Content uploaded by Roland Cochard
Author content
All content in this area was uploaded by Roland Cochard on Oct 31, 2017
Content may be subject to copyright.
REVIEW
Extent and causes of forest cover changes in Vietnam’s
provinces 1993–2013: a review and analysis of official data
Roland Cochard, Dung Tri Ngo, Patrick O. Waeber, and Christian A. Kull
Abstract: Within a region plagued by deforestation, Vietnam has experienced an exceptional turnaround from net forest loss to
forest regrowth. This so-called forest transition, starting in the 1990s, resulted from major changes to environmental and
economic policy. Investments in agricultural intensification, reforestation programs, and forestland privatization directly or
indirectly promoted natural forest regeneration and the setting-up of plantation forests mainly stocked with exotic species.
Forest cover changes, however, varied widely among regions due to specific socio-economic and environmental factors. We
studied forest cover changes (including natural and planted forests) and associated drivers in Vietnam’s provinces from 1993–
2013. An exhaustive literature review was combined with multivariate statistical analyses of official provincial data. Natural
forest regrowth was highest in northern mountain provinces, especially in the period 1993–2003, whereas deforestation con-
tinued in the Central Highlands and Southeast Region. Forest plantations increased most in mid-elevation provinces. Statistical
results largely confirmed case study-based literature, highlighting the importance of forestland allocation policies and agrofor-
estry extension for promoting small-scale tree plantations and allowing natural forest regeneration in previously degraded
areas. Results provide evidence for the abandonment of upland swidden agriculture during 1993–2003, and reveal that spatial
competition between expanding natural forests, fixed crop fields, and tree plantations increased during 2003–2013. While we
identified a literature gap regarding effects of forest management by para-statal forestry organizations, statistical results showed
that natural forests increased in areas managed for protection/regeneration. Cover of other natural forests under the organiza-
tions’ management, however, tended to decrease or stagnate, especially more recently when the organizations increasingly
turned to multi-purpose plantation forestry. Deforestation processes in the Central Highlands and Southeast Region were
mainly driven by cash crop expansion (coffee, rubber) and associated immigration and population growth. Recent data trends
indicated limits to further forest expansion, and logging within high-quality natural forests reportedly remained a widespread
problem. New schemes for payments for forest environmental services should be strengthened to consolidate the gains from the
forest transition, whilst improving forest quality (in terms of biodiversity and environmental services) and allowing local people
to actively participate in forest management.
Key words: forest transition, land allocation policies, reforestation programs, small-scale plantations, forestry organizations,
shifting cultivation.
Résumé : Au sein d’une région affectée par le déboisement, le Vietnam a connu un changement exceptionnel : de la perte
forestière nette a
`la régénération forestière. Cette soi-disant « transition forestière », débutant dans les années 1990, fut le résultat
d’importants changements de la politique environnementale et économique. Des investissements dans l’intensification agri-
cole, les programmes de reboisement et la privatisation des terres forestières ont directement ou indirectement promu la
régénération forestière naturelle et la mise en œuvre de forêts de plantation principalement boisées d’espèces exotiques. Les
changements de la couverture forestière ont cependant varié grandement selon les régions en raison de facteurs socio-
économiques et environnementaux spécifiques. Nous avons étudié les changements de la couverture forestière (y compris les
forêts « naturelles » et « plantées ») et les facteurs associés dans les provinces du Vietnam entre 1993–2013. Un examen exhaustif
de la littérature a été ajouté a
`des analyses statistiques a
`plusieurs variables des données provinciales officielles. La régénération
forestière naturelle a été la plus haute dans les provinces montagneuses du nord, surtout entre 1993–2003, tandis que le
déboisement a continué dans les hauts plateaux du centre et la région du sud-est. Les forêts de plantation ont augmenté le plus
dans les provinces de moyenne altitude. Les résultats statistiques ont fortement confirmé la littérature fondée sur les études de
cas, mettant en évidence l’importance des politiques d’attribution des terres forestières et d’extension agroforestière promou-
vant les plantations d’arbres a
`petite échelle et permettant la régénération naturelle en des zones antérieurement dégradées. Les
résultats fournissent la preuve de l’abandon de l’agriculture de montagne itinérante 1993–2003, et révèlent que la concurrence
spatiale entre l’expansion des forêts naturelles, des champs de culture fixe et la plantation d’arbres a augmenté 2003–2013.
Tandis que nous avons trouvé une lacune dans la littérature concernant les effets de la gestion forestière des organisations de
Received 26 May 2016. Accepted 22 September 2016.
R. Cochard. Institute of Integrative Biology, Swiss Federal Institute of Technology Zurich, Universitätsstrasse 16, 8092 Zurich, Switzerland.
D.T. Ngo. Institute of Resources and Environment, Hue University, Hue City, Vietnam; Forest Management and Development, Department of
Environmental Sciences, Swiss Federal Institute of Technology Zurich, Universitätsstrasse 16, 8092 Zurich, Switzerland.
P.O. Waeber. Forest Management and Development, Department of Environmental Sciences, Swiss Federal Institute of Technology Zurich,
Universitätsstrasse 16, 8092 Zurich, Switzerland.
C.A. Kull. Institute of Geography and Sustainability, University of Lausanne, 1015 Lausanne, Switzerland; Centre for Geography and Environmental
Science, Monash University, 3800 Melbourne, Vic, Australia.
Corresponding author: Roland Cochard (email: r_cochard@bluemail.ch).
Copyright remains with the author(s) or their institution(s). Permission for reuse (free in most cases) can be obtained from RightsLink.
199
Environ. Rev. 25: 199–217 (2017) dx.doi.org/10.1139/er-2016-0050 Published at www.nrcresearchpress.com/er on 29 September 2016.
Environ. Rev. Downloaded from www.nrcresearchpress.com by MONASH UNIVERSITY on 08/03/17
For personal use only.
sylviculture parapubliques, les résultats statistiques ont indiqué que les forêts naturelles ont augmenté dans les zones gérées
pour la protection/régénération. La couverture des autres forêts sous la gestion des organisations a eu, cependant, tendance a
`
diminuer ou a
`stagner, particulièrement plus récemment quand ces organisations se sont tournées de plus en plus a
`la
sylviculture de plantation polyvalente. Les processus de déboisement dans les hauts plateaux du centre et la région du sud-est ont
découlé principalement de l’expansion de la culture commerciale (café, caoutchouc) et de l’immigration et de la croissance
démographique associées. Les tendances des données récentes ont indiqué les limites de nouvelle expansion forestière, et
l’exploitation forestière au sein de forêts naturelles de haute qualité est censément restée un problème répandu. De nouveaux
plans pour « les paiements pour services environnementaux forestiers » devraient être renforcés afin de consolider les gains de
la « transition forestière », en améliorant la qualité des forêts (en matière de services en biodiversité et environnementaux), et en
permettant a
`la population locale de participer activement a
`la gestion forestière. [Traduit par la Rédaction]
Mots-clés : transition forestière, politiques d’attribution des terres, programmes de reboisement, plantations a
`petite échelle,
organisations forestières, cultures itinérantes.
1. Introduction
Vietnam is the only country in Southeast Asia that, in recent
decades, experienced a so-called forest transition from rapid net
deforestation to net reforestation (Meyfroidt and Lambin 2008a).
This transition was largely a consequence of fundamental policy
changes and structural reforms in the agricultural and forestry
sectors, beginning with broader nation-wide economic reforms
(Ð
ôi Mó’i) in 1986, and gaining momentum after the second land
law revision in 1993. The transition has sometimes been heralded
as the successful outcome of pro-forest policies and programs, but
improvements in terms of forest quality varied, particularly as
reforestation largely relied on mono-crop exotic plantations
(McElwee 2016). Furthermore, forest cover changes were far from
uniform across the country (Meyfroidt and Lambin 2008b).
The Vietnam Forestry Development Strategy 2006–2020 (VFDS
2006) set a target to further increase the national forest cover and
to strengthen the inherent forest environmental services (FES). To
reach such a target, the VFDS (2006) promoted payments for FES
(PFES) as a key mechanism for funding, with revenues projected to
bring in around US$2 billion by 2020. Evidence for the actual
contribution of policies and government-sponsored programs to
the forest transition, however, remains limited (McElwee 2012,
2016;Catacutan et al. 2012). Despite many case studies conducted
since the 1990s, the overall picture remains sketchy. There is a
need to bring together and synthesize the disjointed information.
The aim of this paper is to review forest cover dynamics and
their associated drivers in Vietnam’s provinces in the two decades
between 1993 (second land law revision) and 2013 (advent of PFES
program), including both natural forests as well as new plantation
forests. The study was based on an exhaustive literature review
twinned with statistical analyses of official provincial data. Multivar-
iate statistical modelling was used to explore the power of diverse
variables (pertaining to the provinces’ physical, demographic, eth-
nic, economic, and forestry-related political–organizational charac-
teristics) to explain overall changes in Vietnam’s natural/planted
forest cover. As will be shown, the statistical results broadly re-
flect the image provided by the literature. Certain biases in the-
matic and regional focus (and associated analytical limitations)
were however apparent, and the data indicated shifting trends
and limits to forest cover expansions.
2. Background
2.1. Historical perspectives: causes of forest decline in
Vietnam until the 1990s
In Vietnam, deforestation has occurred since colonial times
from logging and land conversion (McElwee 2016). Forest losses of
⬃15%–23% (mainly in Central and Southern Vietnam) occurred
from aerial herbicides and fires during the Indochina wars (1945–
1975) (de Koninck 1999;Brauer 2009). Deforestation rates were
highest, however, post-war. During the 1970s–1980s all forests
were nationalized and managed by more than 400 state forest
enterprises (SFEs) to extract timber and other resources (FSIV
2009;McElwee 2016). Furthermore, land conversion rates multi-
plied, including an expansion of slash-and-burn field cultivation
(swiddening) in ethnic-minority dominated upland forests. The
associated driving factors were diverse and included population
growth due to resettlement programs to mountain areas (the so-
called New Economic Development Zones) and natural increase
(Chi et al. 2013;Jamieson et al. 1998); forced collectivization of
agriculture in the lowlands and valleys, leading to decreasing crop
productivity and added incentives to cultivate individual parcels
in remote areas (Meyfroidt and Lambin 2008b;Castella et al.
2005a,2006;Tachibana et al. 2001); and state appropriation of
forestlands and disruption of customary use rights, leading to
poor regulation and unrestrained resource access, and in some
cases land conflicts (Tachibana et al. 2001;Nguyen et al. 2010;
McElwee 2004;To et al. 2015). As a consequence, national forest
cover declined from perhaps ⬃43% in 1943 to 16%–27% in 1993
(estimates vary; De Koninck 1999). In combination with non-
sustainable agricultural practices, this deforestation resulted in
extensive denuded land areas, especially in the uplands (up to an
estimated 40% national area; Vo and LeThac 1994), with detrimen-
tal knock-on effects such as nutrient losses, erosion, landslides,
siltation of waterways and reservoirs, and recurrent flooding
(Lippe et al. 2011;Clemens et al. 2010;Tran and Shaw 2007;Ziegler
et al. 2004;Wezel et al. 2002a,2002b;Sikor 1995;Cochard 2013). In
the late 1980s the depletion of forest and land resources, rising
costs of natural disasters, and the recognition of environmental
degradation joined with evolving political and economic ideas to
promote a variety of paradigm changes in forestry and land re-
sources management (Jamieson et al. 1998;Vo and LeThac 1994;
McElwee 2016).
2.2. Theorized forest transition pathways: in general and in
the Vietnamese context
A forest transition describes an incisive change from historical
forest loss to forest gain. Two general transition pathways have
been described (Rudel et al. 2005). The first, termed economic
development pathway, posits that technical and socio-economic
changes related to development—such as agricultural intensifica-
tion on the best lands, industrialization, and urbanization—lead
to abandonment of less productive agricultural lands, which then
revert to forest (Meyfroidt and Lambin 2008b). Here forest re-
growth is mainly a side effect of economic growth and modern-
ization.
The second, termed forest scarcity pathway, is dependent on
targeted policies in forestry and natural resource management.
Under this pathway, increasingly pressing deficits of forest goods
and services lead to the recognition that investments in forest
maintenance and restoration would be economically rewarding.
This can lead to both government and private initiatives to protect
existing forests, plant new forests, and manage forests sustain-
ably.
In Vietnam, both pathways played a certain role: economic
modernization was concurrent with major state-led forest initia-
200 Environ. Rev. Vol. 25, 2017
Published by NRC Research Press
Environ. Rev. Downloaded from www.nrcresearchpress.com by MONASH UNIVERSITY on 08/03/17
For personal use only.
tives (Meyfroidt and Lambin 2008a,2008b). The causes of forest
transition were, however, non-trivial and highly contextual
(Pham et al. 2015;Meyfroidt 2013;Lambin and Meyfroidt 2010;
Clement and Amezaga 2008), and as will be illustrated, several
regions have not gone through a forest transition. A large part of
the variation in local or regional outcomes may be explained from
the variation in predisposing environmental factors (Geist and
Lambin 2002; e.g., terrain, elevation, soils, etc.), which differ
markedly along various geographical gradients and distinct re-
gions (Figs. 1a,1b;Sterling et al. 2006;Olson et al. 2001;Epprecht
and Heinimann 2004). Another part may be attributable to locally/
regionally differing underlying forces (e.g., governance, policies,
demographic and economic context; Fig. 1c,Table 1) and associ-
ated proximate causes (e.g., effective state-sponsored reforesta-
tion, land conversion for farming; Ferretti-Gallon and Busch 2014;
Hosonuma et al. 2012). Within historical context and conventional
interpretations, land uses by ethnic minority groups have often
played a significant role (justified or unjustified) to explain forest
dynamics, especially in remote mountainous provinces (Castella
et al. 2006;Fig. 1d). Fifty ethnic minority groups were known to
practice shifting cultivation in forestlands. An estimated 7% of the
Ta
`y, 16% of Nùng, 45% of Thái, and almost 100% of the H’Mong, Dao
and other mountain ethnic minority groups practiced swiddening
in 1989—in total almost 3 million people with ⬃3.5 million ha
under shifting cultivation (Do 1994).
2.3. Transformations in land management, and forest
regrowth since the early 1990s
Several important changes were initiated within the course of
broader national economic policy reforms starting in 1986 (see
policy landmarks, supplementary data, Table SA1)
1
. In the agricul-
tural sector, de-collectivization, market liberalization, and tech-
nical innovations radically increased productivity (Kirk and
Nguyen 2009;Sikor and Vi 2005;Fatoux et al. 2002). Employment
in lowland agriculture and fast growing industries released pres-
sure on marginal lands, contributing to slowing deforestation
(Meyfroidt and Lambin 2008b). Land laws were revised in 1988,
1993, and 2003, which developed and strengthened private prop-
erty rights (Beresford 2008).
The forestry sector was also substantially restructured. Forest
management authorities and responsibilities were transferred
to provincial and district levels. State forest enterprises (SFEs)
were reformed, through a shift in scope and function from pure
extractive to more protection-oriented forestry, and (or) industrial
plantation forestry. Some SFEs were completely dissolved; the
remaining were restructured or were transformed into either
profit-oriented state-owned forest companies (SFCs) or forest pro-
tection management boards (FPMBs) mandated with the manage-
ment of designated protection forests or special use forests (McElwee
2012).
