Next Article in Journal
Flame-Retardant Polyurea Coatings: Mechanisms, Strategies, and Multifunctional Enhancements
Previous Article in Journal
Identifying Critical Fire Risk Transmission Paths in Subway Stations: A PSR–DEMATEL–ISM Approach
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Fire
Volume 8
Issue 8
10.3390/fire8080333
fire-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Wildfires and Climate Change as Key Drivers of Forest Carbon Flux Variations in Africa over the Past Two Decades

by
Lianglin Zhang
1,2 and
Zhenke Zhang
1,2,*
1
School of Geography and Ocean Science, Nanjing University, Nanjing 210023, China
2
Institute of African Studies, Nanjing University, Nanjing 210023, China
*
Author to whom correspondence should be addressed.
Fire 2025, 8(8), 333; https://doi.org/10.3390/fire8080333
Submission received: 23 July 2025 / Revised: 7 August 2025 / Accepted: 18 August 2025 / Published: 20 August 2025
Browse Figures
Versions Notes

Abstract

Forests play a vital role in the global carbon cycle; however, the carbon sink capacity of African forests is increasingly threatened by wildfires, rising temperatures, and ecological degradation. This study analyzes the spatiotemporal dynamics of forest carbon fluxes across Africa from 2001 to 2023, based on multi-source remote sensing and climate datasets. The results show that wildfires have significantly disrupted Africa’s carbon balance over the past two decades. From 2001 to 2023, fire activity was most intense in the woodland–savanna transition zones of Central and Southern Africa. In countries such as the Democratic Republic of the Congo, Angola, Mozambique, and Zambia, each recorded burned areas exceeding 500,000 km2, along with high recurrence rates (e.g., up to 0.7584 fires per year in South Sudan). These fire-affected regions often exhibited high ecological sensitivity and carbon density, which led to pronounced disturbances in carbon fluxes. Nevertheless, the Democratic Republic of the Congo maintained an average annual net carbon sink of 74.2 MtC, indicating a high potential for ecological recovery. In contrast, Liberia and Eswatini exhibited net carbon emissions in fire-affected areas, suggesting weaker ecosystem resilience. These findings underscore the urgent need to incorporate wildfire disturbances into forest carbon management and climate mitigation strategies. In addition, climate variables such as temperature and soil moisture also influence carbon fluxes, although their effects display substantial spatial heterogeneity. On average, a 1 °C increase in temperature leads to an additional 0.347 (±1.243) Mt CO2 in emissions, while a 1% increase in soil moisture enhances CO2 removal by 1.417 (±8.789) Mt. However, compared to wildfires, the impacts of these climate drivers are slower and more spatially variable.

1. Introduction

Global climate change is one of the most pressing environmental challenges of the 21st century, with dynamic carbon cycle processes exerting direct impacts on climate stability [1,2,3]. Forests, as major terrestrial carbon reservoirs, play a pivotal role in regulating carbon fluxes and maintaining ecosystem balance [4,5,6,7]. Forests absorb atmospheric CO2 through photosynthesis, while emissions arise primarily from deforestation, wildfires, and soil respiration [8,9,10,11]. However, climate-induced stressors—such as rising temperatures, shifting precipitation patterns, and increasing climate extremes—have profoundly altered forest carbon fluxes, complicating projections of future forest–climate interactions [12,13,14]. In this context, a comprehensive investigation into the spatiotemporal variations in forest carbon fluxes and the identification of key drivers of forest carbon balance are crucial for enhancing our understanding of the global carbon cycle and developing effective climate change mitigation strategies [15,16,17]. African forests, particularly those in the Congo Basin, serve as key regulators of regional climate and hold the world’s second-largest carbon sequestration capacity—after the Amazon [18,19,20]. However, compared to the Amazon and Southeast Asia, tropical forest research in Africa remains limited [21,22,23,24]. Existing models of global carbon balance often lack precision in the African context due to data scarcity and pronounced spatial heterogeneity [11,25]. Meanwhile, deforestation, agricultural expansion, and urban encroachment have increasingly shifted some African forests from net carbon sinks to carbon sources [26,27,28,29]. Moreover, the response of African forests to climate change remains poorly understood, with substantial variability in carbon sequestration and emissions across distinct ecological zones, including tropical rainforests, savannas, and arid regions [7,21,30,31]. Therefore, a systematic assessment of the spatiotemporal dynamics of forest carbon fluxes in Africa, along with an investigation of their key drivers, is essential. Such research will not only improve the accuracy of global carbon cycle models but also provide critical insights for developing regional forest conservation strategies and carbon management policies.
Despite growing interest in forest carbon dynamics, several persistent challenges hinder a comprehensive understanding of African ecosystems. Global carbon flux research remains geographically imbalanced, with most studies focused on the Amazon and Southeast Asia, while tropical forests in Africa are still underrepresented [22,28,32]. Many existing assessments rely on coarse-resolution or single-source data—such as ground observations or global carbon models—limiting their ability to resolve the spatial heterogeneity and temporal complexity of carbon processes [33,34]. Furthermore, while carbon fluxes are shaped by multiple interacting climatic, soil, and anthropogenic factors, most studies focus narrowly on single drivers and overlook regional variability across different ecosystems [35,36,37]. In Africa, forest carbon sequestration is influenced not only by temperature and precipitation, but also by soil moisture, soil organic carbon, and elevation. However, their joint effects remain poorly quantified [38,39]. To improve the representation of African forests in terrestrial ecosystem models, there is a pressing need for high-resolution, spatially explicit datasets that account for local variability and driver interactions. These improvements are essential for supporting accurate carbon monitoring and sustainable forest management.
To address these gaps, this study provides a systematic, high-resolution analysis of forest carbon fluxes across Africa from 2001 to 2023. Integrating MODIS remote sensing data, ERA5 reanalysis, and forest carbon flux estimates, we apply a Geographically Weighted Regression (GWR) model to explore the spatially varying influence of environmental factors—including temperature, precipitation, soil organic carbon (SOC), soil moisture (SM), and elevation (ELE)—on gross CO2 emissions and removals. Our objectives are as follows: (1) characterize spatiotemporal patterns of forest carbon fluxes and identify major sink/source regions; (2) examine heterogeneity in emissions and removals over time; and (3) quantify the environmental drivers of carbon flux dynamics. The findings provide new insights into regional forest–climate feedbacks and inform improved carbon monitoring and sustainability strategies for Africa.

2. Data and Methods

2.1. Data Sources

This study utilizes a suite of high-resolution, long-term remote sensing and ecological datasets to evaluate the spatial and temporal patterns of forest carbon fluxes and their environmental drivers in Africa from 2001 to 2023. The datasets include vegetation metrics, climate reanalysis products, soil and forest change indicators, and anthropogenic pressure indices. Table 1 summarizes all datasets used, including their sources, spatial resolution, and temporal coverage.

2.1.1. Vegetation and Ecosystem Data

Vegetation indicators are derived from MODIS products, with a spatial resolution of 500–1000 m and annual temporal resolution. The forest cover dataset (MCD12Q1) provides yearly land cover classification using the IGBP scheme. LAI (MOD15A2H, 8-day composite) is used to assess canopy structure and biomass productivity, while NPP (MOD17A3HGF) reflects vegetation carbon assimilation. Kernel NDVI (kNDVI) is adopted to mitigate saturation effects of traditional NDVI in dense tropical canopies, thus enhancing photosynthetic signal quality.

2.1.2. Climate Variables

Precipitation data were obtained from the CHIRPS dataset (0.05° resolution, monthly, 1981–present). Air temperature, relative humidity, and wind speed were sourced from ERA5 monthly means (0.25°, 1979–present). Downward shortwave radiation was approximated using Photosynthetically Active Radiation (PAR) from the GLASS dataset (0.05° resolution), capturing radiation in the 400–700 nm range relevant to photosynthesis.

2.1.3. Soil and Forest Disturbance Variables

Soil organic carbon (SOC) was derived from the ISRIC SoilGrids dataset (250 m resolution), representing organic carbon concentration at 0–30 cm depth. Soil moisture (SM) was obtained from the NASA GLDAS v2.1 Noah Land Surface Model dataset (NASA/GLDAS/V021/NOAH/G025/T3H) for soil moisture analysis (spatial resolution of 0.25°, 2001–2023). Elevation data were obtained from the USGS SRTMGL1 Version 3 dataset, provided by the NASA Jet Propulsion Laboratory (JPL) through the Google Earth Engine platform.

2.1.4. Human Footprint Index

Human pressures were represented using the Global Terrestrial Human Footprint Index, integrating eight anthropogenic indicators such as population density, built infrastructure, and cropland extent [40]. This dataset (1 km resolution, annual) serves to characterize land-use intensity and ecological disturbance across regions.

