Retention Works on the Slope, Unproven at the River: Global Evidence on Water Retention Landscapes

Retention methods measurably cut runoff and sediment, raise soil moisture and recharge groundwater near the structures, yet this review found no study that shows a gauged, controlled rise in catchment dry-season baseflow, and several strong studies show dense retention reducing water downstream.

Overview

Water Retention Landscapes include pits, bunds, terraces, check dams, johads, sand dams, infiltration canals, leaky weirs and beaver dam analogues. At the scale where they are built, the evidence is large and consistent. A review of sub-Saharan plot studies found that cross-slope structures cut runoff by 13 to 71% and soil loss by 39 to 83% (Wolka, Mulder and Biazin 2018). Terraces and check dams were the main driver of the fall in Yellow River sediment from the 1970s to the 1990s (Wang et al. 2016). Under Arvari johads in Rajasthan, groundwater recharge reached 3 to 7 mm per day in the monsoon (Glendenning and Vervoort 2010).

The evidence thins with scale. A water payer would buy more dry-season river flow, and this run found no gauged, controlled demonstration of it. On total yield the best catchment studies point the other way. In the Arvari, past a threshold, more harvesting area added little recharge and “substantially reduces streamflow” (Glendenning 2009). In Murray-Darling study catchments, farm dams cut mean annual streamflow by a median of 13% (Journal of Hydrology 2023). On the Loess Plateau, Feng et al. (2016, Nature Climate Change) found that rising evapotranspiration under the Grain to Green program, running since 1999, caused “a significant (p < 0.001) decrease in the ratio of river runoff to annual precipitation”. They also found that “NPP in this region is found to be already close to this limit” of 400 ± 5 g C m⁻² yr⁻¹. The strongest dry-season result, Peru’s amunas, describes a delay with no added annual volume. Infiltrated water “is retained for an average of 45 d before resurfacing”, and upscaling to Lima’s source areas “can potentially delay 99 × 10^6 m3 yr−1 of streamflow and increase dry-season flows by 7.5% on average” (Ochoa-Tocachi et al. 2019, Nature Sustainability).

The flagship programs carry the weakest independent evidence. Slovakia’s 2010 to 2012 Landscape Revitalization Program has no published hydrological evaluation. Its auditor found invoice fraud, and one of its dams failed fatally in 2021. India’s Comptroller and Auditor General tabled a report on Maharashtra’s Jalyukt Shivar in the state Legislative Assembly in 2020. It found that the scheme, at Rs 9,633.75 crore, “had little impact in achieving water neutrality and increasing groundwater level, and its execution was characterised by lack of transparency” (Outlook India). It also found that storage in 83 of 120 sampled villages did not meet water needs (Republic World). The Mulloon Institute’s monitoring paper says catchment results are still to be published. A BioHub can fund retention on evidence now if it pays for soil moisture, erosion, local wells, sediment and yield, and treats catchment baseflow as a hypothesis to monitor.

This page describes practitioners, NGOs, state programs and research organizations only through their published records and statements. None has yet been invited to comment. Corrections and responses will be recorded on this page. On this site, [MS] marks a claim that rests on the account of the party the claim is about or on the wiki’s own project documents. Money is given in the currency and year reported.

What the syntheses conclude

Synthesis	Conclusion
Wolka, Mulder and Biazin 2018, sub-Saharan Africa	Runoff −13 to −71%, soil loss −39 to −83%; yield effects mixed
Joshi et al. 2005, 311 Indian watersheds	Mean benefit-cost ratio 2.14, IRR 22%; best with strong participation and 700 to 1,000 mm rain
Joshi et al. 2008, 636 watersheds	Benefit-cost ratio about 2; weak institutions limit gains
Glendenning et al. 2012, Indian rainwater harvesting review	Local recharge documented; net watershed effect poorly understood
Larsen, Larsen and Lane 2021, beaver review	Site-scale flood attenuation, altered low flows, raised water tables, more evaporation
Chen et al. 2025, Loess Plateau check dams	58,776 dams by 2019; 5.5 billion m³ sediment retained; 34% annual Yellow River sediment cut 2000 to 2012
Frontiers in Water 2021, sand dams	Local water tables rise; one model met 6% of demand and cut downstream flow 1%
Tigray ecosystem-based adaptation review	Exclosures and stone bunds most successful; water harvesting schemes limited
EU NWRM and Climate-ADAPT	Case factsheets (Lamone, Tamera) give intended functions and modeled storage

None of the syntheses claims a demonstrated catchment-scale rise in dry-season flow. The Indian meta-analyses measure economic returns and do not isolate hydrology. WOCAT and post-2020 EU evaluations were not reached.

The distinctions

• Local versus catchment. Glendenning measured recharge under seven johads and in 29 wells, then modeled the 476 km² catchment, where streamflow fell past a threshold.
• Timing versus volume. Upscaled amunas delay 99 million m³ a year, raising dry-season flow 7.5% and lowering wet-season flow. Makueni sand dams raise evapotranspiration from 84.5% to 89% of rainfall, a volume loss.
• Scale. Each table row records scale. No plot figure is multiplied up.
• Retention versus revegetation. On the Loess Plateau, engineering dominates sediment reduction before the 1990s and vegetation after (Wang et al. 2016). Most Tigray and Indian evaluations do not isolate structures.
• Degraded versus functioning land. The best results come from incised channels, gullied loess and crusted Sahelian soils. New storage on functioning land behaves like Australian farm dams.
• Local climate. Treated under P7. No independent measured study was found. See Bioprecipitation for the wiki’s treatment of the rainfall hypothesis.

