Trading deforestation—why the legality of forest-risk commodities is insufficient

Consumer countries and blocs, including the UK and the EU, are defining legal measures to tackle deforestation linked to commodity imports, potentially requiring imported goods to comply with the relevant producer countries’ land-use laws. Nonetheless, this measure is insufficient to address global deforestation. Using Brazil’s example of a key exporter of forest-risk commodities, here we show that it has ∼3.25 Mha of natural habitat (storing ∼152.8 million tons of potential CO2 emissions) at a high risk of legal deforestation until 2025. Additionally, the country’s legal framework is going through modifications to legalize agricultural production in illegally deforested areas. What was illegal may become legal shortly. Hence, a legality criterion adopted by consumer countries is insufficient to protect forests and other ecosystems and may worsen deforestation and conversion risks by incentivizing the weakening of social-environmental protection by producer countries.


The trade of forest-risk commodities
The way we use the land to produce, trade, and consume food is directly connected to socialenvironmental issues like deforestation, biodiversity loss, human rights violations, climate change, and pandemics (Curtis et al 2018, Brancalion et al 2020, Laroche et al 2020. The production of agricultural commodities is a key driver of deforestation across the globe (Curtis et al 2018). However, deforestation embedded in global supply chains is especially acute in the trade routes between commodity-producing countries in the Global South and commodityimporting countries in the Global North. Recognizing their roles as importers and consumer countries, the United Kingdom and the European Union are considering policy measures to address imported deforestation (Bager et al 2021). In the context of distant connections in food supply chains (Laroche et al 2020), it is crucial to account for greenhouse gas (GHG) emissions, biodiversity loss, and traditional communities' rights embedded in food imports, taking appropriate mitigation measures.
Over the last decade, a wave of voluntary commitments from the private sector and nations (e.g. the Consumer Goods Forum or via the Amsterdam Declarations Partnership) have fallen short in making progress towards deforestation-free supply chains (Garrett et al 2019). Hence, at present, there is growing momentum for bolder actions from both government and private companies. Many discussions are in place on what policies could most efficiently halt the environmental degradation driven by agricultural imports (Bager et al 2021). Part of this debate favors mandatory due diligence by importing countries to verify compliance with legal criteria from exporting countries (Kehoe et al 2020, Bager et al 2021. It is essential to highlight that any legislation to tackle deforestation via a legality-based approach is dependent on the efficacy of local governments and legislation and, ultimately, its alignment with downstream deforestation-free objectives. Considering Brazil as an example of a key agricultural exporter; the country produced around 118 million tons of soybeans in 2018 alone, representing 36% of global soy production (FAO 2021). 57% of Brazil's production in 2018 was exported to China and 11% to Europe, including the UK (Trase 2021). The deforestation risk of this Brazilian soy, which includes deforestation and native vegetation loss in the previous five years that became soy up to 2018, was about 61.4 thousand hectares, emitting slightly over 10 million tons of CO 2 (Trase 2021). About 6.3 thousand hectares of this native vegetation loss (∼10%) and ∼1.1 million tons of emitted CO 2 (∼11%) belong to the EU, including the UK. These volumes refer only to 2018 soybean production, with impacts even higher in previous years (Trase 2021).
Despite global climate and biodiversity crises, Brazil's current environmental legislation authorizes significant amounts of vegetation loss (Rajão et al 2020). This destruction is not necessary from a land-resource standpoint. Brazil has sufficient suitable lands for expanding production without clearing additional hectares of native vegetation (Strassburg et al 2014). Moreover, deforestation jeopardizes overall agricultural production and income due to disruptions in local rainfall patterns driven by deforestation (Leite-Filho et al 2021). Not even the existing legal requirements are adequately enforced. Roughly 20% of Brazil's soy and at least 17% of beef exports to the EU, produced on the Amazon and Cerrado biomes, may be contaminated by illegal deforestation (Rajão et al 2020). Previous studies have shown the limits of Brazil's legislation to tackle illegal deforestation (Azevedo et al 2017), and the actual and potential increases in deforestation stemming from bailouts and revisions in the Forest Code (Sparovek et al 2012, Freitas et al 2018b, Albuquerque Sant'Anna and Costa 2021.