2.1.5. Forest Carbon Flux Products

Gross emissions, gross removals, and net carbon flux were derived from the Global Forest Watch Carbon Monitoring system, which estimates CO2 exchange based on satellite-observed tree cover change and forest type-specific biomass and carbon density models [36]. The dataset covers the period 2001–2023 at a spatial resolution of 30 m. This version is an updated release of the original Forest Carbon Flux dataset developed by Harris [36], incorporating annual updates to key input layers through 2023, including tree cover loss, loss drivers, and burned area estimates. These improvements enhance the accuracy of attributing carbon fluxes to specific land-use transitions and disturbance types (e.g., deforestation, degradation, fire). The dataset provides high-resolution, spatially explicit information for analyzing long-term forest carbon dynamics across the African continent.
The selected datasets provide long-term time-series records with high spatial resolution, ensuring robust data support for this study. Detailed information on data sources is presented in Table 1.

2.2. Method

2.2.1. Estimation of Forest Carbon Fluxes

Forest carbon flux estimates in this study follow the methodological framework proposed by Harris [36], ensuring consistency with international reporting protocols. Initial forest extent is defined as areas with canopy cover ≥30% and tree height ≥5 m [4]. Forest carbon fluxes include gross emissions (from forest loss), gross removals (from forest growth), and net flux (their difference). The fluxes are calculated as follows:
E g r o s s = ∑ i = 1 n A i × E F i
where Egross represents the cumulative gross emissions over n years (Mg CO2), Ai is the forest loss area in year i (ha), and EFi is the carbon emission factor per unit area in year i (Mg CO2 ha−1).
Gross removals originate from forest growth and are estimated based on forest type, climatic region, and age classification. The calculation is expressed as follows:
R g r o s s = ∑ i = 1 n F i × R i
where Rgross represents the cumulative gross removals over n years (Mg CO2), Fi denotes the forest area in year i (ha), and Ri represents the carbon sequestration rate per unit area in year i (Mg CO2 ha−1), typically derived from growth models specific to different forest types and climatic regions.
Net carbon flux refers to the net exchange of carbon between forests and the atmosphere from 2001 to 2023, representing the balance between gross emissions and gross removals. It is calculated as follows:
N f l u x = E g r o s s − R g r o s s
where Nflux represents the net carbon flux over the study period (Mg CO2), Egross is the cumulative gross emissions (Mg CO2), Rgross is the cumulative gross removals (Mg CO2). A positive net flux (Nflux > 0) indicates that forests acted as a net carbon source, meaning that carbon emissions exceeded removals. Conversely, a negative net flux (Nflux < 0) signifies that forests functioned as a net carbon sink, with carbon removals surpassing emissions.

2.2.2. Geographically Weighted Regression Model

To assess the spatially varying influence of environmental factors on forest carbon fluxes, we employed a Geographically Weighted Regression (GWR) model [41]. This model accounts for spatial non-stationarity by allowing regression coefficients to vary by location. The general form of the model is
C E , R , N , i = β 0 ( u i , v i ) + β 1 u i , v i P R E i + β 2 u i , v i T E M P i + β 3 u i , v i R H i + β 4 u i , v i W S i + β 5 u i , v i P A R i + β 6 u i , v i S O C i + β 7 u i , v i S M i + β 8 u i , v i E L E i + β 9 ( u i , v i ) H F P i + ϵ i
where C(E,R,N),i represents the dependent variable, referring to gross emissions, gross removals, or net flux at location i. (ui, vi) represent the latitude and longitude coordinates of the study site, allowing regression coefficients to vary spatially. β1−β9 are the regression coefficients at (ui, vi), which vary spatially to capture local effects of the explanatory variables: PREi (precipitation, mm); TEMPi (air temperature, °C); RHi (relative humidity, %); WSi (wind speed, m/s); PARi (photosynthetically active radiation, W/m2); SOCi (soil organic carbon, g/kg); SMi (soil moisture, %); ELEi (elevation, m); HFPi (human footprint index, dimensionless). The term ϵi represents the residual error. The regression coefficients (β) in the GWR model exhibit spatial variability, meaning that each study site has independently estimated coefficients. By accounting for spatial non-stationarity, the model improves the accuracy of spatially explicit forest carbon flux assessments by identifying regional variations in environmental influences. In the analysis of the results, the estimated regression coefficients are reported as “mean ± standard deviation.” For example, if temperature increases by 1 °C, gross emissions increase by approximately 0.347 (±1.243) Mt CO2. Here, the mean (0.347) represents the average regression coefficient across all study sites, while the standard deviation (±1.243) reflects the spatial variation in the estimated effect of temperature on gross emissions. The regression coefficients of other environmental variables are interpreted in the same manner.
The Gaussian Kernel is used to determine the weights for local regression in the GWR model, formulated as follows:
W i j = e x p ( d i j 2 h 2 )
where Wij is the weight assigned to observation j when estimating the parameters for location i. dij represents the distance between observation points i and j. h is the bandwidth parameter, controlling the spatial extent of the local regression.
The optimal bandwidth (h) is selected using the Cross-Validation (CV) method, defined as follows:
C V h = ∑ i = 1 n ( y i − y − i ∗ ) 2
where yi is the observed value (true value). y − i ∗ is the predicted value for observation i, obtained by training the GWR model on the dataset excluding i. h is the bandwidth parameter, determining the range of local regression in the GWR model. The goal is to select the value of h that minimizes CV(h), ensuring the best trade-off between local specificity and model accuracy. The optimized bandwidth allows the GWR model to accurately capture spatially varying relationships between environmental factors and forest carbon fluxes.

3. Results

3.1. Spatiotemporal Dynamics of African Forests

Forests are primarily distributed along the equatorial belt, especially within the Congo Basin, which remains the most densely forested region in Africa. Major forest types include evergreen broadleaf forests, mixed forests, and deciduous formations. In Figure 1a,b, the maps of forest distribution in 2001 and 2023 highlight forest concentration and the spatial patterns of forest shrinkage. Notably, Madagascar shows a pronounced reduction in forest extent (Figure 1c), reflecting intense anthropogenic pressure and land-use conversion. Vegetation productivity indicators—including the kernel Normalized Difference Vegetation Index (kNDVI), Leaf Area Index (LAI), and Net Primary Productivity (NPP)—demonstrated an overall upward trend during the study period (Figure 1e–g). Specifically, kNDVI values increased in most central and southern forest zones, indicating improved photosynthetic activity (Figure 1e), while LAI expanded in the Congo Basin and southern savanna–forest mosaics (Figure 1f). Conversely, regions such as Madagascar exhibited declines in both kNDVI and LAI, underscoring ecosystem degradation (Figure 1g).
Between 2001 and 2023, the annual mean kNDVI, LAI, and NPP of African forests exhibited a gradual upward trend (Figure 2a–c). Overall, these three indicators showed relatively minor fluctuations, maintaining a stable range. Specifically, the kNDVI values ranged between 0.44 and 0.49, LAI varied from 3.0 to 3.5, while NPP ranged from 1.00 to 1.25. In terms of long-term trends, kNDVI increased from 0.4471 in 2001 to 0.487 in 2023, indicating an enhancement in forest photosynthetic capacity and a degree of vegetation recovery over the past two decades. However, kNDVI declined to 0.4451 in 2012 before gradually rebounding, reaching its highest value of 0.487 in 2023. Similarly, LAI increased from 3.21 in 2001 to 3.47 in 2023, reflecting an overall rise in forest canopy density and vegetation coverage. Notably, LAI experienced a sharp decline to 3.05 in 2012 but recovered after 2015, reaching 3.44 in 2021 and further increasing to 3.47 in 2023. NPP, a key indicator of forest carbon sequestration capacity, represents the growth rate and carbon sink potential of forest ecosystems. During the study period, NPP increased from 1.05 in 2001 to 1.14 in 2023, suggesting an overall improvement in biomass accumulation capacity. Despite interannual fluctuations in vegetation indices, these trends indicate a gradual ecological recovery of African forests over the past two decades.

3.2. Spatial Patterns of Forest Carbon Fluxes

3.2.1. Spatial Distribution of Forest Carbon Emissions

Between 2001 and 2023, the cumulative gross emissions from African forests exhibited a highly uneven spatial distribution, with significant variations in emission levels across countries and regions (Figure 3). Overall, carbon emissions were concentrated in Central and West Africa, particularly in countries with abundant forest resources but high deforestation rates. The Democratic Republic of the Congo was the largest contributor to forest carbon emissions in Africa, with cumulative emissions reaching 124.56 Mt CO2 from 2001 to 2023, far exceeding those of other countries. Madagascar and Ivory Coast also recorded high emissions, at 26.14 Mt CO2 and 20.24 Mt CO2, respectively, indicating substantial forest-related carbon release during this period. Additionally, Mozambique, Angola, Ghana, and Nigeria exhibited considerable emissions, each exceeding 7 Mt CO2. From a regional perspective, Central Africa, particularly the Congo Basin and its surrounding countries, experienced the highest emissions, likely due to the region’s extensive forest cover, increasing forest fragmentation, and land-use changes. Similarly, coastal West African countries, including Ivory Coast, Ghana, and Liberia, showed elevated emissions. In East and Southern Africa, emissions were notably high in Mozambique, Tanzania, and Zambia, reflecting the significant impact of land-use practices on forest carbon fluxes. In contrast, Northern Africa exhibited substantially lower carbon emissions. Countries such as Libya, Mauritania, Tunisia, and Egypt reported cumulative emissions below 1 Mt CO2, with some approaching zero. This is primarily attributed to extremely low forest cover, as the region is dominated by deserts, grasslands, and arid ecosystems, resulting in minimal forest carbon emissions. Overall, the spatial distribution of forest carbon emissions in Africa reflects both forest resource availability and deforestation pressures. High-emission regions correspond to countries with extensive forest cover undergoing intensive deforestation, whereas low-emission areas are primarily found in arid and semi-arid regions, such as Northern Africa and the fringes of the Sahara Desert.

