Grow

Permaculture principles that survive a dry climate

Permaculture is a design system, not a recipe — but most of its published recipes were tested in Britain, the Pacific Northwest, and temperate Australia. Applied uncritically in a semi-arid South African context, several of its most celebrated techniques will drain your time and your water table before they do any good.

David Watts · Filed under Grow

Why permaculture literature mostly does not apply here

The canonical permaculture texts — Mollison's Permaculture: A Designers' Manual, Holmgren's work, and most of what circulates online — were written for places that receive 700 mm of rain or more per year, where it falls in winter or is spread across the year, and where the soil is not actively consumed by termites. Large parts of South Africa receive 400–600 mm annually, much of it in high-intensity summer storms that run off before they infiltrate. Evaporation rates in the Highveld and Karoo regularly exceed 2,000 mm per year — meaning the pan evaporation is three to five times the rainfall. That single number changes almost every design decision.

This is not a reason to discard permaculture. It is a reason to read it critically and test every technique against local conditions before committing labour and water to it. The principles — observe, catch and store energy, obtain a yield, use edges — are sound. Many of the implementations are not.

Swales — where they help and where they fail

A swale is a level trench dug on contour to catch runoff and allow it to infiltrate slowly into the soil. In a 900 mm rainfall region with deep, well-structured soil, a swale can recharge groundwater meaningfully and grow a productive berm. In a 450 mm rainfall region with a hardpan at 400 mm depth, a swale fills in a single storm event, overtops, and erodes the downslope berm. The water does not go where the design assumed it would go.

Swales are worth considering in South Africa under specific conditions: slopes of 1–5%, deep friable soil with no hardpan in the first metre, rainfall above 600 mm per year, and a site that has been observed through at least one full wet season before earthworks begin. On shallow soils, on dolomite, or on slopes above 10%, they are more likely to cause erosion than prevent it. The alternative on shallower soils is a series of small infiltration basins — hand-dug sunken beds positioned at the end of a slope, sized to the catchment above them, and planted immediately so root structure holds the edges.

Earthworks that do work here

The water-harvesting techniques most consistently successful in southern African conditions are also the most modest: half-moon catchments (small semicircular earthworks behind individual trees), keyline subsoiling on contour, and sunken beds at the base of naturally occurring drainage lines. These work because they are scaled to the actual rainfall event — a 25 mm thunderstorm — rather than to a theoretical annual average.

Technique	Best suited to	Avoid when
Swale on contour	Deep soil (>1 m), slope 1–5%, >600 mm rainfall, observed for one season	Hardpan present, dolomitic geology, slope >10%, or <500 mm rainfall
Infiltration basin / sunken bed	Shallow soils, low-slope areas, small catchments, urban and peri-urban sites	Very high clay content with poor drainage; size carefully to catchment
Half-moon catchment	Individual tree establishment, any soil depth, semi-arid sites	Works poorly as sole technique on slopes that sheet-flood
Keyline subsoiling	Compacted Highveld soils, existing pasture or orchard land	Requires tractor; not viable on rocky ground or steep terrain
Gabion check dam	Established drainage lines (dongas), erosion control	Not a primary water-harvesting tool; designed to slow, not store

Food forests and our light levels

The food forest model — a canopy layer, sub-canopy, shrub, herb, ground cover, root, and vine layer stacked to mimic a forest edge — is one of permaculture's most visually compelling ideas and one of its most frequently misapplied ones in South Africa. The problem is not with the stacking concept, which is genuinely useful. The problem is with the species palette imported from temperate literature.

Temperate food forest species — apple, pear, hazel, elderberry, comfrey as a ground cover — are marginal performers in most of South Africa's summer-rainfall interior. They require winter chill hours that many sites do not reliably receive, and they are largely dormant through the hottest, driest months when a South African garden most needs ground cover. A food forest built from subtropical and indigenous species — moringa, subtropical figs, Natal plum (Carissa macrocarpa), sour bush (Searsia species), cowpeas as a nitrogen-fixing ground layer, and pigeon pea as a shrub layer — is a different system with a much higher probability of success.

