Across the world, farmland is replacing some of our most important ecosystems.
We’ve known for years that agriculture is rapidly consuming forests — an issue the world urgently needs to tackle. Yet this problem doesn’t stop at the forest’s edge.
New research from WRI, Land & Carbon Lab, Rainforest Alliance, and the Senckenberg Biodiversity and Climate Research Centre indicates that the world lost as much as 95 million hectares of non-forest natural ecosystems, including grasslands, savannas and wetlands, to annual crops between 2005 and 2020. A comparable area (95 million hectares more) was likely converted to pasture. Together, this is an area nearly as large as Indonesia — and roughly four times the amount of forest that was lost to annual crops and pasture over the same period.
Though sometimes overlooked in conservation efforts, non-forest ecosystems are vital to people and the planet. Grasslands are estimated to hold between 20% and 35% of land-based carbon stores. Wetlands (with and without trees) hold another 20%-30%, despite covering much less area. Non-forest ecosystems provide crucial wildlife habitat, protect soil, sustain fresh water supplies, and underpin food security and livelihoods for over a billion people around the world.
Companies and governments alike have begun taking important steps to address deforestation driven by agriculture. But these policies and commitments often don’t extend to non-forest ecosystems. Having a clearer picture of conversion across all ecosystems is critical to truly understanding and preventing further loss of grasslands, savannas and wetlands as well as forests.
This research offers a starting point, providing the first global look at which commodities are associated with ecosystem conversion outside of forests and where the impacts may be greatest.
What’s Driving Non-Forest Ecosystem Loss?
Crop expansion accounted for about half of the ecosystem conversion we tallied across grasslands, wetlands and savannas, with pasture expansion making up the other half. Understanding which crops are associated with this conversion is key, especially for companies and stakeholders working to assess and address it.
Much of the ecosystem conversion in our study was associated with globally traded commodity crops like soy, corn, rapeseed, cotton and sugarcane. While sometimes eaten, these crops are commonly used for things like animal feed, fiber or biofuel. Companies and governments that purchase or import these products can play an important role in tackling conversion through more sustainable sourcing.
Other crops, such as rice, vegetables, pulses and (in some places) corn, tend to be grown primarily for food and may be central to local food security. Environmental impacts associated with these crops may be better addressed through domestic policy than through trade policy or corporate action. Where corporate action is appropriate, food security implications should be considered.
Dominant crops linked with non-forest ecosystem conversion
While crops have replaced large swaths of grasslands, wetlands and savannas around the world, the types of crops and their uses can vary by region.
Maize (corn) is the crop mostly widely associated with non-forest ecosystem conversion around the world. While corn can be a staple food crop, it’s also used in animal feed, biofuels and chemicals.
Demand for corn has been reinforced by biofuel mandates and subsides, especially in the United States. This means that efforts to curb ecosystem conversion driven by corn should consider not just where it’s grown, but also how policy shapes how it’s used.
Soy is also associated with conversion of non-forest ecosystems on most continents. It is nearly always grown as a commodity crop, used primarily in animal feed, soybean oil and non-food products, including biofuels.
Other commodity crops, like rapeseed, cotton and sugarcane, are not as widespread as corn or soy but are associated with conversion hotspots in specific regions.Rapeseed is linked to extensive grassland conversion in Canada, the U.S. and Australia.
Cotton is expanding alongside ecosystem conversion in India and Benin. Sugarcane is associated with conversion in countries such as Brazil and Mexico.
Because many of these products enter global supply chains, demand for a commodity in one country can often drive ecosystem conversion in another — effectively outsourcing it. For example, demand for soy in China is associated with extensive ecosystem conversion in Brazil.
Factoring in Livestock
Pasture for livestock replaced roughly as much area of non-forest ecosystems between 2005 and 2020 as all the crops we evaluated combined. In some of the countries with the greatest total conversion, including Brazil, Australia and Russia, conversion to pasture was much larger than conversion to cropland.
But pasture is only part of livestock’s land footprint. Cropland expansion also needs to be factored in, because a substantial share is used to grow livestock feed. In this study, over a third (34%) of the total area converted to cropland was used for feed. In some places, such as Brazil, Argentina, the U.S., China and the EU, feed accounted for over half of cropland conversion. International demand plays a major role here: In places like Brazil and Argentina, 80% of the feed grown on converted land was destined for foreign markets like China and Europe.
When pasture and crops used for feed are added together, livestock production is the single biggest replacement of non-forest ecosystems globally. In other words: Meat and dairy simply cannot be ignored in efforts to rein in agricultural expansion and protect natural ecosystems.
What Does This Mean for Sustainable Supply Chains?
As global demand for agricultural commodities takes an ever-bigger toll on the world’s ecosystems, large companies that produce and source these products have a responsibility to identify and minimize impacts linked to their businesses and supply chains. While forest protections have been increasing, policies safeguarding other valuable ecosystems have lagged. This study helps fill in a critical blind spot. With better information about where ecosystem conversion is happening and which products are most associated with it, companies, governments and NGOs can design stronger policies, track progress, and create incentives that reward protecting natural ecosystems.
Information about where ecosystem conversion is happening outside of forests can also help prevent unintended consequences from other policy decisions. For example, there is evidence that increased forest protection in the Amazon in the early 2000s lowered deforestation linked to soy in the region. But it was also associated with increased conversion in the neighboring Cerrado, which is one of the world’s most biodiverse savannas. Including non-forest ecosystem protection in responsible supply chain policies can help prevent displacing agricultural impacts from forests to grasslands, savannas and wetlands.
