Professional Documents
Culture Documents
Editorial
Understanding global forest resource change is more compli- that series – although this is the first time a large portion of the
cated than one might think. This is true irregardless of whether technical analyses have been published in the peer-reviewed
the assessment of change is done with ground-based measure- open literature. This volume was written by over 50 authors from
ments, satellite image analysis, aerial imagery or some combina- some 25 countries from every continent and includes world
tion of these approaches. National governments use a wide experts from many disciplines and institutional homes. Over
variety of approaches in measuring, monitoring and regulating 75% of the authors are from organizations other than the Food
their forests – and have access to widely divergent levels of and Agriculture Organization which organizes and leads the FRA
national forest resource data. Despite the challenges that countries process. It is itself an evolutionary change in how the results dili-
face – our awareness of the global role of forests in climate change, gently provided by 155 governments are aggregated and analyzed
the importance of forest loss or gain and gaps in wood supply have to help us all understand the global forest resource and how it is
over the past 60 years come largely from the Global Forest changing.
Resources Assessment (FRA). Generally the results of FRA 2015 show positive trends –
Too often the FRA is viewed solely for changes in forest area – increased knowledge about the resource, a slowing in the rate of
particularly deforestation rates. While this is important, there are both forest area loss and carbon emissions from forests and
many aspects of the forest resource and forest management that increasingly important actions taken by governments to improve
are perhaps more important. These include indicators of stocking, sustainable forest management.
designated management and ownership, sustainable forest man- The papers in this volume cover a broad range of topics – but
agement and forest disturbance. What happens to the forest and still there is much more to explore from FRA 2015. We encourage
how it is changing is far more than the conversion of forest land readers to look at these other sources for data and analysis of FRA
to agriculture and other uses – the forest that remains as forest 2015 reporting. These include:
is also often changing. Part of this comes from the activities of
man – as it has been since man and forest first met. Part comes 1. The Forest Land Use Data Explorer. This online portal provides
from a changing climate and the stress that this places on forests access to download FRA 2015 data, some of which has been
that have adapted to site conditions that are no longer what they updated since this Special Issue was published. It also allows
once were. Part of this change is also due to the fact that forests exploration of other natural resource data – including the agri-
are generally always in some kind of change – whether from shifts cultural statistics kept by FAO – along with FRA 2015 data.
in management, pest and disease, drought, fire, severe weather or 2. The FAO report: Global Forest Resources Assessment 2015: How
some combination of these. Change is as much a part of the ecolog- are forests changing? This synthesis document represents FAO
ical balance of forests as it is a critical element for foresters to summary reporting of key results from FRA 2015. It and the
understand and manage. Desk Reference are available in all six official UN languages.
The job governments do in monitoring and reporting on their 3. Global Forest Resources Assessment 2015 Desk Reference. This doc-
forest resources is often complex. Multiple jurisdictions, changes ument provides summary tables for nearly all of the quantita-
in monitoring and analysis related technologies and the availability tive variables reported in FRA 2015. It is designed to be
of human and financial resources for monitoring and assessment downloaded, printed and used as a handy paper reference or a
all impact how countries are able to monitor and manage their for- digital reference.
ests. The countries reporting through the FRA account for nearly 4. Country Reports. Since the year 2000 Country Reports have
99% of the world’s forest area – and 91% of this area is reported been the primary source of data reported through the FRA. All
in the top two quality tiers. 155 Country Reports are freely available online as are 79 Desk
This volume reports forest change over the period 1990–2015. Studies that were produced for countries and territories that
The Global Forest Resources Assessment has worked with govern- did not submit reports. Country reports cover nearly 99% of glo-
ments since the mid-1940s to prepare and assemble a global view bal forest area.
of how the world’s forests are changing. This volume is the latest in 5. Background documents, graphics and infographics are all avail-
able on the FRA website (http://www.fao.org/forestry/fra/en).
q
This article is part of a special issue entitled ‘‘Changes in Global Forest Resources The world has learned much about global forests and how they
from 1990 to 2015’’. have changed over the last 25 years. We have discovered as well
http://dx.doi.org/10.1016/j.foreco.2015.06.018
0378-1127/Ó 2015 Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
2 Editorial / Forest Ecology and Management 352 (2015) 1–2
a r t i c l e i n f o a b s t r a c t
Article history: This paper introduces a Special Issue of Forest Ecology and Management that includes a collection of ana-
Received 10 November 2014 lytical results from the 2015 Global Forest Resources Assessment (FRA 2015) covering 25 years of forest
Received in revised form 4 February 2015 change (1990–2015). FRA 2015 builds on a series of global assessments that began in 1948 and covers
Accepted 5 February 2015
change in forest area and type, volume, biomass and carbon stocking, measures of sustainable forest
Available online 7 September 2015
management, biodiversity and conservation, soil and water protective functions, wood production and
a number of socio-economic variables. It covers 234 countries and territories with an emphasis on forest
Keywords:
resource change over a twenty-five year period (1990–2015) and also looks forward to anticipated forest
Global Forest Resources Assessment 2015
FRA
change – both as government targets for forest area and projected change (to 2030) to global production
Global forest statistics and conservation forest area (to the year 2050). This paper describes important contributions of global
Forest resource change forest resource estimates to forest management, the methods used in the collection and analysis of
Forest change FRA 2015 data and provides links to additional information resources. It discusses some of the limitations
of this global dataset, some of the steps taken to improve quality and the characteristics that make this
type of global data most useful. While forest area change dominates public use of the FRA, the state of the
forest resource and management is critical to understanding the ecological and social values of the forest
and forestry. Country level reporting not only provides insights that are only possible through national
reporting but also provides greater national-level understanding and discussion of forest resource
change. The papers that follow in this Special Issue provide analyses of FRA 2015 data covering a wide
range of topics related to sustainable forest management and forest change.
Ó 2015 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
http://dx.doi.org/10.1016/j.foreco.2015.02.006
0378-1127/Ó 2015 Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
4 K.G. MacDicken / Forest Ecology and Management 352 (2015) 3–8
conventions (UN Framework Convention on Climate Change, FRA reports noted the rapid conversion of broadleaf tropical forest
Convention on Biological Diversity and the UN Convention to into agricultural land. This helped prompt some 60 years of
Combat Desertification) that each have expanded requirements increased investment by governments, companies, individuals,
for countries to report on forest resources. Organizations and tech- donor agencies and civil society groups in improving forest man-
nical professionals seek new global information through the FRA agement in the tropics. This investment has in turn contributed
and this interest often translates into new areas of inquiry, despite to the reductions in tropical forest loss rates reported in Keenan
efforts to keep the assessment focused on the forest resource. This et al. (2015) and FAO (2015).
demand for information has been difficult for countries to meet in Forest stock losses due to long-term land use change, especially
large part because the data often do not exist and resources for in the tropics, are an important part of greenhouse gas emissions
additional data collection are scarce. (Settele et al., 2014). Forest area and growing stock changes
Criticisms of FRA data have been primarily focused on detected through monitoring have provided improved understand-
data-poor country reports, the lack of comparable long-term trend ing of forests in the global carbon balance – and much of the early
data (Mather, 2005; Grainger, 2008; Harris et al., 2012) and work on this topic came from the FRA (Detwiler and Hall, 1988;
assumptions that suggest remotely-sensed data are inherently Houghton, 2008). FRA data continues to be used in global estimates
superior to forest statistics reported by sovereign nations of emissions from land use and land use change for the
(Grainger, 2008; Harris et al., 2012; Hansen et al., 2013). While Intergovernmental Panel on Climate Change (IPCC) and climate
the idea of long-term, high quality forest data collected using the change modelling (Smith et al., 2014; Petrescu et al., 2012).
same methods across time, forest type and countries with highly Global reporting over time can help identify knowledge gaps
divergent access to technical and financial resources is attractive, and highlight where improved information on forest resources is
it is also most impractical. At the same time, the assumption that needed. The importance of field-based National Forest
remote sensing provides clear, accurate and precise results for forest Inventories (NFI) has long been demonstrated through FRA report-
change at the global scale is also tenuous. Recent attempts to report ing – both through the value of reported inventory results and the
global forest change have made the mistake of characterizing tree identification of serious data gaps in countries where these inven-
cover change from satellite imagery as forest change (Harris et al., tories do not yet exist. The FRA 2015 dataset adds value to NFI
2012; Hansen et al., 2013) without regard to the processes of natural results by providing a Tier system that integrates data age and
regeneration and reforestation. Both of these studies have confused source classes. The need for updated field-based forest inventory
the distinction between forest and woody horticultural crops and as has become increasingly important for climate change mitigation
a result reflect tree canopy change, but not necessarily forest efforts such as the Reducing Emissions from Deforestation and
change. Neither remote sensing nor country-based reporting pro- Degradation (REDD+) mechanism.
vides perfect answers to forest resource change questions. An anal- The inclusion in FRA global reports of earth observing satellite
ysis of how results from FRA 2015 and remote sensing studies data such as those obtained and derived from the Landsat sensor
compare is found in Keenan et al. (2015). have become a useful adjunct to country reporting on forest extent
Global Forest Resources Assessment (FRA) 2015 statistics and over time. The integration of Landsat in tracking forest cover
analyses reported in this volume inherit many of the same change in the tropics (FRA 1980, 1990, 2000) has helped highlight
strengths and weaknesses as the 1948 assessment, yet it is also a the global value of this unique dataset. At the same time the use of
very different assessment in scope, transparency and quality. remote sensing in national forest inventories on which many FRA
More national forest inventories were available for FRA 2015 than country reports are based has grown substantially (MacDicken
for any other global assessment – 70% of them utilized remote et al., 2015).
sensing for at least a portion of the inventory. In 1948, one hundred
and one countries and territories reported – in 2015, this increased 2. Methods
to 234 countries and territories.
While public interest in forest change presently seems focused 2.1. Characteristics of FRA 2015
on deforestation, many of the world’s forests have changed in other
important ways that are less visible to the public – for example FRA 2015 was organized around 21 key questions grouped into
characteristics such as stocking density, species composition and eight topical categories: forest area and forest characteristics, pro-
diameter distribution (Plumptre, 1996; Dallmeier and Comiskey, duction, protective functions and ecosystems services, biodiver-
1998; Coomes and Allen, 2007). Climate change is predicted to cre- sity/conservation, disturbance, measuring progress toward
ate substantial shifts in tree species distribution and forest struc- sustainable forest management, economics/livelihoods and looking
ture (Scheller and Mladenoff, 2005; Gustafson et al., 2010). These forward (www.fao.org/forestry/FRA2015/Methods). A total of 117
shifts may require even greater monitoring efforts to assess more variables are included, most of which covered the period 1990–
rapid forest change in the future. At the same time, human popu- 2015 (Table 1). FRA 2015 included 37 variables that were not
lations have nearly tripled from 1948 to 2015 and are predicted included in previous assessments. Countries submitted reports
to continue increasing, putting greater pressure on remaining, between October 2013 and July 2014, including projected values
accessible forests to provide goods and services for a growing for the 2015 reporting year.
population. The majority of variables were reported for the years 1990, 2000,
Understanding how global forest resources are changing is far 2005, 2010 and 2015. Future forest area targets were requested
more complicated, and generally far more important than under- for the years 2020, 2030, and some variables were requested for
standing forest area change alone. the latest available year when a specific date was likely to be
unavailable or irrelevant (e.g., monitoring of forest management
1.2. How have FRA global datasets been used? plans stakeholder involvement). In addition to the quinquennial
reports, annual values were reported for wood removals, total
Information on the global state of forests drives policy and burned area and burned forest area. Annual forest area under inter-
resource flows at global, regional and national levels. Forests today national forest management certification were provided by the
– including extent, composition and structure, are in part a result Forest Stewardship Council (FSC) and Programme for the
of the many years spent acting on reported characteristics and for- Endorsement of Forest Certification (PEFC). Projections of future
est change. MacDicken (2014) highlights how, for example, past forest area were requested of countries for 2020 and 2030 and a
K.G. MacDicken / Forest Ecology and Management 352 (2015) 3–8 5
category-based analyses. Analyses by income category provide estimates of key FRA variables along with estimates of precision.
unique insights into where forest is concentrated, where change The Tier designations provide a measure of relative confidence in
is occurring and where the challenges for sustainable forest man- data at the variable level, even though measures of precision are
agement are greatest. not reported. Sample-based remote-sensing values reported at
regional, ecozone and global levels (FAO and JRC, 2012; FAO and
2.2.5. Country report vs. desk study JRC, 2014) provide added value in that the analysis contains calcu-
While the FRA process is dependent on national reporting by lated levels of precision around the means.
governments through their designated National Correspondents,
not all countries/territories have forest or a National Correspondent. 3.1.2. Consistent definitions over time
FAO received 155 country reports for FRA 2015 – representing Consistent definitions allow better comparisons over time and
nearly 99% of total forest area. When country reports were not sub- make reported values more consistent across countries. One of
mitted, FAO made estimates using desk studies based on available the differences between the assessments done prior to the year
literature and expert estimates – a practice also used in previous 1980 is that they contained a variety of forest definitions. While
assessments. Desk studies are analyses done by FAO staff and the pre-1980 definitions essentially defined forests as seen by
consultants based on limited data sources, and are therefore sovereign states, they resulted in countries reporting to different
are as a category less reliable than country reports. Desk study standards. By the year 2000 definitions of variables such as forest
countries are clearly identified in country reports, the online area were stable across all countries and have not changed in the
database (www.fao.org/forestry/FRA2015/database) and the Desk FRA since then. Other definitions have changed, as data limitations
Reference (www.fao.org/forestry/FRA2015/Desk Reference). were demonstrated or needs or data sources changed (www.fao.
org/forestry/FRA/2015/Historicalvariables). Variables included in
2.2.6. Quality tiers FRA 2015 were assembled (and for new variables, defined) through
FRA 2015 used a set of tier categories similar to those used by extensive consultation with a wide range of experts and
the Intergovernmental Panel on Climate Change (IPCC). Tiers were stakeholders.
requested for all variables that potentially had more than one
source of data. Tiers were defined by countries for each of the 3.1.3. Ease of interpretation
included variables for both status (i.e. the most recent report) Given the complexity of forests and forest change, no single
and trend (i.e. for two or more reporting periods). Countries were source of data tells a complete story. Even the definition of forest
asked to assign a Tier value to each qualifying variable: Tier 1 in the FRA can be difficult to communicate. While the FRA defini-
(expert estimate), Tier 2 (low intensity or incomplete surveys, tion of forest may on the surface appear to be a forest cover defini-
older data) or Tier 3 (high reliability, recent sources with national tion it is actually a forest land use definition – land can be
scope). Specific definitions for each tier were provided in the FRIMS temporarily devoid of tree cover and remain forest land for FRA
following this general pattern and are available in the FRA country reporting purposes. This is an important difference between the
reports online. Tier values were requested for status and trend for internationally accepted definition of forest used in the FRA com-
about one-third of all FRA 2015 variables. pared to sources such as Global Forest Watch, which reports
It is important to note that all country reports were indepen- changes in tree cover of all types. Data that are not easy to inter-
dently peer-reviewed by FAO staff, partners in the Collaborative pret are less useful – which requires continuous efforts by FAO
Forest Resources Questionnaire (CFRQ)1 and external experts. and partners to provide easy access, clear definitions and analyses
Peer-review comments were provided to National Correspondents to help inform interpretation.
for inclusion and, where necessary, corrections of individual national
reports were made before incorporation of data into the final FRA
3.1.4. Reasonable potential for practical use
2015 database.
Many users seek FRA data that has practical use. While much is
said about the value of FRA data for policy makers, it is only when
3. Discussion forest policy translates into appropriate changes in management
practices or investment that positive improvements in forest man-
3.1. When are global forest resource data most useful? agement occur. This applies to forestry-related investments by
governments, private companies, donors, NGOs or individuals.
Global datasets are often less detailed and precise than national Making global and national-level forest information relevant –
or sub-national datasets – this is also true for the FRA. Users have and easily accessible – to those who make investment decisions
through stakeholder discussions, internal FRA evaluations and a can help improve decision-making and ultimately improve forest
user survey noted the following desired characteristics for global management and use.
forest resource assessments:
3.2. Challenges with the FRA data include assessments. Analyses of change over time are most reliably done
within the 25 year coverage of FRA 2015 rather than across differ-
Despite efforts since 1948 to improve reporting and data quality ent FRA reports.
there remain constraints to the use of FRA data (FAO (1948, 2010), A summary of changes in data collection and definitions is found
Grainger (2008), Harris et al. (2012)). The following section outli- at http://www.fao.org/forestry/FRA2015/Historicalvariablechange.
nes some of the known limitations of the data collected for FRA Collectively these constraints seem exceptionally limiting – but
2015. this would be an incorrect conclusion for most purposes, particu-
larly when examining change over a 25 year period. The FRA data
3.2.1. Incomplete country reporting point to important changes in global forest resources – not always
The FRA 2015 questionnaire contains a diverse range of as precisely as some users would like, but of substantial quality
requests. In most cases, countries are unable to complete the entire and usefulness.
questionnaire for all years requested. Missing data complicates
time series analysis, particularly those that relate to forest area 4. Conclusions
numbers and requires caution in analysis and interpretation. This
problem will persist in future assessments because – like optical Despite the known limitations of collecting and synthesizing
satellite imagery in areas of perennial cloud cover – data in some information on the state, condition and trends of the world’s for-
countries simply do not exist. ests there is immense value in the data collected and the analyses
generated. Perhaps the best way to summarize the value of global
3.2.2. Inaccurate reporting forest resources information is to ask the question: Would forests
Until FRA 2015, there was no way to assess reporting quality be different today had we not known how they have changed in
without reading individual country reports. In 2015, the tier sys- recent times? The answer is yes. Not knowing would have led to
tem described in this paper was introduced to allow easy assess- piecemeal attempts to focus actions by governments, donors,
ment of timeliness and data source and to select country reports non-government organizations and private companies in improv-
for analysis based on a threshold quality class per variable. ing forest management where they are needed most. The monetary
However, missing documentation for some countries makes it dif- value of actions taken as a result of global knowledge of forest
ficult to assess sources even with tiers assigned because the basis resources may never be precisely known, but it is clear that
for tier selection is not provided. humanity and forest resources are both better off as a result of
the hard work of all who have contributed to understanding
3.2.3. FRA data are not experimental change in world forest resources – including those who have dili-
Some users are accustomed to experimental data that are well gently contributed to this volume.
controlled. The FRA data come from a variety of national sources,
including district forest offices, government registries, remote Acknowledgements
sensing, national inventories and special studies. In some cases,
these sources are combined to provide national statistics – statis- To the 285 FRA National Correspondents/Alternates and the
tics that are generally provided without associated measures of thousands who supported them in collection of the FRA 2015 data
precision. set – a most sincere thanks. This was a complex task often unap-
preciated by those who seek perfection where perfection is gener-
3.2.4. Numbers do not always add up ally impossible! Likewise, the author acknowledges with gratitude
Ideally, forest characteristics such as primary forest, other nat- the useful comments by reviewers and to those diligent FAO staff,
urally regenerated forest and planted forest should add up to the consultants and CFRQ partners who cleaned and spent many long
same number as total forest area. In some cases, they do and in hours analysing and interpreting data. Funding for this work was
others, they do not-in part due to missing data for one or provided from the FAO Regular Program and extra-budgetary con-
more category or the use of more than one source for related tributions from Canada, the European Union, Finland, Japan and
variables. the United States.
Grainger, A., 2014. Pan-tropical perspectives on forest resurgence. In: Hecht, S., Petrescu, A.M.R., Abad-Vinas, R., Janssens-Maenhout, G., Blujdeo, V., Grassi, G., 2012.
Morrison, K., Padoch, C. (Eds.), The Social Lives of Forests. University of Chicago Global estimates of carbon stock changes in living forest biomass: EDGARv4.3 –
Press, p. 512. time series from 1990 to 2010. Biogeosciences 9, 3437–3447.
Gustafson, E.J., Shvidenko, A.Z., Sturtevant, B.R., Scheller, R.M., 2010. Predicting Plumptre, A.J., 1996. Changes following 60 years of selective timber harvesting in
global change effects on forest biomass and composition in south-central the Budongo Forest Reserve, Uganda. Forest Ecol. Manage. 89 (1–3), 101–113.
Siberia. Ecol. Appl. 20 (3), 700–715. Scheller, R.M., Mladenoff, D.J., 2005. A spatially interactive simulation of climate
Hansen, M.C., DeFries, R.S., Townshend, J.R., Carroll, M., Dimiceli, C., Sohlberg, R.A., change, harvesting, wind and tree species migration and projected changes to
2003. Global percent tree cover at a spatial resolution of 500 meters: first forest composition and biomass in northern Wisconsin, USA. Glob. Change Biol.
results of the modis vegetation continuous field algorithm. Earth Interact. 7, 1– 11, 307–321.
15. Settele, J., Scholes, R., Betts, R., Bunn, S., Leadley, P., Nepstad, D., Overpeck, J.T.,
Hansen, M.C., Potapov, P.V., Moore, R., Hancher, M., Turubanova, S.A., Tyukavina, A., Taboada, M.A., 2014. Terrestrial and inland water systems. In: Field, C.B., Barros,
Thau, D., Stehman, S.V., Goetz, S.J., Loveland, T.R., Kommareddy, A., Egorov, A., V.R., Dokken, D.J., Mach, K.J., Mastrandrea, M.D., Bilir, T.E., Chatterjee, M., Ebi,
Chini, L., Justice, C.O., Townshend, J.R.G., 2013. High-resolution global maps of K.L., Estrada, Y.O., Genova, R.C., Girma, B., Kissel, E.S., Levy, A.N., MacCracken, S.,
21st-century forest cover change. Science 342 (6160), 850–853. Mastrandrea, P.R., White, L.L., (Eds.). Climate Change 2014: Impacts, Adaptation,
Harris, N.L., Brown, S., Hagen, S., Saatchi, S., Petrova, S., Salas, W., Hansen, M., and Vulnerability. Part A: Global and Sectoral Aspects. Contribution of Working
Potapov, P., Lotsch, A., 2012. Baseline map of carbon emissions from Group II to the Fifth Assessment Report of the Intergovernmental Panel on
deforestation in tropical regions. Science 336, 1573. http://dx.doi.org/ Climate Change. Cambridge University Press, Cambridge, United Kingdom and
10.1126/science.1217962. New York, NY, USA, pp. 271–359.
Houghton, R.A., 2008. Carbon flux to the atmosphere from land-use changes: 1850– Smith, P., Bustamante, M., Ahammad, H., Clark, H., Dong, H., Elsiddig, E.A., Haberl, H.,
2005. In: TRENDS: A Compendium of Data on Global Change. Carbon Dioxide Harper, R., House, J., Jafari, M., Masera, O., Mbow, C., Ravindranath, N.H., Rice,
Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of C.W., Robledo Abad, C., Romanovskaya, A., Sperling, F., Tubiello, F., 2014.
Energy, Oak Ridge, Tenn., USA. Agriculture, forestry and other land use (AFOLU). In: Edenhofer, O., Pichs-
Iremonger, S., Gerrand, A., 2011. FAO global ecological zones (www.fao.org/forestry/ Madruga, R., Sokona, Y., Farahani, E., Kadner, S., Seyboth, K., Adler, A., Baum, I.,
FRA2015/GEZ). Brunner, S., Eickemeier, P., Kriemann, B., Savolainen, J., Schlömer, S., von
Keenan, R., Reams, G., Achard, F., Freitas, J., Grainger, A., Lindquist, E., 2015. Stechow, C., Zwickel, T., Minx, J.C., (Eds.). Climate Change 2014: Mitigation of
Dynamics of global forest area: results from the FAO Global Forest Resources Climate Change. Contribution of Working Group III to the Fifth Assessment
Assessment 2015. Forest Ecol. Manage. 352, 9–20. Report of the Intergovernmental Panel on Climate Change. Cambridge
MacDicken, K.G., 2014. CFA Newsletter. No. 64, March 2014. Commonwealth University Press, Cambridge, United Kingdom and New York, NY, USA.
Forestry Association. Zon, R., Sparhawk, W.N., 1923. Forest Resources of the World. McGraw-Hill Book
Mather, A.S., 2005. Assessing the world’s forests. Global Environ. Change 15, 267– Co., Inc., New York, vols. 1 and 2.
280.
Forest Ecology and Management 352 (2015) 9–20
Dynamics of global forest area: Results from the FAO Global Forest
Resources Assessment 2015 q
Rodney J. Keenan a,⇑, Gregory A. Reams b, Frédéric Achard c, Joberto V. de Freitas d, Alan Grainger e,
Erik Lindquist f
a
School of Ecosystem and Forest Sciences, University of Melbourne, 221 Bouverie St, Carlton, Vic 3053, Australia
b
USDA Forest Service, National Office, 1601 North Kent Street, Suite 400, Arlington, VA 22209, USA
c
European Commission Joint Research Centre, Via Enrico Fermi 2749, I-21027 Ispra, VA, Italy
d
Brazilian Forest Service, Ministry of Environment, SCEN, AV. L4, Trecho 2, Bloco H, Brasília-DF 70818-900, Brazil
e
School of Geography, University of Leeds, Leeds LS2 9JT, UK
f
UN Food and Agriculture Organization, Viale delle Terme di Caracalla, 00153 Rome, Italy
a r t i c l e i n f o a b s t r a c t
Article history: The area of land covered by forest and trees is an important indicator of environmental condition. This
Received 7 January 2015 study presents and analyses results from the Global Forest Resources Assessment 2015 (FRA 2015) of
Received in revised form 9 June 2015 the Food and Agriculture Organization of the United Nations. FRA 2015 was based on responses to sur-
Accepted 9 June 2015
veys by individual countries using a common reporting framework, agreed definitions and reporting
Available online 7 September 2015
standards. Results indicated that total forest area declined by 3%, from 4128 M ha in 1990 to
3999 M ha in 2015. The annual rate of net forest loss halved from 7.3 M ha y 1 in the 1990s to
Keywords:
3.3 M ha y 1 between 2010 and 2015. Natural forest area declined from 3961 M ha to 3721 M ha between
Forest
Deforestation
1990 and 2015, while planted forest (including rubber plantations) increased from 168 M ha to 278 M ha.
Forest transition From 2010 to 2015, tropical forest area declined at a rate of 5.5 M ha y 1 – only 58% of the rate in the
Land use 1990s – while temperate forest area expanded at a rate of 2.2 M ha y 1. Boreal and sub-tropical forest
Land cover areas showed little net change. Forest area expanded in Europe, North America, the Caribbean, East
Remote sensing Asia, and Western-Central Asia, but declined in Central America, South America, South and Southeast
Asia and all three regions in Africa. Analysis indicates that, between 1990 and 2015, 13 tropical countries
may have either passed through their forest transitions from net forest loss to net forest expansion, or
continued along the path of forest expansion that follows these transitions. Comparing FRA 2015 statis-
tics with the findings of global and pan-tropical remote-sensing forest area surveys was challenging, due
to differences in assessment periods, the definitions of forest and remote sensing methods. More invest-
ment in national and global forest monitoring is needed to provide better support for international ini-
tiatives to increase sustainable forest management and reduce forest loss, particularly in tropical
countries.
Ó 2015 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3. Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.1. The distribution of forest area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.2. Global and regional trends in forest area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.3. Trends by climatic domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.4. Trends by income category. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
q
This article is part of a special issue entitled ‘‘Changes in Global Forest Resources from 1990 to 2015’’.
⇑ Corresponding author.
E-mail addresses: rkeenan@unimelb.edu.au (R.J. Keenan), greams@fs.fed.us (G.A. Reams), frederic.achard@jrc.ec.europa.eu (F. Achard), joberto.freitas@florestal.gov.br
(J.V. de Freitas), a.grainger@leeds.ac.uk (A. Grainger), Erik.Lindquist@fao.org (E. Lindquist).
http://dx.doi.org/10.1016/j.foreco.2015.06.014
0378-1127/Ó 2015 Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
10 R.J. Keenan et al. / Forest Ecology and Management 352 (2015) 9–20
were provided by FAO (2013b). To facilitate the reporting process, Ten countries – the Russian Federation, Brazil, Canada, the USA,
FAO pre-filled the tables with data for 1990–2010 from FRA 2010 China, Democratic Republic of Congo, Australia, Indonesia, Peru
(FAO, 2013a), a practice introduced in FRA 2010. If better estimates and India – account for 67% of total forest area. Six countries or ter-
had since become available, NCs were instructed to update infor- ritories – Aruba, the Faroe Islands, Greenland, Guernsey, Malta and
mation from previous years, using the interpolation and extrapola- Norfolk Island – reported zero forest cover, and there are no entries
tion methods employed in previous FRAs. All values for 2015 were for ten other countries or territories which also have little forest
estimated by NCs using extrapolation, generally based on a linear (Table S2) (FAO, 2015).
projection from previous data. Further details on the methods used
for FRA 2015 may be found in MacDicken (2015).
FRA 2015 received responses from 155 countries, and 79 desk 3.2. Global and regional trends in forest area
studies were conducted for the remaining countries and territories,
giving information for a total of 234 individual countries and terri- Overall, there was a net decrease in global forest area of 3%
tories. Data sources used by countries to report forest area between 1990 and 2015, from 4128 M ha to 3999 M ha, with nat-
included official government statistics, ground-based forest and ural and human-induced deforestation being offset by increases
vegetation inventories, remote sensing-based studies, published in forest area that had both natural and human causes (Table 1).
and refereed studies, ’gray literature’ reports and expert opinion. The annual rate of net forest loss halved over the 25 year period,
Countries were asked to report on the quality of their data for each falling from 7.3 M ha y 1 in the 1990s to 4.6 M ha y 1 between
variable using a three tier system: Tier 1 data were considered the 2000 and 2005, and to 3.4 M ha y 1 and 3.3 M ha y 1 for 2005–10
least reliable and Tier 3 data the most reliable (MacDicken, 2015). and 2010–15, respectively (Table 4). While this reduced rate of
Countries were classified into four climatic domains: Boreal, net forest loss is encouraging, it should not be regarded as equiva-
Temperate, Subtropical and Tropical (FAO, 2013a). If a country’s lent to reduced rates of human-induced deforestation (see
national boundaries encompassed more than one climatic domain, Section 4).
the country was assigned to the domain occupying the largest for- Primary forest accounts for a third of total forest area (see
est area of the country (Table S1). Countries were also divided into Morales et al., 2015), and increased slightly from 1200 M ha in
12 ‘sub-regions’, based on breakdowns of FAO regional groupings; 1990 to 1282 M ha in 2015, mainly because more countries sub-
and into income categories according to the World Bank country mitted data for this statistic. Primary forest, which is regarded as
lending group designation (MacDicken, 2015). undisturbed by human beings, is often reclassified as ’Other natu-
The methods used for the analysis of FRA findings in Section 4 rally regenerated forest’ after disturbance, though entries for this
are outlined in Supplementary Material. statistic are also incomplete, and it has changed little, from
3. Results
Table 2
The trend in forest area from 1990 to 2015 by sub-region (K ha) (FAO, 2015). All totals
This section provides an overview of the most important find- involve rounding.
ings of FRA 2015 on trends in forest area between 1990 and 2015.
Sub-region 1990 2000 2005 2010 2015
Central 26,995 23,448 22,193 21,010 20,250
3.1. The distribution of forest area America
Caribbean 5,017 5,913 6,341 6,745 7,195
East Asia 209,198 226,815 241,841 250,504 257,047
In 2015 forest covers 3999 M ha globally. This is equivalent to
East-Southern 319,785 300,273 291,712 282,519 274,886
31% of global land area, or 0.6 ha for every person on the planet. Africa
A further 1204 M ha are covered by other wooded land (Table 1). Europe 994,271 1,002,302 1,004,147 1,013,572 1,015,482
Forty-four per cent of global forest area is found in countries North Africa 39,374 37,692 37,221 37,055 36,217
classified as tropical and another 8% is in sub-tropical countries. North 720,487 719,197 719,419 722,523 723,207
America
Temperate countries account for 26% of global forest area and bor- Oceania 176,825 177,641 176,485 172,002 173,524
eal countries for 22% (Table 1). Europe (including the Russian South 930,814 890,817 868,611 852,133 842,011
Federation) has more forest than any other geographical America
sub-region (25%), followed by South America (21%) and North South- 319,615 298,645 296,600 295,958 292,804
Southeast
America (16%) (Table 2). Three quarters of all forest is in high
Asia
income and upper middle income countries, with just 25% of the West-Central 346,581 332,407 325,746 318,708 313,000
total in countries classified as having lower middle or low income Africa
(Table 3). The proportion of other wooded land in the tropics (43%) West-Central 39,309 40,452 42,427 42,944 43,511
is similar to that for forest in the tropics, but there are proportion- Asia
Total 4,128,269 4,055,602 4,032,743 4,015,673 3,999,134
ally greater areas of other wooded land in the sub-tropical and
temperate domains (Table 1).
Table 1
Forest and other wooded land from 1990 to 2015 in different global climatic domains (K ha) (FAO, 2015). Note that domains are determined by country classification into a
domain and all forests in the country are included in that domain. All totals involve rounding.
investment in forest inventory capacity in tropical and sub-tropical Our calculations show that natural forest area worldwide
countries, mainly to support more robust estimates of greenhouse declined by 6% from 3961 M ha to 3721 M ha between 1990 and
gas emissions from deforestation and forest degradation in antici- 2015 (Table 7). This was twice the percentage drop in forest area,
pation of payments becoming available under the UNFCCC REDD+ and the net result of a 3% expansion of temperate natural forest,
mechanism (Romijn et al., 2015). On the other hand, estimates for from 529 M ha to 546 M ha, and declines in natural forest in the
11% of global forest area in 2015 were based on data of the lowest other three climatic domains. In 142 tropical countries, the area
(Tier 1) quality. For example, 79 countries or territories, compris- of natural forest declined by 11%, from 1935 M ha to 1713 M ha.
ing 1.2% of global forest area, submitted no country reports and The decline in natural forest area worldwide was offset by a 66%
so their statistics had to be estimated by less accurate desk studies rise in planted forest, from 168 M ha to 278 M ha (Table 7).
(MacDicken, 2015). Furthermore, while the 12 countries which The change in natural forest area between any two time periods
have a forest area of more than 5 M ha and data of Tier 1 quality will be the net effect of forest clearance and conversion to another
only account for 9% of global forest area in 2015, they accounted land use (or deforestation) in some areas, natural forest losses
for 20% of the net decline in global forest area from 1990 to through processes such as fire or drought and natural forest expan-
2015. Ten of these countries are in Africa (Table 6). sion elsewhere. The trend in forest area combines these processes
with afforestation, in which forest is planted or regenerates natu-
4. Analysis rally on previously cleared land, e.g. as intensively managed plan-
tations or restoration forests, and reforestation, in which trees are
A key finding of FRA 2015 is that the net rate of loss of global planted or regenerate naturally on land already classified as forest
forest area halved over the last 25 years. This and other findings (FAO, 2013a). So a report of no net loss in forest area does not mean
are examined in this section to determine their robustness, and that the composition and structure of forest, its habitat value, or its
what they reveal about relationships between human beings and supply of ecosystem services, have stayed the same. Converting an
forests. area of natural forest into an intensively managed plantation of
exotic tree species in the tropics, for example, will increase its tim-
ber production potential but will generally reduce its biodiversity.
4.1. The implications of FRA 2015 statistics for trends in deforestation
FRA 2015 includes statistics on the rate of ’deforestation’ but
coverage is not comprehensive, as the table only includes reports
To grasp the full meaning of FRA 2015 statistics on forest area
from 48 countries. Our calculations on the dynamics of natural for-
trends it is necessary to disaggregate them in two ways: by sepa-
est show that the net rate of loss of natural forest halved, from
rating natural forest from forest plantations, and by separating
11.5 M ha y 1 to 5.8 M ha y 1, between the 1990s and 2010–15
deforestation from afforestation.
(Table 8). This replicated the sharp fall in the net rate of loss of for-
Early FRAs listed areas of natural forest and forest plantations
est (from 7.3 M ha y 1 to 3.3 M ha y 1) over the same period
separately, but since FRA 2000 FAO has combined them in a single
(Table 4). The vast majority of natural forest loss was in the tropics,
statistic, called ’forest’ (Grainger, 2007). Natural forest generally
where the rate of loss fell by 39%, from 10.4 M ha y 1 in the 1990s
describes vegetation that evolved naturally in an area. Planted for-
to 6.4 M ha y 1 from 2010 to 2015. In between these periods, it fell
est includes both intensively managed forest plantations purposely
to 9.1 M ha y 1 in 2000–05 and to 8.1 M ha y 1 in 2005–10. The
established to give priority to wood production that are usually
overall gross rate of loss of natural forest worldwide fell from
composed of a single tree species and forests established for land
11.8 M ha y 1 to 7.2 M ha y 1, but continuing loss in the tropics
conservation, coastal stabilization, biodiversity conservation or
was offset by an almost fivefold rise in the rate of expansion of
other purposes. Fortunately, the inclusion of separate figures for
temperate natural forest, from 0.3 M ha y 1 to 1.4 M ha y 1.
total forest area and planted forest area in FRA 2015 enabled us
Planted forest area has increased in absolute terms and as a pro-
to calculate natural forest area as the difference between the val-
portion of total forest area, though the rate of expansion halved
ues of these statistics (Table S3). FRA 2015, like FRA 2010, includes
from 5.3 M ha y 1 in 2005–10 to 2.5 M ha y 1 in 2010–15 (Table 8).
statistics on the areas of ’primary forest’ and ’other naturally
regenerated forest’, but their coverage for early time periods was
not comprehensive. The ’planted forest’ statistic, introduced in 4.2. Implications for modeling human impacts on forests
FRA 2010, includes both forest plantations and rubber plantations,
but not oil palm plantations and other agricultural plantations. So Changes in land cover and land use from forest to non-forest
replacing an area of rubber plantation by an oil palm plantation have both natural causes, e.g. drought, fire, storms and disease,
will suggest that there has been a reduction in forest plantation
area and hence in forest area too.
Table 7
Table 6 Trends in natural forest area (calculated as total forest area minus planted forest area)
Countries with forest areas greater than 5 M ha in 2015 and data of Tier 1 (lowest) and planted forest area from 1990 to 2015 by climatic domain (K ha) (calculated from
quality (FAO, 2015). FAO, 2015). All totals involve rounding.
Country Forest area in 2015 (K ha) Year 1990 2000 2005 2010 2015
Table 8 Table 9
Rates of change in natural forest area (calculated as total forest area minus planted Mean rates of change in natural forest area in 2000–10 (K ha y 1) in countries with
forest area) and planted forest area from 1990 to 2015 by climatic domain (K ha y 1) the highest mean rates of population growth in 2000–10 (K persons y 1) above
(calculated from FAO, 2015). All totals involve rounding. 1000 K persons y 1.
Period 1990–00 2000–05 2005–10 2010–15 Rate of population Rate of forest area
growth change
Natural forest
Boreal/polar 1.022 1.445 0.261 0.741 India 16,336 3
Temperate 0.279 0.571 0.812 1.385 Nigeria 3,683 625
Sub-tropical 0.338 0.483 1.166 0.034 Indonesia 3,174 1,002
Tropical 10.387 9.127 8.101 6.379 Pakistan 2,932 71
Total 11.467 10.483 8.716 5.770 Ethiopia 2,107 169
Brazil 2,071 3,030
Planted forest
Bangladesh 1,874 4
Boreal/polar 1.073 1.252 1.465 0.657
Philippines 1,579 9
Temperate 2.011 3.086 2.038 0.823
Democratic Republic of the 1,524 467
Sub-tropical 0.273 0.310 0.306 0.123
Congo
Tropical 0.844 1.264 1.493 0.859
Mexico 1,401 187
Total 4.200 5.911 5.302 2.462
Tanzania 1,095 595
past. For example, a test of the 62 tropical countries that had a Transition likely
mean rate of loss of natural forest above zero in 2000–10, and Burundi 289 198 181 253 276
Gambia 442 461 471 480 488
employing logarithmic transformations, resulted in a correlation
Ghana 8,627 8,909 9,053 9,195 9,337
coefficient of r = 0.615 between Log10 (deforestation rate2000–10) Rwanda 318 344 385 446 480
and Log10 (population growth rate2000–10) (Fig. S1, Table S4). Bhutan 2,507 2,606 2,656 2,705 2,755
However, the relationship between the rate of change of natural India 63,939 65,390 67,709 69,790 70,682
Laos 17,645 16,526 16,870 17,816 18,761
forest area and the rate of population growth is now far more com-
Philippines 6,555 7,027 7,074 6,840 8,040
plex than it was in the 1980s and 1990s, owing to the increasing Vietnam 9,363 11,727 13,077 14,128 14,773
dominance of controlling forces and drivers of forest expansion Cuba 2,058 2,435 2,697 2,932 3,200
over drivers of deforestation, the differential influences of urban Costa Rica 2,564 2,376 2,491 2,605 2,756
and rural populations (DeFries et al., 2010), and the effect of time Dominican Republic 1,105 1,486 1,652 1,817 1,983
Puerto Rico 287 450 463 479 496
lags. For example, for the five tropical countries with mean popu-
lation growth rates of 2–4 million persons y 1 in 2000–10 (UN, Transition possible
Cape Verde 58 82 84 85 90
2013), the rate of deforestation varied from 71 K ha y 1
Cote D’Ivoire 10,222 10,328 10,405 10,403 10,401
(Pakistan) to 3030 K ha y 1 (Brazil), and natural forest area actually Sierra Leone 3,118 2,922 2,824 2,726 3,044
rose at a rate of 43 K ha y 1 in India, which had the highest popu- Malaysia 22,376 21,591 20,890 22,124 22,195
lation growth rate in the tropics (16.4 million persons y 1) Thailand 14,005 17,011 16,100 16,249 16,399
Trinidad and Tobago 241 234 230 226 368
(Table 9).
R.J. Keenan et al. / Forest Ecology and Management 352 (2015) 9–20 15
by a change in the method used by its Royal Forest Department to between trends in different FRAs can establish mutual corrobora-
analyse satellite images. tion, or shed light on associated uncertainties (Houghton, 2010).
We examined global trends in natural forest, and trends for a sam-
4.3. Uncertainties about trends in forest area ple of 90 tropical countries in 1990, and how the shape of each
trend varied. By 2015, creation of new states had increased the size
One way to determine the robustness of forest area statistics of this sample to 93 countries, through the emergence of Eritrea,
reported in FRAs is to examine the quality of the latest individual South Sudan and Timor-Leste (Table S5). The set of countries used
estimates of national forest area. In FRA 2005 and earlier reports for long-term comparison corresponds to the set used in FRA 1990
specific information was listed in tables in the FRA Main Report (FAO, 1993), and includes Mexico, Nepal and Pakistan, which in
on the dates and types of the original estimates used to calculate FRA 2015 are allocated to the sub-tropical climatic domain.
the values of statistics, e.g. whether these estimates were recent Sometimes, differences between trends in successive FRAs can
or, say, 20 years old, and whether they were based on satellite or be traced to national factors, but on other occasions they reflect
airborne remote sensing measurements or on ’expert estimates’. changes in overall FRA methods. For example, the FRA 1990 value
The type of measurement (or lack of it) indicated the accuracy of of natural forest area in the tropics in 1990 was 170 M ha lower
each data point. The likely error involved in projecting it forward than the value for the same year in FRA 2000, reflecting a switch
to the latest common FRA reporting year (2015 in the case of from non-linear projection and interpolation methods in FRA
FRA 2015), using the interpolation and extrapolation methods 1990 to linear methods in FRA 2000 (Fig. 1, Table 11). After this
described in Section 2, could be inferred from the date of the esti-
mate and hence the length of the projection (Grainger, 2008). Table 11
In FRA 2015, the quality of the latest survey estimate is reported Trends in tropical forest area and natural forest area (90 countries) and global forest
by countries in terms of Tier quality rankings (from 1, poorer qual- area in Forest Resources Assessments (FRAs) 1990–2015 (M ha).
ity to 3, higher quality) (Table 5). In the template provided for Date FRA 1990 FRA 2000 FRA 2005 FRA 2010 FRA 2015
national forest area reports by FAO, Tier 3 was defined as ‘‘Data
Tropical
sources: either recent (610 years ago) National Forest Inventory
Forest area
or remote sensing, with ground truthing, or programme for 1980 1,928
repeated compatible NFIs’’; Tier 2 as ‘‘Data sources: full cover map- 1990 1,800 1,989 1,978 2,085 2,022
ping/remote sensing or old NFI (>10 years ago)’’; and Tier 1 as 2000 1,866 1,861 1,972 1,923
2005 1,803 1,929 1,872
‘‘Other’’ (FAO, 2013b). The evidence on which these rankings were
2010 1,887 1,847
based will be published in the individual FRA 2015 Country 2015 1,819
Reports, but these reports could not be used for analysis in this
Natural forest area
study. 1980 1,920
We therefore focused instead on examining the uncertainties 1990 1,756 1,926 1,949 2,058 1,995
associated with the aggregate trends in forest area from 1990 to 2000 1,799 1,829 1,934 1,888
2015 that were reported in FRA 2015, by analyzing them in the 2005 1,768 1,885 1,841
2010 1,837 1,799
context of trends for the same period reported in earlier FRAs.
2015 1,767
Since each FRA is undertaken independently, FAO prefers not to
estimate change by comparing, say, the estimate of global forest Global forest area
area in 2015 published in FRA 2015, with the estimate of global 1990 3,963 4,077 4,168 4,128
2000 3,870 3,989 4,085 4,056
forest area in 2005 published in FRA 2005. Instead, each FRA has
2005 3,952 4,061 4,033
presented new historical trends, e.g. from 1990 to 2015 in FRA 2010 4,033 4,015
2015, that are consistent with the estimate for the latest reporting 2015 3,999
year, i.e. 2015 in the case of FRA 2015. Analyzing the relationships
Fig. 1. Trends in tropical natural forest area 1980–2015 in Forest Resources Assessments (FRAs) 1990–2015.
16 R.J. Keenan et al. / Forest Ecology and Management 352 (2015) 9–20
Fig. 2. Trends in global forest area 1990–2015 in Forest Resources Assessments (FRAs) 2000–2015.
disjunction, in FRA 2005 the value of natural forest area in 1990 left as pre-filled. We found that, out of a sample of 93 tropical
only rose by 23 M ha over the value in FRA 2000 (Grainger, countries, in FRA 2010 41 countries had used the same statistics
2008), and this relatively small increment is interesting because for 1990–2005 as in FRA 2005, while in FRA 2015 47 countries
from FRA 2005 onwards NCs had more autonomy to make their had repeated the estimates for 1990–2010 used in FRA 2010. In
reports than in earlier FRAs. The increasing trend continued in contrast, in FRA 2005, the last report before prefilling was intro-
FRA 2010, in which natural forest area for 1990 rose by 109 M ha duced, only 10 tropical countries had used the same statistics for
to 2058 M ha. In FRA 2015, however, this trend was reversed, with 1990–2000 as in FRA 2000 (Table S6). The 47 countries that did
natural forest area in the tropics for 1990 actually falling by not change the prefilled figures in 2015 accounted for 32% of total
63 M ha to 1995 M ha – though this value is still higher than the natural forest area in the sample of 93 countries in 2015
corresponding estimate in FRA 2005. A similar deflation is seen (569 M ha), an area 33% larger than that of the 41 prefilling coun-
in consecutive trends in global forest area (Fig. 2). Whether this ’de- tries in FRA 2010 (428 M ha). When all FRA 2015 Country Reports
flationary’ phenomenon is the result of a greater accuracy of indi- become available, further scientific studies could explore this effect
vidual FRA 2015 estimates, or of other factors, requires further in more detail, and undertake a deeper statistical analysis of the
study. To show how national circumstances can play a role, e.g. uncertainties associated with FRA 2015 trends than was possible
deflation was also visible between tropical forest area trends in in this paper.
FRA 2005 and FRA 2000 (Table 11). A major reason for this was
the reclassification of 90% of forest plantations in India as natural 4.4. Comparison of the findings of FRA 2015 and remote sensing
forests, which inflated natural forest area but not overall forest studies
area (Grainger, 2010).
If FRA 1990 statistics are excluded, as an outlier, the decline of Another way to evaluate the findings of FRAs is to compare
228 M ha in tropical natural forest area between 1990 and 2015, them with the findings of global remote sensing surveys. A funda-
calculated from FRA 2015 statistics (Fig. 1), is 73% greater than mental issue for FRAs is that individual countries undertake their
the 132 M ha range of area estimates of tropical natural forest in forest inventories at different times and frequencies, according to
the year 1990 in FRAs 2000–2015 (Table 11). However, the their own measurement cycles and the availability of funding.
129 M ha decline in global forest area over the same period is only Errors can therefore arise when relatively rudimentary methods
63% of the 205 M ha range of estimates for global forest area in are used in FRAs to adjust estimates of forest area based on the
1990 in FRAs 2000–2015 (Fig. 2), and so could be within the limits results of these national surveys to a common reporting year, e.g.
of error for the trend as a whole, tending to support Mather’s 2015 in FRA 2015, and to estimate rates of forest area change
(2005) conclusions about historic uncertainties in the trend in glo- (see Section 3). Since remote sensing satellites collect data every
bal forest area. year for every part of the Earth’s surface, it is possible, in principle,
A second feature of successive trends that is apparent in Fig. 1 is to use these data to measure global and pan-tropical forest areas in
that while trends in FRAs 1980–2005 were generally parallel to one specific years and the rate at which they changed between these
another, in FRA 2010, and more markedly in FRA 2015, forest area years. In practice, for areas with high cloud cover, or subject to
declines less rapidly after 2005. On the one hand, this reflects the other technical issues associated with image processing, it may
recent reduction in the rate of deforestation, noted above. mean that remotely sensed images are a compilation of images
However, it might also have been influenced by FAO staff ‘‘prefill- from around a target year. In this section we compare FRA 2015
ing’’ the 1990, 2000, 2005, 2010 values of the forest area statistics findings with the results of global remote sensing surveys. We look
in FRA 2015 with those from FRA 2010. As was the case in FRA in turn at estimates of forest area and forest area change, and at
2010, countries were not obliged to fill in all the years in the data- their limitations.
base using the same set of data and the same forward and back- While there are general similarities between FRA and global
ward projection methods. Instead, values for the most recent remote sensing estimates, differences between them can be
years could be estimated separately and figures for previous years explained by a combination of five main factors. First, the
R.J. Keenan et al. / Forest Ecology and Management 352 (2015) 9–20 17
limitations of FRA statistics, in terms of data quality and consis- as the FRA 2015 figure for all tropical forest, which was based on
tency between countries, which are well documented (Grainger, a threshold tree cover of 10%.
2008; Hansen et al., 2013). Second, variation in the methods used Global or pan-tropical remote sensing studies of forests include
to estimate forest area, e.g. FRA main reports involve the aggrega- surveys undertaken outside the FRA process, and those undertaken
tion of national statistics reported by countries, while global within the FRA process itself. To provide an alternative common
remote sensing surveys may employ sampling methods or compre- picture of forest cover FAO has, since 1990, commissioned
hensive (‘‘wall to wall’’) mapping. Each method has its own large-scale uniform Remote Sensing Surveys (RSS) to complement
strengths and limitations. Third, difficulties in using remote sens- the national statistics listed in the FRA main reports. The results of
ing techniques to correctly determine percent tree cover for low these RSS have generally been presented in parallel with data sup-
tree cover densities, i.e. below 30%. Fourth, differences in the set plied by countries. Thus, the findings of RSS 2000 were included as
of countries covered by estimates. a chapter in the FRA 2000 Main Report (FAO, 2001). FAO commis-
The fifth factor, which will have a major impact on the analysis sioned two recent RSS in partnership with the Joint Research
in this section, is the use of different definitions of ’forest’. Lund Centre of the European Commission. These surveys, which were
(2014) lists over 1500 different operational definitions of the term based on a global sample of satellite imagery, were designed to
‘forest’ and over 200 different definitions of the term ‘tree’. Some of provide consistent and comparable estimates of tree cover and for-
these are land cover based (e.g. biophysical) definitions while est land-use from 1990 to 2010 at global and regional scales, to
others are land use based definitions. The FRAs employ a land use complement the increasing number of national statistics in FRA
definition, which means that they can include areas designated main reports that are based on national remote sensing surveys.
for forestry or conservation that are temporarily unstocked with The first, ‘RSS 2010’ (FAO and JRC, 2012), was undertaken as part
trees (see Section 2). Most remote sensing studies, on the other of the FRA 2010 process, but completed and published after the
hand, use a land cover definition (Magdon et al., 2014), because Main Report. The second, here called ‘RSS 2015’ (FAO and JRC,
land use cannot be determined by remote sensing alone. 2014), was part of the FRA 2015 process and extended the
Definitions of forest also vary in the minimum threshold of per- time-series for the same sample sites to 2010.
centage tree cover which they require before an area of land can According to FRA 2015, global forest area declined from
be defined as forest. Consequently, the use of different definitions 4128 M ha in 1990 to 3999 M ha in 2015 (Table 1). In the earliest
can lead to very different estimates of forest area in remote sensing remote sensing survey at this scale, Hansen et al. (2010) found that
estimates. This is illustrated by Hansen et al. (2013), who divided global forest area declined from 3269 M ha to 3168 M ha between
world land area into four ‘‘tree cover’’ classes – 0–25%, 26–50%, 2000 and 2005 (Table 12). The absolute values of area in 2000 and
51–75% and 76–100% – when they undertook wall to wall mapping 2005 were much lower than those in FRA 2015. One possible rea-
using Landsat images. They found that in the tropics, the 76–100% son for this was that Hansen et al. (2010) only mapped areas with
tree cover class, which broadly corresponds with tropical moist tree cover above 25%, whereas FRA 2015 used a 10% threshold;
forest, covered 1324 M ha in the year 2000, while the area above another reason was that they used coarse (0.5 km) resolution
25% tree cover, 2094 M ha, was of the same order of magnitude MODIS images, which may not detect small areas of forest. The
Table 12
Estimates of global and pan-tropical forest area 1990–2015 based on satellite data, compared with FRA 2015 findings (M ha).
Notes.
Hansen et al. (2010) study estimated gross loss from 2000 to 2005.
RSS 2010 study estimated loss from 2000 to 2005.
Achard et al. (2014) study does not include Mexico.
– = not assessed.
MMU = Minimum Mapping Unit.
18 R.J. Keenan et al. / Forest Ecology and Management 352 (2015) 9–20
Hansen et al. (2013) did not attach any error bars to their esti- Tropical forest
mates, but the errors involved in measuring global forest area in FRA 2015 9.5 7.9 6.6
Hansen et al. (2010) – 9.5 –
this way are clear from the range represented by the two Hansen Hansen et al. (2013) – – 8.5
et al. estimates, and another estimate of 3730 M ha for global forest Achard et al. (2014) 8.0 – 7.6
area in 2009 that was also based on Landsat imagery, but with a RSS 2000 8.3 – –
15% tree cover threshold (Gong et al., 2013). RSS 2010 5.7 9.1 –
RSS 2010 reported a drop in global forest area from 3820 M ha Tropical moist forest
to 3790 M ha between 2000 and 2005 (FAO and JRC, 2012), but RSS Achard et al. (2002) 5.8 – –
Hansen and DeFries (2004) 7.2 – –
2015 reported higher area values and but the same rate of loss,
Hansen et al. (2010) – 5.4 –
from 3950 M ha to 3890 M ha between 2000 and 2010 (FAO and Achard et al. (2014) 4.8 – 4.1
JRC, 2014). The relative proximity of these five global surveys to Kim et al. (2015) 4.0 7.0 6.6
the FRA 2015 trend is interesting, given that the two RSS used over
13,000 10 km 10 km sampling blocks within Landsat images,
while the other three surveys used wall to wall sets of Landsat 1150 M ha in 1990 and 1116 M ha in 1997 (Achard et al., 2002);
images (Hansen et al., 2013; Gong et al., 2013) or MODIS images 1094 M ha (Mayaux et al., 2005) and 1156 M ha (Hansen et al.,
(Hansen et al., 2010). 2010) in 2000; and 1257 M ha in 2010 (Kim et al., 2015). The high
Another set of remote sensing surveys can be used to evaluate value of the last of these studies was remarkable, as it was based
the report in FRA 2015 that between 1990 and 2015 tropical forest on wall to wall mapping using Landsat images, and referred to only
area in 142 countries declined by 10% from 1966 M ha to 34 countries and to all forest with more than 30% tree cover.
1770 M ha (Table 1). One estimate of 1870 M ha for all tropical for- Another set of measurements, based this time on samples of
est area in the year 2000 (Hansen et al., 2010) is identical to the Landsat images, did show a decline, from 1043 M ha in 1990, to
FRA 2015 figure for that year, and the RSS 2015 trend may also 1004 M ha in 2000 and 972 M ha in 2010 (Achard et al., 2014).
not be significantly different from the FRA 2015 trend: A more challenging task is to substantiate key findings of FRA
1860 M ha, 1790 M ha and 1730 M ha for 1990, 2000 and 2010 2015 that between the 1990s and 2010–15 the net rates of loss
respectively (FAO and JRC, 2014). On the other hand, the RSS of all forest and tropical forest halved, falling from 7.3 M ha y 1
2010 estimate of 1620 M ha for 2005 (FAO and JRC, 2012) and esti- to 3.3 M ha y 1, and from 9.5 M ha y 1 to 5.5 M ha y 1, respectively
mates by Achard et al. (2014) of 1635 M ha, 1574 M ha and 1514 M (Table 4). One reason for this difficulty is that estimates of forest
for 1990, 2000 and 2010, respectively, are much lower (Table 12). area loss are affected by differences between land use and land
These latter figures resulted from using the same sampling method cover definitions. For example, most forest clearance in the tropics
as RSS 2010 (4000 sample blocks in the tropics), but with a land involves a change in land use, e.g. to agriculture, and the customary
cover definition instead of a land use definition. Interestingly, the forest management practice is selective logging, which does not
Achard et al. (2014) figure for 2000 is almost identical to the esti- involve forest clearance. In temperate and boreal forests, on the
mate of 1571 M ha for all tropical forest in 2000 by RSS 2000 (FAO, other hand, clear-felling is a common forest management practice.
2001), which employed a different (two-stage stratified random) This results in a temporary loss of tree cover but does not lead to a
design for sampling Landsat images. change in land use. Hansen et al.’s (2010) measurement of ’gross
The accuracy of estimates of all tropical forest area is con- global forest loss’, which equated forest with tree cover above
strained by uncertainty about the distribution of open ’savanna’ 25%, concluded that 20.2 M ha y 1 was lost between 2000 and
woodlands in dry areas, which are extensive in Africa, Australia 2005. This was much higher than the corresponding rate in FRA
and Latin America (Bodart et al., 2013; Beuchle et al., 2015). 2015. However, only 47% of this loss (9.5 M ha y 1) occurred in
Open woodlands were thought to contribute 734 M ha, or 38%, to the tropics, while the remaining 53% (10.7 M ha y 1) occurred in
the estimate of 1935 M ha for tropical natural forest area in 1980 temperate and boreal forests, as a result of logging, fire and insect
(FAO, 1982). Yet owing to their low commercial importance they outbreaks. Hansen et al.’s (2010) method involved taking samples
are often not assessed by field surveys, or surveyed regularly by of Landsat images to estimate the rate of forest loss. When Hansen
governments. Measuring their area using remote sensing is also et al. (2013) used the more elaborate approach of classifying wall
difficult, even with Landsat images (Lambin, 1999). Estimates of to wall Landsat images they found that temperate and boreal for-
the area of the even lower tree cover category of ’other wooded ests now accounted for just 38% of global gross forest cover loss
land’ also vary, e.g. the estimate by Achard et al. (2014) of of 19.2 M ha y 1 between 2000 and 2012, and this was offset by
975 M ha for other wooded land in 2010 is 81% higher than the forest gain of 6.7 M ha y 1. RSS 2010 used the same 10% tree cover
538 M ha reported in FRA 2015. This discrepancy is partly threshold used for the FRA main report statistics, but its estimate
explained by the fact that the remotely-sensed class of other of 14.7 M ha y 1 for the rate of global net forest loss between
wooded land may include areas of low density tree cover that 2000 and 2010 was also higher than the corresponding estimate
would be classified as ’forest’ in FRA Main Report statistics. in FRA 2010 (FAO and JRC, 2012). Coulston et al. (2013) shed fur-
Indeed, RSS 2010 showed that differences between regional forest ther light on this issue by showing that estimates of forest land
area estimates in FRA 2015 and RSS 2010 estimates rose as the use extent and forest land cover extent in the southeastern USA
extent of drylands in a region increased (FAO and JRC, 2012). were not correlated; that estimates of net change based on forest
Tropical moist forests – which comprise all closed forests in the land cover and forest land use were only modestly correlated;
humid tropics – are easier to map using satellite data, though they and that net forest land use change estimates were independent
are often hidden from satellite sensors by clouds. They were the of gross forest cover loss. They suggested that changes in forest
initial focus of global remote sensing forest surveys, yet it is diffi- cover are more indicative of a change in forest land use in the trop-
cult to conclude that tropical moist forest area has declined signif- ics than in areas, such as the southeastern USA, where forest regen-
icantly by looking at the time series of available estimates, i.e. eration commonly follows harvesting and disturbance.
R.J. Keenan et al. / Forest Ecology and Management 352 (2015) 9–20 19
A sufficient number of estimates of deforestation rates for trop- policy outcomes. For example, FRA 2015 estimates of forest area
ical moist forest are available for comparison with estimates of include natural forest and planted forest, and a reduction in net
tropical forest area loss in FRA 2015. These estimates tend to be forest loss (which could result from a combination of a loss of nat-
lower, not higher, than corresponding estimates of tropical forest ural forest and a gain in planted forest) is not the same as a reduc-
loss in FRA 2015, not least because they refer to a subset of tropical tion in deforestation.
forests. Thus, estimates of forest loss for the 1990s include The quality of data on which FRA 2015 statistics are based is
5.8 M ha y 1, based on samples of Landsat images (Achard et al. considered to be higher than those in earlier FRAs, with National
(2002); and 7.2 M ha y 1, offset by forest gain of 1.2 M ha y 1, Correspondents reporting that about 60% of global forest area in
using 8 km resolution images (Hansen and DeFries, 2004) 2015 has been estimated using the highest quality data.
(Table 13). The declining trend in FRA 2015 was supported by However, the amount of forest reported using the poorest data
Hansen et al.’s (2010) estimate for 2000–05 of 5.4 M ha y 1, and quality (Tier 1) is still 11%, indicating that a number of countries
estimates by Achard et al. (2014) of 4.8 M ha y 1 between 1990 used out-of-date or incomplete national assessments.
and 2000 and 4.1 M ha y 1 between 2000 and 2010. On the other Independent pan-tropical and global remote-sensing surveys
hand, another set of integrated estimates found that the net rate exhibited differences in estimates of global and tropical forest area
of forest loss in 34 humid tropical countries, based on wall to wall and dynamics, both with FRA 2015 and with each other, because of
mapping using Landsat images, rose from 4.0 M ha y 1 in the 1990s differences in definitions and in measurement methods. This vari-
to 7.0 M ha y 1 in 2000–05, and then declined to 6.1 ha y 1 in ation between studies suggests the need for policy makers to
2005–10 (Kim et al., 2015). understand differences between estimates of forest area derived
Fewer estimates of forest loss are available for all tropical forest. using different methods and based on different definitions when
These include 8.3 M ha y 1 in the 1990s in RSS 2000 (FAO, 2001), establishing policies and targets to reduce deforestation rates
9.5 M ha y 1 between 2000 and 2005 (Hansen et al., 2010), and and increase forest area.
8.5 M ha y 1 between 2000 and 2012 (Hansen et al., 2013). The Increased investment over the last 10 years has improved
combined time series is understandably closer in value to the cor- national forest monitoring capacity in developing countries, in
responding FRA estimates than the time series for tropical moist considerable part to prepare to implement arrangements for the
forest, and supports Kim et al.’s (2015) claim that tropical defor- REDD+ mechanism of the UN Framework Convention on Climate
estation peaked after 2000. Further support is provided by RSS Change. This investment needs to be sustained, and expanded to
2010, which estimated that net tropical forest loss rose from include key African and Latin American countries with currently
5.7 M ha y 1 in the 1990s to 9.1 M ha y 1 in 2000–05 (FAO and deficient inventory systems. A comprehensive monitoring
JRC, 2012). However, Achard et al. (2014) found only a slight drop approach that integrates remotely-sensed data of sufficiently high
from 8.0 M ha y 1 in the 1990s to 7.6 M ha y 1 from 2000 to 2010, resolution with field measurements and observations could pro-
which supports the FRA 2015 trend. vide a sound basis for assessing forest-related greenhouse gas
Overall, the findings of remote sensing studies provide some dynamics and for supporting sustainable forest management at a
support for FRA 2015 findings, especially on trends in global forest range of scales.
area and tropical forest area. However, it is difficult to generalize,
because there are considerable differences between the findings Acknowledgements
of remote sensing studies, owing to differences in assessment peri-
ods, the definitions of forest, remote sensing methods, and country We would like to acknowledge the efforts of the National
coverage. Correspondents from each country and territory in collecting and
presenting data for the 2015 Global Forest Resource Assessment
and the UN FAO staff involved in reviewing and collating data.
5. Conclusions This analysis was completed while RJK was undertaking a Special
Study Program from the University of Melbourne at CATIE in
The results of the latest Global Forest Resources Assessment of Costa Rica and the London School of Economics and Political
the Food and Agriculture Organization of the United Nations (FAO) Science. He would like to acknowledge the support and hospitality
indicate that between 1990 and 2015 total forest area declined by provide by those institutions. Other authors would like to acknowl-
3%, from 4128 M ha to 3999 M ha, and the annual rate of net forest edge the support of their respective host institutions. We also
loss halved from 7.3 M ha y 1 in the 1990s to 3.3 M ha y 1 between thank two anonymous referees for helpful comments and
2010 and 2015. Loss of forest area was largely in the tropics, from suggestions.
1966 M ha in 1990 to 1770 M ha in 2015, while temperate forest
expanded from 618 M ha to 684 M ha over the same period. Net
Appendix A. Supplementary material
tropical forest loss over the last five years was dominated in
South America by Brazil (984 K ha y 1), in Asia by Indonesia
Supplementary data associated with this article can be found, in
(684 K ha y 1), and in Africa by Nigeria (410 K ha y 1). However,
the online version, at http://dx.doi.org/10.1016/j.foreco.2015.06.
the Brazilian and Indonesian loss rates were only about 40% of
014.
the corresponding rates in the 1990s. Forest area expanded
between 2010 and 2015 by 1.5 M ha y 1 in China, and at rates of
301 K ha y 1, 275 K ha y 1 and 240 K ha y 1 in Chile, the USA and References
the Philippines, respectively. Between 1990 and 2015, thirteen Achard, F., Eva, H.D., Stibig, H.-J., Mayaux, P., Gallego, J., Richards, T., Malingreau, J.-
tropical countries may have either experienced national forest P., 2002. Determination of deforestation rates of the world’s humid tropical
transitions from net forest loss to net forest gain, or continued forests. Science 297, 999–1002.
Achard, F., Beuchle, R., Mayaux, P., Stibig, H.-J., Bodart, C., Brink, A., Donnay, F., Lupi,
along the path of forest expansion that follows such transitions.
A., Carboni, S., Desclee, B., Donnay, F., Eva, H.D., Lupi, A., Rasi, R., Seliher, R.,
Forest transitions may be in progress in another six countries. Simonetti, D., 2014. Determination of tropical deforestation rates and related
While the results of FRA 2015 are likely to be used extensively carbon losses from 1990 to 2010. Glob. Change Biol. 20, 2540–2554.
for research and policy formulation, as with previous FRAs, careful Arrow, K.J., Dasgupta, P., Goulder, L.H., Mumford, K.J., Oleson, K., 2012.
