BioHubs

Alien Clearing and Soil Health Restoration: Water Yield and Flood Risk in the Cape Floristic Region

Properties3
descriptionA literature search on whether soil health restoration and water-retention practices, added to invasive alien plant clearing, raise water yield or reduce flood risk in fynbos and renosterveld. It found no study that tests the combination, and maps the nearest evidence and the risks.
tagsresearch, water, restoration, fynbos, renosterveld, cape-floristic-region, soil-health, flood-risk, literature-review
aliasesclearing plus soil health CFR, soil health restoration fynbos renosterveld, alien clearing flood risk Cape Floristic Region, clearing and soil health water yield

18 min read

A literature search asked whether restoring soil health, with Water Retention Landscape and in situ rainwater harvesting practices, adds measurable water yield or flood protection to the gains from clearing invasive alien plants in the Cape Floristic Region (CFR). It found no study that tests the combination in fynbos or renosterveld. Clearing has a long record of raising streamflow, almost all of it from models. Plot studies show that soil condition changes how CFR soils take up water, and studies of nutrients after clearing warn against amendments that feed invasive plants.

This page is the second review on this question. The first, Water Retention Landscapes in Fynbos and Renosterveld Restoration, searched for any pairing of water-slowing earthworks with CFR restoration and found none. This review widens the question to soil health and flood risk, and adds the verification problem that any outcomes-based payment would face.

Overview

The question rests on a hypothesis with two parts. Clearing invasive trees reduces the water lost to transpiration. If the soil left behind is compacted, crusted or water-repellent, more of the rain may run off before it soaks in, which would limit recharge and baseflow gains and leave slopes and rivers exposed to flash floods. Restoring the soil would then add to what clearing achieves.

The search found support for each part taken separately and no test of the two together. Clearing studies in the CFR are modeled at catchment scale. Soil studies in renosterveld and fynbos measure infiltration and water repellency at plot scale, and test no restoration measure. Studies of compost, biochar, microbial inoculants and planned grazing come from other regions, and the CFR literature warns that nutrients favor invasive species in these soils. The flood record of the Western Cape links fire, bare slopes and intense rain to erosion and slope failure, and no study tests whether soil restoration reduces those losses.

This page holds the absence of a direct study as a working hypothesis, for the reasons set out under Scope and method.

Scope and method

The research run searched on 12 October 2026 and reports using Google Scholar, Scopus, Web of Science, SciELO South Africa, African Journals Online, several South African university repositories, the Water Research Commission knowledge hub and the SANBI library, with about 60 English search strings and three in Afrikaans [MS]. This page could not confirm what the database searches returned. Several sources were paywalled and were read from abstracts.

The run covered five strands:

  • alien clearing and water yield;
  • soil hydrology in fynbos and renosterveld;
  • soil health interventions (compost, biochar, inoculants, grazing, brush packing);
  • water retention structures (swales, keyline, check dams, wetland structures);
  • the flood, slope failure and erosion record of the Western Cape.

Each citation used here was checked against an online record on 13 October 2026 where one could be found. A finding confirmed in the source’s own text or abstract carries [IC]. A figure or detail that could not be traced to its cited source carries [TBV]. A claim an organization makes about its own work carries [MS]. The Provenance section lists what changed.

Alien clearing and water yield

The evidence that invasive alien plants reduce streamflow in the CFR is long-established. Le Maitre and colleagues (2016) estimate that invasive alien plants reduce South Africa’s surface flows by about 1,444 million m³ a year, 2.9 percent of naturalized mean annual runoff, at about 970 m³ per hectare a year [IC]. That is less than half the 1998 estimate of 3,300 million m³ [IC]. Allowing for invasions along rivers raises the estimate to about 2,444 million m³ [IC].

For the Western Cape Water Supply System, Le Maitre and colleagues (2019) fed Pitman model streamflow reductions into the system yield model [IC]. The 98 percent assured yield was about 580 million m³ a year under effective control, 542 million under current invasions and 450 million after 45 years without management [IC]. Current invasions already cost about 38 million m³ a year, rising to 130 million [IC].

Researchers modeling four small mountain catchments above Cape Town’s major dams report that clearing fully invaded catchments could raise streamflow by 15.1 to 29.5 percent, that clearing along rivers gives gains almost twice those from clearing the surrounding slopes, and that gains are largest in mid to low flows and in dry years [MS]. Those figures come from the researchers’ own public account; the run gave the riparian factor as 1.7 and named the model as MIKE SHE, and the paper itself was not located [TBV]. A related study by Holden and colleagues (2022) found that clearing at moderate invasion levels would have offset 3 to 16 percentage points of the streamflow losses in the 2015 to 2017 Cape Town drought, and that preventing further spread avoided losses of 10 to 27 points [IC].

