Home> Blog> Resetting the river: site member visit to Holnicote Stage 0 sites

Resetting the river: site member visit to Holnicote Stage 0 sites

Monday, June 29, 2026

Site member visit, Holnicote Estate, Somerset — Tuesday 9th June 2026
Site member visit, Holnicote Estate, Somerset — Tuesday 9th June 2026

The drive across the top of Exmoor down to the North Somerset coast is a beautiful proposition. It isn't as dramatic as some landscapes, but at this time of year it's filled with glorious colour, grazing animals and not much else. For me, coming from the heavily populated South East, the peace and remoteness is a welcome break from the crowds. The RRC has had a number of previous visits to the estate, but it was my first visit which I had been anticipating with excitement.

Holnicote sits on the edge of Exmoor, a 12,500-hectare estate that encompasses coastal saltmarsh and beaches, grazed farmland and upland peat bogs and heathland. The sources of the Rivers Aller and Horner are almost entirely contained within the estate. It is also home to the first UK trial of Stage 0 river restoration — a technique, first pioneered in Oregon, US of effectively resetting a river: filling in a straightened, deepened channel entirely so that water can cut its own path naturally across the floodplain. Known as the Riverlands project, the National Trust team have been delivering this alternative approach to water and landscape management at catchment scale as part of a wider Landscape recovery project encompassing 7,000 Ha in partnership with 40 local land managers. Ben Eardley, Jack Siviter and Mathieu Burtschell of the National Trust team spent the day showing us their stage 0 work and use of ecosystem engineers over recent years.

A potted history of the site

 

It’s useful to consider the past works carried out on site to understand the evolution of restoration practice at Holnicote:

  • 2016 and before — more than 3,000 small natural flood management interventions across the Horner Water and Aller catchments: peat dams up in the gullies, gutter-style cut-offs along tracks, and low bunds on the floodplain to hold water back on the fields.
  • Completed in 2019 — a modest two-week Stage 0 pilot at Selworthy Farm on a tributary of the River Aller, proving the technique at small scale in the UK.
  • to July 2023 — the flagship 1.2km Stage 0 reconnection on the main River Aller, Selworthy farm, creating around 7 hectares of new wetland.
  • 2024 — "one year on" monitoring confirms the headline results: downstream flood peaks on average 38% lower, groundwater up by a metre in places, water turbidity down 41%, aquatic habitat up by roughly 1,780%, and a genuine boom in wildlife.

Currently in 2026 — new floodplain reconnection work completed at Holtball Farm on a tributary of the Aller and beaver release from enclosures, so the estate offers a living timeline of the whole journey from smaller first attempt to large-scale practice.

The bigger picture: Riverlands and Landscape Recovery

This Stage 0 work is one strand of the National Trust's wider Riverlands programme, which sits within the estate's Landscape Recovery Scheme. The aim of the project is to foster healthy rivers and catchments, rich in wildlife, enjoyed and cared for by all — all achieved by working with natural processes rather than against them, and by thinking in decades rather than seasons. In practice, that overarching aim breaks down into several interlinked objectives:

  • A more resilient water system — slowing the flow, storing floodwater and releasing it slowly in dry months so the catchment behaves like a sponge rather than a drainpipe, reducing downstream flood risk and improved drought resilience for farming
  • Improved biodiversity and recovery of species that have all but vanished from our river corridors.
  • A mosaic of habitats — wetland, wet woodland, wood pasture and species-rich grassland  
  • Cleaner water — trapping sediment and nutrients on site instead of flushing them downstream.
  • Carbon storage, restored wetlands lock up carbon.
  • Reconnecting people with nature — these are places to be visited, understood and enjoyed, not just engineered and forgotten.

It's worth holding all of that in mind on a visit, because the brilliance of Stage 0 is that a single intervention can deliver most of these objectives all at once.

Site 1 — Selworthy Farm: the confidence-builder

Our first stop was a first-order tributary of the Aller — an "ordinary watercourse" on Selworthy Farm. Described by the team as a confidence-building exercise: a chance to learn and develop the Stage 0 technique before committing to anything larger scale. The scale of the project is reflected in the figures: it was completed for approximately £5,000, including land drainage consent and an archaeological brief, using local contractors and National Trust staff with a 13-tonne digger and a 10-tonne dumper.

The work itself was straightforward in principle: a six-foot-deep, two-metre-wide main watercourse and its side feeder were filled in using soil scraped from higher ground around the stream and ditch. The site sits on roughly a 6% slope, with greater gradients comes higher risk of incising the newly infilled channel. To mitigate this, wood was placed to deflect flow from the compacted watercourses and encourage flow to drain diffusely across the field. 