Light is not the limiting factor in most South African food forest designs; it is surplus. The understorey in an immature food forest here receives more than enough light for most herb and vegetable production. What limits production is moisture, and the canopy trees that should be creating the canopy are competing for the same limited water. Plant the canopy layer last, not first, after the soil biology is active enough to support it — this is the reverse of the standard permaculture planting sequence and it matters.

Mulch depth and termites

A 150 mm layer of wood-chip mulch is the standard permaculture recommendation for suppressing weeds, retaining moisture, and feeding soil biology. In most of South Africa's summer-rainfall regions, a 150 mm wood-chip layer also creates a rich, moist habitat for subterranean termites, which will colonise it within two to three seasons and begin working on any timber in contact with it — including fence posts, raised bed timbers, and young tree stems.

This does not mean abandoning mulch. It means choosing mulch materials strategically and managing depth. Coarse gravel mulch around tree stems creates a termite-unfriendly collar. Straw and dry grass mulch breaks down faster but is less attractive to Macrotermes than wood chip. A two-layer system — gravel or stone against the stem, organic mulch over the general bed surface — captures most of the moisture-retention benefit with less termite pressure. Keep wood-chip mulch at least 100 mm away from any timber structure and inspect it at the start of each summer.

Biochar mixed into the mulch layer or incorporated into the bed soil serves a different function: it improves water-holding capacity in sandy soils and provides long-term habitat for mycorrhizal fungi without feeding termites. On a smallholding with access to wattle (Acacia mearnsii) or other invasive biomass, making your own biochar is viable and the technique is covered separately on this site.

What to ignore from the northern literature

The following techniques appear in most permaculture design courses and most of the canonical texts. They are not wrong in the contexts for which they were designed. They are wrong for most South African smallholdings and repeating them wastes water, time, and sometimes money.

• Sheet mulching over lawn to create new beds. In a high-rainfall climate, this works because the cardboard and organic matter stay moist long enough to suppress grass and build soil. In a dry summer, the cardboard dries out, the grass breaks through at the edges, and the whole system fails unless you irrigate it — at which point you are using scarce water to suppress grass you could have removed mechanically in an afternoon.
• Ponds as a central design element. Open water in a high-evaporation environment is a significant water loss. A 10 m² pond on the Highveld loses more than 2,000 litres per month to evaporation in summer. A covered or shaded tank stores the same water for a fraction of that loss. If a pond is ecologically justified (fire break, aquatic food production, biodiversity), design it in full knowledge of the evaporation cost.
• Temperate guild planting. The classic permaculture guild — apple, comfrey, yarrow, chives, nasturtium — is designed for a British garden. Comfrey (Symphytum officinale) is marginal in most of our summer-rainfall regions and a declared invasive in some. Yarrow is drought-tolerant and does transfer. Build guilds from local and subtropical species, not from published guild lists.
• Hugelkultur mounds. Buried wood in a high-rainfall system creates long-term water-holding and nutrients as it decomposes. In a semi-arid system, the buried wood dries out before it decomposes, creates a dry, low-density zone that roots avoid, and — again — attracts termites. On high-water-table sites or in wetter areas (KZN midlands, parts of the Western Cape), hugelkultur is worth testing. On the Highveld interior, it is not.
• Abundance framing that ignores evaporation. A great deal of permaculture writing implies that correct design produces abundance without irrigation. In a climate where evaporation exceeds rainfall by a factor of three or more, this is not honest. Correct design radically reduces irrigation need; it does not eliminate it during establishment or during drought.

What permaculture gets right for our conditions

The principles that do transfer, and transfer well, are the ones concerned with observation, design sequence, and soil biology — not the specific earthwork or planting templates.

