1. Introduction #

1.1 Nature crime in the Amazon #

“Nature crime” is used to refer to illegal forms of logging, ecosystem conversion, wildlife exploitation and trafficking (both domestic and international), fishing, and mining, and is often associated with financial offenses, fraud, corruption, organized crime, as well as labor and human rights abuses.¹ In the Amazon, it can be understood as a portfolio of illegal or illegality-tainted activities that harm forests, rivers, wildlife, and local populations,² although its legal definition is subject to variations depending on the environmental regulations of each Amazonian country and what constitutes a violation of these regulations (see Chapter III for a review of legal definitions in each country of the Amazon).

These activities do not operate in isolation. Criminal networks involved in illegal mining, deforestation, and drug trafficking are often involved in more than one activity and frequently share financing channels.³

Establishing a baseline of the extent, distribution and severity of these activities across the entire Amazon is a challenge, as they cannot all be represented spatially with the same precision or consistency. Financial transactions and criminal networks may be geographically dispersed, obscured behind opaque transactions, or impossible to attribute to specific locations. To address this limitation, the analysis presented in this report focuses on manifestations of Nature crime: the physical, observable consequences of illegal activity that leave a detectable footprint — on land cover, in river systems, or in records of violence — and that can be systematically tracked across all countries of the Amazon basin.⁴

This report tracks the following potential manifestations of Nature crime:

• Illegal deforestation: the permanent clearing of forest cover for agriculture, cattle ranching or other uses, that is unlawful in character — whether through the appropriation of public land or through violations of the rules governing land use and conversion in protected areas, Indigenous territories, or private properties.

• Illegal logging: the extraction of timber in violation of national forest management laws — including without a valid management plan, in excess of authorized volumes, in areas where logging is prohibited (such as protected areas and Indigenous territories), or using fraudulent documentation to launder illegally felled timber into legal supply chains. Illegal logging is both a significant source of forest degradation in its own right and a common precursor to full-scale deforestation, as selectively logged areas become more vulnerable to fire and agricultural encroachment.

• Illegal artisanal and small-scale gold mining (land-based and river-based): gold mining has expanded dramatically across the Amazon over the past two decades, driven by rising prices, weakened enforcement, and the entry of organized crime groups that use it as a vehicle for money laundering and territorial control, making it a defining feature of the region’s criminal economy. Illegal mining is both a source of deforestation, through direct removal of vegetation to access the soil where gold is found, and ecosystem degradation, through pollution of river courses from mercury use.

• Mercury pollution: used in artisanal gold mining to bind and separate gold from sediment, mercury enters Amazonian river systems as a persistent, bioaccumulative toxin that travels across national borders through shared waterways and the food chain, reaching dangerous concentrations in fish and in the Indigenous and riverine communities that depend on them.

• Threats and violent deaths of environmental defenders: attacks on individuals who monitor, report, or resist illegal activities in and around areas subject to Nature crime. These incidents are increasingly recognized as a direct and foreseeable consequence of the broader pattern of illegal extraction, and are often perpetrated by or on behalf of the same criminal networks responsible for the crimes being denounced.

  1.2 Why assess the extent and severity of Nature crime across the Amazon? #

The ultimate goal of this analysis is to take stock of data available across Amazonian countries that could enable consistent, systematic tracking of the spatial and temporal variation of Nature crime manifestations across the Amazon — understanding where they are concentrated, intensifying, and appear to be receding. This approach seeks to address a current challenge in our understanding of the dynamics of environmental degradation that stem from criminal or illegal activities: the absence of any single, integrated data source covering all eight countries of the Amazon basin with consistent definitions and methodologies. This results in an inability to assess the aggregate scale of the problem or to identify transboundary dynamics.

Various government and civil society-led platforms generate data relevant to Nature crime monitoring across the Amazon, many of them open-source. Yet most cover only a subset of countries or years, and their fragmented, retrospective character makes it difficult to assess regional trends or the cumulative scale of the problem.

Other organizations have attempted to map Nature crime across the Amazon previously. The Igarapé Institute and InSight Crime’s Amazon Underworld platform and associated policy papers have documented how illegal mining, logging, land clearance, and wildlife trafficking are driven by organized criminal networks across five Amazonian countries, drawing on field investigation and open-source research. The Global Initiative Against Transnational Organized Crime (GI-TOC) published Environmental Crimes in the Amazon: Current Trends and Rising Threats, which identified over 4,000 illegal mining sites across the region in 2023, mapped deforestation hotspots in Brazil, Colombia, and Peru, and documented how criminal organizations have adapted their methods to exploit regulatory gaps. While they serve to raise awareness of the severity and extent of Nature crime as a driver of environmental degradation across the Amazon, they fall short of a systematic assessment of available data across all countries that may be used to establish a baseline of Nature crime and monitor it regularly.

Rather than use in specific legal or enforcement proceedings based in data about individual instances of environmental violations, the primary purpose of such a baseline is to raise awareness among the public of trends in the scale of the issue, to mobilize resources, and to focus policy and enforcement efforts – including contributions from civil society to these efforts – in jurisdictions where the data show that they are most needed. Several audiences stand to benefit from such a sustained, regionally integrated monitoring effort.

• For governments, consistent data on the spatial and temporal variation of Nature crime — where it is intensifying, where it is receding, and what factors are associated with each — enables a more rigorous assessment of the effectiveness of enforcement policies and operations. A regional evidence base is especially valuable for evaluating cross-border interventions and building the case for sustained investment in detection and prosecution capacity.

• For regional governance and cooperation bodies (e.g. ACTO’s public safety commission, Amazonian Police Cooperation Center – CCPI), there is value in tracking nature crimes across the Amazon as a whole, rather than within individual national borders, and across manifestations, rather than focusing on a single issue. Nature crime is increasingly transboundary: border areas are disproportionately affected by illegal activity, and criminal networks routinely exploit jurisdictional gaps and corruption. The same criminal networks also operate across various categories, highlighting the value of datasets that consider more than one single manifestation. A pan-Amazonian evidence base is therefore essential for understanding the aggregate scale of the problem and for enabling the multi-country policy response it demands.

• For Indigenous and local communities, a quantitative baseline of the nature crimes they face provides a foundation for advocacy with governments, donors, and international bodies, establishing more clearly when the level of threats they are facing has increased markedly and warrants further protection.

• For civil society organizations and the donor community, a regional overview helps identify where needs are greatest and where support is most likely to have impact — whether in strengthening monitoring systems, funding enforcement operations, or supporting the communities most exposed to Nature crime.

• For the private sector, spatial data on nature crime is a valuable input to their risk management and due diligence processes. Companies importing into the European Union face supply chain due diligence obligations under voluntary and/or regulatory frameworks (such as the EU Deforestation Regulation – EUDR) and require granular, reliable data on deforestation risk in their sourcing areas. Financial institutions — banks, asset managers, and insurers — need to understand their exposure to assets tainted by illegal land conversion, a risk documented in detail in a 2024 WWF-UK report on financial crimes and land conversion.⁵

Against this backdrop of growing relevance for policy-makers, enforcement agencies, companies, investors and civil society, a question emerges: how well can the data currently available across the Amazon basin support such assessments, and where do the critical gaps lie? Answering this question first requires understanding the type of data that would be required to assess the extent and severity of Nature crime in the Amazon.

1.3 How to assess the extent and severity of Nature crime #

A central difficulty to mapping Nature crime in the Amazon is that it is systematically underreported, under-prosecuted, and under-sentenced. In Brazil — arguably the country in the region with the most developed detection and enforcement systems — only around 45% of illegally deforested areas were issued infraction notices even during periods of heightened enforcement.⁶ A major contributing factor is low reporting. In many areas, more than half of all crimes are never brought to the attention of authorities, largely because communities have little confidence that action will follow.⁷ Basing the analysis on criminal convictions would provide a stronger evidentiary foundation — but conviction rates remain low: in Brazil, formal charges (indiciamento) are filed in only around 25% of environmental crime investigations.⁸

Thus, a picture of where Nature crime is occurring or of its scale cannot be drawn from an analysis of criminal records, which would only show instances of environmental harm actually prosecuted and sentenced. This report instead maps potential manifestations of Nature crime: observable events that combine two elements — (i) evidence of environmental harm, established through remote sensing or other spatially explicit data record; (ii) grounds for characterizing that harm as a violation of applicable law, based on land designation indicating what uses, changes to natural vegetation cover, and activities are authorized. For some offences, the act itself is inherently unlawful and a spatially explicit record of its manifestation is sufficient to establish that violation (for example, release of mercury from illegal mining, or violence against an environmental defender).

This report surveys the data sources currently available across the eight Amazon countries for each of these manifestations of Nature crime, assesses their potential for consistent pan-Amazonian aggregation and presents illustrative analyses that demonstrate how these sources can be combined to shed light on the scale and legality of the harm.

2. Illegal deforestation #

Illegal deforestation is the permanent clearing of forest that is unlawful in character — whether through the appropriation of public land or through violations of the rules governing land use and conversion in protected areas, Indigenous territories, or private properties. Information on the driver of forest cover loss is also particularly relevant to an assessment of legality, as it facilitates the identification of activities that may constitute a violation of applicable regulations protecting against environmental harm.

Deforestation monitoring systems serve three distinct analytical purposes, and distinguishing them clarifies what any individual dataset is or is not suited to show. Historical monitoring reconstructs long-term forest loss trajectories from archived satellite imagery on a consistent methodological basis, typically back to the 1980s or early 2000s; its value lies in establishing reference baselines, tracking multi-decadal trends and anchoring carbon-accounting and REDD+ commitments. Annual monitoring produces one consolidated measurement of deforestation per year — usually with a one- to two-year publication lag to allow for cloud-free imagery compositing and validation — and supplies the figures countries report to international frameworks such as the UNFCCC and against which year-on-year policy performance is assessed. Near-real-time monitoring generates alerts at weekly or sub-weekly frequency, prioritising temporal responsiveness over definitive area estimates, and supports enforcement interception, operational targeting and civil-society monitoring before clearance progresses to full forest loss.

Each cadence plays a distinct role — historical data anchors the baseline against which recent change is read, annual data supports statistical assessment of illegality shares and drivers, and real-time alerts underpin the enforcement response.

Section 2.1 surveys official and civil-society deforestation datasets across the eight Amazon countries and assesses their potential to support a consistent pan-Amazonian baseline. Section 2.2 then presents four analytical approaches that illustrate how existing data can be used to characterize the legality, drivers and economic significance of illegal deforestation to approximate an Amazon-wide baseline for this Nature crime. Section 2.3 concludes on the remaining gaps for a full assessment of illegal deforestation in the Amazon and lists potential ways forward.

2.1 Review of Available Data from Governmental and Non-Governmental Sources #

Official National Monitoring Systems #

Six of the eight Amazonian countries maintain official forest monitoring systems, though they vary significantly in methodological approach, spatial resolution, temporal frequency, and public accessibility.

• Only Brazil, Colombia, and Peru operate monitoring systems with near-real-time alert capability, essential for reactive enforcement against illegal deforestation.

• Bolivia, Brazil, Colombia, and Peru maintain annual datasets of deforested areas that are accessible online

• Ecuador‘s biennial monitoring cycle and Guyana‘s data lag of at least two years mean that neither system can support annual data series.

• Suriname‘s Gonini platform was offline at the time of writing, with its most recent data potentially dating to 2023.

• Venezuela presents the starkest gap: institutional collapse has left the country without a functioning official forest monitoring system since 2011, a period that coincides with a significant expansion of illegal mining and associated deforestation.

• Colombia appears to be the only country in the region to have conducted a classification of its official deforestation data by drivers, although that exercise was last conducted in 2015.

Table 1. Official deforestation and degradation data in Amazonian countries.

The systems also differ in methodological approach, spatial resolution, and temporality (Brazil’s Prodes deforestation year runs from August to July of the following year). In addition, only Brazil’s PRODES and DETER systems, Colombia’s IDEAM, and Peru’s Geobosques routinely publish their data in formats readily available for download that could support cross-border analysis. Aggregating these national datasets into a consistent Amazon-wide annual baseline is currently not feasible as data is not available for several countries and would require significant methodological harmonization where it is available.

Regional and Civil Society Monitoring Platforms #

Several civil society and multi-institutional platforms provide deforestation data at regional scale, in some cases offering the most complete and consistent cross-country coverage available.