How Can Companies Help Curb Ecosystem Loss?
It’s not one or the other — the world needs to protect forests and other vital ecosystems. This doesn’t require reinventing the wheel; often, non-forest ecosystems can be woven into existing sustainability goals.
There are already positive examples. The Accountability Framework initiative provides detailed guidance for companies on setting and implementing commitments to eliminate both deforestation and conversion of other natural ecosystems from their supply chains, also known as “Deforestation and Conversion-Free” (DCF). Nearly a quarter of major soy-sourcing companies and nearly a fifth of major beef-sourcing companies already have commitments that include protection of non-forest natural ecosystems. Through the Science-Based Targets Network (SBTN) and Science Based Targets Initiative (SBTi) — which support companies in setting robust nature and climate goals, respectively — over 300 companies have committed to reducing emissions from conversion or eliminating conversion of all natural ecosystems from their operations and supply chains.
Companies can also begin engaging with their suppliers by communicating clear policies and goals to not purchase products from farms that have cleared natural ecosystems. And they can take steps to incentivize ecosystem protection by producers.
Addressing conversion is especially salient for companies that produce or source meat or dairy products, due to the outsized footprint of livestock grazing and feed production. The Greenhouse Gas Protocol Land Sector and Removals Standard requires that companies with animal products in their value chains account for the land area used for grazing as well as the cropland area used to grow feed, and that companies report on their full land-use-change emissions across all ecosystems and land-based products.
With a finite amount of land on the planet, governments and companies must think about how to use farmland more efficiently and reduce pressure on remaining natural ecosystems. Given that livestock drives such a large share of ecosystem conversion, measures to shift diets high in meat toward more plant-based foods can help lessen the need to convert more land for feed and pasture. Reducing food loss and waste is another major opportunity to boost efficiency; as much as 40% of the food the world produces today goes uneaten.
Finally, using land for crop-based biofuels and other non-food products can increase ecosystem conversion by increasing the total global land area devoted to growing crops. Policies should seek to not further increase the area of cropland dedicated to biofuel production.
This is an early global study of non-forest ecosystem conversion, and there’s a need for additional research and targeted solutions in this arena. More detailed analyses in specific sourcing regions, using contextualized data sets, will be needed to pinpoint specific locations of recent commodity-driven conversion, support engagement and action to reduce ecosystem clearance, and monitor progress over time.
But companies don’t have to wait. They can take action today to help ensure we can feed the world and protect the ecosystems all of us rely on.
]]>Preventing deforestation and habitat loss from agricultural expansion is a priority for companies, governments, and sustainability initiatives around the world. To set and implement deforestation- and conversion-free (DCF) commitments, companies must be able to identify instances of deforestation or conversion in their supply chains. This requires differentiating between natural ecosystems and agriculture using clear definitions.
To provide such clarity, the Accountability Framework includes definitions of terms such as agricultural use, forest, and natural ecosystem. This article provides an overview of these terms, how they are applied, and what this means for companies implementing and monitoring DCF policies.
*Natural ecosystems and agriculture as discrete land uses *
Many international systems that define, monitor, or regulate land use patterns consider agriculture to be mutually exclusive to forests and other natural ecosystems. This includes the UN Food and Agriculture Organization’s Forest Resources Assessments, the European Union Deforestation Regulation (EUDR), and the Accountability Framework.
This mutually exclusive distinction facilitates the detection, mapping, and monitoring of deforestation and conversion due to agricultural expansion. It also enables companies to implement, monitor, and report against no-deforestation or DCF policies throughout their supply chains.
In the large majority of cases, distinctions between natural ecosystems and annual and perennial cropland are very clear. This reflects the reality that the establishment of cropland generally displaces the pre-existing natural ecosystem. Cropland includes annual crops (such as grains, oilseeds, and vegetables), perennial crops (such as coffee, fruit orchards, and oil palms), and crops grown in agroforestry systems, which the FAO and others explicitly define as an agricultural use. There are certain cases where agricultural crops may occur in natural ecosystems, such as wild-harvested coffee in Ethiopia, but these are uncommon and do not contribute heavily to international supply chains.
What about livestock?\ While distinctions between natural ecosystems and cropland are usually clear, this is less frequently the case for livestock production. This is because livestock raising often occurs within natural ecosystems. Such practices are widespread and are an important part of many traditional and modern livestock production systems.
In many contexts, responsibly managed livestock grazing can be compatible with the maintenance or even restoration of natural grasslands, savannahs, or open forests. Most such ecosystems historically included grazing animals, and livestock can serve critical ecological functions and may prevent degrading processes such as the encroachment of trees and shrubs.
The Accountability Framework definitions can be used to determine whether a plot of land with livestock grazing should be considered agriculture or a managed natural ecosystem. In general, where planted pastures have replaced the native vegetation (ie, the main elements of the ecosystem’s composition, structure, and function), the site is considered agriculture. Where this is not the case, the site is considered a managed natural ecosystem if main elements of the ecosystem’s expected species composition, structure, and function remain in place. However, where livestock raising has led to severe and sustained degradation (including major negative effects to composition, structure, or function due to persistent overgrazing), the site is considered to have been converted.
