Quantifying timber illegality risk in the Brazilian forest frontier

Illegal logging remains widespread across the tropics, leading to extensive forest degradation and trade in illegal timber products. By adapting environmentally extended input–output modelling to timber originating from Brazilian native forests, we demonstrate how distinct illegality risks can be mapped and quantified at species-level across the supply chain. We focus on high-value ipê hardwood from the Amazon state of Pará, a leading producer of timber and contested forest frontier. Data on logging permits and state- and national-level Document of Forest Origin licences are used to estimate illegality risks due to missing or invalid logging permits, overstated ipê yields or discrepancies resulting from missing inflows of legal timber. We find that less than a quarter of all ipê entering supply chains between 2009 and 2019 is risk-free and highlight diversified strategies for the laundering of illegal timber across geographies. While legality does not ensure sustainability, this information can be leveraged to this end by supporting improved implementation and enforcement of forest regulations. Illegal logging in the Brazilian Amazon exacerbates existing degradation and pressures, but has been difficult to measure. This paper identifies the risk of entry points for illegally logged forests into formal supply chains to quantify how widespread this issue may be.

Illegal logging remains widespread across the tropics, leading to extensive forest degradation and trade in illegal timber products.By adapting environmentally extended input-output modelling to timber originating from Brazilian native forests, we demonstrate how distinct illegality risks can be mapped and quantified at species-level across the supply chain.We focus on high-value ipê hardwood from the Amazon state of Pará, a leading producer of timber and contested forest frontier.Data on logging permits and state-and national-level Document of Forest Origin licences are used to estimate illegality risks due to missing or invalid logging permits, overstated ipê yields or discrepancies resulting from missing inflows of legal timber.We find that less than a quarter of all ipê entering supply chains between 2009 and 2019 is risk-free and highlight diversified strategies for the laundering of illegal timber across geographies.While legality does not ensure sustainability, this information can be leveraged to this end by supporting improved implementation and enforcement of forest regulations.
The Amazon basin contains one of the world's largest remaining areas of undisturbed forest, but is also among the regions experiencing the fastest loss of intact forest landscapes 1,2 .Forest degradation in the Brazilian Amazon surpasses deforestation in terms of area-amounting to a staggering 34 million hectares (Mha) between 1992 and 2014 3 biodiversity impacts 4 and carbon emissions (estimated to be three times those from deforestation 5 ).However, forest degradation has not received the same scientific or political attention as that given to deforestation 6 .Emissions from forest degradation are not included in national reporting to the United Nations Framework Convention on Climate Change nor in Brazil's nationally determined contribution to the Paris Agreement 7 .
Forest degradation in the Brazilian Amazon is caused directly by logging and fires and indirectly through fragmentation and edge effects due to deforestation 3 .However, while the indirect effects have decreased since the early 2000s-following the reduction in overall deforestation rates-degradation due to logging remains stubbornly high: at about 0.6 Mha yr −1 in 2010-2014, it constituted nearly half of total degradation in that period 3 .Meanwhile, logging remains a poorly quantified and understood driver of global change 6 , partly due to much of it being informal and illicit 8 .
In the wake of the 1992 Earth Summit, Brazil took steps to track timber from native forests (that is, primary and naturally regenerating forests 9 ) and in 2006 a digital system 10,11 for controlling the production, transport and processing of products from such origins was established.The system currently consists of partly overlapping, though not fully harmonized, national-(SINAFLOR 12 ) and state-level systems (SISFLORA-MT in Mato Grosso and SISFLORA-PA in Pará): authorized logging permits detailing areas of extraction (by species) are the source of volume-based credits, which then flow via transport to sawmills, processing industries and traders, all documented through a Forest Guide (Guia Florestal, GF) or Document of Forest Origin (Documento de Origem Florestal, DOF) under state or federal jurisdiction, respectively.
The digital system creates a paper trail that accompanies the timber and other forest products until final processing and export.However, since individual data entries are not systematically linked, a given timber transaction can seldom be traced back to its origin 13 .The ensuing lack of traceability is exploited by actors introducing illegally Analysis https://doi.org/10.1038/s41893-023-01189-3frontiers along the highways BR-163 (running north-south) and the BR-230 (Trans-Amazonian, running east-west), as shown in Fig. 1a.Just three municipalities-Santarém, Juruti and Prainha-in the mid-west of the state together account for nearly 38% of volume logged over the period 2009-2019.
Despite year-to-year variability in production (Fig. 1b), our analysis corroborates the inward movement of the logging frontier towards the heart of the Amazon (Supplementary Fig. 2a,b) reported by other studies 19,31,35 : in 2009 the mid-state municipalities of Anapu and Pacajá accounted for 38% of total production, but by 2019 these municipalities did not even make the top ten.Meanwhile, seven of the top ten producing municipalities in 2019 were located westward of these, with nearly 49% of ipê reportedly coming from the western flank of Altamira (the largest Brazilian municipality, covering a substantial portion of mid-state Pará) and Juruti (bordering the Amazonas state and harbouring the first high-capacity port entering Pará from the west).