Table 2. Civil society datasets relevant for the mapping of illegal deforestation

Global Forest Watch (GFW) / University of Maryland (UMD) annual tree cover loss data provides globally consistent, annual estimates of tree cover loss at 30-meter resolution from 2000 through 2025. The UMD annual product does not directly distinguish between legal and illegal forest loss. A recent release also available on the GFW platform disaggregates detected deforestation by direct driver (agriculture, mining, logging, settlements, other), at 1km resolution.⁹ This methodological step forward is used in Section 2.2 of this report to support a more fine-grained assessment of the relationship between deforestation data and efforts to map Nature crime.

MapBiomas Amazonia stands out as a strong foundation for a consistent pan-Amazonian deforestation baseline. It applies a harmonized methodology across all eight countries, producing land cover and land use transition data at 30-meter resolution from 1985 to 2023, a bigger time lag than the UMD data. Its ability to distinguish between different land use categories (agriculture, pasture, mining, urban) provides insights into deforestation drivers, and its forest degradation layer complements the deforestation detection data. Importantly, it is specific to the Amazon region, meaning that the categories of drivers used may be more in line with the reality of the dynamics on the ground.

RAISG’s pan-Amazonian geospatial platform, meanwhile, provides the most comprehensive available dataset on land designations — protected areas, Indigenous territories, and mining concessions — that are essential for legality assessment. Its coverage is genuinely pan-Amazonian and reflects collaborative data-sharing between civil society organizations across the region, though certain land designations may not be specific enough to establish legality, given the simplification of data associated with producing wall-to-wall coverage across 8 countries, each with their own legal system and land designation categories.

2.2 Relevant analyses and their potential for scaling #

Qualification of deforestation data across Brazil #

A recurring limitation of deforestation monitoring is that detecting where forest loss has occurred tells us little, on its own, about whether that loss was illegal and constitutive of a potential Nature crime. Remote-sensing deforestation data captures the physical fact of clearance; determining its legal status requires additional information on what is permitted under the applicable land designation. Illegality, for example, can be presumed if the clearance has occurred in a Protected Area with strict conservation status, such as a National Park. The illegality of forest cover loss in conservation areas that allow for sustainable use and in Indigenous Territories will depend on the specific rules for that area and the size of the clearance. While small areas of forest loss may indicate sustainable use activities compatible with the designation, large-scale deforestation carries a stronger presumption of illegality.

Over private land, lawful deforestation requires a “vegetation suppression authorization” (Autorização de Supressão de Vegetação, or ASV). The Instituto Centro de Vida (ICV) has pioneered a methodology that systematically crosses georeferenced PRODES deforestation data against ASV records held by SINAFLOR (the federal system for tracking the origin of forest products) and by state environmental agencies. Where cleared areas cannot be matched to a valid, georeferenced authorization, deforestation is classified as unauthorized — a reasonable operational proxy for illegality under the Brazilian Forest Code.

Applied to the most recent PRODES monitoring cycle (August 2024 to July 2025), the analysis found that approximately 90% of deforestation in the Brazilian Amazon — around 4,400 km² — occurred without a recorded authorization. The ten municipalities with the highest share of unauthorized clearance are concentrated in Pará, Amazonas, and Mato Grosso, consistent with longstanding frontier deforestation patterns (See Figure 1 below).

The robustness of this approach depends entirely on the quality and accessibility of ASV data. ICV’s own audits of state environmental agencies reveal that several states do not publish georeferenced authorization records at all, or publish them in non-machine-readable formats that preclude systematic cross-referencing. Only Pará and Mato Grosso currently maintain digitized, publicly accessible ASV databases at a quality that supports reliable illegality assessment.¹⁰ In other states, the estimate of unauthorized deforestation should be treated as a lower bound, as some authorizations exist but cannot be matched due to data fragmentation. Recognizing this, IBAMA launched in 2025 a dedicated panel (SINAFLOR x PRODES) that automates the cross-referencing of federal authorization records against PRODES data, representing a significant institutional step toward systematic illegality monitoring across the Legal Amazon.¹¹ Despite lacking a map visualization similar to ICV’s, the availability of this tool demonstrates the capacity of civil society initiatives to pave the way for official systems.

Figure 1. Authorized deforestation in Amazon and Cerrado biomes in Brazil (Source: Instituto Centro de Vida, 2026).

Figure 2. Sinaflor x PRODES dashboard of illegal deforestation

Potential for replication across the Amazon. ICV’s and Sinaflor’s approach is replicable in principle: it requires a reliable deforestation detection layer (analogous to PRODES), a comprehensive and georeferenced database of vegetation clearance authorizations, and the legal framework that makes unauthorized clearance presumptively illegal. Brazil is unusually well-equipped on all three dimensions.

In most other Amazonian countries, one or more of these conditions is absent or insufficient.

• Colombia and Peru have functional deforestation monitoring systems but lack national clearance authorization databases of equivalent coverage and accessibility. None of the other Amazon countries currently has a fully equivalent, publicly georeferenced vegetation-suppression-authorization registry. Peru, Colombia and Bolivia maintain partial records of forestry and agricultural clearance permits within their respective environmental authorities, but these are not consistently geospatialized or published in machine-readable form. Ecuador, Guyana, Suriname and Venezuela do not publish equivalent registries at all.

• Bolivia’s ABT360 platform does include an up-to-date layer of “Areas designated for controlled grassland burning by the Agricultural Superintendency.” This information forms an integral part of Administrative Resolution No. 147-2005 issued by the Agricultural Superintendency. It could be used in conjunction with remote sensing fire detection data to qualify the legality of these events. However, coverage appears to be limited to pastureland, limiting its usefulness for analysis in conjunction with data on forest cover loss.

Closing these gaps would require both technical investment — in building interoperable ASV-equivalent registries — and legislative and regulatory reform to establish mandatory georeferenced reporting of clearance authorizations at national level. ICV’s Brazilian model offers a proven template; adapting it regionally is a medium-term objective that the ASL program could support through targeted technical assistance to national environmental agencies.

Manual qualification of deforestation alerts in Mato Grosso #

Between 2022 and 2025, Amazon Conservation conducted a four-year monthly monitoring program in Mato Grosso in collaboration with ICV, covering the entire state. Each month, Amazon Conservation identified major new forest loss detections from GLAD alerts (University of Maryland) and confirmed candidate cases as deforestation through visual interpretation of high-resolution satellite imagery.

Confirmed cases were entered into a shared tracking database and classified by driver. ICV then assessed the legal status of each case using land registry records — primarily the Rural Environmental Cadastre (CAR) — to determine whether the clearance occurred on land where deforestation was authorized under the Forest Code. For gold mining cases, legality was assessed by reference to valid mining concession records and the protected status of the land in question. The scope of driver categories tracked evolved across the monitoring period: the 2022 campaign covered soy and cattle; a third category (non-soy crops) was added in 2023; and gold mining was incorporated in 2024, producing a complete four-driver dataset from 2024 onward.

Across four years of monitoring, 479 cases of deforestation were confirmed in Mato Grosso, covering a total area of 131,702 hectares. Of these cases, 67% (321) were illegal by count. Measured by area, 61% of cleared land — approximately 80,000 hectares — constituted potential manifestations of Nature crimes, with the remaining 39% attributable to legal clearance under Forest Code authorizations. The distinction between case count and area reflects the different scales of illegal and legal operations: legal clearances, which require prior authorization and are subject to procedural controls, tend on average to cover larger areas per case than the opportunistic, often small-scale incursions that characterize illegal activity.

For the two years with complete data across all four categories (2024–2025), 336 cases of deforestation were confirmed, of which 75% (252 cases) were illegal. The 336 cases covered a total area of 86,615 hectares, of which 70% (60,553 hectares) was illegal.

Figure 3. Deforestation events in Mato Grosso for the period 2022-2025, classified by driver and assessment of illegality (Source: Amazon Conservation/Instituto Centro de Vida, 2026).

The driver breakdown of illegal areas reveals a contrast between case frequency and spatial impact (Table 3). Gold mining accounted for the largest share of illegal cases (70% of illegal cases) but cattle pasture expansion was responsible for the largest share of illegal deforestation (49% of illegal area, 29,854 ha), followed by non-soy crops (39%, 23,783 ha) and soy (1%, 862 ha). Illegal soy deforestation, while rare in absolute terms, is disproportionately significant from a supply chain accountability standpoint given the traceability mechanisms applied to Brazilian soy exports.¹²

Such disaggregation of both the drivers of deforestation and their legality opens new possibilities of advocacy. One example is pressure on government authorities from supply chain actors that may be bound by strict due diligence requirements. Replicated across the Amazon, it could provide a useful complement to supply chain traceability data such as Trase.earth and enable sector-specific, jurisdictional level assessments of illegal deforestation exposure risk.

Table 3. Driver breakdown of illegal deforestation by case count and area, Mato Grosso, 2024–2025.

Economic impact of illegal deforestation in the Brazilian Amazon #

Illegal deforestation in the Brazilian Amazon generates short-term private benefits for perpetrators while imposing substantial, uncompensated socio-environmental costs on society. The Conservation Strategy Fund’s deforestation impact calculator estimates the full socio-environmental cost of illegal clearance by monetizing carbon emissions, biodiversity loss, water-regulation services and other ecosystem-service values, and comparing these to the restoration and penalty costs that responsible parties would face under full enforcement. The tool was first developed for illegal gold mining and has since been extended to cover five Amazon countries. Economic valuation provides a foundation for assessing the severity of Nature crime across the Amazon through more than a strictly area lens. It can also help calibrate fines and damages to be paid by perpetrators as well as environmental restoration costs, strengthening accountability and environmental justice.

In 2023 alone, approximately 2.9 million hectares of forest and native vegetation were lost in the Brazilian Amazon, of which 2.4 million hectares were converted into pastureland, demonstrating the significant role of cattle ranching as a primary driver of illegal deforestation relative to other drivers of deforestation. Second was conversion for cropland (incl. soy), with 17,000 hectares. An estimated 90.8% of this deforestation is estimated to be illegal, as per the ICV study cited above.

The methodology integrates three valuation approaches:

1. Opportunity Cost (Direct Values): This approach estimates the foregone economic benefits from sustainable forest uses, specifically the responsible extraction of timber and non-timber forest products such as Brazil nuts and rubber.

2. Loss of Ecosystem Services (Indirect Values): This approach values the loss of environmental services provided by standing forests, including carbon sequestration (valued using the Social Cost of Carbon and REDD+ prices), erosion control, and biodiversity conservation.

3. Replacement Cost: This approach estimates the financial cost required to recover forest cover using different restoration techniques, ranging from lower-cost natural regeneration to more complex methods such as direct seeding.

The aggregation of these costs reveals the scale of economic liability associated with illegal deforestation. The total estimated socio-environmental cost of illegal deforestation in the Brazilian Amazon varies between US$ 26 billion and $111 billion, depending on the land-use conversion type, the carbon pricing model applied, and the chosen restoration strategy.

Table 4. Cost of illegal deforestation in Brazilian Amazon (Source: CSF)

The results indicate that while the conversion of forest to agricultural cropland generates higher per-hectare restoration costs due to greater soil disturbance and chemical inputs, the conversion to pastureland represents a substantially higher total financial burden due to the larger scale of converted areas. On a per-hectare basis, the cost is estimated at US$ 2,476/ha for agricultural conversion compared to US$ 1,097/ha for pastureland, assuming natural regeneration is used to restore the land to its original cover. If more costly direct seeding techniques for restoration are used, these costs increase to US$ 6,532/ha and US$ 3,125/ha, respectively.

One significant limitation of this analysis is that it assumes that restoration mandates will be enforced and that responsible parties will bear the full costs of environmental recovery. This assumption does not reflect current implementation realities in the Brazilian Amazon. According to the Termômetro do Código Florestal,¹³ the Amazon contains 9.4 million hectares of “Legal Reserve” deficit (areas that should be conserved under the Forest Code but have been illegally deforested) and approximately 800,000 hectares of “Permanent Protection Area” deficit. The persistence of these substantial deficits reflects weak implementation capacity at state and federal levels and limited enforcement of restoration obligations. This implementation gap means that the restoration costs estimated in this analysis remain largely unrealized, and the economic liability of illegal deforestation continues to accumulate across the region.

From a policy perspective, these findings indicate that preventing illegal deforestation is the most cost-effective strategy. The financial resources required for restoration, particularly for large-scale pastureland conversions, are substantial relative to public budgets. Additionally, because ecological recovery occurs over decades, these monetary estimates likely represent a partial accounting of the long-term losses in biodiversity and ecosystem services.

The deforestation impact calculator is well-suited to replication across Amazon countries, following the precedent of CSF’s mining impact calculator, now covering five countries. Its most important operational use-cases lie in prosecution (establishing quantified damages for criminal and civil proceedings) and in remediation planning (setting the scale and cost of restoration obligations for responsible parties). Expanding the tool to cover all eight ASL countries, with harmonized parameters, would strengthen accountability across the basin.