3.2.2. Spatial Distribution of Forest Carbon Removals

Between 2001 and 2023, the gross removals of African forests exhibited significant spatial variability, with notable differences in carbon sequestration levels across countries (Figure 4). High carbon sequestration areas were primarily concentrated in Central Africa, as well as parts of East and West Africa, whereas Northern Africa and arid and semi-arid regions displayed lower sequestration levels. At the national level, the Democratic Republic of the Congo (DRC) was the largest contributor, with cumulative gross removals reaching 189.26 Mt CO2, far surpassing other countries. Angola and the Central African Republic also demonstrated considerable sequestration, with 40.22 Mt CO2 and 31.80 Mt CO2, respectively. Additionally, Cameroon, Ivory Coast, the Republic of Congo, Mozambique, and Nigeria each recorded over 19 Mt CO2, underscoring the role of their forests in atmospheric carbon fixation. Tropical forest countries in Central and West Africa, including the DRC, Cameroon, Gabon, Ivory Coast, and Ghana, exhibited high carbon sequestration levels, likely due to extensive forest cover and rapid growth rates. Tropical rainforest ecosystems possess strong biomass accumulation capacity, making them crucial in absorbing atmospheric CO2. In Southern Africa, Mozambique, Tanzania, and Zambia also demonstrated relatively high carbon removals, indicating sustained carbon sink capacity. In contrast, Northern Africa and the Sahel region had significantly lower carbon sequestration. Countries such as Libya, Mauritania, Niger, and Egypt reported cumulative carbon removals below 1 Mt CO2, with some approaching zero, primarily due to limited forest cover and arid landscapes. Overall, the spatial distribution of forest carbon removals in Africa is shaped by forest cover, climate zones, and forest growth rates. High sequestration areas are concentrated in tropical rainforests, whereas low sequestration areas correspond to arid and semi-arid regions, reflecting regional differences in forest carbon sink capacity.

3.2.3. Spatial Distribution of Net Carbon Flux

Between 2001 and 2023, the net carbon flux of African forests exhibited significant regional disparities, forming a spatial pattern in which Central Africa and parts of Southern Africa functioned as carbon sinks, while portions of West and East Africa acted as carbon sources (Figure 5). The spatial distribution of carbon sinks and sources was closely associated with forest cover and vegetation growth dynamics. The tropical rainforests of the Congo Basin remained Africa’s most significant carbon sink, particularly in the Democratic Republic of the Congo, Gabon, and the Central African Republic, where CO2 absorption consistently exceeded emissions, ensuring these countries maintained a long-term carbon sink status. In contrast, parts of West and East Africa exhibited carbon source characteristics, primarily due to deforestation and agricultural expansion, which substantially reduced forest carbon sequestration capacity. Some areas even transitioned from carbon sinks to carbon sources over time. In Southern Africa, particularly in Zambia and Tanzania, net carbon flux remained relatively balanced, with some countries sustaining strong carbon sink capacity. Meanwhile, Northern Africa, characterized by low forest cover, exhibited minimal forest-related carbon flux, exerting limited influence on regional and global carbon cycles.
From the perspective of carbon sink regions, Central Africa, parts of West Africa, and certain Southern African countries have consistently maintained their carbon sink status. The Democratic Republic of the Congo remains Africa’s largest forest carbon sink, with a cumulative net carbon flux of −64.7 Mt CO2 from 2001 to 2023, indicating that its forest carbon sequestration capacity exceeded its emissions. Additionally, Angola, the Central African Republic, the Republic of Congo, and Gabon exhibited strong carbon sink capacities, all recording negative net carbon flux values. These regions maintained high carbon absorption rates and relatively low levels of forest carbon loss, sustaining their long-term carbon sink capacity. Gabon, in particular, exhibited stable forest ecosystem dynamics, with a consistently negative net carbon flux throughout the study period. In contrast, carbon source regions were primarily concentrated in parts of West and East Africa. Countries such as Madagascar, South Africa, Sierra Leone, and Liberia recorded positive net carbon flux values, indicating that their carbon emissions exceeded carbon absorption. Madagascar had the highest net carbon flux among them, at 8.5 Mt CO2, followed by South Africa (2.6 Mt CO2), Sierra Leone (1.6 Mt CO2), and Liberia (1.2 Mt CO2). These countries functioned predominantly as carbon sources during the study period. Meanwhile, low carbon flux regions were primarily located in Northern Africa and the Sahel, including Libya, Egypt, Mauritania, Niger, and Chad. These countries exhibited both low carbon emissions and low carbon absorption, reflected in net carbon flux values close to zero. This pattern corresponds with the region’s limited forest cover, where desert and arid ecosystems dominate.

3.3. Drivers of Forest Carbon Fluxes

3.3.1. Direct Impacts of Wildfires on Forest Carbon Fluxes

Between 2001 and 2023, the spatial and temporal patterns of wildfires across Africa and their impact on carbon fluxes exhibited significant regional heterogeneity and ecological feedback variability (Figure 6a,b). Figure 6a shows that wildfires were predominantly concentrated in tropical savannas and woodland ecotones, including the periphery of the Congo Basin, the West African Sahel, the East African Rift grasslands, and inland woodlands of southern Africa. In these areas, wildfire frequency reached up to 23 out of 23 years, indicating near-annual fire occurrences. In contrast, low fire frequencies were observed in the arid zones of North Africa, the humid rainforests of central Congo, and the South African Plateau and coastal regions, reflecting the suppressive role of humid climates, forest cover, and land-use practices on fire occurrence. Figure 6b further reveals the carbon flux responses in burned regions. According to the statistical results, the Democratic Republic of the Congo (DRC), Angola, Zambia, and Mozambique experienced the largest wildfire-affected areas with valid carbon flux data—approximately 934,705, 753,772, 512,719, and 533,245 km2, respectively. These regions generally showed negative average net carbon flux values, indicating considerable wildfire-induced carbon emissions (Table 2). For instance, the DRC had an average net flux of −66.9 Mg CO2e/ha, an annual carbon removal of 92.23 MtC, emissions of 18.02 MtC, and a net sink of 74.2 MtC, the largest among all African countries. Angola, Zambia, and Mozambique also demonstrated substantial net carbon sinks, suggesting strong post-fire ecological recovery and carbon uptake capacity. South Sudan, the Central African Republic, and Nigeria also showed notable annual net carbon sinks of 27.0, 28.7, and 18.8 MtC, respectively. In contrast, countries such as Liberia (−1.01 MtC), Eswatini (−0.25 MtC), and Sierra Leone (−0.88 MtC) exhibited net carbon emissions in fire-affected regions, implying potential ecological degradation or weak post-fire recovery capacity. Moreover, countries with relatively high fire frequency but small burned areas (e.g., Lesotho, Djibouti) contributed minimally to the continental carbon balance. Overall, wildfire-prone regions in Africa can function either as carbon sources or as carbon sinks, depending on a complex interplay of factors such as ecosystem type, fire frequency and intensity, land-use practices, and vegetation recovery potential. Therefore, wildfire management and carbon mitigation strategies should be tailored to regional conditions through targeted and adaptive policy measures.
The scale, intensity, and recurrence characteristics of wildfire activity across African countries show marked spatiotemporal heterogeneity (Table 2). Overall, wildfires in Africa exhibit a distinct pattern of regional concentration and disturbance intensity gradient: the woodland–savanna transition zones in Central and Southern Africa are characterized by both extensive burned areas and high recurrence rates of fire events. In contrast, North Africa and several highland countries experience sparse and spatially fragmented wildfire activity. In terms of total burned area (fire area) and the area affected by carbon flux changes (affected carbon area), the Democratic Republic of the Congo, Angola, Mozambique, and Zambia each exhibit more than 500,000 km2 of wildfire-affected land, making them the most extensively impacted countries. Furthermore, in these countries, the proportion of carbon-affected areas relative to total burned areas generally exceeds 80%, suggesting that wildfires predominantly occur in ecologically sensitive and carbon-rich zones, leading to more pronounced disturbance effects. The recurrence of fire events (Mean Fire Frac) also varies significantly among countries. South Sudan (0.7584), Ghana (0.6471), and Angola (0.6275) demonstrate high annual re-burning rates, indicating ecosystems that are subjected to sustained disturbance regimes, and exhibit heightened ecological vulnerability due to limited recovery windows. In contrast, Tunisia (0.100), Morocco (0.0914), and other North African nations present localized, low-frequency wildfire patterns. Region-specific ecological management and restoration strategies should therefore be prioritized, particularly focusing on regulating land use and assessing fire risks in areas with high fire recurrence, to enhance ecosystem resilience and recovery capacity.