Natural habitat at high risk of legal deforestation
Here we build upon and go beyond Rajao et al (2020) study. We estimate the potential legal deforestation and carbon emissions in Brazil that may take place shortly (until 2025). For this, we combine several spatially explicit databases and a spatial model that estimates the probability that native vegetation will remain until 2025 in the face of several drivers of land use change (see section 6).
We identify 1,114,693 rural establishments holding ∼69.2 million hectares of unprotected native vegetation (i.e. that can be legally deforested), storing ∼5.8 billion tons of CO 2 (tables 1 and S1). Out of these, ∼3.25 million hectares storing ∼152.8 million tons of CO 2 are at high risk of deforestation and native vegetation conversion until 2025. Another ∼26.8 Mha storing 1.1 billion tons of CO 2 are at medium risk (tables 1, S1, figures 1, S1 and S2 available online at stacks.iop.org/ERL/16/124025/ mmedia). This is a plausible extent of short-term future deforestation risk. In 2020 alone, Brazil's Amazon and Cerrado biomes together lost 1.8 Mha (PRODES-INPE 2021), with about 70% of this occurring in private lands according to our own estimates based on the properties map of (Freitas et al 2018a).

The legal basis of deforestation in Brazil
The legislative framework built to protect native vegetation in Brazil comprises two main instruments: protected areas in public lands and mandatory conservation in private properties. Protected areas include conservation units-such as national parks and forests-and traditional peoples lands-such as indigenous communities. On the other hand, mandatory protection in private property is mainly regulated by the Forest Code, introduced in 1934 and most recently revised in 2012. The Brazilian Forest Code, unlike most European forest laws, was designed within the paradigm of an open agricultural frontier, granting rural owners the subjective right to convert forest land into agricultural areas as long as certain limits are respected (de Toledo et al 2017, Rajao et al 2021. These limits are legally defined as Permanent Preservation Areas (natural vegetation in riverbanks, for instance, PPAs) and Legal Reserve (LR), a portion of a given property set aside for conservation or sustainable management (Sparovek et al 2012, Rajão et al 2020. In the forestlands of the Amazon biome, as a general rule, 80% of any medium-to-large private property is overall considered LR. However, this general rule has exceptions since special conditions allow Amazon states and municipalities to reduce this LR area to 50% (Freitas et al 2018b), and medium to large farmers who cleared up to 50% of their forestlands prior to 2001, when the LR was effectively increased from 50% to 80% in the Amazon forestlands, are allowed to have only 50% LRs. Small holders who cleared any amount of LR up to July 2008 have been granted amnesty, therefore not needing to restore these parts (Stickler et al 2013). This level of protection is often used as an example of ambitious legislation, but it also means that 20%-50% of these medium-to-large properties can be lawfully deforested. Considering that properties in the Amazon can    (Freitas et al 2018a). This area is about four times the territory of the United Kingdom. The clear-cut of such regions would mean extra emissions of at least 12.48 billion tons of CO 2-all of which potentially authorized by the current Brazilian legislation (Freitas et al 2018a). This amount is almost 34 times the UK's total GHG emissions in 2018 alone (UCS 2020). In alignment with our short-term modelled estimates (see section 6), this area is what we consider as the total amount of possible legal deforestation. Whilst not all of this is likely to be cleared in the near term due to political, biophysical, and infrastructural constraints for deforestation, some areas in Brazil are likely to be at particularly high risk (figure 1).

Changing legislation over time
As stated above, Brazil's legislation is permissive enough to allow for a substantial amount of legal CO 2 emissions via land use change. Yet, the legal framework to protect native vegetation has been changing in the recent years to allow more legal deforestation and legalize economic activities carried out in former forests and natural habitat that was illegally cleared. The 2012 Forest Code is a stark example of this situation (Sparovek et al 2012). From the 1990s, Brazilian environmental agencies ensured that law enforcement was more stringent than before, and thousands of non-compliant rural producers were fined. This generated political pressure to revise the law, passed in 2012, with several amnesties for illegal deforestation carried out before 2008 (Albuquerque Sant'Anna and Costa 2021), thus effectively legalizing previously illegal deforestation. With the changes in 2012 alone, about 41 Mha of deforested and converted lands that should otherwise have been restored to native habitat were granted amnesty . Another ∼1 Mha is estimated to have been deforested between 2012 and 2017 due to the incentives provided by the revisions to law (Albuquerque Sant'Anna and Costa 2021).
