Biological Conservation 270 (2022) 109554 Contents lists available at ScienceDirect Biological Conservation journal homepage: www.elsevier.com/locate/biocon Policy analysis Vulnerability mapping of 100 priority tree species in Central Africa to guide conservation and restoration efforts Viviana Ceccarelli a, Marius Ekué b,*, Tobias Fremout a,c, Hannes Gaisberger d,e, Chris Kettle d,f, Hermann Taedoumg b,g, Hendrik Wouters h, Eline Vanuytrecht h, Koen De Ridder h, Evert Thomas a,* a Alliance Bioversity International - CIAT, Lima, Peru b Alliance Bioversity International - CIAT, Yaoundé, Cameroon c Division of Forest, Nature and Landscape, KU Leuven, Leuven, Belgium d Alliance Bioversity International - CIAT, Rome, Italy e Department of Geoinformatics, Paris-Lodron University, Salzburg, Austria f Department of Environmental System Science, ETH Zürich, Zürich, Switzerland g Plant Systematic and Ecology Laboratory, University of Yaoundé, Yaoundé, Cameroon h Flemish Institute for Technological Research (VITO), Environmental Modelling Unit, Mol, Belgium A R T I C L E I N F O A B S T R A C T Keywords: Climate change and other anthropogenic threats are increasingly imperilling the diverse biomes of Central Af- African humid forests and savannas rica, which are globally important for biodiversity, carbon storage and people's livelihoods. The objectives of this Climate change paper were to: (i) map the vulnerability of 100 socio-ecologically important priority tree species in Central Africa Ensemble distribution modelling to climate change, fire, habitat conversion, overexploitation, overgrazing and (ii) propose a spatially explicit Fire Habitat conversion strategy to guide restoration and conservation actions. We performed ensemble distribution modelling to predict Overexploitation the present and future distributions of the 100 species, assembled other anthropogenic threat exposure layers, assessed species' sensitivities to the five threats based on their trait profiles, and constructed species-specific vulnerability maps by combining the species' exposure and sensitivity. The results show that these 100 species are vulnerable to the five threats, with an average of 34% of their distribution ranges under high to very high vulnerability and 60% under medium to high vulnerability to at least one threat. Many species identified as most vulnerable in this study are not considered as threatened by the IUCN Red List, suggesting a need to update their conservation status, potentially through integration of the vulnerability mapping methodology we used here. We generated both species-specific maps and summary maps including all 100 species identifying priority areas for a) in-situ conservation, b) ex-situ conservation, and c) active planting or assisted natural regeneration. We present an online platform to enable easy access to the vulnerability and the conservation and restoration priority maps for decision makers and support conservation and restoration planning across Central Africa. 1. Introduction the world's second largest rainforest after the Amazon, accounting for 30% of global rainforest cover (Malhi et al., 2013b). These forests are Climate change and other anthropogenic threats are increasingly crucial for global carbon storage and they sequester more carbon per imperilling the biomes of Central Africa (Abernethy et al., 2016; Réjou- hectare than the Amazon forests (Lewis et al., 2013). On the other hand, Méchain et al., 2021). This region hosts a wide diversity of biomes which the African continent also contains the largest area of tropical savannas are globally important for biodiversity, carbon storage and local people's in the world and, despite the lower tree density, this biome also stores livelihoods, ranging from the humid forests in the Congo Basin and substantial amounts of carbon in vegetation and soil (Grace et al., 2006). western coast, to the savannas in the Sahel region and eastern and Climate change is expected to impact forests and savannas in Central southern borders (Dinerstein et al., 2017). The African humid forests are Africa. Temperatures are predicted to increase by 2–4 ◦C by the end of * Corresponding authors. E-mail addresses: m.ekue@cgiar.org (M. Ekué), e.thomas@cgiar.org (E. Thomas). https://doi.org/10.1016/j.biocon.2022.109554 Received 17 December 2021; Received in revised form 18 March 2022; Accepted 7 April 2022 0006-3207/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by- nc-nd/4.0/). V. Ceccarelli et al. B i o l o g ic a l C o n s e r v a t i o n 270 (2022) 109554 this century in Central Africa (Aloysius et al., 2016), while expected 2. Methods changes in precipitations vary in sign and intensity between different models across most of the region (Aloysius et al., 2016; Dosio et al., 2.1. Study area 2021). A recent study found that current climatic niches in African humid forests associated with specific forest types are predicted to move We performed the analysis in the Central Africa region and the sur- to new areas due to climate change, threatening the survival of such rounding countries. While our main geographical focus was the Central forests and their species (Réjou-Méchain et al., 2021). Several studies African region (Cameroon, Central African Republic, Chad, Republic of have reported that, despite the widespread re-greening of Sahel Congo, Democratic Republic of Congo, Equatorial Guinea, Gabon, São following the long-term droughts in 1970s–1980s (Brandt et al., 2015; Tomé & Principe), we extended the boundaries of the study area to the Eklundh and Olsson, 2003; Herrmann et al., 2005), climate change is extent 0◦ W–35◦ E, − 15◦ S–25◦ N to obtain more reliable suitability decreasing tree diversity and increasing the abundance of drought- distributions of the modelled species, which also occur in surrounding tolerant species in the Sahel and West African savannas (Brandt et al., countries. The study area covers 30 countries and seven biomes (Din- 2015; Gonzalez et al., 2012; Herrmann and Tappan, 2013). erstein et al., 2017) (Fig. S1). For simplicity, from here on we refer to Central African biomes are currently under pressure of anthropo- this whole study area as ‘Central Africa’. genic threats which have increased in unison with rapid population growth during the last decades (Gerland et al., 2014). Population in 2.2. Tree species selection Central Africa is mostly rural and largely relies on subsistence agricul- ture and extraction of forest and savanna resources. Over the past de- For the vulnerability mapping, we assembled a list of 100 socio- cades, deforestation in Central Africa has been mainly driven by ecologically important priority species from several priority lists for vegetation clearance for smallholder agriculture, exploitation of fuel- African tree species (Franzel et al., 2007; IUCN, 2021; Jaenicke et al., wood, and timber logging (Abernethy et al., 2016; Tyukavina et al., 1995; Sacandé and Berrahmouni, 2016; Sacandé and Pritchard, 2004) 2018). Although Central Africa has lower deforestation rates than Latin (Table S1). We selected the species from these priority lists according to America and Asia due to a lower presence of industrial agriculture the following criteria: i) native from Central Africa; ii) socio- (Abernethy et al., 2016; Tyukavina et al., 2018), deforestation in African economically important for timber, fuelwood, edible fruits, forage, or humid forests still contributes to 11% of global forest loss (Malhi et al., other non-wood products; iii) important for conservation or used in 2013a). Fire is another important threat in Central Africa. While natural restoration programs; iv) with at least 30 presence points after fires are very common in African savannas (Andela et al., 2017) and this geographical filtering at 5 arcmin resolution. The 100 species belong to biome is adapted to burning (Veldman et al., 2015), fires represent a 24 families and 70 genera, the most species-rich families being Fabaceae great threat to humid forests in Central Africa as they do not naturally (30 species), Meliaceae (11 species) and Combretaceae (10 species) occur in these forests and consequently, tree species are highly vulner- (Table S1). 