Two Ledgers: Grazing and Browse

#Over-grazed and under-browsed on the same afternoon

A game ranch runs an intensive breeding scenario at 80% conception and ends the season at 234 large stock units on a property whose grass ceiling is 200 LSU. The same herd is comfortably inside the property's browse ceiling.

That farm is 17% overstocked and simultaneously under-utilised, and no single number on a spreadsheet will show you that. Sell 15% of the herd across the board and you fix the grass problem by throwing away browse capacity you already paid to fence. The correct answer was a species-selective destock, and you can only see it if you were keeping two sets of books.

That scenario is from Shepstone and colleagues (2025) in the South African Journal of Animal Science, and it is the reason this lesson exists.

#The units, and where they come from

South African grazing arithmetic is built on the Large Stock Unit: an animal of 450 kg gaining 0.5 kg/day on forage of 55% digestible energy, requiring 75 MJ of metabolisable energy per day (ARC). Grazing capacity is expressed as ha/LSU — the area needed to carry one LSU indefinitely — so a bigger number means poorer veld.

Shepstone et al. extended that same metabolisable-energy logic to wildlife, which is what makes the two ledgers commensurable instead of a matter of opinion.

South African specifics

The four units, all on one energy currency:

Unit	What it is	Energy
LSU — Large Stock Unit	450 kg steer, +500 g/day, 55% digestible forage	75 MJ ME/day
WHU — Wild Herbivore Unit	the wildlife unit the paper sets alongside the LSU	29.71 MJ ME/day
GU — Grazer Unit	a mature 180 kg blue wildebeest	grass ledger
BU — Browser Unit	a mature 180 kg greater kudu	browse ledger

Notice that both reference animals weigh 180 kg. The units differ by feeding class, not by body mass — which is the whole design. And notice the ratio hiding in the first two rows: one LSU is roughly 2.5 wildlife handling units, so game arithmetic and cattle arithmetic will not translate by eye.

#The 2 000 hectare worked example

Take the paper's own property. 2 000 ha, with a grazing capacity of 200 LSU and a browsing capacity of 200 BU. Because a grazer unit is a smaller animal than a large stock unit, those 200 LSU convert to 400 GU.

Now read the three capacities off it:

10 ha/LSU on the grass ledger, in cattle terms
5 ha/GU on the grass ledger, in game terms
10 ha/BU on the browse ledger

Same farm. Same hectares. Three numbers, all correct, and each answering a different question. Fill the grass ledger with cattle and you have 200 animals' worth of demand; fill it with blue wildebeest and you have 400. The browse ledger is untouched by either.

Then the paper puts a 460 kg Cape eland on the property, and the tidy separation breaks:

1.02 LSU · 2.57 WHU · 1.28 GU and 1.28 BU

That single animal draws 1.28 grazer units off the grass ledger and 1.28 browser units off the browse ledger at the same time, because a mixed feeder eats from both. It is not half a grazer and half a browser. It is a full draw on each.

Work the consequence through on the 2 000 ha. Run 50 eland and you have spent 64 GU of a 400 GU grass ledger — 16% — while also spending 64 BU of a 200 BU browse ledger, which is 32%. The browse ledger runs out at roughly 156 eland; the grass ledger, if eland were the only animal, would not run out until about 312. Whichever ledger empties first is your real capacity, and for a mixed feeder that is almost always the browse column. A manager tracking only ha/LSU would have looked at 50 eland on 2 000 ha and seen a nearly empty farm.