Deforestation in Protected Areas and Indigenous Territories as a proxy for illegal deforestation across the Amazon: baseline for the year 2024 #

This section presents a pan-Amazon proxy for illegal deforestation that combines two widely available datasets: (i) the official cartographies of Protected Areas (PAs, here National Parks) and Indigenous Territories (ITs) compiled by RAISG for the eight Amazon countries;¹⁴ and (ii) the University of Maryland’s recently released Mapping Drivers of Primary Forest Loss (2001–2024) dataset, which attributes 1km forest-loss pixels to one of six direct drivers (wildfire, agriculture, hard commodities, logging, settlements and other natural disturbances).¹⁵ Deforestation inside strictly protected PAs and titled ITs is, with limited exceptions such as subsistence agriculture, not authorized under national law and therefore provides a first-order proxy for illegal deforestation. Crossing these land designations with driver attribution allows the analysis to separate losses associated with Nature crime drivers (agriculture, mining, logging, settlements) from those linked to wildfire and natural disturbances.

Over 2001–2024 across the eight Amazon countries and territory, cumulative forest loss inside National Parks reached approximately 3.9 million hectares of non-fire loss and an additional 2.0 million hectares of fire-related loss. Inside Indigenous Territories, the corresponding figures were 2.4 million hectares of non-fire loss and 2.6 million hectares of fire-related loss. 2024 was a record-breaking year in both categories, with fire emerging as the dominant immediate driver in the Brazilian, Bolivian and southern Peruvian Amazon — a pattern consistent with the exceptional drought of that year due to the El Niño phenomenon.

Figure 4. Annual forest loss (2001–2024) inside Protected areas and Indigenous territories across Amazon countries, disaggregated by fire and non-fire. Source: authors’ own analysis

To produce a more tractable pan-Amazon illegality baseline, the analysis was restricted to the 23 National Parks and 86 Indigenous Territories that each recorded more than 1,000 hectares of cumulative forest loss for the year 2024. Within this filtered subset, wildfire accounted for approximately 84.4% of forest loss in the most-affected National Parks and a comparable share in the most-affected Indigenous Territories. The four drivers directly associated with Nature crime — agriculture, hard commodities (mining), logging and settlements — together accounted for approximately 13.2% of forest loss in National Parks and a similar share in Indigenous Territories.

Country-level patterns differ markedly. In Brazil and Bolivia, fire dominates forest loss in both PAs and ITs, with substantial agriculture-driven loss at IT boundaries. In Colombia, agriculture (notably cattle expansion in the northwestern Amazon) is the dominant non-fire driver inside National Parks and titled ITs. In Venezuela, hard commodities — overwhelmingly illegal gold mining — dominate forest loss inside PAs, with Canaima National Park, the Imataca Forest Reserve and Indigenous lands affected by the Arco Minero decree accounting for the bulk of losses. In Peru and Ecuador, agriculture and mining both register as significant drivers, with a marked expansion of mining-related loss in the Ecuadorian Amazon (Punino basin, Podocarpus) and in Madre de Dios (Peru) in 2024.

Figure 5. Dominant direct driver of forest loss inside the 23 National Parks of the Amazon with >1,000 ha cumulative loss (2001–2024). Source: authors’ analysis of UMD Mapping Drivers of Primary Forest Loss.

Figure 6. Dominant direct driver of forest loss inside the 86 Indigenous Territories of the Amazon with >1,000 ha cumulative loss (2001–2024). Source: authors’ analysis of UMD Mapping Drivers of Primary Forest Loss.

High-resolution Planet satellite imagery was used to confirm the drivers in a sub-sample of detections. Agriculture-driven losses were visually confirmed in 12 National Parks and 38 Indigenous Territories, with distinct patterns of pasture expansion and mechanized clearing. Gold-mining-driven losses were confirmed in 6 National Parks and 15 Indigenous Territories. Logging-driven losses were confirmed in a smaller subset, generally along access roads and river networks. As a first-order screening tool, this approach helps distinguish likely illegal forest loss from fire and natural disturbance in 2024, while recognizing that it does not substitute for case‑level legal determination of potential Nature crimes.

2.3 Conclusion on Amazon-scale data availability #

Three conclusions emerge from this review:

First, a direct aggregation of official national data is not feasible for pan-Amazonian analysis. No consistent, up-to-date deforestation baseline can be assembled from official sources alone, as these sources are often outdated or, in some cases, inexistent. Where they are up-to-date, their formats and methodologies vary, making them difficult to compare. Any Amazon-wide analysis will thus need torely on civil society-generated data, which may then limit its acceptability for consideration in official decisions in intergovernmental cooperation fora, such as ACTO.

Consolidating a pan-Amazon deforestation figure will require closer cooperation between the national agencies responsible for these datasets. Aligning definitions of “forest” and “deforestation”, harmonizing minimum-mapping units and driver categories, and agreeing on a common reporting cycle are pre-conditions for aggregation of national datasets. The Amazon Cooperation Treaty Organization (ACTO) is well-placed to broker this coordination, building on the 2023 ministerial decisions of the Belem Declaration and the forthcoming Strategic Cooperation Agenda.

The legality assessment layer remains the main constraint. Even where deforestation detection data is consistent and up-to-date, establishing illegality requires an equally consistent layer of georeferenced government permits (such as mining concessions), land designation data — distinguishing protected areas, Indigenous territories, and private properties and the rules that apply to each — across all eight countries. RAISG’s geospatial platform provides the most comprehensive cross-country foundation for this layer, but supplementary national cadastral data is needed to establish illegality outside of protected areas and Indigenous territories.

Driver-disaggregated data is a significant step up in data availability that helps refine assessments of legality across relevant regulations, bypassing the need for data on deforestation authorizations, for which there are not consolidated and spatially-explicit records in many countries.

In the meantime, the most tractable near-term baseline for illegal deforestation combines UMD driver attribution with RAISG data on protected areas and Indigenous territories. Designations for conservation alone cannot capture the full legality context on private land, but they provide a common unit of analysis across the eight countries that is not available from any official pan-Amazonian system today. Taken together, they represent a defensible proxy for the spatial footprint of illegal deforestation across the Amazon, across several drivers each of them indicative of specific potential manifestations of Nature crime.

3. Illegal logging #

3.1 Review of national-scale data availability #

Monitoring illegal logging presents methodological challenges that are fundamentally distinct from those of deforestation monitoring. Complete forest clearance — the focus of national and regional alert systems — is easily detected in medium-resolution satellite imagery. Selective logging, by contrast, removes individual trees while leaving the canopy largely intact, generating subtle signals that require very high-resolution (VHR) imagery or specialized detection algorithms to detect reliably. Establishing illegality adds a further layer of complexity: it requires cross-referencing detected extraction activity against timber management plan authorizations, concession boundaries, and protected area designations — data layers of uneven completeness and accessibility across the Amazon region. This section reviews the principal governmental and non-governmental data sources available for illegal logging monitoring in each of the eight Amazon countries.

Brazil #

Brazil has the most developed institutional and civil society monitoring infrastructure for illegal logging in the Amazon. The federal timber traceability system, SINAFLOR (Sistema Nacional de Controle da Origem dos Produtos Florestais), operated by IBAMA, is the primary legal instrument for tracking the origin of timber products across the production chain. As of 2023, SINAFLOR is operational in 25 of Brazil’s 27 states and records the issuance and utilization of timber transport documents (DOF — Documento de Origem Florestal), linking authorized volumes to specific management plans. In practice, however, SINAFLOR’s effectiveness is significantly undermined by the fraudulent issuance of timber credits: independent audits have documented that valid DOF credits can be used to launder timber extracted illegally from unauthorized areas, including protected areas and indigenous territories, by attaching them to fictitious or exhausted management plans.¹⁶

The most authoritative independent monitoring system for logging in the Brazilian Amazon is SIMEX (Sistema de Monitoramento da Exploração Madeireira), produced by Imazon and published in collaboration with state environmental agencies. SIMEX uses medium-resolution satellite imagery to detect logging infrastructure — roads, log yards, and skid trails — rather than individual tree removal, and then cross-references detected activity against SINAFLOR authorization records to classify areas as authorized, unauthorized, or of uncertain status. The most recent published data estimate that approximately 35% of all active timber extraction areas in the Brazilian Amazon lack valid authorization, with significant geographic variation by state: illegal extraction is particularly concentrated in Pará, where SIMEX documented a 22% year-on-year increase in illegal logging area in its most recent reporting cycle, reaching approximately 28,100 hectares of unlicensed extraction.¹⁷ SIMEX data for Mato Grosso indicate that 46% of the total illegally exploited area in 2024 was located inside Indigenous Lands or Conservation Units, where commercial logging is prohibited regardless of authorization status.¹⁸

Together, SINAFLOR and SIMEX constitute the most complete publicly available national-level data infrastructure for illegal logging in the Amazon. Its principal weakness is the documented vulnerability of the authorization system itself to fraudulent credit generation, which means that even “authorized” extraction areas identified in SINAFLOR cannot be assumed to be fully legal without additional verification.

Figure 7. Simex Amazonia 2020-2030, SIMEX (Sistema de Monitoramento da Exploração Madeireira), produced by Imazon and published in collaboration with state environmental agencies



Peru #

Peru has two principal governmental bodies with mandates relevant to illegal logging monitoring. OSINFOR (Organismo de Supervisión de los Recursos Forestales y de Fauna Silvestre) is the independent forest supervision authority responsible for verifying compliance with forest management plans; its audit activities cover both concessionaires and Indigenous community forests and represent the primary source of official data on detected non-compliance. SERFOR (Servicio Nacional Forestal y de Fauna Silvestre) administers the national forest information system SNIFFS (Sistema Nacional de Información Forestal y de Fauna Silvestre), which is intended to centralize data on concessions, management plans, and timber transport authorizations nationwide. At the time of writing, the SNIFFS geoportal appeared functional but available data on illegal logging was limited, with only two locations presented.¹⁹

Research by the University of Sheffield, applying satellite-based canopy height and texture analysis to cross-reference against OSINFOR inspection records, found that approximately 37% of all sites reporting legal timber harvesting showed canopy damage patterns inconsistent with authorized selective logging, suggesting that reported extraction volumes significantly understate actual offtake.²⁰ Amazon Conservation’s Monitoring of the Andes Amazon Program (MAAP) has demonstrated the technical feasibility of detecting selective logging impacts using VHR imagery (Planet NICFI and Maxar), establishing a methodology applicable in Peru that could be extended to other countries, though it has not yet been applied systematically to produce national coverage for illegal logging as opposed to deforestation by mining.²¹

Colombia #

In Colombia, official monitoring of forest change is conducted by IDEAM through the Sistema de Monitoramiento de Bosques y Carbono (SMByC), which publishes annual deforestation alerts and end-of-year deforestation figures using a combination of Landsat and Sentinel-2 imagery. IDEAM attribution analyses identify illegal logging as contributing approximately 10% of total annual deforestation, though this figure reflects clearance-scale events rather than selective extraction, and the methodology for attributing deforestation drivers is not fully standardized across reporting cycles.²² For selective logging, no systematic national monitoring exists.

In 2021 Colombia developed its National Forest Traceability System (SNTFC) to track timber and other forest products across the entire supply chain by combining tagging, data registration, and digital monitoring tools. It integrates multiple national information platforms such as the National Environmental Information System (SIAC), Forest Information National System (SNIF), Biodiversity Information System (SIB), and traceability systems from the Instituto Colombiano Agropecuario (ICA), enabling authorities to verify the legal origin of forest products and identify actors involved.

Complementary tools further strengthen this ecosystem: the Single Environmental Procedures System (VITAL) centralizes environmental permits and administrative procedures, improving data consistency and traceability from the point of authorization, while the COVIMA app supports field-level verification and reporting of timber flows, enhancing real-time monitoring and enforcement. Key tools like the online transport permit system (SUNL) and the Forest Operations Logbook (LOFL) have improved transparency, generated updated national statistics on timber flows, and strengthened coordination between public agencies and the private sector. However, the system’s effectiveness still depends heavily on uneven institutional capacity among regional environmental authorities (CAR) –particularly in terms of funding, staffing, turnover, and inter-agency coordination.

