What Was Different About Austin's Thirteen?

On 6 October 1859 the clipper Lightning left England carrying a crate of rabbits. Thomas Austin, a settler near Geelong, had asked his family in Baltonsborough, Somerset, to send “a dozen or so” for hunting. Wild rabbits were not common around the village, so his brother managed six half-grown kittens lifted from nests and made up the number with seven grey rabbits bought off villagers who kept them in hutches. Thirteen animals left England; eighty days later, on Christmas Day, twenty-four came ashore in Melbourne. By 1865 Austin was writing to the Geelong Advertiser that he had killed 20,000 on his own property. Their descendants moved outward at roughly 100 km a year and in fifty years covered an area thirteen times the size of the rabbit’s entire native Iberian range — the fastest colonisation by an introduced mammal on record, and still about A$200 million a year off Australian agriculture, twenty-two times the bill for feral pigs.

The part that should bother you is what came before. Peacock and Abbott’s archival trawl (Aust. J. Zool. 2013) turned up close to 300 records of rabbits imported, sold, moved and released in Australia between 1788 and 1900 — more than ninety of them before Barwon Park, at least 30% deliberate releases into the wild. The first was five domestic rabbits off the First Fleet in 1788. Seventy-one years and ninety-odd shipments produced local warrens and nothing more. Thirteen rabbits in a crate produced a continent. Peacock and Abbott’s own explanation is predation: eastern quolls ate the early releases, and the invasion began when quoll numbers collapsed. But if release from predation were the trigger, the Sydney rabbits sitting there since 1788 should have gone off too. They didn’t. They are still a small, genetically distinct pocket carrying domestic mitochondrial ancestry. Something about these particular animals was different.

Alves and colleagues settled the chain of custody in PNAS in 2022: 187 rabbit exomes at 30.5× coverage, 32.1 Mb of target (1.17% of the genome), 1,987,606 variants. Fifty-seven of 62 mainland Australian rabbits fall into a single cluster spanning thousands of kilometres. Genetic distance tracks geographic distance across it (Mantel r = 0.361, P < 0.001); diversity declines monotonically with distance from Barwon Park across sample sets 72, 979, 1,323 and 2,521 km out; and alleles absent from both the British and the domestic source pools rise in frequency the further you get from Victoria, which is about as clean a field demonstration of allele surfing as exists. Mitochondrial genealogies put the maternal ancestry of the whole mainland sample at five introduced females (95% CI: 3–5). When the authors scored which British county’s private variants the Australian rabbits share, the hits were Hampshire, Dorset and Glamorgan, singleton-sharing falling off with distance from Baltonsborough at r = 0.611. The genomes name the village.

Alves et al. read the difference as wild ancestry: hutch rabbits are tame, wild-caught ones are not. I think their own table supports something narrower and stronger, and they undersell it — the difference was admixture. Line up all five populations they genotyped. Sydney: domestic-dominant, the lowest wild-British ancestry in the dataset — stayed local. Cattai, 50 km away: the most British-like rabbit they sequenced, essentially a pure wild-stock release — also stayed local. The invasive mainland lineage: intermediate on both f₃ axes, wild and domestic, exactly what you would get from thirteen mismatched animals breeding in a crate over eighty days at sea. Tasmania: largely domestic, a nuisance only after mainland stock arrived. New Zealand: British plus domestic plus mainland Australian, strongly structured, and a pest. The two ancestrally near-pure populations are the two failures.

There is a mechanism for this and it is specific to rabbits. Carneiro et al.’s domestication genomics found more than 100 selective sweeps in domestic rabbits but very few fixed differences, with the signal concentrated in conserved noncoding sites near brain and neuronal-development genes. Rabbit tameness is a polygenic allele-frequency shift, not a handful of broken genes. Cross wild with domestic, recombine for a few generations, and selection is not handed a blend — it is handed a distribution, spanning wild flight response through domestic stress tolerance and fecundity, from which the Australian rangelands could pick. That is precisely the mechanism Ellstrand and Schierenbeck proposed in 2000 to explain why invasions so often need a lag and multiple introductions. The caveats are real and I will not bury them: Cattai is n = 1, its release date is unrecorded so the timing confound is not dead, and Rius and Darling are right that admixture is frequently costly rather than helpful.

What it would change: biosecurity risk assessment scores species and propagule pressure. Neither would have flagged Austin’s crate — thirteen animals is noise beside ninety prior importations, and the species had a seventy-one-year track record of failing to establish. The variable that separated invasion from failure was the ancestry composition of thirteen founders, and that is now a cheap thing to measure. So here is the falsifiable version: sequence the remaining relict Australian pockets — Kapunda, Melbourne, the other warrens that never went anywhere — and every population that stayed put will sit near an end of the wild–domestic axis, while every one that spread sits in the middle. An admixed pocket that never went anywhere kills it.

Seed: European rabbit, promoted to Wikipedia Featured Article status in 2026.