• Observe before you act. Walk the site through a full wet season and a full dry season before committing to any earthwork. Where does water run? Where does it pool? Where is the soil already active? This is not patience for its own sake — it is the information on which every other decision depends.
• Start at the highest point in the water cycle. Roof water before ground water. Shade before mulch. Soil structure before planting. Getting this sequence right is worth more than any single technique.
• Perennial over annual where possible. Perennial vegetables and herbs build root systems that access deeper soil moisture and do not require annual soil disturbance. Moringa, cowpea (grown as a perennial), sweet potato vine, and subtropical herbs outperform annual equivalents in most dry-summer conditions.
• Soil biology first. Worm farms, compost, and biochar have a higher return than almost any structural earthwork on a small site. Soil that holds its own moisture and supports fungal networks needs less supplemental water than bare or compacted soil — and this is achievable on a balcony as much as on twenty hectares.
• Multiple functions from every element. A pigeon pea hedge fixes nitrogen, provides a protein-rich seed, creates a windbreak, and produces biomass for mulch or compost. A moringa planted at the edge of a vegetable bed provides shade at the critical mid-morning angle, leaf mulch, and an edible crop. Choosing plants and structures for multiple outputs is permaculture's single most transferable idea.

What went wrong — the failure modes

The most common failure pattern in South African permaculture projects is not bad intentions or bad technique — it is temperate technique applied without adaptation. The second most common is acting before observing: earthworks dug in the first season, before the site's actual drainage patterns were known, that then require expensive repair.

Swales that overtop in a single storm event are the most visible failure. They typically happen because the swale was sized to the design rainfall average rather than to the actual peak event — a 50 mm thunderstorm in a two-hour period, which is routine on the Highveld, will fill a standard swale on a 200 m² catchment almost immediately. If there is no overflow point designed into the system, the berm erodes and the water goes where it always went.

Food forest failures are quieter and take longer to become obvious. Trees planted in an untested guild compete for water through their first two dry seasons, the weakest species die, and the gardener concludes that the species was unsuitable — when the actual problem was establishment watering that stopped too early, or a planting density that made sense on paper but exceeded the site's water budget.

Mulch failures are fastest: a freshly mulched bed that was not watered in, dries to a hydrophobic crust in two weeks of Highveld summer, and subsequent irrigation runs off the surface rather than penetrating. Wet the soil thoroughly before applying mulch, and water the mulch itself until it is fully saturated before relying on it to retain moisture.

What we would do differently

• Spend the first season only observing and mapping — photograph every rain event, mark where water moves, test soil depth with a steel rod at twenty points across the site before touching a spade.
• Build soil biology before any earthwork: start a worm farm and a compost heap in month one, and apply finished compost to a test bed to measure what the baseline soil can actually grow.
• Size every water-harvesting earthwork to the peak storm event, not the annual average — calculate the catchment area above each structure and design for a 50 mm/hour event before designing for the season total.
• Choose a species palette from subtropical and indigenous plants before consulting any temperate food forest guild list; treat the temperate list as a starting point for identifying functions, then find local species that fill each function.
• Apply mulch in two layers: a coarse gravel or stone collar (100 mm wide) against every timber and tree stem, then organic mulch over the broader bed surface — and inspect the junction at the start of each wet season.
• Plant the canopy layer of any food forest last, after the understorey is established and the soil biology is active — the reverse of the standard sequence.
• If swales are part of the design, dig one test bay of 3 m and observe it through a full wet season before committing to the full length; the test bay costs an afternoon, a failed 60 m swale costs much more.

Is it worth it?

Permaculture as a design philosophy is worth the time it takes to learn. The observation discipline, the emphasis on soil biology, and the principle of stacking functions produce better outcomes than conventional gardening in almost any context — including semi-arid South Africa.

Permaculture as a set of published techniques requires more scepticism here than most courses and books will admit. The swale-food-forest-sheet-mulch package that forms the core of most introductory courses was developed in temperate, high-rainfall conditions and transfers poorly to our evaporation rates and soils. Following it without adaptation is how people spend three seasons, significant water, and real labour on a system that does not produce what was promised.

The honest answer is that the design method is worth it; uncritical replication of temperate techniques is not. Adapt every technique to your actual rainfall event, your actual soil depth, and your actual evaporation rate — and observe before you act. That is, ultimately, what the method says to do. The published templates are suggestions, not instructions.