From a policy standpoint, these definitions help safeguard the values of natural ecosystems, while affording ranchers and herders with flexibility to manage their lands responsibly and in a way that responds to shifting environmental conditions. Livestock raising in grassy natural ecosystems is generally compatible with DCF supply chain policies as long as native plant species and vegetation structure are generally retained and severe degradation is avoided. At the same time, application of the definitions protects these livestock-raising systems against forms of conversion that can greatly reduce biodiversity and other values—for instance conversion to cropland or to plantations of non-native tree species.
Putting the definitions into practice\ While there is wide variety in agricultural systems—including livestock raising systems—around the world, the Accountability Framework’s definitions provide a globally-applicable basis for companies to implement and monitor deforestation-free and DCF supply chain policies.
To access Accountability Framework Definitions and our guidance on how to implement them, use the links below:
To learn more about how the Accountability Framework’s definitions of agricultural use are applied on the ground, read about WWF-Australia’s work to help commodity producers and buyers distinguish natural forests from agriculture in Australia here.
]]>Non-forest ecosystems are globally important, but it can be hard to identify and quantify their conversion to agricultural commodity production. This makes it more difficult for companies that produce or source agricultural commodities to implement their commitments to eliminate conversion from their supply chains, and to account for and reduce land use change emissions as part of corporate climate target setting.
Since 2023, teams from the Accountability Framework initiative (AFi) and World Resources Institute (WRI) have been working to close knowledge gaps around non-forest ecosystems, aiming to improve supply chain action both globally and in specific regions. This effort has included convening research at a recent conference and hosting sessions at workshops in North and South America. In March 2025, the AFi, WRI, WWF-Australia, and Griffith University co-hosted a workshop in Brisbane to engage with Australian stakeholders on defining, mapping, and protecting non-forest ecosystems in Australia. Find out more about the AFi and WRI workstream here.
The importance of Australian non-forest ecosystems
According to data from the Australian Bureau of Statistics, non-forest natural ecosystems cover approximately 73% of Australia’s landmass. These include savannahs, grasslands, shrublands, and deserts, which contain high levels of endemism and, all together, store most of Australia’s carbon. Many of these natural ecosystems are currently grazed by livestock, while also providing many of the benefits and services of natural grasslands and savannahs.
Despite their importance, these non-forest ecosystems are vulnerable to conversion due to agricultural expansion. Australian companies and producers can play a significant role in maintaining and enhancing these non-forest ecosystems. However, for effective, appropriate, and coordinated conservation of these ecosystems to occur, there needs to be better understanding of how to define and map commodity-driven ecosystem conversion for the Australian context.
Collaboratively unpacking conversion for the Australian context
During the workshop, stakeholders discussed regional realities and the specific challenges of applying global definitions, data, and tools to local landscapes. Participants from academia, civil society, industry, finance, as well as First Nations representatives explored how Australia’s non-forest ecosystems are defined, monitored, and impacted by land use change.
A panel of First Nations representatives offered vital insights into topics including traditional land stewardship, highlighting perspectives often missing from corporate monitoring approaches. Experts in ecology and remote sensing contributed presentations that helped situate the Australian landscape within broader efforts to map and monitor conversion globally.
Data under the microscope
The group evaluated several global and national land cover datasets, assessing their strengths and limitations in identifying non-forest natural ecosystems and tracking their conversion. These assessments considered both the technical capabilities of the datasets and their practical utility for decision-making in corporate supply chains and conservation strategies. The participants were also able to share feedback with the developers of global products such as the SBTN Natural Lands Map and Global Pasture Watch.
Participants also discussed potential approaches to guide companies in identifying and monitoring ecosystem conversion in Australia. This included early-stage thinking about regionally relevant land cover classifications and indicators, and how these could be operationalised in line with the Accountability Framework.
The first step of many
This workshop provided a crucial first step in reaching a shared understanding of non-forest conversion in Australia. However, fully defining and mapping non-forest ecosystems to support the implementation of no-conversion commitments will require further collaborative engagement in Australia and internationally. Workshop organisers and participants are planning to synthesise workshop insights into a scientific publication and incorporate the workshop’s findings into future guidance.
In addition, the AFi and WRI will also host further workshops to learn more about other regional contexts. The next event will take place in Colombia later this year. As the project progresses, further engagement with local experts and stakeholders will remain essential to shaping guidance that is both technically robust and aligned with on-the-ground realities.
Join the conversation
Capturing the complexity of non-forest ecosystems requires input from a wide range of stakeholders through continued conversations. If you’re involved in land classification, supply chain due diligence, or land use change monitoring, and particularly if your work touches on Australia’s unique ecosystems, then your input can help support this project.
Contact AFi Science and Policy Specialist Sam Levy to get involved!
]]>Brazilian cattle production is one of the leading drivers of tropical deforestation worldwide. This greatly affects the Amazon, the world’s largest and most biodiverse forest, where pasture expansion is the direct driver of the majority of deforestation. However, deforestation has been falling due to programmes like Beef on Track, a multi-stakeholder initiative to prevent deforestation and human rights violations in Amazonian cattle supply chains.
Beef on Track helps companies that produce or buy beef to harmonise and strengthen their supply chain policies and monitoring systems. It also provides them with criteria and methodologies to follow as they implement their commitments to responsible sourcing in the Brazilian Amazon. These include minimum criteria for preventing deforestation, protecting Indigenous Peoples and traditional peoples, and avoiding slave labour abuses. Beef on Track is led by AFi Coalition member Imaflora, along with the Brazilian Federal Prosecutor’s Office, and the programme supports company action in line with the Accountability Framework.