Results by method

Pits and microcatchments (zaï, demi-lunes)

Zaï raised Burkina Faso yields by 300 to 400 kg/ha in dry years and up to 1,500 kg/ha in wet years (Kaboré and Reij 2004). Half-moons with compost or manure gave 900 to 1,600 kg/ha of sorghum, 20 to 39 times unamended half-moons (Zougmoré et al. 2003). Water capture alone failed on nutrient-poor soils. No groundwater or streamflow measurement was found. Cost: labor; no per-hectare figure found.

Bunds, trenches and terraces

Stone bunds cut soil loss 68% on 202 Tigray parcels (Gebremichael et al. 2005). Under 120 mm/h simulated rain on the Loess Plateau, maintained terraces cut runoff 100%; ridgeless terraces 28%. Effectiveness declines as trenches fill (Taye et al. 2015, not re-read). Bunds take cropland and can lower early yields.

Check dams and small dams (johads, stop dams)

Check dams and dam lands cut watershed sediment yield by about 50 to 77% in China, Spain and the United States (Scientific Data 2024). Failures: the Arvari streamflow threshold; Jalyukt Shivar, where 38 of 58 audited villages saw no groundwater rise and none of 120 collected a maintenance cess (CAG 2020); sediment rebounds as Jiuyuangou reservoirs silt (Feng et al. 2026); the 2021 Rudno nad Hronom dam breach in Slovakia, which had no safety inspection or completion approval (STVR 2021). Cost: the CAG counted 6.41 lakh works started in 22,586 villages, of which 6.3 lakh (98%) were completed at Rs 9,633.75 crore (Republic World, citing the CAG). That is about Rs 1.53 lakh per completed work. Kravčík reports a €4 per m³ cap [MS].

Sand dams

Households near Kitui sand dams used about 440 liters a day against 110 elsewhere. The dams stored about 5% of runoff (Lasage et al. 2008). In a 2016 survey, about half of 116 Kitui dams held no water (Springer 2020). In Makueni, 47.6% of reservoir losses evaporated and 3.8% seeped, and baseflow contribution fell (SWAT+, 2026). Lasage et al. (2015) cap Ethiopian catchments at 1,000 and 600 dams to keep flow reduction within 3%.

Infiltration canals (amunas)

At Huamantanga, Peru, dye tracing showed a mean 45-day retention (Ochoa-Tocachi et al. 2019). Modeled local dry-season flow rose 3 to 554%. Basin upscaling is a scenario (E4) that diverts 35% of wet-season flow. Only three communities maintain amunas.

Leaky dams and natural flood management

At Pickering, England, leaky dams, planting and an 85,000 m³ bund cut the Boxing Day 2015 peak 15 to 20%, half from land measures (Environment Agency, 2016). The event was 50 mm over 36 hours. Land measures alone were modeled at 4% for a 1-in-25 peak and 8% for a 1-in-100 peak (Defra 2015). The chair warned they would not prevent a 2007-scale flood. See also Environment Agency (2017), Working with Natural Processes.

Natural Sequence Farming

Mulloon Creek has 14 leaky weirs on 2.5 km, gauges above and below, and floodplain piezometers. The institute reports maintained dry-season flow and a higher water table. Peel et al. (2022) call the pilot monitoring “more ad-hoc” and say catchment results await publication. Tony Coote’s claims of 60% more carrying capacity, outflow above inflow and about 100 mm a year of dew are [MS]. No independent study of Tarwyn Park was found.

Beaver dam analogues

Bridge Creek, Oregon, is a highly degraded incised stream, and BDAs were installed on 3.4 km of it (76 in 2009, 121 working by 2012). Bouwes et al. (2016, Scientific Reports) reported “significant increases in the density, survival and production of juvenile steelhead without impacting upstream and downstream migrations”. Survival rose 52% against a control stream over seven years. The authors call the study “the first large-scale experiment to quantify the benefits of beavers and BDAs to a fish population”. Baseflow was not reported. Natural beaver dams were “generally short lived” before posts were installed to anchor them. Ponds raise evaporation (Larsen et al. 2021).

State programs in Tigray

Abreha we Atsbeha reports a groundwater rise of about 5 m (Gebregziabher et al. 2016), without a control catchment. Nyssen’s 30-year photographs use an unchanged church woodland as a reference against rainfall change (E3).

Results by climate

Climate	Measured success	Measured limits
BSh (Sahel, Rajasthan, Kitui, Maharashtra)	Pits, bunds, johads, sand dams at field and local-well scale	Arvari streamflow loss; Jalyukt nulls; dry sand dams
BSk (Loess Plateau, eastern Oregon)	Catchment sediment; reach habitat	Water limit; siltation; short-lived dams
Cwb highlands (Tigray)	Bunds, exclosures	Trench filling; weak water harvesting gains
Andean seasonal	Amunas springs	Timing only
Cfb (England, NSW tablelands)	Moderate flood peaks	Small effect in large storms; Mulloon unpublished
Cfa/BSk Australia	Farm dam supply	−13% median streamflow
Csa/Csb Mediterranean	Case descriptions only	No measured hydrology found
Dfb (Slovakia, Czechia)	None measured	Failed structures; missed audit targets

The target of five measured Mediterranean cases was not met. That gap is a finding, and it matters for Cape fynbos and renosterveld catchments, which share the Csb class.

