RETRACTED ARTICLE: Great ape abundance and per capita carbon storage in their habitats

Abstract

The ecological importance of great apes is widely recognised, yet few studies have highlighted the role of protecting great apes’ habitats in mitigating climate change, particularly through carbon sequestration. This study used GIS tools to extract data from various sources, including the International Union for Conservation of Nature database, to examine carbon quantity and great ape abundance in African great ape habitats. Subsequently, we employed a generalised linear model to assess the relationship between locally measured great ape populations abundance and carbon storage across areas with different levels of protection. Our findings showed a positive relationship between the abundance of great apes in their habitats and carbon storage, likely since conservation efforts in great apes habitats may be strengthened with higher great ape populations. The results reveal that gorilla habitats exhibited higher carbon storage than chimpanzee habitats. Specifically, the areas inhabited by gorillas are associated with a mean increase of 27.47 t/ha in carbon storage. Additionally, we observed a positive association between highly protected areas and carbon storage within great ape habitats. Our model indicates that highly protected areas increase the mean carbon stored by 1.13 t/ha compared to medium protected areas, which show a reduction of 15.49 t/ha. This highlights the critical role that protected areas play in both species conservation and carbon sequestration, contributing significantly to climate mitigation efforts. Furthermore, our study underscores the significant contribution of great ape habitats, extending beyond protected areas, to carbon storage, highlighting the potential for synergistic conservation strategies targeting both great apes and carbon sequestration. Protecting great apes is vital for reducing carbon emissions from deforestation and boosting tropical forest carbon sinks. Since nearly 90% of great apes live outside protected areas, targeted conservation in these low-protected areas is also crucial.

Introduction

Tropical rainforests are well known for their essential roles in carbon sequestration (i.e., process by which carbon dioxide is captured from the atmosphere and stored in living plants biomass, soils, and other organic materials) [1] and biodiversity conservation, also provide vital ecosystem services such as climate regulation, water cycle management, and support for a wide range of species [1,2,3,4,5,6]. Unfortunately, tropical rainforests, which act as vital carbon sinks, are increasingly threatened by human activities driven by rising global production and consumption demands (e.g. agricultural products, timber, and other natural resources etc.), particularly in Africa [7,8,9]. To safeguard these forests, many conservation programmes aim to implement conservation measures, utilising flagship or indicator species like great apes as focal points for protection efforts [2].

The pressures exerted by humans on tropical forests are mainly linked to the growing demand for land and resources, which manifests itself through the direct exploitation of wildlife [3], the expansion of human infrastructure [4, 5], and the exploitation of natural resources [6]. To counteract these threats to carbon stocks, numerous conservation programs are being devised throughout Africa. We may look at conservation efforts to halt or alleviate some of these threats, drawing motivation from partial successes, such as those achieved by projects like Roopsind et al. (2019), which are linked to the Reducing Emissions from Deforestation and Forest Degradation (REDD +) carbon credit market mechanism [14, 15]. Yet, it is often challenging to monitor the impact of these conservation efforts over time. This difficulty arises from factors such as the long-term nature of ecological processes, the need for consistent and reliable data collection, and potential trade-offs between different conservation goals like the protection of carbon sinks, biodiversity, or ecosystem services (e.g., [8, 9]). For example, promoting fast-growing tree species for carbon sequestration may lead to reduced biodiversity if these species outcompete native flora. Similarly, prioritizing carbon storage through forest protection could sometimes limit land availability for local communities, potentially affecting their livelihoods and food security. Additionally, the complex interactions between environmental, social, and economic factors further complicate the assessment of conservation outcomes. One way to determine spatial and temporal trends and links in carbon and biodiversity is to look for biota that may serve as effective indicators of certain elements of forest habitat (e.g., intactness) as well as biodiversity (e.g., species richness) [10]). Among these biota, great apes could be particularly useful indicators, given their potential to reflect forest quality [11,12,13]. They may serve as umbrella species whose presence indirectly benefits other wild species thanks to the conservation efforts made to protect their habitats [2, 14].

