Main

Biodiversity is an integral part of sustainable development2,3; however, we are rapidly losing biodiversity1 and failing to embed biodiversity management in policy and planning6. One major limitation to achieving conservation goals is the lack of information on the impacts of diverse human activities on biodiversity and resulting ecosystem functions and services2,3. To be useful in national and international decision-making, such information needs to be comparable across spatial and temporal scales and capture changes in biodiversity relevant to sustaining societies and economies2,3. Ecosystem condition or integrity represents the degree to which the composition, structure and function of an ecosystem resembles that of its reference state7. This biodiversity metric is increasingly being adopted in multilateral environmental agreements (for example, the 2030 Global Biodiversity Framework (GBF)) to address these needs. The biodiversity intactness index4 (BII) is an indicator of ecosystem condition that holds promise for mainstreaming biodiversity into policy and planning7. The BII assesses human impacts on the abundance of a wide range of species that contribute diverse functions and capture the multidimensional nature of biodiversity in a way that can be compared across multiple scales and time periods4,8.

However, the limited availability of appropriate data to quantify indicators such as the BII is a major constraint to decision-making, especially in the Global South1,9. Available assessments of ecosystem condition are criticized for being top-down; that is, based on global, decontextualized pressure–impact relationships that extrapolate across data-poor regions and taxa10,11,12. These assessments can have lasting consequences for planning and prioritization10. For example, global assessments of ecosystem condition typically do not differentiate between planted pastureland and untransformed rangeland—a key distinction in the context of sub-Saharan Africa where rangelands predominate13,14—and the validity of such assessments has been questioned15. Large tracts of the supposedly degraded rangelands of the region are inappropriately identified for ‘restoration’ through tree planting, which can undermine both biodiversity and livelihoods16,17. At the same time, the fastest-growing human populations on Earth are in sub-Saharan Africa18. Moreover, the ecosystems of the region are undergoing rapid transformation that could compromise sustainable development into the future in the absence of more context-appropriate biodiversity information to support policy and planning19,20.

The Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES) calls for regional biodiversity information to close these knowledge gaps1,19. Place-based knowledge, which encompasses diverse forms of knowledge—including scientific, experiential and local—is rooted in specific landscapes and contexts and an important potential source of regional biodiversity information21. Mobilizing such knowledge to inform sustainability policy and planning requires approaches that retain social and ecological specificity while enabling comparisons across broader scales12,22. Here we demonstrate a robust approach to mobilizing place-based knowledge to assess the biodiversity intactness of sub-Saharan Africa, one of the most poorly represented regions in global biodiversity datasets and assessments9. Such regional assessments can serve as a bridge between place-based and global sustainability assessments to overcome cross-scale integration challenges through contextualized generalizations21,22.

Our bottom-up approach overcomes critical data gaps and limitations of top-down biodiversity models by quantifying biodiversity intactness using the Biodiversity Intactness Index for Africa (bii4africa), a dataset that we previously co-produced and published with 200 experts in African fauna and flora5. These experts embody place-based African biodiversity knowledge, which holds credibility, legitimacy and saliency for mainstreaming into national decision-making23 and contributes to inclusivity and decoloniality in science24. The bii4africa dataset5 contains standardized estimates by experts of the impact of the predominant land uses in sub-Saharan Africa on diverse functional groupings of species that represent around 50,000 terrestrial vertebrates and vascular plants. Here we integrate ten spatial datasets to map these land uses, which we combine with bioregional lists of indigenous taxa and the associated bii4africa data5 to map the BII across sub-Saharan Africa (Extended Data Fig. 1).

Assessing biodiversity intactness

The BII indicates the average remaining proportion of intact populations of indigenous species in a particular area given the dominant human land uses and activities4. Intactness is defined relative to a reference state: typically, before alteration by modern (industrialized, colonial and post-colonial) society, with large protected or wilderness areas serving as a contemporary reference. The index accommodates as wide a range of species as possible, with all species weighted equally. The BII is spatially explicit and standardized on a scale from 0 to 100%, which reflects completely transformed to intact areas and has the same meaning at all scales to facilitate comparative assessments.

