Land-Cover Change Threatens Tropical Forests and Biodiversity in the Littoral Region, Cameroon M AHMOUD I

Land-cover change threatens tropical forests and biodiversity in the Littoral Region, Cameroon M AHMOUD I. MAHMOUD,MASON J. CAMPBELL,SEAN S LOAN M OHAMMED A LAMGIR and W ILLIAM F. LAURANCE

Abstract Tropical forest regions in equatorial Africa are et al., ; WWF, ). They also provide numerous and threatened with degradation, deforestation and biodiversity valuable environmental services, including carbon storage, loss as a result of land-cover change. We investigated his- protection of threatened ecosystems, hydrological func- torical land-cover dynamics in unprotected forested areas tioning (Abernethy et al., ), and medicinal products of the Littoral Region in south-western Cameroon during (Colfer, ). African forests also provide significant socio- –, to detect changes that may influence this impor- economic resources, including plant and animal products, tant biodiversity and wildlife area. Processed Landsat im- food, medicine, products of cultural value, and building agery was used to map and monitor changes in land use and construction materials. and land cover. From  to  the area of high-value forest Tropical African forests are threatened by a wide array landscapes decreased by c. , ha, and increasing forest of anthropogenic activities that imperil natural ecosystems fragmentation caused a decline of c. %inthelargestpatch and biodiversity. For instance, the few previous studies index. Conversely, disturbed vegetation, cleared areas and conducted in the threatened forests of the Cross–Sanaga urban areas all expanded in extent, by %(c., ha), River region (Fa et al., ), the Gulf of Guinea biodiver- .% (c. , ha) and .% (c. , ha), respectively. sity hotspot (Oates et al., ), and the greater Congo The greatest increase was in the area converted to oil palm Basin (Laurance et al., ) identified deforestation as a plantations (c. , ha), followed by logging and land major factor driving both forest loss and degradation clearing (c. , ha), all of which were the major factors (Hosonuma et al., ; Ordway et al., ; Aleman et al., driving deforestation in the study area. Our findings ). Deforestation, along with other smallholder activities highlight the increasing threats facing the wider Littoral in African tropical forests, is also known to damage impor- Region, which includes Mount Nlonako and Ebo Forest, tant wildlife habitats, leading to biodiversity loss and the both of which are critical areas for regional conservation destruction of forest-based livelihoods. However, the extent and the latter a proposed National Park and the only siz- to which small vs large landholder activities influence land- able area of intact forest in the region. Intact forest in the cover change and conservation in these forests remains Littoral Region, and in particular at Ebo, merits urgent unclear. Known drivers of forested land-cover conversion protection. include industrial logging (Laporte et al., ), agricultural expansion (van Soesbergen et al., ), settlement expan- Keywords Biodiversity loss, Cameroon, deforestation, Ebo sion (Mahmoud et al., ) and road infrastructure expan- Forest, equatorial Africa, logging, oil palm, roads, wildlife sion (Alamgir et al., ; Laurance et al., ; Mahmoud Supplementary material for this article is available at et al., ). These processes can also lead to forest fragmen- https://doi.org/./S tation and degradation, resulting in depauperate forests with significantly reduced biodiversity compared to intact forests (Fahrig, ; Haddad et al., ; Laurance et al., ). Introduction The rate and scale of land-cover change currently occurring in equatorial Africa is unprecedented and unsustain- ’ frica s tropical forests are the most extensive after those able (Tchuenté et al., ). However, there is a paucity of   Aof the Amazon (Cincotta et al., ; Sosef et al., ), studies examining human-driven land-cover change and    covering . million km , with the majority concentrated in threats to conservation at fine spatial scales in unprotect-   the greater Congo Basin. These forests harbour c. , ed yet still highly biodiverse regions of equatorial Africa. .   species of vertebrates and , plant species (Orme Regional studies examining land-cover change have often been at a coarse scale, and many of the findings are based on inferences (Singh et al., ; Thompson et al., ). MAHMOUD I. MAHMOUD (Corresponding author, orcid.org/0000-0002-0432- 0429), MASON J. CAMPBELL,SEAN SLOAN,MOHAMMED ALAMGIR and WILLIAM Consequently, many of these regional studies overlook F. LAURANCE Center for Tropical Environmental and Sustainability Science, challenges associated with localized land-use and land-cover College of Science and Engineering, James Cook University, Cairns, Queensland 4878, Australia. E-mail [email protected] change processes such as selective logging, small-scale log- Received  May . Revision requested  June . ging, land clearing and the establishment of smallholder Accepted  June . First published online  August . oil palm plantations, particularly in areas of high biodiversity