Great apes are often found in intact forests with little human disturbance (exhibiting high carbon storage) but also in non-intact forests where they might face high human disturbance [12]. In Africa, 80 to 90% of the actual number of great apes live outside protected areas [15, 16] where it is not easy to initiate and maintain conservation efforts focused on carbon stocks or great ape populations. In this study, we aim to explore the relationship between great ape abundance and carbon storage in their habitats, particularly in the context of varying levels of habitat protection. Specifically, we hypothesize that due to the conservation efforts mobilized by their presence, great ape abundance will be positively correlated with carbon storage in these habitats, and that this relationship will be influenced by the level of protection the areas receive. We predict that areas with higher levels of protection will store more carbon and exhibit lower emission factors, as reduced disturbances and intact habitats typically result in lower carbon release rates, compared to less protected areas where conservation efforts may be less effective. We use area protection level as an approximation but acknowledge that previous studies have shown that protection status does not necessarily represent actual conservation effort [17, 18].

In this context, understanding the nuances of how protection levels impact both great ape abundance and carbon storage can provide crucial insights for conservation policies. The study also seeks to identify the specific factors within protected areas that contribute most effectively to enhanced carbon sequestration and great ape population stability. By establishing a clearer link between great ape abundance, protection levels, and carbon storage, this study aims to contribute to a more holistic understanding of conservation strategies that benefit both biodiversity and climate goals.

We used existing spatial layers (available in raster format) of the estimated densities of western chimpanzee (Pan troglodytes verus), eastern chimpanzee (Pan troglodytes schweinfurthii), central chimpanzee (Pan troglodytes troglodytes), Nigeria-Cameroon chimpanzee (Pan troglodytes ellioti), and western lowland gorilla (Gorilla gorilla gorilla) as available in the IUCN SSC A.P.E.S. Database ([15], (Appendix). We focus on these African great apes as they play an important role in preserving the health and structure of tropical forest ecosystems [19]. Great apes’ ecological function extends beyond their status as iconic species; they are important agent of biodiversity, affecting forest regeneration and composition through their habitats [20]. They play an important role in forest biodiversity and structure by distributing seeds, ingesting diverse plant materials, and interacting with other specie [12]. This interaction between great ape population and forest dynamics emphasizes their role as ecological engineers capable of maintaining biodiversity while also assisting natural forest regeneration processes that improve carbon storage [12, 21]. While many other forest-dependent primates including bonobos (Pan paniscus) and eastern gorilla (Gorilla beringei) also play similar roles in relation to forest disturbance sensitivity [22], we did not include them in our study due to a lack of data. In addition, abundance estimates at a relatively fine spatial scale (e.g., site-level or local population estimates) are available for these two species (partly because primates are easier to census), data that is typically lacking for most other vertebrates, particularly across African forests while.

Methods

Study area

We used great ape abundance data from locations across 19 African great ape range countries, which include: Angola, Burundi, Cameroon, the Central African Republic, the Republic of the Congo, Equatorial Guinea, Gabon, Ghana, the Republic of Guinea, Guinea Bissau, Ivory Coast, Liberia, Mali, Nigeria, Rwanda, Senegal, Sierra Leone, Tanzania, and Uganda (Fig. 1). The study areas are mixture of savannah and rainforest which comprises of protected and non-protected areas.

Fig. 1

Geographic ranges of the African great apes investigated in this study

Data source

We used existing spatial layers (available in raster format) of the estimated densities of six great ape species (western chimpanzee, eastern chimpanzee, central chimpanzee, Nigeria-Cameroon chimpanzee, and western lowland gorilla) as available in the IUCN SSC A.P.E.S. Database ([15], (see, Appendix for more description).

We used previously published carbon data from a study that combined 2007–2008 ground measurements (measurements of stem diameter), GLAS (Geoscience Laser Altimeter System) LiDAR data, and 500-m resolution MODIS (Moderate Resolution Imaging Spectrometer) imagery, to estimate of carbon density across the tropics [23].

To understand the level of protection of an area, we used the World Database of Protected Areas [24]. The shapefile we obtained comprises various types of protected areas, which we categorised into three levels of protection: high, medium, and low (Table 1). High-level protection denotes areas with stringent legal restrictions on human access to resources. Medium-level protection includes areas where human access is permitted, but resource extraction is regulated. Low-level protected areas include those not classified in the World Database of Protected Areas and lack controlled access or resource use.

Table 1 Classification of the areas by protection level

To understand the level of human pressure in our study area, we used a 1993–2009 world human footprint raster map (1 km x 1 km resolution) [25].

To understand the types of areas that were in our study area, we used a Land Cover Map of Africa at 20 m resolution based on 1 year of Sentinel-2A observations from December 2015 to December 2016 [26]. A Clear description (sources, spatial resolution, layers names, authors and years) of variables cited above are included in the Appendix).

