Inhibition of Amazon Deforestation and Fire by Parks and Indigenous Lands

Abstract

Conservation scientists generally agree that many types of protected areas will be needed to protect tropical forests. But little is known of the comparative performance of inhabited and uninhabited reserves in slowing the most extreme form of forest disturbance: conversion to agriculture. We used satellite-based maps of land cover and fire occurrence in the Brazilian Amazon to compare the performance of large (> 10,000 ha) uninhabited (parks) and inhabited (indigenous lands, extractive reserves, and national forests) reserves. Reserves significantly reduced both deforestation and fire. Deforestation was 1.7 (extractive reserves) to 20 (parks) times higher along the outside versus the inside of the reserve perimeters and fire occurrence was 4 (indigenous lands) to 9 (national forests) times higher. No strong difference in the inhibition of deforestation (p = 0. 11) or fire (p = 0.34) was found between parks and indigenous lands. However, uninhabited reserves tended to be located away from areas of high deforestation and burning rates. In contrast, indigenous lands were often created in response to frontier expansion, and many prevented deforestation completely despite high rates of deforestation along their boundaries. The inhibitory effect of indigenous lands on deforestation was strong after centuries of contact with the national society and was not correlated with indigenous population density. Indigenous lands occupy one-fifth of the Brazilian Amazon-five times the area under protection in parks--and are currently the most important barrier to Amazon deforestation. As the protected-area network expands from 36% to 41% of the Brazilian Amazon over the coming years, the greatest challenge will be successful reserve implementation in high-risk areas of frontier expansion as indigenous lands are strengthened. This success will depend on a broad base of political support.

Full text

Inhibition of Amazon Deforestation and Fire
by Parks and Indigenous Lands
D. NEPSTAD,∗†‡§§ S. SCHWARTZMAN,§B. BAMBERGER,∗†† M. SANTILLI,∗∗ D. RAY,∗
P. SCHLESINGER,∗P. LEFEBVRE,∗A. ALENCAR,† E. PRINZ,‡‡ GREG FISKE,∗AND ALICIA ROLLA∗∗
∗The Woods Hole Research Center, P.O. Box 296, 13 Church Street, Woods Hole, MA 02543, U.S.A.
†Instituto de Pesquisa Ambiental da Amazonia, Av. Nazar´e 669, Centro. 66035-170 Bel´em, PA, Brazil
‡Universidade Federal do Par´a, Nucleo de Altos Estudos Amazˆonicos, Av. Augusto Correa No. 1,
Campus da Universidade—Guam´a, 66.059, Bel´em, Par´a, Brazil
§Environmental Defense, 1875 Connecticut Avenue, NW, Suite 600, Washington, D.C. 20009, U.S.A.
††Yale School of Forestry & Environmental Studies, 205 Prospect Street, New Haven, CT 06511, U.S.A.
‡‡Cooperative Institute for Meteorological Satellite Studies, 1225 W. Dayton Street, Madison, WI 53706, U.S.A.
∗∗Instituto Socioambiental, Av. Higien´opolis, 901, 01238-001 S˜ao Paulo, SP, Brazil
Abstract: Conservation scientists generally agree that many types of protected areas will be needed to protect
tropical forests. But little is known of the comparative performance of inhabited and uninhabited reserves in
slowing the most extreme form of forest disturbance: conversion to agriculture. We used satellite-based maps of
land cover and fire occurrence in the Brazilian Amazon to compare the performance of large (>10,000 ha) un-
inhabited (parks) and inhabited (indigenous lands, extractive reserves, and national forests) reserves. Reserves
significantly reduced both deforestation and fire. Deforestation was 1.7 (extractive reserves) to 20 (parks) times
higher along the outside versus the inside of the reserve perimeters and fire occurrence was 4 (indigenous lands)
to 9 (national forests) times higher. No strong difference in the inhibition of deforestation (p=0.11) or fire (p=
0.34) was found between parks and indigenous lands. However, uninhabited reserves tended to be located away
from areas of high deforestation and burning rates. In contrast, indigenous lands were often created in re-
sponse to frontier expansion, and many prevented deforestation completely despite high rates of deforestation
along their boundaries. The inhibitory effect of indigenous lands on deforestation was strong after centuries of
contact with the national society and was not correlated with indigenous population density. Indigenous lands
occupy one-fifth of the Brazilian Amazon—five times the area under protection in parks—and are currently the
most important barrier to Amazon deforestation. As the protected-area network expands from 36% to 41% of
the Brazilian Amazon over the coming years, the greatest challenge will be successful reserve implementation
in high-risk areas of frontier expansion as indigenous lands are strengthened. This success will depend on a
broad base of political support.