Oryx, 2020, 54(6), 882–891 © 2019 Fauna & Flora International doi:10.1017/S0030605318000881 Downloaded from https://www.cambridge.org/core. IP address: 170.106.35.76, on 25 Sep 2021 at 17:22:11, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0030605318000881 Threat of land-cover change in Cameroon 883

containing disturbance-sensitive wildlife. Understanding tropical forest conservation in the Littoral Region or in equa- underlying land-use and land-cover change processes and torial Africa as a whole. patterns retrospectively at a detailed scale is important for The history of extensive forest loss in the Littoral Region conservation managers, as it can facilitate the development is poorly documented, and its drivers are poorly under- of sustainable socio-ecological management solutions for stood. Here, we provide spatially explicit land-cover change addressing current and imminent environmental challenges analysis and information about forest fragmentation. We in these forests. also discuss the potential influence of these human actions Southern Cameroon is experiencing rapid forest loss on the long-term sustainability of natural forest conserva- (Cheek et al., ; Ingram et al., ). The Littoral tion. To achieve these aims, we addressed the following Region, in south-western Cameroon, has the largest area questions: () How have land surface dynamics changed in of continuous tropical forest, within the Nkam and Sanga- the Littoral Region from  to ?() What are the Maritime areas, bordering the Ebo and Dibamba rivers drivers of land-cover change? () How may land-cover north of Cameroon’s industrial capital Douala. This area in- change imperil future conservation of critical biodiversity cludes the proposed Ebo National Park, which is among the and wildlife habitats? few remaining intact forest landscapes in Africa north of the Sanga River (Potapov et al., ). However, Ebo Forest remains legislatively unprotected despite efforts to designate Study area the area as a National Park. Ebo Forest provides important  We examined a , km study area encompassing Ebo habitat for the Critically Endangered Preuss’s red colobus  Forest (c. , km ) in the Littoral Region of Cameroon Piliocolobus preussi and gorilla subspecies Gorilla gorilla (Fig. ). Mean annual precipitation in the area is ,– diehli and Gorilla gorilla gorilla, the Endangered Nigeria– , mm in the northern and interior regions, and Cameroon chimpanzee Pan troglodytes ellioti and drill ,–, mm in the southern and coastal zones. Mean Mandrillus leucophaeus, and the Vulnerable African forest annual temperature is – °C (Molua, ). The land- elephant Loxodonta cyclotis (Morgan et al., , who also scape is relatively flat, although there are some areas of provide further details of the ecology and geography of undulating terrain, and a few mountains rising to , m the region). Given the unprotected status of Ebo Forest, it inside the boundaries of the proposed Ebo National Park remains particularly vulnerable to anthropogenic pressures (the proposed boundaries are currently under review). such as deforestation and hunting, especially considering Human activities in the Littoral Region, such as unsus- the unprecedented scale and pace of deforestation in the tainable levels of logging, selective logging, land clearing, surrounding areas. These threats remain unchallenged be- farming and hunting, place the regions’ biodiversity (in- cause of delays in securing legal protection, thus threatening cluding that in Ebo Forest) under severe threat (Morgan the long-term integrity of Ebo Forest. et al., ; Whytock et al., ). Large-scale bushmeat The Littoral Region of Cameroon also encompasses wholesalers travel to local villages near Ebo Forest by Mount Nlonako, along the north-western axis of the region. timber trucks, motorcycles and taxis to buy bushmeat Mount Nlonako has a rich diversity of amphibians, birds from hunters to supply the region’s rapidly increasing and reptiles but is subject to human incursions. Both population (from . million in  to nearly . million Mount Nlonako and Ebo Forest are located in relatively in ). In addition, the humid equatorial climate of the close proximity to Douala (Mount Nlonako c.  km to Littoral Region makes it a suitable location for oil palm the north-west of the city, and Ebo Forest c.  km to the plantations. Palm oil production is expanding rapidly, north-east). With its proximity to wildlife habitat, Douala with small- to medium-sized oil palm plantations replacing supports thriving markets for forest products and wildlife, lands previously used for crops such as cocoa and coffee many of which supply bushmeat. The high demand for for- (Feintrenie, ). est and wildlife products is a strong driver of the ongoing depletion of natural forest cover and biodiversity through logging and hunting in the unprotected forests of the Methods Littoral Region and greater Cameroon (Morgan et al., ). It is therefore important to ascertain how human land-use Study overview actions affect unprotected high-value forests at the landscape level across the wider Littoral Region, using com- We assessed land-cover change in the Littoral Region based bined remote sensing and geographical information system on the available historical remote-sensing images from (GIS)-based analysis (Wiens et al., ; Willis, ), in the Landsat- (multispectral scanner,  m resolution) for  interest of environmental sustainability (Cetin, ). There December  and Landsat- (operational land imager, have been no previous detailed studies of anthropogenic  m resolution) for  December . Land-cover data land-use/land-cover changes and their implications for were integrated using GIS overlays of Google Earth and

FIG. 1 Locations of Mount Nlonako, Ebo Forest and various human activities within the study area in the Littoral Region of Cameroon. Data for intact forest were derived from the Intact Forest Landscapes (–) datasets, and data for tree cover loss from the Hansen/ UND/Google/USGS and NASA global dataset.