54 species mostly occur in the savanna biome, 23 in the able (Cochrane, 2003). humid forest biome and 23 in both biomes (Table S1). The list of the 100 In response to the current climate and biodiversity crises, several species, families, main biome, priority lists and main uses is provided in global initiatives have committed to promote conservation and resto- Table S1. We present the species richness map of the 100 selected species ration actions across the world. These include the Bonn Challenge that across Central Africa (i.e. a map indicating the number of species aims to restore 350 million ha of degraded lands and the UN High occurring per grid cell) in Fig. S2. Ambition Coalition for Nature and People that aims to protect 30% of the planet by 2030. In the context of these initiatives, African countries 2.3. Species distribution modelling have pledged to build an 8000 km wall of trees stretching from East to West Africa under the Great Green Wall initiative, and to restore 100 We compiled species presence points from RAINBIO (Dauby et al. million ha of degraded land by 2030 under the African Forest Landscape (2016); https://gdauby.github.io/rainbio/download_page.html), GBIF Restoration Initiative (AFR100). To ensure long-term success, such (Global Biodiversity Information Facility; www.gbif.org), and BIEN conservation and restoration efforts should include careful evaluation of (Botanical Information and Ecology Network; http://biendata.org/). To climate change and other anthropogenic threats (Carwardine et al., reduce spatial bias (Kramer-Schadt et al., 2013), we filtered the presence 2012; Gillson et al., 2013). As threats are not spatially homogeneous and points using both geographical filtering at 5 arcmin and environmental different species have different sensitivities to the same threat, spatially filtering. We used 45 predictor variables, including 19 bioclimatic var- explicit vulnerability assessments can help to define which regions and iables, 5 variables of cloud cover, 6 variables indicating climatic ex- species are most in need of conservation and restoration (Fremout et al., tremes (VITO, 2020, 2021), 10 soil variables (Hengl et al., 2017), and 5 2020; Gaisberger et al., 2017, 2021). terrain variables (Table S2). We removed collinear variables using the The objectives of this paper were to (i) map the vulnerability of 100 Variance Inflation Factors (VIF). We selected pseudo-absence and socio-ecologically important priority tree species in Central Africa to background points using to the target group method described by climate change, fire, habitat conversion, overexploitation, overgrazing Phillips et al. (2009) and Mateo et al. (2010). Distribution modelling was and (ii) propose a spatially explicit strategy to guide species-specific carried out using ensembles with up to nine algorithms using the Bio- conservation and restoration actions. We performed ensemble distri- diversityR package for R (Kindt, 2018), consisting of random forest, bution modelling to predict the present and future distributions of the MAXENT, GBM, GLMSTEP, GAMSTEP, MGCV, FDA, SVM, and EARTH. 100 species, assembled other anthropogenic threat exposure layers, The models were cross-validated with 5 folds and using spatial blocks assessed species' sensitivities to the five threats based on their trait implemented through the blockCV package for R (Valavi et al., 2019), profiles, and constructed species-specific vulnerability maps by and model performance was assessed using the Area Under the receiver- combining the species' exposure and sensitivity. We produced maps operator Curve values cross-validated with spatial blocks (cvAUC). We indicating recommended areas for conservation and restoration actions converted the suitability maps into presence-absence maps using the for each species and for the whole study area. We discuss how the results suitability threshold at which model sensitivity equates to model spec- of this study can be used to guide conservation and restoration actions ificity. The detailed methodology for the species distribution modelling across Central Africa. is presented in Text S1. 2.4. Threat exposure Exposure to each of the five threats (fire, habitat conversion, 2 V. Ceccarelli et al. B i o l o g ic a l C o n s e r v a t i o n 270 (2022) 109554 overgrazing, overexploitation, climate change) was estimated following was assigned a trait weight of 5 for overgrazing, as it is one of the main the methodology described in Fremout et al. (2020) and Gaisberger et al. traits in defining sensitivity to overgrazing, and species with non- (2021). Exposure maps were generated using freely accessible spatial palatable leaves and palatable leaves were assigned partial sensitivity datasets and according to assumptions from literature and expert scores to overgrazing of 0.25 and 1, respectively (Fig. 1, Table S3). knowledge. The exposure maps had values from 0 to 1 (zero to We defined the overall sensitivity of the 100 species to each threat by maximum exposure) and had resolution of 30 arcsec (ca. 0.9 km at the calculating the weighted average of the partial sensitivity scores and the equator). The exposure maps for fire, habitat conversion, overgrazing weights (Fig. 1, Table S3). Some specific trait levels were assigned a and overexploitation represent current exposure levels, while the fixed score: we assigned a sensitivity score to overgrazing of 0.25 to all exposure maps for climate change represent the predicted future expo- species with unpalatable leaves, and a sensitivity score to over- sure. In addition, we performed a sensitivity analysis to assess the exploitation of 0.25 to all species that are not used for firewood nor impact of methodological decisions on the results (see Section 2.7), for timber. We selected a value of 0.25 because these species are not which we complemented the reference exposure maps with best-case completely unsusceptible (e.g., a species with unpalatable leaves may and worst-case exposure maps. The detailed methodology for esti- still be impacted by trampling). mating the reference, best-case and worst-case exposure maps to the five Vulnerability maps for each species were constructed by multiplying threats is presented in Text S2. the threat exposure maps with the sensitivity values for each species. We then categorized these vulnerability maps into five categories: zero 2.5. Species sensitivity and vulnerability (0–0.01), low (0.01–0.25), medium (0.25–0.5), high (0.50–0.75) and very high (0.75–1) vulnerability. Species sensitivity and vulnerability was estimated following Frem- out et al. (2020) and Gaisberger et al. (2021). We estimated the sensi- 2.6. Maps for conservation and restoration