Data extraction

Great apes’ abundance raster data were converted into point using raster pixels to point tool in QGIS (QGIS, 2021) where each raster cell was represented by its center coordinates. Then, using point sampling tool plugin in QGIS we extracted the protection status for each cell by overlying the great ape abundance point layer with the shapefile derived from the World Database on Protected Areas 2023 [24]. Given that great ape abundance points were located at the center of grid cells (approximately 10 km² for Nigeria-Cameroon and eastern chimpanzees, and about 1 km² for western chimpanzees, central chimpanzees, and lowland gorillas), a buffering step was performed prior to data extraction of carbon, human footprints and land cover (for the great apes abundance data that come from raster layer of 1 km² we provided to the point a buffer of 400 m around the point so that the surface can approximately equal to the size of the original raster cell and we provided a buffer of 4000 m around the point that come from great apes abundance data from raster cell that is equal to 10 km²). The buffered points were used to extract carbon quantity and human footprint using Zonal statistics tool in QGIS and with this tool we extracted the mean of carbon quantity and average human footprint corresponding to each buffer zone around each point in the great ape abundance layer. Because we wanted to estimate the proportion of each land cover in each raster cell of great ape abundance, we used the same buffered layer of great ape abundance points and by using Zonal Histogram function in QGIS we extracted the proportion (estimated in ton carbon (tC)) of each land cover for each buffered point.

Quantifying carbon dioxide emissions from above-ground carbon in great ape habitats

To estimate carbon dioxide (CO₂) emission, we applied a molecular calculation [27]. A CO₂ molecule consists of one carbon (C) atom and two oxygen (O) atoms. Given that the atomic weight of C is 12 and the atomic weight of O is 16, the total molecular weight of a CO₂ molecule is 44 (12 from C and 32 from O). Thus, the ratio of C to CO₂ is 44/12 [28, 29]. To calculate CO₂ emissions, we multiplied the quantity of carbon by this ratio, 44/12.

Data analysis

We conducted a Generalized Linear Model (GLM with gaussian family run in R v.4.3.2 [30]) analysis to investigate the relationship between carbon quantity and great ape abundance, human footprint, species diversity and other ecological variables used as a control predictor with carbon quantity as the response variable. The analysis involved two models: a full model including all predictors and interactions, and a reduced model for comparison, with detailed steps provided below. In these models, we used the mean carbon quantity as a response variable and a series of key predictors, including numerical predictors such as great ape abundance, human footprint and, factors predictors such as great apes diversity (i.e., genus Gorilla or genus Pan, hereafter chimpanzees in the model), and protection levels with chimpanzees and low protection used as the reference level in the model. We included control predictors such as tree cover, shrub cover, grassland, cropland, and aquatic or regularly flooded vegetation to account (obtained from land cover map of Africa, 20 m resolution based on 1 year of Sentinel-2A observations from December 2015 to December 2016 as described in Appendix) for the influence of environmental gradients on carbon storage.

Prior to the analysis, human footprints and other values exceeding a score index of 50 were discarded from the analysis as they fell in open water or outside the expected range (0–50) [25] and were considered measurement errors. Similarly, to maintain data accuracy, we also discarded great ape density estimates beyond 10 individuals per square kilometre (inds/km²), as higher values were likely model misspecifications. These values exceed known ecological limits for great ape populations, where typical densities rarely surpass this threshold (see [31, 32]). We also log-transformed human footprint and great ape density to achieve an approximately symmetrical distribution and to avoid potentially influential cases. To avoid convergence issues, we then z-transformed to a mean of zero and a standard deviation of one all keys and controlled predictors [33]. We looked for Pearson correlations between covariates to avoid multi-collinearity in fitting the models, using a threshold of 0.7, above which covariates were considered highly correlated and potentially problematic [34]. Once key assumptions of the GLM model were met, we explored the relationship between carbon and predictors described above by computing two models described as follows: (1) the full model, which includes all the predictors and the interaction between great ape’s abundance and human footprint and (2) the null model, also called the reduced model, lacking the interaction but everything else (see Supplementary file 1 for description of each model).

