RESEARCH ARTICLE
Fo es Loss in P o ec ed A eas and In ac
Fo es Landscapes: A Global Analysis
Ma ias Heino
1
, Ma i Kummu
1
*, Ma ika Makkonen
2
, Ma k Mulligan
3
, Pe e H. Ve bu g
4
,
Mika Jala a
1
, Timo A. Räsänen
1
1Wa e and De elopmen Resea ch G oup, Aal o Uni e si y, Tie o ie 1E, 02150, Espoo, Finland, 2Na u al
Resou ces Ins i u e Finland (Luke), Jokiniemenkuja 1, 01301, Van aa, Finland, 3Depa men o Geog aphy,
King's College London, S and, London, WC2R 2LS, Uni ed Kingdom, 4En i onmen al Geog aphy G oup,
VU Uni e si y Ams e dam, de Boelelaan 1087, 1081 HV, Ams e dam, he Ne he lands
*[email p o ec ed]
Abs ac
In spi e o he high impo ance o o es s, global o es loss has emained ala mingly high du -
ing he las decades. Fo es loss a a global scale has been un eiled wi h inc easingly ine
spa ial esolu ion, bu he o es ex en and loss in p o ec ed a eas (PAs) and in la ge in ac
o es landscapes (IFLs) ha e no so a been sys ema ically assessed. Mo eo e , he impac
o p o ec ion on p ese ing he IFLs is no well unde s ood. In his s udy we conduc ed a con-
sis en assessmen o he global o es loss in PAs and IFLs o e he pe iod 2000–2012. We
used ecen ly published global emo e sensing based spa ial o es co e change da a, being
a uni o m and consis en da ase o e space and ime, oge he wi h global da ase s on PAs’
and IFLs’loca ions. Ou analyses e ealed ha on a global scale 3% o he p o ec ed o es ,
2.5% o he in ac o es , and 1.5% o he p o ec ed in ac o es we e los du ing he s udy
pe iod. These o es loss a es a e ela i ely high compa ed o global o al o es loss o 5%
o he same ime pe iod. The a ia ion in o es losses and in p o ec ion e ec was la ge
among geog aphical egions and coun ies. In some egions he loss in p o ec ed o es s
exceeded 5% (e.g. in Aus alia and Oceania, and No h Ame ica) and he ela i e o es loss
was highe inside p o ec ed a eas han ou side hose a eas (e.g. in Mongolia and pa s o
A ica, Cen al Asia, and Eu ope). A he same ime, p o ec ion was ound o p e en o es
loss in se e al coun ies (e.g. in Sou h Ame ica and Sou heas Asia). Globally, high a ea-
weigh ed o es loss a es o p o ec ed and in ac o es s we e associa ed wi h high g oss
domes ic p oduc and in he case o p o ec ed o es s also wi h high p opo ions o ag icul u al
land. Ou indings ein o ce he need o imp o ed unde s anding o he easons o he high
o es losses in PAs and IFLs and s a egies o p e en u he losses.
In oduc ion
Fo es s play a c ucial ole in sus aining li e on ea h. They main ain ecological di e si y, egu-
la e clima e, s o e ca bon, p o ec soil and wa e and p o ide esou ces and li elihoods o he
wo ld’s popula ion [1–4]. Despi e he inc easing awa eness o he impo ance o hese ecosys-
ems, global de o es a ion a es ha e emained ala mingly high o e he pas decades [2].
PLOS ONE | DOI:10.1371/jou nal.pone.0138918 Oc obe 14, 2015 1/21
OPEN ACCESS
Ci a ion: Heino M, Kummu M, Makkonen M, Mulligan
M, Ve bu g PH, Jala a M, e al. (2015) Fo es Loss in
P o ec ed A eas and In ac Fo es Landscapes: A
Global Analysis. PLoS ONE 10(10): e0138918.
doi:10.1371/jou nal.pone.0138918
Edi o : Madhu Anand, Uni e si y o Guelph,
CANADA
Recei ed: Ma ch 16, 2015
Accep ed: Sep embe 4, 2015
Published: Oc obe 14, 2015
Copy igh : © 2015 Heino e al. This is an open
access a icle dis ibu ed unde he e ms o he
C ea i e Commons A ibu ion License, which pe mi s
un es ic ed use, dis ibu ion, and ep oduc ion in any
medium, p o ided he o iginal au ho and sou ce a e
c edi ed.
Da a A ailabili y S a emen : All ele an da a a e
wi hin he pape and i s Suppo ing In o ma ion iles.
Funding: The wo k was unded by Maa- ja
esi ekniikan uki y. In addi ion, MK was unded by
he Academy o Finland p ojec SCART (g an no.
267463) and MMa by he Academy o Finland p ojec
G een Economy and Policies (g an no. 257919).
PHV hanks he ERC g an ag eemen n . 311819
(GLOLAND) and he FP7 p ojec LUC4C o suppo .
The unde s had no ole in s udy design, da a
collec ion and analysis, decision o publish, o
p epa a ion o he manusc ip .
In yea 2010 o es co e ed a ound 40 million km
2
o 31% o he global land a ea acco ding
o coun y epo s [2]. Es ima es o global o es loss a es wi hin he pas decade a y be ween
130,000 km
2
/y [2] and 177,000 km
2
/y [5]. A he same ime, some e o es a ion and na u al
eg ow h ha e occu ed, leading o ne o es loss epo s anging be ween 52,000 km
2
/y and
115,000 km
2
/y , espec i ely. Al hough o es loss is s ill ema kably high, i has shown some
signs o decline: FAO [2] epo s ha du ing he 1990s, he o es loss a e was 160,000 km
2
/y ;
whe eas be ween 2000–2010, he a e was 130,000 km
2
/y . Acco ding o FAO [2] he opics
we e he only domain whe e he a e o o es loss inc eased in he i s decade o he 21
s
cen-
u y compa ed o 1990s: de o es a ion inc eased in he opics by 2,101 km
2
/y on a e age
ac oss he decade. De o es a ion in he opics accoun ed o 32% o global o es loss wi hin
he pe iod o 2000–2012 [5]. Howe e , he e a e also posi i e signs in he opics: he a e o
o es loss in he B azilian Amazon has declined in ecen yea s [6,7].
The main d i e s o global de o es a ion a e linked o expansion o ag icul u e, wood ex ac-
ion, in as uc u e ex ension, popula ion g ow h, and expansion o ag icul u e [8–13]. The dom-
inan d i e s, howe e , a y among he egions [8–13]. In addi ion o ag icul u e and popula ion
g ow h, a me a-analysis o 117 de o es a ion s udies by Fe e i-Gallon and Busch [9]sugges s
ha de o es a ion is gene ally lowe in high, s eep and we a eas while i is highe in a eas whe e
o es s a e close o oads and u ban a eas. De o es a ion has also shi ed om a dominan ly s a e
ini ia ed o an en e p ise d i en p ocess in he opics be ween 1970 and 2000 [14].
Ala ming de o es a ion a es combined wi h he inc easing awa eness o he impo ance o o -
es s ha e esul ed in exponen ial g ow h o he wo ld’s p o ec ed a eas (PA) o e he pas decades
[15–17]. Schmi e al. [18] es ima e he global o es co e using ea h obse a ion sa elli e da a
om MODIS2005 and he ex en o p o ec ed o es using he Wo ld Da abase o P o ec ed A eas
(WDPA) o he yea 2008. They ind ha 7.7% o global o es co e ell wi hin IUCN’s ou
s ic es p o ec ion ca ego ies (I-IV) and 13.5% wi hin IUCN’s all six p o ec ion ca ego ies (I-VI)
(IUCN p o ec ion ca ego ies a y om s ic ly p o ec ed a eas (I) o p o ec ed a eas wi h sus ain-
able esou ce use (VI); o mo e de ailed ca ego y de ini ion see [19]).Schmi e al.[18] u he
conclude ha o es p o ec ion a ied g ea ly be ween di e en egions and o es ypes, and ha
o es p o ec ion in p io i y a eas, such as biodi e si y ho spo s, was insu icien .
