SEE COMMENTARY

Cropland expansion changes deforestation dynamics

in the southern Brazilian Amazon
Douglas C. Morton*, Ruth S. DeFries*†‡, Yosio E. Shimabukuro§, Liana O. Anderson§¶, Egidio Arai§,
Fernando del Bon Espirito-Santo储, Ramon Freitas§, and Jeff Morisette**
*Department of Geography, 2181 LeFrak Hall, University of Maryland, College Park, MD 20742; †Earth System Science Interdisciplinary Center, University
of Maryland, 2207 Computer and Space Sciences Building, College Park, MD 20742; §Divisão de Sensoriamento Remoto, Instituto Nacional de Pesquisas
Espaciais (INPE), Av. dos Astronautas, 1758 Jardim da Granja, São José dos Campos, SP 12227-010, Brazil; ¶Oxford University Centre for the Environment,
University of Oxford, South Parks Road, Oxford OX1 3QY, United Kingdom; 储Complex Systems Research Center, University of New Hampshire, Morse Hall,
Durham, NH 03824; and **Goddard Space Flight Center, National Aeronautics and Space Administration, Mail Code 614.5, Greenbelt, MD 20771

Contributed by Ruth S. DeFries, July 27, 2006

Intensive mechanized agriculture in the Brazilian Amazon grew by ecosystem services (11), such as carbon storage, and future
>3.6 million hectares (ha) during 2001–2004. Whether this cropland deforestation dynamics.
expansion resulted from intensified use of land previously cleared Amazon deforestation is Brazil’s largest source of CO2 emis-
for cattle ranching or new deforestation has not been quantified sions (12, 13). Carbon fluxes from deforestation are a function
and has major implications for future deforestation dynamics, of the area of forest loss (14–16) and related forest disturbances,
carbon fluxes, forest fragmentation, and other ecosystem services. such as fire (17, 18) and logging (17, 19), variations in forest
We combine deforestation maps, field surveys, and satellite-based biomass across the basin (20), and land use or abandonment after
information on vegetation phenology to characterize the fate of forest clearing (3, 21). Land use after forest clearing remains a
large (>25-ha) clearings as cropland, cattle pasture, or regrowing major source of uncertainty in the calculation of deforestation
forest in the years after initial clearing in Mato Grosso, the Brazilian carbon fluxes because methods to assess deforestation trends in
state with the highest deforestation rate and soybean production Amazonia have not followed individual clearings over time (4, 5,
since 2001. Statewide, direct conversion of forest to cropland 22–28). The relative contributions of smallholder agriculture and
totaled >540,000 ha during 2001–2004, peaking at 23% of 2003 large-scale cattle ranching to annual forest loss have been
annual deforestation. Cropland deforestation averaged twice the inferred from the size of deforestation events (5, 28), but no
size of clearings for pasture (mean sizes, 333 and 143 ha, respec- direct measurements have been available. Rapid growth of
tively), and conversion occurred rapidly; >90% of clearings for large-scale agriculture in Amazonia challenges the historic re-

SUSTAINABILITY
cropland were planted in the first year after deforestation. Area lationship between land use and clearing size.

SCIENCE
deforested for cropland and mean annual soybean price in the year We determine the fate of large deforestation events (⬎25 ha)
of forest clearing were directly correlated (R2 ⴝ 0.72), suggesting during 2001–2004 in Mato Grosso State to provide satellite-
that deforestation rates could return to higher levels seen in based evidence for the relative contributions of cropland and
2003–2004 with a rebound of crop prices in international markets. pasture to increasing forest loss during this period (Fig. 1). Our
Pasture remains the dominant land use after forest clearing in approach combines satellite-derived deforestation data, vegeta-
Mato Grosso, but the growing importance of larger and faster tion phenology information from the Moderate Resolution
conversion of forest to cropland defines a new paradigm of forest Imaging Spectroradiometer (MODIS; ref. 29), and 2 years of
loss in Amazonia and refutes the claim that agricultural intensifi- field observations to establish the spatial and temporal patterns
cation does not lead to new deforestation. of land use after forest clearing.
Direct measurement of land use after deforestation is aided by
agriculture 兩 carbon 兩 land use change 兩 soybean
MODIS, which began near-daily coverage of the entire Amazon
Basin at 250-m to 1-km resolution in February 2000. The higher