Rural households and communities could gain forest tenure
after the 1993 Land Law provided a legal basis for forestland allo-
cation (FLA). The nature of FLA rights, however, depended on
provincial contexts and was typically managed by SFEs or FPMBs
(McElwee 2012;Clement and Amezaga 2008;Dang et al. 2012;Ngo
and Webb 2008;Castella et al. 2006;Nguyen 2008). Initially, FLA
was provided on degraded forestland or barren land to promote
reforestation for land improvements and economic gain. Techni-
cal support was provided via extension through tree nurseries
(mostly using exotic species such as acacia), as well as payments
for labor, food, and equipment. At later stages natural forestland
was made available for FLA, whereby the focus was on protection,
maintenance, and upgrading of natural forests for specific
1
Supplementary data are available with the article through the journal Web site at http://nrcresearchpress.com/doi/suppl/10.1139/er-2016-0050.
Fig. 1. Maps of Vietnam. (a) The overall terrain. (b) The provinces located within eight regions, i.e., 1.) Northwest (4 provinces), 2.) Northeast (11),
3.) Red River Delta (10), 4.) North-Central Coast (6), 5.) South-Central Coast (6), 6.) Central Highlands (5), 7.) Southeast (8), and
8.) Mekong River Delta (13). Northwest and Northeast regions are conjointly referred to as the Northern Mountains Region. Source of maps:
Wikipedia.com. (c) The provinces’ population density (indicated by red colours) and poverty share (bubble size; i.e., percentage in the national
40% income bottom, ranging from 9% in TP HCM to 87% in Lai Châu Province). (d) The relative ethnic composition of the provinces (indicating
the majority Kinh and the five most populous minority groups). The bubble size represents the land area (km
2
) of each province. Data sources:
World Bank (2015) and UNFPA (2011).
Cochard et al. 201
Published by NRC Research Press
Environ. Rev. Downloaded from www.nrcresearchpress.com by MONASH UNIVERSITY on 08/03/17
For personal use only.
Table 1. Regional summaries of relevant statistics of forest extents and changes, socio-economic and agricultural indicators, and statistics of
forest land tenure and contracts.
Note: Included are triangle indices illustrating causal factors, which in the literature were used to explain forest changes in different regions. The relevant literature
references are listed below the table. For more details of literature review see the supplementary data, Table SA2
1
; for data sources see Tables SB1 and SB2
1
. The data
colouring indicates increases (light green) and decreases (red) during the time periods in between the indicated years.
Literature providing information and causal explanations of forest changes: All Vietnam:Meyfroidt and Lambin (2008a,2008b); Vu et al. (2014a,2014b); Northwest
Region: Clement and Amezaga (2008,2009);Clement et al. (2009);Folving and Christensen (2007);Lippe et al. (2011);Meyfroidt (2013);Nguyen et al. (2004);T.T. Nguyen
et al. (2009);Saint-Macary et al. (2010);Sikor (2001,2006);Sikor and Truong (2002);Sikor and Vi (2005);Tachibana et al. (2001);Chi et al. (2013);Wezel et al. (2002a);
Northeast Region: Castella et al. (2005a,2005b,2006); Castella and Dang (2002);Hoang et al. (2014);Jadin et al. (2013);Meyfroidt (2013);Nguyen et al. (2010);Pham et al.
(2015);Sandewall et al. (2010);Sikor (2012);Sikor and Baggio (2014);Tachibana et al. (2001);Trincsi et al. (2014);Turner and Pham (2015); North-Central Coast: Ankersen
et al. (2015);Bayrak et al. (2015);Disperati and Virdis (2015);Jakobsen et al. (2007);Leisz (2009);McElwee (2008,2009);Müller et al. (2014);Thiha et al. (2007);Tran et al.
(2010);Dao et al. (2009); South-Central Coast: Sikor (2012);Sikor and Baggio (2014);Thulstrup (2014,2015); Central Highlands: D’haeze et al. (2005);Dien et al. (2013);
Heinimann (2006);Leinenkugel et al. (2015);Meyfroidt et al. (2013);Müller and Zeller (2002);Sikor and Nguyen (2007);Sikor and Tran (2007);Thanh and Sikor (2006);
Southeast Region: Grogan et al. (2015); Mekong River Delta: Heinimann (2006);Leinenkugel et al. (2015);Tanaka (2001).
a
The mean pulp and paper capacity index combines information of wood processing and transport capacities in the region (see supplementary data, Table SB1c)
1
.
b
Areas contracted by MB-SUFs (management boards for special use forest; cover not shown) corresponded approximately to the areas of nature protected areas.
202 Environ. Rev. Vol. 25, 2017
Published by NRC Research Press
Environ. Rev. Downloaded from www.nrcresearchpress.com by MONASH UNIVERSITY on 08/03/17
For personal use only.
purposes (mainly watershed protection) (Ngo and Webb 2008;
Thiha et al. 2007). Reforestation activities were largely financed
through Program 327 (Greening the Barren Hills Program; 1993–
1998) and Program 661 (Five Million Hectares Reforestation Pro-
gram; 1998–2010). Program 661 mostly aimed at protecting and
upgrading forests in critical watersheds, including forest rehabil-
itation in upland regions (McElwee 2009;To 2007).
As a consequence, national forest cover (as shown by official
data) had by 2013 increased to similar levels as in 1943, i.e., 41% of
land cover (13.96 million ha). Of these forests, 25.5% (3.56 million ha),
however, consisted of plantations with mostly exotic species such as
acacias, eucalypts, pines, and rubber. Only 74.5% (10.40 million ha)
were classified as natural forest (MARD 2015), but it is debatable
how natural these forests really are, as they may range from im-
poverished bush- or thicket-like forests to fairly intact and species-
rich types of secondary forests (Meyfroidt and Lambin 2008a;
Nicolic et al. 2008;Ankersen et al. 2015;Van and Cochard 2016).
Many valuable forests continue to be degraded (Sikor and To 2011;
Meyfroidt 2013;Jadin et al. 2013;McElwee 2004,2016). According
to FAO (2016), the cover of primary forests decreased almost by a
factor of five from 384 000 ha (⬃4.4% of natural forest area) in 1990
to 85 000 ha (⬃0.8%) in 2005, but the cover has since stabilized.
2.4. Forestland allocation (FLA): socio-economic effects and
consequences on forests
FLA programs exerted differing effects on people’s livelihoods
and associated feedbacks on forest cover changes, depending on
the context. FLA and accompanying legal constraints, and state
control of forests via SFEs/SFCs, hindered shifting cultivation
practices and other forest uses by farmers (Clement and Amezaga
2008). Natural forest grew back on abandoned swiddens, but the
decline of land area for cropping in combination with the some-
what experimental introduction of new and more intensive farm-
ing technologies in sensitive environments incurred additional
risks of crop failures and food shortages to the poor (Meyfroidt
2013;Castella et al. 2006;Bayrak et al. 2015;Jakobsen et al. 2007).
This sometimes evoked resistance by traditional swiddener com-
munities and occasionally the suspension of FLA plans (Chi et al.
2013;Alther et al. 2002). In other cases, land registration failed
because new land rights conveyed less security than traditional
systems, which had more layers of social control (Sikor 2006).
As a consequence, top-down FLA programs were often reshaped
bottom-up, and adjusted to accommodate local realities. Tradi-
tional land tenure by relatively strong ethnic groups (e.g., Ta
`y,
Thái, and Mu’ò’ng) were partly recognized (Castella et al. 2006), or
some communities’ autonomous land appropriation (asserted via
exotic tree plantations during pre-FLA legal limbo) were condoned
(Sikor 2012). However, concessions to dominant local stakehold-
ers meant that poor households dependent on open-access natu-
ral resources (e.g., for swiddening), or immigrants not integrated
into local communities, often got the shorter end of FLA (Castella
et al. 2006;Chi et al. 2013;Thiha et al. 2007;Sikor and Nguyen
2007;Gomiero et al. 2000). Many rural poor also lost land resource
access as a consequence of state-sponsored reforestation with ex-
otic plantations (McElwee 2009), rapid expansion of cash-crop
plantations (e.g., coffee or rubber; Meyfroidt et al. 2013), or exclusion
from protected areas (McElwee 2008;Sowerwine 2004;Zingerli et al.
2002).
Households’ labor resources, social networks, and access to fi-
nances were also important to determine whether (or at what
rate) available lands could be planted with tree crops. Many house-
holds who obtained land rights lacked the means to plant trees,
especially given the 3–7 years delay until harvest (Sikor 2012;
Nguyen et al. 2010). In production forest areas, FLA policies, how-
ever, usually required the planting of trees. Consequently, many
poor farmers sold their land and became contracted laborers on
the same lands (Sikor and Baggio 2014;Bayrak et al. 2015). There
were also disparities between communities, some of which were
well-connected to the state apparatus, and quick to innovate and
mobilize resources to set up plantations. Other communities
(often ethnic minority groups) took longer to tap into these new
opportunities (Thulstrup 2015;Meyfroidt et al. 2013).
Resource disfranchisements, in addition to continuing rural
population growth, led to some renewed pressures on natural
forests. While many people found alternative incomes from
emerging industries either locally (e.g., at pulp and paper mills) or
in peri-urban areas, some relocated to weakly controlled upland
areas and resorted to clear new swidden fields, thus reducing
forest regeneration (Castella et al. 2006;Alther et al. 2002;Chi
et al. 2013). Increases in natural and planted forest cover, further-
more, conceal the fact that many natural forests were still being
degraded in quality via illicit selective logging (Sikor and To 2011;
McElwee 2004); in some cases, traditional protection controls of
sacred forests were weakened by FLA interventions (Bayrak et al. 2015).
2.5. Reforestation programs and the role of para-statal
forestry organizations
The achievement of the forest transition came at a major price.
Reforestation programs cost more than US$ 2 billion—funds which
were acquired from the state budget, aid loans, donor support,
and the private sector (McElwee 2009). Furthermore, while com-
mercial logging in Vietnam’s forests was largely banned during
the programs, domestic demand for timber remained high and
was only partially met by increases of plantation forestry. As a
result, much timber exploitation was exported to neighboring
countries, especially Cambodia and Laos (Meyfroidt and Lambin
2009;McElwee 2004,2016).
Within this context, the roles of SFEs, now converted into SFCs
and FPMBs, remained controversial. Endowed with state funds
and official sanction, these para-statal forestry organizations con-
tinued to compete with local communities for land resources,
lucrative afforestation programs, and access to commodity mar-
kets (To et al. 2015). They still drew prolific resources from state
budgets for large-scale forest protection and development proj-
ects (including Programs 327 and 661) but were often seen to be
relatively ineffective with regards to improving land uses and
economic performance (To et al. 2015).
In 2005, SFEs/SFCs controlled ⬃40% of Vietnam’s forestlands
(World Bank 2005). In addition, forestlands were variously con-
tracted out to farmers, whereby SFEs/SFCs managed and issued
specific land contracts, e.g., for purposes such as forest restoration
or protection. Many SFEs/SFCs were also charged with implement-
ing FLA. In several provinces, SFEs/SFCs were noted as notorious
for continuing exploitative and profiteering practices (McElwee
2004). In districts well-endowed with natural forest, SFEs/SFCs
reportedly delayed FLA to extract timber (Thiha et al. 2007;Ngo
and Webb 2008). In other cases, extensive forestlands were as-
signed as protection forests, presumably partly so that SFEs/SFCs
could benefit from state funds provided by reforestation pro-
grams targeted at watershed protection (Clement and Amezaga
2009). Conflicts have recurrently arisen about land contracted to
SFEs/SFCs. In some cases, where local farmers had encroached on
SFCs’ lands, a so-called joint-venture contract (liên doanh) be-
tween SFCs and occupants was negotiated (To et al. 2015). There is
a gap in the literature as to what degree variable efficiencies of
reforestation programs and forest management by SFEs/SFCs or
FPMBs may explain overall forest cover changes.
3. Comprehensive review and data analyses
3.1. Conceptual-analytical approach and compilation of
database
We first conducted a comprehensive literature review on the
extent and causes of forest cover changes in Vietnam (summary
Table SA2)
1
. Insights were used to compile a database with rele-
vant and sufficiently reliable variables available from official
Cochard et al. 203
Published by NRC Research Press
Environ. Rev. Downloaded from www.nrcresearchpress.com by MONASH UNIVERSITY on 08/03/17
For personal use only.
sources. We then examined correlative patterns in the data via
multivariate regression analyses, specifically focusing on how
variables representing agricultural, forestry, and land policy
changes, and predisposing environmental factors, influenced the
extents and changes of natural and planted forest cover in the
provinces in the period 1993–2013. Forest cover (areal extent) was
assessed for 1993 and 2013 in the case of natural forest, and for
1999 and 2013 in the case of planted forest (reliable data on planted
forest for 1993 were unavailable). Relative cover changes were as-
sessed for two periods (1993–2003 and 2003–2013) in the case of nat-
ural forest, and one period (1999–2013) in the case of planted forest (F
variables, Table 2).
Table 2. Summary of the dependent (F) variables and significant predictors (G/P/L/A/C/T) in multivariate models (Table 3). References for
data sources are listed at the table bottom.