The case and study table

Legend: n.r. not reported; a dash means not applicable. Evidence levels: E1 controlled or paired experiment, E2 long record with counterfactual, E3 before-after without control, E4 model, E5 practitioner account; reviews and audits are labeled as such. The same table is available as CSV.

id	case	country	years	climate	soil_geology	slope	method	scale	outcomes	result	effect_size	timescale	persistence	cost	confounders	evidence_level	grade	independent	sources
IN-ARVARI-GV2010	Arvari johads recharge	India, Rajasthan	1985-; 2007-08	BSh	Hard rock	n.r.	Johads	Field, 29 wells	O2	Recharge 3-7 mm/day	3-7 mm/day	20 yr	In use	n.r.	Rain variability	E3	[IC]	Yes	Glendenning & Vervoort 2010
IN-ARVARI-GV2011	Arvari trade-offs	India, Rajasthan	Model	BSh	Hard rock	n.r.	366 structures	Catchment 476 km²	O2,O4,O11	Past threshold, streamflow falls	Direction	Scenario	n.r.	n.r.	Ungauged	E4	[IC]	Yes	Glendenning & Vervoort 2011
IN-RWH-GLEN2012	Indian RWH review	India	to 2011	Mixed	Mixed	Mixed	RWH	Review	O2,O4,O11	Net watershed effect unclear	n.r.	-	n.r.	n.r.	Abstraction	Review	[IC]	Yes	Glendenning et al. 2012
IN-WSD-JOSHI2005	311 watersheds	India	to 2003	700-1,000 mm best	Mixed	Mixed	Watershed dev.	311 sites	O7,O9	BCR 2.14, IRR 22%	BCR 2.14	Varied	n.r.	n.r.	Mixed quality	E3	[IC]	Partly	Joshi et al. 2005
IN-WSD-JOSHI2008	636 watersheds	India	to 2007	Mixed	Mixed	Mixed	Watershed dev.	636 sites	O7,O9	BCR about 2	BCR 2	Varied	Concern	n.r.	Mixed	E3	[IC]	Partly	Joshi et al. 2008
IN-KUMBH-WOTR	Kumbharwadi valuation	India, Maharashtra	to 2012	BSh	n.r.	n.r.	Watershed treatment	171 households	O7,O9	US$5,573-8,172 benefit/ha	BCR 2.28-3.76	15 yr	n.r.	n.r.	Before-after	E3	[MS]	No	WOTR
IN-JSA-CAG2020	Jalyukt Shivar audit	India, Maharashtra	2015-19	BSh/Aw	Basalt	n.r.	Stop dams, ponds	22,586 villages	O2,O9,O10	No rise in 38/58 villages; 29/80 water-neutral	Null 66%	1-4 yr	No cess	Rs 9,633.75 cr	Drought	E3	[IC]	Yes	CAG 2020
IN-JSA-GSDA2018	GSDA depletion	India, Maharashtra	2015-18	BSh/Aw	Basalt	n.r.	As above	State	O2,O11	13,984 villages >1 m drop in 2018 at 73.5% of normal rain vs 13,751 in 2015 at 59.4%; state minister disputed, citing rises in 101 tehsils	+233 villages despite wetter year	3 yr	n.r.	n.r.	Rain; contested figures	E3	[TBV]	Data yes	Hindustan Times 27 Oct 2018 (opposition claim citing GSDA)
IN-ARP-TBS	TBS river revival	India, Rajasthan	1985-	BSh	Hard rock	n.r.	Johads	Catchment	O3,O9	River revived (claim)	n.r.	Decades	Community	n.r.	Ungauged	E5	[MS]	No	TBS accounts
PE-AMUNA-OT2019	Huamantanga amunas	Peru, Lima	2014-17	Andean seasonal	Permeable	Steep	Canals	11 canals; basin	O2,O3	45-day retention; +7.5% dry-season (model)	99 Mm³/yr delayed	Seasonal	3 communities	1/10 gray (interview)	Model	E2/E4	[IC]	Yes	Ochoa-Tocachi 2019
CN-LOESS-WANG2016	Yellow River sediment	China	1950s-2010s	BSk	Loess	Steep	Terraces, check dams	Regional	O6	Engineering main driver to 1990s	About −90% load	Decades	Declining traps	n.r.	Reservoirs	E2	[IC]	Yes	Wang et al. 2016
CN-LOESS-CHEN2025	Check dam overview	China	to 2019	BSk	Loess	Gullies	58,776 dams	Regional	O6	5.5 billion m³ trapped	−34%	Decades	Silting	n.r.	Reservoirs	Review	[IC]	Yes	Chen et al. 2025
CN-LOESS-FENG2016	Water limit	China	1999-2013	BSk	Loess	Mixed	Revegetation	Regional	O4,O7,O11	Near NPP water limit	400 g C/m²/yr	15 yr	n.r.	n.r.	Climate	E4	[IC]	Yes	Feng et al. 2016
CN-LOESS-ZHANG2018	Soil drying	China	post-1999	BSk	Loess	Mixed	Afforestation	Regional	O1,O4,O11	Streamflow and soil water decline	Trees 2x grass	Decades	n.r.	n.r.	Climate	E4	[IC]	Yes	Zhang et al. 2018
CN-LOESS-YIELD2022	Yield not reduced	China	post-1999	BSk	Loess	Mixed	Revegetation	Regional	O4	No yield fall on 82.3% of plateau	None	20 yr	n.r.	n.r.	Feedbacks	E4	[TBV]	Yes	2022 study