T he ‘‘arc of deforestation’’ along the southern and eastern

extent of the Brazilian Amazon is the most active land-use
frontier in the world in terms of total forest loss (1) and intensity
frequency of observations at moderate resolution improves the
problem of persistent cloud cover in high-resolution satellite
data for Amazonia (30) without sacrificing the ability to char-
of fire activity (2). Historically, the dominant pattern of forest acterize land-cover changes in a fragmented forest landscape
conversion has begun with small-scale exploration for timber or (31, 32). Time series of cloud-free composite images at 16-day
subsistence agriculture, followed by consolidation into large- intervals provide vegetation phenology information to identify
scale cattle ranching operations or abandonment to secondary different land-cover types from the unique patterns of vegeta-
forest (3–5). Recent expansion of large-scale mechanized agri- tion greenness for cropland, pasture, and forest (33).
culture at the forest frontier has introduced a potential new
pathway for forest loss, generating debate over the contribution
of cropland expansion to current deforestation dynamics (5–9). Author contributions: D.C.M., R.S.D., and Y.E.S. designed research; D.C.M., R.S.D., Y.E.S.,
L.O.A., E.A., F.d.B.E.-S., and R.F. performed research; D.C.M. and J.M. analyzed data; and
In the nine states of the Brazilian Legal Amazon, mechanized D.C.M. and R.S.D. wrote the paper.
agriculture increased by 36,000 km2,†† and deforestation totaled The authors declare no conflict of interest.
93,700 km2‡‡ during 2001–2004. Recent gains in the area under
Abbreviations: ha, hectares; MODIS, Moderate Resolution Imaging Spectroradiometer;
cultivation and the productivity of locally adapted crop varieties PRODES, Program for the Estimation of Deforestation in the Brazilian Amazon; INPE,
have made Brazil a leading worldwide producer of grains such as Brazilian National Institute for Space Research.
soybeans; the agribusiness sector now accounts for more than See Commentary on page 14261.
one-third of Brazil’s gross national product (10). ‡To whom correspondence should be addressed. E-mail: rdefries@mail.umd.edu.
The state of Mato Grosso alone accounted for 87% of the ††Total area planted in soybeans, corn, cotton, rice, sugarcane, and sorghum from munic-
increase in cropland area and 40% of new deforestation during ipality data on crop production (IBGE Municipal Agricultural Production, www.sidra.ibge.
this period. Whether cropland expansion contributes directly to gov.br).
deforestation activity or occurs only through the intensified use ‡‡Annual deforestation increment from INPE PRODES, www.obt.inpe.br/prodes.
of previously deforested areas has important consequences for © 2006 by The National Academy of Sciences of the USA

www.pnas.org兾cgi兾doi兾10.1073兾pnas.0606377103 PNAS 兩 September 26, 2006 兩 vol. 103 兩 no. 39 兩 14637–14641

Fig. 1. Tropical deforestation for cropland agriculture in Mato Grosso state (2001–2004) is concentrated along the existing agricultural frontier. (Inset) Location
of the study area subset within Mato Grosso state and the Amazon Basin.