Variable name Variable description Source
Geographical and terrain (G) variables
nodelta Whether [0] or not [1] the province lies in the Red River or Mekong Delta GEA
latitude Latitude: north UTM coordinates (m) at approximate centre of the province GEA
elevation An index of the estimated mean elevation (m a.s.l.) of the province GEA
distcoast The nearest distance from the province border to the coast (km) GEA
mtrugged An index of estimated mountainous rugged terrain cover (%) of the province GEA
provarea The provincial area cover (km
2
)WB
Forest cover extent and change (F) variables
natforest93 /03 /13 Natural forest cover (ha) in 1993; respectively 2003; respectively 2013 MAR
natfrch93–03 /03–13 Relative (%) natural forest cover change 1993–2003; respectively 2003–2013 MAR
plantfor99 /13 Planted forest cover (ha) in 1999; respectively 2013 MAR
pltfrch99–13 Relative (%) planted forest cover change 1999–2013 MAR
Population indicator (P) variables
popdens95 Population density (people km
−2
) in 1995 GSO
rurpopd95 /03 /10 Rural population density (people km
−2
) in 1995; respectively 2003; resp. 2010 GSO
rurpch95–03 Rural population change: population density in 2003 as a ratio of density in 1995 GSO
migrat05–14 Mean annual net migration (%) 2005–2014 GSO
Tay,Thai,Muong,H’Mong,ethnother Ta
`y, Thái, Mu’ò’ng, H’Mong, respectively other ethnic mountain minority
populations in provinces (% of total population) in 2009
UNP
Labor/economic development indicator (L) variables
workagri09 Labor force (% of all labor forces) working in agriculture in 2009 WB
hpoverty99 Households (% of population) below poverty line in 1999 WB
bpoverty09 Population (%) in national 40% income bottom in province in 2009 WB
Agricultural productivity (A) and wood processing (C) indicator variables
cerland95 /03 Land area (1000 ha) planted with cereal staple crops in 1995/resp. 2003 GSO
cerlch03–13 Cereal crops land area change: 2013 cover minus 2003 cover GSO
cerlrch95–03 /03–13 Cereal relative (ratio) land area change 1995–2003; respectively 2003–2013 GSO
cerprch95–13 Cereal field productivity change: 2013 output ha
−1
as a ratio of 1995 output GSO
riprch95–13 Rice field productivity change: 2013 output ha
−1
as a ratio of 1995 output GSO
maland03 /13 Land area (1000 ha) planted with maize in 2003; respectively 2013 GSO
mailch03–13 Maize crops land area change: 2013 cover minus 2003 cover GSO
maprch95–03 Maize field productivity change: 2003 output as a ratio of 1995 output GSO
pulpcap04 Index for provincial pulp-and-paper wood processing capacities in 2004a NSY/GEA
Forest land tenure and contract indicator (T) variables
hhften95 /04 Forest tenure (ha) by households and individuals in 1995; respectively 2004 NSY
hhftch95–04 Change in forest area (ha) under household tenure 1995–2004 NSY
hhcften04 Forest tenure (ha) by households, individuals or communities in 2004 NSY
cpcften95 /04 Forest tenure (ha) by communal people’s committees in 1995; resp. 2004 NSY
cpcftch95–04 Change in forest area (ha) under tenure by CPCs 1995–2004 NSY
ecorgften95 /04 Forest tenure (ha) by economic organizations in 1995; respectively 2004 NSY
ecoftch95–04 Change in forest area (ha) under tenure by economic organizations 1995–2004 NSY
forjovften04 Forest tenure (ha) by foreign organizations and joint ventures in 2004 NSY
ecofjvften04 ecorgften04 plus forjovften04 NSY
otherften95 /04 Forest tenure (ha) by other non-specified owners in 1995; respectively 2004 NSY
regenfcon99 Forest area (ha) contracted for zoning for regeneration in 1999 NSY
protfcon99 Forest area (ha) contracted for protection and management in 1999 NSY
prorefcon99 regenfcon99 plus protfcon99 NSY
allfcon04 Forest area (ha) contracted by any contractor (SFE, households, FPMB) 2004 NSY
sfefcon99 /04 Forest area (ha) contracted by SFEs/SFCs in 1999; respectively 2004 NSY
nhhcon95 /99 Number of households contracted by SFEs/SFCs in 1995; respectively 1999 NSY
hhfcon95 /99 /04 Forest area (ha) contracted to households by SFEs/SFCs in 1995; 1999; 2004 NSY
hhfcch95–99 /99–04 Change in area (ha) contracted to households by SFEs/SFCs 1995–1999; 1999–2004 NSY
mbpfcon99 /04 Forest area (ha) contracted by MB-PFs in 1999; respectively 2004 NSY
protect93 /02 Nature protected area (ha) in the province in 1993; respectively 2002 ICE
protch93–02 Change in area (ha) covered by nature protected areas 1993–2002 ICE
Note: List of data sources: GEA, compiled manually from Google Earth
TM
; GSO, GSO (2015); ICE, ICEM (2003); MAR, MARD (2015); NSY, B.N. Nguyen et al.
(2009); UNP, UNFPA (2011); WB, World Bank (2015).
a
Index calculated using original NSY data and UTM coordinates (see supplementary data, Table SB1c)
1
.
204 Environ. Rev. Vol. 25, 2017
Published by NRC Research Press
Environ. Rev. Downloaded from www.nrcresearchpress.com by MONASH UNIVERSITY on 08/03/17
For personal use only.
We tested 135 indicator variables (independent predictors),
which potentially influenced forest cover (dependent Fvariables)
within the 63 provinces. The indicator variables reflected infor-
mation on geography and terrain (Gvariables), population and
ethnicity (P), labor and poverty (L), infrastructural development
(I), education (E), cereal staple crops agricultural land area and
productivity (A), forest resource exploitation (C), and forest land
under tenure or contracts (T). The data were mostly collected from
official sources, but some data were compiled using Google
Earth
TM
. A summary of the 64 variables that turned out to be
significant predictors in statistical models is provided in Table 2,
including listing of data sources. In the supplementary data, a
complete listing of all 144 tested variables is provided (Tables SB1
and SB2)
1
, including a PCA loading plot of important variables
(Fig. SB2)
1
, as well as a discussion of data quality.
3.2. Statistical analyses and interpretation
MS Excel and Minitab 17 statistical software (Minitab Inc., State
College, PA) were used for calculations and multivariate statistical
analyses. Before analyses, all data variables were checked for nor-
mality; if necessary, data were transformed as appropriate (natu-
ral logarithm, square root, Box-Cox or Johnson transformation, or
in some cases normal score function). Multivariate linear regres-
sion (MLR) was used to assess variable interrelationships. To work
out optimal subsets of predictors in MLR models, the best subsets
regression (BSR) tool in Minitab was used. Severe outlier or lever-
age points were identified using residual, Cox’s distance and DFIT
plots, and if considered necessary observations were deleted to
improve MLR models. Non-significant predictors (p-value > 0.05)
were dropped from models in a step-wise mode (if not previously
excluded by BSR selection). Principal components analysis was
another aid used to better interpret data correlations and to sep-
arate relevant from irrelevant factors. While the data of most
variables evidently followed spatial patterns according to Viet-
nam’s geography (Fig. 1), the residuals data of final regression
models did not show statistically significant spatial autocorrela-
tion (determined via testing the correlation of residuals with
between-province spatial distance data). MLR model equations
and relevant statistics are provided in Table 3.
4. Forest change in Vietnam and its causes
4.1. Natural and planted forest extent and change
The nature of the forest transition in Vietnam, even whether it can
be confidently asserted to have occurred or not, can only be partly
resolved from official forestry and land use data due to quality issues
and data gaps (e.g., upland swiddening and land tenure by para-statal
forestry organizations). However, as detailed in our discussion of
data quality (see Part 2, Supplementary data B
1
; see also B.N. Nguyen
et al. 2009;Hieu et al. 2011;McElwee 2004,2016), we deem that the
forest cover (F) data can be trusted to represent passably reliable
figures, especially with regard to their relative inter-provincial vari-
ability. This trust is based on the observations that the Fdata gener-
ally correspond with large scale and case-study based remote sensing
analyses, and that the results of our statistical analyses agree with
trends identified in the literature review.
According to official data, total natural forest cover in Vietnam
increased from 26.1% of land area in 1993 to 31.4% in 2006; net
cover subsequently remained more or less constant (Fig. SB3a)
1
.At
the regional level there were, however, divergent trends (Fig. 2,
Table 1). In the Northern Mountains Region, natural forests cov-
ered only 16.6% in 1993, but in 2013 had more than doubled to
38.0%. Lesser increases were observed in the Central Coast Region.
In contrast, in the Central Highlands and Southeastern Region,
natural forests declined from 49.7% to 35.7%. Remnant forests
(<2% cover) increased marginally in the Red River and Mekong
Deltas 1993–2003, but decreased again 2003–2013 (Fig. 2,Table 1).
Total planted forest cover in Vietnam increased relatively
steadily from an estimated 4.4% of land area in 1999 to 10.6% in
2013. Highest increases were recorded in mid-elevation (midland)
provinces of the Northeast and Central Coast Regions (14%–20%
cover in 2013; Fig. 3,Table 1). Plantation cover in the lowlands
remained minor (<5%); it increased slightly in the Red River Delta
but decreased in the Mekong Delta. Cover more than doubled in
the remote, mountainous Northwest and Central Highlands Re-
gions, but remained below 8% of land area in 2013 (Fig. 3,Table 1).
Forest change patterns, as shown by official data (Figs. 2–3), are
mirrored in the literature. Meyfroidt and Lambin (2008a,2008b)
documented forest cover changes in Vietnam during 1993–2002.
They verified that information on forest cover on official maps
was consistent with the most reliable data from remotely sensed
sources; equally, reported forest changes were largely congruent
with changes reflected in official district data. These patterns
were, furthermore, corroborated by large-scale analyses of NDVI
(normalized difference vegetation index) time series during 1982–
2006 by Vu et al. (2014a,2014b) as well as by a review of 54 publi-
cations on local or regional/provincial case studies, which reported
on forest changes (listed below Table 1; details in Table SA2)
1
.It
should be noted, however, that there were geographical imbalances
in case studies, as 71% were conducted in mostly mountainous prov-
inces in the northern half of Vietnam. Most of these northern studies
(28) reported increases in natural forest cover (at annual rates of
maximally 16%, La
`o Cai Province), but some reported data (6) or qual-
itative observations (5) that indicated decreases. In contrast, only
three studies in southern provinces documented natural forest re-
growth, whereas nine studies reported net deforestation. Changes
of swidden agricultural systems were a subject of 35 studies, and
38 studies described the establishment or expansion of various types
of tree plantations, ranging from intercropping stands of fruit trees
(often at higher altitudes) to small-scale acacia woodlots (mostly at
mid-altitudes), and large rubber plantations (in the south). No studies
reported decreases of plantation forests. Sections 4.2 through 4.9
present our analyses of causes of forest cover distribution and
changes, based on regression models M1-M11 (Table 3), and supported
by the available literature.
4.2. Forest cover in relation to spatial and topographical
factors
The distribution of natural forest largely reflects historic pat-
terns of agricultural expansion. In the flat deltas of the Mekong
River and Red River, native forests have almost entirely disap-
peared. While a few lowland forest remnants persist along rivers
(Vu 2006) or in protected locations on hills, islands, or in home
gardens (Trinh et al. 2001;Kiernan 2010;Van and Cochard 2016),
relict forests were further decimated during 1993–2013, probably
due to pressures emanating from urban growth and development
(Saksena et al. 2014;Castrence et al. 2014). With few exceptions (e.g.,
protected areas of Ha
`Tı˜nh and Kiên Giang provinces; McElwee 2010;
Tanaka 2001) extensive forested areas are now only found at
higher elevations, especially in large upland provinces, where also
the largest changes (in absolute and relative terms) in natural
forest cover have taken place in recent times (+elevation**
(
*
)
,M1–3,
M4a,M5d,M6a-c;+provarea***,M1–4,M5d; see variables in Tables 2
and 3for details)
2
. As already outlined, forests in the northern
mountainous provinces were severely degraded up to the early
1990s, and particularly so on steep and rugged terrain
2
The listed predictor variables (see Table 2 for description) that support the statements in text (elevation and provarea in this case), including the
corresponding regression model numbers (M1-4 and M5d in this case), refer to the results presented in Table 3; the coefficient signs (+/–) and p-value
significance levels (from strongly*** to medium** and marginally significant*; see Table 3) of the predictors in the models are also indicated.
Cochard et al. 205
Published by NRC Research Press
Environ. Rev. Downloaded from www.nrcresearchpress.com by MONASH UNIVERSITY on 08/03/17
For personal use only.
(–mtrugged**
(
*
)
,M1,M3a-b;Meyfroidt and Lambin 2008b). In con-
trast, by 2013 forests had increased mainly in the northern moun-
tains (Fig. 2) and mostly in areas of relatively unproductive and
difficult-to-access, steep terrain (hence, positive association in
M4b,+mtrugged***; not significant in M2 and M4a).
Between 1993 and 2003, major increases in natural forest cover
occurred along latitudinal and poverty/developmental gradients
(+latitude***,+hpoverty99**,M5a). The latitudinal effect (which also
was significant in more elaborate models; +latitude**
(
*
)
,M5a-d,
M6b;+latitude*,M2,M4a) may be partly due to the more mountain-
ous nature of northern provinces, but it may also be explained by
proximity to Vietnam’s political center Ha
`Nខ
ôi. Since the 1990s,
many development and reforestation programs were concen-
trated on the relatively poor and environmentally degraded hin-
terland of Ha
`Nខ
ôi, and specific policies and laws may have been
enforced more stringently than in other regions. In contrast, the
Central Highlands (which were still a major repository of rela-
tively untouched natural forests in 1993; Fig. 2) continued to sup-
ply timber and serve as a frontier for agricultural expansion
(Meyfroidt et al. 2013; see Section 4.8).
The productivity and range of acacia and rubber plantations in
Vietnam is largely determined by soil water and fertility, and
temperature (Hung et al. 2016;Nguyen and Dang 2016). Statistical
results show that plantation forests were mostly set up in the
Table 3. Listing of multivariate regression model equations (M1–M11) for forest cover variables (natural forest in 1993, natforest93, and 2013,
natforest13; correspondingly planted forest in 1999, plantfor99, and 2013, plantfor13), respectively the relative forest cover changes during time
periods 1993–2003 (natfrch93–03), 2003–2013 (natfrch03–13), and 1999–2013 (pltfrch99–13).
