Introduction

The past few decades have seen notable progress acknowledging the role of Indigenous Peoples and Local Communities in combating the biodiversity and climate crises1,2,3,4,5,6. The stewardship of lands by Afro-descendant Peoples (ADP) may also be broadly effective for conserving biodiversity and securing carbon. To date, however, ADP voices and experiences are not widely considered in global and national forums where biodiversity and climate actions are discussed and decided7,8. This is concerning because ADP manage ecosystems crucial for global biodiversity and climate goals. For instance, in Latin America, 133 million people (one in four) self-identify as ADP9, and their presence has been documented across 205 million hectares (ha) of land, much of which is in areas of high conservation importance such as rainforests, mangroves, and savannas10.

Since ADP voices and experiences have been neglected, contributions of ADP resource management practices to biodiversity and carbon storage in biomass have gone unrecognized. Studies highlighting the impacts of ADP land stewardship on climate mitigation remain limited11,12,13. The United Nations’ International Decade for the People of African Descent (2015–2024) played an important role in promoting “recognition, justice and development”14 but lack of documentation of ADP contributions to conservation impedes inclusion of ADP in environmental decision-making8. Documentation of positive environmental outcomes on ADP lands, along with explanations for those benefits in terms of ADP management practices, would therefore be foundational for recognizing ADP as leaders in global conservation initiatives.

Better understanding of practices and outcomes on ADP lands can strengthen the evidence base for recognizing the “efforts and actions” by ADP in nature conservation and their role in implementing biodiversity commitments, emphasized at the Convention on Biological Diversity (CBD) COP1615. Specifically, evidence of environmental outcomes on ADP lands is relevant to Target 3 of the Global Biodiversity Framework (GBF), which seeks to conserve 30% of the world’s lands and waters by 203016, and the related Sustainable Development Goal (SDG) 13 to protect and sustainably use and manage Earth’s ecosystems17. Such evidence can also advance climate commitments under the United Nations Framework Convention on Climate Change (UNFCCC), which seeks to achieve net zero CO2 and greenhouse gas (GHG) emissions by 205018, along with the partnerships required to achieve goals for nature and people as outlined in SDG 17 by strengthening collaboration with crucial allies such as ADP.

To address this evidence need, we posed three research questions: (1) what is the quantity of biodiversity and carbon stocks on ADP lands, (2) do ADP lands exhibit greater levels of avoided deforestation and associated carbon emissions compared to similar control sites nearby, and (3) what ADP management practices with African origins help account for environmental conservation on ADP lands? We focused on legally recognized ADP lands in four tropical countries in South America for which spatial and legal data were accessible: Brazil, Colombia, Ecuador, and Suriname (Supplementary Fig. 1; Supplementary Table 1). To address (1), we used indicators related to biodiversity and carbon storage in biomass, specifically IUCN Red List species status and rarity-weighted species richness (i.e., relative concentration of species accounting for habitat range size) (RWR)19 for terrestrial vertebrates (amphibians, birds, mammals, and reptiles), and irrecoverable carbon (i.e., carbon that, if lost from ecosystem conversion, could not be re-sequestered for at least 30 years20). To address (2), we used a quasi-experimental approach relying on statistical matching to estimate the effect of ADP lands on deforestation. We compared recognized ADP lands to protected areas (PAs) and control sites, while accounting for spatially confounding variables (see Methods). To address (3), we reviewed accounts documenting African settlement histories in the Americas, in particular ADP resource management practices shaped by cultural knowledge, spiritual beliefs, and colonial influence21,22. Our inquiry builds on elements of postcolonial studies pertaining to territorial autonomy and resource use23,24,25, and foundational principles of Black Geographies and Black Ecologies, especially their emphasis on ADP adaptive capacities, place-making, and sustainable practices26,27.

Our findings show that recognized ADP lands have high biodiversity and irrecoverable carbon stocks, and low deforestation compared to control sites. The ADP historical land management systems and practices help explain these positive environmental outcomes indicating ADP’s crucial role in achieving global biodiversity and climate goals.

Results

The extent of recognized ADP lands

Recognized ADP lands encompass different land tenure arrangements per country, from collectively titled lands in Brazil, Colombia, and Ecuador to community concessions and customary lands in Suriname (Supplementary Table 1). ADP in these countries hold management rights to make decisions about their resources on 9.9 million ha of land (0.98% of the total 1 billion ha land area of the four countries) (Supplementary Table 2). Recognized ADP lands represent a small proportion of the total area in Brazil (0.45%), Colombia (5.01%), Ecuador (0.53%), and Suriname (1.05%) (Supplementary Table 2). Recognized ADP lands in Brazil and Colombia represent about 97.1% of total delineated ADP areas in this study (39.1% and 58%, respectively).

The presence of ADP in the study countries

While we focus on legally recognized ADP lands, ADP inhabit many administrative units in Brazil, and most in Colombia, Ecuador, and Suriname (Supplementary Fig. 1; Supplementary Table 3). Based on demographic data, ADP are present in approximately 30.5% of administrative units in Brazil, 97.6% in Colombia, and 100% in Ecuador and Suriname (Supplementary Fig. 1). Colombia has the largest number of administrative units where ≥50% of the population identifies as ADP, followed by Suriname (Supplementary Table 3). Ecuador contains the fewest administrative units with ≥25% ADP populations overall (Supplementary Table 3).

ADP presence within administrative units overlapping recognized ADP lands varies among countries (Supplementary Table 4). In Brazil, 45.5% of the quilombola population lives in administrative units overlapping ADP lands. Colombia has the largest ADP presence, with > 1.1 million ADP living in administrative units overlapping ADP lands. In Ecuador, while only 9% of the 814,495 self-identified national ADP population live in administrative units overlapping ADP lands included in the study, ADP are 39.2% of the total population in these administrative units. Self-identified ADP in Suriname are 9% of the national ADP population and 93.7% of the total population in administrative units overlapping ADP lands.