3.3.2. Environmental Modulators of Forest Carbon Dynamics

In analyzing the drivers of forest carbon emissions, removals, and net carbon flux in Africa from 2001 to 2023 using the GWR model, this study selected nine key factors (Figure 7). These variables were chosen based on their fundamental roles in forest carbon cycling processes. PR, TEMP, RH, and PAR directly influence vegetation growth and photosynthesis. SM and SOC contribute to the stability of soil carbon pools and regulate the decomposition rate of organic matter. WS affects moisture loss and plant transpiration. FL and HFP reflect the impact of human activities on forest carbon fluxes [4]. From a spatial perspective, tropical rainforest regions, such as the Congo Basin, exhibit higher PR, SM, and SOC levels, whereas Saharan and North African regions show relatively lower values for these factors. A positive correlation is observed between ELE and carbon emissions. Conversely, regions with higher forest cover and lower land-use pressure, such as the Central African forest zone, exhibit greater carbon sequestration capacity, showing a positive correlation with carbon removals. All spatial associations discussed here are based on coefficient estimates that are statistically significant at the p < 0.01 level.
The results of the GWR model indicate significant spatial heterogeneity in the effects of different environmental factors on carbon emissions (Figure 8). A 1 °C increase in TEMP is associated with an increase in carbon emissions by approximately 0.347 (±1.243) Mt CO2. For every 1 Mha increase in FL, carbon emissions rise by 0.175 (±0.294) Mt CO2. A 1% increase in SM enhances carbon sequestration capacity by 14.759 (±30.633) Mt CO2, highlighting the critical role of water availability in forest carbon sinks. In 55.69% of the study area, increasing TEMP correlates with higher carbon emissions, whereas in 44.31%, TEMP increases are associated with enhanced carbon sequestration. The highest β value for TEMP (3.46 Mt CO2) was observed at (−6.75° S, 21.45° E), while the lowest β value (−1.89 Mt CO2) occurred at (−11.68° S, 23.99° E). FL is positively correlated with carbon emissions in most areas, although in some regions, its impact is relatively low. The effects of SM exhibit strong spatial heterogeneity. In some regions, increased SM promotes vegetation growth and carbon sequestration, whereas in humid environments, it may accelerate soil organic matter decomposition, leading to higher carbon emissions. RH (β = 0.417 ±0.520 Mt CO2) generally supports carbon sequestration. WS (β = 0.531 ±5.415 Mt CO2) enhances forest transpiration in some regions, improving carbon sequestration, while in others, it may contribute to soil erosion or increased fire risk, leading to higher carbon emissions. HFP (β = −0.077 ±0.181 Mt CO2) affects carbon sequestration in certain areas, potentially reflecting afforestation and ecological restoration activities. PR (β = 0.0005 ±0.0087 Mt CO2) has a minimal direct effect on carbon emissions, primarily influencing carbon cycling through its impact on SM and vegetation growth. SOC (β = 0.217 ±1.012 Mt CO2) generally shows a positive correlation with carbon emissions, suggesting that SOC-rich soils may contribute to greater carbon release through microbial decomposition. PAR (β ≈ −0.001 Mt CO2) promotes carbon sequestration in some areas, though in regions with high PAR and limited water availability, its effects may vary.
The results of the GWR model indicate significant spatial heterogeneity in the effects of different environmental factors on carbon removals (Figure 9). A 1 °C increase in TEMP enhances forest carbon removal capacity by approximately 0.508 (±0.675) Mt CO2. A 1% increase in SM improves carbon sequestration capacity by 1.417 (±8.789) Mt CO2, underscoring the importance of water availability for forest carbon sinks. In 80.89% of the study area, rising TEMP correlates with enhanced carbon removals, with the highest β value (2.44 Mt CO2) observed at (−8.53° S, 22.74° E) and the lowest β value (−1.04 Mt CO2) at (−12.65° S, 24.33° E). The β values for ELE are predominantly positive (78.05% of the area), indicating that forest reduction is generally associated with diminished carbon sequestration capacity. The influence of SM exhibits spatial heterogeneity, with 57.32% of the area showing positive β values, indicating that regions with higher SM have stronger carbon removal capacity. In contrast, 42.68% of the area exhibits negative β values, suggesting that excessively wet conditions may accelerate soil microbial decomposition, leading to carbon loss. RH (β = 0.087 ± 0.159 Mt CO2) generally promotes carbon sequestration in most areas. WS (β = −0.703 ± 2.224 Mt CO2) is negatively correlated with carbon removals in 60.57% of the area, indicating that in wind-eroded regions, increased WS may intensify carbon release. However, in certain humid areas, WS may enhance carbon sequestration by facilitating vegetation spread. HFP (β = 0.049 ± 0.095 Mt CO2) is positively correlated with carbon removals in 67.23% of the area, suggesting that vegetation restoration and land management practices contribute to enhanced carbon sequestration capacity. PR (β = −0.001 ± 0.003 Mt CO2) has a minimal direct effect on carbon removals, primarily influencing carbon dynamics through its regulation of SM and vegetation growth. PAR (β = 0.002 ± 0.006 Mt CO2) promotes carbon removals in some areas, though in regions with high PAR and water limitations, its effects may vary. SOC (β = −0.070 ± 0.496 Mt CO2) is negatively correlated with carbon removals in most areas, suggesting that high SOC levels may enhance microbial decomposition activity, leading to carbon release.

4. Discussion

4.1. Changes in African Forests and Their Ecological Significance

This study reveals that the Congo Basin remains the most intact and resilient forest region in Africa, while parts of West and East Africa have experienced significant forest disturbances. Although our results show a general increase in vegetation indices (such as kNDVI, LAI, and NPP)—indicating localized improvements in forest ecosystem conditions—the widespread impacts of wildfires and climate-related anomalies, particularly temperature and soil moisture, may have led to the degradation of forest integrity and carbon storage capacity in many regions [42,43]. These findings highlight the importance of considering the spatially heterogeneous responses of forest ecosystems to climate change and disturbance regimes when evaluating their carbon sink functions.
Forest loss not only affects carbon balance but also has broader ecological implications. Although this study did not directly analyze hydrological or soil processes, previous studies suggest that reduced forest cover may contribute to localized declines in precipitation, increased aridity, and soil degradation [44]. In regions of West and East Africa that exhibited significant forest disturbance, reduced vegetation cover is likely to impair ecosystem functions such as soil moisture retention and climate buffering capacity. Our findings, consistent with those of Hubau et al. [22], confirm that forests in the Congo Basin continued to act as a net carbon sink during the study period. In contrast, forested areas in West and East Africa showed signs of degradation and net carbon emissions, possibly linked to frequent fire disturbances and climate variability.
From a global perspective, the carbon dynamics of African forests during 2001–2023 differ from trends observed in other tropical regions. For instance, previous studies have documented a gradual shift of the Amazon rainforest from a carbon sink to a carbon source [10]. In contrast, our results indicate that African forests—particularly in the Congo Basin—have largely retained their carbon sink function. This distinction may reflect differences in forest disturbance regimes, recovery capacity, and historical land-use pressures. Nonetheless, the increasing frequency of wildfires and persistent climate anomalies could pose significant risks to the future stability of Africa’s forest carbon sinks. Therefore, sustained monitoring and regionally tailored conservation strategies remain critical for safeguarding the ecological and climatic functions of African forests.

