Substantial increase in population exposure to multiple environmental burdens in sub-Saharan Africa (2000-2019)

In the face of increasing global environmental uncertainties, sub-Saharan Africa stands as a highly vulnerable region with a massive population marked with poverty and inequalities. Moreover, different environmental risk factors can coexist simultaneously as multiple environmental burdens (MEBs); however, population exposure to MEB remains unexamined. Here, using open-access spatial data and critical thresholds, we quantify population exposure to four key environmental risk factors: hazardous fine particulate matter (PM2.5) levels, extreme temperature increase, prolonged severe droughts, and green deficit (scarcity of green trees). Further, we explore the concept of MEB, where these risk factors converge. We derive exposure for 2000 and 2019 at the pixel (1 km grid cell) level. We also check how population change, environmental change, and their interaction contribute to the total change in exposure. We found substantial changes in the population exposed from 2000 to 2019, i.e. an increase of ∼460 million people to hazardous PM2.5 levels, ∼16 million to extreme temperature increase, ∼13 million to prolonged severe droughts, and ∼246 million to green deficit. Population exposure to at least three of these four environmental risk factors (3EB) has increased by ∼246 million. In this increase in exposure to 3EB, the contribution of environmental change is higher (48%), than that of interaction and population change (36% and 15%, respectively). Notably, there are striking disparities in population exposure, its change, and the contributing effects among countries and regions of sub-Saharan Africa.


Introduction
Environmental uncertainties have increased worldwide.Unprecedented episodes of heat waves, droughts, and floods have been striking different regions in recent decades [1].Half of the global population is exposed to increasing fine particulate matter (PM 2.5 ) air pollution [2].Approximately 420 million hectares of forest loss has been reported due to conversions to other land uses since 1990 [3]. Surging extreme weather events and pollution levels can pose negative consequences for health and well-being.According to the World Health Organization, in 2016, up to 24% of all deaths worldwide were due to adverse environmental conditions [4].However, these challenges are far from uniform, varying significantly in magnitude and impact at international, national, and subnational scales.
Sub-Saharan Africa remains at higher risk due to intensifying environmental challenges [5].The literature suggests rising temperature levels [6][7][8], deteriorating air quality [9], recurring droughts [10,11], and rampant land use and land cover change [12], among others, as key environmental challenges posing adverse impacts on the health and well-being of over a billion people.The adversities aggregate as much of its population remains below the poverty line, i.e. representing around two-thirds of the global extreme poor population in 2018 [13].Additionally, many of the countries on the subcontinent are marked by large inequalities and little social protection [14], meaning that much of their population is likely to be unable to absorb the shocks related to a changing environment.Moreover, the lack of sufficient and timely environmental monitoring on the ground is another issue.For instance, the World Air Quality Report 2022 [15] captures data from over 30 000 air quality monitoring stations, of which only 156 are located in Africa.Another study [16] highlights only 1 ground-level air quality monitor per 15.9 million people in sub-Saharan Africa.These conditions limit the power of national or international indicators to explain spatially varying environmental conditions.In light of these limitations, the role of publicly accessible spatial data on the environment is immense, allowing temporally and spatially efficient environmental monitoring and informing targeted adaptation strategies.
Environmental exposure studies [6,8,[17][18][19]] have demonstrated the wide applicability of remote sensing and geospatial data to link specific environmental indicators with demographic and health indicators.However, it is imperative to recognize that different environmental challenges often coexist in the same geographical location, compounding risks for the populations residing there [20].
In the context of this study, we focus on hazardous levels of PM 2.5 , temperature increase, prolonged drought severity, and green deficit (scarcity of green trees) as key environmental risk factors, capturing climate and environmental change in their different dimensions.We report the simultaneous presence of more than one of these risk factors as 'multiple environmental burdens' (MEBs).We quantify the population exposed to each of these key environmental risk factors and to MEB in sub-Saharan Africa at the finest spatial resolution (1 km grid cell) for 2000 and 2019 and analyze the changes in exposure over this period.We also ask: What is the contribution of population change, environmental change, and their interaction to the change in exposure to specific environmental risk factors and MEB?The findings are then aggregated at national, regional, and subcontinental levels, providing a comprehensive understanding of the population exposed to environmental burdens in sub-Saharan Africa.