The Forest Code is not an isolated case. Several bills in Parliament are on the verge of being approved that are likely to bring more amnesties and open additional space for legal deforestation (see the SI, List of Brazilian Congress Bills). These bills are part of an overarching movement by the current Bolsonaro administration to re-shape the socio-environmental legal framework in Brazil, incentivizing further legal destruction of natural habitat and carbon emissions and the legalization of agricultural activities that are currently unlawful.
This social-environmental legal laxation movement by the current administration, however, does not sit in isolation. Within local political constituencies, pro-development and anti-environmental groups are important supporters of the current government and were determinant forces in the election of Bolsonaro for presidency in 2018 (Raftopoulos and Morley 2020, Russo Lopes and Bastos Lima 2020). They are also influential in elections at local levels, mainly in municipalities and states (Rodrigues-Filho et al 2015, Pailler 2018, and are likely to substantially influence the next presidential elections in 2022.

Ways forward
To effectively implement a legality-based sustainability policy, legal deforestation activity should be monitored by producer-country authorities and by the supply chain itself. In ideal circumstances, Brazilianbased tier-1 companies (i.e. sourcing directly from farmers) would be able to conduct basic verification of their suppliers (i.e. farmers), as they do in arrangements such as the Amazon Soy Moratorium (Heilmayr et al 2020, Austin et al 2021, cattle agreements (Gibbs et al 2015), or the High Carbon Stock Approach (Austin et al 2021). However, this verification capacity would not be directly possible for tier-2 or tier-3 companies (e.g. importers, processors and retailers), obliged to meet the incoming United Kingdom's due diligence legislation to place materials on the UK market, since they do not purchase directly from farmers, but from tier-1 companies. Another example of basic verification, in Brazil, is the Federal Environmental Enforcement Agency (IBAMA)'s list of embargoed areas. This list indicates areas that have been illegally cleared. When detected, these areas are embargoed to promote the restoration or regeneration of native vegetation. Nothing can be produced or traded from these areas. Therefore, a basic verification step to be carried out by companies is checking whether their supplying farmers are on that list.
Second, companies can undertake supply chain engagement with upstream suppliers (Austin et al 2021), and request the documentation attesting the legality of production. In Brazil's example, a farmer can only conduct deforestation if they hold an Authorization for Native Vegetation Suppression (ASV, in Brazilian Portuguese acronym), which is issued by state authorities. Therefore, downstream companies might detect deforestation on a supplyingfarm by remote sensing and request the presentation of this permit as proof of legality. Importantly, the existence of this permit system implies that achieving zero-illegal deforestation or native vegetation conversion relies substantially on the discretion of the domestic legislation of producer-countries. The robustness of the domestic legislation is therefore worthy of additional attention.
As said above, in Brazil, for example, to be considered legal, any land use change from native vegetation to agricultural cover in Brazil must be subject to approval by the State Environmental Department before it takes place. The State Environmental Department must assess cases and issue land clearing permits if the requests comply with all legal requirements. Authorizations must identify the geospatial location of the clearance. Nevertheless, most states in Brazil lack any tracking system for these authorizations, which means there is effectively no monitoring of whether the authorized deforestation is being carried out lawfully. Currently, Brazilian agencies lack the technical capacity and political willingness to monitor, verify, enforce and report land use regulation.
Without such local information (i.e. the ASVs), and in the context of the global supply chains in which soy is embedded, it is virtually impossible to attest if deforestation has been carried out legally or not (Valdiones et al 2021). Companies point out technical difficulties in monitoring and verifying legal deforestation to challenge regulation and enforcement by consumer countries (Lambin et al 2020). An ideal scenario would be one where buyers could request the legal permits of clearings and production operations from their upstream suppliers (i.e. farmers). However, this is complicated by technical issues related to Brazilian authorities' incapacity or unwillingness to monitor, verify, enforce, and report legal compliance.