CN-LOESS-JYG2026	Jiuyuangou 59 yr	China	1964-2022	BSk	Loess	Gullied	Check dams, terraces	Watershed	O5,O6,O10,O11	R² 0.98 to 0.008; rebounds in storms	>90% human share	59 yr	Decays	n.r.	Extremes	E2	[IC]	Yes	Feng et al. 2026
CN-LOESS-TERRACE	Terrace simulation	China	n.r.	BSk	Loess	Terraced	Terraces	Hillslope	O5	Runoff −100% maintained, −28% ridgeless	As stated	Event	Ridges decisive	n.r.	Lab	E1	[IC]	Yes	2022 simulation
CN-LOESS-LIU	Liu films	China	1995-	BSk	Loess	Steep	WB measures	Regional	O7,O9	Greening filmed	n.r.	Years	n.r.	n.r.	Many programs	E5	[MS]	No	Liu films
US-BRIDGE-BOUWES2016	Bridge Creek BDAs	US, Oregon	2009-15	BSk	Incised alluvium	Low	BDAs	3.4 km treated	O7, fish	Steelhead survival +52%	+52%	7 yr	Beaver colonized	n.r.	Control stream	E1	[IC]	Partly	Bouwes et al. 2016
US-BRIDGE-DAMLIFE	Dam longevity	US, Oregon	3 decades	BSk	Incised	Low	Beaver dams	Reach	O10,O11	Dams short lived	n.r.	Decades	Washouts	n.r.	Floods	E3	[IC]	Yes	Bouwes et al. 2016
GLOBAL-BEAVER-LARSEN	Beaver review	Global	Lit.	Mixed	Mixed	Low	Beaver dams	Reach	O2,O3,O5,O11	Attenuation; more evaporation	n.r.	-	n.r.	n.r.	-	Review	[IC]	Yes	Larsen et al. 2021
UK-DEVON-PUTTOCK2017	Devon beavers	England	2011-15	Cfb	n.r.	Headwater	Beaver dams	Small catchment	O5	Storm peaks about −30%	−30%	4 yr	n.r.	n.r.	Control	E1	[TBV]	Yes	Puttock 2017, not re-read
AU-MULLOON-SRCMA2008	Mulloon pilot	Australia, NSW	2006-08	Cfb, 600-800 mm	n.r.	Valley	14 leaky weirs	2.5 km reach	O2,O3	Flow held in dry times (reported)	n.r.	2 yr	Repairs	n.r.	No control	E3	[MS]	Partly	SRCMA 2008
AU-MULLOON-PEEL2022	Mulloon design	Australia, NSW	2018-	Cfb	n.r.	n.r.	Leaky weirs	23,000 ha	O10	Results pending	-	-	Flood damage	n.r.	-	Design	[MS]	No	Peel et al. 2022
AU-MULLOON-COOTE	Coote statements	Australia, NSW	2006-18	Cfb	n.r.	n.r.	NSF	Farm	O3,O7,O8	+60% carrying capacity; dew 100 mm	Claimed	12 yr	n.r.	n.r.	No control	E5	[MS]	No	ABC 2016
AU-NSF-TARWYN	Tarwyn Park	Australia, NSW	1970s-	Cfa	n.r.	Valley	NSF	Farm	O3,O7	No independent study	n.r.	n.r.	n.r.	n.r.	n.r.	E5	[TBV]	No	None found
AU-MDB-FARMDAMS2023	Farm dams	Australia, MDB	2018 level	BSk/Cfa	Mixed	Mixed	Farm dams	Basin	O4,O11	Median −13% streamflow; low flows hit	−13%	Long	Permanent	n.r.	Model	E4	[IC]	Yes	J Hydrol 2023
AU-MDB-MALERBA2021	Dam growth	Australia	2021-31	Mixed	Mixed	Mixed	Farm dams	Basin	O4,O11	−180 GL/yr projected	−180 GL	Decade	-	n.r.	Projection	E4	[TBV]	Yes	Malerba 2021
UK-PICKERING-EA2016	Pickering 2015	England	2009-15	Cfb	n.r.	Upland	Leaky dams, bund	Catchment	O5	Peak −15-20%	−15-20%	Months	n.r.	n.r.	One event	E2	[IC]	Partly	EA/STF 2016
UK-PICKERING-DEFRA2015	Pickering model	England	2011-15	Cfb	n.r.	Upland	Land measures	Catchment	O5,O11	−4% (1:25), −8% (1:100); synchronization risk	4-8%	-	-	n.r.	Model	E4	[IC]	Partly	Defra 2015
SSA-SWC-WOLKA2018	SSA barriers review	Sub-Saharan Africa	Lit.	Mixed	Mixed	Mixed	Bunds, terraces	Plot	O1,O6,O7	Runoff −13-71%; soil loss −39-83%	As stated	Seasons	n.r.	n.r.	Designs vary	Review	[IC]	Yes	Wolka et al. 2018
ET-TIGRAY-GEB2005	Tigray stone bunds	Ethiopia	3-21 yr bunds	Cwb/BSh	Mixed	Mixed	Stone bunds	202 parcels	O6	Soil loss −68%	−68%	3-21 yr	Persisted	n.r.	Parcels	E3	[IC]	Yes	Gebremichael 2005
ET-TIGRAY-TAYE2015	Bund decline	Ethiopia	Multi-year	BSh	n.r.	Mixed	Bunds, trenches	Plots	O5,O6,O10	Effect declines as trenches fill	Decline	Years	Maintenance	n.r.	Rain	E1	[TBV]	Yes	Taye 2015, not re-read
ET-ABREHA-NEG2013	Mendae groundwater	Ethiopia	1993-	Cwb	n.r.	Steep	Integrated SWC	Catchment	O2	About tenfold rise (as cited)	10x	20 yr	n.r.	n.r.	No control	E3	[TBV]	Yes	Negusse 2013
ET-ABREHA-GEB2016	Abreha we Atsbeha	Ethiopia	to 2015	Cwb	n.r.	Steep	Bunds, exclosures	Watershed	O2,O9	Groundwater +5 m	+5 m	20 yr	n.r.	n.r.	No control	E3	[IC]	Yes	Gebregziabher 2016
ET-TIGRAY-PHOTO	Repeat photos	Ethiopia	1974-2000s	BSh/Cwb	n.r.	Mixed	Exclosures, bunds	Regional	O3,O6,O7	Springs improved below exclosures	n.r.	30 yr	Maintained	n.r.	Woodland reference	E3	[IC]	Yes	Nyssen et al.
