Water

How a constructed wetland actually treats water

A constructed wetland turns biological waste into clean enough water to irrigate with — no chemicals, no power, and very little maintenance once established. The design that works in Europe needs adjusting before it works here.

David Watts · Filed under Water

What a constructed wetland is

A constructed wetland is a shallow, lined basin filled with gravel or crushed rock, planted with emergent macrophytes — reeds, sedges, or bulrushes — and fed with pre-treated wastewater. Treatment happens in the root zone, where biofilms on the gravel surface and around plant roots break down organic matter, strip nutrients, and reduce pathogens. No electricity is required once the system is running. No chemicals are added. The energy that drives it is biological.

The term covers several configurations. This article focuses on the two that are practical on a smallholding: horizontal subsurface flow (HSSF) and vertical subsurface flow (VSSF). Surface-flow systems — open water with floating vegetation — are used in large municipal systems and are not covered here.

Horizontal versus vertical flow

In a horizontal-flow system, effluent enters at one end of the bed, flows slowly through the gravel horizontally, and exits at the far end through an adjustable standpipe. The water level is kept just below the gravel surface so there is no standing water and no mosquito breeding. Oxygen transfer is limited, which means the system works primarily under anaerobic and anoxic conditions — good at removing BOD (biological oxygen demand) and suspended solids, adequate for ammonia reduction, but not strong on nitrate removal.

In a vertical-flow system, effluent is dosed intermittently from the top and drains downward through the media to a drainage layer at the base. Each dose is followed by a rest period during which air refills the pores. This aerobic environment is better at nitrifying ammonia and treating higher-strength loads. The trade-off is complexity: you need a dosing chamber, a pump or siphon, and a control mechanism.

Characteristic	Horizontal flow (HSSF)	Vertical flow (VSSF)
BOD removal	Very good (>85%)	Very good (>90%)
Ammonia removal	Moderate	Good
Suspended solids	Good	Good
Pathogen reduction	Moderate	Moderate
Power required	None (gravity fed)	Pump or siphon for dosing
Clogging risk	Lower	Higher if overloaded
SA summer drying risk	Low	Moderate (media dries between doses)
Suitable inlet	Septic tank outflow, greywater	Septic tank outflow, stronger loads

For most smallholding applications — treating septic tank outflow or combined greywater before irrigation — a horizontal-flow system is simpler to build, simpler to maintain, and more forgiving of variable loading. Vertical-flow is worth the added complexity when the load is high (more than eight people) or when stricter effluent quality is needed.

How the layers work

The liner sits at the base — a minimum 1 mm HDPE or LLDPE geomembrane, welded at the seams, turned up and over a compacted berm at the perimeter. Everything depends on this layer not leaking. A pinhole will not drain the bed in a day, but it will contaminate the soil beneath over years.

On top of the liner goes a drainage layer of 40–60 mm clean crushed rock, then the main treatment zone of 20–40 mm gravel (100–300 mm depth above the drainage layer), and a thin top layer of coarser material to allow planting. Total bed depth is typically 600–800 mm. The inlet is a perforated pipe running the full width of the bed, distributing flow evenly. The outlet is an adjustable standpipe, usually in a chamber outside the liner, that sets the water table inside the bed.

Plants are not decorative. Their roots extend into the anaerobic zone and create micro-aerobic channels around the rhizosphere — zones of higher oxygen that support different microbial communities than the bulk media. They also take up nutrients directly, particularly nitrogen and phosphorus, and their seasonal dieback adds organic matter that feeds the microbial community over winter.

Sizing per person — and why winter is the constraint

The standard European guidance for horizontal-flow systems is 3–5 m² per population equivalent (PE), where one PE represents one person and approximately 60 g BOD per day. That guidance was developed for temperate climates where evapotranspiration is modest and winter temperatures slow but do not stop biological activity.

In South Africa, two factors push the sizing upward. First, evapotranspiration in summer is high — particularly on the highveld and in the Western Cape — and actual hydraulic loading must account for the water the system will lose to the atmosphere. A bed that handles 150 litres per person per day in July may lose 30–40% of that volume in January before it reaches the outlet. This is useful for water balance but means you are sizing for the worst hydraulic case, which is winter.

Second, winter dormancy of the reeds reduces biological activity. In the highveld, Phragmites australis dies back above ground from May to August. Root activity continues, but at a fraction of the summer rate. The system still treats, but more slowly, and the buffer the roots provide is reduced. Sizing for winter — using 6–7 m² per person rather than 3–5 m² — gives the system the residence time it needs when biology is slow.

