Conservation Evidence strives to be as useful to conservationists as possible. Please take our survey to help the team improve our resource.

Providing evidence to improve practice

Action: Restore/create species-rich, semi-natural grassland Farmland Conservation

Key messages

Read our guidance on Key messages before continuing

Supporting evidence from individual studies


A replicated controlled trial in Denmark, from 1974 to 1978 (Bülow-Olsen 1980) found that re-introducing cattle grazing on abandoned agricultural grassland reduced the frequency of heather Calluna vulgaris and wavy hair grass Deschampsia flexuosa, but did not consistently increase the number of plant species. There were increases in the number of plant species under intensive grazing in the area that had the fewest plant species at the start of the experiment, abandoned 14 years before, but colonization was very slow. Wavy hair grass decreased the most under normal grazing, reducing from 92 to 70% cover on average, intensive grazing had a lesser effect. Heather reduced the most under intensive grazing (from 71% to 6% cover on average). A 59 ha area of hill pasture was subjected to four levels of grazing intensity: ungrazed, lightly grazed, normally grazed and intensively grazed. Grazing took place two to four times a year, in late spring and late summer. There were three fields: one abandoned in 1910 (64 years before), one in 1960 (14 years before) and one grazed every year since 1910. Vegetation was sampled in July each year on 52 x 1 m2 quadrats across all treatments.



A small study of species-poor grassland in the Netherlands (Bakker et al. 1983) found that preferential grazing by sheep gave rise to a ‘macro-pattern’ of various plant communities absent under a hay-making regime. There were short, heavily grazed patches interspersed with taller, lightly grazed patches, a micro-pattern that tended to be stable after initial establishment. Heavily and lightly sheep grazed areas differed little in plant species composition, but abundances of species differed considerably. Heavily grazed areas were characterized by equal amounts of monocots (mainly grasses) and broadleaved species, had higher abundances of rosette species and, to a lesser extent, greater persistence of perennial rye grass Lolium perenne. Lightly grazed patches were dominated by common bent grass Agrostis tenuis and had a large amount of plant litter. Prior to 1972, the grassland was cut for hay and aftermath grazed by cattle. Sheep grazed part of the grassland (3 sheep/ha) from 1972 in July-December and January-July. After 1980, the area was left ungrazed for 2 months each winter. Vegetation was recorded annually within permanent 2 x 2 m quadrats. Grazing intensity was recorded in February as lightly grazed (>70% litter cover), heavily grazed (<30% litter cover), or intermediate. Species abundance was quantified on a dry weight basis. Vegetation was also mapped from October 1979-1982 in a 10 x 10 m² area and recorded as heavily grazed (<5 cm tall) and lightly grazed (>10 cm tall).



A 1985 review of four techniques aimed at reducing soil fertility for the restoration of semi-natural grassland or heath (Marrs 1985) found that there was some success. Addition of inorganic nitrogen to a crop was found to extract more phosphorus and other elements from the soil than treatments with no fertilizer addition. Fertilizer addition almost completely exhausted the soil phosphorus store in long-term wheat experiments. Three studies investigating nutrient removal by stubble burning were reviewed (Allen 1964, Kenworthy 1964, Chapman 1967). Burning decreases phosphorus availability and increases potassium availability. Topsoil stripping reduces fertility, but can result in the loss of a large proportion of the seed bank.

Additional references:

Allen S.E. (1964) Chemical aspects of heather burning. Journal of Applied Ecology, 1, 347-367.

Kenworthy J.B. (1964) A study of the changes in plant and soil nutrients associated with moorburning and grazing. PhD thesis. University of St Andrews.

Chapman S.B. (1967) Nutrient budgets for a dry heath ecosystem in the south of England. Journal of Ecology, 55, 677-689.


A replicated, controlled study of an abandoned arable field in Oxfordshire, UK (Gibson et al. 1987a) found that spring grazing by sheep increased plant species richness. However, this was only evident at intermediate to large sampling scales, suggesting that differences arose from changes in rarer and widely dispersed species. Many annual herbs decreased in ungrazed paddocks in 1985, whilst some annual grasses and other perennial herbs and grasses increased. In grazed (spring or autumn) paddocks, many more species increased their distribution compared to ungrazed plots, particularly annual herbs. The field (12 ha) had been permanent pasture until 1960 and was then cultivated until 1981. In 1985, the central 90 x 90 m blocks of two 1 ha blocks were divided into nine paddocks, i.e. three replicates of three treatments. Vegetation was sampled within various quadrats by several methods in April, July, August and October 1984-1985. This study was part of the same experimental set-up as (Gibson et al. 1987b, Watt & Gibson 1988, Brown & Gibson 1994).