Beef on Track’s Monitoring Protocol 2.0
Beef on Track’s Monitoring Protocol 2.0 for cattle suppliers came into effect in January 2025. The updates build on Beef on Track’s established criteria while strengthening existing rules for the 124 slaughterhouses that have signed Beef on Track, raising the bar for these companies. The new Monitoring Protocol will improve legality and accountability in the sector and increase the confidence for buyers that when they purchase from a company adhering to Beef on Track, they are obtaining ethically-reared beef that is free from illegal deforestation or human rights violations.
The updates to Beef on Track were determined through consultations with experts and stakeholders, including Câmaras Técnica e Social (Technical and Social Advisory Boards) in the Brazilian state of Pará, and in-depth discussions with the Brazilian Public Prosecutor’s Office. The updated Monitoring Protocol was launched in August 2024, giving the cattle sector time to adapt to the new rules before they came into effect this year. The full text is available on the Beef on Track website in English and Portuguese.
Strengthening sustainability in the Brazilian cattle sector
Version 2.0 adds two new criteria to the Monitoring Protocol, and strengthens several of the 12 criteria found in Version 1. Criteria used to block non-compliant producers from selling cattle to Beef on Track signatories are reinforced by including additional public databases of environmental crimes and broadening the list of recognised Indigenous Peoples and traditional peoples. This includes a new criterion that blocks producers whose cattle grazing areas encroach upon Quilombola lands. Quilombola are traditional Afro-Brazilian communities who have been vulnerable to land grabbing.
Potential loopholes have also been closed by blocking supply chain access to irregular producers that alter their property boundaries or deforest their lands incrementally. Of particular importance is a new criterion increasing monitoring to all properties that producers own within the same geographic region. This criterion aims to prevent producers from ‘laundering’ cattle between Beef on Track compliant and non-compliant ranches. Any producers with deforestation or human rights violations on any of their local properties will be blocked from selling to slaughterhouses that are members of the Beef on Track programme.
Implications for beef buyers and financial institutions
Changes to the Beef on Track Monitoring Protocol have important implications for financial institutions and companies that purchase Brazilian beef and leather. As Beef on Track signatories are audited for their compliance with Monitoring Protocol criteria, buyers and investors can easily assess the performance of meatpackers on deforestation and human rights. The rigorous criteria and monitoring, reporting, and verification rules of the programme offer a positive outlook for the sustainability of the Brazilian Amazon’s cattle sector.
Development and implementation of Beef on Track has followed the Accountability Framework’s guidance for responsible commodity supply chains. By closing loopholes, expanding the scope of monitoring and verification rules, and strengthening supply chain management, the companies committed to Beef on Track will be better positioned to reduce their impacts on deforestation and conversion in the Brazilian Amazon.
To learn more about the AFi’s work with Beef on Track and other industry initiatives, explore our suite of case studies.
]]>Non-forest natural ecosystems, including grasslands, shrublands, and wetlands, are vitally important to the earth’s diversity and stability. These ecosystems store vast amounts of carbon and provide key services that support human wellbeing and the global economy. However, these ecosystems have been historically overlooked as meriting and requiring protection, and are currently being lost and degraded at an alarming rate, particularly due to agriculture and livestock production.
Core Principle 1 of the Accountability Framework states that companies should commit to eliminating the conversion of all natural ecosystems, including forests and non-forest biomes, from their operations, supply chains and financial investments. In addition, corporate climate targets require companies to quantify conversion of all ecosystem types that may be associated with their businesses.
The Accountability Framework defines natural ecosystems at the global scale, describing attributes of ecosystems that should be protected from conversion for commodity production. But the ways in which that definition can contextualised, interpreted, and operationalised in many parts of the world is not yet clear. Therefore, companies and other stakeholders still need support regarding how to define, identify, and map conversion of non-forest ecosystems in key regions of commodity production and expansion. This will enable them to effectively set no-conversion commitments and account for carbon emissions from land use change, while protecting biodiverse and carbon-rich non-forest ecosystems.
On the road, learning about non-forest conversion
Definitions and data availability related to non-forest natural ecosystems and their conversion vary greatly across the world. In an attempt to decrease that variability, members of the AFi Backbone Team and colleagues at the World Resources Institute (WRI) have been engaging with diverse stakeholders in key regions in order to better understand how non-forest natural ecosystems and their conversion are defined and mapped.
In July and August of 2024, the project team organised sessions at two regional workshops: one held in the city of Chicago, focusing on the United States, hosted by the Land Use Change Initiative, and another in Pirenópolis, Brazil, concentrating on South America, hosted by the Global Pasture Watch consortium.
These workshops brought together diverse actors, including farmers and ranchers as well as representatives from academia, civil society, industry, and government. During facilitated discussions, participants considered the characteristics and thresholds they felt were most appropriate to define non-forest natural and converted ecosystems. They then examined high-resolution Earth imagery and available maps that differentiated land use or land cover classes in their region and suggested classifications for difficult cases.
Unpacking the dynamics of conversion
Distinguishing what should be considered conversion in non-forest ecosystems is difficult. Participants in both workshops shared insights that will enhance the AFi-WRI team’s understanding of how to effectively define and map non-forest ecosystems for supply chain policies. For example, surveys and discussions highlighted the importance of ploughing or tilling in identifying conversion of grassland ecosystems.
