Forest Status And Threat In Aceh Biology Essay

Published:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Despite substantial international funding to protect rainforests, global deforestation rates show little sign of abatement, suggesting that previous efforts have generally had limited success (Jepson et al., 2001; Whitten et al., 2001). Whilst the ongoing loss of tropical rainforests represents one of the most serious threats to biodiversity (Sodhi, 2008; Sodhi et al., 2010) recent discussions on tropical deforestation have focussed on its contribution to climate change (Achard et al., 2007; DeFries et al., 2007). The International PaneI on Climate Change (IPCC) estimates that destruction of forests contributes around 18% of the greenhouse gas emissions entering Earth's atmosphere (IPCC, 2007). Failure to avoid this deforestation is predicted to greatly accelerate global warming (Fearnside, 2000; Gullison et al., 2007). In response, forest conservation initiatives are considering policy approaches for 'reducing emissions from deforestation and degradation' (REDD), which essentially pays governments to reduce deforestation below an estimated background rate (Blom et al., 2010). These schemes require reliable baseline data on their forest stocks, with varying levels of detail. Identifying the location and rates of forest loss is important information for law enforcement agencies responsible for mitigating this threat.

In Aceh, Indonesia, illegal logging and forest clearance poses a serious threat to ecosystem service functioning and therefore human well-being (van Beukering et al., 2003; Van Beukering et al., 2008; Bradshaw et al., 2010). In the aftermath of the devastating tsunami and protracted civil conflict, there was genuine and legitimate concern about the environmental impacts of the reconstruction and development processes. Also, with peace now having been achieved in Aceh, many former farmlands that had previously been abandoned during the conflict period and since turned back to forest, were being reopened for cultivation. Consequently, Aceh faced an unprecedented demand for its natural resources, such as timber, and its space for creating new farmland. With an increase in demand for timber, in part to support the tsunami reconstruction efforts and in part to provide employment, there has been a significant increase in the number of loggers entering the forests. The forests of Aceh are rich in tropical hardwood trees, such as semaram (Palaquium semaram), merbau (Intsia bijuga) and several species of meranti (Shoreo spec.) , which can obtain a high price on international markets and therefore make logging a lucrative business, for those trading, often outside of Aceh. The Government of Aceh's initiative for Reducing Emissions from Deforestation and Degradation (REDD) in Ulu Masen has brought significant international attention to this protection ecosystem. As proper landscape management and well managed forests provide a safeguard of essential environmental services such bioremediation and disaster relief (floods and landslides), the properties provide one of the strongest arguments for forest protection (Van Beukering et al., 2008).

Remotely sensed data offers an inexpensive and reliable option to estimate forest cover and forest cover change over extended periods. Identifying the location and rates of forest loss is important information for law enforcement agencies responsible for mitigating this threat . Because drivers of deforestation are often site and scale specific ((Lambin & Geist, 2003)) patterns of deforestation should be analyzed at regional scales to predict in situ threats and identify the local area at risk of deforestation (Linkie et al., 2010). A wide range of factors driving deforestation, acting on different spatial scales, have been suggested by various authors. The expansion of oil palm estate has frequently been mentioned as the major factor driving deforestation dynamics across the South East Asian region (Achard et al., 2007; Hansen et al., 2009) as well as in Indonesia (Jepson et al., 2001; Sandker et al., 2007). At a local or sub-national level, a consistent core set of predictors have been found to affect deforestation including: land use and tenure, local administration, soil, elevation, slope, distance to forest edge, distance to roads and distance to nearest settlement (Linkie et al., 2004; Andam et al., 2008; Gaveau et al., 2009b; Gaveau et al., 2009a; Linkie et al., 2010).

To provide a reliable and up-to-date assessment of forest cover change and current threat in and around the forests of Northern Aceh the present and past forest cover in Aceh was estimated for the years 2005-2009 using Landsat satellite imagery. Next, a logistic regression, including anthropogenic as well as topographic parameters was used to identify which factors influence local deforestation processes and to predict local patterns of deforestation.