Equations for natural forest cover extent and changes
M1 [natforest93]
0.18
= −18.05 + 9.05 × [elevation]
0.13
– 0.937 × nscor[distcoast] – 1.69 × nscor[mtrugged] + 2.346 × [provarea]
0.16
(adj. R
2
: 88.0%; 62 observations)
M2 JT[natforest13] = −6.55 + 0.00000016 × [latitude] + 1.054 × [elevation]
0.13
– 0.202 × nscor[distcoast] + 1.153 × [provarea]
0.16
(adj. R
2
: 88.7%; 62 observations)
M3a [natforest93]
0.18
= −11.66 + 5.18 × [elevation]
0.13
– 0.81 × nscor[distcoast] – 1.412 × nscor[mtrugged]+3.79×[provarea]
0.16
+ 0.622 × JT[Tay]
+ 0.462 × (−[Muong]
(−0.33)
) – 0.638 × JT[H’Mong] – 0.9 × ln[cerland95] (adj. R
2
: 93.6%; 60 observations)
M3b [natforest93]
0.18
= −15.61 + 6.86 × [elevation]
0.13
– 2.304 × nscor[mtrugged]+4.03×[provarea]
0.16
– 0.575 × JT[H’Mong] – 1.15 × ln[cerland95]
+ 0.61 × nscor[protect93] (adj. R
2
: 92.4%; 59 observations)
M4a JT[natforest13] = −4.49 + 0.00000018 × [latitude] + 0.742 × [elevation]
0.13
– 0.203 × nscor[distcoast]+0.85×[provarea]
0.16
– 0.17 × JT[rurpopd10] + 0.12 × (−[Muong]
(−0.33)
) (adj. R
2
: 91.3%; 62 observations)
M4b JT[natforest13] = −3.124 + 0.494 × nscor[mtrugged] + 0.954 × [provarea]
0.16
– 0.0068 × [bpoverty09] + 0.084 × (−[Muong]
(−0.33)
)
+ 0.139 × nscor[protect02] (adj. R
2
: 90.5%; 60 observations)
M5a JT[natfrch93–03] = −1.98 + 0.00000098 × [latitude] – 0.269 × JT[hpoverty99] (adj. R
2
: 62.3%; 55 observations)
M5b JT[natfrch93–03] = −0.337 + 0.00000148 × [latitude] – 0.406 × ln[popdens95] – 0.37 × JT[Tay] – 0.224 × nscor[protect93]
– 0.087 × ln([protch93–02] + 1000) + 0.166 × [hhften95]
0.15
– 0.232 × nscor[nhhcon95] (adj. R
2
: 70.5%; 57 observations)
M5c JT[natfrch93–03] = −3.46 + 0.00000135 × [latitude] – 0.28 × JT[rurpopd95] – 0.296 × JT[Tay] – 0.11 × (−[Thai]
(−0.4)
) + 0.126 × [hhcften04]
0.17
– 0.325 × nscor[nhhcon95] (adj. R
2
: 69.5%; 57 observations)
M5d JT[natfrch93–03] = −3.315 + 0.00000055 × [latitude]+1.98×[elevation]
0.13
– 0.284 × nscor[distcoast]+1.1×[provarea]
0.16
– 0.682
×[natforest93]
0.18
– 0.177 × JT[rurpch95–03] + 0.185 × JT[H’Mong] + 0.13 × JT[cerlrch95–03] – 0.074 × JT[maprch95–03] – 0.434 × [cpcften95]
0.08
+ 0.18 × [hhften95]
0.15
– 0.031 × ln[otherften95] + 0.015 × [sfefcon99]
0.26
(adj. R
2
: 97.7%; non-delta provinces, 34 observations)
M6a JT[natfrch03–13] = −2.44 + 1.39 × [elevation]
0.13
+ 0.298 × JT[natfrch93–03] – 0.32 × JT[pltfrch99–13] (adj. R
2
: 39.6%; 60 observations)
M6b JT[natfrch03–13] = −8.215 + 0.000000413 × [latitude]+3.05×[elevation]
0.13
– 0.122 × [natforest03]
0.2
– 0.32 × JT[rurpopd03]
+ 0.34 × ln[cerland03] – 0.38 × JT[cerlrch03–13] – 0.64 × JT[Tay] – 0.26 × (−[Thai]
(−0.4)
) + 0.34 × JT[H’Mong] – 0.28 × JT[ethnother]
+ 0.258 × [ecofjvften04]
0.16
+ 0.119 × [prorefcon99]
0.2
– 0.239 × [sfefcon04]
0.16
+ 0.133 × nscor[nhhcon95] + 0.26 × nscor[hhfcch95–99]
+ 0.137 × nscor[hhfcch99–04] (adj. R
2
: 91.4%; 57 observations)
M6c JT[natfrch03–13] = −18.2 + 1.69 × [elevation]
0.13
+ 0.037 × [bpoverty09] – 0.67 × JT[cerlrch03–13] – 2.15 × ln[cerprch95–13]+1.77×[maland03]
0.22
– 0.51 × JT[Tay] – 0.275 × (−[Muong]
(−0.33)
)–25×[hhftch95–04]
0.1
– 0.07 × ln[cpcftch95–04]+34.8×[ecoftch95–04]
0.0
+0.92×[regenfcon99]
0.08
– 0.324 × nscor[mbpfcon04] (adj. R
2
: 78.9%; only non-delta provinces, 40 observations)
Equations for planted forest cover extent and changes
M7 [plantfor99]
0.33
= −40.0 + 11.02 × [nodelta] + 0.0000071 × [latitude] – 2.73 × nscor[distcoast] – 4.88 × nscor[mtrugged] + 11.66 × [provarea]
0.16
(adj. R
2
: 53.8%; 62 observations)
M8 [plantfor13]
0.3
= −50.7 + 10.35 × [nodelta] + 0.00000646 × [latitude] – 2.22 × nscor[distcoast] – 4.11 × nscor[mtrugged] + 14.56 × [provarea]
0.16
(adj. R
2
: 69.6%; 62 observations)
M9 [plantfor99]
0.33
= −8.16 + 14.89 × [nodelta]+5.72×[provarea]
0.16
+ 7.51 × JT[rurpopd95] + 4.743 × nscor[nhhcon99] + 3.13 × nscor[hhfcon95]
– 2.35 × nscor[mbpfcon99] (adj. R
2
: 75.7%; 59 observations)
M10a [plantfor13]
0.3
= −4.8 + 5.23 × [nodelta] – 3.57 × nscor[mtrugged] + 0.013 × [workagri09]
1.41
+5.27×[riprch95–13] – 7.74 × [maland13]
0.14
+1.91×[hhften04]
0.17
+ 0.205 × [cpcften04]
0.17
+0.57×[otherften04]
0.19
+1.87×[sfefcon04]
0.16
+ 1.4 × nscor[nhhcon99]
(adj. R
2
: 93.2%; 58 observations)
M10b [plantfor13]
0.3
= −2.78 + 6.36 × [nodelta] – 3.01 × JT[migrat05–14] + 3.3 × ln[pulpcap04] + 1.354 × [hhften04]
0.17
+2.24×[sfefcon04]
0.16
(adj. R
2
: 86.8%; 61 observations)
M11 JT[pltfrch99–13] = −2.945 + 0.84 × [nodelta] – 0.467 × nscor[mtrugged] + 0.663 × [provarea]
0.16
– 0.051 × [plantfor99]
0.33
+ 0.0031 × [workagri09]
1.41
+ 0.17 × JT[ethnother] – 0.17 × JT[cerlch03–13]–1.04×[mailch03–13]
0.14
+ 1.16 × ln[cerprch95–13]+0.12×[hhften04]
0.17
+ 0.036 × ln[cpcftch95–04]
– 0.075 × [ecorgften95]
0.14
+0.05×[allcon04]
0.24
+ 0.23 × nscor[hhfcon99] (adj. R
2
: 88.4%; 59 observations)
Note: Statistical significance of the predictors is indicated by underlining as strong (p< 0.0005), medium (0.0005 < p< 0.005), and rather marginally significant
(0.005 < p< 0.05; no underlining). The adjusted R
2
and number of observations used for models are indicated in brackets below the equation. The function nscor
indicates a normal score transformation, JT a Johnson transformation (see supplementary data, Tables SB1 and SB2)
1
.
206 Environ. Rev. Vol. 25, 2017
Published by NRC Research Press
Environ. Rev. Downloaded from www.nrcresearchpress.com by MONASH UNIVERSITY on 08/03/17
For personal use only.
larger mid-elevation provinces less characterized by rugged ter-
rain (dummy variable +nodelta***,M7–11;–mtrugged**
(
*
)
,M7–8,
M10a,M11;+provarea***,M7–8;+provarea*
(
*
)
,M9,M11) but which
were characterized by a relatively high population density (in
1999) and a high share (in 2013) of the population working in
agriculture (+rurpopd95***,M9;+workagri09**
(
*
)
,M10a,M11). Like nat-
ural forests, planted forest cover increased northward (+latitude***,
M7–8), but plantations were more concentrated near the coast
(–distcoast*,M7–8) and near centers of paper and (or) pulp produc-
tion (+pulpcap04**,M10b). Land accessibility and spatial proximity
to wood buyers (paper/pulp facilities, Ha
`Nខˆoi, the Chinese border,
coastal hubs) presumably stimulated the setting-up of planta-
tions (Sikor 2012;Sandewall et al. 2010;Clement et al. 2009). The
predictors pulpcap04, latitude, and distcoast were, however, insignif-
icant in other models, which suggests that over the period 1999–
2013 (and at larger provincial data scales), factors relating to land
tenure and agriculture were more dominant drivers of plantation
establishment (M9,M10a,M11; see Section 4.9).
4.3. Upland forests, ethnic minorities, shifting cultivation,
and protected areas
In addition to topographical/spatial patterns, the distribution
of natural forests in 1993 (but less so in 2013, see Sections 4.4. and
4.6.) appeared to reflect significant agricultural influences. Cover
was negatively associated with land area cultivated permanently
with cereal crops at the time (data available for 1995, –cerland95**
(
*
)
,
M3a-b); the association presumably reflects patterns mainly in the
better developed provinces where most field plots were officially
registered. In contrast, no reliable data were available on the ex-
tent of shifting cultivation fields in upland provinces. A negative
association between forest cover and provincial population share
of the H’Mong ethnic group (–H’Mong**
(
*
)
,M3a-b), however, lends
support to the notion (see Section 2.1) that intensive shifting cul-
tivation practices in the northern uplands had contributed to
deforestation prior to 1993. The H’Mong and other minorities
were noted for subsisting on traditional swidden farming in re-
mote upland mountainous areas at least until the 1990s (Turner
2012;Castella and Dang 2002;Leisz 2009;Folving and Christensen
2007). Caution is, however, advised with “ethnic” interpretations.
Especially during the 1980s, the populations of upland swiddeners
had partly swelled (in addition to natural population growth) and
(or) been displaced by new swiddeners who had immigrated from
the lowlands, and who often ignored local land management
rules (Castella et al. 2006). This often went hand in hand with
uncontrolled logging by SFEs and other powerful networks and
players (McElwee 2004,2016; see Section 2.1). In addition, tradi-
tional swiddening groups such as the H’Mong mainly occupy the
remote northern uplands, which in the 1980s were particularly
poor and underdeveloped and were generally less effectively con-
trolled and administered by the state (Epprecht et al. 2011). Hence,
degradation may be mainly explained by the distance to Ha
`Nខ
ôi
and other administrative centers (possibly reflected by variable
–distcoast***,M3a,M1). Ethnic minorities were often blamed for
deforestation (Castella et al. 2006), even though in earlier times
traditional swiddening was probably practiced in sustainable
ways, with long fallow cycles allowing for forest regeneration (Tran
et al. 2006;Fox et al. 2000,2001;Wangpakapattanawong et al. 2010).
The Ta
`y, Nùng, Thái, and Mu’ò’ng traditionally occupied and
cultivated fertile, alluvial basements and terraces within moun-
tain valleys. These groups were generally less dependent on shift-
ing cultivation (Castella and Dang 2002;Clement and Amezaga
2008). To ensure steady supplies of irrigation water, forests impor-
tant for protecting watersheds were often maintained and man-
aged in relatively sustainable ways. Furthermore, because of their
sedentary lifestyles, communities were often better placed to up-
hold local rules of forest uses within changeable socio-political
contexts (Nguyen et al. 2004;Sikor 2006;Meyfroidt 2013;Folving
and Christensen 2007). This can explain the positive association of
the variables Tay and Muong with natural forest cover in 1993
(+Tay**,+Muong** M3a; lesser so in 2013, +Muong*,M4a-b), as these
Fig. 2. Relative natural forest cover and forest cover changes within Vietnams’ provinces (area of bubbles representing officially reported
absolute values in hectares). From right to left: forest cover in 1993 in the 63 provinces (black bubbles); changes in forest cover during 1993–
2003; and during 2003–2013 (green colour: net forest increase; red colour: net deforestation); and forest cover in 2013. Data sources: MARD
(2015) and B.N. Nguyen et al. (2009).
Cochard et al. 207
Published by NRC Research Press
Environ. Rev. Downloaded from www.nrcresearchpress.com by MONASH UNIVERSITY on 08/03/17
For personal use only.
minorities’ land occupancy presumably provided a certain buffer
against excessive pressures on forests.
Correspondingly, areas protected by the state partly explained
persistent forest cover in the provinces in 1993 (+protect93**,M3b)
and 2013 (+protect02*,M4b). The first protected areas (PAs) were
often set up to maintain the remaining forests in the hinterland of
urban centers and in already largely deforested and intensively
cultivated provinces (ICEM 2003). Notably, this often lead to re-
source conflicts with local traditional land users, whereby timber
resource extraction sometimes significantly increased before a
decently adequate PA management regime could be set in place
(McElwee 2008;Zingerli et al. 2002;Sowerwine 2004;Hoang et al.
2014). In some cases, extensive logging was even conducted by
SFEs in forestlands that were in the process of becoming part of
aPA(McElwee 2004;2016). Despite these impacts, the negative asso-
ciations of PAs with forest cover changes during 1993–2003 (–protect93*,
–protch93–02*,M5b) probably primarily derived from the spatial
limitations to forest increases in provinces with high PA cover,
rather than actual large-scale deforestation within PAs (see spatial
self-limitation effects, Sections 4.4. and 4.6; see forest degradation
effects, Section 4.7).
4.4. The economic development pathway of forest
transition and its limits
Agricultural productivity was stimulated through commodity
market formation and expansion, and increased via technical ad-
vances such as the introduction of new varieties of rice and maize,
subsidies for fertilizers and pesticides, infrastructure improve-
ments, the establishment of new permanent fields on alluvial
lands or fixed terraces, or the improvement of upland fields by
using sloping agricultural land technologies (Sikor 2001;Fatoux
et al. 2002;Sikor and Vi 2005;Pingali et al. 1997;Wezel et al. 2002b;
Folving and Christensen 2007;Jourdain et al. 2014;Leisz 2009).
Agricultural intensification was described as a significant cause
for the abandonment of marginal lands (followed by forest regen-
eration, often natural, sometimes aided) by seventeen studies
(and was implied in a further six studies) conducted in the north-
ern mountain provinces, and by two studies in the south (cause
agro-intensification;Table 1). Since clear causal links were difficult
to establish, substantiation was commonly qualitative in nature
(studies providing data-based indications were Meyfroidt and
Lambin 2008b;Vu et al. 2014b;Müller and Zeller 2002). Nonethe-
less, much literature supported the economic development path-
way of forest transition, especially in remote, underdeveloped
rural mountain areas, where population pressures were high in
relation to low land use potentials. Opposite patterns were re-
ported by seven studies conducted in highly productive regions
(Mekong and Red River deltas, Southeast Region) or frontier prov-
inces (Central Highlands). In these regions, increasing land rents,
spurred by population growth (in an often already densely popu-
lated area) and economic development, exerted increased pres-
sures on remaining forests.
Regional data partly reflected patterns characteristic of forest
transition phases. Despite the fact that the rural population in-
creased during 1995–2010 in most mountain regions (up to 45% in
the Central Highlands; Meyfroidt and Lambin 2008b;Meyfroidt
et al. 2013), the initially high average fertility rates (during 1970s
⬃6, in 1990 ⬃4 children per woman; Allman et al. 1991,GSO 2015)
markedly declined to only 1.8–2.6 children per woman in differ-
ent regions (annual average for 2005–2015, Table 1). While popu-
lations in the Central Highlands and Southeast Region (which
Fig. 3. Relative planted forest cover and forest cover changes within Vietnams’ provinces (area of bubbles representing officially reported
absolute values in hectares). From right to left: forest cover in 1999 in the 63 provinces (blue bubbles); changes in forest cover during 1999–
2013 (green colour: net forest increase; red colour: net deforestation); and forest cover in 2013. Data sources: MARD (2015) and B.N. Nguyen
et al. (2009).
208 Environ. Rev. Vol. 25, 2017
Published by NRC Research Press
Environ. Rev. Downloaded from www.nrcresearchpress.com by MONASH UNIVERSITY on 08/03/17
For personal use only.
includes Hò
ˆChí Minh City metropolitan area) partly grew due to
net immigration spurred by agricultural and industrial develop-
ment (Table 1;Meyfroidt et al. 2013;D’haeze et al. 2005;Grogan
et al. 2015), the other regions experienced net emigration during
2005–2013. Along with population growth, land area under fixed-
field rice and maize crops (the latter especially in mountain re-
gions) expanded and productivity increased. Advances in cereal
yields per hectare were highest in mountainous regions (increases
of 53%–92% compared to 37%–41% in the alluvial lowlands; Table 1;
Müller and Zeller 2002;Vu et al. 2014a;Castella and Dang 2002;
Keil et al. 2008).