Key Words: Brazil, protected areas, tropical forests
Inhibici´on de Deforestaci´on e Incendios por Parques y Terrenos Ind´ıgenas en la Amazonia
Resumen: Los cient´
ıficos de la conservaci´
on generalmente est´
an de acuerdo en que se requerir´
an muchos
tipos de ´
areas protegidas para proteger a los bosques tropicales. Pero se conoce poco del funcionamiento
comparativo de reservas habitadas y deshabitadas en la reducci´
on de la forma m´
as extrema de perturbaci´
on
de bosques: conversi´
on a agricultura. Utilizamos mapas, basados en sat´
elites, de cobertura e incidencia de
incendios en la Amazon´
ıa Brasile˜
na para comparar el funcionamiento de reservas deshabitadas (parques)
grandes (>10,000 ha) y habitadas (terrenos ind´
ıgenas, reservas extractivas y bosques nacionales). Las reservas
redujeron tanto la deforestaci´
on como los incendios significativamente. La deforestaci´
on fue 1.7 (reservas
extractivas) a 20 (parques) veces mayor a lo largo del exterior versus el interior de los per´
ımetros de las
§§email dnepstad@whrc.org
Paper submitted February 26, 2004; revised manuscript accepted March 29, 2005
65
Conservation Biology Volume 20, No. 1, 65–73
C
2006 Society for Conservation Biology
DOI: 10.1111/j.1523-1739.2006.00351.x

66 Inhibition of Amazon Deforestation and Fire Nepstad et al.
reservas y la ocurrencia de incendios fue 4 (terrenos ind´
ıgenas) a 9 (bosques nacionales) veces mayor. No
encontramos diferencias significativas en la inhibici´
on de la deforestaci´
on (p=0.11) o incendios (p=0.34)
entre parques y terrenos ind´
ıgenas. Sin embargo, las reservas deshabitadas tendieron a estar localizadas lejos
de ´
areas con altas tasas de deforestaci´
on y ocurrencia de incendios. En contraste, los terrenos ind´
ıgenas a
menudo fueron creados en respuesta a la expansi´
on de la frontera, y muchos previnieron la deforestaci´
on
completamente a pesar de las altas tasas de deforestaci´
on a lo largo de sus l´
ımites. El efecto inhibidor de los
terrenos ind´
ıgenas sobre la deforestaci´
on fue notable despu´
es de siglos de contacto con la sociedad nacional y
no se correlacion´
o con la densidad de la poblaci´
on ind´
ıgena. Los terrenos ind´
ıgenas ocupan una quinta parte
de la Amazon´
ıa Brasile˜
na—cinco veces el ´
area bajo protecci´
on en parques—y actualmente son la barrera
m´
as importante para la deforestaci´
on de la Amazon´
ıa. A medida que la red de ´
areas protegidas se expanda
de 36 a 41% de la Amazon´
ıa Brasile˜
na en los pr´
oximos a˜
nos, el mayor reto ser´
a la implementaci´
on exitosa
de reservas en ´
areas con alto riesgo de expansi´
on de la frontera agr´
ıcola al mismo tiempo que se refuercen
los terrenos ind´
ıgenas. Este ´
exito depender´
a de una amplia base de soporte pol´
ıtico.
Palabras Clave: ´areas protegidas, Brasil, bosques tropicales
Introduction
There is growing agreement among conservation scien-
tists that many types of protected areas, including those
with resident human populations, are needed for an effec-
tive global strategy to preserve tropical forests. A recent
synthesis concludes that protection of a substantial pro-
portion of the world’s remaining biodiversity is feasible in
part because approximately 2 million km2of tropical for-
est are already protected for indigenous peoples and bio-
diversity (Pimm et al., 2001). Officially recognized indige-
nous lands of the Brazilian Amazon alone comprise half
of this total. Most conservation literature and policy rec-
ommendations are still directed at uninhabited protected
areas. However, these areas differ significantly from in-
habited protected areas such as indigenous lands and ex-
tractive reserves in the processes by which they are cre-
ated, in their long-term management needs, and, hence,
in their role within conservation strategies. One rationale
for this emphasis on uninhabited protected areas is that
the conservation value of indigenous lands is lower than
that of parks because indigenous people ultimately adopt
the cultural values, technology, and patterns of resource
exploitation of their nonindigenous neighbors, a trend
that is exacerbated by population growth within indige-
nous lands (Terborgh 1999, 2000; Redford & Sanderson
2000; Terborgh & van Shaik 2002).
The refinement of biodiversity conservation strategies
in the tropics is hampered by a dearth of comparisons
of the performance of inhabited versus uninhabited pro-
tected areas in slowing the most extreme form of human
disturbance: forest conversion to agriculture. Several re-
searchers (e.g., Redford 1992; Peres 2000a, 2000b; Robin-
son & Bennett 2000) have examined the effects of rural
people on wild game populations and others (Terborgh
1999; S´a & Ferreira 2000; Bruner et al. 2001; Terborgh &
van Schaik 2002) have analyzed the performance of unin-
habited parks in protecting biological diversity. Ferreira
et al. (2005) recently analyzed deforestation rates within
uninhabited reserves, indigenous lands, and unprotected
lands for three Amazon states but did not conduct a rig-
orous comparison of deforestation inhibition by the two
types of reserves. The assumption that the conservation
value of uninhabited parks is higher than reserves with
human residents remains untested.