Recognized ADP lands, ecosystems, and protected areas

Recognized ADP lands overlap critical global ecosystems and PAs (Supplementary Tables 5–6). In Brazil, ADP lands are located within 30 different ecosystems, of which 68% of ADP lands are within tropical and subtropical moist broadleaf forest ecosystems, and overlap 87 PAs, 79 of which are state-designated, including globally important PAs within Amazonia like Parque Nacional do Jaú. In Colombia, about 98% of ADP lands are within tropical and subtropical moist broadleaf forest ecosystems located between the Pacific coast and Cordillera Occidental mountains, including the Chocó biogeographic region, the ninth most biodiverse hotspot globally28 known for driving the regional hydroclimate29. ADP lands in Colombia overlap 36 PAs, 33 of which are state-designated, including Reservas Forestales Protectoras Nacionales del Río Anchicayá and Darién. In Ecuador, ADP lands are primarily within Esmeraldas Province, mainly within tropical and subtropical moist broadleaf forests covering areas between Reserva Ecológica Manglares Cayapas-Mataje along the Pacific coast to the north and Parque Nacional Cotacachi-Cayapas inland to the south, with a small portion of territories, just over 3000 ha, partially overlapping three PAs. In Suriname, ADP lands are entirely within Guianan lowland and highland moist forests in the country’s center and overlap or border Centraal Suriname Natuurreservaat.

The extent and quantity of biodiversity and carbon in ADP lands

Although recognized ADP lands cover less than 1% of the total land area of the study countries, these territories contain disproportionately biodiverse and carbon-rich areas (Fig. 1; Fig. 2). ADP lands overlap the geographic ranges of 4004 terrestrial vertebrate species, 370 of which (9.2%) are listed under IUCN Red List Threatened categories (Supplementary Table 7). This represents about 46% of threatened terrestrial vertebrate species in these countries. Considerable portions of ADP lands harbor high levels of biodiversity as measured by terrestrial vertebrate RWR. For instance, >58% of recognized ADP lands are located within the top 10% and about 72% within the top 20% of biodiverse areas globally (Fig. 1; Supplementary Table 8).

Fig. 1: Terrestrial vertebrate species biodiversity within recognized Afro-descendant Peoples (ADP) lands and per country.
figure 1

This figure shows concentrations of biodiversity (based on rarity-weighted richness (RWR) considering terrestrial amphibians,  birds, mammals, and reptiles) within recognized ADP lands (shown with orange borders) and per country throughout the study area, with the highest 5% of biodiverse areas globally in dark blue. Panels show terrestrial vertebrate biodiversity concentrations within recognized ADP lands in Brazil (a), Colombia (b), Ecuador (c), and Suriname (d). Nearly 92% of recognized ADP lands in Colombia and 99% in Ecuador coincide with areas that are among the highest 5% of biodiverse areas globally.

Full size image
Fig. 2: Irrecoverable carbon areas within recognized Afro-descendant Peoples (ADP) lands and per country.
figure 2

This figure shows the distribution of irrecoverable carbon density (tonnes per hectare ) within recognized ADP lands (shown with blue borders) and per country throughout the study area. Panels show irrecoverable carbon density within recognized ADP lands in Brazil (a), Colombia (b), Ecuador (c), and Suriname (d). Parts of the study area with minimal or no high irrecoverable carbon density (less than 25 t/ha) are displayed in white, and the areas without irrecoverable carbon are shown in gray.

Full size image

Across the study countries, ADP lands average 6.8 tonnes of irrecoverable carbon per hectare (t/ha) compared to 5.2 t/ha of combined national totals (Supplementary Table 9). Recognized ADP lands store nearly 486.2 Mt out of 25,852.9 Mt of irrecoverable carbon total in the study countries, including 51.1 Mt of high ( >25 t/ha) irrecoverable carbon (10.5% of the total irrecoverable carbon within ADP lands) (Fig. 2; Supplementary Table 9). Therefore, ADP lands contain 1.88% of total irrecoverable carbon in the study countries, equivalent to 1784.3 Mt carbon dioxide equivalent (CO2e), or 38.79% of the annual 4600 Mt CO2e potential and almost 1.3% of total natural climate solution protection opportunities in the tropics from 2020 to 205030.

Within recognized ADP lands, irrecoverable carbon is mostly concentrated in tropical and subtropical forest, wetland, and tropical peat ecosystems (Supplementary Table 9). In Brazil, ADP lands contain 172.9 Mt of irrecoverable carbon, mostly in tropical forest, wetland, and peat ecosystems (Supplementary Table 9). In Colombia, ADP lands contain 299.8 Mt of irrecoverable carbon in tropical forest, peat, and mangrove ecosystems concentrated near the Pacific coast. In Ecuador and Suriname, where ADP lands are located away from coastal mangroves, these areas contain smaller amounts of irrecoverable carbon (about 2.4 Mt and 11 Mt, respectively) (Supplementary Table 9).

Overall impact of recognized ADP lands on deforestation

The ecological value of ADP stewardship is evident by avoided deforestation on ADP lands. Overall, rates of forest loss within ADP lands were consistently and significantly lower than rates of forest loss within control cells (Fig. 3; see methods for analytical details). Lower rates of forest loss represent the impact (or additionality) of recognized ADP lands on avoided deforestation, while controlling for potential confounding variables through statistical matching and as covariates in the models (Supplementary Table 10). Both rates of forest loss and impact of ADP land tenure – measured as the difference in forest loss between ADP and controls – varied spatially, depending on whether focal cells were located fully inside, on the edge of, or fully outside of PAs. ADP lands on the edge of or inside PAs had the lowest rates of deforestation, with 46% (Bayesian credible intervals = 39–53%) and 51% (BCI = 49–53%) less deforestation than ADP lands outside PAs, respectively. However, avoided deforestation (ha/yr) was greatest for ADP lands outside PAs, where forest loss was reduced by 36% (BCI = 34–38%) compared to controls, and ADP lands at the edge of PAs, where forest loss was reduced by 55% (BCI = 45–63%) (Fig. 3; Supplementary Table 11). ADP lands fully within PAs were associated with a 29% (BCI = 27–32%) reduction in forest loss compared to controls inside PAs. Patterns of avoided carbon emissions from deforestation mirrored results of deforestation rates, with ADP lands associated with significantly lower carbon emissions regardless of location inside, on the edge of, or outside of PAs (Supplementary Fig. 2; Supplementary Table 12).

Fig. 3: Rates of forest loss (ha/yr) within sampled Afro-descendant Peoples (ADP) lands inside, on the edge of, and outside of protected areas (PAs) compared to control cells for all study countries.
figure 3

Rates of forest loss were consistently lower within ADP lands (purple) compared to controls (green). Points represent mean estimates of rates of forest loss from a Bayesian generalized linear mixed model. Error bars show 95% Credible Intervals. Values of model estimates for all covariates are shown in Supplementary Table 11. The number of sample cells for each category was: ADP outside PAs = 38,166; ADP on PA edge = 435; ADP inside PAs = 23,766; Control outside PAs = 37,748; Controls at PA edge = 431; Controls inside PA = 24,188.