Statistical results appear to provide some evidence of the eco-
nomic development pathway in mountain provinces for 1993–
2003, and to a lesser extent for 2003–2013. During 1993–2003,
population densities and further population increases, appar-
ently still limited natural forest regrowth and (or) contributed to
continued deforestation (–popdens95***,M5b;–rurpopd95*,M5c;
–rurpch95–03*,M5d;Vu et al. 2014a,2014b), but model M5d suggests
that the expansion of fixed agricultural land under cereal staple
crops (mainly rice and maize, but also other cereals, see Ho 2014)
exerted positive effects on forest changes (+cerlrch95–03**), espe-
cially in higher-lying mountain provinces (+elevation***). This may
be a consequence of the direct economic pull of land labor to
more productive sites and (or) programs for agricultural land
reclamation and improvement connected to FLA programs (see
Section 4.5). Since official statistics did not capture information
on swiddens, the negative association of forest changes with in-
creases in maize productivity (–mapch95–03*,M5d) possibly reflects
the circumstance that upland swidden fields were largely main-
tained (not replaced by forest) in provinces where productivity
increases of maize were particularly high (cf. Meyfroidt and
Lambin 2008b;Meyfroidt et al. 2013;Wezel et al. 2002a,2002b;Keil
et al. 2008).
Natural forest cover changes during 2003–2013 (natfrch03–13)
were only weakly explained (adjusted R
2
of 11.2%) by the changes
in the preceding decade (+natfrch93–03**,M6a), indicating that
changes proceeded along largely divergent trajectories. Whereas
labor was retracted from swiddens to more productive fields dur-
ing 1993–2003, this trend apparently weakened during 2003–2013.
Forests still tended to increase in the more mountainous and
comparatively sparsely populated and poorer provinces (+elevation***,
–rurpopd03**,M6b;+elevation**,+poverty09***,M6c). On the other
hand, the expansion of lands under cereal crops no longer exerted
positive effects; in fact, the association with forest cover changes
turned significantly negative (–cerlrch03–13***,M6b-c), and forest
losses (or diminished increases) were especially observed in prov-
inces where productivity of cereal fields was much improved since
1995 (–cerprc95–13**;M6c). This indicates that further expansion of
fixed cereal lands increasingly competed with space that could be
occupied by forests (cf. Jakobsen et al. 2007). Natural forest in-
creases, however, tended to be higher in provinces that were char-
acterized by large areas of cereal (especially maize) crop fields in
2003 (+cerland03***,M6b;+maland03*,M6c). In those provinces the
hunger for additional cereal lands presumably had already been
largely satisfied, and production increases were possible on al-
ready established fields.
From our data there was little overt evidence of direct compe-
tition for space between natural and planted forests (but see
Sections 4.5–4.9). The change ratios of natural forests were, how-
ever, negatively associated with the initial natural forest cover
(–natforest93***,M5d;–natforest03***,M6b) and correspondingly for
planted forests (–plantfor99***,M11). Apart from this spatial self-
limitation effect, there were strong indications that planted for-
ests increasingly competed with expanding agricultural cereal
fields (–cerlrch03–13**,M11), especially where adaptable new variet-
ies of maize (often planted for commercial animal feed; Sikor
2001) allowed the extension of permanent fields on lesser produc-
tive terrain (–mailch03–13***,M11;–maland13***,M10a;Meyfroidt and
Lambin 2008b;Wezel et al. 2002a,2002b;Castella et al. 2005b;
Ankersen et al. 2015,Keil et al. 2008). In contrast, more land was
available for the extension of tree plantations in provinces, where
relatively high improvements in cereal field productivity were
achieved (+cerprc95–13**,M11), especially in intensive rice produc-
tivity (+ripch95–13**,M10a). This lends additional support for the
economic development pathway.
In the fertile rice production regions of the Red River and Me-
kong Deltas, and other alluvial lowland areas, not much space was
left for increases of forest plantations on remaining peripheral
lands (dummy variable +nodelta***,M7–11). Woodlots were often
displaced by increasing urban sprawl and industrial develop-
ments (Saksena et al. 2014;Phuc et al. 2014). Only in former rice
production areas in lesser productive, ephemerally inundated re-
gions in the Mekong River Delta (An Giang and Kiên Giang Prov-
inces; Figs. 2–3) some increases in Melaleuca swamp forests were
possible. These forests partly regenerated naturally, or they were
set up as economic plantations (Tanaka 2001).
4.5. Land allocation, management changes, and overall
effects on natural forests
A major explanation for the abandonment of marginal lands
was the hindrance of shifting cultivation practices in conjunction
with FLA policies. This was mentioned as a cause for natural forest
regrowth by eleven studies (and implied by a further six studies)
conducted in the northern mountain provinces and by one study
in the Central Highlands (cause FLA/swidden stop,Table 1). One
study conducted in the Central Highlands (Sikor and Nguyen
2007) reported how FLA led to an increase in deforestation due to
deteriorating traditional land management regimes and ensuing
resource conflicts. Within the context of FLA policies and refores-
tation programs, laws for forest protection were often set in place
and enforced to various degrees via mechanisms, such as logisti-
cal support and payments for forest patrolling. These measures
were noted by thirteen studies (and implied by a further four
studies) as relevant factors for forest regrowth, whereas three
studies noted negative influences on forests mostly due to weak-
ening traditional forms of management (cause policies/laws/control,
Table 1). Only one known study (Meyfroidt 2013) explicitly inves-
tigated whether perceived qualitative decreases in FES and (or)
newly internalized environmental value systems (e.g., through
government awareness programs) influenced communities to
change their forest management. At least seven other studies,
however, mentioned communities’ changed environmental
perceptions and (or) values, noting potential positive changes to-
wards effective pro-forest management. Any of these policy-
driven causes of forest regrowth may represent the forest scarcity
pathway, but it is not always clear whether changes such as swid-
den abandonment would not have occurred also in the absence of
policy interventions.
Between 1993 and 2003, natural forest cover increased in prov-
inces where large areas of land were allocated for household ten-
ure (i.e., Red Book semi-permanent land use certificates; Barney
2005;B.N. Nguyen et al. 2009), either at the beginning (in 1995;
+hhften95**,M5b,M5d) and (or) during the considered period (FLA
until 2004, including community tenure; +hhcften04**,M5c). Household
forest tenure increased from 1995 (average 3.2%) to 2004 (10.0%),
predominantly in the Northern Mountains Region (19.3% in 2004)
and Central Coast Region (13.5%), but remained very low in the
Central Highlands and Southeast Region (0.8%; Table 1). FLA to
entire communities (often ethnic minority villages) became pos-
sible in 2004 (Tran et al. 2010;Table 1).
In contrast to the period 1993–2003, increases in natural forests
during 2003–2013 could not be related to household tenure (M6b-c).
Indeed, forest changes during 2003–2013 were negatively asso-
ciated with forestland area allocated to households during 1995–
2004 (–hhftch95–04**,M6c). This may be explained by several
developments. FLA focusing on highly degraded forests in moun-
Cochard et al. 209
Published by NRC Research Press
Environ. Rev. Downloaded from www.nrcresearchpress.com by MONASH UNIVERSITY on 08/03/17
For personal use only.
tain areas generally contributed to forest regrowth during 1993–
2003, but FLA programs subsequently focused on relatively intact
natural forests for management and sustainable livelihood im-
provements (see Section 2.3). Legally required protective mea-
sures (e.g., patrolling) were sometimes only maintained during
initial program phases, and regulations were relaxed, occasion-
ally leading to cycles of excessive resource extraction, forest deg-
radation, and piecewise clearance (e.g., Thiha et al. 2007;Dien
et al. 2013). In some cases, FLA led households to directly clear
parts or all of the endowed forest plots for crops or acacia planta-
tions, to secure land entitlement and tangible benefits (McElwee
2009;Thanh and Sikor 2006;Ngo and Webb 2008).
4.6. Changes in land management and natural forests
within provincial contexts
During 1993–2003, and especially 2003–2013, natural forest in-
creases were comparatively smaller (controlled for other factors)
in provinces with large population shares of Ta
`y(–Tay**,M5b-c;
–Tay***,M6b-c), Thái (–Thai*,M5c;–Thai***,M6b) and Mu’ò’ng
(–Muong**,M6c) ethnic minorities. This may be partly explained
by traditional systems of forest tenure and reflect higher initial
provincial forest cover (allowing for lesser subsequent forest in-
crease; see Sections 4.3. and 4.4). Ta
`y, Thái and Muòng communi-
ties were partly successful in maintaining traditional forms of
land management and sometimes obtained corresponding offi-
cial legal tenure via FLA (Castella et al. 2006;Sikor 2006;Clement
and Amezaga 2008). During 1993–2013, paddy/maize fields ex-
panded, and these as well as many other (–ethnother**,M6b; mainly
Central Coast/Highlands Regions, Fig. 1d) mountain ethnic minor-
ities became increasingly engaged in small-scale acacia and rub-
ber forestry, limiting natural forest increase on allocated lands
(Wezel et al. 2002a;Castella et al. 2005a;Clement and Amezaga
2008;Sikor and Vi 2005;Thiha et al. 2007;Thulstrup 2015;Dien
et al. 2013; see Section 4.9). Reduced forest regrowth may partly
also have resulted from the exclusion of immigrant households
from FLA, whereby some of the disfranchised turned again to
swidden agriculture on non-allocated lands (see Section 2.3).
In upland areas, FLA policies impeded the swidden practices of
ethnicities such as the H’Mong and Dao, resulting in natural for-
est regrowth (+H’Mong*,M5d;+H’Mong***,M6b). This effect was
possibly weaker in distant borderlands where FLA policies were
less stringent and some communities maintained swidden prac-
tices (–distcoast***,M5d;–
distcoast**,M2,M4a;Folving and
Christensen 2007;Alther et al. 2002;Jadin et al. 2013;Leisz 2009).
In contrast to 1993 (see Section 4.3), the ethnic composition of
provinces was unimportant to explain forest cover in 2013 (minor
exception +Muong*,M4a-b), but forest cover was still marginally
lower in provinces characterized by comparatively higher rural
population densities (–rurpopd13*,M4a) and higher levels of pov-
erty (–bpoverty09*, excluding variable distcoast,M4b).
4.7. Influences of forest management institutions on
natural forest cover
In addition to forests devolved to households and communities,
extensive forestland areas remained under tenure and manage-
ment by para-statal or (to far lesser degrees) semi-private or pri-
vate forestry organizations. We found no study that specifically
investigated the differential influences of these organizations’
land tenure and management on forest changes, but some quali-
tative inferences can be drawn from the literature (see Section 2.5;
McElwee 2004,2016;To et al. 2015;Clement and Amezaga 2009).
Policy directions since the late 1980s shifted the scope of state
forest enterprises (SFEs) and successor organizations (economi-
cally oriented state forest companies, SFCs, and forest protection
management boards, FPMBs) towards more protection-oriented
forestry, and large investments in reforestation were made via
Programs 327 and 661 (see Section 2.3). None of the reviewed
studies specifically focused on reforestation programs, which
were directly managed by SFEs, SFCs, or FPMBs, but thirteen stud-
ies noted that larger-scale reforestation of watersheds or other
target zones had been undertaken or supervised by respective
para-statal organizations in the study area. On the other hand,
unsustainable exploitation of forest resources reportedly contin-
ued in other forestlands managed by para-statal organizations
with the consent or toleration of responsible authorities. Logging
for high-value timber continued to be a serious issue, even within
protected areas (Sikor and To 2011;McElwee 2004,2008,2016;
Jadin et al. 2013). Twenty-three studies (of which eight actually
documented net increases in forest cover) noted decreases in for-
est quality and integrity due to largely uncontrolled resource ex-
traction (cause forest resource exploitation;Table 1). Furthermore,
conflicts over forest resources between stakeholders (mainly
between different communities or between communities and
para-statal organizations) with negative outcomes for forest man-
agement were reported by six studies (and implied by eight), four
of which were conducted in the Central Highlands.
Large forested areas were managed by SFEs/SFCs/FPMBs, but no
official data has been published on their respective land tenure,
much of which presumably goes under the lumped-up category
“other forest tenure” in Table 1 (B.N. Nguyen et al. 2009). In con-
trast, specific data were available on forests contracted by SFEs/
SFCs (no distinction made between SFE and SFC) or FPMBs (distinction
made between management boards for protection forests, MB-
PFs, and for special use forests, MB-SUFs). As can be seen in Table 1,
“other forest tenure” (average 12.3% in 2004) as well as “forest-
lands contracted” by either SFEs/SFCs (13.2%) or MB-PFs (6.8%) in-
creased from 1995 to 2004 (by ⬃22%–119%), and areas under
tenure/contracts were highest in mountainous provinces, espe-
cially in the Central Highlands and Coast Regions. Data were,
furthermore, available on forestland contracts issued to house-
holds by SFEs/SFCs (Table 1). In most regions forestlands “con-
tracted to households” first decreased (average of 4.5% in 1995;
2.5% in 1999), but then rose again (4.3% in 2004), probably in
conjunction with the dynamics of specific programs for reforesta-
tion and forest management (e.g., Programs 327 and 661; see Sec-
tion 2.3). Relatively fewer households were contracted in the
Central Highlands, resulting in generally larger areas managed
per household (23.6 ha compared to <11.5 ha in other regions;
Table 1). Forests contracted for the purpose of protection (data
from 1999; contractors not specified but presumably mostly SFEs/
MB-PFs) were found at similar levels in all mountain provinces
(average 6.3% land), whereas forests contracted for regeneration
were mostly concentrated in the Northern Mountains Region
(3.7% compared to <1.1% in other regions; Table 1).
In addition to these data on para-statal organizations’ forest
tenure and management (and including sub-contracting to house-
holds), data were available on land tenure by communal people’s
committees (CPCs), economic organizations (EOs) and foreign or-
ganizations or joint ventures (FOs/JVs). CPCs are temporary own-
ers of forestlands that are in the process of being transferred from
other owners (mainly SFEs/SFCs) to households or communities.
While CPCs have tenure rights during the transitional period,
CPC-managed forests (which were seldom of high timber quality
in the beginning) have often become informal open access zones
due to weakly enforced controls (Nguyen 2011). At the beginning
of FLA in 1995, CPCs still held substantial forestlands (7.1% total
land area), especially in the Central Coast Region (17.0%), whereas
CPC tenure decreased to only 1.0% area in 2004 (Table 1). EOs were
mainly operators of hydro-electric dams and providers of freshwa-
ter for large cities, whereas some FOs/JVs were engaged in nature
conservation and ecotourism projects (McElwee and Nguyen
2014). Hence, EOs, FOs or JVs often had an interest in effective
watershed management (steady water supplies, prevention of silt-
ation) and forest protection. Land tenure by EOs varied by region
but decreased nationally from 15.1% in 1995 to 11.9% in 2004
(Table 1). Tenure by FOs/JVs was only 1.5% in 2004.
210 Environ. Rev. Vol. 25, 2017
Published by NRC Research Press
Environ. Rev. Downloaded from www.nrcresearchpress.com by MONASH UNIVERSITY on 08/03/17
For personal use only.