In Colombia, the Ministry of Environment designed the National Forest Traceability System (SNTF) to monitor the timber product supply chain from origin to final destination. The system records the legal origin of timber and is expected to integrate with tools such as the Single Environmental Procedures System (Ventanilla Integral de Trámites Ambientales, Vital) and other administrative registries. The aim is to facilitate the verification of timber origin and support producer formalization. Its implementation has progressed unevenly across regions and faces limitations in connectivity, resources, and coordination among platforms.²³

This scheme is complemented by the platform Choose Legal Timber (Elija Madera Legal), which connects producers in compliance with regulations to buyers, with the purpose of facilitating the distinction between legal and illegal timber in the Colombian market. Its coverage and effective use have not yet been subject to systematic evaluation.

The broader picture of timber legality in Colombia is documented by a combination of NGO research and official enforcement data. Studies cross-referencing timber transport documentation against satellite-derived forest loss estimates find that approximately 47% of all wood sold commercially in Colombia originates from illegal sources, with the illegal timber trade generating an estimated USD 750 million per year.²⁴ Analysis of official export records for the period 2008–2019 found that approximately 40% of Colombia’s timber exports could not be verified as legally sourced.²⁵

Bolivia #

Bolivia’s forest governance is administered by the ABT (Autoridad de Fiscalización y Control Social de Bosques y Tierra), which issues forest management authorizations and has formal enforcement powers over illegal logging. However, the ABT’s enforcement capacity is primarily administrative: Bolivian law does not provide for criminal sanctions for illegal logging, only for administrative penalties.²⁶ Between 2012 and 2018, ABT seized approximately 110,344 cubic metres of illegal timber, a figure that enforcement officials acknowledge substantially understates total illegal extraction given the size of the country’s forested area and the remoteness of key production zones.²⁷

Ecuador #

In Ecuador, the formal authority for forest management and timber control rests with the MAE (Ministerio de Ambiente y Energía), which administers a timber transport permit system and conducts monitoring of forest change. However, MAE’s monitoring and enforcement capacity in the Amazon region is limited, and no national timber traceability system equivalent to Brazil’s SINAFLOR has been established. Estimates drawn from national enforcement data and NGO investigations suggest that up to 70% of timber extracted from some Amazon provinces is of illegal origin, with an annual illegal timber trade value estimated at approximately USD 100 million.²⁸

Suriname #

In Suriname, the principal implementing body is the SBB (Stichting voor Bosbeheer en Bostoezicht/Foundation for Forest Management and Forest Supervision), a semi-autonomous statutory foundation that administers forest concessions, issues timber harvesting authorizations, and conducts compliance monitoring. In 2019, SBB introduced the SFISS (Suriname Forest Information and Supervision System), a platform addressed to individuals or companies involved in the timber logging industry. As such, only registered companies or people get an account to log in the system, while the registration process requires a previous permit and must be done in person at the physical Office SBB. SFISS provides a digital platform for recording concession boundaries, management plan approvals, and timber volumes, enabling more systematic cross-referencing of extraction activity against authorizations. ²⁹ However, the user has only access to their own information according to what is defined in the acquired permit.

The Gonini system is Suriname’s National Land Monitoring System (SLMS) geoportal to provide transparent, geospatial data on the country’s vast forest cover and is also managed by the SBB. As a core component of the National Forest Monitoring System (NFMS), it integrates satellite imagery, near real-time deforestation alerts, and land-use maps to track changes in the forest landscape. Civil society organizations can leverage this platform to monitor illegal logging by analyzing these open-access deforestation layers to identify hotspots, verifying the legality of timber operations through the Sustainable Forestry Information System (SFISS), and collaborating with Indigenous communities to ground-truth satellite alerts. By synthesizing Gonini’s data with community-led monitoring and statistical insights from the KOPI portal (kopi.sbb.sr), NGOs can build robust, evidence-based cases for advocacy, hold authorities accountable, and support national enforcement efforts against illicit logging activities.

Guyana #

Guyana’s forest sector is regulated by the GFC (Guyana Forestry Commission), which issues timber harvesting licenses, maintains a central database of concession records, and operates a log tagging and tracking system to monitor timber flows from forest to processing facility.³⁰

Venezuela #

Venezuela lacks any functioning national forest monitoring and enforcement system for illegal logging. The institutional collapse of MINEA (the Ministry of Ecosocialism, formerly MARN) has eliminated the institutional capacity for systematic forest governance, and no timber traceability or remote sensing monitoring programme has been operational in recent years.

The particular challenge in Venezuela is that the Arco Minero del Orinoco decree, while primarily directed at mining, has also facilitated the expansion of other extractive activities including logging into areas that were previously more effectively protected. The intersection of state-sanctioned extraction, armed non-state actor control of forest territories, and institutional collapse creates conditions under which any legality assessment framework cannot meaningfully be applied.

3.2 Regional Datasets and conclusion on Amazon-scale data availability #

While for deforestation MapBiomas Amazonia and the UMD forest-loss layers provide harmonised pan-Amazon coverage, selective logging, which removes individual trees while leaving the canopy largely intact, is not represented in supra-national datasets.

Taken together with the national-scale review in Section 3.1, three tiers emerge across the eight Amazon countries. Brazil operates the most developed national infrastructure (SINAFLOR combined with Imazon’s SIMEX), albeit undermined by documented fraud in the issuance of timber credits that weakens the “authorised” classification at its source. Guyana and Suriname, and Colombia to some extent, have functional concession-management and log-tracking systems — SFISS in Suriname, the GFC’s log-tagging system in Guyana — but do not publish independent detection layers that can be cross-referenced against them at national scale. Peru, Colombia, Bolivia, Ecuador and Venezuela lack systematic, publicly available data on selective-logging extraction and its legality status: the most granular available evidence in these countries comes from audit samples (OSINFOR’s inspections in Peru), trade-flow comparisons (the 47% illegal-share estimate for Colombia; the 40% unverifiable share of export records for 2008–2019), and one-off research studies (the Sheffield canopy-disturbance analysis in Peru).

Thus, a consistent Amazon-wide measure of illegal logging is not achievable from existing data, and closing the gap is a longer-horizon undertaking than for the other manifestations reviewed in this report. It would require two parallel investments: scaling a SIMEX-type logging-infrastructure detection methodology (roads, log yards, skid trails) to the remaining seven countries, and building interoperable concession and authorisation registries of coverage comparable to Brazil’s SINAFLOR in each.

In the interim, comparability between countries could come from country-level estimates of the illegal share of extraction or through proxy indicators such as GFW-detected road expansion and canopy disturbance inside concessions.

4. Illegal Artisanal and Small-Scale Gold Mining #

4.1 Review of Available Data from Governmental and Non-Governmental Sources #

Monitoring illegal artisanal and small-scale gold mining (ASGM) presents a distinct set of challenges compared to fully land-based manifestations of Nature crime. While land-based mining scars can be detected through satellite imagery, much of the activity occurs on or close to river banks, which are naturally dynamic in the Amazon and complicate the interpretation of remote sensing evidence. Establishing illegality requires cross-referencing detected activity against mining concession records, protected area boundaries, and Indigenous territory designations, for which consistent and updated data across all 8 countries is difficult to obtain. Fully river-based, dredge mining adds a further layer of complexity, as its mobile, small-footprint often falls below the detection threshold of medium-resolution satellites (see below in Section 6). This section reviews the principal data sources available at national and regional level for land-based ASGM monitoring, organized by country and then by regional platform.

Brazil #

Brazil has the most developed monitoring infrastructure for ASGM in the Amazon. The BrasilMAIS detection platform, while only available to Brazilian civil servants, disaggregates between deforestation alerts by driver, including gold mining. MapBiomas Brasil’s dedicated “mining” class shows that small-scale mining area more than doubled between 2012 and 2022, reaching approximately 264,000 hectares before declining by 35% in 2023 and a further 45% relative to the peak in 2024.³¹ Legality assessment indicates that in 2022, 77% of detected garimpos showed explicit signs of illegality, based on location outside valid concession boundaries or within protected areas.³²

Near-real-time monitoring by Amazon Mining Watch (AMW) points to a more complex picture beneath that aggregate trend. In the fourth quarter of 2025 alone, Brazil registered the largest new mining expansion of any Amazonian country at approximately 2,000 hectares.³³ AMW’s resurgence analysis — which flags areas where mining had previously abated but has recently reappeared — identified new mining scars in two protected areas in Mato Grosso: the Igarapés do Juruena State Park, where 2.83 hectares appeared after a three-quarter pause, and the Aripuanã Indigenous Territory, where nearly 60 hectares were recorded in a single quarter.³⁴ These case-level detections illustrate what aggregate annual data can obscure: enforcement suppresses mining in the short term, but without sustained pressure, criminal networks return — often to the same sites.

Peru #

Conservación Amazónica – ACCA implements the Monitoring of the Andes Amazon Program (MAAP), in Peru, where it has produced the most comprehensive country-level dataset on gold mining deforestation in the country. Its most recent report (MAAP #233, February 2026) documents gold mining activity across all nine affected regions of the Peruvian Amazon (Amazonas, Cajamarca, Cusco, Huánuco, Loreto, Madre de Dios, Pasco, Puno, and Ucayali), covering both land-based and river-based operations. Cumulative gold mining deforestation in Peru reached 139,169 hectares by mid-2025, with 97.5% of this total concentrated in the Madre de Dios region.

MAAP’s Peru data represents the most spatially complete and methodologically documented national ASGM dataset in the Amazon, and its multi-source approach is explicitly designed to bridge the gap between satellite detection and ground truth.

Figure 8. National-scale map of ASGM sites (land and river-based) in Peru



Ecuador #

MAAP also conducts systematic monitoring of gold mining in Ecuador in partnership with Fundación Ecocíencia, with particular focus on the Andes-Amazon transition zone where mining activity has expanded rapidly in recent years. Reports cover both the northern and southern sectors of the Ecuadorian Amazon.

In the northern sector (Punino River Basin), MAAP’s MAAP #227 documents cumulative mining deforestation of 1,422 hectares since 2019, with an increase of 420 hectares in 2024 alone. Critically, approximately 90% of this mining deforestation is likely illegal, occurring outside the boundaries of authorized mining areas. In the southern sector (Morona Santiago Province), MAAP #238 documents 856 hectares of affected area by 2024 — a doubling of the impacted area in just four years. Illegal mining has also penetrated formally protected areas: in Sumaco Napo-Galeras National Park, mining deforestation expanded by 125% between July 2023 and September 2024, from 22 to 50 hectares.

Colombia #

In Colombia, the Fundación para la Conservación y el Desarrollo Sostenible (FCDS) is the principal civil society organization conducting field-validated mining monitoring in the Colombian Amazon. FCDS works in close collaboration with MAAP, combining overflight surveys with satellite imagery analysis to document mining activity in remote river corridors.

MAAP #228 reports 29 dredges operating on the Puré River in November 2024 and 27 in March–April 2025, entirely within the Puré National Park, where no mining is permitted. Low-altitude overflights conducted by FCDS in September 2024 provided additional detail on mining methods and infrastructure not visible in satellite imagery. Research drawing on FCDS data and social cartography methodology documented a 1.8% loss of forest cover (990 hectares) in conservation areas between 2020 and 2023, with 70% of forest loss hotspots coinciding with new mining openings. FCDS has also collaborated with Conservation Strategy Fund on applying the Mining Impacts Calculator to Colombian territories, producing socioeconomic cost estimates for affected Indigenous communities.

Bolivia #

Bolivia’s ASGM monitoring is less institutionally developed than that of Brazil, Peru, or Colombia. River-based dredge monitoring on the Bolivian Madre de Dios is being developed by Conservación Amazónica – ACEAA, which has built a barge-tracking workflow based on Planet imagery and manual digitization, producing monthly monitoring reports used as evidence in coordination with the Fiscalia Departamental de Pando. Land-based ASGM in Bolivia is not yet subject to systematic civil society monitoring equivalent to that available in Brazil, Peru, or Ecuador.

Suriname #

For Suriname, the most comprehensive analysis of gold mining deforestation has been produced through a collaboration between the Amazon Conservation Team (ACT) and MAAP, combining 24 years of satellite data with Amazon Mining Watch’s near-real-time AI detection results (MAAP #237). The analysis finds that gold mining has impacted approximately 92,000 hectares of forest in Suriname over the past 24 years, with roughly 25,000 hectares — 28% of the cumulative total — occurring in the four-year period from 2021 to 2024, indicating rapid recent acceleration. Protected areas are increasingly affected: mining deforestation within Brownsberg Nature Park reached 1,274 hectares (8.8% of the park’s area) between 2001 and 2024, and a new mining front appeared inside Brinckheuvel Nature Reserve in March 2025. A confidential version of this analysis was shared directly with the Surinamese government to inform policy discussions.

