ArticlePDF Available

Abstract and Figures

Distinguising between natural forests from exotic tree plantations is essential to get an accurate picture of the world’s state of forests. Most exotic tree plantations support lower levels of biodiversity and have less potential for ecosystem services supply than natural forests, and differencing them is still a challenge using standard tools. We use a novel approach in south-central of Chile to differentiate tree cover dynamics among natural forests and exotic tree plantations. Chile has one of the world's most competitive forestry industry and the region is a global biodiversity hotspot. Our collaborative visual interpretation method combined a global database of tree cover change, remote sensing from high-resolution satellite images and expert knowledge. By distinguishing exotic tree plantation and natural forest loss, we fit spatially explicit models to estimate tree-cover loss across 40 millions of ha between 2000-2016. We were able to distinguish natural forests from exotic tree plantations with an overall accuracy of 99% and predicted forest loss. Total tree cover loss was continuous over time, and the disaggregation revealed that 1,549,909 ha of tree plantations were lost (mean = 96,869 ha/year), while 206,142 ha corresponded to natural forest loss (mean = 12,884 ha/year). Mostly of tree plantations lost returned to be plantation (51%). Natural forests were converted mainly (75%) to transitional land covers (e.g. shrubland, bare land, grassland), and an important proportion of these may finish as tree plantation. This replacement may undermine objectives of increased carbon storage and biodiversity. Tree planting as a solution has gained increased attention in recent years with ambitious commitments to mitigate the effects of climate change. However, negative outcomes for the environment could result if strategies incentivize the replacement of natural forests into other land covers. Initiatives to reduce carbon emissions should encourage differentiating natural forests from exotic tree plantations and pay more attention on protecting and managing sustainably the former.
This content is subject to copyright. Terms and conditions apply.
Environ. Res. Lett. 15 (2020) 124055
Environmental Research Letters
29 June 2020
19 October 2020
13 November 2020
14 December 2020
Original content from
this work may be used
under the terms of the
Creative Commons
Attribution 4.0 licence.
Any further distribution
of this work must
maintain attribution to
the author(s) and the title
of the work, journal
citation and DOI.
Natural forests loss and tree plantations: large-scale tree cover
loss differentiation in a threatened biodiversity hotspot
Adison Altamirano1,2, Alejandro Miranda1,3, Paul Aplin4, Jaime Carrasco5,6, Germ´
an Catal´
Luis Cayuela8, Taryn Fuentes-Castillo9, Angela Hern´
andez10, María J Martínez-Harms11,
Franco Peluso12, Marco Prado1, Rosa Reyes-Riveros1,7, Tracy Van Holt13, Cristian Vergara14,
Carlos Zamorano-Elgueta15 and Carlos Di Bella16,17
1Landscape Ecology and Conservation Lab, Facultad de Ciencias Agropecuarias y Forestales, Universidad de La Frontera, Temuco, Chile
2Butamallin Research Center for Global Change, Facultad de Ciencias Agropecuarias y Forestales, Universidad de La Frontera, Temuco,
3Center for Climate and Resilience Research (CR2), Universidad de Chile, Santiago, Chile
4Department of Geography and Geology, Edge Hill University, Ormskirk, United Kingdom
5Industrial Engineering Department, Universidad de Chile, Santiago, Chile
6Complex Engineering System Institute—ISCI, Santiago, Chile
7Doctorado en Ciencias Agroalimentarias y Medioambiente, Facultad de Ciencias Agropecuarias y Forestales, Universidad de La
Frontera, Temuco, Chile
8Department of Biology, Geology, Physics and Inorganic Chemistry, Universidad Rey Juan Carlos, Madrid, Spain
9Instituto de Geografía, Facultad de Historia, Geografía y Ciencia Política, Pontificia Universidad Cat´
olica de Chile, Santiago, Chile
10 Centro de Investigaci´
on en Ecosistemas de la Patagonia (CIEP), Coyhaique, Chile
11 Center for Applied Ecology and Sustainability (CAPES), Pontificia Universidad Catolica de Chile, Santiago, Chile
12 Instituto de Clima y Agua, Instituto Nacional de Tecnología Agropecuaria (CIRN-INTA Castelar), Hurlingham, Buenos Aires,
13 Center for Sustainable Business, Leonard N. Stern School of Business, New York University, New York, United States of America
14 Laboratorio de Planificaci´
on Territorial, Departamento de Ciencias Ambientales, Facultad de Recursos Naturales—Universidad
olica de Temuco, Coyhaique, Chile
15 Universidad de Aysen, Coyhaique, Chile
16 Departamento de M´
etodos Cuantitativos, Universidad de Buenos Aires, Facultad de Agronomía, Buenos Aires, Argentina
17 Consejo Nacional de Investigaciones Científicas y T´
ecnicas, Argentina
Keywords: remote sensing, augmented visual interpretation, land use and land cover change, tree cover, forest plantation
Supplementary material for this article is available online
Distinguishing between natural forests from exotic tree plantations is essential to get an accurate
picture of the world’s state of forests. Most exotic tree plantations support lower levels of
biodiversity and have less potential for ecosystem services supply than natural forests, and
differencing them is still a challenge using standard tools. We use a novel approach in south-central
of Chile to differentiate tree cover dynamics among natural forests and exotic tree plantations.
Chile has one of the world’s most competitive forestry industry and the region is a global
biodiversity hotspot. Our collaborative visual interpretation method combined a global database of
tree cover change, remote sensing from high-resolution satellite images and expert knowledge. By
distinguishing exotic tree plantation and natural forest loss, we fit spatially explicit models to
estimate tree-cover loss across 40 millions of ha between 2000 and 2016. We were able to
distinguish natural forests from exotic tree plantations with an overall accuracy of 99% and
predicted forest loss. Total tree cover loss was continuous over time, and the disaggregation
revealed that 1 549 909 ha of tree plantations were lost (mean =96869 ha year1), while
206 142 ha corresponded to natural forest loss (mean =12 884 ha year1). Mostly of tree
plantations lost returned to be plantation (51%). Natural forests were converted mainly (75%) to
transitional land covers (e.g. shrubland, bare land, grassland), and an important proportion of
these may finish as tree plantation. This replacement may undermine objectives of increased
© 2020 The Author(s). Published by IOP Publishing Ltd
Environ. Res. Lett. 15 (2020) 124055 A Altamirano et al
carbon storage and biodiversity. Tree planting as a solution has gained increased attention in recen
years with ambitious commitments to mitigate the effects of climate change. However, negative
outcomes for the environment could result if strategies incentivize the replacement of natural
forests into other land covers. Initiatives to reduce carbon emissions should encourage
differentiating natural forests from exotic tree plantations and pay more attention on protecting
and managing sustainably the former.
1. Introduction
Differentiating natural forests from exotic tree plant-
ations at large scales represents a global relevant issue
because different tree covers may produce disparate
estimations of changes in local biodiversity and in rel-
evant ecosystem services such as climate regulation,
carbon storage, and water supply (Hall et al 2012,
Van Holt et al 2012,2016, Viña et al 2016, Lewis et al
2019). However, this still constitutes a technical chal-
lenge for the remote-sensing communities (Zhao et al
2016, Curtis et al 2018).