The influences of different institutions’ forest management var-
ied widely, with some gradual changes over time (see also Section
4.9). Forest cover changes during 1993–2003 were marginally neg-
atively associated with “other” forest tenure in 1995 (–otherften95*),
but positively with forestlands contracted by SFEs/SFCs in 1999
(+sfefcon99*,M5d). This suggests that during the 1990s, natural for-
ests did not fare well under the tenure of para-statal organiza-
tions, except if specific management directives were issued (and
correspondingly, sufficient state funding provided) under forest
protection or reforestation contracts. In particular, forest cover
changes were negatively associated with CPC tenure in 1995
(–cpcften95***,M5d) and with the number of households contracted
by SFEs in 1995 (–nhhcon95*
(
**
)
,M5b-c). Presumably, exploitation of
scarce timber resources was initially still a dominant activity of
SFEs and contractors. Reforestation of bare forestlands (such as
under Program 327, and using contracted labor) was largely un-
dertaken by planting exotic species, rather than allowing the re-
generation and improvement of natural forest. Furthermore, in
many cases where degraded natural forestlands were involved,
time-limited legal rights endowed by contracts, and (or) situations
of quasi-open access (such as under transitional CPC tenure), prob-
ably encouraged the planting of exotic trees, which conveyed
clearer signs of land resource entitlements as compared to native
trees. Hence, some natural forests may have been replaced by
acacias also on lands designated as protection forest (see Section
4.9.; McElwee 2009,2016;Thiha et al. 2007;To et al. 2015;Sikor and
Baggio 2014;Thulstrup 2014).
During 2003–2013, natural forests increased in provinces where
large areas were under contract for purposes of forest regenera-
tion or protection in 1999 (+prorefcon99***,M6b;+regenfcon99**,
M6c). Some of these areas were probably contracted and directly
managed by SFEs/SFCs or FPMBs. The positive association of forest
cover changes with new forestland contracts to households dur-
ing the periods 1995–1999 (i.e., start phase of Program 661;
+hhfcch95–99***) and 1999–2004 (expansion of Program 661; +hhfcch99–
04**,M6b) suggest, however, that the cooperation of households
(many of which were probably already engaged in forestry tasks in
1995; +nhhcon95*,M6b) played an important role in achieving
specific reforestation targets (McElwee 2016). Forest regrowth
(natural, and possibly aided) was apparently also achieved in wa-
tersheds that until 2004 fell under the management EO’s and
FOs/JVs (+ecofjvften04***,M6b;+ecoftc95–04**,M6c). In contrast, nat-
ural forests designated for purposes other than forest regenera-
tion or protection apparently mostly declined in cover under the
management by para-statal organizations: cover changes were
negatively associated with forestlands under contracts by SFEs/
SFCs or MB-PFs in 2004 (–sfefcon04***,M6b;–mbpfcon04***,M6c), and
also with forestlands which came under CPC tenure between
1995–2004 (–cpcftrch95–04**,M6c). This points at an increasing em-
phasis by SFEs/MB-PFs away from management for protection/
restoration of native-species forests towards more lucrative
plantation forestry with exotics (see Section 4.9.; –pltfrch99–13*,
M6a).
4.8. Central Highlands and Southeast Region: Forests,
societal shifts, and cash crops
Our statistical results probably do not adequately reflect the
driving forces behind continuing deforestation in the Central
Highlands and Southeast Region (Fig. 2). In Ð ´
a˘k L´
a˘ k Province,
locally implemented FLA exerted mixed effects. One case study
(Müller and Zeller 2002) reported improving forest conditions
mainly due to agricultural intensification and strengthened con-
trols on forest resources. Another study (Thanh and Sikor 2006;
Sikor and Nguyen 2007), however, described how FLA led to in-
creased pressures on forest resources because of conflicting inter-
ests by new owners and neighboring villagers claiming traditional
use rights. As described by Meyfroidt et al. (2013), the situation in
the Central Highlands is complex, partly owing to historical lega-
cies. The region, originally populated by ethnic minorities (pri-
marily Ê Ðê and M’Nông people), was the focus of resettlement
programs from the late 1970s, often in conjunction with extensive
forest exploitation by SFEs (McElwee 2016). Liberalization in the
1990s spurred unplanned immigration (Table 1) and further mar-
ginalization of many original inhabitants. During the 1990s,
deforestation was mainly driven by a coffee cash-crop boom,
whereby plantations encroached on lands previously cultivated
with staple crops (see also D’haeze et al. 2005;Dien et al. 2013).
Many minority people were displaced and took to clearing new
forestlands (see also Doutriaux et al. 2008). This detrimental cycle
(plantation expansion – social displacement – forest clearing –
plantation expansion) in many areas has continued until recently.
With profits from rubber now exceeding those of coffee the de-
forestation focus has, however, shifted from the Highlands to
lower-lying provinces in the adjacent Southeast Region (Meyfroidt
et al. 2013;Fig. 2). The rubber boom is partly driven by expansion
of small-holder plantations, but state-sponsored and private in-
dustrial plantations play an increasingly important role (Grogan
et al. 2015;To and Tran 2014).
4.9. The plantation boom as evidence for the forest scarcity
pathway
Restrictions on shifting cultivation or contracts issued for forest
protection/regeneration are motivated by the estimation of policy
makers and other stakeholders that forests maintain vital func-
tions and provide important FES. Nonetheless, despite apparent
causal links, it is possible that some forests would have regrown in
degraded areas even in the absence of policy measures. Data on
planted forests, in contrast, provide unequivocal evidence for the
forest scarcity pathway, since actively planted woodlots result
from concrete demands for forest products and services, which
are economically driven and facilitated by land use policies.
FLA and forestland contract schemes significantly facilitated
the planting of fruit tree gardens (especially in remote mountain
areas) or the setting up of small-scale exotic tree plantations
(mainly in the midlands). The factors that spurred the setting-up
of private plantations was the focus of several studies conducted
in the midlands (Sikor and Baggio 2014;Sikor 2012;Thulstrup
2014,2015;Nguyen et al. 2010;Sandewall et al. 2010;Gomiero et al.
2000), but small-holder plantations were also mentioned in other
studies as a factor in planted forest cover increase (overall
26 studies, and implied in another four studies; cause FLA/tree
planting;Table 1). A few studies (6) also reported the setting-up of
large-scale industrial tree plantations (mainly acacia or rubber)
owned by private or state-owned companies.
Land tenure, and especially land contracts (by households, SFEs/
SFCs, CPCs), fostered the fast establishment of plantations. In 1999
planted forest cover was associated strongly positively with the
number of households having forestland contracts issued by SFEs
in 1999 (+nhhcon99***) and with the forestland area under house-
hold contracts in 1995 (+hhfcon95**,M9). The pattern may be ex-
plained by the delay from the issuance of contracts to the setting
up and expansion of plantations. Contracts issued during 1995–
1999 had only partly materialized into plantation cover in 1999
compared to contracts issued earlier until 1995. Contracts issued
until 1999, however, also partly explained plantation increases
during 1999–2013 (+hhfcon99**,M11;+nhhcon99*,M10a).
The establishment of tree plantations by households depends
on many factors, such as security in land tenure, total agricultural/
forest assets and income, household labor and family networks,
availability of loans, agricultural extension support (e.g., local tree
nurseries), and local market prices for wood products (influenced
by factors such as regional market prices, transport access and
vicinity of wood processing industries) (Nguyen et al. 2010;Sikor
and Baggio 2014;Sandewall et al. 2010). As farmers tend to be
budget-constrained and risk-averse, there can be a delay of several
years from FLA to the setting-up of plantations. Furthermore,
Cochard et al. 211
Published by NRC Research Press
Environ. Rev. Downloaded from www.nrcresearchpress.com by MONASH UNIVERSITY on 08/03/17
For personal use only.
re-allocation of land resources from poorer to more powerful
households frequently occurred, partly explaining a time lag be-
fore plantation establishment (Sikor and Baggio 2014;Thulstrup
2014,2015; Section 2.4.).
It is notable that variables of household forestland tenure
(hhften95,hhften99) were not significant predictors of plantation
cover in 1999 (M9). This probably partly reflects the circumstance
that many households that received forestlands via FLA, mainly in
the Northern Mountains Region, were poor compared to other
regions (Fig. 1,Table 1); also agricultural extension programs ini-
tially mainly focused on improving the output of staple crops
such as rice and maize (Jakobsen et al. 2007;Folving and
Christensen 2007;Sikor and Vi 2005;Nguyen et al. 2004;Keil et al.
2008). Furthermore, the remoteness of these regions increased
transport costs of wood products to processing facilities, limiting
profit margins (Nguyen et al. 2010;Alther et al. 2002;Gomiero
et al. 2000). Household tenure (year 2004) was, however, highly
important to explain plantation increases during 1999–2013 and
the cover in 2013 (+hhften04***,M10a-b,M11). Hence, the plantation
boom eventually reached the northern mountains. To some de-
gree the boom also reached various relatively marginalized com-
munities of ethnic minorities in the Central Highlands and Coast
Regions (+ethnother*,M11;Thulstrup 2014,2015;Thiha et al. 2007;
Meyfroidt et al. 2013;Bayrak et al. 2015). Provinces with a high
plantation cover in 2013 were generally characterized by net pop-
ulation emigration (–migrat05–14***, M10b;Table 1), which may
have been partly due to a market-driven redistribution of labor
and land resources. Farmers who successfully gained profits from
plantation forestry were able to acquire additional lands and in-
crease plantations, whereas other farmers became contract labor-
ers or emigrants to industrial centers (Sikor and Baggio 2014).
The plantation boom is also reflected in a change of focus by
para-statal forestry organizations. In 1999, planted forest cover
was marginally negatively correlated with forestlands contracted
by MB-PFs (–mbpfcon99*,M9). This indicates that plantation expan-
sion was possibly limited in provinces where extensive natural
forestlands were assigned for protection under management by
MB-PFs. Similarly, the expansion of plantations during 1999–2013
apparently met limitations by forestlands managed by EOs since
1995 for watershed protection or other purposes (–ecorgften95*,
M11). In contrast, plantation increases after 1999 were positively
related to forestlands contracted to either SFEs/SFCs, FPMBs, or
households (+allfcon04**,M11), and plantations grew in provinces
where until 2004 comparatively large proportions of forestlands
remained under CPC tenure (+cpcftrch95–04**,M11). This suggests
that para-statal organizations (including sub-contracted laborers
and other temporary forestland custodians) increasingly opted for
plantations with fast-growing species to meet reforestation tar-
gets and (or) increasing demands for economic self-sufficiency
(Amat et al. 2010;De Jong et al. 2006;Thiha et al. 2007;Dao et al.
2009;McElwee 2009). Especially in the case of CPC tenure, planta-
tions may also have been set up by local communities (i.e., forest
landholders in spe) to expedite the transfer process and (or) pre-
empt FLA through factual entitlements via plantations (see Sec-
tion 2.4.; Sikor 2012;To et al. 2015). In 2013 the cover of planted
forest was strongly positively associated with forestland area con-
tracted by SFEs/SFCs in 2004 (+sfefcon04***,M10a-b), and to lesser
degrees with forestland area under tenure by CPCs and other
landholders in 2004 (+cpcften04*,+otherften04**,M10a).
Not all planted forests were, however, set up merely for wood
production. In some regions, exotic trees (acacias, eucalypts,
pines) as well as native trees (more than 28 species; Nguyen 2007)
were used to establish forests on sites identified for regeneration.
These were sometimes planted with the purpose to facilitate the
re-colonization of native tree species, and hence ultimately re-
establish quasi-natural forests, mostly managed for watershed
protection and (or) the rehabilitation of soils (De Jong et al. 2006;
Amat et al. 2010;Tran et al. 2005;McNamara et al. 2006;Millet
et al. 2013). The boundaries between planted and natural forests
are thus often somewhat fluid not only in official data, but also in
reality (see Part 2, Supplementary data B)
1
.
5. Final remarks
To explain the particularities of recent forest transitions in
Asian countries, several recent national-level comparative studies
have stressed the importance of private entrepreneurship in for-
est management (facilitated by arrangements for effective forest-
land ownership), sufficient provisioning of timber products to
cover local demands (facilitated by cross-border timber trade and
the setting-up of plantations), shifts in the national economy to-
wards export-oriented manufacturing and service industries (fa-
cilitated by foreign direct investments), and strong pro-forest
state interventions and policies (facilitated by international aid
and expertise, and capacities generated from economic growth)
(Youn et al. 2016;Li et al. 2015;Liu et al. 2016). However, broad
inter-country comparisons focusing on the end state, ignore the
complex, long and dynamically shifting pathways that may effec-
tively lead to a so-called forest transition within any particular
country. Such pathways are strongly intertwined with a country’s
specific history, culture, and geography, and the development
trajectories and sequences may be multifarious within smaller-
scale regional contexts. The outcomes of transition processes are
neither deterministic nor irreversible (Lambin and Meyfroidt
2010;Singh et al. 2015). While country comparisons tend to focus
on differences in current forest governance, each country’s
broader realities and conditions are also singular.
Vietnam’s story is unique within Southeast Asia, and so is the
case of its recent forest cover changes (McElwee 2016). Vietnam’s
forest transition occurred at a time of an increasingly pressing
environmental crisis (Jamieson et al. 1998); in this regard it bears
resemblance to the case of several western countries in the 19th
century (e.g., Switzerland, France, or others; Mather 1992;Mather
et al. 1999;Mather and Fairbairn 2000;Walker 1993). Widely per-
ceived forest resource scarcity (and associated effects) towards the
end of the 1980s played an important role in triggering significant
adjustments in Vietnam’s forest policies and management. Until
now, this is a major difference with other countries in Southeast
Asia (e.g., Indonesia, Malaysia, Laos), where forest resources until
recently were still perceived to be available in relatively ample
quantities and (or) where the economy is characterized by lesser
dependencies on national land resources in upland regions (e.g.,
Thailand). Even within Vietnam, differences between the North-
ern Mountains Region (with fast net forest regrowth) and the
Central Highlands Region (with continuing net deforestation) sug-
gest that the regional abundances of forest resources fundamen-
tally influence patterns of resource uses and management. In this
regard the Kingdom of Bhutan represents a rare exception; it
has been noted as the only country in Asia where stringent
conservation-inspired state forest policies led to an increase in
forest cover under conditions of still-abundant forests (more than
60% cover; Bruggeman et al. 2016).