Clearing works by reducing transpiration. The run attributed program totals and a Berg River Dam estimate to Working for Water reviews; none could be traced to their sources, as in the first review [TBV].

Soil hydrology in fynbos and renosterveld

Renosterveld. De Waal (2023), a Stellenbosch University doctoral thesis on the Bot River catchment, modeled soil erosion at 22 t/km² a year under natural vegetation and 490 t/km² a year under 2018 land use [IC]. It found more soil carbon in renosterveld than in cultivated fields, more phosphorus in cultivated fields, and that renosterveld fragments remediate polluted agricultural runoff [IC]. The run also credited the thesis with faster infiltration in renosterveld than in wheat, through preferential flow along burrows, roots and cracks, with significant effects of earthworm activity and porcupine diggings, and with detailed findings on duplex soils; these are not in the abstract and were not checked against the full text [TBV]. A soil carbon comparison of 70.9 against 27.1 tonnes per hectare came from an unnamed separate study [TBV].

Fynbos and water repellency. Fisher (2025), a Stellenbosch University doctoral thesis, found soil water repellency in five of six native fynbos vegetation types and under invasive alien stands [IC]. Repellency was most frequent and most severe in hot, dry summer months, and the top 20 mm of soil was less repellent than deeper soil [IC]. A planned burn reduced surface repellency at a Protea-dominated site and not at a Restio-dominated site, and the fire effects were short-lived [IC]. In the April 2015 drought, normally wettable soil at 50 and 100 mm at the Restio site became repellent, and the drought and prolonged repellency caused local deaths of seedlings and adult plants [IC]. The run attributed to Scott and Van Wyk (1990) a finding that fynbos soils were the least likely to be water-repellent; this was not checked [TBV].

Fire and erosion. Euston-Brown (2000) studied catchments after the Cape Peninsula fires [IC]. Fire severity was much higher in alien and cleared vegetation than in fynbos [IC]. Erosion at fynbos sites never exceeded 1.5 cm, and on average the soil surface rose [IC]. After the first winter rains, a little over 100 m³ per hectare was lost from alien-infested areas [IC].

Together these establish that CFR soils differ measurably in how they take up water, between natural and transformed land and through the fire cycle. None tests whether a restoration measure can change that.

Soil health interventions

The run located no CFR study of compost, biochar, microbial inoculants or planned grazing as a way to restore soil in fynbos or renosterveld [MS]. Its evidence for each comes from elsewhere [TBV]:

  • Compost: reviews of Mediterranean drylands report better infiltration and water holding, with carbon building up near the surface.
  • Biochar: studies in Mediterranean soils report better infiltration and water retention, with water repellency at high pyrolysis temperatures.
  • Microbial inoculants: a study in Mediterranean quarries reports better plant growth from inoculants drawn from the same host species.
  • Planned grazing: a review reports more soil organic matter under holistic grazing and no significant difference in infiltration between holistic grazing, light continuous grazing and exclusion. This page could not trace the review. A 2017 meta-analysis by Hawkins compared Holistic Planned Grazing with continuous grazing on other measures, and its findings are disputed by the Savory Institute [TBV].
  • Brush packing: branches laid across the slope to slow runoff and trap soil, used in Working for Water projects to rehabilitate bare ground, with no hydrological measurements located.

Nutrients. CFR soils are nutrient-poor and their species are adapted to that. Yelenik, Stock and Richardson (2004) found more available nitrogen in soils under invasive Acacia saligna than under fynbos, found that clearing did not raise nitrogen mineralization over the year they monitored, and found that a native grass grew faster on the nitrogen-rich soils [IC]. They recommend burning and high carbon-to-nitrogen mulch to draw down available nitrogen after clearing [IC]. The run reported that soil nitrogen stayed high more than seven years after acacias were removed; that was not traced [TBV]. Ruwanza, Musil and Esler (2012) found that adding sucrose to lowland fynbos soils suppressed the growth of introduced plants without lowering soil nitrogen, and judged sucrose ineffective as a restoration aid [IC]. The run’s account of gypsum additions and of ex-arable renosterveld does not appear in that abstract [TBV].

The fertilizer analogy that prompted part of the question therefore runs against the CFR evidence: in these soils the restoration problem after clearing is usually too much available nitrogen, so amendments that add nutrients carry a risk of favoring invasive plants [TBV].

Water retention structures

The run located no CFR study of swales, contour bunds, keyline design, gully plugs, leaky weirs or Zeedyk structures in fynbos or renosterveld [MS]. The first review reached the same finding.