Evidence has demonstrated that exotic tree plant-
ations are rather distinct from natural forests in their
role in terms of biodiversity conservation, ecosys-
tem services, and the social impacts that they provide
(Reyes et al 2015, Naudts et al 2016, Jones et al
2017, Martínez-Jauregui et al 2018, Lewis et al 2019).
Therefore, these tree cover types need to be carefully
differentiated, especially in areas with high conserva-
tion values, where it has many implications for biod-
iversity and human wellbeing (Newbold et al 2016).
Current policy discourses related to forest con-
servation, and particularly to restoration, use or
imply the forest definition by the United National
Food and Agricultural Organization (FAO), which
aggregates natural forests and exotic tree plantations
(FAO 2010). Under this definition, exotic tree planta-
tions are under intensified forestry management and
some authors point out that could be classified as
‘tree farms’ (Van Holt and Putz 2017, Curtis et al
2018). However, misclassification of natural forests
and exotic tree plantations could bring misinterpret-
ation in environmental policy and social impact eval-
uation (Van Holt and Putz 2017, Hua et al 2018), as
tree plantations are typically subjected to intensified
forestry management practices that cause environ-
mental impacts similar to those produced by intensive
agriculture (Karp et al 2012, Naudts et al 2016, Lewis
et al 2019, Heilmayr et al 2020, Osuri et al 2020).
Under the FAO definition (2015), not all forests
contribute to climate change mitigation (Naudts et al
2016). Planting trees, and particularly some con-
ifer species for fast growth tree plantations, is not
enough to stave off global warming (Lewis et al 2019).
Given the current global challenge of forest restora-
tion (Chazdon and Brancalion 2019) it is paramount
to accurately discriminate between natural forests
and tree plantations. For instance, China reported a
significant increase in forest cover in around 1.6% of
its territory (Viña et al 2016). However, native forests
are not returning and this forest recovery was mostly
due to exotic tree plantations (Van Holt and Putz
2017). Moreover, while net tree cover in southwest-
ern China grew by 32% (2000–2015), this increase
was mainly due to the conversion of croplands to tree
plantations, but tree plantations also displaced native
forests with a gross loss of 6.6% (Hua et al 2018).
The biggest effort to assess tree cover change is
the global database developed by Hansen et al (2013),
who mapped annual global tree cover loss and gains
from 2001 to 2018 using Landsat satellite program.
Based on this dataset, several studies have assessed
the global and local forest changes in a number of
places/countries (Viña et al 2016, Heilmayr et al 2016,
Potapov et al 2017, Curtis et al 2018, Hua et al
2018). At a global scale, the main driver of tree cover
loss is associated to permanent land use change for
commodity production like forestry and agriculture
(Curtis et al 2018). But not considering the natural
or productive character of these forests can lead to
a substantial misestimation of the natural forest loss
(Tropek et al 2014), compromising its value for local
and global policy decisions.
Identification of different types of forests using
Hansen et al (2013) dataset is mainly based on
tree cover percentage (Tyukavina et al 2018) or in
combination with forest height maps derived from
LIDAR images (Yu et al 2020). However, those cri-
teria do not differentiate natural forests from tree
plantations. In this work, we use a novel approach
to illustrate the importance of accurately discriminat-
ing natural forests from tree plantations when quan-
tifying tree cover losses. We use a method which can
be applied to large-scale land cover monitoring. We
use a collaborative augmented visual interpretation
method that uses the Hansen et al (2013) database,
the Google Earth-Engine platform, high-resolution
satellite images, and expert knowledge through net-
working collaboration (figure 1).
The method is tested in south-central Chile,
which is one of the global leaders in pulp produc-
tion from exotic tree plantations (Cubbage et al 2007),
and also classified as a global biodiversity hotspot
(Mittermeier et al 2004). This area harbors the last
confined and endangered sclerophyllous and tem-
perate forests of South America. Our method was
applied for the period 2000–2016. We discuss the
Environ. Res. Lett. 15 (2020) 124055 A Altamirano et al
Figure 1. Workflow of the methodological framework.
implications for global and large-scale tree cover
monitoring, by providing an accurate, cost-efficient,
and replicable tool, which can be useful for future
biodiversity conservation and climate mitigation
2. Methods
2.1. Study area
The Chilean biodiversity hotspot, also called ‘Chilean
winter rainfall–Valdivian forests’ (25–47S) (Arroyo
et al 2004) covers about 400 000 km2(figure 2). It
comprises half of the temperate forests in the south-
ern hemisphere, but also suffers the greatest land-use-
change pressure in the country due to the high con-
centration of economic activities in Chile (Miranda
et al 2017). This area comprises 79% of the coun-
try’s urban and industrial zones, 94% of its agri-
culture, and 98.7% of the total exotic tree planta-
tion extent (mostly Pinus radiata and Eucalyptus spp.)
(CONAF 2011). In this hotspot, natural forests cover
approximately 9.5 millions of ha (Zhao et al 2016),
which are distributed mainly from 33S southwards
(figure 2(A)). In the country the last official figure of
exotic forest plantations is equivalent to 3.1 millions
of ha (CONAF 2019).
2.2. Data source of tree cover loss
We use the updated tree cover loss database
developed by Hansen et al (2013) (available at This global database
has a high spatial resolution (30 m) of annual tree
cover change. We used the tree cover loss product
(patches lost) for each year between 2000 and 2016.
We applied an image filtering technique to elimin-
ate tree loss patches smaller than 0.27 ha, and the
eight-cells neighborhood rule (Mcgarigal et al 2012).
We use the defined forest loss as a ‘stand replace-
ment disturbance’ (Hansen et al 2013), meaning the
removal or mortality of all tree cover in a Landsat
pixel with more than 50% of tree crown cover.
2.3. The collaborative survey
We performed a random sampling through a sur-
vey to differentiate tree cover loss between natural
forest and exotic tree plantation, and identify the
contributors of change (i.e. the land cover type after
forest loss), for the whole period. The sampling con-
sidered a total of 2623 points in patches of tree cover
loss (figure 1(C)). To get a representative sample
of each loss patch, five points were randomly dis-
tributed in each accumulated forest loss patch for
the whole period which was constrained to the
patch size (minimum distance among sample points
was 30 m). Therefore a total of 1219 patches were
The total sample points were distributed between
ten local experts who were selected given their expert-
ise on remote sensing and on land use/land cover
monitoring. Each expert received a minimum of 250
sample points and all instructions and files to under-
take the survey.
Environ. Res. Lett. 15 (2020) 124055 A Altamirano et al
Figure 2. Map of (A) the current distribution of natural forest and exotic tree plantations (Zhao et al 2016), (B) whole tree cover
loss by Hansen et al (2013) updated to 2016, and (C) the distribution of sampling points on tree cover loss patches in the study
area. Plots represent latitudinal and longitudinal density distribution of land cover-types in each map.