We tested if there was a significant interaction between the factor and the covariate by comparing the fit of the full model to a reduced model (called null model) using chi-square tests which reveal that interaction model is not significant (p < 0.05). We checked model stability using DFBetas, which revealed the null model to be of good stability (see DFBetas values in Table 2). Also, we assess the Generalized Variance Inflation Factors (GVIF) [35] with a threshold set at 3 using the function ‘vif’ of the package ‘car’ (version 3.0–13 [36]) and this revealed no collinearity issues (maximum VIF: 1.37 [37]). Similarly, we check for linearity, homoscedasticity and normality of residuals stability and influentials points. Linearity was assessed through visual inspection of residuals versus fitted values plots. Homoscedasticity was evaluated by checking for constant variance in residuals, and normality of residuals was verified using Q-Q plots while influentials points was assessed by calculating Cook’s Distance, leverage (hat values). To assess the distribution of carbon storage across great ape species and different protection levels, a Kolmogorov–Smirnov test for normality was performed on the carbon mean data for each species and each protection level (high, medium, and low protection levels). The test indicated significant deviations from normality for all levels (p-values < 0.001). Given the large sample size and the independence of observations, non-parametric methods were used for further analysis. The Wilcoxon (W) rank-sum test (also known as the Mann–Whitney U test) was conducted to compare the mean carbon storage between gorilla and chimpanzee habitats and a Kruskal–Wallis rank sum test was conducted on the carbon mean data stratified by protection level.

Table 2 Results of the model with carbon mean as the response (estimates (B) together with, standard errors (SE), significance tests, confidence limits (conf) and range of estimates)

Results

Overall, the analysis of deviance performed to compare the fit of the two models predicting carbon mean suggests that the difference in deviance between the two models was not statistically significant (full-null model comparison χ² = -2695.8, df = 1, p = 0.490), indicating that the interactions do not have a significant effect on carbon storage prediction. Therefore, the null model can provide a comparable fit to the data compared to the full model.

Relationship between carbon quantity and covariates

Across great ape habitats, carbon quantity varied between 0 and 440 tonnes per hectare (t/ha), with a mean of 197 ± 108.1 (t/ha). The null model indicated a positive association between the mean of carbon quantity and great ape abundance (B: 4.27, SE: 0.06, CI [4.15–4.39], p < 0.001; Table 2, Fig. 2), suggesting that areas with a higher abundance of great apes tend to have higher amounts of stored carbon (i.e., 190.69 + 4.28 = 194.97 t/ha).

Fig. 2

Relationship between the great ape’s abundance and carbon stock. Legend: The area of the circle corresponds to the number of data points. All points are binned data points for a fraction on the x and y axis with the mean per fraction. The point is bigger as more data points fall into this fraction. The dashed line depicts the fitted model

The finding of fitted values revealed a strong association between gorilla occurence and carbon storage, with the model suggesting that areas inhabited by gorillas are associated with a mean increase of 27.47 t/ha in carbon stored (i.e., 190.69 + 27.48 = 218.17 t/ha;B: 27.47, SE: 0.12, CI [27.23–27.70], p < 0.001; Table 2) compared to areas inhabited by chimpanzees where the mean quantity of carbon is 190.69 t/ha. However according to carbon raw data, the mean amount of carbon stored in the habitat inhabited by gorilla (243.1507 t/ha) is higher than the mean amount of carbon in the habitat inhabited by chimpanzee (174.0856 t/ha) and the effect of species is statistically significant (W = 7.3745e + 11, p < 0.001, Fig. 3) suggesting that carbon levels vary significantly between these two great ape species habitats.

Fig. 3

Variation of mean carbon quantity according to the great ape’s species (Based on Raw Data). Legend: For each species, boxes show the median, upper value, lower value 25th and 75th percentile

We found significant differences in the mean of carbon values across the different protection levels area (Kruskal–Wallis chi-squared = 13.389, df = 2, p < 0.001, Fig. 4). Based on fitted values the model suggests a positive association between high-protected area and the amount of carbon stored (B: 1.13, SE: 0.19, CI [0.77—1.50], p < 0.001). Specifically, compared to low protected areas which have an average carbon quantity of 190.69 t/ha, highly protected aeras are associated with a mean increase of 1.13 t/ha in carbon storage (ie. 190.69 + 1.13 = 191.82 t/ha) while medium protected areas are associated with a mean reduction of 15.49 t/ha (ie. 190.69–15.49 = 175.2 t/ha ; B: -15.49, SE: 0.19, CI [-15.78 -15.20], p < 0.001; Table 2.