Methods

Study area

Deforestation rates and patterns were investigated using data from the proposed protected are of Ulu Masen covering 739.748 ha of forest across the northern most tip of Aceh. The protected forest area spans six districts (Aceh Barat, Aceh Besar, Aceh Jaya, Bireuen, Pidie and Pidie Jaya) and adjoins the districts of Aceh Tengah and Nagan Raya (Figure 1). Forest cover in 2009 and forest cover change were separately estimated for the six Ulu Masen districts together with Nagan Raya and the eastern forests of Aceh Tengah. The forest boundary used for the analysis consists of forest inside the proposed Ulu Masen boundary (Pasya et al., 2007) and adjacent forest that extends outside the border. Peat swamp and coastal mangrove forest, being subjected to other deforestation dynamics, were omitted from analyses conducted within this study.

Study Area.tif

Figure 1 Overview of the study area.

Forest Classification

Forest classifications were based on 26 orthorectified Landsat 7 ETM+ satellite images, downloaded from the USGS Earth Resources Observation and Science centre (EROS) at http//glovis.usgs.gov. Each Landsat scene consists of seven spectral bands that span 185km x 170km with a 30m x 30m resolution. In order to cover the whole of the Ulu Masen forest block (Fig. 1), three Landsat scenes were downloaded for each year from 2005 to 2009 (Table 1). Due to a technical failure of the Landsat 7 satellite in 2003 and temporal distortion of images resulting from atmospheric hazes and cloud cover, 11 of the images downloaded did not provide a comprehensive coverage for their respective area. So, in order to obtain complete coverage of the study area, 11 additional scenes were downloaded. Shuttle Radar Topography Mission elevation data was downloaded from the Global Land Cover Facility Earth Science Data Interface (http://glcfapp.umiacs.umd.edu:8080/esdi/index.jsp) and used to rectify and calibrate the Landsat images.

Masking

To ensure accurate vegetation classification and to reduce erroneous results caused by noise and biases, anomalies such as water bodies and clouds were masked from the original images. As water is a strong absorber of near infra red waves (Band 4) this band was used to delineate water bodies. Strong atmospheric haze and clouds were extracted from the images using a linear combination of the first and seventh spectral band which define reflectance of blue-green light and mid infrared light, respectively.

Deshading

The Ulu Masen forest block largely covers the northernmost tip of the Bukit Barisan mountain range which generates differences in topographical relief throughout the study area. Consequently, there are various patterns of hill shading on the images. As a result, a single land cover type can produce different patterns of spectral reflectance, or spectral signatures, depending on site-specific topographical orientation and slope. These variations will lead to erroneous results if not controlled for. To remove shade effects from the images, different hill shade models, which mirror the relative surface reflections of a relief given the solar conditions at the moment of satellite passing, were produced for each satellite image. Image specific information on sun angle and azimuth relative to the scanned area and a 30m x 30m digital elevation model were used to produce the models. Next, for each image, every band was regressed against the expected illumination values using a ordinary least squares (OLS) regression (REF). The slope and gain of the regression line were then used to mask illumination effects resulting from a mixture of different aspects and slopes existing within the original image (Fig. 2 for example).

Band selection

To determine which combination of spectral bands most effectively separated variations in reflectance from bare areas, deforested areas and forested areas, a discriminant analysis was performed. Six spectral bands and a Normalized Difference Vegetation Index (NDVI; normalized ratio between red and near infrared light) were used to define the three land cover classes described above.

Image classification

Since different landscape elements reflect and absorb different parts of the solar light spectrum, they produce explicit spectral signatures that can be used for enhanced image classification. Training areas, used to extract spectral signatures for different land cover classes, were manually digitized on screen using vegetation cover data collected by FFI field teams between 2006 and 2009. Vegetation data consisted of 1244 point records which provided information on land cover classes that were encountered during field surveys (e.g. primary/undisturbed forest, secondary/disturbed forest, plantation, garden, grassland or bare soil, as well as relative forest cover classes (0-25%, 25-50%, 50-75% and 75-100%). The data set was split into two equal sized subsets which were subsequently used for image calibration and validation. Each Landsat scene was then separately classified using a maximum likelihood algorithm using R statistical software mlclust package (http://cran.r-project.org/web/packages/mclust/index.html0) and ESRI ArcGIS 9.2 remote sensing software. After classification small scale anomalies, defined as patches of 1 ha or less (<12 cells), were removed by merging them with their respective surroundings. The resulting forest cover estimate for the years 2006 - 2009 were validated using ground truthed control points collected in the field (METODS). For the years 2006, 2008 and 2009, 200 independent records and for the year 2007 another 100 records of forest cover, were used to validate the image interpretations of 'forest' and 'non-forest'. Since no validation data was available for the year 2005, the accuracy of the estimate was assumed to be within the same range of those of the subsequent years.