Full size image

Across the study countries, the direction and magnitude of impacts from ADP lands in reducing forest loss were largely consistent among Brazil, Colombia, and Ecuador (Fig. 4; Supplementary Table 13). By contrast, ADP lands in Suriname were not associated with a consistent reduction in forest loss and were even associated with higher rates of forest loss outside of PAs. Analyses of forest loss at different distance classes – 0–1 km, 1–10 km, and >10 km – from the nearest edge of polygons representing PAs and ADP lands suggest potential displacement and spillover of deforestation under some circumstances (Fig. 5; Supplementary Table 14). Particularly in control cells inside and on the edges of PAs, and in ADP lands inside PAs, deforestation rates were greater within 1 km of the edges of ADP lands.

Fig. 4: Rates of forest loss (ha/yr) within sampled Afro-descendant Peoples (ADP) lands inside on the edge of, and outside of protected areas (PAs) compared to control cells for each study country.
figure 4

Rates of forest loss were typically lower within ADP lands (purple) compared to controls (green) across the study countries, except in Suriname. Points represent mean estimates of rates of forest loss from a Bayesian generalized linear mixed model. Error bars show 95% Credible Intervals. Values of model estimates for all covariates are shown in Supplementary Table 13.

Full size image
Fig. 5: Rates of forest loss (ha/yr) by distance class within sampled Afro-descendant Peoples (ADP) lands inside, on the edge of, and outside of protected areas (PAs) compared to control cells for all study countries.
figure 5

Rates of forest loss change with distance to the edge of ADP lands, suggesting potential displacement of deforestation to areas near edges of PAs and ADP lands inside PAs. Points represent mean estimates of rates of forest loss from a Bayesian generalized linear mixed model. Error bars show 95% Credible Intervals. Values of model estimates for all covariates are shown in Supplementary Table 14.

Full size image

Social-historical assessment of Afro-descendant practices contributing to environmental conservation

Mapping biodiversity and carbon stocks and documenting ADP stewardship impacts through quasi-experimental analysis provide evidence of ADP contributions to biodiversity conservation and climate change mitigation. However, it is also critical to understand ADP management practices that help explain these positive conservation outcomes. We therefore conducted a social-historical assessment of evidence of ADP management practices in the Americas since their arrival in the early 1500s .

Conventional economic and environmental histories of Africans in the Americas focus on plantations as the central representation of production systems throughout the region20. Such work emphasizes contributions of European practices to economic development, without considering African contributions beyond their role as enslaved plantation labor31,32. Yet, there is extensive historical literature on specific resource management practices of enslaved Africans and later ADP in many countries in the Americas exists31,32,33,34,35,36. ADP are referred to by different names throughout the region, including mocambos, quilombos, and cumbes in Brazil, Palenques in Colombia, and Maroons in Suriname37,38,39. In Suriname, descendants of runaway slaves also called themselves bush negroes or loweman pransun40.

Many scholars have called attention to the contributions of Africans in adapting their knowledge of tropical forests and other habitats to those they encountered in the Americas19,21,24. European colonizers and Africans encountered ecosystems that had been cultivated for centuries by Indigenous Peoples of the Americas. However, enslaved Africans emerged as key innovators in landscape management31. Due to domestication efforts in Africa spanning between 3000 and 8000 years, coupled with species exchanges with Asia, African tropical forests underwent a remarkable transformation into “food forests”20,21,22,23,24. These time-tested African practices subsequently crossed the Atlantic Ocean with enslaved individuals, ultimately being applied to the plants and animals of the Americas (Supplementary Table 15). Upon their arrival, both enslaved individuals and Maroon societies implemented management practices that replicated food forests, creating forest canopy structures that constituted functional reservoirs for dietary, medicinal, ritual, and festive purposes32,35. Spaces cultivated by enslaved Africans and ADP became integral components of plantation landscapes and Maroon livelihoods, including those in Colombia’s inter-Andean valleys41,42 and Ecuador’s Valle del Chota43. Similarly, botanical relics near former plantations in Suriname indicate practices employed by enslaved individuals for escape, which encompassed hiding crops, minimizing land clearing, avoiding fire, and cultivating diverse crop varieties. These escape agriculture practices, along with Maroon landraces, persist in the cultivation systems of their descendants44.

These studies offer explanations for the high biodiversity and irrecoverable carbon present within ADP lands. Accordingly, we suggest four key factors that link ADP management practices to sustainability and conservation. The first factor concerns adaptive strategies employed by ADP as they encountered a wide range of ecosystems. ADP often adapted their management practices developed for species brought from Europe and Africa to those they found in the Americas. This involved flexible use of native and introduced species in the constitution of management systems. Similarly, ADP adapted to distinct contexts in implementing their production systems, whether plantations, extractive systems, semi-urban conditions or deep forests.

A second factor concerns diversity among the systems that ADP implemented. Wherever ADP settled, they implemented a variety of sophisticated agricultural and livestock production techniques. Their agricultural production systems ranged from food plots to food gardens and agroforests. Their livestock production systems incorporated a diverse array of animals and foraging practices from tropical Africa. In both types of systems, ADP employed practices to imitate forest structures31,32,33,35,37,45.

The third factor is that ADP created management systems designed to support multiple management goals for survival. Over time, ADP women managed their lands not only by adapting culinary knowledge, but also by identifying species with medicinal properties32,45. Management for such multifaceted goals tended to result in highly agrobiodiverse landscapes. Because ADP management practices simulated forest structure, ADP landscapes constituted refuges for diverse plant and animal species with different uses, encouraging high biodiversity and carbon retention in biomass.

A fourth key factor is that ADP systems incorporated African ethnobotanical knowledge as well as spiritual beliefs about people, land, and forests, which ADP understood as fundamentally interrelated32,37,45. Management practices understood in terms of spiritual relationships served to support species diversity and landscape sustainability in ADP agroecosystems.

The process by which ADP adapted themselves to the opportunities and challenges presented by tropical ecosystems in the Americas played a crucial role in their transition from enslavement to freedom. This process, referred to as “environmental creolization”36, facilitated a degree of economic autonomy for ADP in Brazil and Colombia, leading to their active engagement in transforming tropical landscapes36,46. In their recent history, ADP have continued to adapt and transform landscape management practices that promote forest and biodiversity conservation47.

However, these management practices and production systems are threatened by biologically impoverished production systems, such as monocultures, which echo the colonial-era plantation model29. An example of this transformation from ADP-managed tropical landscapes to monoculture-oriented commodity markets occurred in Monte Oscuro in the northern region of Colombia’s Cauca department. In this region, an agroecological system based on ADP management of cocoa that kept tropical forests intact was replaced by sugarcane monoculture in the mid-20th century. This resulted in considerable deforestation and social deterioration, posing challenges for future development and conservation of ADP lands and traditional “food forests”41,42,48,49.