Results deforested area. Deforested area classified as not in production

The expansion of large-scale mechanized crop production con- diminished in each subsequent year after deforestation, as forest
tributed directly to 2001–2004 deforestation in Mato Grosso, clearings were gradually converted to pastures or cropland.
adding to the existing pressure for forest loss from cattle Cropland deforestation in Mato Grosso during 2001–2004 was
ranching (Fig. 2 Upper). We estimate that the area of tropical concentrated within the Xingu river basin and near the existing
forest converted directly to large-scale crop production during centers of crop production (Sinop, Sorriso, Lucas do Rio Verde,
2001–2004 ranged from 785 to 2,150 km2 per year, peaking at and Nova Mutum) along the Cuiabá-Satarém highway (BR-163)
23% of 2003 annual deforestation in large clearings (⬎25 ha). in the central Mato Grosso State (Fig. 1). Deforestation for
Total cropland deforestation during this period exceeded 5,400 cattle pasture predominated in the northern and western por-
km2 of 33,200 km2 total deforestation in large clearings. tions of the state, and deforestation that retained or regrew
A shift in clearing dynamics occurred between 2002 and 2003 forest cover did not show a specific spatial pattern.
deforestation. The fraction of deforested area converted to cattle The spatial distribution of large clearings during 2002–2005
pasture decreased from 78% to 66%, whereas direct transitions to shows the gradual advance of very large deforestation events into
cropland increased from 13% to 23%, and the amount classified as municipalities in northwest Mato Grosso (Fig. 3). Large clear-
not in production (9–10%) and in small clearings (15–17%) re- ings near the existing mechanized agricultural frontier in central
mained nearly constant. Favorable market conditions for agricul- and eastern Mato Grosso were highest in 2003 and 2004. In 2005,
tural exports, especially for soybeans, may have influenced the soybean prices fell by ⬎25%, and municipalities in eastern Mato
patterns in land use after deforestation. The mean annual soybean Grosso showed a decrease in large deforestation events, yet the
price during 2001–2004 was related to the amount of deforestation central agricultural zone continued to exhibit a similar degree of
for cropland in Mato Grosso (R2 ⫽ 0.72). large forest-clearing activity.
For all years, the average clearing for cropland was more than
twice the size of that for pasture (cropland mean ⫽ 333 ha, SD ⫽ Discussion
459 ha; pasture mean ⫽ 143 ha, SD ⫽ 267 ha; P ⬍ 0.0001). Deforestation for large-scale cropland accounted for 17% of forest
Deforestation for cropland accounted for 28% of the clearings loss in large clearings during 2001–2004 in Mato Grosso, signaling
⬎200 ha in 2003 compared with 6% of clearings ⬍200 ha (Fig. a shift from historic uses of cattle ranching and smallholder
2 Lower). Smaller-size classes showed higher proportions of agriculture. Growth in the number of large deforestation events
clearings not in production and slower conversions than larger (⬎25 ha) was responsible for annual increases in deforestation
deforestation-size classes. during the study period, and the relative contribution of cropland
The transition from forest to cropland occurred rapidly. to large deforestation events was directly correlated with the price
Satellite-based vegetation phenology showed evidence of plant- of soybeans in the year of forest clearing. Pasture remains the
ing on ⬎90% of new cropland areas during the year immediately dominant land use after deforestation in Mato Grosso, but our
after forest clearing. Conversions of forest to cattle pasture results show a general trend of increasing cropland deforestation
occurred more slowly than cropland transitions, such that 72– during 2001–2004 and a continuation of the pattern of large forest
86% of pasture clearings were identified in the year after clearings in the central agricultural region in 2005. The rise in
clearing, with the remainder requiring ⱖ2 years to develop a importance of deforestation for cropland signifies a new paradigm
clear grass phenology component over the majority of the of Amazon deforestation defined by larger clearing sizes and faster

14638 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0606377103 Morton et al.

