Thanks to satellite data, we know how much forest the world is losing, and where. But that’s only part of the story.
Unless we know what’s driving tree cover loss, it’s impossible to know if it’s permanent or temporary; what the impacts are for people, nature and climate; and the solutions to keep forests standing. That’s where new data comes in.
Developed as part of a collaboration between WRI and Google DeepMind and available on Global Forest Watch, the new data provides a more detailed picture than ever before on the local, regional and global causes of tree cover loss. It reveals that 34% of tree cover losses worldwide from 2001-2024 were likely the result of permanent land use change, meaning trees won’t grow back naturally. This percentage nearly doubles in tropical primary rainforests, with 61% of loss likely associated with permanent land use change.
However, the drivers of tree cover loss — and their long-term impacts on forests — vary widely by region and require different solutions. Better understanding them can help aid in the conservation of these critical ecosystems.
Different Drivers of Tree Cover Loss Have Different Impacts
Not all types of tree cover loss are deforestation, or the permanent conversion of forests to other land uses. Some loss can be temporary, though the time it takes for forests to regrow and their condition after regeneration may vary.
About the Data
Researchers at Land & Carbon Lab, Global Forest Watch (GFW), and Google DeepMind developed a new data set — available on GFW — that maps the dominant drivers of forest loss at 1 kilometer resolution from 2001-2024. The new data was developed using an advanced AI model that uses satellite imagery and additional biophysical and population data to predict the dominant driver of tree cover loss. The higher resolution of the new data and the addition of more driver classes, like hard commodities and other natural disturbances, makes this data set the most detailed view of what’s causing forest loss yet. Learn more from Global Forest Watch.
Our analysis found that drivers of tree cover loss likely to cause deforestation — which include “permanent agriculture,” or the removal of tree cover for agricultural activities; “hard commodities,” like mining for minerals or metals and energy infrastructure; or development of settlements and infrastructure — accounted for 34% (177 million hectares) of all tree cover loss globally from 2001-2024. Approximately 95% of this was permanent agriculture, which was associated with the loss of 168 million hectares of trees from 2001-2024, an area of land larger than Mongolia. In tropical primary rainforests, specifically, drivers likely to cause deforestation accounted for 50.7 million hectares, an area nearly the size of Thailand.
Drivers more likely to cause temporary loss include logging, such as cyclical harvesting in timber, pulp, or wood fiber plantations and clearcut or selective logging of natural forests ; shifting cultivation, a type of rotational agriculture where forests are temporarily cleared for cultivation and then abandoned to allow regeneration; wildfires; and natural disturbances like landslides or insect damage. These drivers of temporary loss accounted for 66% of total tree cover loss, or 338 million hectares.
Deforestation often has more severe impacts compared to temporary disturbances, including permanent loss of carbon stocks, profound habitat disruption and loss of ecosystem services. But the impacts of temporary forest disturbances from both human and natural causes can vary widely. Although forests may regrow following these disturbances, they may experience degradation or changes to forest structure and species composition. For example, logging — particularly in primary or old-growth forests — can lead to biodiversity and carbon losses. Some natural disturbances that cause tree cover loss, such as when trees are knocked down due to storms, changing rivers or landslides, can be cyclical features of a forest’s ecology and have ecological benefits. But when these events are more extreme, they can profoundly alter the condition of the ecosystem.
And the lines are blurry. In many cases, climate change has contributed to increasing the extent, frequency or severity of many “natural” disturbance events, including wildfires and pest outbreaks, compromising forests’ overall condition.
Some Drivers Have Outsized Impacts in Specific Locations
Other drivers such as hard commodities, settlements and infrastructure, and other natural disturbances represent a very small proportion of tree cover loss globally, but are important drivers in certain regions.
For example, while hard commodities — which include artisanal to large-scale mining and energy infrastructure like oil drilling — comprise slightly less than 1% of all tree cover loss globally, they’re an important driver in places like Peru, Latin America’s largest gold producer. Both legal and illegal artisanal and small-scale gold mining are widespread throughout the country and can cause long-lasting and acute impacts, especially in Indigenous and local communities. For example, in Madre de Dios, Peru, tree cover loss due to hard commodities comprised 28% of all tree cover loss from 2001-2024 (112,000 hectares), mainly in biodiverse tropical primary forests. Gold mining is one of the top economic activities in this region, but has also negatively affected people’s health due to mercury exposure.