Spatio-temporal analyses

Several different methods to calculate deforestation rates have been proposed over the last decade. Many methods are based on standardized deforestation ratios facilitating comparisons of deforestation rates globally. However, to enable direct interpretation of these results and to then compare these deforestation rates with those from elsewhere in Sumatra and Borneo, annual deforestation rates were calculated as: (1) Percentage forest loss per year (%/yr), defined as the proportion of forest lost against a 2005 baseline forest cover estimate; and (2) Forest hectares loss per year (ha/yr). Forest coverage data layers were overlaid within the GIS to determine the location and rates of deforestation over successive years (i.e. from 2005-2009) for the entire study area and for each focal district.

Deforestation modelling

To investigate deforestation risk, the occurrence of deforestation was analyzed by means of logistic regression using topographic and anthropogenic parameters as predictors. Previous studies have emphasized the importance of area accessibility and human pressure to predict deforestation patterns (Kinnaird et al., 2003; Gaveau et al., 2009b; Linkie et al., 2010). A GIS dataset containing two topographic parameters (elevation, slope), two anthropogenic parameters (distance to nearest village, distance to nearest roads) and distance to forest edge was produced. Elevation data was obtained from the Shuttle Radar Topography Mission (www, which was then used to produce the slope layer. The forest edge information was taken from the 2005 forest cover classification. The position of settlements was obtained from 1:50,000 maps produced by Indonesian National Coordination Agency for Surveys and Mapping. For compatibility, all the spatial data layers were converted UTM47N projection with a 100x100m resolution raster format.

Spatial statistics

The forest risk model was determined using data from 100 forested points that were cleared between 2005 and 2009 and another 100 points that remained forested during this period. Each set of points was randomly selected using the Hawth's tools ArcGIS extension. To reduce the likelihood of spatial autocorrelation, points were selected with a minimum distance of 2 km between points. The GIS was then used to extract the physical covariates values at each of the 200 points. These spatial variables were imported into SPSS v.11 statistical software package (SPSS Inc., Chicago, IL) and log-transformed to prevent outliers from having a disproportionate influence on the result of the analysis. Next, a Spearman's rank correlation was conducted to test for collinearity between the four spatial covariates. Non-independence was identified between slope and elevation, as well as between distance to nearest road and distance to nearest village. Hence, a data reduction technique (PCA) was performed to produce uncorrelated variables for both the combined topographic variables as well as the anthropogenic variables. Factors with an eigenvalue of more than one were extracted and used in the subsequent analysis. This resulted in one factor describing the topographic variation present in the Ulu Masen area (eigenvalue: 1.461; 73% variance explained) an one factor describing the anthropogenic variation (eigenvalue: 1.524; 76% variance explained).

A stepwise logistic regression analyses was performed to determine which parameters, individually and in combination, best explained deforestation across the study area. Models were compared based on the Akaike Information Criterion (AIC; Burnham and Anderson 2002). Models that were within two AIC units (AIC) of the top ranked model with the smallest AIC were considered as plausible candidate models and their results discussed. The performance of a final regression model was then evaluated by calculating the area under the curve of receiver operating characteristics (ROC) plots. The presence of spatial autocorrelation in the model was then tested by calculating Moran's I statistic (Cliff and Ord 1981) using the SAM vs 3.0 software package (Rangel et al.) Subsequently, a spatially explicit deforestation risk model was constructed, using the parameters estimates for each predictor variable included in the final logistic model.

Results

Band selection

The results of the Discriminant analysis show that a combination of Bands 2/3/4/5/7 and the additional NDVI layer, effectively separated bare soil from deforested and forested areas. Spectral bands 3 and 4 (0.63-0.69 µm and 0.76-0.90 µm), which are absorbed by chlorophyll and water, respectively, proved to be the most reliable predictors for separating primary forest from other vegetation types, while Bands 5 and 7 were included to distinguish between vegetation and barren soils (Table1). Because considerable overlap existed between secondary regrowth and undisturbed forest, these groups were merged in the final classification.