Illegality risk from missing or invalid logging permits
Nearly all (98%) of total volume entering the ipê supply chain could be linked to an existing logging permit, rendering the illegality risk from missing permits negligible.However, 16% (0.16 Mm 3 ) of ipê volume entered the crediting system on the basis of invalid permits; that is, permits that had been suspended, cancelled or that were missing an activation date (Fig. 1b).This substantial share lends quantitative support to the existence of loopholes in current systems, where the issuance and subsequent cancellation of permits is used to generate artificial timber credits that can then be used to launder illegally logged timber, a practice hitherto mainly documented for the case of DOFs 14,15,17 .This highlights a shortcoming of the licensing system, where multiple permits linked to the same Sustainable Forest Management Plan (PMFS) area may be issued without documentation of their interrelatedness, often leaving environmental agency analysts with the task of manual cross-checks and validation between licensing and crediting systems 36 .
Three municipalities-Santarém, Juruti and Prainha-together account for 39% of these invalid permits (Supplementary Fig. 2c,d).As shown in Fig. 1b, there is some indication that this risk has decreased over time, potentially as a result of reporting becoming more simplified: until 2014 more detailed information on the reason for permits being invalid (for example, cancelled due to illegality or elaboration failure, among other labels) is displayed, whereas after 2014 permits simply state 'cancelled' or 'suspended'.From 2018 to 2019 only marginal flows were reported coming from invalid permits.Indeed, documented efforts (for example, ref. 37) on technical improvements to the integration of licensing and crediting systems are probably a key driver of this decrease.Nonetheless, uncertainties still exist surrounding the extent and timeline of implementation of these improvements 36,37 .This therefore implies that we cannot disregard the possibility that the trend is due to a decrease in ability and willingness to pursue suspension and cancellations, in line with the broader downward trend in environmental law enforcement that is evidenced by the reduction of embargoes and infractions issued across the Brazilian Legal Amazon and particularly evident for the later years of analysis 38 .