tivity of the 100 species to each of the five threats using a set of 16 traits (Fig. 1, Table S3). The 16 traits mostly refer to biological traits (e.g., Based on the vulnerability maps, we created species-specific maps bark thickness or leaf phenology) but we also included plant uses such as identifying priority conservation and restoration actions, following fuelwood and timber provision. We compiled the trait data for the 100 Fremout et al. (2020) and Gaisberger et al. (2021). For constructing species from an extensive literature search (Table S4). The obtained trait these maps, we analysed vulnerability to climate change and vulnera- dataset had an average of only 5% missing traits per species, ranging bility to current threats separately. We calculated the vulnerability to from 0% for the species Afzelia africana, Pterocarpus angolensis and current threats as the highest among fire, habitat conversion, over- Pentaclethra macrophylla to 13% for Psorospermum febrifugum (Table S5). exploitation and overgrazing. Because different threats often have ad- We defined the relation between each trait and the sensitivity of each ditive or synergistic impacts on vulnerability (Côté et al., 2016), we species to the five threats following the rationale in Table S3. First, each adjusted the values of vulnerability to current threats to ‘very high’ trait was assigned a weight indicating the expected importance for where the vulnerability value to at least three current threats was ‘high’, species sensitivity to any of the five threats, ranging from 1 (very low) to and to ‘high’ where the vulnerability to at least three current threats was 5 (very high importance) (Fig. 1, Table S3). Then, each trait was divided ‘medium’. into several levels linked with a partial score based on the expected Based on the vulnerability to current threats and climate change, we influence on the sensitivity of the species, varying between zero (lowest generated maps indicating priority areas for conservation and restora- sensitivity) and one (highest sensitivity). For example, leaf palatability tion actions for each species. The conservation and restoration actions FIRE OVERGRAZING Maximum height Low Bark thickness Presence of spines Intermediate Thin x No x High Leaf palatabilityIntermediate x Palatable Yes Dispersal type Wood density x Thick Non-palatable Authocory Low Water Intermediate Growth rate x Wind x High Slow x Birds and bats Intermediate Weights Wild mammals x Fast Cattle and/or goats Very high High Intermediate Growth rate Low Slow Very low Intermediate Scores x Fast 1 0.75 0.50 0.25 Germination strategy Recalcitrant x Orthodox Resprouting capacity Resprouting capacity No No x Yes x Yes Species sensitivity to fire = 0.50 Species sensitivity to overgrazing = 0.88 Fig. 1. Illustration of the estimation of the sensitivity of the species Afzelia africana to fire (left) and overgrazing (right). Trait weights are indicated by shades of blue and the partial scores by colours from green to red (see legend in the middle). Overall sensitivity values, estimated as the weighted average of the partial scores with the trait weights, are indicated at the bottom of the figure. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 3 V. Ceccarelli et al. B i o l o g ic a l C o n s e r v a t i o n 270 (2022) 109554 include: a) in-situ conservation, b) ex-situ conservation or assisted 3.2. Species sensitivity and vulnerability migration, c) active planting or assisted natural regeneration. First, in- situ conservation of tree populations and seed collection for tree Fig. 2 summarizes the proportion of the current distribution of each planting activities is prioritized in areas with low vulnerability to cur- of the 100 species under the different vulnerability levels to the five rent threats and climate change. Areas with low vulnerability to current threats. On average 34% (±12 SD) of the grid cells within species dis- threats are likely to have higher genetic variability and lower inbreeding tribution ranges had a high to very high vulnerability to at least one of rates than areas heavily disturbed by humans, while low climate change the five threats, while 60% (±14 SD) had medium to very high vulnerability to climate change ensure that local populations will likely vulnerability. For the individual threats, the average area under high to remain viable and continue producing seed in the future. Second, ex-situ very high vulnerability was 19% (±10 SD) for overexploitation, 10% conservation or assisted migration is prioritized in areas with high (±9 SD) for habitat conversion, 9% (±13 SD) for climate change, 6% vulnerability to climate change. This in order to protect the genetic (±7 SD) for overgrazing, and 5% (±4 SD) for fire. When considering the variability of populations within the same species that currently grow in average area under medium to very high vulnerability, the values rose to an area that are expected to become unsuitable under climate change, of 38% (±14 SD) for overexploitation, 19% (±14 SD) for habitat conver- which the genetic diversity may be lost if they are not conserved ex-situ sion, 18% (±19 SD) for climate change, 12% (±12 SD) for overgrazing, or helped to migrate to areas where they are more likely to persist. and 14% (±8 SD) for fire. The most vulnerable species in terms of pro- Third, restoration activities such as active planting or assisted natural portion of their distribution under high to very high vulnerability were regeneration are prioritized in areas under high to very high current Prunus africana, Cola nitida, Dacryodes macrophylla, Pouteria altissima, threat vulnerability but low vulnerability to climate change, combined and Vachellia gerrardii, with 58–80% of their distribution under high to with interventions to decrease the current anthropogenic threats. Areas very high vulnerability. with a high or very high vulnerability to current threats are the ones that most require restoration interventions, while the low vulnerability to 3.3. Maps for conservation and restoration climate change increases the likelihood that the planted or regenerating trees will be able to survive in future. Finally, the conservation and For each of the 100 species, we generated maps highlighting priority restoration maps also identify areas that are not suitable in present but areas for conservation and restoration actions for a) in-situ conservation; are predicted to become suitable in future under climate change (Text b) ex-situ conservation or assisted migrations; and c) active planting or S2). assisted natural regeneration. Across the 100 species, on average 40% In addition to the species-specific maps, we also constructed sum- (±14 SD) of the distribution ranges were prioritized for in-situ conser- mary maps depicting priority areas for conservation and restoration vation, 22% (±12 SD) for restoration, and 9% (±13 SD) for ex-situ interventions across Central Africa. These maps were generated based conservation, while 11% (±13 SD) of the distribution ranges are ex- on the number of species per grid cell for which the grid cell in question pected to change from unsuitable to suitable under climate change. is recommended for the given conservation and restoration action. Fig. 3 shows the example of the conservation and restoration map for the species Faidherba albida. The conservation and restoration maps for the 2.7. Sensitivity analysis 100 species together with the