We report on the results of a satellite-based comparison
of the inhibitory effects of protected areas that prohibit
human habitation (parks, biological reserves, ecological
stations) and those that permit habitation (indigenous
lands, extractive reserves, and national forests) on defor-
estation and fire within the Brazilian Amazon. Logging
and hunting damage forests, but were omitted from this
analysis because they are difficult to quantify (Nepstad et
al. 1999).
Methods
The quantification of reserve performance in slowing de-
forestation is best measured against a baseline that de-
scribes the trajectory of deforestation in the absence of
the reserve. This trajectory is influenced by the suitabil-
ity of the land within the reserve for agriculture, log-
ging, and other economic activities, by market trends for
agricultural and forest products, by investments in trans-
portation and energy infrastructure, and by agrarian re-
form. A reserve therefore inhibits deforestation only if
it (1) slows the expansion of economic activities (i.e.,
protects natural resources that would otherwise be ex-
ploited), (2) prevents or mitigates the effects of invest-
ments in roads and other infrastructure that cause direct
environmental damages and/or that indirectly foster nat-
ural resource exploitation, and (3) prevents agricultural
settlements—either planned or spontaneous—motivated
by agrarian reform pressures. (The third condition is not
redundant of the first because agricultural settlements are
often planned in places that are not suitable for agricul-
ture.) Within this context, reserves that are far from the
expanding agricultural and logging frontier and are not
Conservation Biology
Volume 20, No. 1, February 2006

Nepstad et al. Inhibition of Amazon Deforestation and Fire 67
slated for infrastructural investments or agricultural set-
tlement have a negligible short-term effect on deforesta-
tion, but may have a very important inhibitory effect as
the frontier grows nearer.
To measure protected-area performance one must dis-
tinguish between local and regional effects. To what ex-
tent is the inhibition of deforestation within a reserve
counterbalanced by an increase in deforestation else-
where? In general, this “leakage” of the inhibitory effect
of reserves should be greatest in young, expanding agri-
cultural frontiers, where land tends to be cheap and abun-
dant but diminishes over time as land suitable for agricul-
ture becomes scarce. We did not address this aspect of
reserve performance.
We provide an initial comparative assessment of reserve
performance in the Brazilian Amazon by using deforesta-
tion and fire occurrence along the reserve perimeter as
a proxy for the threat of imminent deforestation. We as-
sume that reserves with rapid rates of forest conversion
to agriculture and/or high incidence of fire along the out-
side of their perimeters are more likely to suffer forest
clearcutting than reserves with low rates of forest con-
version along their perimeters. Therefore, the ratio of de-
forestation and burning in buffer zones outside versus in-
side the reserve boundary provides a measure of reserve
performance that normalizes the threat. When this ratio
is greater than unity, the reserve has deflected forest con-
version from the baseline trajectory. This approach over-
estimates the inhibition of deforestation in cases where
the reserves were established close to existing roads or
to the boundaries of existing colonization projects.
Interview-based analyses (e.g., Bruner et al. 2001) pro-
vide a qualitative indication of reserve performance, but
are vulnerable to the biases of informants who have a
vested interest in this performance. Park managers and
conservation NGO personnel may have the best informa-
tion about the status of reserves, but they also may have
motives for overstating the success of the reserves. We
therefore measured reserve performance with maps of
land cover developed from satellite images (INPE 2004)
that identify those areas deforested during the period
1997–2000 and maps of active fires derived from a geo-
stationary weather satellite in 1998 (GOES-8, Menzel &
Purdom 1994; E.M. Prins, W. P. Menzel, and J. M. Feltz.
1998. Characterizing spatial and temporal distributions
of biomass burning using multi-spectral geostationary
satellite data. Pages 94–97 in Proceedings of the Ninth
Conference on Satellite Meteorology and Oceanography).
Deforestation is indicated by satellite detection of forest
replacement by cattle pastures and agricultural systems,
and by detection of the fires that are used as part of the
forest-clearing process and in the maintenance of cattle
pastures (Nepstad et al. 2001). Of the four major types of
land-use fire in the Amazon—including fire used to burn
felled forest, fire used to improve forage quality in cat-
tle pastures, accidental cattle pasture fire, and fires that
burn standing forests—only the first three are registered
as “hot pixels” by the thermal channels of satellites (Nep-
stad et al. 2001). Our analysis did not capture understory
fires in standing forests.