In addition to being an indicator of ecosystem condition, the BII has been proposed as a measure of the planetary boundary ‘functional biosphere integrity’25. Our approach aligns with suggestions to address challenges with biosphere integrity in the planetary boundaries framework26,27, including assessments of biome integrity and BII at regional (as opposed to global) scales8,26,28. We significantly advance the original BII approach4 by using data produced through a structured expert-led process to improve rigour29, including data from many more experts, considering the impact of land-use intensity30 and disaggregating the BII into more nuanced functional groups of species5. Our comprehensive assessment of the biodiversity intactness of sub-Saharan Africa provides insights at policy-relevant scales into the human activities that are contributing to the retention and loss of biodiversity across countries, ecoregions and biomes.

Biodiversity intactness of the region

Sub-Saharan Africa has a current estimated biodiversity intactness of 76% (±14%; Fig. 1). This means that indigenous vertebrate and plant populations across the region have on average declined to 76% of their intact reference abundances. The BII of vertebrates is 71% (±9%), which is lower than that of terrestrial vascular plants (79 ± 17%). Mammals have experienced the greatest losses, whereas graminoids (grasses, sedges and rushes) and forbs (non-graminoid flowering plants with no or limited aboveground lignification) have experienced on average the lowest losses (Fig. 1). All reported uncertainties around BII values are based on 95% confidence intervals around average expert estimates of intactness in the bii4africa dataset5.

Fig. 1: The BII across sub-Saharan Africa.

BII scores for terrestrial vertebrates and vascular plants collectively and disaggregated into the constituent species groups. The overall BII score of 76% for the region shows that on average across all indigenous species, 76% of individuals remain compared with intact (pre-modern industrial society) reference populations. Maps were created using ArcGIS Pro v.2.7.0.

Source Data

There is high variability in intactness among functional groups of mammals, which range from 20 to 82% (Extended Data Fig. 2). Large herbivore and carnivore species (>20 kg) have experienced the greatest declines in abundance (BII = 20–52% and 25–51%, respectively), followed by primates (46–65%). These groups are relatively low in species richness and therefore contribute less towards total intactness compared with the more species-rich orders of bats (BII = 64–80%), insectivores (64–74%) and rodents (61–82%), which have retained on average almost double the intactness of larger mammals. There is less marked variability in intactness in the other vertebrate taxa, with birds ranging from 47 to 85%, reptiles from 56 to 77% and amphibians from 55 to 74%. Forest interior and cavity-breeding large savanna birds have been the most affected (BII = 47% and 58%, respectively), whereas grassland birds (except for ground nesters) and aerial feeders have been the least affected (82% and 85%, respectively). Among reptiles, chelonians and large specialist snakes and lizards have experienced the largest declines (BII = 56% and 57%, respectively), whereas small generalist snakes and lizards and rupicolous reptiles have experienced the smallest declines (75% and 77%, respectively). Amphibians that breed in plant or tree hollows or in seep or spray zones have been worst affected (BII = 55%), whereas those that breed in ephemeral streams have been least affected (74%). Plant functional groups have large variability in intactness, ranging from 55 to 91%. Shade-tolerant (forest) and swamp trees and shrubs, together with epiphytes, have suffered the greatest losses (BII = 55–56%), whereas forbs and graminoids that resist disturbance31 have been the most resilient to land-use changes (≥90%).

Variation across nations and ecosystems

Twelve out of the 42 countries in sub-Saharan Africa are estimated to have retained >80% of their biodiversity intactness, with Namibia and Botswana having the highest BII (87%; Fig. 2a). Fifteen countries have retained <70% of their BII, with Rwanda (48%) and Nigeria (53%) having the lowest BII. The remaining 15 countries have retained intermediate levels of BII (70–80%). Sierra Leone and Ethiopia are middle of the range (72–73%).

Fig. 2: The BII across countries and ecoregions of sub-Saharan Africa.

a,b, Average BII scores are depicted in ascending order for countries per African Union region (a) and for ecoregions per biome or biome mosaic (b). Uncertainty around average BII values is based on 95% confidence intervals around average expert estimates of intactness in the bii4africa dataset5. Asterisks indicate countries that are only partially in sub-Saharan Africa. A. savanna, Acacia savanna; DR, Democratic Republic; Fy, fynbos; Th, thicket.