the Planet Labs Inc. application interface (Planet Team, Spatio-temporal analysis and mapping fine-scale ) in conjunction with human land-use interpretations drivers of deforestation (roads, logging, land clearing and oil palm plantations). These data were then used to assess land-use/land-cover We used a three-stage analysis to identify retrospectively the change of natural forest in the study region. Cloud-cover human disturbances (deforestation, land clearing and oil and shadow effects masked parts of the study area, palm plantation development) that threatened forest cover preventing land-cover monitoring of the entire Littoral and biodiversity in the study area of the Littoral Region.   Region through optical remote sensing. Firstly, based on the land-cover maps for and we investigated hindcast land-cover change to derive and quan- tify land-cover type change. The change detection approach Categorization and accuracy assessment was used to compute uniform temporal conversion from land cover in an initial state to a subsequent land use, and We categorized a cloud-free subset of Landsat images from depict irregular variation verified to be human disturbance   and  (covering , km ) that served as the effec- and alteration of natural forest. tive study area for this research (Fig. ). We recognized six Secondly, landscape patterns were quantified using the major land-cover categories: water body, urban area, cleared largest patch index computed from the land-cover maps land, disturbed vegetation, natural forest and cloud cover. of  and . The index provides quantitative informa- Our land-cover categorization was conducted using the tion at the patch, class and landscape levels (Pardini, ). random forest algorithm, which is a machine-learning We focused on the largest patch index as a useful indicator supervised image classifier (Breiman, ). Random homo- of fragmentation or aggregation of the land-cover category geneous training and testing samples were collected manually of interest (in this case natural forest cover). The index is ex- from the  and  images for categorization. pressed as a percentage of land-cover patch. As the value of We optimized the accuracy and area of the  and the largest patch approaches , the size of the corresponding  land-cover maps using the error-adjusted approach patch type decreases. The largest patch index is equal to (Olofsson et al., ). The three steps involved in the % when the entire landscape is made up of a single approach to determine the map accuracies were sampling, patch that corresponds to the patch type (McGarigal & response design and accuracy analysis. As a result, the Marks, ). Hypothetically, a decline in the index from a producer’s accuracy, user’s accuracy, overall accuracy and maximal score of % (equivalent to ) for an intact land- confidence intervals were generated for the final landscape implies that larger fragments are being re-fragmented, cover maps produced. reaching  when all fragments are ultimately converted to a

different land-cover type. At the category level, the largest Drivers of anthropogenic environmental change at the patch index (LPI) is computed as: landscape level n ( ) max j=1 aij We identified four major factors driving land-cover change LPI = A in the Littoral Region: the expansion of paved and unpaved

where aij is the area of patch ij, subscript i indicates the score major and access roads and rail infrastructure, which facili- for an individual patch within a particular group j, A is the tated access to remote forest areas; logging; land clearing; area of the landscape and n is time. and agricultural expansion, especially for palm oil produc- Thirdly, we manually digitized line and polygon features tion (Fig. ). According to our interpretation and digiti- that corresponded to major roads, access roads, logging zation of fine-scale land-use maps, the area of oil palm roads, logging tracks, cleared land and oil palm plantations plantation in the study area is . , ha and the area of between  and , using the historical view of the cleared land is at least , ha (Fig. , Supplementary Fig. a). Google Earth image interface (Supplementary Fig. ). The identified unpaved access roads were . , km in Land-cover maps and relevant land-use layers were inte- length, logging roads and deforestation tracks spanned grated. This approach is similar to the concept of data .  km, and the total length of paved major roads fusion, used to combine multi-source data (Pohl & Van was c.  km (Supplementary Fig. b). Panels A and B Genderen, ). of Fig.  are enlarged in Supplementary Fig. , showing recently logged area, logging roads, cleared land, and oil palm plantations. Results Discussion Land-use and land-cover mapping and accuracy analysis Drivers of land-cover change We identified significant land-cover changes between  and  in the Littoral Region (Fig. ). The confidence of The unregulated expansion of human land use and land- our error-adjusted estimate fell within %, although it cover change in the Littoral Region of Central Africa poses was slightly different from that of the mapped areas using irreversible threats to biodiversity and natural forest the crude land-cover classification maps of  and  ecosystems. Our results show that land-cover change since (Table , Supplementary Table ). The accuracy of our land-  has substantially modified the forest landscapes of the cover maps based on the error-adjusted matrix was % Littoral Region, including one of the remaining intact forest for the  data and % for the  data, the producer’s landscapes, north of Sanga River (Figs ,  & Supplementary accuracy was –%andtheuser’saccuracywas–% Fig. ). We found significant and expanding land-cover (Supplementary Table ). change within  km of the periphery of Ebo Forest and Mount Nlonako, two important conservation areas in the region (Fig. ). Our findings are supported by earlier studies   Spatio-temporal change in fragmentation and landscape (Fa et al., ; Potapov et al., ), which used coarser datasets but also reported significant tree cover loss, deforesta- There was a significant (%) increase in the fragmentation tion and biodiversity degradation in the region. Using the of natural forest in the study area (Supplementary Fig. a). fine-scale data in this study, deforestation was found to be The largest patch index value for natural forest decreased in driven by multiple human land uses, namely oil palm plan- mean forest patch size, from . %in to c. %in. tation, logging and land clearing, all of which were facili- The values for disturbed vegetation increased from .%in tated by expansion of the road network that enabled  to .%in. Similarly, the largest patch index value access to the forest. These findings are similar to elsewhere for cleared land increased from .%in to .%in, in the tropics where anthropogenic deforestation has been and for urban area from .%in to .%in identified as an initial driver of forest degradation leading (Supplementary Fig. b). We found considerable net change to progressive land-use changes and ultimately widespread in the gains or losses of the individual land-use/land-cover land-cover conversion (Marsik et al., ; Renó et al., ). change types (Fig. ) between  and , as determined The expansion of oil palm plantations appears to be the by our spatio-temporal change detection analysis. There was major factor driving deforestation in the studied landscapes a significant loss of c. , ha of high-value natural forest of the southern Littoral Region. There is evidence that, in cover in the study area from  to , and an increase general, logging preceded deforestation and is a major facili- of c. , ha of disturbed vegetation (Fig. ). Similarly, tator of deforestation (Fig. , Supplementary Fig. ) and the urban areas, cleared land, and water all increased in area, initial step in forest degradation. Logging clears the way by ,, , and , ha, respectively (Fig. ). for the land to be used for oil palm plantations or cleared