4.2. Spatial Patterns of Forest Carbon Flux and Key Driving Mechanisms

This study reveals significant spatiotemporal heterogeneity in forest carbon fluxes across Africa. The Congo Basin continues to serve as one of the world’s most vital carbon sink regions, with carbon removals far exceeding emissions. In contrast, several areas in West and East Africa have experienced a shift from net carbon sinks to net carbon sources. The Democratic Republic of the Congo recorded the highest levels of CO2 absorption, whereas emissions in West and East Africa were largely driven by deforestation, agricultural expansion, urban development, and climate variability. In Northern Africa, characterized by sparse forest cover, carbon fluxes remained relatively stable, contributing minimally to the regional carbon budget. These spatial patterns are consistent with findings from recent global carbon budget assessments [14,43,44]. Furthermore, the observed acceleration of carbon emissions in parts of West and East Africa suggests an ongoing transition from weak carbon sinks to active carbon sources in several ecological zones.
In addition to land-use change and climate variability, wildfires have emerged as a critical direct disturbance to forest carbon fluxes in Africa. From 2001 to 2023, fire activity was concentrated in woodland–savanna transition zones, especially in Central and Southern Africa. Countries such as the Democratic Republic of the Congo, Angola, Mozambique, and Zambia experienced both extensive burned areas (over 500,000 km2) and high fire recurrence rates (e.g., 0.7584/year in South Sudan), indicating persistent fire pressure. These fire-prone areas also overlapped with high-carbon-density ecosystems, leading to intensified carbon flux disturbances. While some countries, such as the DRC, demonstrated post-fire carbon recovery capacity and remained net sinks, others—such as Liberia and Eswatini—exhibited net emissions in fire-affected regions, reflecting weaker ecological resilience. These patterns emphasize the importance of incorporating wildfire dynamics into spatial assessments of carbon balance and mitigation planning.
The GWR model quantifies the effects of environmental variables on forest carbon fluxes, identifying TEMP (β = 2.75) and SOC (β = 2.83) as the most influential predictors of carbon balance. Rising temperatures intensify microbial activity and accelerate soil organic matter decomposition, thereby enhancing CO2 emissions—particularly in the arid and semi-arid landscapes of East and West Africa, where reduced PR, soil desiccation, and vegetation degradation compound this effect [45,46,47]. While elevated SOC is generally associated with enhanced carbon sequestration, its effects are ecosystem-specific. For example, in regions such as Gabon and the Central African Republic, high SOC levels support continued carbon removal. In contrast, rapid SOC depletion in West Africa has reinforced carbon source effects and limited recovery capacity.
A comparison between the Congo Basin and the Amazon Rainforest reveals important divergences in recent carbon flux trends. While the Amazon has gradually transitioned from a carbon sink to a carbon source due to declining PR and intensifying human pressures [10,43], the Congo Basin has thus far retained a net sink status, with CO2 removals significantly exceeding emissions. This divergence may be attributed to differences in regional climate stability, land-use regimes, and ecosystem resilience. Higher PR levels in the Congo Basin support forest growth and mitigate drought stress, whereas sustained PR decline in the Amazon has exacerbated regional desiccation. Additionally, deforestation in the Congo Basin is largely associated with commercial logging, whereas the Amazon faces persistent encroachment from large-scale agriculture and livestock expansion. However, the Congo Basin’s carbon sink function remains vulnerable. Ongoing global warming and future PR declines could undermine its current balance, shifting the region towards carbon neutrality or even net emissions.
Protecting carbon sink regions—especially the Congo Basin—remains critical. Currently, forest conservation efforts across Africa heavily rely on REDD+ and other multilateral initiatives. Although pilot REDD+ projects in the Congo Basin have demonstrated some success in reducing deforestation [9,44,48], implementation has been uneven across the continent. In West Africa, for example, weak governance and the continued expansion of subsistence agriculture have impeded REDD+ effectiveness, contributing to a steady erosion of forest-based carbon sinks. Future mitigation strategies should also incorporate fire risk assessment and fire management measures, such as early-warning systems and post-fire ecological restoration, especially in fire-prone savanna–woodland transition zones. To address these challenges, future strategies should prioritize the following: (1) strengthening sustainable forest management in the Congo Basin to prevent large-scale logging; and (2) expanding ecological restoration programs in West and East Africa, including reforestation and land-use optimization. These actions are vital to mitigating the rise of carbon source regions and sustaining Africa’s contribution to global carbon regulation.

4.3. Limitations and Recommendations

Despite systematically assessing the spatiotemporal dynamics and driving mechanisms of African forest carbon fluxes from 2001 to 2023, this study has certain limitations. First, the accuracy of carbon flux estimations may be affected by data precision constraints. This study relies on multi-source remote sensing data, which, while providing long-term temporal coverage and broad spatial representation, have relatively low spatial resolution. This limitation makes it challenging to capture fine-scale forest changes, particularly in forest edge areas and secondary forests, where carbon balance dynamics are more complex. Additionally, variables such as SOC are derived from global datasets, which may introduce uncertainties across different forest types.
Future research should integrate high-resolution remote sensing data, such as GEDI LiDAR, along with ground-based observations, to improve the accuracy of carbon flux estimations. Second, this study primarily focuses on nine key influencing factors, including TEMP, SM, and FL, which may not fully encompass all relevant variables. Future studies should incorporate a broader range of indicators to better capture the complex interactions between natural and anthropogenic factors influencing forest carbon fluxes, providing more precise insights for regional carbon management policies. Lastly, while this study employs the GWR model to reveal spatial heterogeneity in forest carbon fluxes, it does not predict long-term trends or establish causal relationships. Future research should integrate dynamic global vegetation models and deep learning approaches to explore the long-term evolution of forest carbon balance under different climate change scenarios. Additionally, incorporating policy impact assessments could help quantify the potential effects of conservation initiatives on forest carbon sink capacity. Further research should also focus on the impact of extreme climate events, such as droughts and wildfires, on African forest carbon fluxes to enhance understanding of forest carbon cycles. The findings of this study underscore the importance of strengthening forest conservation, optimizing land-use policies, and enhancing the forest carbon sink capacity, providing a scientific basis for Africa carbon neutrality goals.

5. Conclusions

Wildfires, as a major direct disturbance factor, exert significant impacts on forest carbon fluxes across Africa, particularly in central and southern regions. Frequent reburning and extensive burned areas have severely disrupted local carbon dynamics. This study provides a comprehensive assessment of forest carbon fluxes and their environmental drivers across Africa from 2001 to 2023, based on multi-source remote sensing and climate datasets. Cumulative carbon emissions from forest ecosystems reached 328.43 Mt CO2, while removals totaled 545.07 Mt CO2, sustaining a net carbon sink at the continental scale. However, regional disparities remain pronounced. West and East Africa jointly contributed 142 Mt CO2 in emissions—accounting for 43% of the total—indicating a partial shift from net sinks to net sources in certain regions. In contrast, forest-rich zones such as the Congo Basin remained strong carbon sinks despite recurring fire events, while countries such as Liberia and Eswatini exhibited net carbon emissions in fire-affected areas, reflecting lower post-disturbance ecosystem resilience. The GWR model identified TEMP, ELE, and SM as the most influential factors affecting forest carbon fluxes. A 1 °C increase in TEMP was associated with an average rise of 0.347 ± 1.243 Mt CO2 in emissions, and each 1 Mha increase in FL contributed an additional 0.175 ± 0.294 Mt CO2. Conversely, a 1% increase in SM enhanced removals by 1.417 ± 8.789 Mt CO2. These findings underscore the importance of incorporating climate stressors and disturbance regimes—particularly wildfires—into spatially differentiated forest carbon management strategies and long-term ecological monitoring to safeguard Africa’s role in the global carbon cycle.

Author Contributions

L.Z.: conceptualization, data curation, formal analysis, investigation, methodology, resources, supervision, visualization, writing–original draft, writing–review and editing. Z.Z.: funding acquisition, methodology, software, visualization, writing–review and editing. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the National Social Science Fund of China [20FJYA003], the National Natural Science Fund of China “The Interaction Mechanism between the Expansion of Informal Settlements and Urbanization: A Case Study of Typical African Cities”(No.42301227), the Major Research Project of Philosophy and Social Sciences of the Ministry of Education of the People’s Republic of China “Population Opportunities, Challenges, and Policy Research in the Construction of Chinese-style Modernization”.

Data Availability Statement

The datasets utilized in this study are publicly available from various sources. Forest cover data were obtained from the dataset (MCD12) (https://modis.gsfc.nasa.gov/data/dataprod/mod12.php, accessed on 1 January 2025), while LAI data were sourced from the dataset (MOD15A2H) (https://lpdaac.usgs.gov/products/mod15a2hv006/, accessed on 1 January 2025). The kNDVI data were acquired from the GitHub (https://github.com/IPL-UV/kNDVI, accessed on 1 January 2025). Additionally, NPP data were derived from the dataset (MOD17A3HGF) (https://lpdaac.usgs.gov/products/mod17a3hgfv006/, accessed on 1 January 2025). Meteorological variables were compiled from multiple sources. Precipitation data were retrieved from the CHIRPS precipitation dataset (https://www.chc.ucsb.edu/data/chirps, accessed on 1 January 2025), while temperature, relative humidity, and wind speed data were extracted from the ERA5 reanalysis dataset (https://cds.climate.copernicus.eu/datasets/reanalysis-era5-single-levels-monthly-means?tab=overview, accessed on 1 January 2025). PAR data were obtained from the GLASS dataset (https://www.glass.hku.hk/download.html, accessed on 1 January 2025). SOC data were sourced from the global SOC dataset (https://data.isric.org/geonetwork/srv/eng/catalog.search#/metadata/soilgrids-global-soil-organic-carbon-map, accessed on 1 January 2025), while soil moisture data were acquired from the NASA GLDAS v2.1 Noah Land Surface Model dataset (https://disc.gsfc.nasa.gov/datasets/GLDAS_NOAH025_3H_2.1/summary, accessed on 1 January 2025). Elevation data were obtained from the USGS SRTMGL1 Version 3 dataset, provided by the NASA Jet Propulsion Laboratory (JPL) through the Google Earth Engine platform (https://developers.google.com/earth-engine/datasets/catalog/USGS_SRTMGL1_003, accessed on 1 January 2025). Anthropogenic and carbon flux datasets were also employed in this study. Human footprint data were obtained from the annual global terrestrial Human Footprint dataset (https://figshare.com/articles/figure/An_annual_global_terrestrial_Human_Footprint_dataset/16571064, accessed on 1 January 2025). The forest carbon flux data, including gross emissions, gross removals, and net flux, were sourced from the Global Forest Watch Carbon Monitoring dataset (https://www.globalforestwatch.org/blog/data/whats-new-carbon-flux-monitoring/, accessed on 1 January 2025). All datasets used in this study are freely available from the respective sources cited. Further details on data processing and analysis can be found in the Methodology section of this manuscript.