A replicated controlled study of an abandoned arable field in Oxfordshire, UK (Gibson et al. 1987b) found that plant species establishment was better in grazed areas and species richness, diversity, and abundance of individual species were also higher compared to ungrazed controls, however there were no apparent effects of sheep grazing treatment on the likelihood of new plant species arriving in the grassland. Within 18 months, 57% of plant species restricted to patches of old chalk/limestone (calcareous) grassland within 2 km had colonized the study field. Colonization was not thought to have come via the seed bank. In addition, the area grazed April-November contained many component species of mature chalk/limestone (calcicolous) grassland, unlike ungrazed controls. The 12 ha field had been permanent pasture until 1960 and was then cultivated until 1981. In 1985, two blocks of 1 ha were divided into nine paddocks, three replicates of spring grazed, autumn grazed and ungrazed treatments. Additional treatments were grazing from April-November with a short break in summer and continuous grazing August-November. Vegetation was sampled within permanent quadrats (1 m²) from April 1984 to October 1986. This study was part of the same experimental set-up as (Gibson et al. 1987a, Watt & Gibson 1988, Brown & Gibson 1994).


A replicated, controlled, randomized study in 1973-1974 of limestone grassland in Cambridgeshire, UK (Morris & Rispin 1987a) (also described in an additional publication (Morris & Rispin 1987b)) found that cutting tended to benefit plant-eating beetle (Coleoptera) species, whereas most beetles that feed on decaying matter, detritus-feeding beetles, fungus-eating beetles and some predatory beetles were more abundant in uncut plots. Beetle families that feed on decaying matter, and predatory beetle families were next most abundant in May-cut plots. Overall species diversity was highest on uncut plots and then July-cut plots. No beetle species showed a response to treatments in all five sampling periods. A significant reduction in abundance was recorded for 17 species, whilst 12 increased under one or more treatments. Treatments were an annual cut in May or July, or both, and an uncut control, with four replicates (each 16 x 12 m) of each in a randomized block design. The site had previously been unmanaged for several years. Beetles were sampled by vacuum netting 1 m² in each plot at 2-4 weekly intervals from October 1972 to December. Berlese-type funnels were also used on one turf sample (each 0.07 m²) per plot/week over four weeks in each of five sampling periods in 1973-1974.

Additional reference:

Morris M.G. & Rispin W.E. (1987b) Abundance and diversity of the coleopterous fauna of a calcareous grassland under different cutting regimes. Journal of Applied Ecology, 24, 451-465.


A replicated trial in 1986 and 1987 at the Chilworth Research Centre, Hampshire, UK (Fenner & Spellerberg 1988) tested survival of twelve native plant species, either grown in pots or sown as seeds on an existing meadow. Three species, black knapweed Centaurea nigra, St John’s wort Hypericum perforatum and musk mallow Malva moschata always survived well (83% or more survived). Five species survived better when sown as seeds (oxeye daisy Leucanthemum vulgare, self-heal Prunella vulgaris and lady’s bedstraw Galium verum) or when planted as plug plants (betony Betonia officinalis and cowslip Primula veris). Four species had poor survival in both treatments: yellow rattle, Rhinanthus minor, field scabious Knautia arvensis (both 0% survival), harebell Campanula rotundifolia and bulbous buttercup Ranunculus bulbosus (0-8% survival). Plants sown as seeds into cleared plots were almost always larger after two growing seasons than those planted out as plug plants. Sown plots were cleared with herbicide in March 1986. Seeds were covered with 2.5 cm of compost, a perforated polythene cloche, or both. There were three replicates of each treatment. Results showed no difference between the germination or survival of plants under different coverings. Twenty-four plants of each species were planted out into a meadow where the vegetation was 10-15 cm high in July 1986, at 1 m intervals. The growth of sown and planted plants was monitored in September 1987.