Sustainability and the measurement of wealth. Environ. Dev. Econ. 17, 317–353.
interpretation will be needed to ensure that the statistics are used Ausubel, J.H., Wernick, I.K., Waggoner, P.E., 2012. Peak farmland and the prospect
in ways that are consistent with scientific terminology and desired for land sparing. Popul. Dev. Rev. 38 (S), 217–238.
20 R.J. Keenan et al. / Forest Ecology and Management 352 (2015) 9–20
Baker, D.J., Richards, G., Grainger, A., Brown, S., DeFries, R., Gonzalez, P., Held, A., Houghton, R.A., 2010. How well do we know the flux of CO2 from land-use change?
Kellndorfer, J., Ndunda, P., Ojima, D., Skrovseth, P.-E., Souza, C., Stolle, F., 2010. Tellus B 62, 337–351.
Achieving forest carbon information with higher certainty: a five step strategy. Kim, D.-H., Sexton, J.O., Townshend, J.R., 2015. Accelerated deforestation in the
Environ. Sci. Policy 13, 249–260. humid tropics from the 1990s to the 2000s. Geophys. Res. Lett. 2014GL062777.
Beuchle, R., Grecchi, R.C., Shimabukuro, Y.E., Seliger, R., Eva, H.D., Sano, E., Achard, F., http://dx.doi.org/10.1002/2014GL062777.
2015. Land cover changes in the Brazilian Cerrado and Caatinga biomes from Lambin, E.F., 1997. Modelling and monitoring land-cover change processes in
1990 to 2010 based on a systematic remote sensing sampling approach. Appl. tropical regions. Prog. Phys. Geogr. 21, 375–393.
Geogr. 58, 116–127. Lambin, E.F., 1999. Monitoring forest degradation in tropical regions by remote
Bodart, C., Brink, A., Donnay, F., Lupi, A., Mayaux, P., Achard, F., 2013. Continental sensing: some methodological issues. Glob. Ecol. Biogeogr. Lett. 8, 191–198.
estimates of forest cover and forest cover changes in the dry ecosystems of Lambin, E., Turner, B., Geist, H., Agbola, S.B., Angelsen, A., Bruce, J.W., Coomes, O.T.,
Africa for the period 1990–2000. J. Biogeogr. 40, 1036–1047. Dirzo, R., Fischer, G., Folke, C., George, P.S., Homewood, K., Imbernon, J.,
Coulston, J.W., Reams, G.A., Wear, D.N., Brewer, C.K., 2013. An analysis of forest land Leemans, R., Lin, X., 2001. The causes of land-use and land-cover change:
use, forest land cover and change at policy-relevant scales. Forestry 87 (2), 267– moving beyond the myths. Glob. Environ. Change 11, 261–269.
276. Lanly, J.P. (Ed.), 1981. Tropical Forest Resources Assessment Project (GEMS):
DeFries, R.S., Rudel, T., Uriarte, M., Hansen, M., 2010. Deforestation driven by urban Tropical Africa, Tropical Asia, Tropical America, vol. 4. UN Food and Agriculture
population growth and agricultural trade in the 21st-century. Nat. Geosci. 3, Organization/UN Environment Programme, Rome.
178–181. Le Quere, C., Raupach, M.R., Canadell, J.G., Marland, G., 2009. Trends in the sources
FAO, 1982. Tropical Forest Resources. FAO Forestry Paper No. 30. UN Food and and sinks of carbon dioxide. Nat. Geosci. 2, 831–836.
Agriculture Organization, Rome. Lund, H.G., 1999. A ’forest’ by any other name. . .. Environ. Sci. Policy 2, 125–133.
FAO, 1993. Forest Resources Assessment 1990: Tropical Countries. FAO Forestry Lund, H.G., 2002. When is a forest not a forest? J. For. 100, 21–28.
Paper No. 112. UN Food and Agriculture Organization, Rome. Lund, H.G., 2014. Definitions of Forest, Deforestation, Reforestation and
FAO, 1995. Forest Resources Assessment 1990: Global Synthesis. FAO Forestry Paper Afforestation. Forest Information Services, Gainesville, VA, <http://
No. 124. UN Food and Agriculture Organization, Rome. home.comcast.net/%7Egyde/DEFpaper.htm> (accessed 28.06.14).
FAO, 2001. Global Forest Resources Assessment 2000. FAO Forestry Paper No. 140. MacDicken, K.G., 2015. Global Forest Resources Assessment 2015: What, why and
UN Food and Agriculture Organization, Rome. how? For. Ecol. Manage. 352, 3–8.
FAO, 2006. Global Forest Resources Assessment 2005. FAO Forestry Paper No. 147. Magdon, P., Fischer, C., Fuchs, H., Kleinn, C., 2014. Translating criteria of
UN Food and Agriculture Organization, Rome. international forest definitions into remote sensing image analysis. Remote
FAO, 2010. Global Forest Resources Assessment 2010. FAO Forestry Paper No. 163. Sens. Environ. 149, 252–262.
UN Food and Agriculture Organization, Rome. Mather, A.S., 1992. The forest transition. Area 30, 117–124.
FAO, 2013a. Guide for Country Reporting for FRA 2015. FRA 2015 Working Paper. Mather, A., 2005. Assessing the world’s forests. Glob. Environ. Change 15, 267–280.
UN Food and Agriculture Organization, Rome. Mather, A.S., 2007. Recent Asian forest transitions in relation to forest-transition
FAO, 2013b. Template for Country Reports for FRA 2015. FRA 2015 Working Paper. theory. Int. For. Rev. 9, 491–501.
UN Food and Agriculture Organization, Rome. Matthews, E., Grainger, A., 2002. Evaluation of FAO’s global forest assessment from
FAO, 2015. Global Forest Resources Assessment 2015. FAO Forestry Paper No. 1⁄⁄. the user perspective. Unasylva 53 (210), 42–50.
UN Food and Agriculture Organization, Rome. Mayaux, P., Holmgren, P., Achard, F., Eva, H., Stibig, H.J., Branthomme, A., 2005.
FAO, JRC, 2012. Global Forest Land-Use Change 1990–2005. FAO Forestry Paper No. Tropical forest cover change in the 1990s and options for future monitoring.
169. UN Food and Agriculture Organization, Rome. Philos. Trans. Roy. Soc. B 360, 373–384.
FAO, JRC, 2014. Global Forest Land-Use Change 1990–2010. An update to FAO Mena, C.F., Bilsborrow, R.E., McClain, M.E., 2006. Socio-economic drivers of
Forestry Paper No. 169. UN Food and Agriculture Organization, Rome deforestation in the northern Ecuadorian Amazon. Environ. Manage. 37, 802–815.
(Unpublished). Morales, et al., 2015. Status and trends in global primary forest, protected areas, and
Gabler, K., Schadauer, K., Tomppo, E., Vidal, C., Bonhomme, C., McRoberts, R.E., areas designated for conservation of biodiversity from the Global Forest
Gschwantner, T., 2012. An enquiry on forest areas reported to the global forest Resources Assessment 2015 For. Ecol. Manage. 352, 68–77.
resources assessment: is harmonization needed? For. Sci. 58, 201–213. Romijn, E., Lantican, C., Herold, M., Lindquist, E., 2015. Assessing change in national
Gong, P., Wang, J., Yu, L., Zhao, Y., Zhao, Y., Liang, L., Niu, Z., Huang, X., Fu, H., Liu, S., forest monitoring capacities of 99 tropicalcountries. For. Ecol. Manage. 352,
Li, C., Li, X., Fu, W., Liu, C., Xu, Y., Wang, X., Cheng, Q., Hu, L., Yao, W., Zhang, H., 109–123.
Zhu, P., Zhao, Z., Zhang, H., Zheng, Y., Ji, L., Zhang, Y., Chen, H., Yan, A., Guo, J., Yu, Rudel, T., Roper, J., 1997. The paths to rainforest destruction, cross national patterns
L., Wang, L., Liu, X., Shi, T., Zhu, M., Chen, Y., Yang, G., Tang, P., Xu, B., Giri, C., of tropical deforestation 1975–90. World Dev. 25, 53–65.
Clinton, N., Zhu, Z., Chen, J., Chen, J., 2013. Finer resolution observation and Rudel, T.K., Coomes, O.T., Moran, E., Achard, F., Angelsen, A., Xu, J., Lambin, E.F.,
monitoring of global land cover: first mapping results with Landsat TM and 2005. Forest transitions: towards a global understanding of land use change.
ETM+ data. Int. J. Remote Sens. 34, 2607–2654. Glob. Environ. Change 15, 23–31.
Grainger, A., 2007. The role of end users in an international statistical process: the Smith, P., Bustamante, M., Ahammad, H., Clark, H., Dong, H., Elsiddig, E.A., Haberl, H.,
case of tropical forest statistics. J. Off. Stat. 23 (4), 553–592. Harper, R., House, J., Jafari, M., Masera, O., Mbow, C., Ravindranath, N.H., Rice,
Grainger, A., 2008. Difficulties in tracking the long-term global trend in tropical C.W., Robledo Abad, C., Romanovskaya, A., Sperling, F., Tubiello, F., 2014.
forest area. Proc. Natl. Acad. Sci. USA 105, 818–823. Agriculture, Forestry and Other Land Use (AFOLU). In: Edenhofer, O., Pichs-
Grainger, A., 2010. The bigger picture – tropical forest change in context, concept Madruga, R., Sokona, Y., Farahani, E., Kadner, S., Seyboth, K., Adler, A., Baum, I.,
and practice. In: Nagendra, H., Southworth, J. (Eds.), Reforesting Landscapes: Brunner, S., Eickemeier, P., Kriemann, B., Savolainen, J., Schlömer, S., von
Linking Pattern and Process. Springer, Berlin, pp. 15–43. Stechow, C., Zwickel, T., Minx, J.C. (Eds.), Climate Change 2014: Mitigation of
Grainger, A., Obersteiner, M., 2011. A framework for structuring the global forest Climate Change. Contribution of Working Group III to the Fifth Assessment
monitoring landscape in the REDD era. Environ. Sci. Policy 14, 127–139. Report of the Intergovernmental Panel on Climate Change. Cambridge
Hansen, M.C., DeFries, R., 2004. Long-term global forest change using continuous University Press, Cambridge, United Kingdom and New York, NY, USA.
fields of tree-cover maps from 8-km Advanced Very High Resolution UN, 2013. World Population Prospects: The 2012 Revision. DVD Edition. Population
Radiometer (AVHRR) data for the years 1982–99. Remote Sens. Environ. 94, Division, Department of Social and Economic Affairs, United Nations, New York.
94–104. UN, 2014. New York Declaration on Forests. In: Forests: Action Statements and
Hansen, M.C., Stehman, S.V., Potapov, P.V., 2010. Quantification of global forest Action Plans. Climate Summit 2014, 23 September 2014, New York. United
cover loss. Proc. Natl. Acad. Sci. USA 107, 8650–8655. Nations, New York, pp. 2–3.
Hansen, M.C., Potapov, P.V., Moore, R., Hancher, M., Turubanova, S.A., Tyukavina, A., UNCCC, 2014. Report of the Conference of the Parties on its nineteenth session,
Thau, D., Stehman, S.V., Goetz, S.J., Loveland, T.R., Kommareddy, A., Egorov, A., Warsaw, 11–23 November 2013. Addendum. FCCC/CP/2013/10/Add.1. UN
Chini, L., Justice, C.O., Townshend, J.R.G., 2013. High-resolution global maps of Framework Convention on Climate Change, Bonn.
21st-century forest cover change. Science 342, 850–853. World Bank, 2013. Atlas of Global Development: A Visual Guide to the World’s
Holmgren, P., Persson, R., 2002. Evolution and prospects of global forest Greatest Challenges, fourth ed. Collins, Glasgow.
assessments. Unasylva 53 (210), 3–9. Zon, R., 1910. The Forest Resources of the World. Bulletin No. 83. U.S. Department of
Hosonuma, N., Herold, M., De Sy, D., De Fries, R.S., Brockhaus, M., Verchot, L., Agriculture, Forest Service, Washington DC.
Angelsen, A., Romijn, E., 2012. An assessment of deforestation and forest Zon, R., Sparhawk, W.W., 1923. Forest Resources of the World. McGraw-Hill, New
degradation drivers in developing countries. Environ. Res. Lett. 7, 044009. York.
Forest Ecology and Management 352 (2015) 21–34
a r t i c l e i n f o a b s t r a c t
Article history: Forests are important sources of livelihoods to millions of people and contribute to national economic
Received 19 December 2014 development of many countries. In addition, they are vital sources and sinks of carbon and contribute
Received in revised form 25 May 2015 to the rate of climate change. The UN Food and Agriculture Organization has been collecting and present-
Accepted 26 May 2015
ing data on global forest resources and forest cover since 1948. This paper builds on data from FAO’s 2015
Available online 7 September 2015
Global Forest Resource Assessment (FRA) and presents information on growing stock, biomass, carbon
stock, wood removals, and changes of forest area primarily designated for production and multiple use
Keywords:
of the world’s forests.
Forest Resources Assessment 2015
Removals
Between 1990 and 2015, the total growing stock volume has increased in East Asia, Caribbean, Western
Wood fuel and Central Asia, North America, Europe (including the Russian Federation), and Oceania with the highest
Production area relative increase in East Asia and the Caribbean. In all other subregions the total growing stock volume
Carbon decreased. North and Central America, Europe and Asia report forest C stock increases while South
Growing stock America and Africa report strong decreases and Oceania reports stable forest C stocks. The annual rate
of decrease of forest C stock weakened between 1990 and 2015.
The total volume of annual wood removals including woodfuel removals increased between 1990 and
2011, but shows a remarkable decline during the 2008–2009 economic crisis.
Forest areas designated for production purposes differ considerably between subregions. The percent-
age of production area out of total forest area ranges between 16 percent in South America and 53 per-
cent in Europe. Globally about one quarter of the forest area is designated to multiple use forestry.
The balance between biomass growth and removals shows considerable sub-regional differences and
related implications for the sustainable use of forests.
Ó 2015 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction countries. In addition, they are vital sources and sinks of carbon
(C) and contribute to the rate of climate change.
Forests are important sources of livelihoods to millions of peo- Forest ecosystems cover roughly one third of the global land
ple and contribute to national economic development of many area and are among the most biologically rich and genetically
diverse ecosystems on earth. They contribute to soil formation
and water regulation and are estimated to provide direct employ-
q
This article is part of a special issue entitled ‘‘Changes in Global Forest Resources ment to at least 10 million people, apart from being a source of
from 1990 to 2015’’. livelihoods to millions more (FAO, 2010). It is estimated that about
⇑ Corresponding author. 410 million people are highly dependent on forests for subsistence
E-mail addresses: Michael.koehl@uni-hamburg.de (M. Köhl), R.Lasco@cgiar.org and income, and 1.6 billion people depend on forest goods and ser-
(R. Lasco), mcifuentes@catie.ac.cr (M. Cifuentes), Orjan.Jonsson@fao.org
vices for some part of their livelihoods (Munang et al., 2011). Wood
(Ö. Jonsson), kari.t.korhonen@luke.fi (K.T. Korhonen), Philip.mundhenk@uni-
hamburg.de (P. Mundhenk), jnavar@ipn.mx (J. de Jesus Navar), Graham.Stinson@ and manufactured forest products add more than $450 billion to
nrcan-rncan.gc.ca (G. Stinson). the world market economy annually, and the annual value of
http://dx.doi.org/10.1016/j.foreco.2015.05.036
0378-1127/Ó 2015 Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
internationally traded forest products is between $150 billion and 2000, 2005, 2010 and 2015). As expected, there were significantly
$200 billion. more missing values for dead wood, litter and soil C stocks.
Since 1948, the FAO has been conducting the Global Forest We made estimates for missing values in order to evaluate
Resources Assessments (FRA) which are now produced every five sub-regional and global totals. Missing values were categorized
years. It aims to provide a consistent approach to describing the in two main classes: (1) at least one value was reported but time
world’s forests and how they are changing. FRA has become a series is incomplete, and (2) no value was reported for any year.
major source of information on global forests for policy making In order to have complete time series for countries that had
and for the development of international programs and initiatives. reported C stock for at least one year, estimates were made by mul-
In this paper we report on the results of FRA 2015 on the state and tiplying forest area for the missing years by C stock density
trends of forest production, biomass and carbon. Specifically, we (Mg C ha 1) for the closest year reported. Sub-regional mean C
analyze the latest trend in growing stock, biomass, carbon, wood stock densities were then calculated. In order to make estimates
removals, production and multiple use forest area changes. for countries that did not report any value, the sub-regional C stock
density was multiplied by forest area for the corresponding miss-
2. Methods ing year. No country in East Asia reported on C in litter, so the
sub-regional average of South and Southeast Asia was used
FRA 2015 provides reports on 234 countries and territories, of instead. No country in Central America reported on C in soil, so
which 155 come from country reports prepared by national corre- sub-regional average from Caribbean was used instead. Most coun-
spondents nominated by government agencies responsible for for- tries reported soil C to the specified common soil depth of 30 cm. A
estry. The remainder comes from desk studies, which since FRA few countries used other soil depths, but this was not adjusted for
2000 have been used to provide estimated values for forest statis- because it does not affect the trend over time as the same soil
tics in countries which have not provided a country report. While depth was used for all reporting years.
the number of desk studies is high, in total they represent only
0.5% of global forest area. Data for this study were kindly made 2.3. Wood removals
available by FAO.
FRA 2015 utilizes a questionnaire approach to collect forest Wood removals have been reported in the FRA since 2000.
related national statistics. The data collection process for FRA Before 2000, FRA utilized the attribute ‘‘Production’’, which is more
2015 was undertaken collaboratively with Forest Europe, or less synonymous to removals. Wood removals refer to the total
Montréal Process, International Tropical Timber Organization and volume that is actually removed from forests and does not include
l’ Observatoire des Forêts d’Afrique Centrale using a Collaborative the volume of cut trees, logs and logging residuals left in the forest.
Forest Resources Questionnaire (CFRQ) and the data were reported Thus, the volume classified as wood removals is generally smaller
in the Forest Resources Information Management System (FRIMS) than the volume of fellings. However, in rare situations the volume
hosted by FAO in Rome. This information collection process is of removals can exceed the volume of fellings, such as when har-
not a sample based survey and thus does not permit the applica- vesting operations are utilized to remove timber from fellings in
tion of probability based procedures for estimating sampling errors earlier periods or trees killed or damaged by natural causes.
and confidence intervals. Data submission is not mandatory and Wood removed from the forests is either used for energy produc-
the individual countries are responsible for reporting comprehen- tion or as raw material for the production of goods. The term
sive, up-to date, reliable and representative data. For further ‘‘woodfuel removals’’ is used for the wood removed for energy pro-
details on the collection and validation of national data for FRA duction purposes, regardless whether for industrial, commercial or
2015 see MacDicken (2015). domestic use, the term ‘‘industrial roundwood removals ’’
describes the wood removed (volume of roundwood over bark)
2.1. Growing stock for production of goods and services other than energy production
(wood fuel) (FAO, 2010).
184 out of 234 countries and territories reported the growing For the period from 1990 to 2011 countries were provided
stock volume estimates for the reference year 2015. These coun- annual data on wood removals from FAOSTAT and were asked to
tries and territories represent 93.5% of the global forest area. For check and correct whenever necessary. A total of 171 countries pro-
some countries, growing stock data were reported for 2015, but vided information for 1990. The number of countries that reported
missing for some of the other reporting years. We made the follow- information on wood removals consistently increased over time,
ing estimations to get global and sub-regional estimates for grow- with 196 countries provided information for 2011. The global forest
ing stock in 2015. First, in order to have complete time series for area covered by the reporting increase from 92 percent in 1990 to
countries that had reported growing stock for at least one year, 98 percent in 2011. Adjustments for missing data were not applied
estimates were made by multiplying forest area for the missing because wood removals are driven by human interventions that
years by growing stock density (m3 ha 1) for the closest year cannot be approximated using simple imputations.
reported. Sub-regional growing stock densities were then calcu-
lated. Secondly, estimates were made for countries that did not 2.4. Production and multiple use forest area changes
report any value by multiplying the sub-regional mean growing
stock density by forest area for the country and reporting year in The FRA reporting categories differentiate between
question. The estimated data for these latter countries were not
considered in our trend analyses. production forest, which is the forest area designated primarily
for production of wood, fiber, bioenergy and/or non-wood forest
2.2. Biomass and carbon products and
multiple use forest, which is the forest area designated for more
Most Country Reports included reporting on carbon (C) stocks, than one purpose and where none of these alone is considered
but not all included complete reporting. A total of 173 countries as the predominant designated function.
representing 98% of the global forest area reported biomass C
stocks (Fig. 1). Of these, 167 countries representing 84% of the glo- The multiple use concept applied in FRA does not imply that
bal forest area reported a complete time series (including 1990, multiple use forests take into account all of forest functions
(A) Biomass (B) Dead wood
Percent of area with reporng Percent of countries reporng Percent of area with reporng Percent of countries reporng
100% 100% 100% 100%
0% 0% 0% 0%
Europe North and South Africa Asia Oceania World Europe North and South Africa Asia Oceania World
Central America Central America
America America
0% 0% 0% 0%
Europe North and South Africa Asia Oceania World Europe North and South Africa Asia Oceania World
Central America Central America
America America
simultaneously. But these forests are managed to serve more than stock in 1990. Between years 1990 and 2000 the global growing
a single primary designated function. stock volume was slightly decreasing, thereafter slightly increas-
The ‘‘primary designated function’’ refers to the primary func- ing. Between 1990 and 2015, the total growing stock volume has
tion or management objective assigned to a management unit increased in 6 out of the 12 subregions: East Asia, Caribbean,
either by legal prescription, documented decision of the Western and Central Asia, North America, Europe, and Oceania
landowner/manager, or evidence provided by documented studies (Fig. 2). In relative terms, highest increase took place in East Asia
of forest management practices and customary use. No adjust- and the Caribbean, over 40% in both subregions. The increase of
ments were made for non-responses. growing stock volume in the Caribbean subregion is explained by
the increase of forest area by almost 50% between the years 1990
2.5. Categories and 2015. In East Asia, both forest area and mean stocking per hec-
tare increased. In absolute terms, the increase was highest in North
Results are presented for three different stratification America (7.1 billion m3) and Europe (10.9 billion m3). In both of
categories: these sub-regions, the increase of growing stock volume is
explained by increased mean stocking per hectare (Table 2).
(1). Domains (Boreal, Temperate, Subtropics, Tropics) Between 1990 and 2015, the growing stock volume decreased
(2). Subregions (Caribbean, North America, Central America, in 6 sub-regions: Central America, Eastern and Southern Africa,
South America, Northern Africa, Western and Central South and Southeast Asia, Northern Africa, Western and Central
Africa, Eastern and Southern Africa, Europe, Western and Africa, and South America. In relative terms, the highest decrease
Central Asia, East Asia, South and Southeast Asia, Oceania) took place in Central America (16.1%) and Eastern and Southern
(3). Income groups (see Table 1). Africa (12.6%). In absolute terms, the largest decrease took place
in South America (10,843 million m3).
For further details see MacDicken (2015). The average forest growing stock per hectare in 2015 varies by
sub-regions from less than 60 m3 ha 1 in Eastern and Southern
Africa and North Africa to more than 200 m3 ha 1 in Oceania
3. Results
(Table 2). Between 1990 and 2015 the average growing stock
increased in most of the sub-regions and clearly decreased only
3.1. Growing stock and biomass
in South and Southeast Asia. Very slight decrease is observed also
in the Caribbean sub-region and North Africa. In most of those
The total global growing stock estimate is 530.5 billion m3 for
sub-regions where total growing stock volume decreased, the
2015. This is 3.6 billion m3 (0.7%) more than the total growing
decrease has been primarily caused by loss in the forest area, not
Table 1 by degradation or shifts from older to younger age class distribu-
Income groups as defined by the World Bank (Source: http://data.worldbank.org/ tions caused by natural disturbance or harvesting (Fig. 3). Two
about/country-and-lending-groups).
sub-regions, South & South-East Asia and North-Africa show
Income group Index GNI per Capita decrease both in mean stocking and forest area. In North-Africa
Low income economies L 6US$ 1.045 both these changes are nominal, whereas in South & South-East
Lower middle income economies LM US$ 1.046 to US$ 4.125 Asia both the decrease in forest area and decrease in mean stocking
Upper middle income economies UM US$ 4.126 to 12.745 are remarkable leading a clear decrease in total growing stock
High income economies H PUS$ 12.746
(Fig. 2). In 4 sub-regions, Central America, South America,
Fig. 2. Volume of growing stock (million m3) 1990–2015 by sub-regions.
Table 2 Western & Central Africa and Eastern & Southern Africa the forest
Mean volume (m3/ha) of growing stock in forest from 1990 to 2015 by sub-regions. area change is negative but mean stocking positive. In all these
Only countries that have reported both forest area and growing stock volume
included for each reporting year.
sub-regions, the increase in mean stocking is too low to fully com-
pensate the decrease in forest area, leading to decrease in total
Subregion Year growing stock. The positive trend in mean stocking may be caused
1990 2000 2005 2010 2015 by plantations in these sub-regions. In 5 sub-regions, North
Caribbean 78.8 86.7 89.3 78.1 78.0 America, East Asia, Europe, Oceania, Western & Central Asia both
Central America 141.7 140.6 140.5 139.9 158.5 the forest area and mean stocking trends are positive. Naturally,
East Asia 76.8 82.4 84.2 88.5 90.8 the total growing stock has increased in these sub-regions.
Eastern and Southern Africa 56.2 56.1 56.5 56.8 57.2
Caribbean is the only sub-region where the mean stocking has
Europe 104.3 107.4 109.2 111.2 112.9
North America 112.6 115.7 121.4 124.4 128.1 decreased and forest area increased. The changes in mean stocking
Northern Africa 59.2 61.7 58.2 58.0 58.3 is nominal and therefore the total growing stock has slightly
Oceania 194.2 196.4 197.9 200.0 201.7 increased in the Caribbean sub-region (Fig. 2).
South America 172.8 174.4 175.0 176.1 178.2
South and Southeast Asia 102.8 105.1 102.9 100.9 98.5
Western and Central Africa 191.5 193.0 193.8 194.7 195.5
Western and Central Asia 68.6 70.6 70.0 71.8 73.7
3.2. Carbon
Pg C
Pg C
-4 Africa
-0.2
-6 Asia
-0.4
-8
-0.6 Oceania
-10
-12 -0.8 World
-14 -1
1990 2000 2005 2010 2015 1990 2000 2005 2010 2015
2
5
1
0
0
Pg C
Pg C
-1 -5
-2
-10
-3
-15
-4
-5 -20
1990 2000 2005 2010 2015 1990 2000 2005 2010 2015
15
Fig. 4. Forest carbon stock changes since 1990 (1 Pg = 10 g = 1 billion tons).
200 200
120 120
Mg C ha-1
Pg C
100 100
80 80
60 60
40 40
20 20
0 0
Eastern and Southern Africa
Central America
Northern Africa
East Asia
Central America
North America
South America
Northern Africa
North America
South America
Caribbean
Caribbean
East Asia
Europe
Oceania
Europe
Oceania
WORLD
Fig. 5. Forest C stocks and C stock densities in global forest sub-regions, 2015.
America (122.4 Mg C ha 1) and Western and Central Africa 3.3. Wood removals
(120.6 Mg C ha 1). The highest dead wood and litter C densities
were reported in North America (12.1 and 29.5 Mg C ha 1, respec- The total volume of annual wood removals increased from
tively). The highest soil C densities were reported in Europe 2.75 billion m3 to around 3.0 billion m3 from 1990 to 2011 as
(94.6 Mg C ha 1), a sub-region that is dominated by the extremely reported. Fig. 6 presents the development of wood removals for
large forest area in the Russian Federation, making Europe the (a) countries that reported for all points in time and (b) the data
sub-region most dominated by forests of the boreal biome. Forest provided for a specific point in time, which includes countries that
C stocks are largest in Europe, South America and North America missed reporting for one or more reporting periods. The similarity
because these sub-regions host the largest forest areas (26%, 22% in the shape of both curves and their pronounced difference in
and 16% of the world’s forest area, respectively). 1993 suggest that non-response was mainly a problem in the time
Fig. 6. Wood removals between 1990 and 2011.
period from 1990 to 1992. The development of annual wood With reference to domains, the total volume of woodfuel
removals over time was not continuous. Between 1990 and 1993 removals shows the same ranking as wood removals: they are
the total volume of timber declined, followed by a constant highest in the Tropics, followed by the Temperate, Boreal and
increase until 2005. In 2008 and 2009 a reduction was recorded, Subtropics domain. A relatively continuous growth is observed in
coincident with the global economic crisis. In 2010 and 2011 the the Tropics, while moderate changes are found for the other
total volume of wood removals recovered to the levels reported domains (Fig. 8).
for the period 2005 to 2007 for both the countries that reported Fig. 9 gives a more detailed picture. Wood removals were
in every year and for all countries that reported (Fig. 6). highest in North America, Western and Central Asia and South
Woodfuel removals show a different development but it should and Southeast Asia and lowest in the Caribbean, Central
be noted that woodfuel statistics maintained by FAO is largely America, Western and Central Africa, East Asia, and Oceania.
based on estimation methods (Whiteman et al., 2002). For 80% of Compared with the figures on forest areas mainly designated
the reported volume the woodfuel removals were estimated for for production, as presented in Fig. 11, reveals substantial differ-
the year 2011 compared with 84% for the year 2000, over time data ences in area related utilization rates in the sub-regions. Europe
is improving and share of estimates decrease for historical data as takes an intermediate position in the total volume of wood
national data becomes available. They increased almost continu- removals, but reported the largest amount of areas primarily des-
ously between 1990 and 2011, starting at 1134 million m3 in ignated for production. In contrast, North America realizes most
1990 and reaching 1343 million m3 in 2011 (Fig. 7). wood production in multiple use forests (Fig. 13). In Western
Wood removals are highest in the Tropics and lowest in the and Central Asia the level of wood removals is high while only
Subtropics. The global decline in 2008 and 2009 (Fig. 6) can only a relatively small forest area is primarily designated for produc-
be observed in the Boreal and Temperate domains (Fig. 8). While tion forest or multiple use. For most subregions wood removals
the total volume of wood removals shows a notable increase in increased over time. South America shows a decline between
the Tropics, it remains constant in the Subtropics. In the Boreal 1997 and 2000. In Europe a pronounced decline between 1995
domain a clear decline between 1990 and 1996 is followed by a and 1996 is followed by a recovery until 2007. Both subregions,
recovery until 2005. The wood removals in the temperate domain North America and Europe show a downturn in 2007 and 2008,
fluctuated between 1990 and 2007 before the economic crisis of coincident with the economic crisis, while other sub-regions did
2008. not indicate any reduction.
Woodfuel removals are highest in South and Southeast Asia. increasing, with the largest increase in South and Southeast Asia.
Western and Central Africa, Eastern and Southern Africa, South At least in Europe the increase is likely driven by the increasing
America, and Europe take a central position, in all other subregions demand for renewable energies. The EU Renewable Energy
woodfuel removals are low. In Eastern and Southern Africa, Europe, Directive (EU, 2009) requires the EU to fulfill at least 20% of its total
and South and Southeast Asia woodfuel removals are consistently energy needs with renewables by 2020, which should be achieved
Fig. 9. Wood removals (above) and Woodfuel removals (below) by sub-region 1990–2011.
through the attainment of individual national targets. All EU coun- wood removals are slightly higher than for the low income group.
tries must also ensure that at least 10% of their transport fuels In the low and lower middle income group wood removals
come from renewable sources by 2020. increased over time, while in the upper middle and high income
A comparison of wood removals with income groups shows a groups substantial fluctuations can be observed. In the high
mixed pattern (Fig. 10). The level of wood removals increased with income group a depression during the first seven years of the
income groups except for the upper middle income group, where reporting period is followed by an increase and a relatively steady
Fig. 10. Wood removals (above) and Woodfuel removals (below) by income group 1990–2011.
state since 2005. The same pattern is found for the high income development patterns differ. Woodfuel removals decrease from
group, except for the decline in 2007 and 2008. 1990 to 2001 in the upper middle income group followed by a pro-
Woodfuel removals are highest in the low middle income nounced increase between 2004 and 2005 and a slight decrease
group, followed by the low income group. In both groups wood from 2005 to 2011. A downturn between 1991 and 1996 can be
removals are increasing over time. The level of woodfuel removals observed in the upper income group with a subsequent small fluc-
is comparable in upper middle and high income group, but the tuation after 1998. In some countries, wood removals from trees
Fig. 11. Area of production forests by sub-region.
outside of forests (i.e. other lands with tree cover) may be consid- Assigning the area of forests mainly used for production to
erable, but here we make no distinctions between woodfuel income levels gives a relatively uniform picture; most of the forest
removals from forests versus harvesting trees outside of forests. area designated for production is found in higher income countries
(Fig. 12). Production forests in low income countries is roughly
3.4. Production and multiple use forest area changes 100 million hectares while more than 600 million hectares are des-
ignated primarily for production in areas with high income. The
In 2015, 190 countries representing 97 percent of the global for- trends over time indicate only minor differences.
est area reported on production forest area. The number of coun- About 26 percent or 1.049 million hectares of the total forest
tries reporting on multiple use forest area is smaller (183 area are primarily designated for multiple use. Between 1990
countries or 95 percent of the global forest area). There are marked and 2015 the area of forests designated for multiple use
differences between sub-regions in the designation of forest areas decreased in North America, Western and Central Africa, Europe
for production purposes (Fig. 11). In Europe more than 500 million and South and Southeast Asia, while in South America and East
hectares or 52 percent of forests are designated primarily for pro- Asia an increase can be seen. In Eastern and Southern Africa the
duction, a number that is significantly larger than in other decline between 1990 and 2010 could be reversed, so that in
sub-regions. In the Caribbean, Central American, Northern 2015 more areas are designated to multiple use forestry than in
African, Western and Central Asian and Oceania sub-regions the 1990 (Fig. 13).
area of production forests is smaller than 20 million hectares. In
the remaining sub-regions the respective areas range between 4. Discussion
60 million and 120 million hectares. In Western and Central
Africa 30 percent of the total forest area are designated as produc- The results presented are based on the best information that
tion forest and in Eastern and Southern Africa 22.3 percent, in countries can provide. Even so, the quality and reliability of indi-
North America 16.7 percent, in South America 16.2, in Western vidual figures varies considerably because not all countries have
and Central Asia 23.3, in East Asia 44.5, and in South and well established forest inventory and monitoring programs. The
Southeast Asia 45.7 percent. number of countries providing data for the different information
Fig. 11 presents the development over time for the total of the groups requested by the questionnaire varies and thus response
reporting countries (bars) and the countries that reported in all rates show differences between reporting years, sub-regions and
years (lines). In Africa the area of production forests decreased over attributes. The data are weakest for trend analysis, just as they
time, which could be caused by the general reduction of forested were in previous FRA (Marklund and Schoene, 2006; MacDicken,
area. In North and South America as well as in South and 2015). Where changes are driven by biophysical processes, such
Southeastern Asia and Oceania the area of forests primarily desig- as climate, site index or soil carbon, statistical interpolation for fill-
nated for production increased since 1990. In East Asia a decline ing missing values might be an appropriate measure to increase
from 1990 to 2005 was followed by a slight increase until 2015. data completeness. Where developments are mainly a result of
In Europe there is a pronounced decrease between 1990 and human interventions (e.g. wood removals, land-use change, forest
2000, followed by a steady state in 2005, 2010 and 2015. The management), spatial and temporal changes are subject to com-
decrease could be a result of European activities for an increase plex interrelations of a variety of agents. In these situations it is
of areas mainly designated for conservation and protection, such not advisable to fill data gaps by interpolation or other probability
as the EU Habitat Directive on the protection of natural habitats based approaches. Therefore we undertook gap filling for missing
and of wild flora and fauna (EU, 1992). growing stock, biomass and carbon stock data, but not for wood
Fig. 12. Production forest area by income 1990–2015.
removals or forest areas designated to specific uses. Some diagnos- the overall global BCEF of 0.91 in the FRA 2015 data set is almost
tics reveal problematic national reporting data and it is tempting to identical to the 0.92 in FRA 2010 (FAO, 2010). Considering the
apply corrections to these data before analyzing them. For exam- diversity of national circumstances in the more than 200 countries
ple, biomass conversion and expansion factors (BCEF) implied by reporting, it is expected that some of the data will look strange.
the growing stock and biomass C stock data reported by countries Overall, however, the data appear to be plausible and they are use-
do not all appear to be reasonable (Fig. 14). But the largest coun- ful for the purposes of sub-regional and global assessment.
tries that are most influential on sub-regional and global means Between 1990 and 2015, the total growing stock volume has
and totals appear to have reasonable implied BCEF values, and increased in East Asia, Caribbean, Western and Central Asia,
3.50 600
Biomass conversion & expansion factor
3.00
500
2.50
2.00
1.00
300
0.50
FRA 2015
0.00 200
0 100 200 300 400 500 600 700 800 900 FRA 2010
answer may lie in the FRA 2015 data. Like data reported for previ- Change in biomass carbon density
ous FRA, these data have some peculiarities (Marklund and Fig. 17. Country-level forest area and carbon density changes between 1990 and
Schoene, 2006). The quality and reliability of individual figures 2015.
reported by countries varies considerably with many of the
reported data having lower tiers (less reliable). For example, only FRA 2015 data, the average C density of the world’s forests
5% of countries representing 9% of the forest area provided Tier 3 increased by 0.6% between 1990 and 2005 and by 0.1% between
data for soil carbon stocks. But the discrepancy between the sink 2005 and 2015. Pan et al. reported a 2.0% increase in forest C den-
reported by Pan et al. (2011) and the C stock loss calculated from sity between 1990 and 2007, including a 22.5% increase in tropical
FRA 2015 data is not entirely caused by problems in the FRA data. forests that are recovering from past deforestation and logging, and
First, not all CO2 taken up by forests is sequestered in forest ecosys- a 5.0% increase in tropical intact forests. Permanent sample plots in
tem C pools. Some is transferred laterally out of the forest and tropical intact forests are few, but they indicate that these forests
sequestered or emitted back into the atmosphere elsewhere. are accumulating biomass and C (Baker et al., 2004; Phillips
Global wood removals in 2005 totaled 3.0 109 m3, which et al., 2008; Lewis et al., 2009). Pan et al. estimate that tropical
amounts to roughly 0.3 Pg C year 1. Cole et al. (2007) conserva- intact forests are accumulating C at a rate of 1.1 Pg C year 1. In
tively estimate that 1.9 Pg C year 1 are transferred from terrestrial contrast, the FRA 2015 data only report an accumulation of
ecosystems to inland waters, but only a portion of this is trans- 0.1 Pg C year 1 in these forests. This difference accounts for the
ferred from forests. majority of the discrepancy between Pan et al. and FRA in
The second reason for the apparent discrepancy between Pan Fig. 16. Considering our discussion of tropical biomass C stock den-
et al.’s sinks and FRA’s reported C stock losses arises from differ- sities, above, and the equally large (and important) uncertainties
ences in how tropical forest C dynamics are estimated and about dead wood, litter and soil C stocks and stock changes, much
reported. Many countries rely on Intergovernmental Panel on work remains to be done to refine our knowledge of the world’s
Climate Change (IPCC) default expansion factors (IPCC, 2003) to forest C stocks and fluxes.
calculate their C stocks for FRA, or they only have an estimate of The total volume of annual wood removals including woodfuel
forest C stocks from one point in time, which was used by them removals increased between 1990 and 2011. Wood removals show
(or us, see Section 2.2) to calculate C stock density and apply to a remarkable decline in 2009, but recovered until 2011. Forest
the forest areas in the other years to calculate C stocks for those areas designated for production purposes differ considerably
years. Fig. 17 shows these countries clustered along the vertical, between subregions (16.2 percent in South American and 52 per-
zero percent change in biomass C density axis. According to the cent in Europe). Globally about one quarter of the forest area is
2.5
designated to multiple use forestry.
a r t i c l e i n f o a b s t r a c t
Article history: The world’s forests provide fundamental protection of soil and water resources as well as multiple
Received 5 December 2014 ecosystem services and cultural or spiritual values. We summarized the FRA 2015 data for protective
Received in revised form 22 March 2015 functions and ecosystem services, and analyzed increasing or decreasing trends of protective areas.
Accepted 24 March 2015
The global forest area managed for protection of soil and water was 1.002 billion ha as of 2015, which
Available online 7 September 2015
was 25.1% of all global forested areas. Protective forests have increased by 0.181 billion ha over the past
25 years mainly because more countries are now reporting protective forest areas (139 in 2015 vs 114 in
Keywords:
1990). However, average percentage of designated for protective forests did not change significantly from
Sustainable forest management
Multiple functions
1990 to 2015. Global forest area managed for ecosystem services is also now at 25.4% of global total forest
Forest conservation area and has changed little over the past 25 years. Among the twelve categories of protective forests,
Ecosystem value flood control, public recreation, and cultural services increased both in terms of percentage of total forest
Criteria area and the number of reporting countries. Public awareness of the importance of forest resources for
Indicator functions and services other than production continues to increase as evidenced by the increase of pro-
tective forest designations and reporting in many countries. Percentages of total forest area designated
for both protective forests and ecosystem services show a dual-peak distribution of numbers of countries
concentrated at 0% and 100%. This suggests a socio-economic influence for the designations. We exam-
ined five case study countries (Australia, Canada, China, Kenya, and Russia). The most dramatic changes
in the past 25 years have been in China where protective forests for soil and water resources increased
from about 12% to 28% of forest area. The Russian Federation has also increased percentages of forest area
devoted to soil and water resource protection and delivery of ecosystem services. Australia is now report-
ing in more protective forest categories whereas Kenya and Canada changed little. These five countries
have their own classification of forest functions and recalculation methods of reporting for FRA 2015
were different. This demonstrates the difficulty in establishing a universal common designation scheme
for multi-functions of forest. Production of more accurate assessments by further improvements in the
reporting framework and data quality would help advance the value of FRA as the unique global database
for forest functions integrated between forest ecosystems and social sciences.
Ó 2015 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
http://dx.doi.org/10.1016/j.foreco.2015.03.039
0378-1127/Ó 2015 Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
36 S. Miura et al. / Forest Ecology and Management 352 (2015) 35–46
conference spurred people to promote a variety of activities for attention and importance. FAO (2008) conducted an interesting
sustainable forest management. Chapter 11 of Agenda 21 practical experience of compensation mechanisms for water ser-
(‘‘Combating Deforestation’’) is particularly relevant in this vices provided by forests in Central America and the Caribbean,
context. In the summary of this chapter, Keating (1993) writes: however its calculation remains a challenge (FAO, 2004b). Recent
‘‘forests are a source of timber, firewood and other goals. They also advances in Clean Development Mechanism (CDM) and Reducing
play an important role in soil and water conservation, maintaining Emissions from Deforestation and Forest Degradation (REDD) in
a healthy atmosphere and maintaining biological diversity of the 2000s under the UN Framework Convention on Climate
plants and animals. . . there is an urgent need to conserve and plant Change (UNFCCC) also require a reliable reporting of objective for-
forests in developed and developing countries to maintain or ests (REDD Research and Development Center, 2012). Demand for
restore the ecological balance, and to provide for human needs’’. reporting multiple functions of forests has increased in importance
It is generally accepted that forests and trees, in undisturbed based on these rising social concerns in the field of environmental
form, provide the greatest vegetative protection against erosion economics.
from rain, wind, and coastal waves (Broadhead and Leslie, 2007; FRA, the only statistical forest database covering the whole
Hamilton, 2008). Accordingly, they also significantly contribute globe, has attempted to assess the extent of protective forests in
to the reduction of downstream sedimentation (Fu, 1989). The root the world. The inclusion in FRA of the protective function of forests
system of the trees creates increased soil strength (Greenwood gradually developed in parallel with the increasing importance for
et al., 2004; Reubens et al., 2007). Forests and trees contribute to the global community assigned to this function. FRA first intro-
the preservation of a good soil structure thanks to the protection duced a concept of ‘protective function’ of forest as non-wood ben-
against splash erosion (provided the litter layer and the understory efits in FRA 1990 only for developed countries (FAO, 1995) and
vegetation are maintained) and maintenance of robust biological made the first comprehensive report of protective functions of for-
activity in the soil (Binkley and Fisher, 2013). In this context, for- est in FRA 2005 as ‘‘More than 300 million hectares of forests are
ests and trees also contribute to the mitigation of risks of shallow designated for soil and water conservation’’ (FAO, 2006a). In the
landslides. However, deep-rooted mass movements triggered by report of FRA 2015, FAO created separate main categories for pro-
tectonic events cannot be prevented by forests and trees tective functions and selected ecosystem services since 1990 (FAO,
(Hamilton, 1986; Government of Japan, 2002; Dolidon et al., 2009). 2012). An initial evaluation of the status and trend of forest protec-
Clean water is becoming more recognized as one of the most tive functions over the past twenty-five years can provide the basis
important environmental services provided by forests and trees for further detailed analysis of the importance of these forest func-
(FAO, 2013). At least one third of the world’s largest cities draw a tions to the international forestry community and other related
significant proportion of their drinking water from forested areas environmental sciences.
(FAO, 2013). It is also well established that forests play a crucial In this paper we analyze the FRA reported data in two main cat-
role in the hydrological cycle. Forests influence the amount of egories of protective functions and selected ecosystem services. In
water available and regulate surface and groundwater flows while addition, we analyze the status and trend data in several protective
maintaining high water quality (Aust and Blinn, 2004; Hamilton, forest sub-categories. We tested the effects of sub-regional,
2008). Forests and trees contribute to the reduction of water- latitude-affected climatic, and socio-economic differences and
related risks such as floods and droughts and help prevent deserti- temporal changes on the main category and sub-category protec-
fication and salinization (FAO, 2013). However, there is sufficient tive forest variables according the FRA reporting framework
scientific evidence that forests are not able to prevent or even (FAO, 2014a). The trend analyses are based on percentages of total
reduce medium to large scale floods (FAO and CIFOR, 2005; forest area or total land area and not on absolute forest area. We
Hofer and Messerli, 2006; Hamilton, 2008). Policy makers have also discuss, as case studies, the status and trend of protective for-
voiced concern about the effectiveness and limitation of these reg- ests in selected countries located in different regions and climatic
ulating and provisioning services of forests (Cubbage et al., 2007; domains. Finally, we discuss key findings and future recommenda-
Collaborative Partnership on Forests, 2014). tion to FRA for improving the reporting of protective functions and
In the context of climate change and the resulting increased ecosystem services.
incidence of natural hazards, the soil and water protection function
of forests and trees is becoming increasingly important. For the 2. Methods
maintenance and sustainability of this function, forest manage-
ment through a watershed (landscape) approach is very important 2.1. Data source and compilation
(Kammerbauer and Ardon, 1999; Postel and Thompson, 2005).
Watershed management includes the management of all available The FRA 2015 dataset (http://www.fao.org/forestry/fra/fra2015/
natural resources (including forests) in a comprehensive way and en/) is described by MacDicken (2015). We used FRA 2015 data
makes the link between natural resources management and the submitted by countries in response to the question ‘‘How much
improvement of livelihoods. It provides a framework to organize forest area is managed for protection of soil and water and ecosys-
different land-uses (forestry, pasture, agriculture) in an integrated tem services?’’
way (Turner, 1989). Watershed management contributes to the There are two main categories and ten sub-categories of protec-
reduction of risks of natural hazards, such as landslides and local tive forests designated for specific purposes of providing protection
floods, and creates local resilience against climate change as well against events that damage forest resources as well as for provid-
as adaptation options (FAO, 2006b, 2007). ing various types of ecosystem services. The main categories are
The soil and water protection function of forests and trees offers protective forests for soil and water resources and protective for-
significant scope for the establishment of payment for ecosystem ests for delivery of ecosystem services. Within the soil and water
services (PES) schemes. PES has been developing rapidly under resource protection category are protective forests for the sub-
the framework of the Convention of Biological Diversity (CBD) sup- categories of (1) clean water, (2) coastal stabilization (3) desertifi-
ported by Millennium Ecosystem Assessment and environmental cation control, (4) avalanche control, (5) erosion and flood control,
economics (ex. Costanza et al., 1997; Kumar, 2010). In the context and (6) other control. Within the ecosystem services category are
of large economic losses by floods and sediment disasters, the cal- protective forests for the sub-categories of (1) public recreation,
culation of ecosystem values of services related to soil and water (2) carbon storage, (3) cultural services, and (4) other services
protection provided by forests and trees is getting increased (excluding Table 6 for conservation of biodiversity).
S. Miura et al. / Forest Ecology and Management 352 (2015) 35–46 37
2.2. Data analysis subtropical domain), Canada (North and Central American region,
boreal domain), China (Asian region, temperate domain), Kenya
Protective forest area data are reported in kilohectare (kha) (African region, tropical domain), and the Russian Federation
units for each country that provides such information. These area (Europe region, boreal domain). Russia is a special case, geograph-
data range more than six orders of magnitude because of the large ically. It is included in the global region of Europe in the FRA data-
variation in land area and forest areas within a country and are base, but because it stretches across a wide longitudinal area, it
non-normally distributed. To compare protective forest areas also includes large land area in Asia. All five countries report data
among countries of such large size variability, area data were cal- quality in the tier 2 or 3 reliability categories. Thus, countries with
culated as percentages of total land area (remains invariant over the least reliable data tier were excluded from these detailed case
total 25-y reporting period) for each country and as percentages studies.
of total forest area (which may change over reporting period).
This enabled us to avoid too much influence by a few large coun-
3. Results
tries. Protective forest areas were also calculated on a per capita
basis. Expressing the protective forest area data as a % of total land
3.1. Protective forests for soil and water resources
area or forest area basis allows for comparisons among countries of
vastly different sizes and has the added advantage of producing a
Globally, the Earth has about 3.999 billion ha of forest area as of
more normal data distribution. We referred to sum of areas of main
the 2015 report year, a decline of about 0.129 billion ha since 1990
or sub-categories of protective forests and its proportion to total
(Fig. 1 top). Of this total forest area, as of 2015, about 1.002 bil-
forest area or land area in those cases where we need to clarify
lion ha (25.1% of global total forest area) has been designated by
the absolute status of protective forests.
the various countries and territories for the protection of soil and
Statistical data analysis for this paper was generated using SAS
water resources, an increase of 0.181 billion ha since 1990. At least
software, version 9.4 of the SAS System (Ó 2013, SAS Institute Inc.,
part of the increase is due to more countries reporting protective
Cary, NC, USA). The SAS generalized linear mixed model (GLIMMIX)
forests (114 in 1990 vs 139 in 2015 out of 234 recognized countries
software was used to test for differences among global sub-regions
and territories) (Table 1 and Fig. 1 top). In percentage terms, global
(northern Africa, eastern and southern Africa, western and central
forest area has still declined (Keenan et al., 2015) from 32.3% of
Africa, western and central Asia, east Asia, south and southeast
total land area in 1990 to 31.3% of total land area in 2015 (Fig. 1
Asia, Europe, Caribbean, Central America, North America, South
bottom). On the other hand, protective forest area remained fairly
America, and Oceania), climatic domains (polar, boreal, temperate,
constant over the same time interval with a mean of 35.9% of total
subtropical, and tropical), income level (low, low middle, high mid-
forest area (Table 2 and Fig. 1 bottom). Values of the percentage for
dle, and high), and report years (1990, 2000, 2005, 2010, and 2015)
each country showed a U-shaped histogram polarized to 0 or 100%
for each protective forest sub-category with the data expressed in
terms of % of total land area, % of forest area, and on a per capita
basis. Normal distribution of the residuals was tested using the Global total forest and protected forest areas
Pearson graphs panel. The Tukey option was used to test for statis- 4.5e+6 240
tically significant differences among the least-square computed 4.0e+6 220
in terms of ratios (% of land area, % of forest area, per capita), any 200
Total forest area
countries with blank values for the denominator values (land area, 3.0e+6
Total protective forest area 180
forest area, or population) result in an undefined ratio. These were 2.5e+6 Countries reporting forest area
omitted from the data analysis. Some countries report 0 values in Countries reporting protective forest area
160
2.0e+6
the various protective forest categories. Since these are real data
entries, these were included in the data analysis. 140
1.5e+6
Income level classification for testing protective forest differ- 120
1.0e+6
ences among income class is from the 2013 World Bank dataset
and is based on per capita annual income (MacDicken, 2015). 5.0e+5 100
Income classes are low ($1035 USD or less), lower middle ($103 1990 1995 2000 2005 2010 2015
6–$4085 USD), upper middle ($4086–$12,615 USD), and high Report year
($12,616 or more).
We acknowledge that there is considerable variation in data 45
quality from all the reporting countries. FAO has addressed this
problem by data quality tiers to all the main category data from
40
each reporting country (MacDicken, 2015). Two types of tier eval-
Forest area, %
Table 1
Summary of number of reporting countries in each protective forest main category and sub-category for each report year. Nbr > 0 = number of reporting countries with forest area
great than 0 in each category or sub-category.
Report year Soil and water Clean water Coastal Desert control Avalanche Flood control Other control
stabilization control
Total Nbr > 0 Total Nbr > 0 Total Nbr > 0 Total Nbr > 0 Total Nbr > 0 Total Nbr > 0 Total Nbr > 0
(a) Protective forests for
Number of reporting countries
1990 114 90 47 15 56 14 60 10 68 3 55 20 47 5
2000 120 98 50 19 59 16 59 10 68 3 56 22 49 8
2005 124 102 50 19 61 17 61 10 69 3 57 23 49 7
2010 134 113 61 30 68 25 62 10 71 4 63 28 55 12
2015 139 117 62 32 70 27 64 12 74 6 67 32 58 15
Report year Ecosystem services Public recreation Carbon storage Cultural services Other services Forest area
Total Nbr > 0 Total Nbr > 0 Total Nbr > 0 Total Nbr > 0 Total Nbr > 0 Total Nbr > 0
(b) Protective forests for
Number of reporting countries
1990 64 44 51 25 42 3 42 11 37 7 234 224
2000 67 47 54 30 44 4 45 13 38 8 234 224
2005 73 54 59 36 45 6 47 15 41 11 234 224
2010 79 61 63 40 47 6 51 19 45 17 234 224
2015 83 65 65 42 48 10 51 19 45 17 234 224
Table 2 200
Global mean ± std err protective forests (as % of total land area and % of forest area) by
category and sub-category for all report years (1990–2015) and proportions of land
and forest area in each category and sub-category as of 2015 (e.g., Total ha for
Number of countries
protection of soil and water resources/total global land area or forest area in 150
ha 100). Total hectares were obtained by summing the corresponding hectares for
all reporting countries.
100
Protective forest Mean ± std Proportion Mean ± std Proportion
category or sub- err (% of of land area err (% of of forest
category total land in 2015 (%) forest area) area in 2015
area) (%) 50
Soil and water 8.67 ± 0.14 7.85 35.92 ± 0.38 25.06
resources
Clean water 1.14 ± 0.13 0.551 3.42 ± 0.26 1.761 0
Coastal 0.23 ± 0.03 0.210 0.83 ± 0.08 0.671 0 10 20 30 40 50 60 70 80 90 100
stabilization Protective forests for soil and water resources,
Desertification 0.19 ± 0.00 0.041 3.61 ± 0.09 0.132 % of forest area
control
Avalanche 0.04 ± 0.00 0.001 0.36 ± 0.01 0.004
control 200
Flood control 1.46 ± 0.07 0.186 5.10 ± 0.33 0.596
Other control 0.98 ± 0.09 1.948 3.13 ± 0.26 6.221
Number of countries
0
(Fig. 2). The number of countries reporting a median value around 0 10 20 30 40 50 60 70 80 90 100
50% was very small. The U-shaped double peak distribution pattern Protective forests for ecosystem services,
did not change through 25 years. % of forest area
Neither main category of protective forests (for protection of
soil and water resources (Fig. 3) or for delivery of ecosystem ser- Fig. 2. Histogram distribution of numbers of countries reporting protective forests
for soil and water resources as % of total forest area (top) and protective forests for
vices (Fig. 4)) showed significant trends over the reporting time
delivery of ecosystem services as % of total forest area (bottom). Number of
interval (1990–2015) when expressed as a % of total forest area. countries were summed for all five report years.
Sub-category trends in Figs. 3 and 4 are described below under
the various sub-category headings.
No significant differences among global sub-regions for soil and forest area as a % of forest area. Polar countries (only two territo-
water protective forest area were observed when expressed either ries, Greenland (Denmark) and Svalbard and Jan Mayen Islands
as a % of land area or as a % of forest area because of the large vari- (Norway), are classified as polar) did not report any forest area.
ation across countries within a region. Similarly, there were no sig- Per capita protective forest area was also calculated and tested
nificant differences among the climatic domains for protective for significant differences among global sub-regions. None were
S. Miura et al. / Forest Ecology and Management 352 (2015) 35–46 39
Global protective forest area sub-category) among soil and water protection sub-categories
100 (Table 1). These forests in this category are found in all the sub-
Protective forest area, % of forest land
Ecosystem services
Public recreation
and the percentage of Caribbean sub-region was significant higher
Carbon storage than that of South America in 1990 and 2000 (P < 0.001) but was
Cultural services not significant in the most recent ten years due to the increase of
Other services
10 percentage forest area in South American countries.
Few countries in the regions suffering disasters by cyclones,
hurricanes, and typhoons, such as Cuba and Jamaica in Caribbean
or Bangladesh in South Asia have primarily designated forests for
coastal stabilization. Globally, Cuba with its long coastal length
1
has the highest percentage of forest area devoted to coastal stabi-
lization (18–19%) (Table 3). Lithuania ranks second at about 8%.
Table 3
0.1 Countries with the highest percentages of forest area in each of the protective forest
1990 1995 2000 2005 2010 2015
and ecosystem services main and sub-categories.
Report year
Protective forest Countries with highest % of forest area (report
Fig. 4. Mean ± std err global protective forest area (as % of forest area) for delivery category or sub-category % of forest area in year)
of ecosystem services (main category) and for sub-categories of public recreation, protective forest
carbon storage, cultural services, and other services by report year. category
Soil and water resources Austria, Burundi, All at 100% (most to all
Burkina Faso, Bhutan, reporting years)
found. Similarly, no significant differences among income class for
Isle of Man, Jamaica,
soil and water protective forest area were found either as a % of Kenya, Libya, Morocco,
land area, % of forest area, or on a per capita basis. Thus, neither Martinique, Mayotte,
sub-regions, climatic domain, nor income class can account for Saint Pierre &
the variation in protective forest area among countries at the main Miquelon, Thailand,
Tajikistan, Tunisia,
category level.
Yemen
Relationships between soil and water protective forest area (as Clean water Japan 24–37% (1990–2015)
% of forest area or on a per capita basis) and a select group of other Coastal stabilization Cuba 18–19% (2000–2015)
FRA database variables were explored. No relationships were found Desertification control Uzbekistan 80% (1990–2015)
Avalanche control Tajikstan 14% (1990–2015)
between protective forests (% of forest area) and production forest
Flood control Timor-Leste 32% (1990–2015)
area, cumulative forest area damaged by outbreaks (e.g., diseases, Other control United States 61–68% (1990–2015)
insects, severe weather), or forest areas with management plans.
Ecosystem services Belgium, Bhutan, Isle of All at 100% (most to all
On a per capita basis, no relationships were found between protec- Man, Iceland, Jamaica, reporting years)
tive forests and non-wood forest products, forest revenue, or public Kenya, Martinique,
expenditures on forests. Mayotte, United States
Public recreation Saint Lucia 29–31% (1990–2015)
Carbon storage Saint Pierre & 100% (2015)
3.1.1. Clean water Miquelon
Approximately 3.4% of global forest area has been primarily Cultural services Australia 35% (2015)
designated for the protection of clean water (Table 2). The number Other services Australia 59% (2010), 39% (2015)
However, many island countries have not always primarily desig- allowed countries to report a forest area of ‘other control’ by sub-
nated forests for that function. The other 26 countries designated tracting the sum of forest area of sub-categories from the forest
less than 10% of total forest area to protect against coastal erosion. area of the soil and water protection main category. More than half
Among them, Russia designated 24.3 mil. ha (3.0% of total forest of all countries reported an identical number of the ratio of the
area) forests for coastal stabilization, which occupied 90.5% of glo- sum of forest area of sub-categories to the forest area of soil and
bal total forest area in this sub-category as of 2015. However, water protection main category. Thus, many countries reported
Russia has not shown an increasing trend in coastal forest protec- the ‘other control’ sub-category by means of data subtraction.
tion in terms of percentage of total forest area. Other relatively
small countries in the Caribbean, South and Southeast Asia regions 3.2. Protective forests for ecosystem services
and other European countries have contributed increased percent-
ages of total forest area to coastal stabilization. Globally, as of 2015, about 1.018 billion ha of forest (25.4% of
global total forest area) has been designated for ecosystem ser-
3.1.3. Desertification control vices, which is almost the same percentage as global protective for-
Relatively few countries have protective forests designated for est area for soil and water resources (Table 2). Forest area devoted
desertification control. All are in arid regions mainly in parts of to ecosystem services has continuously increased over the past
Africa and in central Asia. About 3.6% of global forest area in the 25 years mainly because of the increasing number of reporting
form of protective forests has been used to check the advance of countries (Table 1) and because of a few countries with distinct
desertification with no significant trend since 1990 (Table 2 and net increase of forest area for ecosystem services (e.g. China,
Fig. 3). Uzbekistan, an arid region country in central Asia, has the Russia, and Ecuador). Of this total forest area managed for ecosys-
highest areal extent of forest land designated to arrest the process tem services, about 48.5% as of 2015 (0.493 billion ha) has been
of desertification (80%) (Table 3). Some other arid region countries primarily designated for sub-category of public recreation, carbon
designate at least 10% or more of their forests to control desertifi- storage, spiritual/cultural services, and the other services. The
cation (e.g., Mauritius, Mauritania, Oman, Sudan, and Tajikstan). average percentage of forest area managed for ecosystem services
Neither regions nor domains were significantly different regardless has not changed from around 29.8 ± 0.6% during this period
of expressing protective forests for desertification control as % of (Table 2). However, the proportion of accumulated forest areas
land area, % of forest area, or on a per capita basis. primarily designated for delivery of ecosystem services sub-
categories was 4.5% in 1990, and then distinctly increased up to
3.1.4. Avalanche control 12.3% as of 2015. This increase is about two times larger than the
About 0.36% of global forest area is primarily designated for increase in the proportion of accumulated forest areas primarily
avalanche control (Table 2). Tajikstan, a mountainous country also designated for protection of soil and water.
in central Asia, has designated 14% of its forests for avalanche con- Canada and United States of America in North America are the
trol to protect villages and transportation corridors (Table 3). top two countries that have the largest forest areas designated
Switzerland, another mountainous country, ranks second at about for ecosystem services. Canada designates 93% of its total forest
7%. Other countries that designate forests for avalanche control use area and the United States of America designates 100% of its forests
less than 0.1% of their forests for that purpose. Obviously, the need for ecosystem services as of 2015. These two countries also desig-
for this category of protected forests depends largely on the clima- nate high percentages for protective forests of soil and water
tology, that is, presence of large and deep layers of snow cover and (Canada, 91%; USA, 61–68%). Another eight countries also desig-
the topography of the countries. The designation of a forest under nate more than 90% for both two main categories. As a natural con-
this category depends also on the potential avalanche risk to pop- sequence, almost all forests for both categories overlap. A
ulations and human assets. This is thus a very specific category of definition modification of removing the definitive adverb ‘primar-
protected forests that is present in very few regions and countries ily’ shifts these countries designation percentages from one end of
in the world and has changed little during these 25 years (Fig. 2). the percentage scale to the other. Thus, the primary designation
issue remains a difficult problem for tracking changes through
3.1.5. Flood control time.
The global mean of percentage of forest area managed for ero- Percentages of total forest area designated for ecosystem ser-
sion and flood protection has been the largest of ca. 5.1% among vices have not shown significant differences in terms of global
this primarily designated sub-category of protection of soil and regions, climatic domains, or time. Other variables in the FRA data-
water (Table 2). Global protective forests designated for erosion set such as production forest area, cumulative forest area damaged
and flood protection have increased three times from ca. 7.0 mil. ha by outbreaks of insects or disease, or forest areas with manage-
(0.17% of global total forest area of 20 countries) in 1990 to ment plans, and economic value data on a per capita basis did
23.8 mil. ha (0.6% of 32 countries) as of 2015 (Fig. 2). The flood con- not have any statistical relationship with forest areas designated
trol and clean water protection sub-categories have the largest for ecosystem services also expressed on a per capita basis.
number of reporting countries (32) (Table 1). This rapid increase Changing trends also did not any relationships with any of the
was due to the increase in Russia in 1990s and is also due to the above variables.
increase of reporting countries in the 2000s. Countries in temper-
ate to sub-tropical and tropical domains designated larger forest 3.2.1. Public recreation
area for erosion and flood control, such as Timor-Leste (32%) About 4.3% of global forest area is primarily designated for pub-
(Table 3), Austria (30%) and Switzerland (28%) as of 2015. lic recreation (Table 2). Forests devoted for public recreation are
Tajikistan, a semi-arid country, also designated a high percentage located in all the regions and sub-regions except North America.
of forest area (25%) for erosion and flood control. The number of reporting countries for this sub-category has been
increasing for 25 years, and it was the largest among ecosystem
3.1.6. Other control services sub-categories (42 countries reporting more than 0 ha
About 3.1% of global forest area has been primarily designated out of 65 total reporting countries as of 2015) (Table 1). Most of
for other controls within the protective forest soil and water con- protective forests for public recreation are located in South
trol category (Table 2). No overall global trend in protective forests America (43%), Oceania (33%) and Europe (20%) as of 2015. South
for other control has been observed (Fig. 2). We note that FAO America, Oceania and Central America were sub-regions which
S. Miura et al. / Forest Ecology and Management 352 (2015) 35–46 41
rapidly expanding its recreational forests. Among all the countries, of soil and water and ecosystem services and cultural and spiritual
Saint Lucia has designated the highest percentage of its forest area values for all 5 reporting years. Thus, areas of protective forests
for public recreation (29–31%) (Table 3). Australia is next highest at change are linked with total forest area. Kenya had decreased for-
21–25% (2010–2015). Forest areas in North America are unques- est area in 1990s due to degazettement of forestland to open up
tionably used for public recreation. However, this reported data areas for agricultural land (FAO, 2014e). Since 2000, there is an
situation is because of the ‘primarily designated’ constraint for increase in forest area both in gazetted and non-gazetted areas
sub-categories. due to rehabilitation of degraded forest, especially through carbon
credit schemes (Ministry of Environment, Water And Natural
3.2.2. Carbon storage Resources, 2014), not reaching, however, the values of 1990. No
About 1.3% of global forest area was designated for carbon stor- sub-division was made of the different protection objectives.
age in 1990 and this has increased to 5.3% as of 2015 (Fig. 4). Saint
Pierre and Miquelon, a self-governing French-aligned territory off 3.3.2. China (Asia)
the coast of Newfoundland, Canada, designated 100% of its forests China has dramatically expanded its protective forests over the
for carbon storage – the only country or territory to so designate past 25 years from 18 mil. ha to 58 mil. ha as of 2015 (Fig. 5). The
thus far (Table 3). Iceland is next highest at 28–37% for 2000– proportion of protective forests to total forest area increased from
2015. The increasing trend is a very positive development for glo- 12% in 1990 to 28% as of 2015, which is also associated with the
bal forests to have an increased role in sequestering more carbon in increase in total forest area (Fig. 5). This incredible increase con-
the future (Fig. 4). However, very few (ten) countries, which was tributed a percentage increase in the proportion of protective for-
the smallest number among four sub-categories of ecosystem ser- ests for soil and water resources in East Asia compared to a global
vices, reported forest area of more than 0 ha for carbon storage as no percentage change.
of 2015 (Table 1). The total global forest area in this sub-category According to the national report on sustainable forest manage-
was 12.5 mil. ha (0.3% of global total forest area). Carbon storage ment in China (State Forestry Administration, 2013), areas suffer-
has been by far the smallest primarily designated sub-category of ing from soil erosion at the end of 1990s extended to 356 mil. ha
ecosystem services (Table 2). Because of the limited dataset, it is and the amount of annual soil losses amounted to 5 billion tons.
difficult to produce a robust statistical analysis. Water erosion covered 165 mil. ha, and wind erosion damaged
191 mil. ha. The degradation of forests in China had continued for
3.2.3. Cultural services several decades until 1990s, owing to rapid population growth
Forests are an important cultural service resource for many coupled with the over-exploitation of forest resources, and subse-
countries. About 1.9% of global forest area is primarily designated quent cultivation on steep slopes (Wenhua, 2004). This caused
for cultural services (Table 2). The forests in this category are dis- serious frequent natural hazards and disasters and produced vast
tributed in all regions of the globe, except for the African continent. human and economic loss. The Chinese government launched a
Most protected forest for cultural services are located in South
America (48%) and North America (32%) as of 2015. The most
noticeable increase in protected forest areas for cultural services Protective forest functions and services in China
occurred in South America, from approximately 11 mil. ha (2.1% 30 30
of total forest area) in 1990 to 108 mil. ha (16.0%) as of 2015. 25 25
This sharp increase was driven by a significant enlargement of cul-
Protective forests,
% of land area
tural services areas in Brazil (2–21%), which correspond to approx- 20 20
imately 97% of the protected areas in the region. Likewise, most of
15 15
the protected areas in the region of North America are in the USA
(21.0% of total forest area) and Canada (2.0%) where figures have 10 10
been fairly stable since the 1990s. Significant protected areas for
cultural services are also present in Oceania (35% of Australia), 5 Soil and water resources 5
Total forest area
which contributes 19.3% as of 2015 to the global figure for this cat- 0 0
egory. Australia began using the primary designation of forests for
cultural services in 2000 as the first country in the region to do so. 2.5
Europe (0.6% of global total forest area) and Asia (1.5%) has con-
Protective forests,
2.0
% of forest area
series of top-to-bottom afforestation projects and implemented comprises forests for clean water protection, coastal stabilization,
natural forest conservation projects, and increased shelter forests and flood control (Fig. 6). While small change occurred in area des-
associated with total forest area (State Forestry Administration, ignated for coastal stabilization, protective forests for clean water
2013). This caused a major annual increase in forested areas from increased from 4.1% (33 mil. ha) to 6.3% (52 mil. ha) in the period.