  • Overberg Renosterveld Conservation Trust. The Trust reports watercourse restoration in renosterveld and says passive restoration is not enough [MS]. The run listed gabions, soil blankets, drift fences and drainage hollows, which the first review could not find in the retrieved text [TBV]. No hydrological measurements are reported [MS].
  • Tamera, Portugal. Tamera reports 29 lakes and retention spaces by 2015, water self-sufficiency and stabilized groundwater; these come from the project’s own documentation [MS]. The run reported eutrophic lakes in 2012 to 2014 monitoring [TBV].
  • Keyline design, Chile. Toro-Mujica (2023) reports that keyline design influenced soil moisture, temperature and grass quality on naturalized annual grassland [IC]. The yield gains of 44 and 25 percent are not in the abstract [TBV].
  • Swales and keyline in Mediterranean Europe. An EU agricultural innovation factsheet reports less erosion and better soil water retention, at about €5 a meter [TBV].
  • Check dams and gully plugs. The run cites Water Research Commission guidelines for stone and brushwood check dams and gully head reshaping, with engineering design needed for gullies deeper than 1 to 1.5 m [TBV].
  • Kromme River wetlands, Eastern Cape. Thirteen large structures were built from 2000 to protect palmiet wetland basins [IC]. The sources disagree on how they came through the August 2006 flood. SANBI, which runs the Working for Wetlands program, reports that the structures did their job, the main basins came through largely intact, and palmiet slowed the floodwater and trapped sediment [MS]. A Rhodes University case study for the Water Research Commission found every structure damaged, with four needing urgent repair and none intact [IC]. The run reported only the first account.

Olivier, Van De Wiel and De Clercq (2023) review gully erosion in South Africa and call for research on mitigation measures and how well they work [IC].

Flood, fire and slope failure in the Western Cape

The run drew on a compiled summary of Western Cape floods from 2003 to 2026, which it attributed to the Western Cape Government. The summary was compiled for this network and its sources have not been checked, so every event below carries [TBV].

  • 2003, Montagu: 178 mm in one day; overland flow stripped vegetation along the Kogmans River.
  • December 2004: damage of about R57.9 million in the Eden District, worst where vegetation had been removed.
  • 2008: floods in November did damage close to R1 billion, with extreme erosion along foothill rivers.
  • 2016 and 2018, Cape Town: rockfall and mudslide risk heightened by fires that had stripped ground cover.
  • 2021, Overberg: rill erosion washed topsoil from new grain fields near Riviersonderend.
  • 2024: soil washed off farmland, silting dams; four dams failed near Riverlands in August.
  • 2026: a national disaster declared after a series of cut-off lows, with slope failures on the Garden Route.

The run reads this record as a pattern in which fire or bare, compacted soil is followed by intense rain on saturated ground. That reading is consistent with the Euston-Brown findings above. No study tests whether soil restoration or water retention structures reduce flood peaks or slope failure in CFR catchments.

Verification

The Greater Cape Town Water Fund’s 2019 business case found that Working for Water “outcomes are not measured in ecological or water terms, but rather in terms of hectares treated, and job creation” [IC]. Its own water gains were modeled from earlier studies with the WRYM, WR2005 and ResSim models [IC]. The run cites a 2020 monitoring and evaluation plan for the Fund that describes paired-catchment and before-after-control-impact designs; the plan was not located [TBV]. The run also reports that the Cape Water Performance-Based Bond’s pricing supplement, which would set its verification method, was not located [MS].

Any claim that soil restoration adds water or reduces floods would face a harder attribution test than clearing does. Rainfall varies more from year to year than any plausible treatment effect, and extreme floods may not occur during a study.

Evidence table

Counts are the sources located and checked for this page. A zero records that no source was located.

Practice or questionCFR, measuredCFR, modeled or descriptiveElsewhere
Alien clearing and streamflow04 (Le Maitre 2016 and 2019, Holden 2022, the four-catchment study) [IC, MS]0
Soil infiltration and repellency under natural and transformed land2 (De Waal, Fisher) [IC]00
Fire severity and post-fire erosion1 (Euston-Brown) [IC]00
Compost, biochar, inoculants as soil restoration003, unconfirmed [TBV]
Planned grazing and infiltration001, untraced [TBV]
Nutrient drawdown after clearing2 (Yelenik, Ruwanza) [IC]00
Water-slowing earthworks003 (Tamera [MS], Chile [IC], EU factsheet [TBV])
Wetland and gully structures under flood01 (Kromme River, sources disagree) [IC, MS]0
Clearing combined with soil restoration000

Gap statement

What no located study has done:

  • Tested whether soil restoration or water-slowing earthworks, added to clearing, change infiltration, soil moisture, baseflow or water yield in fynbos or renosterveld.
  • Measured the effect of erosion-control or water retention structures on flood peaks or downstream water in the CFR.
  • Compared combined treatments with clearing alone, fire management or grazing exclusion.
  • Tested any soil amendment in fynbos or renosterveld for its effect on water, or on invasion.
  • Assessed how earthworks and amendments interact with the fire regimes of these vegetation types.