ET-TIGRAY-GULLY	Gully diversion	Ethiopia	Experiment	BSh	n.r.	Hillslope	Diversion	Hillslope	O2,O5,O6	Percolation up to 850 mm	850 mm	Seasons	n.r.	n.r.	One site	E3	[IC]	Yes	Nyssen et al.
ET-TIGRAY-EBA	EBA review	Ethiopia	Lit.	Mixed	Mixed	Mixed	Mixed	Regional	O7,O9,O11	Water harvesting limited	n.r.	-	n.r.	n.r.	-	Review	[IC]	Yes	Springer review
KE-KITUI-LASAGE2008	SASOL sand dams	Kenya	2000s	BSh	Sandy	Low	Sand dams	3 dams	O2,O9	440 vs 110 l/day; 5% runoff stored	4x use	Years	n.r.	n.r.	Non-random	E3	[IC]	Partly	Lasage 2008
KE-SANDDAM-HUT2008	Sand dam groundwater	Kenya	Campaign	BSh	Sandy	Low	Sand dams	1 dam	O2	Quick rise; depends on use	n.r.	Seasons	n.r.	n.r.	Abstraction	E3	[IC]	Yes	Hut 2008
KE-SANDDAM-QUINN2019	3-dam balance	Kenya	Field	BSh	Sandy	Low	Sand dams	3 dams	O2,O11	Specific yield below assumed	n.r.	Seasons	n.r.	n.r.	Parameters	E3	[IC]	Yes	Quinn 2019
KE-SANDDAM-VLD2006	Sand dam model	Kenya	Model	BSh	Sandy	Low	Sand dams	Catchment	O4,O9	6% of demand; −1% downstream	−1%	-	-	n.r.	Model	E4	[IC]	Yes	Van Loon 2006
KE-KITUI-SITING2020	116 dams survey	Kenya	2016	BSh	Sandy	Low	Sand dams	116 dams	O9,O10,O11	About half dry	50% dry	Years	Siting	n.r.	One survey	E3	[IC]	Yes	Springer 2020
KE-MAKUENI-SWAT2026	Makueni SWAT+	Kenya	Model	BSh	Sandy	Low	22 dams	22.35 km²	O3,O4,O11	ET/P 84.5 to 89%; baseflow share falls	+4.5 pts	-	-	n.r.	Ungauged	E4	[IC]	Yes	Frontiers 2026
ET-SANDDAM-LAS2015	Dam ceiling	Ethiopia	to 2050	BSh	n.r.	n.r.	Sand dams	Catchments	O4,O11	1,000 and 600 dam caps	3% limit	2050	-	n.r.	Model	E4	[IC]	Yes	Lasage 2015
BF-ZAI-KABORE2004	Yatenga zaï	Burkina Faso	1980s-2000s	BSh	Crusted	Gentle	Zaï + manure	Field	O7,O9	+300 to +1,500 kg/ha	As stated	Seasons	Maintained	n.r.	Manure	E3	[IC]	Partly	Kaboré & Reij 2004
BF-HALFMOON-ZOUG2003	Half-moons	Burkina Faso	Trials	BSh	Nutrient-poor	Gentle	Half-moons + compost	Plot	O7	900-1,600 kg/ha	20-39x	Seasons	n.r.	n.r.	Nutrients	E1	[IC]	Yes	Zougmoré 2003
NE-ZAI-RINAUDO	Niger zaï	Niger	n.r.	BSh	n.r.	n.r.	Zaï	87 villages	O7	Millet <350 to 1,000-2,000 kg/ha	3-6x	Seasons	Adopted	n.r.	No control	E5	[MS]	No	Rinaudo, ECHO
SK-LRP-SELF	Slovak program claims	Slovakia	2010-12	Dfb	Flysch	Mixed	Small check dams	488 municipalities	O2,O5	100,000 structures; 10 Mm³	Claimed	18 mo	n.r.	€4/m³ claimed	No monitoring	E5	[MS]	No	Kravčík
SK-LRP-NKU2014	NKÚ audit	Slovakia	2014	Dfb	-	-	Spending	46 municipalities	O11	Duplicate invoices; €32,181.17	€32,181	2-4 yr	-	n.r.	-	Audit	[IC]	Yes	Korzár 2014
SK-LRP-DENNIKN2020	Denník N check	Slovakia	2020	Dfb	n.r.	n.r.	Wooden dams	3 sites	O10,O11	Half damaged at Rudno; no expert analysis exists	50%	9 yr	Cleaned 1-2 yr	€20,000/33 structures	Reporters	E5	[IC]	Yes	Denník N 2020
SK-LRP-SVP2013	SVP statements	Slovakia	2013	Dfb	-	-	Check dams	National	O10,O11	Many dams destroyed, nonfunctional	n.r.	1-2 yr	Destroyed	n.r.	Critic	E5	[TBV]	Yes	Pravda 2013
SK-LRP-RUDNO2021	Rudno breach	Slovakia	2011-21	Dfb	n.r.	Steep	6 m dam	1 structure	O11	Breach; one death	1 death	10 yr	Never handed over	n.r.	Long rain	Official	[IC]	Yes	STVR, TA3 2021
SK-LRP-VTEI2022	Gully monitoring	Slovakia	2014-17	Dfb	n.r.	Gully	7 dams	Gully	O6,O10	Only measured study found	n.r.	2-5 yr	n.r.	n.r.	One gully	E3	[IC]	Yes	VTEI 2022
EU-CBA-SUSNIK2022	Cost-benefit model	DE, CZ, SK	Model	Dfb	-	-	Retention	Regional	O8,O9	Benefits exceed costs	n.r.	-	-	Modeled	Co-author	E4	[MS]	No	Sušnik 2022
CZ-NKU-2026	Czech audit	Czechia	2020-24	Dfb	-	-	Small water programs	CZK 7.1 bn	O10,O11	10 of 15 targets missed	10/15	4 yr	-	CZK 7.1 bn	-	Audit	[IC]	Yes	NKÚ ČR 2026
PT-TAMERA-EEA	Tamera	Portugal	2007-	Csa	n.r.	Valley	Lakes, swales	Valley	O7	Intended outcomes only	n.r.	Years	n.r.	n.r.	No data	E5	[MS]	Partly	Climate-ADAPT
AT-KRAMETERHOF	Krameterhof	Austria	1960s-	Dfb	n.r.	Steep	Terraces, ponds	Farm	O7	No independent study	n.r.	n.r.	n.r.	n.r.	n.r.	E5	[TBV]	No	None found