Do this on the farm

Run the ledgers like this, in this order, every planning year:

Get your grass capacity from the map, then discount it. The DALRRD long-term grazing capacity map, published under Regulation 10 of the Conservation of Agricultural Resources Act 43 of 1983 (CARA), is the legal starting number, in ha/LSU. Biome means are Grassland 6, Savanna 12, Nama-Karoo 25 ha/LSU (DALRRD). Discount for your own veld condition.
Do not use that number for browse. The DALRRD map expressly excludes tree crown cover and browsing capacity. It is a grass map. Your browse ledger has to be estimated separately, from your own woody cover.
Convert every animal on the property into both columns. Cattle, sheep, goats, horses, donkeys and resident game all go in. A grazer scores in the grass column only; a browser in the browse column only; a mixed feeder scores a full entry in both.
Compare each column against its own ceiling. Two answers, not one.
Where a column is over, destock that column's species — not the herd across the board.

#Does adding game create capacity, or only revenue?

This is where imported "multi-species means more carrying capacity" material either holds or falls over, and the honest answer is conditional.

The rule: complementarity only creates capacity where a feeding class is genuinely unused. Adding kudu to a cattle operation in Limpopo bushveld can add real production, because the browse ledger was sitting untouched. Adding blesbok to a cattle farm on Highveld grassland mostly spends the same grass ledger twice, in a slightly different-shaped animal.

Contested — here is both sides

Before teaching that rule as absolute, steelman the other side. Three qualifications that are genuinely true:

One — feeding class is a coarse bin. A blesbok and a cow are both "grazers", and they do not take the identical bite. They differ in sward height preference, tuft selectivity and tolerance of mature grass. Some real gain is available from within-class partitioning, and the metabolisable-energy ledger does not capture any of it. But no South African study quantifying that gain on Highveld grassland could be found, and the effect is smaller than the marketing suggests.

Two — grassland is not zero-browse. Drainage lines, kloofs, termitaria and encroached patches carry woody material even inside the Grassland Biome. The browse ledger there is small, not nil — and on a bush-encroached property it may be substantial, in which case a browser stops being a competitor and starts being a management tool.

Three — income diversification is not carrying capacity. Game can raise the revenue earned per grazing unit without adding a single kilogram of forage to the farm. That is a perfectly good business reason and a bad ecological one. Say which claim you are making. Most arguments about game on livestock farms are two people making different claims at each other.

#One wrong feeding class wrecks the whole ledger

The arithmetic above is only as good as the species classification underneath it, and this is where an honest course has to correct itself in public.

An earlier draft of this course's own research listed springbok as a grazer. It is not. The species account treats springbok as a mixed feeder, browse-leaning in arid systems, taking shrubs, forbs and young succulents and selecting flowers, seeds and leaves for their moisture (Cain, Krausman & Germaine 2004, as summarised in the species account; the primary account could not be opened, so no grass:browse percentage is quoted here).

Follow that error through a ledger. Classify 200 springbok as grazers and they appear in one column. Classify them correctly and they appear, at some rate, in both — so the browse ledger you thought was empty is being drawn down, and the grass ledger you thought was full is being drawn down too. Two ceilings move at once, in opposite directions from what you planned.

On the Highveld, correcting the ecology changes the reason rather than the conclusion. Springbok still compete with cattle for grass, and the thin Highveld browse ledger still cannot carry the browse half of their diet, so they remain a poor fit — but now for a reason you can defend rather than one that happens to be wrong. Blesbok is a genuine short-grass grazer; black wildebeest is too, and needs open grassland and drinks daily; gemsbok is predominantly a grazer with a meaningful browse, fruit and tuber component in arid systems.

Careful — this one bites

Per-species LSU, GU and BU substitution values for springbok, blesbok, black wildebeest, gemsbok, impala, zebra and red hartebeest sit in Appendices 1 and 2 of Shepstone et al. (2025), rendered as images that could not be extracted for this course. Do not quote a per-species substitution value from memory or from a neighbour. Take it from that paper's appendices or from the substitution-ratio literature in the African Journal of Range & Forage Science.

Buffalo change the legal character of the whole property and are not a ledger problem at all. They fall under a DALRRD Buffalo Veterinary Procedural Notice made under the Animal Diseases Act 35 of 1984, requiring disease-status certification, testing, permits and quarantine. That notice is a scanned document whose testing intervals and controlled-disease list could not be read for this course. If you are not prepared to run a state-vet-audited disease programme, do not buy buffalo.

