The harder half is why. Without knowing what is driving a loss, it is impossible to say whether it is permanent or temporary, what it means for people, nature, and climate, or what would actually keep the forest standing.
New data on Global Forest Watch finally answers that question at scale. And it carries a lesson that extends well beyond conservation: a single forest-loss number tells an investor almost nothing.
What the New Data Shows
Developed through a collaboration between the World Resources Institute and Google DeepMind, the dataset maps the dominant driver of tree cover loss at one-kilometer resolution from 2001 to 2025. It uses an AI model trained on satellite imagery, biophysical data, and population data to classify why each loss occurred — not just that it did.
The top-line finding reframes two decades of forest reporting:
- About 34% of tree cover loss worldwide over that period was likely permanent land-use change — forest converted to something else, with no natural regrowth.
- In tropical primary rainforests, that figure roughly doubles to 60%.
- Of the permanent loss, the overwhelming majority — around 94% — was driven by agriculture, accounting for an area of trees larger than Mongolia.
The remaining two-thirds of loss came from drivers more likely to be temporary: logging cycles, shifting cultivation, wildfire, and natural disturbance.
Same satellite footprint. Radically different meaning.
Why "Forest Loss" as a Single Number Misleads
This is the point that should matter most to anyone allocating capital.
Two hectares of tree cover loss can register identically from orbit and mean opposite things on the ground. One can be a managed Swedish timber stand that will be replanted and harvested again on a cycle. The other can be primary Amazonian rainforest cleared for pasture that will never return. The pixel is the same. The financial, regulatory, and ecological reality is not.
This is the structural problem Resōno has written about across every domain of environmental data: aggregation hides signal. When you collapse forest loss into one figure — a percentage, a hectare count, a single deforestation score — you erase exactly the information that makes it actionable.
The new data does the opposite. It disaggregates. And in doing so, it turns a headline into intelligence.
The Driver Is the Signal
What makes this dataset valuable is that the drivers vary enormously by place, and each driver carries a different meaning for risk.
- In Latin America and Southeast Asia, permanent agriculture dominates — 72% and 65% of loss respectively — the kind of conversion most exposed to supply-chain scrutiny and deforestation regulation.
- In much of Africa, shifting cultivation leads, a largely temporary practice with a very different long-term profile, though one that becomes permanent when it pushes into untouched primary forest.
- Across Russia and the Asian mainland, wildfire is the leading driver, increasingly amplified by a warming, drying climate.
- In Europe, logging accounts for roughly 90% of loss — overwhelmingly managed harvest cycles rather than conversion.
A holding exposed to permanent agricultural conversion in Latin America sits squarely in the path of demand-side rules like the EU Deforestation Regulation. A holding tied to managed European timber does not. An investor working from a single regional loss figure cannot tell these apart. An investor working from driver-level data can.
The number tells you something happened. The driver tells you what it costs.
What This Means for Investors
Land-use change is one of the core dimensions of nature-related financial risk, and forest conversion is its sharpest expression. The new data makes that risk legible in a way it has never been before:
Operational exposure
Supply chains that run through regions of permanent conversion face input volatility and reputational scrutiny that managed-harvest regions do not.
Regulatory exposure
Deforestation-linked rules apply to specific drivers and geographies, not to forest loss in general.
Transition exposure
Assets dependent on conversion that is becoming ecologically and legally untenable carry stranding risk that an aggregate score cannot surface.
Granular, spatially explicit driver data is what separates a narrative ("deforestation is a problem") from a decision ("this position carries regulatory and supply-chain exposure to permanent conversion in a specific place, and this one does not").
That is the difference between reporting and intelligence.
From Forest Data to Decision-Grade Intelligence
A dataset like this is a foundation, not a finished answer. On its own it is a remarkable map. It becomes investment-grade only when it is interpreted relative to a portfolio — linked to holdings, weighted by materiality, traced to its source, and surfaced in a form an analyst can act on and an auditor can verify.
That translation is the work Resōno does. We integrate spatial forest and land-use data into nature intelligence that is traceable, comparable across assets, and aligned with how capital decisions are actually made — so the question stops being "how much forest was lost" and becomes "what does this loss mean for what I hold." Resōno does not ask investors to become ecologists. It gives them the tools to see what the satellite already recorded but the headline number threw away.
Looking Ahead
The science of forest monitoring is moving from measuring loss to explaining it. Finance will have to follow.
As nature-related disclosure matures and deforestation regulation tightens, the tolerance for portfolio-level estimates built on aggregate forest figures will fall the same way tolerance for opaque ESG scores has fallen. The investors who gain an edge will be the ones who can read the driver, not just the number — and who built the data infrastructure to do it before they were required to.
Forest loss was never a single story. The data finally proves it. The advantage now belongs to whoever can read the difference.