Fig. 4

Variation of carbon quantity according to the level of protection (Based on Raw Data). Legend: LP, low-protected area, MP, medium protected area, HP, high-protected area. For each species, boxes show the median, upper value, lower value, 25th and 75th percentile

The mean of carbon quantity was negatively correlated to human footprint (B: -25, SE: 0.06, CI [-25.12–24.89], p < 0.001, Table 2, Fig. 5), suggesting that an increase of one unit of human footprint corresponds to a decrease of carbon amount by 25.

Fig. 5

Relationship between the carbon stock and human footprint. Legend: The area of the circle corresponds to the number of data points. All points are binned data points for a fraction on the x and y axis with the mean per fraction. The point is bigger when more data points fall into this fraction. The dashed line depicts the fitted model

Link between great ape abundance and aboveground carbon stock

The aboveground carbon stock showed considerable spatial variation. Independently of the type of protected area, area inhabited by chimpanzees store 20246.66 tC per capita which was lower than the one stored in area inhabited by gorillas, which is 25178.13 tC per capita (Table 3). However, highly protected areas such as national parks store a large amount of aboveground carbon (20154.55 tC) independent of the type of great ape species living in that area. The dashed line depicts the fitted model.

Table 3 Link between per capita aboveground carbon stock and area protection level

Link between great ape abundance and mean carbon emission

The emission factors are defined here as the estimated mean amount of CO₂ released in the atmosphere due to the loss of one great ape within a habitat that has a specific protection status. When considering great ape species living within a highly protected area, the emission factor is 19066.70 tCO2 for the loss of one chimpanzee, rising to 22973.57 tCO2 for the loss of one gorilla (Table 4). Compared to an area with a low protection status, the emission factor is 26557.41 tCO2 for chimpanzees, 47342.61 tCO2 for gorillas.

Table 4 Link between great apes’ abundance and mean aboveground carbon emission

For more details about the variation of above ground carbon and emission factor per great apes and countries see Supplementary files 2, 3 and 4.

Discussion

We found a strong positive correlation between the mean carbon storage in tropical forests and great ape abundance (Fig. 2), highlighting the role of African great apes as key indicators of forest quality and umbrella species [2, 11,12,13]. Furthermore, our study revealed that great apes can also serve as a proxy of the level of protection and therefore the amount of carbon stored within their habitat as an increase in the number of great ape individuals significantly boosted the amount of carbon sequestered. This is likely because the presence of great apes often triggers enhanced conservation efforts, which in turn help preserve the stability of these carbon sinks. This finding supports our hypothesis that the abundance of great apes is positively correlated with the amount of carbon that can be sequestered by their habitat. We do not necessarily suggest that great ape’s abundance is a direct causal factor in carbon storage, instead, we emphasize the essential role of protecting great ape and their habitats in preserving carbon stocks within tropical forests. Given that great ape-rich habitat also tends to harbour significant carbon reserves, focusing attention on this co-benefit in conservation efforts, for example, through projects associated with the REDD + carbon credit mechanism, can make a substantial contribution to climate change mitigation. The fact that habitats with gorillas store more carbon than habitat with chimpanzees (Fig. 3) may stem from the fact that the former are primarily found in dense tropical rainforests [38, 39], whereas chimpanzees are known to live in rainforests and savannahs [40, 41]. This is in accordance with previous findings showing that higher tree cover and sizes increases carbon sequestration capacity [42,43,44].

Our study reveals that the level of protection in an area significantly influences the amount of carbon stored in great apes’ habitats (Fig. 4). On one hand, we observed a positive relationship between highly protected habitats (i.e., habitats located in areas that meet the IUCN Category I and II standards, such as Natural Parks, Integral Nature Reserves, National Forest Parks, National Parks, Presidential Reserves, and Strict Nature Reserves) and the amount of stored carbon. These findings align with general expectations, as highly protected areas are better shielded from intrusions [45] and deforestation [46] maintaining relatively intact forest habitats, resulting in an increase in carbon storage. On the other hand, we found that medium-protected areas store significantly less carbon compared to the reference category of low-protected areas, suggesting that the protection status of an area alone may not be a reliable indicator of the amount of carbon stored within a specific area. This challenges conventional wisdom regarding the association between conservation status and carbon storage. The apparent contradiction might arise from medium protected areas being designated in response to threats, potentially resulting in protective measures implemented rather reactively [47]. Many of medium-protected areas, such as classified forests, game reserves, hunting areas, and hunting reserves, are designated for protection but not in a strict manner. These areas permit human activities, such as timber extraction and hunting, to a certain extent but humans frequently exceed their permitted access. Consequently, it becomes challenging to differentiate between legal and illegal activities in medium protected areas making them difficult to manage effectively [48, 49]. Forests within these medium protection levels may exhibit less efficient carbon storage, primarily because of poorly controlled exploitative activities. Furthermore, some protected areas (including highly protected aeras) might exist as “paper parks,” lacking effective measures for resource safeguarding despite legal protection [50]. Indeed, it is more the conservation efforts undertaken in areas inhabited by great apes that favour the maintenance of carbon stocks, rather than the protection status itself [51]. These efforts, which are generally intensified with the abundance of great apes, make this variable an effective proxy for carbon stocks. Thus, the abundance of great apes can be safely used as a surrogate for ‘conservation effort’ in this context.