Source Data

Biodiversity intactness varies considerably across ecoregions, from an average of 37% in the Lowland Fynbos and Renosterveld to 92% in the Etosha Pan halophytics (Fig. 2b). With each species considered equally in the BII, plants contribute more towards BII than vertebrates in most ecoregions given their higher species richness (Extended Data Fig. 3). The exceptions are most desert ecoregions, where vertebrates are more species-rich than plants, and grasslands and Acacia savannas, where plants and vertebrates have similar species richness.

Comparisons of the major biomes of sub-Saharan Africa show that BII is highest in the more arid biomes (86% in desert and 83% in shrubland), and lowest in the fynbos (a Mediterranean-type ecosystem and biodiversity hotspot; 56%) and grassland (68%) biomes (Fig. 2b and Extended Data Fig. 4). On average, BII is lower for vertebrates than plants across most biomes (grassland, thicket, humid savanna, Acacia savanna and shrubland; Extended Data Fig. 4). The exceptions are fynbos and desert, where vertebrates fare better than plants, and forest, where BII is similar for both species groups.

The impact of land-use intensity

The average BII is 95% (±8%) across strictly protected lands, 79% (±14%) across unprotected untransformed lands, 48% (±16%) across croplands, 43% (±18%) across tree croplands, 34% (±15%) across settlements and 29% (±17%) across timber plantations (Fig. 3b). The variation in BII scores in these land uses is caused by spatial variation in both land-use intensity (Fig. 3a) and species composition (Extended Data Fig. 3), as BII varies among species groups (Fig. 3c). All species groups have the lowest levels of intactness in settlements and timber plantations (Fig. 3c). Plants tend to have higher intactness in croplands (for example, maize and wheat) compared with tree croplands (for example, coffee and fruit), whereas reptiles tend to have higher intactness in tree croplands than croplands. There are greater differences in intactness between protected and unprotected untransformed lands for vertebrates compared with plants.

Fig. 3: Land use across sub-Saharan Africa and its influence on the BII.

a, Six distinct land uses are predominant across the region, with notably variable intensity in four of these land uses: settlements, tree crops, crops and untransformed (unprotected) lands. b,c, Absolute BII scores in each land use for all plants and vertebrates collectively (b) and the major species groups (c). Boxplots show median BII scores across pixels, interquartile ranges and maximums and minimums within 1.5× the interquartile range. Variability in the BII in a land use arises from differences in species composition and land-use intensity. The map in a was created using ArcGIS Pro v.2.7.0.

Source Data

Land-use intensity has a notable impact on biodiversity intactness in the two most extensive land uses. In unprotected untransformed lands, the highest intensity rangelands have an average BII of 51% compared with 85% in the lowest intensity ‘near-natural’ lands (Extended Data Fig. 5a). In croplands, the average BII is 26% in the highest intensity croplands, which is notably less than in the lowest intensity, smallholder croplands (54%; Extended Data Fig. 5b). The distribution of land-use intensity is right-skewed across sub-Saharan Africa (Extended Data Fig. 5c,d). In other words, non-intensive activities are more common than intensive activities in each of these two land uses, which has a substantial impact on the BII of the region given the extent of these land uses.

Directing conservation efforts

Our results highlight which land uses make the largest relative contributions to lost and remaining biodiversity intactness, and those that contribute disproportionately given their extent (Fig. 4a). Notably, the majority (84%) of remaining BII across sub-Saharan Africa occurs in unprotected, largely untransformed lands, which cover 80% of the region. Given their vast extent, these areas also contribute the most (68%) to the total BII that has been lost across the region. These findings highlight the critical importance of sustainably managing these areas. Strictly protected lands contribute disproportionately to remaining biodiversity intactness, comprising only 6% of the area of the region but contributing 7% of the remaining BII and just 1% of the lost BII. Croplands contribute a larger amount (9%) to the remaining BII than protected lands, but cover over double the area of protected lands (14%) and are responsible for 29% of the lost BII across the region. Settlements, tree croplands and timber plantations each cover <1% of the region, support <1% of remaining BII and are responsible for ≤1% of lost BII, respectively.