There are several methods to trace and verify gross (i.e. rather than just legal) zero-deforestation and natural habitat conversion in export supply chains at the jurisdictional level of production (Green et al 2019, Escobar et al 2020, Austin et al 2021, Lathuillière et al 2021. Platforms such as (Trase 2021) and Mapbiomas (2021) are examples of publicly available supply chain and land use data that companies can use for risk assessments. These tools do not yet show deforestation at the individual farm-level in the case of Brazil. This level of detail would, for example, require integration with official government data, such as with the Rural-Environmental Registry (CAR, in the Brazilian-Portuguese acronym), and the ASV state spatial datasets but these datasets are still pending validation and not publicly available in all states (Valdiones et al 2021). Despite the limitations of publicly available data that link supply chain activities to gross clearance, it is arguably the case that-due to concerns about local monitoring and enforcement necessarily to demonstrate legalitydownstream companies can assess risk exposure to gross deforestation at least as easily as to illegal deforestation.
Ultimately, companies can-and do-have mechanisms to audit and verify whether their supply chains are deforestation-free via supplier engagement processes, such as voluntary certification schemes such as the Roundtable on Sustainable Soy (RTRS) and the Forest Stewardship Council (FSC). However, a missing link to ensure a level playing field and promote industry-wide action is mandatory regulation that applies to all companies for all forest-risk commodities, with requirements for precise cut-off dates, past, present, or future, to mark when no conversion will be allowed in supply chains (Garrett et al 2019). These certification tools can, therefore, be important learning and information experiences to construct new mandatory systems.
Since companies and supply chain actors have been failing to define commitments and cut-off dates voluntarily (Garrett et al 2019), clear and stringent mandatory regulation demanding zero gross deforestation and natural habitat conversion by consumer-country authorities appears to be an effective measure. While the introduction of downstream legislation covering all forms of deforestation may appear politically infeasible in the short term (Bager et al 2021), it should not be dismissed as an ultimate goal. Nor should other potential policy mechanisms which have the potential to encompass broader protections be de-prioritized.
In sum, in addition to being insufficient to prevent deforestation activity in all forms, regulation based solely on the criterion of legality carries the risk that it may be detrimental to the protection of forests. In countries where legislation can be easily modified, as is demonstrably the case for Brazil, increased demand for products with legal origins can increase pressure for legislative changes that aim to legalize agricultural activities located in illegally deforested areas. These legislative changes, in turn, also pave the way for more legal deforestation. Therefore, while welcoming the steps made by governments to introduce regulation to respond to the global threat of deforestation, we urge consumer-nation policymakers to consider zero-gross deforestation and zero native vegetation conversion criteria in their initiatives to address imported deforestation, GHG emissions, and biodiversity loss, considering that this is technically viable and potentially effective despite the short-term political hurdles that would need to be overcome to implement these advanced measures.

Methods and data
First, we use a property-level boundaries spatial dataset, which includes the amounts of LR, PPAs, and the surplus of LRs (i.e. that can be legally cleared) (Freitas et al 2018a). Second, we use an aboveground carbon density map at 50 m pixel resolution for Brazil (Englund et al 2017). Third, we use spatial projections of unprotected native vegetation in private lands, considering rural establishments that can conduct legal clearings (Freitas et al 2018a). Fourth, we employ a spatially-explicit modelling approach that estimates the probability of the existence of native vegetation until 2025 (Fendrich et al 2020). This model considers topography, soil properties, climate data, distance to transportation corridors, and legally protected areas, including indigenous lands. This unprecedented combination of spatial datasets allows us to estimate the potential future legal deforestation with different levels of risk. For the purpose of alignment with climate policy, we use the 44/12 conversion factor to estimate the potential carbon dioxide (CO 2 ) emissions based on the total carbon stocks in the areas at risk of legal deforestation (Federici et al 2015). Therefore, we report the results directly in potential CO 2 emissions.