58 rows. Null, negative, harm or non-lasting rows: 22. African rows: 21.

The predictions

P1. Soil moisture and infiltration rise at plot scale. For: runoff −13 to −71% (Wolka 2018); terraces −100% at 120 mm/h; 850 mm percolation in a Tigray gully diversion. Against: effects fade as structures fill. Grade [IC]. Status documented-fact. Wiki line: cross-slope structures and pits reduce runoff and erosion on plots, with effects that depend on maintenance.

P2. Groundwater rises and springs revive locally. For: 8 of 12 rows with well or spring records report a rise. Against: none used a control catchment; the CAG found no rise in 38 of 58 villages. Grade [IC] locally, [TBV] program-wide. Status academic-inference. Wiki line: structures recharge groundwater nearby where geology allows.

P3. Catchment dry-season baseflow rises. For: amunas model +7.5%; Mulloon reports [MS]. Against: zero gauged, controlled cases; Makueni and farm dam results show lower low flows. Grade [TBV]. Status working-hypothesis. Wiki line: no controlled gauged study demonstrates it.

P4. Total downstream yield rises. For: one model finds no yield loss from Loess Plateau revegetation, a vegetation result. Against: Arvari threshold; −13% farm dams; Loess Plateau limit; sand dam −1% and 3% caps. Grade [IC] against. Status academic-inference. Wiki line: dense retention reduces downstream yield.

P5. Floods and sediment fall. For: check dams −34% sediment; Pickering −15 to −20%; Devon beavers −30%. Against: Pickering land measures −4 to −8% modeled; Jiuyuangou storm rebounds; Robichaud (2008) log barriers work only in small storms. Grade [IC] for sediment and moderate floods, [TBV] for large floods. Status academic-inference.

P6. Effects persist. For: Tigray bunds aged 3 to 21 years still cut soil loss 68%. Against: half of Rudno dams damaged; fatal breach; half of Kitui dams dry; Jiuyuangou silting. Grade [IC] that persistence depends on maintenance. Status academic-inference.

P7. Cooling and rainfall. For: Kravčík thermal images and Coote dew claims [MS]. Against: no independent measured study found. Grade [TBV]. Status working-hypothesis for cooling, speculative-projection for rainfall.

P8. Community-led outperforms state-led. For: Joshi links participation to higher returns. Against: community-built Slovak and Kitui structures failed; state-engineered Loess Plateau has the strongest sediment record. Grade [TBV]. Status working-hypothesis. Wiki line: maintenance and siting predict outcomes better than who leads.

P9. Practitioner claims match measurement. For: amunas tracers; Pickering’s “up to 20%”. Against: Slovak, Jalyukt, TBS and Mulloon headline figures lack independent confirmation. Grade [IC] that a gap exists. Status academic-inference.