Function

Eigenvalues

Structure coefficients

Significance

 

Class centroids

 

Score

(%)

R

 

b2

b3

b4

b5

b7

NDVI

 

Wilks

P

 

bare

dist.

forest

1

4.98

(77.0)

0.91

-0.78

-0.83

-0.23

-0.46

-0.50

0.47

0.07

<0.0001

-5.08

-1.69

1.58

2

1.49

(23.0)

0.77

 

-0.10

0.32

-0.77

-0.41

-0.11

-0.82

 

0.40

<0.0001

 

2.32

-1.74

0.39

Tabel 1 Results of the multivariate Discriminant analysis using six Landsat derived spectral bands to distinguish 3 land cover classes: bare, disturbed forest (dist.) and forest. The eigenvalues and the canonical correlation coefficient (R) as well as the structure coefficients between each function and the original variables are given.

The classification of forest and non-forest was found to be highly accurate (> 90%) with the kappa statistic being >0.81 statistic for all years (table2). For every year, the proportion of control points correctly predicted as non-forest land cover (specificity) was generally lower as compared to the proportion of points correctly predicted as forest (sensitivity).

Year

Sensitivity

Specitivity

Total Correct

Kappa

2009

92%

91%

92%

0.83

2008

98%

88%

94%

0.88

2007

98%

81%

93%

0.83

2006

98%

80%

92%

0.82

Table 1. Accuracy of the predicted forest cover maps based on a maximum likelihood estimations of Landsat data. Sensitivity: correctly predicted forest cover (eg true positives). Specificity: correctly predicted non-forest (eg true negatives). Kappa: chance corrected proportional agreement.

Spatial patterns of forest cover change

Comparing the 2005 and 2009 forest cover maps showed that a total of 36600 ha of forest had been cleared during that period, equivalent to a mean deforestation rate of 1.11%/yr ±0.513 (±95% C.I.) or 11310 ha/yr ±5.33. As a result forest covered 992470 ha in 2009 (Fig. 2, Table 4). Comparing deforestation rates across the study area districts revealed that the mean rates recorded in the non-Ulu Masen districts of Aceh Tengah (1.34%/yr±0.68) and Nagan Raya (1.18%/yr±1.809), but also Pidie Jaya (1.41%/yr±0.754), were higher than the study area average (1.11%/yr±0.513; Fig. 2, Table 4). The lowest deforestation rates were recorded in Aceh Besar (0.78%/yr ±0.437), then Bireuen (1.02%/yr±0.521) and Pidie (1.10%/yr±0.466). The most rapidly cleared forest type was lowland (2.1%/yr), followed by sub-montane (0.6%/yr), hill (0.4%/yr) and then montane (0.3%/yr).

Study Area_Defor.tif

Figure 2 Deforestation in the northern forest of Aceh between 2005 and 2009

District

2005

2006

2007

2008

2009

Total Deforestation

Forest Cover

Deforested

%

Forest Cover

Deforested

%

Forest Cover

Deforested

%

Forest Cover

Deforested

%

Forest Cover

Deforested

%

Deforested

%

Average annual deforestation % year-1

(*1000 ha)

(ha)

(*1000 ha)

(*1000 ha)

(*1000 ha)

(*1000 ha)

(*1000 ha)

(*1000 ha)

(*1000 ha)

(*1000 ha)

(*1000 ha)

(95% CI)

Aceh Barat

119.8

-

-

119.1

0.7

0.6

117.2

1.9

1.6

114.8

2.4

2.1

114.3

0.5

0.42

5.5

4.6

1.18 (±0.788)

Aceh Besar

136.6

-

-

136.1

0.5

0.4

135

1.1

0.8

133.1

1.9

1.4

132.4

0.7

0.52

4.3

3.1

0.78 (±0.437)

Aceh Jaya

254.6

-

-

252.8

1.8

0.7

250.1

2.7

1.1

246.8

3.3

1.3

243.6

3.2

1.3

11

4.3

1.1 (±0.277)