The conversations also provided insights into how varying environmental conditions (such as rainfall or fire dynamics) and land use histories (such as planting of exotic grasses) inform the extent to which grazing and pasture management constitute ecosystem conversion. For instance, some high fertility grasslands depend on grazing to prevent woody encroachment and promote seed dispersal, while in others grazing may rapidly lead to degradation and loss of ecosystem function.
The team will hold further discussions with stakeholders to unpack context-specific dynamics and learn more about the capabilities of available datasets to identify conversion. In both regions, the team identified and engaged with ongoing efforts to map natural non-forest ecosystems and their conversion, with potential applications for supply chain contexts.
More stakeholders, more geographies
Building on these workshops, the project team is identifying next steps to both look more closely at key elements of data availability and quality in these two regions, and to hold similar engagement sessions in other regions.
In North America, the team is already beginning new collaborations with local experts to help identify the best available datasets and definitions for conversion of natural non-forest ecosystems.
In South America, the team is continuing conversations with geospatial experts to determine whether additional data development or guidance could improve the ability of companies and other stakeholders to set and monitor supply chain policies in poorly understood grassland ecosystems. Future discussions will centre on other regions of the world where non-forest conversion may be relevant. The next workshop will be held in Australia in early 2025.
Finally, insights and perspectives gathered through these workshops – as well as from literature reviews and ongoing discussions with regional experts – will be synthesised to improve a global definitional framework for non-forest conversion, in alignment with the existing definitions of the Accountability Framework.
Talk to us about conversion
Standardised approaches to mapping and monitoring non-forest ecosystem conversion are only possible with the input and collaboration of experts, practitioners, and stakeholders around the world. If you work on land classification or land use change monitoring, or if you represent a company working to develop no-conversion policies or land use change assessments, then your input can support this project. Contact Sam Levy at slevy@ra.org to get involved!
Non-forest natural ecosystems, such as grasslands, shrublands, and savannahs are vitally important but poorly understood. Compared to forests, these open ecosystems have been historically overlooked and understudied. Despite increasing recognition of their importance to global carbon stocks and biodiversity, these ecosystems are being converted and degraded rapidly, with losses of non-forest ecosystems exceeding that of forests in many countries, including Brazil.
Limited research on how to define, map, and protect non-forest ecosystems makes it hard for companies to identify, quantify, and prevent conversion in their supply chains. This hinders the ability of companies to implement commitments around no-conversion as recommended by Core Principle 1 of the Accountability Framework, as well as to account for and reduce land use change emissions as part of corporate climate target setting.
Gathering cutting-edge science on non-forest conversion
To bring together the newest research on non-forest ecosystems and commodity-driven conversion, AFi Backbone Team (BBT) members, Sam Levy and Leah Samberg, organised two sessions at the Global Land Programme (GLP)’s 5th Open Science Meeting. The GLP is an interdisciplinary network of experts working on science and policy related to land systems and land use change. The meeting ran over 4-8 November 2024, in Oaxaca, Mexico with the overarching goal of improving understanding of the land system and generating science-based solutions for local- to global-scale challenges. Achieving ethical commodity supply chains was a major theme of the meeting.
The two sessions, organised by the AFi BBT, featured experts in ecology, geospatial monitoring, and policy analysis from across the globe. This included scientists working in non-forest conversion hotspots, such as the Brazilian Cerrado, Paraguayan Chaco, and the US Great Plains. Several speakers from the World Resources Institute (WRI) presented new work on mapping and defining non-forest ecosystems. WRI is an AFi Coalition member and key partner in AFi’s ongoing work related to defining and mapping non-forest natural ecosystems.
New datasets illuminate the scale of conversion, but also offer potential solutions
Several speakers presented studies quantifying the scale and location of non-forest natural ecosystem loss due to the expansion of commodities such as maize, soybeans, and livestock. These preliminary results consistently showed high levels of non-forest ecosystem loss, with the area of non-forest ecosystem conversion estimated to be potentially twice the size of the area deforested globally. Key areas were also identified with especially high levels of non-forest ecosystem conversion, such as the Brazilian Cerrado, South American Pampas, North American Great Plains and grasslands, and savannahs of Australia.
Several mapping products were also presented that are intended to help companies and other key stakeholders to identify potential ecosystem conversion in their supply chains. WRI presented the recently published Natural Lands Map V1, which offers an open source platform to identify natural ecosystems present in 2020. This map supports companies setting no-conversion targets via the Science Based Targets Network (SBTN), using the AFi’s recommended cut-off date to identify sourcing areas that are compliant with a no-conversion commitment. WRI also presented work related to Global Pasture Watch, a project to identify natural/semi-natural and cultivated grasslands globally at high spatial resolution. Ongoing work was also presented to identify areas in the USA that have not been plowed in recent history to establish the most ancient and undisturbed grasslands in the country. This will enable greater clarity about which areas are primary or secondary grasslands, helping indicate which areas may be suitable for crop production under a no-conversion commitment. These new projects will make it easier for companies to adopt and implement no-conversion commitments in the future, which will hopefully reduce non-forest ecosystem conversion from commodity production.
