Wilderness and biodiversity conservation
R. A. Mittermeier*
†
, C. G. Mittermeier*, T. M. Brooks
‡
, J. D. Pilgrim
‡
, W. R. Konstant*, G. A. B. da Fonseca
‡§
,
and C. Kormos*
¶
*Conservation International and
‡
Center for Applied Biodiversity Science at Conservation International, 1919 M Street NW, Suite 600, Washington,
DC 20036;
§
Departamento de Zoologia, Universidade Federal de Minas Gerais, 31270-901 Belo Horizonte, Minas Gerais, Brazil; and
¶
International Wilderness Leadership (WILD) Foundation, P.O. Box 1380, Ojai, CA 93024
Edited by Edward O. Wilson, Harvard University, Cambridge, MA, and approved July 2, 2003 (received for review April 24, 2003)
Human pressure threatens many species and ecosystems, so con-
servation efforts necessarily prioritize saving them. However,
conservation should clearly be proactive wherever possible. In this
article, we assess the biodiversity conservation value, and specif-
ically the irreplaceability in terms of species endemism, of those of
the planet’s ecosystems that remain intact. We find that 24 wil-
derness areas, all >1 million hectares, are >70% intact and have
human densities of less than or equal to five people per km
2
. This
wilderness covers 44% of all land but is inhabited by only 3% of
people. Given this sparse population, wilderness conservation is
cost-effective, especially if ecosystem service value is incorporated.
Soberingly, however, most wilderness is not speciose: only 18% of
plants and 10% of terrestrial vertebrates are endemic to individual
wildernesses, the majority restricted to Amazonia, Congo, New
Guinea, theMiombo–Mopanewoodlands, and the North American
deserts. Global conservation strategy must target these five wil-
dernesses while continuing to prioritize threatened biodiversity
hotspots.
T
he concept of wilderness is ancient. The word itself is derived
from the Norse will (uncontrolled) and deor (animal), evolv-
ing to its biblical use as ‘‘uncultivated’’ (1). The term began to
gain positive connotations through the Romantic and Transcen-
dentalist writers and Hudson River School of landscape painters
in the 19th century, fledging into a conservation movement with
the writings of Muir, Audubon, and others. The concept first
entered a regulatory context in 1929, building up to the U.S.
Wilderness Act of 1964, which established the standards for
protection of wilderness on federal lands. Countries as diverse as
Australia, Canada, Finland, and South Africa now have similar
wilderness legislation. A wilderness area is defined in TheWorld
Conservation Union (IUCN) Framework for Protected Areas as
‘‘a large area of unmodified or slightly modified land and
or sea,
retaining its natural character and influence, which is protected
and managed so as to preserve its natural condition’’ (2).
Building from ‘‘good news areas’’ (3) and ‘‘major tropical
wilderness areas’’ (4), we now expand the focus of the wilderness
concept beyond specific protected areas to inform global con-
servation strategies.
The units of analysis for this study were based largely on the
world’s terrestrial ecoregions (5). Where these ecoregions could
be combined into broader biogeographic units, such as Amazo-
nia, we aggregated them into single units. To select only eco-
systems of global significance, we set a minimum size for
inclusion in the analysis as 10,000 km
2
. As a preliminary assess-
ment of ecoregions for inclusion as wilderness areas, we overlaid
a binary classification of human population density data (6)
outside of urban areas as greater than and less than five people
per km
2
(rounding down), retaining only the latter for subse-
quent analysis (we subsequently also identified areas with ap-
proximately one person or less per km
2
). We then conducted an
extensive literature search and contacted
200 specialists on
these potential wilderness areas, compiling data on intactness,
biodiversity, human populations, threats, and existing conserva-
tion initiatives. We used intactness (7) as a further criterion for
inclusion, stipulating that an area must retain at least 70% of its
historical habitat extent (500 years ago) to be considered a
wilderness area. These data are less precise than the other data
used here, with sources ranging from detailed remote sensing
assessments through to the opinion of regional specialists, and so
we use only one significant figure for intactness throughout.
In total, this analysis identified 24 wilderness areas (Fig. 1).
These fall into nine terrestrial biome types (8): Tropical Humid
Forest (three areas, 12% of total area); Tropical Dry Forests and
Tropical Grasslands (three areas, 4% of total area); Warm
Deserts
Semideserts and Cold-Winter Deserts (seven areas,
30% of total area); Mixed Mountain Systems, Temperate Rain-
forests, and Temperate Needleleaf Forests (five areas, 23% of
total area); and Tundra Communities and Arctic Desert (two
areas, 30% of total area); plus four relatively small wetland
wildernesses (1% of total area). Six additional regions (the
Appalachians, the European mountains, the Sudd swamp, the
Serengeti, the Brazilian Caatinga, and the Peruvian and Chilean
coastal deserts) came close to but failed to meet the thresholds.