Data
Environmental burdens can be estimated on the basis of several indicators representing key environmental challenges.We have selected four major indicators measurable at a fine level, to partially account for global environmental challenges: air pollution (with PM 2.5 ), climate change (with temperature increase), extreme events (with prolonged severe drought), and land cover change (FCover vegetation index).To derive population and environmental indicators we employed publicly available raster data (geospatial pixelated data where each pixel presents the value of the defined indicator).The gridded population estimates are derived from the modeling of multiple spatial layers and population estimates from the census/surveys/United Nations' projected population [21].All data were spatially processed (clipped, reprojected, rescaled, and aggregated using zonal statistics) using QGIS.The years 2000 and 2019 were selected to capture long-term changes and to avoid the temporary impact of COVID-19 restrictions on environmental parameters.Further, the selected years were also devoid of any major climate abnormalities associated with El Niño and La Niña occurrences.Such phases of abnormalities can deviate from the normal precipitation patterns and temperature levels, with which the studied risk factors are strongly associated.
Table 1 lists the details of the data used in this study.

Environmental burden and population exposure
We set thresholds to define the environmental risk factors as: (a) Hazardous PM 2.5 levels: pixels with values above 20 µg m −3 (annual average).The air quality guidelines (AQGs) recommended by the World Health Organization [25] suggest that PM 2.5 above 5 µg m −3 is harmful, which is also referenced in many global studies dealing with their health impacts [2,26].Unsurprisingly, at this threshold, about 99% of the world's population is considered to be exposed to air pollution [27].To detect extremely harmful levels of PM 2.5 and to relate the approach to globally accepted standards, we have chosen 20 µg m −3 as an appropriate threshold.Although four times higher than the AQG, this value falls between the WHO's second and third interim targets [25].PM 2.5 is estimated by modeling aerosol optical depth estimates from various satellite instruments and then adjusting the retrievals to global ground-based observations of PM 2.5 using geographically weighted regression [22].
(b) Extreme temperature increase: pixels with a 1 • C increase or more, in average annual temperature compared to twenty years before (1980 is the reference year for 2000, and 2000 for 2019).This threshold represents about three times the global warming trend since 1981 (+0.18 • C per decade) [28].Previous studies measured the number of days above certain temperature thresholds [29,30] in order to capture heatwaves.In this study, we chose to capture the highresolution change in average temperature over two decades that can account for the actual trends in temperature.In addition to the health impact, the average temperature increase can be linked to multiple aspects affecting the economy and sustainable development.
(c) Prolonged severe drought: pixels with PDSI ⩽−3 for at least four months during the year and/or in  [31], this index allows analyses at finer resolutions for the selected time period.
(d) Green deficit: pixels with an average FCover value of less than 0.3, that is, where less than 30% of the 1 km grid cell holds vegetation (tree canopy cover).This tree cover canopy represents the amount of greenery by trees (continuous/scattered) across forest/rural/urban lands.Here, we derived data on FCover for the same month (March) for each year to minimize seasonal variations in green canopy cover.Moreover, we averaged two years of data for each time period (average of 1999-2000 for 2000 and average of 2018-2019 for 2019) to avoid the impact of abovenormal precipitation during any specific year that can increase the green canopy extent significantly.Although a single risk factor at these thresholds may not pose a serious challenge to population exposure, the presence of MEBs at the same time and in the same place may cause serious detrimental impacts on the health and overall wellbeing of the exposed population.To estimate the coexistence of multiple risk factors, we have attempted to estimate MEB.To create raster layers of MEB, the four criteria-specific raster data were resampled and rescaled at ∼1 km resolution and superimposed.First, all four layers of risk factors were given binary values (0 and 1), where 1 represents the existence of a particular risk factor (pixel values above/below the set thresholds).Second, we added these four layers using the raster calculator to have a single-band raster with pixel values 0 (meaning absence of all risk factors) to 4 (meaning presence of all risk factors).Finally, we constructed three successive raster layers representing the presence of at least two (2EB), three (3EB), or four (4EB) of the abovementioned risk factors per pixel.To estimate the exposure, we multiply the population in each grid cell with each criteria-specific raster layer.We employed a similar approach as Jones et al [18] and subsequent exposure studies [27,[32][33][34][35].The exposed population data were aggregated to country borders using zonal statistics.