FIG. 2 Land-cover maps of the Littoral Region of Cameroon from (a)  and (b) . Numbers in (a) depict areas with apparent land-cover changes, labelled in (b).

for other human use (usually agricultural; Fig. , Supple- typically of , . ha, interspersed within subsistence plots, mentary Fig. ). Land clearing in most instances also pre- degraded forests and fallows; % are within  km of ceded oil palm plantations (Fig. ), with plantations again roads. However, because of limitations of the time-series being the final land-cover type in the sequence of land-use data for the satellite imagery, it is impossible to confirm conversion (Fig. , Supplementary Figs.  & ). In  oil whether oil palm expansion was the direct cause of the palm plantations comprised at least half the area of dis- increase in disturbed vegetation since  or simply turbed forests in all parts of the Littoral Region. The dynam- the latest land use at these locations. Nevertheless, the fact ics of oil palm expansion and loss of tropical forest in this that smallholder oil palm complexes occur predominantly region appear to be distinct from those in other tropical re- within remnants of native vegetation suggests that oil palm gions, such as Indonesia (Pirker et al., ; Austin et al., ) may be a direct driver of natural forest conversion. Our and Colombia (Garcia-Ulloa et al., ), where oil palm finding is supported by previous studies in the region, expansion often occurs at an industrial scale. which identified oil palm plantations as one of the major dri- In the Littoral Region the majority of oil palm cultivation vers of deforestation (Tamungang et al., ; Mitchell et al., (c. %) occurs within smallholder agricultural complexes. ). The conversion of natural forests to landscapes domi- These smallholder complexes are visually identifiable by nated by oil palm may also lead to a cascade of ongoing nega- mosaics of small, irregularly shaped oil palm stands, tive conservation outcomes in the region, as evidenced

  TABLE 1 Proportions of various land-cover types in the effective study area (,. km cloud-free of the , km total study area) in the Littoral Region of Cameroon in  and .

1975 2017 Land use/land Mapped area Estimated area Confidence interval Mapped area Estimated area Confidence interval cover (%) (%) (%) (%) (%) (%) Water 3.0 0.5 −0.1–1.6 2.7 3.0 −0.1–1.6 Urban area 0.1 0.2 −0.1–0.01 5.0 5.0 −0.1–0.1 Cleared land 0.2 0.5 −0.2–0.2 2.2 3.0 −0.2–0.2 Disturbed 31.0 16.0 −2.0–2.0 37.0 38.0 −0.1–0.1 vegetation Natural forest 63.0 72.0 −0.1–0.3 53.1 51.0 −0.1–0.1

within  km of a forest edge, near roads. These observations combined suggest that forest degradation and deforestation are facilitated by the increased access to intact forests that the roads provide (Fig. ). Moreover, roads have facilitated the expansion of human habitations, which are now present within  km of Ebo Forest (Figs. b, , & Supplementary Fig. ); a process increasingly threatening biodiversity and wildlife habitats and their conservation (Brandão, Jr & Souza, Jr, ; Pfaff et al., ).