Illegality risk from overstated ipê yields
While about 76% of ipê volume entering the supply chain can be matched to valid state-level logging permits (that is, excluding invalid or missing federal jurisdiction permits and inflows from other states), this timber is not necessarily free from illegality risks.Comparing reported volumes entering the supply chain with the areas licensed for extraction in the corresponding logging permits, we found an average ipê yield of 1.8 m 3 ha −1 (s.d., 2.2 m 3 ha −1 ).This can be contrasted with field observations from Pará 18 , where yields average 0.7 m 3 ha −1 (Fig. 2a).While we should expect logging permits to be concentrated in forests with higher than average ipê stocking rates (number of trees per hectare), the differences are quite striking: 28% of valid permits display logged timber into the system.This 'laundering' of illegal timber has been documented by Brazil's public prosecutor's office 14,15 , environmental journalists 16 , NGOs 17 and scientists 13,18 .Recent estimates suggest that as much as 44-68% of timber extraction from native forests in the top producing Brazilian states is illegal 19 .Despite this, Brazil remains a leading supplier of tropical timber to the European Union, United States and China 20,21 , notwithstanding efforts to halt trade in illegal timber such as the EU Timber Regulation and the US Lacey Act.
Addressing loopholes in the current traceability system for Brazilian timber is crucial for effectively addressing the environmental, economic and social impacts of logging 8 ; while legality does not ensure sustainability 22,23 nor fairness 24,25 , the widespread illegal exploitation of Brazilian forest resources precludes effective governance to these ends and has profound consequences, from excessive forest degradation 26 to undermining the viability of sustainable forest management 27 and fuelling environmental violence 28,29 .
Here we present an analysis that aims to overcome inadequacies in the current traceability system and improve our understanding of practices for laundering timber in Brazil.We make use of comprehensive logging permit and timber transaction data for the state of Pará for the period of 2009-2019 to identify illegality risks across the supply chain using the case of ipê (Handroanthus spp.), a group of high-value hardwood species.While accounting for only about 2% of total roundwood production from native forests between 2012 and 2017, ipê fetches the highest value among all species per volume, with estimates showing 92% is destined for export-mainly to the United States (35%) and the EU (44%) 30 .
Ipê, slow-growing and occurring in low densities, has been characterized as the 'new big-leaf mahogany' 18,31 and was added to the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) Appendix II in November 2022 for its recognized threats 32 .It is also one of the species with the most prevalent documentation of fraudulence 18,33 , and a driver of the westward movement of the logging frontier in Brazil, towards the heart of the Amazon 31 .Pará is a contested forest frontier, accounting for 28% of Brazil's total timber production from native forests (second only to Mato Grosso) and 43% of ipê harvests during 2012-2017 30 .
By compiling data from 1,262 logging permits and 309,198 records on the transport of ipê, we quantify different sources of illegality risk and identify at what stage in the supply chain illegally logged timber is potentially laundered (Supplementary Fig. 1).At the point of logging, we assess illegality risks in two ways: first, we match timber transport data to logging permits substantiating volumes credited and quantify the share of timber with missing or invalid logging permits; second, like Brancalion et al. 18 , we quantify the practice of overstating ipê yields, comparing volumes entering the supply chain per area authorized for logging with ipê occurrence estimates from the RADAM forest inventory 34 , with discrepancies between stated yields and natural occurrences of ipê indicating illegality risk.Downstream laundering of timber is quantified by comparing the inflows and outflows of timber, converted to roundwood equivalents, by individual actors; where sales of timber volumes exceed those documented as received by an actor, the discrepancy risks being of illegal origin.Finally, all three quantified sources of illegality risk-missing or invalid logging permits, overstated yields and missing inflows-are traced downstream to apparent consumption through an actor-level physical input-output model.This model uses a mass balance principle and an assumption of proportionality between an actor's inflows and outflows to overcome the lack of full traceability, allowing us to quantify the upstream embodied illegality risks associated with the consumption of ipê originating from Pará.

Results
Our compilation of logging permits (see Methods for details) used to substantiate volume entering the SISFLORA-PA and SINAFLOR timber credit systems shows that most ipê in Pará originated from forest Analysis https://doi.org/10.1038/s41893-023-01189-3yields that exceed the RADAM plots' 99% percentile class (2.4 m 3 ha −1 ).More starkly, as shown in Fig. 2b, these permits account for nearly half (48%, 0.36 Mm 3 ) of the volume entering the supply chain under valid permits, but only 14% (0.1 Mha) of the licensed area (Supplementary Fig. 3).In total, valid permits with yields exceeding the 99th percentile make up approximately 36% of all ipê entering the supply chain.
Three additional observations strengthen the suspicion that the reportedly high yields reflect an avenue for introducing illegally logged timber: first, we use a conservative threshold for overstated yields, based on simulated yield distributions for the smallest occurring logging plot in our sample (about 19 ha), but 83% of ipê volume comes from authorized areas larger than 370 ha, for which expected stocking rates are more narrowly distributed around the mean (Supplementary Fig. 4).In fact, even if we exclude all timber from the smallest quartile of authorized areas (19-66 ha)-which are most likely to include sites selected for high occurrence of ipê-the estimated illegality risk of laundering through inflated yields would only drop marginally (0.02 Mm 3 ) to 45%.Second, as shown in Fig. 2b, areas from federal-licensed enterprises display yields much more aligned with volumes derived from RADAM plots, with an average of 0.68 m 3 ha −1 (s.d., 0.87 m 3 ha −1 ).While the SINAFLOR database includes only 29 permits ( just 3% of the number from state level), and there is inconsistency in reporting between  All classes aside from 'National jurisdiction' and 'Undetermined' come from state-level permits.The black line serves as reference for total annual roundwood production from Pará.