A replicated, controlled study in 1986 and 1987 of an abandoned arable field in Oxfordshire, England (Watt & Gibson 1988) found that sheep grazing increased seedling establishment compared to ungrazed plots. The most heavily grazed treatment had the highest levels of seedling establishment, whereas few new seedlings were recorded on ungrazed paddocks. Treatments with some autumn grazing had a peak of seedling establishment the following spring. Seedling survival was not affected by grazing treatment or gap size. The two short-grazing treatments (lasting 10 days) had the least bare ground whilst April-November grazed areas had the most. Insecticide use increased seedling establishment in October in ungrazed and spring-grazed paddocks but decreased establishment in autumn-grazed paddocks. In 1985, three treatments were applied in six replicate (30 x 30 m) paddocks: 10 days grazing in spring or autumn, and ungrazed controls. Two 3 x 3 m permanent quadrats were treated weekly with malathion insecticide in each paddock and four permanent 1 x 1 m sampling quadrats were established. Another two paddocks were grazed from April-November with or without a short break in summer, twelve 1 x 1 m quadrats were established in each. Gap type and seedling sampling was undertaken in all quadrats seven times from April 1986 to July 1987. Vegetation height was recorded in September 1986. This study was part of the same experimental set-up as (Gibson et al. 1987a, Gibson et al. 1987b, Brown & Gibson 1994).



A replicated, controlled study of grassland restoration within four geological regions in southern Germany (Kaule & Krebs 1989) found that in the first year, natural regeneration following topsoil removal had higher species diversity than sown treatments (47 vs 33-38 species), but sowing tended to reduce pioneer species and perennial weeds. In the first year, the most successful establishment was of nutrient-rich meadows and tall herbaceous vegetation in existing arable soils (no soil removal). Topsoil removal enhanced establishment of semi-dry grassland species (37 vs 33 species), but not other communities. Complete soil removal favoured pioneer, semi-dry grassland and neutral (mesotrophic) edge communities. Isolated plots within fields had lower numbers of species than plots adjacent to existing edge habitats for all treatments. Six communities were sown in plots of 10 x 5 m: control (existing seed bank), semi-dry grassland, nutrient-rich grassland, neutral edge communities, pioneer vegetation and tall herbaceous vegetation. Seeds were collected within a 10 km radius. Establishment methods were complete soil removal, topsoil removal or existing arable soil. Results are from year one of 15 years of monitoring.



A controlled, replicated trial from 1982 to 1993 on an abandoned ex-arable field at Oxford University Farm at Wytham, Oxfordshire, UK (Brown & Gibson 1994) found that 10 years after abandonment, heavily grazed treatments (particularly spring-and-autumn grazing, at 3-6 sheep/ha) more closely resembled target ancient chalk/limestone (calcicolous) communities than lightly grazed or ungrazed treatments. Over the experimental site, 250 plant species colonized, 77 of which were typical of chalk/limestone grassland. However, the species composition of the site differed markedly from that of nearby ancient chalk/limestone grassland (where later-successional and stress-tolerator species were more common), indicating that restoration may take decades. Arable cultivation was abandoned in 1982 and five grazing treatments began in 1985. Three treatments were replicated six times in 30 x 30 m paddocks (ungrazed control, short-period spring and short-period autumn grazing) and two treatments were applied in larger areas (spring-and-autumn grazing and long-period autumn grazing, not replicated). Plants were surveyed four times a year in 12 quadrats (1 m2) in each replicate and in nearby ancient grassland patches. This study was part of the same experimental set-up as (Gibson et al. 1987a, Gibson et al. 1987b, Watt & Gibson 1988).



A replicated controlled trial from 1984 to 1990 at Little Wittenham Nature Reserve, Oxfordshire, UK (Bullock et al. 1994) found that plant composition on a previously improved pasture hardly changed in response to reduced sheep grazing intensity and no fertilizer. Plant species diversity was still low after six years. The vegetation remained dominated by perennial grasses, with four species making up 80% of records. Herbaceous plants (non-grasses) made up just 0.4% of records. Seventy percent of seedlings growing in artificial gaps in the grass cover were of two grass species, perennial rye grass Lolium perenne and meadow barley Hordeum secalinum. Only 4% of seedlings were non-grass species, and none were species not already found in the paddocks. There was no evidence of a seed bank (gaps with original topsoil did not differ from gaps with topsoil replaced by sterile soil). There were eight levels of sheep grazing: summer grazing to a height of either 3 cm (more intensive) or 9 cm (less intensive), with or without winter and/or spring grazing, but grazing intensity had only small effects on the vegetation. Each treatment was replicated in two 50 x 50 m paddocks. Plants were surveyed using a point quadrat at 64 points/paddock in 1990. Vegetation and topsoil or vegetation-only were removed in September 1990 in five 10 cm diameter circles/paddock and seedlings growing in these areas counted and removed regularly until January 1992.