Conside able e o s o p o ec ion a e a ge ed o he p ima y o es s and la ge in ac o es
landscapes (IFLs, i.e. unb oken expanse o na u al ecosys ems). These o es a eas play c ucial
oles in sus aining ecological di e si y [2,20]. FAO [2] es ima es ha p ima y o es s accoun ed
14 million km
2
(36% o he global o al o es s) in yea 2010, ha ing dec eased ala ming
400,000 km
2
o e he pe iod o 2000–2010 (annual a e 0.4%). Po apo e al. [20] epo ha
he ex en o IFL is 13.1 million km
2
. Vas majo i y o he IFLs a e ound ei he in dense T opi-
cal and Sub-T opical o es s (45% o wo ld o al) o in Bo eal Fo es s (44%). The lowes p o-
po ion o IFL was ound in Tempe a e B oadlea and Mixed Fo es s. Po apo e al. [20]
u he ind ha 18.9% o IFLs a e unde p o ec ion o IUCN p o ec ion ca ego ies I-VI and
only 9.7% o IFLs is s ic ly p o ec ed unde IUCN p o ec ion ca ego ies I-III.
The p o ec ion e ec o PAs is, howe e , ques ioned. The PAs a e conside ed in many cases
o be biased in hei loca ion, meaning ha PAs a e loca ed in a eas ha a e unlikely o ace
land con e sion p essu es [21]. Joppa and P a [22] e eal ha a majo i y o he PA ne wo ks
a e loca ed in high ele a ions, s eep slopes and a om oads and ci ies. Joppa and P a [21]
a gue ha his bias has esul ed in o e es ima ions in he p o ec ion e ec o PAs. Local case
s udies (e.g. [23]) suppo hese global indings. Joppa and P a [24] use ‘ma ching’app oach
ha a emp s o a oid he o e es ima ion o p o ec ion e ec o PAs in 147 coun ies by com-
pa ing p o ec ed and non-p o ec ed a eas wi h simila land cha ac e is ics. They ind ha
ma ching educed he p o ec ion e ec in 80% o he coun ies compa ed o an assessmen
wi hou ma ching. Al oge he hey ind ha he p o ec ion educed con e sion o na u al land
P o ec ed and In ac Fo es Loss
PLOS ONE | DOI:10.1371/jou nal.pone.0138918 Oc obe 14, 2015 2/21
Compe ing In e es s: The au ho s ha e decla ed
ha no compe ing in e es s exis .
co e in 75% o he assessed coun ies. P o ec ed a eas a e also epo ed o become inc easingly
isola ed, especially in opics [25]. This is ala ming because smalle p o ec ed a eas a e o en
unde g ea h ea [26] and isola ion o hese a eas es ic s he habi a size, i.e. limi s he su -
i al o a g ea numbe o auna and lo a species (e.g. [27,28]).Al hough he unde s anding o
o es loss a a global scale wi h an inc easing spa ial esolu ion is g owing apidly [5], he
global o es loss in PAs and in IFLs a e no assessed wi h de ailed and uni o m da ase s ha
allow consis en o es ex en compa isons o e space and ime. So a only egional analyses
exis a his le el, o example, o Indonesia [29]. Fu he mo e, he success o p o ec ion in
p ese ing he IFLs is no ye well unde s ood. The e o e, in his s udy, we aim o conduc a
consis en and spa ially explici assessmen o he global o es loss, pa icula ly wi hin p o-
ec ed and in ac o es s by ocusing on he a ia ion among he coun ies be ween 2000 and
2012. Addi ionally, we aim o s udy whe he socio-economic indica o s can o e po en ial
explana ion o hese obse ed global o es losses in PAs and IFLs. We hypo hesised ha i) he
ex en o o es loss wi hin PAs and IFLs a ies s ongly among he coun ies bu is always less
mani es ed wi hin PAs han ou side o hem, and ii) coun y le el indica o s o popula ion size,
land use change and s a e economy can o e po en ial explana ion o he obse ed global o es
losses in PAs and IFLs. The key e minology used in his s udy is explained in Table 1.
Ma e ials and Me hods
To conduc he assessmen , we combined ou global da ase s (Fig 1;Table 2): o es ex en
based on he Global Fo es Change (GFC) da a [5], o es loss based on he GFC da a [5], he
Wo ld Da abase o P o ec ed A eas (WDPA) [30] and he global La ge In ac Fo es Land-
scapes (IFL) da a [20]. F om GFC we assessed he o es ex en and o es loss while WDPA
and IFL da ase s we e used o de e mine how much o ha loss ook place in p o ec ed a eas
and in ac o es landscapes espec i ely. The esul ing spa ial da a we e agg ega ed o coun y-
scale and analysed wi h Weigh ed Leas -Squa es (WLS) eg ession analyses o assess whe he
one o a se ies o socio-economic indica o s a e associa ed o o es loss pa e ns, possibly
explaining hei occu ence. The o es loss esul s a e gi en a na ional and global scale and by
geog aphical egions, while eg ession analyses we e conduc ed a na ional scale o co espond
wi h he socio-economic indica o da a and he pu sued o es go e nance le el. Below he
da a and me hods a e desc ibed in mo e de ail.
Da a
We used he GFC da ase [5] o map he o es ex en and o es loss (Fig 1A and 1B;Table 2).
GFC da a a e based on Ea h obse a ion sa elli e da a and ha e a esolu ion o ~30 me e s a
Table 1. Key e minology used in he s udy.
Te m Defini ion
Fo es Vege a ion alle han 5 m wi h 20% ee co e canopy h eshold using Global
Fo es Change (GFC) da a by Hansen e al. [5].
Fo es loss ”S and eplacemen dis u bance o comple e emo al o o es canopy”as
defined by Hansen e al. [5] in GFC da a. Fo es gain was no aken in o accoun .
P o ec ed a ea (PA) P o ec ed a ea as in Wo ld Da abase on P o ec ed A eas (WDPA) [30].
P o ec ed o es Fo es wi hin PA (i.e. p o ec ed a ea).
In ac o es landscape
(IFL)
”An unb oken expanse o na u al ecosys ems wi hin a eas o cu en o es
ex en ,wi hou signs o significan human ac i i y,and ha ing and a ea o a leas
500 km
2
”as in IFL da ase by Po apo e al. [20].
In ac o es Fo es wi hin IFL (i.e. In ac o es landscape).
P o ec ed in ac o es In ac o es wi hin PA (i.e. p o ec ed a ea).
doi:10.1371/jou nal.pone.0138918. 001
P o ec ed and In ac Fo es Loss
PLOS ONE | DOI:10.1371/jou nal.pone.0138918 Oc obe 14, 2015 3/21
he equa o (i.e. 1 a c-second). The o es ex en in he GFC da a a e gi en as canopy co e pe -
cen age pe ou pu g id cell o all ege a ion alle han 5 me e s o he yea 2000 [5]. The o -
es loss in he GFC da ase is de ined as s and- eplacing dis u bance, o comple e emo al o
ee canopy, which has occu ed be ween yea s 2000 and 2012 [5]. In his s udy, we used he
same de ini ion o o es loss (Table 1). The GFC da ase p o ides g oss and ne o es loss
da a. We chose o ocus on g oss o es loss ins ead o ne o es loss, as we belie e he o me
e lec s be e he occu ence o o es dis u bances in p o ec ed and in ac o es s han he la -
e . The use o ne o es loss could ha e masked some o he dis u bances h ough o es gain.
