In September 2025 New Zealand’s Department of Conservation announced it would spend the next two years poisoning rats across the South Island, on account of something the beech trees had not done yet.
The forecast was for a mast — a year when whole forests flower and seed at once, then stop — the largest in seven years. It arrived on schedule. Rodents ate the seed all winter, stoats ate the rodents, and when the seed ran out both turned to the birds. The poisoning is split into two rounds, because while the floor is carpeted in beech seed no rat will touch bait. The mast satiates the poison too.
Notice what that forecast requires: not heavy seeding — that is ordinary — but that they all do it in the same year, across hundreds of kilometres, with nothing passing between them.
Last time the subject was an insect defended by nothing but how many of it arrive at once, and the question was the floor beneath the crowd. Trees run the same economy — a flood too large to eat — but raise the harder question. Cicadas at least count: each nymph tracks its own tree’s seasons underground, and a brood arrives together because every member counts the same thing. A beech tree counts nothing.
It is not agreement. It is obedience.
The rule, in the department’s own words: if the summer just ended was warmer than the one before it, the trees flower in spring and drop seed the following autumn. They call the gap between those summers the DeltaT, and map it across the country every year.
The important word is difference. A hot summer means nothing by itself. What moves the tree is a comparison between this summer and the last — a subtraction, performed separately inside every tree, on two numbers the whole country shares.
That one choice does an extraordinary amount of work.
It makes the alternation automatic. To mast twice running, a forest would need each summer hotter than the one before, indefinitely — so a big year all but forbids the next. You can watch this happen in public: having called 2026 correctly, the department has already published its forecast for 2027, and it is nothing.
It makes the rule immune to a trend. Add two degrees to every summer and the difference between consecutive summers is unchanged. That was the point of the 2013 paper that fixed the pattern across twenty-six datasets: a differential cue makes masting insensitive to climate change. Hold on to that sentence. It does not survive the end of this issue.
And it leaves a forest legible from outside — two thermometer readings and a subtraction.
The private half
The signal is public; the sensitivity is not. As a plant’s stored resources accumulate it becomes more responsive to the cue — a depleted tree sits through a warm summer unmoved, a loaded one is on a hair trigger. Everybody hears the same thing at a different volume.
There is also exactly one real connection between these trees, and it is not a message. It is pollen: a tree that flowers when its neighbours do not has wasted the flowers. Break that coupling and the agreement goes. Among fragmented Californian valley oaks, many with no neighbour of their own species within five hundred metres, eleven years of counting found the isolated trees made fewer acorns — worst in the biggest years — and barely synchronised at all.
The other story — trees wired together underground, coordinating through the fungal network — is not needed here: an audit in 2023 found the claim that mature trees send resources and warnings through those networks had no published evidence.
The famine is the synchronised part
Thirty-six years of European seed records, four hundred sites, seven species: the years of failure are more tightly synchronised across space than the mast years are.
That is the reverse of how the story is usually told, and it is the right way round. Satiation does not depend on one enormous crop. It depends on the seed-eaters being kept few. You may be generous locally. You must be stingy everywhere.
It is coming apart, and not as you would guess
Warming is dismantling this, and not by making crops smaller. Thirty-nine years of English beech records show the opposite: warming raised seed production while flattening the year-to-year variation and pulling individual trees out of step. More seed, more evenly spread — precisely the diet a seed predator wants. Pollination grew less efficient, losses rose, and the extra seed benefits the seed predators, not the trees.
The trees got richer and came out behind. What was working was never the quantity. It was the variance.
What this is not
No. 8 was fireflies, where synchrony comes from coupling — oscillators dragging each other into step. This is nearly the opposite. Take away the connections between these trees and most of the synchrony survives: it was never made between them. It was handed to them.
Nor is it No. 120’s cicada, carrying its clock inside and counting alone for seventeen years. A tree keeps no count. It reads a number off the sky, subtracts last year’s, and acts.
≈
One loop I’m watching
Next: a lake that turns itself over twice a year because water is heaviest at four degrees above freezing — and the fish at the bottom spend the whole summer on one breath drawn last spring. Next time.