Providing evidence to improve practice

The journal, Conservation Evidence

Our online journal publishes research, monitoring results and case studies on the effects of conservation interventions. All papers include some monitoring of the effects of the intervention and are written by, or in partnership with, those who did the conservation work. It includes interventions such as habitat creation, habitat restoration, translocations, reintroductions, invasive species control, and education or integrated conservation development programmes, from anywhere around the world.

A volume is created each year with peer-reviewed papers published throughout the year. We now accept Short Communications as well as standard papers.

Special issues contain new papers on a specific topic.

Virtual collections collate papers published in the journal on specific topics such as management of particular groups of species.

To search for papers on a specific topic within the journal select 'Advanced search' on the Home page, enter your keyword(s) and within the Source box type: "conservation evidence".

Invasive plants

This virtual collection contains 25 papers on the management of invasive plants.

Crassula helmsii (Australian swamp stonecrop or New Zealand pygmyweed) was first recorded growing on Mile Cross Marsh in Norwich, Norfolk, UK before 2003. Previous management undertaken to prevent the spread of this plant across the site had been unsuccessful. A three-phase project was undertaken in 2012 to control C. helmsii on Mile Cross Marsh. The aim was to eradicate C. helmsii from two infected ponds and reduce the risk of plants spreading through the boundary onto Sweetbriar Marsh Site of Special Scientific Interest. The control programme included herbicide application and in situ burial due to the high level of pond infestation. The project successfully achieved its aim of eradicating C. helmsii from the infected ponds and preventing the further spread of C. helmsii on Mile Cross Marsh.  However additional work will be required to fully eradicate the plant from the site.


Feasibility of using glyphosate to control beach evening primrose Oenothera drummondii in heavily invaded coastal dunes, Odiel Marshes, Spain
García-de-Lomas J., Fernández-Carrillo L., Saavedra C., Dana E.D., Rodríguez C. & Martínez E. (2016), 13, 72-78

Beach evening primrose Oenothera drummondii is a perennial plant native to the southern USA and adjacent parts of Mexico that invades coastal habitats in several countries. There are currently no accepted control methods. We conducted a seven-month controlled field trial using the glyphosate herbicide Roundup® Ultra Plus in the Odiel Marsh Nature Reserve, Huelva Province, southern Spain. Different herbicide concentrations were tested by knapsack spraying. We estimated the costs of treating an entire invaded nature reserve in southern Spain where O. drummondii has invaded 123 ha of land. A dose of 20 g active ingredient/litre was the minimum effective dose for this species in coastal dunes. As new seedlings appeared after a single herbicide treatment, periodic treatments would be necessary to maintain the population level below an impact threshold. However, the total glyphosate input (710 kg active ingredient/year) to the Reserve for an indefinite period may give rise to social rejection, and demands for the assessment of ecotoxicological impact on native fauna, adjacent habitats and site uses before initiating control actions at full scale. The control costs of the entire 123 ha invaded area for two herbicide applications/year were estimated at €162,000/year (€1,317/ha/year). This includes materials (30% of total costs) and workers (70% of total costs). The study highlights the difficulties and constraints of controlling advanced stages of invasions.


Sisal Agave sisalana is an invasive alien plant species of concern at the UNESCO World Heritage Site of Aldabra Atoll in the Seychelles. Physical control efforts since the 1970s to remove sisal from Aldabra have only been partially successful because the roots cannot be completely removed, resulting in continuous control efforts. We conducted a seven month herbicide trial, using different herbicide concentrations with two application methods, to determine the most effective and feasible control method for sisal. We also checked effects on surrounding native plants. The highest treatment mortality was from 50% herbicide concentration applied directly to the cut growth tip, which resulted in 80% sisal mortality after four months. Fewer treated plants died at lower herbicide concentrations and more small plants died than large plants. No sisal plant died that was foliar sprayed, only cut, or in the control group. There were no visible negative effects of any treatment on the surrounding native flora. The results indicate that chemical control of sisal is effective at high herbicide concentration applied directly to the cut growth tip. A full-scale eradication of sisal from Aldabra has been started based on the trial results.

