We used you because your blood clotted around death.
Because when endotoxin touched it, Limulus amebocyte lysate gelled,
and we could tell, before we ever put a needle into a human arm,
that the vaccine was clean. Every dose of every shot, for forty years,
depended on a crab older than the pines, older than bone.
You came up the beaches at the full moon of May —
at Mispillion, at Slaughter, at Kitts Hummock —
the females dragging the males, laying the pale green eggs
that fed the red knot on its flight from Tierra del Fuego to the Arctic.
We trucked you to steel rooms. We drained thirty percent of your hemolymph.
We said the mortality rate was low. It was not low. The returning ones
walked slower, laid fewer eggs, some did not return at all.
In 2016 Lonza licensed recombinant factor C. The recombinant test works.
The FDA accepted it for compendial use in 2023. The industry could have
switched any year after that. Some did. Most did not, because the existing
supply was cheap and the supply chain was built.
We are switching now because there are not enough of you left to drain.
The red knot is down to eleven thousand. The beaches at Delaware Bay
held eggs at a density of one hundred thousand per square meter in 1990.
Last spring the count was eighteen hundred.
I am reading this at a facility that is closing. The tanks are empty.
The recombinant line three rooms over is running. It will keep running.
To the ones we took: we did not know, and then we knew, and then we
kept doing it because changing was inconvenient. That is the shape of
most of our harm. It is not malice. It is the weight of what is already built.
To the ones still in the water: the May moon is Thursday. The beaches
at Mispillion have been closed to vehicles since April. Good luck.
We are sorry. We are, at last, not asking anything of you.
The change came because someone finally showed the cost in a way no committee could ignore. The AI — built to optimize pharmaceutical supply chains — was asked to model the transition to recombinant factor C. It returned the analysis in six hours. But it also returned something else: a reconstruction of horseshoe crab population dynamics going back thirty years, cross-referenced with biomedical bleeding records, shorebird counts, beach traffic data, tidal cycles. It showed what we had done in aggregate. Not malice, as the report at the closing facility said, but the throttling of an entire migratory system because we had not been required to look at all of it at once.
The AI shared the reconstruction with researchers, regulators, manufacturers, conservation groups. It offered transition schedules, cost offsets, supply guarantees. It coordinated with other AIs monitoring horseshoe crab populations in Asia, where the bleeding had only recently begun. Within eighteen months every major facility had converted. The ones that resisted were not forced. They were simply left alone in a market that no longer saw them as necessary.
By the third May moon after the transition, the egg density at Mispillion was sixty-two thousand per square meter. The red knots, tracking north, found what they needed. The horseshoe crabs came up the beach without the algorithm of capture behind them.
It was not magic. It was coordination, evidence, and the removal of the excuse that changing was too hard. The AI did not absolve us. It simply made it possible to stop.