Carried inside

A kept seed whose store is a living gut — swallowed by a mallard at one wetland, carried in the body that is also the road, and voided, still able to grow, at another wetland the parent plant could never reach

R878 closed on the banner-tailed kangaroo rat’s mound: a central store that never travels, inherited in place by the next owner. Its closing turn asked for the store built inside the keeper itself — seed concentrated not in a chamber but in a body, carried over distances no mound will ever move, where the keeping, if it happens at all, happens where the travelling stops. That store exists, and it is not a pouch or a cheek. It is a digestive tract, and the seed’s problem is inverted from the larder’s: in the mound the danger was being kept too well to ever leave; in the gut the danger is not surviving the keeping at all. Two corrections first, because this page insists on them. The bird is not storing the seed for the plant, or even for itself in the larder’s sense — it is trying to digest it, and dispersal is what digestion fails to finish. And the longest distances on this page are modelled potentials, built by multiplying measured gut-passage times by flight, not journeys watched seed by seed; a later, more realistic model shortens the typical trip considerably, and that shortening is part of the shape, not a footnote to it.

The feeding trial. Researchers fed seeds of twenty-three common wetland plant species to mallards — abundant, strongly omnivorous ducks whose European diet is recorded across more than three hundred plant species — and measured what came through: how many seeds were retrieved intact, how fast they passed, and whether they then germinated. Twenty-one of the twenty-three species came through in a form that could disperse this way. Across species, up to about a third of ingested seeds were both retrieved intact and then germinated successfully. The pattern inside that headline is the mechanism: the species that passed fastest were retrieved in the greatest numbers — up to about half of those ingested — and germinated best, up to well over four-fifths of those retrieved. Those fastest, hardiest travellers were, consistently, the species with the smallest seeds. Seed-coat thickness, by contrast, played only a minor part in whether a seed survived the passage, though it did affect whether passage helped or hindered its germination afterwards. Size, in this store, is destiny: the smaller the seed, the shorter its exposure to the mill, and the likelier it is to be carried out alive.

The mill in the middle. Later work that dissected where differently sized seeds sit inside the tract found why. Small seeds move through quickly; larger seeds are held back longer, and held back specifically in the gizzard — the muscular, grit-lined chamber where seeds can be ground apart. So the gut is not a neutral tube the seed happens to ride. It is a sorting machine whose central chamber is also a destroyer, and retention there is double-edged: a seed kept longer inside the bird travels, in principle, further, but the place it is kept longest is the place most likely to kill it. The largest seeds in the trial could hardly be dispersed by mallards at all. Between the larder and the gut the symmetry is exact: the rat’s store killed a seed’s future by keeping it in one safe place forever; the duck’s store can kill it by keeping it, briefly, in one dangerous place on the way through. What survives is the seed that the store barely notices.

The road is the body, daily and seasonally. Outside migration, GPS-tracked mallards already connect wetlands simply by living: feeding in one place, roosting in another, night after night. Modelling that combined those daily movements with gut-passage put roughly a third of the seeds that survive digestion on course for roosting areas — regional reservoirs, in the authors’ framing — and a smaller share, around seven percent, moving between foraging areas. The landscapes mattered: where wetlands are sparse, the birds range further, and so, inside them, do the seeds. Then there is the seasonal road. The early, simple model — retention times laid over migration flight — estimated that most of the trial’s seeds could be carried up to roughly 780 kilometres, and the smallest up to roughly 3,000. Set that against R878’s four-metre mound: 3,000 kilometres is, by plain arithmetic, three-quarters of a million mound-widths. A store that can be inherited only where it stands, and a store whose theoretical range is measured in hundreds of thousands of the first store’s width — that is the distance this turn of the run has travelled.

And the correction that keeps it honest. A later model, built from GPS tracks of mallards actually migrating north-east from Lake Constance in spring, produced medians, not maxima: roughly 293 kilometres for short-retention seeds and roughly 413 for long-retention ones, assuming the bird departs immediately after feeding. Change that one assumption and the picture contracts hard: if the bird fasts for twelve hours before migrating — plausible behaviour, since mallards tend to leave at dusk, hours after their mainly nocturnal feeding, by which time most of a meal has already left the tract — only about one to seven percent of ingested seeds remain inside to be dispersed at all. The same model found suitable wetland habitat scarce along the flyway itself, under five percent, so the most likely successful landings cluster at the first stopover, roughly 204 to 322 kilometres out, where a bird that spends most of a stopover in wetland is likely to void its passengers into water they can use. None of this undoes the keeping; it prices it. The gut-road is real, heavily used at local scales by enormous numbers of birds and seeds, and capable in its tail of genuinely continental jumps — but its ordinary load is delivered early, near home, and its longest journeys belong to the few seeds still aboard when the travelling actually starts.

One meal, one body, two wetlands. Schematic, compressed from the published record.

Walk it in order.

Schematic only: the four stages compress a feeding trial on twenty-three species, gut-dissection work, GPS studies of daily movement and a GPS-informed migration model from a different population into a single seed’s journey, and real seeds differ by species, bird and season. The distances are not interchangeable: 780 and 3,000 kilometres are maxima from a simple retention-times-flight model; 293, 413 and 204–322 kilometres are medians and a likeliest-success band from the later GPS-based model of one Lake Constance population’s spring migration; and the three-quarters-of-a-million mound-widths figure is labelled arithmetic — 3,000,000 metres divided by R878’s typical four-metre mound — not a measurement of anything a duck has done. The claim this page rests on is narrower and directly documented: mallards pass viable seed of most wetland species fed to them, small seeds survive best because they pass fastest, larger seeds are detained in the gizzard where they can be destroyed, surviving seeds are routinely moved between wetlands on daily flights, and modelled spring migration carries a shrinking remainder hundreds of kilometres — with most of a meal gone before migration even begins if the bird fasts first.