0.27% (1970s–90s) to 3.27% (6.2 mil. ha/y, 1999–2008) of land area. The most notable increase in the allocation of protective forests is
Areas of shelter forests have reached 83.1 mil. ha in late 2000s, evidenced for erosion and flood protection; the forest area in this
occupying 45.8% of forest area and 8.7% of total national land. category grew from 0.2% (1.6 mil. ha) in 1990 to 1.3% (10.4 mil. ha)
Greater parts of shelter forests are for water supply conservation as of 2015 (Fig. 6). Forest designated for the protection of ecosys-
forest (30.6 mil. ha) and water and soil conservation forest tem services, cultural and spiritual values increased from 9.1%
(43.7 mil. ha), whereas 3.0 mil. ha are for windbreak and sand- (74 mil. ha) in 1990 to 13.4% (109 mil. ha) as of 2015 (Fig. 6). It is
fixing forest. Although the area of shelter forests for arid and interesting to note that the forest designated for public recreation
semi-arid region is not large, it is apparent that many forests are has significantly decreased in the period 1990–2000. In any case, a
explicitly managed to control desertification in China, but were slight change of percentage for a sub-category in Russia strongly
not reported in FRA 2015. China stated that it did not report affects any increase or decrease in trend for Europe and even influ-
sub-categories of protection of soil and water because of difficul- ences the entire global forest assessment as we can see for clean
ties of translation between the designation system of FAO and that water or flood control.
of China own (FAO, 2014d).
3.3.4. Canada (Americas)
Canada has reported all publicly owned forests as designated
3.3.3. Russian Federation (Europe)
for protection of soil and water (FAO, 2014c). At 347 mil. ha
Russian territory spreads over the most northern part of the
(2015 report year), Canada ranks 3rd globally (behind Russia and
north hemisphere. Thus most Russian forests are located in the
Brazil) in terms of total forested land area (38% of total land area
boreal domain (88%) (FAO, 2014f). About 50% of land area is cov-
in Canada is forested). Despite ranking 3rd in total forest area,
ered by forest (815 mil. ha, the world’s largest) as of 2015. Russia
Canada leads all nations in terms of total forest area designated
designated forests by its own classification system, as four main
for protection of soil and water resources (317 mil. ha as of
categories: protected forests (2.2%), protective forests (24%), oper-
2015). Canada reported protective forest area in both main cate-
ational forests (50.9%) and reserve forests (22.9%) in 2013 (FAO,
gories – soil and water protection (91.4% of forested land area in
2014f). Protected forests (18 mil. ha) and protective forests
all report years) and ecosystem services (93.4% of forest area in
(215 mil. ha) correspond to FRA protective forests and conserva-
all report years). Canada only reported protective forest areas in
tion forests for biodiversity (main category 6). Russia reclassified
the cultural services sub-category (about 2% of forest areas).
and divided these two national categories of forests to adapt FRA
definitions of protective forests for soil and water (FAO, 2014f).
3.3.5. Australia (Oceania)
According to the new FRA 2015 dataset, forests designated for
Australia (2015 report year) has a total forested land area of
the protection of soil and water have increased from 7.3%
124 mil. ha (16% of total land area) and ranks 7th behind China
(59 mil. ha) in 1990 to 10.6% (86 mil. ha) as of 2015 (Fig. 6).
and Congo. It has reported protective forest area in the two main
However, Russia did not show the details of the calculation in
categories – soil and water protection (23.9% of total forest area
the country report (FAO, 2014f). The area under this category
as of 2015) and ecosystem services (98.4%) (Fig. 7). Australia is
one of the largest dry continent in the world. Most of Australia’s
Protective forest functions and services in Russia land area is classified as rangelands, which encompass some 75%
12 (570 mil. ha) of the continent. One third of the continent has extre-
mely low and variable stream-flow compared to other continents
10
and large areas of the arid hinterland produce almost no run-off.
Soil and water resources
Protective forests,
% of forest area
12 Carbon storage
Cultural services tection or ecosystem services have been included in any reporting
10
Other services year. Though Australia operates indeed dune protection program
8
for coast care, integrated package for desertification control and
6
managements for erosion and flood controls, it has not provided
4
nationally inconsistent information for these functions (FAO,
2 2014b).
0 Ownership of forests affects designation of protective forests in
1990 1995 2000 2005 2010 2015 Australia. Importantly, forest areas in public nature conservation
Report year reserves are ‘managed primarily for protective functions’, which
includes the conservation of biodiversity as well as protection of soil
Fig. 6. Mean ± std err protective forest area (as % of forest area) for soil and water and water values. It is worth noting that, across most of Australia’s
resources (main category); for sub-categories of clean water, coastal stabilization,
desertification control, avalanche control, flood control, and other control (top); for
forest management jurisdictions, the preservation of soil and water
ecosystem services (main category); and subcategories of public recreation, C is usually one of several forest management objectives included in
storage, cultural services, and other services (bottom) for Russia by report year. the management of multiple-use public forests.
S. Miura et al. / Forest Ecology and Management 352 (2015) 35–46 43
Protective forest functions and services distribution although we could not clarify the exact reason. A gov-
in Australia ernment could hesitate to designate or increase protective forests
30 in such a situation. For instance, stakeholders of forests in the coun-
try which has a regulation against production activities may hesitate
25 Soil and water resources to designate protective forests. In the opposite case, they have no
Protective forests,
Clean water reason to hesitate to designate 100% of their forests as protective for-
% of forest area
20 Other control ests. Historical development of forest ownership may also influence
15 this situation such as in Canada or Australia. For example, in
Australia, significant proportions of multiple-use forests are informal
10 reserves where wood harvesting is not permitted. In addition, differ-
ences among state-based jurisdiction rules in Australia could be
5 obstacles to adjust a unified reclassification of multiple-use forests
within one country. Such possible political considerations influenc-
0 ing the designation of protective forests are apparent from country
reports for FRA 2015 and other country reports from the Montreal
100
Process for four countries, China, Russia, Canada and Australia.
Those countries reports are indispensable sources of understanding
Protective forests,
80
% of forest area
Percentage forest area in FRA 2005 was divided into six exclusive of FE, introduced general principles to designate protective func-
categories of functions; (1) Production, (2) Protection of soil and tions with legal basis, long-term commitment and explicit designa-
water, (3) Conservation of biodiversity, (4) Social services, (5) tion of biodiversity, landscape, specific natural element or
Multiple purpose, (6) None or unknown function. Category 1 protective function of forest and other wooded land (MCPFE,
through 4 was applicable if a forest is primarily designated as 2003). This exclusive designation rule for protective functions is
one of four functions. Otherwise, category 5 or 6 was assigned. similar to the definition of FRA 2005 (FAO, 2004a). This kind of des-
All the forest area must be within one of six categories and the ignation scheme was intended to be shared internationally in early
sum of percentages of all six categories must add up to 100%. 2000’s. However, FE also could not completely overcome the prob-
Using this definition in FRA 2010 (FAO, 2010a), the United States lem of overlapping designation for protective forests. FE had to
of America did not report in the category of ‘primarily function acknowledge that there exists overlapping designations of indica-
as’ and no primarily designated protective forests were reported tors 5.1 Protective forests of FE – soil, water and other ecosystem
for the North American sub-region. Reasons how and why this functions and 5.2 Protective Forest – infrastructure and managed
was done were described in the main report of FRA 2010 (FAO, natural resources in 2011 (MCPFE, 2011). Thus, we can say that
2010b). there still exists potential to further develop the framework of
This potentially-misleading definition was modified in FRA definitions for reporting multiple forest functions to establish
2015. The word ‘primarily’ designated was not used for the two long-term robust variables as a whole forestry sector. Tracing
main categories of protection of soil and water and selected and examining details of recalculation formulas and comments
ecosystem services and was applied only to sub-categories. In con- for the three Tables 4–6 of FRA 2015 would be a good start. We
sequence, the global level of protective forests as a proportion of could learn from the diversity of the methods and original defini-
total forest area abruptly jumped to 25.1% in FRA 2015 from 8% tions for designation of protective forests in each country as we
in FRA 2010. Sum of forest area of primarily designated sub-cate- illustrated with the five case study countries. On the other hand,
gories of protection forest in 2010 was 369 mil. ha (9.2% of total movement of developing tools for sustainable forest management
forest area) in FRA 2015 report. This shows a fairly good correspon- is a sweeping trend of forestry sector. Regional initiatives such as
dence to 8% in FRA 2010 report considering that each country was ITTO, FE, MP have lead the establishment of criteria and indicators
allowed and encouraged to make a retroactive modified report in (McDonald and Lane, 2004). Strengthening collaborations with
previous reporting years of 1990 to 2010. those activities would also be another effective solution for this
The problem of ‘primarily’ designated for protective forests for issue. The beginning of such international collaboration has been
soil and water has been resolved as described above for the two seen in the partnership activities of the Collaborative Forest
main categories. However, the overall strategy of how FRA reflects Resources Questionnaire (MacDicken, 2015) for FRA 2015. The
multiple functions of forests still lacks consistency and harmoniza- overall strategy of FRA for reporting multiple forest functions
tion across reporting periods. Ecosystem services of forests are should be reconciled and improved.
divided into conservation of biodiversity (FRA 2015, Table 6) and
other services (FRA 2015, Table 5b) (FAO, 2014a). Carbon storage 4.3. Data quality
or sequestration is assigned a sub-category of selective ecosystem
services, though biomass carbon and below ground carbon are These are largely qualitative designations that while providing a
reported within the main category of production (FRA 2015, guide to overall data quality, do not facilitate a more quantitative
Table 3e). In the meantime, there is an opinion such as that of assessment of data reporting errors. For protective forests for soil
China (FAO, 2014d) and Russia (FAO, 2014f) that all forests make and water resources, tier 1 countries (least reliable data quality)
a significant contribution to carbon storage or sequestration outnumber tier 3 countries (most reliable data quality) by a large
because every forest provides a function of carbon sequestration margin (Table 4). For example, there are twice as many tier 1 coun-
to some extent. Variables reporting multiple functions in FRA tries in the protective forests for soil and water category as there
2005 and 2010 showed a clear stance of FAO to report the multiple are tier 3 countries. Only about 59% of all countries have tier des-
functionality of forests within a table even if it introduced a new ignations for the soil and water resources protective forests cate-
problem as described above. However, in FRA 2015, FAO separated gory. Only 35% of all countries have assigned data quality tiers in
a part of forest functions into three different tables (4. production, the ecosystem services delivery category. For future editions of
5. protection, 6. biological conservation). Thus, the FAO approach FRA, we recommend a renewed emphasis on encouraging coun-
to understand multiple forest functions was still developing. tries in the lowest tier 1 to take steps to move to tier 2 and tier 2
Forest Europe (FE), the pan-European political process for the countries to move to tier 3. In the current FRA reporting cycle, only
sustainable management of the continent’s forests, has also been about 18% of reporting countries are in tier 3 for the protective for-
facing to this kind of dilemma of designation between a primary est category and only about 13% are in tier 3 for the ecosystem ser-
function and multi-functionality of forest. MCPFE, the predecessor vices category.
Table 4
Numbers of countries (% of total countries) in each data reliability tier for the main protective forest categories of soil and water resources and ecosystem services. Tier 1 = least
reliable data quality, tier 2 = moderately reliable data quality, tier 3 = most reliable data quality.
All but 10 countries and territories report forest area greater Appendix A. Supplementary material
than 0 ha. Half the world’s countries now report some forest area
(>0 ha) for protection of soil and water resources (117 in 2015) Supplementary data associated with this article can be found, in
(Table 1). Only about a quarter of all countries (28%) have actual the online version, at http://dx.doi.org/10.1016/j.foreco.2015.03.
forest area (>0 ha) reported for ecosystem service delivery. In all 039.
the protective forest and ecosystem service sub-categories, the
number of reporting countries is a minor fraction of all possible
reporting countries. It ranges from as few as 6 countries with pro-
tective forest areas designated for avalanche control in 2015 to 42 References
countries reporting forest area for public recreation in 2015.
Aust, W.M., Blinn, C.R., 2004. Forestry best management practices for timber
Because of the low numbers of reporting countries with hectares harvesting and site preparation in the eastern United States: an overview of
>0 in each of the sub-categories, both the total forest area and per- water quality and productivity research during the past 20 years (1982–2002).
Water Air Soil Pollut. Focus 4, 5–36. http://dx.doi.org/10.1023/
centages of forest area designated as protective forests remains
B:WAFO.0000012828.33069.f6.
small on a total global basis. There is opportunity for expansion Binkley, D., Fisher, R., 2013. Ecology and Management of Forest Soils, fourth ed.
in many of these protective forest sub-categories. Wiley-Blackwell, Hoboken, 362 pp.
Bosch, J.M., Hewlett, J.D., 1982. A review of catchment experiments to determine
the effect of vegetation changes on water yield and evapotranspiration. J.
Hydrol. 55 (1), 3–23.
5. Conclusions
Broadhead, J., Leslie, R., 2007. Coastal protection in the aftermath of the Indian
Ocean tsunami: what role for forests and trees? In: Proceedings of an FAO
FRA was originally founded to build a global inventory of forest Regional Technical Workshop, Khao Lak, Thailand, 28–31 August 2006, FAO,
resources as the name implies. The global community has come to Thailand.
Collaborative Partnership on Forests. SFM and the Multiple Functions of Forests.
expect FRA to report forest functions and services other than forest <http://www.cpfweb.org/32819-045ba23e53cbb67809cef3b724bef9cd0.pdf>
productivities towards the end of the 20th century, where threats of (Retrieved 12.11.14).
global warming and degradation of biological diversity to human Costanza, R., d’Arge, R., de Groot, R., Farberparallel, S., Grasso, M., Hannon, B.,
Limburg£star, K., Naeem, S., O’Neill, R.V., Paruelo, J., Raskin, R.G., Sutton, P., van
societies became obvious (McCarthy, 2001; WHO, 2005; Kirilenko den Belt, M., 1997. The value of the world’s ecosystem services and natural
and Sedjo, 2007). We cannot assess FRA data as a controlled water- capital. Nature 387, 253–260.
shed experiment like Bosch and Hewlett (1982). However, FRA data Cubbage, F., Harou, P., Sills, E., 2007. Policy instruments to enhance multi-functional
forest management. Forest Policy Econ. 9, 833–851. http://dx.doi.org/10.1016/
include above and below ground biomass and carbon. Forest soil j.forpol.2006.03.010.
carbon storage, for instance, could be a good indicator of a develop- Diaz-Balteiro, L., Romero, C., 2008. Making forestry decisions with multiple criteria:
ment and maturing of a forest ecosystem (Post and Kwon, 2000). a review and an assessment. For. Ecol. Manage. 255, 3222–3241. http://
dx.doi.org/10.1016/j.foreco.2008.01.038.
FRA 2015 started to provide information about data quality using Dolidon, N., Hofer, T., Jansky, L., Sidle, R., 2009. Watershed and forest management
the tier concept of assessment. We now have foundational informa- for landlside risk mitigation. In: Sassa, K., Canuti, P. (Eds.), Landslides – Disaster
tion from FRA to assess forest functions, which can lead to imple- Risk Reduction. Springer, pp. 633–649.
FAO, 1995. Forest Resources Assessment 1990 – Global Synthesis. FAO Forestry
mentation of sound forest management. What has happened in
Paper 124. Rome, Italy.
the forests of China or Russia where protective forests have FAO, 1997. State of the World’s Forests, 1997. Rome, Italy.
increased rapidly? Are the changing directions and strengths in FAO, 2000. On Definitions of Forest and Forest Change. FRA Working Paper 33.
those countries different from changes in Canada where a stable Rome, Italy. 14 p.
FAO, 2001. Global Forest Resources Assessment 2000 – Main Report. FAO Forestry
forest management has persisted? Although beyond the scope of Paper 140. Rome, Italy. <http://www.fao.org/forestry/site/7949/en/>.
this paper, this type of analysis incorporating socio-economic vari- FAO, 2004a. Global Forest Resources Assessment Update 2005. Terms and
ables in FRA may spur further investigation. Definitions. FRA Working Paper 83. Rome, Italy. 34 p.
FAO, 2004b. Payment for Environmental Services in Watersheds. Land and Water
FRA is likely the only and largest database with which we can Discussion Paper 3. Rome, Italy. 74 p.
conduct analysis of the relationships between forest managements FAO, 2006a. Global Forest Resources Assessment 2005 – Progress Towards
and forest functions. We can say that FRA 2015 has succeeded in Sustainable Forest Management. FAO Forestry Paper 147. Rome, Italy. <http://
www.fao.org/forestry/fra2005/en/>.
building a long-term forest database having both socio-economic FAO, 2006b. The New Generation of Watershed Management Programmes and
and environmental aspects. This could be a good start point for a Projects. Forestry Paper #150. Rome. 128 p.
global database for evaluation of the humanities and sciences of FAO, 2007. Why Invest in Watershed Management? Rome. 31 p.
FAO, 2008. Practical Experiences of Compensation Mechanisms for Water Services
the environment. Maintaining and improving this database would Provided by Forests in Central America and the Caribbean. National Forest
contribute immeasurably. Linkage between forestry and social Programme FACILITY. Rome, Italy. 51 p.
sciences such as environmental economics or behavioral eco- FAO, 2010a. Global Forest Resources Assessment 2010. Terms and Definitions. FRA
Working Paper 144. Rome, Italy. 27 p.
nomics would be primarily of importance to utilize FRA data for
FAO, 2010b. Global Forest Resources Assessment 2010 – Main Report. FAO Forestry
functional assessment of forest. Devising a sophisticated handling Paper 163. Rome, Italy. <http://www.fao.org/forestry/fra/fra2010/en/>.
scheme for multiple functions of forest and introducing measur- FAO, 2012. Terms and Definitions. Forest Resources Assessment Working Paper 180.
able variables to assess exact functioning of soil and water protec- Rome, Italy. 31 p.
FAO, 2013. Forests and Water – International Momentum and Action. Rome, Italy.
tion or ecosystem services would provide a breakthrough for 76 p.
sustainable forest management. Improving reliability of reporting FAO, 2014a. Guide for Country Reporting for FRA 2015. On-line FRA 2015 Working
through capacity building should be the minimum requirement Paper. <http://www.fao.org/forestry/fra>. Rome, Italy. 52 p. 5 appendices.
FAO, 2014b. Country Report: Australia. Global Forest Resources Assessment 2015.
for future FRA reports. Fostering the database with a long-term Rome, Italy. 130 p.
strategy is one of the most important missions of the forestry FAO, 2014c. Country Report: Canada. Global Forest Resources Assessment 2015.
sector. Rome, Italy. 94 p.
FAO, 2014d. Country Report: China. Global Forest Resources Assessment 2015.
Rome, Italy. 111 p.
FAO, 2014e. Country Report: Kenya. Global Forest Resources Assessment 2015.
Acknowledgement
Rome, Italy. 78 p.
FAO, 2014f. Country Report: Russian Federation. Global Forest Resources
We thank to Kenneth MacDicken and FRA team of Food and Assessment 2015. Rome, Italy. 87 p.
Agriculture Organization of the United Nations (FAO) to provide FAO and CIFOR, 2005. – Drowning in Fiction or Thriving on Facts? RAP Publication
2005/03. Bangkok. 30 p.
us an opportunity to analyze FRA 2015 data. This paper is a volun- Fu, B., 1989. Soil erosion and its control in the loess plateau of China. Soil Use
tary work of all authors. Manage. 5, 76–82. http://dx.doi.org/10.1111/j.1475-2743.1989.tb00765.x.
46 S. Miura et al. / Forest Ecology and Management 352 (2015) 35–46
Government of Japan, 2002. International Expert Meeting on Forests and Water, Post, W.M., Kwon, K.C., 2000. Soil carbon sequestration and land-use change:
Shiga. Proceedings of the meeting. Tokyo. 361 p. processes and potential. Glob. Change Biol. 6, 317–327. http://dx.doi.org/
Greenwood, J.R., Norris, J.E., Wint, J., Greenwood, J.R., Norris, J.E., Wint, J., 2004. 10.1046/j.1365-2486.2000.00308.x.
Assessing the contribution of vegetation to slope stability. Proc. ICE – Geotech. Postel, S.L., Thompson, B.H., 2005. Watershed protection: capturing the benefits of
Eng. 157, 199–207. http://dx.doi.org/10.1680/geng.2004.157.4.199. nature’s water supply services. Natl. Resour. Forum 29, 98–108. http://
Hamilton, L.S. 1986. Towards clarifying the appropriate mandate in forestry for dx.doi.org/10.1111/j.1477-8947.2005.00119.x.
watershed rehabilitation and management. East-West Environment and Policy Raison, R.J., Flinn, D.W., Brown, A.G., 2001. Application of criteria and indicators to
Institute, Reprint No 94, East-West Center, Honolulu. 51 p. support sustainable forest management: some key issues. In: ‘Criteria and
Hamilton, L.S. 2008. Forests and water – a thematic study prepared in the Indicators for Sustainable Forest Management’. IUFRO Research Series No. 7,
framework of the Global Forest Resources Assessment 2005. Forestry Paper CABI, Wallingford, pp. 5–18. doi: http://dx.doi.org/10.1079/9780851993928.
#155. Rome, Italy. 78 p. 0005.
Hofer, T., Messerli, B., 2006. Floods in Bangladesh – History, Dynamics and Rametsteiner, E., Simula, M., 2003. Forest certification—an instrument to promote
Rethinking the Role of the Himalayas. United Nations University Press, Tokyo, sustainable forest management? J. Environ. Manage. 67, 87–98. http://
468 p. dx.doi.org/10.1016/S0301-4797(02)00191-3.
Kammerbauer, J., Ardon, C., 1999. Land use dynamics and landscape change pattern REDD Research and Development Center, 2012. REDD-plus cookbook – How to
in a typical watershed in the hillside region of central Honduras. Agric. Ecosyst. measure and monitor forest carbon. In: Hirata, Y., Takao, G., Sato, T., Toriyama, J.
Environ. 75, 93–100. http://dx.doi.org/10.1016/S0167-8809(99)00071-7. (Eds.), Forestry and Forest Products Research Institute, Japan, 151pp.
Keating, M., 1993. The Earth Summit’s Agenda for Change – A Plain Language Reubens, B., Poesen, J., Danjon, F., Geudens, G., Muys, B., 2007. The role of fine and
Version of Agenda 21 and the Other Rio Agreements. Published by the Centre coarse roots in shallow slope stability and soil erosion control with a focus on
for Our Common Future, Geneva, 35 p. root system architecture: a review. Trees 21, 385–402. http://dx.doi.org/
Keenan, R.J., Reams, G.A., Achard, F., de Freitas, J., Grainger, A., Lindquist, E., 2015. 10.1007/s00468-007-0132-4.
Dynamics of Global Forest Area: Results from the FAO Global Forest Resources SOFR (Montreal Process Implementation Group for Australia and National Forest
Assessment 2015. Forest Ecol. Manage. 352, 9–20. Inventory Steering Committee), 2013. Australia’s State of the Forests Report.
Kirilenko, A.P., Sedjo, R.A., 2007. Climate change impacts on forestry. Proc. Natl. Acad. Canberra: ABARES; 2013. <http://www.daff.gov.au/ABARES/forestsaustralia/
Sci. U.S.A. 104, 19697–19702. http://dx.doi.org/10.1073/pnas.0701424104. Documents/sofr2013-web2.pdf> (accessed 23.07.14).
Kumar, P. (Ed.), 2010. The Economics of Ecosystems and Biodiversity: ecological and State Forestry Administration, 2013. National Report on Sustainable Forest
economic foundations. UNEP/Earthprint. Management. China Forestry Publishing House, Beijing, China, 232 p.
MacDicken, K.G., 2015. Global Forest Resources Assessment 2015: What, why and Turner, M.G., 1989. Landscape ecology: the effect of pattern on process. Annu. Rev.
how? Forest Ecol. Manage. 352, 3–8. Ecol. Syst. 20, 171–197. http://dx.doi.org/10.1146/
McCarthy, J.J. (Ed.), 2001. Climate Change 2001: Impacts, Adaptation, and annurev.es.20.110189.001131.
Vulnerability: Contribution of Working Group II to the Third Assessment van Kooten, G.C., Nelson, H.W., Vertinsky, I., 2005. Certification of sustainable forest
Report of the Intergovernmental Panel on Climate Change. Cambridge management practices: a global perspective on why countries certify. Forest
University Press. Policy Econ. 7, 857–867. http://dx.doi.org/10.1016/j.forpol.2004.04.003.
McDonald, G.T., Lane, M.B., 2004. Converging global indicators for sustainable forest Wang, X.M., Zhang, C.X., Hasi, E., Dong, Z.B., 2010. Has the Three Norths Forest
management. Forest Policy Econ. 6, 63–70. http://dx.doi.org/10.1016/S1389- Shelterbelt Program solved the desertification and dust storm problems in arid
9341(02)00101-6. and semiarid China? J. Arid Environ. 74, 13–22. http://dx.doi.org/10.1016/
MCPFE, 2003. State of Europe’s Forests 2003. Ministerial Conference on the j.jaridenv.2009.08.001.
Protection of Forests in Europe, Liaison Unit, Vienna. Wenhua, L., 2004. Degradation and restoration of forest ecosystems in China. For.
MCPFE, 2011. State of Europe’s Forests 2011. Ministerial Conference on the Ecol. Manage. 201, 33–41. http://dx.doi.org/10.1016/j.foreco.2004.06.010.
Protection of Forests in Europe, Forest Europe, Liaison Unit, Oslo. WHO, 2005. Ecosystems and human well-being: Health Synthesis. Island Press,
Ministry of Environment, Water and Natural Resources, 2014. Forest Policy, 2014. Washington, DC.
Ministry of Environment, Water and Natural Resources, Republic of Kenya, 13 p. Yu, K., Li, D., Li, N., 2006. The evolution of greenways in China. Landsc. Urban Plann.
Perlin, J., 1989. A Forest Journey: The Role of Wood in the Development of 76, 223–239. http://dx.doi.org/10.1016/j.landurbplan.2004.09.034.
Civilization. W.W. Norton & Company, New York, p. 445.
Forest Ecology and Management 352 (2015) 47–56
a r t i c l e i n f o a b s t r a c t
Article history: Sustainable forest management (SFM) is many things to many people – yet a common thread is the pro-
Received 10 December 2014 duction of forest goods and services for the present and future generations. The promise of sustainability
Received in revised form 4 February 2015 is rooted in the two premises; first that ecosystems have the potential to renew themselves and second
Accepted 5 February 2015
that economic activities and social perceptions or values that define human interaction with the environ-
Available online 7 September 2015
ment are choices that can be modified to ensure the long term productivity and health of the ecosystem.
SFM addresses a great challenge in matching the increasing demands of a growing human population
Keywords:
while maintaining ecological functions of healthy forest ecosystems. This paper does not seek to define
Sustainable forest management
SFM
SFM, but rather provides analyses of key indicators for the national-scale enabling environment to gain
Forest policy a global insight into progress in implementing enabling and implementing SFM at the national and opera-
Forest management planning tional levels. Analyses of the Global Forest Resources Assessment 2015 (FRA) country report data are used
National forest inventory to provide insights into the current state of progress in implementing the enabling conditions for SFM.
Forest certification Over 2.17 billion ha of the world’s forest area are predicted by governments to remain in permanent for-
est land use, of which some 1.1 billion ha are covered by all of the SFM tools investigated in FRA 2015. At
the global scale, SFM-related policies and regulations are reported to be in place on 97% of global forest
area. While the number of countries with national forest inventories has increased over that past ten
years from 48 to 112, only 37% of forests in low income countries are covered by forest inventories.
Forest management planning and monitoring of plans has increased substantially as has forest manage-
ment certification, which exceeded a total of over 430 million ha in 2014. However, 90% of internationally
verified certification is in the boreal and temperate climatic domains – only 6% of permanent forests in
the tropical domain have been certified as of 2014. Results show that more work is needed to expand
the extent and depth of work on establishing the enabling conditions that support SFM over the long term
and suggests where those needs are greatest.
Ó 2015 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
http://dx.doi.org/10.1016/j.foreco.2015.02.005
0378-1127/Ó 2015 Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
48 K.G. MacDicken et al. / Forest Ecology and Management 352 (2015) 47–56
Saxonian forest regulation was used as an important principle of Operational scale progress toward SFM
forest management in the early 18th century. Hartig (1795)
described sustainable yield based on the quantity of increasing What is the area of forest under a forest management plan and
timber volume which was an important step in the quantitative how is this monitored?
regulation of harvest volumes. In the early 20th century, Gifford How are stakeholders involved in management decisions for
Pinchot recognized that clear and convincing evidence was needed publicly-owned forests?
to demonstrate that sustainable forest management would return What is the area of forest under one or more independently
a profit. He also noted that sustainable forestry was not possible verified forest certification schemes?
without the consent and active participation of the public
(Schmithüsen, 2013). Zon and Sparhawk (1923) demonstrated
2. Methods
how globally available data on forest resources provides profes-
sionals and the public with information vital to effective strategies
The methods, definitions and approaches used in FRA 2015 are
for sustaining forest values. They also note the gaps in this data
discussed and referenced in greater detail in MacDicken (2015)
that constrain management – some of which remain unfilled in
and www.fao.org/forestry/FRA/2015/Methods. FRA 2015 data were
the early 21st century. The Global Forest Resources Assessment
extracted from the Forest Resources Information System (FRIMS) as
(FRA) was created to provide a continuing assessment of forest
described in www.fao.org/forestry/fra2015. All data used in this
resources and how they are changing (MacDicken, 2015).
paper except international forest management certification was
SFM has been encouraged as an important guiding principle in
provided by countries or through desk studies carried out by FAO.
managing forests (ITTO, 2006; EC, 2003). The concept provides
Other than for international certification, country reports from gov-
guidance on how to manage forests to provide for today’s needs
ernment-appointed National Correspondents contributed data
(as best as possible) and not compromise (i.e. reduce) the options
representing some 99% of global forest area. International certifica-
of future generations (Forest Principles, UN Rio, 1992). The tools
tion data was provided by the Forest Stewardship Council (FSC) and
available for encouraging SFM begin with policy and regulations
the Programme for Endorsement of Forest Certification (PEFC) for
that support those who are practicing forest management. They
July in each of the reporting years1. Enabling environment-related
also include inventories, monitoring, forest management
data were collected on policies, legislation and regulations support-
certification, stakeholder involvement and forest management
ing SFM, presence of a national stakeholder platform and the types
plans. Where there is a clear understanding of the ecological
of forest resource monitoring information and progress reporting
circumstances of the forests being managed an appropriate
available. Progress at the operational level was measured as forest
regulatory framework can establish the enabling conditions for
area under Forest Management Plans (FMP), including an assessment
SFM.
of how the content of these plans are monitored and how frequently
Criteria and indicators (C&I) of SFM have been developed
and if stakeholder inputs are required and the extent of both interna-
through the work of many actors – including governments, research
tional and domestic forest management certification. Statistical
organizations, non-governmental organizations and private com-
summaries and analyses were done for all variables using Microsoft
panies (MCPFE, 2001; Prabhu et al., 1998). This includes work by
Excel and Systat (Ver. 13) and relationships described by national
countries involved in the Montréal Process (MP), FOREST EUROPE
income category, climatic domain and sub-region (see MacDicken,
(FE), the International Tropical Timber Organization (ITTO) and
2015). Reported values were clustered into four nested categories:
the Food and Agriculture Organization (FAO). These C&I are used
legal framework, national data reporting, management planning
to define SFM and to measure and report progress towards its
and stakeholder involvement plus certified forest area.
implementation (Canadian Council of Forest Ministers, 2008).
These international and regional initiatives and research efforts
have made good progress in using science, commerce and social 3. Results and discussion
values to devise their indicators. Supportive national legal, policy
and institutional frameworks can make SFM practices cost-effective 3.1. When do the conditions exist to enable sustainable forest
and when effectively applied encourage the practices needed for management?
SFM (Keeton and Crow, 2009; FAO, 2010; Lovrc et al., 2010).
Forest management certification provides independent, third-party It depends on where you set the threshold – if the presence of a
verification of adherence to a defined set of management standards regulatory framework is deemed adequate, then the conditions
that promote and measure SFM (CEPI, 2006). exist when policies and regulations are in place. The most rigorous
The main focus of this paper is to present factors that are set of enabling conditions includes the legal framework, national
needed for and provide support to long-term sustainable forest data reporting, the availability of quality forest inventory data,
management. By presenting a sub-set of data relevant to SFM management planning, effective stakeholder involvement and
derived from the Global Forest Resources Assessment 2015 regular monitoring and reporting. Measuring and reporting these
(www.fao.org/forestry/fra) the reader is provided with information data at the national scale and sharing the results through the
that can help determine where and how much progress is being FRA is an important step in understanding progress to SFM and
made towards establishing and maintaining the enabling where it or is not occurring.
conditions for SFM. FRA 2015 was designed in part to provide this Using the SFM related data collected through FRA 2015, it is
information by asking questions under two broad categories: possible to begin with the area of permanent forest land2 and
Enabling environment (national scale): evaluate how much forest land is covered by successive indicators.
Fig. 1 presents this progression of the application of these ‘‘SFM
What forest policy and regulatory framework exists to support
implementation of sustainable forest management? 1
July was used as a mid-point for the annual data and is important because
Is there a national platform that promotes stakeholder partic- monthly certification values change throughout the year as additional forest area is
certified or previously certified forest are decertified.
ipation in forest policy development? 2
Permanent forest land is defined in FRA 2015 as the area of state-owned forest
What is the forest area intended to be in permanent forest land designated to remain permanently as forest (i.e. the permanent forest estate) plus an
use now and how has it changed over time? estimate by governments of the portion of privately-owned forest expected to remain
How is progress toward SFM measured and reported? permanently in forest land use.
K.G. MacDicken et al. / Forest Ecology and Management 352 (2015) 47–56 49
Fig. 2. The area of permanent forest use as modified by other elements of the SFM
enabling environment by climatic domain: (A) Permanent forest use; (B) Policies;
Fig. 1. The area of permanent forest use as modified by other elements of the SFM (C) Legislation; (D) Stakeholder platform; (E) Forest inventory; (F) National
enabling environment: (A) Permanent forest use; (B) Policies; (C) Legislation; (D) reporting; (G) Forest management plans; (H) Soil and water conservation; (I)
Stakeholder platfor; (E) Forest inventory; (F) National reporting; (G) Forest High conservation value forest; (J) Social engagement; (K–M) Stakeholder involve-
management plans; (H) Soil and water conservation; (I) High conservation value ment in operational planning, operations and review.
forest; (J) Social engagement; (K–M) Stakeholder involvement in operational
planning, operations and review.
tools’’. It begins from the left, showing how forest area changes with 3.2. Legal framework
the level and presence of enabling conditions for SFM. Moving from
left to right the area of permanent forest subject to each of the 3.2.1. Permanent forest land
indicators listed on the horizontal axis is given. This reduction is The area of permanent forest land indicates the area of forest
based on the global extent of supportive legal, data availability, likely to remain as forest, including both the permanent forest
management planning and stakeholder involvement in operations. estate and privately owned forest land expected by governments
In those countries with permanent forest, some 98% of permanent to remain in forest land use (see www.fao.org/forestry/FRA2015/
forest land is covered by policies, laws and regulations in support Terms and Definitions for additional details). The permanent forest
of SFM. This indicates a broad intention on the part of governments estate is state-owned and is mandated to remain as forest. The
to support SFM into the future on some 2.2 billion ha (55% of global term permanent forest land is used to describe the sum of govern-
forest area in 2015). When all of the SFM tools are included, the area ment estimates of non-state owned land intended to remain in
decreases to 1.11 billion ha globally. permanent forest use plus the permanent forest estate. The intent
When examining the application of each of the SFM tools to the to retain these areas as forest is the starting point for forest area
forest by climatic domain (Fig. 2), all domains experience a that is managed sustainably – they are in principle those forest
decrease in area as the number of SFM tools applied increases lands for which a commitment has been made to sustaining forest
(from left to right on the x axis). Overall the temperate domain values well into the future.
shows the least decrease in forest area as the number of SFM tools In 1990 the area reported as permanent forest estate was 1.4
increases, followed by the boreal, tropical and sub-tropical domain billion ha (FRA 2010) or 34% of total forest area while in by 2010
in order. It can be noted that each decrease identifies a need to this went to 1.7 billion or 43% of total forest area (as assessed in
increase investment required to apply a given tool(s). FRA 2015). The total amount of forest land intended to be perma-
The area under FMP in the Boreal domain drops by nearly 200 nent forest was 2.17 billion ha or 54% of total forest area in 2010.
million ha once soil and water conservation objectives are included The difference between these two numbers comes from private
as requirement in forest management plans. In the tropical forest lands intended to remain in forest. Fig. 3 shows the dis-
domain, the permanent forest area covered by inventories is 309 tribution of permanent forest area by income category – of which
million ha less than the area covered by legal frameworks support- some 82% is in upper middle and high income countries. Privately-
ing SFM. The tropical domain area that does not include national owned forest is important in sustaining future forest area in the
reporting, nor management plans nor stakeholder involvement high, upper middle and lower middle income category and of
drops this by an additional 123 million ha. In the subtropics only almost no importance in low income countries where state
14% of permanent forest area is covered by forest reporting, ownership dominates.
management plans and stakeholder involvement in operations.
This difference between domains illustrates the need for greater 3.2.2. Policies and legislation supporting SFM
investment in forest reporting, management planning and The practice of sustainable forest management is enhanced by
stakeholder involvement in operations if the area of SFM is to be the effective application of a national legal, policy and institutional
increased. framework. Countries that had either policies and or legislation
The following discussion covers each of the SFM related that support sustainable forest management cover 99% of global
indicators measured through the FRA with analyses by one or more forest area (Table 1). This is an increase in the area reported in
of the categorical variables: climatic domain, national income FRA 2010 which reported coverage of about 70% of global forest
category or sub-region. area under forest policy and legislation. Countries reporting only
50 K.G. MacDicken et al. / Forest Ecology and Management 352 (2015) 47–56
Table 2
Proportion of forest area covered by policies and or legislation supporting SFM by
region and sub-region.
area. This demonstrates the degree with which devolution and Oceania Oceania 33 97
decentralisation of regulatory framework for forest management South America South America 93 100
has occurred.
There were no major differences in the proportion of countries
and forests covered in terms of presence of policies versus legisla-
tion (Table 2). The presence of policies and legislation were sharply
skewed to the national level, particularly in the Caribean, Central
America, Europe, South and Southeast Asia and West Central Asia
where the proportion of countries investing in subnational policies
and or legislation was well below 50%. More than 90% of the
countries in Africa indicated they had policies and/or legislation
supporting SFM covering 94% of the forests in the region. Each
sub-region had more than 90% of the forests covered by policies
and/or legislation except for West and Central Asia that was just
below 80%.
Analysis by climatic domain revealed no difference in the
degree of SFM-supportive policy and legislation across domains. Fig. 4. Forest area covered by policies supporting SFM by income category and level
Presence of legislation and or policies did not vary significantly of decentralization.
across income groups. The geopolitical analysis revealed that coun-
tries in the lower income categories invested more in national level
and success of these platforms certainly varies substantially. In
policies while higher income category countries investing in SFM
part the inclusion of stakeholders is the result of widespread pro-
policies across all levels of governance (Fig. 4).
motion of stakeholder participation in forest policy development
(FAO, 2009). Stakeholder engagement can help mitigate conflicts,
3.2.3. National stakeholder platforms enhance cooperation across stakeholder groups and improve the
FRA 2015 defined national stakeholder platforms (NSP) as the quality of national forest policy – although it can be a difficult
presence of a recognized procedure that a broad range of stake- and complex process to implement effectively (FAO, 2009). The
holders can use to inform the national policy process through their functionality of reported NSPs was not investigated but their pres-
opinions, suggestions and analysis. Significant progress has been ence is a positive starting point for most countries.
made in allowing or encouraging stakeholder inputs into forest
policy processes – countries indicated that 94% of global forest area
is covered by national stakeholder platforms, although the extent 3.3. Data availability and reporting
Geographically the degree of NFI coverage varies across Sub-region % of sub- Area of sub- Total forest Area
sub-regions with most of the world’s forest area (77%) in region area region area of sub- inventoried as
sub-regions with high levels of NFI coverage (Table 3). Four percent with forest inventoried region % of global
inventory (000 ha) (000 ha) forest area
of the world’s forest were in sub-regions with lower (40–79%) NFI
coverage. West and Central Africa have coverage of only 10% of Sub-regions with (80% or more NFI coverage)
Central America 90 76,614 86,290 1.9
sub-regional forest area, presenting a significant challenge to
East Asia 93 239,509 257,047 6
future governments to enhance forest management based on Europe 97 993,878 1,015,482 24.9
NFI data. Considering the importance of the forests in these North America 100 655,789 657,167 16.5
sub-regions to the well-being of people and the biodiversity found Oceania 96 166,863 173,376 4.2
South America 80 678,096 842,132 17.1
there, these data support the need for strategic investment in those
South and 80 234,374 292,804 5.9
sub-regions most lacking in NFI coverage – particularly countries Southeast Asia
in West and Central Africa. Total 3,045,123 3,324,298 77
There was a very strong relationship between national income Sub-regions with (40–79% NFI coverage)
and forest area covered by forest inventory – as national income Caribbean 48 3530 7328 0.1
increases, so does the proportion of forest area covered by a East and Southern 49 126,317 274,866 3.2
national forest inventory (NFI) (Table 4). High income countries Africa
North Africa 53 19,651 36,217 0.5
reported 98% of the forest area covered by national forest inven-
West and Central 50 19,898 39,711 0.5
tories. Low income countries had national inventories covering Asia
37% of the forest area within this grouping suggesting this should Total 169,396 358,122 4
be considered as a priority group of countries for additional NFI Sub-region with <40% NFI coverage
investment due to the highest annual forest loss rates ( 0.6%) West and Central 10 32,678 312,997 0.8
reported for any income category (Keenan et al., 2015) and a lack Africa
of reliable inventory data on forest change.
the national level. The least helpful in determining progress to Inventory/sampling type % of total area under NFI
SFM are reports in the ‘‘other’’ category. National periodic state Continuous forest inventory 36
of the forest type reports can be more helpful providing data Periodic forest inventory 10
needed in determining progress to SFM by providing consistent Combination of continuous and periodic 48
details about the forest resource over time. C&I reports are Single inventory event 6
purposefully designed to provide comprehensive information on Ground plots used 96
social, economic and environmental aspects of a nation’s forest Permanent ground plots 37
Temporary ground plots 8
that, when considered together over time, can provide insight into
Both permanent and ground plots 51
that country’s progress towards SFM. No use of ground plots 4
The data show a significant commitment by countries to report
Aerial imagery or remote sensing used 70
on their forests. Most countries (60%) use a periodic state-of-the- Sample based 29
forest type of report, globally covering 89% of forest area Full coverage 19
(Table 6). Criteria and indicator reports were drafted by 86 coun- Combination of sample and full coverage 22
tries that together covered 77% of total forest area and sixty-nine
countries used ‘‘other’’ report formats mostly in combination with
the other reporting types. the 22% of the global forest area that was not the subject of C&I
There were 41 countries that produced all three types of report, reporting is located within 107 countries, many of which have little
representing 67% of global forest area. It is important to note that or no forest area. The 11% of the global forest area not having per-
iodic state of the forest reporting was within 77 countries and the
3
NFI support has included Algeria, Angola, Bangladesh, Brazil, Cameroon, Comoros 28% of the global forest area not reported within ‘‘other’’ reports
Island, Congo, Costa Rica, Ethiopia, Gambia, Guatemala, Honduras, Kenya, Kyrgyzstan, were found across 124 countries. No reports were produced by
Lebanon, Nicaragua, Panama, Paraguay, Peru, Philippines, Tanzania, Uruguay, Viet
38 nations that represented 5% of the global forest area.
Nam, Zambia (Source: National Forest Monitoring Programme, FAO). UN REDD has
provided country support for forest monitoring in 58 countries (I. Jonckheere, pers. Nearly all of the forest area in the high, upper middle and lower
com.). middle income countries were covered by C&I reporting (Table 7).
52 K.G. MacDicken et al. / Forest Ecology and Management 352 (2015) 47–56
Table 6
Forest reporting types by number of countries and global forest area coverage.
Table 7
Forest area reported through C&I reporting by income category.
Table 8
Forest area with FMP by climatic domain.
Domain Forest with FMP FMP for production FMP for conservation
Area (000 ha) % of domain forest area Area (000 ha) % of domain forest area Area (000 ha) % of domain forest area
Tropical 509,761 28.2 191,267 10.6 203,787 11.3
Temperate 424,971 63.1 175,516 26.1 209,428 31.1
Boreal 1,073,801 87.7 442,734 36.1 401,497 32.8
Boreal without Russia 258,656 63.1 21,243 5.2 7,852 1.9
Sub-tropical 91,131 28.5 36,505 11.4 28,678 8.9
Total 2,099,664 846,021 843,391
K.G. MacDicken et al. / Forest Ecology and Management 352 (2015) 47–56 53
(Fig. 7). All of North America and most of Asia require soil and
50 water, social considerations and HCVF delineation while Europe,
South America, Africa and Oceania have countries with varying
40 requirements.
30
3.5. How are stakeholders involved in the management decision
20
making stages for public forests?
Fig. 7. Countries with required inclusion of soil and water management, stakeholder inputs and high conservation value forest designations in forest management plans.
54 K.G. MacDicken et al. / Forest Ecology and Management 352 (2015) 47–56
Acknowledgements
Kimmons, J.P., 1997. Forest Ecology: A Foundation for Sustainable Management. Rametsteiner, E., Simula, M., 2003. Forest certification—an instrument to promote
Prentice-Hall, New Jersey, USA, 596 pp. sustainable forest management? J. Environ. Manage. 67 (2003), 87–98.
Lovrc, M., Martinic, I., Lovric, N., Landekic, M., Sporcic, M., 2010. Assessment of Schmithüsen, F., 2013. Three hundred years of applied sustainability in forestry.
progress of forests in Croatia towards sustainable forest management SEEFOR Unasylva 240, vol. 64, 2013/1.
(South-east European forestry), vol. 1 No. 2 December 2010. UNFF, 2007. Highlights from Enabling Sustainable Forest Management: Strategies for
MacDicken, K.G., 2015. Global Forest Resources Assessment 2015: What, why and equitable development, for forests, for people United Nations New York <http://
how? Forest Ecol. Manage. 352, 3–8. www.un.org/esa/forests/pdf/publications/Enabling_SFM_highlights.pdf>.
MCPFE, 2001. Criteria and indicators for sustainable forest management of the United Nations Conference on Environment and Development, 1992. Agenda 21, Rio
MCPFE: review of development and current status. International expert Declaration, Forest Principles. New York: United Nations.
meeting on monitoring, assessment and reporting on the progress towards Walter, M., 2008. Analysis of the FSC and PEFC Systems for Forest Certification using
sustainable forest management, Yokohama, Japan. the FCAG.
Montréal Process, 2009. Criteria and Indicators for the Conservation and Sustainable West, T.A.P., Vidal, E., Putz, F.E., 2014. Forest biomass recovery after conventional and
Management of Temperate and Boreal Forests. Technical Notes on reduced-impact logging in Amazonian Brazil. Forest Ecol. Manage. 314, 59–63.
Implementation of the Montreal Process Criteria and Indicators. Zon, R., Sparhawk, W. N., 1923. Forest resources of the world (Vols 1 and 2).
Prabhu, R., Colfer, C., Shepherd, G., 1998. Criteria and Indicators for Sustainable McGraw-Hill Book Company, New York. 986 pages.
Forest Management: New Findings from CIFOR.s Forest Management Unit Level
Rural Development Forestry Network Paper 23a ODI.
Forest Ecology and Management 352 (2015) 57–67
a r t i c l e i n f o a b s t r a c t
Article history: This paper focuses on an analysis of planted forests data from the 2015 Forests Resources Assessment of
Received 13 March 2015 the U.N. Food and Agriculture Organisation (FRA 2015). It forms one of a series of papers in the FRA 2015
Received in revised form 15 June 2015 special issue of this journal.
Accepted 17 June 2015
While total forest area decreased from 4.28 billion hectares to 3.99 billion hectares from 1990 to 2015,
Available online 7 September 2015
with percent global forest cover dropping from 31.85% to 30.85%, the area of planted forests increased
from 167.5 to 277.9 million hectares or 4.06% to 6.95% of total forest area. Increase was most rapid in
Keywords:
the temperate zone, and regionally in East Asia, followed by Europe, North America, and Southern and
Planted forests
Global trends
Southeast Asia.
Climate However the annualised rate of increase in area of planted forests slowed in the 2010–2015 period to
Population 1.2%, below the 2.4% rate suggested is needed to supply all of the world’s timber and fibre needs.
The majority of planted forests comprised native species with only 18–19% of the total area being of
introduced species. Introduced species were dominant in the southern hemisphere countries of South
America, Oceania and Eastern and Southern Africa where industrial forestry is dominant.
Twenty countries accounted for 85% of planted forest area and a different 20 countries for 87% of
planted forest roundwood supply. As with forest area, roundwood supply from planted forests also
showed an increasing trend although this was based on minimal data. There was a mismatch in compo-
sition and rankings of the top 20 countries with top forest area and roundwood production suggesting
that there are substantial opportunities to increase roundwood production in the future, especially in
China which has the largest area but is currently ranked 3rd in roundwood production.
Outlook statements were developed for the FAO sub regions based on past changes in planted forest
area, population growth, and climate and forest health risks to identify key issues for the future. The over-
all view from this study suggests that climate impacts, especially from extreme climatic events will affect
planted forests in the future and that forest health impacts can also be expected to increase. Outlooks
vary regionally. Europe and North America are likely to be most concerned with climate and health risks;
Asia will experience population pressure that will impact on land availability for new forests and risks
from extreme weather events, and will need to make the most of its existing forests; Africa will need
to increase planted forest area to offset continuing deforestation and rapid population growth; and
Oceania, the Caribbean, Central and South America are likely to be most concerned with climate impacts.
To ensure the continued contribution of planted forests, a number of responses will be required to both
maintain existing and also to develop new forests. Intensification of production in existing forests will
lessen the need for greater forest areas and offset any land use conflicts related to food security; climate
adaptation strategies will need to be developed as a matter of urgency, and forest health focus must
remain a priority for research. Establishment of new forests will be eased through greater community
q
This article is part of a special issue entitled ‘‘Changes in Global Forest Resources
from 1990 to 2015’’.
⇑ Corresponding author at: Scion, Private Bag 3020, Rotorua, New Zealand.
E-mail address: tim.payn@scionresearch.com (T. Payn).
http://dx.doi.org/10.1016/j.foreco.2015.06.021
0378-1127/Ó 2015 Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
58 T. Payn et al. / Forest Ecology and Management 352 (2015) 57–67
and stakeholder engagement. Application of models such as WWF’s New Generation Plantations, which
recognises the importance of society and the need to consider the full range of forest products and ser-
vices within the wider landscape and spectrum of land uses, will be important.
We recommend that to enable deeper analysis related to planted forests future FRA Assessments con-
sider ways to better gather data specific to planted forests such as productivity so that this important
component of global forests can be better understood.
Ó 2015 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
will therefore be somewhat conservative. We also calculated 2012 Greener Future’’ (http://www.iufro.org/science/task-forces/-
roundwood per unit of planted area in 1990 as a crude estimate of planted-forests/).
productivity across the climate domains and sub regions.
3. Results
2.3.2. FAO planted forests thematic study
The FRA dataset apportions forests depending on their proposed
3.1. Total forest trends (1990–2015)
use; production, multi-use, and protection categories. These statis-
tics are for all forests combined. The intended use of planted for-
Total forest area decreased between 1990 and 2015 from 4.12
ests is important to know when looking at future timber supply
to 3.99 billion hectares. The Tropics and Subtropics both decreased
so we sourced data from the thematic study of planted forests
in area while Boreal and Temperate increased (Table 1).
(Del Lungo et al., 2006) where 61 countries were surveyed in detail
Changes in total forest cover varied regionally (Table 2). Central
as to their planted forests. We used this dataset to analyse the pro-
America, East Africa, North Africa, Oceania, South America,
portion of planted forests intended for production vs protection
Southern and Southeast Asia, and West and Central Africa
use on the 2015 top 20 planted forest countries (see Section 2.1)
decreased while the Caribbean, East Asia, Europe, North America,
identified.
and West and Central Asia increased. Overall percentage of land
covered by forest decreased from 31.85% to 30.85%.
2.4. Outlook analysis: climate, population pressure, and forest health
While total forest area declined, forest productivity increased
over the period from an average 4.10 m3 ha 1 yr 1 to 5.26 m3 -
An aim of this paper was to present an outlook for planted for-
ha 1 yr 1 with the changes varying by climate domain and region.
ests that relates to the two key global drivers – climate and popu-
Temperate zone productivity increased the most (1.91 m3 ha 1 -
lation change. Another aspect that is hard to extract from the FRA
yr 1) with the boreal zone also increasing slightly. These data
dataset is forest health status but one that is very important to
should be taken as indicative as only 45% of countries surveyed
consider.
provided data for this variable.
Population data are well covered in the FRA dataset and we
analysed various planted forests variables in relation to population
trends. However climate trends are not covered in the FRA dataset. 3.2. Planted forest trends
To evaluate the impact of climate on the planted forest outlook we
utilised two sources on information: 1. The IPCC 5th Assessment 3.2.1. Planted forest area
report (IPCC, 2014); and 2. The Germanwatch climate risk index Global planted forest area increased from 1990 to 2015 from
(Kreft and Eckstein, 2013). 167.5 million ha to 277.9 million ha with the increase varying by
The IPCC 5th Assessment report outlines climate related drivers region and climate domain. Annualised rates of increase in area
e.g. warming trends, drying trends, extreme precipitation; and were highest in the 1990–2000 period (2.0%) and the 2000–2005
associated key risks e.g. drought, storm and flood damage, period (2.7%) but dropped in 2005–2010 (1.9%) and further in
decreased crop production. We mapped these drivers and risks to 2010–2015 to 1.2%. This drop may be of concern given that a study
the FAO subregions. by WWF and IIASA have suggested that a rate of increase of 2.4% is
The Germanwatch risk index is developed from historic data needed to meet future demands and supply all of the world’s tim-
analyses regarding the extent to which countries have been ber and fibre and thus offset deforestation impacts on wood supply
affected by the impacts of weather-related loss events (storms, (WWF and IIASA, 2012).
floods, heat waves, etc.). The 2014 index is based on data from Of the 277.9 million ha of planted forests in 2015 56% are in the
1993 to 2012. It provides a country level index which we compiled temperate zone, 15% boreal, 20% tropical, and 9% subtropical
to give a population weighted average for each of the FAO subre- (Fig. 1). All zones showed an increase in area, with the largest
gions. It serves to provide an indication of past impacts that may increase in area in the temperate zone (93.4–154.4 million ha) fol-
be extrapolated to the future where these climate impacts will lowed by tropical, boreal, and subtropical (Fig. 2).
remain the same or worsen under climate change. The subregional Overall East Asia and Europe had the largest areas of planted
range was from 40 (highest risk) to 111 (lowest risk). forests, followed by North America and Southern and Southeast
Forest health data in the FRA dataset shows area affected by the Asia (Fig. 3). All regions, except Central America, showed an
10 most important extensive pest and disease impacts/outbreaks increase in planted forest area over the period with the greatest
for each country. We summed the total area affected and calcu- absolute increases in these top four regions followed by South
lated percentage of total forest area affected and averaged this by America. Central America showed a very slight decrease between
region. Only 76 countries reported this variable so values must 1990 and 2015, probably within the margin of error. The trend of
be taken as only indicative. Additionally, and as for other variables, a slowing of expansion rate (annualised% change in area) noted
this does not relate to planted forests per se but could provide a globally was also apparent at the subregional level (Table 3).
perspective on current forest health impacts by country and Most regions showed a peak in annualised percent change in either
region. To augment these data we considered published studies 2000–2005 or 2005–2010 followed by a decrease in expansion rate
to develop the regional outlook for health risk. in 2010–2015. The only exception was East Africa where the
Utilising the forest area, population, climate and health infor-
mation we developed a subregional outlook from a three class
Table 1
(low, medium and high) ranking of current forest areas (planted
Change in total forest area by climate domain from 1990 to 2015.
and natural), population and population density; and changes in
population and natural and planted forest area between 1990 Year Forest area by climate domain (Million ha)
and 2015. This information was then considered in relation to Boreal Temperate Subtropics Tropics Total
the climate drivers and risks, and forest health state, and an out- 1990 1219.3 618.0 325.4 1965.5 4128.3
look statement developed. This outlook was developed in terms 2000 1219.8 640.9 324.8 1870.1 4055.6
of level of future challenge facing planted forests. These outlook 2005 1218.9 659.2 323.9 1830.8 4032.7
2010 1224.9 673.4 319.6 1797.8 4015.7
statements were subject to review by planted forest experts in
2015 1224.5 684.5 320.1 1770.2 3999.1
the new IUFRO task force ‘‘Sustainable Planted Forests for a
60 T. Payn et al. / Forest Ecology and Management 352 (2015) 57–67
Table 2
Change in total forest area by FAO sub region from 1990 to 2015.
15% 100
91.8
20% 90
56%
154.5
2000
160
2005
composition of the planted forests, with only between 18% and
140 19% of the planted forests comprising exotic or introduced species.
Planted forest area (Mha)
2010
2015 The high proportion of native species in planted forests is not well
120 appreciated by those opponents of planted forests who cite as a
100 major concern the use of introduced species.
When considering the distribution regionally we can see that
80 South America, Oceania, and East and Southern Africa are the
60
56.8 regions dominated by plantings of introduced species with 88%,
41.9 75% and 65% of plantings with introduced species respectively
40 (Fig. 5). North America, West and Central Asia, and Europe are at
24.7
20
the other end of the spectrum with 1%, 3% and 8% of the area
planted in introduced species.
0
Tropics Subtropics Temperate Boreal
3.2.3. Country level trends
Climate zone
For country-level data, planted forest area for the top 20 coun-
Fig. 2. Trends in planted forest area 1990–2015 by climate domain. tries is shown in Fig. 6. These top 20 countries accounted for 85% of
planted forest area and were distributed across most regions bar
Central America, the Caribbean, and West, Central, East and
percentage change increased in each five year period peaking at
Southern Africa. In 2015, China had by far the largest planted forest
2.65% in the 2010–2015 period.
area at 91.8 million ha followed by the USA (26.4), the Russian
Federation (19.8) and Canada (15.8).
3.2.2. Planted forest contribution to global forest area None of the 20 countries showed a decrease in planted forest
Deforestation is a major concern globally, and total forest area area though Japan and Germany were static over the period.
has decreased over the 1990–2015 period (Table 1). Planted forests China showed by far the largest absolute increase in area (37.0 mil-
play a role in offsetting the pressure and negative impacts on nat- lion ha) followed by Canada (11.2), the USA (8.4) and Russia (7.2).
ural forests. Analysis of total global area and the relationship The next tier were India and Sweden (6.3), Brazil (2.8) and Finland
between natural and planted forest area shows that the decrease (2.4). Data for Indonesia were missing for 1990 but this country
in natural forest area is offset by the increase in planted forest area also appears to be of this second tier order of increase.
to an increasing degree with time. Planted forests increased from These countries also exhibited the slowdown in expansion rate
4.1% to 7.0% of the total forest area over the period, or 1.29% to noted at the global and regional level (Table 3). Generally, the per-
2.14% of total land area1 (Fig. 4). Also of interest was the species iod 2000–05 had the most rapid increase in planted forest area
with the rate slowing in 2005–10 and slowing again in 2010–15,
1
Based on land area from UN Statistics Division. although for some countries the peak was between 2005 and
T. Payn et al. / Forest Ecology and Management 352 (2015) 57–67 61
Table 3
Annualised rate of change in planted forest area (%) (shaded cells are peak values).
2010. The decline in planting rate was common across most coun-
Planted forest Planted forest
tries with only 3 of the 20 countries (Chile, Sudan, and Ukraine)
(introduced (native species)
species) Natural forest
planting at a greater rate in the last 5 years (2010–2015) than in
33% the 5 previous years.
32%
90%
the global industrial roundwood production from planted forests,
80%
75%
totalling about 675 million m3. The country mix and rankings of
these top 20 countries differs from the top 20 area rankings
70% (Fig. 6) reflecting opportunities for large increases in production
65%
in countries such as China where area is high and production cur-
60% rently relatively low.
50%
3.3.2. Temporal trends in roundwood supply 1990–2015
42%
The FRA 2015 dataset showed an overall increase in roundwood
40%
production from 1990 to 2015 though data was sparse. The
time-series data on the industrial roundwood production from 17
30%
25% countries reported in the Jürgensen et al. (2014) study showed that
21%
18%
for many countries in Latin America and Asia (Chile, Brazil, China,
20%
Indonesia, Malaysia, Myanmar, Thailand, Uruguay and Vietnam),
12%
10%
8%
the production had increased considerably since 2000; while in
10%
3%
Argentina, Australia, New Zealand and the United States of
1%
America, production had been increasing at a considerably slower
0%
pace. In European countries (Portugal, Spain, and Turkey) and in
South Africa, the trend in industrial roundwood production had
basically been stagnant since 2000, with some noticeable ups
and downs during this period.
Subregion
3.4. Population, climate and forest health trends
Fig. 5. Proportion of planted forests with introduced species by FAO subregion.
3.4.1. Population
Over the period 1990 to 2015 global population increased from
79.0
just over 5bn to just over 7bn an annualised rate of increase of
80
1.46%. Highest 2015 population densities were in Southern and
70 1990 Southeast Asia, Caribbean and East Asia (Table 5). All regions show
Planted forest area (Mha)
2000
2005 increasing population but rate of population increase varies
60 2010
2015
regionally. Southern and Southeast Asia, Western and Central
50
44.1
Africa, and East Asia showed the largest rate of increase with
Europe the lowest rate. Comparing annualised population growth
40
rate with annualised change in planted forest area showed that
30 26.4 globally the rates are similar, but regionally these vary. Europe
19.8 for example, had minimal population growth, but a 37% increase
20 15.8
13.7
12.0
10.3
in planted forest area while Central America has shown a 45%
7.7 6.8
10 6.1 5.3 4.9 4.9
4.0 3.7 3.4 3.0 2.9 increase in population with only a 17% increase in planted forest
2.1 2.0
0
area. Southern and Southeast Asia by contrast shows large
increases in both population and planted forest area. We used
these relationships to infer pressure on future supply of timber
Country regionally – high pressure where annualised population growth
rate was greater than annualised forest area increase and vice
Fig. 6. Planted forest area and trends (1990–2015) for the top 20 countries by versa for low pressure.
planted forest area.
3.4.2. Climate
The regional analysis of the Germanwatch climate risk index
must be reflected by a corresponding increase in wood production showed a range of risk indices across regions and also across cli-
in future decades. This all points to a greater proportion of wood mate zones (Table 5). Highest risk regions (lowest scores) were
being supplied from planted forests in the future. Already in Central America, the Caribbean, East Asia, and Southern and
2012, almost as much (46.3%) industrial wood comes from planted Southeast Asia; and lowest risk regions Western and Central
than natural forest when fuel wood is excluded from the analysis. Africa, Western and Central Asia, North Africa, and Eastern and
An analysis of the data by region indicated that the production Southern Africa. In terms of planted forest regions this suggests
of industrial roundwood in planted forests in 2012 was close to that Eastern Asia and Southern and Southeast Asia with large and
200 million m3 in South America (193 million m3), followed by increasing areas of planted forests are most at risk. This analysis
Asia (151 million m3) and North and Central America (104 mil- gives a retrospective view and future risk distributions may
lion m3). Oceania, Europe and Africa produced considerably less, change.
ranging from 26 million to 47 million m3. Productivity and per- The IPCCC AR5 findings suggest that all regions will be affected
centage roundwood sourced from planted forests was highest in by climate change and that there may well be effects on planted
South America (24.0 m3 ha 1 yr 1; 92.4%) and Oceania followed forests. IPCC AR5 projections (IPCC, 2014) for climate change sug-
by East and Southern Africa and Southern, Southeast, and East gest increased risks and impacts globally, both from direct climatic
Asia. These regions host countries with highly developed inten- events such as storms, but also indirectly from increased risks from
sively managed tree plantations and fast growing exotic species. fire, pests and diseases, or spread of invasive species. As most
T. Payn et al. / Forest Ecology and Management 352 (2015) 57–67 63
Table 4
Planted forest roundwood production in 2012 by region and climate domain adapted from Jürgensen et al., 2014.