Fig. 4: Contributions to remaining and lost BII in sub-Saharan Africa.

a, The relative contributions of six distinct land uses to the total lost (y axis) and remaining (circle size) BII scores across the region compared with their contributions to the total extent in sub-Saharan Africa (x axis). Land uses above the diagonal line contribute disproportionately to losses relative to their extent. b, The contributions of land uses in each biome to remaining (top) and lost (middle) BII relative to their extent (bottom), differentiating between low and medium to high land-use intensities in settlements, tree crops, crops and untransformed lands. The ‘All’ bars on the left depict all biomes collectively (that is, the full extent of sub-Saharan Africa). Individual biomes are otherwise ranked on the basis of decreasing remaining BII. c, The average BII of each country relative to the proportion of its land extent that is transformed (that is, covered by settlements, timber, tree crops or crops). Country names corresponding to each two-letter code referenced here are depicted in Fig. 2a.

Source Data

At the biome scale, the highest relative contributions to remaining BII are similarly made by unprotected untransformed lands: predominantly near-natural lands in forests, savannas and arid biomes, and rangelands in thickets, grasslands and fynbos (Fig. 4b). In desert and fynbos, strictly protected lands also make major contributions to remaining BII (41% and 23%, respectively), whereas their lower contributions (5–10%) in other biomes largely reflect their more limited extent in those biomes. Croplands are responsible for notable losses in BII across the grassy biomes (grassland, Acacia savanna and humid savanna) and fynbos, with less-intensive croplands being more common in the savannas compared with more-intensive croplands (largely in South Africa) in the grassland and fynbos (Fig. 4b). Rangelands are the major driver of lost BII in the thicket biome (Fig. 4b). Degradation of near-natural lands contributes more to lost BII in forests than in the other biomes. Deforestation to make way for rangelands and croplands also contributes to losses in the forest biome.

As with biomes, countries with a higher proportion of their land transformed (mostly to cultivated lands) tend to have notably lower remaining biodiversity intactness (Fig. 4c). However, there is variability in this relationship. For example, Burundi has the third lowest country-level BII but is notably less transformed than the two countries with the lowest BII (Nigeria and Rwanda; Fig. 4c). The degree of land transformation in Burundi is comparable with Tanzania and Zimbabwe, which have considerably higher BII scores. This variability reflects national differences in the intensity of both transformed and untransformed land and in how the species in a country respond to those pressures.

Validation and uncertainties

A challenge with broad-scale biodiversity assessments is the feasibility of performing independent validations to document the degree of error11, particularly in data-poor regions. Errors can arise from biases of the experts, which we mitigated by adopting evidence-based guidelines to improve elicitation rigour29. However, it was not possible to eliminate potential error arising, for example, from knowledge gaps and potential systematic biases in the expert group or from data limitations in our land-use mapping (see the section ‘Caveats’ in the Methods). Such biases and errors may have had a directional impact on our assessment (that is, leading to consistent overestimation or underestimation). To assess these potential errors, we critically evaluated our assessment in multiple ways, including the degree of consensus between experts and corroboration between our results and other assessments of human pressure and threat.

The structured expert elicitation process that forms the basis of our results included a critical review of the results by participating experts as a validity check embedded in the process5,29 (Extended Data Fig. 1). The uncertainty we report around BII scores reflects the degree of consensus among experts, which highlight taxa and land uses for which knowledge is currently more uncertain or disputed; therefore the risk of error may be greater. There was higher uncertainty for plants than vertebrates and for cultivated lands (timber, tree croplands and croplands) than for other land uses (Extended Data Table 1). These findings can guide future research to close knowledge gaps4.

Correlations between our BII map and three human-pressure maps followed the expected directions, with lower BII in biodiversity hotspots (areas of exceptional endemism that have lost ≥70% of their primary vegetation32; Extended Data Fig. 6a–d). Although these datasets cannot be considered entirely independent of our own, given that they all rely on (imperfect) land-cover data, the existing global BII map13 showed unexpected relationships with these datasets, which raises concerns about its validity15. Although expected correlations corroborated our BII assessment to some extent, the variability is also important. That is, the BII provides insights into how diverse species groups respond to different human pressures and is therefore not synonymous with aggregated human-pressure indices.

