The Delta Farms Worked Plan and Its Upper Bound

#One hundred hectares, and a number you should not put in a loan application

Here is the arithmetic for a real property: 100 hectares at Meyerton, in the Midvaal Local Municipality of Sedibeng, Gauteng. Grassland Biome, Highveld, summer rainfall, hard frosty winter, the Vaal system nearby and the country's densest concentration of meat demand up the road.

The answer, worked below, is roughly fourteen and a half large stock units, about nine breeding cows, about R46 700 a year, about R467 a hectare.

That number is a ceiling — not a forecast, not a target, not a base case a good year improves on. Every assumption underneath it was made in the farm's favour, and several inputs are still unsettled. A farmer who leaves with the number and not the reasoning has taken the one thing on the page least likely to be true.

#Three site facts nobody has settled

Rainfall. Two commercial climate portals give Meyerton about 520 mm a year and about 732 mm a year. That is not a rounding difference. It is roughly a 40% spread, and it straddles the sourveld boundary at around 650 mm. Both are model output — the lower one reanalysis weighted heavily to Vereeniging — not gauge records, and nobody has pulled the South African Weather Service station record. The course frame assumed 600–700 mm. The mean is not the number that matters anyway: variability, not average, is what breaks South African farms.

The vegetation unit. The property sits in the Mesic Highveld Grassland bioregion, most likely Soweto Highveld Grassland, with Rand Highveld Grassland and Vaal-Vet Sandy Grassland as live alternatives and an alluvial or wetland unit possible near the river. Nobody has checked that against the SANBI vegetation layer for the farm's coordinates. It matters because the unit sets the benchmark your veld condition is scored against.

The CARA polygon. Grazing capacity is a number held on a topo-cadastral map at the executive officer's office under Regulation 10(1) of the Conservation of Agricultural Resources Act 43 of 1983 (CARA). Nobody has read the value for this cadastral unit. Gauteng is reported to run 4 to 20 ha/LSU — and a 4 ha/LSU polygon and a 20 ha/LSU polygon are both "Gauteng", five times apart.

#The base case, worked

Nine cows is not a business. A great deal of peri-urban livestock enthusiasm survives only by never doing this sum. The question is not how to make nine cows work; it is what 100 ha at Meyerton must become before it earns anything — more output per hectare, a different species mix, more of the value chain, or non-grazing enterprises on the same ground.

#What the base case borrowed and cannot pay back

The R5 186.69 per cow is lifted from a budget for a 1 000-LSU, 625-cow operation. Three of its lines do not scale down, and they are the three that matter most on a smallholding.

• Fixed labour: two workers spread over 625 cows — R185.25 per cow. Nine cows still need someone to check water, move the mob, handle at the crush and be there at calving. No version of that job costs R1 667 a year.
• Veterinary: two visits a year for the whole herd — R6.40 per cow. One call-out to a peri-urban smallholding costs more than that budget's entire annual vet line.
• Fuel: 4 000 km a year at R48 per cow. The bakkie does not get cheaper because you own fewer animals.

And the budget treats the capital as someone else's problem — handling facility, water system, fence, crush, all for nine cows.

#Winter is the enterprise

Strip the Land Bank beef budget down and one line dominates: winter lick is 55.2% of total variable cost, and bought supplement of all kinds is 75.5%. Fencing and water, which farmers argue about endlessly, are under 1% as an annual repair line.

In this farm's units: at R934.81 per LSU of winter lick over 6.8 hectares per LSU, the lick bill is about R137 per hectare per year against a gross margin of about R470 per hectare. Nearly thirty cents in every rand of margin leaves in one line, in one season, to fix one problem — the protein collapse on Highveld sourveld from roughly May to September, when the standing crop is abundant and nearly worthless.

That is the economic argument for regenerative management here, and it needs no claim about carbon. Every practice that carries quality forage into July — a deferred camp held as standing hay, a small planted block taking winter pressure off the veld, calving timed so peak cow demand meets spring growth — attacks the biggest number on the income statement. The fodder-flow shortfall is the enterprise. The grazing chart is downstream of it.