Total forest Planted forest Annualised% change in Planted forest Planted forest % of 2012 Planted forest
area 2015 area 2015 planted forest industrial roundwood total industrial roundwood productivity
(Million ha) (Million ha) area 1990–2015 2012 (1000 m3) roundwood index based on 1990
area (m3 ha 1 yr 1)
World 3999.1 277.9 2.0 770,200 46.3 4.6
Tropical 1770.1 56.8 2.5 255,300 63.7 8.4
Subtropical 320.0 24.7 1.2 69,600 64.7 3.8
Temperate 1031.5 154.4 2.0 410,100 45.2 4.4
Boreal 877.3 41.9 2.0 35,200 13.9 1.4
South America 842.0 15.0 2.5 193,000 89.8 24.0
Oceania 173.5 4.3 1.9 47,500 84.0 17.2
East and Southern Africa 274.8 4.6 1.2 20,700 64.7 6.0
Caribbean 7.1 0.7 2.4 300 24.7 0.7
East Asia 257.0 91.8 2.2 78,700 46.9 1.5
Central America 86.2 0.4 0.6 1600 18.0 4.3
West and Central Africa 313.0 3.2 3.2 5100 14.1 3.5
Southern and SE Asia 292.8 29.9 3.4 82,700 52.0 6.4
North Africa 36.2 8.4 0.9 400 15.7 0.1
Europe 1015.4 70.4 1.3 166,200 33.4 3.2
West and Central Asia 43.5 6.7 2.1 3900 19.1 1.0
North America 657.1 42.1 2.5 170,100 36.0 7.6
131.9
120
CO2) will continue to exacerbate the establishment and spread of
100
pests, vectors and pathogens, and negatively impact production
80 systems such as planted forests (Robinet and Roques, 2010).
64.2
60
Globally, biomass and soil carbon stocks in forest ecosystems are
43.1
38.4
currently increasing but are vulnerable to loss to the atmosphere
40
28.6 28.5 27.5 as a result of rising temperatures, droughts, and fires projected in
20.4 19.2 18.9
20 15.9 14.6 13.7 13.5
12.5 11.6 11.6 10.0 9.6 the 21st century. Measurements of increased tree growth over
the last several decades, a large sink for carbon, are consistent with
0
this (Settele et al., 2014), but confounding factors such as N depo-
sition, increasing area of productive planted forests, and forest
management practices make attribution of these trends to climate
Country change difficult.
Fig. 7. The world’s top 20 producers of industrial roundwood from planted forests
in 2012. 3.4.3. Forest health
Only 76 countries supplied data for this variable covering a total
forest area of 2.8 billion hectares or 30% of global forests. Averaged
terrestrial ecosystems, planted forests are vulnerable to climate
across the 76 countries, 8.5% of forest area was affected by pest and
change projected even under low to medium-range warming sce-
disease outbreaks. The percentage area affected was highly vari-
narios (RCP22.6 to RCP6.0) as defined in IPCC AR5; in the second half
able and ranged from very close to zero to 82%. Average percent
of the 21st century, climate change is projected to be a powerful
area affected varied regionally (Table 5) from a minimum of
stressor specially under high-warming scenarios such as RCP6.0
0.37% in Western and Central Africa to 12.04% in Europe. Trend
and RCP8.5 (Settele et al., 2014). Increases in the frequency or inten-
data were not available, but it seems that the total area affected
sity of disturbances such as droughts, wind storms, fires and pest
has increased from previous FRA surveys – total area affected in
outbreaks have been detected in many parts of the world where
2015 was 141 million ha, as compared to 68 million ha in 2005
planted forests are located and in some cases are attributed to cli-
(FRA 2005/thematic study/chapter4).
mate change (medium confidence in IPCC AR5). Also, the establish-
ment, spread and survival of populations of invasive species have
increased (IPCCAR5), mainly due to increased dispersal opportunities 4. Discussion
or to increased disturbances rather than climate change.
The consequences for the provision of timber and other wood 4.1. Outlook
products are projected to be highly variable between regions and
products, but might induce an increased demand on wood supply The FRA 2015 data analysed in this study suggest that in
from planted forests. Decreased production from planted forests is planted forests, a new timber resource is continuing to be created
expected in already dry forest regions where increasing water def- and that it will contribute significantly not only to future wood and
icit is projected such as the south-western part of Europe, USA or energy supplies but can also meet a range of wider social and envi-
ronmental benefits (ecosystem services). Planted forests are likely
2
RCPs (Representative Concentration Pathways) are identified with the radiative
to continue to supply an increasing proportion of the world’s wood
forcing by 2100; four main RCPs scenarios have been used in AR5 (8.5, 6.0, 4.5, and requirements; the trend is supported by this new dataset. The abil-
2.6 W m 2). ity of planted forests to increase supply will depend on global and
Table 5
64
Summary of forest, population trends, and climate impacts and risks by FAO sub-region.
Subregion Population % Change % Change % Change IPCC key risks IPCC climate related Climate % Forest area Outlook
density since since since 1990 drivers Risk affected by
2015 1990 1990 Population Index outbreaks
(n/km2) Natural Planted
Forest Forest
South and 276 14 132 49 Drought, decreased crop Warming trend, drying 40 1.2 High population density, rapid population increase, large natural
Southeast Asia productivity, storm and trend, extreme forest loss, but large planted forest expansion though rate
flood damage, drought temperatures, extreme slowing, climate impacts high. Population pressure and climate
precipitation, cyclones impacts likely to put pressure on further planted forest area.
Focus on increasing productivity from existing resource,
adaptation to extreme climate impacts
Europe 33 0 37 2 Storm and flood damage, Warming trend, drying 59 12 Medium density but stable population, stable natural forest, low
decreased crop trend, extreme planted forest expansion, medium climate impacts and high
production, wildfires temperatures, extreme disease impacts. Focus on climate adaptation and forest health
precipitation management to maintain status
Caribbean 188 40 79 25 Storm and flood damage, Warming trend, drying 44 2.1 High population density, increased area of natural forest,
decreased crop trend, extreme medium increase in planted forest area, low population growth,
country policies, the sustainable forest management and trade competition for forest land, both natural and planted, and requires
requirements, and the development of supply chains and markets forestry and farming practices that produce more with less land
(Freer-Smith and Carnus, 2008). The FRA dataset has allowed us to and water. Regions with the highest population pressure (mainly
develop a global and regional picture of the extent and trends in a Asia) will have the highest level of competition and future expan-
number of planted forests variables over the period 1990 to 2015. sion may not be an option or may be very difficult.
Overall planted forests are playing a very significant role in 2015. The focus may then move to intensification of management and
While deforestation has continued to occur, the area of planted for- increasing productivity in existing forests. Well managed planted
ests has increased to 277 million hectares (now making up 6.95% of forests usually have higher yields of wood than natural, unman-
forest area) and this has slowed deforestation. While expansion aged forests, with commercial plantations in the tropics having
has continued over the period, the rate of that expansion has annual growth rates of 10–30 m3 ha 1 compared to 1–5 m3 ha 1
dropped in the decade from 2005 to 2015. So while general signals for natural forests (Evans and Turnbull, 2004). Annual growth rates
concerning the state of planted forests are good – increase in area, could not be assessed for planted forests within the FRA 2015 data
roundwood supply approaching that from natural forests – there set, but the planted forests thematic study (Del Lungo et al., 2006)
are some potential concerns regarding the ability to meet future gives a more comprehensive picture that complements Evans and
demands, but also some opportunities. Outlook statements devel- Turnbull. There are a number of examples of improvements in
oped for the FAO sub regions based on past changes in planted for- planted forest productivity in the literature, often related to
est area, population growth, and climate and forest health risks to genetic improvement, e.g., Bouffier et al. (2008). Brazil has shown
identify key issues for the future are summarised in Table 5. a steady increase in productivity; in the 1970s, the average yield
Overall, we suggest that climate impacts, especially from extreme was 13 m3 ha 1 yr 1, and it currently exceeds 40 m3 ha 1 yr 1
climatic events will affect planted forests in the future and that for- (Gonçalves et al., 2013; IBA, 2015). New Zealand can demonstrate
est health impacts can be expected to increase. Expansion of a 25% improvement in productivity through use of improved geno-
planted forest area will be at a slower rate than previously and this types in its radiata pine forests (Kimberley et al., 2015).
may put pressure on roundwood supply; however there will be a Intensification is a feature of many national research programmes,
lag between increase in area and supply so this may not be a real e.g., www.gcff.nz.
concern over time. It is clear from this study and from the associated literature that
Regionally there are variations in the outlook for planted for- pest (including insects and pathogens) problems are giving rise to
ests. The two most stable regions are Europe and North America. increasing levels of loss in planted forests. This includes both those
Both these regions have well established and relatively stable, or that are established based on native and non-native species. Native
increasing, areas of planted forests. They also have stable natural species in planted forests are often very seriously damaged by
forest areas, and populations are either stable or only growing native pests but they are also subject to increasing problems due
slowly. The main issues for these two regions will be climate to accidental introductions of non-native organisms. Plantations
change, adaptation to extreme events and managing forest health of non-native, or introduced, species such as Eucalyptus, Pinus
impacts. By contrast Asia – South and Southeastern, Western and and Acacia have typically been relatively free of pest problems dur-
Central, and to a lesser degree Eastern – has a high and increasing ing the early years of establishment due to a separation from their
population pressure, recent rapid expansion of planted forest area natural enemies (Wingfield et al., 2008, 2010). In all areas studied,
and relatively low roundwood production from these new forests this situation has changed dramatically as pests are accidentally
and high climate impacts and risk index. These regions will proba- introduced, but also where native organisms have adapted the
bly face a number of challenges both from population pressure ability to infect/infest trees (Wingfield, 2003; Wingfield et al.,
slowing any future increase in planted forest area and from some 2010). Given the dramatic increases in the movement of pests
very significant impacts from extreme climate events. To respond globally, largely driven by anthropogenic factors (Liebhold et al.,
to these pressures, focus on enhancing productivity from existing 2012; Santini et al., 2013), forest health will be an increasingly
forests and developing robust climate adaptation strategies is important constraint to forest productivity although the applica-
likely to be a high priority. The dominant risk to South America, tion of modern breeding and other technologies can offset losses
including the planted forest powerhouses of Brazil and Chile, (Wingfield et al., 2012). The impacts of future climate related
appears to be climate impacts. Africa has a major issue with con- events remain poorly understood although there is good evidence
tinuing loss of natural forest area, population growth and only for increased pest problems in some situations (Sturrock et al.,
low to moderate increase in planted forest areas. This suggests that 2011). These will require increased emphasis on forest health man-
focus will need to be on expanding planted forest areas, but that agement and also adaptation strategies to deal with a changing cli-
this will have to take into account the high risk of future droughts mate. Investment in forest health research is globally significant
and also pests and diseases on the success of these new forests. and organisations such as IUFRO (www.iufro.org) are coordinating
Other smaller regions – Oceania, the Caribbean, Central America major global initiatives on forest health including for example
– will face their main challenges around the impacts of extreme newly established task forces on these topics.
climate events. One of the most important challenges to the further expansion
of planted forests and indeed the on-going management of existing
4.2. Challenges and opportunities facing planted forests forests is that of societal expectations. In the past there was serious
concern about conversion of natural forest systems to plantations
The challenges facing planted forests will come from population and thus a strong negative perception of planted forests.
growth and climate impacts, but also other issues such as However, with the development of independent third party audit
Governance (Cubbage et al., 2014) which can affect forest invest- certification schemes such as FSC (www.fsc.org) and PEFC (www.
ment and management. The noted decrease in expansion rate of pefc.org) in the 2000s this perception is changing. These schemes
planted forest area, while unexplained here, is of concern. disallow conversion of natural forests to plantations, and together
Competition for land is likely to be a major factor affecting future with the development of WWF’s New Generation Plantations plat-
expansion. The UN is projecting that an increase in food production form and its concept of well-managed and designed, inclusive and
of 60% will be required by 2050 (UN 2013). A more recent study profitable plantations (NGP, 2014), the negative views of planted
(Ray et al., 2013) suggests it will not be possible to meet this target forests will most likely change. Acceptability does appear to be
by boosting crop yields on existing land. This will lead to increasing with a more positive view from bodies such as FSC
66 T. Payn et al. / Forest Ecology and Management 352 (2015) 57–67
where plantations are now fully integrated into global standards strategies to mitigate climate change impacts. Additionally the
(Maunder 2014 pers comm). This will make it easier to develop increasing involvement of communities with planted forest issues
new forests in the future. Community involvement and dialogue and the development of socially based programmes such as forest
between forestry companies and stakeholders is becoming increas- certification and New Generation Plantations will aid in future
ingly important. Initiatives such as The Forests Dialogue (The management of existing forests and development of new ones.
Forests Dialogue, 2015), convenes exchanges between forestry
companies and civil society organisations, provides an ongoing, Acknowledgements
multi-stakeholder dialogue focused on developing mutual trust, a
shared understanding, and collaborative solutions to sustainable The authors acknowledge the support of their organisations for
forest management across a range of topics such as ‘‘Intensively the development of this paper; the support and encouragement of
Managed Planted Forests (IMPF)’’, and ‘‘Food, Fuel, Fibre and the FAO, and especially Ken MacDicken, in making available the
Forests (4F’s)’’. 2015 Global Forest Resources Assessment dataset for the analysis
Increasingly, planted forests will need to be recognised within undertaken here; and the comments of three anonymous review-
the wider community for the range of values provided, not just ers on an earlier draft.
the timber. While this analysis has shown that the predominant
use of planted forests by the top 20 countries was for ‘production’, References
planted forests are established and managed for a wide range of
objectives not solely for intensive wood production. Historically Bateman, I.J., Harwood, A.R., Mace, G.M., Watson, R.T., Abson, D.J., Andrews, B.,
trees and woodlands have been planted for landscape, protection Binner, A., Crowe, A., Day, B.H., Dugdale, S., Fezzi, C., Foden, J., Hadley, D.,
Haines-Young, R., Hulme, M., Kontoleon, A., Lovett, A.A., Munday, P., Pascual, U.,
(against snow avalanche, landslip and soil erosion), hunting and Paterson, J., Perino, G., Sen, A., Siriwardena, G., Van Soest, D., Termansen, M.,
other socio economic objectives. In the East Asian region for exam- 2013. Bringing ecosystem services into economic decision-making: land use in
ple the protection which planted forests provide against soil ero- the United Kingdom. Science 341 (6141), 45–50.
Bouffier, L., Rozenberg, P., Raffin, A., Kremer, A., 2008. Wood density variability in
sion and flooding have been important drivers of woodland successive breeding populations of maritime pine. Canadian J. Forest Res. 38 (8),
creation. In China alone, the Natural Forest Protection Program 2148–2158.
(NFPP) and the Conversion of Cropland to Forest Program (CCPF), Boyd, I.L., Freer-Smith, P.H., Gilligan, C.A., Godfray, H.C.J., 2013. The consequence of
tree pests and diseases for ecosystem services. Science 342 (6160).
triggered mainly by the flooding disaster of 1998, have generated Buongiorno, J., Zhu, S., 2014. Assessing the impact of planted forests on the global
afforestation area of more than 32.5 million hectares. forest economy. NZ J. Forest Sci. 44 (Suppl 1), S2.
In Europe and North America the increasing areas of planted Carle, J., Holmgren, P., 2008. Wood from planted forests: a global outlook 2005–
2030. Forest Prod. J. 58, 6–18.
forests are likely to have even broader explanations. The
Cubbage, F., Mac Donagh, P., Balmelli, G., Morales Olmos, V., Bussoni, A., Rubilar, R.,
Millennium Ecosystem Assessment and subsequent research and De La Torre, R., Lord, R., Huang, J., Afonso Hoeflich, V., Murara, M., Kanieski, B.,
policy initiatives have been informed by the ecosystem services Hall, P., Yao, R., Adams, P., Kotze, H., Monges, E., Hernandez Perez, C., Wikle, J.,
concept. This categorises the benefits/services of ecosystems as: Abt, R., Gonzalez, R., Carrero, O., 2014. Global timber investments and trends,
2005–2011. NZ J. Forest. Sci. 44 (Suppl 1), S7.
Supporting Services such as primary production, soil formation, Del Lungo, Ball, J., Carle, J., 2006. Global planted forests thematic study: results and
nutrient and water cycling which provide the basic infrastructure analysis FAO. Planted Forests and Trees Working Paper 35b , p. 38.
of life; Provisioning Services which are the goods such as food, fuel Evans, J., 2009. Planted forests: uses, impacts and sustainability. Rome: CAB
International and Food and Agriculture Organization of the United Nations
and fibre; Regulating Services such as climate and hazard regula- (FAO).
tion (prevention of erosion, carbon storage, water regulation, ava- Evans, J., Turnbull, J.W., 2004. Plantation forestry in the tropics: the role, silviculture
lanche protection, etc.); and Cultural Services such as recreational and use of planted forests for industrial, social, environmental and agroforestry
purposes, OUP Oxford.
use, benefits to health and spiritual well-being. The important pol- Food and Agriculture Organization of the United Nations (FAO), 2010. Global Forest
icy development has been the recognition that as populations Resources Assessment 2010. <http://www.fao.org/docrep/013/i1757e/i1757e.
increase, there will be a need to manage rural and urban land- pdf>.
FAO, 2012. FRA 2015 Terms and Definitions. Forest Resources Assessment Working
scapes so as to benefit from the full range of services which wood- Paper 180, p. 31.
land and trees can provide. Relative to other land-uses, and Freer-Smith, P., Carnus, J.M., 2008. The sustainable management and protection of
particularly when compared to food production, the benefits from forests: analysis of the current position globally. Ambio 37 (4), 254–262.
Gonçalves, J.L.M., Alvares, C.A., Higa, A.H., Silva, L.D., Alfenas, A.C., Stahl, J., de Barros
planted forests especially their protection functions are often
Ferraz, S.F., de Paula Lima, W., Brancalion, P.H.S., Hubner, A., Bouillet, J.D.,
uncosted and may be enjoyed by stakeholders other than those Laclau, J., Nouvellon, Y., Epron, D., 2013. Integrating genetic and silvicultural
who own the land (Boyd et al., 2013). It has become recognised strategies to minimize abiotic and biotic constraints in Brazilian eucalypt
that land-use policies focussed solely on say agriculture or wood plantations. For. Ecol. Manage. 301, 6–27.
IBA, 2015. Brazilian Tree Industry 2015: a report of the Brazilian Tree Industry,
production, give lower overall benefits relative to policies which Brasilia, p. 62. <http://www.iba.org/images/shared/iba_2015.pdf>.
consider the full range of ecosystem services (Bateman et al., International Congress on Planted Forests (ICPF), 2013. Planted forests are a vital
2013). The drivers for establishment of new forests in the future resource for future green economies. Summary report of the 3rd International
Congress on Planted Forests. <http://www.fao.org/forestry/37902-
may therefore be through new opportunities from voluntary car- 083cc16479b4b28d8d4873338b79bef41.pdf>.
bon offset accreditation schemes or on a wider scale through the IPCC, 2014. Summary for Policy Makers. Climate Change 2014: Impacts, Adaptation,
FCCC REDD + scheme rather than just timber alone. and Vulnerability. Part A: Global and Sectoral Aspects. Contribution of Working
Group II to the Fifth Assessment Report of the Intergovernmental Panel on
In this study we have shown that planted forests will continue Climate Change, p. 32.
to be a major source of timber and other forest products that on an Jürgensen, C., Kollert, W., Lebedys, A., 2014. Assessment of industrial roundwood
area basis far outweigh production from natural forests. A recent production from planted forests. <http://www.fao.org/3/a-i3384e.pdf>.
Kimberley, M.O., Moore, J., Dungey, H., 2015. Quantification of realised genetic gain
paper (Buongiorno and Zhu, 2014) noted that planted forests in radiata pine and its incorporation into growth and yield modelling systems.
reduced harvesting from natural forests globally by 26% and had Can. J. For. Res. (accepted for publication).
significant ecological benefits. We have also shown that in some Kreft, S., Eckstein, D., 2013. Global Climate Risk Index 2014. Who Suffers Most from
Extreme Weather Events? Weather-Related Loss Events in 2012 and 1993 to
regions, expansion rates of planted forests are decreasing and that
2012. Germanwatch e.V, p. 28. <http://www.germanwatch.org/en/cri>.
climate impacts and population pressures on land availability will Liebhold, A.M., Brockerhoff, E.G., Garrett, L.J., Parke, J.L., Britton, K.O., 2012. Live
have an impact on planted forests in the future. We suggest these plant imports: the major pathway for forest insect and pathogen invasions of
pressures can be managed by development of more intensive man- the US. Front. Ecol. Environ. 10, 135–143.
New Generation Plantations (NGP), 2014. New generation plantations: review 2014.
agement regimes in existing forests, improved forest health man- <http://newgenerationplantations.org/multimedia/file/12b486cb-ea24-11e3-
agement techniques, and the development of adaptation 9f9e-005056986313>.
T. Payn et al. / Forest Ecology and Management 352 (2015) 57–67 67
Paquette, A., Messier, C., 2009. The role of plantations in managing the world’s Group II to the Fifth Assessment Report of the Intergovernmental Panel of
forests in the Anthropocene. Front. Ecol. Environ. 8 (1), 27–34. Climate Change. Cambridge University Press, Cambridge, United Kingdom and
Pawson, S.M., Brin, A., Brockerhoff, E.G., Lamb, D., Payn, T.W., Paquette, A., Parrotta, New York, NY, USA, pp. 271–359.
J.A., 2013. Plantation forests, climate change and biodiversity. Biodivers. Sturrock, R.N., Frankel, S.J., Brown, A.V., Hennon, P.E., Kliejunas, J.T., Lewis, K.J.,
Conserv. 22 (5), 1203–1227. Worrall, J.J., Woods, A.J., 2011. Climate change and forest diseases. Plant. Pathol.
Payn, T., Carnus, J., Freer-Smith, P., Orazio, C., Nabuurs, G.-J., 2014. Third 60, 133–149.
international congress on planted forests: planted forests on the globe – The Forests Dialogue, 2015. The Forests Dialogue: Overview. <http://
renewable resources for the future. NZ J. Forest. Sci. 44 (Suppl 1), S1. theforestsdialogue.org/about>.
Ray, D.K., Mueller, N.D., West, P.C., Foley, J.A., 2013. Yield trends are insufficient to Vira, B., Wildburger, C., Mansourian, S., 2015. Forests, trees and landscapes for food
double global crop production by 2050. PLoS ONE 8 (6), e66428. http:// security and nutrition. A global assessment report. IUFRO World Ser. 33, 1–172.
dx.doi.org/10.1371/journal.pone.0066428. Wingfield, M.J., 2003. Increasing threat of diseases to exotic plantation forests in the
Robinet, C., Roques, A., 2010. Direct impacts of recent climate warming on insect Southern Hemisphere: lessons from Cryphonectria canker. Australas. Plant
populations. Integrat. Zool. 5, 132–142. Pathol. 32, 133–139.
Santini, Ghelardini, A.L., De Pace, C., Desprez-Loustau, M.L., Capretti, P., Chandelier, Wingfield, M.J., Slippers, B., Hurley, B.P., Coutinho, T.A., Wingfield, B.D., Roux, J.,
A., Cech, T., Chira, D., Diamandis, S., Gaitniekis, T., Hantula, J., Holdenrieder, O., 2008. Eucalypt pests and diseases: growing threats to plantation productivity.
Jankovsky, L., Jung, T., Jurc, D., Kirisits, T., Kunca, A., Lygis, V., Malecka, M., Southern Forests 70, 139–144.
Marcais, B., Schmitz, S., Schumacher, J., Solheim, H., Solla, A., Szabò, I., Tsopelas, Wingfield, M.J., Slippers, B., Wingfield, B.D., 2010. Novel associations between
P., Vannini, A., Vettraino, A.M., Webber, J., Woodward, S., Stenlid, J., 2013. pathogens, insects and tree species threaten world forests. New Zealand J.
Biogeographical patterns and determinants of invasion by forest pathogens in Forest Sci. 40, S95–S103.
Europe. New Phytol. 197, 238–250. Wingfield, M.J., Roux, J., Slippers, B., Hurley, B.P., Garnas, J., Myburg, A.A., Wingfield,
Settele, J., Scholes, R., Betts, R., Bunn, S.E., Leadley, P., Nepstad, D., Overpeck, J.T., B.D., 2012. Established and new technologies reduce increasing pest and
Taboada, M.A., 2014. Terrestrial and inland water systems. In: Field, C.B., Barros, pathogen threats to Eucalypt plantations. Forest Ecol. Manage. 301, 1–10.
V.R., Dokken, D.J., et al. (Eds.), Climate Change 2014: Impacts, Adaptation, and http://dx.doi.org/10.1016/j.foreco.2012.09.002.
Vulnerability. Part A: Global and Sectoral Aspects. Contribution of Working WWF, IIASA, 2012. Living Forests Report. Gland, Switzerland.
Forest Ecology and Management 352 (2015) 68–77
Status and trends in global primary forest, protected areas, and areas
designated for conservation of biodiversity from the Global Forest
Resources Assessment 2015 q
David Morales-Hidalgo a, Sonja N. Oswalt b, E. Somanathan c,⇑
a
Food and Agriculture Organization of the United Nations, Forestry Department, Viale delle Terme di Caracalla, 00153 Roma, Italy
b
USDA Forest Service Southern Research Station, Forest Inventory and Analysis, 4700 Old Kingston Pike, Knoxville, TN 37922, USA
c
Indian Statistical Institute, Economics and Planning Unit, 7 Shaheed Jeet Singh Marg, New Delhi 110016, India
a r t i c l e i n f o a b s t r a c t
Article history: The global community has recognized the importance of forests for biodiversity, and has prioritized the
Received 18 December 2014 preservation of forest biodiversity and ecosystem functions through multiple multilateral agreements
Received in revised form 5 June 2015 and processes such as the Convention on Biodiversity’s Aichi Targets and the Millennium Development
Accepted 8 June 2015
Goals. The Global Forest Resources Assessment (FRA) provides one mechanism for tracking progress
Available online 7 September 2015
toward such goals in three particular areas: primary forest area, protected forest areas, and areas desig-
nated for the conservation of biodiversity. In this paper, we quantify current area and trends in forest
Keywords:
areas designated for the conservation of biodiversity, protected forest areas, and primary forests by coun-
FRA 2015
Biodiversity conservation
try and biome; and examine the association between total forest area and measures of protection,
Global Forest Resource Assessment per-capita income, and population. The overall findings suggest that countries are increasingly protecting
Primary forest forests of ecological significance at the global scale (7.7% of forests were protected in 1990 rising to 16.3%
Protected areas in 2015), with a strong upward trend in protected areas in the tropical domain (from 12% in 1990 to 26.3%
in 2015). However, primary forest area has declined by 2.5% globally and by 10% in the tropics over the
period 1990–2015 (using data for countries that reported in all years). Given that many species in the
tropics are endemic to primary forests, losses in that climatic domain continue to be of concern, although
the rate of decline appears to be slowing.
Using multiple regression analysis, we find that a 1% increase in protected area or area designated for
biodiversity conservation within a country is associated with an increase in total forest area in that coun-
try of about 0.03% (p < 0.05). A 1% within-country increase in population density and per capita GDP are
associated with a decrease in forest area of about 0.2% (p < 0.01) and an increase in forest area of about
0.08% (p < 0.05) respectively. Our findings also indicate that, since FRA is used as one mechanism for
tracking progress toward goals like the AICHI Biodiversity Targets, country correspondents may require
additional assistance toward reporting on primary forest, protected forest, and biodiversity conservation
statistics.
Ó 2015 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction biomass and providing habitat for over half of the world’s known
terrestrial plant and animal species (Shvidenko et al., 2005; Aerts
Forests provide critical ecosystem goods and services such as and Honnay, 2011).
food, water, shelter, and nutrient cycling among others, and play The global community has recognized the importance of forests
a fundamental role in conservation of biodiversity. According with for biodiversity, and has prioritized the preservation of forest bio-
recent studies, forests cover nearly 30 percent of the Earth’s land diversity and ecosystem functions through multiple multilateral
area (Keenan et al., 2015), containing 80 percent of terrestrial agreements and processes. For example, the Aichi Biodiversity
Targets established by the Convention on Biological Diversity
(CBD) in its strategic plan include halving the rate of loss of natural
q
habitats including forests (target 5) and conserving 17% of terres-
This article is part of a special issue entitled ‘‘Changes in Global Forest Resources
from 1990 to 2015’’.
trial areas through effectively and equitably managed, ecologically
⇑ Corresponding author. representative and well connected systems of protected areas (tar-
E-mail address: David.Morales@fao.org (E. Somanathan). get 11) (SCBD, 2006). Despite global acknowledgement of the
http://dx.doi.org/10.1016/j.foreco.2015.06.011
0378-1127/Ó 2015 Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
D. Morales-Hidalgo et al. / Forest Ecology and Management 352 (2015) 68–77 69
importance of forests, recent data show that forest area has contin- adjust their estimates as needed to bring them in line with the FAO
ued to shrink (Keenan et al., 2015) as agricultural land continues to definitions [see more details about the definitions and reporting
expand in 70% of countries (FAO, 2003). Deforestation is declining, process in MacDicken (2015)].
globally from 16 million ha annually in the 1990s to 13 million ha The global FRA offers one of the few, if not only, opportunities to
annually between 2000 and 2010 (FAO, 2010), but the current collate and analyze data on the above metrics across a long time
0.08% rate of loss reported by Keenan et al. (2015) for the years period at a global scale. When analyzed in relation to other metrics
2010–2015 is still a matter of concern because the loss is occurring (e.g., socioeconomic data), trends in these measures of forest area
in areas with particularly high ecological value. may provide insights into the factors that influence conservation
Given that a majority of known terrestrial species live in forests, of forests at a broad scale and offer possibilities for improvement
and 9 percent of tree species alone are currently threatened with both in forest conservation and in tracking the status of forests
extinction, it is clear that achieving success in meeting the Aichi believed to be particularly important for conservation. This explo-
Biodiversity Targets and Millennium Development Goals is linked ration of FRA data should reveal strengths and weaknesses in the
to slowing or reversing deforestation (JLG, 2008). While the global metrics currently reported in the global assessment.
community appears to agree on the importance of biological diver-
sity and forest cover, quantifying progress toward meeting biodi-
2. Methods
versity and forest degradation targets proves challenging.
Currently, designating protected areas is one of the primary
2.1. Sources
strategies for conserving biodiversity. Watson et al. (2014) discuss
the increase in protected areas over the past century; however,
FAO has been monitoring the world’s forests since 1946, ini-
they find that many key biodiversity areas are not adequately cov-
tially at 10-year intervals, and every 5 years since 2000. Since its
ered by protected area status. Joppa and Pfaff (2009) reached sim-
inception, FAO has produced eleven Global Forest Assessments.
ilar conclusions.
This paper is intended to serve as one component of the 12th glo-
Beyond forests that are specifically protected, countries may
bal FRA, dated 2015.
recognize certain forest areas as important for the conservation
The FRA provides a consistent approach to describing the
of biodiversity. The Global Forest Resource Assessment (FRA) asks
world’s forests and how they are changing. The assessment is
countries to report on forests that are designated for conservation
based on two primary sources of data: (a) Country and territory
of biodiversity with the understanding that these forests may
reports prepared by national correspondents and (in some cases)
include those in protected areas, but may also encompass other
desk studies by the FRA team and (b) remote sensing analysis con-
areas. That additional forest area may contribute to meeting
ducted by FAO together with national focal points and regional
Aichi Biodiversity Target 11, which calls for including 17% of ter-
partners. This paper excludes the remote-sensing analyses and is
restrial area in protected areas and other effective area-based con-
based only on the data collected for the country and territory
servation measures (SCBD, 2006).
reports and contained within the FRA database (available online).
While forests are recognized as important for biodiversity and
The FRA report compiles information from 234 countries and
protected areas are doubtless important for biodiversity conserva-
territories around the world. Since 2000, each participating coun-
tion, not all forests are equal in terms of the diversity they support
try has designated an official national correspondent who is
and the quality of the ecological functions they provide. Primary
charged with providing the official national statistics for their
forests are globally irreplaceable with unique qualities that make
country. For FRA 2015, in addition to the national reports, ancillary
significant contributions to biodiversity conservation, climate
data such as gross domestic product (GDP), per capita income level,
change mitigation, and sustainable livelihoods (Foley et al., 2007;
and population were incorporated using information from World
Gibson et al., 2011), and with particular importance in tropical
Bank (World Bank, 2013). For more detail regarding the methods,
areas. Barlow et al. (2007) found that 25% of species in
classifications and definitions used in FRA 2015 please refer to
Amazonian Brazil were unique to primary forests, and almost 60
MacDicken (2015).
percent of tree and liana genera were recorded only in primary for-
In this paper, whenever trends and changes in areas are
ests. In North America, primary forest (sometimes referred to as
described, the figures include only countries and territories that
‘‘old growth’’ or ‘‘mature forest’’) is characterized by high struc-
reported in all years, so as not to bias trends by including a chang-
tural heterogeneity as well as biodiversity. Biodiversity in North
ing set of countries. The numbers thus presented may, therefore,
American primary forests often includes a wide range of lichens,
not match the corresponding area reported in a given year.
fungi, insects, bats, spiders, and other organisms found only in
For data analysis, descriptive statistics per country, sub-region
structurally complex mature forests (Spies, 2003).
or climatic domain were used. In addition, multiple regression
Given the global recognition of the importance of biodiversity,
analysis was used to investigate the relationship between socioe-
the link between biodiversity and forests, the need for protected
conomic variables such as population density, per-capita GDP,
areas to preserve biodiversity, and the contribution of primary for-
and the impacts of conservation efforts on forest cover.
ests (or structurally complex mature forests) to biodiversity within
A descriptive analysis of data reliability was performance using
countries, tracking those resources through time is integral to
the tier system introduced by MacDicken (2015), in order to have a
maintaining target goals. To this end, the Food and Agriculture
better understanding of the information provided.
Organization of the United Nations (FAO) includes in the FRA
instrument questions requesting countries to report forest area
within formally established protected areas independent of the 2.2. Examining the effects of protected area and biodiversity
purpose for which the protected areas were established; forest conservation area on forest area
area designated primarily for conservation of biological diversity,
including but not limited to areas within protected forests; and While primary forest area is assumed to relate to biodiversity at
‘primary forest’ – naturally regenerated forests of native species, the global scale, protected area and biodiversity conservation area
where there are no clearly visible indications of human activities can be thought of as measures of conservation effort. Below we
and the ecological processes are not significantly disturbed. report the results of multiple regression analysis directed at exam-
Countries are asked to review their national definitions for the rel- ining the impacts of these two measures of conservation effort as
evant metrics in light of the definitions specified in the FRA, and to well as other variables (population density and per capita GDP)
70 D. Morales-Hidalgo et al. / Forest Ecology and Management 352 (2015) 68–77
Fig. 1. Extent of country reporting to the FRA by year on total forest area, forest area in protected areas or designated for biodiversity conservation and primary forest area.
Left panel: Percent of 234 countries or territories with data on the relevant variables. Right Panel: Percent of global forest area in countries or territories with data.
Central America
Oceania
North Africa
East Asia
Caribbean
North America
TOTAL
Sub-regions
Fig. 3. Change in forest area in protected areas by sub-region and over time according country data reported to FRA 2015.
Table 2
Top 15 countries reporting the largest amounts of forest included in protected areas (ordered by area in year 2015).
Table 3
Top 15 countries reporting area for conservation of biodiversity to the FRA report
2015 by year (ordered by year 2015).
Table 4
The 15 countries reporting largest primary forest area (in 1000 ha) to FRA 2015 (representing 90% of the global primary forest area reported in FRA 2015).
36%
1,800,000 Primary forest area
1,400,000
Area in 1000 ha
1,200,000 31%
1,000,000 31%
800,000
600,000
15%
400,000
200,000
0
Sub Tropical Boreal Template Tropical
Fig. 5. Trends in primary forest area by climatic domain (Countries that did not
report in all years are excluded from the domain totals in this figure). The
Climatic domain
consistently reporting countries included in the domain totals accounted for about
88–89% of the global forest area in each year. Fig. 6. Total and primary forest area and percentage of primary forest by climatic
domain reported for all countries in year 2015 to FRA.
Fig. 6 shows the total and the primary forest area per climatic Table 5
The 15 countries reporting to FRA 2015 with the greatest loss of primary forest area
domain in the year 2015. The tropical domain had the highest total
between 1990 and 2015 (area ‘000 ha).
area and with the highest percentage of primary forest area, fol-
lowed by the template domain. The subtropic domain had the low- Country Primary % of the change % of Global
forest area at country level primary forest
est percentage of primary forest area (15%).
Change (1990 baseline area (1990
Table 5 shows the countries that reported the largest losses of 1990–2015 year) baseline year)
primary forest area over the 25-year reporting period. This table
Brazil 15,549 7.1 1.3
excludes Indonesia because 1990 data were not reported for that Papua New Guinea 13,730 43.8 1.1
country. Indonesia reported a loss of primary forest of about Gabon 8130 38.8 0.7
3.4 million hectares in the period 2000–2015, the fourth-largest Mexico 6387 16.2 0.5
after Papua New Guinea, Brazil and Gabon. There was considerable Bolivia, Plurinational 4640 11.4 0.4
State of
variability in the rate of loss of primary forest reported across Peru 3842 5.5 0.3
countries, with Nigeria reporting that nearly all its primary forest Guyana 3000 31.7 0.2
was lost. Congo, the Democratic 2503 2.4 0.2
Eighty-two countries also reported zero primary forest area in Republic of the
Ecuador 2119 14.5 0.2
all reporting years. Most of these countries and territories have
Central African 1912 49.0 0.2
only small forest areas, but some have substantial forest areas. Republic
The latter include several European countries – for example Guatemala 1617 54.8 0.1
Spain (16 million ha), Germany (11 million ha), Greece, and the Nigeria 1536 98.7 0.1
UK. There were also several African countries with forest area in Suriname 967 6.5 0.1
Malawi 882 51.1 0.1
the millions of hectares that reported zero primary forest in all
Canada 444 58.0 0.0
years. These included Angola with 59 million ha of forest (averaged
over all years), Mozambique (40 million ha), Tanzania (50 mil-
lion ha), Zambia (50 million ha), and several others. In most of least reliable level according with the Tier definitions for FRA
these countries primary forest data were reported as Tier 1, the 2015 (MacDicken, 2015).
74 D. Morales-Hidalgo et al. / Forest Ecology and Management 352 (2015) 68–77
We now examine to what extent an increase in protected and Model 1 Model 2 Model 3
biodiversity conservation area are associated with an increase in Dependent variable: Log(forest area)
forest area, all else remaining the same. Log(Population density) 0.180*** 0.203** 0.118
The first column of Tables 6 and 7 show the estimated slope (0.0557) (0.0939) (0.0920)
coefficients for Model 1 (described in Section 2.2 above) using Log(real GDP per capita) 0.0792** 0.0669* 0.0666*
(0.0312) (0.0377) (0.0368)
Protected Area in Table 6 and Biodiversity Conservation Area in Log(Protected Area) 0.0372** 0.0338* 0.0240*
Table 7.3 Since the variables are in logs, a slope coefficient is an elas- (0.0173) (0.0175) (0.0136)
ticity – the percentage change in forest area induced by a 1% change Boreal domain trend (5-year unit) 0.0333
in an explanatory variable. We see from Table 6, Model 1, that if pop- (0.0354)
Subtropical trend (5-year unit) 0.0219*
ulation density were to increase by 1% in a country and the other
(0.0113)
variables (GDP and Protected Area) were held constant, then the Temperate trend (5-year unit) 0.00401
model implies a 0.18% decrease in forest area in that country. The (0.00732)
effect of an increase in GDP on forest area is positive – a 1% increase Tropical trend (5-year unit) 0.0111
in GDP is associated with a 0.08% increase in forest area. The effect of (0.0109)
Global trend (Unit = 5 years) 0.00427
an increase in Protected Area is also positive, but quite small, a 1%
(0.00973)
increase in Protected Area is associated with a 0.037% increase in for- Constant 6.562***
6.751*** 6.875***
est area. In Table 7, Model 1, in which Protected Area is replaced by (0.482) (0.561) (0.570)
Biodiversity Conservation Area, the results are quite similar. Observations 586 586 586
R-squared 0.114 0.116 0.174
The results reported in the columns headed ‘Model 2’ in Tables
Number of countries 137 137 137
6 and 7 pertain to de-trended as well as de-meaned data. The coef-
ficients on population density, GDP per capita, and Protected Area Note: All models include country fixed effects. Reported coefficients are the per-
centage change in forest area associated with a 1% change in population density, per
and Biodiversity Conservation Area are not very different from
capita GDP, and protected area, and with a five-year change in the others. Robust
those in Model 1. standard errors, clustered by country, in parentheses.
As noted earlier, Model 3 eliminates the possibility that differ- ***
p < 0.01.
ent trends in each domain – tropical, subtropical, temperate and **
p < 0.05.
*
boreal – rather than simply a common global trend, could bias p < 0.1.
the slope coefficients. Once again, the elasticity of forest area with
respect to the different variables are quite similar to those in the
Table 7
previous two models. Only in Model 3 of Table 6 does the coeffi- Effects of biodiversity conservation area on forest area.
cient of population density become distinctly smaller and not sta-
tistically significant at even the 10% level. However, the estimated Model 1 Model 2 Model 3
effect of population density on forest area remains quite similar Dependent Variable: Log(Forest Area)
and statistically significant even in Model 3 in Table 7 where Log(Population density) 0.242*** 0.256** 0.223*
(0.0593) (0.0996) (0.122)
Biodiversity Conservation Area is used instead of Protected Area. Log(real GDP per capita) 0.0817** 0.0751** 0.0740**
When these models are run after excluding boreal and temper- (0.0329) (0.0339) (0.0336)
ate countries, very similar coefficients are obtained (not reported Log(Biodiversity Conservation area) 0.0292** 0.0269* 0.0232*
to save space), although the smaller sample results in their being (0.0119) (0.0144) (0.0126)
Boreal domain trend (5-year unit) 0.00317
less statistically significant for the most part.
(0.00547)
This evidence is suggestive of a positive effect of an increase in Subtropical trend (5-year unit) 0.0182
protection on total forest area, although the effect is not very large (0.0154)
– a slope coefficient of 0.03 implies that a doubling of protected Temperate trend (5-year unit) 0.00472
area leads to an increase in forest area of about 2%. (0.00650)
Tropical trend (5-year unit) 0.00442
A similar regression analysis was attempted to examine the (0.0138)
effects of these variables on primary forest area using 99 countries Global trend (Unit = 5 years) 0.00252
for which data on all variables are available. However, variation in (0.00944)
primary forest area over time is very limited – 82 countries Constant 6.490*** 6.592*** 6.656***
(0.528) (0.528) (0.532)
reported consistently zero primary forest area. Many countries also
Observations 664 664 664
reported no or small changes in primary forest area. R-squared 0.171 0.172 0.192
Unsurprisingly, then, the results of the econometric analysis were Number of countries 146 146 146
inconclusive. The estimated effects of the independent variables
Note: All models include country fixed effects. Reported coefficients are the per-
on primary forest area are not statistically different from zero. centage change in forest area associated with a 1% change in population density, per
Therefore, the tables of results were not included. capita GDP, and biodiversity conservation area, and with a five-year change in the
others. Robust standard errors, clustered by country, in parentheses.
***
p < 0.01.
3.5. Data reliability **
p < 0.05.
*
p < 0.1.
The FRA uses a tier rating system for countries to self-report
data quality, ranging from tier 3 (data collected less than 10 years
ago using National Forest Inventory or remote sensing with ground tier 1, which is designated only as ‘‘other’’ and could represent a
truthing (highest quality data) to tier 2 (full cover mapping/remote respected professional opinion (MacDicken, 2015). Tiers are
sensing or National Forest Inventory >10 years old) and finally to reported for both current status and for trends. Fig. 7 shows the
percentage of area by tiers and categories.
3 While 45% of countries reported using the lowest tier for forest
Some countries are missing in Tables 6 and 7 because they showed no variation in
Protected Area or Biodiversity Conservation Area between FRA years, so they cannot area, those countries represent only 11% of total forest area. Only
be used in the estimation that uses only within-country variation. 28% reported using tier 3 data, but those countries represent 59%
D. Morales-Hidalgo et al. / Forest Ecology and Management 352 (2015) 68–77 75
Conservation of biodiversity
Protected forest
Primary forest
Forest
Conservation of
Forest Primary forest Protected forest
biodiversity
Tier 3 59% 33% 34% 53%
Tier 2 30% 11% 53% 24%
Tier 1 11% 57% 13% 23%
Fig. 7. Percentage of countries reporting by tier and category and percentage of area in each category by tier.
of forest area. Fifty-five percent of countries representing 57% of definition as it applied to their specific forests, and/or how coun-
forest area reported using tier 1 data for primary forest, while only tries arrived at their estimates. For example, some increases in pri-
33% of area was represented by tier 3 data, suggesting that primary mary forest area over the years analyzed in this paper appeared to
forest may be a less accurate variable. The largest percent of pro- result from new definitions of forest area or primary forest; other
tected forest area was represented by tier 2 and 3 data, at almost increases appeared to be due to the use of new technologies
87% of reported area in those two tiers. Finally, 51% of countries (e.g.: high resolution and multispectral imagery, UAV, others)
reported using tier 2 and 3 data for conservation of biodiversity, and approaches for analysis.
which accounted for 77% of area in that category. Continuing loss of primary forests in both the tropics (62 M ha)
and the subtropics (5 M ha) mirrors patterns of total forest area
loss in those biomes. Global primary forest loss of 1 percent per
4. Discussion decade (and 4 times that in the tropics) is potential cause for con-
cern. While ecologists debate the conservation value of ‘‘sec-
The designation of areas specifically meant to protect forests ondary’’ versus ‘‘primary’’ forest, and differing ideas about what
and for the conservation of biodiversity is one of the primary constitutes a primary forest make it difficult to assess the true
mechanisms currently used to help meet biodiversity targets out- impact of these changes, some authors have attempted to assess
lined in such strategic programs as the Aichi Biodiversity Targets. impacts of primary forest loss. Gibson et al. (2011) reviewed 138
Similarly, tracking primary forest area is one method of capturing studies of primary forests to determine impacts of primary forest
change in forests that are assumed to be particularly important loss and/or disturbance on overall biodiversity in the tropics and
ecologically. The global FRA serves as a unique instrument for found that secondary forests have demonstrably lower overall bio-
tracking changes in areas that countries self-report as primary for- diversity and that disturbance effects were ‘‘essentially universal,’’
est, protected forest, or forests set aside for the conservation of leading the authors to conclude that primary forests are irreplace-
biodiversity. able reserves of tropical biodiversity. In many cases, the persis-
Our study showed a loss of 128 M ha of total global forest area tence of threatened and endangered species appears to be
from 1990 to 2015, with the majority of that occurring in the trop- dependent upon the availability of intact primary forest. For exam-
ics. Both FRA and Hansen et al. (2013) show that global forest loss ple, Gregory et al. (2012) found that orangutan population size was
is highest in the tropical domain, an area with high biological bio- heavily influenced by the degradation of primary forests, and esti-
diversity and increasing concern over forest protection. Global for- mated a reduction in orangutan populations of 40–80 percent
est area change and further comparisons with other datasets is without measures toward protection of these forests. There is some
further explored in this special issue by Keenan et al. (2015). hope, as tropical primary forest loss appears to be slowing over
In FRA 2015, primary forest area was defined as: ‘‘naturally time. Additionally, the lack of complete country reporting may
regenerated forest of native species where there are no clearly vis- underestimate the total amount of global primary forest, and
ible indications of human activities and the ecological processes may impact overall global trends.
are not significantly disturbed. Key characteristics include: they Our results were similar to those reported by Mackey et al.
show natural forest dynamics, such as natural tree species compo- (2014) in that almost 98% of primary forest was found within 25
sition, occurrence of dead wood, natural age structure, and natural countries, where half of this percentage belong to developed coun-
regeneration processes; the area is large enough to maintain its tries (Russian Federation, Canada, USA, Australia, Japan, Republic of
natural characteristics; there has been no known significant Korea). We also found that a large proportion of primary forest
human intervention or the last significant human intervention (75%) reported by countries was contained within 7 countries,
was long enough ago to have allowed the natural species composi- including Indonesia. Indonesia alone reported a 3.4 M ha loss of
tion and processes to have become re-established’’. However, there primary forest from 2000 to 2015, which is half the 6 M ha
appeared to be some variability in how countries interpreted the Landsat-derived primary forest loss estimate for Indonesia
76 D. Morales-Hidalgo et al. / Forest Ecology and Management 352 (2015) 68–77
reported by Margono et al. (2014). In this particular case, Indonesia showed that protected areas experienced deforestation due to agri-
reports that their primary forest variable for FRA 2015 was derived cultural expansion, human population growth, and infrastructure
using the highest FRA tier rating, implying that the variable was development, along with economic activities outside of forests,
determined from a national forest inventory sample or remotely and that reforestation was more common in community forests
sensed data coupled with ground truthing. While some of the dis- (Porter-Bolland et al., 2012). Other studies similarly report that
crepancy between the 3.4 M ha reported to FRA and the 6 M ha protected areas may not be successful at preventing deforestation
derived from landsat is certainly methodological or definitional (Andam et al., 2008; Nagendra, 2008; DeFries et al., 2005).
and hinders direct correlation between the two, it’s clear from both Tittensor et al. (2014) project based on current trends that pro-
that Indonesia is losing primary forest at an alarming rate. Kessler tected areas will increase significantly by 2020, meeting terrestrial
et al. (2005) reported distinct declines in species richness of both Aichi Targets (which includes forests). However, they also note
vegetation and wildlife when Indonesian forests were converted that declared protection does not always mean adequate
from primary states to other forested states, suggesting that bio- protection.
logically, even if the loss of primary forest is not a loss of forest Human population pressure has long been understood as one of
cover in general, in Indonesia it represents a distinct loss of many drivers in the conversion of natural habitats to other land
biodiversity. uses. We found that increases in population density within coun-
Protected area establishment and setting forest area aside for tries are negatively associated with increases in forest area while
biological diversity conservation are two ways in which both glo- per capita income shows a positive association. This is the case
bal forest area losses and loss of primary forest can be slowed or globally and also in the subsample of tropical and sub-tropical
reversed and/or important functions and resources within forests countries. This is broadly consistent with DeFries et al. (2010)’s
can be maintained. Interestingly, though primary forest area is finding that urban population density and agricultural exports
declining we found that, in countries that report regularly, forests were correlated with deforestation measured using satellite data
in protected areas have increased. The increase in protected areas between 2000 and 2005 in a sample of tropical countries (although
was highest in the tropics, where – as noted above – primary forest this study uses between, rather than within-country comparisons).
area loss is highest. Inconsistent reporting rates across years Mills Busa (2013) notes that high-income countries import more
severely limit the ability to report on and draw meaningful conclu- wood than do countries with lower GDP, suggesting that ‘‘rich
sions from trends in protected forests and forest set aside for bio- countries practice preservation within borders but appropriate
logical conservation. For example, only 77 percent of the forest resources from poorer countries to sustain consumption’’ leading
area reported as protected in FRA 2015 could be examined in this her to the conclusion that consumption reduction is as important
paper in terms of trends. a strategy as forest protection. Meyfroidt et al. (2010) also note
Forest area set aside to conserve biodiversity also grew in all that countries that experience increased forest cover have often
domains with the highest increases in the tropics. This begs the redistributed their resource use to other countries; i.e., exploita-
question of whether high-productivity primary forests are being tion of resources elsewhere is facilitating reforestation locally.
protected at the same rate as secondary forests. Indeed, Indonesia – the country with the largest primary forest loss
Our analysis found that increases in protected area and in bio- as mentioned earlier – is among the countries absorbing the dis-
diversity conservation area within countries are associated with placed agricultural demand from countries experiencing reforesta-
increases in forest area but the effect is not large. These relations tion (Meyfroidt et al., 2010).
persist in the smaller sample of tropical and sub-tropical countries, Among studies that used only within-country variation, Scrieciu
more strongly in the case of biodiversity conservation area which is (2007) and Culas (2007) failed to find statistically significant
reported by more countries. It is possible that the beneficial effect effects of income and population density on deforestation, but
of protection could to some extent be masked in the data by spurts were limited by having data on fewer countries over a shorter time
of deforestation leading to an increase in protected area. It could span in the late 20th century. In the Democratic Republic of the
also be true that countries that increased protection simultane- Congo, forest loss intensity was associated with areas of high pop-
ously adopted other policies to reduce deforestation, so that the ulation density (Potapov et al., 2012).
estimated effect of protection could also be overstated. Andam Drawing meaningful conclusions from the FRA data requires an
et al. (2008) found that from the 1960s to the 1990s, deforestation assessment of the quality and credibility of data at such a large
in Costa Rica was decreased as a result of protected areas, but that scale. While many countries used tier 1 data in reporting, most
the impact of protected areas on deforestation had previously been of the area represented in each category was covered by tier 2
overestimated (by as much as 65 percent). Wendland et al. (2015) and 3 data, suggesting that data quality comes from fairly reliable
found that protected areas in European Russia had little impact on sources, overall. The exception seems to be primary forest, which
rates of forest disturbance. This may be somewhat unsurprising, may be a reflection of the difficulty in applying the definition of
given that the Millennium Ecosystem Assessment (2005) reported primary forests in each country. Assessment of the impact of tier
that many protected forest areas are specifically situated because 1 data on the overall quality of the results of our study is infeasible
they are unsuitable for other use by humans. Despite the afore- given that there is little or nothing with which to compare the data
mentioned limitations in their effectiveness, the increase in pro- reported for countries using tier 1 information.
tected forest area and forests designated for biodiversity Additional challenges exist in analyzing trends in primary for-
conservation is encouraging, particularly given that the largest est, protected area, and area set aside for conservation of biodiver-
increase in protected areas was in the tropical domain where total sity. First, country response rates are variable, both among years
global forest loss is highest. and among countries, limiting the types of analyses that we were
In the Democratic Republic of the Congo, forest cover loss able to conduct as well as the sensitivity of our analyses. Second,
within protected areas was more than two times lower than the we noted that challenges exist in the harmonization of country
national average, but forest cover loss still increased by 64% data to FRA definitions, suggesting that additional conversations
between 2000–2005 and 2005–2010 in protected areas (Potapov may be necessary amongst national correspondents to FRA as
et al., 2012). In a meta-analysis of published case studies assessing regards some variables, particularly primary forest. The overall
tropical forest cover change, Porter-Bolland et al. (2012) found that findings of our study suggest that, while some progress has been
community managed forests resulted in lower deforestation rates made in the protection of forests at the global scale in that last
in the tropics than the declaration of protected forests. Their study 25 years, and while tropical countries are increasingly protecting
D. Morales-Hidalgo et al. / Forest Ecology and Management 352 (2015) 68–77 77
forests, primary forests of ecological significance in the tropical Chini, Justice, C.O., Townshend, J.R.G., 2013. High-resolution global maps of
21st-century forest cover change. Science 342 (6160), 850–853. http://
domain are still declining. The decline is decreasing, however,
dx.doi.org/10.1126/science.1244693.
which allows for some optimism as global conversations and Joint Liaison Group of the Rio Conventions (JLG), 2008. Forests: CLIMATE CHANGE,
awareness continue. Our findings also indicate that, if FRA is to BIODIVERSITY and Land Degradation. <http://unfccc.int/resource/docs/
be used as one mechanism for tracking progress toward goals like publications/forest_eng.pdf>.
Joppa, L.N., Pfaff, A., 2009. High and far: biases in the location of protected areas.
the AICHI Biodiversity Targets, country correspondents may PLoS ONE 4 (12), e8273. http://dx.doi.org/10.1371/journal.pone.0008273.
require additional assistance toward reporting on primary forest, Keenan, R., Reams, G., Freitas, J., Lindquist, E., Achard, F., Grainger, A., 2015.
protected forest, and biodiversity conservation statistics. Dynamics of global forest area: results from the 2015 Global Forest Resources
Assessment. Forest Ecol. Manage. 352, 9–20.
Kessler, M., Kessler, P.J., Gradstein, S.R., Bach, K., Schmull, M., Pitopang, R., 2005.
Acknowledgements Tree diversity in primary forest and different land use systems in Central
Sulawesi, Indonesia. Biodivers. Conserv. 14 (3), 547–560.
MacDicken, K.G., 2015. Global Forest Resources Assessment 2015: What, why and
The authors gratefully acknowledge Val Kapos for her extensive how? Forest Ecol. Manage. 352, 3–8.
input, Swagatam Sinha for research assistance, and the Food and Mackey, B., DellaSala, D.A., Kormos, C., Lindenmayer, D., Kumpel, N., Zimmerman, B.,
Agriculture Organization of the United Nations for access to the Hugh, S., Young, V., Foley, S., Arsenis, K., Watson, J.E.M., 2014. Policy options for
the World’s Primary Forests in multilateral environmental agreements.
data and input throughout the process. Conserv. Lett. http://dx.doi.org/10.1111/conl.12120.
Margono, B.A., Potapov, P.V., Turubanova, S., Stolle, F., Hansen, M.C., 2014. Primary
forest cover loss in Indonesia over 2000–2012. Nat. Climate Change.
Meyfroidt, P., Rudel, T.K., Lambin, E.F., 2010. Forest transitions, trade, and the global
References displacement of land use. Proc. Natl. Acad. Sci. U.S.A. 107 (2010), 20917–20922.
Millennium Ecosystem Assessment, 2005. Ecosystems and Human Well-being:
Aerts, R., Honnay, O., 2011. Forest restoration, biodiversity and ecosystem Synthesis. Island Press, Washington, DC. <http://www.unep.
functioning. BMC Ecol. 11, 29. org/maweb/documents/document.356.aspx.pdf>.
Andam, K.S., Ferraro, P.J., Pfaff, A., Sanchez-Azofeifa, G.A., Robalino, J.A., 2008. Mills Busa, J.H., 2013. Deforestation beyond borders: addressing the disparity
Measuring the effectiveness of protected area networks in reducing between production and consumption of global resources. Conserv. Lett. 6 (3),
deforestation. Proc. Natl. Acad. Sci. 105 (42), 16089–16094. 192–199.
Barlow, J., Gardner, T.A., Araujo, I.S., Avila-Pires, T.C., Bonaldo, A.B., Costa, J.E., Nagendra, H., 2008. Do parks work? Impact of protected areas on land cover
Esposito, M.C., Ferreira, L.V., Hawes, J., Hernandez, M.I.M., Hoogmoed, M.S., clearing. AMBIO: A J. Human Environ. 37 (5), 330–337.
Leite, R.N., Lo-Man-Hung, N.F., Malcolm, J.R., Martins, M.B., Mestre, L.A.M., Porter-Bolland, L., Ellis, E.A., Guariguata, M.R., Ruiz-Mallén, I., Negrete-Yankelevich,
Miranda-Santos, R., Nunes-Gutjahr, A.L., Overal, W.L., Parry, L., Peters, S.L., S., Reyes-García, V., 2012. Community managed forests and forest protected
Ribeiro-Junior, M.A., da Silva, M.N.F., Silva, Motta C., Peres, C.A., 2007. areas: an assessment of their conservation effectiveness across the tropics. For.
Quantifying the biodiversity value of tropical primary, secondary, and Ecol. Manage. 268, 6–17.
plantation forests. Proc. Natl. Acad. Sci. 104 (2007), 18555–18560. Potapov, P.V., Turubanova, S.A., Hansen, M.C., Adusei, B., Broich, M., Altstatt, A.,
Billor, N., Hadi, A.S., Velleman, P.F., 2000. BACON: Blocked adaptive computationally 2012. Quantifying forest cover loss in Democratic Republic of the Congo, 2000–
efficient outlier nominators. Comput. Stat. Data Anal. 34, 279–298. 2010, with Landsat ETM+ data. Remote Sens. Environ. 122, 106–116.
Culas, R.J., 2007. Deforestation and the environmental Kuznets curver: an Secretariat of the Convention on Biological Diversity (SCBD), 2006. Global
institutional perspective. Ecol. Econ. 61, 429–437. Biodiversity Outlook 2. Montreal, 81 + vii pages. <http://www.cbd.int/doc/
DeFries, R., Hansen, A., Newton, A.C., Hansen, M.C., 2005. Increasing isolation of gbo/gbo2/cbd-gbo2-en.pdf>.
protected areas in tropical forests over the past twenty years. Ecol. Appl. 15 (1), Scrieciu, S.S., 2007. Can economic causes of tropical deforestation be identified at a
19–26. global level? Ecol. Econ. 62, 603–612.
DeFries, R.S., Rudel, T., Uriarte, M., Hansen, M.C., 2010. Deforestation driven by Shvidenko, A., Barber, C.V., Persson, R., Gonzalez, P., Hassan, R., 2005. Forest and
urban population growth and agricultural trade in the twenty-first century. Nat. woodland systems. In: Millennium Ecosystem Assessment (Ed.), Ecosystems
Geosci. 3, 178–181. and Human Well-Being: Current State and Trends. Island Press, Washington,
Food and Agricultural Organization of the United Nations (FAO), 2003. State of the DC, pp. 585–622.
World’s Forests 2003. FAO, Rome, Italy, 151p. Spies, T., 2003. New finding about old-growth forest. PNW Science Update Series.
Food and Agricultural Organization of the United Nations (FAO), 2010. Global Forest U.S. Department of Agriculture Pacific Northwest Research Station. pp 10.
Resources Assessment 2010. Main report. FAO Forestry Paper No. 163, Rome, <http://www.fs.fed.us/pnw/pubs/science-update-4.pdf>.
Italy, 378p. Tittensor, D.P., Walpole, M., Hill, S.L., Boyce, D.G., Britten, G.L., Burgess, N.D.,
Foley, J.A., Asner, G.P., Costa, M.H., Coe, M.T., DeFries, R., Gibbs, H.K., Howard, E., Visconti, P., 2014. A mid-term analysis of progress toward international
Olson, S., Patz, J., Ramankutty, N., Snyder, P., 2007. Amazonia revealed: forest biodiversity targets. Science 346 (6206), 241–244. http://dx.doi.org/
degradation and loss of ecosystem goods and services in the Amazon Basin. 10.1126/science.1257484.
Front. Ecol. Environ. 5, 25–32. Watson, J., Dudley, N., Segan, D., Hockings, M., 2014. The performance and potential
Gibson, L., Lee, T.M., Koh, L.P., Brook, B.W., Gardner, T.A., Barlow, J., Peres, C.A., of protected areas. Nature 515, 67–73. http://dx.doi.org/10.1038/nature13947,
Bradshaw, C.J., Laurance, W.F., Lovejoy, T.E., Sodhi, N.S., 2011. Primary forests 06 November 2014.
are irreplaceable for sustaining tropical biodiversity. Nature 478 (7369), 378– Wendland, K.J., Baumann, M., Lewis, D.J., Sieber, A., Radeloff, V.C., 2015. Protected
381. area effectiveness in European Russia: a postmatching panel data analysis. Land
Gregory, S.D., Brook, B.W., Goossens, B., Ancrenaz, M., Alfred, R., Ambu, L.N., Econ. 91 (1), 149–168.
Fordham, D.A., 2012. Long-term field data and climate-habitat models show World Bank, 2013. Atlas of Global Development: A Visual Guide to the World’s
that orangutan persistence depends on effective forest management and Greatest Challenges, fourth ed. Collins, Glasgow.
greenhouse gas mitigation. PLoS ONE 7 (9), e43846.
Hansen, M.C., Potapov, P.V., Moore, R., Hancher, M., Turubanova, S.A., Tyukavina, A.,
Thau, D., Stehman, S.V., Goetz, S.J., Loveland, T.R., Kommareddy, A., Egorov, A.,
Forest Ecology and Management 352 (2015) 78–88
Global forest area disturbance from fire, insect pests, diseases and severe
weather events q
Pieter van Lierop a,⇑, Erik Lindquist a, Shiroma Sathyapala a, Gianluca Franceschini b
a
Forestry Department, Food and Agriculture Organization, Rome, Italy
b
Natural Resources Department, Food and Agriculture Organization, Rome, Italy
a r t i c l e i n f o a b s t r a c t
Article history: Reliable global data on forest degradation and disturbances due to fire, insect pests, diseases and severe
Received 10 March 2015 weather are important to understand ecosystem health and condition, safeguard production of goods and
Received in revised form 5 June 2015 services and avoid negative impacts on human livelihoods. This paper presents a global analysis of forest
Accepted 8 June 2015
area affected by fire, significant insect pest outbreaks, diseases and severe weather reported by countries
Available online 7 September 2015
as part of the Global Forest Resources Assessment 2015. Between 2003 and 2012, approximately 67 mil-
lion hectares (1.7%) of forest land burned annually, mostly in tropical South America and Africa. In a sim-
Keywords:
ilar reporting period, in total 142 million hectares of forest land were affected by other disturbances than
Forest disturbances
Fire
fire. Insect pests affected more than 85 million hectares of forest, of which a major part was in temperate
Insect pests North America. Severe weather disturbed over 38 million hectares, mostly in Asia. About 12.5 million
Weather events hectares were reported to be disturbed by diseases, mostly in Asia and Europe. There were strong corre-
Partial canopy cover reduction lations between burned forest area and the area of partial canopy cover reduction, as well as between
burned forest area and net forest loss. Partial canopy cover reduction is used as a proxy for forest degra-
dation, although it also includes land under management that is not degraded. A decreasing trend in
burned forest area was found, largely accounted for by decreased area burned within the last ten years
in tropical South America. However, an increasing trend in burned forest area was found in the boreal
climatic domain. The data on other disturbances was not suitable for determining any year on year cor-
relations and should be improved in future data collection exercises.
Ó 2015 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
http://dx.doi.org/10.1016/j.foreco.2015.06.010
0378-1127/Ó 2015 Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
P. van Lierop et al. / Forest Ecology and Management 352 (2015) 78–88 79
To assess the state and trend of forest disturbance globally, the requested specifically for these years because they are by nature
Food and Agriculture Organization of the United Nations (FAO) has episodic. Countries were asked to report annual figures for number
sought to aggregate information on forest disturbances in its peri- of fires and area burned from 2003 to 2012. The area of forest
odic Global Forest Resources Assessment (FRA). FRA 2000 (FAO, affected by insects, diseases and severe weather was reported
2001) revealed a lack of national data sources capable of generat- non-systematically according to the year the disturbance was
ing global estimates of land and forest area burned. FRA 2005 detected. The area of forest with reduced canopy cover was calcu-
(FAO, 2007) provided data on global and regional areas burned lated on an annual basis, but annual figures were summed to pro-
based on 12 regional working papers prepared within the Global duce a single statistic representing the total area affected between
Wildland Fire Network of the United Nations International 2000 and 2010.
Strategy for Disaster Reduction (UNISDR). In FRA 2010, FAO National reporting of burned area and PCCR was supplemented
attempted to quantify the impacts of many factors that affect the through the use of remote-sensing data (see Section 2.2).
health and vitality of forests by collecting data on the impact of fire Remote-sensing based values for total burned area, burned forest
together with insect pests and diseases and other biotic and abiotic area and PCCR were provided to the countries as pre-filled vari-
factors. ables, and country experts were asked to accept, reject or modify
Early attempts to assemble global-level information on forest the supplied area estimates with better quality, nationally derived
insect pests and diseases included several international meetings data where available (MacDicken, 2015). This was done in an
held in the 1960s and 1970s (e.g. FAO, 1965, 1976), but since then attempt to make global estimates of burned area more consistent.
most of the information available at the global level has been pest PCCR was summed for all the years between 2000 and 2012. Thus
specific. As part of FRA 2005, FAO asked countries to report on area there is no time series of estimates to determine changes in rates
affected by insect pests, diseases and other disturbances. This or locations of PCCR.
information was supplemented by a thematic study reviewing for-
est pests in 25 countries (FAO, 2009). In FRA 2010, countries were 2.2. Derivation of pre-filled data for burned area
again asked to report on the impact of insect pests and diseases.