Guyana #

Dedicated country-level civil society monitoring of ASGM for the country is not available, and data for the country is limited to regional datasets (see below Section 5.1.2)

Venezuela #

Venezuela presents the most complex monitoring environment in the Amazon due to the institutional collapse of its environmental agencies and the active role of the state in promoting extractive activities within the Arco Minero del Orinoco — a decree-defined zone covering approximately 12% of the country’s territory. In this context, SOS Orinoco, a Venezuelan civil society organization, has become the principal source of documented evidence on mining impacts.

SOS Orinoco’s monitoring relies on satellite imagery analysis combined with field reports from affected Indigenous communities. Its data documents sustained and accelerating environmental destruction: within Canaima National Park (a UNESCO World Heritage Site), mined areas grew by more than 1,300% between 2000 and 2023, from 122 to 1,582 hectares, with a further 73 hectares added in 2024 ³⁵. In the four years following the Arco Minero decree (2016), over 2,821 km² of forest were destroyed, half of it in protected territories.³⁶

Regional Datasets #

Amazon Mining Watch (AMW). Amazon Mining Watch is a partnership between Earth Genome, the Pulitzer Center’s Rainforest Investigations Network, and Amazon Conservation. It is the first platform to provide near-real-time, AI-powered detection of land-based gold mining deforestation across all eight Amazonian countries. The platform updates quarterly, enabling systematic monitoring of new mining fronts. In 2025, it detected 37,109 hectares of recent gold mining expansion across the Amazon, with Brazil (15,538 ha), Peru (6,511 ha), and Guyana (4,942 ha) recording the largest areas, and Venezuela, Suriname, Ecuador, and Bolivia each exceeding 2,000 hectares (MAAP #235).

Accuracy assessments are ongoing to refine the model based on known high-accuracy national datasets such as those maintained by Conservación Amazónica – ACCA in Peru and Fundación Ecociencia in Ecuador, and revised data for the entire annual series from 2018 to 2025 is set to be published in 2026. Regular updates alerting of new incursions into Protected Areas and Indigenous Territories previously untouched by mining, cases of resurgence of mining in areas where mining had stopped, and areas with persistent and accelerating mining deforestation will continue to be provided in 2026.³⁷

A key feature of AMW is its legality assessment layer, which cross-references detected mining areas against protected area and Indigenous territory boundaries and, in five countries (Bolivia, Brazil, Colombia, Ecuador, and Peru), against mining concession records to produce a presumption-of-illegality classification.³⁸

According to this analysis, Colombia and Brazil are the countries with the highest rates of illegality in ASGM, followed by Ecuador and Peru (See Table 5). Taken together, MapBiomas’ mining class and the SOS Orinoco, MAAP and WWF/ACTO datasets constitute the most complete available baseline of the extent and distribution of illegal ASGM across the Amazon. Maintaining these datasets on an annual cycle and consolidating them into a common time series is essential for assessing the robustness of policy efforts to curb illegal ASGM, tracking enforcement strength, and monitoring displacement effects across the basin.

Table 5. Presumption of illegality of ASGM in 5 countries of the Amazon, in % of total area (Source: Amazon Mining Watch)

Presumption of illegality:	Very High Activity is occurring without permit and within a protected area that doesn’t allow for any kind of resource exploitation	High Activity is occurring outside of any explicit concession for doing so	Medium Activity is happening within a concession, but active status of the concession could not be verified.	Low Activity is happening within an active concession (assuming compliance with all applicable exploration requirements)	Indices of illegality – all categories
Bolivia	15.3%	2.6%	1.4%	80.8%	19.2%
Brazil	2.1%	56.7%	20.0%	21.2%	78.8%
Colombia	7.0%	33.7%	53.6%	5.7%	94.3%
Ecuador	1.1%	38.7%	21.2%	39.0%	61%
Peru	1.0%	23.5%	29.2%	46.3%	53.7%

Figure 9. Graph representing the presumption of illegality in 5 countries of the Amazon in % of total area.



MapBiomas Amazonia. MapBiomas Amazonia produces annual land cover and land use maps at 30-meter resolution from 1985 to 2023 across all eight Amazon countries. The platform includes a dedicated “mining” class that covers both artisanal and industrial operations, The most recent year available is 2023 and the data cannot be filtered for ASGM, limiting its usefulness for an assessment of recent ASGM extent and distribution across the Amazon, and legality analysis against known ASGM concessions.

WWF also partnered with ACTO on a basin-wide study that identified more than 4,000 illegal mining sites across all nine Amazonian countries — one of the first attempts at a comprehensive Amazon-wide illegal mining count.³⁹ The study is from 2023 and has not been repeated since.

RAISG. The Amazon Network of Georeferenced Socio-environmental Information (RAISG) maintains the most comprehensive pan-Amazonian spatial dataset on protected areas, Indigenous territories, and mining concessions, an essential input to legality analysis of known ASGM extent. The network also produced a layer of illegal ASGM in 2020, based on a combination of mining concessions and MapBiomas data on mining areas. At the time of writing, this Amazon-wide layer was in the process of being updated based on different data sources, including MapBiomas and Amazon Mining Watch, with a view to provide authoritative data on the issue.

4.2 Challenges in consistent detection of river-based mining #

River-based artisanal and small-scale gold mining (ASGM) — using dredge platforms anchored directly on Amazonian rivers — has emerged as one of the most rapidly expanding forms of Nature crime in the basin.

Unlike land-based ASGM, which leaves visible scars on forest cover detectable through satellite-based deforestation monitoring, river-based dredge mining presents a fundamentally different detection challenge. The physical footprint of a mining dredge — a floating barge or raft operating on a watercourse — may be as small as a few square meters and, critically, is mobile. Dredges move along river systems daily, following seasonal hydrological patterns, relocating in response to gold concentrations, or dispersing in the wake of enforcement operations.

These structural monitoring gaps carry direct implications for enforcement. The return of approximately 130 dredges to the Madeira River just five months after a major enforcement operation in August 2024⁴⁰ illustrates both the resilience of criminal networks and the limits of enforcement cycles not grounded in continuous monitoring.

In January 2026, Amazon Conservation convened a workshop in Bogotá with representatives of civil society organizations across seven countries of the Amazon⁴¹ to discuss, among other topics, the availability of consistent monitoring data on river-based mining and potential ways forward. The discussion identified persistent monitoring gaps, available technologies, and priority investments for a regional monitoring architecture.

The workshop confirmed four persistent monitoring gaps:

• ​Dredge mobility and counting methodology: near-daily monitoring cadence is needed for reliable dredge counts, but no standard methodology exists to reduce counting error for mobile objects, and recording practices remain inconsistent across organizations and jurisdictions.

• ​Image resolution and satellite tasking: VHR optical imagery (sub-0.5 m) is essential for smaller dredges but expensive and dependent on advance tasking — which presupposes prior intelligence about mining locations, systematically excluding undocumented fronts.

• ​Seasonal hydrology: dry-season operations cluster around exposed sandbars while wet-season activity disperses into floodplains, requiring monitoring protocols adapted to shifting river morphology that are not yet standardized at regional level.

• ​Cross-border displacement: dredging networks routinely cross jurisdictional boundaries to evade enforcement — a dynamic documented in the Yanomami Indigenous Territory, where miners displaced by Brazilian operations migrated toward Venezuelan and Surinamese river corridors — yet cross-border monitoring coordination remains absent.

Participants also identified a number of available and emerging detection technologies:

• Synthetic Aperture Radar (SAR) imagery — notably the European Space Agency (ESA)’s Sentinel-1 constellation — represents a significant technical advance for dredge detection. SAR penetrates cloud cover and operates day or night, addressing two critical constraints on optical monitoring in the humid tropics. U.S. Geological Survey research demonstrated that semi-automated detection of riverine dredges using Sentinel-1 SAR is technically feasible on the Madeira River, with a local detection method offering the best balance between sensitivity and precision; the strong radar backscatter of metallic dredge structures on water allows them to be extracted using statistical thresholds adapted from ocean vessel detection.⁴²

• Automated object detection algorithms applied to satellite imagery offer a scalable complement. Amazon Mining Watch’s AI-powered quarterly detection of land-based mining scars has demonstrated regional-scale feasibility;⁴³ A comparable approach for river-based dredges requires labelled training datasets across diverse river environments and seasons, identified by workshop participants as a high-priority technical investment.

• Drone and light aircraft overflights provide high-resolution ground-truth for satellite analyses and are well-established among several participating organizations. Their limitation is scalability: they are resource-intensive, cannot be deployed continuously, and carry security risks in areas controlled by criminal networks.

• Mobile applications allow community members and field monitors to log dredge observations with GPS coordinates in near-real-time — valuable for river segments inaccessible to aerial or satellite observation. A key constraint is interoperability: current apps used by different organizations cannot share data, preventing regional aggregation of what are individually substantial national datasets.

• Among newer satellite capabilities, Planet’s Pelican constellation — capable of 0.3-m VHR optical imaging with onboard AI object detection —⁴⁴ was identified as potentially transformative if access costs can be reduced for civil society organizations. Planet’s Tanager hyperspectral constellation (400–2,500 nm)⁴⁵ offers additional potential for detecting the spectral signatures of mercury contamination and suspended sediment plumes associated with dredge operations.⁴⁶ ESA’s forthcoming BIOMASS mission and thermal infrared detection — exploiting the heat signatures of diesel engines powering dredges — were also flagged as candidate complementary approaches.

• River sedimentation and water turbidity monitoring represents a methodologically distinct approach that can serve as an indirect proxy for dredge activity. Research has shown that artisanal gold mining causes measurable, satellite-detectable changes in riverine suspended sediment concentrations, including seasonal inversions distinguishable from natural variability.⁴⁷ While this approach does not identify individual dredges, it can flag disturbed river segments and guide targeted VHR tasking. Combined with nighttime light detection — useful where dredges operate after dark to evade daytime coverage — sedimentation screening could serve as a first-filter layer in a multi-source monitoring architecture.

The workshop converged on four priority investments for moving from fragmented, nationally-specific monitoring toward a consistent Amazon-wide baseline:

• ​Standardized reporting framework: agreed definitions of dredge types and operational status, common georeferencing protocols, and interoperable data formats allowing field observations, drone surveys, and satellite analyses to feed a single regional monitoring layer.

• Security protocol with standardized program guidelines about public appearances and comments about mining detection work, which can generate threats.

• ​Regional geodatabase: a rolling, multi-source database of river-based mining activity accessible to all contributing organizations, governed by information security protocols to prevent mining location data from being exploited by criminal networks.

• ​Dredge movement analysis: systematic modeling of daily and seasonal displacement ranges by river type, with explicit cross-border tracking for frontier river systems shared by two or more countries.

• ​Technical training: investment in satellite image analysis (optical and SAR), mobile app deployment, and standardized data protocols to broaden the network of national and subnational organizations contributing to a common monitoring system.

  4.3 Assessment: Coverage Gaps and Implications for a Regional Baseline #

Taken together, the national and regional data sources reviewed above represent a substantially more developed monitoring landscape for ASGM than existed five years ago, driven largely by civil society investment in satellite-based detection and the launch of Amazon Mining Watch as a regional platform. However, significant gaps remain.

At national level, systematic monitoring coverage is concentrated in Brazil and Peru, with increasingly credible civil society datasets emerging for Ecuador, Colombia, and Suriname. Bolivia lacks a land-based ASGM monitoring system equivalent to those in neighboring countries. Venezuela is monitored almost exclusively by SOS Orinoco under conditions of severely constrained access, and the state’s active promotion of the Arco Minero makes standard legality assessment frameworks difficult to apply. Guyana has no dedicated civil society monitoring capacity for ASGM despite being identified as a high-impact country by regional satellite data.

At regional level, Amazon Mining Watch provides the most promising foundation for a consistent pan-Amazonian ASGM baseline, combining broad coverage, near-real-time frequency, and a nascent legality assessment layer. While useful to assess overall trends at large scales and detect new incursions given the high frequency of updates, the platform is not accurate or high-resolution enough to be used as evidence in reports to authorities, showing the complementarity with national systems maintained by civil society organizations with strong local bases and direct rapport with enforcement agencies.

With regards to river-based mining the priorities identified at the Bogota workshop shed light on the challenge of monitoring this fast-growing and destructive manifestation of Nature crime. Translating that expertise into a formal regional monitoring architecture will require substantial institutional support and deliberate mechanisms for integrating civil society monitoring data into governmental enforcement and prosecutorial workflows.

