The NZ pastoral sheep system, properly modelled.

New Zealand exports more sheep meat than any other country on earth — half a billion lambs through Alliance, Silver Fern Farms and AFFCO every year. The system runs almost entirely on pasture, predominantly Romney and Coopworth maternal genetics, on hill country where land cost and pasture growth are the binding constraints. Number of Lambs Weaned per ewe joined drives everything else.

Why NLW is the load-bearing axis.

In nearly every other sheep system on earth, growth, carcass yield, or wool quality dominate the selection objective. In New Zealand the dominant trait is Number of Lambs Weaned per ewe joined — NLW. The reason is structural: most NZ ewes wean two lambs and a fraction wean three. Moving the herd average from 1.45 to 1.65 lambs weaned per ewe is a 14% revenue uplift on a roughly fixed cost base. Nothing else in the trait list comes close to that on a per-ewe-year economic axis.

The closed-form bioeconomic derivation reflects this. The NLW component of the weight derivation is:

On a typical Romney hill-country operation, this resolves to roughly NZ$95-120 per additional lamb weaned. That number is the load-bearing input to every other trait weight; the platform's production-system modifier "nz_pastoral_sheep" automatically captures it.

What's different about New Zealand.

• Wool is residual revenue, not the business. Most NZ sheep are crossbred or shedding (Wiltshire, Wiltipoll); wool returns ~NZ$3-8/ewe/year on a coarse crossbred clip. On a $150/ewe gross-margin operation, that's noise. The platform's production-system modifier "nz_pastoral_sheep" zeroes wool quality weights and lowers fleece weight weight by 80%.
• Hill country drives the cost structure. NZ sheep predominantly run on steep, undulating hill country where pasture growth is highly seasonal and supplementary feed is largely impractical. Mature ewe size matters more than in flatland systems because every extra kg of ewe weight is fed across a 12-month grazing year.
• Beef + Lamb NZ Genetics is the breed-evaluation pipeline. Two flagship evaluation systems coexist: SIL (Sheep Improvement Limited) carries the pedigree-and-performance database; B+L NZ Genetics performs the BLUP evaluation and publishes the trait BVs. The platform reads both natively (Tier 1 ingestors).
• The processor grid is simpler than US/EU. Most lambs are sold on a per-head basis with a discount/premium for fat depth (GR), conformation grade (Y / P / T), and carcass weight banding. There's no marbling axis; minimal premium tiering.

Trait terminology, side by side.

The NZ trait names are mostly English-based but the canonical mapping for cross-country reading:

Two structural points worth flagging. NZ and AU share most ASBV/BV trait codes (a legacy of Sheep Genetics covering trans-Tasman work), but the underlying reference populations and BLUP calibrations are different — a Romney ASBV in AU is not the same as a Romney BV in NZ. Second, UK Signet uses a different trait nomenclature with no exact NLW equivalent; the translator handles the semantic gap.

What that looks like on a real-shape Romney operation.

The numbers below are from a real-shape Hawke's Bay hill-country operation — 3,800 mixed-age Romney/Coopworth ewes, 1,150 ewe hoggets, ram team of 22 (mix of Romney maternal sires and Suffolk/Texel terminal sires), 1.62 lambs weaned per ewe joined, all lambs sold on Silver Fern Farms PrimeAdvance schedule.

The pattern is what the closed-loop calibration loop is designed to find. The producer's specific contract economics, their hill-country cost structure, and their realised slaughter outcomes shift the trait weights away from the national average — but in interpretable, defensible directions. The same Romney bull catalogue feeds through this re-weighting and produces a different top-10 list than it would on a national-average operation.

One trait does most of the work. The other six do real work too. Industry indexes either nail the first part and miss the second, or get the second right and dilute the first. The per-farm engine doesn't have to choose.

The terminal-sire overlay.

Most NZ commercial sheep operations run a split mob strategy: maternal ewes mated to a maternal ram team to produce replacements (Romney × Romney, Coopworth × Coopworth), and the remainder of the ewe flock joined to terminal sires (Suffolk, Texel, Poll Dorset, Hampshire) to produce slaughter progeny. The terminal-sire animals' offspring are sold; no daughters retained.

The platform exposes this as a distinct production-system modifier — nz_terminal_sire — that re-weights the bull's index to ignore maternal traits entirely. A Suffolk ram joined over Romney ewes has its NLW BV zeroed (he's never measured for it; his daughters aren't retained), and the carcass-trait weights amplified. The setup-flow asks the producer what proportion of their ewe flock goes maternal vs terminal, and the engine derives the right mix.

What's wired natively for NZ sheep.

• B+L NZ Genetics BVs — full trait set from the official B+L NZ Genetics feed.
• SIL pedigree — Sheep Improvement Limited's pedigree-and-performance database for inbreeding control and mate allocation.
• AgriHQ daily lamb and mutton prices — NZ farmgate indicator, North and South Island, in NZD/kg cwt.
• Silver Fern Farms / Alliance / AFFCO processor schedules — branded-program premiums where the producer is contracted.
• NIWA climate — National Institute of Water and Atmospheric Research, daily temperature, rainfall, soil moisture per producer's nearest station.
• NZ ETS carbon price — pulled into the methane economics layer (see the methane economics piece — NZ has had agriculture in the ETS since 2025).

Where the platform can go further for NZ producers.

• Methane EBV ingestion. B+L NZ Genetics is the most active publisher of methane EBVs globally — the platform reads them natively today. The frontier is composite multi-trait methane indices that weight methane-efficiency alongside conventional productivity. The composite-index work on the 2027 horizon directly serves this.
• Hill-country pasture-growth models. The platform reads NIWA climate today; the next step is integrating regional hill-country pasture-growth models (Mead, Sutherland, Massey University research) so the engine can project feed-deficit windows by farm location.
• Cross-Tasman ASBV conversion. Many NZ studs sell into AU and vice versa. The platform handles the semantic translation via the AI translator today; native ASBV ↔ NZ BV conversion tables are on the Q4 2026 roadmap.