Discussion

Our analysis showed that there are high levels of biodiversity and carbon stocks within recognized ADP lands in Brazil, Colombia, Ecuador, and Suriname, and linked those findings to several historical processes by which ADP adapted and developed sustainable management practices. These findings highlight the crucial role of ADP contributions to global biodiversity conservation and climate commitments. At the global level, countries are seeking strategies to expand PAs to achieve the GBF target of conserving 30% of the world’s lands by 203016. In that context, given our findings, ADP lands should be recognized as important contributors, whether as other effective area-based conservation measures (OECMs) or other forms of protection promoted by ADP. The recognition of Afro-descendant contributions and ancestral knowledge during COP16 in Cali, Colombia15 marked a major step towards a more inclusive conservation model. This acknowledgment within the Convention on Biological Diversity (CBD) paves the way for ADP involvement in global biodiversity governance.

Our findings add to emerging evidence that ADP lands coincide with areas of high vertebrate species richness, similar to many Indigenous lands50,51,52. Although our study was limited to examining biodiversity within ADP lands in four countries, our research aligns with a previous study of 16 countries that found that ADP lands are often located in biodiversity hotspots and ecosystems of critical importance to conservation10. Our study builds on evidence highlighting the biodiversity contributions of ADP lands53,54,55,56 by examining terrestrial vertebrate species richness across multiple taxa and IUCN threat classifications. Nearly half of the threatened vertebrate species in the four countries examined have potential habitat within ADP lands.

Additionally, we provide key evidence to support further actions to advance the recognition of ADP roles within global climate policies. Countries worldwide are actively formulating strategies to attain carbon neutrality by 2050. For many countries in Latin America with extensive forest cover, deforestation remains a primary contributor to GHG emissions. This is particularly pronounced in countries such as Colombia, Ecuador, and Suriname, where land-use changes and forestry activities constitute the largest source of GHG emissions by sector57. In Brazil, the land-use and forestry sector ranks second in emissions, following agriculture57. Protecting forests is paramount for countries with extensive forest cover to achieve their Nationally Determined Contributions (NDCs) outlined in the Paris Agreement. By extending beyond national PAs, ADP lands are positioned as valuable areas that could contribute towards meeting national and international climate commitments. Since ADP lands in the study countries are covered by tropical and subtropical forests, have considerable irrecoverable carbon, and have experienced 29%–55% less deforestation compared to controls, supporting ADP resource management practices may represent an effective approach for reducing deforestation throughout the region. ADP practices also offer valuable insights for broader implementation of community-based sustainable forest management.

We quantified irrecoverable carbon in ecosystems that are important for climate change mitigation and within areas under communal tenure regimes. While a previous study determined that one third of global irrecoverable carbon is within Indigenous and Community lands20, it did not explicitly consider land tenure status nor quantify irrecoverable carbon within ADP lands. Our study addressed this gap by mapping and estimating irrecoverable carbon specifically within recognized ADP lands, highlighting their crucial role in protecting high carbon ecosystems like tropical forests58. Our findings complement previous research in Brazilian forests which shows that ADP lands are effective at sustaining stored carbon and reducing carbon emission rates from forest loss59.

The dual importance of ADP in biodiversity conservation and climate change mitigation also aligns ADP with global developmental goals, particularly SDGs 13 (climate action) and 14 (life on land), making a case for strong partnerships (SDG 17) with ADP to achieve these and other SDGs. Moreover, our socio-historical analysis suggests that traditional ADP practices have sustained resilient ecosystems for centuries. By valuing these practices, our study reveals their potential to inform solutions promoted by SDGs, GBF, and the Paris Climate Agreements. It opens opportunities to integrate ADP practices that simultaneously enhance equity, biodiversity, and climate change mitigation into global policies.

Results from the quasi-experimental analysis suggest that strengthening the role of ADP in policy formulation and securing ADP tenure rights are important mechanisms for advancing global biodiversity conservation and climate change mitigation. We found that ADP lands with tenure recognition were, on average, associated with significant avoided deforestation and avoided carbon emissions compared to controls. Notably, rates of forest loss were lowest on ADP lands overlapping PAs. This outcome could either represent opportunities to recognize ADP stewardship under the GBF or pose risks of undermining land rights, depending on the type of PA governance. More than half of the PAs overlapping ADP lands were governed by federal or national agencies, while less than a quarter were governed by Local Communities, according to the World Database of Protected Areas (WDPA) (Supplementary Fig. 3). While on-the-ground research is needed to better understand governance practices where ADP lands and PAs intersect, there is growing evidence that community-managed PAs and OECMs can be effective in achieving multiple environmental and social outcomes60,61, provided community land rights and priorities are respected. Outside and on the edges of PAs, ADP lands were associated with the greatest reductions in forest loss (36%–55%) observed in this study. The impact of ADP lands on deforestation outside of PAs was consistent across space – we found no evidence of deforestation being displaced to areas adjacent to ADP lands or spillover into areas just within the borders of ADP lands outside of PAs. These results mirror recent pantropical and national-level studies demonstrating significant avoided deforestation on lands of Indigenous Peoples62,63, whereas current evidence for ADP lands is limited12,13,64. This study indicates that ADP land stewardship and tenure recognition also stand to contribute to biodiversity and climate goals.

The social-historical assessment provides context for positive environmental outcomes in ADP lands documented through mapping and quasi-experimental analysis. Throughout the Americas, the African diaspora managed landscapes in ways that have conserved forests and biodiversity31,35,65,66. Numerous ADP cultural practices, rooted in both colonial and post-colonial survival strategies, included agrobiodiverse production systems that mimicked forest ecosystems that support sustainability. The innovations of ADP management practices and their contribution to cultural resilience alongside environmental conservation remains a vital contemporary solution to combat biodiversity loss and the climate crisis.

Similar to a study by the Rights and Resources Initiative (RRI) and Afro-organizations, our study underscores the urgent need for securing ADP territorial rights as a means to ensure environmental conservation and cultural survival10. Our socio-historical approach complements RRI and partners’ study by highlighting the contributions of ADP management practices that have been instrumental to enduring conservation of biodiversity and irrecoverable carbon. While previous studies provide a valuable lens on Black Geographies and Ecologies, they primarily focus on the United States context concerning racial and ecological equity26,27. We expand this scope by examining recognized ADP territories in four tropical countries in Latin America and socio-historical practices of ADP that originated in Africa and during the colonial period. Our work introduces additional dimensions to a Black Geographies and Black Ecologies framework by merging spatial and impact analysis with historical-environmental assessment, offering a more comprehensive understanding of ADP contributions to biodiversity and ecosystem conservation.