A study of a former improved grassland over 16 years in the Netherlands (Van Der Woude et al. 1994) found that plant species richness was lower in the fertilized compared to unfertilized plot, even after 16 years. One plot was treated with nitrogen fertilizer in an attempt to remove nutrients, other than nitrogen, from the grassland via hay removal. Although there was a peak in above-ground biomass in the fertilized plot in 1977 (900 g/m² vs 400 g/m²), in 1986, biomass in the fertilized plot started to decrease and by 1990 was the same as in the unfertilized plot (300 g/m²). There were no significant differences in soil chemical variables after 16 years. The authors conclude that fertilizer application as a conservation measure does not seem appropriate for restoring species-rich grassland. In 1972 the grassland was taken out of production, fertilizer addition ceased and vegetation was mown and removed in late July-early August. Two adjacent 20 x 10 m plots were established in 1973, one received nitrogen-fertilizer (50 kg/ha/year) and the other was unfertilized. Plant species composition and above-ground standing crop was sampled from 1972 onwards.



A randomized, replicated controlled trial from 1993 to 1994 in Hampshire, UK (Stevenson et al. 1995) found that hand-sowing chalk grassland plants on rotavated ex-arable plots created a community partly resembling the target plant community after two years. Computer analysis showed a 45-62% fit to a UK National Vegetation Classification scheme chalk/limestone grassland community (CG2a – sheep’s fescue Festuca ovina- meadow oat-grass Avenula pratensis grassland, dwarf thistle Cirsium acaule-squinancy wort Asperula cynanchica subcommunity). Higher sowing rates gave a better fit, and a higher percentage cover of chalk grassland plants (from 10% cover at the lowest sowing rate to 100% at the highest rate). The three higher sowing rates had similar numbers of chalk grassland species in the plots (28, 30 and 31 species respectively), by 1994. Control plots and plots sown at 0.1 g/m2 had around six and 20 species chalk grassland species respectively. Seed was sown at 0.1, 0.4, 1 or 4 g/m2. The mixture contained 22 grass/sedge species and 25 herb species. Each rate was sown in four replicate 6 m2 plots, and four control plots were not sown. Plots were rotavated in March 1993, sown by hand, raked and left unmanaged (lightly grazed by rabbits). Plant cover was measured in two 1 m2 quadrats/plot in August 1993 and 1994.



A trial at the Crichton Royal Farm, Dumfries, Scotland (Blake et al. 1996) found more ground beetle (Carabidae) species at the restored wildflower sites (18-26 species/line of traps) than in more intensively managed grassland (17-23 species) or unmanaged grassland (15-16 species). However, an index of ground beetle diversity was highest at the unmanaged grassland nature reserve site. Beetles found there were larger, and a different set of species. The authors concluded that the beetle community found in natural grassland habitat in the area had not re-colonized the restored species-rich grasslands, even after five years. Two fields were ploughed and sown with 17 plant species in August 1987 (five grasses, two clovers Trifolium spp. and twelve other broadleaved flowering plant species). They were managed without fertilizers, cut once in July and grazed in autumn and winter. Ground beetles were sampled in 18 pitfall traps (laid out in two lines) in each treatment area, between April and September in 1989 and again in 1993. Ground beetles were also sampled at sites with continuously grazed perennial rye grass Lolium perenne, perennial rye grass fertilized with cattle slurry and mineral fertilizer and cut three times a year, and an unmanaged grassland on a nearby National Nature Reserve.


A 1997 review of experimental evidence (Nösberger & Kessler 1997) concluded that a combination of management changes and introducing plant species is necessary to restore species-rich grassland. A four year study from 1991 to 1995 (Steinegger & Koch 1997) in Switzerland found that a late hay cut and no fertilization did not restore a species-poor rye grass Lolium spp. sward to a species-rich meadow. The number of plant species gradually increased to 40 on plots where additional plant species were repeatedly sown in. Another study  (Lehmann et al. 1996) found removing the existing grass sward and re-sowing with a tailored mix of at least 30 species (12 grasses, 5-7 legumes and 13-16 herbs) was the most reliable and quickest way to restore species-rich grassland.