Conflicts of Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

References

  1. Anderegg, W.R.L.; Trugman, A.T.; Badgley, G.; Anderson, C.M.; Bartuska, A.; Ciais, P.; Field, C.B.; Freeman, J.; Goetz, S.J.; Hicke, J.A.; et al. Climate-driven risks to the climate mitigation potential of forests. Science 2020, 368, eaaz7005. [Google Scholar] [CrossRef] [PubMed]
  2. Busch, J.; Bukoski, J.J.; Cook-Patton, S.C.; Griscom, B.; Kaczan, D.; Potts, M.D.; Yi, Y.; Vincent, J.R. Cost-effectiveness of natural forest regeneration and plantations for climate mitigation. Nat. Clim. Change 2024, 14, 996–1002. [Google Scholar] [CrossRef]
  3. Ke, P.; Ciais, P.; Sitch, S.; Li, W.; Bastos, A.; Liu, Z.; Xu, Y.; Gui, X.; Bian, J.; Goll, D.S.; et al. Low latency carbon budget analysis reveals a large decline of the land carbon sink in 2023. Natl. Sci. Rev. 2024, 11, nwae367. [Google Scholar] [CrossRef]
  4. Hansen, M.C.; Potapov, P.V.; Moore, R.; Hancher, M.; Turubanova, S.A.; Tyukavina, A.; Thau, D.; Stehman, S.V.; Goetz, S.J.; Loveland, T.R.; et al. High-resolution global maps of 21st-century forest cover change. Science 2013, 342, 850–853. [Google Scholar] [CrossRef]
  5. Doughty, C.E.; Metcalfe, D.B.; Girardin, C.A.J.; Farfán Amézquita, F.; Galiano Cabrera, D.; Huaraca Huasco, W.; Silva-Espejo, J.E.; Araujo-Murakami, A.; da Costa, M.C.; Rocha, W.; et al. Drought impact on forest carbon dynamics and fluxes in Amazonia. Nature 2015, 519, 78–82. [Google Scholar] [CrossRef]
  6. Li, W.; Ciais, P.; Wang, Y.; Yin, Y.; Peng, S.; Zhu, Z.; Bastos, A.; Yue, C.; Ballantyne, A.; Broquet, G.; et al. Recent changes in global photosynthesis and terrestrial ecosystem respiration constrained from multiple observations. Geophys. Res. Lett. 2018, 45, 1058–1068. [Google Scholar] [CrossRef]
  7. Su, Y.; Li, X.; Zhang, C.; Yan, W.; Ciais, P.; Cook-Patton, S.C.; Phillips, O.L.; Shang, J.; Cescatti, A.; Chen, J.-M.; et al. Carbon accumulation rate peaks at 1,000-m elevation in tropical planted and regrowth forests. One Earth 2025, 8, 101147. [Google Scholar] [CrossRef]
  8. Pan, Y.; Birdsey, R.A.; Fang, J.; Houghton, R.; Kauppi, P.E.; Kurz, W.A.; Phillips, O.L.; Shvidenko, A.; Lewis, S.L.; Hayes, D. A Large and Persistent Carbon Sink in the World’s Forests. Science 2011, 333, 988–993. [Google Scholar] [CrossRef]
  9. Pugh, T.A.M.; Lindeskog, M.; Smith, B.; Poulter, B.; Arneth, A.; Haverd, V.; Calle, L. Role of forest regrowth in global carbon sink dynamics. Proc. Natl. Acad. Sci. USA 2019, 116, 4382–4387. [Google Scholar] [CrossRef]
  10. Gatti, L.V.; Basso, L.S.; Miller, J.B.; Gloor, M.; Gatti Domingues, L.; Cassol, H.L.G.; Tejada, G.; Aragão, L.E.O.C.; Nobre, C.; Peters, W.; et al. Amazonia as a carbon source linked to deforestation and climate change. Nature 2021, 595, 388–393. [Google Scholar] [CrossRef]
  11. Feng, Y.; Ciais, P.; Wigneron, J.-P.; Xu, Y.; Ziegler, A.D.; van Wees, D.; Fendrich, A.N.; Spracklen, D.V.; Sitch, S.; Brandt, M.; et al. Global patterns and drivers of tropical aboveground carbon changes. Nat. Clim. Change 2024, 14, 1064–1070. [Google Scholar] [CrossRef]
  12. Büntgen, U.; Krusic, P.J.; Piermattei, A.; Coomes, D.A.; Esper, J.; Myglan, V.S.; Kirdyanov, A.V.; Camarero, J.J.; Crivellaro, A.; Körner, C. Limited capacity of tree growth to mitigate the global greenhouse effect under predicted warming. Nat. Commun. 2019, 10, 2171. [Google Scholar] [CrossRef] [PubMed]
  13. Pennisi, E. Tropical forests store carbon despite warming. Science 2020, 368, 813. [Google Scholar] [CrossRef] [PubMed]
  14. Friedlingstein, P.; O’Sullivan, M.; Jones, M.W.; Andrew, R.M.; Gregor, L.; Hauck, J.; Le Quéré, C.; Peters, G.P.; Peters, W.; Pongratz, J.; et al. Global Carbon Budget 2022. Earth Syst. Sci. Data 2022, 14, 4811–4900. [Google Scholar] [CrossRef]
  15. Falkowski, P.; Scholes, R.J.; Boyle, E.; Canadell, J.; Canfield, D.; Elser, J.; Gruber, N.; Hibbard, K.; Högberg, P.; Linder, S.; et al. The global carbon cycle: A test of our knowledge of Earth as a system. Science 2000, 290, 291–296. [Google Scholar] [CrossRef]
  16. Qin, Y.; Xiao, X.; Wigneron, J.-P.; Ciais, P.; Brandt, M.; Fan, L.; Li, X.; Crowell, S.; Wu, X.; Doughty, R.; et al. Carbon loss from forest degradation exceeds that from deforestation in the Brazilian Amazon. Nat. Clim. Change 2021, 11, 442–448. [Google Scholar] [CrossRef]
  17. Gauci, V.; Pangala, S.R.; Shenkin, A.; Teh, Y.A.; Richards, M.; Moore, S.; Malhi, Y. Global atmospheric methane uptake by upland tree woody surfaces. Nature 2024, 631, 796–800. [Google Scholar] [CrossRef]
  18. Lewis, S.L.; Sonké, B.; Sunderland, T.; Begne, S.K.; Lopez-Gonzalez, G.; van der Heijden, G.M.F.; Phillips, O.L.; Affum-Baffoe, K.; Baker, T.R.; Banin, L.; et al. The role of Central African forests in regional climate regulation and global carbon cycling. Philos. Trans. R. Soc. B Biol. Sci. 2013, 368, 20120300. [Google Scholar] [CrossRef]
  19. Baccini, A.; Walker, W.; Carvalho, L.; Farina, M.; Sulla-Menashe, D.; Houghton, R.A. Tropical forests are a net carbon source based on aboveground measurements of gain and loss. Science 2017, 358, 230–234. [Google Scholar] [CrossRef]
  20. Dimobe, K.; Kouakou, J.L.N.; Tondoh, J.E.; Zoungrana, B.J.-B.; Forkuor, G.; Ouédraogo, K. Predicting the potential impact of climate change on carbon stock in semi-arid West African savannas. Land 2018, 7, 124. [Google Scholar] [CrossRef]
  21. Ciais, P.; Bombelli, A.; Williams, M.; Piao, S.L.; Chave, J.; Ryan, C.M.; Henry, M.; Brender, P.; Valentini, R. The carbon balance of Africa: Synthesis of recent research studies. Philos. Trans. R. Soc. A Math. Phys. Eng. Sci. 2011, 369, 2038–2057. [Google Scholar] [CrossRef]
  22. Hubau, W.; Lewis, S.L.; Phillips, O.L.; Affum-Baffoe, K.; Beeckman, H.; Cuní-Sanchez, A.; Daniels, A.K.; Ewango, C.E.N.; Fauset, S.; Mukinzi, J.M.; et al. Asynchronous carbon sink saturation in African and Amazonian tropical forests. Nature 2020, 579, 80–87. [Google Scholar] [CrossRef] [PubMed]
  23. Daigneault, A.; Baker, J.S.; Gu, Y.; Liu, Y.; McKeown, A. How the future of the global forest sink depends on timber demand, forest management, and carbon policies. Glob. Environ. Change 2022, 76, 102582. [Google Scholar] [CrossRef] [PubMed]
  24. Aishwarya, B.; Jyotish, P.B. Impact of forest governance and enforcement on deforestation and forest degradation at the district level: A study in West Bengal State, India. Reg. Sustain. 2023, 4, 441–452. [Google Scholar] [CrossRef]
  25. Poulsen, J.R.; Rosin, C.; Meier, A.; Mills, E.; Nuñez, C.L.; Koerner, S.E.; Blanchard, E.; Callejas, J.; Moore, S.; Sowers, M. Ecological consequences of forest elephant declines for Afrotropical forests. Conserv. Biol. 2017, 32, 559–567. [Google Scholar] [CrossRef]
  26. Tyukavina, A.; Hansen, M.C.; Potapov, P.V.; Stehman, S.V.; Smith-Rodriguez, K.; Okpa, C.; Aguilar, R. Congo Basin forest loss dominated by increasing smallholder clearing. Sci. Adv. 2018, 4, eaat2993. [Google Scholar] [CrossRef]