Maps of reserves for the Brazilian Amazon were ac-
quired from the Instituto Socioambiental (Capobianco
et al. 2001; Fig. 1). We refer here to those federal land
designations that prohibit resource exploitation by peo-
ple (national parks, biological reserves, ecological re-
search stations) as “parks.” Indigenous lands, extractive
reserves, and national forests permit human residents and
subsistence agricultural activities, but partially restrict
deforestation. Deforestation in extractive reserves, for ex-
ample, is restricted to 10% of the forest area. Indigenous
peoples have permanent use rights to their land, but reg-
ulations on resource use are ambiguous. In contrast, pri-
vate landholders can clear up to 20% of the forests on
their properties, although enforcement of this legislation
is largely ineffective.
Our metric for reserve performance—the ratio of defor-
estation and fire occurrence rates along the outside versus
inside of the reserve perimeter—is sensitive to coregistra-
tion errors between land cover maps of different years,
and between land cover and park boundaries. We there-
fore excluded from the deforestation analysis those re-
serves that were small (<10,000 ha) and therefore had
proportionally high image coregistration errors. We also
inspected the superimposed land cover maps and reserve
locations for each of the reserves and excluded from the
analysis those for which coregistration errors could be de-
tected visually. Reserves established after 1997, reserves
established by state governments, images with clouds or
missing data accounting for >20% of the inside or outside
reserve buffer, reserves with <10% forest cover in 2000,
and reserves for which satellite data were not available
were also excluded from the analysis. Based on these se-
lection criteria, we were able to quantify deforestation
rates of 15 parks, 121 indigenous lands, 10 extractive re-
serves, and 18 national forests, representing 40%, 35%,
58%, and 39%, respectively, of the total area of each re-
serve type under federal jurisdiction.
Fire inhibition was quantified for reserves that were at
least 50,000 ha and that had <20% classified as cerrado—
Brazil’s savanna woodland vegetation that burns naturally.
The spatial distribution of fires was compiled for 1998, a
severely dry year. This sample consisted of 11 parks, 87
indigenous lands, 4 extractive reserves, and 12 national
forests, representing 35%, 51%, 74%, and 34% of the area
of these federal reserves, respectively.
We measured deforestation inhibition by comparing av-
erage annual deforestation rates from 1997 to 2000 within
10-km-wide strips of land located along the inside and out-
side of the reserve perimeter. Deforestation rates were
calculated as
r=(F1997/F2000 )1/t−1,(1)
Conservation Biology
Volume 20, No. 1, February 2006

68 Inhibition of Amazon Deforestation and Fire Nepstad et al.
Figure 1. Reserves and
human-caused disturbance in the
Brazilian Amazon. (a) Reserves,
highways (paved in gray,
unpaved in dashed black and
white lines), and deforestation
(as of 2000) of the Brazilian
Amazon (letters indicate
indigenous lands referred to in
the text: Go, Governador; IC,
Igarap´
e do Caucho; Ka, Kayap´
o;
KK, Katukina/Kaxinaw´
a; MM,
M˜
ae Maria; Ma, Maraiwatsede;
and Xi, Xing´
u Indigenous Park.
(b) Reserves (polygons) and
active fires (hot pixels) registered
by the GOES satellite for 1998.
where ris the annual deforestation rate (%), F1997 and
F2000 are forest cover in 1997 and 2000, respectively, and
tis time in years. The influence of reserves on fire occur-
rence was measured by comparing fire density (number
of fires per square kilometer in 1998) within 20-km-wide
strips along the inside and outside of the reserve perime-
ter. We used larger buffer areas for the fire data because
of the coarser spatial resolution of these data—the GOES
pixels are 4 km wide, whereas the Landsat thematic map-
per pixels are 30 m wide. Spatial accuracy of GOES fire
detection is within one pixel (Menzel & Purdom 1994).
Only those hot pixels detected at mid-day and classified
as having a high probability of being associated with fire
were included in the analysis (GOES-8, Menzel & Purdom
1994; E.M. Prins, W. P. Menzel, and J. M. Feltz. 1998). For
reserves <200,000 ha in size, fire density of the entire
reserve was compared with that along the outside of the
perimeter.
Nonparametric statistics were used to make compar-
isons among reserve types due to the highly skewed distri-
bution of the deforestation rate data. First, we compared
the inhibition of deforestation and fire by the reserves
(outside buffer vs. inside buffer values) within each re-
serve type with Wilcoxon signed ranks test for dependent
Conservation Biology
Volume 20, No. 1, February 2006

Nepstad et al. Inhibition of Amazon Deforestation and Fire 69
samples. We used Kruskal–Wallis one-way analysis of vari-
ance for independent samples to compare inhibition of
deforestation and fire across reserve types. The depen-
dent variable for this analysis was the ratio between inside
and outside buffer values. We used the same procedure to
test for differences between parks and inhabited reserves
(for two samples the test reduces to the Mann–Whitney
Utest).