I should be no ed ha he GFC da ase does no speci y he easons behind o es loss and
hus, he epo ed o es loss inside p o ec ed a eas migh be due o di ec human ac ions (e.g.
illegal logging, managemen s a egies) as well as na u al causes (e.g. o es i es, diseases, pes s
Fig 1. Agg ega ed (A-B) and con e ed (C-D) da ase s used o he o es loss analyses. A) Fo es ex en in pe cen age o g id a ea a 1 km esolu ion,
based on Global Fo es Change (GFC) da a [5]; B) Fo es loss in pe cen age o g id a ea a 1 km esolu ion, based on GCF da a [5]; C) P o ec ed a eas
based on Wo ld Da abase o P o ec ed A eas (WDPA) [30], con e ed o 1 km esolu ion; and D) La ge In ac Fo es Landscapes (IFLs) based on da ase by
Po apo e al [20], con e ed o 1 km esolu ion.
doi:10.1371/jou nal.pone.0138918.g001
Table 2. A desc ip ion o he da ase s used in his s udy. GFC s ands o Global Fo es Change da ase .
Da ase Spa ial ex en Time pe iod Spa ial esolu ion Re e ence
GFC: Fo es loss La : 80N - 60S Lon:
90E - 90W
2000–2012 30 m (a equa o ; 1 a c-second x 1 a c-second) Hansen e al. [5]
GFC: Fo es ex en La : 80N - 60S Lon:
90E - 90W
2000 30 m (a equa o ; 1 a c-second x 1 a c-second) Hansen e al. [5]
Wo ld Da abase o P o ec ed
A eas (WDPA)
Global 2010 ( e e ence
yea )
Vec o da a wi h a ying esolu ion IUCN and
UNEP-WCMC [30]
La ge In ac Fo es Landscapes
(IFL)
Global 2000 Vec o da a wi h app oxima e esolu ion o
1 km (scale: 1:1,000,000)
Po apo e al. [20]
doi:10.1371/jou nal.pone.0138918. 002
P o ec ed and In ac Fo es Loss
PLOS ONE | DOI:10.1371/jou nal.pone.0138918 Oc obe 14, 2015 4/21
and s o ms) [2,5,31,32]. Fo example, na u al o es i es a e epo ed o be he mos signi ican
cause o o es loss in he bo eal o es s [33].
We used he WDPA da ase o map he p o ec ed a eas o he globe (Fig 1C;Table 2). The
da ase is a join p ojec o IUCN and UNEP and i is s a ed o be he mos ex ensi e da ase on
p o ec ed a eas con aining bo h na ionally and in e na ionally designa ed p o ec ed a eas [30].
The da ase is based on bo h au ho i y and communi y sou ces. We included only he p o-
ec ed a eas epo ed by au ho i y sou ces in o ou assessmen .
Fo he in ac o es a eas we used he global ex en o IFLs (Fig 1D;Table 2)[20]. The da a-
se is c ea ed using a ious emo ely sensed da ase s and maps, including da a om MODIS
and Landsa . They de ine IFL as an unb oken a ea o na u al ecosys ems wi hin o es s, wi hou
signs o signi ican human ac i i y, and ha ing an a ea o a leas 500 km
2
. The e e ence yea
o his da ase is 2000, i.e. he s a ing yea o ou analysis.
Me hods: o es ex en and o es loss
We conduc ed he analyses on 1 km esolu ion (a he equa o ; i.e. on 30 a c-second esolu-
ion). All da ase s we e ei he agg ega ed ( o es ex en and o es loss) o con e ed (PA and
IFL) o his esolu ion. The con e sion p ese ed he o iginal esolu ion o he inpu da a
(Table 2) and he agg ega ion o coa se esolu ion was ca ied ou in a way ha o iginal in o -
ma ion was no dis o ed (see below o mo e de ails).
Be o e any da a agg ega ion, we ans o med he o iginal GFC o es co e da a in o boolean
ype a 30 m esolu ion. Fo he ans o ma ion we used 20% ee canopy co e h eshold, simi-
la ly o Po apo e al. [33], o de ine whe he he 30 m esolu ion g id cells a e classi ied as o es
o no- o es . These boolean o es co e da a we e mul iplied by a su ace a ea o 30 m cell in
ques ion. The o es ex en a ea da a (in km
2
) we e hen agg ega ed in o 1 km esolu ion by sum-
ming he su ace a ea o o es om 30 m g id cells wi hin each 1 km g id cell (no e: he a ea o
30 m g id cells a y o e la i udes in he o iginal WGS84 p ojec ion and his was aken in o
accoun in a ea calcula ions). The GFC o es loss da a a e boolean da a in na u e (loss o no
loss). These o es loss boolean da a we e mul iplied by a su ace a ea o 30 m cell in ques ion and
agg ega ed o 1 km esolu ion simila ly as o es ex en (see abo e). Possible o es loss was only
accoun ed in hose 30 m g id cells ha had ee canopy co e o e he 20% h eshold in he o es
ex en da ase . As a esul we go he o es ex en and o es loss in km
2
o each 1 km g id cell
wi hou losing any in o ma ion in he agg ega ion p ocess. To illus a e he agg ega ed da ase s,
he o es ex en and o es loss in pe cen ages o g id cell a ea a e p esen ed in Fig 1A and 1B.
IFL and WDPA da ase s (Fig 1C and 1D;Table 2) we e used o calcula e he ex en and loss
o p o ec ed o es , in ac o es and p o ec ed in ac o es . We i s con e ed he ec o da a
o bo h da ase s o as e da a wi h 1 km esolu ion, being he o iginal esolu ion o IFL da a
while esolu ion o WDPA da a a ies be ween he en ies. A e his we used he agg ega ed
GFC da a o calcula e he a ea o o es ex en and o es loss inside each ype o spa ial c i e ia
wi hin an analysis uni (coun y, egion o global): i) o al o es ex en and loss, ii) o es ex en
and loss wi hin p o ec ed a eas, iii) o es ex en and loss wi hin in ac o es landscapes, and
i ) o es ex en and loss wi hin p o ec ed in ac o es landscapes. We epo bo h absolu e
and ela i e alues o o es ex en and loss.
By using hese esul s, we analysed he o es p o ec ion e ec o p o ec ed a eas. The e ec
was es ima ed by calcula ing anomaly a ios o ela i e loss o p o ec ed and non-p o ec ed
o es , in ac and non-in ac o es , and o p o ec ed in ac and non-p o ec ed in ac o es .
The anomaly a ios we e calcula ed a na ional scale by di iding he ela i e o es loss in he
p o ec ed a eas in ques ion by he ela i e o es loss in he co esponding non-p o ec ed a eas,
e.g. [( ela i e loss o p o ec ed in ac o es / ela i e loss o non-p o ec ed in ac o es )–1].
P o ec ed and In ac Fo es Loss
PLOS ONE | DOI:10.1371/jou nal.pone.0138918 Oc obe 14, 2015 5/21

Thus he anomaly a ios p o ide in o ma ion on he e ec o p o ec ed a eas on o es loss as
hey indica e whe he he ela i e o es loss is lowe (<0) o highe (>0) in he p o ec ed o es
han in he non-p o ec ed o es .
We u he iden i ied simila i ies in o es loss pa e ns be ween coun ies by clus e ing he
coun y scale esul s wi h k-mean clus e ing me hod. In his me hod he clus e ing o he
obse a ions (i.e. coun y- alues) is based on minimizing he sum o he squa ed Euclidean dis-
ances be ween each obse a ion and he clus e -cen oid- alue. We assessed goodness o i in
e ms o a ia ion wi hin clus e s. To de ine he op imal numbe o clus e s we used he SSE
cu e. We pe o med wo di e en clus e analysis, each conside ing ou di e en pa ame e s.
In he i s clus e analysis, we iden i ied clus e s based on absolu e and ela i e losses in o al
o es and p o ec ed o es a eas while in he second clus e analysis, we conside ed absolu e
and ela i e o es losses in in ac o es landscapes and p o ec ed in ac o es landscapes.