Based on their experience each local expert iden-
tified at each sample point the land-cover type prior
to tree cover loss (i.e. natural forest and tree planta-
tion), but was not possible to distinguish the species
of origin. After that, at the same sample point, the
expert identified the land-cover type in the last year
of the period (2016), among the following options:
natural forest, exotic tree plantation, cropland, grass-
land, settlement, wetland, shrubland, bare land, and
other land (appendix 1, which is available online
Local experts also identified errors associated to no
presence of tree cover loss in Hansen et al (2013)
database, and all these points were discarded from
the final analysis and the spatial and temporal vari-
ations of cover loss.
We tested the sensitivity of our results to assess
the variation of land cover identification according
to the sample size (appendix 3). Furthermore, we
assessed the accuracy of the local experts to differen-
tiate between ‘natural forest’ and ‘tree plantation. We
collected a total of 450 independent points from field-
work on natural forest and tree plantation. We ran-
domly selected 100 of these field points (50% of each
land cover). All local experts received the same sample
points and identified the land-cover type on each
one. We compared the relationship between known
reference data (field points) and the results of the
local expert identification for accuracy estimation.
Additionally, we measured agreement among experts
using Fleiss’ kappa (K), which measures reliability
among a group of experts. We calculated Kusing the
R software package irr (Gamer et al 2014).
The survey was designed through Augmen-
ted Visual Interpretation and implemented in
‘Open Foris Collect’ software (www.openforis.
org/tools/collect.html) (Bey et al 2016). This platform
is open source and provides a flexible solution for data
management, allowing full customization of survey
structure, variables, and data checks. To apply the sur-
vey, we used ‘Collect Earth’, which is a free and open
source tool for land monitoring that uses Google
Earth and Google Earth Engine, and was developed
by FAO (Bey et al 2016). For land cover identification
of each local expert we mostly used historical Google
Earth images but in years with no availability of these
images we used Landsat images.
2.4. Natural forest and tree plantation losses
We estimated the tree cover loss area in the whole hot-
spot distinguishing between tree plantation and nat-
ural forest loss for the entire period using boosted
regression trees. For the training data set we use one
sample per patch (discarding all duplicate samples
in a patch) which is equivalent to a total of 1219
samples. Then, on that way the samples are relatively
spatially independent. The decision trees were gen-
erated first by linking the potential explanatory vari-
ables to the response variables. As response variable
we used the binary variable of forest loss (natural
forest =1, and tree plantation =0). We specified
three parameters to fit the model: tree complexity
(5), learning rate (0.01), and bag fraction (0.8). We
constructed a set of environmental variables maps as
explanatory variables from spatially-explicit data on
geography, landscape characteristics, and tree cover
loss patch metrics (appendix 2). The environmental
explanatory variables were: latitude, longitude, elev-
ation, slope, distance to cities, distance to villages,
distance to roads, and different patch-lost met-
rics (year loss, area, perimeter, area/perimeter, and
perimeter/area). We thus examined the correlation
matrix of all these explanatory variables and excluded
Environ. Res. Lett. 15 (2020) 124055 A Altamirano et al
Figure 3. Annual tree cover loss in the Chilean hotspot according to Hansen et al (2013) updated database.
those that were highly correlated (|r| > 0.6) to avoid
This technique generates many regression trees
that are combined into one ultimate regression tree
model, boosting the ultimate model’s accuracy and
predictive performance (Elith et al 2008). After train-
ing the model, a validation accuracy score estimates
the performance of the model on an independent
dataset (20%). When the dataset of observations are
divided into kdisjoint subsamples (or folds), then is
taken a group as a hold out or test dataset and the
remaining groups as a training dataset, this proced-
ure is known as K-fold cross-validation. In our study,
we adopted the latter procedure (with K=5) to val-
idate, to avoid overfitting and to estimate the aver-
age classification. Then, the majority of model pre-
dictions were applied across all study area.
We calculated usual measures of model perform-
ance as the Area under the ROC curve (AUC), the cor-
relation between the observed and predicted values,
sensibility and sensitivity (Shabani et al 2016). Sensit-
ivity is the percentage of positive observations that are
correctly classified whereas sensibility is the percent-
age of negative observations that are correctly identi-
fied. AUC assess the overall accuracy of the classifier’s
performance. AUC value near 0.5 means that the pre-
dictive ability of the model is completely random and
a value of 1.0 represents a perfect prediction without
We assessed the uncertainty of the model estima-
tions across the latitudinal gradient of the study area.
The model estimates the relative influence of each
explanatory variable. We thus chose those explanat-
ory variables with 4% of influence in BTR mod-
els (a strong relationship with the response variable).
The influence was based on the number of times a
variable was selected for splitting, weighted by the
squared improvement to the model as a result of each
split, and averaged over all trees (Elith et al 2008). We
fitted the model using the dismo package implemen-
ted in R(R Development Core Team 2016).
3. Results
Total tree cover loss in the entire Chilean biodiversity
hotspot for the period 2000–2016 was 1 756 052 ha
(109 753 ha year1in average). Overall, there was
a continuous increment in tree cover loss dur-
ing the whole period, ranging from approximately
70 000 ha year1in 2001–160 000 ha year1in
2016 (figures 2(B) and 3). The main contributors
of change (i.e. the land-cover types after tree cover
loss) were tree plantation (44%), bare land (36%),
and shrubland (11%).
3.1. Land cover classification accuracy assessment
The global assessment of local experts showed high
accuracy values for differentiation between natural
forest and exotic tree plantation, as well as for the
other land covers. Mean global accuracy from the
independent field samples applied to the ten local
experts was 99% (natural forest and exotic tree plant-
ation rised 99% and 98% respectively). This res-
ult was consistent with Fleiss’ Kappa analysis where
K=0.95. Sensitivity analysis showed that the influ-
ence of sample size variation on land-cover type iden-
tification prior and after the tree cover loss decreases
as sample size increases (appendix 3).
3.2. Tree cover losses
Based on the sampling, most of tree cover loss
turned to exotic tree plantations in the whole time
period (2000–2016), explaining 85% of the tree
patches lost, while only 15% of the samples were nat-
ural forests loss. These tree plantations lost mostly
returned to be plantation (51%), even in some cases
they were changed to bare land (38%), grassland
(5%), shrubland (5%), and other land-cover types
(1%). On the contrary, natural forests were conver-
ted mainly to shrubland (40%), bare land (27%),
grassland (11%), cropland (10%), tree plantation
(7%), and other land covers (5%) (see examples in
appendix 4).
Environ. Res. Lett. 15 (2020) 124055 A Altamirano et al
Figure 4. Map of sample points by type of tree cover loss and period (red points =tree plantation, green points =natural forest).
Table 1. Model performance statistics for boosted regression tree
model of disaggregated tree forest loss prediction. s.e. =standard
error for cross-validation.
Indicator Training Cross-validation s.e.