Analysis
https://doi.org/10.1038/s41893-023-01189-3state and national level, plus the fact not all volumes may yet have been reported for concessions (given the 2019 cut-off date), this still indicates that national jurisdiction enterprises are less susceptible to this type of illegality risk.Third, the logging permits displaying yields above the 99th percentile class are highly concentrated geographically: the municipalities of Juruti and Santarém together account for 31% of volumes with potentially overstated yields (0.11 Mm 3 ; 11% of the entire roundwood production) (Supplementary Fig. 5).

Illegality risk from production-consumption flows discrepancies
According to the transport records (GF and DOF) analysed, nearly 1.0 Mm 3 of ipê roundwood was reported as leaving forests across Pará between 2009 and 2019, with an additional 0.17 Mm 3 roundwood equivalent (RWE) entering the supply chain from federal units.The figures contrast with the amount of timber originated in the state being consumed nationally and abroad, which, according to the same data, amounts to nearly twice as much: 2.1 Mm 3 RWE (1.8-2.6 Mm 3 RWE using a high (53.9%)and low (35%) overall sawmill efficiency in timber processing, respectively; see Methods).Consequently, 37-54% (0.65-1.39 Mm 3 ) of the total ipê volume entering the supply chain in Pará cannot be traced to a forest of origin (or to an inflow from other states) (Fig. 3).
Belém, Breu Branco and Itaituba-all key port cities and commercial hubs-together account for 36-38% (0.23-0.53 Mm 3 ) of total discrepancies and this increases to nearly half of all discrepancies-and a fifth of all ipê being transported in the Pará supply chain-if also accounting for metropolitan and adjacent cities.The Belém metropolitan region (which includes origin municipalities of Belém, Ananindeua, Benevides, Castanhal and Santa Bárbara do Pará), together with the neighbouring port city of Barcarena-none of which produce any ipê, but from where most timber exports leave the state-account for a fifth to a fourth (19-27%; 0.13-0.37Mm 3 ) of all discrepancies.

Mapping illegal timber flows through the supply chain
The results above create a picture of diverse strategies for the laundering of illegal timber across the state.Overall, only about 19% of the ipê volume transacted as licit in Pará was found to be entirely risk-free.On the eastern and more consolidated frontier (Fig. 3a), the predominant illegality risk comes from flows that cannot be traced to a forest of origin.Relatively low signs of overestimation exist, indicating that here illegal timber entry points are associated with downstream processors and intermediaries, rather than in the first stage of the supply chain.Moving westwards, patterns of illegality risk become more diversified in line with the opportunities offered by larger producing municipalities.
Such diversity can be best exemplified by the contrast between Itaituba and Juruti, both of which are significant points of origin of ipê located at the border with the state of Amazonas.Nearly 80% of timber originating from Itaituba cannot be accounted for by own production and inflows, suggesting the more prominent risk here is at the processing and intermediaries level.Meanwhile, more than half of the illegality risk for timber originating in Juruti comes from overstated yields.A significant volume of timber from invalid permits has also entered the supply chain through this municipality, indicating a higher risk for the entry point of illegal timber at the first stage of the supply chain in this municipality.
Despite the diversity of entry points, Belém metropolitan region and neighbouring Barcarena exert a clear consumer pull on supply of ipê from across the state (Fig. 3b), largely due to export demand.We found 69-65% (2.59-1.78Mm 3 ; low-high sawmill efficiency) of the entire ipê volume bought and sold along the supply chain is destined for export, while the remaining 31-35% can be interpreted as domestic final consumption or consumption by intermediaries (that may be exported after processing).Only a relatively modest share of ipê is destined to other municipalities within Brazil, with two other port cities in the south of the country (Paranaguá and Itajaí) making the top ten municipalities in terms of apparent consumption.