Finally, we used rank correlation analysis to test for a
relationship between deforestation in indigenous lands
as a function of time since first contact with nonindige-
nous groups and reserve population density. The ratio
between the annual deforestation rates, inside and out-
side the reserve, determined between 1997 and 2000,
provided the dependent variable for these analyses. All
significance tests were carried out at the α=0.05 level
in SyStat software (SPSS 2000).
Results
On average, deforestation from 1997 to 2000 was 1.7 (ex-
tractive reserves) to 20 (parks) times higher along the
outside of the reserves than along the inside (Figs. 1a,
2a, & 3a). This inhibitory effect was significant for all re-
serve types except extractive reserves, for which the sam-
ple size was smallest, as follows: indigenous lands (Z=
8.053; p<0.000), parks (Z=3.076; p=0.002), extrac-
tive reserves (Z=0.771; p=0.441), and national forests
(Z=3.061; p=0.002). Eighty-five percent of indigenous
lands (97 of 114, with 7 ties) had higher deforestation
rates in the outer than inner buffer; this figure was 92%
for parks (12 of 13, with 2 ties). Differences among the
four reserve types were indicated (K–W=11.390; p=
0.010), perhaps because of the large difference between
parks and extractive reserves in the ratio of deforestation
in the outer and inner buffers. Inhibition of deforestation
was highest for the parks (20-fold), intermediate for the
national forests (9.5-fold) and indigenous lands (8.2-fold),
and lowest for extractive reserves (1.7-fold) (Fig. 2a; 3a).
A similar inhibitory effect was found for fire (Figs. 1b;
2b; 3b). The average density of fires was 3.7 to 9.4 times
higher along the outside of the reserves than along the in-
side (Figs. 2b & 3b). This effect was highly significant for
indigenous lands (Z=6.84; p<0.000), parks (Z=2.94;
p=0.002), and national forests (Z=2.76; p=0.003).
For fires we also detected a marginally significant effect
for the small number of extractive reserves (n=4) in-
cluded in the sample (Z=1.604; p=0.055). Each of these
reserve types exerted a similar degree of control over fire
occurrence (K–W=3.253; p=0.354, Figs. 2b & 3b).
Indigenous lands strongly inhibited deforestation in the
active agricultural frontier. Thirty-three of 38 indigenous
lands with high annual deforestation rates (>1.5%/year)
along the outside of their perimeters had inner deforesta-
Figure2. Inhibition of (a) deforestation and (b) fire
by individual reserves (inside and outside is inside
and outside reserves). Boundary effectiveness is
illustrated relative to the 1:1 line, which represents the
null effect. Reserve codes defined in Fig. 1 legend.
tion rates of 0.75% or lower (Fig. 2a). When expressed
on an area basis, 31% of indigenous lands, representing
39% of the total land area for this reserve type in our
sample, were exposed to this level of outside pressure.
Fewer parks have been established within the active agri-
cultural frontiers of eastern and southern Amazonia (Fig.
1a), in part because of the historical tendency of park
planners to avoid the conflicts and conservation threats
Conservation Biology
Volume 20, No. 1, February 2006

70 Inhibition of Amazon Deforestation and Fire Nepstad et al.
Figure3. Reserve performance in slowing Amazon
deforestation and fire. (a) Average annual
deforestation rates (Eq. 1) from 1997 to 2000 within
10-km strips of land along the inside and outside of
each reserve boundary of 121 indigenous lands, 15
parks, 10 extractive reserves, and 18 national forests.
(b) Cumulative fire density for 1998 within 20-km
strips of land along the inside and outside of each
reserve boundary (87 indigenous lands, 11 parks, 4
extractive reserves, 12 national forests). Fire data were
restricted to one fire/day/16-km2pixel.
associated with expanding frontier regions (Peres & Ter-
borgh 1995). Only four of the 15 parks in our sample,
representing 21% of the land area, were exposed to sim-
ilar outside deforestation pressure (>1.5% deforestation
per year in the outer buffer). The parks and indigenous
lands included in this analysis appeared similar in their
capacity to inhibit deforestation (U=259.5; p=0.113).
The proximity of indigenous lands to the active fron-
tier is also reflected in the fire data (Figs. 1b, 2b, & 3b).
The average density of fires was nearly two times greater
along the outside of indigenous lands than it was along
park perimeters (Fig. 3b). Through their location in lower
risk regions of the Amazon, the park network has had a
proportionally smaller effect on frontier expansion. In-
digenous groups, in contrast, often live in the path of
expanding frontiers, and fight to win legal recognition of
their land rights while defending their forests from clear-
ing by outsiders.
Discussion
The high variability of reserve performance can be traced
to individual reserve histories. High rates of deforestation
in indigenous lands were generally associated with ex-
ploitation or invasions from nonindigenous populations
that had occurred prior to reserve demarcation. The in-
digenous lands with the highest interior deforestation
in our sample (Igarape do Caucho, Acre State, Katuk-
ina/Kaxinawa, Acre State, Governador, Maranh˜ao State
and Maraiwatsede, Mato Grosso State) either border ur-
ban areas or are on roads already opened when the re-
serves were recognized, or both.