vulnerability maps can be visualized on- line at: https://tree-diversity.shinyapps.io/vulnerability_central_africa/ Finally, we carried out a sensitivity analysis to estimate how meth- and can be downloaded at: https://doi.org/10.6084/m9.figshare odological decisions impact the results of the conservation and resto- .19635996. ration maps, following Fremout et al. (2020). We included three factors: Fig. 4 shows the number of species per grid cell recommended for a) methodological decisions used to construct threat exposure maps; b) conservation and restoration actions across Central Africa. Priority areas trait weighting schemes chosen to calculate sensitivity values; and c) for conservation for 20 to 50 species (representing 90–100% species missing trait values. For each factor, we applied two ‘treatments’ for the occurring in the areas) are concentrated in the humid forests of Gabon vulnerability assessment in addition to the ‘reference’ treatment (i.e., and southern Cameroon and in the savannas in southern Chad, northern the original or reference maps). The two treatments for a) correspond to Central African Republic and western South Sudan, and occur both in- the best-case and worst-case exposure map described in Section 2.4 and side and outside protected areas (Fig. 4a). Priority areas for restoration Text S2, while the details for the treatments for b) and c) are explained in for 20 to 50 species (representing 90–100% species occurring in the Text S3. We performed the sensitivity analysis on the conservation and areas) are concentrated in the humid forests in southern Nigeria and in restoration maps and we generated for each species six versions of these the savannas from Togo, Benin, Nigeria to northern Cameroon, and they maps, corresponding to the six treatments. For each of the six treat- are mostly within converted areas (Fig. 4b). ments, we calculated the percentage of grid cells within distribution of each species that changes their priority actions recommended as 3.4. Sensitivity analysis compared to the reference map. Table S7 summarizes the results of the sensitivity analysis. The 3. Results conservation and restoration maps are robust against the trait weighting schemes and missing traits (average change 2–9%), while they are more 3.1. Species distribution modelling influenced by the methodological decisions used to construct threat exposure maps (average change 23–25%). The mean cvAUC value of the distribution models of the 100 species was 0.81. The cvAUC ranged from 0.64 to 0.96 with only seven species 4. Discussion out of 100 with cvAUC < 0.70, indicating good to very good distribution models. The list of cvAUC values for the 100 species is provided in In this study, we quantified the vulnerability of 100 socio- Table S6. The average cvAUC value of the individual modelling algo- ecologically important priority tree species across Central Africa to rithms ranged from 0.73 ± 0.21 SD (GBMSTEP) to 0.80 ± 0.07 SD climate change, fire, habitat conversion, overexploitation, overgrazing. (random forest). The ensemble model was the most accurate for 39 of Our results show that several species are threatened, with an average of the 100 species, followed by random forest (18 species) and SVM (18 34% of their distribution ranges under high to very high vulnerability species). and 60% under medium to high vulnerability to at least one threat. Considering the commitment of African countries towards AFR100 and other international initiatives, it is essential that conservation and 4 V. Ceccarelli et al. B i o l o g ic a l C o n s e r v a t i o n 270 (2022) 109554 Habitat Maximum Fire conversion Overexploitation Overgrazing Climate change threat Prunus africana Cola nitida Vulnerability Dacryodes macrophylla Pouteria altissima Vachellia gerrardii Lannea microcarpa Gambeya africana Senegalia dudgeonii Combretum molle Prosopis africana Khaya anthotheca Ficus exasperata Pterocarpus erinaceus Pterocarpus lucens Diospyros mespiliformis Combretum adenogonium Parkia biglobosa Borassus aethiopum Combretum nigricans Garcinia lucida Adansonia digitata Sterculia setigera Carapa procera Afrostyrax lepidophyllus Pentadesma butyracea Khaya senegalensis Craterispermum schweinfurthii Combretum aculeatum Combretum glutinosum Isoberlinia doka Annona senegalensis Trichilia emetica Anogeissus leiocarpa Sclerocarya birrea Vachellia hockii Pterocarpus soyauxii Afzelia africana Lovoa trichilioides Tamarindus indica Parinari curatellifolia Detarium microcarpum Erythrophleum ivorense Vitex doniana Combretum collinum Terminalia macroptera Ximenia americana Entandrophragma candollei Daniellia oliveri Vitellaria paradoxa Psorospermum febrifugum Khaya ivorensis Garcinia livingstonei Terminalia superba Entandrophragma cylindricum Garcinia kola Senegalia polyacantha Dalbergia melanoxylon Alchornea cordifolia Mansonia altissima Irvingia gabonensis Afzelia pachyloba Tetrapleura tetraptera Bauhinia rufescens Vachellia nilotica Faidherbia albida Voacanga africana Commiphora africana Crossopteryx febrifuga Erythrophleum suaveolens Senegalia senegal Chrysophyllum lacourtianum Afzelia quanzensis Milicia excelsa Senegalia ataxacantha Lophira alata Cola acuminata Garcinia epunctata Maerua crassifolia Diospyros crassiflora Nauclea latifolia Entandrophragma angolense Balanites aegyptiaca Piptadeniastrum africanum Guarea cedrata Annickia chlorantha Nauclea diderrichii Picralima nitida Dacryodes edulis Trichoscypha acuminata Pterocarpus angolensis Baillonella toxisperma Terminalia mollis Xylopia aethiopica Entandrophragma congoense Ricinodendron heudelotii Pentaclethra macrophylla Aucoumea klaineana Cylicodiscus gabunensis Vachellia tortilis Pausinystalia johimbe 0% 50% 100% 0% 50% 100% 0% 50% 100% 0% 50% 100% 0% 50% 100% 0% 50% 100% (caption on next page) 5 V. Ceccarelli et al. B i o l o g ic a l C o n s e r v a t i o n 270 (2022) 109554 Fig. 2. Summary of sensitivity and vulnerability estimates of the 100 tree species for the five threats. Black dots indicate the species sensitivity values to the five threats. The cumulative bar plots indicate the relative frequency of species vulnerability values to each of the five threats (columns 1–5) and the relative frequency of the maximum vulnerability values (column 6). The bars show the proportions of the current distribution range of each species with a zero, low, medium, high and very high vulnerability to the five threats; and are indicated with colours from light yellow to dark red. The relative frequency of the maximum vulnerability values refers to the highest vulnerability among the different threats within a grid cell. The species are ordered according to decreasing proportion of the distribution range with high or very high vulnerability to at least one of the five threats (column 6). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) et al., 2013). The lower vulnerability to habitat conversion is in line with Faidherbia albida the fact that deforestation in Central Africa is mostly driven by small- scale vegetation clearance for smallholder agriculture which leads to lower deforestation rates, while large-scale deforestation driven by in- dustrial agriculture is still limited (Abernethy et al., 2016; Tyukavina et al., 2018). However, the recent expansion of both industrial agricul- ture (Feintrenie, 2014; Ordway et al., 2017) and subsidence agriculture (Herrmann et al., 2020; Tyukavina et al., 2018) may increase the threat of habitat conversion in the near future. Considering the high impact of overexploitation, it is fundamental that at least this threat is included in conservation and restoration planning in Central Africa, in addition to the more commonly considered threats of climate change and habitat conversion (e.g. Bomhard et al., 2005; Gomes et al., 2019; Triviño et al., 2018). The overall high vulnerability to anthropogenic threats of the selected 100 species poses concerns about the conservation status of tree species in Central Africa. The most vulnerable species identified in this study should be considered high priority species for conservation ac- tions. The five most vulnerable species had 58–80% of their area under high to very high vulnerability. Yet of these only Prunus africana is classified as ‘vulnerable’ according to the Global IUCN Red List, while Cola nitida and Pouteria altissima are considered ‘not threatened’ and Dacryodes macrophylla and Vachellia gerrardii are not assessed. This suggests that there may be a need to re-evaluate the current IUCN Red List assessments for these and other species, potentially through inte- gration of the vulnerability mapping methodology we used here. Eighty- five out of the 100 species considered in this study have been assessed in the Global IUCN Red List, but among the countries in the Central African region, only Cameroon currently has a National Red List for Vascular Plants (Onana and Cheek, 2011). It is essential to have country-level assessments to evaluate the conservation status of tree species in in situ conservation of seed sources different countries and conserve the genetic variation that exists across active planting or assisted regeneration their distributions. Of the 98 species occurring in Cameroon among the ex situ conservation of seed sources 100 species analysed in this study, only 31 were included in the National expected to become suitable Red List of Cameroon, with 11 species classified as threatened while the no priority actions remaining 67 were not assessed (Table S8). Furthermore, 4 out of the 20 species considered as not threatened and 17 out of 67 not assessed by the Fig. 3. Map of restoration and conservation priority areas for Faidherbia albida, National Red List of Cameroon have more than 50% of their distribution indicating priority areas for in-situ conservation of seed sources (blue), active area within Cameroon under high to very high vulnerability (Table S8). planting or assisted natural regeneration (green), ex-situ conservation of seed This illustrates the urgent need in Central African countries to develop sources (dark red), areas expected to become suitable (yellow), and areas with National Red Lists, to which our vulnerability mapping methodology no priority actions (grey). Countries of Central African region are indicated can contribute. with thick borders. (For interpretation of the references to colour in this figure The vulnerability mapping methodology used in this study can legend, the reader is referred to the web version of this article.) contribute to IUCN Red List assessments and to inform large-scale con- servation and restoration planning. The methodology can be useful to restoration plans in the region include a careful evaluation of these complement the IUCN Red List assessments, as it uses a spatially-explicit threats. estimation of the impact of current anthropogenic threats and includes Our vulnerability assessment suggests that overexploitation repre- the future impact of climate change. The methodology has already been sents the highest threat to the selected tree species in the study area applied in tropical forests in South America (Fremout et al., 2020) and (19% of the species distribution areas under high to very high vulner- Asia (Gaisberger et al., 2021), and future studies could further improve ability on average), followed by habitat conversion and climate change the method used to create the exposure maps and to estimate trait-based (9% each), while overgrazing and fire have lower importance (5–6%). sensitivity values, in order to make it applicable to other regions and The lower expected impact of climate change at least in the humid forest ecosystems as well. Regarding the exposure maps, while the exposure biome corroborates previous studies which postulated that African maps of fire, habitat conversion, overgrazing and climate change were humid forests are more resilient to climate change than other tropical calculated using databases directly related to the respective threat forests (Asefi-Najafabady and Saatchi, 2013; Bennett et al., 2021) exposure, the exposure map of overexploitation was constructed by largely due to a history of unstable post-Pleistocene climate which combining the proxies of human population density and accessibility to already led to the selection of more climate-resilient species (Willis cities (Text S2). If available in a study area, estimates of exposure to 6 V. Ceccarelli et al. B i o l o g ic a l C o n s e r v a t i o n 270 (2022) 109554 outside protected areas converted areas within protected areas non-converted areas Fig. 4. Summary maps for priority conservation and restoration areas of the 100 socio-economically important tree pecies. The maps indicate the number of species per grid cell recommended for a) in-situ conservation and b) active planting or assisted natural regeneration actions. In a) we indicate in different colours priority areas falling within protected areas (shades of green) and outside protected areas (shades of red), while in b) we indicate priority areas falling within non-converted areas (shades of blue) and converted areas (shades of purple). Countries of Central African region are indicated with thick borders. (For interpretation of the ref- erences to colour in this figure legend, the reader is referred to the web version of this article.) overexploitation could be improved by including spatial data describing these ecosystems characterized by low tree density (Tölgyesi et al., other determinants of overexploitation, such as law enforcement and use 2021; Veldman et al., 2019). of forest resources (Fremout et al., 2020). On the other hand, depending On the other hand, priority areas for conservation are concentrated on the study area, other threats in addition to the five used in this study in the humid forests of Gabon and southern Cameroon and in the sa- could be added. For instance, in Central Africa, other relevant threats vannas in southern Chad, northern Central African Republic and western include mining (Edwards et al., 2014) and conflict zones (Mirzabaev South Sudan. Considering the overall low expected impact of climate et al., 2021). Regarding the trait-based sensitivity estimates, it would be change on the species of interest across the region, there are no specific important to standardize the set of traits selected for each threat ac- areas prioritized for ex-situ conservation of many species simulta- cording to latest studies and possibly also include ‘hard’ traits directly neously. However, individual species such as Dacryodes macrophylla, linked to the functional mechanisms determining species vulnerability Prunus africana and Pouteria altissima may be severely impacted by (Fremout et al., 2020) such as leaf flammability for fire or xylem hy- climate change and may require collection of seeds for ex-situ conser- draulic conductivity for climate change, at least for species for which vation or assisted