I should be no ed ha while o o es ex en and in ac o es da ase s he e e ence yea is
2000, o p o ec ed a eas he co e age in yea 2010 was aken. The e e ence yea 2010 o p o-
ec ed a eas was chosen because ha moniza ion o he e e ence yea was no easible. The
WDPA da ase does no epo he es ablishmen yea o all p o ec ed a eas. This howe e did
no comp omise he analyses as we ound ha he WDPA da a om he yea 2010 adequa ely
e lec he p o ec ion s a us o he o es o e he s udy pe iod: only 8% o he wo lds p o ec ed
a eas (whe e an es ablishmen yea is p o ided) we e es ablished a e he yea 2000.
Me hods: he ela ionship be ween o es loss and socio-economic
indica o s
To s udy he ela ionship be ween obse ed o es loss and socio-economic a ibu es, we
employed WLS (i.e. Weigh ed Leas Squa es) eg ession analysis, which allowed he o al ex en
o o es ype in ques ion o be used as weigh s. The deg ee o which he socio-economic a i-
bu es can accoun o he ela i e o es loss was analysed o each o he ou o es loss mea-
su es conside ed in his s udy: o al o es loss, loss in p o ec ed o es , loss in in ac o es , and
loss in p o ec ed in ac o es . Al oge he 11 commonly used socio-economic indica o s we e
selec ed and hey a e lis ed in Table 3. The socio-economic indica o s we e, mo eo e , selec ed
Table 3. The socio-economic indica o s used in he eg ession analysis. The indica o s we e used as independen a iables in he assessmen o he
o es loss d i e s wi h Weigh ed Leas -Squa es (WLS) eg ession analysis.
Va iable Da a ype Time pe iod Re e ence
Popula ion densi y pe squa e km 2010 UN [34]
Popula ion densi y g ow h % 2000–2010 UN [34]
GDP g ow h % 2000–2010 Wo ld Bank [35]
GDP pe capi a (PPP) In e na ional USD 2010 Wo ld Bank [35]
Ru al popula ion o o al popula ion % 2010 Wo ld Bank [35]
Popula ion g ow h % 2000–2010 Wo ld Bank [35]
Co up ion pe cep ion index
a
0.. 10 2010 T anspa ency In e na ional [36]
Poli y index
b
-10.. 10 2010 Ma shall and Gu [37]
Human De elopmen Index (HDI)
c
0.. 1 2012 UNDP [38]
Ag icul u e o o al land a ea % 2010 Wo ld Bank [35]
Ag icul u al land a ea g ow h % 2000–2010 Wo ld Bank [35]
a
Low co up ion pe cep ion index co esponds wi h high co up ion and ice e sa
b
Low poli y index co esponds wi h low democ acy and ice e sa
c
High HDI co esponds wi h high human de elopmen and ice e sa
doi:10.1371/jou nal.pone.0138918. 003
P o ec ed and In ac Fo es Loss
PLOS ONE | DOI:10.1371/jou nal.pone.0138918 Oc obe 14, 2015 6/21
o ep esen he widely acknowledged causes o global de o es a ion: popula ion size, land-use
change, s a e go e nance and s a e economy. The selec ion was ca ied ou wi h a equi emen
o a ailable homogeneous da a a he global scale. We b ie ly desc ibe he WLS eg ession
analysis below while s ep-by-s ep desc ip ion is gi en in S3 Appendix.
The dependen a iables we e log
10
– ans o med p io o analyses o ensu e he no mali y
o hei dis ibu ions. Simila ly, he independen a iables we e ans o med wi h a link unc-
ion in cases i imp o ed he linea i y o he ela ion wi h he analysed o es loss measu es.
The link unc ion o each independen a iable was de ined sepa a ely o each ou dependen
a iables. Despi e he ans o ma ions, he da a included some s a is ical ou lie s. Exclusion o
hese ou lie s did no a ec he signi icance o he esul s and hus he epo ed esul s we e
gained wi hou omi ing he s a is ical ou lie s. The mul iple eg ession models we e buil by
s a ing wi h he ull model comp ising o all 11 independen a iables and ending wi h a
model whe e no mul icollinea i y was de ec ed. Mul icollinea i y was de e mined by he Va ia-
ion In la ion Fac o (VIF, a se limi o VIF <4) and he collinea independen a iable wi h
he lowes explana o y powe was always excluded om he model.
Resul s
Ou analyses based on GFC da a [5] e ealed ha o es co e ed one hi d o he o al land a ea
(app oxima ely 43 million km
2
) in he yea 2000, and 19% o his o es (o e 8 million km
2
)
was unde some o m o p o ec ion (Table 4). Acco ding o ou analyses 25% (11 million km
2
)
o global o es s we e in ac , o which almos 35% was p o ec ed. A eas ha a e bo h in ac and
p o ec ed co e ed hus a o al o almos 3.7 million km
2
(o 8.7% o he o al o es ex en )
(Table 4).
The global o es loss based on GFC da a was 2.14 million km
2
be ween he yea s 2000 and
2012, being o e 5% o he o al o es ex en in yea 2000 (Table 4). This global o es loss a e
co esponds o calcula ions by Hansen e al. [5]. Fu he , acco ding o ou analyses, o e 10%
o he o al o es loss occu ed in p o ec ed a eas: 219,000 km
2
o p o ec ed o es (3% o he
o al p o ec ed o es ) was los wi hin ha ime pe iod. We u he ound ha 269,000 km
2
(2.5%) o in ac and 55,000 km
2
(1.5%) o p o ec ed in ac o es we e los du ing he yea s
2000–2012 (Table 4). In he ollowing sec ions he esul s a e p esen ed in mo e de ail. The
nume ic esul s a e also gi en o each coun y in S2 Table o he supplemen a y in o ma ion.
I is wo h o highligh ha ou es ima es on o al o es ex en and o al o es loss on a global
scale p esen ed in his sec ion a e gi en only o alida e ou analyses (i.e. compa ison o Han-
sen e al. [5]) and o pu ou esul s on PAs and IFLs in o a con ex .
Fo es ex en s and hei p o ec ion s a us
In yea 2000 he absolu e o es ex en was la ges in Russia, China, Indonesia, Democ a ic
Republic o he Congo, B azil, Uni ed S a es, and Canada, exceeding 1 million km
2
(Fig 2A).
Toge he hese coun ies con ained o e 60% o he global o es ex en . When examining ela-
i e o es ex en (Fig 2B), pa icula ly Scandina ia and Finland (No dic coun ies om he e
on), and coun ies in cen al A ica and Sou heas Asia had a ema kable po ion o hei e i-
o y unde o es . The coun ies wi h he la ges p o ec ed o es a eas we e mainly he same
coun ies wi h he g ea es ex en o o es (Fig 2C). Howe e , when assessing he a io o p o-
ec ed o es ex en and he o al o es ex en , he la ges p o ec ed o es a eas we e loca ed in
Cen al Eu ope, Aus alia, and in some coun ies o Sou h Ame ica, Sub-Saha an A ica and
he Middle Eas (Fig 2D), whe e o e 30% o he o es s we e p o ec ed ( hough hese o en ep-
esen small a eas o o es ).
P o ec ed and In ac Fo es Loss
PLOS ONE | DOI:10.1371/jou nal.pone.0138918 Oc obe 14, 2015 7/21
The coun ies wi h he la ges o al o es ex en had also he la ges a eas o in ac o es
(o e 400,000 km
2
), excep Indonesia, Uni ed S a es and China (Fig 2E). On he o he hand, a
la ge pa o Eu opean, Middle Eas e n, No h A ican and Sou he n A ican coun ies did no
ha e any in ac o es landscapes (Fig 1D). When compa ing he a ea o in ac o es o he
a ea o o al o es , we ound ha in many coun ies o No h and Sou h Ame ica, Sou heas
Asia, Middle A ica and Aus alia sha es o in ac o es o o al o es we e e y high (Fig 2F).
In con as o hese coun ies, No dic coun ies had a la ge p opo ion o hei land unde o -
es (Fig 2B) bu only less han 2% o ha o es was classi ied as in ac o es (Fig 2F).
