A kept seed carried out of the scatter and into one defended pile — the banner-tailed kangaroo rat’s mound larder in the Chihuahuan Desert, where seeds from a whole foraging ground lie together, deep and crowded, and almost none of them ever grow where they lie
R877 closed on the black palm and its agoutis: a seed planted again and again because every thief was itself a scatter-hoarder, re-burying its loot one seed to a hole. Its closing turn asked for the opposite thief — a concentrator: a pilferer that carries what it takes not to a scatter of small holes but to one central store, where seeds pile up far too crowded and far too deep to grow, so that any keeping could only happen at the store’s ruin. That store exists, and it has been excavated, mapped and cultured. But two corrections first, because this page insists on them. The concentrator is mostly not a thief of other animals’ caches at all: it harvests the open seed shadow for itself, pouchload by pouchload, and the pilfering in this record runs the other way — at the pile, by the scatter-hoarder next door. And no single dramatic spill, no flood or collapse that scattered a larder back into plantings in one event, is documented in the sources used here. The larder’s exits are slower and less generous than the opening imagined, and stating that plainly is part of the shape.
Two neighbours, two geometries. In the same desert, Merriam’s kangaroo rat does what the last two pages celebrated: it scatter-hoards, small clumps of seed in numerous shallow holes dug across its foraging ground. Its larger neighbour, the banner-tailed kangaroo rat, does the reverse. It is solitary and strongly territorial, living alone in a large, semi-permanent mound — typically about four metres across and thirty centimetres high, a labyrinth of tunnels and chambers on up to four levels reaching more than ninety centimetres down — and it larder-hoards substantial quantities of seed inside that mound, defending both mound and store vigorously. New mounds are rarely built in an established population; a mound is instead occupied by subsequent generations, a house and a pantry inherited together. The contrast is a genuine partition of one landscape: the same seed rain, split between a thousand shallow hiding places nobody can guard and one deep pile that one animal guards with its life.
Inside the pile. When researchers mapped where the banner-tail actually puts its caches, the pattern was precise: most caches sat about a third of the way from the mound’s centre to its edge — on a four-metre mound, a ring roughly two-thirds of a metre out — favouring the north and northwest side, at depths mostly near thirty centimetres or below fifty. Set that beside the nutcracker’s one-to-three-centimetre caches of R876: the bottom of this store lies something like seventeen times deeper than the deepest of those plantings, in soil no seedling of these small desert seeds could climb out of. Nor is depth the only seal. A burrow larder is warm and moist compared with a surface pit, and mycological analysis of these caches found at least twenty-three species of fungi in them, colonists most abundant and diverse in the caches made mid-season, when seed was plentiful. The store is not inert; it is a crowded, damp, living dark, and the keeping in it is done by microbes as much as by the rat. Tellingly, over a four-month watch the seeds stored early and late were significantly drawn down, but the mid-season caches — laid in at the peak of abundance — were not depleted at all: at the top of the year, even the owner does not get back to the bottom of its own pile.
For the seed, concentration is close to a terminus. The general verdict in this literature is blunt: larder-hoarded seeds are unlikely to germinate, and to the extent that piling seed in a defended burrow works for the rodent, it works against the plant. A scatter cache is a wager spread thin — lose ninety-nine holes, win one. A larder is the wager consolidated into a single vault, and the vault’s whole design — depth, crowding, one vigilant owner, a resident flora of fungi — is aimed at making sure nothing leaves except through the owner’s mouth. Where the last two pages found trees planted by surplus and by theft, this page finds the anti-case: a keeping so good at keeping that it keeps the future out.
The exits that do exist. Two, on this record, and neither is the ruin the opening pictured. First, the breach: a concentrated store is a standing temptation, and the cost side of larder-hoarding, as the literature lists it, is precisely that it attracts other rodents that steal. A century ago the first life-history of this species already suspected — its authors said the point was hard to prove — that Merriam’s kangaroo rats, in whose burrows no food store was ever found, were visiting the banner-tails’ mounds to pilfer, and trappers on the mounds caught a large share of Merriam’s. Whatever seeds a pilferer carries out and scatter-hoards in shallow holes have been returned, by theft, to the only geometry in which they can grow: the larder’s leak is the scatter’s supply. Second, the succession: when an owner dies, the mound is rarely left empty for the rain to sort out. It is taken over, by the next generation of the same species, store and all — so the pile passes mouth to mouth down a line of owners, an inheritance eaten rather than spilled. The ruin, when it finally comes, comes as abandonment and slow erosion of an emptied house, not as a granary burst open over the desert. In the terms of this run, it is a keeping whose concentration defeats its dispersal: the seed is kept the way money is kept in a vault, and a vault, unlike a field, grows nothing.
One seed shadow, two geometries, and a pile that keeps too well. Schematic, compressed from the published record.
Walk it in order.
Schematic only: the four stages compress two species, a century of life-history work, one cache-mapping study and a general principle into a single seed’s journey, and real larders differ — both kangaroo rats in fact use both strategies at times, the banner-tail predominantly larder-hoarding and Merriam’s predominantly scatter-hoarding rather than each keeping to one geometry absolutely. The arithmetic on this page is labelled as such: the cache ring at about two-thirds of a metre is the published one-third-of-the-way placement applied to a typical four-metre mound, and the roughly seventeen-fold depth contrast is fifty centimetres divided by the nutcracker’s three-centimetre maximum, not a measured germination threshold for these seeds. The claim this page rests on is narrower and directly documented: the banner-tail larder-hoards substantial quantities inside a defended, inherited mound; its caches sit mostly near thirty centimetres or below fifty, carry a rich fungal flora, and can go undepleted even by their owner across a season of abundance; larder-hoarded seed in general is unlikely to germinate; and pilferage out of such stores, long suspected in this very species pair, is the documented route by which concentrated seed re-enters the scatter. No single-event spill of a banner-tail larder into a stand of seedlings was found in these sources, and this page does not invent one.