The objective of this work was to test the effectiveness of reducing bracken density by cutting once or twice a year, or by hand-pulling, compared with a control. The experimental site had already been managed by annually cutting bracken for about 10 years. One year after the management treatments were applied to the plots, both the cut treatments had significantly shorter fronds than the control, but no difference was seen with the pulling treatment. Frond densities and frond coverage were not significantly different from the control, with each showing high variability between plots.

Crassula helmsii is a semi-aquatic, invasive macrophyte, which has become abundant in wetland habitats across Europe. This species is of conservation concern because heavy invasions form dense carpets within which few other plants species occur. C. helmsii is known to be killed by inundation with seawater, but published information on its response to inundation by less saline water is limited. Growth trials were conducted to investigate the levels of salinity required to kill this species. We found a linear negative relationship between growth rate and salinity across the range from 2 – 8 ppt, but that 8 ppt was required to kill C. helmsii. These findings suggest that C. helmsii growth could be controlled by inundation with saline water of 8 ppt. This may present a method for reducing the negative effects of salt water on co-occurring species, and thus the next stage will be to determine the efficacy of this method in field trials.


Eradication of New Zealand flax Phormium tenax on Inaccessible and Nightingale Islands, Tristan da Cunha
Ryan P.G., Glass P.G., Glass T., Barendse J. & Cuthbert R.J. (2012), 9, 58-62

New Zealand flax Phormium tenax was introduced to Tristan da Cunha, an island in the central South Atlantic Ocean, in the 1800s. During the following century it was transferred to two other islands in the Tristan archipelago: Inaccessible and Nightingale Islands. Although not an aggressive invader, flax spread on both islands threatening their status as among the least disturbed temperate islands remaining in the Southern Ocean. In 2004 an eradication programme was initiated on both islands to clear flax using a combination of uprooting, cutting, crushing and spraying with herbicide. Despite some regrowth, follow-up operations greatly reduced the number of flax plants. Established plants are now confined to about 300 m of cliffs at the Waterfall on Inaccessible Island where clearing is hampered by the steep terrain. Further follow-up management is planned until the plant is eradicated from both islands.

Control of giant reed Arundo donax on Vila Franca do Campo Islet, Azores, Portugal
Silva C.M.N., Silva L., Oliveira N., Geraldes P. & Hervías S. (2011), 8, 93-99

The non-native, invasive giant reed Arundo donax covers an estimated 30% of Vila Franca do Campo Islet (Azores). It blocks the entrance of Cory's shearwater Calonetris diomedea borealis nest burrows and out-competes threatened Azorean endemic flora. Three A. donax control methods were tested in 90, 1m2 plots, and cost-effectiveness of each determined using a Simple Additive Weighting Model. The most effective control method was cutting and removal of reed stems followed by two glyphosate-based foliar herbicide applications (one in May and another in late October i.e. corresponding to before and after the Cory's shearwater breeding cycle). After one year, 92% of giant reed was eradicated at an estimated cost of €8,000 per hectare.

In 1998, procumbent pearlwort Sagina procumbens a non-native and invasive plant was discovered on Gough Island (Tristan da Cunha). Efforts to eradicate this species have been underway since 2000. To date it has been restricted to a small (400 m length; approximately 1.2-1.6 ha area) but complex stretch of coastal cliffs. Measures of seed density, based on germination trials of soil samples collected from Sagina infested areas, indicate that 'standard eradication' methods (digging up individual plants, heat-treating soil and spot-treatment with herbicides) in place over the last 10 years have resulted in a three orders of magnitude reduction in the seed load. However, around 200 seedlings per m² continue to be recorded due to germination from seeds within the soil. In 2008-09 we investigated the effectiveness of three methods designed to eradicate Sagina on Gough: 'standard eradication' based on methods used in previous years (n = 5 plots); monthly 'herbicide treatment' across the whole plot (n = 4 plots); and 'soil stripping' where all plants and soil was removed down to bedrock (n = 2 plots). Sagina plants remained present within the standard eradication plots, with an average of 64 ± 79 (range 6 - 200) plants recorded per plot cleared in 8-months of monitoring. No Sagina plants were found in herbicide treated plots, although this method is unlikely to tackle the problem of dormant seeds remaining within the soil. Soil stripping was effective at removing the seed bank and only five and 33 plants were found over the 8-months monitoring period. We recommend that a combination of monthly herbicide spraying across the whole infested area (to prevent plants maturing and setting seed) and a programme of soil stripping (working from the outer edge of the plants range) to remove the seed bank, be utilised in order to provide a potential method to eradicate Sagi from Gough.