The Vietnamese forest transition also bears resemblances to
earlier examples of transitions in Europe and North America (as
well as more recent transitions in neighboring provinces of China;
He et al. 2014;Xu et al. 2007;Mather 2007), as it coincided with the
onset of strong national economic growth and modernization;
this facilitated the efforts of the state and landholders to achieve
a turn-around in forest management. Vietnam was characterized
by levels of poverty similar to Ethiopia in the early 1980=s, but the
country has since achieved middle income status and currently
features the second-fastest economic growth rate worldwide
(after China) (Economist 2016). Compared to historical Euro-
American forest transitions, however, significant differences exist
in terms of the country’s integration within globalized economic,
political and ideological networks, including having access to
212 Environ. Rev. Vol. 25, 2017
Published by NRC Research Press
Environ. Rev. Downloaded from www.nrcresearchpress.com by MONASH UNIVERSITY on 08/03/17
For personal use only.
ideas, tested experiences, technologies and sources of informa-
tion that are now much advanced (Kull et al. 2007;Lambin and
Meyfroidt 2010;Li et al. 2015). For example, various agricultural
innovations such as improved genetic varieties of key crops
(including rice, maize, and acacia trees; Tran and Kajisa 2006,Keil
et al. 2008;Griffin et al. 2015), combined with high demands for
agricultural and forestry products on international markets (e.g.,
staple crops, rubber, wood chips; To and Tran 2014;Barney 2005),
contributed to rapid land use intensification, which (as demon-
strated in this and other studies) facilitated reforestation of mar-
ginal lands, especially in upland regions. Hence, as was earlier
conjectured by Mather (2007), and as has recently been stressed by
FAO (2016), the so-called Borlaug hypothesis (which posits that
agricultural intensification provided room for reforestation) is
partly confirmed for the case of Vietnam. Our study, however,
showed that agricultural modernization and land use transforma-
tions proceeded in contextually highly disparate ways, with cer-
tain regions actually experiencing continuing net deforestation as
a result of expanding cash crop plantations (especially coffee and
rubber).
More recently, emerging trends in forest changes raise ques-
tions as to what degree the targets set by VFDS (2006) for the year
2020 can be reached. A further increase of the forest cover to 48%
of national land area is presumably only possible in a scenario of
furthered expansion of mainly exotic-species planted forests and
the upholding of tight restrictions on swidden farming. It remains
unclear how a forest transition strongly shaped by exotic planta-
tions can achieve targets to improve forest biodiversity and FES
(including aspects pertaining to improvements of food security of
the rural poor). Compared to natural forests, exotic plantations
hold little value for biodiversity conservation (Wang et al. 2011;
McShea et al. 2009); they also differ in terms of ecosystem func-
tions and services (Chazdon 2008;Sidle et al. 2006;Jackson et al.
2005), and single-species plantations may be vulnerable to dis-
eases and pests (Nambiar et al. 2015). Information on the composi-
tion and status of Vietnam’s natural forests remains fragmentary;
hence it will be difficult to adequately estimate FES over large
areas—despite contrary assertions made by new PFES schemes
(McElwee et al. 2014,2016;Ankersen et al. 2015). This situation is
particularly disconcerting given additional uncertainties arising
from impacts of climate change, invasive species, and biodiversity
losses (Cochard 2011,2013;Richardson et al. 2015;Rijal and
Cochard 2016).
The high expenditures for forest management, including in-
vestments for reforestation such as under Programs 327 and 661,
are probably difficult to maintain in the long term. Forest man-
agement (and the associated organizations) will therefore need to
become more cost effective to maintain and consolidate the gains
made during the recent forest transition phase. This needs to be
achieved within the context of newly rising pressures on forests.
Even though natural forest degradation may potentially be re-
duced by substituting wood extraction from natural forests with
wood cultivated in timber plantations (Pirard et al. 2016), natural
forests in Vietnam are still under considerable logging pressure
(Sikor and To 2011). With an increased pressure on forest re-
sources in neighboring countries (Laos and Cambodia) the appe-
tite for timber in Vietnam’s remaining old-growth forests may
intensify again. The population is still increasing in several up-
land regions, and our results indicate that the economic develop-
ment pathway of forest transition may have reached a limit.
Furthermore, the implementation of FLA has not led to satisfac-
tory outcomes everywhere, with inequalities arising within as
well as between certain communities. The resulting renewed pres-
sures on forest resources may not be easily controlled and man-
aged (McElwee 2016).
PFES schemes can play important roles in increasing incentives
for the protection and qualitative improvement of natural forests.
Given current information gaps, new programs should, however,
be accompanied by research that can foster a better grounding of
PFES programs in ecological and socio-economic understanding.
More information is, for example, needed about the sustainability
of exotic plantation forestry, in particular alien trees’ long-term
impacts on soils and local climate, including changes in ecological
dynamics and risks of invasion and fire. More research is needed
on the effects of natural forest fragmentation on biodiversity and
vegetation stability, novel biological threats such as tropical vine
invasions that smother forest regeneration, and the functions of
exotic plantations to serve as wildlife corridors (Gérard et al. 2015;
Le et al. 2012;Cochard 2011;Dickinson and Van 2006). Better pro-
tection should be afforded to remaining primary forests within
conservation areas, and particular efforts should be directed to
the last remnants of lowland forests; these forests are composed
of unique sets of potentially highly endangered species (Sterling
et al. 2006;McElwee 2010,2016;Cochard 2016;Webb and Kabir
2009). Research should also address the question of how local
people can be adequately empowered in conservation-oriented
forest management and in how fast-changing traditional land use
systems can contribute to natural forest protection (Boissière et al.
2009;McElwee 2008;Bayrak et al. 2015). In many constellations,
investment in the building of permanent field terraces is likely
more cost-effective and sustainable than shifting cultivation
(Jourdain et al. 2014). If it is found, however, that some forms of
traditionally practiced shifting cultivation represent a sustainable
and locally appropriate form of land management (Fox et al. 2000;
Jakobsen et al. 2007), FLA schemes could also be adjusted to legal-
ize certain flexible forms of land ownership and management.
Acknowledgements
This research was supported financially by the University of
Lausanne, Switzerland. Sabine Güsewell (ETH Zürich) provided
helpful advice on issues of statistical analyses. Michael Epprecht
(CDE, University Berne) kindly made available poverty data on
Vietnam. Two anonymous reviewers provided constructive com-
ments on the manuscript.
References
Allman, J., Nhan, V.Q., Thang, N.M., San, P.B., and Man, V.D. 1991. Fertility and
family planning in Vietnam. Stud. Family Plann. 22(5): 308–317. doi:10.2307/
1966685.
Alther, C., Castella, J.-C., Novosad, P., Rousseau, E., and Tran, T.H. 2002. Impact of
accessibility on the range of livelihood options available to farm households
in mountainous areas of northern Viet Nam. In Doi Moi in the mountains.
Land use changes and farmers’ livelihood strategies in Bac Kan Province, Viet
Nam. Edited by J.-C. Castella and D.Q. Dang. The Agricultural Publishing
House, Hanoi, Vietnam. pp. 121–146.
Amat, J.-P., Phung, T.B., Robert, A., and Tran, H.N. 2010. Can fast-growing species
form high-quality forests in Vietnam, examples in Thua Thien-Hue province.
Bois For. Trop. 305(3): 67–76.
Ankersen, J., Grogan, K., Mertz, O., Fensholt, R., Castella, J.-C., Lestrelin, G., et al.
2015. Vietnam’s forest transition in retrospect: Demonstrating weaknesses in
business-as-usual scenarios for REDD+. Environ. Manage. 55: 1080–1092. doi:
10.1007/s00267-015-0443-y. PMID:25588807.
Barney, K. 2005. Central plans and global exports: tracking Vietnam’s forestry
commodity chains and export links to China. Forest Trends, Washington D.C.
Available from http://www.cifor.org/publications/pdf_files/research/
governance/foresttrade/Attachment7-KBarney-VietnamReport.pdf [accessed
20 January 2016].
Bayrak, M.M., Tran, N.T., and Burgers, P. 2015. Formal and indigenous forest-
management systems in Central Vietnam. Implications and challenges for
REDD+. In Shifting cultivation and environmental change: Indigenous peo-
ple, agriculture and forest conservation. Edited by M.F. Cairns. Earthscan,
Routledge, London. pp. 319–334.
Beresford, M. 2008. Doi Moi in review: The challenges of building market socialism
in Vietnam. J. Contemp. Asia, 38(2): 221–243. doi:10.1080/00472330701822314.
Boissière, M., Sheil, D., Basuki, I., Wan, M., and Le, H. 2009. Can engaging local
people’s interests reduce forest degradation in Central Vietnam? Biodivers.
Conserv. 18(10): 2743–2757. doi:10.1007/s10531-009-9627-1.
Brauer, J. 2009. War and nature. The environmental consequences of war in a
globalized world. AltaMira Press, Plymouth, UK.
Bruggeman, D., Meyfroidt, P., and Lambin, E.F. 2016. Forest cover changes in
Bhutan: Revisiting the forest transition. Appl. Geogr. 67: 49–66. doi:10.1016/
j.apgeog.2015.11.019.
Cochard et al. 213
Published by NRC Research Press
Environ. Rev. Downloaded from www.nrcresearchpress.com by MONASH UNIVERSITY on 08/03/17
For personal use only.
Castella, J.-C., and Dang, D.Q. (Editors). 2002. Doi Moi in the mountains. Land use
changes and farmers livelihood strategies in Bac Kan Province, Vietnam. The
Agricultural Publishing House, Hanoi, Vietnam.
Castella, J.-C., Gevraise, V., and Novosad, P. 2005a. Centralized planning and
economic reforms in a mountainous region of Vietnam. J. Contemp. Asia,
35(2): 166–182. doi:10.1080/00472330580000111.
Castella, J.-C., Pham, H.M., Kam, S.P., Villano, L., and Tronche, N.R. 2005b. Anal-
ysis of village accessibility and its impact on land use dynamics in a moun-
tainous province of northern Vietnam. Appl. Geogr. 25(4): 308–326. doi:10.
1016/j.apgeog.2005.07.003.
Castella, J.-C., Boissau, S., Nguyen, H.T., and Novosad, P. 2006. Impact of forest
land allocation on land use in a mountainous province of Vietnam. Land Use
Policy, 23(2): 147–160. doi:10.1016/j.landusepol.2004.07.004.
Castrence, M., Nong, H.D., Tran, C.C., Young, L., and Fox, J. 2014. Mapping urban
transitions using multi-temporal Landsat and DMSP-OLS night-time lights
imagery of the Red River Delta in Vietnam [online]. Land, 3(1): 148–166. doi:
10.3390/land3010148.
Catacutan, D., Pham, T.T., Dam, V.B., Simelton, E., To, T.H., Enright, A., et al.
2012. Major challenges and lessons learnt from payments for forest environ-
mental services (PFES) schemes in Vietnam. Policy Brief. World Agroforestry
Centre, Hanoi, Vietnam.
Chazdon, R.L. 2008. Beyond deforestation: restoring forests and ecosystem ser-
vices on degraded lands. Science, 320(5882): 1458–1460. doi:10.1126/science.
1155365. PMID:18556551.
Chi, V.K., Van Rompaey, A., Govers, G., Vanacker, V., Schmook, B., and Hieu, N.
2013. Land transitions in Northwest Vietnam: an integrated analysis of bio-
physical and socio-cultural factors. Hum. Ecol. 41(1): 37–50. doi:10.1007/s10745-
013-9569-9.
Clemens, G., Fiedler, S., Nguyen, D.C., Nguyen, V.D., Schuler, U., and Stahr, K.
2010. Soil fertility affected by land use history, relief position and parent
material under a tropical climate in NW-Vietnam. Catena, 81(2): 87–96. doi:
10.1016/j.catena.2010.01.006.
Clement, F., and Amezaga, J.M. 2008. Linking reforestation policies with land
use change in northern Vietnam: Why local factors matter. Geoforum, 39(1):
265–277. doi:10.1016/j.geoforum.2007.05.008.
Clement, F., and Amezaga, J.M. 2009. Afforestation and forestry land allocation
in northern Vietnam: analyzing the gap between policy intentions and out-
comes. Land Use Policy, 26(2): 458– 470. doi:10.1016/j.landusepol.2008.06.003.
Clement, F., Orange, D., Williams, M., Mulley, C., and Epprecht, M. 2009. Drivers
of afforestation in Northern Vietnam: assessing local variations using geo-
graphically weighted regression. Appl. Geogr. 29(4): 561–576. doi:10.1016/j.
apgeog.2009.01.003.
Cochard, R. 2011. Consequences of deforestation and climate change on biodi-
versity. In Land use, climate change and biodiversity modeling: perspectives
and applications. Edited by Y. Trisurat, R. Shrestha, and R. Alkemade. IGI
Global, Hershey, U.S.A. pp. 30–55.
Cochard, R. 2013. Natural hazards mitigation services of carbon-rich ecosystems.
In Ecosystem services and carbon sequestration in the biosphere. Edited by
R. Lal, K. Lorenz, R.F. Hüttl, B.U. Schneider, and J. von Braun. Springer, Heidel-
berg, Germany. pp. 221–293.
Cochard, R. 2016. Scaling the costs of natural ecosystem degradation and biodi-
versity losses in Aceh Province, Sumatra. In Redefining diversity and dynam-
ics of natural resources management in Asia. Volume I. Edited by G. Shivakoti,
U. Pradhan, and Helmi. Elsevier, Amsterdam, Netherlands. pp. 233–273.
Dang, T.K.P., Turnhout, E., and Arts, B. 2012. Changing forestry discourses in
Vietnam in the past 20 years. For. Policy Econ. 25: 31–41. doi:10.1016/j.forpol.
2012.07.011.
Dao, M.T., Kono, Y., Yanagisawa, M., Leisz, S.J., and Kobayashi, S. 2009. Linkage of
policies and programs with land cover and land use changes in the Northern
Mountain Region of Vietnam: A village-level case study. Southeast Asian
Stud. 47 (3): 244–262. Available from http://ci.nii.ac.jp/naid/110007529888
[accessed 10 August 2015].
De Jong, W., Do, D.S., and Trieu, V.H. 2006. Forest rehabilitation in Vietnam.
Histories, realities and future. Center for International Forestry Research
(CIFOR), Bogor, Indonesia. Available from http://www.cifor.org/library/2106/
forest-rehabilitation-in-vietnam-histories-realities-and-future/ [accessed
8 November 2015].
De Koninck, R. 1999. Deforestation in Vietnam. International Development Re-
search Centre, Ottawa, Canada.
D’haeze, D., Deckers, J., Raes, D., Phong, T.A., and Loi, H.V. 2005. Environmental
and socio-economic impacts of institutional reforms on the agricultural sec-
tor of Vietnam. Land suitability assessment for Robusta coffee in the Dak Gan
region. Agric. Ecosyst. Environ. 105(1): 59–76. doi:10.1016/j.agee.2004.05.009.
Dickinson, C.J., and Van, N.T. (Editors). 2006. An Assessment of the fauna and
flora of the Green Corridor Forest Landscape, Thua Thien Hue Province,
Vietnam. Report No 7: Green Corridor Project, WWF Greater Mekong, and
Vietnam Country Programme and FPD Thua Thien Hue Province, Vietnam.
Dien, V.T., Bay, P.N., Stephen, P., Chau, V.T., Grais, A., and Petrova, S. 2013. Land
use, forest cover change and historical GHG emission. Lam Dong Province,
Vietnam. United States Agency for International Development (USAID), Low-
ering Emissions in Asia’s Forests (LEAF), Hanoi. Available from http://
www.leafasia.org/library/usaid-leaf-land-use-forest-cover-change-and-
historical-ghg-emission-english-version [accessed 20 January 2016].
Disperati, L., and Virdis, S.G.P. 2015. Assessment of land-use and land-cover
changes from 1965 to 2014 in Tam Giang-Cau Hai Lagoon, central Vietnam.
