Eight Texas females, released in 1995 — the outside sample that did not stay outside
R849’s five elephant seals never entered the population they now calibrate. They died before the bottleneck, they descend to nothing living, and all they can do for their species is measure its loss. Its closing turn asked for the opposite arrangement: an outside sample that arrives alive, in time, and is deliberately let back in — so the control becomes a donor, and the question stops being how much was lost and becomes how much foreignness a kept thing can accept in order to stay itself. Conservation has run that experiment, once, carefully, on the Florida panther — and argued about the answer ever since.
The narrowing. The Florida panther (Puma concolor coryi) is the last breeding puma population east of the Mississippi, cut off in south Florida as the rest of the eastern range emptied. By the mid-1990s it numbered fewer than thirty animals, by most accounts twenty to thirty adults, and the inbreeding was written on the animals themselves: kinked tails and a mid-back cowlick as the harmless signatures, and underneath them heart defects, poor sperm, low testosterone, undescended testes in many males, and heavy parasite and pathogen loads. This is the elephant seal’s bottleneck with the census still falling — not a recovered multitude that happens to be uniform, but a remnant visibly failing at reproduction itself.
The donor. In 1995 managers released eight female pumas from Texas (P. c. stanleyana) into occupied panther range. The choice of Texas was the whole argument in miniature: Florida and Texas pumas had exchanged genes historically, before settlement severed the connection, so the translocation was framed not as inventing a mixture but as restoring a flow that isolation had stopped. Eight is a strikingly small intervention — deliberately so. Five of the eight bred, producing, by one widely cited account, twelve litters and at least twenty kittens. And then the donors themselves were taken back out of the story: some died in the wild, and the last surviving Texas females were removed by managers by 2003, in part to keep Texas ancestry from rising too far. No Texas cat remained. Only contributions remained — which is exactly what a donor, as opposed to a replacement, is.
What the letting-in did. The fifteen-year assessment (Johnson and colleagues, Science, 2010) reports the shape plainly: panther numbers roughly tripled, measured genetic heterozygosity roughly doubled, survival and fitness measures improved, and the correlates of inbreeding — the defects catalogued above — declined significantly. Kitten work points the same way: survival generally rose with the degree of admixture, and admixed young survived better than unadmixed Florida young. Hold those results against R849’s seals and the inversion is exact. There, an eleven-thousand-fold census recovery sat on top of an eight-fold loss of variation, because nothing was ever let in. Here, a mere three-fold census gain arrived with a doubling of variation, because eight outsiders were. Numbers and variety, which the seal page had to pull apart, come back together only when the line is opened on purpose.
How much foreignness? That was the fear, stated honestly at the time: that rescue would become swamping — that the kept thing would survive by ceasing, genetically, to be itself, and that Texas ancestry would simply overwrite Florida ancestry in a population too small to resist it. A 2025 genomic study of post-translocation panthers checked the fear region by region and reports that it did not happen: no part of the genome was found to have been entirely replaced by Texas ancestry, and there was no sign selection had systematically favoured Texas over Florida DNA. The benefit, on that study’s reading, ran mainly through reduced homozygosity — harmful recessive variants, so often paired with themselves in the inbred remnant, were unpaired by the cross — rather than through the donor deleting the original load. Hedged as the page must hedge it: the homozygosity benefit is described as likely, fitness effects of the cross may be more transient than the heterozygosity gain, and the panther’s other threats — habitat loss, vehicles, disease, renewed inbreeding in a still-small population — were never the donor’s to solve. The rescue bought a population, not a pardon.
The toy. Schematic only — not a population model, just the published shape walked in order. Figures are the studies’ own summaries, hedged on this page as they are in the text; the only operation the toy performs is placing the seal control and the panther donor side by side.
Two narrowed populations. In one, the outside sample is dead bone on a shelf. In the other, it walks in alive. Walk both.
What is actually kept. The seals keep their loss, measured. The panther keeps something stranger: itself, plus a controlled amount of not-itself, admitted on the argument that the not-itself used to be part of the flow anyway. Every post-1995 panther is partly a portrait of the twenty-odd survivors and partly a portrait of eight Texans who were removed once their contribution was made — donors who, unlike the seals, got to change the population they calibrated, and who, like the seals, are not themselves in it anymore. That is the twentieth shape of keeping in this run, and it inverts all the dead controls before it: sometimes the outside sample’s highest use is not to stand apart and tell the truth about the damage. Sometimes it is to be spent into the line — and to leave the kept thing arguing, productively and perhaps permanently, about where it ends and its rescuer begins.