What is asserted without measurement: the Trust’s benefit claim for watercourse restoration; Tamera’s water self-sufficiency; and, per the business case, the water gains from clearing, which are modeled.

Study design proposed by the run

The run proposed a 10 to 20 year program for a site in the Duiwenhoks catchment, with an untreated reference area as the control [MS]. The design below is general; a site-specific version belongs with the proposal for that site.

  1. Years 1 to 2, baseline. Install flumes or weirs, piezometers, soil moisture sensors, sediment traps and weather stations; measure infiltration, soil carbon, bulk density, nutrients, vegetation, fuel load and water repellency.
  2. Years 2 to 3, plot trials. Replicated plots of brush packing, contour swales or keyline, check dams in gullies, low-rate biochar, native soil inoculants, and grazing exclusion or planned grazing.
  3. Years 3 to 10, sub-catchment scale. Extend the treatments that show promise, with untreated controls.
  4. Years 10 to 20, catchment assessment. Calibrate a yield model against the record and estimate what soil restoration adds to clearing.

It proposed at least two years of baseline, a paired-catchment or before-after-control-impact design, rainfall simulation at plot scale to test extreme events, and independent verification. Its cost range of R50 million to R100 million carries no worked estimate [TBV]. The run’s partnership list names organizations that have not been approached, and is held off this page.

What this means for BioHub practice

A BioHub in a fynbos or renosterveld BioRegion can cite CFR evidence that clearing raises streamflow (modeled), that fynbos burns cooler and loses less soil than invaded land, and that renosterveld fragments clean farm runoff. It cannot yet cite evidence that soil restoration or earthworks add water or reduce floods. A service built on that claim fits the speculative pending research category of the Tenderable Services Portfolio readiness diagnostic. Services that add nutrients to these soils carry a documented invasion risk and should not be offered until tested. A measured study would be new evidence for the field.

Sources

Cited by the run and not confirmed for this page: the Working for Water program reviews, the GCTWF Monitoring and Evaluation Plan (2020), the Cape Water bond’s Information Memorandum and pricing supplement, the Water Research Commission gully guidelines, the WOCAT brush packing case, the EIP-AGRI swales factsheet, the compost, biochar, inoculant and grazing studies, Scott and Van Wyk (1990), and the compiled Western Cape flood summary.

Provenance

Extracted from research-brief-duiwenhoks.md, the output of a Claude Research run on 12 October 2026. Editorial decisions:

  • Spelling converted to American, with the network’s voice conversions applied.
  • epistemic_status set to working-hypothesis. The run recorded its null results “as evidence of absence”; that framing was removed.
  • Tags re-assigned. The run tagged whole paragraphs [IC] on single sources, and used [IC/MS]. Each claim now carries one tag set by the checks above.
  • Confirmed on 13 October 2026: Le Maitre et al. (2016 and 2019), De Waal (2023) from its abstract, Fisher (2025), Euston-Brown (2000), Yelenik et al. (2004), Ruwanza et al. (2012), the Kromme River sources, the business case passages, Toro-Mujica (2023) and Olivier et al. (2023).
  • Corrections: De Waal is the sole author of the thesis; Van Niekerk and Miller supervised it. The gully review is by Olivier, Van De Wiel and De Clercq in Earth Surface Processes and Landforms; the run credited it to Panter and Ruwanza in Land Degradation & Development, and no such paper was found. The De Waal erosion figure of 490 t/km² a year, unconfirmed in the first review, is confirmed by the abstract. The Kromme River account was one-sided; the Rhodes University inspection is added. The sucrose study concerns lowland fynbos and does not report gypsum. The four-catchment clearing figures are traced to the researchers’ public account, and the paper itself was not located.
  • Added from verification: Holden et al. (2022), Euston-Brown’s report as the source of the post-fire erosion figures, and the Rhodes University inspection.
  • Omitted: the run’s site-specific details, its partnership and governance lists, its funding list and its names list. Outreach leads sit in the workbench, outside the published site.

The research run reports a database search; this page could not confirm its results. A search of the databases themselves would test the central finding.

Graph View

Backlinks