The relationship between the protection level of areas and carbon storage is reinforced by our examination of carbon storage per great ape in 1 km² across their ranges. As a result, the presence of great apes might contribute to forest regeneration, even in areas with a low level of protection; however, caution is warranted, as the presence of great apes does not necessarily lead to more effective carbon storage in unprotected areas. Areas lacking proper protection may face significant hunting pressure; thus, such areas need special attention, particularly for great apes, of which 80–90% live outside protected areas [15, 16, 52]. A strategic approach to forest carbon protection from outside protected areas will bring additional improvements. Specifically, we advocate for the creation and preservation of wildlife corridors that, by their characteristics [53], connect protected areas and offer a tangible and applied approach to simultaneously safeguarding great apes and preserving biodiversity, which can maximise the carbon sequestration potential of tropical forests. At the same time, great apes have already been exposed to climate change impacts, and these impacts will likely be exacerbated in the future [54]. Wildlife corridors could support great apes and sympatric biodiversity to adapt to climate change impacts by facilitating dispersal. This integrated conservation approach not only addresses immediate threats to great ape populations but also aligns with broader climate change adaptation efforts, making it a valuable tool for sustainable forest management and global environmental health.

Putting our study results into perspective, according to the United States Environmental Protection Agency [24] a typical passenger vehicle releases on average about 4.6 metric tonnes of CO2 per year meaning that losing the habitat equivalent to one chimpanzee could be translated into using 5379 passenger cars per year, and similarly, losing one gorilla could be translated into using 6689 passenger cars per year (refer to Table 4 for the values uses for the calculations). Such heightened human activities could lead to habitat disturbance and, to some extent, habitat destruction, subsequently contributing to higher CO2 emissions. Our results align with this insight, showing that highly protected areas have lower carbon emission factors than medium and low-protected areas (Table 4) which face more degrading human activities, like large-scale habitat conversion and land-use planning. Conversely, highly protected areas, with less disturbance from human activities, play a crucial role in controlling deforestation [55,56,57]. Effective forest management policies can mitigate the impacts of human activity on carbon emissions. It is imperative to develop strategies to mitigate CO2 emissions, especially considering that these emissions are even more detrimental to biodiversity. Indeed, it is estimated that 2.3 × 10^-7 species are at risk of inevitable extinction per tCO2e emitted, meaning that with each emission of 4.3 MtCO2e, a species could disappear [58]. Protecting great apes in tropical forests, both inside and outside protected areas, may contribute to mitigating carbon emissions resulting from deforestation and human activities while expanding the carbon sink capacity of these areas [56, 59].

In summary, our study underscores the crucial link between great ape abundance and carbon storage in tropical forests, highlighting the importance of protecting great ape habitats for effective climate change mitigation. Carbon levels varied with great apes’ habitat and species, favouring gorilla habitats. Contrary to highly protected areas, medium protected areas show lower carbon levels, challenging prevailing assumptions, possibly due to the weak management of those areas and the presence of ineffective “paper parks.” Our analysis suggests that even in areas with lower protection levels, the abundance of great ape can lead to conservation efforts that help safeguard carbon sinks. However, caution is needed due to hunting and other human pressure in unprotected areas. Targeted conservation outside protected areas is crucial, considering nearly 90% of great apes live outside protected areas. Our results support the idea that protecting great apes helps mitigate carbon emissions from deforestation, thereby enhancing the carbon sink capacity of tropical forests. The practical application of our findings extends beyond protected areas, advocating for wildlife corridors connecting these zones to not only safeguard great ape populations and biodiversity but also contribute to effective climate change mitigation and, offering a concrete strategy for sustainable forest management on a global scale. We encourage conservation programmes such as those linked to the REDD + carbon credit market mechanism to prioritise the joint preservation of great apes and carbon stocks. Considering the association between great ape abundance and carbon storage while designing conservation strategies offers the opportunity to maximise conservation efforts and simultaneously mitigate climate change.