Mairawatsede, a somewhat special case, was to have
been returned in 1992 to a group of Xavante who had
been forcibly removed from the area by the military in
1966. When local politicians became aware of the plan,
they fomented large-scale invasions by colonists. The area
was recognized as indigenous land by the Justice Ministry
in 1993, but the government never resettled the colonists.
The Xavante have not returned and the colonists remain.
At the same time, established indigenous lands are virtu-
ally the only places in active frontiers traversed by roads
without deforestation (M˜ae Maria, Par´a State, Xingu Park,
Mato Grosso State). The M˜ae Maria reserve of the Gavi˜oes
is cut by the Par´a-332 road, a power line, and borders the
Caraj´as Iron Ore Railway, but strong Gavi˜oes leadership
has negotiated compensation for the works and exercises
vigilance over reserve boundaries.
Indigenous lands that successfully inhibited deforesta-
tion within the active agricultural frontier were often in-
habited by tribes who actively enforce legal restrictions
on natural resource exploitation by outsiders. The Kayap´o
people have successfully defended their ancestral lands,
expelling ranchers and settlers who invade their reserve
(Zimmerman et al. 2001), maintaining deforestation rates
at close to zero (Schwartzman et al. 2000; Zimmerman et
al. 2001) (Figs. 1a & 2a). In recent years, various indige-
nous groups have taken intruders hostage to reinforce
demands for reserve demarcation and government assis-
tance in protecting boundaries.
It has been postulated that the tendency of indigenous
people to protect their forests from deforestation is lost
as these groups adopt the values of a market-based soci-
ety, and as their population densities increase (Terborgh
2000; Redford & Sanderson 2000; Terborgh & van Shaik
2002). We tested this prediction by examining the re-
sponse of deforestation inhibition by indigenous lands to
population density within the reserve and the time since
first contact with nonindigenous groups. The relationship
between indigenous land population density and defor-
estation inhibition was highly variable (r=0.19) and was
significant at the 90% confidence level (p<0.06; Fig.
4a). Although indigenous land performance in slowing
deforestation also varied greatly with time since contact
Conservation Biology
Volume 20, No. 1, February 2006

Nepstad et al. Inhibition of Amazon Deforestation and Fire 71
Figure4. The relationship between deforestation
inhibition by indigenous lands and (a) population
density and (b) time since contact with white
populations. The ratio between the deforestation rates
inside and outside reserves determined between 1997
and 2000 provided the dependent variable for these
analyses. Reserve codes defined in Fig. 1 legend.
(r=0.40; Fig. 4b), the lands continued to substantially
inhibit deforestation in almost all cases up to 400 years
after contact with the national society. A negative corre-
lation was found over the 200-year gradient ( p<0.05),
indicating a slight tendency for the inhibitory effect of
indigenous lands on deforestation to decline following
contact with nonindigenous people. Factors that we did
not analyze may be more significant for this variation than
time, such as whether or not reserves were invaded be-
fore their legal recognition. (A similar test for parks is not
possible because most were created in the last 50 years.)
The high level of variability of both of these relation-
ships (the dependence of deforestation inhibition on time
since contact and population density) indicates that con-
tact with the national society, population growth, and
resource degradation are not inevitably linked. The eco-
logical integrity of the indigenous lands will ultimately
depend on cultural factors and on the economic alterna-
tives that are available to indigenous peoples.
We did not measure reserve performance in protecting
forests from impoverishment through logging and hunt-
ing. Many indigenous groups have opened their lands to
mahogany extraction. Logging is also common in those
parks that are located in the active frontier. The Panar´a
people have actively expelled mahogany loggers from
their area. The A’Ukre group of the Kayap´o people has ini-
tiated a promising resource management system that may
have prevented depletion of game species (Zimmerman
et al. 2001). The 13-million-ha reserve of the Kayap´o and
the Upper Xingu peoples in south-central Par´a and Mato
Grosso is larger than any tropical forest park in the world
and is the main barrier between the forest and business-
as-usual frontier expansion in the heavily settled eastern
Amazon (Fig. 1a, b). These indigenous groups have been
strengthened through collaboration with conservation or-
ganizations (Zimmerman et al. 2001).
Indigenous lands occupy one-fifth of the closed-canopy
forests of the Brazilian Amazon, which is twice the area
targeted by the Brazilian government for preservation in
parks (http://www.mma.gov.br/port/sbf/dap/parqbras.
html), and five times the area currently designated as
parks (Fig. 1). They also contain larger blocks of forest
than parks, as represented in our sample (Fig. 5). In Ama-
zonia, where 84% of the forest is still standing, protection
of the regional forest-climate system is critical to the long-
term protection of biodiversity, and this will demand for-
est cover over most of the region (Nobre et al. 1991; Silva
Dias et al. 2002). Recent advances in the enforcement of
Figure5. Reserve size versus deforestation rate in the
10-km-wide buffer (i.e., outside) along the outer
perimeter of reserve boundaries. Indigenous lands are
larger than other reserves and many have high rates
of deforestation along their perimeters. Reserve codes
defined in Fig. 1 legend.