Appl. Geogr. 58: 48–64. doi:10.1016/j.apgeog.2014.12.012.
Do, D.S. 1994. Shifting cultivation in Vietnam: its social, economic and environ-
mental values relative to alternative land use. International Institute for
Environment and Development (IIED), London, UK. Available from http://
pubs.iied.org/7505IIED.html [accessed 20 January 2016]
Doutriaux, S., Geisler, C., and Shively, G. 2008. Competing for coffee space:
development-induced displacement in the Central Highlands of Vietnam.
Rural Sociol. 73(4): 528–554. doi:10.1526/003601108786471422.
Economist. 2016. Vietnam’s economy: the other Asian tiger. Vietnam’s success
merits a closer look. Available from http://www.economist.com/news/
leaders/21703368-vietnams-success-merits-closer-look-other-asian-tiger
[accessed 20 August 2016].
Epprecht, M., and Heinimann, A. 2004. Socio-economic atlas of Vietnam. A
depiction of the 1999 housing census. Swiss National Centre of Competence
in Research (NCCR) North-South, University of Berne. Available from http://
www.north-south.unibe.ch/content.php/publication/id/1712 [accessed
20 January 2016]
Epprecht, M., Müller, D., and Minot, N. 2011. How remote are Vietnam’s ethnic
minorities? An analysis of spatial patterns of poverty and inequality. Ann.
Regional Sci. 46(2): 349–368. doi:10.1007/s00168-009-0330-7.
FAO. 2016. State of the world’s forests 2016. Forests and agriculture: land-use
challenges and opportunities. Food and Agriculture Organization of the
United Nations (FAO), Rome, Italy. Available from http://www.fao.org/3/
a-i5588e.pdf [accessed 10 September 2016]
Fatoux, C., Castella, J.-C., Zeiss, M., and Pham, H.M. 2002. From rice cultivator to
agroforester within a decade: The impact of Doi Moi on agricultural diversi-
fication in a mountainous commune of Cho Moi District, Bac Kan Province,
Viet Nam. In Doi Moi in the mountains. Land use changes and farmers’ liveli-
hood strategies in Bac Kan Province, Viet Nam. Edited by J.-C. Castella and
D.Q. Dang. The Agricultural Publishing House, Hanoi, Vietnam. pp. 73–97.
Ferretti-Gallon, K., and Busch, J. 2014. What drives deforestation and what stops
it? A meta-analysis of spatially explicit econometric studies. Working Paper
361. Center for Global Development, Washington D. C. Available from http://
papers.ssrn.com/sol3/papers.cfm?abstract_id=2458040 [accessed 20 August
2015]
Folving, R., and Christensen, H. 2007. Farming system changes in the Vietnam-
ese uplands – using fallow length and farmers’ adoption of Sloping Agricul-
tural Land Technologies as indicators of environmental sustainability. Geogr.
Tidsskr. 107(1): 43–58. doi:10.1080/00167223.2007.10801374.
Fox, J., Dao, M.T., Rambo, A.T., Nghiem, P.T., Le, T.C., and Leisz, S. 2000. Shifting
cultivation: a new old paradigm for managing tropical forests. BioScience,
50(6): 521–528. doi:10.1641/0006-3568(2000)050[0521:SCANOP]2.0.CO;2.
Fox, J., Leisz, S., Dao, M.T., Rambo, A.T., Nghiem, P.T., and Le, T.C. 2001. Shifting
cultivation without deforestation: a case study in the mountains of North-
western Vietnam. In GIS and remote sensing applications in biogeography
and ecology. Edited by A.C. Millington, S.J. Walsh, and P.E. Osborne. Springer,
Heidelberg. pp. 289–307.
FPD. 2014. Monthly data on violation of forest law enforcement from November
2007 to March 2014 [online]. Forest Protection Department (FPD), Vietnam.
Available from http://kiemlam.org.vn/ [accessed 30 May 2015].
FSIV. 2009. Vietnam Forestry Outlook Study. Working Paper No. APFSOS II/WP/
2009/09. Forest Science Institute of Vietnam (FSIV). FAO Regional Office,
Bangkok. Available from http://www.fao.org/docrep/014/am254e/am254e00.pdf
[accessed 20 January 2016].
Geist, H.J., and Lambin, E.F. 2002. Proximate causes and underlying driving
forces of tropical deforestation. BioScience, 52(2): 143–150. doi:10.1641/0006-
3568(2002)052[0143:PCAUDF]2.0.CO;2.
Gérard, A., Ganzhorn, J.U., Kull, C.A., and Carrière, S.M. 2015. Possible roles of
introduced plants for native vertebrate conservation: the case of Madagascar.
Restor. Ecol. 23(6): 768–775. doi:10.1111/rec.12246.
Gomiero, T., Pettenella, D., Trieu, G.P., and Paoletti, M.G. 2000. Vietnamese
uplands: Environmental and socio-economic perspective of forest land allo-
cation and deforestation process. Environ. Dev. Sustain. 2(2): 119–142. doi:10.
1023/A:1011443731674.
Griffin, A.R., Nghiem, Q.C., Harbard, J.L., Do, H.S., Harwood, C.E., Price, A., et al.
2015. Breeding polyploid varieties of tropical acacias: progress and prospects.
South. For. 77(1): 41–50. doi:10.2989/20702620.2014.999303.
Grogan, K., Pflugmacher, D., Hostert, P., Kennedy, R., and Fensholt, R. 2015.
Cross-border forest disturbance and the role of natural rubber in mainland
Southeast Asia using annual Landsat time series. Remote Sens. Environ. 169:
438–453. doi:10.1016/j.rse.2015.03.001.
GSO. 2015. Statistics. General Statistics Office of Vietnam (GSO), Vietnam
[online]. Available from https://www.gso.gov.vn/default.aspx?tabid=714
[accessed 5–25 May 2015].
He, J., Lang, R., and Xu, J. 2014. Local dynamics driving forest transition: insights
from upland villages in Southwest China [online]. Forests, 5: 214–233. doi:10.
3390/f5020214.
Heinimann, A. 2006. Patterns of land cover change in the Lower Mekong Region.
The relevance of mesoscale approaches. Ph.D. thesis, Centre for Develop-
ment and Environment (CDE), Geographical Institute, University of Berne,
Switzerland.
Hieu, P.S., Thuy, V.H., and Van, L.T. 2011. Main characteristics of statistical data
214 Environ. Rev. Vol. 25, 2017
Published by NRC Research Press
Environ. Rev. Downloaded from www.nrcresearchpress.com by MONASH UNIVERSITY on 08/03/17
For personal use only.
and the statistical system for wood and wood-processing products in Viet-
nam. Small-scale For. 10(2): 185–198. doi:10.1007/s11842-010-9138-7.
Ho, D.H. 2014. The edible plants in Vietnam. Vietnamese Cereal Grains [online].
Available from http://www.edibleplantsinvietnam.com/vietnamese-cereal-
grains.html [accessed 24 February 2016].
Hoang, H.T.T., Vanacker, V., Van Rompaey, A., Vu, K.C., and Nguyen, A.T. 2014.
Changing human-landscape interactions after development of tourism in
the northern Vietnamese Highlands. Anthropocene, 5: 42–51. doi:10.1016/j.
ancene.2014.08.003.
Hosonuma, N., Herold, M., De Sy, V., De Fries, R.S., Brockhaus, M., Verchot, L.,
et al. 2012. An assessment of deforestation and forest degradation drivers in
developing countries [online]. Environ. Res. Lett. 7: 044009 (12 pp.). doi:10.
1088/1748-9326/7/4/044009.
Hung, T.T., Almeida, A.C., Eyles, A., and Mohammed, C. 2016. Predicting produc-
tivity of Acacia hybrid plantations for a range of climates and soils in Viet-
nam. For. Ecol. Manage. 367: 97–111. doi:10.1016/j.foreco.2016.02.030.
ICEM. 2003. Vietnam national report on protected areas and development. Re-
view of protected areas and development in the Lower Mekong River Region.
International Centre for Environmental Management (ICEM), Indooroopilly,
Queensland, Australia. Available from http://cmsdata.iucn.org/downloads/
vietnam_pad.pdf [accessed 20 May 2015].
Jackson, R.B., Jobbagy, E.G., Avissar, R., Roy, S.B., Barrett, D.J., Cook, C.W., et al.
2005. Trading water for carbon with biological carbon sequestration. Sci-
ence, 310(5756): 1944–1947. doi:10.1126/science.1119282.
Jadin, I., Vanacker, V., and Hoang, T.T.H. 2013. Drivers of forest cover dynamics
in smallholder farming systems: the case of Northwestern Vietnam. Ambio,
42(3): 344–356. doi:10.1007/s13280-012-0348-4. PMID:23001944.
Jakobsen, J., Rasmussen, K., Leisz, S., Folving, R., and Nguyen, V.Q. 2007. The
effects of land tenure policy on rural livelihoods and food sufficiency in the
upland village of Que, North Central Vietnam. Agr. Syst. 94(2): 309–319.
doi:10.1016/j.agsy.2006.09.007.
Jamieson, N.L., Le, T.C., and Rambo, T.A. 1998. The development crisis in Viet-
nam’s mountains. East-West Center Special Reports, 6: 1–32.
Jourdain, D., Boere, E., van den Berg, M., Dang, Q.D., Cu, T.P., Affholder, F., and
Pandey, S. 2014. Water for forests to restore environmental services and
alleviate poverty in Vietnam: a farm modelling approach to analyze alternative
PES programs. Land Use Policy, 41: 423–437. doi:10.1016/j.landusepol.2014.
06.024.
Keil, A., Saint-Macary, C., and Zeller, M. 2008. Maize boom in the uplands of
Northern Vietnam: economic importance and environmental implications.
Department of Agricultural Economics and Social Sciences in the Tropics and
Subtropics, University of Hohenheim, Stuttgart, Germany. Research in De-
velopment Economics and Policy, Discussion Paper No. 4/2008.
Kiernan, K. 2010. Human impacts on geodiversity and associated natural values
of bedrock hills in the Mekong Delta. Geoheritage, 2(3): 101–122. doi:10.1007/
s12371-010-0015-8.
Kirk, M., and Nguyen, D.A.T. 2009. Land-tenure policy reforms. Decollectiviza-
tion and the Doi Moi system in Vietnam. IFPRI Discussion Paper 00927. Inter-
national Food Policy Research Institute (IFPRI), Washington, D.C., U.S.A.
Available from http://www.ifpri.org/publication/land-tenure-policy-reforms
[accessed 20 January 2016].
Kull, C.A., Ibrahim, C.K., and Meredith, T.C. 2007. Tropical forest transitions and
globalization: Neo-liberalism, migration, tourism, and international conserva-
tion agendas. Soc. Natur. Resour. 20: 723–737. doi:10.1080/08941920701329702.
Lambin, E.F., and Meyfroidt, P. 2010. Land use transitions: Socio-ecological feed-
back versus socio-economic change. Land Use Policy, 27: 108–118. doi:10.1016/
j.landusepol.2009.09.003.
Le, B.T., Nguyen, T.L.T., and Adkins, S. 2012. Damage caused by Merremia
eberhardtii and Merremia boisiana to biodiversity of Da Nang City, Vietnam.
Pak. J. Weed Sci. Res. 18: 895–905.
Leinenkugel, P., Wolters, M.L., Oppelt, N., and Kuenzer, C. 2015. Tree cover and
forest cover dynamics in the Mekong Basin from 2001 to 2011. Remote Sens.
Environ. 158: 376–392. doi:10.1016/j.rse.2014.10.021.
Leisz, S.J. 2009. Dynamics of land cover and land use changes in the Upper Ca
River Basin of Nghe An, Vietnam. Southeast Asian Studies, 47(3): 287–308.
Available from http://ci.nii.ac.jp/naid/110007529894 [accessed 30 August 2015].
Li, L., Liu, J., Long, H., de Jong, W., and Youn, Y.-C. 2015. Economic globalization,
trade and forest transition – the case of nine Asian countries. For. Policy Econ.
(In press). doi:10.1016/j.forpol.2015.12.006.
Lippe, M., Thai Minh, T., Neef, A., Hilger, T., Hoffmann, V., Lam, N.T., and
Cadisch, G. 2011. Building on qualitative datasets and participatory processes
to simulate land use change in a mountain watershed of Northwest Vietnam.
Environ. Modell. Softw. 26(12): 1454–1466. doi:10.1016/j.envsoft.2011.07.009.
Liu, J., Liang, M., Li, L., Long, H., and de Jong, W. 2016. Comparative study of the
forest transition pathways of nine Asia-Pacific countries. For. Policy Econ.
(In press). doi:10.1016/j.forpol.2016.03.007.
MARD. 2015. Annual decisions by the Minister of Ministry of Agriculture and
Rural Development (MARD) on annual forest cover in Vietnam [online]. Avail-
able from http://tongcuclamnghiep.gov.vn/ [accessed 30 May 2015].
Mather, A.S. 1992. The forest transition. Area, 24(4): 367–379.
Mather, A.S. 2007. Recent Asian forest transitions in relation to forest-transition
theory. Int. For. Rev. 9(1): 491–502. doi:10.1505/ifor.9.1.491.
Mather, A.S., and Fairbairn, J. 2000. From floods to reforestation: the forest transi-
tion in Switzerland. Environ. Hist. 6: 399–421. doi:10.3197/096734000129342352.
Mather, A.S., Fairbairn, J., and Needle, C.L. 1999. The course and drivers of the
forest transition: the case of France. J. Rural Stud. 15(1): 65–90. doi:10.1016/
S0743-0167(98)00023-0.
McElwee, P. 2004. You say illegal, I say legal. The relationship between ‘illegal’
logging and land tenure, poverty, and forest use rights in Vietnam. J. Sustain-
able For. 19(1–3): 97–135. doi:10.1300/J091v19n01_06.
McElwee, P. 2008. Forest environmental income in Vietnam: household socio-
economic factors influencing forest use. Environ. Conserv. 35(2): 147–159.
doi:10.1017/S0376892908004736.
McElwee, P. 2009. Reforesting “bare hills” in Vietnam: Social and environmental
consequences of the 5 million hectare reforestation program. Ambio: J. Human.
Environ. 38(6): 325–333. doi:10.1579/08-R-520.1. PMID:19860156.
McElwee, P. 2010. Resource use among rural agricultural households near pro-
tected areas in Vietnam: the social costs of conservation and implications for
enforcement. Environ. Manage. 45(1): 113–131. doi:10.1007/s00267-009-9394-5.
PMID:19924473.
McElwee, P. 2012. Payments for environmental services as neoliberal market-
based forest conservation in Vietnam: Panacea or problem? Geoforum, 43(3):
412–426. doi:10.1016/j.geoforum.2011.04.010.
McElwee, P. 2016. Forests are gold. Trees, people and environmental rule in
Vietnam. University of Washington Press, Seattle, U.S.A.
McElwee, P., and Nguyen, C.T. 2014. Assessment of the 3 year implementation of
PFES policy in Vietnam 2011-2014. Winrock International, Vietnam Forests
and Deltas, Hanoi.
McElwee, P., Nghiem, T., Le, H., Vu, H., and Tran, N. 2014. Payments for environ-
mental services and contested neoliberalisation in developing countri