  27. Chen, W.X.; Gu, T.C.; Fang, C.L.; Zeng, J. Global urban low-carbon transitions: Multiscale relationship between urban land and carbon emissions. Environ. Impact Assess. Rev. 2023, 100, 107076. [Google Scholar] [CrossRef]
  28. Ge, J.M.; Lin, B.Q. Convergence or divergence? Unraveling the global development pattern of forest carbon sink. Environ. Impact Assess. Rev. 2024, 105, 107442. [Google Scholar] [CrossRef]
  29. Grieco, E.; Vangi, E.; Chiti, T.; Collalti, A. Impacts of deforestation and land use/land cover change on carbon stock dynamics in Jomoro District, Ghana. J. Environ. Manag. 2024, 367, 121993. [Google Scholar] [CrossRef]
  30. Walid, C. Assessment of vegetation cover changes and the contributing factors in the Al-Ahsa Oasis using Normalized Difference Vegetation Index (NDVI). Reg. Sustain. 2024, 5, 100111. [Google Scholar] [CrossRef]
  31. Zhao, Z.; Ciais, P.; Wigneron, J.-P.; Santoro, M.; Brandt, M.; Kleinschroth, F.; Lewis, S.L.; Chave, J.; Fensholt, R.; Laporte, N.; et al. Central African biomass carbon losses and gains during 2010–2019. One Earth 2024, 7, 506–519. [Google Scholar] [CrossRef]
  32. Bayramli, G.; Karimli, T. Driving factors of CO2 emissions in South American countries: An application of Seemingly Unrelated Regression model. Reg. Sustain. 2024, 5, 100182. [Google Scholar] [CrossRef]
  33. Besnard, S.; Koirala, S.; Santoro, M.; Weber, U.; Nelson, J.; Gütter, J.; Herault, B.; Kassi, J.; N’Guessan, A.; Neigh, C.; et al. Mapping global forest age from forest inventories, biomass and climate data. Earth Syst. Sci. Data 2021, 13, 4881–4896. [Google Scholar] [CrossRef]
  34. Liu, L.C.; Hu, X.T.; Li, L.X.; Sun, Z.X.; Zhang, Q. Understanding China’s agricultural non-carbon-dioxide greenhouse gas emissions: Subnational insights and global trade dynamics. Environ. Impact Assess. Rev. 2024, 106, 107487. [Google Scholar] [CrossRef]
  35. Smith, W.K.; Dannenberg, M.P.; Yan, D.; Herrmann, S.; Barnes, M.L.; Barron-Gafford, G.A.; Biederman, J.A.; Ferrenberg, S.; Fox, A.M.; Hudson, A.R.; et al. Remote sensing of dryland ecosystem structure and function: Progress, challenges, and opportunities. Remote Sens. Environ. 2019, 233, 111401. [Google Scholar] [CrossRef]
  36. Harris, N.L.; Gibbs, D.A.; Baccini, A.; Birdsey, R.A.; de Bruin, S.; Farina, M.; Fatoyinbo, L.; Hansen, M.C.; Herold, M.; Houghton, R.A.; et al. Global maps of twenty-first century forest carbon fluxes. Nat. Clim. Change 2021, 11, 234–240. [Google Scholar] [CrossRef]
  37. Rigatti, E.; McDermid, S.S.; Cook, B.I.; De Kauwe, M.G. The impact of drought on terrestrial carbon in the West African Sahel: Implications for natural climate solutions. J. Geophys. Res. Biogeosci. 2024, 129, e2024JG008143. [Google Scholar] [CrossRef]
  38. Valentini, R.; Arneth, A.; Bombelli, A.; Castaldi, S.; Gatti, R.C.; Chevallier, F.; Ciais, P.; Grieco, E.; Hartmann, J.; Henry, M.; et al. A full greenhouse gases budget of Africa: Synthesis, uncertainties, and vulnerabilities. Biogeosciences 2014, 11, 381–407. [Google Scholar] [CrossRef]
  39. Hall, J.; Sandor, M.E.; Harvey, B.J.; Parks, S.A.; Trugman, A.T.; Williams, A.P.; Hansen, W.D. Forest Carbon Storage in the Western United States: Distribution, Drivers, and Trends. Earth’s Future 2024, 12, e2023EF004399. [Google Scholar] [CrossRef]
  40. Mu, H.; Li, X.; Wen, Y.; Huang, J.; Du, P.; Su, W. An annual global terrestrial Human Footprint dataset from 2000 to 2018. Sci. Data 2021, 9, 176. [Google Scholar] [CrossRef]
  41. Brunsdon, C.; Fotheringham, A.S.; Charlton, M.E. Geographically Weighted Regression: A method for exploring spatial nonstationarity. Geogr. Anal. 1996, 28, 281–298. [Google Scholar] [CrossRef]
  42. Ranjan, A.K.; Gorai, A.K. Assessment of global carbon dynamics due to mining-induced forest cover loss during 2000–2019 using satellite datasets. J. Environ. Manag. 2024, 371, 123271. [Google Scholar] [CrossRef]
  43. Mostefaoui, M.; Ciais, P.; McGrath, M.J.; Peylin, P.; Patra, P.K.; Ernst, Y. Greenhouse gas emissions and their trends over the last 3 decades across Africa. Earth Syst. Sci. Data 2024, 16, 245–275. [Google Scholar] [CrossRef]
  44. Ernst, Y.; Archibald, S.; Balzter, H.; Chevallier, F.; Ciais, P.; Fischer, C.G.; Gaubert, B.; Higginbottom, T.; Higgins, S.; Lawal, S.; et al. The African regional greenhouse gases budget (2010–2019). Glob. Biogeochem. Cycles 2024, 38, e2023GB008016. [Google Scholar] [CrossRef]
  45. Bond-Lamberty, B. New techniques and data for understanding the global soil respiration flux. Earth’s Future 2018, 6, 1176–1180. [Google Scholar] [CrossRef]
  46. Mitchard, E.T.A. The tropical forest carbon cycle and climate change. Nature 2018, 559, 527–534. [Google Scholar] [CrossRef]
  47. Chen, B.; Xu, X.; Wang, S.; Yang, T.; Liu, Z. Carbon dioxide fertilization enhanced carbon sink offset by climate change and land use in Amazonia on a centennial scale. Sci. Total Environ. 2024, 955, 176903. [Google Scholar] [CrossRef]
  48. Lawrence, D.; Vandecar, K. Effects of tropical deforestation on climate and agriculture. Nat. Clim. Change 2015, 5, 27–36. [Google Scholar] [CrossRef]
Fire 08 00333 g001
Figure 1. Spatiotemporal distribution of African forests and Vegetation Index. (a) kNDVI in 2001; (b) LAI in 2001; (c) NPP in 2001; (d) Major distribution patterns of African forests; (e) kNDVI in 2023; (f) LAI in 2023; (g) NPP in 2023.
Figure 1. Spatiotemporal distribution of African forests and Vegetation Index. (a) kNDVI in 2001; (b) LAI in 2001; (c) NPP in 2001; (d) Major distribution patterns of African forests; (e) kNDVI in 2023; (f) LAI in 2023; (g) NPP in 2023.
Fire 08 00333 g001
Fire 08 00333 g002
Figure 2. Temporal variation in kNDVI (a), LAI (b), and NPP (c) in African forests.
Figure 2. Temporal variation in kNDVI (a), LAI (b), and NPP (c) in African forests.
Fire 08 00333 g002
Fire 08 00333 g003
Figure 3. Cumulative carbon emission density from 2001 to 2023.
Figure 3. Cumulative carbon emission density from 2001 to 2023.
Fire 08 00333 g003
Fire 08 00333 g004
Figure 4. Cumulative carbon removal density from 2001 to 2023.
Figure 4. Cumulative carbon removal density from 2001 to 2023.
Fire 08 00333 g004
Fire 08 00333 g005
Figure 5. Spatial distribution of cumulative net carbon flux from 2001 to 2023.
Figure 5. Spatial distribution of cumulative net carbon flux from 2001 to 2023.
Fire 08 00333 g005
Fire 08 00333 g006
Figure 6. Spatiotemporal characteristics of wildfires and their impact on forest carbon fluxes in Africa (2001–2023). (a) Wildfire frequency (number of years with fire occurrence); (b) Carbon flux in burned areas (Mg C·CO2/ha).
Figure 6. Spatiotemporal characteristics of wildfires and their impact on forest carbon fluxes in Africa (2001–2023). (a) Wildfire frequency (number of years with fire occurrence); (b) Carbon flux in burned areas (Mg C·CO2/ha).
