Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Protected Areas: Mixed Success in Conserving East Africa’s Evergreen Forests


In East Africa, human population growth and demands for natural resources cause forest loss contributing to increased carbon emissions and reduced biodiversity. Protected Areas (PAs) are intended to conserve habitats and species. Variability in PA effectiveness and ‘leakage’ (here defined as displacement of deforestation) may lead to different trends in forest loss within, and adjacent to, existing PAs. Here, we quantify spatial variation in trends of evergreen forest coverage in East Africa between 2001 and 2009, and test for correlations with forest accessibility and environmental drivers. We investigate PA effectiveness at local, landscape and national scales, comparing rates of deforestation within park boundaries with those detected in park buffer zones and in unprotected land more generally. Background forest loss (BFL) was estimated at −9.3% (17,167 km2), but varied between countries (range: −0.9% to −85.7%; note: no BFL in South Sudan). We document high variability in PA effectiveness within and between PA categories. The most successful PAs were National Parks, although only 26 out of 48 parks increased or maintained their forest area (i.e. Effective parks). Forest Reserves (Ineffective parks, i.e. parks that lose forest from within boundaries: 204 out of 337), Nature Reserves (six out of 12) and Game Parks (24 out of 26) were more likely to lose forest cover. Forest loss in buffer zones around PAs exceeded background forest loss, in some areas indicating leakage driven by Effective National Parks. Human pressure, forest accessibility, protection status, distance to fires and long-term annual rainfall were highly significant drivers of forest loss in East Africa. Some of these factors can be addressed by adjusting park management. However, addressing close links between livelihoods, natural capital and poverty remains a fundamental challenge in East Africa’s forest conservation efforts.


Tropical evergreen forests represent around 6% of the terrestrial surface in Eastern Africa, being found mainly in Eastern Congo, Rwanda and Burundi and Eastern Tanzania. They provide goods and services (i.e. natural capital) to rural and urban communities [1], [2], [3], are rich in species and local endemics [4], [5] and are vital carbon sinks, storing from 70 to more than 300 tonnes of carbon per ha, depending on structure, climate and location [6], [7].

East Africa’s evergreen forests also exhibit marked congruence with the most densely populated areas of Africa [8], [9], and may therefore be susceptible to habitat conversion. High human population growth [2], [10] coincides with the expansion of cropland, grazing land and forest plantations at the expense of natural forests [9]. Remaining forests are known to be degraded and declining, particularly in easily accessible coastal areas [11], [12], near main cities [13] and at low altitudes [7]. Towards the eastern edge of the Congo forests and further towards the coast, forests within and outside PAs are increasingly accessible through a network of roads. Additional pressure on forest resources is exerted by commercial timber trade supplying both urban expansion and growing demand from abroad [14]; part of this logging is illegal and thus unregulated [13], [15].

Global and regional analyses suggest that protected areas may be able to stop land clearing and to mitigate logging, hunting, fire and grazing [16][19]. Thereby, mixed-used PAs may be as effective, or more effective, than strict PAs in preventing forest fires and forest loss, especially in less remote areas [18], but see Burgess et al [20]. In East Africa, National Parks have firm restrictions on resource use and strong law enforcement [21], although there are exceptions (e.g. Mago National Park in Ethiopia; [22]). Nature Reserves are intended to protect biodiversity, but law enforcement is sporadic and they are often understaffed [21]. Game Parks are largely designed for conservation of large mammals [23] and sport hunting, and are only occasionally patrolled (predominantly during the hunting season). Forest Reserves (commonly gazetted as multi-resource use areas) are often located in areas with valuable timber stocks and used for extractive forestry. They may allow extractive resource use by adjacent communities; this extraction may be permit-regulated, but law enforcement is typically weak [21].

Figure 1. National forest trends in East Africa.

Shown are overall forest trends independent of protection status (a) and forest trends depending on protection status (b). Note that only Kenya, Tanzania, Uganda, Rwanda and Burundi are fully covered by the study area.

Analyses of PA effectiveness stress the importance of park management [18], [20] and accessibility [24], [25]. Processes outside PAs (i.e. encroachment of invasive species, regional and local pollution, and socio-economic pressures) may shape processes within PAs [26], [27]. A widespread lack of integration of PAs with local development and community needs can lead to land alienation [28]. Fear of conservation-related ‘land grabs’ may accelerate ‘defensive farming’, as local communities struggling to meet their resource needs expand the land under cultivation to formalize land tenure and gain land use security [29]. Also, ‘leakage’ may offset forest protection within parks by elevating forest loss in areas nearby, as demands for food and fuel still need to be met [30], [31].