Unanswered questions remain
The sessions showcased increased knowledge of the importance of non-forest ecosystems, as well as new products that are making it easier to map conversion within agricultural supply chains. However, it was also clear that defining natural non-forest ecosystems and differentiating them from converted lands remains challenging. This is especially true in boundary cases such as overgrazed natural pastures, intensification of natural grasslands (eg, via fertilisers), and regeneration of abandoned cropland.
Two presentations were given on this topic, both of which are part of the AFi and WRI’s ongoing work related to improving definitions for non-forest ecosystems. This work reflected the diversity of non-forest natural ecosystems and the challenge in consistently defining them for the context of commodity supply chains. While the session participants provided valuable input, it was clear that additional feedback from experts, practitioners, and stakeholders with knowledge of specific non-forest ecosystems is needed. This will help to develop actionable definitions and guidance for commodity sourcing in non-forest natural ecosystems.
Want to get involved? Get in touch!
If you work on land classification or land use change monitoring, or if you represent a company working on a no-conversion policy or land use change assessment, then your input would help address these remaining unanswered questions. If you would like to give input to our ongoing work related to non-forest ecosystem conversion, please contact Sam Levy.
The production of agricultural commodities such as beef cattle, soy, palm oil, or cocoa, drives around two-thirds of all deforestation in the tropics. Between 2001 and 2015, seven ‘forest-risk commodities’ replaced around 72 million hectares of forest globally - an area larger than the Iberian Peninsula. Companies who use these goods in their products have made ‘zero-deforestation commitments’ (ZDCs) to eliminate deforestation from their supply chains in response to NGO pressure. ZDCs therefore seek to convert consumer and civil society pressure, often in locations far from the sites of production, into behavior change that promotes conservation. In doing so, they aim to take advantage of the telecouplings inherent in commodity supply chains that link land use change to consumers across the world. However, ZDCs face a host of challenges, including the difficulty of reshaping opaque commodity markets into transparent and collaborative supplier relationships, tracing complex supply chains, as well as the continued existence of alternative ‘leakage’ markets for unsustainable products. As members of ETH Zurich’s Environmental Policy Lab, led by PI Professor Rachael Garrett, and with funding from the US and Swiss National Science Foundations as well as ETH’s World Food System Center, postdoc Dr. Janina Grabs and PhD candidate Sam Levy study how effective such attempts to govern telecoupled deforestation are in practice. In this blog post, we share some first insights from our work on the effectiveness of zero deforestation commitments in the Brazilian Amazon’s cattle sector and the Indonesian palm oil sector. The blog post summarises our work as presented in a webinar hosted by the GLP Telecoupling Research Working Group in December 2020.
Brazil, and in particular the Amazon region, has a long history of using zero-deforestation commitments to manage telecoupled deforestation, beginning in the soy sector in 2006 and spreading to the cattle sector in 2009. Collective agreements and individual company commitments are widespread in the soy and cattle sectors of Brazil and more than 85% of exported beef and 60% of exported soy are covered by a ZDC. Although cattle ranching has been linked to 80% of forest clearing in the Brazilian Amazon in the past, it is still unclear if the ZDCs implemented to govern cattle-driven deforestation in the region have been effective at changing this relationship.
A key barrier that is likely very important in reducing the effectiveness of ZDCs is the failure of commitments present in the Brazilian Amazon to capture the entire market, particularly in certain regions. There are two widespread ZDCs in the cattle sector of the Brazilian Amazon; the G4 Agreement signed between the four largest slaughterhouse companies in the Amazon and Greenpeace and TAC agreements, a Brazilian Public Prosecutor’s Office (MPF) led campaign to force cattle buying companies to be more sustainable. Collectively the two cattle agreements control the majority of the market, but their facilities are concentrated in some regions. In areas where the market share of ZDC companies is low, producers who wish to avoid ZDCs and continue deforesting land for cattle production can likely do so with little difficulty.
Understanding the true impacts of ZDCs requires an understanding of the actors in the sector, what kind of zero-deforestation policies they have and also their sourcing behavior - both the quantities sourced and the regions sourced from. Through the work of myself, Dr. Federico Cammelli and Professor Rachael Garrett in collaboration with the Gibbs’ Land Use and Environment Lab, we have been able to collect these data to determine the municipal market share of cattle ZDCs in the Brazilian Amazon over time. By using this estimate of ZDC market share as an explanatory variable in first-difference panel regressions, applied at the municipal scale across the key cattle producing states of the Brazilian Amazon, we are able to estimate quantitatively the impact of changes in the market share of ZDC companies on deforestation within the Amazon region. Our preliminary results suggest that where the market share of ZDC companies is higher, forests are better protected and deforestation is reduced. This suggests that although the telecouplings between consumers and producers have been successful at reducing deforestation in the Amazon, further work to strengthen these bonds and increase the pressure on non-committed cattle companies is required to eliminate deforestation from the Brazilian cattle sector.
In the palm oil sector, so-called “No Deforestation, No Peat, No Exploitation” policies have been in effect since around 2011. They now cover a large share of the market, with over three-quarters of the Southeast Asian refining capacity - the most highly concentrated part of the supply chain - covered by such commitments. Interviews with over 50 stakeholders collected in 2020 show that while significant progress has been made in mapping a previously inscrutable supply chain and reacting to the largest transgressions, companies are reaching the limits of their influence, while palm-driven deforestation continues at the margins and in the shadows. Interviewees agree that, in an impressively rapid change of norms, commitment to NDPE practices have become a new standard of doing business in the international palm oil market. Banks have come on board, high-profile cases of supply chain exclusion have shocked large growers into compliance, and technological advances in satellite-assisted monitoring further the perception that it is becoming more difficult to hide.