Only B azil had o e 1 million km
2
o p o ec ed in ac o es (Fig 2G). Russia, Canada,
Uni ed S a es and no he n coun ies o Sou h Ame ica had o e 100,000 km
2
o hei in ac
o es s unde p o ec ion. When examining he ela i e ex en o p o ec ed in ac o es (ex en
o p o ec ed in ac o es e sus ex en o o al in ac o es ), we ound ha majo i y o in ac
o es , o e 70%, we e p o ec ed in coun ies such as Thailand, New Zealand, Japan, Madagas-
ca , and E hiopia as well as No dic coun ies (Fig 2H). On he con a y, in some coun ies less
han 20% o he in ac o es was unde p o ec ion, including Russia, Canada and Democ a ic
Republic o he Congo (Fig 2H).
Loss o o al and p o ec ed o es
The la ges absolu e o es loss alues we e encoun e ed in B azil, Canada, Uni ed S a es, Rus-
sia and Indonesia. Each o hem expe ienced o es losses o o e 100,000 km
2
du ing he
pe iod o 2000–2012 (Fig 3A). The la ges ela i e losses, o e 10% o hei o es ex en , we e
Table 4. Fo es ex en in 2000 and i s loss o e he pe iod o 2000–2012. The o es ex en and loss a e calcula ed o geog aphical egions in absolu e
[10
3
km
2
] and ela i e alues [%].
Fo es ex en [10
3
km
2
] Fo es loss [10
3
km
2
]
Region To al
ex en
P o ec ed
ex en (o o al
ex en )
In ac
ex en (o
o al
ex en )
P o ec ed in ac
ex en (o o al
in ac ex en )
To al loss
(o o al
ex en )
P o ec ed loss
(o o al
p o ec ed
ex en )
In ac loss
(o o al
in ac
ex en )
P o ec ed in ac
loss (o o al
p o ec ed in ac
ex en )
Aus alia and
Oceania
1,101 251 (23%) 256 (23%) 71 (28%) 48 (4%) 12 (5%) 4 (2%) 2 (3%)
Cen al Ame ica 1,038 221 (21%) 60 (6%) 47 (78%) 54 (5%) 11 (5%) 1 (1%) 0.4 (1%)
Eas e n Asia 2,203 193 (9%) 32 (1%) 10 (33%) 70 (3%) 5 (3%) 1 (5%) 0.4 (4%)
Eas e n Eu ope
and Cen al Asia
8,759 1,067 (12%) 2,039
(23%)
300 (15%) 357 (4%) 43 (4%) 72 (4%) 10 (3%)
La in Ame ica 9,665 3,322 (34%) 4,400
(46%)
2,415 (55%) 531 (5%) 41 (1%) 31 (1%) 9 (0.4%)
Middle and Sou h
A ica
6,768 1,054 (16%) 998 (15%) 212 (21%) 201 (3%) 22 (2%) 3 (0.3%) 0.4 (0.2%)
Middle Eas 188 12 (6%) 4 (2%) 0.4 (9%) 3 (2%) 0.1 (0.4%) 0.004
(0.1%)
0.0003 (0.1%)
No h A ica 219 15 (7%) - (-) - (-) 2 (1%) 0.2 (1%) - (-) - (-)
No h Ame ica 7,298 837 (11%) 2,408
(33%)
418 (17%) 513 (7%) 47 (6%) 152 (6%) 32 (8%)
Sou h Asia 592 84 (14%) 31 (5%) 8 (26%) 10 (2%) 1 (1%) 0.1 (0%) 0.03 (0.3%)
Sou heas e n
Asia
3,234 657 (20%) 478 (15%) 203 (42%) 271 (8%) 23 (4%) 3 (0.7%) 1 (0.3%)
Wes e n Eu ope 1,481 353 (24%) 11 (0.8%) 9 (81%) 83 (6%) 12 (3%) 0.03 (0.2%) 0.01 (0.1%)
Global 42,562 8,068 (19%) 10,717
(25%)
3,693 (34%) 2,144 (5%) 219 (3%) 269 (2.5%) 55 (1.5%)
doi:10.1371/jou nal.pone.0138918. 004
P o ec ed and In ac Fo es Loss
PLOS ONE | DOI:10.1371/jou nal.pone.0138918 Oc obe 14, 2015 8/21
ound in some coun ies in no h-wes e n and sou he n A ica, and Sou heas Asia (Fig 3B). In
hese coun ies he ela i e losses we e high in hei p o ec ed o es s oo (Fig 3D). Rela i ely
high (>5%) p o ec ed o es losses occu ed in Aus alia, Uni ed S a ed and some Eu opean
coun ies. Sou h Asia, Middle Eas and Cen al Asia showed, on he o he hand, small numbe s
o bo h absolu e and ela i e p o ec ed o es losses (Fig 3C and 3D).
Fig 2. Absolu e and ela i e o es ex en by coun y in yea 2000. Absolu e o es ex en s (le column) a e p esen ed in km
2
and ela i e o es ex en s
( igh column) a e in %. A and B) To al o es ex en ; C and D) P o ec ed o es ex en ; E and F) In ac o es ex en ; G and H) P o ec ed in ac o es ex en .
doi:10.1371/jou nal.pone.0138918.g002
P o ec ed and In ac Fo es Loss
PLOS ONE | DOI:10.1371/jou nal.pone.0138918 Oc obe 14, 2015 9/21
assessed coun ies (76%), ela i e o es loss was smalle in p o ec ed a eas han ou side hem
(anomaly a io <–0.125: Fig 5A). The p o ec ion, mo eo e , seemed o be pa icula ly e ec i e
o educing o es loss in Sou h Ame ica, Sou heas Asia and Sub-Saha an A ica (Fig 5A).
Ne e heless, he e ec o p o ec ion on o es loss was he opposi e in some coun ies in Mid-
dle A ica, Eu ope, Middle Eas and Mongolia in Asia and hus ejec s ou i s hypo hesis o a
uni ied posi i e p o ec ion e ec . On coun y scale, ou indings sugges posi i e news o B a-
zilian Amazon whe e p o ec ion was ound o be e y e ec i e in p e en ing o es loss. This is
in ag eemen wi h he ea lie esea ch indings on he B azilian Amazon [6,39]. In he case o
in ac o es , he global esul s we e simila , i.e. p o ec ion was e ec i e in 43 o 60 o assessed
coun ies (72%) (Fig 5B). La ges de ia ions o he global majo i y we e shown by Mongolia,
Nepal and Aus alia. I is wo h o acknowledge ha he PAs a e no necessa ily always aimed
o p o ec he o es s ands pe se, ins ead p o ec ion can ha e o he goals. P io o ou s udy,
he e ec i eness o p o ec ion is globally analysed by Joppa and P a [24]. Ou s udy u he
s eng hens hose indings, ye ou indings a e e ie ed da a-wise in mo e consis en manne
by using empo ally compa able o es ex en da ase s (Table 2), which only became a ailable
a e he s udy by Joppa and P a [24].
D i e s o o es loss
Impo an ly, we also explo ed he d i e s o o es loss by looking a he s a is ical linkage
be ween socio-economic indica o s and weigh ed o es loss (Table 5). Weighing he coun y
speci ic o es losses by hei o al o es ex en in ques ion allowed ou analysis o indica e he
d i e s ha a e exp essi e o he global o es ex en ins ead o ocusing on d i e s explaining
he in e -coun y a ia ion. This globally uni o m assessmen showed o es loss ha ing a
s ong connec ion wi h ag icul u al land ex en . Mo eo e ag icul u al land expansion is ecog-
nised as one o he mos impo an p ocesses causing o es loss (see e.g. [8–13]) and ou analy-
sis ag ees well wi h his, pa icula ly conce ning he p o ec ed in ac o es s (Table 5).
Howe e , i is wo h o ecognise ha ag icul u e ela ed o es loss in ol es much mo e com-
plex dynamics han ou eg ession analyses a e able o cap u e (see e.g. [40]). We u he ound
s ong connec ion be ween losses in p o ec ed and/o in ac o es s and popula ion densi y and
GDP o ully suppo ou second hypo hesis.