Discussion
Over the past three decades, Brazil has invested significant resourcesat federal and state level-in developing legal frameworks, institutional and technical capacity, as well as supply chain actor's buy-in, towards a system for controlling timber extraction from native forests.Our analysis shows that data from these systems-while far from perfectcan be leveraged to identify and quantify illegality risks at the actor level in different stages of the supply chain, from timber harvesting to processing, transport and export.Given previous evidence of the effectiveness of field-based enforcement in controlling deforestation in the Brazilian Amazon 39 , this points to great potential in using available data from these systems to reduce forest degradation.To date, this potential remains largely untapped: the SISNAMA (National Environmental System) institutions tasked with overseeing environmental enforcement in Brazil have largely ignored the issue of illegal timber harvesting, lacking-as previously highlighted in the case of deforestation 40 -accurate statistics on the magnitude of illegality.This omission is amplified by the recent national downward trend in embargoes and infractions issued by environmental authorities 38 and the curtailment in resources availability for SISNAMA institutions 41 under the Bolsonaro regime.
In response to the lack of capacity and political will for environmental monitoring and enforcement 38,42 , civil society initiatives (for example, SIMEX Network 19 and Imaflora's Timberflow), as well as private sector solutions (for example, BVRio), are currently trying to fill information gaps and support due diligence in the Brazilian timber supply chain.Nonetheless, the lack of transparency, quality and completeness of information made available by SISNAMA institutions makes this a daunting task 19 .The most obvious shortcomings of the current traceability system are the lack of systematic connection of processed products with forest of origin, the fact that finished products are captured too broadly (or nor at all) beyond certain categories of sawnwood and decking, and the inaccessibility of data.Notably, the public-facing SISFLORA-PA (https://monitoramento.semas.pa.gov.br/sisflora/) and DOF System (https://servicos.ibama.gov.br/ctf/modulos/dof/consulta_dof.php)search engines, for instance, still only allow for single GF or DOF searches, thus providing limited support for end-to-end due diligence from a buyer's perspective.While the assumption may be that internally such systems operate with lesser constraints, audit documentation from the Federal Court of Accounts 36 suggests otherwise: even after a decade of technical cooperation, analysts of the Brazilian Institute of the Environment and Renewable Natural Resources (IBAMA) still could not access key information from the tax department of the National Registry of Legal Entities (CNPJ) to check if timber transactions of enterprises amounted to a proportionate level of reported economic activity, and had to resort to single-searching platforms and manual cross-checks.
Importantly, poorly integrated state-and national-level systems continue to provide opportunities for the laundering of illegal timber.The lack of integration is evident at two levels: between the licensing (SIMLAM) and the transport (SISFLORA) systems, and between the state and national level of transport systems.In 2014, an agreement on integration across these levels was signed and in 2020 a normative instruction 43 again sought to make states comply with SINAFLOR.However, full integration remains to be accomplished, with Pará and Mato Grosso data 'still missing for the entirety of the period' in the Brazilian Open Data portal.On the one hand, states have stepped up to devise the first comprehensive timber traceability systems, but lack the national oversight needed for full traceability.On the other hand, national agencies with the mandate have failed to offer a replacement system that incorporates hard-won lessons and improvements made by states (for example, certain features are oversimplified at national level, such as the lack of DOF type, barcode tracing and registrations that may be important for internal monitoring by states).
While the analysis presented above tries to address some of these limitations-through careful compilation and deduplication of data across state-and federal-level systems and constructing an actor-level input-output model to trace timber from origin to consumption-it is no substitute for a full-fledged traceability system.The fragility of the existing licensing and timber crediting systems and the lack of official estimates on illegality, together with the lack of enforcement, currently do little more than add a veneer of legality to a system known to be riddled with loopholes, as our (and other 18 ) results for ipê in Pará have clearly shown.Nothing makes the shortfall of the current system more evident than the fact that, while our analysis suggest that over three-quarters of all ipê originating from Pará is associated with some degree of illegality risk, most of that timber is still being exported to markets in the United States and EU countries 33 that have legal provisions in place which prohibit the imports of illegally logged timber.
Recent developments, however, give some reasons for hope: as of 5 December 2022, a new 'DOF+ Traceability' (DOF+ Rastreabilidade) system has come into force 44 , which automatically generates a persistent traceability code to products being credited into the system, at last allowing for full traceability from timber harvest to consumption or export.While this system-which is meant to replace the current federal-and state-level DOF systems-would constitute a major improvement in traceability for Brazilian timber, our analysis also highlights the need for going beyond just improved reporting.If entered data is not cross-checked for credibility and consistency and used as a basis for monitoring and enforcement, actors are likely to continue to use existing loopholes or find new ones for laundering illegally logged timber.