Figure 2. Forest trends within individual parks of four different protection categories between 2001 and 2009 as function of initial forest size in 2001 (log10-scale).

For graphical display of forest trends we excluded (very small) PAs that increased their forests by more than 300%. Thus, we excluded five Forest Reserves: Mukugodo FR in Kenya (3.9 km2, 822%; forest cover in 2001 and forest change), Ngaia FR in Kenya (0.2 km2, 900%), Geita FR in Tanzania (0.2 km2, 600%), Vumari FR in Tanzania (0.6 km2, 433%), Mwalugulu FR in Tanzania (0.4 km2, 450%). On this basis, we also excluded four National Parks: Rubondo NP in Tanzania (0.4 km2, 800%), Murchison Falls NP in Uganda (14.5 km2, 391%), Mago NP in Ethiopia (0.4 km2, 3550%) and Ruma NP in Kenya (1.1 km2, 440%).

In our analyses, we evaluate the success of East Africa’s PA network for the conservation of evergreen forests (classified as broad-leaved evergreen tropical forest; International Geosphere Biosphere Programme classification) between 2001 and 2009. We quantify forest trends within PAs, in three buffer zones (B01∶0–1 km from park edge, B15∶1–5 km from park edge, B510∶5–10 km from park edge) and in the unprotected surrounding matrix. By focusing our buffer trend analyses on distances of up to 10 km from park boundaries, we minimize confounding problems of overlap between neighboring PAs. Unlike global scale analyses, which are typically restricted to IUCN category parks [27], our analyses cover all kinds of state managed reserves in East Africa (Forest Reserves, Game Parks, National Parks and Nature Reserves); however, we do not have accurate spatial data for community based management approaches (e.g. Wildlife Management Areas and Village Land Forest Reserves in Tanzania), so these reserves are excluded from analyses. To interpret our findings for forest management, we test for significant dependencies of forest trends on protection status and indicators of human pressure, i.e. population density, road networks, and distance to major towns and fire events.

Joppa et al. [32] have previously evaluated changes in natural vegetation and forest fragmentation across four moist tropical forests: the Amazon and Congo ‘wilderness forests’ and the high-biodiversity forests at the Atlantic coast and in West Africa. They find large geographic variation, however their analyses are static (using a 1 km spatial resolution dataset from 2002) and ignore effects of the matrix at larger spatial scales. In our study, we quantify forest trends at higher spatial resolution (500 m), to produce an assessment of park effectiveness over eight years, based on temporal evidence of forest change within and outside park boundaries, whilst also comparing forest trends within parks and buffers to regional background forest loss (BFL).

Table 2. Trends in forest cover across protection categories between 2001 and 2009.


Forest Trends

Between 2001 and 2009, forest cover in East Africa decreased in all countries, except the Southern Sudan region (Figure 1), and most severely in the (previously) forest-rich countries of Uganda and Rwanda. Forest decrease was strongest outside protected areas (background forest loss, hereafter BFL) with a decrease of 17,167 km2 (−9.3%) over the eight year period. Quantifying forest trends separately for ‘2001 to 2004’ (Period 1) and ‘2004 to 2009’ (Period 2) shows that areal forest loss slowed during Period 2 in some countries (Kenya, Ethiopia, Somalia, Zambia), while in others forest was gained during Period 2 following loss during Period 1 (Rwanda, Burundi, Congo), a pattern broadly consistent inside and outside protected areas (Table S1).

Figure 3. Satellite-derived estimates of forest trend within and around three PAs in East Africa.

Forest cover increased (green), decreased (red) or remained constant (orange). Some parks show significant loss in forest cover within their three buffer zones (0–1 km, 1 to 5 km, and 5–10 km). Other land cover transitions are white.

Table 3. Percentage of forest change in buffer zones around PAs between 2001 and 2009.

Table 4. Forest trends in buffer zones (B01, B15, and B510) around Effective National Parks (i.e. parks that increased or maintained their forest area between 2001 and 2009).