Additional references:

Lehmann J., Rosenberg E., Bassetti P. & Mosimann E. (1996) Standardmischungen fur den Futterbau. Agrarforschung, 3, 489-500.

Steinegger R. & Koch B. (1997) Naturschutzerische optimierung okologischer ausgleichsflachen. Agrarforschung, 4, 35-38.


A trial near Clavering in Essex, UK (Snow et al. 1997) found that plant species richness increased dramatically on an ex-arable field following ten years of traditional hay meadow management. The number of plant species in the restored meadow increased from 19 species in 1988 to 42 species in 1994, after planting with a commercial rye grass Lolium perenne and white clover Trifolium repens seed mix in 1984. The meadow changed from being dominated by sown rye grass to including grasses such as soft brome Bromus hordeaceus and crested dog’s tail Cynosurus cristatus. Species usually found on species-rich grasslands, black medick Medicago lupulina, not found in 1988, became dominant. Two orchid species colonized (common spotted orchid Dactylorhiza fuschii and bee orchid Ophrys apifera). The 3 ha site was cut once in July, aftermath grazed with around five sheep/ha from August to October and had no fertilizer added. Plants were monitored each June from 1988 to 1994, in 40 randomly placed 0.25 m2 quadrats. Importantly, the site was immediately adjacent to two established hay meadows (present in 1941), in which the number of plant species also increased from 26 in 1989 to 48-57 species in 1994.



A replicated study in 1975-1995 in pastureland in the Archipelago National Park, southwest Finland (Kotiluoto 1998) found that restoration methods including removing trees and shrubs, grazing, pollarding trees and mowing increased the average number of plant species/plot from 32 to 41. This was mostly due to an increase in common species, but the number of old meadow indicator species also increased slightly. Three out of four locally endangered plant species increased in cover but none colonized new areas. Grasses benefited more than broadleaved flowering plants, increasing their overall average cover from 19% before restoration to 26% after. The study used 41 permanent 10 x 10 m plots, 16 of which were in grazed areas, 11 in thinned areas (some of which were also grazed) and 14 in thinned, mowed and grazed areas. Monitoring took place every three to eight years. Older plots were assessed six times and newer ones twice.



A 1998 review of case studies in France gathered from published and unpublished literature on grasslands (Muller et al. 1998) found that the restoration process generally takes eight years or more and does not always work as expected. Fifty-one case studies were identified across France (including 16 on wet grassland). Most were not discussed in detail. On chalk grassland, one study showed an increase in a plant species diversity index over three years in response to mowing (Dutoit 1996). On meadows degraded by engineering work or ski tracks, seven studies were found demonstrating that grasslands can be artificially re-created, but that it takes many years. One study, for example, demonstrated that achieving vegetation similar to the original vegetation took nine years (Bédécarrats 1991). Another demonstrated how, over eight years, sown species gradually disappeared to be replaced by a newly created grassland with a different plant community from the original (Maman 1985, Coin 1992).

Additional references:

Maman L. (1985) La dynamique de la vegetation sur les ouvrages des aménagements hydroélectriques du Rhône. PhD thesis. University of Grenoble, Grenoble, France.

Bédécarrats A. (1991) Dynamique des enherbements des pistes de ski en Savoie et leur gestion pastorale. Pages 77-80 in: IVth International Rangeland Congress, Association Française de Pastoralisme, Montpellier, France.

Coin R. (1992) Variabilité  spatio-temporelle des communautés végétales artificielles sur les ouvrages des aménagements hydroélectriques: enseignements en vue d’améliorer les techniques de végétalisation. PhD thesis. University of Grenoble, Grenoble, France.

Dutoit T. (1996) Dynamique et gestion des pelouses calcaires de Haute-Normandie. Presses Universitaires de Rouen, Rouen, France.