Methods
We propose three complementary approaches for quantifying sources of illegality risk of Brazilian timber from native forests, reflecting three different stages in the timber supply chain.First, for roundwood being Analysis https://doi.org/10.1038/s41893-023-01189-3transported from forests, we match unique identifiers in logging permits and state-and federal-level documents of forest origin (GFs and DOFs) to distinguish timber with valid and invalid logging permits.Second, at the timber-harvesting stage, we quantify the documented practices 18 for inflating the harvest of valuable species to allow illegal timber to enter the supply chain, by contrasting implicit ipê yields with stocking rates from the RADAM forest inventory.Third, we assess inconsistencies between downstream actors' input and output volumes to identify the introduction of illegally logged timber, using an adapted environmentally extended input-output model.This model also allows for the connection between localities of production and consumption, supporting an analysis of illegality risks from a consumer perspective.The next sections first introduce the unique data used for the analyses and then detail the three steps outlined above (see Supplementary Fig. 1 for an overview of the methodological approach and the input data used in each step).The analyses have been performed in R and Python and leveraged several open source libraries referenced in Supplementary Note 5.

Timber production and traceability systems in Brazil
Timber originating from Brazilian native forests first enters the supply chain via logging permits generically known as Authorization for Forest Exploitation (Autorização de Exploração Florestal, AUTEX).Once the timber is felled, the issuance of a DOF is obligatory for the transport, reception, processing and stocking of products of native origin 45,46 .The digital DOF System, instituted in 2006 10,11 , is in practice a forest products credit system based on units of volume: credits are first created in connection with the authorized area, where extracted volumes (by species explored) are credited to the forest owner or administrator.Credits then flow via transport to sawmills, processing industries and traders.The DOF creates a paper trail that accompanies the timber and other forest products until its final domestic processing and export, thus constituting the primary instrument and data source for Brazilian timber traceability and origin accountability.
While this is a simplified national view, implementation of timber origin control has been historically complex and diverse.The National System for Control of Origin of Forest Products (SINAFLOR), created in the aftermath of the 2012 Brazilian Forest Code, sought to integrate the varying sub-national control systems for forest product origins, as well as improve and systematize processes related to licensing and monitoring of forest exploitation activities 12 .However, Mato Grosso and Pará still maintain their own state-level systems-SISFLORA-MT and SISFLORA-PA, respectively-that are deemed compatible with SINAFLOR 10 and in many ways are precursors to the latter.Despite broad compatibility, notorious inconsistencies between these systems still exist and states have been pressured to join the national system 43 .
A key difference relevant for this study is the use of the Forest Guide (GF), the equivalent of the DOF for internal transport in SISFLORA-PA (and SISFLORA-MT) and used alongside DOF in international due diligence systems (for example, BVRio).Variables in GFs and DOFs are not collected or reported equally (see Supplementary Note 4 and Fig. 6 for further details).Additionally, SISFLORA-PA underwent an upgrade in 2016, resulting in slightly differing data structures for GFs under what is now known as SISFLORA-PA 1.0 (2006-2016) and SISFLORA-PA 2.0 (2016-).The environmental licensing under which logging permits are issued also differs between state and federal level, with the logging permit (AUTEX) having its Pará counterpart called AUTEF.In practice, this implies that the national-and state-level environmental licensing and timber transport systems are only partially overlapping.
Despite varied nomenclature in licensing between states and national systems, the sources of roundwood production can be broadly grouped into four categories: (1) harvest of isolated trees (accounting for 0.2% of Brazilian timber production in 2012-2017); (2) forest plantations (4.3% of volume); (3) Sustainable Forest Management Plans (PMFS) (87.6% of volume); and (4) vegetation suppression (legal deforestation; 7.9% of volume) 30 .Each encompasses its own permit issuance requirements and protocols, ultimately constituting different types of logging permits.Requirements for forest management plans were first put forward in 1987, following which exploitation of native forests has been regulated by legal parameters 47 .Current sustainable forest management requirements broadly encompass a 25-35-year harvest cycle, with an overall upper limit of 30 m 3 ha −1 (about 3 trees per ha), with special cases applying for lower intensities and smaller areas 45,48 .A minimum of 10% of trees per species should be retained, ensuring that at least three individuals with diameter at breast height above the minimum harvest diameter (50 cm) are kept in every 100 ha, occurrence below which precludes the exploitation of the species altogether (including differentiated rates for species listed as vulnerable 49 ).Legislation underpinning PMFS licensing across jurisdictions include provisions for differentiated harvest intensities according to species.Nonetheless, it is widely recognized that these legal parameters are set at odds with species-specific thresholds for sustainable harvest 22,50 -particularly for sought-after species such as ipê 31 .The PMFS category under which most (96.9%)ipê is produced 30 , therefore, includes everything from actors pursuing best practices (for example, reduced impact logging), to those just fulfilling the legal minimum, or worse.