While ADP management practices are of immense cultural and ecological importance, ADP and their lands are under great pressure. ADP lands are often contiguous with other lands with diverse owners and differing land uses. Adjacent lands usually include plantation monocultures oriented to agroexports20,21,22,23,24,25,26,27,28,29 and extensive livestock systems. These agro-industrial production systems, managed for efficiency and expansion, predominate throughout the Americas67. Such biologically impoverished production systems exert pressure on ADP lands68 and threaten their high levels of biodiversity and irrecoverable carbon. A survey of ADP in Brazil showed that nearly all their lands faced pressures from infrastructure development, mining, and overlapping land claims69. In Colombia and Ecuador, ADP face escalating threats from violence, illicit economies, and the encroachment of armed and extractive actors endangering the lives, cultures, and livelihoods of ADP communities, particularly women and youth70,71,72. These pressures jeopardize the region’s fragile ecosystems and transform its biodiversity, cultural identity, and local economies through both licit and illicit monoculture production systems71. The matching analysis we conducted indicated that forest loss in ADP lands is lower but increasing, notably in Suriname, where rates of forest loss were relatively high within ADP lands outside PAs. Deforestation in ADP lands in Suriname has been attributed to logging and mining concessions as these lands lack formal recognition73. Small-scale and artisanal gold mining, livelihood sources for many Maroon communities in Suriname, are also known to drive deforestation74. Forest loss within ADP lands weakens the sustainability of ADP management systems and resilience of the forests and biodiversity they encompass. Further, forest loss undermines the material basis for ADP sustainable livelihoods and cultural beliefs and practices75, thus threatening ADP societies that have existed for generations.

Along with societal pressure, ADP face extreme climatic events like droughts and floods, which negatively impact their subsistence activities and livelihoods. In a dialogue with Conservation International about their ancestral territory, the ecosystems where they sustain their livelihoods, and the associated biodiversity, Saamaka people from the Pinkinslee community – one of six Maroon tribes of Suriname – identified over 15 varieties of Oryza glaberrima (African rice). However, when discussing the status of each variety, three were identified as being no longer cultivated. These varieties, despite their remarkable genetic diversity, are disappearing from the community’s biodiverse agroecosystems due to increasingly prolonged droughts, which prevent their propagation. ADP living in coastal areas of Ecuador face frequent and intense floods with negative impacts on health, food security, and production systems. Flooding increases the prevalence of infectious diseases (cholera, diarrhea, flu)76, limits ADP fishing capacity77, damages crops located on riverbanks, and threatens poultry farming76.

Based on empirical findings and urgency of threats, we propose policy and research recommendations to support ADP and their lands. Our first recommendation is to support legal recognition of ADP lands. Over 9.1 million ha of quilombo lands in Brazil and 1.76 million ha of ADP lands in Colombia remain unrecognized78. In Suriname, unrecognized land claims of the Maroon and Indigenous Peoples combined are estimated to exceed 10.5 million ha78. By demonstrating that legally recognized ADP lands contribute to significant reductions in rates of deforestation (29%–55%) compared to control sites, our study provides evidence of positive environmental outcomes associated with formal recognition of ADP land claims. Legal recognition of ADP lands has the potential to secure ADP resource rights for conservation purposes, thereby strengthening the position of ADP to make decisions regarding the use of their lands, including declaring them as OECMs or implementing other forms of protection. Legal recognition of ADP lands can improve ADP negotiating positions when facing outside economic interests, allowing ADP to ensure that their life plans and cultural practices are respected and sustained. Legal recognition of ADP lands can be anchored within SDG 1.4.2 and GBF Target 22 indicators, which track land tenure-related progress.

The second recommendation involves platforms like CBD and UNFCCC integrating sustainable ADP land management practices into conservation and climate policies. ADP practices are rooted in their ancestral knowledge, cultural identity, and spiritual belief systems. Recognition of their management practices would elevate ADP visibility in terms of acknowledging their rights and leadership in the conservation of biodiversity79 and irrecoverable carbon as well as increase representation of ADP through these policies. This recommendation can be implemented under GBF Target 21, which seeks inclusion of Indigenous and community knowledge into biodiversity action and would in turn support NDCs and National Biodiversity Strategies and Action Plans.

Lastly, we recommend support for researching ADP management practices and ecosystem dynamics to inform conservation and climate strategies. A gap remains in the current state of ADP knowledge systems and their link to contemporary management practices in relation to environmental conservation and climate change mitigation. It is crucial to identify the needs and priorities of ADP land management and consider future biodiversity loss, climate risks, and community vulnerabilities. Future research should map societal and climate-related threats specific to ADP lands to identify the most vulnerable locations and communities. This could be an important tool for prioritizing and channeling funds to ADP in areas needing urgent actions in conservation and climate adaptation. It is important to include both recognized and unrecognized lands in future analyses to more comprehensively represent the extent of ADP lands, and their potential for conserving biodiversity and irrecoverable carbon in the Americas.

We argue that ADP are important cultural and ecological stewards of their lands. Their sustainable management practices support positive conservation outcomes for biodiversity and climate. The documented management strategies of ADP, based in large part on adaptation across distinct contexts, make ADP vital environmental conservation partners against climate change. Support for ADP is urgent due to external threats to ADP lands and ADP cultural practices. Global environmental agreements, multilateral agencies, and governments should include ADP in environmental decision-making, and fund research and action to support ADP cultural practices that have led to environmental sustainability on their lands.

Methods

Mapping recognized Afro-descendant lands

We first collated and combined spatial datasets delineating ADP lands with legal tenure recognition. In the context of this paper, we use ADP lands, recognized ADP lands and recognized Afro-descendant lands interchangeably; all refer to delineated areas in which ADP communities possess recognized tenure (at minimum management rights). We considered ADP to have management rights if they can make decisions about their lands and resources within those lands.

We collected legal tenure information through which ADP land rights are recognized in each study country. We identified the legislation that first recognized the tenure system and rights of Afro-descendant communities to lands and waters and the years enacted, as well as the bundle of rights (access, use, management, exclusion, alienation) conferred to the communities through legislation, for each country (Supplementary Table 1). Legislation information was obtained primarily from FAOLEX80,81,82,83. Secondary supporting documents, such as peer-reviewed papers or NGO reports clarifying nuances or to better understand the legal frameworks and history of rights conferred, were obtained from other sources. ADP may also have other property rights described in the bundle of rights concept84.

