Report — Field Note
Greener, and Slower to Recover
For twenty-four years, satellites have watched the land over the Arctic's frozen ground turn greener. A new study finds that since 2008 the same vegetation has been recovering more slowly from the knocks it takes.
Permafrost is ground that stays frozen all year round, across much of the far north and the high Tibetan Plateau. Since 2000, satellites have recorded the vegetation above it growing greener. A team led from the University of Hong Kong took those records, cut into roughly two million squares five kilometers on a side, and asked a second question. Greenness tells you how much is growing. It does not tell you how well that growth comes back when something knocks it down.
Sourced · Guo et al., Nature Plants (3 Nov 2026): MODIS satellite records 2000–2023; 1.97 million 5-km squaresThe paper defines resilience as speed of return. A drought, a fire, or a bad season knocks the vegetation away from its usual state; resilience is how quickly it comes back. Picture a ball resting in a bowl. Push it and it rolls back to the bottom. The steeper the bowl, the faster it returns.
Sourced · Guo et al., Nature Plants (3 Nov 2026) (definition) · the bowl is the standard picture in this fieldNo one can watch every knock across two million squares, so the team measured something easier: how much each reading of greenness resembles the one before it. A system that recovers fast shakes off small shocks quickly, so one reading says little about the next. A system that recovers slowly carries each shock forward, and its readings start to look alike. Statisticians call that resemblance autocorrelation. Here it works as a measure of memory, and more memory means slower recovery.
Sourced · Guo et al., Nature Plants (3 Nov 2026): temporal autocorrelation of 16-day greenness readings, used as a stand-in for resilienceAs a bowl gets shallower, the ball takes longer to roll back, and the readings grow more alike. Ecologists call this critical slowing down and treat it as an early warning: a system that recovers ever more slowly may be nearing a point where one push sends it somewhere else. The team first removed the effects of seasons and weather, so that a longer growing season would not be mistaken for memory.
Sourced · Guo et al., Nature Plants (3 Nov 2026) (critical slowing down; early-warning reading; seasonal and climate adjustment)From 2000 to 2008, memory fell, which means recovery was getting faster. After 2008 it rose, and recovery slowed, through 2023. Both trends were strong. The turn appeared in every vegetation type and every permafrost zone, sharpest in eastern Siberia and on the eastern Tibetan Plateau and weaker in North America. Through all of it, the land kept getting greener.
Sourced · Guo et al., Nature Plants (3 Nov 2026), Fig. 1 and text · chart shape is schematicWhat changed around 2008? The closest match was in the air. Warmer air can hold more water vapor, and the amount it could hold kept climbing with temperature. The amount it actually held rose more slowly after 2008. The gap between the two, called vapor pressure deficit, is a measure of how drying the air is, and it widened quickly after 2008. The authors call it the leading statistical correlate. They suggest changes in ocean evaporation helped set the timing. Neither is shown to be the cause.
Sourced · Guo et al., Nature Plants (3 Nov 2026), Fig. 2 and Discussion · chart shape is schematicFire made things worse locally. Burned ground had lower resilience than matched unburned ground nearby, but both showed the same turn in 2008, so fire did not drive the wider reversal. The damage compounded: the more often a place burned, the lower its resilience. The authors suggest repeated fires may hold land in arrested succession, an early stage of regrowth that never reaches the mature state the fires interrupted.
Sourced · Guo et al., Nature Plants (3 Nov 2026), Fig. 3 (paired burned and unburned squares; fire recurrence)The authors tested fifteen of the computer models used to project how vegetation will change. Two reproduced the turn. As resilience fell, the carbon taken up by the plants each year also grew more variable. The authors' suggested reading is that the vegetation was damping out ordinary swings in the weather less and passing more of them on. They conclude that current projections probably overestimate how stable this northern carbon store will be.
Sourced · Guo et al., Nature Plants (3 Nov 2026), Figs. 4–5 (15 TRENDY models; OCN and SDGVM reproduced the reversal) · lines are schematicThe limits are the authors' own. Weather records are thin at high latitudes, where stations are sparse. Ground towers that measure carbon exchange directly support the satellite pattern, but there are too few with records on both sides of 2008 to test its timing on their own. Drying air is the strongest match the team found, and they present it as an association.
Two measures have come apart. Greenness says how much is growing; resilience says how fast it returns after a shock. For two decades the first rose while the second fell after 2008, and the one that kept rising is the one a satellite map shows at a glance. Repeated fire shows the other half of the picture: ground can stay in one condition for a long time because each disturbance arrives before it can leave. A green map is a record of growth, not of how much that growth can take. The term to search is critical slowing down.
Sourced — Guo, Wang, Smith, Li, Zhang, Wu, Zhou & Gong, "Widespread reversal of vegetation resilience in permafrost ecosystems", Nature Plants (3 Nov 2026). Charts and landscapes in the diagram are schematic; directions, dates and counts are the paper's.