Evidence the claims must face

1. Rainfall. Bouwes used a control stream; Nyssen a woodland reference. Mulloon and Slovak claims have no rainfall control.
2. Upstream-downstream transfer. Arvari, Murray-Darling interception policy, Ethiopian sand dam caps.
3. Evaporation. 47.6% of Makueni reservoir losses; 57% of Kitui depletion; beaver ponds.
4. Abstraction rebound. Irrigation expansion lowered Arvari resilience; the river parliament lacked aquifer data to enforce limits.
5. Failure and silting. Rudno, Kitui, Jiuyuangou, Tigray trenches.
6. Survivorship. No independent gauged study of Hiware Bazar, Ralegan Siddhi, Laporiya or Tarwyn Park was found.
7. Self-reported totals. Labeled [MS] in the table.

Practitioner claims against independent evidence

Program	Stated result	Independent evidence	Gap
Slovak program	100,000 structures, 10 Mm³ [MS]	No evaluation; audit fraud; dam failures	Large
Tarun Bharat Sangh	Arvari perennial [MS]	Local recharge; modeled flow loss	Unverified
Jalyukt Shivar	Villages drought-free	CAG nulls	Large
Mulloon NSF	Outflow above inflow [MS]	Results pending	Open
Pickering	Up to 20%	15-20% in a 50 mm event	Small
Amunas	Dry-season springs	45-day tracers	Small
Zaï	Yields doubled	Yield gains with manure	Moderate
Sand dams	Reliable water	Half dry in Kitui	Siting
Tamera, Krameterhof	Restored water cycle	None found	Untested

What a credible retention monitoring design looks like

Pair the treated catchment with a control (Bridge Creek against Murderers Creek). Collect at least two years of baseline. Gauge above and below the works, run a well or piezometer network (29 wells in the Arvari), keep rain gauges in the catchment, use tracers for residence time (amunas), and log abstraction. Measure evaporation from new storage. Run for ten years or more, through at least one large event, with structure-condition surveys at years 1, 5 and 10. Publish data independently of the implementer, as the Environment Agency did at Pickering. For performance-based water bonds, this supports paying for O1, O2, O6 and O7 now and contracting O3 only with this monitoring.

Summary table

Claim	Grade	Status	Safe
Structures cut plot runoff and erosion	[IC]	documented-fact	Yes
Local recharge near structures	[IC]	academic-inference	With caveat
Catchment baseflow rises	[TBV]	working-hypothesis	No
Dense retention cuts downstream yield	[IC]	academic-inference	Yes
Check dams cut sediment	[IC]	documented-fact	Yes
Moderate flood peaks fall	[IC]	academic-inference	With caveat
Large floods prevented	[TBV]	speculative-projection	No
Retention cools land	[TBV]	working-hypothesis	With caveat
Retention raises rainfall	[TBV]	speculative-projection	No
Community-led outperforms	[TBV]	working-hypothesis	No
Slovak 10 Mm³ stored	[MS]	speculative-projection	Only as claim

Related

• Water Retention Landscapes
• Retention Logic
• Performance-Based Water Bonds
• Outcome Payments for Community Restoration
• Bioprecipitation
• Community-Led Landscape Restoration in Sub-Saharan Africa
• Kravčík, Water Recovery
• Environment Agency (2017), Working with Natural Processes