5. Mercury pollution #

Mercury pollution from artisanal and small-scale gold mining (ASGM) represents one of the most pervasive and least visible manifestations of Nature crime in the Amazon. Unlike deforestation — which leaves a permanent, satellite-detectable scar — mercury enters river systems as an invisible contaminant, cycling through water, sediment, and living organisms, across national boundaries and with no regard for jurisdictional divides. Estimates of annual mercury releases from ASGM in the Amazon range from 200 to over 500 tonnes, with cumulative releases over four decades potentially exceeding 8,000 metric tonnes⁴⁸.

5.1 Available datasets #

Existing datasets on mercury in the Amazon can be grouped into three categories: peer-reviewed scientific studies, institutional monitoring programs, and regulatory national inventories. Taken together, they constitute a substantial but geographically uneven and methodologically inconsistent body of evidence.

Peer-reviewed datasets provide the most rigorous available data but are heavily concentrated in Brazil and, to a lesser extent, Peru. A 2025 meta-analysis of mercury contamination across social groups and environmental contexts in the Amazon basin identified 25 peer-reviewed studies covering Brazil, 13 covering Peru, and 6 covering Colombia; Ecuador and Venezuela had no representation in that review, and the available literature more broadly contains little published data from Guyana, Suriname, and Bolivia.⁴⁹ The most spatially comprehensive scientific dataset is the SERAFM probabilistic model, applied to 8,259 sub-basins across the Branco, Tapajós, and Xingu river systems, which generates projected mercury distribution and fish bioaccumulation estimates at sub-basin resolution.⁵⁰

Mercury bioaccumulates through the food chain, reaching its highest concentrations in large, long-lived carnivorous fish. Mean mercury concentrations in fish across key basins illustrate the scale of contamination: 0.47 μg/g in the Madeira, 0.69 μg/g in the Tapajós, and 0.72 μg/g in the Madre de Dios (Peru) — the latter two substantially above the WHO guideline of 0.5 μg/g for most species.⁵¹

A risk assessment covering the Branco and Tapajós basins found that approximately 49.8 percent of the assessed population faces extremely high mercury exposure risk through fish consumption.⁵² In the city of Santarém, at the confluence of the Tapajós and the Amazon, 75 percent of residents show elevated blood mercury levels, with some carrying four times the WHO safety threshold.⁵³

At the institutional level, the most significant ongoing effort to map mercury contamination is the regional mercury overview study coordinated by the Amazon Cooperation Treaty Organization (OTCA/ACTO) through its Amazon Basin Project.⁵⁴ Covering all eight Amazon countries, this study applies the UNEP Mercury Inventory Toolkit to develop a standardized, multi-country assessment of mercury sources and contamination levels — the first systematic attempt at a truly Amazon-wide baseline.⁵⁵ OTCA’s Amazon Regional Observatory (ORA) is developing an integrated regional information platform that will aggregate environmental and social data across member countries, including water quality indicators relevant to mercury monitoring.

In addition, PAHO/WHO tracks health-side mercury exposure through national health systems and signed a dedicated cooperation agreement with OTCA in October 2024 covering Indigenous health, climate-related health risks, and access to health services in the Amazon region.⁵⁶

​Finally, the World Bank’s Amazon Sustainable Landscapes Program and Alianza Amazonica para la Reduccion de los Impactos de la Mineria de Oro (AARIMO) bring together Bolivia, Brazil, Colombia, Ecuador, Guyana, Peru, and Suriname around shared evidence generation and multi-country dialogue on gold mining impacts, serving as a practical platform for translating scientific data into policy recommendations.⁵⁷

National regulatory inventories under the Minamata Convention on Mercury provide a third layer of data. All Amazon countries, with the notable exception of Venezuela, are parties to the Convention and have ratified it. The convention requires any country where ASGM is “more than insignificant” to develop a National Action Plan (NAP) containing national mercury inventories, sector-specific emission estimates, and reduction targets.⁵⁸ In practice, however, the quality, completeness, and methodological consistency of these NAPs vary considerably across the region. They represent a floor of nationally self-reported data rather than an independently verified regional baseline, and their utility for cross-country comparison is limited by differing scope and update cycles.

Amazon Conservation’s Amazon Targets Tracker platform monitors the implementation status of several international conventions across the 8 countries of the Amazon, including the Minamata Convention. This simple criteria-based analysis reveals large disparities, with Brazil, Bolivia and Ecuador having yet to present a NAP, despite significant ASGM activity on their territory. Both initial assessments and National Action Plans may provide useful data for aggregation into a regional baseline, but remain unavailable for 2 countries, including Brazil.

Table X. Status of implementation of the Minamata Convention across countries of the Amazon

The following table summarizes the implementation status of the Minamata Convention across eight Amazonian countries. The assessment is based on six criteria: Party Status, Ratification, Initial Assessment, Action Plan, Forest Targets, and Amazon Targets. The full results can be found on the Amazon Targets Tracker.

Country	Total Score out of 6 criteria	Justification / Implementation Details
Colombia	6/6	Full implementation. Ratified on 26/08/2019. Initial assessment completed in 2017. Action plan established in 2024. Both forest-specific and Amazon-specific targets are available.
Guyana	5/6	Full implementation. Ratified on 24/09/2014. Initial assessment completed in 2016. Action plan established in 2021. Forest targets are available. No Amazon-specific target is available but the whole country is part of the Amazon.
Peru	4/6	Advanced implementation. Ratified on 21/01/2016. Initial assessment completed in 2020. Action plan established in 2019. Forest-specific targets and Amazon-specific targets are not specified.
Suriname	4/6	Advanced implementation. Ratification date N/A. Initial assessment completed in 2020. Action plan established in 2024. Forest-specific targets and Amazon-specific targets are not specified.
Bolivia	3/6	Moderate progress. Ratified on 26/01/2016. Initial assessment completed in 2017. Action plan, forest targets, and Amazon targets are not specified.
Ecuador	3/6	Moderate progress. Ratified on 29/07/2016. Initial assessment not available. Action plan established in 2020. Forest targets and Amazon targets are not specified.
Brazil	2/6	Limited progress. Ratified on 08/08/17. Initial assessment, action plan, forest targets, and Amazon targets are not specified.
Venezuela	0/6	No implementation. Not a party to the convention. Ratification, initial assessment, action plan, forest targets, and Amazon targets are all not available.

On the civil society and research side, Brazil’s FIOCRUZ has produced extensive community-level exposure data. This data is published, alongside other sources, on the Observatorio do Mercúrio na Amazônia, an online tool developed by WWF-Brazil, FIOCRUZ and Cincia. The Mercury Observatory is an online platform that compiles data on mercury contamination across the Pan-Amazon region, launched to monitor the impacts of illegal gold mining. It georeferences all available studies on mercury emissions, fish contamination and risks to human health, highlighting high levels of contamination in Indigenous and riverine communities. ⁵⁹

Figure 10. Mercury-Observatory portal, an online platform that compiles data on mercury contamination across the Pan-Amazon region (2026).

5.2 Conclusion on potential for Amazon-wide, consistent monitoring #

The data gap remains significant on mercury pollution. No systematic, long-term monitoring network exists on the basin’s major transboundary rivers that would enable consistent tracking of mercury flux across national borders and no integrated database currently exists that combines data on pollution through water-column mercury measurements and sediment concentrations, as well as contamination through fish tissue data, human blood or hair mercury levels, across all eight Amazon countries using consistent methodologies. ⁶⁰

The institutional foundations for such a system are partly in place: the OTCA regional overview study will establish the first shared multi-country methodological baseline⁶¹ and Minamata Convention NAPs, if strengthened with harmonized reporting requirements, could anchor national contributions to a regional platform. The Conservation Strategy Fund, with support from the World Bank’s Amazon Sustainable Landscapes program, has developed the Gold Mining Impact Calculator — a tool that estimates socio-environmental costs of mining damage, including from mercury contamination.⁶² This tool has been used by Brazilian prosecutors to estimate damages from illegal ASGM operations and set fines or reparations, and could pave the way for an Amazon-wide assessment of the social and environmental cost of Mercury pollution as a Nature crime, building on the example provided in 2.2.3 with illegal deforestation in the Brazilian Amazon.

6. Violence against environmental defenders #

Environmental defenders play an important role in monitoring environmental conditions and reporting illegal activities in the region. Retaliation against them has often been associated with the expansion of environmental crimes, including those perpetrated by organized criminal groups.

The connection between environmental protection and public safety has been increasingly highlighted by different authorities in the region. Notably, in June 2025, the Amazon Cooperation Treaty Organization (ACTO) established the Special Commission on Public Security and Transnational Illicit Activities in the Amazon Region (CESPIT), dedicated to strengthening regional safety and cooperation against transnational crimes affecting the integrity of the Amazon. The move is part of OTCA’s wider security agenda, which explicitly links the protection of the Amazon to the defense of human rights and the safety of the region’s populations.⁶³

Improved monitoring of attacks against defenders allows governments, local and crossborder authorities to understand the pressures associated with environmental crimes and to support evidence-based policy interventions to both protect the biome and those who defend it. Such monitoring also helps identify how different groups of defenders (including women, indigenous peoples and children) may be disproportionately affected.

6.1 Data availability #

Data for this type of manifestation of Nature crime typically originates from government agency or civil society records, often including GPS coordinates or jurisdiction names that allow mapping. Illegality is treated as inherent to the act — which is also likely to be directly connected to the illegal extractive activities the defender was reporting or resisting. The principal analytical challenge in this category is not establishing that an act was unlawful but confirming that it was specifically motivated by the victim’s environmental work: a determination that requires careful, case-by-case qualification of each recorded incident.

This qualification and the compilation of cases of violence against environmental defenders is largely developed and organized by civil society organizations. The table below outlines the datasets available in each country and organizations that maintain them:

Country	Local coverage
Bolivia	Medium	Locally, non-profit organizations, Centro de Documentación y Información Bolívia (CEDIB) and Coordinadora Nacional de Defensores de Territorios Indigenas Originarios, Campesinos y Áreas Protegidas (CONTIOCAP), maintain a dataset documenting events from 2017 to present. Although the dataset provides detailed information regarding each recorded incident, the data is not publicly available in open format.
Brazil	Medium	The country has relatively well-established initiatives dedicated to collecting data about violations against human rights defenders in general, with indirect reference to environmental defenders. Comissão Pastoral da Terra (CPT), a Catholic-affiliated civil-society organization that monitors rural land conflicts, has been gathering historic data on rural violence since the 1980s. Conselho Missionário Indígena (CIMI), an Indigenous rights organization also linked to the Catholic Church, has focused on documenting cases of violence against Indigenous peoples since the 1990s. While both datasets provide a similar level of detail, in the case of CIMI the absence of data in an open, machine‑readable format limited its potential for further analysis or compilation.
Colombia	Medium	The country has relatively well-established initiatives dedicated to collecting data about violations against human rights defenders in general, with indirect reference to environmental defenders. Historic data mostly concern conflicts involving armed groups. One non-profit organization, Instituto de Estudios para el Desarrollo y la Paz (INDEPAZ), reports on the assassinations of social leaders, including a specific category for environmental defenders, from 2016 to present. In some of its publications, INDEPAZ classifies additional groups under the broader umbrella of environmental defenders (such as local leaders and indigenous groups). As a result, it creates a potential overlap between categories and limits the comparability of the data from a regional perspective.
Ecuador	Low	Country-specific data is limited. Local initiatives are not systematic or comprehensive, and focus on qualitative analysis but lack systematic quantitative datasets.
Peru	Medium	Locally, two organizations independently document cases: Instituto del Bien Común (IBC) and Coordinadora Nacional de Derechos Humanos (CNDDHH). CNDDHH, an institution comprised of several human rights organizations, documents cases since 2002 and has a specific category for environmental and Indigenous rights defenders. IBC, on the other hand, provides information on broader trends and aggregated data starting in 2000. Similarly, the Peruvian Ombudsman’s Office publishes information on broader numbers rather than providing details on each specific case. Both sources provide aggregated data and do not detail information about each incident.
Guyana & Suriname	Low	The countries lack country-specific, quantitative data on violence against environmental defenders. Research for these countries largely relies on the information compiled by global organizations.
Venezuela	Low	The country faces significant limitations in the availability of data. Local non-profit groups, namely SOS Orinoco, VE360 and Infove360, have documented cases since at least 2013, though they are not presented in detail, with figures being presented broadly, without individual details.

In addition to these locally-managed datasets, Global Witness and Tierra de Resistentes maintain global datasets whose coverage include Amazon countries.

Global Witness defines land and environmental defenders as people who take a stand and carry out peaceful action against unjust, discriminatory, corrupt or damaging exploitation of natural resources or the environment. The dataset documents attacks where there is a reasonable and suspected link to an individual’s activism. The database also includes forced disappearances of land and environmental defenders where the individual remains missing after at least six months.