Our study area covers four South American countries (Brazil, Colombia, Ecuador, and Suriname) where ADP communities have legally recognized land tenure rights to communal governance and management of their lands, and where we could obtain spatial datasets. In these countries, except for Suriname, ADP communities not only have legally recognized rights to manage lands but also possess or are in the process of receiving (such as in the case of certain quilombos in Brazil) collective titles through which they communally own delineated territories. In Suriname, ADP lack full legally recognized rights and thus are unable to obtain territorial ownership through collective titling or other means. However, we include ADP lands in Suriname in this study because we consider ADP in Suriname to at least have partial legally recognized tenure: through the Forest Management Act of 1992 (Wet Bosbeheer 1992, No. 80), they can have management rights to community forestry concessions for logging and customary land surrounding their villages. Recognized ADP lands in Suriname in our study represent areas that we could determine these management rights apply.

We obtained the recognized ADP lands datasets from the governmental web portals of Instituto Brasileiro de Geografia e Estatística (IBGE)85 for Brazil and Agencia National de Tierras (ANT)86 for Colombia; from EcoCiencia for Ecuador87; and from Conservation International for Suriname, which co-mapped spatial datasets in collaboration with and in support of the Matawai community (Supplementary Table 1). Country boundary polygons for all study countries were obtained from Global Administrative Areas (GADM) (version 4.1, 2022)88 and Esri89. Administrative unit boundary polygons were obtained from IBGE90 for Brazil, from Departamento Administrativo Nacional de Estadística (DANE)91 for Colombia, and from GADM (version 4.1, 2022)88 for Ecuador and Suriname.

Calculating the extent of ADP lands

To calculate areas of ADP lands, we used the ArcGIS Pro Python site package ‘ArcPy’ to automate spatial data processing and ensure that the ADP tenure spatial datasets would be uniformly processed. All ADP lands datasets were processed in ArcGIS Pro (version 3.2) and projected to WGS 1984 Web Mercator Auxiliary Sphere (WKID 3857) to maximize utility in calculating area numbers, determining areas of overlap across platforms, and overlaying other datasets. For consistency, all area values were then calculated using R Studio or Google Earth Engine (GEE).

There are limitations inherent to spatial analyses, which may occur due to approaches used to quantify the extent and presence of ADP lands. Limitations may be caused by different levels of accuracy for each input ADP lands dataset depending on how datasets were mapped by their creators, by the need to standardize ADP lands datasets to one coordinate system from different initial coordinate systems, and by differences between software and platforms used for processing and analysis (ArcGIS Pro, R, GEE). These limitations may also apply to other input datasets used.

The extent of recognized ADP lands we calculated does not reflect the extent of customary or ancestral territories of ADPs, much of which remains unrecognized. As such, our study only shows a conservative estimate of the extent of ADP lands in the four countries we examined, whereas ADP presence in other countries of the Americas and the Caribbean is documented elsewhere9,92. Further limitations to mapping the extent and presence are detailed in the Supplementary Note 1 section of this paper.

Mapping Afro-descendant presence

Information on the number and percentage of self-identified Afro-descendants in Brazil, Colombia, Ecuador, and Suriname (Supplementary Tables 3, 4) were derived from tabular datasets containing demographic information within level-2 administrative units from each country’s most recent census93,94,95,96. For Brazil and Colombia, Afro-descendant presence specifically within ADP lands, in addition to per administrative unit, was included as a component of population data for both censuses. For Ecuador and Suriname, since Afro-descendant presence was only included per administrative unit, we determined which administrative units overlap ADP lands by intersecting administrative unit88 and ADP land data we obtained for both countries. The Afro-descendant presence map (Supplementary Fig. 1) was created by joining census data with level-2 administrative unit data for each country (see administrative unit data sources in the Data Availability section below)88,90,91.

ADP lands, ecosystems, and protected areas

To provide additional context about the locations of ADP lands, we calculated the extent to which ADP lands overlap ecosystems and PAs within the study countries ArcGIS Pro (version 3.4). We determined which biomes and ecosystems overlap ADP territories in each country by projecting the RESOLVE Ecoregions and Biomes layer97 to WKID 3857, pairwise intersecting with the ADP territories layer, then calculating areas of overlap (ha). We determined which PAs overlap ADP territories by filtering the WDPA polygon layer98 by country (including Brazil, Colombia, Ecuador, and Suriname) and designation status (excluding Proposed), projecting to WKID 3857, pairwise intersecting with the ADP land layer, then calculating areas of overlap (ha). Since areas of overlap were determined through spatial analysis of polygon extents, we did not repeat this analysis for the WDPA points layer.

Calculating biodiversity within ADP lands

To estimate the spatial distribution of different levels of global importance for biodiversity in terms of terrestrial vertebrate species (amphibians, birds, mammals, and reptiles), we used a global RWR raster layer at 30-km resolution. This biodiversity metric shows the relative importance of a grid cell by accounting for both the number of species potentially present and the extent of their total ranges. The RWR raster, produced by IUCN using the Red List version 2023-1, represents where thousands of terrestrial vertebrate species are potentially present99,100,101. This metric is based on habitat ranges mapped by experts and informed by occurrence data, but values at a given point do not represent confirmed occurrences. Therefore, values represent the potential number of species present, given the overlap of geographic ranges with suitable habitat. The level of detail in the global database may not sufficiently capture local variations, and the accuracy of the database may vary by geographic region, as the quality of expert information could differ. These limitations have been captured by previous studies102.

Based on the global RWR raster, we estimated the spatial distribution of different levels of global biodiversity by calculating 5%, 10%, 15%, 20%, and 50% threshold values and masked the global raster to values at or above the corresponding thresholds. We clipped these masked global RWR rasters by the spatial extents of each region of interest (ROI): recognized ADP lands as well as overall areas nationally across the four study countries (Fig. 1). We then estimated how much of each ROI extent has high biodiversity by calculating the total number of pixels within each extent. The continent of Antarctica was removed from the global RWR raster before performing this analysis.

For additional biodiversity analysis, we evaluated the intersection of the spatial extent of each ROI with each species habitat range to generate lists of all terrestrial vertebrate species with ranges overlapping each ROI, as well as total species per taxonomic group (amphibians, birds, mammals, and reptiles) and IUCN Red List category (Critically Endangered, Endangered, Vulnerable, Lower Risk, Near Threatened, Least Concern, and Data Deficient)100,101. To account for ranges of certain species covering multiple countries in the study area, we analyzed each ROI individually within each country and across all study countries. With these output lists, for each ROI we determined the composition of terrestrial vertebrate species by their taxonomic groups and IUCN Red List categories and compared lists for all ROIs. All biodiversity spatial analysis was implemented in R statistical software version 4.3.1103, primarily through the’terra’ package version 1.7.39104.