AUC 0.98 0.96 0.02
Correlation 0.94 0.87 0.02
Sensibility (%) 98 92 1.84
Sensitivity (%) 97 85 1.71
Total tree cover loss was mostly concentrated in
the north-central area of the hotspot (the peak at
37S) (figure 4). However, there was a skewed lon-
gitudinal pattern towards the coastal range, which is
consistent with the peak of spatial distribution of tree
plantation (figure 2). Natural forest loss was more
scattered throughout the region in most time periods,
though it was more evident further south.
Overall, tree plantation and natural forest showed
an increasing pattern of tree cover loss during the
whole period (figure 4, appendix 5). Tree plantation
clear cutting was relatively constant until the period
2012–2013. However, from the period 2014–2015
onwards it increased substantially. Natural forest loss
remained relatively constant with minor fluctuations
until 2010, then increased, first steadily, and after
2013–2014 more sharply (appendix 5).
3.3. Disaggregating tree cover losses
Our model to predict disaggregated tree cover loss in
the whole study area obtained high accuracy under
different indicators (table 1). It is especially relevant
the high AUC using train and cross-validation data.
Model performance results were consistent due to the
high values of correlation, sensibility and sensitivity,
the last ones over 85%.
Our initial model was reduced to five main
explanatory variables (appendix 6), with far latitude
the most influencing variable (73%). Other relev-
ant variables were longitude (10%), elevation (9%),
perimeter/area relationship (4%), and distance to
cities (4%).
The disaggregation of total tree cover loss
in the study area revealed that from the total
1 756 052 ha, 1 549 909 ha corresponded to tree
plantation loss (mean =96 869 ha year1), while
206 142 ha corresponded to natural forest loss
(mean =12 884 ha year1). The spatial pattern of
disaggregated tree cover loss predictions (figure 5)
showed the same pattern of the sample points, which
also indicated the consistence of our results. Uncer-
tainty of estimations across the whole latitudinal
range of the study area showed low error. Higher
error are located in the southern area, however
the maximum error remains being lower than 10%
(appendix 7).
4. Discussion
4.1. Natural forest and tree plantations losses
We present an application to test disaggregation of
tree cover loss into natural forests and tree plantations
in a representative place, where forestry is a relevant
economic activity. Chile is one of the top ten coun-
tries in the world in terms of land dedicated to forestry
based on exotic tree plantations and the fifth in the
Americas (Cubbage et al 2007), with 3.2 million ha
(CONAF 2019). Moreover, forestry has been estim-
ated as the main driver of tree cover loss. In particular,
in Europe, North America, Russia/China/South Asia,
and Australia/Oceania it represents 99%, 56%, 41%,
and 29% of tree cover loss respectively (Van Holt et al
2016, Curtis et al 2018).
We differentiated tree-cover loss among natural
forest and exotic tree plantation with high accuracy.
This is not an easy task, especially at large scales given
Environ. Res. Lett. 15 (2020) 124055 A Altamirano et al
Figure 5. Map of spatial predictions of tree cover loss
disaggregated by forest plantation and natural forest.
the difficulties of remote sensing techniques and data.
The main reason of that is because tree plantations are
easily visible on the satellite images given its homo-
geneous structure (Wang and Huang 2012, Van Holt
et al 2016) which local experts are able to identify.
Our method takes advantage of freely available data
and remote sensing techniques combined with expert
local knowledge that have the potential to reproduce
the analysis for any region in the world. Moreover,
our results illustrate well the advantages of combining
remote-sense measurements and expert knowledge
than to use remote-sense technology alone (Cayuela
et al 2006, El Hajj et al 2009, Huang and Jia 2012,
Mialhe et al 2015). In this sense, we provide trans-
parent, comprehensive, confident and a cost-efficient
data-set given its several advantages of this approach
as stability, replicability, easy to share, testable and
low cost.
Based on the sampling, we also found that the
most important contributor of whole tree cover loss
is tree plantation, which account for 44% of total
tree cover loss. When we separate the tree cover loss,
more than 50% of tree plantation loss finished as tree
plantation at the end of our assessment period, and
this pattern is consistent with last global assessment
(Curtis et al 2018). After tree cover loss some land
cover types are transitional (i.e. bare land and grass-
land), but usually and in particular in Chile these land
cover types represent a stage in the intensive harvest-
ing activities of the tree plantation dynamic (Aguayo
et al 2009, Patterson and Hoalst-Pullen 2011). There-
fore, it would be likely that most tree plantation
remain over time with the same land use type. Also is
important to highlight that the main drivers of forest
plantations loss can be associated to both harvest-
ing and fires which we have not differentiated. Fur-
ther research is needed about this topic and especially
about the underlying causes of tree plantation and
natural forest loss given that both are related to dif-
ferent dynamics of change.
Natural forest loss continues to be an important
concern in one of the most endangered areas world-
wide, where our results show that approximately
13 000 ha year1are replaced by shrublands, bare
lands, grasslands, croplands, and exotic tree plant-
ations. This result contradicts some research point-
ing out that the expansion of planted forests has
the potential to reduce pressure on natural forests
(Köhlin and Parks 2001, Kauppi et al 2006). Other
studies support our findings and shows that exotic
tree plantation expansion has resulted in a contrac-
tion of natural forests (Heilmayr 2014, Sloan and
Sayer 2015, Van Holt et al 2016, Miranda et al
2017, Hua et al 2018) and can potentially increase
deforestation in certain regions (Pirard et al 2016).
After natural forest loss, transitional land cover types
(e.g. shrubland, bare land, grassland) can be found
which will finish as permanent ones (e.g. agricultural
crops, tree plantation). These transitional land cover
types account for more than 75% of natural forest
loss. An important proportion of these may finish
as tree plantation as illustrated by several examples
(Aguayo et al 2009, Patterson and Hoalst-Pullen 2011,
Altamirano et al 2016, Austin et al 2019).
4.2. Implications for public policies on forest
management and restoration
Inconsistences in terms of the information and
applied monitoring methods are recognized by FAO
for national forest monitoring systems (Macdicken
2015). This is particularly important in Chile, where
the Global Forest Resources Assessment report
indicated a net increase of natural primary forests
from 4 631 000 (1990) to 5 355 000 ha (2015), and
other natural regenerated forest from 8925 000
to 9 336 000 ha (FAO 2014). Our results high-
light the implications of information misinter-
pretation, and represent an opportunity for local,
but also global policies related to forest man-
agement and conservation and large-scale forest
Increasing the world’s forest cover have been
settled as the most important goal for fighting against
and adapting to climate change (Chazdon 2014,
Bastin et al 2019, Lewis et al 2019, Carey 2020).