Tierra de Resistentes similarly documents episodes of violence against land and environmental defenders, understood as individuals or groups who strive to protect and promote human rights related to the environment. Covering the period between 2009 and 2018, the database was built through information requests, desktop research and on-the-ground reporting, ensuring that information provided by different sources was adequately corroborated.

Amazon Conservation compiled both datasets, in addition to Comissao Pastoral da Terra’s for Brazil – the only national dataset readily available for data analysis – and reconciled records to remove duplicates and standardize categories used by each database. The result, presented in more detail in the next section, is the first known attempt at a pan-Amazonian, consolidated and spatially explicit record of violence against environmental defenders.

6.2 A consolidated Amazon-wide database of violence against environmental defenders #

The data presented in this section examines incidents of violence against environmental defenders within the Amazon bioregion, from three data sources spanning 2016-2024. The data reveals concentrated geographic hotspots, predominantly lethal violence driven by land disputes, and disproportionate impacts on Indigenous Peoples and small-scale rural producers.

The spatial clustering observed in the data reflects the geography of resource extraction pressures, particularly in frontier zones where agribusiness expansion, illegal mining, and logging operations intersect with Indigenous territories and protected areas. The spatial pattern observed confirms that violence functions as a deliberate mechanism to facilitate resource extraction and land appropriation, targeting those who resist these processes. Unsurprisingly, land disputes account for 333 incidents (58%), making it by far the dominant driver of violence, before forest, mining and water-related disputes.

The dataset documents 574 incidents across eight categories of violence, with murder representing the most severe form (346 incidents, 60% of total), followed by threats (149 incidents, 26%), direct attacks (36 incidents, 6%), legal harassment (14 incidents, 2%), disappearance (9 incidents, 2%), criminalization/stigmatization (9 incidents, 2%), displacement (3 incidents, 1%), and other forms (8 incidents, 1%). The dominance of murder as a documented form of violence reflects the extreme danger faced by environmental defenders in the Amazon. However, it is important to note that threats—the second most common form—likely represent only a fraction of actual intimidation and harassment incidents, as many incidents go unreported or unrecorded. This suggests that the true scope of violence against defenders is substantially larger than the documented figures indicate.

The violence documented in this dataset disproportionately affects specific communities, with Indigenous Peoples representing 227 incidents (40%) of the total, followed by small-scale rural producers (182 incidents, 32%). Indigenous territories have been recognized as valuable figures for Amazon conservation but this role comes at a cost, with Indigenous leaders and communities facing systematic violence from those seeking to exploit the resources within their territories.

Figure 11. Record of violence against environmental defenders across the Amazon. Source: authors own analysis, based on consolidation of data from Global Witness, Tierra de Resistentes and Commissão Pastoral da Terra.
This data is also available as a dynamic geoportal, which allows filtering by categories and consultation of the details of specific data points





6.3 Conclusion on potential for Amazon-wide consistent data collection #

Data availability varies across jurisdictions as the report of incidents is contingent on other factors. For instance, countries that face heightened levels of violence or are subject to a biased judicial system may record a smaller number of offenses given the uncertainty that filing a complaint represents for the victims. Underreporting may also stem from difficulty of data verification and the remoteness of many attacks. Data collection initiatives thus usually focus on assassinations, as these are more easily identified and verified, with non-lethal attacks, threats and harassment often underreported.

Data is also limited by time-related constraints, and historic information rarely extends beyond the 2000s. Other challenges include restraints associated with data accessibility, as information is often published in non-machine-readable formats, such as PDF or Word files, which limits the potential for large scale data analysis and cross-referencing.

The data architecture needed for a sustained regional layer on violence against defenders is lighter than for the other manifestations reviewed in this report: there is no remote-sensing barrier, and the consolidated 574-incident database already functions as a working prototype. What would be required is a shared case-definition protocol adopted across participating civil-society organizations — an effort that could build on the Escazú Agreement’s 2025–2026 Action Plan⁶⁴ commitments on defender monitoring and on ACTO’s CESPIT — together with sustained resourcing for the verification and reconciliation work currently carried out on a project basis.

7. Conclusion #

The five manifestations of Nature crime reviewed in this report — illegal deforestation, illegal logging, illegal artisanal and small-scale gold mining, mercury pollution, and violence against environmental defenders — share an underlying physical geography and a common set of criminal networks, but they are tracked today by largely separate data systems, on different cadences, with uneven country coverage.

This fragmentation is the single most important finding of this technical assessment. It means that even where individual manifestations can be credibly documented — illegal deforestation in the Brazilian Amazon at roughly 90% of total clearance (Section 2.2.1), gold mining in Peru’s Madre de Dios at 139,169 cumulative hectares (Section 4.1), a consolidated record of 574 incidents of violence against defenders (Section 6.2) — the aggregate scale and spatial distribution of Nature crime across the basin cannot be read from any single source, official or otherwise.

Four patterns emerge from the review and point to what must change. #

While the five manifestations share an underlying geography and a common set of transnational criminal networks, they are tracked by largely separate systems, on different timescales and frequencies, with coverage that reflects institutional capacity rather than the severity of the problem. Four patterns emerge consistently from the evidence:

The monitoring map is the inverse of the crime map. Brazil and Peru — the two countries with the most developed monitoring infrastructure — are not the countries where Nature crime is most out of control; they are the countries where sustained civil society and institutional investment has made the problem legible. Venezuela, where environmental institutions have effectively ceased to function since 2011, is simultaneously the country with the most acute and accelerating mining-driven destruction and the one generating the least reliable data about it. Bolivia and Guyana sit in the same blind spot. The practical consequence is that the Amazon’s most dangerous criminal frontiers are also its least monitored — a pattern that enforcement resources, donor priorities, and intergovernmental attention all risk reproducing.

Criminal networks, unlike monitoring systems, are transboundary. Criminal networks routinely operate across national borders. Yet, every national dataset reviewed in this chapter stops at the border. The responsibility for tracking these displacements — and for maintaining any coherent picture of basin-wide dynamics — has fallen to civil society organizations operating without a formal regional mandate or stable funding to match that responsibility.

Civil society is providing the data that governments cannot — or will not. MapBiomas Amazonia, RAISG, Amazon Mining Watch, MAAP, the Observatório do Mercúrio, and the Global Witness and Tierra de Resistentes databases are all civil-society products. Across every manifestation reviewed here, civil society monitoring is more consistent, more current, and more spatially complete than official systems — not because governments lack interest, but because they lack capacity, face political constraints on what they can publish, or operate systems that were never designed for cross-border aggregation. This is a significant institutional vulnerability: the basin-wide picture depends on organizations that are project-funded, understaffed relative to the territories they cover, and whose data carries uncertain standing in intergovernmental decision-making forums where national delegations may be reluctant to accept that better data is available from civil society than from their official institutions.

Legality assessment remains the main bottleneck. Detection has improved dramatically over the past decade. The harder problem is establishing what is illegal; that requires georeferenced, up-to-date cartographies of protected areas, Indigenous territories, mining and forestry concessions, and private-property authorisations — a combination that is most closely achieved in Brazil, and there only partially, given documented fraud in authorization systems. Outside Brazil, the operational proxy for illegality often reverts to simplification of occurrence inside protected areas and Indigenous territories. The five-country ASGM assessment in Section 4 demonstrates that aggregation of legality data is possible using overarching categories of presumption rather than definitive judgement, which materially changes what the data can say to enforcement agencies, prosecutors, supply chain actors, and financiers.

Together, these four patterns define both the problem and the opportunity. The data exists — imperfect, uneven, and largely civil society generated — to establish a credible regional baseline for Nature crime across the Amazon. What does not yet exist is the architecture to make it consistent, sustained, and authoritative enough to drive policy. Chapter 2 examines the civil society organizations that are building that architecture from the ground up: who they are, what they have achieved, where they are under-resourced, and where the gaps in coverage are greatest.

Notes #

1. Barber, C. V., K. Winfield, and Y. Aspinall. People. Planet. Justice. Understanding and countering nature crime. Report. Washington, DC: World Resources Institute. 2024. Available online at doi.org/10.46830/wrirpt.22.00038.

2. Funari, Gabriel; Cote, María Antonia; Aponte, Andrés; Ríos, Lina María Asprilla. Environmental crimes in the Amazon: current trends and rising threats. Global Initiative Against Transnational Organized Crime. 2025. Available at: https://globalinitiative.net/analysis/environmental-crimes-in-the-amazon/

3. Waisbich, L. T., Risso, M., Husek, T., & Brasil, L. (2022). The ecosystem of environmental crime in the Amazon: An analysis of illicit rainforest economies in Brazil. Strategic Paper 55. Instituto Igarapé. Available at: igarape.org.br.

4. This report considers all 8 Amazonian countries covered by the World Bank’s Amazon Sustainable Landscape Program: Bolivia, Brazil, Colombia, Ecuador, Guyana, Peru, Suriname and Venezuela. These countries are also members of the Amazon Cooperation Treaty Organization.

5. WWF-UK (2024), Financial Crimes and Land Conversion: Uncovering Risk for Financial Institutions. Available at https://www.wwf.org.uk/sites/default/files/2024-04/WWF-UK-Financial-Crimes-and-Land-Conversion-Uncovering-Risk-for-Financial-Institutions.pdf.

6. Schmitt, J. (2015). Crime sem castigo: a efetividade da fiscalização ambiental para o controle do desmatamento ilegal na Amazônia. PhD thesis, Universidade de Brasília. https://repositorio.unb.br/bitstream/10482/19914/1/2015_JairSchmitt.pdf . See also Nunes et al. (2024), Lessons from the historical dynamics of environmental law enforcement in the Brazilian Amazon, Scientific Reports 14: https://www.nature.com/articles/s41598-024-52180-7.

7. See Mongabay (2022), “Government inaction sees 98% of deforestation alerts go unpunished in Brazil”, https://news.mongabay.com/2022/05/governmet-inaction-sees-98-of-deforestation-alerts-go-unpunished-in-brazil/ ; Washington Post (2022), “Brazil fails to protect world’s largest rainforest”, https://www.washingtonpost.com/world/interactive/2022/brazil-amazon-deforestation-enforcement/.

8. https://www1.folha.uol.com.br/ambiente/2025/01/75-dos-inqueritos-por-queimada-e-desmatamento-na-policia-federal-nao-indiciam-ninguem.shtml#:~:text=Mas%20ressaltou%20que%20a%20sua,n%C3%A3o%20exclusivamente%20queimadas%20e%20desmatamento).

9. Sims, M.J., R. Stanimirova, A. Raichuk, M. Neumann, J. Richter, F. Follett, J. MacCarthy, K. Lister, C. Randle, L. Sloat, E. Esipova, J. Jupiter, C. Stanton, D. Morris, C. M. Slay, D. Purves, and N. Harris. 2025. “Global Drivers of Forest Loss at 1 Km Resolution.” Environmental Research Letters 20 (7): 074027. doi:10.1088/1748-9326/add606.

10. Instituto Centro de Vida, Imaflora, UFMG and WWF-Brasil (2024), Estudo inédito aponta falta de transparência e ilegalidade em 94% do desmatamento na Amazônia e Matopiba. Available at https://www.icv.org.br/noticias/estudo-inedito-aponta-falta-de-transparencia-e-ilegalidade-em-94-do-desmatamento-na-amazonia-e-matopiba/.

11. IBAMA (2025), Sinaflor x Prodes: ferramenta do Ibama distingue desmatamento legal de irregularidades na Amazônia e no Cerrado. Available at https://www.gov.br/ibama/pt-br/assuntos/noticias/2026/ferramenta-do-ibama-distingue-desmatamento-legal-do-potencialmente-ilegal

12. Regulation (EU) 2023/1115 of 31 May 2023 on the making available on the Union market and the export from the Union of certain commodities and products associated with deforestation and forest degradation. Consolidated text at https://eur-lex.europa.eu/eli/reg/2023/1115/oj. On the Amazon Soy Moratorium see Gibbs, H. et al. (2015), Brazil’s Soy Moratorium, Science 347: https://www.science.org/doi/10.1126/science.aaa0181.

13. Observatório do Código Florestal (2024), O “eco” passivo de Reserva Legal no país ultrapassa 16 milhões de hectares — 9,4 milhões deles no bioma Amazônia. Available at https://observatorioflorestal.org.br/o-eco-passivo-de-reserva-legal-no-pais-ultrapassa-16-milhoes-de-hectares/ .

14. RAISG (Amazon Network of Georeferenced Socio-Environmental Information) (2024), Pan-Amazon cartographies of Protected Areas and Indigenous Territories. Available at https://www.raisg.org/en/maps/.