 SEE COMMENTARY
to higher corn, rice, cotton, and sorghum production from Mato
Grosso during 2000–2004, but soybean yield was 10% lower in
2004 than peak production in 2002 based on Brazilian Institute
for Geography and Statistics (IBGE) Municipal Agricultural
Production agricultural census data. Declining soybean yields
may reflect expansion of cropland into less-productive sites,
reductions in soil fertility, or lower harvests because of soybean
rust (Phakopsora pachyrhizi; ref. 37) and other crop pathogens.
Declining yields could be either an incentive or a disincentive to
clear more land, but it is not possible to make this distinction
from our analysis.
Continued expansion of cropland production in Amazonia is
possible. Large areas of the Amazon Basin are projected to have
suitable soils, climate, and topography for large-scale mecha-
nized agriculture (U.S. Department of Agriculture, www.fas.
usda.gov兾current2003.html, January 23, 2003; ref. 38), and many
other regions of the world face a shortage of arable land for
additional cropland expansion (39). Recent and planned future
development of critical infrastructure, such as roadways and
ports, is also intended to support ranching and farming opera-
tions by reducing the cost of transporting agricultural products
to markets (6, 24). The new paradigm of Amazon deforestation
makes farmers and ranchers flexible to future opportunities;
once an area is cleared to bare soil for mechanized agriculture,
it is highly fungible in terms of future land use. The rise and fall
of profits for different crops, beef, plantation timber, and other
resources will therefore determine future land use on both new
deforestation and previously cleared areas.

Implications for Carbon Fluxes from Deforestation. Deforestation

dynamics in Mato Grosso during 2001–2004 highlight the need

SUSTAINABILITY
to understand land use after deforestation, rather than just the

SCIENCE
Fig. 2. Trends in land use after 2001–2004 deforestation events ⬎25 ha in total area of forest loss, to characterize the timing and magnitude
Mato Grosso state, Brazil. (Upper) Summary of conversion dynamics by post- of carbon losses from forest clearing. Carbon losses per area
clearing land cover from satellite-based phenology information in the years deforested for cropland are potentially greater than other types
after forest clearing. A preliminary estimate of 2005 deforestation is shown in
of forest conversion because of the rapid and complete removal
gray (INPE PRODES). Inflation-adjusted prices per 60-kg sack of soybeans for
the same period as the annual deforestation increment (September–August)
of above-ground biomass and woody roots to permit tractor
are plotted on the right-hand axis in Brazilian Reais (R$).§§ (Lower) Fate of planting, with little or no net carbon offset from subsequent crop
2003 deforestation events by clearing size. production. Unlike previous estimates of carbon losses during
conversion of forest to pasture (3, 21, 40, 41), decomposition may
contribute very little to the total carbon lost during the conver-
rates of forest conversion than previous pathways of forest loss for sion of forest to cropland, because trunks, stumps, and woody
pasture or smallholder agriculture. Our findings challenge previous roots are completely combusted in multiple fire events during
assumptions about the fate of carbon after deforestation (3, 21), the clearing process. Stratifying land use after deforestation in
economic drivers of land-use change in Amazonia (4, 28, 34), and terms of clearing size, biomass removal, and duration enables
the possibility for land sparing through crop intensification (7, 35). more accurate estimates of interannual variation in deforestation
carbon fluxes from Amazonia than previously available.
Implications for Future Deforestation Dynamics. Mechanization of
both forest clearing and crop production has encouraged simul- Application to Deforestation Monitoring. Characterizing the fate of
taneous expansion and intensification of land use at the forest individual clearings over time provides input for programs to
frontier. Although the growth of high-yield mechanized agricul- reduce deforestation (5), projections of future deforestation
ture can be a land-sparing option compared with lower-yield (42), and efforts to identify priority areas for conservation (43).
methods (35), our results suggest that intensification of crop A similar approach as presented here that integrates moderate
production in the Brazilian Amazon to meet global demand for and high spatial resolution satellite data was established to
feed crops (8, 9, 36) does not necessarily lead to local land identify deforestation events in the Brazilian Amazon in near-
sparing. Growing production of soybeans and other crops in real time [Brazilian National Institute for Space Research
Amazonia is also a function of expansion into nonforest cover (INPE) Program for the Estimation of Deforestation in the
types (33) and increased yields (ref. 7; Fig. 4). Conversion of Brazilian Amazon (PRODES) and Program for Real-Time
planted pastures and natural grasslands accounted for 36% of Detection of Deforestation (DETER)¶¶ programs]. Linking
new cropland area in Mato Grosso between 2001 and 2004, and vegetation phenology data from MODIS with other types of
an additional 30% of cropland expansion statewide replaced change monitoring, such as logging (19), could be done to
Cerrado savanna兾woodland vegetation (33). Improved yields led characterize the fate of other forest disturbances over time.