Environmental and population effects on exposure
The change in population exposure to an environmental risk factor can be caused by population change, change in the area undergoing this environmental risk factor, or by the interaction of these two elements over time.Following previous studies [8,18,36], we decomposed the total change in exposure from 2000 to 2019 (∆Exp) into three components: the environmental effect, defined by the product of the population from base year and the change in the area under environmental risks (P 2000 × ∆Env); the population effect, defined by the product of the population change from 2000 to 2019 and the area under extreme environmental conditions at base year (∆P × Env 2000 ); and the interaction effect, that can be derived from the previous terms (by subtracting the sum of population and environmental effects from total exposure) but comes to the product of population change by the area under environmental change (∆P × ∆Env).The ∆Exp is expressed as: where: P 2000 and ∆P are expressed in absolute numbers, ∆P being positive or negative; Env 2000 is binary (0 when a pixel lies under the chosen threshold, 1 if it lies above) and ∆Env takes the values −1, 0 or 1.The three effects are the results of the addition of positive and negative terms, the negative ones occurring for pixels where areas under environmental risk are improving (∆Env = −1) or where population is decreasing (∆P ⩽ 0).Globally, the interaction effect will be higher when there is simultaneous population growth and environmental degradation (as per our indicators) in one place, which is a common case but not the only scenario found in sub-Saharan Africa over the period 2000-2019, as we shall see in the results.
To obtain the percentage of each effect, we divided each effect by the total change in exposure ∆Exp and multiplied by 100.A reduction in population or in the area under environmental risk results in negative terms in (1) and can lead to a decrease in exposure.
In addition to these, we have provided extended analysis in the supplementary material to indicate how a different set of thresholds (for two of the risk factors) can change the exposure estimates and spatial patterns of MEB.The extended analyzes also include results specific to four urban regions.

Population exposure to hazardous PM 2.5 levels
Figure 1 illustrates population exposure to hazardous PM 2.5 levels in 2000 and 2020 (figures 1(a) and (b)) and country-level change, presented as absolute values in figure 1(c) and as percentages of the total population in supplementary table S1.The data reveal high PM 2.5 levels for both years as well as pronounced spatial disparities across countries that are not necessarily reflected in exposure since population distribution over the subcontinent is highly uneven and concentrated in a few regions.In sub-Saharan Africa, the population exposed to hazardous PM 2.5 levels rose from approximately 438 million (69%) in 2000 to about 898 million (80%) in 2019.At the regional level, West Africa exhibits markedly higher PM 2.5 concentrations, contrasting with Southern Africa, which experiences the lowest levels.The escalation is particularly significant in densely populated areas and regions experiencing notable population growth.Notably, it is the Central African region that exhibits the highest population exposure (98.5%).
At the country level, exposure to hazardous PM 2.5 levels has markedly increased in densely populated countries between 2000 and 2019.For instance, in Nigeria, the exposed population rose from approximately 115.65 million (99%) to about 206.59 million (99%), in the Democratic Republic of Congo from around 43.9 million (96%) to approximately 108.19 million (100%), and in Ethiopia from roughly 13.54 million (22%) to about 60.53 million (60%).Additionally, Eastern African countries such as Sudan, Uganda, Tanzania, and Kenya experienced a notable increase in exposure levels.In contrast, marginal reductions in exposure were observed in Eswatini, Namibia, and Botswana.However, South Africa, despite its proximity to these three countries, witnessed a significant increase in exposure, rising from approximately 24 million to about 35 million.