Yet, while large integrated palm oil companies have adjusted their own practices and deforestation on official palm oil concessions is trending down, deforestation outside of known concessions is increasing. The responsibility for such trends is alternatively attributed to smallholder farmers, absentee landowners, or even land speculators and investors who might be informally linked to larger companies. The continued opacity of on-the-ground land ownership, property boundaries, and actual land occupancy makes effective monitoring and compliance enforcement challenging. In addition, countervailing industrial policies by the Indonesian government, including expanding a biodiesel feed-in mandate and supporting the construction of more in-country refineries, have turned the local palm oil market from a buyers’ to a sellers’ market. This makes it more difficult for international companies to assert effective supply chain pressure, given that Indonesia is itself the largest consumer of domestic palm oil, absorbing one-third of internal production.
Hence, as international telecoupled deforestation might be decreasing, it is time to hone in on more locally driven feedbacks to initial telecoupling connections. These might require radically different governance approaches, including fostering a more fundamental change in local priorities as well as the identification of alternative development pathways. These insights also show that telecoupled drivers of the marketization of land and its use - for instance the EU and US biofuel policies which accelerated oil palm planting in Southeast Asia - are difficult to correct once unleashed, as agro-industrial development creates domestic interest groups and entrenched interests to protect industry in producing countries, including by boosting domestic markets.
There is compelling evidence that complex, telecoupled deforestation systems can be governed by policies such as ZDCs, but that there is still a long way to go to fully achieve this goal. From our two case studies, and the wider body of work on such policies, we see a number of cross-cutting insights that are important to acknowledge and address for improving such policies. First, it will always be possible for some actors to avoid the rules of ZDCs if full traceability of production down to the production base is not achieved. However, achieving this transparency without excluding the most vulnerable informal actors and smallholders is a hitherto unresolved challenge. Second, these policies can also be avoided due to the presence of potential buyers that lack a deforestation policy; it is only where the market share of all ZDC companies collectively is high that we expect a significant effect on deforestation. Indeed, without significant market penetration, including at local scales, opportunities will always exist for producers to avoid ZDCs. Third, under the scenario where ZDC companies do not control the entire local market, positive incentives for compliance are necessary to reduce the motivation of producers to avoid ZDC policies. Yet, incentive-setting in the supply chain for deforestation-free products is still widely underdeveloped, as actors disagree over who should burden such costs. Correctly targeting such incentives is furthermore difficult and risks reinforcing existing inequalities on the ground. Hence, ZDCs are a work in progress - but have still contributed to wide-reaching changes in how commodity supply chains work.
]]>The expansion of commodity production is one of the largest drivers of tropical deforestation across the globe. In response, buying companies have adopted zero-deforestation commitments and other forest focused supply chain policies (FSPs). Such policies hold the potential to reduce deforestation by excluding deforestation-linked products from the market and thereby reducing the incentives to expand production in forested areas. However, in practice, these policies are hard to implement. When they are successful, they may leave behind poorer producers who are less able to comply with no-deforestation restrictions or take advantage of policy loopholes. ETH’s Environmental Policy Lab, led by Prof. Dr. Rachael Garrett, a member of the World Food System Center (WFSC), seeks to understand the effects of these policies at achieving conservation goals and altering human wellbeing in the targeted regions. To understand different perspectives about these policies and how they may affect producers on the ground, we (Federico Cammelli, Sam Levy and Janina Grabs from the Environmental Policy Lab) conducted semi-structured interviews with representatives of groups involved at all stages along the beef cattle and palm oil supply chains in the Brazilian Amazon and Indonesia with the aid of funding from the WFSC.
We conducted our fieldwork in the Brazilian Amazon entirely on the ground before the pandemic began. In Brazil, our goal was to understand the complexities of FSPs in the Brazilian cattle sector, where supply chains are often long and complex – cows usually travel between several farms before they are sent to a slaughterhouse. This required the identification, mapping and interviewing of all formal slaughterhouses sourcing from our study area, as well as tracing producers selling to these slaughterhouses all the way back to their farms (and the farms of the producers they buy from) – making our way across challenging terrain.
Whenever possible we interviewed suppliers and buyers that traded with each other in order to gain different perspectives on the same supply chain relationship. This method allowed for surprising insights. For example, in the remote town of Ulianópolis, many smallholder farmers used to sell for slaughter to a local butcher. More recently, they had switched to a fattener who was laundering (the process of making deforestation-linked production appear to be deforestation free) the smallholders’ cattle on his property pretending it was born and raised there, and then selling it for slaughter on the market regulated by FSPs. The local butcher lamented that he was going to be outcompeted by supermarkets because many other smallholder farmers had largely abandoned cattle ranching and had rented or sold their land (reportedly because they were forced to) to larger cattle and soy operations, the latter seeking arable land on which they could expand production in compliance with FSPs requirement for soy and cattle.
By interviewing many individuals at different stages in the cattle supply chain we were able to capture the different ways farmers can be affected by (and try to avoid) FSPs, indicating that the livelihood impacts of these policies will potentially be affected by where you are in the supply chain. During fieldwork we also collected a large farm-level dataset that we are using to test this and many other potential channels in which FSPs may affect farmers’ livelihood.