Forest fragmentation FIG. 3 Change in land cover (water, urban area, cleared land,  disturbed vegetation, natural forest) in the , km study   There was a significant increase in forest fragmentation area in the Littoral Region of Cameroon (Fig. ) between   and . between and in the Littoral Region. For instance, there was a pronounced decline in inter-patch habitat connectivity in the forested area, as indicated by the significant decline in the extent of natural forest (Fig. ). Connectivity in South-east Asian landscapes where oil palm plantations is considered to be essential for conserving biodiversity have had negative effects on the abundance and occurrence at the landscape scale (Hodgson et al., ) and facilitating of a wide range of taxa, including birds, invertebrates and the movement of fauna (Vogt et al., ), and therefore the mammals (Fitzherbert et al., ; Yue et al., ). Oil isolation of many forest patches in the study area threatens palm expansion in the south of the Littoral Region landscape biodiversity (Fahrig, ) and the genetic integrity of the is a major driver of current and ongoing regional remnant species (Young et al., ). If the current trend deforestation, although Ebo Forest is protected to some in logging and farming activities is sustained there is likely degree by its steep and hilly terrain, which makes it less to be ongoing isolation of natural forest in patches, which accessible. may lead to further biodiversity loss. This is because larger Logging and land clearing are two other major drivers of patches of natural forest provide better protection for a deforestation we observed in the north-west of the Littoral variety of habitats, and withstand anthropogenic threats Region. The proximity of a government-sanctioned active and degradation better than smaller, fragmented forests -year logging concession (FMU--), adjacent to (Laurance et al., ). The current trajectory of human Ebo Forest in the north-east of the Littoral Region land use in the region suggests that significant conservation (Fig. ), is a potential threat to the biodiversity of Ebo areas, such as Ebo Forest, are increasingly at risk from frag- Forest and is a concern for unprotected forests in the mentation and isolation of forest patches. A comparison of region, which are also threatened by the extensive areas of the  and  land-cover maps, for instance, revealed sanctioned logging and mining concessions (Fig. ). that substantial fragmentation occurred in the forest land- The infiltration of roads into the studied forests appears scapes of the Littoral Region as a result of deforestation to have contributed substantially to the progressive defor- (Fig. ). The observed fragmentation and deforestation estation in the region; for example, roads were found in all began in degraded areas and expanded gradually into nat- human-modified land uses, including deforested areas. ural forest, resulting in an increase in the area of disturbed Additionally, most of the identified deforestation occurred vegetation (Fig. ).

FIG. 4 Spatially explicit drivers of deforestation overlain on a  land-cover map (as base  layer) in the , km study area encompassing Ebo Forest and Mount Nlonako, Cameroon (Fig. ). For detailed views of panels A–E, see Supplementary Figs –.

Implications for conservation ). Deforestation processes therefore pose substantial risks to the integrity and long-term sustainability of forests Deforestation of natural forests and the conversion of through numerous mechanisms. forested land for smallholder agricultural practices in the Outright conversion and loss of natural tropical forest Littoral Region of Cameroon have numerous negative increases net greenhouse gas emissions (Houghton & implications for regional biodiversity and wildlife conserva- Nassikas, ), contributing to human-induced climate tion, ecosystem functioning, environmental sustainabil- change. The ongoing conversion of the forests in the ity and livelihoods. Regarding biodiversity and wildlife, Littoral Region, which is occurring at a landscape level, multiple anthropogenic influences are threatening forests, will therefore contribute to greenhouse gas emissions in with many direct impacts, depending upon the extent and Cameroon and global human-induced climate change. type of forest altered. For instance, the observed deforesta- Long-term human well-being in the Littoral Region is tion and the replacement of natural forest with oil palm also threatened by the loss of natural forest cover. The for- plantations is likely to result in severe biodiversity loss ests in the Littoral Region harbour numerous medicinally   (Fitzherbert et al., ; Azhar et al., ). In South-east useful species, which are used by local people and are in de- Asia forest conversion to oil palm cultivation has had severe mand from global pharmaceutical companies (Colfer, ; negative effects on biodiversity across a wide range of taxa, Watson et al., ). The potential extinction of these medi- including birds, invertebrates and mammals (Peh et al., cinal plants as a result of deforestation in the Littoral Region    ; Koh & Wilcove, ; Vargas et al., ; Yue et al., has significant implications for the local populace, and  ). Hence, our findings forewarn of potential loss of legislative protection of the forests is needed to avoid this biodiversity and wildlife habitat in the Littoral Region as a scenario. Otherwise, if the expansion of oil palm cultivation, result of deforestation for oil palm cultivation. mining, logging, land clearing and hunting towards the With the current trajectory of land-cover change, we infer boundary of Ebo Forest is not halted, the combined effects that functioning of forested ecosystems in the Littoral will deny people medicinal support, decrease the quality Region is under threat as a result of the landscape-level of life in the region, and have significant socio-economic loss of forest cover, which can alter biogeochemical cycles impacts. and induce climate change. Moreover, degradation of ecosystem functioning in the forests of the Littoral Region is ex- pected to continue as anthropogenic land use gradually Conclusions encroaches on areas once occupied by natural forest (Gérard et al., ; Nicholas et al., ). There are likely Land-cover conversion is an insidious and increasing threat to be other cascading implications of forest land-cover to the forests and biodiversity in the Littoral Region of change, including soil degradation, in the Littoral Region. Cameroon. Significant deforestation has occurred in the Removal of natural forest cover exposes soils, which results study region, reducing natural forest cover from .%in in potentially irreversible soil erosion, degradation and  to .%in. This forest loss occurred predomi- threats to soil biota (Danielsen et al., ; Beca et al., nantly within  km of major roads, and is now occurring