While other natural disturbances such as droughts, floods and pests represent only 1.4% of all tree cover loss globally, they can have a substantial impact on forests in certain places.
For example, bark beetles are a native insect in North American conifer forests. However, over the past three decades, severe outbreaks fueled by climate change have occurred across North America, Europe and Russia, threatening the health of these regions’ forests. In the United States, Colorado’s forests have been extensively impacted by bark beetle outbreaks, with natural disturbances representing 27% of all tree cover loss there from 2001-2024 (140,000 hectares). While forests can recover following pest outbreaks, research finds that when wildfires occur in the first few years after a severe outbreak, conifers may not be able to recover, shifting forests’ species composition as they become dominated by different tree species that are able to successfully regrow.
Different Drivers of Forest Loss Require Different Solutions
Because the natural and human management dynamics behind the drivers of tree cover loss differ across regions, there is no single solution to eliminate deforestation or degradation and sustainably manage the world’s forests. However, this new data provides accurate, spatially detailed and globally consistent information to support policymakers, land managers, researchers and others in identifying the causes of disturbances and the most appropriate interventions.
The most effective mix of policies and management interventions will vary according to the local context:
- Permanent agriculture: Voluntary corporate commitments and demand-side regulations, such as the E.U. Deforestation Regulation, can play an important role in reducing deforestation associated with international supply chains. However, strengthening forest and land-use governance, including granting property rights to Indigenous communities, is also critical, particularly when deforestation is tied with domestic demand for agricultural goods, land speculation, or land tenure insecurity and conflict. When tree cover loss is associated with permanent smallholder agriculture, policies should support vulnerable farmers and their livelihoods.
- Shifting cultivation: While shifting cultivation is often a temporary forest disturbance, the environmental and livelihood impacts of this practice are highly dependent on context. Policies should balance food security with forest conservation objectives, carefully considering the social, economic and environmental context in these landscapes, as well as the potential impact of alternative management systems.
- Wildfires: Managing wildfires and mitigating wildfire risk in a changing climate can involve a variety of strategies depending on the type of forest ecosystem, including fuel management, wildfire monitoring, land and fire management practices, improving capacity to respond to wildfires, and more.
- Logging: In managed forests or timber plantations, managing rotation cycles and tree species diversity can improve health and resilience of these forests and enhance carbon sequestration. In primary forests and those with high conservation value, establishing protections and preventing illegal logging within existing protected areas through enforcement or traceability systems can help ensure these areas are protected for years to come.
- Hard commodities: Protecting forests and nearby communities from expansion of hard commodities requires a number of strategies, including establishing land and resource rights, monitoring and enforcement, and demand management, among others.
- Settlements and Infrastructure: Managing or preventing tree cover loss associated with settlements and infrastructure expansion will require consideration of forests and trees in land use and urban planning, ensuring that residents can benefit from the ecosystem services that trees provide.
- Other natural disturbances: Ensuring long-term forest health after a natural disturbance will be highly dependent on the type of event. Interventions to reduce the risk of insect outbreaks or diseases and/or promoting natural recovery or restoration following disturbances are some options.
It is also important to consider that local land use dynamics are shaped by global market forces. In a globalized economy, sustainable management of the world’s resources as a whole is crucial for our collective future. This new data provides a sharper picture of our progress toward global goals to end deforestation — but to meet them, we must effectively address the underlying causes of tree cover loss.
This dataset has been updated since the original publication to include tree cover loss from 2023-2024.
The authors would like to acknowledge The Sustainability Consortium, who contributed to the early stages of this research.
Data visualization by Sara Staedicke.
Footnotes
1 The wildfire class includes tree cover loss due to fire with no visible human conversion or agricultural activity afterwards. Forest clearing for agriculture that involves the burning of vegetation are included under the relevant agricultural class (permanent agriculture or shifting cultivation). ‘Runaway’ or ‘escaped’ fires that are started as part of the process to clear vegetation for agriculture but spread into surrounding forests that are not cleared for agriculture are included as wildfire.