The total historical area of the 24 wildernesses was 76 million
km
2
, 52% of the Earth’s land area, of which 65 million km
2
remains intact, covering 44% of the planet (Table 1, and see
Table 3, which is published as supporting information on the
PNAS web site, www.pnas.org). The total human population of
these areas is 204 million (3% of the global total), which is
reduced to 83 million (1.4%) when urban areas are excluded
(Table 1 and see Table 3), giving an average rural population
density of 1.1 people per km
2
.
Considering only those 16 wilderness areas with rural human
population densities of less than approximately one person per
km
2
, the results are even more striking. Together, these areas
covered 66 million km
2
(45% of the land’s surface), of which
90% remains intact, accounting for 57 million km
2
(39% of the
land’s surface), an area equivalent to the world’s six largest
countries combined. A third of this area is under permanent ice,
making habitation impossible; in total, this vast area holds just
43 million people, or 0.7% of Earth’s human population.
This study provides an assessment of the biodiversity value of
remaining wilderness areas. About 55,000 vascular plant species
(18% of the global total) and 2,800 terrestrial vertebrate species
(10%) are endemic to the wilderness areas (Table 2, and
see Table 4, which is published as supporting information on
the PNAS web site). The disparity between the plant and the
vertebrate percentages is probably explained by the much
smaller mean range size of plants (9). However, even for plants,
and even given that many wilderness areas are poorly studied
(10), this percentage is clearly far lower than would be expected
were endemic species distributed across ecoregions in propor-
tion to land area. Further, the vast majority of these species are
concentrated into just five high-biodiversity wilderness areas:
Amazonia, the Congo forests of Central Africa, NewGuinea, the
Miombo–Mopane woodlands of Southern Africa (including the
Okavango Delta), and the North American desert complex of
northern Mexico and the southwestern U.S. The intact portion
of these five wildernesses covers 8,981,000 km
2
(76% of their
This paper was submitted directly (Track II) to the PNAS office.
†
To whom correspondence should be addressed. E-mail: r.mittermeier@conservation.org.
www.pnas.org
cgi
doi
10.1073
pnas.1732458100 PNAS  September 2, 2003  vol. 100  no. 18  10309–10313
E
C
O
L
O
G
Y

Table 1. Extents of the 24 wilderness areas, their human populations, and their levels of protection
Biome and wilderness Area,* km
2
Intact,* %
Population
†
People per km
2†
Protected areas,
‡
%Total Minus urban
Tropical humid forest
Amazonia 6,683,926 80 21,430,115 7,355,126 1.1 8.3
Congo forest 1,725,221 70 16,000,000 10,000,000 5.8 8.1
New Guinea 828,818 70 6,000,000 4,197,200 5.1 11
Tropical dry forests and grasslands
Chaco 996,600 70 2,810,000 648,693 0.65 7.5
Miombo–Mopane 1,176,000 90 5,839,000 3,816,000 3.2 36
Australian savannas 585,239 100 60,730 24,188 0.041 11
Mixed mountain, temperate rain, and temperate needleleaf forest
Rocky Mountains 570,500 70 1,574,986 1,035,174 1.8 17
Pacific Northwest 315,000 80 770,000 597,095 1.9 48
Magellanic forests 147,200 100 253,264 34,501 0.23 72
Tasmanian WHWA 13,836 90 8 8 0.000058 100
Boreal forests 16,179,500 80 30,337,925 15,438,546 0.95 3.8
Wetland
Llanos 451,474 80 4,444,243 1,065,956 2.4 15
Pantanal 210,000 80 1,125,200 81,200 0.38 2.7
Ban˜ados del Este 38,500 80 200,000 40,000 1.0 2.8
Sundarbans 10,000 80 3,000 3,000 0.30 31
Warm and cold-winter deserts
North American deserts 1,416,134 80 15,348,342 4,509,403 3.2 23
Patagonia 550,400 70 800,000 200,000 0.36 4.1
Sahara 7,780,544 90 35,187,620 10,273,595 1.3 2.8
Kalahari–Namib 714,700 80 1,422,700 425,900 0.60 25
Arabian deserts 3,250,000 90 47,000,000 15,000,000 4.6 8.3
Central Asian deserts 5,943,000 80 9,000,000 5,500,000 0.93 2.8
Australian deserts 3,572,209 90 400,000 285,000 0.080 9.4
Tundra
Arctic tundra 8,850,000 90 4,288,613 2,385,713 0.27 20
Antarctic 13,900,000 100 1,000 1,000 0.000072 0.025
Total 75,908,801 90 204,296,746 82,917,298 1.1 7.5
WHWA, World Heritage Wilderness Area.
*Area of each ofwilderness area andpercentage intactness (one significantfigure). Areasmust exceed 10,000 km
2
and 70% intactness to qualify aswildernesses.
†
Human population, human population outside of urban areas, and human population density (two significant figures) of each wilderness area.
‡
Percentage (two significant figures) of each wilderness area under protected area status (IUCN categories I–IV).
Fig. 1. Overall map showing wilderness areas, human population density less than or equal to five people per km
2
, with biomes shaded, and the five
high-biodiversity wilderness areas outlined in red.
10310  www.pnas.org
cgi
doi
10.1073
pnas.1732458100 Mittermeier et al.