BFL differed among countries (Table S2), being high in Rwanda (−79.3%; −4,159 km2) and Burundi (−82.9%; 1,750 km2). Relative BFL was moderate in Uganda (−36.3%), although areal BFL was higher than in any of the other countries (4,609 km2). Relative BFL was very low in the Eastern Congo (−0.9%), although that still translated to 1,325 km2 of forest lost between 2001 and 2009. Note that relative BFL was also high or moderate in Somalia (85%), Zambia (82%), Ethiopia (56%) and Mozambique (45%). However, our analyses covered only small parts of these countries and areal BFL was low in Ethiopia and Zambia.

Figure 4. Changes in human population densities with increasing distance from parks across the study area in East Africa.

Patterns of human population densities within buffer zones of protected areas differ between protection categories.

Table 5. Significant drivers of forest trends (0: no forest loss, 1: forest loss) modelled using general linear models with logit link functions.

Countries differ in the presence and abundance of reserves within each protection category, as well as in the proportion of IUCN parks (Table 1). Overall, PAs lost comparatively little of their forest cover (378 km2; −0.6%), and in Uganda, Congo and Malawi there was an increase in forest cover within protected area boundaries (Figure 1). Game Parks were least effective, losing 24.4% of their 2,078 km2 forest cover present in 2001. Nature Reserves and Forest Reserves lost 5.3% and 3.5% of their respective 8,686 km2 and 11,337 km2 forest area. Only National Parks, often the best protected and funded PAs, increased their forest area, by 3.2% from 15,679 km2, although this is primarily driven by the success of National Parks in Tanzania (Figure 2).

PAs differed in their effectiveness depending on protection status, initial forest size and location of the park (Figure 2). The Wilcoxon rank-sum test with Bonferroni adjustment of P values (hereafter referred to as Pbonf) indicates significantly higher areal forest loss from Game Parks compared to Forest Reserves and National Parks (Pbonf <0.01), and significantly higher relative forest loss compared to National Parks, Nature Reserves and Forest Reserves (Pbonf <0.01).

Forest conservation success differed considerably within protection categories: nearly 50% of National Parks, 50% of Nature Reserves, 61% of Forest Reserves and 92% of Game Parks were Ineffective (Table 2), some of them losing more than 50% of their forests, especially when initial forest extent was small (Table 2). Parks encompassing smaller forest patches experienced stronger relative forest loss (linear model of relative change in forest cover as a function of initial forest size, including only Ineffective parks: National Parks: p<0.01; Forest Reserves: p<0.001; Game Parks: p  = 0.066), some losing their forests entirely (61 of 204 Ineffective Forest Reserves, 8 of 22 Ineffective National Parks, 5 of 24 Ineffective Game Parks) (Table 2).

Community benefits programs and resourcing are among the weakest points in protected area management effectiveness [33]. Detailed information on local management of selected PAs in our study area (see Tables S2 and S3) suggests that involving local communities in forest management improves forest conservation outcomes. Mukogodo FR in Kenya, for example, has largely been managed and conserved by the local indigenous community, with little interference of the government following initial monitoring and training [34]. Participatory forest management is also used in Vumari FR in Tanzania, whose legal status is listed as ‘excellent’, and benefits from regular council-funded patrols and conservation interventions such as licensing charcoal burning and pole-cutting. The extent to which PAs can conserve their forests are also likely to be governed by the trade-off between benefits associated with conservation and opportunity costs resulting from forsaken access to forest resources [35]. Economic value (e.g. presence of commercial timber, firewood and charcoal) and local use (e.g. charcoal, honey, grazing, thatching) of forests in Mukogodo FR for example, are low [36].


Ideally, information on forest cover before and after imposing land use restrictions would be used to determine the extent to which gazetting a new PA affects deforestation rates [37]. These analyses are beyond the scope of this study, but we can show that forest loss in the vicinity of some PAs exceeds the 9.3% BFL in East Africa (Figure 3, Table 3). In the B01 buffer, this is the case for 16 National Parks, 7 Nature Reserves, 67 Forest Reserves and 9 Game Parks; in the B15 buffer it is true for 13, 7, 85 and 8 and in the B510 buffer for 16, 7, 78 and 7 National Parks, Nature Reserves, Forest Reserves and Game Parks, respectively. Forest loss in buffer zones of some Effective National Parks is higher than BFL in East Africa (Table 4). Country-specific BFL is more severe in most countries than the overall BFL in East Africa (Table S4). There are still some PAs for each protection category and country, where forest loss within buffer zones exceeds country-specific BFL (Table S4).