§§Datasources: Monthly price paid to soybean producers, Fudaçao Getúlio Vargas Agro- ¶¶INPE Detecção de Desmatamento em Tempo Real, or Program for Real-Time Detection of

analysis; deflator, IBGE Extended National Consumer Price Index. Prices are shown per Deforestation, was started in 2003 to provide regular updates of new deforestation ⬎25
60-kg sack of soybeans to maintain consistency with the common unit of soybean ha in the Brazilian Amazon using data from MODIS sensors and CBERS-2, the Chinese–
production. Brazilian Environmental Satellite. Data can be accessed at www.obt.inpe.br兾deter.

Morton et al. PNAS 兩 September 26, 2006 兩 vol. 103 兩 no. 39 兩 14639
 Fig. 3. Spatial distribution of 2002–2005 deforestation events larger than 20 MODIS 250-m pixels (⬇125 ha) for municipalities in Mato Grosso (32).

Our ability to fully explore the interannual variability in to global market demand for soybeans and other crops have
deforestation dynamics and place recent trends in the context of reduced the remoteness of the forest frontier, and the potential
historic patterns of forest conversion is somewhat limited by the exists for a return to higher deforestation in Mato Grosso as seen
short duration of the MODIS time series. Results showing less in 2003–2004 with a rebound of crop prices. Initiatives such as
regrowth after forest clearing than previous studies (15, 44), the certification schemes for environmental best practices that apply
short interval between forest clearing and production, and market pressure to ranching and soybean production at the forest
limited secondary land-use transitions after forest conversion frontier (9) would augment existing efforts to reduce illegal
(forest–pasture–cropland) merit further investigation with the deforestation through satellite-monitoring programs. Increasing
growing MODIS data record. The approach is potentially ap- incentives for intensified use of unproductive pastures or other
plicable in other areas undergoing conversion to mechanized existing cleared lands will also be essential to balance economic
agriculture but could be limited by absence of high-resolution benefits from increasing crop production with ecosystem ser-
deforestation maps, clearing sizes too small for isolation of vices from intact forest and Cerrado habitat.
vegetation phenology information with MODIS 250-m resolu- Data and Methods
tion data, or land uses after forest clearing without distinct
Remote-Sensing Analysis. We combine field observations with
phenologies.
satellite-based data on annual deforestation and vegetation
In summary, our findings refute the claim that new crop
phenology to classify the fate of new forest clearings ⬎25 ha in
production in Amazonia is occurring only through intensified
Mato Grosso State, Brazil. Field data on the location and
use of lands previously cleared for cattle ranching rather than condition of deforested areas, pastures, and cropland were
adding a new pressure for forest loss (45, 46). The large clearings collected during June 2004, March 2005, and July 2005, and
and complete removal of above-ground biomass indicate per scaled from Global Positioning System point observations to
area carbon emissions to the atmosphere greater than previous polygon training data by digitizing feature boundaries on near-
clearing for cattle ranching and fewer forest fragments on the coincident Landsat Thematic Mapper (TM) data. Landsat TM
landscape as habitat and suggest rapid loss of forest as infra- data were provided by INPE before each field campaign and
structure develops for large-scale agriculture. Growing linkages georeferenced to existing Landsat Enhanced Thematic Mapper
Plus (ETM⫹) data provided by the Global Land Cover Facility
with a spatial error of less than one pixel (30 m). We used
PRODES digital results of the annual deforestation increment
mapped using Landsat TM data from approximately August of
2001–2004 for the state of Mato Grosso to identify the location
and size of new clearings and summarize total deforested area,
limiting our analysis to new clearings ⬎25 ha based on the
moderate resolution (250 m) of the MODIS sensor (29, 31, 32).
Estimates of 2005 deforestation in Mato Grosso State were
generated from MODIS red reflectance data contained in the
MODIS兾Terra Vegetation Indices 16-day L3 Global product at
250-m resolution (MOD13Q1, version 4; ref. 47) and forest
information from the PRODES 2004 deforestation analysis
following methods outlined in ref. 33.
Before generating phenology metrics for land-cover classifi-
cation, we implemented a two-stage method to remove cloud
contamination in annual time series of normalized difference
vegetation index (NDVI) and enhanced vegetation index (EVI)
data from the MOD13Q1 product from 2000–2005 for three 10°
⫻ 10° spatial tiles (h12v10, h12v09, and h13v10). Clouds, cloud
shadows, high aerosols, or other data artifacts were identified by
Fig. 4. Relationship between cropland expansion and deforestation in Mato using the Quality Assessment layer available with the MODIS
Grosso, Brazil, during 2001–2004. Estimates of forest conversion directly to data product and replaced with a predicted value by fitting the
cropland range from 4,670 (33) to 5,463 km2 (this study). Expansion of large- remaining high-quality data in each pixel’s time series with a
scale mechanized agriculture was estimated from annual land cover maps of
cubic spline function. Second, the resulting annual time series
Mato Grosso derived from MODIS-based phenology information; only tran-
sitions from forest, Cerrado, or pasture兾grasslands to double-cropping sys-
were fit with zero to third-order harmonic functions to identify
tems are included in this estimate (33). Estimated cropland expansion from and eliminate any clouds not captured by the image-quality data
agricultural census data of total planted area is nearly two times the area layer (48). We derived 36 metrics from the cloud-free time series:
derived from satellite data, because individual fields are counted separately NDVI and EVI minimum, maximum, mean, median, amplitude,
for each crop rotation in the agricultural census. and standard deviation for annual (yearn⫺1: day 273⫺yearn: day

