Feeding an isolated New Zealand: the land, the fuel, and what it would actually take.

Feeding all 5.1 million New Zealanders for a year — meeting the full energy and protein need with one food alone — takes less land than you'd guess. The wheat pathway needs just 24% of existing grain land. The dairy pathway needs only 37% of the land dairy already farms. Used optimally, the study finds New Zealand's existing arable land could theoretically feed 43.2 million people — more than eight times its population — and 16.8 million even in a severe nuclear winter.

These are minimums — survival arithmetic on energy and protein for one food at a time (wheat OR potatoes OR dairy, never added together). Real life wants variety, and variety costs more land and fuel. §3.1 and Table 3 of the paper.

All of that land is farmed with diesel — and New Zealand makes almost none. The country consumes 3.7 billion litres a year, nearly all imported. The mandated national reserve holds 213 million litres: about 21 days of normal use (28 days counting government-procured storage). Here is the good news hiding inside that alarming number: feeding the nation takes almost none of it. The minimum food-production diesel is 5.4 million litres a year on the wheat pathway — about 0.15% of what the country burns. Rationed strictly for food, the existing reserves alone could stretch two years or more. Fuel is only one of the shortages an isolated country would juggle — but it is one we can measure precisely, and the measurement is hopeful.

Consumption, reserve and off-road figures from Table 2 (EECA 2021; MBIE 2022); the 2+ years and 0.15% from §4.2. "Consumption", not "supply" — the paper's own wording.

To brew that minimum food-fuel at home, forever: about 4,600 hectares of canola — 0.9% of existing grain land, six and a half times today's tiny canola crop — and refining capacity roughly equal to one plant New Zealand already operates commercially (Pure Oil NZ in Rolleston processes ~14 million litres' worth a year). Even the severe nuclear-winter case needs about 35,000 ha and 5–17 million litres — still a rounding error against the country's cropland. The paper's conclusion names the pieces plainly: wheat and canola seed, urban-adjacent arable land, harvesting and processing infrastructure, and biofuel refining should be treated as strategic national assets — before they're needed, not after.

Areas, seed tonnage, refinery comparison and "strategic national assets" wording from §4.1–4.2 and Table 6. The recursive fuel-loop is counted once — ≈+5% at baseline, rising in severe scenarios (a first approximation, as in the paper, p.7).

This study deliberately measures a slice: the land to grow the food, and the fuel to farm and move it as far as processing. A real catastrophe would strain much more, and none of it is modelled here: every other claim on the same diesel — health, freight, emergency services — competing for the reserves; spare parts and machinery that are also imported; fertiliser and agrichemicals (the paper flags these); processing energy and distribution beyond the factory gate; and the many non-food faces of isolation. Solving the slice does not solve the catastrophe.

But that is exactly why the slice matters: on the piece we can quantify with published science behind it, the arithmetic is promising and the preparation is small. That is not a solution — it is evidence that one hard piece of a hard problem is tractable, and an invitation to quantify the rest.

Everything above uses the study's published figures, unchanged. But the whole model is open — and it scales from the nation down to a town.