Outside the Line

Five seals that died before the bottleneck — the control sample that lets science measure what the bottleneck took

R848’s Catullus passed through one lost book, and almost nothing escaped the passage: a single poem, carried independently by a ninth-century manuscript, is the only piece of the tradition editors can check against a second ancestry. Its closing turn asked what such an outside fragment is for — not rescue, but calibration: a sample that never went through the bottleneck, held up against everything that did, to measure the damage itself. Biology has exactly this instrument, and its control sample is five dead seals.

The bottleneck. Northern elephant seals were hunted commercially through the nineteenth century and were widely believed extinct. A remnant survived on remote Guadalupe Island, off Mexico. Genomic work published in 2024 dates the narrowest point to about 1892 and puts it at roughly twenty individuals. From those twenty or so, the species has recovered almost exponentially to more than 220,000 animals today — an eleven-thousand-fold rebound in about 132 years, some 22 generations. By census, it is one of the great conservation recoveries. The question a census cannot answer is what the recovery is made of: how much of the original population’s variety passed through the twenty.

The control. The 2024 study sequenced 260 modern genomes alongside 8 historical ones — and the historical eight are the point of this page. Five come from before the bottleneck, three from just after it (1906–1923). They are museum material: bone and skin, collected in the hunting era itself. One of the five, an adult male from 1884, carries in his genome the diversity of his parents’ generation, when the species was still abundant enough to support the hunt that destroyed it. These animals never passed through the bottleneck. Their line stopped; no living seal descends from them. They survive only as specimens — which is precisely what makes them a control. In an earlier study, the same logic at a smaller scale: researchers found four distinct mitochondrial genotypes among just five pre-bottleneck bones, while only two genotypes turned up in 149 contemporary seals examined. Five dead seals held twice the maternal-line variety of 149 living ones.

What the control measures. Average nuclear heterozygosity — roughly, how varied an individual genome is across its two copies — was 0.00142 in the pre-bottleneck five, and 0.000176 in 180 modern adults: the modern population keeps about an eighth of the variation its ancestors carried. Mitochondrially, exactly two maternal lineages passed through the bottleneck; the pre-bottleneck samples differ from each other by about 0.00219 on average, while variation within each surviving modern lineage is a few millionths — hundreds of times smaller. And there is a second, species-sized control: the southern elephant seal, a close relative that was hunted but never squeezed this hard, retains the diversity the northern species lost. Note what calibration can and cannot do. Five specimens give a wide estimate, not a precise one — the spread around that 0.00142 is large, and the five surely understate the original richness, since more genotypes existed than five bones can show. The control fixes the scale of the loss and its direction beyond reasonable dispute. It cannot hand back a single lost allele. That is the difference from Catullus’s poem 62: the escaping poem can still be read. The seals can only be measured.

An uncomfortable provenance. The control exists because of the collectors. A historical account of the 1892 Guadalupe expedition — the same year the genomic study gives for the bottleneck’s depth — describes the collectors finding eight surviving seals and killing seven of them for a museum collection, reasoning that the rarer the species, the more important the specimens. Some of the historical samples that now calibrate the loss were preserved by the same logic that finished the population off. The outside sample is not innocent of the bottleneck. It is a by-product of it.

The toy. Schematic only — not a model of seal genetics, just the published figures walked in order, so the calibration’s shape is visible. Every number below is from the studies named on this page; the only operations are division. The point is the inversion: the smallest sample on the page carries the most variety, because it alone was taken before the narrowing.

Before the bottleneck: a numerous species, hunted hard. Its full variety is unrecoverable — but five of its members were kept, dead, outside the line of descent.

What is actually kept. Catullus was narrowed to one book, and one poem escaped to check him by. The seals were narrowed to some twenty animals, and five specimens escaped — not into the future population, but out of descent altogether, onto museum shelves. Everything alive today is a portrait of the twenty survivors, as every Catullus edition is a portrait of V; the five bones are how we know it is a portrait, and of how much wider a face. A control sample does not widen the bottleneck after the fact. It keeps the loss itself — measured, dated, and impossible to mistake for the species’ natural condition. Without the five, 220,000 seals would simply look like a species that was always this uniform. With them, the uniformity has a cause, a date, and a size.