14640 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0606377103 Morton et al.

 SEE COMMENTARY
288), wet season (yearn⫺1: day 273⫺yearn: day 112), and dry results for each year after deforestation to eliminate spurious
season (yearn: day 113–273) time periods. Harmonic equations pasture–cropland transitions in the first 2 years after deforestation
provided three additional phenology amplitude and phase met- (Table 2, which is published as supporting information on the PNAS
rics for the classification process. web site). The largest contribution to estimated deforestation for
A decision-tree classifier was developed with training data cropland from the trajectory corrections was a single year of pasture
from field observations in July 2004 and MODIS time series classification followed by 1–3 years of cropland classification. Use
metrics from 2003-273 to 2004-288 (Fig. 5, which is published as of deforested lands as pasture for a single year is unlikely, given the
supporting information on the PNAS web site). Validation of the high cost of fencing material (49). Trajectory-based modifications
classification was done by using field data from March and July accounted for 20–27% of the total deforestation for cropland in
2005 and time series metrics from 2004-273 to 2005-288 (Table 2001–2003. We report statistics and trends based on the corrected
1, which is published as supporting information on the PNAS trajectories for deforestation in 2001–2003 and unadjusted results
web site). Following accurate classification of 2005 validation for 2004, because only one postclearing MODIS classification
data (overall accuracy, 89%), the same classification rules were (2005) was available.
applied to each year of MODIS metrics. We define the fate of
deforested areas as cropland, pasture, or not yet in production Soybean Price Data. To estimate the influence of crop prices on
using the majority land-cover class within each deforestation deforestation, we adjusted the Fudaçao Getúlio Vargas Agroanaly-
polygon based on the improvement in classification accuracy for sis monthly price paid to soybean producers for inflation using a
cropland and cattle pasture with this method. Deforestation standard consumer price index (46), IBGE Extended National
classified as forest and degraded forest was combined into a Consumer Price Index. Average annual prices were calculated from
single class, not in production, encompassing damaged forests monthly data for the same period as PRODES deforestation
that were never fully cleared (e.g., logged or burned forest), edge calculations, September through August of each year, and August
effects from adjacent forest cover, and regrowth. 2005 was the reference month for inflation adjustment.
Interpretation of Remote-Sensing Results. The classification was We thank Jan Dempewolf, Carlos Klink, Greg Asner, and three review-
highly accurate in separating double-cropping systems and pasture. ers for helpful comments on a previous version of this manuscript. This
However, the annual phenological patterns of fallow agricultural study was supported by grants from the National Aeronautics and Space
cycles or single-crop rotations are similar to a pasture phenology Administration Large Scale Biosphere–Atmosphere Experiment in the
and could be misclassified as such. To correct for these land-use Amazon (Projects LC-22 and LC-23), Land Use Land Cover Change,
patterns, we established a land-use trajectory from classification and Terrestrial Ecology Programs.