While se e al global scale s udies ha e ound ha highe popula ion densi y inc eases o es
loss (e.g. [41,42]), hese all epo o al o es loss. We did no ind signi ican ela ionship
be ween o al o es loss and popula ion densi y, whe eas ou indings o losses in in ac and
p o ec ed in ac o es s (Table 5) show opposi e di ec ion compa ed o hese exis ing s udies
on o al o es loss. Ou indings also seem o disag ee wi h a me a-analysis by Po e -Bolland
e al. [43]. Howe e , a di ec compa ison be ween ou esul s and hose o Po e -Bolland e al.
[43] is discou aged by he di e en scopes and app oaches in he s udies. Unlike ou s udy, he
me a-analysis assessed he e ec i eness o only 40 p o ec ed a eas in he opics. Addi ionally,
popula ion densi ies a e measu ed by di e en manne s: in ou s udy popula ion densi ies
we e coun y a e ages while he me a-analysis based i s da a on popula ion densi y in he
immedia e neighbou hood o p o ec ed a eas. Ou desi e o single s a is ical analysis including
mul iple explana o y a iables de e mined he selec ion o coun y a e ages on he popula ion
me ic o ma ch wi h he a ailable le el o he o he a iables. All in all, ou indings oge he
wi h hose by Po e -Bolland e al. [43] poin o a need o u he esea ch on he ole o popu-
la ion dynamics in o es loss.
In e ms o GDP, we did no ind a s a is ically signi ican ela ionship wi h o al o es loss
(Table 5). This inding is in line wi h some p e ious s udies [44–46], whe eas some o he s ud-
ies sugges ha lowe GDP could indica e highe o es loss (e.g. [47,48]). While he e ec o
P o ec ed and In ac Fo es Loss
PLOS ONE | DOI:10.1371/jou nal.pone.0138918 Oc obe 14, 2015 16 / 21

GDP on o es loss in p o ec ed a eas has no been assessed p e iously on a global scale, i is
e alua ed by Nagend a [49] on 56 p o ec ed a eas. He indings indica e ha he a es o land
co e clea ing do no di e signi ican ly be ween coun ies wi h di e en le els o GDP pe
capi a. We ound, in u n, high GDP pe capi a o indica e high weigh ed o es loss a e in p o-
ec ed a eas (Table 5). This ela ion be ween GDP and weigh ed o es loss on p o ec ed a eas
can possibly indica e di e en managemen s a egies in coun ies wi h a ying GDP, i.e. coun-
ies wi h highe GDP may ha e mo e e sa ile managemen s a egies such as, hose ha aim
o s eng hen biodi e si y o p o ide habi a s o speci ic species, compa ed o coun ies wi h
lowe GDP.
Ou assessmen , i.e. looking a he o es loss d i e s globally, na u ally used a he limi ed
se o socio-economic indica o s on a coa se scale. This cons ic ion on indica o s was needed
o he sake o homogeneous da a. Ano he s o y would be e ealed by mo e local and egional
s udies ha could o e ui ul insigh s o his global s udy, as is o example shown by he
global me a-analysis looking a he d i e s behind he e ec i eness o p o ec ion managemen
[50]. Ou s udy p o ides a global empi ical assessmen , weigh ed by he o al o es ex en s in
ques ion, ha can be used as a e e ence poin o mo e de ailed s udies a o he spa ial scales.
Taking u he –subna ional esul s and compa ison among da ase s
We acknowledge ha he coun y le el esul s p esen ed in his s udy do no e eal ine le el
he e ogenei y. The e o e, as an example, we explo ed how he o es loss in PAs a ies wi hin a
coun y. Taking Colombia as an example, in ou esul s we showed ha he mean a e o o es
loss wi hin all p o ec ed a eas was 0.08% pe yea . In a ine scale assessmen using he same
da ase (i.e. GFC) we ound ha he o es loss a ied among he indi idual Colombian PAs
om ze o o 1.48%, he obse ed a es being low o e la ge PAs in he Amazon and gene ally
highe o p o ec ed a eas in he Andes. The mean a e o annual o es loss pe PA was 0.12%
wi h s anda d de ia ion o 0.19%. This highligh s he impo ance o s udy u he he impac o
p o ec ion on o es loss in lowe spa ial scales.
Fu he mo e, we ca ied ou he same analysis by using an al e na i e da ase , Te a-I [51],
o explo e possible di e ences in a ailable da ase s. Te a-i da ase de ia es om GFC in i s
use o MODIS (250 m esolu ion) ins ead o Landsa (30 m esolu ion) and 2004–2014 ins ead
o 2000–2012. The ange in o es loss in he indi idual p o ec ed a eas (0% o 1.72%) was simi-
la o he inding made by GFC bu he mean annual o es loss a e wi hin he PAs was one
hi d (mean a e o 0.043% wi h s anda d de ia ion 0.14%) o he ound losses compa ed o
GFC da ase indings. This compa ison e lec s di e ences in da a ype, me hodology and
pe iod bu also indica es he unce ain y associa ed wi h hese da ase s and, hus calls o u -
he compa isons be ween he widely used da ase s.
Me hodological limi a ions and ways o wa d
I is impo an o expand he discussion owa ds he limi a ions c ea ed by he used da ase s
ha also ou s udy is na u ally subjec ed o. Fo example, o he o es co e and loss we used
he GFC da ase [5], which can gi e de ia ing esul s compa ed o o he da ase s mainly due o
he di e ences in me hods such as: de ini ion o o es , sepa a ion o an h opogenic om na u-
ally occu ing o es loss, and handling o a mosphe ic dis o ions (e.g. cloud co e ) in map-
ping he o es co e . Al hough, di ing deepe in o hese di e ences and he e ec s hey ha e
on he indings is beyond he scope o his analysis, i would o e a ui ul opic o a compa a-
i e s udy. In addi ion, Lee [52] aises a conce n on he IFL da ase [20] ha i s de ini ion o
in ac ness o o es may no cap u e he socio-ecological a ia ion be ween di e en geog aphi-
cal egions and may hus in oduce inaccu acies in coun y speci ic analyses. In e -compa ison
P o ec ed and In ac Fo es Loss
PLOS ONE | DOI:10.1371/jou nal.pone.0138918 Oc obe 14, 2015 17 / 21
o global scale da ase s as well as mo e de ailed analyses a he coun y/ egional le el would be
needed o un a el he s a e o global classi ica ion o IFLs.
I should be no ed ha ou s udy ocused on o es loss, o mo e p ecisely on s and- eplac-
ing dis u bance, igno ing he possible o es gain. This ocus ga e us mo e space o explo e he
loss o o iginal habi a s wi hin p o ec ed a eas and in ac o es landscapes and he d i e s
behind hese losses. A compa ison o ou indings o indings ha would be based on he ne
o es loss, i.e. a loss assessmen including he plan ing o o es s and na u al o es egene a-
ion (see [2,5] and e e ences he ein o u he in o ma ion), would allow acking o he o -
es canopy co e pe se.
Ou es ima ions on he e ec i eness o p o ec ion o educe o es loss wi hin PAs com-
pa ed o he o es loss ou side hose a eas p o ide use ul in o ma ion he p o ec ion e ec a
he coun y le el. A he coun y le el, we we e ne e heless no able o assess he leakage e ec
o indi idual p o ec ed a eas, as done by e.g. Oli ei a e al [53]. As highligh ed by Ewe s and
Rod igues [54] a compa ison o o es loss in he icini y o and inside p o ec ed a eas is essen-
ial in unde s anding he e ec s o o es p o ec ion on a local scale. I is also wo hwhile o ec-
ognise ha o es ed a eas migh ha e been p o ec ed o o he easons han p o ec he o es
s ands pe se. The global da ase s used in his s udy could be u he employed o his kind o
assessmen , compa ing he leakage e ec wi hin and among egions and u he mo e ex ending
i all he way o global le el. Mos impo an ly, global da a on o es managemen goals pe
each p o ec ion a ea would be highly needed o suppo mo e de ailed assessmen on he
impac o p o ec ion on o es loss.