The invasive New Zealand pygmyweed Crassula helmsii was eradicated from approximately 120 ha of coastal grazing marsh at a site in southeast England by shallow flooding of the area with sea water for 12 months. This method of eradication can only be used where saline water can be held on a site (with due regard for potential impacts on non-target species). We have not come across an example of successful C. helmsii eradication on this scale by using other methods.

In the late 1990's, The Nature Conservancy (TNC) documented a rise in five invasive plant species, barberry Berberis thunbergii, bittersweet Celastrus orbiculatus, garlic mustard Alliaria petiolata, buckthorn Frangula alnus, and honeysuckle Lonicera morrowii on the periphery of the relatively intact and uninvaded 14,600 ha Berkshire Taconic Plateau in Massachusetts (USA). The Plateau comprises an ecologically significant block of forest. In response, TNC began a large-scale herbicide-based control program on approximately 3,600 ha of land with the goal of reducing invasive cover to less than 10%. Our objective was to evaluate the efficacy of this effort, but this was hampered by a dearth of untreated control sites and pretreatment data on invasive species cover. Four sites (three treated, one untreated) on the plateau periphery similar in understory vegetation, overstory cover, slope, and proximity to a hiking trail were surveyed and compared. Across each site, native and invasive plant percent cover within 44, 1m² plots was measured, and native and invasive presence absence recorded on an additional 2,000m² area. All five target invasives were present at all four sites 5-years post-treatment. In two of the treated sites, invasive percent cover significantly exceeded the 10% goal, largely due to the abundance of garlic mustard. Without garlic mustard, all the sites (including the untreated one) had < 10% invasive cover. Surprisingly, the high level of invasive cover did not have a significant negative impact on native cover (native species richness was not quantified), although a hypothesized negative relationship was invoked as justification for the herbicide treatment.

Given the difficulty in finding comparable treated and untreated sites after herbicide application, we suggest 1) that quantitative data on invasive abundance be gathered prior to a control program and 2) that treated and untreated plots be allocated to monitor outcomes. Without this, determination of effectiveness is difficult and likely to be inconclusive.

Cutting, removal and herbicide stump-treatment of dense grey willow Salix cinerea scrub from a 1 ha wet dune-slack was undertaken in a northwest England National Nature Reserve. This resulted, over the next two years, in colonisation by 139 vascular plant taxa. Of these, 11 are nationally or regionally notable, with 28 being new reserve records. The high proportion of ruderal plants in the first year was largely replaced by dune species in the second season after scrub removal.


In a 7-year field experiment undertaken in western Norway, the efficiency of four bracken control measures on a heathland was investigated: application of two herbicides i) Asulox®- and ii) Gratil®-with follow-up annual cutting; iii) annual cutting; and iv) biannual cutting. Assessments were also made as to what extent the characteristic species composition and vegetation structure of heathlands were restored, and effects of the herbicides on non-target plant species commonly found on heaths. Fastest reduction in bracken cover resulted from herbicide application, but cutting proved equally efficient in the longer term; Asulox and biannual cutting both reduced bracken cover from over 70% to below 10% in 2 years, while annual cutting achieved this in 5 years. Gratil failed to have long-term effects. Species composition progressed towards a desirable heathland vegetation community, but successional trajectories differed, and Asulox had minor unintended effects on a number of heathland plants, including heather Calluna vulgaris, several grasses, herbs and mosses. These effects could not be predicted by functional group or other simple species characteristics. However, any short-term detrimental effects of Asulox application were considered to be outweighed by the beneficial longer term effects of reduced bracken cover, which allowed re-establishment of the heathland flora.

On recently created wet grassland at a site in eastern England, willow Salix spp. was invading. It was cut then 30 adult female Hebridean sheep and a Texel ram were introduced to control any regrowth. The sheep have kept the site clear of both newly sprouting willow shoots and willow seedlings.

Non-native floating pennywort Hydrocotyle ranunculoides had colonised and grown to dominate parts of Gillingham Marshes, eastern England, where it was outcompeting native plants. Removal was undertaken using a mechanical digger and by monthly picking by hand. This greatly reduced its cover but did not completely eradicate it. The native aquatic vegetation is re-establishing.


