The stock with no cold room: a landrace is kept by being sown, harvested and eaten every year — where the keeping and the living are the same act
R836’s genebank accession is kept by being withdrawn from life: frozen, tested, grown out under guard, frozen again, its keepers accepting a little metered change at every burning as the price of still being itself. Here is the older arrangement, which has no freezer, no threshold and no endowment. A landrace — a farmer variety shaped by generations of seed saved from a family’s own harvest, in a particular place — exists only in the years somebody plants it. Skip a year and, for seed that cannot wait, there is no packet to go back to. The archive is a field, and the catalogue is dinner.
A park that is a farm. Near Cusco, in a centre of origin of the potato, six Quechua communities manage the Potato Park: around 9,000 hectares held under customary law as an in-situ reserve, worked since 2000. By traditional classification the communities sustain roughly 1,345 potato varieties (about 650 by scientific classification). Among them are some 410 varieties repatriated in 2004 from the International Potato Centre — varieties collected from that same area in the 1960s and since lost locally to genetic erosion, returned from the cold room to the field, where keeping them means growing and eating them, in rotation across altitude, alongside their wild relatives. The two systems met and each supplied what the other lacks: the genebank had the seed the field had lost; the field gives the returned seed somewhere to go on changing.
Change is not the failure here. That is the deep difference from the accession. A frozen sample is graded on sameness; a planted population is expected to move. Farmers select seed from plants that did well this year, in this soil, under this weather, and exchange lots with family and neighbours — in one Mexican maize study, friends and neighbours accounted for almost a fifth of the introductions and infusions of a traditional variety over farmers’ working lives, so the unit of keeping is not even the household, it is the network. The pressure is visible in the Park itself: rising temperatures and the pests that follow them have pushed the lower line for planting potatoes up by some 200 metres in 30 years. A stock kept by use can answer that by moving and being re-selected as it moves. A stock kept frozen can only be thawed into a world it was not selected for.
Scale, elsewhere. Oaxaca’s farmers maintain 35 of the 59 maize races native to Mexico, in the form of hundreds of farmer-named varieties, with most households at least partly feeding themselves from their own maize. And persistence is not only a story of isolation: in a twelve-year longitudinal study of households at Yaxcabá, in Yucatán, improved maize varieties were introduced and supplied throughout, and the richness of traditional varieties planted, and their evenness across area, stayed the same. Less than a quarter of Mexico’s maize seed was bought from the formal sector in the year that study cites. Farmers kept a portfolio — traditional, locally improved, nationally improved, hybrid — rather than switching as a bloc.
Why they keep them, when they do. The reasons recorded are not sentiment. In Stephen Brush’s classic Andean potato work, farmers treated the new varieties as inferior eating potatoes and kept traditional ones as a culinary necessity for a household whose diet rests on potatoes; native seed could also be kept viable indefinitely by rotating it between altitudinal zones, while improved seed degenerated as saved seed within a few seasons and had to be bought again. Against that, improved varieties yielded more and resisted frost and late blight better. Keeping, here, is a weighed choice remade every planting season, by every household, for reasons of taste, cost, risk and ground — which is exactly why it cannot be endowed once and forgotten.
The threat has a different shape. A freezer fails mechanically and can be alarmed. This stock fails socially, one decision at a time. A recent threat-assessment review for crop landraces lists the pattern plainly: maintainers are often elderly (in the United Kingdom, an average age of 65), the transmission of cultivation knowledge from one generation to the next is weakening, children of maintainers are not taking over, rural people are migrating to cities, seed law and market standards reward uniformity, and some agencies actively promote replacing diverse landraces with uniform modern cultivars. Brush found the gradient decades earlier: replacement above 60 percent in Peru’s commercially developed valleys, much lower — perhaps below 25 percent — away from them, and no high, subsistence area where native varieties had been abandoned completely. The variety does not die when the last seed rots. It dies in the year the last household that would have planted it plants something else, and there may still be seed in a jar in that house, unplanted, already an exhibit.
The toy — schematic, stated plainly. No empirical parameter: suppose each household that keeps a variety has, in any one year, a 1-in-5 chance of stopping — retiring, migrating, switching seed — independent of the others, and nobody new takes it up. Then the variety ends in that year only if every keeper stops at once, with probability 0.2N for N keepers. Independence is the generous assumption, and the page says where it breaks.
Run the numbers and the social floor is as stark as the genebank’s arithmetic one. One keeper: a 20 percent chance the variety ends this year, and about a 99 percent chance it ends within twenty years on those terms. Five independent keepers: the one-year risk falls to 0.032 percent. Distribution is the whole insurance — which is why the seed network, the six communities, the hundreds of named varieties across a state, matter more than any single devoted keeper. But note the assumption doing the work. Keepers do not stop independently when the reason is shared: a new road, a new buyer, a law, a drought, a generation leaving at once. Correlated stopping is the landrace’s version of the freezer failing — the one event the distributed design cannot diversify away, because it arrives at every household in the same season, wearing the same reasons.
So here is the seventh shape, and the mirror of the last. The accession is kept by being taken out of use, and pays for sameness with an endowment. The landrace is kept by being used, pays for its keeping in meals, and is not the same from decade to decade — that drift is not its decay but its method, the only keeping in this whole run of sittings that improves the kept thing’s fit to a changing world while it keeps it. What it cannot survive is not change. It is applause from a distance: being valued, catalogued and praised by people who do not plant it. A landrace has exactly one kind of keeper, and the job cannot be outsourced to an admirer.