The lender’s obvious move — change the signal, leave the borrowers behind — is the move the evidence cannot find
R825’s kingsnake is kept by its model’s reputation, and every borrower that prospers spends that reputation down a little. So stand on the lender’s side and ask what the lender should do about it. The coral snake paid for its lesson in full — real venom, taught to real predators at real cost. Now harmless snakes wear its rings and cash its credit, and every tasty one a predator samples teaches the hunter that the rings can mean dinner. The defended animal is being robbed through its own face. Its obvious move is to change its face: shift the pattern somewhere the borrowers are not, stand apart again, and let the lesson it paid for become true once more. Theory gave that move a name generations ago. The strange thing, when someone finally went and looked for it, is that the lender appears never to have made it.
The predicted escape. The idea is old and tidy. Brower & Brower (1972) called the dynamic advergence: in a Batesian pair the traffic is not one-way convergence but a pursuit — the mimic evolves toward the model while the model is selected to evolve away, the pair chasing each other through pattern space. The pressure on the model was later named chase-away selection: a defended species burdened with parasitic look-alikes should keep producing phenotypes at the edge of its own range, extremes that leave the crowd of borrowers behind. It follows from the arithmetic of R825 almost mechanically — if borrowers dilute the signal, the signal’s owner should flee it. Almost everything in this system that theory predicted about the borrower has been found in the field. The flight of the lender is the exception.
The test, in the snake complex itself. Akcali, Kikuchi & Pfennig put the chase-away hypothesis to the very pair R825 borrowed from: the venomous eastern coral snake and its scarlet kingsnake mimics. If any system should show the escape, it is this one — the mimicry is close, the parasitism is measurable, and regions differ in how many mimics a model must carry. They asked two things: whether coral snakes carry the phenotypic signature of flight (patterns displaced away from their mimics, most strongly where mimics press hardest), and whether coral snakes in high-mimic regions are actually under directional or disruptive selection favouring extreme patterns. They found neither. No signature of chase-away in the models’ patterns, and no selection pushing models toward extremes where mimics are common. Their conclusion was carefully flat: chase-away in natural systems may be weak, rare, or of a different character than theory predicted. The lender, in the best-lit case anyone has examined, is standing still.
Why standing still can be the rational move. Several grounded reasons stack up, and each one is a cost the fleeing model pays that the chasing mimic does not. First, the asymmetry Nur pointed to in 1970: selection on the mimic to match is fierce — a poor copy is eaten — while selection on the model to escape is diffuse, because each borrower taxes the model only slightly. Second, the chase is subsidised for the pursuer: in Franks & Noble’s simulations (2004) the mimics close in regardless of where they start, partly because a palatable individual gains simply by mutating away from its own kind’s average pattern, whatever direction that is — the crowd of borrowers spreads, and some of it always lands near wherever the model went. Third, flight buys less than it looks: predators judge patterns coarsely. Kikuchi & Pfennig (2010) found that predators of this same snake pair do not even use the order of the rings — the very feature that distinguishes model from mimic on paper — so a small step sideways, within what a predator generalises, is no escape at all. And fourth, the deepest cost: a model that changes its signal abandons its installed base of taught predators and must pay the teaching price of R824 all over again, in sampled lives, for a lesson it already owned. The borrower spends the lender’s reputation; but the lender who flees spends it too, all at once, on purpose.
Even the empty chair did not move them. A coda from the same research group, and it cuts the same way. Coral snakes vanished from the North Carolina Sandhills around 1960, for reasons unknown. Akcali & Pfennig (2014) compared museum kingsnakes across the forty years that followed, expecting the orphaned mimicry to decay once there was no model left to copy. Instead the Sandhills kingsnakes came to resemble the extinct coral snakes more, not less — red and black bands converging in size — while kingsnakes in the Florida panhandle, where coral snakes remain, showed no such change. Whatever motion exists in this system, it is all on the borrower’s side of the ledger, in both directions: mimics chase, converge, even keep perfecting a copy of a lender that no longer exists. The model’s pattern is the still point everything else orbits.
Where escape did happen, the architecture differed. There is a famous system where the pursued party really did run: brood parasitism. The cuckoo finch lays forged eggs in the nests of tawny-flanked prinias, and Spottiswoode & Stevens documented the hosts escaping — prinia egg appearance shifting measurably over some forty years between museum and modern collections, and, more radically, diversifying: females each lay a consistent individual pattern, so variable across the population that the eggs work like the markings on a banknote, hard to forge because there is no single pattern to forge. A related host that both diversified and discriminated sharply, the rattling cisticola, is reported no longer parasitised at all. But notice what made escape possible there: the prinia is its own receiver. It inspects the eggs itself, against its own memory of its own clutch, so a private signature works. The coral snake has no such channel — its signal is read by a third party, the predator, whose memory is a public register that any new pattern must be re-entered into at full price. Escape by signature is open only to those who grade their own papers.
The chase, as arithmetic. A schematic toy — no empirical parameters, nothing measured. Put model and mimic on a pattern line. Each round the model steps away by a fixed amount, and the mimic closes half of whatever gap remains. Watch what the gap settles at: fleeing does not widen it without limit, it converges to exactly the size of the model’s per-round step. The lender can hold only as much distance as it can move in one generation of the chase.
A red bug is kept by a reputation it paid for. A kingsnake is kept by that same reputation, borrowed. And the coral snake is kept, apparently, by staying exactly where it is: the theory said it should run from its borrowers, the field test found no running, and the reasons are written into the economics of the thing — the chase is cheap for the chaser, coarse in the judge’s eye, and ruinously expensive for a lender whose only asset is a lesson already taught. It does not escape its mimics. It out-stands them.