Household size	Minimum area (summer)	Recommended area (winter constraint)	Approximate bed length at 1.2 m width
2 people	10 m²	14 m²	12 m
4 people	20 m²	28 m²	23 m
6 people	30 m²	42 m²	35 m
8 people	40 m²	56 m²	47 m
A	s	s	u	m	e	s		1	5	0		L	/	p	e	r	s	o	n	/	d	a	y		g	r	e	y	w	a	t	e	r		o	r		s	e	p	t	i	c		o	u	t	f	l	o	w	.		W	i	d	t	h		o	f		1	.	2		m		i	s		a		p	r	a	c	t	i	c	a	l		m	i	n	i	m	u	m		f	o	r		u	n	i	f	o	r	m		f	l	o	w		d	i	s	t	r	i	b	u	t	i	o	n	;		w	i	d	e	r		b	e	d	s		r	e	d	u	c	e		c	l	o	g	g	i	n	g		r	i	s	k	.

Why this is different in South Africa

Indigenous species that work here

Phragmites australis (common reed) is the most studied species globally and performs well in South African conditions. Use locally sourced rhizomes rather than imported cultivars. It is vigorous, tolerates the full range of SA wastewater strengths, and recovers quickly after dormancy.

Typha capensis (bulrush or cat-tail) is fully indigenous, establishes well, and has similar treatment performance to Phragmites. It tolerates heavier loading and higher phosphorus concentrations. Its biomass is also useful — dried leaves can be woven or composted. It spreads by rhizome and seed and will need cutting back from the berms seasonally.

Cyperus papyrus (papyrus) is indigenous to the eastern and northern regions and works well in warmer climates. It does not tolerate hard frost and is not suitable for the highveld interior. In KwaZulu-Natal and Limpopo it is an excellent choice.

Do not use Arundo donax (giant reed), which is a Category 1b invasive listed under the National Environmental Management: Biodiversity Act. Despite its excellent treatment performance, planting it constitutes an offence under the Alien and Invasive Species Regulations.

Species	Common name	Climate suitability	Dormancy	Notes
Phragmites australis (local ecotype)	Common reed	All regions	Highveld winter	Most documented; use local rhizomes only
Typha capensis	Bulrush	All regions	Partial	Higher phosphorus tolerance; useful biomass
Cyperus papyrus	Papyrus	Warm regions only	Frost-sensitive	Not suitable below -2°C
Schoenoplectus validus	Soft-stem bulrush	Temperate and warm	Partial	Good in shallower beds
Arundo donax	Giant reed	— Do not use —	—	Category 1b invasive; illegal to plant

What it will and will not treat

A well-designed horizontal-flow constructed wetland receiving septic tank outflow (primary-treated) will reliably achieve:

• BOD reduction of 80–90% (from roughly 200–300 mg/L to below 30 mg/L)
• Suspended solids reduction of 70–90%
• Ammonia reduction of 40–60% (higher in vertical-flow systems)
• 1–2 log reduction in faecal coliforms (from 10⁶ to 10⁴ CFU/100 mL, approximately)

That effluent quality is appropriate for sub-surface drip or furrow irrigation of non-food-contact crops — trees, pasture, woodlots, flowers. It is not appropriate for spray irrigation of vegetables eaten raw, or for any contact with potable water systems.

What a single-stage horizontal-flow system will not reliably achieve:

• Pathogen reduction to WHO Guidelines for unrestricted irrigation (requires additional treatment or 30-day storage)
• Nitrate removal (requires either a vertical-then-horizontal sequence, or a dedicated anoxic stage)
• Heavy metal removal (constructed wetlands accumulate metals in sediment but do not reliably remove them from the water column at the concentrations found in industrial or mining-affected water)
• Treatment of undiluted chemical waste, solvents, or concentrations of detergent above what a normal household produces
• Primary sewage treatment — raw sewage must pass through a septic tank or primary settlement tank first

When registration is required

The National Water Act 36 of 1998 defines water use broadly. Section 21 lists activities that constitute water use, including "disposing of waste in a manner that may detrimentally impact on a water resource" and "altering the bed, banks, course or characteristics of a watercourse".

A constructed wetland that receives greywater or septic outflow and discharges only to irrigation within the property boundary does not, in practice, trigger a registration obligation under the general authorisations for small-scale non-commercial domestic use. The Department of Water and Sanitation's General Authorisation (GN 509 of 2009 and its successors) permits small volumes of irrigation use from treated domestic wastewater without a licence, provided no discharge reaches a watercourse.