Conservation Biology
Volume 20, No. 1, February 2006

72 Inhibition of Amazon Deforestation and Fire Nepstad et al.
environmental legislation in the Brazilian Amazon (Nep-
stad et al. 2002; Fearnside 2003; Soares et al. 2006) dem-
onstrate the potential feasibility of maintaining forest over
most of this region. Government regulation of deforesta-
tion, fire, and logging is essential to this strategy, as is an
expanding network of forest reserves.
The establishment of parks in regions that are largely
inaccessible to humans is an important component of
a long-term strategy to defend nature in places like the
Amazon, but such risk-avoiding reserves must be comple-
mented by reserves within the active frontier. The Brazil-
ian government’s efforts to expand the network of parks
in the Amazon will have the greatest conservation value
if protected areas are successfully established in active
frontier regions, where >20,000 km2of forest are cur-
rently being converted to agriculture each year. This will
require strong political support for such initiatives (Bran-
don 2002; Dourojeanni 2002; van Shaik & Rao 2002) and
improvements in government capacity to protect such
reserves.
Recent experiences in the 7-million-ha Terra do Meio
(Land in the Middle) region north of the Kayapo reserve
complex (Fig. 1a) demonstrate that uninhabited parks
can gain broad political support within active frontier re-
gions of the Brazilian Amazon if they are advanced within
the context of a regional conservation and development
planning process that addresses the needs and aspira-
tion of local indigenous groups, agroextractivist popula-
tions (e.g., rubber tappers), and colonist farmers. From
November 2004 through February 2005, 7 million ha of
new reserves were declared in the Brazilian Amazon, in-
cluding 5 million ha created in the Terra do Meio mo-
saic. The creation of these reserves brought to fruition
a 7-year series of conflicts and campaigns, initiated by
a smallholder organization of the Transamazon highway,
the Movimento pelo Desenvolvimento do Transamazon-
ico e Xingu (MDTX–Movement for the Development of
the Transamazon and Xingu), and supported by regional
(Instituto Socioambiental, Instituto de Pesquisa Ambien-
tal da Amazonia, Comiss˜ao Pastoral da Terra, Conselho Na-
cional dos Seringueiros) and international (Environmen-
tal Defense, Greenpeace) conservation and land-rights
organizations. These new reserves may represent the
biggest conservation achievement in the history of tropi-
cal conservation. The prospects for further conservation
successes on this scale will improve to the extent that
conservationists recognize existing and potential roles of
diverse alliances of the Amazon’s rural populations as pro-
tagonists of nature conservation (Diegues 1992).
Acknowledgments
This study was funded by the Gordon and Betty Moore
Foundation through grants to the Conservation Interna-
tional Center for Applied Biodiversity Science and Woods
Hole Research Center, National Aeronautics and Space Ad-
ministration (Large-Scale Biosphere Atmosphere Experi-
ment, LBA-ECO), the U.S. Agency for International De-
velopment, and a Brazilian Visiting Professor Fellowship
from CAPES.
Literature Cited
Brandon, K. 2002. Putting the right parks in the right places. Pages 443–
467 in J. Terborgh, C. van Schaik, L. Davenport, and M. Rao, editors.
Making parks work. Island Press, Washington, D.C.
Bruner, A. G., R. E. Gullison, R. Rice, and G. A. Fonseca. 2001. Effective-
ness of parks in protecting tropical biodiversity. Science 291:125–
128.
Capobianco, J. P. R., A. Ver´ıssimo, A. Moreira, D. Sawyer, I. Santos, and
L. P. Pinto. 2001. Biodiversidade na Amazˆonia Brasileira: avalia¸c˜ao e
a¸c˜oes priorit´arias para a conserva¸c˜ao, uso sustent´avel e reparti¸c˜ao de
benef´ıcios. Instituto Socioambiental, Esta¸c˜ao Liberdade, S˜ao Paulo.
Diegues, A. 1992. O Mito da Natureza Intocada. Universidade de S˜ao
Paulo, S˜ao Paulo.
Dourojeanni, M. J. 2002. Political will for establishing and managing
parks. Pages 230–334 in J. Terborgh, C. van Schaik, L. Davenport,
and M. Rao, editors. Making Parks Work. Island Press, Washington,
D.C.
Fearnside, P. 2003. Deforestation control in Mato Grosso: a new model
for slowing the loss of Brazil’s Amazon Forest. Ambio 32:342–345.