But the current forest restoration strategies at land-
scape scales (and even larger), including different
activities (e.g. from strict restoration to monocul-
ture of tree plantations), may have different impacts
Environ. Res. Lett. 15 (2020) 124055 A Altamirano et al
on biodiversity, carbon, water, and eventually on
human wellbeing (Chazdon and Brancalion 2019,
Lewis et al 2019). Therefore, negative outcomes for
the environment could result if strategies incentiv-
ize exotic tree plantations establishment. In this
context, differentiating the cover dynamics of nat-
ural forests and exotic tree plantations is highly
Current sectorial policies supporting forest res-
toration ignore the links between biodiversity, water,
soil retention and timber-production (Latawiec et al
2015). With so many multiple benefits, regrowing
forests would be seen as a means for achieving
goals related to sustainability and human livelihoods
(Chazdon et al 2017). But to achieve this, new forest
visions should be encouraged based on a more com-
prehensive understanding of the ecological landscape
impacts of managing natural forest and tree planta-
tions, and eventually built a better base for developing
more efficient economic compensating mechanism to
ensure the multiple functions and benefits these tree
covers may provide (Chazdon and Brancalion 2019).
Our results can be useful to build capacity for
land monitoring and to improve our collective under-
standing of forest loss dynamics at global scale, and
even more it can be expanded to other conflicts of
land use and land cover change. For instance, it could
be used to check FAO statistics in places where we
are unsure or/and accessibility is limited. Currently,
forest certification covers an important area of world
managed forests and tree plantations (FSC 2018) but
it requires accurate monitoring systems. The current
climate change crisis and the related forestry agendas
(e.g. REDD+and Aichi Targets for 2020) require crit-
ical revision in global policy discussions, and at the
same time an accurate and specific land monitor-
ing system which can help to prevent the growing
problem of green grabbing in land use (e.g. Zhao et al
2014, Scheidel and Work 2018).
Chile has currently proposed its goal of National
Determined Contribution to face the climate crisis,
but the current proposal for reducing greenhouse
gas emissions inadequately addresses forest manage-
ment mainly through tree plantations (Chazdon and
Brancalion 2019, Duran and Barbosa 2019, Rudel et al
2019). The proposal considers to plant 200000 ha of
forests mainly oriented to tree plantations (approx-
imately 130 000 ha). Tree plantation is not a perman-
ent land cover as our results demonstrate, and this has
serious implications for climatic goals given harvested
systems, and carbon loss sequestered. Additionally,
this target results clearly insufficient to counterbal-
ance the forest loss area reported in this study. Ini-
tiatives to reduce carbon emissions should encour-
age differentiating natural forests from exotic tree
plantations and pay more attention on protecting and
managing sustainably the former. To advance towards
a global monitoring system, effectively differentiate
global tree cover loss should be an urgent goal as a
climate change mitigation action and to face the cur-
rent environmental global challenges.
Data availability statement
The data that support the findings of this study
are openly available at the following URL/DOI:
This research was supported by funding from
Fondecyt Grant No. 1171445 and Direcci´
on de
on of Universidad de La Frontera. AM and
CZ are thankful to ANID/FONDAP/15110009. AH
thanks the support of ANID NREDI170329 grant.
Figures were refined by V Sontag (@illusscientia).
Adison Altamirano
Paul Aplin
Rosa Reyes-Riveros
Aguayo M, Pauchard A, Az´
ocar G and Parra O 2009 Cambio del
uso del suelo en el centro sur de Chile a fines del siglo XX.
Entendiendo la din´
amica espacial y temporal del paisaje Rev.
Chil. Hist. Nat. 82 361–74
Altamirano A, Cely J P, Etter A, Miranda A, Fuentes-Ramirez A,
Acevedo P, Salas C and Vargas R 2016 The invasive species
Ulex europaeus (Fabaceae) shows high dynamism in a
fragmented landscape of south-central Chile Environ. Monit.
Assess. 188 495–509
Arroyo M T K, Marquet P, Marticorena C, Simonetti J,
Lohengrin C, Squeo F and Rozzi R 2004 Chilean winter
rainfall-Valdivian forest Hotspots Revisit pp 99–103
Austin K G, Schwantes A, Gu Y and Kasibhatla P S 2019 What
causes deforestation in Indonesia? Environ. Res. Lett.
14 024007
Bastin J-F, Finegold Y, Garcia C, Mollicone D, Rezende M,
Routh D, Zohner C M and Crowther T W 2019 The global
tree restoration potential Science 366 76–79
Bey A et al 2016 Collect earth: land use and land cover assessment
through augmented visual interpretation Remote Sens.
Carey J 2020 The best strategy for using trees to improve climate
and ecosystems? Go natural Proc. Natl Acad. Sci. USA
117 4434–8
Cayuela L, Golicher J D, Rey J S and Benayas J M R 2006
Classification of a complex landscape using
Dempster–Shafer theory of evidence Int. J. Remote Sens.
27 1951–71
Chazdon R L 2014 Second Growth: The Promise of Tropical Forest
Regeneration in an Age of Deforestation (Chicago, IL:
University of Chicago Press)
Chazdon R L, Brancalion P H S, Lamb D, Laestadius L, Calmon M
and Kumar C 2017 A policy-driven knowledge agenda for
global forest and landscape restoration Conserv. Lett.
10 125–32
Chazdon R and Brancalion P 2019 Restoring forests as a means to
many ends Science 364 24–25
Environ. Res. Lett. 15 (2020) 124055 A Altamirano et al
CONAF 2011 Catastro de los Recursos Vegetacionales Nativos de
Chile. Monitoreo de cambios y actualizaciones. Perıodo
1997–2011 (Santiago)
CONAF 2019 Cifras oficiales Catastros Usos de Suelo y Recursos
Vegetacionales noviembre de 2018 (available at: http:// (Consultado el 20 de agosto de 2019)
Cubbage F et al 2007 Timber investment returns for selected
plantations and native forests in South America and the
southern United States New For. 33 237–55
Curtis P G, Slay C M, Harris N L, Tyukavina A and Hansen M C
2018 Classifying drivers of global forest loss Science
361 1108–11
Duran A P and Barbosa O 2019 Seeing Chile’s forest for the tree
plantations Science 364 1388
El Hajj M, B´
e A, Guillaume S and Martin´
e J F 2009 Integrating
SPOT-5 time series, crop growth modeling and expert
knowledge for monitoring agricultural practices—the case
of sugarcane harvest on Reunion Island Remote Sens.
Environ. 113 2052–61
Elith J, Leathwick J R and Hastie T 2008 A working guide to
boosted regression trees J. Animal Ecol. 77 802–13
FAO (Food and Agriculture Organization of the United Nations)
2010 Evaluaci´
on de los recursos forestales mundiales 2010
informe nacional (Roma, Italia) p 68
FAO (Food and Agriculture Organization of the United Nations)
2014 Global forest resource assessment 2015, country report
Chile (Roma, Italia) p 99
FAO (Food and Agriculture Organization of the United Nations)
2015 FRA 2015 Terms and Definitions Forest Resources
Assessment Working Paper 180 (available at: https://
FSC 2018 Reporte anual 2017 Forest Stewarship Council
Gamer M, Lemon J, Fellows I and Singh P 2014 Package ‘irr’
(available at:
Hall J M, van Holt T, Daniels A E, Balthazar V and Lambin E F
2012 Trade-offs between tree cover, carbon storage and
floristic biodiversity in reforesting landscapes Landscape
Ecol. 27 1135–47
Hansen M C et al 2013 High-resolution global maps of
21st-century forest cover change Science 342 850–4
Heilmayr R 2014 Conservation through intensification? The
effects of plantations on natural forests Ecol. Econ.