We adopt a conservative approach, excluding from this analysis the areas that Freitas et al (2018a) projected as potential private properties to fill spatial gaps in the Brazilian territory. We only consider the rural establishments officially registered and identified at Brazilian land databases. Therefore, while Freitas et al (2018a) estimated ∼101 Mha of unprotected native vegetation within private lands, we only considered ∼69.2 Mha. We use (Fendrich et al 2020) land cover model to classify these unprotected areas according to their risk of conversion.
The land tenure database presented in Freitas et al (2018) (Freitas et al 2018a) and updated by Freitas et al (2018) (Freitas et al 2018b) was used as starting point for the analysis presented here. This database comprises public and private properties in Brazil such as Indigenous Lands, Conservation Units, Quilombola Territories, and private rural properties from the Rural Environmental Registry (Portuguese acronym CAR) and the georeferenced properties of the National Agrarian Reform Institute (Portuguese acronym SIGEF). The database also includes information on the compliance of these rural properties with the Brazilian Forest Code. It has information related to the area of native vegetation and the amount of aboveground carbon stock within each rural property (Englund et al 2017, Freitas et al 2018a. With the existent variables, it is also possible to identify the part of the native vegetation (or its carbon stocks) that is both legally protected or has the potential to be legally deforested (the latter called hereafter as unprotected native vegetation or unprotected carbon stocks). Fendrich et al (2020) presented maps of the probability of the existence of native vegetation in a given pixel for the years 2017 and 2025. The probabilities are calculated based on a spatially explicit regression model that explores the relation of land cover maps of the Mapbiomas project (2019) (Mapbiomas 2021) and spatial variables (drivers of land cover change), such as topography, soil properties, climate data, distance to transportation corridors and legally protected areas, including indigenous lands. The land cover maps were reclassified to four classes in the model, namely, native vegetation, pasture, agriculture, and other uses. The existence of these four land cover classes was analyzed for every pixel in the entire period (from 1985 to 2017). The model captured the relation of the land cover classes and the spatial variables. Based on alternative future climate and policy scenarios (S1-aggressive, S2-business as usual, and S3-conservative scenarios), Fendrich et al (2020) estimated the probability of the existence of native vegetation, agriculture, and pasture in Brazil for the year 2025.
The Fendrich et al (2020)'s scenarios are based on the following. The S1 scenario, aggressive, considers no mitigation policy and additional environmental protection is adopted in Brazil. Furthermore, transport and energy national expansion plans are fully implemented, the population and the economy grows untapped according to national projections. The S2, business as usual, considers some environmental policies and decisions are implemented in Brazil, competing with economic growth. The S3, conservative, assumes that in Brazil there are largescale and active restoration of native vegetation and significant improvements in environmental policies (Fendrich et al 2020).
Here we used the S2-business as usual scenario, to generate a map of the variation of native vegetation probability (VNVP map) between 2017 and 2025, where positive values represent pixels with an increase in the probability of native vegetation existence until 2025. In contrast, negative values represent pixels with a decrease in the probability of native vegetation existence until 2025. Further, we have combined the rural properties with unprotected native vegetation and the VNVP map to extract the average probability for each of these properties. These probability values are relative and averaged over all pixels per property. Therefore, this value presents the average probability of natural vegetation occurring on properties with LR surpluses being lost. For example, if the LR surplus of a property is composed equally by pixels with VNVP of 2% and 5%, then its average VNVP is 3.5%, therefore it has no risk of being lost until 2025. If the value is negative, then there is risk of legal deforestation. Figure S1 expresses the distribution of the average probabilities within rural properties.
The distribution of the VNVP shows that 99.6% of the properties with unprotected native vegetation have VNVP between −5% and +5%, with the majority of the properties (79.2% of the total) presenting negative values (figure 2). Based on the distribution of the VNVP, we defined four classes of risk of the deforestation of unprotected native vegetation: • Properties with VNVP lower than −3% = High risk • Properties with VNVP between −3% and −1%= Medium risk • Properties with VNVP between −1% and 0% = Low risk • Properties with VNVP higher or equal than 0% = No risk.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files). An extra dataset is available at: https:// zenodo.org/record/5675649#.YY0eomDMLIV.