1. Food and Agriculture Organization of the United Nations (2006) Global Forest 29. Justice CO, Vermote E, Townshend J, DeFries RS, Roy DP, Hall DK,

SUSTAINABILITY
Resources Assessment 2005: Progress Towards Sustainable Forest Managment Salomonson VV, Privette JL, Riggs G, Strahler A, et al. (1998) IEEE Trans

SCIENCE
(Food and Agriculture Organization, United Nations, Rome, Italy). Geosci Remote Sens 36:1228–1249.
2. Giglio L, Csiszar I, Justice CO (2006) J Geophys Res 111:G02016. 30. Asner GP (2001) Int J Remote Sens 22:3855–3862.
3. Houghton RA, Skole DL, Nobre CA, Hackler JL, Lawrence KT, Chomen- 31. Anderson LO, Shimabukuro YE, DeFries RS, Morton DC (2005) IEEE Geosci
towski WH (2000) Nature 403:301–304. Remote Sens Lett 2:315–318.
4. Lambin EF, Geist HJ (2003) Environment 45:22–36. 32. Morton DC, DeFries RS, Shimabukuro YE, Anderson LO, del bon Espı́rito-
5. Fearnside PM (2005) Conserv Biol 19:680–688. Santo F, Hansen MC, Carroll M (2005) Earth Interact 9:1–22.
6. Fearnside PM (2001) Environ Conserv 28:23–38. 33. Morton DC, DeFries RS, Shimabukuro YE (2006) in Cerrado Land Use and
7. Brown JC, Koeppe M, Coles B, Price KP (2005) Ambio 34:462–469. Conservation: Assessing Trade-Offs Between Human and Ecological Needs, eds
8. Naylor R, Steinfeld H, Falcon W, Galloway J, Smil V, Bradford E, Alder J, Klink C, DeFries RS, Cavalcanti R (Conservation International, Washington,
Mooney H (2005) Science 310:1621–1622. DC), in press.
9. Nepstad DC, Stickler CM, Almeida OT (2006) Conserv Biol, in press. 34. Kaimowitz D, Mertens B, Wunder S, Pacheco P (2004) Hamburger Connection
10. Empresa Brasileira de Pesquisa Agropecuária (2004) Criação de Empregos Pelo Fuels Amazon Destruction: Cattle Ranching and Deforestation in Brazil’s Amazon
Complexo Agroindustrial da Soja (Ministry of Agriculture, Brasilia, Brazil). (Center for International Forestry Research, Jakarta, Indonesia).
11. Mooney H, Cropper A, Reid W (2005) Nature 434:561–562. 35. Green RE, Cornell SJ, Scharlemann JPW, Balmford A (2005) Science 307:550–555.
12. Ministério da Ciência e Tecnologia (2004) Brazil’s Initial National Communi- 36. Kaimowitz D, Smith J (2001) in Agricultural Technologies and Tropical Defor-
cation to the United Nations Framework Convention on Climate Change (Min- estation, eds Angelsen A, Kaimowitz D (CABI Publishing, Wallingford, UK),
istry of Science and Technology, Brasilia, Brazil). pp 195–212.
13. Santilli M, Moutinho P, Schwartzman S, Nepstad DC, Curran LM, Nobre CA 37. Yorinori JT, Paiva WM, Frederick RD, Costamilan LM, Bertagnolli PF,
(2005) Clim Change 71:267–276. Hartman GE, Godoy CV, Nunes J, Jr (2005) Plant Dis 89:675–677.
14. DeFries RS, Houghton RA, Hansen MC, Field CB, Skole DL, Townshend J 38. Jasinski EW, Morton DC, DeFries RS, Shimabukuro YE, Anderson LO,