105 204–10
Heilmayr R, Echeverría C, Fuentes R and Lambin E 2016 A
plantation-dominated forest transition in Chile Appl. Geog.
75 71–82
Heilmayr R, Echeverría C and Lambin E 2020 Impacts of Chilean
forest subsidies on forest cover, carbon and biodiversity Nat.
Sustainability 3701–9
Hua F, Wang L, Fisher B, Zheng X, Wang X, Yu D W, Tang Y, Zhu J
and Wilcove D S 2018 Tree plantations displacing native
forests: the nature and drivers of apparent forest recovery on
former croplands in Southwestern China from 2000 to 2015
Biol. Conserv. 222 113–24
Huang Z and Jia X 2012 Integrating remotely sensed data, GIS
and expert knowledge to update object-based land use/land
cover information Int. J. Remote Sens. 33 905–21
Jones J et al 2017 Forests and water in South America Hydrol.
Process. 31 972–80
Karp D S, Rominger A J, Zook J, Ranganathan J, Ehrlich P R and
Daily G C 2012 Intensive agriculture erodes β-diversity at
large scales Ecol. Lett. 15 963–70
Kauppi P E, Ausubel J H, Fang J, Mather A S, Sedjo R A and
Waggoner P E 2006 Returning forests analyzed with the
identity Proc. Natl Acad. Sci. USA 103 17574–9
Köhlin G and Parks P J 2001 Spatial variability and disincentives
to harvest: deforestation and fuelwood collection in South
Asia Land Econ. 77 206–18
Latawiec A E, Strassburg B B N, Brancalion P H S, Rodrigues R R
and Gardner T 2015 Creating space for large-scale
restoration in tropical agricultural landscapes Front. Ecol.
Environ. 13 211–8
Lewis S L, Wheeler C E, Mitchard E T A and Koch A 2019
Restoring natural forests is the best way to remove
atmospheric carbon Nature 568 25–28
Macdicken K G 2015 Global forest resources assessment
2015: what, why and how? For. Ecol. Manage.
352 3–8
Martínez-Jauregui M, Soliño M, Martínez-Fern´
andez J and
Touza J 2018 Managing the early warning systems of
invasive species of plants, birds, and mammals in natural
and planted pine forests Forests 916–19
Mcgarigal K, Cushman S and Ene E 2012 FRAGSTATS v4: spatial
pattern analysis program for categorical and continuous
maps Amherst: Computer Software Program Produced by
the Authors at the University of Massachusetts (available at:
Mialhe F, Gunnell Y, Ignacio J A F, Delbart N, Ogania J L and
Henry S 2015 Monitoring land-use change by combining
participatory land-use maps with standard remote sensing
techniques: showcase from a remote forest catchment on
Mindanao, Philippines Int. J. Appl. Earth Obs. Geoinf.
36 69–82
Miranda A, Altamirano A, Cayuela L, Lara A and Gonz´
alez M
2017 Native forest loss in the Chilean biodiversity
hotspot: revealing the evidence Reg. Environ. Change
17 285–97
Mittermeier R A, Gil P, Hoffmann M, Pilgrim J, Brooks T,
Mittermeier C G, Lamoreux J and da Fonseca G A B 2004
Hotspots Revisited: Earth’s Biologically Wealthiest and Most
Threatened Ecosystems (M´
exico: CEMEX)
Naudts K, Chen Y, Mcgrath M J, Ryder J, Valade A, Otto J and
Luyssaert S 2016 Forest management: Europe’s forest
management did not mitigate climate warming Science
351 597–9
Newbold T et al 2016 Has land use pushed terrestrial biodiversity
beyond the planetary boundary? A global assessment Science
353 288–91
Osuri A M, Gopal A, Raman T R S, Defries R, Cook-Patton S C
and Naeem S 2020 Greater stability of carbon capture in
species-rich natural forests compared to species-poor
plantations Environ. Res. Lett. 15 034011
Patterson M W and Hoalst-Pullen N 2011 Dynamic equifinality:
the case of south-central Chile’s evolving forest landscape
Appl. Geogr. 31 641–9
Pirard R, Dal Secco L and Warman R 2016 Do timber plantations
contribute to forest conservation? Environ. Sci. Policy
57 122–30
Potapov P et al 2017 The last frontiers of wilderness: tracking loss
of intact forest landscapes from 2000 to 2013 Sci. Adv.
R Development Core Team 2016 R: a language and environment
for statistical computing R Foundation for Statistical
Computing, Vienna, Austria (available at: www.
Reyes R, Nelson H, Navarro F and Retes C 2015 The firewood
dilemma: human health in a broader context of well-being
in Chile Energy Sustain. Dev. 28 75–87
Rudel T K, Meyfroidt P, Chazdon R, Bongers F, Sloan S, Grau H R,
Van Holt T and Schneider L 2019 Whither the forest
transition? Climate change, policy responses, and
redistributed forests in the twenty-first century Ambio
49 74–84
Scheidel A and Work C 2018 Forest plantations and climate
change discourses: new powers of ‘green’ grabbing in
Cambodia Land Use Policy 77 9–18
Shabani F, Kumar L and Ahmadi M 2016 A comparison of
absolute performance of different correlative and
mechanistic species distribution models in an independent
area Ecol. Evol. 65973–86
Sloan S and Sayer J A 2015 Forest Resources Assessment of 2015
shows positive global trends but forest loss and degradation
persist in poor tropical countries For. Ecol. Manage.
352 134–45
Environ. Res. Lett. 15 (2020) 124055 A Altamirano et al
Tropek R, Sedl´
cek O, Beck J, Keil P, Musilov´
a Z, Šímov´
a I and
Storch D 2014 Comment on ‘High-resolution global maps
of 21st-century forest cover change’ Science 344 981
Tyukavina A, Hansen M, Potapov P, Parker D, Okpa C,
Stehman S, Kommareddy I and Turubanova S 2018 Congo
Basin forest loss dominated by increasing smallholder
clearing Sci. Adv. 4eaat2993
Van Holt T, Binford M W, Portier K M and Vergara R 2016 A
stand of trees does not a forest make: tree plantations and
forest transitions Land Use Policy 56 147–57
Van Holt T, Moreno C A, Binford M W, Portier K M, Mulsow S
and Frazer T K 2012 Influence of landscape change on
nearshore fisheries in southern Chile Glob. Change Biol.
18 2147–60
Van Holt T and Putz F E 2017 Perpetuating the myth of the return
of native forests Sci. Adv. 31–4
Viña A, Mcconnell W J, Yang H, Xu Z and Liu J 2016 Effects of
conservation policy on China’s forest recovery Sci. Adv.