However, most countries were unable to provide reliable quantita- Burned area estimates for FRA 2015 were derived from the
tive information because they did not systematically monitor these Moderate Resolution Imaging Spectroradiometer (MODIS)
disturbances or had limited access to data (FAO, 2011). FRA 2015 Collection 5 Burned Area product (MCD45A1) (Roy et al., 2008) dis-
responded by modifying the data request to indicate significant tributed as part of the MODIS fire products suite (Justice et al.,
outbreaks only. This resulted in more precise reporting of impor- 2002). The MODIS sensor is aboard the AQUA and TERRA satellite
tant disturbance than in previous years. platforms and is capable of imaging nearly the entire Earth’s sur-
This paper updates the state of knowledge on global forest dis- face daily with a pixel size of up to 250 m. The algorithm for map-
turbances from fire, insects, diseases and severe weather. It analy- ping burned areas is based on the spectral, temporal and structural
ses the status and trend of these various disturbance agents and changes that characterize the land surface after a fire occurs (Roy
events as reported by countries for FRA 2015. Total land area et al., 2005). It detects the approximate date of burning at 500 m
burned annually is included, but the analysis and discussion by locating the occurrence of rapid changes over a long time series
focuses, as much as possible, specifically on disturbed forest area. of daily land-surface reflectance observations.
First presented are the aggregated results of the areal extent of for- Estimates of burned forest area were obtained by spatially
est disturbances globally, by broad climatic domain and by region. intersecting the burned area with forest area. Forest areas were
The article describes the trend in burned forest area over time and determined using Collection 4, Version 3 (Hansen et al., 2006) of
compares the amount of forest area burned with net deforestation. the MODIS Vegetation Continuous Fields (VCF) product
Further, a new disturbance-related variable in FRA 2015, called (MOD44B) (Hansen et al., 2003). The VCF product is a globally con-
partial canopy cover reduction (PCCR), is introduced and analyzed sistent depiction of per-pixel percent cover for three types of veg-
as a proxy indicator of forest degradation. PCCR is summarized by etative cover – woody vegetation, herbaceous vegetation and bare
climatic domain and region and related with burned forest area ground – with a pixel size (spatial resolution) of 250 250 m. VCF
and net forest loss for each. data have been produced annually since 2000. Forest was distin-
guished from non-forest by applying a 30% threshold to the contin-
uous VCF values for woody vegetation. Values of less than 30%
2. Data and methods were considered non-forest, and those of 30% or greater were con-
sidered forest. It should be noted that this threshold is different
2.1. The Global Forest Resources Assessment than the 10% used for FRA 2015 and that the MODIS definition of
woody vegetation includes areas that are not forest using the
The results presented in this paper are based on data from FRA FRA forest definition.
2015. FRA relies on responses to a standardized questionnaire sub-
mitted by countries through a network of official national corre- 2.3. Derivation of pre-filled data for partial canopy cover reduction
spondents. For the history and details of the FRA reporting
process see MacDicken (2015). National correspondents were Partial canopy cover reduction (PCCR) was defined as a detect-
asked to respond to questions in the following six categories relat- able modification of canopy cover, at the 250 250 m MODIS pixel
ing to the type, number and areal extent of forest disturbances in size, that resulted in a partial loss of tree cover relative to a prede-
their countries: number of fires per year; total land and forest area termined reference time period, in this case the year 2000. Partial
burned per year; area affected by insect outbreaks; area affected by loss was determined on the basis of a change in per-pixel percent
disease outbreaks; area affected by severe weather events; and tree canopy cover as estimated from the annual VCF time series.
area of forest with reduced canopy cover. Where possible, coun- Pixels with PCCR between 2000 and 2012 were identified using
tries were asked to report on specific types of pests, diseases and three main criteria: an initial VCF value greater than 30%; an over-
weather disturbances as well as on the species or forest type all decrease in percent canopy cover greater than 20% between
affected. 2000 and 2012; and a negative slope (> 1) of the line formed by
The target reporting years for FRA 2015 were 1990, 2000, 2005, the linear regression of the VCF percent tree cover over the period
2010 and 2015. Disturbance variables, however, were not from 2000 to 2012. Pixels where VCF values dropped and stayed
80 P. van Lierop et al. / Forest Ecology and Management 352 (2015) 78–88
Table 1 Absolute area calculations with MODIS pixels are prone to sys-
Land and forest areas burned globally, 2003–2012. tematic over- or underestimation, because the MODIS pixel size of
Year Burned % of Burned Burned Total % of 6.25 ha is a limitation to detecting truly fine-scale change. Much
land land forest forest as % forest total change due to PCCR will take place at a scale not detectable by
(000 ha) burned (000 ha) of burned (000 ha) forest the MODIS pixel unless the change is of a large, diffuse nature. A
land burned
single MODIS pixel flagged as PCCR can contain either large, diffuse
2003 355 300 2.8 68 592 19.3 4 042 531 1.7 partial canopy cover loss or areas of complete tree cover removal
2004 384 798 3.0 75 636 19.7 4 038 801 1.9
within a matrix of dense tree cover (e.g. smallholder clearing for
2005 369 817 2.9 78 497 21.2 4 035 072 1.9
2006 337 990 2.6 64 250 19.0 4 031 342 1.6 agriculture in an otherwise forested pixel). These clearings, at a
2007 381 897 3.0 84 965 22.2 4 027 613 2.1 finer scale of analysis, may meet the definition of deforestation.
2008 322 705 2.5 56 820 17.6 4 023 883 1.4 Researchers are currently exploring the potential of high spatial
2009 308 935 2.4 53 728 17.4 4 020 154 1.3
resolution satellite data for measuring and monitoring forest
2010 325 229 2.5 65 225 20.1 4 016 424 1.6
2011 322 228 2.5 55 479 17.2 4 012 695 1.4 degradation, but this effort is largely in its early phases (Gascon
2012 297 137 2.3 63 141 21.2 4 008 965 1.6 and Eva, 2014; GFOI, 2013).
Average 340 604 66 633
Note: Total land area for all years: 12 908 461 (000 ha). It should be noted that in 2.4. Analysis of variables
the case of repeated fires on the same area, it has been accounted for each time.
The areal extent of forest disturbances was summarized glob-
ally and by smaller geographic region including climatic domain,
below the threshold that defines forest (e.g. 10%) were considered geographic region and income category. Climatic domains were
deforestation and removed from the analysis. Pixels within the delineated into four global ecological zones: boreal, temperate,
bounds of an intact forest landscape (Potapov et al., 2008) or subtropical and tropical (FAO, 2012; MacDicken, 2015). If a coun-
wetland (Carroll et al., 2009) were also eliminated from the try’s national boundaries, as defined by the Global Administrative
analysis. Unit Layers (GAUL) (FAO, 2008), encompassed more than one cli-
matic domain, the country was assigned to the domain occupied
by the largest part of the country’s forest area. Countries were also
2.3.1. Limitations to the method for calculating PCCR divided into five regions defined in the FRA process and into low,
Forest degradation is a term that suffers from many disparate medium–low, medium–high and high-income categories accord-
definitions that are often stakeholder dependent (FAO, 2011). ing to the World Bank income-group classification. Income cate-
Finding a single indicator that serves each of these definitions gories are defined by gross national income per capita per year:
requires a generic approach. FRA 2015 chose a proxy variable of low (USD 1045 or less), lower middle (USD 1046–4125), upper
partial canopy cover reduction (PCCR) to estimate such a proxy. middle (USD 4126–12 745) and high (USD 12 746 or more)
Detecting PCCR from remotely sensed data, especially with the (MacDicken, 2015).
most commonly used forms of medium spatial resolution data, is
difficult because forest canopy change almost always takes place
at sub-pixel resolution scale when MODIS-sized pixels are used 2.5. Missing data
in the analysis. That is to say that the nature of canopy cover
change affects areas smaller than the detection capability of the A few very large forested countries (Australia, Brazil and India)
remotely sensed pixel. The methods described here can best be did not provide annual estimates of burned area for all years.
considered a ‘‘hot-spot’’ approach to the detection of partial Without these countries included, the burned area analysis would
canopy cover removal. be incomplete for the purposes of this paper. Accordingly, data for
the missing years were imputed from the available values in the
annual records of these countries. The area of burned forest for
Burned Land and Forest, 2003 - 2012
missing years was calculated using the average percentage forest
450 000
burned obtained from the data reported by the country.
400 000 Brazil reported burned area directly for 2005–2010. The burned
area reported by Brazil was between 3 and 16 times greater than
350 000
the MODIS burned area product. The reason for this discrepancy
300 000 is not known; it may be attributable to a large burned area under
y = -8,065.18x + 16,531,443.93
R² = 0.63 dense tree canopy which may go undetected by the MODIS burned
250 000 area algorithm (Giglio et al., 2006). The reported burned area, how-
Area
200 000 ever, exhibited a high correlation (R2 = 0.71) with the number of
fires detected by the MODIS sensor and published by the
150 000 y = -1,916.54x + 3,914,086.13 Instituto Nacional de Pesquisas Espaciais (INPE, 2015). The burned
R² = 0.31 area for the years that were not reported directly were imputed by
100 000
using the equation formed by the linear regression of number of
50 000 fires and burned area for the years directly reported.
Australia reported burned area annually for 2006–2011. The
00 000
2003 2005 2007 2009 2011 reported burned area exhibited a very high correlation (R2 = 0.99)
Year with the MODIS-derived pre-filled values for those years. Thus,
Total Burned Land (000 ha) the FRA pre-filled values were used for the years not directly
reported by Australia.
Burned Forest (000 ha) India supplied no burned area values for any years in its report
Fig. 1. Total area of burned land and burned forest from 2003 to 2012. The linear
to FRA 2015. Values for India’s annual burned land and forest area
trend over time is shown as a solid black line. The equation describing the line and were taken directly from the FRA pre-filled values as these were
the goodness of fit (R2) are shown. the best available information.
P. van Lierop et al. / Forest Ecology and Management 352 (2015) 78–88 81
Table 2
Forest area, burned forest and proportion burned forest by climatic domain, 2003–2012.
Burned Forest Area by Climatic Domain, 2003-2012 countries, representing >90% of the total land area and >99% of
80 000 the world’s total forest area, reported burned area.
Only 75 countries reported on the areal extent of disturbances
70 000
from insect pests, diseases or severe weather. In total, these 75
Area ('000 hectares)
60 000 countries represented 70% of the total global forest area. PCCR val-
50 000 ues of >0 were found in 133 countries and territories, of which 29%
were desk studies and were not subject to review by national
40 000 authorities.
30 000
y = -1,768.33x + 3,603,037.18
20 000 R² = 0.22 3.2. Land area and forest area burned
10 000
3.2.1. Land area and forest area burned globally
00 000 The results over the ten-year period from 2003 to 2012 indicate
2003 2005 2007 2009 2011 2013 that an average of 341 million hectares, or 2.6% of all land area, was
Year burned annually (Table 1). This is in line with the estimate by
Boreal Temperate Chatenoux and Peduzzi (2013) of 327 million to 391 million hec-
Sub Tropical Tropical tares per year from 2000 to 2011 based on SPOT (Satellite Pour
l’Observation de la Terre) Vegetation satellite sensors and
Fig. 2. Annual burned forest area by climatic domain, 2003–2012. The linear trend
MODIS; the 350 million hectares estimated for the year 2000 by
over time is shown for the tropical domain as a solid black line. The equation
describing the line and the goodness of fit (R2) are shown.
the Joint Research Centre of the European Commission, based on
MODIS products (JRC, 2005); and the estimate by Roy et al.
Annual forest area figures used in burned area analyses were (2008) of 280 million to 290 million hectares for the first consecu-
taken from Keenan et al. (2015) for the nominal FRA reporting tive 12 months (2001–2002) of the MODIS global burned area
years. Forest area for years not reported in FRA was imputed lin- product.
early and separately at the global, regional and climatic domain In the same time period, an average of about 67 million hectares
levels. These are the best estimates available for the missing years; or 1.7% of forest land burned each year (Table 1). About 20% of all
however, they will not sum correctly and are subject to rounding. area burned annually was in forests as compared with 80% outside
Significance of the trend in burned forest area over time was forests.
assessed using the Mann–Kendall test for monotonic trend in Total land area burned over the 10 years analyzed in this paper
time-series data. Significance was tested at each level of aggrega- exhibits a variable but decreasing trend (Fig. 1). This result is sim-
tion twice, once with raw data and a second time with values ilar to findings reported by Giglio et al. (2010), who used a combi-
smoothed by applying a moving average. Statistical tests were per- nation of moderate and coarse resolution satellite imagery sources
formed using the statistical software R (R Core Team, 2014) and the to analyze trends and variability in burned area globally from 1996
‘rkt’ package (Marchetto, 2015). to 2009. The amount of forest area burned is also decreasing annu-
The availability of quantitative data for insect pests, diseases ally, but at a lesser rate than total land area.
and severe weather events was limited – although this can indicate
either a lack of data or a lack of significant disturbance. The aggre- 3.2.2. Land area and forest area burned by climatic domain
gated data on insect pests, diseases and severe weather events are By climatic domain, the largest area of land and forest burned
based on latest occurrence so they refer to different reporting years was in the tropics, where more than 290 million hectares of land
and sometimes to multiple years (e.g. intervals). Thus it was only burned annually (between 79% and 91% of total global burned
possible to summarize all incidences, for any year and for any kind area). Of the total area burned in the tropics, over 53 million hec-
of insect pest, diseases or severe weather event. tares were in forest land (Table 2). In the subtropical domain,
nearly 33 million hectares of land burned each year, of which
3. Results approximately 7 million hectares were forest land. The temperate
and boreal domains accounted for nearly 15 million hectares of
3.1. Countries reporting land burned, of which approximately 7 million hectares were for-
est land.
In total, 155 country reports were examined for disturbance The highest proportion of forest area burned was also in the
from fire, pests, disease and severe weather. Of these, over 140 tropics, where almost 3% of forest burned annually during the
82
P. van Lierop et al. / Forest Ecology and Management 352 (2015) 78–88
Table 3
Forest area, burned forest and proportion burned forest by region, 2003–2012.
Year Africa Asia Europe North and Central America Oceania South America
Forest Burned % of Forest Burned % of Forest Burned % of Forest Burned % of Forest Burned % of Forest Burned % of
area forest forest area forest forest area forest forest area forest forest area forest forest area forest forest
(000 ha) (000 ha) burned (000 ha) (000 ha) burned (000 ha) (000 ha) burned (000 ha) (000 ha) burned (000 ha) (000 ha) burned (000 ha) (000 ha) burned
2003 660 827 18 658 3 574 207 1093 0.2 1 004 469 3039 0.6 748 586 3660 0.5 186 428 4177 2 878 054 38 190 4
2004 657 723 16 321 2 576 025 1342 0.2 1 005 448 1251 0.1 748 758 6365 0.9 186 091 8270 4 874 796 42 120 5
2005 654 679 15 449 2 580 868 2813 0.5 1 004 147 1727 0.2 747 953 5450 0.7 176 485 3436 2 868 611 49 785 6
2006 651 515 18 383 3 579 660 1279 0.2 1 007 407 4207 0.4 749 102 6038 0.8 185 418 8472 5 868 280 25 794 3
2007 648 411 21 205 3 581 478 2216 0.4 1 008 386 2080 0.2 749 274 5013 0.7 185 081 6663 4 865 022 47 758 6
2008 645 307 17 186 3 583 296 2270 0.4 1 009 365 2595 0.3 749 446 3415 0.5 184 745 5826 3 861 764 25 484 3
2009 642 203 14 503 2 585 114 1078 0.2 1 010 345 3975 0.4 749 618 2970 0.4 184 408 10 236 6 858 506 20 921 2
2010 638 282 18 986 3 589 406 1377 0.2 1 013 572 2614 0.3 750 278 3892 0.5 172 002 2557 1 852 133 35 815 4
2011 635 995 14 929 2 588 749 0502 0.1 1 012 303 1710 0.2 749 963 4324 0.6 183 735 8494 5 851 990 27 243 3
2012 632 891 14 290 2 590 567 0612 0.1 1 013 283 2659 0.3 750 135 3833 0.5 183 398 7679 4 848 732 35 848 4
P. van Lierop et al. / Forest Ecology and Management 352 (2015) 78–88 83
Burned Forest Area by Region, 2003-2012 3.2.4. Land and forest area burned by income category
60 000 The largest area of land and forest burned annually was in coun-
tries in the upper-middle income category, 123 million hectares
y = -1,405.80x + 2,857,045.25 and 40 million hectares, respectively (Table 4). The low-income
50 000 R² = 0.19 countries had the next largest area of land burned annually, or
Area ('000 hectares)
Table 4
Forest area, burned forest and proportion of forest burned by income category, 2003–2012.
Burned forest area by income category, 2003 - 2012 3.3.2. Total forest area affected by other disturbances by region
60 000 Insect pests were the leading cause of forest disturbance other
y = -1,429.83x + 2,909,942.44 than fire in nearly all geographic regions. In North America, they
50 000 R² = 0.21 are the leading cause of forest disturbance. North America reported
Area ('000 hectares)
Fig. 4. Annual burned forest area by income category, 2003–2012. The linear trend
over time is shown for upper-middle income category as a solid black line. The 3.4. Partial canopy cover reduction
equation describing the line and the goodness of fit (R2) are shown.
3.4.1. Total forest area exhibiting PCCR by climatic domain
The tropical domain was the climatic domain with the most
3.3. Forest area affected by other disturbances
PCCR detected, over 156 million hectares or 9% of the domain’s for-
est area as reported in 2010 (Table 8), as compared with 1–2% in
In the 75 countries that reported on insect pests, diseases and
the other domains.
severe weather events, the total forest area affected by these dis-
The study found a strong linear correlation between PCCR and
turbances, over all years reported, was 141.6 million hectares.
net deforestation by climatic zone (r2 = 0.92).
This represents 5% of the total forest area in these countries
(2807 million hectares). Insect pests accounted for the greatest
part of the area damaged, followed by severe weather events 3.4.2. PCCR by region
(Table 6). However, many factors limit the ability to report on for- Asia was the region with the highest amount of indicated PCCR,
est damaged by disease, including the nature of disease cycles, spa- almost 54 million hectares. Africa was next, followed by South
tial distribution of pathogens and complexities in assessment of America (Table 9).
disease in standing trees (Burdon, 1987; Sturrock et al., 2011). As The study found a strong linear correlation between PCCR and
a result, reported disease damage may be underestimated. net deforestation by region (excluding Asia) (r2 = 0.83).
Table 5
Significance of trend (columns 1–4) and smoothed trend (columns 5–10) in total land area and forest area burned between 2003 and 2012 by climatic domain, region and income
group. The sign (+/ ) indicates the direction of the slope of the line formed by burned area over time. Significance (p < 0.05) of the trend as determined from the Mann–Kendall
test is indicated as (n.s.) if no significance and (⁄) if significant. Column ‘Avg’ indicates the number of years included in the moving average to smooth the trend.
Aggregation Trend burned p < 0.05 Trend burned p < 0.05 Trend burned area, Avg p < 0.05 Trend burned forest area, Avg p < 0.05
area forest area smoothed smoothed
Climatic domain
⁄ ⁄ ⁄
Tropical n.s. 3 3
Subtropical + n.s. + n.s. + 3 n.s. + 3 n.s.
⁄ ⁄
Temperate n.s. n.s. 3 4
⁄
Boreal + n.s. + n.s. + 3 n.s. + 4
Region
⁄ ⁄
Africa n.s. 3 3 n.s.
⁄
Asia n.s. n.s. 3 3 n.s.
Europe + n.s. + n.s. + 3 n.s. + 3 n.s.
North and Central n.s. n.s. 3 n.s. 3 n.s.
America
Oceania + n.s. + n.s. + 3 n.s. + 3 n.s.
⁄ ⁄
South America n.s. n.s. 3 3
Income group
High + n.s. n.s. + 3 n.s. 3 n.s.
⁄ ⁄
Upper middle n.s. n.s. 3 3
⁄ ⁄
Lower middle n.s. 3 3 n.s.
⁄ ⁄
Low n.s. 3 3 n.s.
P. van Lierop et al. / Forest Ecology and Management 352 (2015) 78–88 85
Table 7
Forest area affected by other disturbances by region as reported to FRA 2015.
Region Forest areaa (000 ha) Forest area disturbed (000 ha) % of forest area disturbed
Insect pests Diseases Weather Unknown Total
Africa 104 392 9518 1042 2563 5 13 128 12.6
Asia 399 796 7406 5251 17 814 307 30 778 7.7
Europe 941 992 10 449 4681 4963 4774 24 867 2.6
North and Central America 738 287 57 024 1260 13 087 61 71 432 9.7
Oceania 10 158 2 196 6 0 204 2
South America 612 358 1120 24 0 12 1156 0.2
Total 2 806 983 85 519 12 454 38 433 5159 141 564 5.0
a
Aggregated 2010 forest area of the countries that reported on these disturbances to FRA 2015.
86 P. van Lierop et al. / Forest Ecology and Management 352 (2015) 78–88
Resources Assessment. The authors would also like to thank Gascon, L., Eva, H., 2014. Field Guide for Forest Mapping with High Resolution Data:
Monitoring Deforestation and Forest Degradation in the Context of the UN-
Andrea Perlis for providing proofreading and editorial suggestions,
REDD Programme Tanzania REDD+ Initiative. JRC Technical Report. Publications
Orjan Jonsson for his continues support with data analysis and Office of the European Union, Luxembourg.
Leticia Piña for her valuable suggestions. The authors also wish GFOI, 2013. Integrating Remote-sensing and Ground-based Observations for
to thank the very valuable suggestions of two reviewers whose Estimation of Emissions and Removals of Greenhouse Gases in Forests:
Methods and Guidance from the Global Forest Observation Initiative. Group
comments greatly improved the manuscript. In addition, authors on Earth Observations, Geneva, Switzerland.
are also very grateful to Kenneth MacDicken for his constant guid- Gibbs, H.K., Ruesch, A.S., Achard, F., Clayton, M.K., Holmgren, P., Ramankutty, N.,
ance and useful inputs to improve and finalize this paper. Foley, J.A., 2010. Tropical forests were the primary sources of new agricultural
land in the 1980s and 1990s. Proc. Natl. Acad. Sci. U.S.A., 1–6
Giglio, L., van der Werf, G.R., Randerson, J.T., Collatz, G.J., Kasibhatla, P., 2006. Global
References estimation of burned area using MODIS active fire observations. Atmos. Chem.
Phys. 1, 957–974.
Giglio, L., Randerson, J.T., van der Werf, G.R., Kasibhatla, P.S., Collatz, G.J., Morton,
Alencar, A., Asner, G.P., Knapp, D., Zarin, D., 2011. Temporal variability of forest fires
D.C., Defries, R.S., 2010. Assessing variability and long-term trends in burned
in eastern Amazonia. Ecol. Appl. 21, 2397–2412.
area by merging multiple satellite fire products. Biogeosciences 7, 1171–1186.
Avissar, R., Werth, D., 2005. Global hydroclimatological teleconnections resulting
Gregoire, J.M., Eva, H.D., Belward, A.S., Palumbo, I., Simonetti, D., Brink, A., 2012.
from tropical deforestation. J. Hydrometeorol. 6 (2), 134–145.
Effect of land-cover change in Africa’s burnt areas. Int. J. Wildland Fire 22, 107–
Ayres, M.P., Lombardero, M.J., 2000. Assessing the consequences of global change
120.
for forest disturbance from herbivores and pathogens. Sci. Total Environ. 262,
Hansen, M.C., DeFries, R.S., Townshend, J.R.G., Carroll, M., Dimiceli, C., Sohlberg, R.A.,
263–286.
2003. Global percent tree cover at a spatial resolution of 500 meters: first
Brink, A.B., Eva, H.D., 2009. Monitoring 25 years of land cover change dynamics in
results of the MODIS Vegetation Continuous Fields algorithm. Earth Interact. 7
Africa: a sample based remote sensing approach. Appl. Geogr. 29 (4), 501–512.
(10).
Buma, B., Wessman, C.A., 2011. Disturbance interactions can impact resilience
Hansen, M., DeFries, R., Townshend, J.R., Carroll, M., Dimiceli, C., Sohlberg, R., 2006.
mechanism of forests. Ecosphere. http://dx.doi.org/10.1890/ES11-00038.1
Vegetation Continuous Fields MOD44B, 2001 Percent Tree Cover, Collection 4.
(2:art64).
University of Maryland, College Park, Maryland, USA.
Burdon, J.J., 1987. Diseases and Plant Population Biology. Cambridge University
Hicke, J.A., Johnson, M.C., Hayes, J.L., Preisler, H.K., 2012. Effects of bark beetle-
Press, Cambridge, UK.
caused tree mortality on wildfire. For. Ecol. Manage. 271, 81–90.
Carmenta, R., Parry, L., Blackburn, A., Vermeylen, S., Barlow, J., 2011. Understanding
INPE, 2015. Monitoramento de queimadas e incindios. <www.inpe.br/queimadas/
human-fire interactions in tropical forest regions: a case for interdisciplinary
estatisticas.php> (accessed 18.02.15).
research across the natural and social sciences. Ecol. Soc. 16 (1), 53, <www.
JRC, 2005. SAFARI 200 Global Burned Area Map, 1-km, 2000 Data Set. Joint Research
ecologyandsociety.org/vol16/iss1/art53/>.
Centre, European Commission, Ispra, Italy.
Carroll, M.L., Townshend, J.R., DiMiceli, C.M., Noojipady, P., Sohlberg, R.A., 2009. A
Justice, C., Giglio, L., Korontzi, S., Owens, J., Morisette, J., Roy, D., Descloitres, J.,
new global raster water mask at 250 m resolution. Int. J. Digit. Earth 2 (4), 291–
Alleaume, S., Petitcolin, F., Kaufman, Y., 2002. The MODIS fire products. Rem.
308. http://dx.doi.org/10.1080/17538940902951401.
Sens. Environ. 83, 244–262.
Chambers, J.Q., Fisher, J.I., Zeng, H., Chapman, E.L., Baker, D.B., Hurtt, G.C., 2007.
Kasischke, E., Turetsky, M., 2006. Recent changes in the fire regime across the North
Hurricane Katrina’s carbon footprint. Science 318, 1107.
American boreal region—spatial and temporal patterns of burning across
Chatenoux, B., Peduzzi, P., 2013. Biomass Fires: Preliminary Estimation of
Canada and Alaska. Geophys. Res. Lett. 33, L09703. http://dx.doi.org/10.1029/
Ecosystems Global Economic Losses. Background Paper Prepared for the
2006GL025677.
Global Assessment Report on Disaster Risk Reduction 2013. UNISDR, Geneva,
Keenan, R., Reams, G.A., Achard, F., Freitas, J., Grainger, A., Lindquist, E.J., 2015.
Switzerland.
Dynamics of global forest area: results from the FAO Global Forest Resources
Cochrane, M.A., Barber, C.P., 2009. Climate change, human land use and future fires
Assessment 2015. For. Ecol. Manage. 352, 9–20.
in the Amazon. Glob. Change Biol. 15 (3), 601–612.
Lambin, E.F., Geist, H.J., Lepers, E., 2003. Dynamics of land-use and land-cover
Cochrane, M.A., Laurence, W., 2008. Synergisms among fire, land use and climate
change in tropical regions. Annu. Rev. Environ. Resour. 28 (1), 205–241.
change in the Amazon. AMBIO: J. Hum. Environ. 37 (7), 522–527.
Lewis, S.L., 2006. Tropical forests and the changing earth system. Philos. Trans. R.
Dajoz, R., 2000. Insects and Forests: The Role and Diversity of Insects in the Forest
Soc. Lond. B: Biol. Sci. 361 (1465), 195–210.
Environment. Lavoisier, London, UK.
Logan, J., Régnière, J., Powell, J.A., 2003. Assessing the impacts of global warming on
Dale, V.H., Joyce, L.A., McNulty, S., Neilson, R.P., Ayres, M.P., Flannigan, M.D., Hanson,
forest pest dynamics. Front. Ecol. Environ. 1, 130–137.
P.J., Irland, L.C., Lugo, A.E., Peterson, C.J., Simberloff, D., Swanson, F.J., Stocks, B.J.,
Lund, H. Gyde, 2014. Definitions of Forest, Deforestation, Reforestation and
Wotton, M., 2001. Climate change and forest disturbances: climate change can
Afforestation. Online Report. Forest Information Services, Gainesville, Virginia,
affect forests by altering the frequency, intensity, duration, and timing of fire,
USA. <http://home.comcast.net/%7Egyde/DEFpaper.htm> (accessed 28.06.14).
drought, introduced species, insect and pathogen outbreaks, hurricanes,
MacDicken, K.G., 2015. Global Forest Resources Assessment 2015: what, why and
windstorms, ice storms, or landslides. Bioscience 51 (9), 723–734.
how? For. Ecol. Manage. 352, 3–8.
Dawson, I.K., Leakey, R., Clement, C.R., Weber, J.C., Cornelius, J.P., Roshetko, J.M.,
Marchetto, A., 2015. rkt: Mann–Kendall Test, Seasonal and Regional Kendall Tests. R
Vinceti, B., Kalinganire, A., Tchoundjeu, Z., Masters, E., Jamnadass, R., 2014. The
Package Version 1.4. <http://CRAN.R-project.org/package=rkt>.
management of tree genetic resources and the livelihoods of rural communities
Margono, B.A., Turubanova, S., Zhuravleva, I., Potapov, P., Tyukavina, A., Baccini, A.,
in the tropics: non-timber forest products, smallholder agroforestry practices
Goetz, S., Hansen, M.C., 2012. Mapping and monitoring deforestation and forest
and tree commodity crops. For. Ecol. Manage. 333, 9–21. http://dx.doi.org/
degradation in Sumatra (Indonesia) using Landsat time series data sets from
10.1016/j.foreco.2014.01.021.
1990 to 2010. Environ. Res. Lett. 7 (3), 034010.
FAO, 1965. FAO/IUFRO Symposium on Forest Diseases and Insects. Unasylva, 78
Michaelian, M., Hogg, E.H., Hall, R.J., Arsenault, E., 2011. Massive mortality of aspen
(entire issue).
following severe drought along the southern edge of the Canadian boreal forest.
FAO, 1976. Second FAO/IUFRO World Technical Consultation on Forest Diseases and
Glob. Change Biol. 17, 2084–2094.
Insects, New Delhi (India), 7–12 April 1975. Conclusions and
Morton, D.C., DeFries, R.S., Shimabukuro, Y.E., Anderson, L.O., Arai, E., del Bon
Recommendations. Rome, Italy.
Espirito-Santo, F., Freitas, R., Morisette, J., 2006. Cropland expansion changes
FAO, 2001. Global Forest Fire Assessment 1990–2000. Forest Resources Assessment
deforestation dynamics in the southern Brazilian Amazon. Proc. Natl. Acad. Sci.
Programme Working Paper 5. Rome, Italy.
U.S.A. 103 (39), 14637–14641.
FAO, 2005. Global Forest Resources Assessment Update 2005 – Terms and
Oris, F., Asselin, H., Adam, A., Finsinger, W., Brrgeron, Y., 2013. Effect of increased
Definitions. Forest Resources Assessment Programme Working Paper 83.
fire activity on global warming in the boreal forest. Environ. Rev. 22 (3), 206–
Rome, Italy.
219. http://dx.doi.org/10.1139/er-2013-0062.
FAO, 2007. Fire Management – Global Assessment 2006. FAO Forestry Paper 151.
Pan, Y., Birdsey, R.A., Fang, J., Houghton, R., Kauppi, P.E., Werner, A., et al., 2011. A
Rome, Italy.
large and persistent carbon sink in the world’s forests. Science 14 (July), 1–11.
FAO, 2008. FAO GEONETWORK. Global Administrative Unit Layers (GAUL)
Potapov, P., Yaroshenko, A., Turubanova, S., Dubinin, M., Laestadius, L., Thies, C.,
(GeoLayer). <http://data.fao.org/ref/f7e7adb0-88fd-11da-a88f-000d939bc5d8.
Tsybikova, E., 2008. Mapping the world’s intact forest landscapes by remote
html?version=1.0> (accessed 10.03.15) (latest update 18.02.14).
sensing. Ecol. Soc. 13 (2).
FAO, 2009. Global Review of Forest Pests and Diseases. A Thematic Study Prepared
R Core Team, 2014. R: A Language and Environment for Statistical Computing. R
in the Framework of the Global Forest Resources Assessment 2005. FAO
Foundation for Statistical Computing, Vienna, Austria. <http://www.R-project.
Forestry Paper 156. Rome, Italy.
org/>.
FAO, 2011. Abiotic Disturbances and Their Influence on Forest Health. By B. Moore
Roy, D.P., Jin, Y., Lewis, P.E., Justice, C.O., 2005. Prototyping a global algorithm for
and G. Allard. Forest Health and Biosecurity Working Paper FBS/35E. Rome,
systematic fire-affected area mapping using MODIS time series data. Rem. Sens.
Italy.
Environ. 97, 137–162. http://dx.doi.org/10.1016/j.rse.2005.04.007.
FAO, 2012. Global Ecological Zones for FAO Forest Reporting: 2010 Update. By S.
Roy, D.P., Boscchetti, L., Justice, C.O., Ju, J., 2008. The collection 5 MODIS burned area
Iremonger and A. Gerrand. Forest Resources Assessment Working Paper 179.
product – global evaluation by comparison with the MODIS active fire product.
Rome. <www.fao.org/docrep/017/ap861e/ap861e00.pdf>.
Rem. Sens. Environ. 112, 3690–3707.
Federici, S., Tubiello, F.N., Salvatore, M., Jacobs, H., Schmidhuber, J., MacDicken, K.,
Schowalter, T.D., 2012. Insect responses to major landscape-level disturbance.
2015. New estimates of CO2 forest emissions and removals: 1990–2015. For.
Annu. Rev. Entomol. 57, 1–20.
Ecol. Manage. 352, 89–98.
88 P. van Lierop et al. / Forest Ecology and Management 352 (2015) 78–88
Spracklen, D.V., Arnold, S.R., Taylor, C.M., 2012. Observations of increased tropical Overview Surveys of Forest Health Conducted from 1999 through 2012 and the
rainfall preceded by air passage over forests. Nature 489 (7415), 282–285. BCMPB Model (year10). British Columbia Ministry of Forests, Lands and Natural
Sturrock, R.N., Frankel, S.J., Brown, A.V., Hennon, P.E., Kliejunas, J.T., Lewis, K.J., Resource Operations, Victoria, British Columbia.
Worrall, J.J., Woods, A.J., 2011. Climate change and forest diseases. Plant. Pathol. Yang, G., Tian, H., Tao, B., Ren, W., Kush, J., Liu, Y., Wang, Y., 2014. Spatial and
60 (1), 133–149. temporal patterns of global burned area in response to anthropogenic and
van der Werf, G.R., Morton, D.C., DeFries, R.S., Olivier, J.G.J., Kasibhatla, P.S., Jackson, environmental factors: reconstructing global fire history for the 20th and early
R.B., Collatz, G.J., Randerson, J.T., 2009. CO2 emissions from forest loss. Nat. 21st centuries. J. Geophys. Res.: Biogeosci. 119.
Geosci. 2 (11), 737–738. http://dx.doi.org/10.1038/ngeo671. Zhuravleva, I., Turubanova, S., Potapov, P., Hansen, M., Tyukavina, A., Minnemeyer,
Veblen, T., Hadley, K., Nel, E., Ktzberger, T., Villalba, R., 1994. Disturbance regime S., Laporte, N., Goetz, S., Verbelen, F., Thies, C., 2013. Satellite-based primary
and disturbance interaction in a Rocky Mountain subalpine forest. J. Ecol. 1994 forest degradation assessment in the Democratic Republic of the Congo, 2000–
(82), 125–135. 2010. Environ. Res. Lett. 8 (2), 024034. http://dx.doi.org/10.1088/1748-9326/8/
Walton, A., 2013. Provincial-level Projection of the Current Mountain Pine Beetle 2/024034.
Outbreak: Update of the Infestation Projection Based on the Provincial Aerial
Forest Ecology and Management 352 (2015) 89–98
a r t i c l e i n f o a b s t r a c t
Article history: Using newly available data from the 2015 Forest Resources Assessment (FRA), we refined the informa-
Received 2 January 2015 tion, currently available through the IPCC AR5 and FAOSTAT, on recent trends in global and regional
Received in revised form 16 April 2015 net CO2 emissions and removals from forest land, including from net forest conversion (used as a proxy
Accepted 22 April 2015
for deforestation) and forest remaining forest. The new analysis is based on the simplified forest carbon
Available online 7 September 2015
stock method of the FAOSTAT Emissions database, equivalent to a Tier 1, Approach 1 IPCC methodology,
limited to biomass carbon stocks. Our results indicated that CO2 emissions from net forest conversion
Keywords:
decreased significantly, from an average of 4.0 Gt CO2 yr1 during 2001–2010 to 2.9 Gt CO2 yr1 during
Forest
CO2 emissions
2011–2015. More than half of the estimated reductions over the last five years, some 0.6 Gt CO2 yr1, took
CO2 removals place in Brazil. Detailed analyses further indicated that remaining forests continued to function as a net
Deforestation carbon sink globally, with an average net removal of 2.2 Gt CO2 yr1 during 2001–2010, and
Forest degradation 2.1 Gt CO2 yr1 during 2011–2015. Annex I Parties represented the bulk of this sink, contributing 60%
REDD+ of the total in 2011–2015, down from 65% in 2001–2010. Compared to previous FAOSTAT assessments
Carbon for the period 2001–2010, based on the 2010 FRA and published in the IPCC AR5, the use of FRA 2015
Stock change data led to estimates of net forest conversion that were consistent with previous ones (4.0 vs.
FAOSTAT
3.8 Gt CO2 yr1), while the estimated forest sinks were 22% larger (2.2 vs. 1.8 Gt CO2 yr1). The net
FRA
contribution of forests to anthropogenic forcing based on FRA2015 data was thus smaller than previously
estimated by the IPCC AR5. Finally, we separated for the first time net emissions and removals from forest
land into a sink component and a degradation component. Results indicated that, contrary to CO2 emis-
sions from deforestation, CO2 emissions from forest degradation increased significantly, from
0.4 Gt CO2 yr1 in the 1990s, to 1.1 Gt CO2 yr1 in 2001–2010 and 1.0 Gt CO2 yr1 in 2011–2015.
Emissions from forest degradation were thus one-fourth of those from deforestation in 2001–2010,
increasing to one-third in 2011–2015.
Ó 2015 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction GHG Inventories (IPCC, 2006). Since 2012, FAO has used FRA data
to estimate net CO2 emissions and removals associated with C
The Food and Agriculture Organization of the United Nations stock changes of the biomass C pools, related to forest land and
(FAO) makes available online data on forests, originally submitted net forest conversion, the latter used as a proxy for net deforesta-
by its Member States and analyzed through the Global Forest tion (FAOSTAT, 2015; FAO, 2014). The FAO data, considered equiv-
Resource Assessment (FRA) (MacDicken, 2015). FRA data, available alent to a Tier 1, approach 1 estimate using the carbon stock
over the period 1990–2015, include, among others, estimates of difference method of the IPCC guidelines, were published in the
forest area and of carbon stocks in aboveground and belowground IPCC AR5 (Smith et al., 2014).
biomass carbon pools. The latter data categories are needed in The FRA is the only global database that provides the value and
standard carbon cycle computations of forest dynamics, in line historical trend of C stock changes in forest for each country, using
with methods provided in the 2006 IPCC Guidelines for National official data provided by countries to FAO. An additional repository
of data on forest carbon dynamic is the database of the UN
Framework Convention on Climate Change (UNFCCC), which con-
q
This article is part of a special issue entitled ‘‘Changes in Global Forest Resources tains data submitted by its Parties (http://unfccc.int/national_re-
from 1990 to 2015’’. ports/items/1408.php).
⇑ Corresponding author at: Food and Agriculture Organization of the United
Nations, Via Terme di Caracalla, Rome 00153, Italy.
There are important differences between the FRA and UNFCCC.
E-mail address: francesco.tubiello@fao.org (F.N. Tubiello). First, the UNFCCC database currently does not cover all UN
http://dx.doi.org/10.1016/j.foreco.2015.04.022
0378-1127/Ó 2015 Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
90 S. Federici et al. / Forest Ecology and Management 352 (2015) 89–98
Member States. In coming years, new country submissions of 2010, and 2015, including total national forest area, A, and its three
Biennial Update Reports1 to UNFCCC by non-Annex I Parties2 will subcategories: primary forest, other naturally regenerated forest, and
increase coverage, while adding valuable information on manage- planted forest; as well as total woody biomass carbon stock, B (defined
ment practices, area, carbon stocks and their changes. as above and below-ground biomass)7. FRA data also included some
Second, even though UNFCCC requires its Parties to provide information on carbon stocks in litter, deadwood and soils. Coverage
data on forest land that are consistent with those they submit to by country was rather incomplete however, so that this information
FAO3, often the data provided to FAO are numerically different from was not used herein.
those submitted to UNFCCC, stemming from discrepancies in forest FRA 2015 data presented country data gaps on forest area by
definitions applied4, and from the fact that different national focal subcategory, especially for the first two mentioned above, and on
points are responsible for FAO and UNFCCC submissions. woody biomass. These data gaps were filled as follows. For forest
The FAOSTAT Emissions database is one of three independent area, missing total forest area data were linearly extrapolated or
sources used in the recent IPCC Fifth Assessment Report (AR5) to interpolated using existing FRA data, as needed. When the propor-
estimate anthropogenic GHG emissions and removals from tion of forest area among sub-categories was missing we used
Agriculture, Forestry and Other Land Uses (AFOLU) (Tubiello information on shares by forest subcategory from neighboring
et al., 2015). In particular, for CO2 emissions and removals from countries with similar forest conditions, and applied these shares
forest, FAOSTAT provided estimates of net forest carbon stock to total forest area. For total woody biomass, we used the relevant
change at national, regional and global levels to the IPCC AR5. sub-regional weighted average value8, and multiplied it by the total
Using data from the FRA 2010 (FAO, 2010), and with reference to country forest area.
the period 2001–2010, FAO estimated average decadal CO2 emis- All data were linearly interpolated to obtain a yearly time series
sions from deforestation of 3.8 Gt CO2eq yr1 and a net carbon sink over the period 1990–2015. As described below in more detail,
in forest land of 1.8 Gt CO2eq yr1. Additional GHG emissions annual data were used as input to compute net CO2 emissions
from terrestrial C pools estimated by FAO to the IPCC AR5, but and removals on forest, including forest land area (hereafter
not discussed herein, included emissions from drained peatlands referred to as forest land) and forest area converted to other land
(0.9 Gt CO2eq yr1) and from biomass fires (0.3 Gt CO2eq yr1). uses (hereafter referred to as net forest conversion or deforestation).
This paper provides FAOSTAT updates of CO2 emissions and The terminology used herein is in line with the relevant database
removals from forests, based on FRA 2015 data. Furthermore, this categories of the FAOSTAT Emissions database—Land use
paper introduces an improved methodology to assess separately, (http://faostat3.fao.org/browse/G2/⁄/E). It may be considered con-
within the forest domain, net CO2 emissions from forest degrada- sistent with REDD+, albeit representing an over-simplification of
tion and net CO2 removals from forest re-growth, thus providing current GHG national reporting practice9.
for the first time in the literature, an assessment of both of these
2.1. Derived input data
important terrestrial carbon fluxes.
only for total forest, hence no further stratification between the Table 1
two above forest strata was made. Rather, each was assigned Subdivisions of Forest land categories and forest strata used in the FAOSTAT
methodology.
with the same value b, with limitations discussed below.
Forest Forest land Natural Stable forest area in the inventory
2.2. Stock difference equations forest year (SFA1)
New net forest area in the inventory
year (FAC1)
We estimated annual C stock change, limited to changes in car- Planted Stable forest area in the inventory
bon in woody biomass, and thus net CO2 emissions and removals, forest year (SFA2)
over the period 1990–2015, associated with the following process New net forest area in the inventory
year (FAC2)
types: (i) net forest area loss with associated net C stock loss
(reported under net forest conversion or deforestation); and (ii) net Net forest Natural Net deforestation in the inventory
conversion forest year (FAC3)
forest area gain with associated net C stock gain; and (iii) net
Planted Net deforestation in the inventory
changes in C stock density over the forest land area. The last two forest year (FAC4)
process types were reported together under forest land)10. The
stratification of the forest area and equations applied followed the
IPCC (2006) guidelines and are summarized for convenience in
Table 1 and 2. Forest land area : Ai ðtÞ ¼ SFAi ðtÞ þ Max½FAC i ðtÞ; 0; ð3Þ
In general, net C stock change was computed by applying:
Deforestation area : Min½FAC i ðtÞ; 0 ð4Þ
Approach 1 (IPCC, 2006, Ch. 3, Vol. 4) for land representation. For each of the process type j defined above, the carbon stock
Only net forest area changes were considered, since FRA data change factor CSCF j ðtÞ was computed as follows:
did not allow for quantification of gross area changes. 8
IPCC stock-difference equation (IPCC, 2006; equation 2.5, Vol. 4) < bðtÞ bðt 1Þ; for j ¼ SFA;
>
for the calculation of average C stock density change. CSCF j ðtÞ ¼ bðt 1Þ; for j ¼ FAC < 0 ð5Þ
>
:
bðtÞ; for j ¼ FAC > 0
More specifically, net CO2 emissions and removals11 were com-
puted for each of the two forest strata i (natural forest and planted Note that under Eq. (5), the carbon stock density of deforested land,
forest) each of the three process types j and for each year t, as follows as well as the previous woody biomass carbon stock density of land
(see also Table 2): that was converted to forest, are both assumed to be zero. Similarly,
the woody biomass carbon stock density of new forested land is
44 assumed to be that of the remaining forest. This is a consequence
netCO2 emissioni;j ðtÞ ¼ ai;j ðtÞ CSCF j ðtÞ 103 ð1Þ
12 of the methodology applied herein, leading to either under and over
where: estimates of carbon stock changes reported for natural forest and
netCO2emission is expressed in Gg CO2 (or equivalently, kt CO2); planted forest under net forest conversion and under forest land,
ai,j (t) is the area underpinning the process type j in strata i, at depending on national forest circumstances. Errors however cancel
year t, expressed in ha (see below); out once estimates are summed up for the category forest, since the
CSCF j ðtÞ, or carbon stock change factor, is the woody biomass net C stock change calculated for forest corresponds exactly to the
carbon stock density change relative to process type j, estimated difference in the woody biomass C stocks reported by countries
at time t, expressed in metric tonnes C ha1. The factor 44/12 con- under FRA, and thus are not a source of bias at the level of total for-
verts C to CO2; 103 converts metric tonnes to Gg. The minus sign est area.
allows to assign emissions (positive sign) to carbon stocks losses,
and removals (negative sign) to carbon stocks gains. 2.3. Definition of net emissions from forest degradation
For each of the forest strata i defined above, the area ai;j ðtÞ was
defined as follows: Using Eqs. (1)-(3) and (5), net CO2 emissions and removals esti-
mated over forest land area at country level, netCO2emission ðtÞ cor-
SFAi ðtÞ
ai;j ðtÞ ¼ ð2Þ responded to net annual decreases or increases in forest woody
FAC i ðtÞ biomass C stock density. Irrespective of the underlying processes,
where SFAi ðtÞ is the stable forest area of strata i at time t, defined as which could not be further analyzed using FRA data, whenever
the forest area remaining forest between two successive years (t1) netCO2emission ðtÞ > 0, we counted net forest emissions as forest
and t: degradation. Similarly, whenever netCO2emission ðtÞ < 0, we
counted them as net forest removals. This allowed us to separate
SFAi ðtÞ ¼ Min ½Ai ðtÞ; Ai ðt 1Þ ð2aÞ estimates of net forest sinks at global and regional level into a
And FAC i ðtÞ is forest area change of strata i at time t: degradation component and into a net removal component.
Table 2
Equations applied for estimating the components A and CSCF of Eq. (1).
Subdivision A CSCF
Forest land Natural forest Stable forest area in the inventory year (SFA1) Min[A(t),A(t1)] CSCF = b(t) – b(t1)
New net forest area in the inventory year (FAC1) Max[A(t)–A(t1),0] CSCF = b(t)
Planted forest Stable forest area in the inventory year (SFA2) Min[A(t),A(t1)] CSCF = b(t) – b(t1)
New net forest area in the inventory year (FAC2) Max[A(t)-A(t1),0] CSCF = b(t)
Net forest conversion Natural forest Net deforestation in the inventory year (FAC3) |Min[A(t)-A(t1),0]| CSCF = b(t1)
Planted forest Net deforestation in the inventory year (FAC4) |Min[A(t)-A(t1),0]| CSCF = b(t1)
litter, soil organic carbon, and harvested wood products. This likely C stock change (+13.4%), including for Annex I (+10.2%) and
impacted the magnitude of estimated net C stock change, as the non-Annex I Parties (+20.4%).
inclusion of other C pools would probably have increased the mag- Regarding the quality of country data, FRA 2005 estimated
nitude of estimated changes, but not their direction of change, i.e., uncertainties of growing stock at ±8% for industrialized countries
the identification of a net sink or a net source, since the dynamics (Annex I) and ±30% for non-industrialized countries (non-Annex
of C pools are strictly linked. Soil carbon pools in forest over I). The uncertainty of biomass density for estimating C stock losses
drained organic soils represent a notable exception, since they from net deforestation was larger, and was set by expert judge-
may be characterized by long-term C losses even as carbon bio- ment at ±80%. Uncertainties for wood density data were estimated
mass and litter pools increase. in the range 10–40%. Additionally, FAOSTAT estimates uncertain-
By excluding dead wood, litter and soil organic carbon, ties of area data at ±10% in general, and possibly as low as ±3% in
FAOSTAT estimates of net CO2 emissions and removals were likely industrialized countries. Additional uncertainties were linked to
under-estimates of actual net CO2 emissions. By analysing UNFCCC the use of conversion factors needed to estimate total living bio-
country submissions, we assessed the magnitude of this underesti- mass stocks from nationally reported wood volumes. These factors
mate to be as large as one-third of the potential C stock change include a biomass conversion and expansion factor (BCEF = 1.3), a
from all pools. CO2 Emissions from deforestation are possibly fur- root-to-shoot ratio (R:S = 0.2), and a biomass-to-carbon conversion
ther underestimated, since they were based on net, rather than factor (CF = 0.5). The IPCC associates uncertainties of ±30% to these
gross, deforested area. factors. The resulting overall uncertainty12 at global level, com-
Additional methodological uncertainties were linked to the use puted by applying relevant 2006 IPCC equations for error propaga-
of average C stock densities, i.e., not differentiated by forest strata. tion, was about ±30% for CO2 emissions from deforestation and
This would lead to an over or under-estimation of net emissions ±15% net CO2 emissions and removals from forest land. It should be
and removals for forest strata with C density significantly above noted that FRA applies the conversion factors described above
or below the average values. Because the information on C density equally across all countries. This simplification introduced additional
is however correct when considered over the sum of forest and uncertainties, which we did not know how to quantify.
deforested areas, these errors cancel out at the level of total net
C stock change of forest. 4. Results and discussion
Finally, our estimates included CO2 gas only. Including CH4 and
N2O emissions would have increased our emission estimates, espe- Updated annual net CO2 emissions and removals from forest and
cially in countries where forest fires play a significant role in defor- from its two sub-components forest land and net forest conversion
estation and forest degradation. Analyses we performed with the (deforestation), were computed and made available via the
FAOSTAT Emissions database, however, showed that such addi- FAOSTAT database (http://faostat3.fao.org/faostat-gateway/go/
tional non-CO2 emissions were typically one order of magnitude to/download/G2/GF/E)13. Results were summarized globally and
lower than the CO2 estimated values. disaggregated by Annex I and non-Annex I Parties as averages for
the periods 1991–2000, 2001–2010, 2011–2015, and 1991–2015
(Table 3). Only mean values were reported.
3.2. Uncertainties associated with FRA data
4.1. Global trends
Additional uncertainties stemmed from: (i) Heterogeneity in
country data quality; and (ii) FRA data projections within given
Results indicated that over the period of study, 1991–2015, for-
FRA cycles.
est land was a net source of CO2 emissions globally, averaging
The latter type of uncertainty was related to the fact that, for
1.52 Gt CO2 yr1. These corresponded to emissions from
any FRA cycle, the last years of data are in fact projections rather
than actual measurements. The impact of this on estimated emis-
12
From IPCC 2006: Lack of knowledge of the true value of a variable that can be
sions and removals was evaluated herein by comparing results for
described as a probability density function (PDF) characterising the range and
years 2009–2010, which were projections in FRA 2010 data, likelihood of possible values. The uncertainty in the mean (e.g. the uncertainty
revised using actual measurements in FRA 2015. This was a simpli- associated with the calculation of the average biomass C density across the country) is
fied exercise, recognizing that periods longer than two years may estimated as plus or minus k (or approximately 2) multiples of the standard error
in fact represent projections in any given FRA cycle, and acknowl- divided by the sample mean times 100, where the standard error is the sample
standard deviation divided by the square root of the sample size. The uncertainty in
edging that uncertainties may grow by comparing periods further
the individual (e.g. the uncertainty associated with the use of the average biomass C
apart in time. Our analyses comparing net CO2 emissions and density for estimating C stock losses of a single unit of land) is estimated as plus or
removals for 2009–2010 made with FRA2015 to FRA2010, indi- minus k (or approximately 2) multiples of the sample standard deviation divided by
cated small relative effects on global deforestation estimates the sample mean times 100. Both calculations are based on an assumption of a
normal distribution, and k is obtained from the student’s t-distribution.
(0.6%), including for non-Annex I Parties (5.8%), and large rela- 13
The Forest sub-domain in FAOSTAT contains data by country, with a range of
tive effects in Annex I Parties (+43.8%), although it should be con- regional aggregations, including Annex I and non-Annex I groups. Data for download
sidered that in the latter case absolute values were very small. are net CO2 emissions/removals in GgCO2; C stock change in GgC; implied emission
Similarly, we found small effects on estimates of global forest net factors; and area.
S. Federici et al. / Forest Ecology and Management 352 (2015) 89–98 93
Table 3
Estimates of annual net CO2 emissions and removals from forest (woody biomass C stock changes only). Mean values only are reported.
deforestation of 4.04 Gt CO2 yr1, counterbalanced by net removals the global net forest sink decreased, from 2.99 on average during
in forest of 2.52 Gt CO2 yr1 (Table 3) Importantly, global emis- 1991–2000, to 2.24 in 2001–2010 and then to 2.15 Gt CO2 yr1
sions from forest degradation were also estimated, for the first time in 2011–2015.
in the literature, at 0.80 Gt CO2 yr1 on average over 1991–2015. As a result of the above, forest was a net source of CO2 into the
This implies that the actual forest sink was 3.32 Gt CO2 yr1. atmosphere over the entire period of study, albeit one that signif-
Historical trends were particularly significant (Table 4). Net icantly decreased over time (Table 4). Average CO2 emissions from
emissions from deforestation decreased over the period 1991– forest slowed from rather steady rates of 1.70 Gt CO2 yr1 over the
2015, from an average of 4.68 Gt CO2 yr1 in 1991–2000, to 3.95 period 1991–2010, to roughly half this value, or 0.79 Gt CO2 yr1,
in 2001–2010 and then to 2.94 Gt CO2 yr1 in the last five years in 2011–2015.
2011–2015. The most recent rates represented a reduction of over
one-fourth over the preceding decade.
4.2. Annex I and non-Annex I Parties and regional trends
During the same time periods, net emissions from forest degra-
dation increased three-fold, from 0.35 on average during 1991–
Disaggregation by country groups added significant insight into
2000, to 1.15 in 2001–2010 and 0.99 Gt CO2 yr1 in 2011–2015.
largely different trends between Annex I and non-Annex I Parties.
It should be noted that sudden shifts in CO2 emissions from defor-
On average during the entire study period of 1991–2015, Forest
estation to forest degradation were observed in some countries with
was a net source in non-Annex I Parties, averaging
significant forest area, after the year 2000. In such cases, aggregat-
2.58 Gt CO2 yr1, while it generated net CO2 removals in Annex I
ing net CO2 emissions at the level of forest would be preferable in
Parties, with an average of 1.06 Gt CO2 yr1. Natural forest and
order to provide for unbiased trends.
planted forest contributed equally to the estimated overall net
Finally, alongside the estimated increases in forest degradation,
CO2 removal in Annex I Parties. In non-Annex I Parties, natural for-
and mostly due to a decrease in non-Annex I Parties sink strength,
est were a significant net source (3.12 Gt CO2 yr1), largely due to
deforestation, while planted forest acted as a net sink
Table 4 (0.54 Gt CO2 yr1) (Table 3).
Summary estimates of annual net CO2 emissions and removals for forest, showing
deforestation and forest land contributions, limited to woody biomass C stock
Deforestation took place mainly in natural forest in non-Annex I
changes. The latter is further separated into a forest degradation component and a net Parties, emitting on average 3.75 Gt CO2 yr1, while deforestation
CO2 removal component. rates in Annex I Parties were minimal, at 0.26 Gt CO2 yr1.
Decreases in deforestation rates in non-Annex I Parties followed
Average annual Gt CO2 yr1
the global trends reported above. In addition, net CO2 emissions
1991– 2001– 2011–
from forest degradation were almost entirely located in
2000 2010 2015
non-Annex I Parties, which were responsible for as much as 90%
net CO2 sink in forest land 2.99 2.24 2.15
of the total net CO2 emissions from degradation estimated herein.
net CO2 removals in forest land 3.34 3.38 3.14
net CO2 emissions from forest 0.35 1.15 0.99 Over the period 1991–2015, planted forest, representing 7% of
degradation total forest area, accounted for a global average sink that was com-
CO2 emissions from net deforestation 4.68 3.95 2.94 parable to the sink in natural forest (1.08 vs. 1.44 Gt CO2 yr1)
(Table 3). The sink in planted forests was driven by continuous
Forest total net CO2 source 1.69 1.71 0.79
increases in total area.
94 S. Federici et al. / Forest Ecology and Management 352 (2015) 89–98
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
0
-500,000
-1,000,000
Gg CO2
-1,500,000
-2,000,000
-2,500,000
Year
Global Annex I Parties non-Annex I Parties
Fig. 1. Annual total net CO2 emissions and removals in the Natural forest subdivision of category Forest land.
Forest: annual total net CO2 emissions and removals in Planted forest
100,000
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
-100,000
-300,000
-500,000
Gg CO2
-700,000
-900,000
-1,100,000
-1,300,000
-1,500,000
Year
Global Annex I Parties non-Annex I Parties
Fig. 2. Annual total net CO2 emissions and removals in the Planted forest subdivision of category Forest land.
S. Federici et al. / Forest Ecology and Management 352 (2015) 89–98 95
4,500,000
4,000,000
3,500,000
3,000,000
Gg CO2
2,500,000
2,000,000
1,500,000
1,000,000
500,000
0
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
Year
Global Annex I Parties non-Annex I Parties
Fig. 3. Annual total CO2 emissions in the Natural forest subdivision of category Net Forest Conversion.
60,000
50,000
Gg CO2
40,000
30,000
20,000
10,000
0
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Year
Global Annex I non-Annex I
Fig. 4. Annual total CO2 emissions in the Planted forest subdivision of category Net Forest Conversion.
96 S. Federici et al. / Forest Ecology and Management 352 (2015) 89–98
Forests in Annex I Parties represented 52% of the global esti- discontinuities in the order of 1 Gt CO2 yr1 were observed across
mated carbon sink, consistently with their share of global forest years 2000 and 2001, in non-Annex I Parties, while smaller discon-
area coverage (45%), and increases in biomass C density, from tinuities were observed in Annex I Parties, across the years 2005
44.9 t C ha1 in 1990 to 48.2 t C ha1 in 2015, accounted for 87% and 2006. Remarkably, they were largely determined by trends
of the forest sink. The remaining 13% of the sink was linked to in a few countries. In fact, the large discontinuity in non-Annex I
increases in forest area, from 1.81 to 1.84 billion hectares in Parties trends was due to one country alone (Indonesia). Its forests
1990 and 2015, respectively. were estimated to be a significant net sink before the year 2000,
By contrast, the net sink in non-Annex I Parties was determined becoming a large net source afterwards. The smaller discontinuity
exclusively by forest area expansion from 1991 to 2015. Indeed, in Annex I Parties between 2005 and 2006 was characterized by
despite overall area decreases in non-Annex I Parties of some large increases in forest sinks after 2006. This trend was deter-
154 million ha, area gains of 104 million ha were observed in some mined by three countries: Australia, where forests were a net
countries, established through natural forest expansion (56 million source before 2005 but became a net sink afterwards; Canada,
ha) and afforestation (48 million ha). By contrast, biomass C stock where the net source in forest land was halved after 2005, and
density decreased, from 97.5 in 1990 to 96.3 t C ha1 in 2015. the Russian Federation, where the net sink almost doubled after
2005.
4.3. Critical national trends For planted forest, the estimated annual net uptake of CO2
decreased in Annex I Parties during 2011–2015, due to significant
Net CO2 emissions and removals for natural forest were com- decreases in newly planted area, i.e., from an average of 2 Mha yr1
puted annually in order to assess critical dynamics (Fig. 1). Large prior to 2010 to 1 Mha ha yr1 during 2011–2015 (Fig. 2). In
Table 5
Twenty-five year long-term average analysis of forest net carbon emissions and removals from FAOSTAT, compared to Annex I and available non-Annex I Parties submissions to
UNFCCC (woody biomass C stock changes only).
Comparison between FRA and FAOSTAT data GHG data (non-Annex I Parties) GHG Inventory
UNFCCC data (% difference) (Annex I Parties)
This analysis UNFCCC website
Average 1991–2010 – Gg CO2 net emissions
Grand total 3.46% 2,557,116 2,648,718
Total non-Annex I Parties 7.93% 1,215,833 1,126,553
Total Annex I Parties 11.88% 1,341,283 1,522,165
Annex I Parties
FAOSTAT/UNFCCC Comparisons, 20 years average annual biomass C stock change 1991-2020
120,000
100,000
FAOSTAT Tier 1 Estimates, Forest land (kt C)
80,000
60,000
40,000
20,000
0
-20,000 0 20,000 40,000 60,000 80,000 100,000 120,000
-20,000
UNFCCC National Inventories, Forest Land (kt C)
Fig. 5. Comparisons of net CO2 emissions and removals from Forest land in Annex I Parties. FAOSTAT estimates vs. UNFCCC reports.
S. Federici et al. / Forest Ecology and Management 352 (2015) 89–98 97
non-Annex I Parties, by contrast, the net carbon sink of planted Despite the simplified carbon stock change methodology
forest continued to increase over the entire analysis period, due applied in this study, it is worth noting that the estimates of global
to increased area. CO2 emissions from deforestation provided herein were fully con-
Our annual estimates clearly indicated the critical role of defor- sistent with those summarized by the IPCC AR5 WGI
estation in natural forest of non-Annex I Parties, with a minor con- (4.03 ± 2.93 Gt CO2 yr1). The latter were based on a range of mod-
tribution from planted forest (Figs. 3 and 4). Similarly to previous elling techniques, including some using FRA data in conjunction
cases, observed discontinuities were determined by only two coun- with more complex bookkeeping methods (i.e., Houghton et al.,
tries, as follows: 2012).
Decreases by almost two-thirds of net deforestation in 5.2. Suggested methodological and data improvements
Indonesia after 2000.
Decreases by almost one-third of net deforestation in Brazil The quality of FAO estimates for forest discussed herein could
after 2005. benefit from a number of improvements in FRA reporting. These
Decreases by almost one-half of net deforestation in Brazil after include:
2010.
1. Enhanced country data to cover carbon stock gains and carbon
stock losses separately, and disaggregated by forest type.
4.4. Comparisons of FAOSTAT estimates and UNFCCC data
2. Improved coordination at national level, especially between
FRA and UNFCCC focal points, to ensure consistency, complete-
We compared 1991–2015 global averages of FAOSTAT esti-
ness and accuracy of reported information.
mates for forest land14, limited to countries with forest area larger
than 1 Mha, to similar data reported by countries to the UNFCCC15
Looking forward to the 2020 FRA, with respect to point 1 above,
(Table 5).
significant possible improvements would be the inclusion of data
We found that FAOSTAT estimates were within 5% of UNFCCC
on woody biomass stocks by forest strata,, i.e. primary forest, other
data, thus statistically similar, given the large uncertainties
naturally regenerated forest, and planted forest, allowing for com-
discussed previously. Disaggregation between Annex I and
putation of different carbon stock densities and thus a more pre-
non-Annex I Parties also showed good agreement. Specifically,
cise estimation of CO2 emissions and removals. Separation of
for non-Annex I Parties, FAOSTAT estimated a sink 4% larger than
losses from harvesting of industrial round-wood and fuel-wood
implied by UNFCCC data. For Annex I Parties, FAOSTAT estimated
would also be extremely useful to permit implementation using
a sink 12% smaller than implied by UNFCCC data (Fig. 5). Results
the gain-loss method.
for the Russian Federation were responsible for most of the differ-
Further improvements to enhance progress toward a harmo-
ence observed between the two data sources. This country
nized and complete dataset of regional and global C balance esti-
reported a rather stable forest area and forest C stock to FRA,
mates of forest, while recognizing the difficulty of gathering such
resulting in minor net CO2 removals. By contrast, its UNFCCC data
information, would be data on forest disturbance, as well as better
indicated a large net carbon sink, of some 400 Mt CO2 yr1 on
data on C stock gains associated with stock increments, ideally dis-
average.
aggregated by strata. We further note that, although the new FRA
2015 questionnaires requested information on gross forest expan-
5. Conclusions sion (topic 1.6) and information on gross deforestation (FRA topic
1.7), few countries reported such information. Improved reporting
5.1. Main findings of gross area changes is very difficult, yet it would be needed to
increase the accuracy of computations of annual C stock changes
Our analysis quantified the global relevance of forest-related associated with forest area changes.
CO2 emissions and removals, making available estimates based Importantly, countries should be given an opportunity, depend-
on official country data and international methodologies. Our find- ing on their capacity and timing of forest survey/census cycles, to
ings quantified for the first time the overall net forcing of forests on communicate updated forest data to FAO between FRA reporting
the atmosphere over the period 1991–2015, (1.52 Gt CO2 yr1), periods, in order to avoid incongruities between actual national
highlighting the significant role of deforestation as a net source data and FRA projections, especially for the last two years of a
(4.04 Gt CO2 yr1), and the importance of remaining forests as given FRA cycle.
net sink (2.52 Gt CO2 yr1). Furthermore, this study quantified Finally, consistency between FRA and UNFCCC data should be
global CO2 emissions form forest degradation, alongside those from improved. FRA questionnaires could explicitly ask countries to
deforestation, providing a useful basis for quantifying emission report to the two UN agencies the same set of values on forest area
sources under REDD+. The value estimated for forest degradation and area change, volume increment, harvesting quantities, grow-
over 1990–2015, 0.80 Gt CO2 yr1, was about one-fourth of defor- ing stock, other C stocks, etc.
estation emissions. Importantly, while emissions from deforesta-
tion significantly decreased in non-Annex I Parties over time, e.g., 5.3. Mitigation potential of forests
from 4.68 in 1991–2000 to 2.94 Gt CO2 yr1 in 2011–2015, emis-
sions from forest degradation increased three-fold, from 0.35 in Forests can play a significant role in the mitigation of climate
1991–2000 to 0.99 Gt CO2 yr1 in 2011–2015. change. Mitigation through forest may be achieved either reducing
net C stock losses or increasing of long-term average C stocks.