#The revenue side, and what it does not tell you

Game earns from hunting fees, live sales and meat. The best public price data on the hunting side is the North-West University TREES survey of local biltong hunters.

What it costs, what it earns

Average prices paid by the hunter in the 2022/3 season (NWU TREES 2023): springbok R1 238, impala R1 705, black wildebeest R3 240, blue wildebeest R4 238, blesbok R1 967, zebra R5 093, gemsbok R5 619, kudu R5 879, eland R10 105, buffalo R58 329.

Three health warnings, all of which matter more than the numbers:

This is what the hunter paid, not what the landowner banked. An outfitted hunt splits that fee. It is gross revenue and it is not a margin.
The species table is survey-respondent totals — respondents accounted for 23 333 animals across all species — not national offtake. It describes who answered the survey.
Game prices are the most volatile animal prices in South African agriculture. Auction and trophy analysis over 1991–2022 shows a sharp plains-game rise in 2015–2017 followed by a decline from 2018, with kudu up more than 1 000% between 2012 and 2015 and Cape buffalo trophy prices swinging +89%, −64% and −41% in different single years (Muller et al. 2025). A business case built on a 2015 price was already wrong by 2019.

And now the gap that no amount of arithmetic closes: no published South African gross-margin budget per game species per hectare could be located for this course. That is the single biggest economics hole in this whole domain. Every rand above is gross revenue or a trade quote; none of it has been costed against a real farm's fence maintenance, capture, veterinary, permit, marketing and outfitter share. Nor is the scale of the sector settled — Agribook records roughly 20 million ha under wildlife ranching and gives no ranch count at all, while other peer-reviewed figures put it at over 9 000 farms on more than 10 million hectares. Those two differ by a factor of two and both are cited seriously.

#What to do on Monday

Open a spreadsheet with two columns, not one, and put every animal on the place into whichever columns it actually eats from — including the horses, the donkeys and the game that came with the farm. Then compare each column against its own ceiling and see which one runs out first. If the answer is that grass runs out at 234 against a 200 ceiling while browse sits half empty, you do not have a stocking problem. You have a species mix problem, and the fix costs you a truckload of the wrong animal rather than a fifth of your herd.

#Check yourself

3 questions — answers explained as you go

1Your 2 000 ha property has a grazing capacity of 200 LSU and a browsing capacity of 200 BU. You buy 100 Cape eland at roughly 460 kg. What have you spent?
Why: A 460 kg Cape eland is 1.28 GU and 1.28 BU — a full draw on each ledger, not a split. That is the defining property of a mixed feeder and the reason two ledgers are needed at all. Budget it as a split and you will systematically under-count demand on the resource that runs out first, which for a mixed feeder is usually browse.
2Why can the DALRRD long-term grazing capacity map not be used to set your browse budget?
Why: The exclusion is stated in the department's own background document, which also assumes veld in relatively good condition at a 40–50% utilisation factor. The map is the legal starting point for grazing capacity in ha/LSU, and it is silent on woody material by design. A mixed cattle-and-game operation that reads only the map will overstock one axis while understocking the other — which is exactly how a property ends up over-grazed and under-browsed at the same time.
3A neighbour argues that adding blesbok to his Highveld cattle farm "raises the carrying capacity". What is the honest response?
Why: Complementarity creates capacity only where a feeding class is genuinely unused, which is a savanna and bushveld story rather than a grassland one. The 20–25% figure comes from North American extension material and has not been quantified in South Africa. And grassland browse is small rather than nil — drainage lines, kloofs, termitaria and encroached patches all carry woody material. The distinction worth holding onto is between an ecological claim and a business claim: both can be legitimate, and confusing them is what makes these arguments unresolvable.