Site 1 Selworthy Farm
Site 1 Selworthy Farm

In high summer there's no obvious surface water until you reach the lowest points of the field — but the real tell is the grass as to the wetness of the site. It was growing in tall, lush swards up the gentle slope, and the ground stayed noticeably wet at the surface in lower reaches despite a few dry months and a little recent rain. For a farmer in a changing climate, this site provided observable evidence of a clear benefit: resilient grassland that keeps producing in dry periods. 

Tenant and non-tenant farmers alike have been keen to understand and trial the approach, with extensive, low-density grazing of cattle and sheep planned around the restoration. At this first site pigs to rootle through the soil — making space for wildflower seed to germinate and carving out their own scrapes and wallows in the wetter ground, another example of an ecosystem engineer.

The proof-of-concept lessons learned here were:

  • Compact and fill the ditch higher than bank level — then compact it more.
  • Sub-surface field drainage wasn't removed here, so it still pipes water straight back into the watercourse.
  • Plant more aquatic plug plants after infilling than you think you'll need. There is no downside.

Site 2 — the main River Aller: the flagship stage 0 project

From there we dropped into the valley floor and the project everyone comes to see: the 1.2km Stage 0 reconnection on the main River Aller, completed in July 2023. This was a different order of undertaking altogether — around £600,000, with far more complex consenting because it's a main river (a full Flood Risk Activity Permit needed), and the added task of surveying, trapping and relocating a couple of hardy water voles before work began.

The River Aller before restoration was pretty bleak in habitat terms: a completely straight, four-metre-wide, five-foot-deep (<3:1 width to depth ratio), plane-bed, silty channel with very little habitat availability. Post restoration, three years later, there are few signs of the mass engineering works at all, due to the density of vegetation which has grown up on site. The huge volume of wood that was added to roughen the floodplain, divert water and slow the flow is now mostly hidden in the growth. 

Site 2: Flagship stage 0 site
Site 2: Flagship stage 0 site

The most obvious part of the restoration were the criss-crossing log structures set near the target elevation. Target elevation is set to match the outflowing channel downstream of the restoration works, where the wide, diffuse multi-threaded valley floor joins back in to a single channel, out of the restoration site. These structures guard against the river head-cutting back through the infilled channel. 

Ben showed us construction photos of the bare, scraped ground used to infill the channel and flatten the floodplain — and made the point that despite several big flood events since, only a handful of logs have needed pinning back into place. Wood movement through the site is desirable to some degree and will occur in higher flows, however where wood is at risk of exiting the site, some intervention is needed to protect downstream landowner infrastructure (fences etc.). 1.2 km of river have been reconnected with their floodplain, increasing the width of the wetted channel from 4 metres to 147 metres at its widest point.

Viewing site photos during construction and planned earthworks
Viewing site photos during construction and planned earthworks

The scale of the intervention is striking: around 5,000 cubic tonnes of soil removed, 3,500 wetland plug plants put in (Greater Tussock sedge among them, which can grow well over head height), and some 25,000 trees planted across the site — aspen, black poplar, buckthorn, willow and bird cherry — with alder regenerating naturally in the riparian corridor. Beavers have been present on site, in enclosures since 2020, however the Trust recently received their license for wild release and the enclosures were breached in Feb 2026, set for full removal this Autumn. Planting species favoured by beaver in particular areas will encourage these ecosystem engineers into the area and speed up the retention of water. Upstream of the site, trees were hinged into the channel to add roughness and habitat variation and to slow the water entering the floodplain. Monitoring of water levels is being carried out across the site by the universities of Exeter and Nottingham, and Umeå in Sweden to build a picture of how well water is being retained. Current data suggests that the flood peak has reduced by 38% downstream of the restoration site. Other biological monitoring and terrestrial monitoring (e.g. moth traps) are being undertaken to understand the uplift in biodiversity on site. 

From a regulatory viewpoint, the major concerns with the Stage 0 approach were around fish passage and sediment transport — reasonable worries about a technique infilling a river reach. However, the fact that the project was signed off says a great deal about how the technique is evolving to shift mindsets about restoration of rivers in the UK.