Drivers of Forest Loss

Our analyses show that, on average, human density peaks at 1 to 2 km distance from park boundaries for Nature Reserves, Forest Reserves and Game Parks, but increases with distances from the boundaries of National Parks (Figure 4). Nature Reserves, Forest Reserves and Game Parks may be perceived by local communities to represent more viable resources for exploitation (legal or illegally), leading to population clustering in their immediate vicinity, although assessments of perceptions over access to resources need to be carried out within communities to support such conclusion.

As elsewhere [25], [38], human population growth and forest accessibility are significant drivers of the observed forest trends in East Africa. Logit models of the binary response variable ‘no forest loss’/‘forest loss’ show significant correlations with human population density, but also vegetation burning, slope and distance to road networks and towns (Table 5).

Steeper slopes are associated with lower deforestation pressures, presumably because they are less suitable for agriculture and also less accessible compared to gentler slopes [39], [40], which has been linked to lower opportunity costs and lower necessary spend for effective forest protection [40], [41]. Steepness (mean and minimum of slope) was significantly less in the B01 buffer zones of Ineffective PAs compared to slope in the B01 buffer of Effective PAs, but only in the case of Forest Reserves (Wilcoxon rank sum test: P<0.005 and P<0.01) and Game Parks (Wilcoxon rank sum test: P<0.05 and P<0.05). Thus, hampered forest accessibility appears to play a role in reducing deforestation in parks that allow extractive resource use and are generally less well-protected, but is less important in well-protected parks.

We also found a significant effect of long-term rainfall (Table 5), with forests in drier regions appearing more susceptible to habitat conversion. Possible explanations are that dry climates may affect the ability of forests to regenerate after disturbance, that dry forests burn more readily and more extensively [42] and are more suitable for production of charcoal and extraction of firewood. Also, remaining moist forests are often found on challenging mountain terrain.


Large areas of evergreen forests have been lost from East Africa during the 20th century [43], [44], [45] resulting in carbon emissions [6], reduced habitat for forest dependent biodiversity [5], [15], and reduced availability of essential ecosystem services [2], [46]. Initial conservation efforts in East Africa, like elsewhere, focused on creating PAs [20]. However, PAs - worldwide - have faced challenges imposed by inadequate park budgets, varying public and political support and development pressures beyond park boundaries [33], [47].

The mandate for PAs in some East African countries (e.g. Tanzania, Kenya) has changed dramatically in past decades from prioritizing areas for large mammal conservation, to protecting biodiversity in the 1980s and 1990s, and more recently to alleviate poverty and support livelihoods, both key objectives in Tanzania’s Participatory Forest Management (PFM) policy [48]. The proliferation of PFM in Tanzania, legally underpinned by the 1998 National Forest Policy and the 2002 Forest Act, increases decision-making powers of villages on the use of forest resources and empowers them to declare, own and manage their forests [49].

As demonstrated in the case of Mukogodo FR as well as other Effective Nature Reserves and Forest Reserves (S2), PFM can significantly improve forest conservation outcomes. However, human pressure is also shown to lead to deforestation encroachment to within park boundaries, especially among less well-protected parks, and some countries (i.e. Uganda, Burundi, Zambia) fare worse than others. Forest conservation success varies considerably within and between protection categories and within and between countries. And, National Parks perform better than other protection categories in terms of protecting forests.

There is evidence for ‘leakage’ around protected areas (Tables 3, 4, S4), which needs to be analyzed in more detail using higher-spatial resolution satellite images going back to the 1970s, combined with targeted fieldwork for parks established after 1980. Our approach could easily be used to identify ‘sorrow parks’ (i.e. parks that show high deforestation within boundaries and/or within their immediate buffer zones) that should be prioritized for management adjustment. We acknowledge that each of these Ineffective parks will require a slightly different, individual management approach. But, comparing management and surrounding matrix traits of these parks to those of Effective parks in future analyses can be used to reveal the presence of general mechanisms controlling park effectiveness across geographical scales.