Pidie

207.7

-

-

207

0.7

0.4

204

3

1.4

201.1

2.9

1.4

198.8

2.4

1.19

9

4.3

1.1 (±0.466)

Pidie Jaya

58.9

-

-

58.6

0.3

0.5

57.6

1

1.7

56.3

1.3

2.3

55.6

0.6

1.15

3.3

5.6

1.41 (±0.754)

Bireuen

66.6

-

-

66.1

0.5

0.8

65.7

0.4

0.6

64.5

1.2

1.8

64.5

0.6

0.89

2.1

3.2

1.02 (±0.521)

Nagan Raya

80.4

-

-

79.39

0.5

0.6

79.8

0.1

0.2

76.6

3.1

3.9

76.6

0

0.01

3.8

4.7

1.18 (±1.809)

Aceh Tengah

112.6

-

-

111.5

1.2

1

109.8

1.7

1.5

107.3

2.4

2.2

106.7

0.6

0.61

5.9

4.7

1.34 (±0.68)

Total (UM)

777.6

-

-

773.6

4

0.51

763.9

9.7

1.25

752.1

11.8

1.54

744.6

7.4

0.99

33

4.2

1.07 (±0.429)

Total (No-UM)

259.6

-

-

256.99

2.2

0.85

255.3

2.2

0.86

248.4

6.7

2.62

247.9

1.2

0.5

11.8

4.5

1.21 (±0.941)

Grand total

1037.2

 

 

1030.59

6.2

0.6

1019.2

11.9

1.15

1000.5

18.5

1.82

992.5

8.6

0.86

44.8

4.3

1.11 (±0.513)

Table 4. Remaining forest cover (ha) and annual forest loss (ha and %) for the Ulu Masen study area and two adjacent districts from 2005-2009.

Drivers of deforestation

Based on the AIC value, a logistic model including distance to forest edge and both the anthropogenic and topographic factors as predictor variables best explained the observed pattern of deforestation (table 2). This model received high support (AIC = 7.96 wi=0.98) and had a good fit (Hosmer-Lemeshow test: χ2= 12.839, p = 0.117) and high accuracy (ROC = 0.933±0.017; Table 2). Deforestation was most strongly related to the distance to forest edge and significantly related to both anthropogenic and topographic variables (table 3). Between 2005 and 2009, an average deforestation rate of 1.1%/yr was recorded in the Ulu Masen forest Block.

Logistic Model

-2log(L)

K

ΔAIC

wi

HL-test

Sig.

ROC±SE

Dist. Forest edge + Anthropogenic + Topographic

124.5

4

0

0.9816

12.839

0.117

0.933±0.017

Dist. Forest edge + Anthropogenic

134.46

3

7.96

0.0184

15.46

0.051

0.923±0.018

Dist. Forest edge

190.21

2

61.71

0

5.19

0.736

0.896±0.024

Table 2 Overview of logistic regression models ranked according to their AIC

Deforestation was closely related to anthropogenic pressure and topographic constraints (Table 3) . Areas subjected to a high level of anthropogenic pressure, corresponding to forest closer to settlements and roads, and low topographic constraints relating to forest occurring at lower elevations and on flatter land being more likely to be cleared (Table 3). Deforestation risk was most strongly related to the distance to forest edge, emphasizing the importance of forest access. The final regression model explained 86.0% of the original observations and was not affected by spatial autocorrelation (Moran's I =-0.005, P = 0.132). The spatially explicit forest risk model (figure 3), which was based on the results of the final regression model (Table 2), was found to accurately predict deforestation that occurred between 2005 and 2009 with of kappa = 0.72.

Best logistic Model

B

S.E.

Wald

df

Sig.

Exp(B)

Distance forest edge

-3.24

0.67

23.76

1

0

0.04

Anthropogenic

-1.67

0.31

29.25

1

0

0.19

Topographic

-0.8

0.27

8.73

1

0

0.45

Constant

-0.65

0.31

4.4

1

0.04

0.52

Table 3. Parameter estimates and significance under the best performing logistic regression model describing the relationships between landscape variables and deforestation patterns across the northern forest of Aceh.