Ferreira, L. V., E. Venticinque, and S. Almeida. 2005. O desmatamento na
Amazˆonia e a importˆancia das ´areas protegidas. Estudos Avan¸cados
19:1–10.
Instituto Nacional de Pesquisas Espaciais (INPE). 2004. Monitoramento
da floresta amazˆonica brasileira por satelite: projeto PRODES. INPE,
S˜ao Paulo. Available from http://www.obt.inpe.br/prodes/index.
html (accessed July 2004).
Menzel, W. P., and J. F. W. Purdom. 1994. Introducing GOES-I: the first of
a new generation of geostationary operational environmental satel-
lites. Bulletin of the American Meteorological Society 75:757–781.
Nepstad, D. C., et al. 1999. Large-scale impoverishment of Amazonian
forests by logging and fire. Nature 398:505–508.
Nepstad, D. C., G. Carvalho, A. C. Barros, A. Alencar, J. P. Capobianco,
J. Bishop, P. R. Moutinho, P. A. Lefebvre, U. L. Silva, and E. Prins.
2001. Road paving, fire regime feedbacks, and the future of Amazon
forests. Forest Ecology and Management 154:395–407.
Nepstad, D., D. G. McGrath, A. Alencar, A. C. Barros, G. Carvalho, M.
Santilli, and M. d. C. Diaz. 2002. Frontier governance in Amazonia.
Science 295:629–631.
Nobre, C. A., P. J. Sellers, and J. Shukla. 1991. Amazonian deforestation
and regional climate change. Journal of Climate 4:957–988.
Peres, C. A. 2000a. Effects of subsistence hunting on vertebrate commu-
nity structure in Amazonian forests. Conservation Biology 14:240–
253.
Peres, C. A. 2000b. Evaluating the impact and sustainability of subsis-
tence hunting at multiple Amazonian forest sites. Pages 83–115 in J.
G. Robinson and E. L. Bennett, editors. Hunting for sustainability in
Tropical Forests. Columbia University Press, New York.
Peres, C. A., and J. W. Terborgh. 1995. Amazonian nature reserves: an
analysis of the defensibility status of existing conservation units and
design criteria for the future. Conservation Biology 9:34–46.
Pimm, S. L., et al. 2001. Can we defy nature’s end? Science 293:2207–
2208.
Redford, K. H. 1992. The empty forest. BioScience 42:412–422.
Redford, K. H., and S. E. Sanderson. 2000. Extracting humans from na-
ture. Conservation Biology 14:1362–1364.
Robinson, J. G., and E. L. Bennett. 2000. Hunting for sustainability in
tropical forests. Columbia University Press, New York.
Conservation Biology
Volume 20, No. 1, February 2006

Nepstad et al. Inhibition of Amazon Deforestation and Fire 73
S´a, R. M. L., and L. V. Ferreira. 2000. ´
Areas protegidas em espa¸cos
amea¸cados: o grau de implementa¸c˜ao e vulnerabilidade das unidades
de conserva¸c˜ao federal brasileiras de uso indireto. World Wildlife
Fund, Brazil, and Serie T´ecnica III, Bras´ılia.
Silva Dias, M. A. F., et al. 2002. Cloud and rain processes in biosphere-
atmosphere interaction context in the Amazon region. Journal of
Geophysical Research 107:8072, doi:10.1029/2001JD000335.
Schwartzman, S., A. G. Moreira, and D. C. Nepstad. 2000. Rethinking
tropical forest conservation: perils in parks. Conservation Biology
14:1351–1357.
Soares, B., D. Nepstad, L. Curran, G. Cerqueira, R. Garcia, C. Ramos, E.
Voll, A. McDonald, P. Lefebvre, P. Schlesinger, and D. McGrath. 2006.
Modelling conservation in the Amazon Basin. Nature: in press.
Terborgh, J. 1999. Requiem for nature. Island Press, Washington, D.C.
Terborgh, J. 2000. The fate of tropical forests: a matter of stewardship.
Conservation Biology 14:1358–1361.
Terborgh, J., and C. van Schaik. 2002. Why the world needs parks. Pages
3–14 in J. Terborgh, C. van Schaik, L. Davenport, and M. Rao, editors.
Making Parks Work. Island Press, Washington, D.C.
SPSS. 2000. SYSTAT. Version 10. SPSS, Chicago, Illinois.
van Schaik, C., and M. Rao. 2002. The frontier model of development
and its relevance to protected area management. Pages 424–440 in
J. Terborgh, C. van Schaik, L. Davenport, and M. Rao, editors. Making
Parks Work. Island Press, Washington, D.C.
Zimmerman, B., C. A. Peres, J. R. Malcolm, and T. Turner. 2001. Conser-
vation and development alliances with the Kayap´o of south-eastern
Amazonia, a tropical forest indigenous people. Environmental Con-
servation 28:10–22.
Conservation Biology
Volume 20, No. 1, February 2006