Pragmatic (human-centered) approaches to forest conservation emphasize the importance of conservation in human-modified landscapes [50]. Managing the human-forest interaction across the landscape is politically more feasible than excluding communities from forest resources. However, it may not provide wanted outcomes in regions of rapid human population growth [2], [17], [51], [52] and in regions where the relationship between ‘poor’ and ‘forest use’ is transformed by further interventions (i.e. impact of logging or mining companies, influx of newcomers interested in land for crops and livestock) [53].

While there is an ongoing need to assess ‘fitness-for purpose’ across the PA network, increasing and enforcing existing protection status is likely to remain best-practice on the ground for a while to come. This forest-centered approach to slow or reverse forest loss in East Africa could be combined with (i) establishing and managing multiple-use buffers (e.g. foster tree planting for firewood extraction) around existing PAs [3], [27], (ii) reconnecting local communities to their forests [54], (iii) establishing payments for ecosystem services schemes managed to provide local benefits, and (iv) to identify motivating factors driving resource extraction locally to subsequently provide sustainable and feasible alternatives for services provided by forests [3], [55].

Materials and Methods

Study Area

Our study area in Eastern Africa (3,882,887 km2; bounded by N6, S-15, W27.5, E42.5), covers Uganda, Kenya, Tanzania, Rwanda and Burundi and extends to partially cover neighboring countries including Somalia, Ethiopia, South Sudan, Congo, Zambia, Malawi and Mozambique.

Classification of Pas

We defined five land management categories of decreasing protection status: National Parks > Nature Reserves > Forest Reserves > Game Parks > Unprotected land [20]. Boundaries of PAs were derived from the World Database on Protected Areas [56]. Parks were reclassified into National Parks, Nature Reserves, Game Parks (includes Game Reserves, Game Management Areas, Game Controlled Areas, Game Sanctuary, Hunting Reserve and Controlled Hunting Area), and Forest Reserves (includes Village, District and Nationally Managed Forest Reserves). Buffer zones were created around individual PAs after merging parks within protection categories that were located ≤10 km from one another.

Extraction of Forest Distribution Data

The distribution of evergreen broadleaved forests in the study area was extracted from MODIS Type 1 land cover grids (discussed in Pfeifer et al. [57]; downloaded from, which provide information on vegetation cover at 500 m spatial resolution. The MODIS algorithm calculates the probability of class membership (PM) for each land cover pixel in each year (via boosting using a base learning algorithm and high spatial resolution Landsat TM imagery). PM is high for evergreen forests (2001 and 2009: PMmedian: 96%, PMmajority: 100%). Background forest loss was calculated as forest cover change between 2001 and 2009 relative to the amount of forest area present in 2001.

Analysis of Forest Trends

Forest cover trends for the various spatial subsets (PA, PA buffers, national, regional) were computed from maps of evergreen forests between 2001 and 2009. We computed background forest loss for East Africa and separately for each country. We assessed forest trends within individual PAs and their buffers for each protection category, and compared forest trends within park buffers to overall background forest change.

Environmental Variables in Deforestation Models

Total fire frequency was computed from MODIS active fire hotspot data between 2001 and 2009. We concentrated on fire locations with a reported accuracy ≥50%, accepting that this may result in underestimating fire frequencies. Fire data were converted to 1 km grids, indicating whether a pixel was burned or not in a given year. Fire frequency grids were computed as fire sums between 2001 and 2009 (e.g. fire frequency per pixel could range from 0 to 9). MODIS Burned Area maps (MCD45A1) between 2001 and 2009 were downloaded from (discussed in Pfeifer et al. [57]). These grids were transformed into annual presence/absence maps indicating whether a pixel got burned or not. The derived maps were subsequently used to compute grids of pixel-specific annual burning probabilities. Road and town distribution data were derived from the Africover project [58]. Spatial analyses were carried out using ArcGIS v9.3 software. Statistical models and graphics were computed using R v2.11.1 statistical software environment.

Logit Models of Deforestation

We modelled forest change in East Africa as a function of accessibility (distance to roads, distance to towns, protection status and slope), vegetation burning (annual burning probabilities and pixel-specific fire frequency at annual resolution), and human population density (persons per 500 m cell) derived from density maps at 1 ha spatial resolution [59]. The binary response variable in the logit models of forest change was derived by selecting 5000 points randomly from pixels with forest loss and 5000 points from pixels maintaining their forests. The points were placed more than 1 km distance from one another to minimize spatial autocorrelation [60], [61].