Wang H C and Huang C Y 2012 Investigating the spatial
heterogeneity of a subtropical montane cloud forest
plantation with a QuickBird image Int. J. Remote Sens.
33 7868–85
Yu Z et al 2020 Mapping forest type and age in China’s plantations
Sci. Total Environ. 744 140790
Zhao Y et al 2016 Detailed dynamic land cover mapping of Chile:
accuracy improvement by integrating multi-temporal data
Remote Sens. Environ. 183 170–85
Zhao Y, Huang C M and Zhang H J 2014 Ecological risk
assessment of provincial land-use overall
planning in China Hum Ecol. Risk Assess.
20 1491–506
... Estos planes de manejo autorizados por CONAF fueron calificados como ilegales por el Dictamen de la Contraloría General de la República, la Corte de Apelaciones de Santiago rechazó el recurso de protección presentado por la Sociedad Nacional de Agricultura, y la Corte Suprema confirmó la sentencia de la Corte de Apelaciones. La pérdida sostenida de bosques nativos por cambios de uso de suelo es una realidad (Heilmayr et al., 2016;Miranda et al., 2016;Altamirano et al., 2020). ...
... Para el período 2000-2016, la pérdida de bosque nativo en el hotspot mundial de Biodiversidad, entre los 25°S y 47°S (Arroyo et al., 2004), fue de 206.142 hectáreas, con un promedio de pérdida de 12.884 hectáreas anuales. El suelo donde estaban esos bosques nativos se transformó principalmente en matorrales, tierras desnudas o praderas, que posteriormente pueden ser transformadas en plantaciones forestales (Altamirano et al., 2020). CONAF no actúa apegada a la legislación vigente cuando desarrolla licitaciones para financiar plantaciones y manejo de plantaciones de pino y eucalipto (Licitaciones en Mercado Público: 1090-3-LR19, 2573-2-LQ19, 1035-9-LQ19, 2574-2-LP19, 2134-1-LQ19, 1035-9-LE20, 1090-2-LR20, 2573-1-LE20, 2134-5-LE20, 2575-4-LE20, 1092-61-L119), ya que no existe ningún cuerpo legal vigente que le otorgue tal facultad. ...
Technical Report
Full-text available
Propuesta científica desarrollada para la Comisión del Futuro de los Ecosistemas. Comisión del Futuro. Senado de Chile
... In such a way, the diversity of resources associated with heterogeneity on the local scale could compensate for the low habitat quality (Fontúrbel et al., 2017;González-Ancin et al., 2022). Nevertheless, at the landscape scale, the combination of different processes, for example, land-use change, deforestation, or urbanization, could increase habitat heterogeneity, making them more prone to isolation and decline (Altamirano et al., 2020). ...
Full-text available
Anthropogenic disturbance has dramatically degraded and reduced the extension of the temperate rainforests of southern South America, negatively affecting forest animals that depend on habitat attributes at local and landscape scales. We conducted a multi-scale assessment (from 1 to 4000 m) to understand better how local and landscape attributes influence forest animal abundance in an anthropogenic disturbance gradient. We selected five forest-dependent animal species to assess the effects of habitat alteration: an arboreal marsupial (Dromiciops gliroides) and four forest birds (Pteroptochos tarnii, Scytalopus magellanicus, Scelorchilus rubecula, and Sylviortho-rhynchus desmursii). We recorded forest animal abundances in four different habitat types (old-growth native, secondary, and logged native forests and forestry plantations). We measured local attributes in the field and characterized landscape attributes remotely. We evaluated marsupial abundance using camera traps and forest bird abundance using point counts, which were analyzed with Generalized Linear Mixed Models. Locally, canopy cover positively predicted marsupial abundance, while the number of fallen logs positively predicted bird abundance. At the landscape scale, native forest cover positively affected marsupial abundance, with significant effects at all levels. Conversely, plantation cover negatively affected forest bird abundance, while landscape heterogeneity negatively affected both groups. Our results showed that the abundance of the forest animals assessed here depends on multi-scale determinants. At the local scale, we advise greater canopy cover and maintaining woody debris. On the landscape scale, maintaining native forest cover should prioritize biodiversity management in the southern South America temperate forests. It is also crucial to control the expansion of forest plantations and reduce forest fragmentation to guarantee the persistence of forest-dependent species.
... Changes in tree cover at a landscape level are typically gradual and long-term, and therefore monitoring tree cover change at annual or longer intervals is the rule (e.g., Potapov et al., 2017;Altamirano et al., 2020). To assess mangrove development in our study basin we mapped mangrove cover once a year based on Google Earth satellite images. ...
Full-text available
We studied nearshore diurnal fish catches and fisheries development in the early stages of mangrove rehabilitation around the village of Timbulsloko on the central north coast of Java, Indonesia. Mangrove rehabilitation was part of a Nature-based Solutions project to re-establish ecological and economic resilience by combining coastal engineering measures with ecological recovery in conjunction with sustainable land and resource use. Creel surveys were conducted during the onset of the monsoon season October–December 2017 to document yields of the three main fishing gears targeting mangrove finfish in a 419 ha mangrove backwater basin. These were accompanied by structured interviews to obtain fisher views on developments in these fisheries. Analysis of satellite images and GIS-mapping were used to follow developments in mangrove coverage and effort in the estuarine lift-net fishery. Mangrove recovery between 2005 and 2018 achieved 8.5% of its maximum possible cover potential in the basin and followed an exponential growth curve. The increase in lift-net installations targeting finfish lagged in comparison to mangrove increase, remaining virtually zero till 2014, after which it rapidly increased. A baseline study in 2015 found no mangrove-associated finfish fisheries occurring in our study area. The 51 fishers surveyed in 2017, indicated that fishing activity in the area had strongly increased since 2015, with 45% of fishers stating to having started fishing this area a year earlier or even more recently. A significant majority of 87% of respondents with more than one year of experience at this location, stated that their catches had changed in terms of either fish size, quantity or composition since they started fishing, while 86% indicated improvement in terms of either size or quantity. Fishing generated about 1.05 ± 1.11 (SD) USD/hr worth of catch to professional fishers using either of the two net-types studied. As per 2017, fishing had become profitable in our study area, whereas this kind of fishery practically did not exist prior to 2014. We suggest that higher profitability may partially explain the rapid growth seen in fishing activity in the mangrove rehabilitation area. However, for 12 of the 18 larger species caught, the mean size in the catch was lower than mean size of maturation, indicating that these fisheries were principally targeted towards immature nursery fish. The results highlight the need to manage this currently developing fishery, otherwise any benefits to the local community might be nullified by overfishing.
... In-depth information regarding native forests and industrial plantations composition are available at CONAF (2017b). cAltamirano et al. (2020) mean estimation between 2000 and 2016. Tree cover loss in industrial plantations mostly returned to be industrial plantations, whilst native forests were converted to transitional cover land vegetation (e.g. ...