14
Net forest conversion could not be directly compared with data on gross According to the 2015 FRA data, natural forests lost their bio-
deforestation reported to the UNFCCC mass stocks at a rate of 2.5% per year during the period 1991–
15
For Annex I Parties, data were taken from their 2014 national GHG Inventory, 2015. Our estimates helped to clarify that, if reversed, this would
Common Reporting Format (CRF) Table 5A: biomass net C stock change (cell M10): correspond to a REDD+ mitigation potential of about
http://unfccc.int/national_reports/annex_i_ghg_inventories/national_invento-
ries_submissions/items/8108.php. For non-Annex I Parties, data were taken from
4 Gt CO2 yr1 from avoided deforestation and 1 Gt CO2 yr1 from
their most recent National Communications (http://unfccc.int/ghg_data/items/3800. avoided forest degradation. Additional mitigation potential would
php), category ‘‘Changes in Forest and Woody Biomass Stocks’’. have to be achieved through new afforestation of degraded lands.
98 S. Federici et al. / Forest Ecology and Management 352 (2015) 89–98
Therefore, future FRA efforts should focus on providing more Series, 14/01. UN FAO, Rome, Italy, <http://www.fao.org/docrep/019/i3671e/
i3671e.pdf>.
detailed information on forest types, strata, and C stocks, in order
FAOSTAT, 2015. FAOSTAT online database at <http://faostat3.fao.org/faostat-
to allow for better quantification of such significant potential. gateway/go/to/browse/G2/*/E>.
Houghton, R.A., House, J.I., Pongratz, J., van der Werf, G.R., DeFries, R.S., Hansen,
Acknowledgments M.C., Le Quéré, C., Ramankutty, N., 2012. Carbon emissions from land use and
land-cover change. Biogeosciences 9, 5125–5142. http://dx.doi.org/10.5194/bg-
9-5125-2012.
The Climate, Energy and Land Tenure Division, the Forestry IPCC, 2006. 2006 IPCC Guidelines for National Greenhouse Gas Inventories,
Division and the Statistics Division of FAO supported this work Prepared by the National Greenhouse Gas Inventories Programme. In:
Eggleston, H.S., Buendia, L., Miwa, K., Ngara, T., Tanabe, K., (Eds), IGES,
under the MAGHG Project. Generous funding was provided by Hayama, Japan, <http://www.ipcc-nggip.iges.or.jp/public/2006gl/index.html>.
Norway and Germany, via Trust Funds GCP/GLO/GER/286 and MacDicken, K.G., 2015. Global Forest Resources Assessment 2015: What, why and
GCP/GLO/NOR/325. We thank A. Gordon, F. Grita, S. Murzilli and how? Forest Ecology and Management. 352, 3–8.
Smith, P., Bustamante, M., Ahammad, H., Clark, H., Dong, H., Elsiddig, E.A., Haberl, H.,
G. Barbaglia of the FAO Statistics Division for providing essential Harper, R., House, J., Jafari, M., Masera, O., Mbow, C., Ravindranath, N.H., Rice,
data and IT support. We are thankful to K.G. MacDicken of the C.W., Robledo Abad, C., Romanovskaya, A., Sperling, F., Tubiello, F., 2014.
FAO Forest Management Division for providing us with the FRA Agriculture, Forestry and Other Land Use (AFOLU). In: Edenhofer, O., Pichs-
Madruga, R., Sokona, Y., Farahani, E., Kadner, S., Seyboth, K., Adler, A., Baum, I.,
2015 data and for many helpful suggestions and insights. We thank Brunner, S., Eickemeier, P., Kriemann, B., Savolainen, J., Schlömer, S., von
two anonymous reviewers for their useful comments, which signif- Stechow, C., Zwickel, T., Minx, J.C., (Eds.), Climate Change 2014: Mitigation of
icantly improved this manuscript. Climate Change. Contribution of Working Group III to the Fifth Assessment
Report of the Intergovernmental Panel on Climate Change, Cambridge
University Press, Cambridge, United Kingdom and New York, NY, USA.
References Tubiello, F.N., Salvatore, M., Ferrara, A.F., House, J., Federici, S., Rossi, S., Biancalani,
R., Condor Golec, R.D., Jacobs, H., Flammini, A., Prosperi, P., Cardenas-Galindo, P.,
FAO, 2010. Global Forest Resources Assessment 2010, FAO Forestry Paper 163, Schmidhuber, J., Sanz Sanchez, M.J., Srivastava, N., Smith, P., 2015. The
Rome, <http://www.fao.org/forestry/fra/fra2010/en/>. contribution of agriculture, forestry and other land use activities to global
FAO, 2014. Agriculture, Forestry and Other Land Use Emissions by Sources and warming, 1990–2012. Glob. Change Biol. http://dx.doi.org/10.1111/gcb.12865.
Removals by Sinks: 1990–2011 Analysis. FAO Statistics Division Working Paper
Forest Ecology and Management 352 (2015) 99–108
a r t i c l e i n f o a b s t r a c t
Article history: This paper presents results of analyses of data from the 2015 Global Forest Resources Assessment on
Received 19 December 2014 changes in forest ownership, public income and expenditure on forestry and forestry employment.
Received in revised form 1 April 2015 Forest ownership continued to show less state control and ownership of forests. This was due to
Accepted 8 April 2015
private-sector investment and, in some countries, the transfer of public forests to the private-sector
Available online 7 September 2015
(including local communities). This contrasts somewhat with results on public income and expenditure,
which indicate that public expenditure on forestry has increased dramatically over the last decade, while
Keywords:
income has increased by very little. Global employment in forestry has not changed much over the last
Forestry employment
Forest ownership
two decades and has remained at about 12.7 million people, with the majority of these employed in infor-
Forestry revenue and expenditure mal activities, particularly in Asia. While production of many if not most forest goods and services has
increased, labour productivity has improved at the same time, leading to this result. Comparing the
results for groups of countries at different income levels, it appears that higher income countries tend
to have a relatively high proportion of private forest ownership, high levels of labour productivity and
high levels of public spending (per hectare) on forestry. However, apart from these very general differ-
ences in outcomes related to income levels, there does not appear to be strong correlation between these
socioeconomic variables and other forest-related variables collected in the FRA.
Ó 2015 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
http://dx.doi.org/10.1016/j.foreco.2015.04.011
0378-1127/Ó 2015 Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
100 A. Whiteman et al. / Forest Ecology and Management 352 (2015) 99–108
individuals and enterprises (e.g. corporate, payroll or income taxes, compared to 77%) and it is highest of all in low income countries
land and property taxes, sales or value-added taxes), import taxes (at 96%). The result for high income countries largely reflects the
or duties levied on forest products and income from publicly presence of the Russian Federation in this country group, which
owned business entities. accounts for over half of the total forest area in this domain and
Income and expenditure data were reported for 2000, 2005 and almost all of this forest is owned by the state. Excluding this coun-
2010 (Table 1). For the countries providing information about only try, the other high income countries would also follow this pattern
income or expenditure, it was much more common for countries to with a proportion of forests in public ownership of 62%, slightly
report expenditure, although it is quite likely that the missing val- lower than in upper-middle income countries.
ues (about income) are either zero or very low. Thus, the compar- The area where forest ownership is unclear or unknown is rel-
ison of income and expenditure is not likely to be biased atively small at the global level and the figures presented above
significantly by the absence of paired observations, although this largely reflect the situation in two of the 21 countries that reported
remains a source of uncertainly in this analysis. any areas of unknown ownership (Brazil and Mexico2). Given that it
is commonly believed that land and forest ownership and tenure is
quite uncertain in many countries (especially tropical countries), it
2.3. Employment in forestry
would appear that the data reported to the FRA 2015 may not be
capturing the complexity of the situation with respect to forest
It is vital to note that FRA 2015 defined employment in forestry
ownership.
as ‘‘employment in activities related to production of goods derived
For example, land tenure and ownership is very clear in almost
from forests’’, which corresponds to the International Standard
all high income countries. The legal framework for land and forest
Industrial Classification (ISIC, 2008) Activity A02 (forestry and log-
ownership has been established for many decades and
ging) and is referred to as the forestry sub-sector in the remainder
record-keeping is effective in many of these countries, so there
of this text.
are relatively few areas of forest where ownership is uncertain,
The ISIC does not distinguish between formal and informal or
unknown or contested. In contrast, in many low income countries
between legal and illegal production. According to ISIC Rev. 4,
privately owned land is clearly identified and recorded, but most
Activity A02, forestry and logging include the following Groups:
other land is owned by the state as a default position in the
021 – silviculture and other forestry activities; 022 – logging;
absence of any other legally defined or recognised types of land
023 – gathering of non-wood forest products; and 024 – support
ownership. In reality, while technically owned by the state, many
services to forestry. This means that forest sector employment in
of these areas are likely to be used by local people for lots of differ-
other than these activities is not included in this analysis.
ent purposes, meaning that the state does not always have exclu-
Employment data were reported for 1990, 2000, 2005 and 2010
sive control and use rights (as stated in the definition of
and were expressed in thousand Full Time Equivalents (1000 FTE)
ownership given above) (Banerjee, 1997).
(see Table 1).
Countries that report some areas of unknown ownership are
For the year 1990 in FRA 2015, information about employment
likely to be those that recognise that state ownership without real
was collected from 105 countries representing 32% of the global
control over such areas is untenable in the long-run and many of
forest area (Table 1). Several large forest area countries did not
these countries have reached a level of development where they
report (Democratic Republic of Congo, Mexico, Colombia, Angola,
are willing and able to address such issues. It is also worth noting
Papua New Guinea and the Republic of Congo).
that many of the countries with areas of unknown ownership are
Ninety-one countries reported employment for all four years,
in Latin America, where socio-economic and historical factors have
with these countries representing approximately 30% of the global
not led to state ownership as the default position for much of the
forest area. The low response rate is because some countries with
land area in these countries.
large forest areas (Russian Federation, Brazil, USA, the Democratic
Globally, the forest area owned by the state has fallen from
Republic of Congo, Australia and Indonesia) did not provide infor-
2.6 billion ha to 2.5 billion ha over the period and the area of pri-
mation for the entire time series. Thus, due to the low number of
vately owned forest has increased by about 100 million ha, from
countries reporting the whole time series, the trend analysis from
430 million ha to 530 million ha (Table 2). This means that the pro-
1990 to 2010 is likely to be very imprecise. The same is true for
portion of forest owned by the state has fallen slightly from 85% in
female employment, where only 29 countries reported for the
1990 to 82% in 20103 and privately owned forests have increased in
whole time series, representing approximately 16% of the total for-
importance from 14% to 18% of the total forest area over the same
est area. This shows clearly the challenge to collect
period (Fig. 2). However, while these two trends at first appear to
sex-disaggregated data (SDD).
be ‘‘equal and opposite’’ at the global level, the regional pattern of
changes in forest ownership are much more complicated and the
3. Results and discussion trends at this level have been driven in different places by several
different factors.
3.1. Forest ownership For example, looking at the different ecological domains, almost
all of the change in the structure of forest ownership has occurred
Forest ownership in 2010 was divided as follows: 2,964 million in the temperate zone, where the proportion of forests owned by
ha in public ownership (76%); 772 million ha in private ownership the state has declined from 63% to 52% over the period. The struc-
(20%); and 141 million ha of unknown or unclear ownership (4%). ture of ownership in the tropical and boreal domains is the same in
Fig. 1 shows that most forests are owned by the state in all cat- 2010 as it was in 1990 with only a 1% change in the subtropical
egories. In many cases, public ownership accounts for the greatest domain (see Fig. 2).
share of all forest by far, but this dominance is lower in the temper-
ate and subtropical domains, where public ownership accounts for
2
only 52% of all forest area in both cases. Brazil (Tropical) and Mexico (Subtropical) are upper-middle income countries.
3
With respect to income, the dominance of public ownership Note – the figure for 2010 is slightly different to that presented earlier because
the calculation only includes the countries that provided data for all of the four years:
generally declines at higher levels of income. For example, the pro- 1990; 2000; 2005 and 2010. The area of forest of unknown ownership is not
portion of forest that is publicly owned is lower in upper-middle mentioned here because, in the absence of trend data for Brazil and Mexico, the
income countries than in lower-middle income countries (66% figures for the remaining countries are insignificant.
102 A. Whiteman et al. / Forest Ecology and Management 352 (2015) 99–108
Fig. 1. Forest areas in different ownership categories by climatic domain and income category in 2010.
Fig. 2. The proportion of forest owned by the state by climatic domain and income category (1990–2010).
the Fig. 3 below shows that there are some significant differences income and expenditure in these two domains (although a figure
between public sector income and expenditure in countries, lar- for income in 2010 was not available for the United States of
gely related to income levels in countries. America, which reported income of USD 1.3 billion in 2000).
Looking at the climatic domains, expenditure in boreal and The figure for income and expenditure by income category
tropical forests was relatively low at USD 1.5 billion and USD clearly shows how public expenditure is much higher in higher
2.9 billion respectively and income in each of these domains income countries and is significantly higher than income in those
was just under USD 6 billion. In these countries, forestry activities countries. At low and lower-middle incomes, both public expendi-
result in significant net income for the state, although these fig- ture and income are much lower, although they do also appear to
ures largely reflect the results in a small number of countries be approximately the same, suggesting that forestry in these coun-
(Sweden, Finland, the Russian Federation, Canada, Brazil, tries is not a net cost to the government. The data for these coun-
Malaysia, Gabon, India and Papua New Guinea) and while the tries is very partial, especially with the absence of any figures for
results for the boreal region are comprehensive, many tropical many tropical countries. However, it is suspected that income
countries did not provide any data. and expenditure is probably also very low for many of the missing
For the subtropical and temperate zones, public sector expendi- countries (and may also be in balance).
ture is far higher than income, at USD 8.3 billion and USD 25.2 bil- What is perhaps the most interesting result of this analysis is
lion respectively and with corresponding public sector income that expenditure on forest management per hectare is significantly
figures of USD 0.5 and USD 2.8 billion. The figures for these two higher in higher income countries; at about USD 12 per hectare in
domains included the results for many countries in Europe, plus high and upper middle income countries compared to USD 2 per
China, so they are likely to be quite representative of public sector hectare in low and lower middle income countries.
Fig. 3. Public sector income and expenditure on forestry by climatic domain and income category in 2010.
104 A. Whiteman et al. / Forest Ecology and Management 352 (2015) 99–108
Fig. 4. Distribution of global forestry employment by income category and by climatic domain in 2010.
A. Whiteman et al. / Forest Ecology and Management 352 (2015) 99–108 105
80
80
Fig. 5. Employment, forest designation and production by income category in 2010. Note: the size of the bubbles represent the amount of production, with woodfuel shown
on the left (A) and industrial roundwood shown on the right (B).
14 13
13 12
12 11
11
Employees (million)
Employees (million)
10
10 9
9 8
8
7
7
6
6
5 5
4 4
3 3
2 2
1 1
0 0
2000 2005 2010 2000 2005 2010
Total employment High Total employment Boreal
Low Low middle Sub tropical Temperate
Upper middle World average Tropical World average
Generally, from Fig. 6 above it can be observed that global An opposite trend (of increasing employment) can be observed in
employment in forestry and logging has decreased very slightly over some other regions, such as: South America, Western and Central
the decade towards 2010, with very little difference in these trends Africa and South and Southeast Asia. Some countries in these
between income or climatic domain categories. However, there are regions have considerably increased employment in forestry (and
differences between selected sub-regions (Fig. 7). also wood removals in many cases) with, for example an increase
Although these trends in sub regions have been strongly influ- of 200% in Uruguay, 136% in Peru, 96% in Suriname and 26% in
enced by a few countries in each case, two main trends can be Brazil from 2000 to 2010. According to the Forest Producer’s
observed. The first is that there is a group of sub regions where Society of Uruguay (2011), forestry there has been growing due to
employment has declined in the last decade (including: Europe; increasing exports of forest products, afforestation and reforesta-
North America; Oceania5; East Asia; and Western and Central tion, as well as industrial development. In Brazil, afforestation has
Asia), although the reasons behind these declines differ somewhat consistently increased the forest plantation area since the
between the sub regions. mid-1960s and these resources now support a major pulp and paper
In the case of Europe and North America (generally high income industry, as well as the development of furniture and other value
countries), employment has been drastically affected by the global added industries (Biani et al., 2009). Both of these countries are good
economic downturn of 2008–2009, when the construction industry examples of where legislation to promote development of the for-
(a major consumer of forest products) entered a period of deep and estry sector has led to significant long-term employment benefits.
prolonged recession (Ma et al., 2009). China accounts for most of In contrast, employment growth in South and Southeast Asia
the trends in East Asia6 and employment there has also reduced by and West and Central Africa has been strongly influenced by
21% from 2000 to 2010, due to a decline in export markets for processed growth in the numbers of people involved in the collection of wood
forest products (such as furniture) and a decline in house building fuel and NWFPs in places such as India, Togo and Mali.
(UNECE/FAO, 2009). In contrast, employment also decreased in
Australia and New Zealand at the same time that production increased,
showing that declining employment in the sub sector can also be due to 3.3.1. Female employment in 2010
rising productivity (in terms of production per employee). In 2010, female employment reached 1.40 million employees.
The countries with the highest female employment were
5
Bangladesh with 600 thousand, China, with 300 thousand, Mali
Oceania is represented in this analysis by 3 countries; Tonga, Australia and New
Zealand.
with 180 thousand and Brazil with 90 thousand.
6
In this analysis, there are 3 East Asian countries reporting for the 3 years: China, Sixty-nine countries provided information on female employ-
Japan and Mongolia. ment in 2010 (of which nine reported this as zero) and the share
106 A. Whiteman et al. / Forest Ecology and Management 352 (2015) 99–108
200
190
180
170
160
150
Employees (thousand)
140
130
120
110
100
90
80
70
60
50
40
30
20 2000
10
0 2005
Caribean 2010
Central
North
America Northern
America Oceania
Africa Western
and Central
Asia
100
90
Share of female employment (%)
80
70
60
50
40
30
20
10
0
of female employment in total forestry employment was calcu- The high proportion of women employed in forestry in some
lated and is shown in Fig. 8. The countries with the highest share countries could also be partly explained by progress in policy
of female employment are Mali with 90%, Mongolia and Namibia and legislation in places like in Bangladesh, which has updated
with 45% and Bangladesh with 40%. its forest policy and legislation to enhance women’s participation
The countries that did report female employment included in social forestry development.7
countries from a mixture of the different climatic domains and
levels of development and if these are considered to be globally 3.3.2. Productivity
representative, they suggest that females may account for about The data on employment can also be used to analyse trends in
30% of employment in the sub-sector. This figure is similar to the productivity (i.e. the amount of output per employee) (Fig. 9). In
level reported in FAO (2014), which estimated that women general, productivity has increased over the last 10 years, by 23%
accounted for about 25% of all employment in formal forestry and at the global level from 2000 to 2005 and 20% from 2005 to 2010.
logging activities. The analysis of the additional information in As this figure shows, boreal countries have the highest levels of
the country reports suggested that, because of a lack of options, productivity and it has also increased in these countries more than
women are more involved in non-paid and subsistence activities
(e.g. collection of fuelwood and NWFPs) rather than in salaried 7
National Forest Policy 1994 ‘‘Women will be encouraged to participate in
jobs, so the slight difference between these two figures could be homestead and farm forestry, and participatory afforestation programs’’. Laskar
explained by the inclusion of informal activities in some countries Muqsudur Rahman. Gender-positive Changes in Benefits-Sharing in Social Forestry
in the FRA 2015 data. Projects in Bangladesh. ICIMOD, 2012.
A. Whiteman et al. / Forest Ecology and Management 352 (2015) 99–108 107
2500
2250
2000
1750
m3 per employee
1500 Boreal
1250 Temperate
Sub tropical
1000
Tropical
750
Global average
500
250
0
2000 2005 2010
Years
anywhere else, but there are some differences between these coun- 4. Conclusions
tries. For example, Canada and Finland have reduced the number of
employees in forestry and logging since 2000, but they have also The FRA data on forest ownership, public income and expendi-
reduced wood removals at a higher rate (partly due to the recent ture and employment in forestry show huge variations between
recession). A similar trend to this is present in most temperate countries that cannot be easily explained. Apart from the general
countries. In contrast, the Russian Federation (which accounts for observation that some of these variables appear to vary according
most of the boreal region) has reduced employment by an even to income levels, they are not strongly related to other forest vari-
greater amount at the same time as wood removals have increased. ables where some degree of correlation might have been expected.
In the tropical domain, the productivity trend is strongly influ- This suggests that factors outside the sector may have more of an
enced by India, Brazil and Indonesia. While India shows a level of influence on the way that forests develop over time and in different
productivity far below the global average (47 and 67 m3 per places.
employee in 2000 and 2010 respectively), Brazil and Indonesia It is also worth noting that, for many of these variables, often
are nearer the global average. Productivity in Indonesia has drasti- only a small number of countries account for a huge proportion
cally increased in the last 10 years from 25 to 522 m3 per of the global total and tend to have a major impact on results at
employee, although this may be due to inaccurate information the global level (e.g. China and the USA in the case of public expen-
about wood removals. diture, Russia, Canada and the USA in the case of ownership struc-
Subtropical countries show relatively stable productivity, with a tures and India in terms of employment). A similar issue was noted
slight decreased from 2005 to 2010. This is the result of gains in in the previous FRA. Despite some improvements in the dataset,
some countries and losses in others. For instance, Australia has there still remain a number of significant gaps and uncertainties
reduced employment by almost 50%, resulting in a productivity in the data, particularly related to public expenditure and income
increase of 84%, although the sector is of growing importance to in the sector and, to a lesser extent, employment.
Australia and has received intensive policy focus since the early The analysis presented here has used fairly simple techniques to
1990s (Low and Sinniah, 2010). try to identify relationships between different variables by exam-
From the analysis of employment, it can be concluded that data ining changes over time or differences between countries, but
quality is a crucial issue for understanding the contribution of the the results have been fairly inconclusive. Given the complexity of
forestry sub-sector to employment in a country. In many cases, the situations in different countries it may be more useful in the future
figures gathered by FRA 2015 included besides the formal employ- to explore similarities rather than differences between countries,
ment, also part time/temporary jobs that may be partly informal. using techniques such as cluster analysis. While less useful for
Therefore, the data and analysis cannot be compared with other examining causal relationships in forestry development, such tech-
studies based on formal employment statistics. However, although niques could provide insights in other areas such as the forest tran-
this is a difficulty, the inclusion of informal employment helps sition hypothesis and may make better use of the many different
greatly to understand the realities in the sector, where many of dimensions of the data collected in the FRA.
the employment benefits are to be found in informal activities,
especially in less developed countries.
The data on female employment showed that there are a high Acknowledgements
number of women involved in forestry activities but there is a
need to improve information about the roles that they play, as Special thanks to Canada, the European Commission, Finland,
the few figures that are available suggest that they are largely Food and Agriculture Organization of the United Nations, Japan
involved in subsistence activities, especially in developing coun- and the United States for funding provided to the Global Forest
tries. With this apparent lack of employment opportunities pre- Resources Assessment. Thanks to the CFRQ partners for the com-
sented to women, this suggests that further action is needed to mitment and hard work on harmonization of FRA 2015 data. The
promote more equal opportunities for women to access training authors would also like to thank Safia Aggarwal for her valuable
and paid employment. contributions to understand forest land ownership data, to
108 A. Whiteman et al. / Forest Ecology and Management 352 (2015) 99–108
Arvydas Lebedys for his support in the harmonization of data on Kastenholz, E., 2011. Criterion 6: maintenance of other socio-economic functions
and conditions. State of Europe’s Forest 2011. <http://www.unece.org/forests/
forest employment and to Orjan Jonsson for his continues support
fr/outputs/soef2011.html>.
with data analysis. In addition, authors are also very grateful to Krott, M., 2008. Forest government and forest governance within a Europe in
Kenneth MacDicken for his constant guidance and useful inputs change. Institute for Forest Policy, Forest History and Nature Conservation,
to improve and finalize this paper. Georg-August-University, Goettingen, Germany. The Multifunctional Role of
Forests- Policies, Methods and Case Studies. European Forest Institute
Proceedings. No 55.
Laskar Muqsudur Rahman, 2012. ICIMOD. Gender-positive changes in benefits-
sharing in social forestry projects in Bangladesh. National Forest Policy- 1994
References ‘‘Women will be encouraged to participate in homestead and farm forestry, and
participatory afforestation programs’’.
Agrawal, A., Cashore, R., Gill, H., Benson, C., Miller, C., 2013. Background paper 1: Low, K., Sinniah, M., 2010. Australian Bureau of agricultural and resource economics
economic contributions of forests. United Nations Forum on Forests. Tenth (abare). Future directions for the Australian forest industry. Issues insights 10.1.
Session. 8–19 April 2013. Istanbul. Turkey. March 2010. <http://daff.gov.au/SiteCollectionDocuments/abares/outlook/
Banerjee, A., 1997. Asia-pacific forestry sector outlook study. Decentralization and 2010/a1.pdf>.
Devolution of Forest Management in Asia and the Pacific. Working Paper No.: MacDicken, K.G., 2015. Global forest resources assessment 2015: what, why and
APFSOS/WP/21. how?. Forest Ecol. Manage. 352, 3–8.
Biani, J., Ingrassia L., Lanese L, Nusshold M., 2009. CEMUPRO. Estudio comparativo Phels, J., Webb, E., Agrawal, A., 2010. Does REDD+ threaten to recentralize forest
del sector forestal. Argentina, Chile, Uruguay y Brasil. <http://www. governance? Land Use Policy Forum Sci. 328 16, <http://www.sciencemag.org>.
cemupro.com.ar/wp-content/uploads/2010/11/trabajo_sector_forestal_comparativo. Ma, Q., Liu, J., Du, W., 2009. Food and agriculture organization of the united nations
pdf>. (FAO). How Chinese forestry is coping with the challenges of global economic
Cañares, M., 2009. Asia-pacific forestry sector outlook study II. Working Paper No. downturn. Unasylva. FAO, 2009. vol 60. ISSN 0041-6436. <http://www.fao.
APFSOS II/WP/2009/27. Macro – Economic Trends and their Impacts on Forests org/docrep/012/i1025e/i1025e09.htm>.
and Forestry in Asia and the Pacific to 2020. Ribot, J., Agrawal, A., Larson, A., 2006. Recentralizing while decentralizing: how
Cerutti, P., Tacconi, L., 2006. Forest, illegality, and livelihoods in Cameroon. Working national governments reappropriate forest resources. World Development, vol.
Paper No. 35. <http://www.cifor.org/publications/pdf_files/wpapers/wp-35. 34. Elsevier Ltd, pp. 1864–1886, No. 11.
pdf>. Sociedad de Productores Forestales del Uruguay, 2011. El sector forestal en el
Contreras, A., 2000. The Underlying causes of forest decline. Occasional Paper No. Uruguay. 2011. <http://www.fagro.edu.uy/~forestal/Archivos/Sector%20Forestal%
30. ISSN 0854–9818. 20Uruguayo%20marzo%202011.pdf>.
FAO, 2014. State of the world’s forests 2014. Enhancing the Socioeconomic Benefits The Organization for Economic Co-operation and Development (OECD), 2014. Better
from Forests. <http://www.fao.org/3/a-i3710e.pdf>. Life Index, Canada 2014. <http://www.oecdbetterlifeindex.org/countries/
FAO, 2010. Global forest resources assessment 2010. Main Report. FAO Forestry canada/>.
Paper No. 163. <http://www.fao.org/docrep/013/i1757e/i1757e.pdf>. Tomter, S., 2011. Criterion 6: 6.1 maintenance of other socioeconomic functions and
FAO, 2012. Forest resources assessment. Terms and Definitions. Working Paper No. conditions. State of Forest Europe 2011.
180. United Nations Economic Commission for Europe and Food and Agriculture
Forest Europe, 2014. Green economy and social aspects of sustainable forest Organization of the United Nations (UNECE/FAO), 2009. The importance of
management. Workshop Report. China’s forest products markets to the UNECE region. Geneva Timber and Forest
Indufor, 2013. Forest financing in African countries. Second Macro-Level Paper. Discussion Paper No. 57. <http://www.unece.org/fileadmin/DAM/timber/
International Standard Industrial Classification of All Economic Activities (ISIC), publications/dp-57.pdf>.
2008. Statistical papers, series M No. 4/Rev. 4. <http://unstats.un.org/unsd/cr/
registry/isic-4.asp.
Forest Ecology and Management 352 (2015) 109–123
a r t i c l e i n f o a b s t r a c t
Article history: Monitoring of forest cover and forest functions provides information necessary to support policies and
Received 28 November 2014 decisions to conserve, protect and sustainably manage forests. Especially in the tropics where forests
Received in revised form 29 May 2015 are declining at a rapid rate, national forest monitoring systems capable of reliably estimating forest
Accepted 1 June 2015
cover, forest cover change and carbon stock change are of vital importance. As a large number of tropical
Available online 7 September 2015
countries had limited capacity in the past to implement such a system, capacity building efforts are now
ongoing to strengthen the technical and political skillsets necessary to implement national forest moni-
Keywords:
toring at institutional levels. This paper assesses the current status and recent changes in national forest
Tropical forests
Forest cover change
monitoring and reporting capacities in 99 tropical countries, using the Food and Agriculture Organization
Monitoring systems of the United Nations (FAO) Forest Resources Assessment (FRA) 2015 data, complemented with FRA 2010
Carbon reporting and FRA 2005 data. Three indicators ‘‘Forest area change monitoring and remote sensing capacities’’,
Capacity building ‘‘Forest inventory capacities’’ and ‘‘Carbon pool reporting capacities’’ were used to assess the countries’
REDD+ capacities for the years 2005, 2010 and 2015 and the change in capacities between 2005–2010 and
2010–2015. Forest area change monitoring and remote sensing capacities improved considerably
between 2005 and 2015. The total tropical forest area that is monitored with good to very good forest
area change monitoring and remote sensing capacities increased from 69% in 2005 to 83% in 2015.
This corresponds to 1435 million ha in 2005 and 1699 million ha in 2015. This effect is related to more
free and open remote sensing data and availability of techniques to improve forest area change monitor-
ing. The total tropical forest area that is monitored with good to very good forest inventory capacities
increased from 38% in 2005 to 66% in 2015. This corresponds to 785 million ha in 2005 and 1350 mil-
lion ha in 2015. Carbon pool reporting capacities did not show as much improvement and the majority
of countries still report at Tier 1 level. This indicates the need for greater emphasis on producing accurate
emission factors at Tier 2 or Tier 3 level and improved greenhouse gases reporting. It is further shown
that there was a positive adjustment in the net change in forest area where countries with lower capac-
ities in the past had the tendency to overestimate the area of forest loss. The results emphasized the effec-
tiveness of capacity building programmes (such as those by FAO and REDD+ readiness) but also the need
for continued capacity development efforts. It is important for countries to maintain their forest monitor-
ing system and update their inventories on a regular basis. This will further improve accuracy and reli-
ability of data and information on forest resources and will provide countries with the necessary input to
refine policies and decisions and to further improve forest management.
Ó 2015 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction
q
About one third of the earth’s land surface is covered by for-
This article is part of a special issue entitled ‘‘Changes in Global Forest Resources
from 1990 to 2015’’.
ests which store about 45% of the world’s terrestrial carbon in
⇑ Corresponding author. wood, leaves, roots and soil. Almost half of this area consists
http://dx.doi.org/10.1016/j.foreco.2015.06.003
0378-1127/Ó 2015 Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
110 E. Romijn et al. / Forest Ecology and Management 352 (2015) 109–123
of tropical forests, which, on average, can store 50% more carbon inventories provide data on forest cover and forest cover change
in the trees than other types of forests (Houghton, 2005; Bonan, at national scale. At least a time series of Landsat-type remote
2008). In addition to playing a critical role in regulating the sensing data with 30 m spatial resolution should be used for
world’s climate, forests provide a variety of functions for people national deforestation monitoring. An example of operational
and the planet, including ecological, economic, social and aes- national and regional remote sensing monitoring is Brazil’s
thetic functions (Miura et al., 2015). Moreover, forests contribute PRODES system for monitoring deforestation in the Brazilian
largely to livelihood security and provide fuelwood and charcoal, Legal Amazon region. This system uses Landsat, DMC and CBERS
which are major energy sources in developing countries (FAO, satellite data at 20–30 m resolution (GOFC-GOLD, 2014). National
2014a). forest inventories provide data that are needed to estimate the car-
Humans are continuously changing the land use to get access to bon content of different forest types. India has a long history of
the planet’s resources through clearance of forests for agricultural national forest inventories. The new national forest inventory,
activities and urban expansion. Land use and land cover change established in 2001, generates national level estimates of growing
have a climate forcing effect and play a major role in changing stock at similar time intervals as the biennial forest cover assess-
the world’s climate (Pielke, 2005; Hosonuma et al., 2012). ment which is based on 23.5 m resolution IRS P6 satellite imagery.
Deforestation is a global threat, not only because it causes habitat Using a sampling design, estimates of growing stock are developed
fragmentation and loss of biodiversity, but it also degrades envi- for 14 physiographic zones based on physiographic features
ronmental conditions and has an impact on global greenhouse including climate, soil and vegetation. Every two years different
gas emissions (GHG) by releasing CO2 to the atmosphere. This districts are inventoried (GOFC-GOLD, 2014).
causes changes in the global carbon cycle and alters the surface Remote sensing technologies are continuously in development
energy and water balance. As a consequence, the release of carbon and new available satellite and airborne sensors, analysis and
affects climatic patterns and causes changes in environmental con- methods are emerging at a constant pace (De Sy et al., 2012).
ditions and ecosystems (Cramer et al., 2004; Foley et al., 2005). Evolving technologies include the use of LIght Detection And
Avoiding deforestation could reduce GHG emissions significantly. Ranging (LIDAR) and Synthetic Aperture Radar (SAR) observations
Forest management, including reducing and preventing deforesta- for forest characterization and biomass estimation. These tech-
tion is an important climate mitigation strategy and helps to niques can help to overcome the challenge of cloud cover in the
secure the different forest functions (Bonan, 2008; Salvini et al., tropics. Unmanned Aerial Vehicles (drones) can be used for local
2014). Numerous studies used global remote sensing data to high- scale validation studies. New techniques for acquiring, processing
light the fact that during the last decades forests in the tropics have and managing vast amounts of satellite remote sensing data
been rapidly declining (DeFries et al., 2002; Achard et al., 2004; include cloud-based databases and data processing platforms
Hansen et al., 2010, 2013; FAO and JRC, 2012). FRA 2015 data show which offer space for large datasets and computational resources
that the annual rate of net forest loss in the tropics has decreased for processing (GOFC-GOLD, 2014). An example of new and forth-
compared to the 1990s (9.5 M hectare per year) and 2000s (7.2 M coming initiatives includes the Copernicus program with a constel-
hectare per year). Recent estimates indicate a decline of 5.5 M hec- lation of earth observation satellites (Sentinels) and in-situ sensors
tare of forests per year between 2010 and 2015 in the tropics for monitoring the earth.
(Keenan et al., 2015). Capacities of tropical Non-Annex I countries to monitor forests
Monitoring forests over time allows countries to observe and forest cover change were limited in the past. However, through
changes. Regular and accurate monitoring of forest cover, forest capacity building efforts capacities are strengthening at technical,
cover change and drivers of change provides the necessary infor- political and institutional levels (Herold and Skutsch, 2011;
mation to support policies and management practices to protect, Romijn et al., 2012). A few Non-Annex I countries like Mexico
conserve and sustainably manage forests and to ensure the differ- and India have well developed national forest monitoring systems.
ent functions of forests (Mayaux et al., 2005; Achard et al., 2007; Other countries are in the process of developing capacities and are
MacDicken, 2015). Consistent information on forest resources is at various stages of development; they need considerable capacity
needed for developing these policies and monitoring should be improvements before they are able to produce accurate estimates
done at a national scale to properly assist localized land manage- of forest area, forest area change and carbon stock change
ment decisions. At the global level, the importance of monitoring (Tulyasuwan et al., 2012).
forest cover change and forest functions is reflected in environ- The aim of this paper is to assess the capacity status and the
mental conventions such as the United Nations Framework changes in national forest monitoring and reporting capacities in
Convention on Climate Change and the Convention on Biological 99 tropical Non-Annex I countries for the years 2005, 2010 and
Diversity. 2015, using the FAO FRA data. The specific objectives are to explain
Main focus of forest observation systems is on monitoring forest the change in forest monitoring and reporting capacities and to
area and changes in forest area and on monitoring forest carbon investigate the effectiveness of capacity building initiatives for
stocks and changes. Earth observation has a key role in monitoring improving national forest monitoring systems. Additional objec-
tropical forests. It should be noted that different studies use differ- tives are to assess the effect of increased capacities on the area of
ent definitions of forests and earth observation may provide vary- tropical forest that is monitored with accurate and reliable data
ing estimates of ‘‘forest area’’ depending on the definition and and methods, and to assess the effect of increased capacities on
method that is used. For example, the FRA (FAO, 2015) uses a forest reported FRA numbers of net change in forest area for similar time
land use definition. The JRC/FAO Remote Sensing Survey (FAO and periods.
JRC, 2012) uses a forest cover definition, while Hansen et al. (2013)
measures tree cover change. This results in varying estimates of
forest extent for a similar area. A detailed explanation on this mat- 2. Materials and methods
ter can be found in Keenan et al. (2015).
Several papers discuss the technical requirements for imple- 2.1. Data
menting national forest monitoring systems in the context of
REDD+ (Mayaux et al., 2005; Achard et al., 2007; DeFries et al., This study focuses on 99 tropical Non-Annex I countries. These
2007; Herold and Johns, 2007; De Sy et al., 2012). Earth observing include countries that are located in the sub-tropical or tropical
satellite data analyses, together with field-based national forest domain, as defined in the FRA 2015 datasets (FAO, 2015) and
E. Romijn et al. / Forest Ecology and Management 352 (2015) 109–123 111
China, which in the Southern part is covered by tropical forests. biomass, while the third indicator focuses on reporting on biomass
The main data sources used for this study were the FAO FRA coun- and carbon stocks and changes in the five different carbon pools of
try reports and global tables for 2005, 2010 and 2015. Table 1 spec- forest: aboveground biomass (AGB), belowground biomass (BGB),
ifies which sections from the country reports were used. National soil organic matter (carbon in mineral and organic soils, including
data on forest extent and forest monitoring capacities were peat), dead wood and litter, using different Tiers for estimating
extracted from the report sections on area of forest and other emission factors as set out in IPCC guidelines (IPCC, 2003, 2006,
wooded land; growing stock; and carbon stock (including Tiers 2014). Tier 1 uses the default emission factors provided by IPCC
as specified by FRA). Data for each individual country were com- for carbon stock change estimation. Tier 2 employs
piled in a database which allowed to make global maps and sys- country-specific data for the most important land uses and activi-
tematic global comparisons. ties, making use of forest volume or biomass values from existing
Additional sources of information were used to explain the sta- forest inventories and/or ecological studies. Tier 3 is most demand-
tus of current country capacities and to investigate what drives the ing and uses repeated measurements of trees from plots and/or
recent change in capacities. This information was found in (online) calibrated process models, tailored to address national circum-
reports and articles concerning capacity building initiatives related stances. For each indicator a score was calculated based on the
to tropical forest monitoring from bilateral initiatives (e.g. USAID, characteristics of the country, see Table 1. Because qualitative data
AUSAID, Norway, Japan), multilateral initiatives (UN-REDD, sources were used for this analysis, the indicator outcomes were
World Bank FCPF, FAO NFMA), international initiatives determined on an ordinal scale. This enables to make systematic
(COMIFAC) and by digging deeper into the individual country global comparisons and to observe relative differences between
reports. the 99 countries. There are a few points of consideration in assign-
ing ordinal values to the indicators. For example, forest area
change monitoring and remote sensing capacities were deemed
2.2. Methodology as good or very good when one or multiple forest cover maps were
produced by the country itself on a regular basis. This however, did
The countries’ national forest monitoring and reporting capaci- not take into account the quality of the maps, as there were no
ties and the change in capacities were assessed by examining three accuracy estimates available for all of these maps within our anal-
different indicators. For each of the 99 tropical Non-Annex I coun- ysis. Nevertheless, it does reflect the progress that a country has
tries data were compiled for the three indicators ‘‘Forest area made in performing forest area change assessments indepen-
change monitoring and remote sensing capacities’’, ‘‘Forest inven- dently. The same applies for the other two indicators. The analysis
tory capacities’’ and ‘‘Carbon pool reporting capacities’’. These are did not take into account the accuracy of the forest biomass and
similar to the indicators used in the study about assessment of carbon estimates in the forest inventories, but assignment of val-
national forest monitoring capacities of Non-Annex I countries in ues to indicators was based on the information that was available
the context of REDD+ by Romijn et al. (2012). The first indicator from the country reports.
reflects the capacities of a country to monitor forest area and forest In addition, the change in capacities for each of the indicators
area changes and its ability to produce forest cover and forest cover was calculated as follows:
change maps using time series of remote sensing data. The second
and third indicators reflect the capacities to perform a carbon stock Change 2005—2010 ¼ Indicator score 2010 Indicator score 2005
assessment. The focus of the second indicator is on performing a Change 2010—2015 ¼ Indicator score 2015 Indicator score 2010
national forest inventory for collecting data on forest species and
Table 1
Indicators used to assess a countries’ national forest monitoring capacities, the data sources that were used to gather information for each indicator, and the scoring system.
Global maps and tables were created with GIS and spreadsheet 3. Results and discussion
software, showing the capacities of the three indicators ‘‘Forest
area change monitoring and remote sensing capacities’’, ‘‘Forest 3.1. Forest area change monitoring and remote sensing capacities
inventory capacities’’ and ‘‘Carbon pool reporting capacities’’ for
the years 2005, 2010 and 2015, as well as the change in capacities Forest area change monitoring and remote sensing capacities
for the periods 2005–2010 and 2010–2015. Capacities were improved significantly between 2005 and 2015. Fig. 1 shows the
expressed in relation to the percentage of total tropical forest area forest area change monitoring and remote sensing capacities in
of the 99 countries that is monitored with this level of capacities. all 99 tropical Non-Annex I countries for the assessment years
Information from several capacity building programmes was gath- 2005, 2010 and 2015. Fig. 2 shows the changes in capacities for
ered in order to observe the effect of these programmes on the periods 2005–2010 and 2010–2015. In 2015, 54 countries
improvements of capacities. Reported FRA numbers of net change had good to very good capacities to monitor changes in forest area
in forest area (data from 2005, 2010 and 2015) for a similar period using remote sensing data, which means they were able to produce
(i.e. 2000–2005 or 2005–2010) were compared for countries with their own forest change maps. Overall, capacities in Latin American
increased capacities to see the actual effect of capacity improve- and South-East Asian countries are now very well developed. In
ments on reported numbers. Africa, capacities improved over time, with a few countries (DRC
(a) Forest area change monitoring and remote sensing capacities 2005
(b) Forest area change monitoring and remote sensing capacities 2010
(c) Forest area change monitoring and remote sensing capacities 2015
Fig. 1. Forest area change monitoring and remote sensing capacities in 2005 (a), 2010 (b) and 2015 (c). The outcomes, ranging from ‘‘low’’ to ‘‘very good’’, reflect the indicator
score for each of the 99 tropical Non-Annex I countries, based on analysis of FAO FRA country reports.
E. Romijn et al. / Forest Ecology and Management 352 (2015) 109–123 113
Fig. 2. Change in forest area change monitoring and remote sensing capacities between the years 2005–2010 (a) and 2010–2015 (b).
and Kenya) now having very good capacities to monitor changes in 3.2. Forest inventory capacities
forest area using remote sensing. Between 2005 and 2010 capaci-
ties improved slightly, with most changes taking place in Central Forest inventory capacities in 2005, 2010 and 2015 and changes
African countries. Between 2010 and 2015 considerable capacity in capacities between 2005–2010 and 2010–2015 are presented in
improvements took place, mainly in Central African countries Figs. 3 and 4. Indicator values for all 99 tropical Non-Annex I coun-
and Latin American countries. In many African countries the maps tries can be found in Appendix A. In 2015, 40 countries had good to
are now made by the country itself instead of by external research- very good capacities which means they performed national forest
ers and the countries in Latin America are now able to produce inventories on a regular basis. In South East Asian countries capac-
maps to assess forest area change on a more regular basis. ities were well developed, where 13 out of 17 countries had good
Between 2005 and 2015 the number of countries with low moni- or very good capacities in 2015. In Latin America and Africa, capac-
toring capacities reduced from 35 to 24 countries, while the num- ities were less well developed and considerable capacity improve-
ber of countries with good and very good capacities grew from 25 ments are needed before countries will be able to perform regular
to 31 and from 12 to 23 respectively (see Table 2). In 2015, one national forest inventories. Nevertheless, in these two continents
quarter of the countries still had low capacities. Especially in capacities have improved over time. Between 2005 and 2010
Africa, there is a large number of countries with low (17 countries), capacities improved especially in Africa. In many cases this was a
limited (6 countries) or intermediate (8 countries) capacities and change from limited to intermediate capacities, which means that
considerable improvements are needed. Appendix A contains the they produced more forest inventories, but these were still done by
indicator values of forest area change monitoring and remote sens- external researchers. From 2010 to 2015 increases in capacities can
ing capacities for all 99 tropical Non-Annex I countries. be seen in Latin American countries, Central African countries and
in Indonesia. Some of these countries had a medium or large
increase in capacities (i.e. Colombia, Ecuador, Zambia, Tanzania)
and now have good forest inventory capacities which means their
Table 2
inventories are done in-country instead of by external researchers.
Change in forest area change monitoring and remote sensing capacities between In a few cases, a decrease in capacities can be seen. This can be due
2005, 2010 and 2015. The numbers in the table refer to the number of countries to altering national circumstances or unstable conditions due to
which fall into each category for a certain year. the political situation in a country. The number of countries with
Forest area change monitoring 2005 2010 2015 low or limited capacities decreased from 63 in 2005 to 40 in
and RS capacities 2015 (see Table 3). The number of countries with intermediate
Low 35 31 24 capacities increased from 7 in 2005 to 19 in 2015 and the number
Limited 12 11 11 of countries with good capacities increased from 23 in 2005 to 31
Intermediate 15 14 10 in 2015. The number of countries with very good capacities
Good 25 30 31
increased only slightly, from 6 countries in 2005, to 8 countries
Very good 12 13 23
in 2010 to 9 countries in 2015.
114 E. Romijn et al. / Forest Ecology and Management 352 (2015) 109–123
3.3. Carbon pool reporting capacities reporting capacities. The largest group of countries had low to
intermediate capacities, as can be seen from Table 4. Half of the
Fig. 5 shows the carbon pool reporting capacities for the years countries (50) were ‘‘stranded’’ with intermediate capacities.
2005, 2010 and 2015. Fig. 6 shows the capacity changes between They were able to report on various carbon pools, at least on
the years 2005–2010 and 2010–2015. Indicator values for all 99 AGB and soil, but were not able to use Tier 2 or Tier 3 methods
tropical Non-Annex I countries can be found in Appendix A. for biomass conversion. A large number of countries (17), still
Overall, carbon pool reporting capacities improved over time. had low capacities and were not able to report on carbon in the five
However, most improvements imply changes from low, to limited different pools. This mainly concerns countries and small island
and intermediate capacities. This means countries started to report states in the Caribbean and Pacific. Also, a large number of coun-
on more carbon pools but still used Tier 1 emission factors. tries (17) had limited carbon pool reporting capacities and were
In many African countries, modest improvements took place only able to report AGB at Tier 1 level. An example of a country that
between 2005 and 2010. This was a change from only reporting steadily improved over time is Mexico. The country had low capac-
AGB carbon stocks, to reporting AGB and soil carbon stocks, but ities in 2005, increased to intermediate capacities in 2010 and to
still using Tier 1 methods. Similar modest improvements took very good capacities in 2015. In 2005, no carbon stocks were
place in a few Latin American and South East Asian countries reported. In 2010 AGB, BGB and soil carbon stocks were reported
between 2005 and 2010. From 2010 to 2015 more improvements at Tier 1 and in 2015, AGB, BGB and dead wood carbon stocks were
took place. In 2015, 14 countries had very good carbon pool reported at Tier 3.
Fig. 3. Forest inventory capacities in 2005 (a), 2010 (b) and 2015 (c). The outcomes, ranging from ‘‘low’’ to ‘‘very good’’, reflect the indicator score for each of the 99 tropical
Non-Annex I countries, based on analysis of FAO FRA country reports.
E. Romijn et al. / Forest Ecology and Management 352 (2015) 109–123 115
Fig. 4. Change in forest inventory capacities between the years 2005–2010 (a) and 2010–2015 (b).
Table 3
Change in forest inventory capacities between 2005, 2010 and 2015. The numbers in
a wall-to-wall mapping and analysis of its forests and land cover
the table refer to the number of countries which fall into each category for a certain for three epochs: 1990, 2000 and 2010; using ALOS AVNIR and
year. Landsat images. This allowed the country to map forest and land
cover changes for the periods 1990–2000, 2000–2010 and 1990–
Forest inventory capacities 2005 2010 2015
2010. The analysis was conducted by a team comprising of repre-
Low 31 22 21
sentatives from the Kenya Forest Service, Department of Resource
Limited 32 29 19
Intermediate 7 17 19 Survey and Remote Sensing with the technical back-stopping of
Good 23 23 31 PASCO consulting company (Pasco Consultants, 2013).
Very good 6 8 9 Indonesia is currently revisiting its national land cover change
assessment and is extending the analysis which started in 1990,
to 2013 (Indonesian Ministry of Forestry, 2013). The land cover
information was based on visual interpretation of mosaic Landsat
Overall, progress has been made, but reporting is still very rudi- TM/ETM/LDCM data generated by the Ministry of Forestry of
mentary. Many countries (84) were reporting at Tier 1 level in Indonesia (Romijn et al., 2013). The country now has more than
2015. Several studies showed that Tier 1 does not adequately rep- two decades of deforestation and forest cover change estimates,
resent national circumstances and may have uncertainties of up to and subsequently is capable of analyzing drivers of deforestation
±70% from the mean (Meridian institute, 2009). At least for signif- and forest degradation over these periods. This explains why
icant pools such as AGB and BGB, reporting should be done at Indonesia was classified among countries with very good capacity
higher Tiers which use allometric equations or models that are for monitoring forest area change and remote sensing competence.
specific for the biomes and tree species in the country and have The slight increase in forest inventory capacities from 2010 to 2015
lower uncertainties (Wertz-Kanounnikoff et al., 2008; Baker was mainly due to improvements of national forest inventory (NFI)
et al., 2010; GOFC-GOLD, 2014). sampling design. The sampling density increased to more than
3000 inventory plots across the entire nation. In the near future,
3.4. How can the change in forest monitoring and reporting capacities Indonesia is ready to report forest reference emissions levels
be explained? (FRELs) not only for the AGB pool but also for other carbon pools
(Krisnawati et al., 2014). Indonesia’s first national FREL submission,
3.4.1. A few country cases incorporating the refined NFI product, was planned during the
The ‘‘very large’’ improvement in forest area change monitoring UNFCCC Conference of Parties (COP) in Lima, in December 2014.
and remote sensing capacities in Kenya can be explained by recent In some cases, a decrease in capacities occurred between the
acquisition and analysis of detailed satellite imagery. In 2013, FRA reporting years. Reasons may vary, but some reoccurring
Kenya, through support from the Government of Japan, completed motives were found in this study.
116 E. Romijn et al. / Forest Ecology and Management 352 (2015) 109–123
Both Cameroon and Congo showed a decline in forest inventory soil pool was omitted from all the analyses. The reason for this is
capacities between 2010 and 2015. The decrease was from ‘‘very unclear, but again this example shows that maintaining capacities
good’’ in 2010 to ‘‘good’’ in 2015. The 2010 FRA country reports is very important to ensure consistent updates for the five year
showed that both countries performed regular national forest reporting cycle.
inventories, with the most recent inventory less than five years
before. However, in the 2015 report no new inventories were men- 3.4.2. Capacity building initiatives for improving national forest
tioned, so it was classified as ‘‘good’’. This indicates the risk that monitoring capacities
countries face when they are not maintaining their forest monitor- Table 5 gives an indication of the effectiveness of the different
ing system and update their inventories regularly. Once a country international capacity building initiatives related to forest moni-
has good capacities, it needs to keep investing in the national forest toring. It shows which initiatives are present in the countries
monitoring programme in order to maintain its capacities. where improvements have taken place for performing forest area
Another obvious decline concerned the decrease in carbon pool change assessments and national forest inventories, but also which
reporting capacities from ‘‘intermediate’’ to ‘‘limited’’ in Cape initiatives are present in the countries without improvements.
Verde (2010–2015) and Panama (2005–2010). In the first year, in Please note that the total number of countries does not add up to
Cape Verde, carbon was reported for AGB, BGB, litter and soil. In 99 because a few countries already had very good capacities in
Panama carbon was reported for all pools. However, in the second 2005, 2010 and 2015 for both assessing change in forest area and
year, carbon was reported for fewer pools in each country, and the for performing a forest inventory. Therefore, China, India,
Fig. 5. Carbon pool reporting capacities in 2005 (a), 2010 (b) and 2015 (c). The outcomes, ranging from ‘‘low’’ to ‘‘very good’’, reflect the indicator score for each of the 99
tropical Non-Annex I countries, based on analysis of FAO FRA country reports.
E. Romijn et al. / Forest Ecology and Management 352 (2015) 109–123 117
Fig. 6. Change in carbon pool reporting capacities between the years 2005–2010 (a) and 2010–2015 (b).
Table 4 Table 5
Change in carbon pool reporting capacities between 2005, 2010 and 2015. The International initiatives to support capacity development with respect to national
numbers in the table refer to the number of countries which fall into each category for forest monitoring in tropical Non-Annex I countries. For each initiative the
a certain year. number and percentage of countries that showed improvements and the number
and percentage of countries that did not show improvements are indicated in the
Carbon pool reporting capacities 2005 2010 2015 table.
Low 28 17 17
Countries with Countries without
Limited 56 26 17
improvements improvements
Intermediate 12 54 50
Good 1 1 1 Number Percentage Number Percentage
Very good 2 1 14 (%) (%)
Total number of countries 51 54 43 46
UN-REDD national 15 79 4 21
Malaysia, Mexico and Myanmar were not included in this table.
programmes
Malaysia, Mexico and Myanmar are UN-REDD partner countries UN-REDD partner 16 57 12 43
and Mexico is also a WB FCPF REDD+ country participant. countries
Targeted forest monitoring programmes like the FAO NFMA FAO NFMA project 6 86 1 14
project (Saket et al., 2010) seem to be most effective: 86% of the completed
FAO NFMA project 8 67 4 33
participating countries improved their national forest monitoring ongoing
capacities. For countries where the FAO NFMA project is still ongo- FAO Capacity Building for 6 55 5 45
ing, already 67% of the countries showed improvements and more REDD+ NFMS
results can be expected at a later stage. 55% of the countries sup- Support from FAO only 8 – – –
mentioned in country
ported by FAO capacity building for REDD+ NFMS (FAO, 2014b,
reportsa
accessed online) showed improvements. The UN-REDD national WB FCPF REDD+ country 27 61 17 39
programme (UN-REDD, 2014, accessed online) is also very effec- participant
tive; 79% of the participating countries showed capacity improve- WB FCPF REDD+ country 4 80 1 20
ments. On the contrary, UN-REDD partner countries did not show candidates
Other support mentioned 10 – – –
this obvious success. However, these countries are new within in country reports (e.g.
the UN-REDD programme and may request to receive funding for bilateral/COMIFAC)a
a national programme in the future. World Bank (WB) FCPF Total number of initiatives 100 69 44 31
REDD+ programme (WB FCPF, 2014, accessed online) is supporting
a
49 countries in building up their capacities; 61% of the country par- For these two categories, the column ‘‘countries without improvements’’
remains blank, as the information was taken from the FRA reports for countries that
ticipants and 80% of the country candidates showed progress in showed an improvement in capacities.
their monitoring capacities. These countries receive support
118 E. Romijn et al. / Forest Ecology and Management 352 (2015) 109–123
Fig. 7. Percentage of total tropical forest area of the 99 countries that is monitored with ‘‘low’’, ‘‘limited’’, ‘‘intermediate’’, ‘‘good’’ or ‘‘very good’’ forest area change monitoring
and remote sensing capacities (A) and forest inventory capacities (B) in the assessment years 2005, 2010 and 2015.
Fig. 9. Difference in reported numbers of annual net change in forest area for the periods 2000–2005 (A) and 2005–2010 (B) in 1000 ha/yr between two different reporting
years (2005 and 2010; 2005 and 2015; 2010 and 2015). The difference is indicated for countries that showed an increase in forest area change monitoring and remote sensing
capacities between 2005–2010 and 2010–2015.
reported for the same time period (for example 2000–2005). The 4. Conclusions
figure depicts the change that happened in reported numbers of
net change in forest area for a similar time period between two dif- Major improvements can be seen in forest area change monitor-
ferent reporting years. It only includes the countries with capacity ing capacities and in forest inventory capacities. The total tropical
improvements. The dark gray bars in the figure sum up the total forest area that is monitored with good to very good forest area
negative change between two reporting years for all countries change monitoring and remote sensing capacities increased from
included. The light gray bars in the figure sum up the total positive 69% (1435 million ha) in 2005 to 83% (1699 million ha) in 2015.
change between two reporting years for all countries included. Fifty-four of the 99 countries now have good to very good forest
For example when looking at the net change in forest area area change monitoring and remote sensing capacities. Free and
between 2005 and 2010 (two columns at the right side of the fig- open source high resolution satellite data such as Landsat remain
ure), there is a difference in reported numbers of net change for an important data source for assessing historical forest cover
this period in the FRA 2010 and FRA 2015 reporting. All countries change and have been an asset to allow countries to take the steps
that had negative change in net change in forest area between for improving their national forest monitoring. The total tropical
those two reporting years together had a difference of forest area that is monitored with good to very good forest inven-
433.000 ha/yr and all the countries that had a positive change tory capacities increased from 38% (785 million ha) in 2005 to 66%
in net change in forest area between those two reporting years (1350 million ha) in 2015. This concerns 40 countries with good to
together had a difference of +651.000 ha/yr. very good capacities. Continued capacity building investments are
The figure clearly shows that reported numbers tend to be more needed to ensure that the remaining countries will be able to accu-
positive (increasing forest area) when capacities increase. This rately monitor the tropical forest area with sufficient level of
indicates a trend that countries with lower capacities in the past capacity. Carbon pool reporting capacities did not increase as dra-
had the tendency to overestimate the net area of forest loss. The matically and at this moment fifteen countries have good to very
reason for this trend is not very clear but it points at an important good capacities and are able to report on one or more carbon pools
issue that national estimations based on low quality data and using Tier 2 or Tier 3 methods. The majority of the countries report
expert judgements do not balance in their effects of over- or under- at Tier 1, which may contain large uncertainties up to 70%. Priority
estimating the forest area change but can result in biases in forest now needs to go to improving carbon pool reporting. More
loss estimations in large area assessments. This effect is most pro- country-specific data of carbon stocks are needed to be able to
nounced for the previous estimates from FRA 2005 where the report at Tier 2 or Tier 3 in order to represent the specific biomes
investments in better data and national capacities resulted in and tree species and to have lower uncertainties. In most countries,
lower forest loss numbers in FRA 2010 and FRA 2015 which are the data for reporting on carbon in different pools are present from
based on more robust data. forest inventories, however more capacity building and training is
120 E. Romijn et al. / Forest Ecology and Management 352 (2015) 109–123
needed so countries can actually perform the reporting. Some building programmes and the need for further capacity develop-
countries showed a decrease in capacities within a five-year cycle. ment. Further investments will enable countries to obtain accurate
It is important for countries to keep the level of acquired capacities and reliable data and information on forest area and forest
by maintaining their forest monitoring system and updating their resources which provides the necessary input to refine policies
maps inventories on a regular basis. and decisions to track drivers of deforestation, to conserve forests
Increased capacities had an effect on the annual net change rate and to further improve forest management.
in forest area that was reported in the FRA. Overall, there was a
positive change in the net change in forest area that was reported Acknowledgements
between two reporting years for a similar time period. This means
that countries with lower capacities in the past had the tendency The authors gratefully acknowledge the Food and Agriculture
to overestimate the net area of forest loss and that use of low qual- Organization of the United Nations for providing access to the glo-
ity data and expert judgements resulted in biases in forest loss esti- bal Forest Resources Assessment (FRA) data from 2015 at an early
mations in large area assessments. stage to enable the performance of this FRA-based research. The
The results demonstrate that capacity building programmes study was supported by NORAD (grant agreement no.
have proven to be successful. Countries which increased in capac- QZA-10/0468) and AusAID (grant agreement no. 46167) within
ities participated in more capacity building initiatives than coun- CIFOR’s Global Comparative Study on REDD+. This work was fur-
tries that did not increase in capacities. Targeted programmes, ther supported by the German Federal Ministry for the
such as those from FAO projects seem to be very effective with a Environment, Nature Conservation and Nuclear Safety (BMU)
success rate of 86%. Also, the engagement in REDD+ capacity devel- International Climate Initiative (IKI) through the project ‘‘From
opment initiatives had a positive impact on country forest moni- Climate Research to Action under Multilevel Governance:
toring capacity. This clearly shows the importance of capacity Building Knowledge and Capacity at Landscape Scale.’’
Appendix A
This table contains the values for the three indicators ‘‘Forest area change monitoring and remote sensing capacities’’, ‘‘Forest inventory
capacities’’ and ‘‘Carbon pool reporting capacities’’ for all 99 tropical Non-Annex I countries of this study. The codes correspond to the fol-
lowing indicator values: – – Low; – Limited; 0 Intermediate; + Good; ++ Very good
Country Forest area change Forest inventory capacities Carbon pool reporting
monitoring & remote capacities
sensing capacities
2005 2010 2015 2005 2010 2015 2005 2010 2015
Angola –– –– –– – – – – 0 0
Antigua and Barbuda –– –– –– –– –– –– –– –– ––
Argentina + + ++ + + ++ 0 0 0
Bahamas –– –– –– – – – –– –– ––
Bangladesh – + ++ + + + – – 0
Belize + + + –– – – – – –
Benin – – 0 –– + + –– 0 0
Bhutan 0 0 + + + + 0 0 0
Bolivia ++ ++ ++ + + + – 0 0
Botswana + + + – – – – 0 0
Brazil ++ ++ ++ 0 0 0 + + ++
Burkina Faso –– –– – 0 0 0 – 0 0
Burundi –– –– –– –– 0 0 –– 0 0
Cambodia + + + – – – – 0 0
Cameroon 0 0 0 + ++ + – 0 0
Cape Verde –– –– –– – – + – 0 –
Central African Republic –– –– – – – – – 0 0
Chad –– –– –– – – – – 0 0
China ++ ++ ++ ++ ++ ++ – – –
Colombia ++ ++ ++ –– – + – – –
Comoros – – – + + + 0 0 ++
Congo + + + + ++ + – 0 0
Costa Rica ++ ++ + – – + – – ++
Côte d’Ivoire + + + 0 0 0 0 0 0
Cuba –– –– –– + + + ++ – ++
Democratic Republic of the Congo 0 + ++ – 0 0 0 0 0
Dominica – – – –– –– –– –– –– ––
Dominican Republic –– + + –– –– –– – – 0
Ecuador + + ++ –– –– + –– –– ––
El Salvador + + + –– –– –– –– –– ––
Equatorial Guinea –– –– –– – – – – 0 0
E. Romijn et al. / Forest Ecology and Management 352 (2015) 109–123 121
Appendix A (continued)
Country Forest area change Forest inventory capacities Carbon pool reporting
monitoring & remote capacities
sensing capacities
2005 2010 2015 2005 2010 2015 2005 2010 2015
Eritrea –– –– –– –– –– –– –– –– ––
Ethiopia + + + –– –– –– – 0 0
Fiji –– –– –– –– –– –– –– –– ––
Gabon –– –– –– – – – – 0 0
Gambia – – – – – + – 0 0
Ghana + + + – 0 0 – 0 ++
Guatemala ++ ++ ++ – – – – – ++
Guinea –– –– –– –– 0 0 – – –
Guinea-Bissau 0 0 0 –– –– –– – 0 0
Guyana –– –– ++ –– – 0 – 0 ++
Haiti –– –– –– –– –– –– 0 0 0
Honduras –– –– –– –– – – –– – –
India ++ ++ ++ ++ ++ ++ ++ ++ ++
Indonesia ++ ++ ++ + + ++ – – –
Jamaica + + + – – 0 0 0 ++
Kenya –– –– ++ – 0 + – 0 0
Lao People’s Democratic Republic + + + + + + – – –
Lesotho 0 0 0 –– –– –– –– 0 0
Liberia 0 0 0 – – – – 0 0
Madagascar 0 0 + + + + – 0 0
Malawi – – 0 – – – – 0 0
Malaysia ++ ++ ++ ++ ++ ++ – – –
Mali –– –– –– 0 0 0 – 0 0
Mauritania –– –– –– – – – – 0 0
Mauritius –– –– –– – – – – 0 0
Mexico ++ ++ ++ ++ ++ ++ –– 0 ++
Micronesia (Federated States of) –– + + – 0 0 –– – –
Mozambique 0 – – – – – – 0 0
Myanmar ++ ++ ++ ++ ++ ++ – – +
Namibia – 0 0 + + + – 0 0
Nepal + + + + + + 0 – ++
Nicaragua + + + – + + – – ++
Niger –– –– –– 0 0 0 0 0 0
Nigeria –– –– –– – – – – 0 0
Pakistan –– –– – – – – – – –
Palau 0 0 0 – 0 0 –– – ++
Panama + ++ ++ + + + 0 – –
Papua New Guinea 0 0 + 0 0 + –– – –
Paraguay 0 0 + –– –– –– –– –– ––
Peru ++ ++ ++ + + + –– 0 0
Philippines –– + ++ ++ + + – 0 0
Rwanda –– –– + + + + 0 0 0
Saint Lucia 0 0 0 0 0 0 –– –– ––
Saint Vincent and the Grenadines –– – – –– –– –– –– –– ––
Samoa + + + –– –– –– –– –– ––
Sao Tome and Principe –– –– –– – – 0 – 0 0
Senegal + + + + + + – – 0
Sierra Leone –– –– –– –– – –– –– 0 0
Singapore –– –– – –– –– –– –– –– ––
Solomon Islands –– –– –– –– 0 0 –– – –
Somalia –– –– –– –– –– –– – – –
South Africa + + + –– –– –– – 0 0
Sri Lanka + + ++ + + ++ – – –
Sudan 0 0 + – 0 0 – 0 0
Suriname + + ++ – – 0 – 0 0
Swaziland – – – + + + – 0 0
Thailand + + + + + + – – –
Appendix A (continued)
Country Forest area change Forest inventory capacities Carbon pool reporting
monitoring & remote capacities
sensing capacities
2005 2010 2015 2005 2010 2015 2005 2010 2015
Timor-Leste + + + –– –– –– –– –– ––
Togo –– –– –– –– –– –– –– –– ––
Trinidad and Tobago + + + + + + – 0 0
Uganda – + + + + + – 0 0
United Republic of Tanzania 0 0 + – – + – 0 0
Uruguay + + + –– –– – –– –– ––
Vanuatu – – – –– –– –– –– –– ––
Venezuela + + ++ –– –– –– –– –– ––
Viet Nam – – ++ + ++ ++ 0 0 0
Zambia – – + – – + – 0 ++
Zimbabwe 0 0 0 – 0 0 – 0 0
References Herold, M., Johns, T., 2007. Linking requirements with capabilities for deforestation
monitoring in the context of the UNFCCC-REDD process. Environ. Res. Lett. 2.