migration to preserve the genetic variability of pop- these data are available. ulations that grow in areas that are predicted to become unsuitable The general conservation and restoration maps that we generated under climate change. (Fig. 4) can help prioritize conservation and restoration actions. Large Apart from prioritizing the most suited tree species and areas, to areas in the humid forests in southern Nigeria and in the savannas from achieve successful restoration, it is also critical to ensure that functional Togo, Benin, Nigeria to northern Cameroon are indicated as priority for seed systems are put in place (Atkinson et al., 2021). Such systems are restoration activities, as the tree populations in these regions are under crucial to obtain sufficient quantities of genetically diverse and locally higher threat especially from overexploitation and habitat conversion adapted planting material, capable to persist under climate change and compared to less populated forests in the Congo Basin (Fig. S3). Some of able to meet the diverse restoration goals. Such seeds should be sourced these priority areas for restoration activities in northern Nigeria and from areas identified as priority for in-situ conservation in Fig. 4 northern Benin are already part of the Great Green Wall initiative, but whenever possible. In countries with large-scale restoration needs, it additional efforts are needed in other areas. Large-scale restoration will be critical to protect remaining seed sources, such as forest frag- projects with active planting and assisted natural regeneration should ments but also trees on farms or even in cities (Rimlinger et al., 2021). focus on degraded forests which have not been converted to agriculture Further, there may be a need for multilateral collaboration within and yet in southern Nigeria, while promoting agroforestry in farmed land- across countries to facilitate seed exchange. To ensure long-term suc- scape might be a better option for converted areas in Togo, Benin, cess, it is also imperative to involve local people in the decision making northern Nigeria, northern Cameroon. Considering the high impact of of projects (Mansourian and Berrahmouni, 2021). overexploitation, restoration projects should promote planting of spe- The maps indicating vulnerability and conservation and restoration cies that are important to local people for timber, fuelwood, fruits, and actions for the 100 species are available online to facilitate their use by other non-wood products. Importantly, restoration interventions in sa- forest practitioners and policy makers: https://tree-diversity.shinyapps. vannas and grasslands should focus on restoring the original tree cover io/vulnerability_central_africa/ and can be downloaded at: https://doi. of these ecosystems rather than afforestation, which could instead org/10.6084/m9.figshare.19635996. To further support restoration threaten the unique biodiversity and ecosystem service provision of projects in the region, the distribution maps generated in this study were 7 V. Ceccarelli et al. B i o l o g ic a l C o n s e r v a t i o n 270 (2022) 109554 also integrated into the Diversity for Restoration (D4R) tool, which References provides location-specific information on suitable tree species and seed sources for restoration projects: https://www.diversityforrestoration. Abernethy, K., Maisels, F., White, L.J.T., 2016. Environmental issues in Central Africa. org/ (Fremout et al., 2021). The D4R tool is currently available for Annu. Rev. Environ. Resour. 41, 1–33. https://doi.org/10.1146/annurev-environ- 110615-085415. Cameroon and Burkina Faso and will be expanded to other countries in Aloysius, N.R., Sheffield, J., Saiers, J.E., Li, H., Wood, E.F., 2016. Evaluation of historical Central Africa. We hope that the maps and results from this study can and future simulations of precipitation and temperature in central Africa from contribute to inform conservation and restoration projects in Central CMIP5 climate models. J.Geophys.Res.Atmos. 121 (1), 130–152. https://doi.org/ 10.1002/2015JD023656. Africa. Andela, N., Morton, D.C., Giglio, L., Chen, Y., Van Der Werf, G.R., Kasibhatla, P.S., DeFries, R.S., Collatz, G.J., Hantson, S., Kloster, S., Bachelet, D., Forrest, M., 5. Conclusions Lasslop, G., Li, F., Mangeon, S., Melton, J.R., Yue, C., Randerson, J.T., 2017. A human-driven decline in global burned area. Science 356 (6345), 1356–1362. https://doi.org/10.1126/science.aal4108. In this study, we assessed the vulnerability to climate change, fire, Asefi-Najafabady, S., Saatchi, S., 2013. Response of African humid tropical forests to habitat conversion, overexploitation, overgrazing of 100 socio- recent rainfall anomalies. Philos.Trans.R.Soc.BBiol.Sci. 368 (1625) https://doi.org/ 10.1098/rstb.2012.0306. ecologically important priority tree species and generated maps identi- Atkinson, R.J., Thomas, E., Roscioli, F., Cornelius, J.P., Zamora-Cristales, R., Franco fying priority areas for conservation and restoration actions across Chuaire, M., Alcázar, C., Mesén, F., Lopez, H., Ipinza, R., Donoso, P.J., Gallo, L., Central Africa. We found that on average 34% of the distribution ranges Nieto, V., Ugarte, J., Sáenz-Romero, C., Fremout, T., Jalonen, R., Gaisberger, H., Vinceti, B., Kettle, C., 2021. Seeding resilient restoration: an indicator system for the of the 100 species is under high to very high vulnerability and 60% analysis of tree seed systems. Diversity 13 (8), 1–14. https://doi.org/10.3390/ under medium to high vulnerability to at least one threat, which calls for d13080370. actions to protect these species. Many species identified as most Bennett, A.C., Dargie, G.C., Cuni-Sanchez, A., Mukendi, J.T., Hubau, W., Mukinzi, J.M., vulnerable in this study are not considered as threatened by the IUCN Phillips, O.L., Malhi, Y., Sullivan, M.J.P., Cooper, D.L.M., Adu-Bredu, S., Affum- Baffoe, K., Amani, C.A., Banin, L.F., Beeckman, H., Begne, S.K., Bocko, Y.E., Red List and there are no national assessments for tree species in the Boeckx, P., Bogaert, J., Lewis, S.L., 2021. Resistance of African tropical forests to an Central African countries apart from Cameroon, suggesting a need to extreme climate anomaly. Proc. Natl. Acad. Sci. U. S. A. 118 (21), 1–12. https://doi. update of the IUCN Red List for these countries. The conservation and org/10.1073/pnas.2003169118. Bomhard, B., Richardson, D.M., Donaldson, J.S., Hughes, G.O., Midgley, G.F., restoration priority maps are available on a platform online in order to Raimondo, D.C., Rebelo, A.G., Rouget, M., Thuiller, W., 2005. Potential impacts of contribute to AFR100 and other conservation and restoration initiatives future land use and climate change on the Red List status of the Proteaceae in the in Central Africa. The vulnerability mapping methodology used in this Cape Floristic Region,South Africa. Glob. Chang. Biol. 11 (9), 1452–1468. https:// doi.org/10.1111/j.1365-2486.2005.00997.x. study can complement IUCN Red List assessments and inform large-scale Brandt, M., Mbow, C., Diouf, A.A., Verger, A., Samimi, C., Fensholt, R., 2015. Ground- conservation and restoration planning. and satellite-based evidence of the biophysical mechanisms behind the greening Sahel. Glob. Chang. Biol. 21 (4), 1610–1620. https://doi.org/10.1111/gcb.12807. Carwardine, J., O’Connor, T., Legge, S., Mackey, B., Possingham, H.P., Martin, T.G., CRediT authorship contribution statement 2012. Prioritizing