Supporting Information

Table S1.

Comparison of forest loss (in km2 and %) within and outside PAs in two time periods (P1∶2001–2004, P2∶2004–2009).


Table S2.

Information on management of eight randomly selected effective protected areas in East Africa.


Table S3.

Information on nine randomly chosen Ineffective protected areas in East Africa.


Table S4.

Comparison of forest loss within PA and within their buffer zones to country-specific background forest loss (BFL).



We are grateful to Minnie Wong (University of Maryland) for provision of MODIS fire hotspots data.

Author Contributions

Conceived and designed the experiments: MP NDB PJP RM. Performed the experiments: MP. Analyzed the data: MP PJP RDS. Wrote the paper: MP NDB PJP SW RM.


  1. 1. Masozera MK, Alavalapati JRR (2004) Forest Dependency and Its Implications for Protected Areas Management: A Case Study from the Nyungwe Forest Reserve, Rwanda. Scand J For Res 19: 85–92.
  2. 2. Lung T, Schaab G (2010) A comparative assessment of land cover dynamics of three protected areas in tropical Eastern Africa. Environ Monit Assess 161: 531–548.
  3. 3. Fisher B, Lewis SL, Burgess ND, Malimbwi RE, Munishi PK, et al. (2011) Implementation and opportunity costs of reducing deforestation and forest degradation in Tanzania. Nature Clim Change 1: 161–164.
  4. 4. Myers N, Mittermeier RA, Mittermeier CG, da Fonseca GAB, Kent J (2000) Biodiversity hotspots for conservation priorities. Nature 403: 853–858.
  5. 5. Burgess ND, Butynski TM, Cordeiro NJ, Doggart NH, Fjeldså J, et al. (2007) The biological importance of the Eastern Arc Mountains of Tanzania and Kenya. Biol Conserv 134: 209–231.
  6. 6. Lewis SL, Lopez-Gonzalez G, Sonke B, Affum-Baffoe K, Baker TR, et al. (2009) Increasing carbon storage in intact African tropical forests. Nature 457: 1003–1007.
  7. 7. Hall J, Burgess N, Lovett J, Mbilinyi B, Gereau RE (2009) Conservation implications of deforestation across an elevational gradient in the Eastern Arc Mountains, Tanzania. Biol Conserv 142: 2510–2521.
  8. 8. Rutagarama E, Martin A (2006) Partnerships for protected area conservation in Rwanda. Geogr J 172: 291–305.
  9. 9. Balmford A, Moore JL, Brooks T, Burgess N, Hansen LA, et al. (2001) Conservation Conflicts Across Africa. Science 291: 2616–2619.
  10. 10. UN (2011) World Population Prospects: the 2010 revision. February 2011.
  11. 11. Tabor K, Burgess ND, Mbilinyi BP, Kashaigili JJ, Steininger MK (2010) Forest and woodland cover and change in coastal Tanzania and Kenya, 1990 to 2000. J East Afr Natural History 99: 19–45.
  12. 12. Timberlake J, Goyder D, Crawford F, Burrows J, Clarke GP, et al. (2011) Coastal dry forests in northern Mozambique. Plant Ecol Evol 144: 126–137.
  13. 13. Ahrends A, Burgess ND, Milledge SAH, Bulling MT, Fisher B, et al. (2010) Predictable waves of sequential forest degradation and biodiversity loss in spreading from an African city. Proc Natl Acad Sci USA 107: 14556–14561.
  14. 14. Cotula L, Vermeulen S, Leonard R, Keeley J (2009) Land grab or development opportunity. Agricultural investment and international land deals in Africa. IIED, FAO & IFAD, London/Rom. ISBN. pp. 978–1-84369-741-1.
  15. 15. Milledge SAH, Gelvas IK, Ahrends A (2007) Forestry, Governance and National Development: Lessons Learned from a Logging Boom in Southern Tanzania. TRAFFIC East/Southern Africa, Tanzania Development Partners Group, Tanzania Ministry of Natural Resources and Tourism, Dar es Salaam. Accessed April 2012.
  16. 16. Bruner AG, Gullison RE, Rice RE, da Fonseca GAB (2001) Effectiveness of Parks in Protecting Tropical Biodiversity. Science 291: 125–128.