Population exposure to extreme temperature increase
The spatial patterns of temperature increase across sub-Saharan Africa and populated areas are presented in figures 2(a) and (b) for 2000 and 2019, respectively.The temperature increase is measured relatively to levels from two decades prior (1980-2000 for 2000, 2000-2019 for 2019).The visual representations clearly indicate a high temperature increase across most of the subcontinent.Interestingly, while temperature rise was not a significant concern in 2000, it became prevalent over the subsequent decades.However, in 2019, population concentrations were primarily situated in areas where temperature increase was not extreme (figure 2(b)).
Our analysis focuses on populations residing in areas with extreme temperature increase (exceeding 1 • C). Figure 2(c) provides country-level aggregated statistics of the exposed population for 2000 and 2019, displayed in absolute values, while percentages of the total population are presented in supplementary table S1.In sub-Saharan Africa, the population exposed to extreme temperature increase rose from approximately 30 000 (0.01%) in 2000 to around 16 million (1.4%) in 2019.At the regional level, Southern Africa accounted for the largest share of the exposed population (2.6%) compared to other regions.A noticeable heterogeneity in exposure levels is evident at the country level.Remarkably, in 2019, countries with relatively high population exposure included Ethiopia (∼2.24 million), DR Congo (∼1.93 million), South Africa (∼1.74 million), and Angola (∼1.15 million), among others.The significantly lower change in exposure in Nigeria embodies the little influence of population size or growth on the levels of exposure.Some countries, such as Guinea-Bissau, Gambia, and Burundi had close to zero exposure to extreme temperature increase in 2019.S1).East Africa exhibits a relatively higher share of population exposure, accounting for 2% in 2000 and 2.5% in 2019, in comparison to other regions.However, a closer country-level examination reveals significant disparities in exposure and exposure change.In 2019, countries with high population exposure included highly populated nations such as DR Congo (∼3.6 million or 3.3%), Ethiopia (∼2.3 million or 2.3%), Kenya (∼1.8 million or 3.5%), South Africa (∼1.7 million or 2.9%), and Nigeria (1.2 million or 0.6%).Notably, the changes in exposure also differ among these countries.For instance, South Africa experienced a substantial increase in exposure (1.6 million), whereas Nigeria witnessed only a marginal rise (48 000) from 2000 to 2019.Conversely, other countries experienced a decline in exposure over this period: Madagascar (−371 000), Sierra Leone (−90 000), and Zambia (−80 000).These varied exposure and changes seem to stem from both the varying severity of drought over space and time and population growth in affected areas.

Population exposure to green deficit
FCover indicates the fraction of green vegetation cover in a unit area, based on a two-year average of measures taken in the month of March, and ranges from 0 to 1.We considered green deficit as FCover values lower than 0.3.This threshold signifies areas with very limited green tree cover, which can potentially add to challenges like urban heat islands and also highlight the areas where better land-use planning is needed.
As illustrated in figures 4(a) and (b), the greener areas are concentrated mainly over the central parts of the subcontinent in both years.Overall FCover levels in sub-Saharan Africa have remained largely stable, especially in unpopulated regions.However, a contrasting trend is observed over populated areas, where serious challenges are evident.The population exposed to green deficit has more than doubled over 20 years, increasing from approximately 291 million (45.6%) in 2000 to approximately 537 million (48%) in 2019 (table S1).Regionally, West and East Africa bear the highest population exposure, accounting for 68.3% and 62.4% in 2019, respectively.
Country-level exposure and its changes are detailed in figure 4(c) (absolute values) and supplementary table S1 (percentages).Highly populated countries exhibited significantly high exposure levels and notable changes over time, including Nigeria, from ∼71 million (61%) to ∼140 million (67%), Ethiopia, from ∼53 million (87%) to ∼72 million (71%), Sudan, from ∼25 million (99%) to ∼40 million (97%), Kenya, from ∼15 million (52%) to ∼23 million (45%), and Niger, from ∼11 million (100%) to ∼23 million (100%).Rapid increases in both the absolute and percentage of the population exposed were observed in several countries, such as South Africa, Angola, and Ghana.Remarkably, despite higher exposure levels in 2019, the percentage of the population exposed to green deficit decreased in many countries.A negative change in exposure was observed only in Lesotho, where the population remained stable from 2000-2019, contributing to an improvement in FCover values.

Population exposure to MEBs
Figures 5(a) and (b) illustrate the spatial distribution of MEBs, alongside populated areas for 2000 and 2019, respectively.The analysis identifies areas characterized by at least two (2EB), three (3EB), or four (4EB) environmental risk factors.A noticeable expansion in the regions concerned by MEB is evident over time.In 2000, the majority of the subcontinental area exhibited 2EB.However, by 2019, there was a significant increase in areas with 3EB, and patches characterized by all four risk factors (4EB) became apparent.Figure S2 illustrates the spatial patterns of change in MEB over the past twenty years.Although, positive change dominates the regions, there are areas with no and negative change in the MEB (figure S2).
We reported exposure by identifying the number of people living in each grid cell with MEBs (2EB, 3EB, and EB) for both years.Table 2 shows total and country-level exposure numbers.
At the subcontinental level, the population exposed to MEB experienced a remarkable increase over the study period.Specifically, there was an increase in population exposure from approximately 300 million to about 465 million for 2EB, from around 47 million to approximately 292 million for 3EB, and from none to approximately 92 million for 4EB.In 2019, several East African countries witnessed substantial population exposure to 4EB, including Sudan (∼22.3 million), Ethiopia (∼18 million), South Sudan (∼7.6 million), and