threat management for biodiversity conservation. Conserv. Lett. 5 (3), 196–204. https://doi.org/10.1111/j.1755-263X.2012.00228.x. V.C.: conceptualization, data collection (threat exposure data, Cochrane, M.A., 2003. Fire science for rainforests. Nature 421 (6926), 913–919. https:// doi.org/10.1038/nature01437. occurrence data, trait data), methodology, formal analysis, writing – Côté, I.M., Darling, E.S., Brown, C.J., 2016. Interactions among ecosystem stressors and original draft; M.E.: data collection (occurrence data, trait data), writing their importance in conservation. Proc. R. Soc. B Biol. Sci. 283 (1824), 1–9. https:// – review and editing, funding acquisition, project administration, su- doi.org/10.1098/rspb.2015.2592. pervision; T.F.: conceptualization, formal analysis, methodology, Dauby, G, Zaiss, R, Blach-Overgaard, A, Catarino, L, Damen, T, Deblauwe, V, Dessein, S, Dransfield, J, Droissart, V, Duarte, MC, 2016. RAINBIO: a mega-database of tropical writing – review and editing; H.G.: conceptualization, data collection African vascular plants distributions. PhytoKeys 74 (1). https://doi.org/10.3897/ (threat exposure data), methodology, writing – review and editing; H. phytokeys.74.9723. W., E.V., K.D.R.: preparation of climatic layers; C.K.: writing – review Dinerstein, E., Olson, D., Joshi, A., Vynne, C., Burgess, N.D., Wikramanayake, E., Hahn, N., Palminteri, S., Hedao, P., Noss, R., Hansen, M., Locke, H., Ellis, E.C., and editing, funding acquisition; H.T.: data collection (occurrence data, Jones, B., Barber, C.V., Hayes, R., Kormos, C., Martin, V., Crist, E., Saleem, M., 2017. trait data); E.T.: conceptualization, methodology, writing – review and An ecoregion-based approach to protecting half the terrestrial realm. Bioscience 67 editing, funding acquisition, supervision. (6), 534–545. https://doi.org/10.1093/biosci/bix014. Dosio, A., Jury, M.W., Almazroui, M., Ashfaq, M., Diallo, I., Engelbrecht, F.A., Klutse, N. A.B., Lennard, C., Pinto, I., Sylla, M.B., Tamoffo, A.T., 2021. Projected future daily Data availability statement characteristics of African precipitation based on global (CMIP5, CMIP6) and regional (CORDEX, CORDEX-CORE) climate models. Clim. Dyn. 57 (11–12), 3135–3158. https://doi.org/10.1007/s00382-021-05859-w. We stored the data in the figshare platform and this is available at Edwards, D.P., Sloan, S., Weng, L., Dirks, P., Sayer, J., Laurance, W.F., 2014. Mining and https://doi.org/10.6084/m9.figshare.19635996. the African environment. Conserv. Lett. 7 (3), 302–311. https://doi.org/10.1111/ conl.12076. Eklundh, L., Olsson, L., 2003. Vegetation index trends for the African Sahel 1982–1999. Declaration of competing interest Geophys. Res. Lett. 30 (8), 1–4. https://doi.org/10.1029/2002GL016772. Feintrenie, L., 2014. Agro-industrial plantations in Central Africa, risks and The authors have no conflict of interests to declare. opportunities. Biodivers. Conserv. 23 (6), 1577–1589. https://doi.org/10.1007/ s10531-014-0687-5. Franzel, S., Akinnifesi, F.K., Ham, C., 2007. Setting priorities among indigenous fruit tree Acknowledgments species in Africa: examples from Southern, Eastern and Western Africa regions. In: Indigenous Fruit Trees in the Tropics: Domestication, Utillization And This work was supported by the European Centre for Medium-Range Commercialization. CAB International, in association with the World Agroforestry Centre (ICRAF), pp. 1–27. https://doi.org/10.1079/9781845931100.0001. Weather Forecasts and from the Forest Trees and Agroforestry program Fremout, T., Thomas, E., Gaisberger, H., Van Meerbeek, K., Muenchow, J., Briers, S., of CGIAR. Gutierrez-Miranda, C.E., Marcelo-Peña, J.L., Kindt, R., Atkinson, R., 2020. Mapping tree species vulnerability to multiple threats as a guide to restoration and conservation of tropical dry forests. Glob. Chang. Biol. 26 (6), 3552–3568. https:// Appendix A. Supplementary data doi.org/10.1111/gcb.15028. Fremout, T., Thomas, E., Taedoumg, H., Briers, S., Gutiérrez-Miranda, C.E., Alcázar- Supplementary data to this article can be found online at https://doi. Caicedo, C., Lindau, A., Hubert, M.K., Vinceti, B., Kettle, C., Ekué, M., Atkinson, R., Jalonen, R., Gaisberger, H., Elliott, S., Brechbühler, E., Ceccarelli, V., Krishnan, S., org/10.1016/j.biocon.2022.109554. Vacik, H., Muys, B., 2021. Diversity for Restoration (D4R): guiding the selection of tree species and seed selection for climate-resilient restoration of tropical forest landscapes. J. Appl. Ecol. 59 (3), 664–679. https://doi.org/10.1111/1365- 2664.14079. Gaisberger, H., Fremout, T., Kettle, C.J., Vinceti, B., Kemalasari, D., Kanchanarak, T., Thomas, E., Serra-Diaz, J.M., Svenning, J.C., Slik, F., Eiadthong, W., Palanisamy, K., 8 V. Ceccarelli et al. B i o l o g ic a l C o n s e r v a t i o n 270 (2022) 109554 Ravikanth, G., Bodos, V., Sang, J., Warrier, R.R., Wee, A.K.S, Elloran, C., Ramos, L. Mansourian, S., Berrahmouni, N., 2021. In: Review of Forest And Landscape Restoration T., Henry, M., Hossain, M.A., Theilade, I., Laegaard, S., Bandara, K.M.A., in Africa. https://doi.org/10.4060/cb6111en. Weerasinghe, D.P., Changtragoon, S., Yuskianti, V., Wilkie, P., Nghia, N.H., Mateo, R.G., Croat, T.B., Felicísimo, Á.M., Munoz, J., 2010. Profile or group Elliott, S., Pakkad, G., Tiansawat, P., Maycock, C., Bounithiphonh, C., Mohamed, R., discriminative techniques? Generating reliable species distribution models using Nazre, M., Siddiqui, B.N., Lee, S.L., Lee, C.T., Zakaria, N.F., Hartvig, I., Lehmann, L., pseudo-absences and target-group absences from natural history collections. Divers. David, D.B.D., Lillesø, J.P.B., Phourin, C., Yongqi, Z., Ping, H., Volkaert, H.A., Distrib. 16 (1), 84–94. https://doi.org/10.1111/j.1472-4642.2009.00617.x. Graudal, L., Hamidi, A., Thea, S., Sreng, S., Boshier, D., Tolentino, E., Ratnam, W., Mirzabaev, A., Sacandé, M., Motlagh, F., Shyrokaya, A., Martucci, A., 2021. Economic Aung, M.M., Galante, M., Isa, S.F.M., Dung, N.Q., Hoa, T.T., Le, T.C., Miah, M.D., efficiency and targeting of the African Great Green Wall. Nat.Sustain. https://doi. Zuhry, A.L.M., Alawathugoda, D., Azman, A., Pushpakumara, G., Sumedi, N., org/10.1038/s41893-021-00801-8. Siregar, I.Z., Nak, H.K., Linsky, J., Barstow, M., Koh, L.P., Jalonen, R., 2021. Tropical Onana, J.M., Cheek, M., 2011. Red data book: the flowering plants of Cameroon. In: and subtropical Asia’s valued tree species under threat. Conserv. Biol. e13873 IUCN Global Assessments. Kew Publishing, Kew Royal Botanic Gardens. https://doi.org/10.1111/cobi.13873. Ordway, E.M., Asner, G.P., Lambin, E.F., 2017. Deforestation risk due to commodity crop Gaisberger, H., Kindt, R., Loo, J., Schmidt, M., Bognounou, F., Da, S.S., Diallo, O.B., expansion in sub-Saharan Africa. Environ. Res. Lett. 12 (4), 044015 https://doi.org/ Ganaba, S., Gnoumou, A., Lompo, D., 2017. Spatially explicit multi-threat 10.1088/1748-9326/aa6509. assessment of food tree species in Burkina Faso: a fine-scale approach. PLoS ONE 12 Phillips, S.J., Dudík, M., Elith, J., Graham, C.H., Lehmann, A., Leathwick, J., Ferrier, S., (9), e0184457. https://doi.org/10.1371/journal. pone.0184457. 2009. Sample selection bias and presence-only distribution models: implications for Gerland, P., Raftery, A.E., Ševčíková, H., Li, N., Gu, D., Spoorenberg, T., Alkema, L., background and pseudo-absence data. Ecol. Appl. 19 (1), 181–197. https://doi.org/ Fosdick, B.K., Chunn, J., Lalic, N., Bay, G., Buettner, T., Heilig, G.K., Wilmoth, J., 10.1890/07-2153.1. 