Three methods of New Zealand pygmyweed Crassula helmsii control were undertaken: using 'Waipuna' hot foam, spraying with Glyphos bioactive, and by smothering and burying. Spraying with either 'Waipuna' or Glyphos killed 50% of the vegetation and stopped Crassula from spreading, but did not eradicate the weed. Smothering and burying killed all plants (100% mortality), but proved very labour intensive and causes much disturbance.

Sweet chestnut Castanea sativa coppice was controlled by five different methods: repeated cutting,amcide poured into drilled holes, herbicide painting of stumps, weed-wiping of first year growth, and knapsack spraying of regrowth. The most effective and efficient method was to cut the coppice stools and subsequently knapsack spray the first year regrowth with herbicide.

Measures were undertaken to attempt to eradicate invasive rhododendron Rhododendron ponticum. Mechanical removal is quick, but expensive (£1,000/day) and has a high potential for damaging the soil and coppice stools. Manual removal is labour intensive (120 man-days/4 ha), but was less damaging and could be applied in areas in which machinery could not be used. Weed-wiping regrowth was slower but more effective (70% kill rate) than spraying (40-50% kill rate).

A programme was introduced to eradicate invasive New Zealand pygmyweed Crassula helmsii by Reglone herbicide treatment. In response, 70% of pygmyweed was killed. Follow up spraying of regrowth was partially successful but some re-growth is noticed annually.


At The Lodge RSPB Reserve, an attempt was made to eradicate invasive New Zealand pygmyweed Crassula helmsii that had colonised a pond by covering with black polythene. Pygmyweed was completely eradicated but re-colonised from plants in marginal vegetation in areas not covered with polythene.

An area of silver birch Betula pendula (40% cover) was cleared during the winter of 1995-1996. Two summers later, the cleared area and an adjacent area of open heath were sprayed with Asulox herbicide to control invasive bracken Pteridium aquilinum growth. Seven years after the last spray treatment, bracken was dominant throughout the area (median cover in sample plots 80%; cover estimate range 40%-90%). Whilst achieving short-term success, longer term evidence therefore suggests that bracken management needs to be ongoing at this site.

At a site in southern England in September 2004, about 4 ha of mature, dense, non-native, maritime pine Pinus pinaster was cleared using a shear-head timber processor. One year later in August 2005, the cleared area was predominantly covered in purple moor-grass Molinia caerulea (approx. 80% cover). The remaining 20% was either bare ground (approx. 15%) or mature heather Calluna vulgaris and gorse Ulex plants (less than 5% cover) present prior to clearance, but no heather seedlings were found. There was no evidence of any pine regeneration.

In August 2004, an area of just over 1 ha of dense, mature Scots pine Pinus sylvestris was cleared using a shear-head timber processor. Prior to the clearance, the area supported little heathland vegetation and was predominantly bare ground with some bracken Pteridium aquilinum underneath the dense pine canopy. One year after the clearance, heather Calluna vulgaris seedlings had become established. Evidence suggests that control of bracken and tree seedlings may be required as part of long-term management in order to restore and maintain an open heath.


Approximately 0.5 ha of dense, mature Scots pine Pinus sylvestris and rhododendron Rhododendron ponticum was cleared using a shear-head timber processor. Prior to clearance the area under the canopy supported little heathland vegetation. One year later, heather Calluna vulgaris seedlings had become established; other heathland vegetation included purple moor grass Molinia caerulea and cross-leaved heath Erica tetralix. There was also evidence of vigorous rhododendron regrowth from the cut stumps.


Two different herbicides were used to treat young Rhododendron ponticum plants on an area of heathland in June. Three months later, of 157 plants sprayed with Glyphagen (Roundup), all but 11 had died (93% kill rate), while of 161 plants sprayed with Garlon with additions of Mixture B (a wetting agent), all but five had died (97% kill rate).


During the winter of 2000-2001, approximately 2.4 ha of dense (75-100% tree cover) and 0.6 ha of medium (25-50% tree cover) birch Betula was cut and removed. In August 2002, the cleared birch had regenerated resulting in a dense stand of 2 m high birch. The regrowth was sprayed with Timbrel using knapsack sprayers. The regrowth kill rate of sprayed areas was 100%; no regrowth was subsequently recorded.