Every EUDR obligation eventually points at one date. "Deforestation-free" means the relevant commodities were produced on land that has not been subject to deforestation after 31 December 2020, and for wood, that harvesting did not induce forest degradation after that same date. The cutoff predates the regulation itself by years, and that is deliberate. This guide explains the definitions hanging off the date, why it was fixed in the past, and how the test actually runs against a plot.
The definitions doing the work
- Forest: land over half a hectare with trees above five metres and canopy cover over ten percent (or trees able to reach those thresholds), excluding land predominantly under agricultural or urban use. This is the FAO-derived definition, and it is broader than rainforest: dry woodland like the denser Gran Chaco and cerradão formations qualifies.
- Deforestation: conversion of forest to agricultural use, whether or not human-induced. The agricultural qualifier surprises people: forest converted to a reservoir or a mine is not "deforestation" under this regulation (other laws apply), while forest converted to pasture or cropland is, even by wildfire followed by planting.
- Forest degradation: structural conversion of primary or naturally regenerating forest into plantation forest or other wooded land. This limb binds wood products only, and it is why a legal clear-cut-and-replant rotation can still fail EUDR if it converted natural forest to plantation after the cutoff.
Why a date in the past
A cutoff at the regulation's application date would have rewarded pre-deadline clearing: every season of legislative delay would have become a clearing window. Fixing 31 December 2020 (the date international zero-deforestation commitments converged on) froze the baseline before the law's incentives could touch it. The consequence operators must internalise: the application dates moving to 2026 and 2027 changed nothing about the cutoff. Land cleared in 2021 was already permanently out of EU supply chains before most companies had heard of the regulation.
How the test runs against a plot
The claim a DDS makes is testable per plot: this geometry, forest status at end-2020, no conversion to agriculture since. Satellite archives answer it. Screening establishes the end-2020 baseline from tree-cover data, then looks for post-cutoff loss and its successor land use, reading several independent datasets together (the convergence-of-evidence approach described in the risk assessment guide). Three recurring subtleties:
- Tree crops on old land pass. A plot cleared in 1995 that grows coffee today is fine; the test is post-2020 conversion, not the presence of any historical clearing.
- Replanting is not deforestation. Rotating rubber or oil palm reads as "loss" in naive datasets; commodity-aware layers separate replanting from conversion.
- Natural disturbance needs its successor checked. Fire or storm damage is not conversion by itself; fire followed by pasture is. The definition says "whether human-induced or not" about the conversion, so what the land became matters more than how the trees died.
Production dates complete the logic
Geolocation alone cannot prove a consignment clean: the DDS pairs each plot with the date or range of production. Commodities produced on a plot before its post-2020 clearing could in principle be compliant while later harvests are not, and conversely a plot that deforested in 2024 taints everything produced there afterwards, forever. There is no rehabilitation mechanism in the regulation: post-cutoff deforestation permanently disqualifies the land. This permanence is what makes early screening an economic decision, as Brazilian soy buyers discovered first: a flagged plot found before contracting is a sourcing choice, found after loading it is a stranded cargo.
Edge cases worth documenting
Clearing that straddles the date (imagery shows loss "2020-2021") resolves on evidence: alerts, higher-frequency imagery, planting records. Plots smaller than the datasets' resolution warrant boundary polygons rather than points so the analysis has edges to work with. In every ambiguous case the file you keep, per the record-keeping guide, is what converts a defensible judgement into a defended one.