2014. World population stabilization unlikely this century. Science 346 (6206), Réjou-Méchain, M., Mortier, F., Bastin, J.-F., Cornu, G., Barbier, N., Bayol, N., 234–237. https://doi.org/10.1126/science.1257469. Bénédet, F., Bry, X., Dauby, G., Deblauwe, V., Doucet, J.-L., Doumenge, C., Gillson, L., Dawson, T.P., Jack, S., McGeoch, M.A., 2013. Accommodating climate Fayolle, A., Garcia, C., Kibambe, J.-P., Gourlet-Fleury, S., 2021. Unveiling african change contingencies in conservation strategy. Trends Ecol. Evol. 28 (3), 135–142. rainforest composition and vulnerability to global change. Nature 593 (7857), https://doi.org/10.1016/j.tree.2012.10.008. 90–94. https://doi.org/10.1038/s41586-021-03483-6. Gomes, V.H.F., Vieira, I.C.G., Salomão, R.P., ter Steege, H., 2019. Amazonian tree species Rimlinger, A., Avana, M.L., Awono, A., Chakocha, A., Gakwavu, A., Lemoine, T., threatened by deforestation and climate change. Nat. Clim. Chang. 9 (7), 547–553. Marie, L., Mboujda, F., Vigouroux, Y., Johnson, V., Vinceti, B., Carrière, S.M., https://doi.org/10.1038/s41558-019-0500-2 Amazonian. Duminil, J., 2021. Trees and their seed networks: the social dynamics of urban fruit Gonzalez, P., Tucker, C.J., Sy, H., 2012. Tree density and species decline in the African trees and implications for genetic diversity. PLoS ONE 16 (3), e0243017. https://doi. Sahel attributable to climate. J. Arid Environ. 78, 55–64. https://doi.org/10.1016/j. org/10.1371/journal.pone.0243017. jaridenv.2011.11.001. Sacandé, M., Berrahmouni, N., 2016. Community participation and ecological criteria for Grace, J., José, J.S., Meir, P., Miranda, H.S., Montes, R.A., 2006. Productivity and carbon selecting species and restoring natural capital with native species in the Sahel. fluxes of tropical savannas. J. Biogeogr. 33 (3), 387–400. https://doi.org/10.1111/ Restor. Ecol. 24 (4), 479–488. https://doi.org/10.1111/rec.12337. j.1365-2699.2005.01448.x. Sacandé, M., Pritchard, H.W., 2004. Seed research network on African trees for Hengl, T., de Jesus, J.M., Heuvelink, G.B.M., Gonzalez, M.R., Kilibarda, M., Blagotić, A., conservation and sustainable use. For. Genet. Resour. 31, 31–35 ftp://ftp.fao.org/ Shangguan, W., Wright, M.N., Geng, X., Bauer-Marschallinger, B., 2017. docrep/fao/008/y5901e/y5901e00.pdf. SoilGrids250m: global gridded soil information based on machine learning. PLoS Tölgyesi, C., Buisson, E., Helm, A., Temperton, V.M., Török, P., 2021. Urgent need for ONE 12 (2), e0169748. https://doi.org/10.1371/journal.pone.0169748. updating the slogan of global climate actions from ‘tree planting’ to ‘restore native Herrmann, S.M., Anyamba, A., Tucker, C.J., 2005. Recent trends in vegetation dynamics vegetation’. Restor. Ecol., e13594 https://doi.org/10.1111/rec.13594. in the African Sahel and their relationship to climate. Glob. Environ. Chang. 15 (4), Triviño, M., Kujala, H., Araújo, M.B., Cabeza, M., 2018. Planning for the future: 394–404. https://doi.org/10.1016/j.gloenvcha.2005.08.004. identifying conservation priority areas for Iberian birds under climate change. Herrmann, S.M., Brandt, M., Rasmussen, K., Fensholt, R., 2020. Accelerating land cover Lanscape Ecology 659–673. https://doi.org/10.1007/s10980-018-0626-z. change in West Africa over four decades as population pressure increased. Commun. Tyukavina, A., Hansen, M.C., Potapov, P., Parker, D., Okpa, C., Stehman, S.V., EarthEnviron. 1 (1), 1–10. https://doi.org/10.1038/s43247-020-00053-y. Kommareddy, I., Turubanova, S., 2018. Congo Basin forest loss dominated by Herrmann, S.M., Tappan, G.G., 2013. Vegetation impoverishment despite greening: a increasing smallholder clearing. Sci. Adv. 4 (11), eaat2993 https://doi.org/10.1126/ case study from central Senegal. J. Arid Environ. 90, 55–66. https://doi.org/ sciadv.aat2993. 10.1016/j.jaridenv.2012.10.020. Valavi, R., Elith, J., Lahoz-Monfort, J.J., Guillera-Arroita, G., 2019. Block CV: an R IUCN, 2021. The IUCN Red List of Threatened Species. Version 2021-3. Accessed on 10 package for generating spatially or environmentally separated folds for k-fold cross- December. www.iucnredlist.org. validation of species distribution models. Methods Ecol. Evol. 10 (2), 225–232. Jaenicke, H., Franzel, S., Boland, D.J., 1995. Towards a method to set priorities amongst https://doi.org/10.1111/2041-210X.13107. species for tree improvement research - a case study from West Africa. J. Trop. For. Veldman, J.W., Aleman, J.C., Alvarado, S.T., Anderson, T.M., Archibald, S., Bond, W.J., Sci. 7 (3), 490–506 https://www.jstor.org/stable/43581842. https://www.jstor. Boutton, T.W., Buchmann, N., Buisson, E., Canadell, J.G., de Sá Dechoum, M., Diaz- org/stable/43581842. Toribio, M.H., Durigan, G., Ewel, J.J., Fernandes, G.W., Fidelis, A., Fleischman, F., Kindt, R., 2018. Ensemble species distribution modelling with transformed suitability Good, S.P., Griffith, D.M., Zaloumis, N.P., 2019. Comment on “The global tree values. Environ. Model Softw. 100, 136–145. https://doi.org/10.1016/j. restoration potential”. Science 366 (6469), 1–5. https://doi.org/10.1126/science. envsoft.2017.11.009. aaz0111. Kramer-Schadt, S., Niedballa, J., Pilgrim, J.D., Schröder, B., Lindenborn, J., Veldman, J.W., Buisson, E., Durigan, G., Fernandes, G.W., Le Stradic, S., Mahy, G., Reinfelder, V., Stillfried, M., Heckmann, I., Scharf, A.K., Augeri, D.M., 2013. The Negreiros, D., Overbeck, G.E., Veldman, R.G., Zaloumis, N.P., Putz, F.E., Bond, W.J., importance of correcting for sampling bias in MaxEnt species distribution models. 2015. Toward an old-growth concept for grasslands, savannas, and woodlands. Divers. Distrib. 19 (11), 1366–1379. https://doi.org/10.1111/ddi.12096. Front. Ecol. Environ. 13 (3), 154–162. https://doi.org/10.1890/140270. Lewis, S.L., Sonké, B., Sunderland, T., Begne, S.K., Lopez-Gonzalez, G., van der VITO, 2020. Downscaled bioclimatic indicators for selected regions from 1950 to 2100 Heijden, G.M.F., Phillips, O.L., Affum-Baffoe, K., Baker, T.R., Banin, L., Bastin, J.F., derived from projections. In: Product User Guide. ECMWF, Copernicus Climate Beeckman, H., Boeckx, P., Bogaert, J., De Cannière, C., Chezeaux, E., Clark, C.J., Change Service https://cds.climate.copernicus.eu/cdsapp#!/dataset/sis- Collins, M., Djagbletey, G., Zemagho, L., 2013. Above-ground biomass and structure biodiversity-cmip5-regional?tab=doc. of 260 African tropical forests. Philos.Trans.R.Soc.BBiol.Sci. 368 (1625), 20120295. VITO, 2021. Downscaled bioclimatic indicators for selected regions from 1979 to 2018 https://doi.org/10.1098/rstb.2012.0295. derived from ERA5 reanalysis. In: Product User Guide. ECMWF, Copernicus Climate Malhi, Y., Adu-Bredu, S., Asare, R.A., Lewis, S.L., Mayaux, P., 2013a. African rainforests: Change Service, pp. 1–44 https://cds.climate.copernicus.eu/cdsapp#!/dataset/sis- past, present and future. Philos.Trans.R.Soc.BBiol.Sci. 368 (1625), 20120312. biodiversity-era5-regional?tab=doc. https://doi.org/10.1098/rstb.2012.0312. Willis, K.J., Bennett, K.D., Burrough, S.L., Macias-Fauria, M., Tovar, C., 2013. Malhi, Y., Adu-Bredu, S., Asare, R.A., Lewis, S.L., Mayaux, P., 2013b. The past, present Determining the response of african biota to climate change: using the past to model and future of Africa's rainforests. Philos.Trans.R.Soc.BBiol.Sci. 368 (1625), the future. Philos.Trans.R.Soc.BBiol.Sci. 368 (1625), 1–9. https://doi.org/10.1098/ 20120293. https://doi.org/10.1098/rstb.2012.0293. rstb.2012.0491. 9