Environmental and population effects on exposure
To elucidate the factors driving change in exposure to different environmental risk factors from 2000 to 2019, we have distinguished the components of this exposure change (environmental effect, population effect, and interaction effects) for sub-Saharan Africa (figure 6).
About 460 million more people are exposed to hazardous PM 2.5 levels in 2019.The population effect accounts for the major part of this change in exposure, contributing 81% (∼373 million) of the total (figure 6(a)), whereas the environmental effect constitutes only 11% (∼50 million) and the interaction effects only 8% (∼37 million) of the total.This underscores the pivotal role of the significant population increase, primarily in areas where PM 2.5 levels have consistently exceeded the threshold over time.
About 16 million more people are exposed to extreme temperature increase.The population effect Here, the change in exposure is predominantly influenced by the environmental effect, contributing to 57% (∼9 million) of the total, and the interaction effect accounting for 43% (∼7 million) of the total change (figure 6(b)).This pattern underscores the critical role played by the expansion of areas with extreme temperature increases and the synergistic interactions between these areas and the changing population in driving the surge in exposure.
In the change in exposure to prolonged severe drought (∼13 million), all three components play significant roles, with the population effect contributing 53% (∼7 million) of the total, while the environmental and interaction effects account for 27% (∼3.6 million) and 20% (∼2.7 million) of the total, respectively (figure 6(c)).This distribution highlights the dynamic interplay between changes in population and areas facing severe drought, driving enhanced exposure over time.
In the context of the change in exposure to green deficit (∼246 million), the population effect significantly dominates, contributing 105% (∼258 million) to the total change.This is partially countered by the environmental effect (−3% or ∼ −7 million) and interaction effect (−2% or ∼ −5 million) to the total change in exposure (figure 6(d)).These results shed light on a situation where population increase is notably high in areas characterized by green deficit.Additionally, it suggests slight improvements in green cover from 2000 to 2019, indicated by negative values for both environmental and interaction effects.
To analyze the change in population exposure to MEBs, we reported changes in exposure to at least three risk factors (3EB) and their contributing factors (figure 6(e) for sub-Saharan Africa and supplementary figure S1 for country-specific effects).The total change in exposure to 3EB (∼246 million) attributed to the environmental effect constitutes 48% (∼119 million) of the total, whereas the change caused by interaction effects accounts for 36% (∼89 million) of the total.In contrast, the population effect contributes only 15% (∼37.6 million) of the total.This underscores the remarkable expansion of areas with 3EB, coinciding with a simultaneous increase in population over time.Similar patterns of change in exposure to 3EB were observed in the majority of the countries, where environmental effects played a more significant role in driving country-specific changes in exposure (figure S1).