Achard, F., Eva, H.D., Mayaux, P., Stibig, H.-J., Belward, A., 2004. Improved estimates http://dx.doi.org/10.1088/1748-9326/2/4/045025.
of net carbon emissions from land cover change in the tropics for the 1990s. Herold, M., Skutsch, M., 2011. Monitoring, reporting and verification for national
Global Biogeochem. Cycles 18. http://dx.doi.org/10.1029/2003GB002142. REDD+ programmes: two proposals. Environ. Res. Lett. 6, 1–10.
Achard, F., DeFries, R., Eva, H., Hansen, M., Mayaux, P., Stibig, H.-J., 2007. Pan- Hosonuma, N., Herold, M., De Sy, V., De Fries, R.S., Brockhaus, M., Verchot, L.,
tropical monitoring of deforestation. Environ. Res. Lett. 2. http://dx.doi.org/ Angelsen, A., Romijn, E., 2012. An assessment of deforestation and forest
10.1088/1748-9326/2/4/045022. degradation drivers in developing countries. Environ. Res. Lett. 7, 1–12.
Baker, D.J., Richards, G., Grainger, A., Gonzalez, P., Brown, S., DeFries, R., Held, A., Houghton, R.A., 2005. Tropical deforestation as a source of greenhouse gas
Kellndorfer, J., Ndunda, P., Ojima, D., Skrovseth, P., Souza Jr., C., Stolle, F., 2010. emissions. In: Moutinho, P., Schwartzman, S. (Eds.), Tropical Deforestation
Achieving forest carbon information with higher certainty: a five-part plan. and Climate Change. IPAM – Instituto de Pesquisa Ambiental da Amazônia,
Environ. Sci. Policy 13, 249–260. Environmental Defense, Belém, Pará, Brazil, Washington DC – USA (Chapter 1).
Bonan, G.B., 2008. Forests and climate change: forcings, feedbacks, and the climate Indonesian Ministry of Forestry, 2013. Digital Land Cover Map of Indonesia.
benefits of forests. Science 320, 1444–1449. Directorate General of Forestry Planning – Ministry of Forestry of Indonesia.
Cramer, W., Bondeau, A., Schaphoff, S., Lucht, W., Smith, B., Sitch, S., 2004. Tropical <http://nfms.dephut.go.id/ipsdh/> (accessed 05.11.14).
forests and the global carbon cycle: impacts of atmospheric carbon dioxide, IPCC, 2003. Good practice guidance for land use, land-use change and forestry. In:
climate change and rate of deforestation. Phil. Trans. R. Soc. Lond. B 359, 331– Penman, J., Gytarsky, M., Hiraishi, T., Krug, T., Kruger, D., Pipatti, R., Buendia, L.,
343. Miwa, K., Ngara, T., Tanabe, K., Wagner, F. (Eds.). Prepared by the National
De Sy, V., Herold, M., Achard, F., Asner, G.P., Held, A., Kellndorfer, J., Verbesselt, J., Greenhouse Gas Inventories Programme. IGES, Japan.
2012. Synergies of multiple remote sensing data sources for REDD+ monitoring. IPCC, 2006. 2006 IPCC guidelines for national greenhouse gas inventories. In:
Curr. Opin. Environ. Sustainab. 4 (6), 696–706. Eggleston, H.S., Buendia, L., Miwa, K., Ngara, T., Tanabe, K. (Eds.), Prepared by the
DeFries, R.S., Houghton, R.A., Hansen, M.C., Field, C.B., Skole, D., Townshend, J., 2002. National Greenhouse Gas Inventories Programme. IGES, Japan.
Carbon emissions from tropical deforestation and regrowth based on satellite IPCC, 2014. 2013 Supplement to the 2006 IPCC guidelines for national greenhouse
observations for the 1980s and 1990s. PNAS 99 (22), 14256–14261. gas inventories. In: Hiraishi, T., Krug, T., Tanabe, K., Srivastava, N., Baasansuren,
DeFries, R., Achard, F., Brown, S., Herold, M., Murdiyarso, D., Schlamadinger, B., de J., Fukuda, M., Troxler, T.G. (Eds), Wetlands. IPCC, Switzerland.
Souza Jr., C., 2007. Earth observations for estimating greenhouse gas emissions Keenan, R.J., Reams, G., De Freitas, J., Lindquist, E., Achard, F., Hirata, Y., Odeke, D.E.,
from deforestation in developing countries. Environ. Sci. Policy 10 (4), 385–394. Grainger, A., 2015. Dynamics of global forest area: results from the 2015 UN
FAO, 2014a. State of the World’s Forests. Enhancing the Socioeconomic Benefits FAO global forest resource assessment. For. Ecol. Manage. 352, 9–20.
from Forests. Food and Agriculture Organization of the United Nations, Rome, Krisnawati, H., Adinugroho, W.C., Imanuddin, R., Hutabarat, S., 2014. Forest Biomass
Italy. Estimate for Predicting CO2 Emissions in Central Kalimantan Indonesia. Forest
FAO, 2014b. Capacity Building for REDD+ NFMS. <http://www.fao.org/forestry/ Research and Development Agency, Ministry of Forestry of Indonesia, Bogor,
nfms-for-redd/85205/en/> (accessed 26.11.14). Indonesia, 36pp..
FAO, 2015. Global Forest Resources Assessment 2015. Food and Agriculture MacDicken, K., 2015. Global Forest Resources Assessment 2015: What, why and
Organization, Rome, Italy. how? For. Ecol. Manage. 352, 3–8.
FAO & JRC, 2012. Global Forest Land-Use Change 1990–2005, by E.J. Lindquist, R. Mayaux, P., Holmgren, P., Achard, F., Eva, H., Stibig, H.-J., Branthomme, A., 2005.
D’Annunzio, A. Gerrand, K. MacDicken, F. Achard, R. Beuchle, A. Brink, H.D. Eva, Tropical forest cover change in the 1990s and options for future monitoring.
P. Mayaux, J. San-Miguel-Ayanz, H.-J. Stibig. FAO Forestry Paper No. 169. Food Phil. Trans. R. Soc. B 360, 373–384.
and Agriculture Organization of the United Nations and European Commission Meridian Institute, 2009. Reducing Emissions from Deforestation and Forest
Joint Research Centre. Rome, Italy. <http://www.fao.org/docrep/017/i3110e/ Degradation (REDD): An Options Assessment Report. Prepared for the
i3110e.pdf> (accessed 25.22.14). Government of Norway, by Angelsen, A., Brown, S., Loisel, C., Peskett, L.,
Foley, J.A., DeFries, R., Asner, G.P., Barford, C., Bonan, G., Carpenter, S.R., Chapin, F.S., Streck, C., Zarin, D. <http://www.REDD-OAR.org>.
Coe1, M.T., Daily, G.C., Gibbs, H.K., Helkowski, J.H., Holloway, T., Howard, E.A., Miura, S., Amacher, M., Hofer, T., San Miguel, S., Ernawati, Thackway, R., 2015.
Kucharik, C.J., Monfreda, C., Patz, J.A., Prentice, I.C., Ramankutty, N., Snyder, P.K., Protective functions and ecosystem services of global forests in the past
2005. Global consequences of land use. Science 309, 570–574. quarter-century. For. Ecol. Manage. 352, 35–46.
Giri, C., Pengra, B., Long, J., Loveland, T.R., 2013. Next generation of global land cover Pasco Consultants, 2013. Report on National Forest Resource Mapping and Capacity
characterization, mapping, and monitoring. Int. J. Appl. Earth Obs. Geoinf. 25, Development (NFRMCD) For The Republic of Kenya, vol. 2.
30–37. http://dx.doi.org/10.1016/j.jag.2013.03.005. Pielke Sr., R.A., 2005. Land use and climate change. Science 310, 1625–1626.
GOFC-GOLD, 2014. A Sourcebook of Methods and Procedures for Monitoring and Romijn, E., Herold, M., Kooistra, L., Murdiyarso, D., Verchot, L., 2012. Assessing
Reporting Anthropogenic Greenhouse Gas Emissions and Removals Associated capacities of non-Annex I countries for national forest monitoring in the context
with Deforestation, Gains and Losses of Carbon Stocks in Forests Remaining of REDD+. Environ. Sci. Policy 19–20, 33–48.
Forests, and Forestation. GOFC-GOLD Report Version COP20-1. GOFC-GOLD Romijn, E., Ainembabazi, J.H., Wijaya, A., Herold, M., Angelsen, A., Verchot, L.,
Land Cover Project Office, Wageningen University, The Netherlands. Murdiyarso, D., 2013. Exploring different forest definitions and their impact on
Hansen, M.C., Stehman, S.V., Potapov, P.V., 2010. Quantification of global gross developing REDD+ reference emission levels: a case study for Indonesia.
forest cover loss. PNAS 107 (19), 8650–8655. Environ. Sci. Policy 33, 246–259.
Hansen, M.C., Potapov, P.V., Moore, R., Hancher, M., Turubanova, S.A., Tyukavina, A., Roy, D.P., Ju, J., Mbow, C., Frost, P., Loveland, T., 2010. Accessing free Landsat data via
Thau, D., Stehman, S.V., Goetz, S.J., Loveland, T.R., Kommareddy, A., Egorov, A., the Internet: Africa’s challenge. Remote Sens. Lett. 1 (2), 111–117. http://
Chini, L., Justice, C.O., Townshend, J.R.G., 2013. High-resolution global maps of dx.doi.org/10.1080/01431160903486693.
21st-century forest cover change. Science 342, 850–853.
E. Romijn et al. / Forest Ecology and Management 352 (2015) 109–123 123
Saket, M., Branthomme, A., Piazza, M., 2010. FAO NFMA—support to developing WB FCPF, 2014. REDD+ Countries. <https://www.forestcarbonpartnership.org/redd-
countries on national forest monitoring and assessment. In: Tomppo, E., countries-1> (accessed 26.11.14).
Gschwantner, T., Lawrence, M., McRoberts, R.E. (Eds.), National Forest Wertz-Kanounnikoff, S., Verchot, L.V., Kanninen, M., Murdiyarso, D., 2008. How can
Inventories—Pathways for Common Reporting. Springer, New York, pp. 583–594. we monitor, report and verify carbon emissions from forests? In: Angelsen, A.
Salvini, G., Herold, M., De Sy, V., Kissinger, G., Brockhaus, M., Skutsch, M., 2014. How (Ed.), Moving Ahead with REDD: Issues Options and Implications. CIFOR, Bogor,
countries link REDD+ interventions to drivers in their readiness plans: Indonesia, pp. 87–98 (Chapter 9).
implications for monitoring systems. Environ. Res. Lett. 9 (7), 074004. Wulder, M., Masek, J.G., Cohen, W.B., Loveland, T.R., Woodcock, C.E., 2012. Opening
Tulyasuwan, N., Henry, M., Secrieru, M., Jonckheere, I., Federici, S., 2012. Issues and the archive: how free data has enabled the science and monitoring promise of
challenges for the national system for greenhouse gas inventory in the context Landsat. Remote Sens. Environ. 122, 2–10. http://dx.doi.org/10.1016/
of REDD+. Greenhouse Gas Meas. Manage. 2 (2–3), 73–83. j.rse.2012.01.010.
UN-REDD, 2014. UN-REDD Programme Regions and Partner Countries. <http://
www.un-redd.org/Partner_Countries/tabid/102663/Default.aspx> (accessed
26.11.14).
Forest Ecology and Management 352 (2015) 124–133
a r t i c l e i n f o a b s t r a c t
Article history: There is strong interest in gaining an informed view of changes likely to occur in forest area and the
Received 2 December 2014 impacts of these changes on production forestry and forest conservation. Despite the complexity of
Received in revised form 19 February 2015 underlying causes, it is largely accepted that deforestation is mainly focused in the tropics and driven
Accepted 8 March 2015
by conversion to agriculture. Similarly, energy demand and GDP are largely determining wood consump-
Available online 7 September 2015
tion and production. Based on these assumptions, we built a model predicting natural forests and planted
forests’ evolution in the next 15 years, and compared the results of the modelling with survey results
Keywords:
from country expertise. The results suggest that on a global level, forest resources loss is likely to slow
Forest area projection
Spatial modelling
down. The forests that are most at risk of conversion were clearly identified within the tropical domain,
Agricultural expansion while the forest under protected areas showed very little risk of being converted to other land uses in the
Wood consumption near future.
Natural forests Ó 2015 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
Planted forests creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction et al., 2010). The underlying causes determine the degree of direct
causes resulting in land-use change.
The world’s population is growing rapidly: the UN predicts a Despite the complexity of deforestation causes, it is generally
15% population increase in the next 15 years to a total of 8.4 billion accepted that deforestation is primarily occurring in the tropics
people (UN 2012). Per capita consumption is increasing as well, (FAO and JRC, 2012) and the largest direct cause of deforestation is
especially in fast-growing economies, resulting in an unprece- agricultural expansion as 70–95% of forests lost in the tropics are
dented demand for resources. In response to increased forest loss converted to agriculture (Holmgren, 2006; Hosonouma et al., 2012).
over the past decades, international decisions have set global Hosonouma et al. (2012) used information reported in REDD+
targets like the Aichi targets for biodiversity (CBD, 2010) and Readiness Preparation Proposals from various countries and the
incentives to reduce emissions from deforestation and forest Global Forest Resource Assessment (FRA) (FAO, 2010) to suggest
degradation in developing countries are being negotiated that agriculture (cropland and pasture) is by far the largest direct
(UNFCCC, 2014). An informed vision on future forest area dynamics cause of deforestation; according to their estimations between
may help guide and prioritize international decisions aimed at 70% and 80% of forest conversion is to agriculture in Africa, around
reducing forest loss; this paper explores projected forest area 70% in subtropical Asia and >90% in Latin America. Other studies
change and its potential effect on the production and conservation equally indicate agricultural expansion as the largest direct cause
functions of forests towards 2030. of deforestation in Africa, Asia and Latin America (Nepstad et al.,
Deforestation is the result of many processes driven by multiple 2008; Guitierrez-Velez et al., 2011).
causes. We can distinguish underlying and direct causes of land Forest gains, on the other hand, are driven by two main factors:
conversion; underlying causes can include economic development, natural forest regrowth on abandoned agricultural land (Baumann
demographic trends and technology factors, and direct causes can et al., 2011) and tree planting for consumption, either as timber
include cropland, pasture land or urban development expanding (Antweiler et al., 2012) or energy wood. Many studies suggest that
on, and replacing, forest land (Geist and Lambin, 2001; Smith wood is indeed increasingly used as an energy source at the global
level, not only in developing countries (Smeets et al., 2007; IEA,
2011) but also in developed economies (UNECE and FAO, 2009;
q
This article is part of a special issue entitled ‘‘Changes in Global Forest Resources USEIA, 2014). As a consequence, the regional and global patterns
from 1990 to 2015’’. of wood production have changed in the last few decades, with a
⇑ Corresponding author. Tel.: +44 7 9464 35 698.
rapid and significant increase in the area of planted forests and
E-mail address: remi.dannunzio@fao.org (R. d’Annunzio).
http://dx.doi.org/10.1016/j.foreco.2015.03.014
0378-1127/Ó 2015 Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
R. d’Annunzio et al. / Forest Ecology and Management 352 (2015) 124–133 125
the growing importance of these resources for wood supply (Carle comprised of both primary forest and naturally regenerated forest
and Holmgren, 2008; Whiteman, 2014). However, technological and planted forests are established through planting and/or delib-
development in wood processing and the use of other bio-energy erate seeding.
sources does not imply that the increased energy demand results Three sets of data on forest area change, compliant with the FAO
in an equal increase in wood demand. For instance, Buongiorno definitions, were used to determine and discuss potential forest
and Zhu (2014) looked at changes in technology and found that projection towards 2030:
wastepaper is increasingly used to replace virgin fibre in pulp
and paper production, buffering the effect of increased global – A tabular data country-based model that produces regional and
bio-energy demand on global wood demand. global trends (explained in 2.2–2.5),
In order to capture these different patterns of losses and gains – a spatially explicit model using the quantitative projections
of forest in the next 15 years, we built a model based on several from the first model and spatially allocating these losses on a
hypotheses: the first is that the changes occurring in agricultural global forest map based on a historical trend analysis of loca-
land and natural forest land are meaningfully correlated. This is tions of past forest loss (2.6),
despite the fact that magnitude of changes occurring in agricul- – a set of country-specific predictions provided as expert judg-
tural land and natural forest land are not directly and linearly com- ment for the 2015 Global Forest Resource Assessment (FRA
parable. Our second hypothesis is that we can use wood 2015) user survey (2.7).
consumption projections to build a model predicting planted forest
projections. To test the hypotheses we looked at the correlation 2.1. Forest projection model: GFRM
between past natural forest area change and past arable land area
change on the one hand, and wood consumption and planted for- The Global Forest Resources Model (GFRM), developed in this
ests area on the other hand. study, is based on tabular data per country and projects forest area
Further on, forest area change per country was projected based change using exogenous projections of arable land and wood
on a historical trend analysis of FAOSTAT (2013) and FRA country demand. Because of the strong prevalence of agricultural expansion
reported data on forest and agricultural area, combined with as the main determinant of past deforestation, we built our mod-
exogenous global cropland projections to 2030 (Alexandratos and elling choices on the assumption that future forest loss is likely to
Bruinsma, 2012) and global wood demand projections to 2030 be strongly determined by future agricultural expansion.
(as in Buongiorno et al., 2012, harmonized with FAO data). Agricultural expansion is approximated with arable land projec-
To assess the impacts of forest area changes on production and tions up to 2030 by Alexandratos and Bruinsma (2012), which are
conservation functions, a global forest map with information on mainly driven by projections of Gross Domestic Product (GDP)
forest functions was produced and the forest area loss projected and population expansion, the main exogenous drivers for most
by the first model was spatially allocated based on an analysis of global land-use change models (Fischer et al., 2005; Van Vuuren
socio-economic and biophysical characteristics of past forest loss. et al., 2007; Schmitz et al., 2014). The arable land projections also
Finally, we compared the outcomes of the model with regional include projected changes in agricultural intensification which
change estimates based on country expectations on future forest include policy assumptions that potentially provide an enabling
area changes reported in the global forest resources assessment environment and some major assumptions on future commodity
(FRA) 2015 user survey and provided possible explanations for trade (Conforti, 2011; Alexandratos and Bruinsma, 2012). The
diverging expectations. The results are presented by region and model assumes forest gain to be determined by forest regrowth
income level and are limited to the countries that reported data on a share of abandoned agricultural land and an increase in forest
in the FRA 2015 survey, answering to the question, ‘‘What is forest planting driven by wood demand (timber and fuelwood).
area likely to be in the future?’’ The GFRM consists of projections of natural and planted forest
as described in Fig. 1.
2. Material and methods
2.2. Arable land and natural forests
In this paper, the definitions of forests, planted forests and other
land uses follows the FRA Terms and definitions (FAO, 2012b). In The GFRM projection for natural forest change is a linear
particular, forest land use excludes any agricultural use (i.e. oil relationship between arable land change projection and forest
palm plantations are not considered forests), natural forests are change projection:
Fig. 1. Schematic representation of the modelling approach used in the study: the Global Forest Resources Model (GFRM).
126 R. d’Annunzio et al. / Forest Ecology and Management 352 (2015) 124–133
½Natural forest change projection – Where there is no correlation and the arable land increases, the
¼ a ½Arable land change projection ð1Þ model assumes the full arable land expansion to occur on forest
(a ¼ 1Þ:
In Eq. (1), a is a parameter that depends on the correlation, at the
country or sub-regional level, between recent arable land change Fig. 2 summarizes the modelling process of arable land and
(ALC) available in FAOSTAT (2013) and recent natural forest area natural forest change projections as a decision tree, with the num-
change (NFAC) available in FAO (2010). The a parameter is deter- ber of countries concerned at each node and the share of the global
mined by the following circumstances: forests they represent.
– Where there is a correlation, the natural forest change projec-
tion is simulated proportionally to the ratio between the vari-
2.3. Projections of arable land
ables a ¼ NFACALC 0
0
. If the correlation is country specific, the
country ratio is taken, if the correlation is only sub-regional, FAO’s arable land projections are driven by exogenous assump-
the sub-region ratio is taken. tions on population and GDP, in simplified terms described as more
– Where there is no correlation between ALC and NFAC and the people will consume more agricultural products, and richer people
arable land decreases, the GFRM simulates that half of the aban- will consume more agricultural products up to a certain extent and
doned arable land will grow back to forest evaluating the coun- will have different diets, i.e. eat more meat. Increased production
try’s potential area for forest a ¼ Forest2 Pot . The potential is demand in the model is met by arable land expansion and intensi-
calculated from the global ecological zone map from FAO fication, either through increased cropping intensity or increased
(2012a) as the proportion of zones where the biophysical condi- yield intensity. A more detailed description of FAO’s agricultural
tions are such that the estimated vegetation would be forest in production, yield and arable land projections is derived from
the absence of human induced or natural disturbances. Alexandratos and Bruinsma (2012) and Conforti (2011).
Fig. 2. Flowchart representing the decision tree of the Arable Land/Natural Forest module of GFRM. Explanations of the main equation and parameters are found in the text.
At each node, number of countries concerned and the share of the global forests they represent are given.
R. d’Annunzio et al. / Forest Ecology and Management 352 (2015) 124–133 127
The Intact Forest Landscapes (IFLs) dataset (Potapov et al., 2008) change determining the locations of projected loss. The coincident
defines intact forest as unbroken expanses of natural ecosystems use of accessibility data to determine the spatial distribution of
within the zone of current forest extent, with an area of at least both location of production forest and location of projected forest
500 km2 and minimal signs of human activity in the year 2000. loss may have led to overestimation of production forest at risk
In 2010, intact forest covered over 1 billion hectares of the global of being lost, and this must be considered while analyzing the
forest area and the IFL map is used as a proxy for 2010 primary for- results.
est area.
3.2. Global and regional forest projections
2.7. Country-specific predictions to FRA2015 survey
As observed in Fig. 3, global forest area is projected to continue
The question asked in the survey was ‘‘What is forest area likely to decrease over the next 15 years. However, the rate of overall loss
to be in the year 2030?’’ and was answered by 91 countries, con- is projected to slow down, going from 0.13% per year at the begin-
taining 65% of the world’s forests. The answers were estimated ning of the century to 0.06% per year by 2030. This is the result of
by FRA national correspondents, based on varying level of data the decrease in the rate of natural forest loss (0.26% per year to
quality and they represent the official positions of the countries. 0.19% projected per year by 2030) combined with the decrease of
They essentially reflect country-specific expectations and helped the rate of the planted forest gains (2.36% per year to 2.0% pro-
take national circumstances into account in the discussion of the jected per year by 2030).
model’s results. Our projections fall within global forest area projections found
in the literature (MEA, 2005, UNEP, 2007, and OECD, 2012) that
show a range of outcomes from recovered forest area numbers
3. Results and discussion resulting in no net change in global forest area or slight area
increase, to a substantial loss (>15%) up to 2030 compared to the
3.1. Validation of the model assumptions year 2010.
Global loss of forest area is projected to be the net result of for-
The first assumption of the model involved trends in agricul- est area increase in some regions and forest area decrease in others
tural land change and forest land change that were correlated in (Table 4). The regional forest area changes are also the result of
the past. Table 3 shows the comparison between forest area change increases in some sub-regions and decreases in other sub-regions.
and arable land change. In all sub-regions where forest area was For instance, Asia shows increases in East Asia compensated by
decreasing between 1990 and 2010, there was a negative correla- losses in Southeast Asia, resulting in a net forest increase for the
tion with arable land change. In two sub-regions where the area region. Following the model, South America is projected to con-
of natural forest was increasing, again a negative correlation with tinue undergoing the largest net forest area loss over the next
arable land dynamics appears, while in a third (the Caribbean) this 15 years.
trend is absent. Therefore, we assume the first assumption to be As explained under Section 2.1, the projections of arable land
valid. dynamics include some assumptions on policies and national cir-
The second assumption in the model was also corroborated by cumstances, especially concerning assumptions on agricultural
the systematic expected correlation, for all sub-regions, between intensification in Africa (Alexandratos and Bruinsma, 2012). No
wood demand and planted forests (Tables 1 and 2). assumptions have been made on future forest policies though,
The accessibility data was used to determine the locations of whose inclusion could be considered highly speculative. Provided
production forest and was also used as a driver of future forest the limited consideration of future forest policies, the results
should be handled with caution and understood as a business as
usual scenario, as only global considerations on prices have been
Table 3
incorporated in the model. Policy measures such as future climate
Correlation between natural forest area change as estimated in FRA 2010 and Arable change mitigation or future land use planning are not integrated in
land change in FAOSTAT (2013). the model considerations, and can possess an influential effect on
Subregiona Natural forest area change Negative correlation b the forestry trajectories. For instance, Arima et al. (2014) showed
(1990–2010) with arable land? that the decline in rate of deforestation in the Brazilian Amazon
forest is the result of two simultaneous processes, stagnation of
Central America Decreasing Yes
Eastern and Decreasing Yes global demand on agricultural prices and enforcement of policy
Southern Africa regime to cut down deforestation while Dalla-Nora et al. (2014)
Northern Africa Decreasing Yes discussed the limitations of models as they often fail to capture
South America Decreasing Yes
the real trajectories of land use change that are strongly influenced
South-east Asia Decreasing Yes
Western and Decreasing Yes by policies.
Central Africa This could indeed change the picture of global forest evolution:
East Asia Increasing Yes at the global level, non-legally binding political declarations are
Europe Increasing Yes regularly made to strive to end deforestation by 2030 (UN
Caribbean Increasing No
climate summit, 2014). If these measures are realized by countries,
North America Stablec No
Oceania Stable No the modelling approach we use would no longer be valid.
South Asia Stable No Forest policies have only been passively considered if they had
Western and Stable No an effect on forest area change before 2010. Therefore, climate
Central Asia
change policies such as the mechanism for reducing emissions
a
Only countries for which FAOSTAT data from 1990 was available are included, from deforestation and forest degradation (REDD+) are only incor-
e.g. Europe does not include the Russian Federal Republic. porated based on early actions. Most country-driven actions under
b
Given the small amount of dates compared here (4) we say a negative REDD+ are expected after 2010, since the Warsaw Framework for
correlation exists when the correlation factor <0.7. The average correlation for the
subregions where natural forest area is decreasing is 0.93 (a perfect negative
REDD+ (UNFCCC, 2014) was adopted in 2013, setting out the guide-
correlation would be 1). lines for developing country parties to receive results-based pay-
c
Stable is defined as a <5% change in natural forest area between 1990 and 2010. ments for emissions reductions in the forest sector. An example
R. d’Annunzio et al. / Forest Ecology and Management 352 (2015) 124–133 129
4,200,000
4,100,000
4,000,000
3,900,000
'000 ha
3,700,000
Natural forest
3,600,000
3,500,000
3,400,000
3,300,000
3,200,000
1990 2000 2010 2020 2030 2040 2050
year
Fig. 3. Total forest area and natural forest area as projected by the GFRM (full line: data reported to FRA 2010, dotted line: data projected with the model).
Table 4
Regional forest areas in 2010 and their projections towards 2030.
forests (which mainly concerns natural forest with a production
function).
Forest area (1000 ha) However, according to the spatial modelling at the global scale,
Regions 2010 2030 most primary forests are not at high risk of loss with less than 1%
Africa 674,000 646,000 projected loss. Over 95% of this primary forest loss is projected to
Asia 593,000 604,000 occur in the tropical climatic domain only.
Europe 1,005,000 1,039,000 The results can also be used to reveal trends in projected loss in
N&C America 705,000 717,000
production forest areas and protected forest areas, for all regions
Oceania 191,000 190,000
South America 864,000 788,000 and climatic domains.
South America has the largest proportion of projected gross loss
The results in this table come from the GFRM modelling exercise.
of production forest, losing 26% of production forest area by 2030,
though this does not take into account neither law enforcement
nor the increased production function of planted forest. Similarly
of early REDD+ action reflected in the model is Brazil where the Africa has very high rates of production forest loss with 15% of
deforestation rate started to drop as early as 2005 (Brazil, 2014). its production forest area lost by 2030. Europe (including the
This trend is therefore considered in the modelling results, though Russian Federation) is the region with the largest area of produc-
regionally compensated by increasing deforestation trends in other tion forest and least projected production forest loss.
countries. When assessing the trends by climatic domain it is apparent
Furthermore, the model functions on the assumption that the that the tropical climatic domain has the highest risk of future for-
global conditions will remain the same during the period of interest. est conversion. Production forest in the tropical domain are pro-
Yet, we acknowledge that the occurring climate change will have jected to have about 15% area loss. Sub-tropical production forest
effects up to at least the end of the 21st century (IPCC, 2013).The area losses are projected to be 5%. Temperate forests are projected
effect of climate change on forest projection has been quantified in to lose less than 1% of production forest area and boreal forests are
various studies and ranges from limited impact component projected to experience virtually no loss between 2010 and 2030.
(Thompson et al., 2011) to game-changer scenario: Kreileman and Protected areas make up a smaller proportion of total forest
Alcamo (1998) for instance, project forests towards 2100 under vari- area than production forest. South America, North America and
ous modalities and show that a scenario including only land-use Oceania all have large areas designated for protection and small
change would lead to global forest loss, whereas the inclusion of cli- projected loss within those forests. Africa and Asia are projected
mate change in their model switches the results to substantial forest to suffer the highest portion of protected areas loss (4%).
gains. Depending on how climate changes in the next decades, the Tropical protected forests have the highest risk of conversion to
effect of climate change could become more significant than pre- non-forest between 2010 and 2030. The tropical climatic domain is
sented here. Given the relatively short span of time addressed by this projected to lose 3% of its protected areas between 2010 and 2030.
study, we believe the assumption is still acceptable. The model projects very little to no change for the subtropical,
The overall pattern of our model shows that global loss will temperate, and boreal climatic domains.
keep on by 2030, while slowing down, and we further explored The results can finally be compared with specific country expec-
the projections stratified by regions, to examine what type of forest tations towards 2030 as a mean to verify the model validity and
was at risk of being lost. discuss some of the assumptions.
3.3. Functions of forests at risk of being lost 3.4. Country specific predictions for 2030
The results of the projection of productive and protective areas of Out of the 234 countries that reported data to FRA 2015 only 91
forests with GEOMOD are summarized in Table 5. The area compar- countries (containing 65% of the global forests) reported data to the
ison showed that 32 countries presented a risk of seeing their pro- question ‘‘What is forest area likely to be in the future?’’ This is
tection/conservation forest area be threatened by deforestation, clearly not sufficient to generate sensible global or regional trends,
while 36 countries presented a similar risk for their production especially in Africa, North and Central America or Oceania where
130 R. d’Annunzio et al. / Forest Ecology and Management 352 (2015) 124–133
Table 5
Projected area of forest at risk of being lost by 2030 within production, protection and primary forests, by climatic domain and FRA region.
Area of forest at risk of being lost (2010– Climatic domains FRA regions
2030). . .
Tropical Subtropical Temperate Boreal Africa Asia Europe N&C Oceania South
(%) (%) (%) (%) (%) (%) (%) America (%) (%) America (%)
Production Proportion of 2010 15 5 0.80 0 15 5 0.07 0.30 3 26
forest production forest
Proportion of 2010 total 4 2 0.50 0 5 2 0.04 0.10 0.50 5
forest area
Protected Proportion of 2010 protected 3 1 0.10 0 4 4 0.20 0.20 1 2
forest forest area
Proportion of 2010 total 0.30 0.10 0.02 0 0.30 0.30 0.01 0.02 0.30 0.20
forest area
Primary Proportion of 2010 primary 2 0 0 0 1 1 0 0 0 2
forest forest area
Proportion of 2010 total 0.10 0 0 0 0.08 0.02 0 0 0 0.20
forest area
The results in this table come from the combination of the GFRM modelling exercise and spatial modelling.
Table 6
reporting countries represent less than half of the actual forests Expected forest area change (losses, gains and resulting net change) for 2015–2030
(Table 6). (1000 ha) summed up by region and income level.
For that reason, the results of the country specific prediction Expected forest change Nb Share of total
cannot be extrapolated to regional or global estimate and were within reporting countries, countries forest
not directly compared with the results of the modelling. For 2015–2030 (1000 ha) area (%)
instance, the net resulting gain of circa 78 Mha expected for all Loss Gain Net
the reporting countries only concerns 65% of the forests of the Africa 2298 21,686 19,389 22 26
world and cannot be compared to the expected global continuing Asia 5615 95,656 90,041 21 80
forest loss coming from the model. Europe 557 12,820 12,263 28 92
However the country-specific predictions aid in understanding N&C America 6626 6626 5 51
Oceania 50 1 49 2 19
the vision and target that the reporting countries intend to meet
South America 50,145 13,327 36,819 13 76
by the year 2030; these were used to discuss the projections of High 6122 11,986 5865 29 68
the model. Upper medium 10,193 44,136 33,942 29 73
There are strong contrasts by region and income level in pro- Lower medium 47,446 78,844 31,398 23 76
jected forest area change. For instance, Asia reported to expect Low 1529 8,523 6994 10 16
Total 65,291 143,490 78,199 91 65
the highest gain with 90 million hectares of forest area increase
whereas South America reported to expect the highest forest loss The results in this table come from the FRA 2015 user survey analysis. The number
with over 35 million hectares of forest area decrease. Little change of reporting countries and share of the total forest area represented by the reporting
countries are given in the last two columns. Because the reporting countries are not
was reported to be expected in the high and low income categories,
enough to adequately represent the regions (e.g. Africa with less than 26% of the
while the middle categories comprised both strong gains and forests covered by the survey), results are only summed-up and not extrapolated.
losses. This is particularly true when comparing the high income
and upper medium income categories: for the same number of
reporting countries and a sensibly equal forest area represented,
the latter expect 6 times more change to happen in the next Table 7
15 years, indicating an economic dynamism in Medium countries Countries expecting strong change in their forest area by 2030.
(more conversion of forest to agriculture but also more planting) Forest change 2015–2030
that is not detected in High income countries. The responses for
(1000 ha) (%)
the Low income category countries are not representative enough
Argentina 12,947 48
(only 16% of forests represented) to draw any practical conclusion.
Bolivia 45,764 84
Regarding the magnitude of expected forest area change, 56% of China 22,079 11
the reporting countries estimated less than 10% of forest area Indonesia 21,330 23
change to occur between 2015 and 2030. However, a small number India 37,798 53
of countries, listed in Table 7, presumed strong changes towards Nigeria 13,720 196
Togo 1510 803
2030. These few countries strongly influence upward the entire
Russia 10,070 1
dataset as it essentially concerns gains, with the exception of
Bolivia which suspects increased forest loss in the next 15 years The results in this table come from the FRA 2015 user survey analysis.
(84% of its current forest area is expected to be lost by 2030).
The figures for China and Russia are in line with the trends In the case of Russia, regrowth of forest on abandoned arable
observed in the past, and hence would tend to confirm the predic- land is a past trend that is expected to continue, as described in
tion of the model. China has substantially increased its forest area the Russian outlook study (FAO, 2012c).
and forest stock volume since the early 1990s because it made The case of Indonesia illustrates the will of the country to
increasingly significant, effective and large-scale programmatic engage in REDD+ (UN-REDD, 2014) process and re-convert a large
efforts during the past three decades to enhance afforestation share of its land to forests. This is another example of political deci-
and reforestation. These programs have benefited from sustained sion-making that tends to counter-effect past trends in the tropics
and substantial allocation of fiscal and other resources by central and could, if indeed enforced, invalidate the conclusions of the
and local governments (Antweiler et al., 2012). modelling decisions taken in this study.
R. d’Annunzio et al. / Forest Ecology and Management 352 (2015) 124–133 131
To explore that point in more details, we compared trends in The other countries are expecting an inversion in their forest
forest area over the past 15 years with the expected trend for the trends: Bhutan, Belarus, Iran and the USA for instance, have seen
next 15 years (Fig. 4). Most countries (58 out of 91) are presuming their forest area increase over the past 15 years but they are
the trend to remain the same, e.g. continuing loss is presumed in assuming a minor forest loss in the next 15 years. Conversely,
Brazil and Mali and continuing gain in India, China and Russia, countries like Argentina, Indonesia, Nepal, Nigeria, Tanzania and
though the rate of loss/gain may be expected to change.
Fig. 4. Plot of forest area change for 2015–2030 as expected by countries, in absolute (1000 ha) against relative values (% of the 2015 forest area). Countries in the upper right
quarter are expecting gains in the future, while countries in the lower left are expecting losses. The past trend is showed in colors, with countries that experienced losses
during 2000–2015 in red and countries that experienced gains in blue. This helps to underline countries that expect an inversion of their current trend (e.g. USA showing up in
blue in the lower left quarter or Nigeria showing up in red in the upper right quarter). (For interpretation of the references to colour in this figure legend, the reader is referred
to the web version of this article.)
1.4
Relave forest area of the reporng countries (base 2000)
Caribbean
1.3
South and Southeast Asia
East Asia
1.2
West and Central Africa
0.9 Oceania
East Africa
0.8
North Africa
Central America
0.7
South America
0.6
1990 1995 2000 2005 2010 2015 2020 2025 2030 2035
Fig. 5. Relative forest area (year 2000 is taken as the base) as reported in FRA 2015 (for 1990, 2000, 2010 and 2015, full lines) and projected by countries (for 2030, dotted
lines). Results are aggregated by sub-region but note that only reporting countries are included (see Table 6) and should hence be handled with caution for underrepresented
sub-regions. This graphic is essentially displayed to spot out sub-regions where trends are likely to diverge from the modelling linear solution presented in the GFRM (South
and South East Asia and West and Central Africa).
132 R. d’Annunzio et al. / Forest Ecology and Management 352 (2015) 124–133
1,200,000
1,000,000
200,000
0
1990 2000 2010 2020 2030 2040 2050
Fig. 6. Forest area by region as (i) estimated in FRA2010 for the period 1990–2010 (full lines), (ii) projected with GFRM towards 2050 (dotted lines) and (iii) extrapolated from
the country specific aspiration for 2030 as reported in FRA 2015 (points). Note that the divergence for Asia and Africa is strongly influenced by the low representation of
reporting countries and the change of trend expected from countries within these regions.
Thailand are assuming the loss observed in the past will be However, the relatively smaller global annual net change in for-
reverted to gain in forest. est area in 2030 compared to 2015 masks large regional differ-
The trends in forest area translate into sub-regional patterns as ences; in some regions, forests are projected to continue to
can be seen in Fig. 5. In most sub-regions, the trend in forest area decline at alarming rates. Furthermore, even though global forest
change observed from 1990 to 2015 is estimated by the reporting loss is projected to slow down, the rate of biodiversity loss may
countries to remain similar in the future. The trends slow down not display a similar levelling trend since loss of natural forest is
with lesser gains and lesser losses, which corresponds to the pro- partially off-set by expansion of planted forests. Additionally, the
jections we made using the GFRM. However, two sub-regions stand impacts on biodiversity are not fully captured because forest habi-
out of that converging picture. In South and Southeast Asia as well tat losses in the tropics cannot be directly compensated for forest
as in West and Central Africa, countries are foreseeing a clear gains in other ecological zones (Pereira et al., 2010). The productiv-
inversion of the expanding forest trend. This still holds for North ity of planted forests is estimated to increase and that may have a
and East Africa, but to a lesser extent. trade-off in diminished richness of biodiversity. The forest areas
One should keep in mind that these trends are strongly influ- that are the most at risk of conversion were identified as forests
enced by outlier countries (e.g. Indonesia and Nigeria), so that under multiple uses, within the tropical domain. The forest under
the extrapolation for these sub-regions should be used cautiously protected areas showed very little risk of being converted to other
and only for discussion purposes. We hence used the limited data land uses in the near future.
coming from the expectations of countries and extrapolated them The conclusions of the modelling effort generally align with the
to the regions to compare with the results of the model (Fig. 6). estimations of the FRA2015 user survey, at least for the regions
Overall, the model results and the country expectations are in where enough countries have reported their expectations.
agreement for North and Central America, Europe, South America This study helped identify countries whose forest policy and/or
and Oceania, but diverge for Africa and Asia. The difference for aspiration for the future might curb the actual trend. If these
Asia might be explained by the fact that Asia’s planted forest objec- aspirations prove to be true, the projected loss from the model
tives are possibly not entirely driven by a demand for wood but might be lower than expected. On the other hand, the assumptions
may consider soil restoration, climate change mitigation and other of the model that productivity will increase might not prove strong
conservation related objectives whereas the model is only driven enough to reduce pressure on forests; this, in turn, may lead to
by wood demand. For Africa, the reporting countries are clearly more forest loss than projected.
not representative of the whole region (26% only) and the extrap- The discrepancies between the modelling exercise and the
olation should not be considered valid. country estimations reveal relevant aspects that would need to
Finally, these discrepancies also show that country specific poli- be incorporated in a further global projection exercise, accounting
cies could still influence the pattern of continuing forest loss pro- for the global-scale effort to curb deforestation put into place by
jected to occur. The diverging results reflect the intentions and the international community.
expectations of countries regarding their forest policies as relevant
items that would need to be incorporated in further global projec-
tion exercise. References
Alexandratos, N., Bruinsma, J., 2012. World agriculture towards 2030/2050: the
2012 revision. ESA Working paper 12-03. FAO, Rome.
4. Conclusion Antweiler, P., Wei, L., Liu, Y., 2012. Ecological rehabilitation in China. Achievements
of Key Forestry Initiatives. Asia Pacific Network for Sustainable Forest
Both the modelling results as well as the country predictions Management and Rehabilitation. China Forestry Publishing House.
Arima, E.Y., Barreto, P., Araújo, E., Soares-Filho, B., 2014. Public policies can reduce
suggest that, at the global level, forest resource loss is likely to con- tropical deforestation: lessons and challenges from Brazil. Land Use Policy 41,
tinue but slow down by 2030. 465–473.
R. d’Annunzio et al. / Forest Ecology and Management 352 (2015) 124–133 133
Baumann, M., Kuemmerle, T., Elbakidze, M., Ozdogan, M., Radeloff, V., Keuler, N., IPCC, 2013: Summary for Policymakers. In: Stocker, T.F., Qin, D., Plattner, G.-K.,
Prishchepov, A., Kruhlov, I., Hostert, P., 2011. Patterns and drivers of post- Tignor, M., Allen, S.K., Boschung, J., Nauels, A., Xia, Y., Bex, V., Midgley, P.M.
socialist farmland abandonment in Western Ukraine. Land Use Policy 28, 552– (Eds.), Climate Change 2013: The Physical Science Basis. Contribution of
562. Working Group I to the Fifth Assessment Report of the Intergovernmental
Brazil, 2014. Brazil’s submission of a Forest Reference Emission Level (FREL) for Panel on Climate Change. Cambridge University Press, Cambridge, United
reducing emissions from deforestation in the Amazonia biome for REDD+ Kingdom and New York, NY, USA.
results-based payments under the UNFCCC. <http://unfccc.int/land_use_ Kreileman, E., Alcamo, J., 1998. The distribution of future global forests as affected
and_climate_change/redd/items/8414.php>. by climate change and land use, Chapter 4.5. In Carbon Dioxide Mitigation in
Buongiorno, J., Zhu, S., 2014. Technical change in forest sector models: the global Forestry and Wood Industry, pp 353–371.
forest products model approach. Scand. J. For. Res. 30, 30–48. http://dx.doi.org/ MEA, 2005. Millennium ecosystem assessment: ecosystems and human well-being
10.1080/02827581.2014.957238. (volume 2). Findings of the Scenarios Working Group, Millennium Ecosystem
Buongiorno, J., Zhu, S., Zhang, D., Turner, J., Tomberlin, D., 2003. The Global Forest Assessment, Island Press, Washington, DC.
Products Model (GFPM): Structure, Estimation, and Applications. Academic Nepstad, D.C., Stickler, C.L., Soares-Filho, B., Merry, F., 2008. Interactions among
Press, San Diego, p. 301. Amazon land use, forests and climate: Prospects for a near-term forest tipping
Buongiorno, J., Zhu, S., Raunikar, R., Prestemon, J., 2012. Outlook to 2060 for world point. Philos. Trans. R. Soc. Lon. B Biol. Sci. 363, 1737–1746.
forests and forest industries: A technical document supporting the forest OECD, 2012. Environmental outlook to 2050: the consequences of inaction.
service 2010 RPA Assessment. General Technical Report SRS-151. Southern Organization for Economic Co-operation and Development, France, 2012.
Research Station, USDA Forest Service. Pereira, H.M., Leadley, P.W., Proença, V., Alkemade, R., Scharlemann, J.P., Fernandez-
Carle, B.J., Holmgren, P., 2008. Wood from planted forests: a global outlook 2005– Manjarrés, J.F., Araújo, M.B., Balvanera, P., Biggs, R., Cheung, W.W.L., Chini, L.,
2030. For. Prod. J. 58 (10469), 6–18. Cooper, H.D., Gilman, E.L., Guénette, S., Hurtt, G.C., Huntington, H.P., Mace, G.M.,
CBD, 2010. COP 10 decision X/2 adoption of strategic plan for biodiversity 2011– Oberdorff, T., Revenga, C., Rodrigues, P., Scholes, R.J., Sumaila, U.R., Walpole, M.,
2020, including the Aichi targets. In: Tenth meeting of the Conference of the 2010. Scenarios for global biodiversity in the 21st century. Science 330 (6010),
Parties to the Convention on Biological Diversity held from 18 to 29 October in 1496–1501.
Nagoya, Japan. Pontius Jr, R.G., Cornell, J.D., Hall, C.A., 2001. Modeling the spatial pattern of land-
Conforti, P., 2011. Looking ahead in world food and agriculture: Perspectives to use change with GEOMOD2: application and validation for Costa Rica. Agric.
2050. FAO, Rome. <http://www.fao.org/docrep/014/i2280e/i2280e.pdf>. Ecosyst. Environ. 85 (1), 191–203.
Dalla-Nora, E.L., Aguiar, A.P.D., Lapola, D.M., Woltjer, G., 2014. Why have land use Potapov, P., Yaroshenko, A., Turubanova, S., Dubinin, M., Laestadius, L., Thies, C.,
change models for the Amazon failed to capture the amount of deforestation Zhuravleva, I., 2008. Mapping the World’s Intact Forest Landscapes by remote
over the last decade? Land Use Policy 39, 403–411. sensing. Ecol. Soc. 2 (13).
Day, R.H., 1973. Recursive programming models: a brief introduction. In: Judge, Samuelson, P.A., 1952. Spatial price equilibrium and linear programming. Am. Econ.
G.G., Takayama, T. (Eds.), Studies in Economic Planning over Space and Rev. 42 (3), 283–303.
Time: Contributions to Economic Analysis. American Elsevier, New York, pp. Schmitz, C., van Meijl, H., Kyle, P., Nelson, G.C., Fujimori, S., Gurgel, A., Havlik, P.,
329–344. Heyhoe, E., Mason d’Croz, D., Popp, A., Sands, R., Tabeau, A., van der
Dudley, N., (Ed.), 2008. Guidelines for Applying Protected Area Management Mensbrugghe, D., von Lampe, M., Wise, M., Blanc, E., Hasegawa, T., Kavellari,
Categories. In: Stolton, S., Shadie, P., Dudley, N., 2013. IUCN WCPA Best practice A., Valin, H., 2014. Land-use change trajectories up to 2050-insights from a
guidance on recognizing protected areas and assigning management categories global agro-economic model comparison. Agr. Econ. 45 (1), 69–84.
and governance types, Best practice protected area guidelines. 21, Gland, Smeets, E., Faaij, A., 2007. Bioenergy potentials from forestry in 2050, an assessment
Switzerland, p. 143. of the drivers that determine the potentials. Clim. Change 81, 353–390.
EROS, 1996. GTOPO30: Global digital elevation model (DEM). U.S. Geological Smith, P., Gregory, P.J., van Vuuren, D., Obersteiner, M., Havlík, P., Rounsevell, M.,
Survey’s EROS Data Center, Sioux Falls, South Dakota, USA. <http://edcdaac. Woods, J., et al., 2010. Competition for land. Philos. Trans. R. Soc. Lon. B Biol. Sci.
usgs.gov/gtopo30/gtopo30.html> (accessed 01.02.2014). 365 (1554), 2941–2957.
FAO, 2010. Global Forest Resources Assessment 2010. FAO Forestry Paper 163, Thompson, J., Foster, D., Scheller, R., Kittredge, D., 2011. The influence of land use
Rome. <http://www.fao.org/docrep/013/i1757e/i1757e.pdf>. and climate change on forest biomass and composition in Massachusetts, USA.
FAO, 2012a. Global ecological zones for FAO forest reporting: 2010 update. Forest Ecol. Appl. 21, 2425–2444, <http://dx.doi.org/10.1890/10-2383.1>.
Resources Assessment Working Paper 179, FAO, Rome. <http://www.fao. UN, 2012. Probabilistic Population Projections based on the World Population
org/docrep/017/ap861e/ap861e00.pdf>. Prospects: The 2012 Revision. United Nations Department of Economic and
FAO, 2012b. FRA2015 Terms and Definitions, Forest Resources Assessment Working Social Affairs, Population Division, Population Estimates and Projections
Paper 180, FAO, Rome. <http://www.fao.org/docrep/017/ap862e/ap862e00. Section. <http://esa.un.org/unpd/wpp/Excel-Data/population.htm> (accessed
pdf>. 01.02.2015).
FAO, 2012c. The Russian Federation Forest Sector Outlook Study to 2030. FAO, UN, 2014. United Nations Climate Summit 2014, New-York declaration on Forests,
Rome. Action statements and action plans. UN Headquarters, New-York, Sept. 23,
FAO, 2013. Population density map for the year 2000, extracted from FAO’s 2014. www.un.org/climatechange/summit/ (accessed 01.02.2015).
GeoNetwork, <http://www.fao.org/geonetwork/srv/en/> (accessed 01.02.2014). UNECE and FAO, 2009. Joint Wood Energy Enquiry. <http://www.unece.org/forests/
FAO and JRC, 2012. Global forest land-use change 1990–2005. FAO Forestry paper jwee.html> (accessed 01.02.2015).
169. FAO, Rome. <www.fao.org/docrep/017/i3110e/i3110e.pdf>. UNEP, 2007. Global environment outlook 4 (GEO-4). United Nations Environment
FAOSTAT, 2013. www.data.fao.org (accessed 01.02.2014). Programme, Nairobi. <http://www.unep.org/geo/geo4.asp> (accessed 01.02.
Fischer, G., Shah, M., Tubiello, F.N., Van Veldhuizen, H., 2005. Socio-economic and 2015).
climate change impacts on agriculture: an integrated assessment, 1990–2080. UNFCCC, 2014. Report of the Conference of the Parties on its nineteenth session,
Phil. Trans. Royal Soc. Bio. Sci. 360, 1463(2067–2083). held in Warsaw from 11 to 23 November 2013. Part two: Action taken by the
Fischer, G., Nachtergaele, F.O., Prieler, S., Teixeira, E., Tóth, G., van Velthuizen, H., Conference of the Parties at its nineteenth session, Decisions adopted by the
Verelst, L., et al., 2010. Global Agro-ecological Zones: Model Documentation. Conference of the Parties, 2014.
IIASA and FAO, Laxenburg and Rome. UN-REDD, 2014. <http://www.un-redd.org/CountryActions/Indonesia/tabid/987/
Geist, H.J., Lambin E.F., 2001. What drives tropical deforestation? LUCC Report language/en-US/Default.aspx> (accessed 01.02.2015).
Series 4. LUCC, Louvain La Neuve, Belgium. <http://www.pik-potsdam.de/ USEIA, 2014. Increase in wood as main source of household heating most notable in
~luedeke/lucc4.pdf>. the Northeast, U.S Energy Information Administration. <http://www.eia.gov/
Gutierrez-Velez, V.H., de Fries, R.S., Pinedo-Vasquez, M., Uriarte, M., Padoch, C., todayinenergy/detail.cfm?id=15431> (accessed 01.02.2015).
Baethgen, W., Fernandes, K., Lim, Y., 2011. High-yield oil palm expansion spares Van Vuuren, D.P., Lucas, P., Hilderink, H., 2007. Downscaling drivers of global
land at the expense of forests in the Peruvian Amazon. Environ. Res. Lett. 6 environmental change. Enabling use of global SRES scenarios at the national and
(044029), 5. grid levels. Glob. Environ. Change 17, 114–130.
Holmgren, P., 2006 Global Land Use Area Change Matrix: Input to GEO-4. FAO, WDPA, 2012. World Database on Protected Areas, available at www.
Rome. <ftp://ftp.fao.org/docrep/fao/010/ag049e/ag049e00.pdf>. protectedplanet.net/ (accessed 01.02.2014).
Hosonouma, N., Herold, M., De Sy, V., De Fries, R.S., Brockhaus, M., Verchot, L., Whiteman, 2014. Global Trends and Outlook for Forest Resources. In: Challenges
Angelsen, A., Romijn, E., 2012. An assessment of deforestation and forest and Opportunities for the World’s Forests in the 21st Century, For. Sci. vol. 81,
degradation drivers in developing countries. Environ. Res. Lett. 7 (4), 044009, pp. 163–211.
12. World Bank, 2009. Reshaping Economic Geography, World Development Report
IEA, 2011. World Energy Outlook, 2011. International Energy agency, SBN: 978–92- 2009. Accessibility map available at <http://bioval.jrc.ec.europa.eu/products/
64-12413-4, pp. 696. www.worldenergyoutlook.org/publications/weo-2011/. gam/index.htm> (accessed 01.12.2012).
Forest Ecology and Management 352 (2015) 134–145
a r t i c l e i n f o a b s t r a c t
Article history: The Global Forest Resources Assessment 2015 shows that deforestation has slowed and afforestation has
Received 16 March 2015 increased globally during 1990–2015. Planted forests have increasingly provided goods and services hith-
Received in revised form 7 June 2015 erto derived from natural forests, and mosaic forests in agricultural landscapes are increasing. Forest gain
Accepted 8 June 2015
is occurring at higher latitudes and in richer countries whilst forest loss continues in poor countries in the
Available online 7 September 2015
tropics. Some middle income tropical countries are now also transitioning to forest gain. These transition
countries are characterised by reforms to forest management and improvements in agricultural practices
Keywords:
but also by significant expansions of planted forest, which account for 25–100% of gains. Forest-area
Global forest trends
Forest transitions
estimates of the FRA align with satellite-derived estimates, with deviations of 6±7% globally and
Sustainable forest management 6±17% for the tropics. Mosaics comprised of trees outside forests, remnant forest patches, and young
Plantation forests regenerating forests constitute a modest proportion of the tropical forest estate and are seemingly well
Planted forests inventoried by the FRA. Extensive areas of forest experienced partial canopy cover reduction since 2000,
Forest pests and diseases particularly in the tropics where their area is 6.5 times that deforested since 1990. The likelihood of the
Canopy cover reduction eventual loss of these forests and a decline in their capacity to provide goods and services is a matter of
concern. Demand for industrial wood and fuelwood increased 35% in the tropics since 1990, principally in
poorer countries, and growth in demand will accelerate into the future, particularly in the Asia-Pacific
region. Notwithstanding significant increases in forests within protected areas since 1990 to 517 Mha
(16.3%) globally and 379 Mha (26.6%) in the tropics, increasing demands for ecological services, forest
products, and climate change mitigation is likely to be met from an expanding area of planted forests
more than from the declining area of natural forests, particularly in Africa. The global rate of
planted-forest expansion since 1990 is close to a target rate of 2.4% per annum necessary to replace wood
supplied from natural forests in the medium term, though the expansion rate has declined to 1.5% since
2005. Multiple-use forests permitting both production and conservation account for 26% of the global for-
est area and 17% of the tropical forest area, and have increased by 81.8 Mha or 8.5% globally since 1990,
with most gains in the tropics. Sustainable forest management in low-income and tropical countries
remains modest, with only 37% low-income country forests covered by forest inventories.
International support has proven effective at increasing this coverage since 2010.
Ó 2015 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
2. The FRA and realities on the ground. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
3. Patterns of forest change, 1990–2015. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
4. Meeting the world’s needs for forest goods and services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
4.1. The growing role of planted forests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
q
This article is part of a special issue entitled ‘‘Changes in Global Forest Resources from 1990 to 2015’’.
⇑ Corresponding author.
E-mail address: sean.sloan@jcu.edu.au (S. Sloan).
http://dx.doi.org/10.1016/j.foreco.2015.06.013
0378-1127/Ó 2015 Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
S. Sloan, J.A. Sayer / Forest Ecology and Management 352 (2015) 134–145 135
highlights instances of national transitions from loss to gain. renewed debate over the utility of the FRA for tracking global forest
Section 4 profiles the growing demands for forest products and change (Friedl et al., 2010; Hansen et al., 2010, 2013; Tateishi et al.,
environmental services and discusses potential responses for 2010; FAO and JRC, 2012, 2014; Sexton et al., 2013; Achard et al.,
large-scale forest management. Section 5 concludes by emphasis- 2014; Kim et al., 2015; Keenan et al., 2015). This is exemplified
ing the growing challenge faced by large-scale forest inventories by assertions of large discrepancies between forest-change esti-
such as the FRA in capturing forest extent and services in heteroge- mates of the FRA and satellite observations (Hansen et al., 2013;
neous forest landscapes. Kim et al., 2015). Estimates derived from different data, methods,
and definitions of forest inevitably diverge, and this is also true
of estimates derived from different satellite observations and even
2. The FRA and realities on the ground the same satellite sensor (Keenan et al., 2015), complicating asser-
tions of superior accuracy.
The FRA 2015 gives a more precise and consistent picture of Global forest-area estimates defined by five remotely-sensed
realities on the ground than previous assessments (Romijn et al., studies for approximately 1990, 2000, 2005, and/or 2010 (Hansen
2015). Estimates of forest area for 1990–2010 have been expressed et al., 2010, 2013; FAO and JRC, 2012, 2014; Gong et al., 2013)
in metrics that allow comparisons with those of 2015 (Keenan range from 21.4% to +2.2% of the estimates from the FRA 2015
et al., 2015; MacDicken, 2015). Modern remote sensing has for the same years (Keenan et al., 2015: Table 11). This range nar-
increased capacities to detect forest changes and 70% of countries rows considerably to 6.5% to +2.2% upon omitting Hansen et al.
are now deploying remote-sensing surveys as part of the FRA (2010) from consideration, and narrows further still to 6.5% to
forest-inventory exercise. The FRA 2015 saw 17 additional tropical 2.6% upon considering only the two remotely-sensed estimates
countries qualify as having ‘good’ or ‘very good’ capacities for for- which are consistent with the FRA in defining forests as a land
est monitoring and remote sensing compared to the FRA 2005, an use2 rather than as general tree cover (FAO and JRC, 2012, 2014).
increase of 30%, with a corresponding increase in the proportion of The reported accuracies of remotely-sensed estimates vary in the
tropical forest area monitored to such a level from 69% to 83% order of 80–95% plus a margin of error dependent on methodological
(Romijn et al., 2015). Globally the number of countries with such factors including errors in classification, interpretation and sam-
capacities rose to 54, out of the 99 tropical countries surveyed by pling. No comparable accuracy assessment exists for FRA
Romijn et al. (2015). Only 11% of global forest area for 2015 forest-area estimates.
remains in the lowest of three tiers of data quality whilst 59% of In the tropical domain the range of forest-area estimates
the global forest area qualifies for the highest of the tiers defined by six remote-sensing exercises for approximately 1990,
(Keenan et al., 2015: Table 5). This uppermost tier is defined as 2000, 2005, and/or 2010 (FAO, 2001: Ch. 46; Hansen et al., 2010,
forest-area data sourced from either repeated national forest 2013; FAO and JRC, 2012, 2014; Achard et al., 2014) is centred rel-
inventories or from remote sensing or a national forest inventory atively evenly at 16.8% to +12% of the FRA 2015 estimate for the
completed 610 years ago and entailing ground truthing (FAO, same years. The range for the tropics is no more narrow than the
2015). Many poor countries still lack capacity for autonomous range for global estimates in spite of the fact that changes in forest
high-quality forest monitoring and reporting and international use (as reported by the FRA) are believed to correspond most clo-
efforts to improve this situation should be pursued (MacDicken, sely to changes in forest cover (as reported by satellite observa-
2015). The 12 countries with >5 Mha of forest of a Tier 1 grade tions) in the tropics (Coulston et al., 2013). The range for the
accounted for only 9% of the global forest area but for 20% of global tropics similarly narrows to between 16% and 3.8% upon con-
forest loss since 1990. sidering the three estimates which use forest definitions consistent
Successive FRAs have been criticised as imprecisely and incon- with the FRA (FAO, 2001: Ch. 46; FAO and JRC, 2012, 2014), again
sistently estimating forest area, particularly by analysts seeking to suggesting that FRA estimates align well with reality at large
compare forest area between countries and over time (Grainger, scales. Comparisons of rates of forest loss within the tropical
1996, 2008; Grainger, 2010; MacDicken, 2015). But the FRA is domain suggest more significant deviations between the FRA and
now assessing 120 different variables concerning forest resources remotely-sensed estimates (Keenan et al., 2015). However, differ-
and attributes for which data are provided by a range of national ent periods of observation and forest definitions preclude confi-
authorities. Methods, sampling intensity, and competence vary dent comparisons of rates of loss, especially since deforestation
amongst countries. The term ‘forest’ describes everything from rates have varied over time (Keenan et al., 2015) and slowed con-
scattered trees in dry landscapes to dense, closed canopy old siderably in some global sub-regions between successive FRA peri-
growth forests in high rainfall areas (Lund, 2006). The FRA has con- ods (Payn et al., 2015).
tributed to the adoption of globally comparable approaches to Forests which are naturally regenerating or subject to direct
defining and categorising forests but total uniformity of national human influence are becoming increasingly important compo-
approaches is likely to be elusive. In Indonesia, for instance, much nents of tropical landscapes (Lugo and Helmer, 2004; Chazdon
land officially designated as forest has been converted to agricul- et al., 2009; Meyfroidt and Lambin, 2011; Chazdon, 2014b,a;
ture, often illegally, and only some of these lands have been subse- Schnell et al., 2015). Tree cover of >10% canopy cover increased
quently re-designated as non-forest land (Indrarto et al., 2012; by 13% (49 Mha) within nominally agricultural landscapes in
Sloan, 2014). This discrepancy partially reflects tensions between South America over the 2000s according to satellite estimates
the central forest administration seeking to retain influence over (Zomer et al., 2014) whilst total forest area declined over this same
the forest estate and decentralised administrative bodies seeking period. In South America as well as South East Asia and Central
greater freedom to allocate land for agricultural development. America tree cover of >10% canopy occupies well over half of the
Indonesia is struggling to reconcile conflicting maps of land allo- total agricultural extent (Zomer et al., 2014), and may account
cated to forests and agriculture and to adopt a single national
map of forest and agricultural cover (Unit Keria Presiden4, 2012;
2
Sloan, 2014). These national dynamics may affect Indonesia’s The FRA defines forests as land use. This definition excludes lands predominantly
report of forest-change to the FRA and complicate comparisons dedicated to other land uses, such as agriculture, even where these have appreciable
forest cover. Similarly it includes lands dedicated to forest production or conservation
with otherwise similar countries, such as Malaysia. even where these are temporarily destocked. This definition is therefore distinct from
The recent production of several independent global one of forest as a land cover, that is, the simple presence of tree cover above a certain
satellite-based estimates of forest area and forest change has density.
S. Sloan, J.A. Sayer / Forest Ecology and Management 352 (2015) 134–145 137
for 16–21% of national above-ground tree biomass generally FRA estimates may be less robust for countries where the forest
(Schnell et al., 2015: Table 3). It is therefore noteworthy that a pro- area is proportionally small, agricultural landscapes relatively
portion of the discrepancies between the FRA and satellite esti- dynamic, and extensive subsistence agricultural is widely prac-
mates reflect the fact that FRA forest-area estimates exclude tree ticed. El Salvador is illustrative, with only 14% forest area according
cover in landscapes nominally managed for agricultural purposes to the FRA and ubiquitous agro-pastoral mosaics. Satellite observa-
even where canopy cover exceeds the 10% threshold applied else- tions indicate that El Salvador has experienced net gain in tree
where. Tree cover in agricultural and other non-forest landscapes cover since the early 1990s following widespread regrowth on
was reported to the FRA by only 58% of countries for 2015 because abandoned pastures (Hecht and Saatchi, 2007). In contrast, El
many countries lack data (MacDicken pers. comm.). When such Salvador reported steady deforestation over 1990–2015 to the
tree cover is reported it is aggregated with non-forest tree cover FRA, mostly in areas of dense remnant forests (Hansen et al.,
such as orange groves or oil palm and recorded as ‘trees outside 2013). El Salvador’s so-called ‘invisible forests’ are therefore allo-
forests’ (TOF) separately from total forests area (FAO, 2012a,c). cated largely to the FRA category ‘trees outside of forest’ (Hecht
The FRA 2015 reports 284 Mha of TOF globally, of which et al., 2006), where observed. This tendency has also been noted
214 Mha or 75% are in the tropics. In tropical Asia community for- with FRA 2010 for Africa (Hansen et al., 2013), although there
ests and home-garden mosaics with dense tree cover occupy vast dry-forest contexts and extensive agro-pastoral systems introduce
areas officially designated as agriculture (Collins et al., 1991; notable uncertainties in the remote detection and categorisation of
Zomer et al., 2014). The African context is similar, with extensive forest area, particularly where forest-canopy thresholds are close
areas of orchard-bush in sub-humid areas. In Latin America there to 10% (FAO and JRC, 2012: 23; Beuchle et al., 2015).
are large areas of spontaneous forest regrowth on The current FRA converges with remotely-sensed studies more
semi-abandoned agro-pastoral lands (Hecht and Saatchi, 2007; than previous FRAs. Many countries have improved forest monitor-
Asner et al., 2009; Aide et al., 2013). According to Aide et al. ing and remote-sensing capacities relative to previous FRAs, most
(2013) some 22–36 Mha of net forest cover gain occurred on aban- notably in Africa, and in so doing have enhanced the degree to
doned agricultural lands over 2000–2010 across Latin America. The which they capture recovering or disperse forest mosaics in man-
correct categorisation of such tree cover is challenging where agri- aged landscapes. Countries that improved their monitoring capac-
cultural landscapes and tree cover therein are rapidly changing in ities reported net forest gains for the FRA 2015, whereas previously
their extent, form, and usage. The FAO provides countries with they reported net forest losses (Romijn et al., 2015: Fig. 9). This
guidance on defining landscapes managed for agricultural pur- trend is consistent with an enhanced detection and reporting of
poses (FAO, 2012a,c) but the degree and manner to which regener- mosaic and recently regenerated forests as forest in the current
ating and managed mosaic forest cover therein are recorded is FRA. There are also indications that improved capacity amongst
variable and not well understood by data users. forestry institutions correlates with better outcomes for forests
The extent of forest mosaics and recent forest regeneration in (Keenan et al., 2015; MacDicken et al., 2015).
nominally agricultural landscapes appears to be modest relative The time series FRA data has permitted numerous advances in
to the FRA estimates of tropical forest extent or reasonably well the science and practice of forest management, including the doc-
inventoried otherwise. No direct independent measures exist, umentation and understanding of recent declines in rates of trop-
however, and comparisons of FRA and other estimates of forest ical forest loss (e.g., Rudel, 1998). The FAO has invested heavily in
area and agricultural landscapes are complicated by differences negotiating forest definitions and norms with member countries
in definitions, forest detection capacities and other methodological for which definitions are politically sensitive and reflect the views
factors. For instance, the FRA 2015 tropical forest area having >10% of diverse stakeholders – there are more than 1500 definitions of
canopy cover (1770 Mha) is 324 Mha less than the relatively inclu- the term ‘‘forest’’ (Lund, 2006). The current FRA definition of forest
sive tropical tree-cover estimates of Hansen et al. (2013) having has been in use for 15 years, but it is inevitable that this definition
>25% canopy covey (2094 Mha). This negative discrepancy sug- does not reflect the understanding of the term forest as understood
gests a first-order estimate of such mosaic and regenerating forests by all users of the FRA. The FRA definition of forest has nonetheless
in the tropical domain. However, this estimate of 324 Mha over- brought clarity to international debates on changes in forest
looks the fact that the FRA reports some of the tree cover observed resources. The FRA does reflect realities on the ground to the extent
by Hansen et al. (2013) as ‘other wooded land’3 (OWL) rather than that these may be captured by aggregate national data, notwith-
as forest. As with the FRA forest category, OWL excludes nominally standing uncertainties concerning the categorisation of newly
agricultural landscapes. Unlike with the forest category, OWL established forest cover in semi-abandoned agricultural land-
includes an unknown and potentially large proportional area with scapes. The FRA arguably remains the best dataset for tracking
canopy cover between 5% and 10%. Given 517 Mha of OWL in the large-scale forest change trends since 1990 – a period for which
tropics, the combined area of FRA tropical forest and OWL there are very few large-scale, high-resolution remote-sensing
(2287 Mha) is actually 193 Mha greater than the Hansen et al. estimates available – and significant changes in its methodology
(2013) estimate, but this positive discrepancy would be reduced sig- might actually prove disadvantageous because they would entail
nificantly and potentially become negative if the unknown propor- interruption of the unique FRA time series. To this end, in consid-
tion of OWL with <10% canopy cover was excluded. Therefore eration of the fact that others will continue to produce global
whilst no precise discrepancy can clearly indicate the extent to satellite-based estimates of tree cover, it is important that the
which the mosaic and regenerating forests in question are reported FRA continue to provide long-term, authoritative series of esti-
as TOF rather than as forest or OWL, its potential extent appears to mates according to its existing standards. As illustrated below,
be modest relative to the many hundreds of millions of hectares of such time series provide important insights into divergent trajecto-
forest cover of >10% canopy in agricultural landscapes ries of forest change.
pan-tropically (Zomer et al., 2014).