Conclusions
Ou global scale analysis indica es ha subs an ial a eas o p o ec ed (3%) and in ac o es s
(2.5%) we e los o e he pas decade, and la ge pa o ha loss occu ed in coun ies wi h
well-es ablished p o ec ion (e.g. in Aus alia and Oceania, and No h Ame ica). In e ms o
compa ison be ween o es loss inside and ou side o PAs and IFLs we could no con i m ou
i s hypo hesis, which s a ed ha o es loss would always be less mani es ed wi hin PAs han
ou side o hem. Al hough he ela i e o es loss was smalle inside he PAs and p o ec ed IFLs
in global a e ages and in majo i y o coun ies, he e we e nume ous coun ies ha showed an
opposi e e ec . Also ou second hypo hesis was no en i ely con i med: while popula ion size
and o he indica o s we e able o explain di e ences in weigh ed global o es loss wi hin PAs
and IFLs, a la ge pa o he a ia ion emained unexplained. Mo eo e , he s a is ical associa-
ions did no always co espond wi h ou causal unde s anding o o es loss p ocesses. Ou
indings hus highligh ha p o ec ion o o es ed a eas does no always gua an ee a lowe a e
o o es loss. The e is, indeed, a high geog aphical a ia ion in he e ec i eness o p o ec ion
agains o es loss, a a ia ion ha is p obably a leas o some ex en explained by coun ies’
di e en means and in ensi ies o comba agains o es loss.
Suppo ing In o ma ion
S1 Appendix. C oss-co ela ion ables. C oss-co ela ion ables o socio-economic indices in
cases o di e en o es loss ca ego ies.
(PDF)
S2 Appendix. Compa ison o ou indings o p e ious s udies: o es ex en and loss.
(PDF)
S3 Appendix. Weigh ed Leas Squa es (WLS) eg ession analysis s eps.
(PDF)
P o ec ed and In ac Fo es Loss
PLOS ONE | DOI:10.1371/jou nal.pone.0138918 Oc obe 14, 2015 18 / 21
S1 Table. Socio-economic da a a coun y le el. Socio-economic da a used in he WLS eg es-
sion analysis.
(XLSX)
S2 Table. Coun y le el esul s. Fo es ex en and o es loss esul s p esen ed a coun y le el.
(XLSX)
Acknowledgmen s
We a e g a e ul o CSC—IT Cen e o Science o p o iding he compu a ional acili ies o
he analyses. We also hank he e iewe s o hei cons uc i e commen s and P o . Olli Va is,
D . Joseph Guillaume, and o he membe s o ou esea ch eam a Aal o Uni e si y o hei
encou agemen s and ui ul discussions.
Au ho Con ibu ions
Concei ed and designed he expe imen s: MH MK M. Mulligan TAR. Pe o med he expe i-
men s: MH MK M. Mulligan MJ TAR. Analyzed he da a: MH MK M. Makkonen M. Mulligan
MJ TAR. Con ibu ed eagen s/ma e ials/analysis ools: MH MK M. Makkonen M. Mulligan
MJ TAR. W o e he pape : MH MK M. Makkonen M. Mulligan PHV TAR.
Re e ences
1. Bonan GB (2008) Fo es s and Clima e Change: Fo cings, Feedbacks, and he Clima e Bene i s o Fo -
es s. Science 320: 1444–1449 doi: 10.1126/science.1155121 PMID: 18556546
2. FAO (2010) Global o es esou ce assessmen 2010. Rome: Food and Ag cul u e O ganisa ion o he
Uni ed Na ions.
3. Gibson L, Lee TM, Koh LP (2011) P ima y o es s a e i eplaceable o sus aining opical biodi e si y.
Na u e 478: 378–381. doi: 10.1038/na u e10425 PMID: 21918513
4. Pan Y, Bi dsey RA, Fang J, Hough on R, Kauppi PE, Ku z WA, e al. (2011) A La ge and Pe sis en Ca -
bon Sink in he Wo ld’s Fo es s. Science 333: 988–993 doi: 10.1126/science.1201609 PMID:
21764754
5. Hansen MC, Po apo PV, Moo e R, Hanche M, Tu ubano a SA, Tyuka ina A, e al. (2013) High- eso-
lu ion global maps o he 21s -cen u y o es co e change. Science 342: 850–853. doi: 10.1126/
science.1244693 PMID: 24233722
6. Neps ad D, Soa es-Filho BS, Me y F, Lima A, Mou inho P, Ca e J, e al. (2009) The End o De o es a-
ion in he B azilian Amazon. Science 326: 1350–1351. doi: 10.1126/science.1182108 PMID:
19965742
7. Goda J, Ga dne TA, Tizado EJ, Pacheco P (2014) Ac o -speci ic con ibu ions o he de o es a ion
slowdown in he B azilian Amazon. P oceedings o he Na ional Academy o Sciences 111: 15591–
15596.
8. DeF ies RS, Rudel T, U ia e M, Hansen M (2010) De o es a ion d i en by u ban popula ion g ow h and
ag icul u al ade in he wen y- i s cen u y. Na u e Geoscience 3: 178–181.
9. Fe e i-Gallon K, Busch J (2014) Wha d i es de o es a ion and wha s ops i ? A me a-analysis o spa-
ially explici econome ic s udies. Washing on, DC: Cen e o Global De elopmen .
10. Geis HJ, Lambin EF (2002) P oxima e causes and unde lying d i ing o ces o opical de o es a ion.
BioScience 52: 143–150.
11. Hosonuma N, He old M, De Sy V, De F ies RS, B ockhaus M, Ve cho L, e al. (2012) An assessmen
o de o es a ion and o es deg ada ion d i e s in de eloping coun ies. En i onmen al Resea ch Le -
e s 7.
12. Rudel TK (2005) T opical Fo es s. Regional Pa hs o Des uc ion and Regene a ion in he La e Twen i-
e h Cen u y. New Yo k: Columbia Uni e si y P ess.
13. Rudel TK, DeF ies R, Asne GP, Lau ance WF (2009) Changing d i e s o de o es a ion and new
oppo uni ies o conse a ion. Conse a ion Biology 23: 1396–1405. doi: 10.1111/j.1523-1739.2009.
01332.x PMID: 20078640
P o ec ed and In ac Fo es Loss
PLOS ONE | DOI:10.1371/jou nal.pone.0138918 Oc obe 14, 2015 19 / 21
14. Rudel TK (2007) Changing agen s o de o es a ion: F om s a e-ini ia ed o en e p ise d i en p ocesses,
1970–2000. Land Use Policy 24: 35–41.
15. Ju e-Bignoli D, Bu gess ND, Bingham H, Belle EMS, de Lima MG, Deguigne M, e al. (2014) P o ec ed
Plane Repo 2014. Camb idge, UK: UNEP-WCMC.
16. Naugh on-T e es L, Holland MB, B andon K (2005) The ole o p o ec ed a eas in conse ing biodi e -
si y and sus aining local li elihoods. Annu Re En i on Resou 30: 219–252.
17. Radelo VC, Beaud y F, B ooks TM, Bu sic V, Dubinin M, Kuemme le T, e al. (2013) Ho momen s o
biodi e si y conse a ion. Conse a ion Le e s 6: 58–65.
18. Schmi CB, Bu gess ND, Coad L, Beloku o A, Besançon C, Bois obe L, e al. (2009) Global analysis
o he p o ec ion s a us o he wo ld’s o es s. Biological Conse a ion 142: 2122–2130.
19. Dudley N (2008) Guidelines o Applying P o ec ed A ea Managemen Ca ego ies. Gland, Swi ze land:
IUCN.
20. Po apo P, Ya oshenko A, Tu ubano a S, Dubinin M, Laes adius L, Thies C, e al. (2008) Mapping he
wo ld's in ac o es landscapes by emo e sensing. Ecology and Socie y 13.