Appendix

Description of variables included in the analysis and their sources

Category	Variables	Layer name	Description	Year	Spatial resolution	Sources	Data Limitations
Carbon	Carbon quantity	Vegetation carbon stock	The layer characterizes the global distribution of above-ground biomass in the tropics. The study used multi-sensor satellite data to estimate above ground live woody vegetation carbon density for pan-tropical ecosystems	2013	500 m x 500 m	Baccini et al., 2012 https://www.nature.com/articles/nclimate1354
Great ape	Great ape abundance	Western chimpanzee density distribution	Modeled density distribution of western chimpanzees based on 52 nest count datasets from eight countries in West Africa	2015	1 km x 1 km	Heinicke et al., 2019 https://iopscience.iop.org/article/10.1088/1748-9326/ab1379
		Western lowland gorilla and Central chimpanzee density distribution	Modeled density distribution of chimpanzees and gorillas in Western Equatorial Africa (WEA), based on field survey data from 59 sites surveyed between 2003 and 2013, in five countries (Cameroon, Gabon, CAR, Congo Rep. and Equatorial Guinea)	2013	1 km x1km	Strindberg et al., 2018 https://www.science.org/doi/10.1126/sciadv.aar2964	The layer is not recommended for predicting great ape abundance in areas of small sizes
		African great ape density distribution	Modeled density distribution of African great apes based on great ape abundance estimates from 156 sites in 18 countries	2015	85.4 km2	Ordaz-Németh et al., 2022 https://doiorg.publicaciones.saludcastillayleon.es/10.1002/ajp.2338	Density predictions for DRC were very high leading the authors to remove this country from the predictions
	Taxon/subspecies range	IUCN Taxon range	Layer shows the geographic distribution of each great ape species, e.g. where great apes were sighted or where their presence is confirmed through indirect signs (sleeping nests, feces, vocalization, footprint, etc.)	2017–2018	Polygon	IUCN Red List (2023) https://www.iucnredlist.org/
Contextual layers	Country	Africa—Admin Level 0	Layer show the surface and the boundaries of each country	2013	Polygon	ICPAC GeoPortal (http://geoportal.icpac.net/layers/geonode%3Aafr_g2014_2013_0)
	Level of protection	World database on protected areas (WDPA)	Presents information on all protected areas. Protected areas are classified into various categories (National park, Forest reserve, Biosphere reserve, Hunting reserve, Community forests, etc.)	2023	Polygon	WDPA (protectedplanet.net) https://www.protectedplanet.net/en/thematic-areas/wdpa?tab=WDPA
	Habitat type	S2 prototype LC 20 m map of Africa 2016	Land Cover map of Africa at 20 m resolution based on 1 year of Sentinel-2A observations from December 2015 to December 2016. The layer comprises 10 generic classes: trees cover areas, shrubs cover areas, grassland, cropland, vegetation aquatic or regularly flooded, lichen and mosses / sparse vegetation, bare areas, built-up areas, snow and/or ice and open water	2016	20 m x 20 m	ESA, 2016 https://2016africalandcover20m.esrin.esa.int/
	Human footprint	Last of the Wild Project, Version 3 (LWP-3): 2009 Human Footprint, 2018 Release	This raster map used data that were taken in 1993 and 2009 on eight variables that can measure the direct and indirect human pressure on environments. The variables that were used are the following: (1) extent of built environments; (2) crop land; (3) pasture land; (4) human population density; (5) night-time lights; (6) railways; (7) roads; and (8) navigable waterways. These pressures were weighted and given score indexes according to estimates of their relative levels of human pressure following Sanderson et al., 2002. The score was between 0 to 50 where 0 means no human pressure and 50 means very high human pressure	1993–2009	1 km x 1 km	Venter et al. 2018 https://sedac.ciesin.columbia.edu/data/set/wildareas-v3-2009-human-footprint/data-download Venter et al. 2016 https://doiorg.publicaciones.saludcastillayleon.es/10.1038/ncomms12558