Carbon runs the same way. At AgriCarbon's own optimistic terms — 2 tonnes per hectare per year, from US$10 a credit — the gross is about R330 per hectare per year, on an assumed R16.50 to the dollar that must be re-priced before anyone uses it. Not trivial: about 2.4 times the winter lick bill, roughly 70% of the whole gross margin. But it is gross, before aggregator share, soil sampling, verification, registry fees and buffer-pool deductions; it is a claim, not a measured yield on this farm; it pays after a monitoring period is verified rather than annually; and it commits the land use for decades. One South African grassland project's issued volumes imply closer to 0.9 t/ha/yr. And the organisation with the longest track record in market-linked communal grazing here has earned R134 million from market access against R2.7 million from carbon over a decade. Build the plan on the first number.

#The peri-urban operating context

Market proximity is real. Gauteng concentrates the country's meat demand and its logistics, so direct-to-consumer distribution that is impossible from the Northern Cape is merely hard here. But the route is legally narrow: under the Meat Safety Act 40 of 2000 no meat may be sold unless the animal was slaughtered at a registered abattoir. You sell live, or toll-slaughter and sell abattoir-slaughtered meat. The binding unknown is which abattoir near Meyerton takes small lots on a schedule, and at what cost per head — the most valuable missing number for this farm, three phone calls away.

Disease density is the peri-urban tax. Gauteng carried 227 open FMD outbreaks at 31 March 2026, from a SAT2 introduction in April 2025 and a separate SAT1 event at a Gauteng feedlot that October. Only state veterinary services may vaccinate. African swine fever and notifiable avian influenza are controlled diseases in their own right under the Animal Diseases Act 35 of 1984, so a pig or poultry enterprise on these hectares inherits a second control regime rather than sharing the first. Treat movement restriction as the base case, and ask the state vet what applies today.

Stock theft is a cost line, not a risk. Published per-head values from an Eastern Cape study across 2018–2024 run at R12 876.35 for cattle, R2 021.90 for goats and R1 675.10 for sheep — recorded values well below 2026 replacement cost. Recovery ran at 17.2% of animals and about 7% of value; cattle were 14.7% of animals stolen and most of the money lost. A single high-density mob in one small camp is a concentrated target — an argument against naive high-density grazing that appears in no imported curriculum.

Zoning and encroachment. Confirm with Midvaal Local Municipality whether agriculture, intensive animal farming and any farm-stall or processing use are permitted, consent or prohibited before building anything. And watch for Seriphium plumosum: in Mesic Highveld Grassland, dense stands at 10 000 or more individuals per hectare can cut grass production by up to 75%, and the plant is allelopathic and avoided by stock. The same study found the intermediate density, around 1 500 per hectare, carried the highest plant diversity — so the goal is balance, not eradication.

#Why this is a ceiling, and which three facts move it

Every assumption leans generous. The capacity figure assumes veld in relatively good condition at a 40–50% utilisation factor, mapped at 1:250 000, excluding tree crown cover and browsing capacity. The per-cow margin comes from a herd a hundred times larger. The income side is at 2023/24 prices with no theft, no drought, no excess mortality, no movement restriction. Land, interest and drawings are absent.

Three site facts would move that ceiling further than anything a farmer could do in year one.

1. The CARA polygon value for this cadastral unit. Between 4 and 20 ha/LSU the answer ranges from about 25 LSU to about 5. Nothing else moves it that far, and this costs a visit to the executive officer.
2. A South African Weather Service gauge record for the nearest long-run station, with its variability. The 520-to-732 mm spread is the difference between a mixed-veld property and a sour one, and it sets the fodder-flow gap you must bridge.
3. A veld condition survey against the benchmark for the correct vegetation unit. The map value assumes good condition. Below benchmark, plan below the map; genuinely above it, Reg 10(3) is the route to a farm-specific determination, substantiated by a survey from a registered pasture scientist.

Notice what is not on that list: breed, grazing system, and every argument the internet has about density. All downstream of those three numbers.

The farm that comes out of this course is not the one in the brochure. It is a hundred hectares with an unverified capacity, an unresolved rainfall figure, a winter that eats a third of the margin and a district with 227 open FMD outbreaks — and a farmer who knows it on day one instead of in year three.