Study Area_Defor_Risk.tif

Figure 3. Estimated deforestation risk across the northern forest of Aceh.

Discussion

The current estimates of forest cover and deforestation for the northern forest of Aceh comprise the first step in realizing a framework for the implementation of REDD in Aceh. The estimated forest cover and hence deforestation rate encompass a guideline to assess the total forest estate available and loss for to be used for REDD purposes. Yet, small-scale forest disturbances (~100 m2) encountered on the ground cannot be distinguished using satellite data (~30x30m resolution). For that reason, these results reflect the total amount of forest cover and clearance rather than forest degradation as a result of selective logging. Lower specificity scores in relation to specialization scores (table 1) suggest that ground observations of non-forest land cover have more often been erroneously classified than was the case if forest was observed on the ground. This discrepancy between the land cover classes observed on the ground and the predicted forest cover maps is likely to be a result of the method used. Removing small-scale (eg 1ha) anomalies from the predicted forest cover maps increased accuracy of the forest cover maps at the cost of introducing a small bias towards forested land cover. However, one can argue whether small-scale alterations of forest canopy integrity should de facto be interpreted as true non-forest land cover, as these are a naturally occurring phenomenon in tropical rainforests (REF). Considering that the omission of small-scale gaps has been introduced independent of the year being estimated, the deforestation rates are consistent between consecutive years. The overall predictions of forest cover for all consecutive years proved highly accurate, therefore they are believed to realistically reflect spatial patterns of deforestation within the Northern Forest of Aceh.

Comparing deforestation rates across Indonesian regions revealed a marked variation between and within the islands of Sumatra and Borneo (Table 4). For example, the central Sumatra region had the highest deforestation rates reaching up to 5.50%/yr. From the case studies found, the Ulu Masen and Leuser regions in Aceh had some of the lowest deforestation rates that were much lower the average rate recorded from the selected case studies (Fig. 4).

Location

Year

Deforestation rate (%/yr)

Source

Central Sumatra region

1990-1997

3.20 - 5.50

Achard et al. (2002)

Bengkulu province, Sumatra

1985-2002

1.41

Linkie et al. (in press)

Southern Sumatra region

1972-2006

0.64 -2.86

Gaveau et al. 2007

Riau province, Sumatra

1982-2007

1.68

Uryu et al. (2008)

Sumatra island

1990-2000

2.56

Gaveau et al. (2009)

Borneo island

2002-2005

1.70

Langer et al (2007)

Ulu Masen-Aceh, Sumatra

2005-2008

1.11

This study

West-central Sumatra region

1995-2001

0.96

Linkie et al. (2008)

Leuser region-Aceh, Sumatra

2006-2009

0.90

AFEP-LIF (2009)

Table 4. Overview of comparable deforestation rates recorded in Indonesia over the period 1990-2009.

Figure 4. Deforestation rates across the islands of Sumatra and Borneo.

Over the period 2005-2009 a small increase in the annual deforestation rate was observed over the study area. This finding agrees with the observed deforestation rates reported by Hansen et al (2008). Yet, the temporal resolution of one-year intervals used in this study did not allow to make inferences about the statistical accuracy of these estimates and hence cannot be used to draw conclusions about deforestation trends over time.

Spatial deforestation patterns

Ultimate and proximate causes driving deforestation processes differ between various regions and spatial scales (REF). The spatial patterns of deforestation across the forest of northern Aceh highlighted the critical role of accessibility, with the importance of distance to forest edge having the largest influence on predicting deforestation (table 2). Topographical constraints, limiting the access to forest growing at higher elevations or in terrain that is more rugged, further reduced deforestation. This also explained why forests located closer to the forest edges and to settlements than hill forest, tended to be at a greater risk to clearance than hill forest (Fig 3). Deforestation levels were generally higher around settlements, presumably because travel time and cost are considerably lower when transporting timber across shorter distances. However, most of these settlements are also located at lower elevations adding the advantage of relatively easy access to lowland forests due to the lack of topographic barriers such as steep slopes of high elevation gradients, making it most susceptible to clearance. Whilst this highlights the importance of providing alternative livelihood opportunities and attractive incentives to reduce illegal logging and overexploitation by local communities living near the forest edge (Linkie et al. 2008), part of any solution will involve active forest protection.