Registration is required if:

• The outlet discharges to a stream, drainage line, pan, or dam
• The system handles wastewater from more than a single domestic dwelling (commercial or communal systems)
• The system is within a water resource protection area or a reserve as defined under the Act
• The local municipality's by-laws impose stricter requirements — check with the relevant district municipality before construction

When in doubt, a pre-application meeting with the relevant Catchment Management Agency costs nothing and provides clarity that avoids enforcement action later. The DWS office for your catchment is listed on the DWS website.

The maintenance nobody mentions

Constructed wetlands are sometimes sold as "maintenance-free". They are not. They are low-maintenance, which is different.

• Inlet inspection, monthly: Check that the distribution pipe is not blocked. Grease and hair accumulate at the inlet zone. The first 500 mm of the bed near the inlet is where most clogging begins.
• Outlet level, quarterly: Confirm the outlet standpipe is at the design level. A pipe that has shifted or been knocked will either waterlog the bed (surface ponding) or under-fill it (insufficient contact time).
• Reed cutting, annually: Cut stems to ground level in late winter (August on the highveld) before new growth begins. Remove the cut material from the bed — leaving it to rot in place adds organic load and can contribute to clogging over years.
• Berm inspection, after heavy rain: Check that the liner has not been undercut or that rodents have not burrowed through the berm. A breach here drains the bed to the soil below.
• Sludge accumulation, every 8–15 years: The inlet zone accumulates sludge that eventually reduces the effective pore volume. Systems that have been running for a decade or more will need the inlet zone excavated and the media replaced. This is not an emergency maintenance task but should be planned for.

The most common reason a constructed wetland underperforms in its second or third year is inlet clogging from inadequate pre-treatment. If the septic tank is not being desludged every three to four years, solids that should settle in the tank are carried into the reed bed instead, blocking the media progressively from the inlet end. The septic tank and the reed bed are one system — maintaining one means maintaining both.

What people get wrong

The failures in constructed wetlands are predictable and almost always trace back to one of four decisions made before the first spade went in.

• Undersizing for winter. Using European tables without adjusting for dormancy produces a bed that works well in summer and overloads in winter, when biological activity slows and rainfall-driven hydraulic load is often highest.
• No liner, or a poor one. A clay-lined basin is not equivalent to a geomembrane. Clay shrinks and cracks in a dry highveld winter. Any system that relies on natural soil impermeability will leak.
• Skipping pre-treatment. A reed bed is a polishing stage, not a primary treatment device. Feeding raw sewage directly into gravel media clogs the inlet within months.
• Using exotic species. Planting Arundo donax is illegal. Importing Phragmites cultivars bred for European conditions produces plants that perform poorly in local conditions and may not be authorised for release.
• Setting the outlet level too low. If the outlet standpipe is set so the water table inside the bed is well below the surface, the system runs aerobically and treats differently than designed. Horizontal-flow systems need the water table at 50–100 mm below the surface — not at the base of the media.

What we would do differently

• Commission a soil and geohydrology report before excavation, particularly on any property with dolomitic or expansive clay soils. The cost of this step is small relative to the cost of a liner failure.
• Size at 7 m² per person from the start. Retrofitting a bed that is too small means either rebuilding or adding a second cell in series — both are more disruptive than building correctly once.
• Install the outlet chamber with at least three standpipe positions at 50 mm increments. Being able to adjust the water table without cutting pipe is worth the extra fittings.
• Use Typha capensis as the primary species on heavier-load systems. Its tolerance for variable loading and high phosphorus makes it more forgiving of the inconsistent inputs typical of a smallholding.
• Plant at twice the recommended density and thin after two seasons rather than waiting for the bed to fill in at low density. A densely rooted bed establishes treatment performance faster.
• Build a simple weekly log — inlet and outlet appearance, any surface ponding, reed condition. Two years of entries will tell you when the inlet zone is approaching the end of its working life, well before it becomes a problem.

Is it worth it?

For a smallholding already running a septic tank, a constructed wetland is the lowest-cost, lowest-maintenance upgrade available for the treated effluent. It converts a disposal problem into an irrigation resource, requires no power, and once established, runs for a decade or more with predictable and modest inputs of time.

It is not worth it if:

• The only available site is within 50 m of a watercourse and the registration pathway looks uncertain
• The household is fewer than two people — the capital cost of a properly lined bed does not change much with size, and a system sized for one person produces effluent volume too small to be useful for irrigation
• The inlet load is highly variable — a guesthouse or event venue with irregular guest numbers will stress a fixed-size bed in ways that are difficult to manage without the dosing control a vertical-flow system provides

Where it is worth it, it is genuinely worth it. There are constructed wetlands on South African properties that have been running since the early 2000s with nothing more than annual reed cutting and periodic desludging of the upstream septic tank. That track record is difficult to match with any mechanical or chemical treatment alternative.