  17. 17. Naughton-Treves L, Alix-Garcia J, Chapman CA (2011) Lessons about parks and poverty from a decade of forest loss and economic growth around Kibale National Park, Uganda. Proc Natl Acad Sci USA 108: 13919–13924.
  18. 18. Nelson A, Chomitz KM (2011) Effectiveness of Strict vs. Multiple Use Protected Areas in Reducing Tropical Forest Fires: A Global Analysis Using Matching Methods. PLoS ONE 6: e22722.
  19. 19. DeFries R, Hansen A, Newton AC, Hansen MC (2005) Increasing Isolation of Protected Areas in Tropical Forests over the past Twenty Years. Ecol Appl 15: 19–26.
  20. 20. Burgess N, Küper W, Mutke J, Brown J, Westaway S, et al. (2005) Major gaps in the distribution of protected areas for threatened and narrow range Afrotropical plants. Biodivers Conserv 14: 1877–1894.
  21. 21. Caro T, Gardner TA, Stoner C, Fitzherbert E, Davenport TRB (2009) Assessing the effectiveness of protected areas: paradoxes call for pluralism in evaluating conservation performance. Divers Distrib 15: 178–182.
  22. 22. Enawgaw AC, Deksios AD, Timer AG (2011) Existing Challenges: Plantation Development versus Wildlife Conservation in the Omo-Tamo-Mago Complex. Accessed April 2012.
  23. 23. Gardner TA, Caro T, Fitzherbert EB, Banda T, Lalbhai P (2007) Conservation Value of Multiple-Use Areas in East Africa. Conserv Biol 21: 1516–1525.
  24. 24. Cropper M, Puri J, Griffiths C (2001) Predicting the location of deforestation: The role of roads and protected areas in North Thailand. Land Econ 77: 172–186.
  25. 25. Freitas SR, Hawbaker TJ, Metzger JP (2009) Effects of roads, topography, and land use on forest cover dynamics in the Brazilian Atlantic Forest. For Ecol Manage 259: 410–417.
  26. 26. Lovejoy TE (2006) Protected Areas: a prism for a changing world. Trends Ecol Evol 21: 329–333.
  27. 27. DeFries R, Rosenzweig C (2010) Towards a whole-landscape approach for sustainable land use in the tropics. Proc Natl Acad Sci USA 107: 19627–19632.
  28. 28. Naidoo R, Balmford A, Costanza R, Fisher B, Green RE, et al. (2006) Global mapping of ecosystem services and conservation priorities. Proc Natl Acad Sci USA 105: 9495–9500.
  29. 29. Davis A (2011) Ha! What is the Benefit of Living Next to the Park? Factors Limiting In-migration Next to Tarangire National Park, Tanzania. Conservation and Society 9: 25–34.
  30. 30. Ewers RM, Rodrigues ASL (2009) Estimates of reserve effectiveness are confounded by leakage. Trends Ecol Evol 23: 113–116.
  31. 31. Fearnside PM (2009) Carbon benefits from Amazonian forest reserves: leakage accounting and the value of time. Mitig Adapt Strateg Glob Change 14: 557–567.
  32. 32. Joppa LN, Loarie SR, Pimm SL (2008) On the protection of “protected areas” Proc Natl Acad Sci USA 105: 6673–6678.
  33. 33. Leverington F, Costa KL, Pavese H, Lisle A, Hockings M (2010) A Global Analysis of Protected Area Management Effectiveness. Environmental Management 46: 685–698.
  34. 34. Indigenous Information Network INN (2004) Accessed April 2012.
  35. 35. Naidoo R, Ricketts TH (2008) Mapping the Economic Costs and Benefits of Conservation. PLOS Biology 4: e360. DOI:
  36. 36. Wass P (1995) Kenya’s Indigeneous Forests: Status, Management, and Conservation. IUCN.
  37. 37. Oliveira PJC, Asner GP, Knapp DE, Almeyda A, Galván-Gildemeister R, et al. (2007) Land-Use Allocation Protects the Peruvian Amazon. Science 317: 1233–1236.
  38. 38. Wittemeyer G, Elsen P, Bean WT, Burton ACO, Brashares JS (2008) Accelerated human population growth at protected area edges. Science 321: 123–126.