within  km of important biodiversity and wildlife habitats. BRANDÃO,JR, A.O. & SOUZA,JR, C.M. () Mapping unofficial It is clear that the cumulative effects of human actions such as roads with Landsat images: a new tool to improve the monitoring of the establishment and expansion of oil palm plantations, log- the Brazilian Amazon rainforest. International Journal of Remote Sensing, , –. ging, land clearing, and expansion of roads and settlements BREIMAN,L.() Random forests. Machine Learning, , –. are driving deforestation in the periphery of important CETIN,M.() Evaluation of the sustainable tourism potential of a forested areas, particularly those of Ebo Forest and Mount protected area for landscape planning: a case study of the ancient Nlonako. There are numerous negative implications of the city of Pompeipolis in Kastamonu. International Journal of  – observed environmental change in the study area, including Sustainable Development and World Ecology, , . CHEEK, M., CABLE, S., HEPPER, F.N., NDAM,N.&WATTS,J.() habitat and biodiversity loss, a decline in ecosystem services, Mapping plant biodiversity on Mount Cameroon. In The and a decrease in the availability of medicinal forest products. Biodiversity of African Plants (eds L.J.G. van der Maesen, X.M. van In particular, the current rate of land conversion, and its der Burgt & J.M. van Medenbach de Rooy), pp. –. Kluwer proximity, suggests that Ebo Forest is highly threatened, Academic Publishers, Dordrecht, The Netherlands.  given its current lack of legislative protection. It is not an CINCOTTA, R.P., WISNEWSKI,J.&ENGELMAN,R.( ) Human population in the biodiversity hotspots. Nature, , –. overstatement to say that the continued pace of human ac- COLFER, C.J.P. () Human Health and Forests: a Global Overview of tions and environmental change in the areas surrounding Issues, Practice and Policy. Routledge, London, UK. Ebo Forest forewarn of serious ecological and conservation DANIELSEN, F., BEUKEMA, H., BURGESS, N.D., PARISH, F., BRÜHL, threats to the forest if legislative action is not taken swiftly, C.A., DONALD, P.F. et al. () Biofuel plantations on forested including the implementation of a proactive environmental lands: double jeopardy for biodiversity and climate. Conservation Biology, , –. management plan for the biologically diverse unprotected – FA, J.E., SEYMOUR, S., DUPAIN, J., AMIN, R., ALBRECHTSEN,L.& forests in the Littoral Region and Cross Sanga River MACDONALD,D.() Getting to grips with the magnitude of landscapes. exploitation: bushmeat in the Cross–Sanaga rivers region, Nigeria and Cameroon. Biological Conservation, , –. Acknowledgements Financial support for this study was provided FAHRIG,L.() Effects of habitat fragmentation on biodiversity. by the Arcus Foundation and James Cook University. We thank Annual Review of Ecology, Evolution, and Systematics, , –. Pontus Olofsson for verification of the error-adjusted area estimates, FAHRIG,L.() Rethinking patch size and isolation effects: the and the Editor and reviewers for their comments and suggestions. habitat amount hypothesis. Journal of Biogeography, , –. FEINTRENIE,L.() Oil palm in Cameroon: risks and opportunities. Author contributions Study conceptualization, data analysis and Nature & Faune, , –. writing: MIM; revision: all authors. FITZHERBERT, E.B., STRUEBIG, M.J., MOREL, A., DANIELSEN,F., BRÜHL, C.A., DONALD,P.F.&PHALAN,B.() How will oil palm Conflicts of interest None. expansion affect biodiversity? Trends in Ecology & Evolution, , –. Ethical standards This research complied with the Oryx guidelines GARCIA-ULLOA, J., SLOAN, S., PACHECO, P., GHAZOUL,J.&KOH, L.P. on ethical standards. () Lowering environmental costs of oil‐palm expansion in Colombia. Conservation Letters, , –. GÉRARD, A., WOLLNI, M., HÖLSCHER, D., IRAWAN, B., SUNDAWATI, References L., TEUSCHER,M.&KREFT,H.() Oil-palm yields in diversified plantations: initial results from a biodiversity enrichment ABERNETHY, K., MAISELS,F.&WHITE, L.J.T. () Environmental experiment in Sumatra, Indonesia. Agriculture, Ecosystems & issues in Central Africa. Annual Review of Environment and Environment, , –. Resources, , –. HADDAD, N.M., BRUDVIG, L.A., CLOBERT, J., DAVIES, K.F.,  ALAMGIR, M., CAMPBELL, M.J., SLOAN, S., GOOSEM, M., CLEMENTS, GONZALEZ, A., HOLT, R.D. et al. ( ) Habitat fragmentation ’  G.R., MAHMOUD, M.I. & LAURANCE,W.F.() Economic, and its lasting impact on Earth s ecosystems. Science Advances, , socio-political and environmental risks of road development in the e. tropics. Current Biology, ,R–R. HODGSON, J.A., MOILANEN, A., WINTLE, B.A. & THOMAS, C.D. () ALEMAN, J.C., JARZYNA, M.A. & STAVER, A.C. () Forest extent Habitat area, quality and connectivity: striking the balance for and deforestation in tropical Africa since . Nature Ecology & efficient conservation. Journal of Applied Ecology, , –. Evolution, , –. HOSONUMA, N., HEROLD, M., DE SY, V., DE FRIES, R.S., BROCKHAUS,  AUSTIN, K.G., MOSNIER, A., PIRKER, J., MCCALLUM, I., FRITZ,S.& M., VERCHOT, L. et al. ( ) An assessment of deforestation and KASIBHATLA, P.S. () Shifting patterns of oil palm driven forest degradation drivers in developing countries. Environmental deforestation in Indonesia and implications for zero-deforestation Research Letters, , . commitments. Land Use Policy, , –. HOUGHTON, R.A. & NASSIKAS, A.A. () Global and regional fluxes – AZHAR, B., LINDENMAYER, D.B., WOOD, J., FISCHER,J.&ZAKARIA, of carbon from land use and land cover change . Global M. () Ecological impacts of oil palm agriculture on forest Biogeochemical Cycles, , –. mammals in plantation estates and smallholdings. Biodiversity and INGRAM, V., EWANE, M., NDUMBE, L.N. & AWONO,A.() Conservation, , –. Challenges to governing sustainable forest food: Irvingia spp. from  – BECA, G., VANCINE, M.H., CARVALHO, C.S., PEDROSA, F., ALVES, southern Cameroon. Forest Policy and Economics, , .  R.S.C., BUSCARIOL, D. et al. () High mammal species turnover KOH, L.P. & WILCOVE, D.S. ( ) Is oil palm agriculture really in forest patches immersed in biofuel plantations. Biological destroying tropical biodiversity? Conservation Letters, , Conservation, , –. –.