Site 3 — Holt Ball Farm: joining the floodplains

Our third stop was the most recent floodplain reconnection completed in 2025 at Holt Ball Farm, where the aim was to link two floodplains and create a mosaic of wetland and wood pasture. Because this is again an ordinary watercourse rather than the main stem, the consenting was less onerous, land drainage consent for infilling the watercourses, and no consent needed to break through the subsurface drains. At around £30,000, it shows how much cheaper this work becomes once you're off the main river and up the system. Coir rolls and large wood were utilised to form the floodplain roughness features intended to slow the flow and spread it out across the floodplain. 

Here, learning from their first attempt at stage 0, the subsurface drains were smashed through, sealed with clay, then backfilled with broken pipe and earth and compacted down. Finding the main drain first, simplified blocking the smaller feeder drains. Traditional dowsing rods and old field drain maps along with Lidar were used to detect field drains. Some shallow ponds and scrapes are visible containing surface water to the south end of the site. The trees planted on site here, including rare black poplars, are protected from browsing deer with spiral "cactus" guards, a heavy duty tree protection not observed prior to this site visit. 

Site 3 - recent 2025 stage 0
Site 3 - recent 2025 stage 0

This site also contains another breached beaver enclosure. A 45-metre beaver dam and lodge were built within this enclosure in 2022/23, prior to the birth of the pairs first kits. The dam is blocking the majority of flow into the small stream and the water backed up on site has created rich biodiverse areas such as, lakes, backwaters and wet woodland. There are currently six pairs of beaver located on the Holnicote estate, with an ultimate plan for 24 individuals, drawn from a mixed Bavarian, Polish and Swedish gene pool. In order to ensure that genetic variation and integrity is maximised, 

Beaver dam, 45m long
Beaver dam, 45m long

the Beaver Trust keeps a studbook to maximise genetic variability when pairing. Anecdotal evidence shows water now stands in parts of the farm that were previously bone dry for most of the year, building drought- and flood-resilience into the water systems here.

The beavers aren't the only reintroduction, either. Around 300 water voles were released at the bottom of the catchment in 2018/2019; given their roughly two-year lifespan, the voles trapped and moved off the Site 2 restoration were almost certainly the descendants of those original animals.

Site 4 — Holnicote Meadow: an optional extra

The day finished with an optional bonus stopping at Holnicote Meadow, and it was a lovely note to end on. Historical ponds had been dug here some time ago, but it's the beavers that have brought a new lease of life: by damming the Aller tributary, they've raised the water level in the over deep channel by around a metre, enabling water to spill over and connect up with the old lakes in the floodplain. The result is a stretch of open water alive with insects and wildfowl, including a pair of swans that have recently taken up residence. It was a repeat demonstration of the same principle running through the whole estate, albeit instigated entirely by beaver ecosystem engineers — give water room and space, and species will naturally thrive. 

Optional visit, holnicote meadow
Optional visit, holnicote meadow
Beaver activity and old enclosure
Beaver activity and old enclosure
Downstream of the beaver dam
Downstream of the beaver dam

Conclusions: take away messages for Stage 0 technique

Stage 0 rewards boldness tempered by learning. Each Holnicote site improved on the last, and the practical lessons stack up neatly for anyone thinking of attempting it. These are set out here:

  • Fill high and compact hard — then compact some more. Settling soil needs to be higher than the original channel bank top following compaction to ensure limited incision.
  • Remove/block buried field drains - Sub-surface field drains will quietly undo your work by piping water straight back to the channel; break them, seal with clay, backfill and compact.
  • Over-plant aquatic species - You cannot put in too many wetland plug plants into bare soil, and planting beaver friendly food sources upstream encourages ecosystem engineering
  • Make the flow as diffuse as possible, especially on a gradient — fully redirect water away from the infilled channel and spread it wide across the floodplain.
  • Work at catchment scale — improve roughness and habitat upstream with tree hinging, and let ecosystem engineers (pigs, beavers) carve complexity you'd otherwise have to dig by hand.
  • Cost scales with order of watercourse. Around £5,000 on a first-order tributary, roughly £30,000 to join floodplains on an ordinary watercourse, and around £600,000 on the main river — the larger the river the greater amount of earth works and consent/permits required.

And one last observation worth stating; it helps enormously to be the landowner. As the National Trust, the team can plan grazing, drainage, planting and reintroductions across the estate, with the help of tenant farmers, and with that stability comes the ability to take a much longer view of the entire system. Stage 0 is replicable — but the institutional freedom to reset a whole catchment is part of what makes Holnicote such a remarkable place to learn from. Many thanks for a thoroughly enjoyable and informative visit from start to finish!