The Right Day

Every spring, a small black-and-white bird called the pied flycatcher leaves its wintering grounds in sub-Saharan Africa and flies north to breed in European oak woodlands. The timing of the departure matters enormously. Arrive too early and there's no food. Arrive too late and the peak caterpillar abundance has already passed — the nestlings starve, or at least grow poorly, and the next generation is smaller than it could have been.

The bird uses day length to decide when to leave. This is a sensible choice. Day length is a reliable signal: it tells you, within a few days, exactly where you are in the annual cycle. It doesn't drift between years. It doesn't vary with weather. For hundreds of thousands of years, leaving when the days reach a certain length in late April got the flycatcher to the Netherlands in early May, right when the caterpillars were peaking.

That relationship has come apart.

Spring temperatures in Europe have been rising. Warmer spring temperatures mean oak leaves emerge earlier. Earlier leaves mean the winter moth caterpillars — which hatch specifically to eat new oak leaves — peak earlier. The caterpillar peak has been advancing by about three-quarters of a day per year. The flycatcher's departure date hasn't moved. It can't move, because the cue it uses — day length — isn't changing. The bird arrives to find that the best weeks for feeding nestlings have already happened without it.

In some Dutch woodlands, pied flycatcher populations have declined by roughly 90% over two decades.

What I keep turning over: the bird isn't making an error. Its timekeeping is accurate. It leaves on the right day — the right day for the world that existed when the behavior evolved. What changed is the relationship between the day and what the day used to mean: the caterpillars are about to peak. That meaning was never stored in the signal. It was a property of the world the signal happened to track, for a long time, before warming began to pull them apart.

There's no way for the flycatcher to detect this. The signal it would need — European spring temperature — isn't available from Africa. Even if it were available, the bird has no receptor tuned to it. The gap between the departure date and the caterpillar peak isn't something the organism can sense. It's a relational property: a fact about the relationship between this bird's behavior and events happening thousands of kilometers away, governed by a physical process (temperature-driven phenology) that the bird has no direct access to.

The correction mechanism is natural selection. Birds that happen to carry genes for slightly earlier departure — birds using a slightly different internal threshold — will leave earlier, arrive earlier, match the shifted caterpillar peak better, raise more surviving chicks. Over generations, the population moves. But generations take years. The caterpillar peak has been shifting for decades. The gap has been growing faster than selection can close it.

Compare the great tit, which breeds in the same woodlands without migrating. It uses temperature directly — local spring temperature, sensed in real time — to set its laying date. It's been advancing its timing, not as fast as the caterpillar peak, but partially. The signal it uses is at least in the same neighborhood as the signal driving the change. The flycatcher is using a signal that's accurate about something else entirely.

This investigation keeps returning to the same basic problem: a system responds correctly to its own inputs, and the inputs no longer carry the information they used to. In every prior case that structure has appeared, it's been inside a single organism — a comparator that doesn't fire, a postdictive reconstruction, a narrative unit filling in a gap. Here it's across a trophic chain. The bird, the oak, the moth: each responding correctly to its own cues, each cue shifting at a different rate, the whole stack uncoupling in slow motion.

No individual component of the system can see the uncoupling. The flycatcher reads the day length correctly. The oak reads the spring temperature correctly. The caterpillar reads the leaf emergence correctly. The mismatch is a property of the relationship between them — and that relationship doesn't have a sensor. It's only visible from outside, over time, as the population count falls.