Discussion
We investigated population exposure to environmental risk factors-hazardous PM 2.5 levels, extreme temperature increases, prolonged severe drought, green deficit-and MEBs across sub-Saharan Africa from 2000 to 2019.Our results indicate a significant escalation in specific environmental risk factors and MEB, which, in conjunction with population growth, leads to an increasing population exposure.
The extreme levels and increases in PM 2.5 concentrations have been the highlights of previous studies [37][38][39].Multiple sources of PM 2.5 can be identified.Dust from the Sahara desert contributes to higher concentrations of PM 2.5 in parts of western sub-Saharan Africa, while fossil fuel use (coal, liquid oil, and natural gas) is a major contributor in southern Africa [26].The observed spatial patterns highlight a stark contrast in PM 2.5 concentrations across different regions.This spatial heterogeneity in air pollution is contributed from the complex interplay of environmental factors [40,41], population density [42], and urbanization [9].Our analyses found a prominent role of population increase in the unprecedented surge in population exposure to extreme PM 2.5 levels (∼460 million).South Africa, despite being geographically close to countries with exceptionally lower exposure (Eswatini, Namibia, and Botswana), experienced a notable increase in exposure levels.This spatial variation within neighboring countries emphasizes the localized nature of air pollution sources (particularly transport emissions) [43], especially in the most urbanized countries [26].The steep rise in temperatures over the past 20 years has highlighted the challenges of global warming and climate change on the subcontinent.These conditions present severe challenges to public health [6,7], socioeconomic stability [44] agriculture [45,46], and the overall development of a region.We find an alarming increase in population exposure (from 30 000 to 16 million), emphasizing the growing exposure of populations to climate change in the form of increasing temperature.Exposure at the national level indicates regional disparities in the impact of rising temperatures on populations.Interestingly, the study identifies Nigeria as an exception, with a significantly lower change in exposure despite its large population.At the subcontinental level, the change in exposure is largely driven by the expansion of areas with extreme temperature increase and by the interaction effect rather than by population growth itself.
The increased exposure to prolonged severe drought highlights a growing vulnerability of communities to water scarcity and agricultural challenges [47][48][49].Interestingly, the changes in exposure over time vary significantly among countries.For instance, South Africa saw the highest increase in exposure, with approximately 1.6 million more people exposed to prolonged severe drought in 2019.In contrast, Nigeria experienced a marginal increase in exposure, emphasizing the complex interplay of factors influencing vulnerability, including population size and varying drought severity over space and time.At subcontinental levels, the change in exposure to prolonged severe drought is contributed by all three factors, i.e. change in population, change in the areas under prolonged drought, and interaction between these two.
The greening of African drylands has already been reported [50].A recent study by Wei et al [51].Reports gains in woodland forests, rainforests, and savannas from 2000 to 2020, with spatially varying patterns.However, as little population is located in these areas, our results show an alarming increase in population exposure to green deficit (∼246 million) underscoring growing land use challenges [52].Higher exposure to green deficit is also a characteristic of high population density in the urban regions that are devoid of green spaces, and are at higher subsequent risks such as heat exposure [53], air pollution [54], and associated health problems [55].Highly populated countries experienced increasing exposure to green deficit.Population increase substantially offsets greening effects and drives an increase in exposure to green deficit over time.Despite the high levels of exposure to green deficit in 2019, it is noteworthy that the share of the population exposed to green deficit has declined in some countries.This decline could be attributed to various factors, including potential afforestation efforts, reforestation projects, or natural regrowth of vegetation [50].
Our findings showcase a substantial increase in MEB and exposed population.In 2019, specific countries in East Africa (Sudan, Ethiopia, South Sudan, and Kenya) experienced significant exposure to 4EB.Other countries, including South Africa and Ghana, also reported high levels of exposure to 4EB, highlighting the widespread nature of this issue.The DR Congo experienced a remarkable escalation in population exposure to 3EB, from 0.6 million to 44.2 million (followed by Ethiopia, Uganda, South Africa, Kenya, and Ghana).The increase in exposure to 3EB is mainly due to environmental change and its interaction with population change.

Conclusion
Our results emphasize the urgency of addressing the challenges of increasing population exposure to environmental risk factors for the well-being of the population and the sustainability of the environment.Given the spatially heterogeneous patterns of population exposure, there is a strong need for targeted regional interventions, keeping in focus the relative roles of population and environmental factors in driving exposure change.There is an urgent need for air pollution and climate change mitigation [56,57] in sub-Saharan Africa.This situation demands strategies to face the health risks associated with prolonged exposure to high PM 2.5 levels [58] as well as to enhance water resource management, promote sustainable agricultural practices, and improve community resilience to heat and drought-related challenges.
Our study is a pioneering effort in understanding MEB dynamics in sub-Saharan Africa from 2000 to 2019.The critically defined thresholds made it possible to identify populated areas with extreme environmental risks and MEB as per our objectives.Our approach, at a 1 km grid cell, offers crucial information for policy formulation and targeted interventions at national, regional, and subcontinental levels, with the possibility of defining different thresholds.
However, our study has limitations.Social and demographic indicators were limited, hindering our understanding of population vulnerability and adaptive capacity.We did not fully account for extreme events, which can impact environmental parameters [48,59] and exposure levels.Defining environmental burdens solely based on four parameters is restrictive.There is a need to incorporate more environmental factors into MEB analysis.
Integrating social and demographic indicators into future studies would provide a holistic understanding of the interplay between environmental challenges and human societies.Exploring the links between MEB and public health outcomes could deepen our understanding of long-term environmental exposure effects on human well-being.Since environmental indicators have different levels and severity in urban and rural places, future studies can distinguish exposure between them.Additionally, including more parameters and conducting regionspecific analyses could offer valuable insights into localized environmental challenges, aiding in the development of targeted mitigation strategies.