  39. 39. Gaveau DLA, Wandono H, Setiabudi F (2006) Three decades of deforestation in southwest Sumatra: Have protected areas halted forest loss and logging, and promoted re-growth? Biol Conserv 134: 495–504.
  40. 40. Ferraro PJ, Hanauer MM, Sims KRE (2009) Conditions associated with protected area success in conservation and poverty reduction. Proc Natl Acad Sci USA 108: 13913–13918.
  41. 41. Joppa LN, Pfaff A (2009) High and Far: Biases in the Location of Protected Areas. PLOS One 4: e8273.
  42. 42. Ewers RM, Kliskey AD, Walker S, Rutledge D, Harding JS, et al. (2006) Past and future trajectories of forest loss in New Zealand. Biolog Conserv 133: 312–325.
  43. 43. Klein Goldewijk K (2001) Estimating global land use change over the past 300 years: The HYDE database. Global Biogeochem Cy 15: 417–433.
  44. 44. FAO (2005) State of the world’s forests 2005. Rome: FAO.
  45. 45. Baranga D, Chapman CA, Kasenene JM (2009) The structure and status of forest fragments outside protected areas in central Uganda. Afr J Ecol 47: 664–669.
  46. 46. Swetnam RD, Fisher B, Mbilinyi BP, Munishi PKT, Willcock S, et al. (2011) Mapping socio-economic scenarios of land cover change: A GIS method to enable ecosystem service modelling. J Environ Manag 92: 563–574.
  47. 47. Lilieholm RJ, Weatherly P (2009) Kibale Wild Forest Coffee: Challenges to Market-based Conservation in Africa. Conserv Biol 24: 924–930.
  48. 48. Blomley T, Iddi S (2009) Participatory Forest Management in Tanzania: 1993–2009. Lessons learnt and experiences to date. Ministry of Natural Resources and Tourism Forestry and Beekeeping Division, Tanzania.
  49. 49. Harris D, Domingo P, Sianga C, Chengulla E, Kavishe CB (2011) The Political Economy of Social Accountability in Tanzania: Anti-logging Advocacy and the Mama Misitu Campaign. Overseas Development Institute.
  50. 50. Gardner TA (2009) Prospects for tropical biodiversity in a human-modified world. Ecol Lett 12: 561–582.
  51. 51. Andrew G, Masozera M (2010) Payment for ecosystem services and poverty reduction in Rwanda. J Sustain Develop Afr 12: 122–139.
  52. 52. Tumusiime DM, Vedeld P, Gombya-Ssembajjwe W (2011) Breaking the law? Illegal livelihoods from a Protected Area in Uganda. Forest Policy Econ 13: 273–283.
  53. 53. Wunder S (2001) Poverty Alleviation and Tropical Forests – What Scope for Synergies? World Development 29: 1817–1833.
  54. 54. Pfund JL, Watts JD, Boissière M, Boucard A, Bullock RM, et al. (2011) Understanding and Integrating Local Perceptions of Trees and Forests into Incentives for Sustainable Landscape Management. Environ Manage 48: 334–349.
  55. 55. Tallis H, Pagiola S, Zhan W, Shaikh S, Nelson E, et al. (2011) in NATURAL CAPITAL. Theory and Practice of Mapping Ecosystem Services, eds. Kareiva P, Talllis H, Ricketts TH, Daily GC, Polasky S (Oxford University Press, Oxford), 278–295.
  56. 56. IUCN & UNEP (2009) The World Database on Protected Areas (WDPA) UNEP-WCMC. Cambridge, UK. Version 2010. Accessed February 2011.
  57. 57. Pfeifer M, Disney M, Quaife T, Marchant R (2011) Terrestrial ecosystems from space: a review of earth observation products for macroecology applications. Global Ecol Biogeogr. Early View Online. DOI:
  58. 58. FAO (2011) Africover. Accessed July 2011.
  59. 59. Tatem AJ, Noor AM, von Hagen C, Gregorio AD, Hay SI (2007) High Resolution Population Maps for Low Income Nations: Combining Land Cover and Census in East Africa. PLoS ONE 2: e1298.
  60. 60. Linkie M, Smith RJ, Leader-Williams N (2004) Mapping and predicting deforestation patterns in the lowlands of Sumatra. Biodivers Conserv 13: 1809–1818.
  61. 61. Koenig WD (1999) Spatial autocorrelation of ecological phenomena. Trends Ecol Evol 14: 22–26.