Perennial grains

An annual cereal has one reproductive event and no future to fund. It can move nearly everything it has fixed into grain, and a century of breeding has pushed that fraction — the harvest index — to roughly half of above-ground biomass in modern wheat. A perennial cannot do this. It must finish the season with enough carbon, nitrogen and structure in crowns, rhizomes and roots to survive dormancy and regrow, and that reserve is subtracted from seed before anything else is decided.

The trade-off is real, not an artefact of poor breeding#

This is the honest version of the argument. It is often said that perennial grains yield little only because they have had decades of selection rather than millennia, and there is truth in that: the crops in question are recently domesticated from wild species, with all the wild traits still attached — shattering, small seeds, uneven maturity, low seed set, self-incompatibility and awkward polyploid genomes.

But an allocation constraint sits underneath the domestication gap and does not disappear with more cycles. Selection for grain and selection for persistence pull in opposite directions, and stands that are pushed hard toward seed yield tend to thin and fail earlier. A perennial grain therefore has a ceiling set by how much reserve the stand needs to come back, and the breeding target is not maximum yield but the best yield compatible with survival — a constrained optimisation, which is a different and slower problem than a straight selection response.

Where Kernza stands#

Intermediate wheatgrass, sold as Kernza, is the most developed perennial grain and the reference case. Grain yields in field trials are a small fraction of annual wheat on the same land, seed weight is several times lower than a wheat kernel, and — the point most often left out — yield falls after the first two harvest years as the stand ages and tillers compete, so multi-year averages are worse than first-year figures. The agronomic case for it does not rest on grain: continuous living roots reduce erosion, capture nitrate that would otherwise leach in the months when an annual field is bare, and extend the period of photosynthetic capture. Those are the benefits to argue from.

Perennial sorghum, from crosses with rhizomatous Sorghum halepense, illustrates a different hazard: the rhizome that provides perenniality is also what makes johnsongrass one of the world’s worst weeds, so the trait cannot be introgressed without addressing weediness.

The exception worth naming#

Perennial rice is the case that shows the constraint is quantitative rather than absolute. Lines derived from crosses between cultivated rice and the rhizomatous wild species Oryza longistaminata, developed in Yunnan and released in China, produce successive harvests from one planting at yields comparable to annually replanted rice, with the labour and seed cost of establishment paid once. Performance still declines over several years and the stand must eventually be renewed. It works where it works because rice is grown in a system that already suits regrowth — and because the wild parent supplied a functioning rhizome, which no amount of selection within cultivated wheat can produce. This is a decades-long breeding programme in every case, not an engineering target.