As luck would have it, our fieldwork in Indonesia was planned for March to May 2020 – just when Covid-19 became a global concern that shut down international flights and imposed strict stay-at-home orders. While I was lucky enough to meet with my collaborators, make first connections and venture out into the field in East Kalimantan (on the island of Borneo) to meet with palm oil farmers, after 3 weeks my in-country adventure came to an abrupt end as borders started closing and all foreign residents were called home.
This meant that my interviews, like so much else, moved into the virtual realm. Still, thanks to the relationships built in the field, remote-based interviews worked surprisingly well and allowed to capture the truly global nature of supply chains, with interviewees’ home bases covering the United States, Europe, Malaysia, Singapore, as well as Indonesia. Of course, it needs to be said that such methods have a bias against interviewees who can’t be easily reached electronically – which means we are itching to return to fieldwork in order to hear more from smallholders directly, rather than only via their representatives and advocacy groups.
One central insight from these interviews is that FSPs in Indonesia coincided with a period of low palm oil prices and little expansion pressure by large growing companies, which meant that interviewees acknowledged a general trend toward lower deforestation rates that may have been influenced by the supply chain commitments. Still, many plantation companies felt that such policies are unfairly imposed on them, and likened this pressure (much of which comes from European companies) to past experiences of colonialism. As one representative argued, “We are seen as the loser and victim of neo-imperialism, neo-colonialism. In the past they used weapons, now they are using trade”. This is a common view in forest-rich countries of the Global South that also contributes to diplomatic tensions. New framings, such as ‘forest-positive’ commitments and initiatives, currently aim to change this narrative to identify greater common ground between environmental and economic development objectives.
As different as these supply chains and contexts are, we also found that both are struggling to successfully integrate smallholder farmers. In Brazil the FSPs for soy and cattle were potentially weakening the smallholder farmers’ position in two ways: forcing them to launder their cattle before slaughter and lowering their bargaining power and increasing pressure for land concentration. Many smallholder farmers had left, potentially increasing deforestation pressure at the forest frontier and reducing the overall FSP effectiveness. In Indonesia, smallholder farmers often lack both the resources and knowledge to identify the areas on their land that international buyers require to set aside for conservation. This has limited the reach of palm oil FSPs, as the question of who will pay for the capacity building of thousands of mills and millions of smallholder farmers – who account for 40% of Indonesia’s production area – remains unanswered.
]]>The loss of tropical forests is a key driver of global biodiversity loss and a major contributor to climate change. Current forest loss is driven by the expansion of agricultural land for food products like beef, soybeans, and palm oil, all products which are then exported to consumers across the world. A handful of companies control the global trade of these products, and increasing pressure has resulted in them signing policies known as “zero deforestation commitments.” These corporate policies aim to stop deforestation by excluding producers who clear new land through deforestation and incentivizing sustainable practices.
If successful, zero deforestation commitments could reduce deforestation substantially, since the affected food commodities are the largest driver of deforestation in most tropical regions, including the Amazon. But determining whether these policies are actually changing farmers’ decisions to deforest is not so simple. Supply chains are long and complex, and even if a company knows the entire chain, they are private entities with no obligation and little incentive to be transparent. Cattle production in the Amazon is a classic example, with cattle often moving between farms before sale and companies only knowing the farm from which the livestock was bought.
In order to therefore assess if these private zero deforestation commitments are driving change, numerous sources need to be triangulated to pinpoint who sources from where, what is the impact associated with this sourcing, and how farmers are responding to new signals from companies on what behavior is acceptable.
To provide this assessment, the Environmental Policy Lab at ETH Zurich synthesizes and analyzes company policy documents, supply chain transaction records, satellite images of crops and forests, and local household surveys. Sam Levy and Federico Cammelli recently returned from the Brazilian Amazon, where they successfully implemented this assessment approach to understand zero deforestation commitments in the Brazilian cattle sector. They interviewed 307 farmers, who control an area of over one million hectares of land (more than five times the size of canton Zurich). For each farm they visited, satellite data was also examined to understand deforestation patterns on that property.
“What makes the approach so exciting is that we were able to use commodity transport data to actually link farmers to the companies that buy their products and then go and talk to these same farmers,” says Sam Levy. “This means we not only know where zero deforestation commitments are implemented, but how farmers responded to them and why.”
The Environmental Policy Lab believes that understanding why farmers respond the way they do, instead of just how they respond, is both the most important and the hardest part of understanding private environmental policies, including zero deforestation commitments. By understanding why farmers respond either positively or negatively to rules made by companies, researchers can provide better predictions for how such policies will be received when implemented in new commodities or locations.
Rafael Peniche Ferreira, research assistant who helped implement the project and now a doctoral student at the Federal University of Pará, comments “the experience convinced me that the diverse ways in which this research is conducted is key to finding solutions to the issues facing the Amazon region, which are so important due to the scale of the region, the biodiversity of the ecosystems, and the diversity of social groups that make up the biome.”
To further test and refine their work from the Brazilian Amazon, the group will implement their multifaceted approach in the Brazilian Cerrado, Indonesian Borneo and Sumatra, and the West African Guinean Forests in the coming year. By systematically determining the behavior of companies and farmers, across countries and commodities, the researchers hope to provide public assessments that can lead to more sustainable forest policy.
]]>