(2002) Proc Natl Acad Sci USA 99:14256–14261. Hansen MC (2005) Earth Interact 9:1–18.
15. Skole DL, Tucker C (1993) Science 260:1905–1910. 39. Food and Agriculture Organization (2003) World Agriculture: Towards 2015兾
16. Achard F, Eva HD, Stibig, H-J, Mayaux P, Gellego J, Richards T, Malingreau, 2030 (Earthscan Publications, London, UK).
J-P (2002) Science 297:999–1002. 40. Fearnside PM, Leal N, Jr, & Fernandes FM (1993) J Geophys Res 98:16733–16743.
17. Nepstad DC, Verı́ssimo A, Alencar A, Nobre CA, Lima E, Lefebre P, Schlesinger 41. Carvalho JA, Higuchi N, Araújo TM, Santos JC (1998) J Geophys Res
P, Potter C, Moutinho P, Mendoza E, et al. (1999) Nature 398:505–508. 103:13195–13199.
18. Cochrane MA (2003) Nature 421:913–919. 42. Soares-Filho BS, Nepstad DC, Curran LM, Cerqueira GC, Garcia RA, Ramos
19. Asner GP, Knapp DE, Broadbent EN, Oliveira PJC, Keller M, Silva JN (2005) CA, Voll E, McDonald A, Lefebre P, Schlesinger P (2006) Nature 440:520–523.
Science 310:480–482. 43. Nepstad DC, Schwartzman S, Bamberger B, Santilli M, Ray D, Schlesinger P,
20. Houghton RA, Lawrence KT, Hackler JL, Brown S (2001) Glob Change Biol Lefebre P, Alencar A, Prins E, Fiske G, et al. (2006) Conserv Biol 20:65–73.
7:731–746. 44. Roberts DA, Numata I, Holmes K, Batista G, Krug T, Monteiro A, Powell B,
21. Hirsch AI, Little WS, Houghton RA, Scott NA, White JD (2004) Glob Change Chadwick OA (2002) J Geophys Res 107, 40:41–48.
Biol 10:908–924. 45. Mueller CC (2003) in Série Textos Para Discussão, Working Paper, eds Bugarin
22. Hecht SB (1993) Bioscience 43:687–695. M, Mueller CC (University of Brasilia, Department of Economics, Brasilia,
23. Pfaff ASP (1999) J Env Econ Mgmt 37:26–43. Brazil), no 306.
24. Nepstad DC, Carvalho GO, Barros AC, Alencar A, Capobianco JP, Bishop J, 46. Brandão ASP, de Rezende GC, Marques RW (2005) in Texto Para Discussão
Moutinho P, Lefebre P, Silva UL, Prins E (2001) For Ecol Manag 154:395–407. (Instituto de Pesquisa Econômica Aplicada, Rio de Janeiro, Brazil), no 1103.
25. Geist HJ, Lambin EF (2002) Bioscience 52:143–150. 47. Huete AR, Didan K, Miura T, Rodriguez EP, Gao X, Ferreira LG (2002)
26. Cardille JA, Foley JA (2003) Remote Sens Environ 87:551–562. Remote Sens Environ 83:195–213.
27. Chomitz KM, Thomas TS (2003) Am J Agric Econ 85:1016–1028. 48. Roerink GJ, Meneti M, Verhoef W (2000) Int J Remote Sens 21:1911–1917.
28. Margulis S (2004) World Bank Working Paper No. 22: Causes of Deforestation 49. de Mendonça MJC, Vera Diaz MdC, Nepstad DC, Motta RS, Alencar A,
of the Brazilian Amazon (World Bank, Washington, DC). Gomes JC, Ortiz RA (2004) Ecol Econ 49:89–105.

Morton et al. PNAS 兩 September 26, 2006 兩 vol. 103 兩 no. 39 兩 14641