Calculating irrecoverable carbon within ADP lands

We also intersected the spatial extent of each ROI with a global irrecoverable carbon20 raster to estimate total tonnes of irrecoverable carbon as well as generate lists of tonnes of irrecoverable carbon per ecosystem (mangroves, tropical and subtropical grasslands, tropical and subtropical forests, tropical and subtropical wetlands, and tropical peatlands). We generated these estimates for each ROI. We then calculated the proportion of irrecoverable carbon that is high irrecoverable carbon (>25 t/ha) following a previous publication20, and the average irrecoverable carbon (t/ha) within every ROI. Irrecoverable carbon values were generated using GEE.

We visualized levels of high irrecoverable carbon within recognized ADP lands and per country using the Irrecoverable Carbon 2018 dataset20 to map global irrecoverable carbon across ecosystems. We assigned the same irrecoverable carbon thresholds as the intersection analysis by a previous publication20 to the dataset’s pixel values, where areas containing high irrecoverable carbon have pixel values > 25 (t/ha).

The study by Noon et al. (2022) on mapping irrecoverable carbon does acknowledge certain limitations in quantifying irrecoverable carbon, including uncertainty in global estimates of irrecoverable carbon due to variability in data quality and availability across different ecosystems. This uncertainty can impact the precision of mapping. On smaller scales and at the pixel-level, this uncertainty falls within reasonable ranges; thus, we do not anticipate that this global uncertainty will significantly affect this study or the relative comparison of high irrecoverable carbon concentrations within ADP lands.

Quasi-experimental design and analysis: sampling grid

We subsampled areas inside and outside of ADP lands using a 300 m resolution sampling grid. This cell size was chosen to balance considerations of sample size, computation time, and spatial coverage of ADP lands (e.g., data exploration showed that larger cell sizes would exclude coverage of many smaller ADP polygons). To produce the sampling grid for statistical matching and quantifying forest cover change, we first overlaid ADP lands from Brazil, Colombia, Ecuador, and Suriname with country boundaries and removed island regions (Providencia Island, Colombia, and Galápagos Islands, Ecuador) that did not intersect with our spatial dataset of ADP lands. To generate treatment cells – i.e., sample cells within ADP lands – we created a 300 m grid covering the extent of ADP lands within each country. From this grid, we then randomly selected 10% of cells that fully overlapped ADP polygons. We subsetted in this way to (1) improve computation time at all steps in the data and analytical pipeline and (2) to reduce the potential spatial autocorrelation among grid cells. To generate a pool of potential control cells for matching, we first masked out ADP lands (including a 500 m buffer around ADP polygons) from country polygons and also excluded areas within 5 km of country borders. We then generated a 300 m resolution spatial grid from this masked area and randomly sampled cells within each country at a percentage that yielded a 20:1 control:treatment ratio. This ratio of control:treatment cells was chosen to balance computation limitations with the need for an excess of potential control cells to improve matching outcomes. This entire process was then repeated with the added step of masking out Indigenous Peoples and Local Communities (IP and LC) lands (Supplementary Table 16) (including a 500 m buffer around polygons) from control cell grids to create a complementary, IP and LC-exclusive sampling grid. The same seed was used to randomize subsetting of treatment cells in both the IP and LC-inclusive and IP and LC-exclusive sampling grids, meaning that treatment cells did not differ between the two sets. All spatial data processing for the quasi-experimental analyses was conducted in R, primarily using the terra package104 (version 1.7-55).

Quasi-experimental design and analysis: spatial covariates

We then extracted covariate information for each cell in the sampling grids by intersecting grids with multiple spatial layers representing factors that are known to influence the outcomes of forest cover, forest cover change, and associated carbon emissions. We first extracted PA information for each cell from the WDPA98. PAs represented only by point data were first buffered by their respective reported areas as recommended for analyses by maintainers of the WDPA at UNEP-WCMC. We excluded PAs for which designation status was Proposed or for which status designations were not available. Data extracted included original WDPA layer information (i.e., PA name, governance type, designation, IUCN category, etc.), the proportion of the cell that overlapped a PA, and whether the information was derived from a polygon or buffered point. PA information was preferentially extracted where cells overlapped spatially explicit PA polygons. For cells that overlapped multiple distinct PAs, preference was given to the PA with maximal overlap. In cases with equal areas of overlap, information from all relevant PAs was extracted. We also calculated the proportion of overlap of cells with IP and LC lands and territories.

We extracted six additional covariates from raster layers for each cell, including climate variables, elevation, and variables that characterize human modification of lands and access to markets and services, including the Human Impact Index, population density, and travel time to cities (Supplementary Table 14). We also calculated the distance between the centroid of each cell in the sampling grid and the nearest ADP polygon boundary, to examine distance dependence of potential ADP effects on forest loss. Sampling grids were then uploaded to GEE, which was used to extract yearly (2001–2021) forest cover as the difference between forest cover in the baseline year (2000) and the amount of forest loss occurring each subsequent year using the Global Forest Change (GFC) dataset105. To estimate the CO2e associated with forest loss, we first converted biomass data for each cell. Below-ground biomass (BGB) was estimated from above-ground biomass (AGB) using the equation below from a previous study106:

\({BGB}=0.489\times {{AGB}}^{0.89}\)

Next, we summed AGB and BGB to obtain the total biomass. This total biomass was then converted to biomass carbon (t C ha–¹) by multiplying by 0.5. Finally, we converted biomass carbon to carbon dioxide equivalent (t CO2e ha–¹) using the conversion factor of 3.67, based on the relationship that 1 tonne of carbon is equivalent to 44/12 tonnes of carbon dioxide. We then calculated rates of forest loss as the slope of a linear model fit to forest cover across all years in each grid cell and calculated total CO2e associated with forest loss as the sum of CO2e across all years for each cell.