21. Joppa LN, P a A (2010) Reassessing he o es impac s o p o ec ion: The challenge o non andom
loca ion and a co ec i e me hod. Annals o he New Yo k Academy o Sciences 1185: 135–149. doi:
10.1111/j.1749-6632.2009.05162.x PMID: 20146766
22. Joppa LN, P a A (2009) High and a : Biases in he loca ion o p o ec ed a eas. PloS ONE 4: e8253–
e8273.
23. Ve bu g PH, O e ma s KP, Huigen MGA, de G oo WT, Veldkamp A (2006) Analysis o he e ec s o
land use change on p o ec ed a eas in he Philippines. Applied Geog aphy 26: 153–173.
24. Joppa LN, P a A (2011) Global p o ec ed a ea impac s. P oceedings o he Royal Socie y B 278:
1633–1638. doi: 10.1098/ spb.2010.1713 PMID: 21084351
25. DeF ies R, Hansen A, New on AC, Hansen MC (2005) Inc easing isola ion o p o ec ed a eas in opical
o es s o e he pas wen y yea s. Ecological Applica ions 15: 19–26.
26. Maio ano L, Falcucci A, Boi ani L (2008) Size-dependen esis ance o p o ec ed a eas o land-use
change. P oc R Soc B June 7, 2008 275 1640 1297–1304 275: 1297–1304.
27. B ook BW, Sodhi NS, B adshaw CJA (2008) Syne gies among ex inc ion d i e s unde global change.
T ends in Ecology & E olu ion 23: 453–460.
28. Tu ne IM (1996) Species loss in agmen s o opical ain o es : A e iew o he e idence. Jou nal o
Applied Ecology 33: 200–209.
29. Ma gono BA, Po apo PV, Tu ubano a S, S ollle F, Hansen MC (2014) P ima y o es co e loss in
Indonesia o e 2000–2012.
30. IUCN, UNEP-WCMC (2014) The Wo ld Da abase on P o ec ed A eas (WDPA). In: UNEP-WCMC, edi-
o . Camb idge, UK. A ailable: www.p o ec edplane .ne . Accessed 7 July 2014.
31. Bonda WJ, Keeleyb JE (2005) Fi e as a global ‘he bi o e’: he ecology and e olu ion o lammable eco-
sys ems. T ends in Ecology and E olu ion 20: 387–394. PMID: 16701401
32. Logan JA, Régniè e J, Powell JA (2003) Assessing he impac s o global wa ming on o es pes
dynamics. F on ie s in Ecology and he En i onmen 1: 130–137.
33. Po apo P, Hansen MC, S ehman SV, Lo eland TR, Pi man K (2008) Combining MODIS and Landsa
image y o es ima e and map bo eal o es co e loss. Remo e Sensing o En i onmen 112: 3708–
3719.
34. UN (2014) Wo ld Popula ion P ospec s: The 2012 Re ision. UN Popula ion Di ision. A ailable: h p://
esa.un.o g/unpd/wpp/Excel-Da a/popula ion.h m. Accessed 7 Augus 2014.
35. Wo ld Bank (2014) Wo ld Bank De elopmen Indica o s. Wo ld Bank. A ailable: h p://da a.wo ldbank.
o g/indica o . Accessed 7 Augus 2014.
36. T anspa ency In e na ional (2010) Co up ion Pe cep ion Index. T anspa ency In e na ional. A ailable:
h p://www. anspa ency.o g/cpi2010/. Accessed 14 June 2014.
37. Ma shall MG, Gu T (2014) Poli y IV P ojec , Poli ical Regime Cha ac e is ics and T ansi ions, 1800–
2013. Cen e o Sys ema ic Peace. A ailable: h p://www.sys emicpeace.o g/insc da a.h ml.
Accessed 14 June 2014.
38. UNDP (2014) Human De elopmen Index 2012. Uni ed Na ions De elopmen P og amme. A ailable:
h ps://da a.undp.o g/da ase /Table-1-Human-De elopmen -Index-and-i s-componen s/wxub-qc5k.
Accessed 7 Augus 2014.
39. Nol e C, Ag awal A, Sil ius KM, Soa es-Filho BS (2013) Go e nance egime and loca ion in luence
a oided de o es a ion success o p o ec ed a eas in he B azilian Amazon. PNAS 110: 4956–4961.
doi: 10.1073/pnas.1214786110 PMID: 23479648
P o ec ed and In ac Fo es Loss
PLOS ONE | DOI:10.1371/jou nal.pone.0138918 Oc obe 14, 2015 20 / 21
40. Pagnu i C, Bauch CT, Anand M (2013) Ou look on a Wo ldwide Fo es T ansi ion. PLoS ONE 8:
e75890. doi: 10.1371/jou nal.pone.0075890 PMID: 24130750
41. Ma he AS, Needle CL (2000) The ela ionships o popula ion and o es ends. Geog aphical Jou nal
166: 2–13.
42. Ma he AS, Needle CL, Fai bai n J (1998) The human d i e s o global land co e change: he case o
o es s. Hyd ological P ocesses 12: 1983–1994.
43. Po e -Bolland L, Ellis EA, Gua igua a MR, Ruiz-Mallen I, Neg e e-Yankele ich S, Reyes-Ga cia
(2012) Communi y managed o es s and o es p o ec ed a eas: An assessmen o hei conse a ion
e ec i eness ac oss opics. Fo es Ecology and Managemen 268: 6–17.
44. Dame e O, Delaco e P (2012) On he economic ac o s o de o es a ion: Wha can we lea n om quan-
ile analysis? Economic Modelling 29: 2427–2434.
45. Koop G, Tole L (1999) Is he e an en i onmen al Kuzne s cu e o de o es a ion? Jou nal o De elop-
men Economics 58: 231–244.
46. Tole L (1998) Sou ces o De o es a ion in T opical De eloping Coun ies. En i onmen al Managemen
22: 19–33. PMID: 9419282
47. Ewe s RM (2006) In e ac ion e ec s be ween economic de elopmen and o es co e de e mine de o -
es a ion a es. Global En i onmen al Change 16: 161–169.
48. Smi h RJ, Mui RDJ, Walpole MJ, Balm o d A, Leade -Williams N (2003) Go e nance and he loss o
biodi e si y. Na u e 426: 67–70. PMID: 14603318
49. Nagend a H (2008) Do pa ks wo k? Impac o p o ec ed a eas on land co e clea ing. Ambio 37: 330–
337. PMID: 18828277
50. Le e ing on F, Lemos Cos a KL, Pa ese P, Lisle A, Hockings M (2010) A Global Analysis o P o ec ed
A ea Managemen E ec i eness. En i onmen al Managemen 46: 685–698. doi: 10.1007/s00267-
010-9564-5 PMID: 20859627
51. Reymondin L, Ja is A, Pe ez-U ibe A, Tou al J, A go e K, Rebe ez J, e al. (2012) A me hodology o
nea eal- ime moni o ing o habi a change a con inen al scales using MODIS-NDVI and TRMM. A ail-
able: www. e a-i.o g. Accessed 16 Decembe 2014
52. Lee PG (2009) Cau ion agains using in ac o es -landscapes da a a egional scales. Ecology and
Socie y 14: 1. A ailable: h p://www.ecologyandsocie y.o g/ ol14/iss11/ esp11/.
53. Oli ei a PJC, Asne GP, Knapp DE, Almeyda A, Gal án-Gildemeis e R, Keene S, e al. (2007) Land-
Use Alloca ion P o ec s he Pe u ian Amazon. Science 317: 1233–1236. PMID: 17690260
54. Ewe s RM, Rod igues ASL (2008) Es ima es o ese e e ec i eness a e con ounded by leakage.
T ends in Ecology and E olu ion 23: 113–116. doi: 10.1016/j. ee.2007.11.008 PMID: 18280005
P o ec ed and In ac Fo es Loss
PLOS ONE | DOI:10.1371/jou nal.pone.0138918 Oc obe 14, 2015 21 / 21