Illegality risk due to missing or invalid logging permits
In principle, all roundwood being transported from forest of origin should have a valid logging permit that substantiates the volume credited to corresponding GFs and DOFs 11 .However, fraud schemes starting in this node of the supply chain are well-known 13,14,17,18,51 .A common practice for laundering illegal timber is the issuance and subsequent cancellation of documentation, which generates illicit credits that are used to enter illegally logged timber into the system 17 .This practice has mainly been documented 14,15 in the case of DOFs 17 .In addition, GFs and DOFs may cite logging permits that are either nonexistent or that lack required information.However, the lack of a comprehensive database of logging permits that detail the origins of forest products have precluded an assessment of the extent of these practices.
To overcome this, we compiled a consolidated logging permits database for the state of Pará by combining official data made available publicly or through other publications (see Supplementary Note 1 for details and sources).A total of 1,292 permits have been used by supply chain actors transporting roundwood to report place of ipê origin: 1,215 of these were backed up by logging permits issued at state level and 47 at federal level, while the remaining 30 could not be backed up (original permit data could not be downloaded or found).
Logging permit information and timber flows entering the supply chain (as documented through GF and DOF records) were uniquely matched by permit.We used GF data from the SISFLORA-PA system accessed via the Timberflow initiative (http://timberflow.org.br/).DOF data were accessed via the SINAFLOR 'transport' module (https:// dados.gov.br/dados/conjuntos-dados/dof-transportes-de-produtosflorestais)(see the 'Production vector' below and Supplementary Note 3 for details).By extracting authorization status and other key information (for example, permit number, activation date, geolocation, area authorized) from logging permits, we identified timber flows with (1) missing logging permits and (2) logging permits that for different reasons had been suspended, cancelled or missed crucial information (for example, activation date).This analysis provides data both on localities of timber production and respective volumes entering the supply chain, as well as indicators of illegality risk (missing or invalid logging permits) that were then fed into the input-output model to trace this risk through the supply chain to downstream consumers.

Illegality risk from overstated ipê yields
High-value species are particularly susceptible to yield overestimation, as demonstrated by Brancalion et al. 18 when contrasting volume

Fig. 1 |
Fig. 1 | Forest of origin.a, Spatial distribution of issued logging permits and associated volume entering the supply chain.Unit: million m 3 (Mm 3 ).b, Status of issued logging permits used to transport roundwood out of forest of origin.