suggests that the errors involved could be of the same order as the amount of forest said to be cleared in each 10-year period. I identify various sources of error which may be involved. One is the error associated with estimates of the area of open forest, which accounts for 40 percent of all tropical forest and is not surveyed very frequently. To remove these latter errors, I produced a second time series for tropical moist forest only, and this is the series that shows no apparent long-term decline.

Considerable errors are associated with this tropical moist forest time series, too. Just as these errors may obscure deforestation, so it seems logical to hypothesize that natural reforestation is taking place on a wider scale than previously assumed. I am careful in my paper to show, by means of various references, that others have been studying this phenomenon for about 10 years on a sub-national scale. I also cite pan-tropical estimates that natural reforestation is equivalent to between 10 and 20 percent of deforestation on an annual basis. What I infer from the second time series is that the actual proportion may well be higher than this.

Miller-McCune: In general, what has been the response to your findings from your peers who study this issue?

Grainger: Since publication, responses from colleagues have been positive. I'm sure that some will want to raise questions about the study, but that is how science works. However, I have been extremely careful about making statements and inferences, as it is difficult to be too definite when such large errors are involved.

In retrospect, one could view my paper as questioning the validity of previous statements by others that have been too categorical. When I submitted the paper, I hoped it would persuade colleagues who use FAO data to pay much more attention to their quality than many of them have done in the past. Analyzing data quality is an essential scientific practice, and I am surprised by how little attention has been paid to this in previous studies. I also hope that editors of scientific journals will insist on this when papers are submitted for review. That many have not done so in the past is also surprising.

This is why I recommend at the end of my paper that all researchers who have used FRA data should check the findings of previous studies to see how they have been influenced by errors.

Miller-McCune: Do you think we’ll see researchers looking back at and revising their previous studies?

Grainger: I cannot predict the overall effect of this re-evaluation but it will be considerable, as FRA data have been used extensively in global change and land change sciences. Forest area estimates have been used directly to model trends in biodiversity. Climate change and land change studies have used estimates of deforestation rates. With regard to climate change, if the actual scale of natural reforestation does prove to be very different from that assumed until now, then this will have an impact on carbon budget calculations.

I say in my paper that re-evaluating previous studies that used FRA data deserves to be the subject of a separate paper. I think that was rather over-optimistic. In reality, owing to the scale of the task, it will take many years of research and several Ph.D. projects. My own paper is the result of years of research in itself, but I'm afraid that the consequences of its findings could keep others busy for far longer.

Of course, this would not have been necessary if we had not blundered into the practice of global change and land change sciences. If instead we had developed them in a systematic way, then we would never have ignored the importance of analyzing the quality of our data. There are many reasons for this haphazard approach, not least the compartmentalization of natural sciences from social sciences, and their mutual neglect of the crucial overlap area between them, which is the province of global change and land change sciences.

Miller-McCune: It would seem, on the whole, that your findings could be seen as encouraging: After all, we have more tropical forest than we thought we did. Are you at all concerned that this study might lead lawmakers or scientists to treat this problem with less urgency — that is to say, dismissing the declining rainforests as just another “green myth”?

Grainger: I am afraid that I do not find the situation encouraging at all. My findings are actually a condemnation of the world's inability to monitor long-term trends in these marvellous forests with any degree of accuracy. I have called for a long-term global monitoring program for tropical forests for 25 years, so my own efforts have clearly failed.

But the implications of this are far wider. Back in the 1980s, it seemed entirely self-evident to me that if we were to practice a true global change science, then the whole planet should be our laboratory and we should collect data on a global scale. While this view may well have taken hold in atmospheric research, the fact that many of my colleagues still depend on national forest statistics suggests that this is far from the case for terrestrial studies.

I fervently hope that my findings are not used to substantiate the argument that stories of declining tropical forests are just a “green myth." I go out of my way in the paper to state quite firmly that my findings do not show that deforestation is not happening. They merely show that we have not been able to measure forest change with sufficient accuracy.

Miller-McCune: In your mind, how would a World Forest Observatory actually work? What would be the monitoring technique?

Grainger: A primary aim of the World Forest Observatory will be to survey all tropical forests in the same year. This will reduce many of the correction errors I identify in my paper. The more frequent pan-tropical monitoring is, the easier it will be to estimate the true extent of natural forest regeneration. This is obscured when monitoring is infrequent.

I see the WFO bringing together under one umbrella the leading teams around the world who are already engaged in monitoring tropical forests using satellite remote sensing methods. Their work will naturally be complemented by extensive collection of ground data. The governing body of the WFO will channel the additional funds that these teams will need to expand their existing efforts, and ensure that all teams use the same approach for monitoring.

Miller-McCune: What would be the first steps toward effectively creating such an organization?

Grainger: At the end of my paper, I refer briefly to the institutional challenges of establishing a feasible global monitoring system for tropical forests.

I am convinced from this analysis that the World Forest Observatory must be a non-governmental initiative — somewhat like the Forest Stewardship Council

I see the work of the World Forest Observatory running in parallel with FAO's Forest Resources Assessments. FAO will continue to collect statistics from its member states as it has always done. The World Forest Observatory could easily take over the parallel remote sensing survey, commissioned from outside scientific bodies, that was included in FRA 2000 and is planned for FRA 2010.

Miller-McCune: Are we likely to see such an institution mandated in the next version of the Kyoto Protocol?

Grainger: A very interesting question. It is quite clear that if developing countries participate in this international agreement, and a Reduced Emissions from Deforestation and Degradation scheme is introduced, a very precise and accurate global forest monitoring system will be essential.

We have the technology and the scientists to design and implement such a system.

The Intergovernmental Panel on Climate Change has done marvellous work in advancing global change science and presenting the results of this to governments. Yet why is it that every time it releases a report we see negotiating sessions stretching long into the night? It is because of the IPCC's hybrid structure, by which government representatives have the right to challenge how scientific findings are summarized in the Policymakers' Summary. I do not question the rights and wrongs of this, I simply use it as an example of how the relationship during science and government is by no means simple.

One important lesson from all of this is that global change and land change scientists must become far more aware of how their work, and even the data they use, is institutionally framed. Another lesson is that conflicts between the credibility of scientific data and the sovereignty of member states, which is fundamental to the whole U.N. system, suggest that the U.N. urgently needs to re-evaluate the role of science in its operations.