Historic patterns of deforestation

The government sponsored and spontaneous transmigrations from Java to the northern parts of Sumatra in the early 1990s led to massive amounts of forest being converted to small-scale farmland. The deforestation pattern spread from the lowland coastal areas, where most transmigrants initially settled, inwards up the mountain slopes and higher forest plateaus. The 1990's featured a considerable economic growth in the South East Asia region resulting in a vast expansion of estate crops and the development of oil palm. In Aceh this has led to a reduction in forest cover of no less than 60.4% of total forest loss between 1984 and 1997 (Holmes 2002).

Following the collapse of the authoritarian Suharto regime in 1999, there was a burst of anti-government political activity in Aceh (Schulze, 2003). The economic crash and the renewed violence in the province eventually led to the armed conflict and military operations in 2001 (McCullogh, 2003; Schulze, 2003). Resultantly, deforestation rates halted and areas previously opened for agricultural purposes developed secondary forests. Only after the invoked peace agreement of 2006 and the booming demand for resources following rehabilitation of the province after the 2004-tsunami an sharp increase in illegal logging and conversion of forest for farmland were observed. Our deforestation estimates did not include the forest degradation caused by illegal timber trade and therefore represent a conservative estimate of the deforestation. Nevertheless, with the removal of the most accessible export-quality timber from our study area, many loggers would have turned their attention back to agriculture (e.g. small-scale farming or plantations), thereby contributing to the inflated deforestation rate.

Achard, F., DeFries, R., Eva, H., Hansen, M., Mayaux, P. & Stibig, H.J. (2007) Pan-tropical monitoring of deforestation. Environmental Research Letters, 2

Andam, K.S., Ferraro, P.J., Pfaff, A., Sanchez-Azofeifa, G.A. & Robalino, J.A. (2008) Measuring the effectiveness of protected area networks in reducing deforestation. Proceedings of the National Academy of Sciences of the United States of America, 105, 16089-16094.

Blom, B., Sunderland, T. & Murdiyarso, D. (2010) Getting REDD to work locally: lessons learned from integrated conservation and development projects. Environmental Science & Policy, 13, 164-172.

Bradshaw, C.J.A., Giam, X. & Sodhi, N.S. (2010) Evaluating the Relative Environmental Impact of Countries. Plos One, 5, 16.

DeFries, R., Achard, F., Brown, S., Herold, M., Murdiyarso, D., Schlamadinger, B. & de Souza, C. (2007) Earth observations for estimating greenhouse gas emissions from deforestation in developing countries. Environmental Science & Policy, 10, 385-394.

Fearnside, P.M. (2000) Global warming and tropical land-use change: Greenhouse gas emissions from biomass burning, decomposition and soils in forest conversion, shifting cultivation and secondary vegetation. Climatic Change, 46, 115-158.

Gaveau, D.L.A., Linkie, M., Suyadi, Levang, P. & Leader-Williams, N. (2009a) Three decades of deforestation in southwest Sumatra: Effects of coffee prices, law enforcement and rural poverty. Biological Conservation, 142, 597-605.

Gaveau, D.L.A., Epting, J., Lyne, O., Linkie, M., Kumara, I., Kanninen, M. & Leader-Williams, N. (2009b) Evaluating whether protected areas reduce tropical deforestation in Sumatra. Journal of Biogeography, 36, 2165-2175.

Gullison, R.E., Frumhoff, P.C., Canadell, J.G., Field, C.B., Nepstad, D.C., Hayhoe, K., Avissar, R., Curran, L.M., Friedlingstein, P., Jones, C.D. & Nobre, C. (2007) Tropical forests and climate policy. Science, 316, 985-986.

Hansen, M.C., Stehman, S.V., Potapov, P.V., Arunarwati, B., Stolle, F. & Pittman, K. (2009) Quantifying changes in the rates of forest clearing in Indonesia from 1990 to 2005 using remotely sensed data sets. Environmental Research Letters, 4

IPCC (2007) Findings of the IPCC Fourth Assessment Report: Climate Change 2007:synthesis report. In: Climate Change (ed. I.P.O.C. Change)

Jepson, P., Jarvie, J.K., MacKinnon, K. & Monk, K.A. (2001) The end for Indonesia's lowland forests? Science, 292, 859-+.

