The text that follows is a PREPRINT.


Please cite as:


Laurance, W.F., M.A. Cochrane, P.M. Fearnside, S. Bergen, P. Delamonica, S. D’Angelo, T. Fernandes, C. Barber. 2001. Author response to S. Schwartzman and R. Bonnie. Science dEbates. 31 May 2001.


ISSN: 0036-8075


Copyright: American Association for the Advancement of Science (AAAS)


The original publication is available at  








Debating the future of the Brazilian Amazon


We are pleased to respond to Schwartzman and Bonnie, who assert that , whose criticisms bear an uncanny similarity to those made earlier by members of the competing IPAM team.  In fact, most of these criticisms fall apart on close inspection. 


Perhaps most astonishing is Schwartzman and Bonnie’s assertion that the IPAM study by the Instituto de Pesquisas Ambientais na Amazônia (IPAM) (1, 2) is more conservative than ours (3) and provides a better basis for policy conclusions.  While we were careful to give due credit to the pioneering work of IPAM, we believe that we have built a better mousetrap and dispute both of Schwartzman and Bonnie’s assertions.  The IPAM estimates are not “more conservative” but only less complete.  


Like us, the IPAM group evaluated historical deforestation along Amazonian highways and then extrapolated these results into the future.  However, the IPAM effort was based on a highly subset of only four highways that had caused especially heavy deforestation.  We used a far more reliable method, which involved assessing deforestation along all Amazonian highways, including several that had caused only limited deforestation.  Our extrapolations calculations were therefore more conservative and robust than those of IPAM.  Because of this important bias, the IPAM study actually projects a greater increase in future deforestation rates (400,000-900,000 ha yr-1) than does our study (269,000-506,000 ha



In addition, the IPAM study is far from comprehensive, because it fails to account for the effects of infrastructure projects and unpaved roads on Amazonian forests.  Some roads, such as the Northern Perimeter Road, will carve large swaths across the Amazon, strongly influencing deforestation, logging, mining, and other activities.  Infrastructure projects such as powerlines, gaslines, and hydroelectric reservoirs also contribute directly to forest-degrading activities because they require road networks for construction and maintenance.  Obvious eExamples of this can be seen in the Ecuadorian and Brazilian Amazon, where roads associated with gas- and powerlines and reservoirs have led to dramatic rises in slash-and-burn farming, logging, market hunting, and land speculation (4, 5).  Our assumption that major infrastructure projects will behave like unpaved roads—because they cannot be constructed without first making roads—therefore is logical and defensible.


The IPAM study has other key limitations.  It does not consider vast forested lands that would be inundated by planned hydroelectric reservoirs in the Amazon.  It also fails to consider the influence of protected and semi-protected areas (such as national parks, national forests, and indigenous reserves) on spatial patterns of forest loss and degradation.  Finally, it distinguishes only between forested vs. deforested lands—despite the fact that this is an obvious oversimplification.  Many activities, such as selective logging, forest fragmentation, surface fires, wildcat mining, and overhunting, can degrade forest ecosystems without causing deforestation per se.  Thus, the failure of the IPAM study to predict the extent of forest degradation significantly reduces its utility.     


While most of Schwartzman and Bonnie’s assertions can be easily rebutted, they do raise a valid point.  A debatable aspect of our models is the assumption that river-channelization projects would likely lead to increased logging, deforestation, and other degrading activities along rivers, comparable to those caused by unpaved roads.  This was, admittedly, educated guesswork, because nNo such projects exist in the Amazon on which to base  such extrapolationsprojections.  While our remote-sensing analyses suggest that forests near rivers with heavy boat traffic are especially prone to deforestation (5), further studies are needed to predict the impacts of river-channelization on Amazonian forests.    Contrary to Schwartzman and Bonnie’s suggestion, however, our analysis does not exaggerate impacts by double-counting deforestation from river channeling (or other) projects in already deforested areas because our geographic information system automatically tracks the status of each point in the landscape, preventing any one from being deforested twice. 


In summary, many of the large infrastructure projects included in our study—such as the Porto Velho-Urucu gasline, which will penetrate into the “pristine” heart of the Amazon—are likely to have dramatic impacts on the pattern and pace of forest conversion.  While predictive models such as ours can always be improved, ignoring such projects in the name of waiting for better data would be to neglect one of the most important features of Avança Brasil.  Our Policy Forum article helped to initiate a vigorous debate about the Avança Brasil program, and we regard this as a very healthy and timely development.


William F. Laurance1,2, Mark A. Cochrane3, Philip M. Fearnside4, Scott Bergen5, Patricia Delamonica2, Sammya D’Angelo2


1Smithsonian Tropical Research Institute, Apartado 2072, Balboa, Panamá

2Biological Dynamics of Forest Fragments Project, National Institute for Amazonian

   Research (INPA), C.P. 478, Manaus, AM 69011-970, Brazil

3Basic Science and Remote Sensing Initiative, Michigan State University, East Lansing,

   MI 48823, USA

4INPA Ecology Department, C.P. 478, Manaus, AM 69011-970, Brazil

5Department of Forest Science, Oregon State University, Corvallis, OR 97331, USA



1.  D. C. Nepstad et al. Avança Brasil: Cenários Futuros para a Amazônia (Instituto de

     Pesquisa Ambiental da Amazônia, Belém, Brazil, 2000).

2.  G. Carvalho, A. C. Barros, P. Moutinho, and D. C. Nepstad, Nature 409, 131 (2001).

3.  W. F. Laurance et al., Science 291, 438 (2001).

4.  B. Holmes, New Scientist 151(2048), 43 (1996).

5.  S. Bergen et al., The Future of the Brazilian Amazon: Development Trends and

     Deforestation (



Word count: main text: 662 words; overall: 810 words