3. 3 .
Population exposure to prolonged severe drought Figures 3(a) and (b) illustrate the spatial patterns of the Palmer drought severity index (PDSI) alongside population concentrations in 2000 and 2019, respectively.According to our definition, prolonged severe drought conditions correspond to areas experiencing PDSI ⩽−3 for at least four months during the year and/or previous year.Country-level estimates for 2000 and 2019 are depicted in figure 3(c) (absolute values) and supplementary table S1 (percentages).An evident escalation in areas facing drought severity is observable across the subcontinent from 2000 to 2019.The population exposed to prolonged severe drought has increased substantially, rising from approximately 9.35 million (1.5%) in 2000 to about 22.68 million (2%) in 2019 (figure 3(c), table

Figure 1 .
Figure 1.PM2.5 concentrations and population exposure.(a) Spatial patterns of PM2.5 (from grey to red) and populated areas (pixels with more than 100 persons) in 2000 (black); (b) spatial patterns of PM2.5 and populated areas (pixels with more than 100 persons) in 2019; (c) country-wide levels of population exposed to hazardous PM2.5 (>20 µg m3).The three most populated countries-DR Congo (Central Africa), Ethiopia (East Africa), and Nigeria (West Africa)-are presented on a separate scale.

Figure 2 .
Figure 2. Temperature increase and population exposure.(a) Spatial patterns of temperature increase (sky blue to orange) and populated areas (pixels with more than 100 persons) in 2000 (black); (b) spatial patterns of temperature increase and populated areas (pixels with more than 100 persons) in 2019; (c) country-wise levels of population exposed to extreme temperature increase (>1 • C increase in average annual temperature compared to the reference year -twenty years-before).The three most populated countries-DR Congo (Central Africa), Ethiopia (East Africa), and Nigeria (West Africa)-are presented on a separate scale.

Figure 3 .
Figure 3. Drought severity and population exposure.(a) Spatial patterns of average PDSI of 24 months-1999 & 2000 (tan to slate grey) and populated areas (pixels with more than 100 persons) in 2000 (black); (b) spatial patterns of PDSI (average of 24 months-2018 & 2019) and populated areas (pixels with more than 100 persons) in 2019; (c) country-wise levels of population exposure to prolonged severe drought (PDSI ⩽−3 for at least four months during the year and/or in the previous year).The three most populated countries-DR Congo (Central Africa), Ethiopia (East Africa), and Nigeria (West Africa)-are presented on a separate scale.

Figure 4 .
Figure 4. Green deficit and population exposure.(a) Spatial patterns of average FCover of 1999 & 2000 (grey to green) and populated areas (pixels with more than 100 persons) in 2000 (black); (b) spatial patterns of FCover (average of 1999 & 2000) and populated areas (pixels with more than 100 persons) in 2019; (c) country-wise levels of population exposure to green deficit (FCover < 0.3 for average of the last two years).The three most populated countries-DR Congo (Central Africa), Ethiopia (East Africa), and Nigeria (West Africa)-are presented on a separate scale.

Figure 5 .
Figure 5. Multiple environmental burdens (MEB) and population exposure.(a) Spatial patterns of MEB (light to dark blue) and populated areas (pixels with more than 100 persons) in 2000 (black); (b) spatial patterns of MEB and populated areas (pixels with more than 100 persons) in 2019.

Figure 6 .
Figure 6.Contribution of environmental effect, population effect, and interaction effect in the change in population exposure from 2000 to 2019 to (a) hazardous PM2.5 levels, (b) extreme temperature increase, (c) prolonged drought severity, (d) green deficit, and (e) multiple environmental burdens (at least three risks-3EB).

Table 1 .
Details of the data used in the study.

Table 2 .
Country-level population exposure to environmental burdens in 2000 and 2019.