As president of the Climate Change Conference of the Parties, Chile has advocated for developing ambitious commitments to mitigate greenhouse gas emissions to achieve carbon-neutrality by 2050. However, Chile’s motivations and ambitious push to reach carbon-neutrality are complicated by a backdrop of severe drought, climate change impacts (i.e., wildfires, tree mortality), and the use of industrial plantations as a mitigation strategy. This has become more evident as widespread and severe wildfires have impacted large areas of industrial plantations, transforming the land-use, land-use change, and forestry sector from a carbon sink to a net carbon source. Consequently, Chile must diversify its climate actions to achieve carbon-neutrality. Nature-based solutions, including wetlands-peatlands and oceans, represent alternative climate actions that can be implemented to tackle greenhouse gas emissions at a national level. Diversification, however, must guarantee Chile’s long-term carbon sequestration capacity without compromising the ecological functionality of biodiverse treeless habitats and native forest ecosystems.
Full-text available
In response to the important benefits forests provide, there is a growing effort to reforest the world. Past policies and current commitments indicate that many of these forests will be plantations. Since plantations often replace more carbon-rich or biodiverse land covers, this approach to forest expansion may undermine objectives of increased carbon storage and biodiversity. We use an econometric land use change model to simulate the carbon and biodiversity impacts of subsidy driven plantation expansion in Chile between 1986 and 2011. A comparison of simulations with and without subsidies indicates that payments for afforestation increased tree cover through expansion of plantations of exotic species but decreased the area of native forests. Chile’s forest subsidies probably decreased biodiversity without increasing total carbon stored in aboveground biomass. Carefully enforced safeguards on the conversion of natural ecosystems can improve both the carbon and biodiversity outcomes of reforestation policies.
Full-text available
Tree plantations and forest restoration are leading strategies for enhancing terrestrial carbon (C) sequestration and mitigating climate change. While it is well established that species-rich natural forests offer superior C sequestering benefits relative to short-rotation commercial monoculture plantations, differences in rates of C capture and storage between longer-lived plantations (commercial or non-commercial) and natural forests remain unclear. Using a natural experiment in the Western Ghats of India, where late-20th century conservation laws prohibited timber extraction from monodominant plantations and natural forests within nature reserves, we assessed forests and plantations for aboveground C storage and the magnitude and temporal stability of rates of photosynthetic C capture (gross primary production). Specifically, we tested the hypothesis that species-rich forests show greater temporal stability of C capture, and are more resistant to drought, than monodominant plantations. Carbon stocks in monodominant teak (Tectona grandis) and Eucalyptus (Eucalyptus spp.) plantations were 30%-50% lower than in natural evergreen forests, but differed little from moist-deciduous forests. Plantations had 4%-9% higher average C capture rates (estimated using the Enhanced Vegetation Index-EVI) than natural forests during wet seasons, but up to 29% lower C capture during dry seasons across the 2000-18 period. In both seasons, the rate of C capture by plantations was less stable across years, and decreased more during drought years (i.e. lower resistance to drought), compared to forests. Thus, even as certain monodominant plantations could match natural forests for C capture and storage potential, plantations are unlikely to match the stability-and hence reliability-of C capture exhibited by forests, particularly in the face of increasing droughts and other climatic perturbations. Promoting natural forest regeneration and/or multi-species native tree plantations instead of plantation monocultures could therefore benefit climate change mitigation efforts, while offering valuable co-benefits for biodiversity conservation and other ecosystem services.
Full-text available
The restoration of trees remains among the most effective strategies for climate change mitigation. We mapped the global potential tree coverage to show that 4.4 billion hectares of canopy cover could exist under the current climate. Excluding existing trees and agricultural and urban areas, we found that there is room for an extra 0.9 billion hectares of canopy cover, which could store 205 gigatonnes of carbon in areas that would naturally support woodlands and forests. This highlights global tree restoration as our most effective climate change solution to date. However, climate change will alter this potential tree coverage. We estimate that if we cannot deviate from the current trajectory, the global potential canopy cover may shrink by ~223 million hectares by 2050, with the vast majority of losses occurring in the tropics. Our results highlight the opportunity of climate change mitigation through global tree restoration but also the urgent need for action.
Full-text available
Forest transitions occur when net reforestation replaces net deforestation in places. Because forest transitions can increase biodiversity and augment carbon sequestration, they appeal to policymakers contending with the degrading effects of forest loss and climate change. What then can policymakers do to trigger forest transitions? The historical record over the last two centuries provides insights into the precipitating conditions. The early transitions often occurred passively, through the spontaneous regeneration of trees on abandoned agricultural lands. Later forest transitions occurred more frequently after large-scale crisis narratives emerged and spurred governments to take action, often by planting trees on degraded, sloped lands. To a greater degree than their predecessors, latecomer forest transitions exhibit centralized loci of power, leaders with clearly articulated goals, and rapid changes in forest cover. These historical shifts in forest transitions reflect our growing appreciation of their utility for countering droughts, floods, land degradation, and climate change.
Full-text available
We investigate the causes of deforestation in Indonesia, a country with one of the highest rates of primary natural forest loss in the tropics, annually between 2001 and 2016. We use high spatial resolution imagery made available on Google Earth to characterize the land cover types following a random selection of deforestation events, drawn from the Global Forest Change dataset. Notorious in the region, large-scale oil palm and timber plantations together contributed more than two-fifths of nationwide deforestation over our study period, with a peak in late aughts followed by a notable decline up to 2016. Conversion of forests to grasslands, which comprised an average of one-fifth of national deforestation, rose sharply in dominance in years following periods of considerable fire activity, particularly in 2016. Small-scale agriculture and small-scale plantations also contributed one-fifth of nationwide forest loss and were the dominant drivers of loss outside the major islands of Indonesia. Although relatively small contributors to total deforestation, logging roads were responsible for a declining share of deforestation, and mining activities were responsible for an increasing share, over the study period. Direct drivers of deforestation in Indonesia are thus spatially and temporally dynamic, suggesting the need for forest conservation policy responses tailored at the subnational level, and new methods for monitoring the causes of deforestation over time.
Forest age serves as an essential factor in determining the accuracy of historical and future carbon (C) uptake quantifications, which is especially critical for China since the forest C stock dynamics are sensitive to the fast-growing, young-age plantations. However, a spatially explicit forest age maps with specific focus on forest plantations is not available yet. In this study, we developed a 1-km resolution age and type maps of forest plantations, and quantified their uncertainties spatially using field-measured data, national forest inventory data, digitalized forest maps, and remote sensing-based forest height maps. Simulation results showed forest plantations were 16.5 years old at national scale in 2005, which is close to the age of 16.6 years old derived from the 7th national inventory data using medium age in each forest plantation group with weighted area. Interestingly, we found that human management played an important role in forest age map reconstruction, which has not yet been considered in former studies. We also suggest that forest age and type maps should be used consistently in C stock simulations to avoid biases from mismatch information. Large uncertainty found in this study suggests further endeavors are required for improving the forest age and type maps.
Plans to triple the area of plantations will not meet 1.5 °C climate goals. New natural forests can, argue Simon L. Lewis, Charlotte E. Wheeler and colleagues.