Table 1
Change in the Area of Planted Forest, Natural Forest, and Total Forest, 1990–2015, by FRA Period and National Income Level.
Income per capita Period Change in Natural Change in Planted Change in Total Ratio of Change of Planted
Forest Area (1000s of ha) Forest Area (1000s of ha) Forest Area (1000s of ha) Forest to Natural Forest
High 1990–2000 8650 16,819 8171 1.94
(>$12,746) 2000–2005 6773 8986 2212 1.33
2005–2010 316 6 9237 12,404 2.92
2010–2015 1138 4795 3656 4.21
1990–2015 13,396 39,836 26,443 2.97
Upper-Middle 1990–2000 29,209 14,616 14,593 0.50
($4125–$12,745) 2000–2005 18,872 14,718 4153 0.78
2005–2010 13,214 9373 3841 0.71
2010–2015 7172 7954 782 1.11
1990–2015 68,467 46,661 21,805 0.68
Lower-Middle 1990–2000 13,661 3473 10,172 0.25
($1046–$4125) 2000–2005 5656 3352 2308 0.59
2005–2010 5762 3259 2507 0.57
2010–2015 6843 2820 4022 0.41
1990–2015 31,922 12,904 19,010 0.40
Low 1990–2000 29,371 527 28,835 0.02
(<$1045) 2000–2005 12,673 585 12,096 0.05
2005–2010 13,545 910 12,635 0.07
2010–2015 12,721 303 12,002 0.02
1990–2015 68,310 2326 65,568 0.03
Source: Data from the FRA of 2015. National income levels defined by World Bank (2013). See http://data.worldbank.org/about/country-and-lending-groups.
Notes: N = 126 countries. All countries were considered except those for which: (i) income level was not recorded (typically small island nations), (ii) FAO ‘desktop’ estimate
were given in lieu of nationally-reported estimates, and (iii) valid values were not reported for the tabled variables for all FRA periods (notable countries include Australia,
Cameroon, Great Britain, Indonesia, New Zealand, and Venezuela).
that forest loss is now occurring almost exclusively in the tropics, economic incentives for forest production and conservation
with either stable or expanding forest in other domains (Keenan (Foster and Rosenzweig, 2003) whilst also phasing out the
et al., 2015). Half of the world’s forest area is within sub-regions exploitation of natural public forests (Jürgensen et al., 2014).
where forest extent has been declining since 1990 (Central and Costa Rica pioneered the use of payments for environmental ser-
South America, South and South East Asia, and all three vices to landholders in order to encourage reforestation and forest
sub-regions of Africa), whilst the other half is within sub-regions conservation (Pagiola, 2008), and recently introduced laws to pro-
where forest extent is stable or increasing. tect regenerating forests from re-clearance (Fagan et al., 2013).
National wealth is an apparent determinant of this dichotomy. Puerto Rico underwent rapid urbanisation as a result of its special
Since 1990, richer countries have registered forest gains, poorer economic and migratory relationship with the USA (Rudel et al.,
countries forest losses, and many middle-income countries have 2000). Such shifts in the culture and practice of forest and land
transitioned from forest loss to forest gain (Table 1) (Keenan management have only indirect relationships with wealth per se.
et al., 2015). This is consistent with forest-transition theory Much of their effect on forest expansion was indirect and acciden-
(Rudel, 2005; Sloan, 2015) as well as environmental Kuznets tal, and arguably stemmed from changes to underlying economic
curves (Mather et al., 1999; Culas, 2012). Both conceptualisations conditions unamenable to forest conservation and regeneration.
of forest change anticipate deforestation giving way to forest This raises the question as to whether the resultant expansion of
expansion as ‘development’ proceeds. Nevertheless, in the tropics forest area was simply a fortuitous ‘one-off’ phenomenon (Müller
at least, the gain-loss dichotomy is probably driven less by wealth et al., 2014; Sloan, 2015). Ongoing forest expansion in northern
than by the improved ‘rational’ allocation of land amongst forest nations and the parallels between temperate and tropical forest
and agricultural uses, often following economic shifts and transitions gives some reason to believe forest expansion in tropi-
forward-looking government interventions (cf. Whiteman et al., cal countries with growing economies will continue.
2015). National wealth is not especially high amongst the 13 trop- The literature on the socio-economic forces underlying shifts to
ical countries identified as undergoing forest transitions since 1990 forest expansion and conservation has neglected the role of
by the FRA 2015 (Keenan et al., 2015: Table 9). Only two of these planted forests and the economics of forest production
countries have per capita GDP greater than the 75th percentile (Meyfroidt and Lambin, 2011). Yet the FRA 2015 clearly illustrates
for all 142 tropical countries (Costa Rica, Puerto Rico), and only a that planted forests4 account for much of the world’s total
further three have a per capita GDP greater than the 50th per- forest-cover change, particularly in countries with positive or low
centile (Cuba, Dominican Republic, India). rates of forest change. Indeed planted forest account for 25–100%
Since the 1990s most tropical countries apparently undergoing of the gains in forest area since 1990 in many of the tropical forest
forest transitions have also undergone processes that impacted on transition countries (Fig. 1), suggesting that planted-forest expan-
the national culture and practice of forest management and land sion is a significant and overlooked factor in forest transitions.
use, thus creating conditions wherein forest expansion could occur. Globally, countries in each of the four per capita income categories
Vietnam and Cuba underwent economic liberalisation entailing recognised by the FRA expanded their planted-forest extents over
more efficient, flexible, and privately-owned agricultural enter- all four FRA periods since 1990, with expansion rates peaking in
prises resulting in the re-establishment of forests on lands hitherto
dedicated to inefficient, often quasi-collective agriculture. Vietnam
also invested heavily in industrial plantation forestry (Rosset and 4
‘Planted forests’ encompass a variety of forest types. The FRA 2015 defines them
Benjamin, 1994; Rosset, 1998; Meyfoidt and Lambin, 2008; Payn as ‘‘predominantly composed of trees established through planting and/or deliberate
seeding’’ (www.fao.org/fra/fra2015). Planted forests are inclusive of all types of
et al., 2015). India instituted its Joint Forest Management programs
plantings, including semi-natural forests, that meet the FRA definition of forest
and similarly encouraged private forestry, giving communities (Jürgensen et al., 2014).
S. Sloan, J.A. Sayer / Forest Ecology and Management 352 (2015) 134–145 139
Fig. 1. Change in Planted Forest Area as a Percent of Change in Total Forest Area, for Tropical Forest-Transition Countries Identified by the FRA 2015, by FRA Period, 1990–
2015.Source: Data from the FRA of 2015. Notes: Graphed data pertain only to countries and FRA periods for which total forest-area change was positive. Negative percentages
indicate that a loss of planted forest area coincided with a gain of total forest area. Negative measures are not strictly percentage measures per se but are graphed here for
consistency. Philippines: The unseen negative measure for 2000–2005 is 587%. Vietnam: The sharp downward trend after the 2005–2010 FRA period owes to a relatively
small decline ( 160,000 ha) in planted-forest area between the periods 2005–2010 and 2010–2015, coincident with a continued increase in total forest area. Total forest-area
change was slightly negative for 2005–2010 ( 0.89%) but the graphed measure was retained in to permit an unbroken trend line. Puerto Rico, Gambia, Laos, and Costa Rica:
Have measures of 0% for the time series (or for 2000–2005 and 2005–2010 for Costa Rica), reflecting largely natural forest gain over these periods.
Fig. 2. Changes in the Area (Mha) of Natural Forest and Planted Forest, by Income Level and Country, 1990–2015.Source: Data from the FRA of 2015. Notes: Countries were
selected in the same manner as described for Table 1.
the mid-2000s (Table 1). Only the ‘high’ and ‘upper-middle’ income (excepting China, where natural forests also increased (Liu et al.,
categories experienced net forest gains, however, and only these 2008)) (Fig. 2). In most other high and upper-middle income coun-
countries consistently established planted forests at higher rates tries both natural-forest and planted-forests areas increased simul-
both absolutely and relative to rates of total forest change taneously, distinguishing such countries from lower-income
(Table 1). The increase in planted-forest area for high and countries (Fig. 2).
upper-middle income countries was driven predominantly by The historical situation whereby richer, temperate countries
China, Canada, the USA, Russia and Sweden, where gains in contained most planted forests is giving way to the reality that
planted-forest area more than offset losses of natural forest expanding populations and economies in the global South are
140 S. Sloan, J.A. Sayer / Forest Ecology and Management 352 (2015) 134–145
Table 2
Multiple-use Forest Area by Climate Domain, as Percent Total Forest Area, 1990 and 2015.
rapidly establishing forests in response to market opportunities. forests are under explicit ‘nature-oriented’ production regimes
Whilst the expansion of planted-forest area since 1990 was great- (Nabuurs et al., 2007, 2014), multiple-use forests account for 58%
est in temperate countries (including China), planted-forest expan- and 24% of the total forest estate, respectively, whereas in South
sion in East Asia (led by China) and South and South-East Asia (led America they account for only 13% despite having expanded
by India, Thailand, and Indonesia) was equal to or larger than 5.5-fold there since 1990. Parallel differences are also apparent
expansion in Europe (including Russia) and North America (Payn amongst the climatic domains, with the tropics having a much
et al., 2015). Globally planted forests are increasingly playing a role smaller area of multiple-use forests as a proportion of total forest
in balancing competing demands for forest conservation, amenity, area (17%) compared to other domains (Table 2). Although the pro-
and production. portional area of multiple use forests is low in the tropics the abso-
lute area of tropical multiple-use forests is still relatively large, at
287 Mha (Table 2).
4. Meeting the world’s needs for forest goods and services
Conservationists are concerned with the limited extent of pri-
mary and quality secondary forest habitat, particularly in humid
Much environmental discourse implicitly presupposes that
and seasonally-dry tropical forests (Jepson et al., 2001; Miles
maximising forest extent will provide the greatest global benefit
et al., 2006; Ribeiro et al., 2009; Portillo-Quintero and
(cf. Griscom, 2014; Planet Experts, 2014; Unger, 2014).
Sánchez-Azofeifa, 2010) where most of the world’s biodiversity
Maximising forest area is a concern amongst conservationists
resides. Agricultural expansion threatens these forests with con-
(Edwards et al., 2011; Gibson et al., 2011) and a major issue in
version (Laurance et al., 2014b). Modelling by d’Annunzio et al.
the climate change debate. Political and some corporate interests
(2015) suggests that agricultural expansion to 2030 threatens a
are committing to ‘zero deforestation’ (Butler, 2013; Lister and
small fraction (2%) of the world’s largest expanses of intact primary
Dauvergne, 2014), and significant international financing is assist-
tropical forests, thereby implying that most threatened forests will
ing tropical countries to develop and implement REDD+ programs
be relatively fragmented, degraded, and proximate to human activ-
and related forest monitoring schemes to reduce deforestation and
ity. This estimate by d’Annunzio et al. (2015) may be conservative
increase afforestation (e.g., Satgas REDD+, 2012). The governments
given the 6% loss of total primary tropical forest area since 2000
of some tropical countries may subscribe to this rhetoric, but many
(Morales-Hidalgo et al., 2015: Fig. 5) and the fact that modelling
stakeholders, including the resource poor people who live at the
by d’Annunzio et al. (2015) does not account for the penetration
forest margins, often have very different perspectives on what con-
of forests by new roads. Sayer and Cassman (2013) argue that
stitutes the optimal extent and type of forest cover (Poore and
future agricultural demand for land could be met without signifi-
Sayer, 1991). The sufficiency of the global forest estate is better
cant forest loss if agricultural innovations were deployed.
considered in terms of the provision of multiple services, e.g., car-
Even if tropical forest loss could be halted the current area and
bon, timber and biodiversity protection, which the 2015 FRA has
distribution of quality forest habitat is already insufficient for many
attempted to inventory. From this perspective there are indeed
forest species. The global biodiversity hotspots retain only 15% of
signs of an impending shortfalls of tropical forest area on multiple
intact natural vegetation (Sloan et al., 2014). One response to this sit-
fronts as increasingly more services are demanded of a decreasing
uation has been a dramatic increase in protected area coverage since
forest estate (Millennium Ecosystem Assessment, 2005). We are
1990. The proportion of forest area within protected areas has risen
witnessing the transformation of forest management, conserva-
to 7.7–16.3% (308–517 Mha) globally and 10.8–26.6% (128.5–
tion, and production so as to satisfy multiple goals simultaneously
379 Mha) in the tropics5 (Schmitt et al., 2009: Table 1;
from the same landscape (Sayer et al., 2008, 2009).
Morales-Hidalgo et al., 2015) (see also Watson et al., 2014). Yet, due
An appreciable portion of the global forest estate is already des-
to the tendency for protected areas (PAs) to be situated far from threats
ignated for multiple use allowing both production and conserva-
to forests and biodiversity (Joppa and Pfaff, 2009; Sloan et al., 2012),
tion without prioritising either. Globally 1.049 Mha or one
and given their inevitably limited extent, this expansion of
quarter of the forest estate were designated as multiple-use forests
protected-forests has been inadequate to safeguard biodiversity and
in 2015, an increase of 81.8 Mha ha (8.5%) since 1990 (Köhl et al.,
ecological services generally (Millennium Ecosystem Assessment,
2015). In contrast, the area of protective forests assessed by the
2005; Watson et al., 2014; Miura et al., 2015). Some 17% of threatened
FRA – those dedicated to protecting soils, water, and certain eco-
logical and cultural values – is at least as extensive as
multiple-use forests but did not meaningfully increase since 5
Upper estimates are from the 2015 FRA, and lower estimates derived from
1990, potentially because the protective-forest designation may remotely-sensed data tabled in Schmitt et al. (2009). As with all comparisons of FRA
preclude forest exploitation in some countries (Miura et al., and remotely-sensed data, differences in forest definitions must be considered.
Schmitt et al. (2009) attempted to align their definition with the FAO’s definitions and
2015). Multiple-use forests registered the greatest gains in South
defined forest as >10% tree cover according to 2005 MODIS satellite imagery. However
America (87.5 Mha), Oceania (52.5 Mha), and East Asia these definitions are still not totally comparable as FAO defines forest as a land use,
(48.3 Mha), and registered the greatest declines in South and not a land cover. Protected areas in Schmitt et al. (2009) have IUCN categories I-VI
Southeast Asia ( 56.5 Mha), Europe ( 31.9 Mha), and North exclusively, and whereas the FRA’s definition of protected areas is the same in this
America ( 19.6 Mha). Regional differences in the area of regard it may also encompass other areas (FAO, 2012a). Some countries, the USA for
example, report forest reserves as protected areas whereas other countries do not.
multiple-use forests relative to total forest area are great and The extent of the ‘tropics’ in Schmitt et al. (2009: Table 1) is defined with respect to
reflect differences in forest production and management practices. climate and geography whereas FAO allocates countries to a single climate domain as
In North America and Europe (including Russia), where many per FAO (2012b) even though many countries span several climatological regimes.
S. Sloan, J.A. Sayer / Forest Ecology and Management 352 (2015) 134–145 141
bird, mammal, and amphibian species (n = 4118) are not found in any The extensiveness of reduced-canopy forest cover in the tropics is
PA (n > 162,906), and 85% are so poorly represented across PAs that of concern for various reasons. Degraded forests generally provide
they may not survive in the long-term (Venter et al., 2014), a situation fewer environmental services such as biodiversity than intact natu-
that has deteriorated over the past decade (Rodrigues et al., 2004). ral forests (Gibson et al., 2011). More significantly the inclusion of
Human pressures have increasingly isolated PAs and other intact for- reduced-canopy cover forests within the FRA forest category may
ests (DeFries et al., 2005; Potapov et al., 2008). PAs are often remnants encourage the mistaken impression that the natural forest extent
left after other land use needs have been met and are inadequate to off- is adequate, or at least that the forest conservation situation is better
set a much more generalised decline in the integrity and extent of the than it actually is, leading to complacency. As yet the PCCR estimates
larger forest estate. do not provide time-series trends for forest degradation so the long
Increasing demand for forest and agricultural products threatens term fate of PCCR forests is not known. Of more immediate concern is
to undermine efforts to arrest biodiversity decline and maintain the the fact that several prominent environmental NGOs (e.g.,
integrity of the forest estate (Laurance et al., 2014b). Demographic Greenpeace) and the agri-businesses they have influenced (e.g.,
and economic growth have historically driven deforestation, forest Golden Agri-Resources, Syme Derby, PT Smart) have begun assessing
exploitation, and agricultural demand (South, 1999; Wright and forests’ conservation value and priority for conversion according to
Muller-Landau, 2006; Cohen, 2014a,b), and such growth will con- their carbon stock in efforts to direct conversion towards degraded
tinue rapidly in the developing world. Just as European consumption forests (Dinerstein et al., 2014). The application of High Carbon
of forest products increased 50% with increasing prosperity in the Stock (HCS) assessments in particular is being promoted as a precon-
latter half of the 20th Century6, emergent economies are rapidly dition for tropical forest conversion (Greenpeace SE Asia, 2014). The
increasing their consumption today. Combined industrial and fuel- problem here lies in the fact that forests subject to relatively minor
wood removals in the tropics increased by 35% (3,928,650 m3) over disturbances such as selective logging or long-cycle shifting agricul-
1990–2015 or 1.4% per annum whilst either holding constant or ture may still retain high conservation value and will regain much of
declining slightly in the other climatic domains, so that tropical their former biodiversity value if allowed to recover (Meijaard et al.,
domain is currently the greatest source of removals globally. 2005; Edwards et al., 2011). Reduced-canopy cover tropical forests
Similarly, industrial and fuelwood removals over 1990–2015 are already extensive and agricultural corporations now have a per-
increased most rapidly in lower-middle and lower income countries verse incentive to degrade forests so as to then legitimately obtain
(Köhl et al., 2015), where economic and demographic growth has been them for conversion. Stakeholders should be cautious in considering
greatest (Wilson et al., 2010; United Nations, 2013). China increased the merits of conversion on the single criterion of reduced carbon
its share of global log imports nearly three-fold to 37–50% over the last stock.
decade (Dieter, 2009; Cohen, 2014b), and demand for wood products Increasing demand for forest products and services as well as
in the Asia-Pacific region is projected to rise 80% by 2030 (FAO, 2008, for forest land for agriculture is adding to the difficulty of mitigat-
2009). Economic growth in the developing world is projected to dou- ing global climate change via REDD+ initiatives. Analyses of FRA
ble global consumption of forest products by 2030 (WWF and IIASA, 2015 data (Federici et al., 2015; Köhl et al., 2015) and FRA 2010
2012: Ch4. p. 9). Demand for industrial forest products in data (Pan et al., 2011; Reich, 2011) indicate that, despite the loss
Asia-Pacific and Africa7 may exceed forest production by 89 mil- of 10% (195 Mha) of the tropical forest estate since 1990, tropical
lion m3 by 2020, whereas Latin America may enjoy a modest surplus forests are only marginally a net source of atmospheric carbon
of 17 M m3, albeit still accompanied by forest conversion to meet agri- emissions. Carbon emissions from deforestation have been off-
cultural demand (FAO, 2008, 2009). Indeed, projections to 2030 sug- set almost entirely by carbon sequestration due to forest regener-
gest that 26% and 15% of current production forests in Latin America ation as well as increases in forest carbon density as a result of
and Africa respectively are likely to be converted for agricultural ‘carbon fertilisation’ (Fang et al., 2014). Significant proportions of
(d’Annunzio et al., 2015: Table 5). total national-level above-ground woody biomass occur as tree
The apparent contradiction of continued globally-significant cover in non-forest lands such as nominally agricultural landscapes
declines in forest biodiversity despite increases in protected areas such that estimates of carbon flux may be sensitive to definitions of
highlights the shortcomings of a narrow focus on forest extent. The forest (Schnell et al., 2015). However it remains difficult to predict
area of forest allocated for conservation, which is relatively easily whether the reported gap between gross tropical carbon emissions
measured, is important but equally important is the degree to and removals would broaden or narrow upon incorporating trees
which these forests are intact, include representative forest types, outside of forests. Tropical forests are now the only forests yielding
are inter-connected and subject to effective protection. For the first net carbon emissions according to FRA data and could become a
time the FRA 2015 has provided global figures on partial canopy net carbon sink if sustainable forest management were applied
cover reduction (PCCR), defined as the loss of >20% of tree cover only slightly more widely (Pan et al., 2011; Reich, 2011). The role
between 2000 and 2012, presumably a result of forest degradation of tropical forests in the global carbon budget will depend mainly
by fire, wood removals, small clearances, insect damage, and so on. on the rate of future conversion of forests to agriculture, the extent
Van Lierop and Lindquist (2015) report that the area subject to of which will be determined by delicate and unforeseeable com-
PCCR in the tropical domain is 6.5 times that deforested since promises between competing forces (FAO, 2008, 2009; WWF and
1990, and up to 15 times greater for South and Southeast Asia. IIASA, 2012; Laurance et al., 2014b). The projected conversions of
Whilst the ratio of PCCR to deforestation is actually higher in bor- production forests to agriculture and a corresponding intensifica-
eal and subtropical domains than in the tropical domain, this is tion of production forestry in Latin America and Africa would
because deforestation in the former domains was negligible, not almost certainly entail net emissions, whereas achieving the
because PCCR was extensive. national targets of net forest-area increases across Asia and
Africa would entail the contrary (see Jürgensen et al., 2014 for
6
Whilst consumption increased 54% over 1964–2000, natural forest exploitation China; d’Annunzio et al., 2015: Table 6). Tropical forests still have
increased only 9%, and the total forest supply of timber increased, largely due by far the smallest proportion of their total area under an interme-
significant increases in the efficiency of forest harvest, forest-product manufacture, diate or good level of sustainable forest management, at 23%
recycling as well as forest protection (Nabuurs et al., 2007).
7
(MacDicken et al., 2015: Fig. 2) High standards of sustainable man-
Production and consumption in Africa is much lower than in the Asia-Pacific
region. The African contribution to the 2020 production deficiency reported here is
agement greatly increase the likelihood that a forest will serve as
probably under-estimated because these figures do not consider fuelwood, which carbon sink (Putz and Pinard, 1993) so there is much scope for gain
accounts for 88% of total wood removals in Africa (FAO, 2010: Table 5.11). in this respect. Large areas of tropical forest are legally designated
142 S. Sloan, J.A. Sayer / Forest Ecology and Management 352 (2015) 134–145
for sustainable permanent use but in practice remain vulnerable to (MacDicken et al., 2015). All forest areas are subject to national
unchecked exploitation (Blaser et al., 2011; MacDicken et al., policies and legislation which promote SFM but low-income coun-
2015), and indeed virtually all tropical forest loss since 1990 has tries are in general failing to enact policy and legislation to the
occurred in state lands (Whiteman et al., 2015: Table 2). local scale, with potential implications for local conservation and
production. National forest inventories (NFIs) and forest manage-
ment plans (FMPs) appear critical for practical near-term improve-
4.1. The growing role of planted forests ment in SFM in such countries. Presently 309 Mha of permanent
tropical forest is subject to policy and legislation but not covered
Growing demand for forest products makes a strong case for by NFIs, a far greater area than in other climatic domains abso-
expanding planted forests. Planted forests may reduce pressure on lutely and relative to total forest area. Fortunately NFIs and FMPs
natural forests (South, 1999), support biodiversity conservation are responsive to support from the international community and
(Sayer et al., 2004; Sayer and Elliot, 2005; Brockerhoff et al., 2008), are efficacious means of advancing various aspects of SFM. The
and actively remove atmospheric carbon (Nilsson and number of countries with NFIs more than doubled to 112 since
Schopfhauser, 1995). There is significant scope to increase produc- 2010 thanks largely to support by UN-REDD+ and the FAO in
tion from planted forests in several subregions according to the best poorer countries, although coverage remains low in Africa.
available sub-regional data on timber production from planted for- Where implemented, FMPs have explicitly considered
ests and from all forests combined (natural and planted), which col- high-conservation value forest and community involvement in
lectively account for 60% of global timber production from natural over 90% of their area, suggesting they may serve as portals to
and planted forests. Planted forests contribute a low to moderate broad and effective SFM generally.
proportion of total production in Africa (10–30%, depending on the Care must be taken that any concentration of production in a
sub-region), Central America (34%), and South East Asia (49%) (see planted-forest estate does not devalue natural forests and discour-
also Jürgensen et al., 2014; d’Annunzio et al., 2015: Table 1). age investment in their sustainable management. In Borneo the des-
Modelling based on FRA and FAOSTAT data indicates that if the glo- ignation of natural production forest has historically prevented
bal extent of planted forests were to increase at 2.4% per annum forest conversion as effectively as protected forests (Curran et al.,
between 2010 and 2050 planted forests could replace natural forests 2004; Gaveau et al., 2013). With the depletion of many natural pro-
as a source of timber and fibre (WWF and IIASA, 2012: Ch.4). This duction forests following decades of encroachment, fire, and logging,
assumes that 80% of current planted forests are dedicated to produc- forest royalties declined correspondingly (ITS Global, 2011; Ministry
tion.8 This target rate it is not dramatically different to the 2.05% per of Forestry, 2012; Gaveau et al., 2014). The Indonesian government
annum increase in planted forest expansion reported by the FRA for has since enacted policies to encourage the establishment of forest
1990–2015. However, the actual global rate of planted-forest estab- plantations, and in many cases these have replaced degraded natural
lishment has fallen to 1.5% per annum since 2005. Target rates are production forests (Gaveau et al., 2014: Table S5). Agricultural
higher in the Asia-Pacific region (excluding China) and Latin expansion has similarly often occurred on poorly stocked or other-
America, at 3.2% and 3% per annum respectively, as compared to their wise degraded natural forests (Abood et al., 2015). Should fibre pro-
still appreciable actual rates of planted-forest expansion of 2.25% and duction increasingly derive from planted forests investments in
2.4% per annum for 1990–2015, underscoring the magnitude of the sustainable forest management in natural forests must arguably give
challenge of meeting wood demands exclusively from planted forests. increased priority to environmental, rather than economic, values,
An expanded planted-forest estate is ultimately only one means as by payments for environmental services.
of many necessary to achieve better natural forest conservation FRA data on public expenditures on forests are mute in regards to
and management. There is a time lag between establishing planted the degree to which expenditure is directed towards natural or
forests and achieving significant wood production and carbon planted forests or towards conservation or production, as these tar-
sequestration at a national to regional scale (Nilsson and gets are highly intertwined at the national scale. The case of China is
Schopfhauser, 1995; Nabuurs et al., 2007, 2014). An extensive or illustrative. Expenditures in temperate countries including China
rapidly expanding planted-forests estate must complement sus- increased significantly over 2000–2010 from $9.6 billion to $25 bil-
tainable forest management more generally to yield functional, lion whilst public revenues from forests were much lower and
sustainable landscapes providing a full range of forest products increased relatively modestly from $1.1 billion to $2.7 billion
and environmental services (Sayer et al., 2004). In South America, (Whiteman et al., 2015). Superficially this dominance of expendi-
where planted forests meet 77–88% of total wood demand tures over revenues, which is by far the greatest in the temperate
(Jürgensen et al., 2014; d’Annunzio et al., 2015: Table 1), deforesta- domain, suggests significant investment in forest management and
tion is still greater than in other continents both absolutely and rel- monitoring apart from production per se. China accounts for 62%
atively in part because only 15% of forests there are subject to an of this growth in expenditure in the temperate domain and just less
intermediate level of sustainable forest management or better than half of its total expenditure in 2010. In China forest area
(MacDicken et al., 2015: Fig. 5). Modest areas of multiple-use for- increased since 2000 through state-sponsored initiatives such as
ests and lower grades of sustainable forest management in many the National Forest Conservation Program and ‘Grains for Green’
low-income and tropical countries suggest that improved manage- programs entailing natural forest restoration, payments for environ-
ment could allow production from natural forests to increase with- mental services, reduced natural forest harvests and planted-forest
out undue threats to forest conservation, integrity, and establishment for environmental objectives (Zhang et al., 2000; Xu
biodiversity. et al., 2007; Liu et al., 2008). The programs also entailed massive sub-
Low-income countries and the tropics have by far the greatest sidised expansions of timber plantations for industrial production
scope to improve sustainable forest management (SFM) (Fig. 2). In China at least, the significant economic value invested in
planted forests for production is inseparable from the significant
8
Using earlier FAO data Sayer and Elliot (2005) report that 46% of the global non-economic values recently assigned to natural forests following
planted-forest estate is explicitly dedicated to industrial production (e.g. timber, their widespread degradation (Liu et al., 2008). In the tropical
fibre), 26% is dedicated to nonindustrial uses (e.g., fuelwood, soil protection), and the domain, where planted forests are less extensive, forest revenues
remaining 26% has an unclear purpose, almost certainly being a mix of industrial and
nonindustrial uses. Similarly, FAO (2010: 339) observes that ‘productive plantations’
exceeded expenditures and both increased far more modestly than
accounted for 79% of the global planted-forest estate in 2005, with the balance in the temperate domain, at +$2.9 billion to $5.5 billion and at
accounted for by protective planted forests. +$0.5 billion to $1.1 billion, respectively (Whiteman et al., 2015).
S. Sloan, J.A. Sayer / Forest Ecology and Management 352 (2015) 134–145 143
5. Novel forests landscapes degradation is mostly a result not of economic and demographic
forces per se but of institutional failures to contain such forces.
In its earliest days the FRA focussed narrowly on the manage- Information alone cannot resolve the tensions between values
ment of forests as sources of industrial timber but has since pro- attributed by different stakeholders or the power differentials
gressively adapted its scope to reflect a much broader range of between local people and corporations. But the FRA continues to
forest attributes and values. This trend must continue as the have an essential role in reconciling such tensions, both by stimu-
demands that societies place upon forests continue to grow and lating discussion amongst stakeholders nationally during the FRA
vary. Forests are ever more scarce, contested, and valuable, and for- exercise (MacDicken, 2015) and by providing an over-arching
est landscapes are ever more diverse and novel as they react to dif- framework under which global, national, and local management
ferent societal needs and occur in changing environmental initiatives may flourish. The ultimate challenge is not to know
conditions (Lugo, 2009). The extent of natural and semi-natural how much forest exists but rather to what degree forests are meet-
forests is growing in many human-dominated, multi-functional ing the varied needs of humanity.
landscapes (Aide et al., 2013; Zomer et al., 2014), and these forests
are playing a growing role in sustaining production and environ- Acknowledgements
mental services (Brockerhoff et al., 2008; Sayer et al., 2013).
Landscapes are emerging as an important organising framework Sean Sloan is supported by an ARC – Australia Laureate
for the provision of multiple forest services and are the object of Fellowship awarded to Prof. William Laurance. We thank Ken
significant international debate. Recently global landscape fora MacDicken of the FAO and two anonymous reviewers for valuable
have taken place in association with the UNFCCC Conferences of comments on an earlier draft of this paper.
the Parties and several international bodies have committed them-
selves to a Global Programme for Forest Landscape Restoration
(Global Landscapes Forum, 2014). These novel forest landscapes References
will be difficult to accommodate within the current FRA reporting
framework and will similarly present challenges for Abood, S.A., Huay Lee, J.S., Burivalova, Z., Garcia-Ulloa, J., Koh, L.P., 2015. Relative
remotely-sensed estimates. Metrics for assessing the diverse con- contributions of the logging, fibre, oil palm and mining industries to forest loss
in Indonesia. Conserv. Lett. 8, 58–67.
servation and production values of trees and forests in mosaic Achard, F., Beuchle, R., Mayaux, P., Stibig, H.J., Bodart, C., Brink, A., Carboni, S.,
landscapes are currently inadequate and urgently needed. Desclée, B., Donnay, F., Eva, H.D., 2014. Determination of tropical deforestation
Tensions exist between the discourse emphasising global forest rates and related carbon losses from 1990 to 2010. Glob. Change Biol. 20, 2540–
2554.
conservation benefits and the need to alleviate the poverty of local Aide, T.M., Clark, M.L., Grau, H.R., López-Carr, D., Levy, M.A., Redo, D., Bonilla-
forest dwellers (Kremen et al., 2000). These tensions signal Moheno, M., Riner, G., Andrade-Núñez, M.J., Muñiz, M., 2013. Deforestation and
‘wicked’ problems where achieving agreement on even the nature reforestation of Latin America and the Caribbean (2001–2010). Biotropica 45,
262–271.
of the problem is difficult and where solutions for one stakeholder
Angelsen, A., Rudel, T.K., 2013. Designing and implementing effective REDD+
create problems for another (Giller et al., 2008; Stewart et al., policies: a forest transition approach. Rev. Environ. Econ. Policy 7, 91–113.
2011). There has been a recent tendency to decentralize the locus Asner, G.P., Rudel, T.K., Aide, T.M., DeFries, R., Emerson, R., 2009. A contemporary
assessment of global humid tropical forest change. Conserv. Biol. 23, 1386–
of decision making on forests towards the local scale (MacDicken
1395.
et al., 2015), typically with a good measure of improvisation as Asner, G.P., Llactayo, W., Tupayachi, R., Luna, E.R., 2013. Elevated rates of gold
decision makers struggle with notions of ‘balance’ (Sayer et al., mining in the Amazon revealed through high-resolution monitoring. Proc. Natl.
2008). Countries will make their own decisions on how much for- Acad. Sci. USA 110, 18454–18459.
Beuchle, R., Grecchi, R.C., Shimabukuro, Y.E., Seliger, R., Eva, H.D., Sano, E., Achard, F.,
est to preserve how it should be managed as their needs evolve in 2015. Land cover changes in the Brazilian Cerrado and Caatinga biomes from
the 21st Century. The FRA has provided the main 1990 to 2010 based on a systematic remote sensing sampling approach. Appl.
inter-governmental framework for debate on forest needs and val- Geography 58, 116–127.
Blaser, J., Sarra, A., Poore, D., Johnson, S., 2011. Status of Tropical Forest
ues and it should continue to do so as those values evolve over Management 2011. ITTO Technical Series, No. 38. The ITTO. Yokohama, Japan.
time. Forest goods and services in future are likely to derive from Brockerhoff, E.G., Jactel, H., Parrotta, J.A., Quine, C.P., Sayer, J., 2008. Plantation
an increasingly wider range of land cover types and jurisdictions forests and biodiversity: oxymoron or opportunity? Biodivers. Conserv. 17,
925–951.
than presently. Private forests, community forests, planted forests, Butchart, S.H., Walpole, M., Collen, B., van Strien, A., Scharlemann, J.P., Almond, R.E.,
and trees outside forests will need to be distributed in Baillie, J.E., Bomhard, B., Brown, C., Bruno, J., 2010. Global biodiversity:
multi-functional landscapes in ways that optimise the provision indicators of recent declines. Science 328, 1164–1168.
Butler, R., 2013. The beginning of the end of deforestation in Indonesia? <http://
of the goods and services that societies will need. The challenge
news.mongabay.com/2013/0205-app-forest-policy.html> (accessed December
in the future will not be to quantify forest-area changes per se 2014).
but rather to assess the ability of landscapes to meet a diversity Chazdon, R.L., 2014a. PARTERS: Building a Socio-Ecological Understanding of
Reforestation in the Tropics. <http://peoplefoodandnature.org/blog/partners-
of societal needs. Difficult choices will need to be made at all spa-
building-a-socio-ecological-understanding-of-reforestation-in-the-tropics/>
tial scales, from local to global, and metrics that measure the abil- and <http://partners-rcn.uconn.edu/page.php?4> (accessed 06.01.15).
ity of the broader landscape to provide a balance of goods and Chazdon, R.L., 2014b. Second Growth: The Promise of Tropical Forest Regeneration
services will enable better decisions on these choices. in an Age of Deforestation. University of Chicago Press, Chicago.
Chazdon, R.L., Harvey, C.A., Komar, O., Griffith, D.M., Ferguson, B.G., Martínez-
Ramos, M., Morales, H., Nigh, R., Soto-Pinto, L., Van Breugel, M., 2009. Beyond
reserves: a research agenda for conserving biodiversity in human-modified
5.1. An expanding need for Forest Resources Assessments tropical landscapes. Biotropica 41, 142–153.
Cohen, D., 2014a. Drivers of future forest use. Paper presented at The XXIV IUFRO
World Congress. ’Sustaining Forets, Sustaining People: The Role of Research’,
The greatest challenge to improved global forest management is Salt Lake City, USA, 7 October 2014.
ultimately not the information provided by the FRA and similar Cohen, D., 2014b. Forces shaping the globe’s forests and forest use. In: Nikolakis, W.,
Innes, J. (Eds.), Forests and Globalization: Challenges and Opportunities for
sources but rather the capacity of institutions and civil society in
Sustainable Development. Routledge, New York (Ch. 2).
low-income countries to exploit this information. The FRA is the Collins, N.M., Sayer, J.A., Whitmore, T.C., 1991. The Conservation Atlas of Tropical
apex of a hierarchy of management tools extending down to Forests: Asia and the Pacific. Macmillan Press Ltd.
sub-national inventories and local consultations. Where forests Coulston, J.W., Reams, G.A., Wear, D.N., Brewer, C.K., 2013. An analysis of forest land
use, forest land cover and change at policy-relevant scales. Forestry, cpt056.
continue to decline in poorer tropical countries there is often a Culas, R.J., 2012. REDD and forest transition: tunneling through the environmental
weak link between national and local levels. Forest loss and Kuznets curve. Ecol. Econ. 79, 44–51.
144 S. Sloan, J.A. Sayer / Forest Ecology and Management 352 (2015) 134–145
Curran, L.M., Trigg, S.N., McDonald, A.K., Astiani, D., Hardiono, Y.M., Siregar, P., Grainger, A., 1996. An evaluation of the FAO tropical resource assessment 1990.
Caniago, I., Kasischke, E., 2004. Lowland forest loss in protected areas of Geography J. 162, 73–79.
Indonesian Borneo. Science 303, 1000–1003. Grainger, A., 2008. Difficulties in tracking the long-term global trend in tropical
d’Annunzio, R., Sandker, M., Finegold, Y., Min, Z., 2015. Projecting global forest area forest area. Proc. Natl. Acad. Sci. USA 105, 818–823.
towards 2030. For. Ecol. Manage. 352, 124–133. Grainger, A., 2010. Uncertainty in the construction of global knowledge of tropical
de Wasseige, C., Devers, D., De Marcken, P., Eba’a Atyi, R., Nasi, R., Mayaux, P., 2010. forests. Prog. Phys. Geogr. 34, 811–844.
The Forests of the Congo Basin: State of the Forest 2008. Publications Office of Greenpeace SE Asia, 2014. High Carbon Stock Approach – WWF, GP, RAN, FPP Joint
the European Union, Luxembourg. Briefer. What is the HCS Approach? <http://www.greenpeace.org/seasia/Press-
DeFries, R., Hansen, A., Newton, A.C., Hansen, M.C., 2005. Increasing isolation of Centre/Press-Releases/HCS-Briefer/> (accessed April 2015).
protected areas in tropical forests over the past twenty years. Ecol. Appl. 15, 19– Griscom, B., 2014. A Fully Loaded, Double-Barreled Forest Climate Solution,
26. Accessed March 2015, http://blog.nature.org/science/2014/09/24/forest-
Dieter, M., 2009. Analysis of trade in illegally harvested timber: Accounting for nadine-unger-plant-trees-climate/.
trade via third party countries. For. Econ. 11, 600–607. Gutiérrez-Vélez, V.H., Ruth, D., Miguel, P.-V., María, U., Christine, P., Walter, B., Katia,
Dinerstein, E., Baccini, A., Anderson, M., Fiske, G., Wikramanayake, E., McLaughlin, F., Yili, L., 2011. High-yield oil palm expansion spares land at the expense of
D., Powell, G., Olson, D., Joshi, A., 2014. Guiding agricultural expansion to spare forests in the Peruvian Amazon. Environ. Res. Lett. 6, 044029.
tropical forests. Conserv. Lett. http://dx.doi.org/10.1111/conl.12149. Hansen, M., Stehman, S.V., Potapov, P.V., 2010. Quantification of global forest cover
Durán, A.P., Rauch, J., Gaston, K.J., 2013. Global spatial coincidence between loss. Proc. Natl. Acad. Sci. USA 107, 8650–8655.
protected areas and metal mining activities. Biol. Conserv. 160, 272–278. Hansen, M.C., Potapov, P.V., Moore, R., Hancher, M., Turubanova, S.A., Tyukavina, A.,
Edwards, D.P., Larsen, T.H., Docherty, T.D.S., Ansell, F.A., Hsu, W.W., Derhé, M.A., Thau, D., Stehman, S.V., Goetz, S.J., Loveland, T.R., Kommareddy, A., Egorov, A.,
Hamer, K.C., Wilcove, D.S., 2011. Degraded lands worth protecting: the Chini, L., Justice, C.O., Townshend, J.R.G., 2013. High-resolution global maps of
biological importance of Southeast Asia’s repeatedly logged forests. Proc. R. 21st-century forest cover change. Science 342, 850–853.
Soc. Lond., B 278, 82–90. Hecht, S.B., 2005. Soybeans, development and conservation on the Amazon Frontier.
Edwards, D.P., Sloan, S., Weng, L., Sayer, J., Dirks, P., Laurance, W.F., 2014. Mining Dev. Change 36, 375–404.
and the African environment. Conserv. Lett. 7, 302–311. Hecht, S.B., Saatchi, S.S., 2007. Globalization and forest resurgence: changes in forest
Fagan, M.E., DeFries, R.S., Sesnie, S.E., Arroyo, J.P., Walker, W., Soto, C., Chazdon, R.L., cover in El Salvador. Bioscience 57, 663–673.
Sanchun, A., 2013. Land cover dynamics following a deforestation ban in Hecht, S.B., Kandel, S., Gomes, I., Cuellar, N., Rosea, H., 2006. Globalization, forest
northern Costa Rica. Environ. Res. Lett. 8, 034017. resurgence and environmental politics in El Salvador. World Dev. 34, 308–323.
Fang, J., Kato, T., Guo, Z., Yang, Y., Hu, H., Shen, H., Zhao, X., Kishimoto-Mo, A.W., Indrarto, G.B., Murharjanti, P., Khatarina, J., Pulungan, I., Ivalerina, F., Rahman, J.,
Tang, Y., Houghton, R.A., 2014. Evidence for environmentally enhanced forest Prana, M.N., Resosudarmo, I.A.P., Muharrom, E., 2012. The context of REDD+ in
growth. Proc. Natl. Acad. Sci. USA 111, 9527–9532. Indonesia. Working Paper 92. Center for International Forestry Research
FAO, 2001. Global Forest Resource Assessment 2000. Forestry Paper No. 140, Food (CIFOR). Bogor, Indonesia.
and Agricultural Organization of the United Nations. Rome. ITS Global, 2011. The Economic Contribution of Indonesia’s Forest-Based Industries
FAO, 2008. Global Forest Product Projections, by R. Jonsson and A. Whiteman. The (Draft Report). ITS Global, Melbourne, p. 49. <http://www.itsglobal.net/sites/
Food and Agricultural Organization of the United Nations (FAO). Rome, Italy. default/files/itsglobal/ITS_Indoforest_Economic_Report.pdf>.
FAO, 2009. State of the World’s Forests 2009. The Food and Agricultural Jepson, P., Jarvie, J.K., MacKinnon, K., Monk, K.A., 2001. The end for Indonesia’s
Organisation of the United Nations (FAO). Rome, Italy. lowland forests? Science 292, 859–861.
FAO, 2010. Global Forest Resources Assessment 2010: Main Report. FAO Forestry Joppa, L.N., Pfaff, A., 2009. High and far: biases in the location of protected areas.
Paper, No. 163. The Food and Agricultural Organization of the United Nations PLoS One 4, e8273.
(FAO). Rome. Jürgensen, C., Kollert, W., Lebedys, A., 2014. Assessment of Industrial Roundwood
FAO, 2012a. FRA 2015 Terms and Definitions. Forest Resources Assessment Working Production from Planted Forests. . Planted Forests and Trees Working Paper
Paper 180. The Food and Agricultural Organization of the United Nations (FAO). Series, No. FP/48/E. The Food and Agricultural Organization of the United
Rome. Nations (FAO). Rome.
FAO, 2012b. Global Ecological Zones for FAO Forest Reporting: 2010 Update. Forest Keenan, R.J., Reams, G.A., Achard, F., de Freitas, J., Grainger, A., Lindquist, E., 2015.
Resources Assessment Working Paper 179. Forest Resources Assessment Dynamics of global forest area: results from the FAO Global Forest Resources
Working Paper 179. The Food and Agricultural Organization of the United Assessment 2015. For. Ecol. Manage. 352, 9–20.
Nations (FAO). Rome. Kim, D.-H., Sexton, J.O., Townshend, J.R., 2015. Accelerated deforestation in the
FAO, 2012c. Guide for Country Reporting for FRA 2015. FRA 2015 Working Paper, humid tropics from the 1990s to the 2000s. Geophys. Res. Lett., 2014GL062777
the Food and Agricultural Organization of the United Nations (FAO). Rome. Köhl, M., Lasco, R., Cifuentes, M., Jonsson, O., Korhonen, K., Mundhenk, P., de Jesus
FAO, 2015. Global Forest Resources Assessment 2015. FAO Forestry Paper No. 1. The Navar, J., Stinson, G., 2015. Changes in forest production, biomass and carbon:
Food and Agricultural Organization of the United Nations (FAO). Rome. results from the 2015 UN Global Forest Resources Assessment. For. Ecol.
FAO, JRC, 2012. Global Forest Land-Use Change 1990–2005. FAO Forestry Paper No. Manage. 352, 21–34.
169. The Food and Agricultural Organization of the United Nations (FAO) with Kremen, C., Niles, J.O., Dalton, M.G., Daily, G.C., Ehrlich, P.R., Fay, J.P., Grewal, D.,
the E.U. Joint Research Centre (JRC). Rome. Guillery, R.P., 2000. Economic incentives for rain forest conservation across
FAO, JRC, 2014. Global Forest Land-Use Change 1990–2010. An update to FAO scales. Science 288, 1828–1832.
Forestry Paper No. 169. The Food and Agricultural Organization of the United Lambin, E.F., Meyfroidt, P., 2011. Global land use change, economic globalization,
Nations (FAO) with the E.U. Joint Research Centre (JRC). Rome. and the looming land scarcity. Proc. Natl. Acad. Sci. USA 108, 3465–3472.
Federici, S., Tubiello, F.N., Salvatore, M., Jacobs, H., 2015. New estimates of CO2 Laurance, W.F., Clements, G.R., Sloan, S., O/’Connell, C.S., Mueller, N.D., Goosem, M.,
forest emissions and removals: 1990–2015. For. Ecol. Manage. 352, 89–98. Venter, O., Edwards, D.P., Phalan, B., Balmford, A., Van Der Ree, R., Arrea, I.B.,
Foster, A., Rosenzweig, M.R., 2003. Economic growth and the rise of forests. Q. J. 2014a. A global strategy for road building. Nature 513, 229–232.
Econ. 118, 601–637. Laurance, W.F., Sayer, J., Cassman, K.G., 2014b. Agricultural expansion and its
Friedl, M.A., Sulla-Menashe, D., Tan, B., Schneider, A., Ramankutty, N., Sibley, A., impacts on tropical nature. Trends Ecol. Evol. 29, 107–116.
Huang, X., 2010. MODIS collection 5 global land cover: algorithm refinements Lister, J., Dauvergne, P., 2014. Voluntary zero net deforestation. In: Nikolakis, W.,
and characterization of new datasets. Remote Sens. Environ. 114, 168–182. Innes, J. (Eds.), Forests and Globalization: Challenges and Opportunities for
Gaveau, D.L.A., Kshatriya, M., Sheil, D., Sloan, S., Wijaya, A., Wich, S., Ancrenaz, M., Sustainable Development. Routledge, New York (Ch. 2).
Hansen, M., Broich, M., Molidena, E., Guariguata, M., Pacheco, P., Potapov, P., Liu, J., Li, S., Ouyang, Z., Tam, C., Chen, X., 2008. Ecological and socioeconomic effects
Turubanova, S., Meijaard, E., 2013. Reconciling forest conservation and logging of China’s policies for ecosystem services. Proc. Natl. Acad. Sci. 105, 9477–9482.
in Indonesian Borneo. PLoS One 8, e69887. Lugo, A.E., 2009. The emerging era of novel tropical forests. Biotropica 41, 589–
Gaveau, D.L.A., Sloan, S., Molidena, M., Husnayanem, Wijaya, A., Ancrenaz, M., Nasi, 591.
R., Wielaard, N., Meijaard, E., 2014. Four decades of forest loss and degradation Lugo, A.E., Helmer, E., 2004. Emerging forests on abandoned land: Puerto Rico’s new
in Borneo. PLoS One 9, e101654. forests. For. Ecol. Manage. 190, 145–161.
Gibson, L., Lee, T.M., Koh, L.P., Brook, B.W., Gardner, T.A., Barlow, J., Peres, C.A., Lund, H.G., 2006. Definitions of forest, deforestation, afforestation, and
Bradshaw, C.J.A., Laurance, W.F., Lovejoy, T.E., Sodhi, N.S., 2011. Primary forests reforestation. For. Inf. Services.
are irreplaceable for sustaining tropical biodiversity. Nature 478, 378–381. MacDicken, K.G., Sola, P., Hall, J., Sabogal, C., Tadoum, M., de Wasseige, C., 2015.
Giller, K.E., Leeuwis, C., Andersson, J.A., Andriesse, W., Brouwer, A., Frost, P., Global progress toward sustainable forest management. For. Ecol. Manage. 352,
Hebinck, P., Heitkönig, I., Van Ittersum, M.K., Koning, N., 2008. Competing 47–56.
claims on natural resources: what role for science? Ecol. Soc., 13 (Article 34) MacDicken, K.G., 2015. Global Forest Resources Assessment 2015: what, why and
Global Landscapes Forum, 2014. Global Landscapes Forum. Convened in parallel how? For. Ecol. Manage. 352, 3–8.
with the UNFCCC COP20 meeting. 2014 Global Landsdcapes Forum: Landscapes Masters, W.A., Djurfeldt, A.A., De Haan, C., Hazell, P., Jayne, T., Jirström, M., Reardon,
for a New Climate and Development Agenda, Lima, Peru, 6–7 December 2014. T., 2013. Urbanization and farm size in Asia and Africa: implications for food
Gong, P., Wang, J., Yu, L., Zhao, Y., Zhao, Y., Liang, L., Niu, Z., Huang, X., Fu, H., Liu, S., security and agricultural research. Global Food Security 2, 156–165.
Li, C., Li, X., Fu, W., Liu, C., Xu, Y., Wang, X., Cheng, Q., Hu, L., Yao, W., Zhang, H., Mather, A.S., Needle, C.L., Fairbairn, J., 1999. Environmental Kuznets curves and
Zhu, P., Zhao, Z., Zhang, H., Zheng, Y., Ji, L., Zhang, Y., Chen, H., Yan, A., Guo, J., Yu, forest trends. Geography 84, 55–65.
L., Wang, L., Liu, X., Shi, T., Zhu, M., Chen, Y., Yang, G., Tang, P., Xu, B., Giri, C., Meijaard, E., Sheil, D., Nasi, R., Augeri, D., Rosenbaum, B., Iskandar, D., Setyawati, T.,
Clinton, N., Zhu, Z., Chen, J., Chen, J., 2013. Finer resolution observation and Lammertink, A., Rachmatika, I., Wong, A., Soehartono, T., Stanley, S., O’Brien, T.,
monitoring of global land cover: first mapping results with Landsat TM and 2005. Life after Logging: Reconciling Wildlife Conservation and Production
ETM+ data. Int. J. Remote Sens. 34, 2607–2654. Forestry in Indonesian Borneo. CIFOR, Bogor, Indonesia.
S. Sloan, J.A. Sayer / Forest Ecology and Management 352 (2015) 134–145 145
Meyfoidt, P., Lambin, E., 2008. Forest transitions in Vietnam and its environmental Sayer, J., 2009. Reconciling conservation and development: are landscapes the
impacts. Glob. Change Biol. 14, 1–18. answer? Biotropica 41, 649–652.
Meyfroidt, P., Lambin, E.F., 2011. Global forest transition: prospects for an end to Sayer, J., Cassman, K.G., 2013. Agricultural innovation to protect the environment.
deforestation. Ann. Rev. Environ. Resour. 36, 9.1–9.29. Proc. Natl. Acad. Sci. 110, 8345–8348.
Miles, L., Newton, A.C., DeFries, R.S., Ravilious, C., May, I., Blyth, S., Kapos, V., Gordon, Sayer, J., Elliot, C., 2005. The role of commercial plantations in forest landscape
J.E., 2006. A global overview of the conservation status of tropical dry forests. J. restoration. In: Mansourian, S., Vallauri, D., Dudley, N. (Eds.), Forest Restoration
Biogeogr. 33, 491–505. in Landscapes: Beyond Planting Trees. Springer, New York, pp. 379–383, 54.
Millennium Ecosystem Assessment, 2005. Ecosystems and Human Well-Being: Sayer, J., Chokkalingam, U., Poulsen, J., 2004. The restoration of forest biodiversity
Biodiversity Synthesis. United Nations Environmental Program and the World and ecological values. For. Ecol. Manage. 201, 3–11.
Resources Institute, Washington, D.C. Sayer, J., Bull, G., Elliott, C., 2008. Mediating forest transitions:’Grand
Ministry of Forestry, 2012. Statistik Kehutanan Indonesia 2011 (Forestry Statistics design’or’Muddling through’. Conserv. Soc. 6, 320.
of Indonesia 2011). Kementerian Kehutanan Direktorat Jenderal Planologi Sayer, J., Sunderland, T., Ghazoul, J., Pfund, J.-L., Sheil, D., Meijaard, E., Venter, M.,
Kehutanan (Ministry of Forestry Directorate General for Forestry Planning). Boedhihartono, A.K., Day, M., Garcia, C., 2013. Ten principles for a landscape
Jakarta. approach to reconciling agriculture, conservation, and other competing land
Miura, S., Amacher, M., Hofer, T., San-Miguel-Ayanz, J., Ernawati, Thackway, R., uses. Proc. Natl. Acad. Sci. USA 110, 8349–8356.
2015. Protective functions and ecosystem services of global forests in the past Schmitt, C.B., Burgess, N.D., Coad, L., Belokurov, A., Besançon, C., Boisrobert, L.,
quarter-century. For. Ecol. Manage. 352, 35–46. Campbell, A., Fish, L., Gliddon, D., Humphries, K., Kapos, V., Loucks, C., Lysenko,
Morales-Hidalgo, D., Oswalt, S.N., Somanathon, E., 2015. Conserving forests for I., Miles, L., Mills, C., Minnemeyer, S., Pistorius, T., Ravilious, C., Steininger, M.,
biodiversity: status and trends from the Global Forest Resources Assessment Winkel, G., 2009. Global analysis of the protection status of the world’s forests.
2015. For. Ecol. Manage. 352, 68–77. Biol. Conserv. 142, 2122–2130.
Müller, D., Sun, Z., Vongvisouk, T., Pflugmacher, D., Xu, J., Mertz, O., 2014. Regime Schnell, S., Altrell, D., Ståhl, G., Kleinn, C., 2015. The contribution of trees outside
shifts limit the predictability of land-system change. Global Environ. Change 28, forests to national tree biomass and carbon stocks—a comparative study across
75–83. three continents. Environ. Monit. Assess. 187, 1–18.
Nabuurs, G.-J., Pussinen, A., Van Brusselen, J., Schelhaas, M., 2007. Future harvesting Sexton, J.O., Song, X.-P., Feng, M., Noojipady, P., Anand, A., Huang, C., Kim, D.-H.,
pressure on European forests. Eur. J. Forest Res. 126, 391–400. Collins, K.M., Channan, S., DiMiceli, C., Townshend, J.R., 2013. Global, 30-m
Nabuurs, G.-J., Schelhaas, M.-J., Orazio, C., Hengeveld, G., Tome, M., Farrell, E.P., resolution continuous fields of tree cover: landsat-based rescaling of MODIS
2014. European perspective on the development of planted forests, including vegetation continuous fields with lidar-based estimates of error. Int. J. Digital
projections to 2065. NZ J. Forest. Sci. 44, S8. Earth 6, 427–448.
Nilsson, S., Schopfhauser, W., 1995. The carbon-sequestration potential of a global Sloan, S., 2014. Indonesia’s moratorium on new forest licenses: an update. Land Use
afforestation program. Climatic Change 30, 267–293. Policy 38, 37–40.
Pagiola, S., 2008. Payments for environmental services in Costa Rica. Ecol. Econ. 65, Sloan, S., 2015. Development-driven recovery of tropical forests: the tropical forest
712–724. transition with implications for REDD+. Ecol. Econ. 116, 1–11.
Pan, Y., Birdsey, R.A., Fang, J., Houghton, R., Kauppi, P.E., Kurz, W.A., Phillips, O.L., Sloan, S., Edwards, D.P., Laurance, W.F., 2012. Does Indonesia’s REDD+ moratorium
Shvidenko, A., Lewis, S.L., Canadell, J.G., 2011. A large and persistent carbon sink on new concessions spare imminently-threatened forests? Conserv. Lett. 5,
in the world’s forests. Science 333, 988–993. 222–231.
Pandit, M., Sodhi, N.S., Koh, L.P., Bhaskar, A., Brook, B.W., 2007. Unreported yet Sloan, S., Jenkins, C.N., Joppa, L.N., Gaveau, D.L.A., Laurance, W.F., 2014. Remaining
massive deforestation driving loss of endemic biodiversity in Indian Himalaya. natural vegetation in the global biodiversity hotspots. Biol. Conserv. 117, 12–24.
Biodivers. Conserv. 16, 153–163. South, D.B., 1999. How can we feign sustainability with an increasing population?
Payn, T., Kollert, W., Liu, S., Rodriguez, L., Neves Silva, L., Wingfield, M., Michel- New Forest. 17, 193–212.
Carnus, J., Orazio, C., Freer-Smith, P., 2015. Changes in planted forests and future Stewart, R.E., Desai, A., Walters, L.C., 2011. Wicked Environmental Problems:
global implications. For. Ecol. Manage. 352, 57–67. Managing Uncertainty and Conflict. Island Press.
Piketty, T., Goldhammer, A., 2014. Capital in the Twenty-First Century. Belknap Tateishi, R., Uriyangqai, B., Al-Bilbisi, H., Ghar, M.A., Tsend-Ayush, J., Kobayashi, T.,
Press. Kasimu, A., Hoan, N.T., Shalaby, A., Alsaaideh, B., Enkhzaya, T., Gegentana, Sato,
Planet Experts, 2014. Climate Scientist Respond to Dr. Nadine Unger. <http:// H.P., 2010. Production of global land cover data – GLCNMO. Int. J. Digital Earth 4,
www.planetexperts.com/climate-scientists-respond-dr-nadine-unger/> 22–49.
(accessed March 2015). Unger, N., 2014. To Save the Planet, Don’t Plant Trees, The New York Times. Op Ed,
Poore, D., Sayer, J., 1991. The Management of Tropical Moist Forest Lands: Ecological September 19. <http://www.nytimes.com/2014/09/20/opinion/to-save-the-
Guidelines. IUCN. planet-dont-plant-trees.html>.
Portillo-Quintero, C.A., Sánchez-Azofeifa, G.A., 2010. Extent and conservation of Unit Keria Presiden4, 2012. ONE MAP Indonesia. Paper presented at Presentation by
tropical dry forests in the Americas. Biol. Conserv. 143, 144–155. Nirarta Samadhi, Chair of Satgas REDD+ Moratorium Working Group, at Mid-
Potapov, P., Yaroshenko, A., Turubanova, S., Dubinin, M., Laestadius, L., Thies, C., Way Moratorium Workshop, 2012, Borodubur Hotel, Jakarta, 4–5 September,
Aksenov, D., Egorov, A., Yesipova, Y., Glushkov, I., Karpachevskiy, M., Kostikova, 2012.
A., Manisha, A., Tsybikova, E., Zhuravleva, I., 2008. Mapping the world’s intact United Nations, 2013. Millennium Development Goals Report 2013. The United
forest landscapes by remote sensing. Ecol. Soc. 13, 51. Nations. New York.
Putz, F.E., Pinard, M.A., 1993. Reduced-impact logging as a carbon-offset method. Van Lierop, P., Lindquist, E., 2015. Global forest area disturbance from fire, insect
Conserv. Biol. 7, 755–757. pests, diseases and severe weather events. For. Ecol. Manage. 352, 78–88.
Reich, P.B., 2011. Taking stock of forest carbon. Nat. Climate Change 1, 346–347. Venter, O., Fuller, R.A., Segan, D.B., Carwardine, J., Brooks, T., Butchart, S.H.M., Di
Ribeiro, M.C., Metzger, J.P., Martensen, A.C., Ponzoni, F.J., Hirota, M.M., 2009. The Marco, M., Iwamura, T., Joseph, L., O’Grady, D., Possingham, H.P., Rondinini, C.,
Brazilian Atlantic Forest: how much is left, and how is the remaining forest Smith, R.J., Venter, M., Watson, J.E.M., 2014. Targeting global protected area
distributed? Implications for conservation. Biol. Conserv. 142, 1141–1153. expansion for imperiled biodiversity. PLoS Biol. 12, e1001891.
Rodrigues, A.S., Andelman, S.J., Bakarr, M.I., Boltani, L., et al., 2004. Effectiveness of Watson, J.E., Dudley, N., Segan, D.B., Hockings, M., 2014. The performance and
the global protected area network in representing species diversity. Nature 428, potential of protected areas. Nature 515, 67–73.
640–643. Weng, L., Boedhihartono, A.K., Dirks, P.H., Dixon, J., Lubis, M.I., Sayer, J.A., 2013.
Romijn, E., Lantican, C.B., Herold, M., Lindquist, E., Ochieng, R., Wijaya, A., Mineral industries, growth corridors and agricultural development in Africa.
Murdiyarso, D., Verchot, L., 2015. Assessing changes in national forest Global Food Security 2, 195–202.
monitoring capacities of 99 tropical non-Annex I countries. For. Ecol. Manage. Whiteman, A., Wickramasinghe, A., Pina, L., 2015. Global trends in forest ownership,
352, 109–123. public income and expenditure on forestry and forest employment. For. Ecol.
Rosset, P.M., 1998. Alternative agriculture works: the case of Cuba. Mon. Rev. (July– Manage. 352, 99–108.
August), 21–28 Wilson, D., Kelston, A., Ahmen, S., 2010. Is this the ’BRICs decade’? BRICs Mon.
Rosset, P.M., Benjamin, M., 1994. The Greening of the Revolution: Cuba’s 10, 1–4.
Experiment with Organic Agriculture. Ocean Press, Hoboken, NJ. World Bank, 2013. Atlas of Global Development: A Visual Guide to the World’s
Rudel, T.K., 1998. Is there a forest transition? Deforestation, reforestation and Greatest Challenges, fourth ed. Collins, Glasgow.
development. Rural Sociology 63, 533–552. Wright, J.S., Muller-Landau, H.C., 2006. The future of tropical forest species.
Rudel, T.K., 2005. Tropical Forests: Regional Paths of Destruction and Regeneration Biotropica 38, 1–15.
in the Late Twentieth Century. Columbia University Press, New York. WWF, IIASA, 2012. Living Forests Report. WWF and the IIASA. Gland, Switzerland.
Rudel, T.K., 2007. Changing agents of deforestation: from state-initiated to Xu, J., Yang, Y., Fox, J., Yang, X., 2007. Forest transition, its causes and environmental
enterprise driven processes, 1970–1990. Land Use Policy 24, 35–41. consequences: an empirical evidence from Yunnan of Southwest China. Tropical
Rudel, T.K., 2013. The national determinants of deforestation in sub-Saharan Africa. Ecol. 48, 1–14.
Philos. Trans. R. Soc. Lond., B, 368. Zhang, P., Shao, G., Zhao, G., Le Master, D.C., Parker Jr., G.R., Dunning, J.B., Li, Q., 2000.
Rudel, T.K., Perz-Lugo, M., Zichal, H., 2000. When fields revert to forests: China’s forest policy for the 21st century. Science 288, 2135–2136.
development and spontaneous reforestation in post-war Puerto Rico. Prof. Zomer, R.J., Trabucco, A., Coe, R., Place, F., van Norordwijk, M., Xu, J.C., 2014. Trees on
Geographer 52, 386–397. Farms: An Update and Reanalysis of Agroforestry’s Global Extent and Socio-
Satgas REDD+, 2012. REDD+ National Strategy. Indonesian REDD+ Task Force Ecological Characteristics. ICRAF Working Papers, No. 179. World Agroforestry
(Satgas REDD+). Jakarta. Centre (ICRAF). Bogor, Indonesia.