When considering the International Union for Conservation of Nature (IUCN) Red List, we found that the threat status of a vertebrate species is a significant predictor of its BII across its sub-Saharan African range (Extended Data Fig. 6e). Critically endangered species have significantly lower BII than endangered, vulnerable and near-threatened species, which in turn have significantly lower BII than least-concern species. These broad trends demonstrate the robustness of our approach, although the Red List may share some of the unknown biases inherent in the BII given the central role of expert knowledge in both assessment processes. Moreover, large within-category variation results in relatively small absolute differences in mean BII between threat categories. A review of Red List assessments for a random sample of outlier species indicated that this variation arises from differences in the purpose of assessments of intactness versus threat of extinction, as well as knowledge gaps (Supplementary Table 1). Such gaps include, for example, assessment inaccuracies for poorly known species, potential BII overestimates for localized species in ineffective de jure protected lands and potential BII underestimates for species prevalent in de facto protected lands. Of note, the BII is not intended as a species-level index, and caution should be exercised when considering species-level results beyond general trends.

Compared with previous assessments of BII (Extended Data Fig. 7), our BII estimate of 80% for the southern African sub-region is a plausible decrease from the 84% estimated for the sub-region in 2005 through a simplified expert elicitation and mapping process4. However, both our and the 2005 BII estimates for southern African are higher than the 74% predicted for the sub-region by the global BII model in 2016 (ref. 13). Considering the full region, the global model estimated a higher BII for sub-Saharan Africa (84%) than our approach (76%). These differences are due to the global model estimating lower BII in lower-rainfall biomes (desert, shrubland and Acacia savanna) and higher BII in higher-rainfall biomes (forest, humid savanna and grassland) than our assessment (Extended Data Fig. 7c).

Taken together, these diverse comparisons corroborate our BII assessment, which is in contrast to the existing global BII model, which lacks such corroboration15. The reported uncertainty around our BII scores gives an indication of uncertainty in the underlying expert scores. We also note that this does not fully account for potential systematic biases and other unknown potential sources of error in our assessment, some of which may be shared with the corroborating datasets.

Discussion

Our assessment of biodiversity intactness across sub-Saharan Africa integrates the place-based knowledge of 200 experts5 into a regional measure that can be consistently applied at multiple policy-relevant scales to address longstanding cross-scale integration challenges in sustainability science and practice3,12,21. This bottom-up approach accounts for context-specific complexities to help overcome critical data gaps that limit the availability of credible biodiversity information for national policy and planning20,23,33.

We estimate that sub-Saharan Africa has lost just under a quarter of its pre-industrial biodiversity intactness. A notable finding is that >80% of the remaining wild organisms in the region persist in unprotected and largely untransformed natural forests and rangelands where people coexist with and depend on biodiversity. Conserving and restoring biodiversity, while working towards just and sustainable development, requires a focus on these working lands that sustain more than 500 million people17,34,35,36. Our results indicate nuanced differences in both the threats to biodiversity and the resilience of different species groups to human activity across the region and point to land-use approaches and policies that can support more sustainable coexistence between nature and people.

Large tracts of high-integrity humid forest remain in Central Africa37,38, which contribute to relatively high BII for Central African countries and ecoregions. By contrast, much of the West African forest is highly degraded37,38, thereby contributing to very low BII. Degradation of near-natural lands (for example, through faunal and floral overharvesting37,39) is a major cause of diminished intactness in these forested ecosystems, with forest-dependent species incurring some of the greatest regional BII losses. West African humid savannas have also been extensively degraded19,38, with smallholder croplands and rangelands contributing to low BII. Policies that promote sustainable harvesting or alternative livelihood opportunities are key to retaining biodiversity across these systems19,37,40. Our findings are congruent with other assessments of degradation across the African tropics19,37,38 and contrast with the high estimates across West Africa in the existing global BII model13.

Croplands are regarded as the greatest threat to biodiversity across sub-Saharan Africa41,42. The two countries with the greatest crop cover (Nigeria and Rwanda) have the lowest BII scores in the region. In contrast to this finding, the existing global BII model[13](https://www.nature.com/articles/s41586-025-09781-7#ref-CR13 “Newbold, T. et al. Has land use pushed terrestrial biodiversity beyond the planetary boundary? A global assessment. Science 351, 600–60