LAPORTE, N.T., STABACH, J.A., GROSCH, R., LIN, T.S. & GOETZ, S.J. PEH, K.S.H., SODHI, N.S., DE JONG, J., SEKERCIOGLU, C.H., YAP, () Expansion of industrial logging in Central Africa. Science, C.A.M. & LIM, S.L.H. () Conservation value of degraded , . habitats for forest birds in southern Peninsular Malaysia. Diversity LAURANCE, W.F., SLOAN, S., WENG,L.&SAYER, J.A. () and Distributions, , –. Estimating the environmental costs of Africa’s massive PFAFF, A., ROBALINO, J., WALKER, R., ALDRICH, S., CALDAS, M., REIS, ‘development corridors’. Current Biology, , –. E. et al. () Road investments, spatial spillovers, and LAURANCE, W.F., CAMPBELL, M.J., ALAMGIR,M.&MAHMOUD, M.I. deforestation in the Brazilian Amazon. Journal of Regional Science, () Road expansion and the fate of Africa’s tropical forests. , –. Frontiers in Ecology and Evolution, , –. PIRKER, J., MOSNIER, A., KRAXNER, F., HAVLÍK,P.&OBERSTEINER, LAURANCE, W.F., CAMARGO, J.L.C., FEARNSIDE, P.M., LOVEJOY, T.E., M. () What are the limits to oil palm expansion? Global WILLIAMSON, G.B., MESQUITA, R.C.G. et al. () An Amazonian Environmental Change, , –. rainforest and its fragments as a laboratory of global change. PLANET TEAM () Planet Application Program Interface: in Space Biological Reviews, , –. for Life on Earth. Planet Team, San Francisco, USA. Https://api. MAHMOUD, M.I., DUKER, A., CONRAD, C., THIEL,M.&AHMAD, H.S. planet.com [accessed  December ]. () Analysis of settlement expansion and urban growth POHL,C.&VAN GENDEREN, J.L. () Multisensor image fusion in modelling using geoinformation for assessing potential impacts remote sensing: concepts, methods and applications. International of urbanization on climate in Abuja city, Nigeria. Remote Sensing, Journal of Remote Sensing, , –. , . POTAPOV, P., HANSEN, M.C., LAESTADIUS, L., TURUBANOVA, S., MAHMOUD, M.I., SLOAN, S., CAMPBELL, M.J., ALAMGIR, M., IMONG, YAROSHENKO, A., THIES, C. et al. () The last frontiers of I., ODIGHA, O. et al. () Alternative routes for a proposed wilderness: tracking loss of intact forest landscapes from  Nigerian superhighway to limit damage to rare ecosystems and to . Science Advances, ,e. wildlife. Tropical Conservation Science, , –. RENÓ, V.F., NOVO, E.M.L.M., SUEMITSU, C., RENNÓ, C.D. & SILVA, MARSIK, M., STEVENS, F.R. & SOUTHWORTH,J.() Amazon T.S.F. () Assessment of deforestation in the Lower Amazon deforestation: rates and patterns of land cover change and floodplain using historical Landsat MSS/TM imagery. Remote fragmentation in Pando, northern Bolivia,  to . Progress Sensing of Environment, , –. in Physical Geography: Earth and Environment, , –. SINGH, S.K., SRIVASTAVA, P.K., SZABÓ, S., PETROPOULOS, G.P., MCGARIGAL,K.&MARKS, B.J. () Fragstats: Spatial Pattern GUPTA,M.&ISLAM,T.() Landscape transform and spatial Analysis Program for Quantifying Landscape Structure. USDA metrics for mapping spatiotemporal land cover dynamics using Forest Service General Technical Report PNW-GTR-. Earth Observation data-sets. Geocarto International, , –. MITCHELL, A.L., ROSENQVIST,A.&MORA,B.() Current remote SOSEF, M.S.M., DAUBY, G., BLACH-OVERGAARD, A., VAN DER BURGT, sensing approaches to monitoring forest degradation in support of X., CATARINO, L., DAMEN, T. et al. () Exploring the floristic countries measurement, reporting and verification (MRV) systems diversity of tropical Africa. BMC Biology, , . for REDD+. Carbon Balance and Management, , . TAMUNGANG, S.A., CHEKE, R.A., MOFOR, G.Z., TAMUNGANG, R.N. & MOLUA, E.L. () Accommodation of climate change in coastal OBEN,F.T.