Fire 08 00333 g006
Fire 08 00333 g007
Figure 7. Spatial distribution of environmental driving factors in Africa. (a) Precipitation (mm); (b) Temperature (°C); (c) RH (%); (d) Wind Speed (m/s); (e) PAR (W/m2); (f) SOC (g/kg); (g) SM (kg/m2); (h) Elevation (m); (i) HFP.
Figure 7. Spatial distribution of environmental driving factors in Africa. (a) Precipitation (mm); (b) Temperature (°C); (c) RH (%); (d) Wind Speed (m/s); (e) PAR (W/m2); (f) SOC (g/kg); (g) SM (kg/m2); (h) Elevation (m); (i) HFP.
Fire 08 00333 g007
Fire 08 00333 g008
Figure 8. Analysis of factors influencing cumulative carbon emissions. Note: all spatial patterns are based on statistically significant relationships at the p < 0.01 level.
Figure 8. Analysis of factors influencing cumulative carbon emissions. Note: all spatial patterns are based on statistically significant relationships at the p < 0.01 level.
Fire 08 00333 g008
Fire 08 00333 g009
Figure 9. Analysis of factors influencing cumulative carbon removals. Note: all spatial patterns are based on statistically significant relationships at the p < 0.01 level.
Figure 9. Analysis of factors influencing cumulative carbon removals. Note: all spatial patterns are based on statistically significant relationships at the p < 0.01 level.
Fire 08 00333 g009
Table 1. Overview of data sources.
Table 1. Overview of data sources.
VariablesAbbreviationsData NameData Sources
Leaf Area IndexLAI https://lpdaac.usgs.gov/products/mod15a2hv006/ (accessed on 1 January 2025)
Kernel Normalized Difference Vegetation IndexkNDVIhttps://github.com/IPL-UV/kNDVI (accessed on 1 January 2025)
Net Primary ProductivityNPPhttps://lpdaac.usgs.gov/products/mod17a3hgfv006/ (accessed on 1 January 2025)
Wildfire Frequency and Burned AreaFireFreq
FireDummy		Fire Datahttps://developers.google.com/earth-engine/datasets/catalog/MODIS_061_MCD64A1 (accessed on 1 January 2025)
PrecipitationPREMeteorological Datahttps://www.chc.ucsb.edu/data/chirps (accessed on 1 January 2025)
TemperatureTEMPhttps://cds.climate.copernicus.eu/datasets/reanalysis-era5-single-levels-monthly-means?tab=overview (accessed on 1 January 2025)
Relative HumidityRHhttps://cds.climate.copernicus.eu/datasets/reanalysis-era5-single-levels-monthly-means?tab=overview (accessed on 1 January 2025)
Wind SpeedWINDhttps://cds.climate.copernicus.eu/datasets/reanalysis-era5-single-levels-monthly-means?tab=overview (accessed on 1 January 2025)
Photosynthetically Active RadiationPARhttps://www.glass.hku.hk/download.html (accessed on 1 January 2025)
Soil Organic CarbonSOCSoil and Forest Change Datahttps://data.isric.org/geonetwork/srv/eng/catalog.search#/metadata/soilgrids-global-soil-organic-carbon-map (accessed on 1 January 2025)
Soil MoistureSMhttps://disc.gsfc.nasa.gov/datasets/GLDAS_NOAH025_3H_2.1/summary (accessed on 1 January 2025)
ElevationELEhttps://developers.google.com/earth-engine/datasets/catalog/USGS_SRTMGL1_003 (accessed on 1 January 2025)
Human FootprintHFPTerrestrial Human Footprinthttps://figshare.com/articles/figure/An_annual_global_terrestrial_Human_Footprint_dataset/16571064 (accessed on 1 January 2025)
Carbon EmissionsEmissionsCarbon Flux DataForest Carbon dataset (https://www.globalforestwatch.org/blog/data/whats-new-carbon-flux-monitoring/) (accessed on 1 January 2025)
Carbon RemovalsRemovals
Net Carbon FluxNet Flux
Table 2. Summary of wildfire-affected carbon flux metrics across African countries (2001–2023).
Table 2. Summary of wildfire-affected carbon flux metrics across African countries (2001–2023).
CountryFire Area
(km2)		Affected Carbon Area (km2)Mean Fire FracAnnual Emission
(MtC)		Annual Removal
(MtC)
Congo, Democratic Republic of the940,912.3934,704.80.546618.016492.2315
Angola904,603.2753,7730.62756.953945.0181
Zambia610,765.6512,7190.59486.463920.1749
Mozambique587,616.1533,244.90.56449.424122.3988
South Sudan555,296.1328,643.20.75840.145827.1581
Central African Republic465,633422,605.50.61850.484929.1352
Tanzania459,216.4376,533.80.46795.386419.951
Nigeria427,690.7192,9650.36072.909221.671
Botswana374,426.75105.4920.28060.00030.4789
Sudan321,524.142,689.390.41220.01664.4818
Chad316,271.771,286.620.51080.62454.7825
South Africa315,869.3120,737.80.22864.93486.512
Mali249,196.977,881.610.47040.21658.2113
Madagascar241,693.4184,100.20.3087.22227.895
Namibia205,768.34238.8290.32570.02440.2784
Guinea189,796.2172,229.70.39294.64226.4092
Ethiopia168,281126,705.40.54350.221810.7384
Cameroon166,420.3126,836.40.38960.95359.2499
Ghana154,100.175,019.490.64710.68097.4128
Zimbabwe152,259.881,345.920.37110.84095.236
Côte d’Ivoire128,898.8118,9790.33131.41078.7812
Senegal114,103.825,217.20.61620.01462.7154
Burkina Faso112,592.642,222.40.37180.00394.8604
Congo103,093.692,458.380.50590.322512.6518
Uganda98,362.6770,9390.48040.16326.4849
Benin73,906.0343,875.140.42320.50414.2038
Sierra Leone50,865.0850,644.520.31062.24511.3674
Togo41,536.1527,231.80.43670.21352.6126
Kenya36,244.1210,292.410.1460.16920.7658
Malawi29,629.4126,113.970.31590.40421.2111
Mauritania26,327.521.90480.1900.0001
Niger25,832.5714,74.7720.208400.1007
Guinea-Bissau19,257.8817,101.30.42050.25961.3956
Algeria15,491.6611,381.460.11750.15490.5599
Gabon14,586.812,028.110.41760.02071.7085
Liberia14,37114,255.410.09951.1790.1642
Gambia7351.4884731.0650.52670.00310.4412
Lesotho5806.155706.96780.16170.00040.0417
Swaziland (Eswatini)5668.51656,21.6380.24370.55080.3016
Egypt4479.151949.8460.18770.00060.0581
Somalia4054.975677.28490.10740.00450.0191
Eritrea3254.457161.32110.179500.0136
Burundi2504.6232479.8390.27140.00620.1841
Rwanda1720.6081386.9680.34440.00420.1451
Tunisia1374.646648.06340.10.00940.031
Morocco1298.805938.03890.09140.0330.0317
Libya257.542348.39160.084700.0028
Comoros213.9589164.37170.11020.00010.0042
Mauritius75.526169.00740.10080.00020.0013
Djibouti35.035800.08800
Reunion29.649827.79550.08590.00010.0006
Equatorial Guinea8.9338.9330.08330.00010.0005
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Zhang, L.; Zhang, Z. Wildfires and Climate Change as Key Drivers of Forest Carbon Flux Variations in Africa over the Past Two Decades. Fire 2025, 8, 333. https://doi.org/10.3390/fire8080333

AMA Style

Zhang L, Zhang Z. Wildfires and Climate Change as Key Drivers of Forest Carbon Flux Variations in Africa over the Past Two Decades. Fire. 2025; 8(8):333. https://doi.org/10.3390/fire8080333

Chicago/Turabian Style

Zhang, Lianglin, and Zhenke Zhang. 2025. "Wildfires and Climate Change as Key Drivers of Forest Carbon Flux Variations in Africa over the Past Two Decades" Fire 8, no. 8: 333. https://doi.org/10.3390/fire8080333

APA Style

Zhang, L., & Zhang, Z. (2025). Wildfires and Climate Change as Key Drivers of Forest Carbon Flux Variations in Africa over the Past Two Decades. Fire, 8(8), 333. https://doi.org/10.3390/fire8080333

Article Metrics

Back to TopTop