15. Sims, M., Stanimirova, R., Raichuk, A., Neumann, M. et al. (2025), Global Drivers of Forest Loss at 1 km Resolution, Environmental Research Letters 20: https://doi.org/10.1088/1748-9326/add606. Data: https://datasets.wri.org/datasets/dominant-drivers-of-tree-cover-loss-at-1km.

16. Mongabay (2024), Amazon Fraud 101: How timber credits mask illegal logging in Brazil. Available at https://news.mongabay.com/2024/08/amazon-fraud-101-how-timber-credits-mask-illegal-logging-in-brazil/

17. Imazon (2024), SIMEX Pará bulletin — Sistema de Monitoramento da Exploração Madeireira, as reported in Brasil de Fato (2024), “Illegal logging in the state of Pará rose by 22% in a year”: https://www.brasildefato.com.br/2024/08/16/illegal-logging-in-the-state-of-para-rose-by-22-in-a-year/.

18. Instituto Centro de Vida (2024), Boletim SIMEX Mato Grosso — agosto de 2022 a julho de 2023: 219.032 ha de exploração madeireira mapeados. Summary coverage in CenárioMT: https://cenariomt.com.br/mato-grosso/quem-lucra-com-a-madeira-ilegal-em-mato-grosso-cadeias-ocultas-e-falhas-na-fiscalizacao/

19. https://sniffs.serfor.gob.pe/monitoreo/dgioffs/tala.html

20. https://sheffield.ac.uk/news/new-technology-helping-fight-against-illegal-logging-perus-valuable-rainforest

21. MAAP #125 (2020), Detecting Illegal Logging with Very High Resolution Satellites: https://www.maapprogram.org/high-res-satellites/ ; MAAP #139 (2021), Using Satellites to Detect Illegal Logging in the Peruvian Amazon: https://www.maapprogram.org/peru_logging/ .

22. IUCN NL (2022), Drivers of deforestation in the Colombian Amazon: Illegal logging, summarizing IDEAM estimates. Available at https://www.iucn.nl/en/publication/drivers-of-deforestation-in-the-colombian-amazon-illegal-logging/ .

23. https://igarape.org.br/wp-content/uploads/2026/04/ENG_Timber-Report-Markets-and-Forest.pdf

24. Colombian Ministry of Environment and Sustainable Development and WWF-Colombia, as cited in WWF (2023), Illegal logging in the Amazon Basin: What could countries do to fight it? https://www.wwf.org.co/en/?380470%2FIllegal-logging-in-the-Amazon-Basin-What-could-countries-do-to-fight-it=.

25. Environmental Investigation Agency (2025), Decking the Forest: How Colombia’s unlawful timber enters US and EU supply chains. Available at https://eia.org/wp-content/uploads/2025/05/EIA-DECKING-THE-FOREST-5.8.pdf.

26. InsightCrime (2019), Bolivia Forestry Officials Profit from Harvest of Illegal Wood, summarizing ABT (Autoridad de Fiscalización y Control Social de Bosques y Tierra) reporting. Available at https://insightcrime.org/news/brief/bolivia-forestry-officials-profit-harvest-illegal-wood/.

27. InsightCrime (2019), Bolivia Forestry Officials Profit from Harvest of Illegal Wood, https://insightcrime.org/news/brief/bolivia-forestry-officials-profit-harvest-illegal-wood/ ; Amazon Underworld (2024), Forest On the Run: Timber Trafficking Devouring Bolivian Forests, https://amazonunderworld.org/selva-en-fuga-el-trafico-de-madera-que-devora-los-bosques-bolivianos/.

28. Ecuadorian Environment Ministry officials as reported by InsightCrime (2015), Ecuador Deforestation Spurred by Illegal Logging, https://insightcrime.org/news/brief/ecuador-deforestation-spurred-by-illegal-logging/ ; see also Eurasia Review (2018), “Illegal Logging: An Organized Crime That Is Destroying Latin American Forests”, https://www.eurasiareview.com/21052018-illegal-logging-an-organized-crime-that-is-destroying-latin-american-forests/.

29. Government of Suriname — Stichting voor Bosbeheer en Bostoezicht (SBB), National Forest Monitoring System. See REDD+ Suriname: https://sbb.sr/sfiss-info/

30. Guyana Forestry Commission, Log-tagging and timber traceability system. Summary in Forest Policy / Preferred by Nature Guyana risk tool: https://forestpolicy.org/risk-tool/country/guyana .

31. Siqueira-Gay, J. et al. (2024), Uncontrolled Illegal Mining and Garimpo in the Brazilian Amazon, Nature Communications 15: https://www.nature.com/articles/s41467-024-54220-2.

32. Siqueira-Gay, J. et al. (2024), Uncontrolled Illegal Mining and Garimpo in the Brazilian Amazon, Nature Communications 15: https://www.nature.com/articles/s41467-024-54220-2.

33. Amazon Mining Watch Panorama QUARTERLY REPORT / October – December 2025. https://www.amazonconservation.org/wp-content/uploads/2026/03/Amazon-Mining-Watch-Panorama-Issue-1.pdf

34. ibid.

35. SOS Orinoco, Canaima National Park reports: https://sosorinoco.org/en/reports/2025-world-heritage-watch-canaima-report/ . See also Americas Quarterly (2022), “The Destruction of Venezuela’s Amazon Is Going Virtually Unnoticed”: https://americasquarterly.org/article/the-destruction-of-venezuelas-amazon-is-going-virtually-unnoticed/.

36. Global Forest Watch (2021), “Arco Minero Destroys Venezuelan Forests”: https://www.globalforestwatch.org/blog/users-in-action/arco-minero-venezuela-gold-mining/.

37. https://amazonminingwatchpanorama.org/

38. www.amazonminingwatch.org. Extension of this layer to the remaining Amazonian countries is planned for 2026

39. WWF-Brasil / Amazon Cooperation Treaty Organization (2023), Amazon has more than 4,000 illegal mining sites — ACTO-WWF study. Available at https://www.wwf.org.br/?86681%2FAmazon-has-more-than-4000-illegal-mining-sites-shows-ACTO-study-with-WWF-Brazil=.

40. Mongabay (February 2025). Mining dredges return to Amazon River’s main tributary, months after crackdown. https://news.mongabay.com/2025/02/mining-dredges-return-to-amazon-rivers-main-tributary-months-after-crackdown/

41. Fundación EcoCiencia (Ecuador), the Fundación para la Conservación y el Desarrollo Sostenible (FCDS, Colombia), Conservación Amazónica – ACEAA (Bolivia), Conservación Amazónica – ACCA (Peru), Federación Nativa del Río Madre de Dios y Afluentes (FENAMAD), Amazon Conservation Team (ACT, Suriname and Guyana), Instituto Centro de Vida (ICV, Brazil) and Instituto Socioambiental (ISA, Brazil)

42. Alessi, M.A. et al. (2023). Artisanal Mining River Dredge Detection Using SAR: A Method Comparison. Remote Sensing 15(24), 5701. https://www.mdpi.com/2072-4292/15/24/5701

43. Amazon Conservation Association (2024). Amazon Mining Watch: AI-Powered Platform Detects Gold Mining Deforestation in All Amazonian Countries for the First Time. https://www.amazonconservation.org/amazon-mining-watch-launch-press-release/

44. Planet (2026). Pelican — Next-generation tasking constellation (0.3 m resolution, onboard AI object detection). https://www.planet.com/constellations/pelican/

45. Planet (2024). Tanager — Hyperspectral imagery constellation (400–2,500 nm). https://www.planet.com/constellations/tanager/

46. Vasquez et al. (2024). Mercury Dynamics and Bioaccumulation Risk Assessment in Three Gold Mining-Impacted Amazon River Basins. Toxics 12(8), 599. https://pmc.ncbi.nlm.nih.gov/articles/PMC11359172/

47. Lopes et al. (2019). Heightened levels and seasonal inversion of riverine suspended sediment in a tropical biodiversity hotspot due to artisanal gold mining. PNAS 116(43). https://www.pnas.org/doi/10.1073/pnas.1907842116

48. WWF (2018), Healthy Rivers Healthy People: Addressing the mercury crisis in the Amazon: https://www.wwf.org.uk/sites/default/files/2018-11/WWF%20-%20Healthy%20Rivers%20Healthy%20People.pdf; IPEN (2023), Massive Amounts of Mercury for Gold Mining are Smuggled from Mexico: https://ipen.org/news/massive-amounts-mercury-gold-mining-are-smuggled-mexico-polluting-amazon-and-threatening.

49. Costa et al. (2025). Mercury Scenario in Fish from the Amazon Basin: Exploring the Interplay of Social Groups and Environmental Diversity. Toxics 13(7), 580. https://pmc.ncbi.nlm.nih.gov/articles/PMC12298900/

50. Vasquez et al. (2024). Mercury Dynamics and Bioaccumulation Risk Assessment in Three Gold Mining-Impacted Amazon River Basins. Toxics 12(8), 599. https://pmc.ncbi.nlm.nih.gov/articles/PMC11359172/

51. Vasquez et al. (2024). Mercury Dynamics and Bioaccumulation Risk Assessment in Three Gold Mining-Impacted Amazon River Basins. Toxics 12(8), 599. https://pmc.ncbi.nlm.nih.gov/articles/PMC11359172/

52. Azevedo, B.M.S. et al. (2023). Risk Assessment of Mercury-Contaminated Fish Consumption in the Brazilian Amazon: An Ecological Study. International Journal of Environmental Research and Public Health. https://pmc.ncbi.nlm.nih.gov/articles/PMC10535031/

53. Dórea, J.G. et al. (2024). Trends in Mercury Contamination Distribution among Human and Animal Populations in the Amazon Region. Toxics 12(3), 204. https://pmc.ncbi.nlm.nih.gov/articles/PMC10974390/

54. OTCA / Amazon Basin Project. Overview of Mercury Pollution in the Amazon Region. https://aguasamazonicas.otca.org/environmental-and-water-resources-monitoring/regional-overview-of-mercury-contamination/

55. OTCA / Amazon Basin Project. Overview of Mercury Pollution in the Amazon Region. https://aguasamazonicas.otca.org/environmental-and-water-resources-monitoring/regional-overview-of-mercury-contamination/

56. PAHO/WHO (October 2024). PAHO and OTCA sign agreement to promote health and sustainable development in the Amazon region. https://www.paho.org/en/news/3-10-2024-paho-and-otca-sign-agreement-promote-health-and-sustainable-development-amazon

57. World Bank (October 2023). Regional Collaboration to Address the Impacts of Mercury Pollution in the Amazon (ARAIMO). https://www.worldbank.org/en/news/feature/2023/10/27/colaboraci-n-regional-para-abordar-los-impactos-de-la-contaminaci-n-por-mercurio-en-la-amazon-a

58. UNEP / Minamata Convention Secretariat. National Action Plans — Artisanal and Small-Scale Gold Mining. https://www.unep.org/globalmercurypartnership/what-we-do/artisanal-and-small-scale-gold-mining/national-action-plans

59. https://experience.arcgis.com/experience/90af50053d514c6eb5c7d07d5e2bc616 https://storymaps.arcgis.com/stories/c14f2a58f2c4422aaf3388243040f288

60. Grotto et al. (2025). Mercury Contamination and Co-exposures in the Amazon Basin: At the Center of the Planetary Environmental Crisis. Annals of Global Health 91(1). https://pmc.ncbi.nlm.nih.gov/articles/PMC12315690/

61. OTCA / Amazon Basin Project. Overview of Mercury Pollution in the Amazon Region. https://aguasamazonicas.otca.org/environmental-and-water-resources-monitoring/regional-overview-of-mercury-contamination/

62. World Bank (October 2023). Regional Collaboration to Address the Impacts of Mercury Pollution in the Amazon (AARIMO) https://www.worldbank.org/en/news/feature/2023/10/27/colaboraci-n-regional-para-abordar-los-impactos-de-la-contaminaci-n-por-mercurio-en-la-amazon-a

63. See ACTO’s ministerial resolution RES/XIV MRE-OTCA which mentions “The need to strengthen and expand police and intelligence cooperation for the prevention, suppression and investigation of illegal activities, including environmental crimes and violations of the rights of human rights defenders, the rights of indigenous peoples and socio-environmental rights affecting the Amazon Region;”

64. The Escazú Agreement is an international treaty signed by 24 Latin American and Caribbean nations concerning the rights of access to information about the environment, public participation in environmental decision-making, environmental justice, and a healthy and sustainable environment for current and future generations.