Quasi-experimental design and analysis: matching approach

The purpose of statistical matching here is to reduce the influence of location bias on our estimates of the effects of ADP lands on rates of deforestation63. The locations of both ADP lands and geographic patterns of forest loss in the study region are likely non-random with respect to the distributions of human populations and land use, PAs, infrastructure, and accessibility. We performed statistical matching to improve the balance in the multivariate distributions of these potentially confounding variables between samples inside and outside ADP lands, allowing for a more accurate estimation of the effect of ADP lands on rates of forest loss. We first assessed correlations among possible matching variables (removing temperature as a matching variable, which was highly correlated with precipitation) and removed any grid cells without forest cover (<0.01 ha forest cover in 2000 in 9 ha cells, or < 0.11% forest cover). We then conducted matching separately for each country – i.e., ADP cells were only matched to controls within the same country – to account for socioeconomic and political drivers (e.g., GDP and national governance) that can vary substantially among countries. We matched each ADP cell to one control cell without replacement. The matching function minimized multivariate Mahalanobis distances between treatment and control cells for the following matching variables: the proportion of (maximum) PA coverage, forest cover in 2000, mean monthly precipitation, mean elevation, Human Impact Index, the natural log of population density, and the travel time to city (Supplementary Table 10). Matching was carried out using the R package MatchIT107 (version 4.5.5).

We performed an initial round of matching without using calipers (a parameter imposing minimum difference between matched variables), thereby keeping every ADP cell, along with an equal number of matching control cells, regardless of match quality. We then conducted a second round of matching, imposing minimal calipers determined through trial and error to keep the absolute standard mean difference of each variable in each country’s matched set below 0.25. This resulted in more closely matched cells at the cost of excluding some treatment cells for which there were not sufficiently similar controls in environmental space. Matching sets from each country were merged back together, resulting in four final, matched datasets: a calipered and non-calipered matched set each from both the IP and LC-inclusive and IP and LC-exclusive sampling grids. We fit models (as described below) with each of these datasets to assess the robustness of results to data processing decisions. We found that results were qualitatively consistent across datasets. We present results from the matched dataset with calipers and masking of IP and LC lands from the pool of controls.

Quasi-experimental design and analysis: avoided deforestation and carbon emissions

To estimate the effect of ADP lands on rates of forest loss and associated carbon emissions, as well as variation across countries and distance classes, we fit multiple Bayesian Generalized Linear Mixed models. We specified a Gamma probability distribution for each model to accommodate the right skewed and positive, continuous distributions of the response variables. We analyzed variation in two response variables: rates of annual forest loss and total CO2e associated with forest loss in each cell. To assess the overall impact of ADP lands on the response variables, we fit models with a categorical explanatory variable representing the six-level factorial combination of treatment (ADP vs. control cells) and PA category (cells fully inside PA, on PA edges, and fully outside PA). In these models, we also included the additive covariates of country, forest cover in 2000, mean monthly precipitation, mean elevation, Human Impact Index, and the natural log of travel time to city to control for any leftover variation in these variables post-matching (Supplementary Table 10). All continuous covariates were centered and scaled before modeling. Spatial correlograms indicated some spatial autocorrelation among grid cells at distances up to 200 km in some cases, likely reflecting large-scale, regional variation in forest loss. To account for spatial non-independence, we generated a 200 km spatial grid and grouped the 300 m sample cells that occurred within the same 200 km grid cell. We then included the 200 km grid cell ID as a varying intercept in the models, thereby accounting for potential non-independence of sample cells at this scale. To evaluate variation in the impacts of ADP lands across countries, we fit the same model structures described above, except we included the interaction between the country and the ADP-PA factor. To evaluate rates of forest loss at different distances from ADP polygon boundaries, we created a composite variable that reflected the 18-level factorial combination of treatment, PA category, and distance to ADP borders ( < 1 km, 1–10 km, and >10 km). We fit all models described above in Stan using the ‘brms’ package version 2.21.0108 to interface with R. For each model, we specified normal, uninformative priors for main effects parameters and ran eight chains for a total of 5000 iterations, discarding 500 iterations as burnin, and sampling every 20 iterations. We ensured that the chains mixed adequately and converged by inspecting traceplots and the Gelman-Rubin statistic (all ~1.0).

Potential limitations of this analysis include the possibility that unknown and unmeasured confounders may bias estimates of impact, even though we accounted for multiple potential confounders through matching. This is an inherent limitation of all large-scale quasi-experimental analyses that rely on remotely sensed datasets. In addition, there was limited forest cover change data prior to tenure recognition of the ADP polygons, which did not allow for robust analysis of before-after comparisons of tenure recognition. Therefore, our inferences are limited to control-impact comparisons across space. Codes related to the biodiversity, irrecoverable carbon, and quasi-experimental analysis are deposited in Zenodo109.

Social-historical assessment

In our study, we use a mixed methods approach with a concurrent convergence design110 to combine various research methods. This interdisciplinary approach aims to provide a comprehensive understanding of the contributions of ADP to conservation and climate solutions.

For the social-historical assessment, we drew from multiple academic disciplines and sources to illuminate the factors influencing the situation under study through qualitative methods: the settlement patterns of enslaved individuals in the Americas and their management of tropical ecosystems, with a specific focus on livelihood practices. These two categories guided the content analysis and facilitated the collection and analysis of descriptive and contextual data from the consulted texts. We reviewed interdisciplinary sources, including works by historians, geographers, anthropologists, botanists, paleontologists, and environmental sociologists. We extensively reviewed Chapter 13, titled “African Presence in the Amazon: A Glance” from the Scientific Panel for the Assessment of the Amazon which included 100 references33. Additionally, we used the Web of Science digital library due to its provision of high-quality bibliographic records and identified 35 more references using the snowball technique to address our selected categories effectively.

The snowball technique led us to bibliographic references focused on regions of origin and domestication of numerous plant and animal species introduced to the Americas. A small but important body of scholarship—primarily from non-Spanish-speaking countries and published in English, Portuguese, and Dutch—has documented the African and Asian origins of many of these species. Although not directly tied to global economic systems, these species played a vital role in the subsistence and food practices of newly arrived populations to the Americas, including ADP. As a result, origin and domestication emerged as a third analytical category, emphasizing Afro-descendant ecosystem management practices and revealing their contributions to biodiversity and environmental stewardship, particularly in Suriname, Brazil, and Colombia. We found fewer references addressing this topic in the case of Ecuador. These scholars also trace the introduction and ecological adaptation of such species within Afro-descendant settlements—ranging from plantations and mining zones to forested regions where Maroons sought refuge. The association between species domestication and these settlement patterns enabled us to trace practices developed in Africa and adapted to the ecological conditions of the American tropics.

There is a linguistic and geographic gap in the literature we reviewed. Many key contributions of ADP to the Americas are documented in non-Spanish-speaking countries, often in English, Portuguese, and Dutch languages. This language barrier limited the analysis and likely excluded valuable insights. Together with spatial and statistical analysis, future research should explore non-Spanish and non-English literature and highlight ADP current practices from underrepresented regions and countries to gain a fuller understanding of ADP’s role in conservation.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.