() Conservation concern for the deteriorating areas of Cameroon: selection of household-level protection geographical range of the grey parrot in Cameroon. International options. Mitigation and Adaptation Strategies for Global Change, Journal of Ecology, , . , . TCHUENTÉ, A.T.K., ROUJEAN, J.-L. & DE JONG, S.M. () MORGAN, B.J., ADELEKE, A., BASSEY, T., BERGL, R.A., DUNN, A., Comparison and relative quality assessment of the GLC, FOTSO, R. et al. () Regional Action Plan for the Conservation of GLOBCOVER, MODIS and ECOCLIMAP land cover data sets at the Nigeria–Cameroon Chimpanzee (Pan troglodytes ellioti). IUCN/ the African continental scale. International Journal of Applied Earth SSC Primate Specialist Group and Zoological Society of San Diego, Observation and Geoinformation, , –. California, USA. THOMPSON, P.L., RAYFIELD,B.&GONZALEZ,A.() Loss of habitat NICHOLAS, K.M., FANZO,J.&MACMANUS,K.() Palm oil in and connectivity erodes species diversity, ecosystem functioning, Myanmar: a spatiotemporal study of how industrial farming affects and stability in metacommunity networks. Ecography, , –. biodiversity loss and the sustainable diet. Global Health: Science and VAN SOESBERGEN, A., ARNELL, A.P., SASSEN, M., STUCH, B., Practice, , –. SCHALDACH, R., GÖPEL, J. et al. () Exploring future agricultural OATES, J.F., BERGL, R.A. & LINDER, J.M. () Africa’s Gulf of Guinea development and biodiversity in Uganda, Rwanda and Burundi: Forests: Biodiversity Patterns and Conservation Priorities. Center for a spatially explicit scenario-based assessment. Regional Applied Biodiversity Science, Conservation International, Environmental Change, , –. Washington, DC, USA. VARGAS, L.E.P., LAURANCE, W.F., CLEMENTS, G.R. & EDWARDS,W. OLOFSSON, P., FOODY, G.M., HEROLD, M., STEHMAN, S.V., () The impacts of oil palm agriculture on Colombia’s WOODCOCK, C.E. & WULDER, M.A. () Good practices for biodiversity: what we know and still need to know. Tropical estimating area and assessing accuracy of land change. Remote Conservation Science, , –. Sensing of Environment, , –. VOGT, P., FERRARI, J.R., LOOKINGBILL, T.R., GARDNER, R.H., ORDWAY, E.M., ASNER, G.P. & LAMBIN, E.F. () Deforestation risk RIITTERS, K.H. & OSTAPOWICZ,K.() Mapping functional due to commodity crop expansion in sub-saharan Africa. connectivity. Ecological Indicators, , –. Environmental Research Letters, , . WATSON, J.E.M., EVANS, T., VENTER, O., WILLIAMS, B., TULLOCH, A., ORME, C.D.L., DAVIES, R.G., BURGESS, M., EIGENBROD, F., PICKUP, STEWART, C. et al. () The exceptional value of intact forest N., OLSON, V.A. et al. () Global hotspots of species richness are ecosystems. Nature Ecology & Evolution, , –. not congruent with endemism or threat. Nature, , –. WHYTOCK, R.C., BUIJ, R., VIRANI, M.Z. & MORGAN, B.J. ()Do PARDINI,R.() Effects of forest fragmentation on small mammals large birds experience previously undetected levels of hunting in an Atlantic Forest landscape. Biodiversity and Conservation, , pressure in the forests of Central and West Africa? Oryx, , –. –.

WIENS,J.,SUTTER,R.,ANDERSON,M.,BLANCHARD,J., WWF () Congo Basin. Https://www.worldwildlife.org/places/ BARNETT,A.,AGUILAR-AMUCHASTEGUI,N.etal.() congo-basin [accessed  November ]. Selecting and conserving lands for biodiversity: the role YOUNG, A., BOYLE,T.&BROWN,T.() The population genetic of remote sensing. Remote Sensing of Environment, , consequences of habitat fragmentation for plants. Trends in Ecology –. & Evolution, , –. WILLIS, K.S. () Remote sensing change detection for ecological YUE, S., BRODIE, J.F., ZIPKIN, E.F. & BERNARD,H.() Oil palm monitoring in United States protected areas. Biological plantations fail to support mammal diversity. Ecological Conservation, , –. Applications, , –.