Kinnaird, M.F., Sanderson, E.W., O'Brien, T.G., Wibisono, H.T. & Woolmer, G. (2003) Deforestation trends in a tropical landscape and implications for endangered large mammals. Conservation Biology, 17, 245-257.

Lambin, E.F. & Geist, H.J. (2003) Regional differences in tropical deforestation. Environment, 45, 22-+.

Linkie, M., Smith, R.J. & Leader-Williams, N. (2004) Mapping and predicting deforestation patterns in the lowlands of Sumatra. Biodiversity and Conservation, 13, 1809-1818.

Linkie, M., Rood, E.J.J. & Smith, R.J. (2010) Modelling the effectiveness of enforcement strategies for avoiding tropical deforestation in Kerinci Seblat National Park, Sumatra. Biodiversity and Conservation, 19, 973-984.

McCullogh, L. (2003) Greed: the silent force of the conflict in Aceh In: Building human security in Indonesia. Humanitarian Poicy and Conflict reserach, Brussels

Pasya, R., Usher, G. & O Niles, J. (2007) Reducing Carbon emissions from Deforestation in the Ulu Masen Ecosystem, Aceh, Indonesia. In: A Triple benefit Project Design Note for CCBA Audit. Flora & Fauna International

Rangel, T.F., Diniz, J.A.F. & Bini, L.M. SAM: a comprehensive application for Spatial Analysis in Macroecology. Ecography, 33, 46-50.

Sandker, M., Suwarno, A. & Campbell, B.M. (2007) Will forests remain in the face of oil palm expansion? Simulating change in Malinau, Indonesia. Ecology and Society, 12

Schulze, K.E. (2003) The Struggle for an Independent Aceh: The Ideology, Capacity, and Strategy of GAM. Studies in Conflict & Terrorism, 26, 241 - 271.

Sodhi, N.S. (2008) Invited views in basic and applied ecology - Tropical biodiversity loss and people - A brief review. Basic and Applied Ecology, 9, 93-99.

Sodhi, N.S., Posa, M.R.C., Lee, T.M., Bickford, D., Koh, L.P. & Brook, B.W. (2010) The state and conservation of Southeast Asian biodiversity. Biodiversity and Conservation, 19, 317-328.

van Beukering, P.J.H., Cesar, H.S.J. & Janssen, M.A. (2003) Economic valuation of the Leuser National Park on Sumatra, Indonesia. Ecological Economics, 44, 43-62.

Van Beukering, P.J.H., Grogan, K., Hansfort, S.L. & Seager, D. (2008) An Economic Valuation of Aceh's forests: The road towards sustainable development. In. Vrije Universiteit, Amsterdam

Whitten, T., Holmes, D. & MacKinnon, K. (2001) Editorial: Conservation Biology: A Displacement Behavior for Academia? Conservation Biology, 15, 1-3.

Writing Services

Essay Writing
Service

Find out how the very best essay writing service can help you accomplish more and achieve higher marks today.

Assignment Writing Service

From complicated assignments to tricky tasks, our experts can tackle virtually any question thrown at them.

Dissertation Writing Service

A dissertation (also known as a thesis or research project) is probably the most important piece of work for any student! From full dissertations to individual chapters, we’re on hand to support you.

Coursework Writing Service

Our expert qualified writers can help you get your coursework right first time, every time.

Dissertation Proposal Service

The first step to completing a dissertation is to create a proposal that talks about what you wish to do. Our experts can design suitable methodologies - perfect to help you get started with a dissertation.

Report Writing
Service

Reports for any audience. Perfectly structured, professionally written, and tailored to suit your exact requirements.

Essay Skeleton Answer Service

If you’re just looking for some help to get started on an essay, our outline service provides you with a perfect essay plan.

Marking & Proofreading Service

Not sure if your work is hitting the mark? Struggling to get feedback from your lecturer? Our premium marking service was created just for you - get the feedback you deserve now.

Exam Revision
Service

Exams can be one of the most stressful experiences you’ll ever have! Revision is key, and we’re here to help. With custom created revision notes and exam answers, you’ll never feel underprepared again.