Sources

• Bouwes, N. et al. (2016). Ecosystem experiment reveals benefits of natural and simulated beaver dams to a threatened population of steelhead. Scientific Reports 6: 28581. https://doi.org/10.1038/srep28581
• CAG of India (2020). Report No. 3 of 2020, Government of Maharashtra: Implementation of Jalyukta Shivar Abhiyan. Tabled 8 September 2020. Also summarized via India Water Portal. https://www.indiawaterportal.org/articles/jalyukt-shivar-abhiyan-low-impact-high-expenditure-scheme-cag
• Chen et al. (2025). Check dams on the Loess Plateau. River. https://onlinelibrary.wiley.com/doi/10.1002/rvr2.118
• Defra (2015). Slowing the Flow at Pickering, Phase II, executive summary. https://cdn.forestresearch.gov.uk/2022/02/fr_slowingtheflowphase2executivesummary.pdf
• Denník N (2020). Pred desiatimi rokmi začali vznikať hrádzky proti povodniam. https://dennikn.sk/1941823/
• Feng, L. et al. (2026). Journal of Soils and Sediments 26: 117. https://link.springer.com/article/10.1007/s11368-026-04308-7
• Feng, X. et al. (2016). Revegetation in China’s Loess Plateau is approaching sustainable water resource limits. Nature Climate Change 6: 1019. https://doi.org/10.1038/nclimate3092
• Frontiers in Water (2021). Sand dam review. https://doi.org/10.3389/frwa.2021.651954
• Frontiers in Water (2026). Makueni sand dams, SWAT+ modeling. https://doi.org/10.3389/frwa.2026.1922866
• Glendenning, C. J. (2009). Evaluating the impacts of rainwater harvesting in a case study catchment: the Arvari River, Rajasthan, India. PhD thesis, University of Sydney. https://ses.library.usyd.edu.au/handle/2123/5835
• Glendenning, C. J., and Vervoort, R. W. (2011). Hydrological impacts of rainwater harvesting in a case study catchment: the Arvari River, Rajasthan, India. Part 2. Agricultural Water Management 98: 715. https://doi.org/10.1016/j.agwat.2010.11.010
• Glendenning, C. J. et al. (2012). Review of rainwater harvesting in India. Agricultural Water Management. https://www.sciencedirect.com/science/article/abs/pii/S0378377412000273
• Joshi, P. K. et al. (2005). Meta-analysis to assess impact of watershed program and people’s participation. ICRISAT. https://oar.icrisat.org/2371/
• Journal of Hydrology (2023). Farm dam impacts on Murray-Darling streamflow. https://www.sciencedirect.com/science/article/pii/S0022169423011277
• Korzár (2014), reported via Lesmedium. https://www.lesmedium.sk/o-com-sa-pise/hradzky-stale-kamenom-svaru
• Larsen, A., Larsen, J. R., and Lane, S. N. (2021). Dam builders and their works: beaver influences on the structure and function of river corridor hydrology, geomorphology, biogeochemistry and ecosystems. Earth-Science Reviews 218: 103623. https://doi.org/10.1016/j.earscirev.2021.103623
• Lasage, R. et al. (2008). Potential for community based adaptation to droughts: sand dams in Kitui, Kenya. https://www.researchgate.net/publication/254419733
• Lasage, R. et al. (2015), cited via Concern Worldwide sand dam brief. https://fscluster.org/sites/default/files/2025-04/CWW%20sand%20dam%20brief%20v2.pdf
• Mulloon Institute; SRCMA (2008). Natural Sequence Farming pilot project. https://mullooninstitute.org/projects/natural-sequence-farming-pilot-project/
• NKÚ ČR (2026), reported via Epoch Times CZ. https://www.epochtimes.cz/2026/03/16/protipovodnova-opatreni-za-miliardy-nesplnila-cile-zadrzovani-vody-uvedl-nejvyssi-kontrolni-urad/
• Ochoa-Tocachi, B. F. et al. (2019). Potential contributions of pre-Inca infiltration infrastructure to Andean water security. Nature Sustainability 2: 584. https://doi.org/10.1038/s41893-019-0307-1
• Peel, L. et al. (2022). The Mulloon Rehydration Initiative: the project’s establishment and monitoring framework. Ecological Management & Restoration. https://doi.org/10.1111/emr.12549
• Pravda (2013). Hrádzky skončili, povodne má zastaviť betón. https://spravy.pravda.sk/domace/clanok/257694-hradzky-skoncili-povodne-ma-zastavit-beton/
• Republic World (2020), reporting the CAG audit of Jalyukt Shivar. https://www.republicworld.com/india/politics/thackeray-govt-targets-another-fadnavis-project-probe-jalyukt-shivar
• Scientific Data (2024). Check dam sediment dataset. https://www.nature.com/articles/s41597-024-03198-z
• Slowing the Flow Partnership (2016). Slowing the Flow at Pickering. https://www.forestresearch.gov.uk/research/slowing-the-flow-at-pickering
• Springer, Sustainable Water Resources Management (2020). Kitui sand dam siting survey. https://doi.org/10.1007/s40899-020-00417-4
• STVR (2021). Budaj: hrádza v Rudne nad Hronom nemala dostatočné bezpečnostné previerky. https://spravy.stvr.sk/2021/05/budaj-hradza-v-rudne-nad-hronom-nemala-dostatocne-bezpecnostne-previerky/
• Sušnik, J. et al. (2022). Land Degradation & Development. https://doi.org/10.1002/ldr.4373
• Wang, S. et al. (2016). Reduced sediment transport in the Yellow River due to anthropogenic changes. Nature Geoscience 9: 38. https://doi.org/10.1038/ngeo2602
• Wolka, K., Mulder, J., and Biazin, B. (2018). Effects of soil and water conservation techniques on crop yield, runoff and soil loss in Sub-Saharan Africa: a review. Agricultural Water Management 207: 67. https://ideas.repec.org/a/eee/agiwat/v207y2018icp67-79.html
• Zhang, S. et al. (2018). Vegetation cover change and soil water on the Loess Plateau. JGR Biogeosciences. https://doi.org/10.1002/2017JG004038
• Also cited from secondary summaries: Zougmoré, R. et al. (2003); Kaboré, D., and Reij, C. (2004); Gebremichael, D. et al. (2005); Hut, R. et al. (2008); Nyssen, J. et al., Tigray repeat photography (https://www.researchgate.net/publication/330441021).

Provenance

Written on 28 October 2026 under version 0.1 of wrl-success-evidence-deep-research.md, as a follow-on to the WRL and fynbos/renosterveld run of 12 October 2026 and the “Clear and Retain” run of 28 October 2026. The research run made 18 web searches and one delegated search on the Slovak program, in English, Slovak, Czech, Spanish and French.

The research tools could not open the wiki corpus pages during the run, so the brief was written without reading the current text or epistemic_status of Water Retention Landscapes or the practitioner pages. The CAG and NKÚ audits were read through press reports. Taye et al. (2015) and Puttock et al. (2017) were cited from earlier knowledge and not re-read. Brazil, Mexico, Jordan, the Negev, the Baviaanskloof, keyline and stage-zero restoration sites were not reached. Reviews, audits and official records sit outside the E1 to E5 ladder and are labeled by type. No place-specific material from the Overberg or Ward 2 was used.