A replicated, controlled study in the winters of 1994-1997 in southern England (Wakeham-Dawson & Aebischer 1998) (same study as (Wakeham-Dawson et al. 1998)) found that Eurasian skylark Alauda arvensis, corn bunting Miliaria calandra and meadow pipit Anthus pratensis were consistently more abundant on arable fields reverted to species-rich chalk grassland than on land reverted to permanent grassland (sown with agricultural grasses), intensively managed permanent grassland or winter wheat fields (25-230 skylarks/km2 on reverted chalk grassland vs 0-11 on other field types, 0.9-4.7 corn buntings/km2 on reverted chalk grassland vs 0-1 on other field types, 3.7-6.1 meadow pipits/km2 on reverted chalk grassland vs 0-4.3 on other field types). Densities of rooks Corvus frugilegus and species richness of plant seeds did not differ across field types. However, there were significantly more plant species on reverted chalk grassland than the other field types (7.8-9.2 species/quadrat vs 1.4-5.1 species/quadrat). Reverted chalk grassland fields were sown with species such as Festuca spp. and Bromus spp. grasses.  Fields on forty farms were surveyed. Birds were surveyed once during December and January on 217 fields in winter 1994-1995, repeated on 205 fields in winter 1995-1996 and on 225 fields in winter 1996-1997. The numbers of grassland birds and types of grazing livestock were recorded. In November in the winters of 1995-1996 and 1996-1997, seeds lying on the ground in 31 fields, were sampled in two 0.25 m2 quadrats/field and identified to species. Plant species were surveyed in four 0.25 m2 quadrats/field in July-August 1994 and 1996 in 121 and 72 fields.


A replicated, controlled study in spring and summer from 1994 to 1996 in southern England (Wakeham-Dawson et al. 1998) (same study as (Wakeham-Dawson & Aebischer 1998)) found that arable fields reverted to species-rich chalk grassland consistently had higher densities of Eurasian skylark Alauda arvensis than land reverted to permanent grassland (sown with agricultural grasses), intensively managed permanent grassland or winter wheat fields (12-23 skylarks/km2 on reverted chalk grassland fields (16-35 fields) vs 3-12 skylarks/km2 on 16-82 fields of other types). Densities of carrion crows Corvus corone and rooks C. frugilegus were not consistently higher on any field type (0.5-1.9 crows/km2 and 0-14 rooks/km2 on chalk grassland vs 0-1.8 crows/km2 and 0-90 rooks/km2 on other fields). Reverted chalk grassland fields were sown with Festuca spp. and Bromus spp. grasses. The other field types studied were arable fields reverted to permanent grassland, downland turf (close-cropped, nutrient-poor grassland), permanent grassland, winter wheat, barley, oilseed rape and set-aside. Fields on forty farms were surveyed. The number and location of singing skylarks were recorded in April-May and June-July in 1994 and 1996 and in May-June 1995. The locations of foraging carrion crows and rooks were also recorded in 1994-1996.



A replicated, controlled trial on six Environmentally Sensitive Area sites in England and Wales (Hopkins et al. 1999) found that sowing 35 to 40 plant species increased the number of broadleaved plant species on all sites relative to the control and the number of grass species increased on five of the six sites by the second year. The sowing treatments with the most soil disturbance; rotavating and de-turfing, established the most species. In year one, there was an average of seven broadleaved species and seven grass species on the control plots, compared to sixteen broadleaved species and ten grass species on the deturfed treatment, which had the strongest increase in species richness. The trial was carried out in 1994 on 6 x 4 m plots monitored for the following two years, with four replicates of each treatment on each farm. Seven species were successfully introduced by some or all treatments on all sites: yarrow Achillea millefolium, oxeye daisy Leucanthemum vulgare, self-heal Prunella vulgaris and ribwort plantain Plantago lanceolata, black knapweed Centaurea nigra, bird’s-foot trefoil Lotus corniculatus and cat’s ear Hypochoeris radicata. Other species, including yellow rattle Rhinanthus minor, failed to establish.



A replicated, controlled trial in Trawsgoed Research Farm, Aberystwyth, Wales (Jones & Hayes 1999) (partly the same study as (Hayes & Tallowin 2007)) found that seedlings established best, and survived best in plots that were cut twice with aftermath grazing by sheep in winter. The lowest rates of seedling establishment and plant survival (lower than the control) were in plots cut twice but without grazing. The authors conclude that winter grazing is very important when re-introducing plants to restore hay meadows. By September 1996, seeds from the local area; yarrow Achillea millefolium, purple betony Stachys officinalis and self-heal Prunella vulgaris had survived better than non-local seeds, with no difference in two other species (black knapweed Centaurea nigra, ribwort plantain Plantago lanceolata). The five plant species were sown in October 1994, two years after the management experiment began. Fifty seeds of each species and provenance were sown in each of three 1 m2 quadrats/plot. Seeds were either gathered from within 8 km or purchased from a seed supplier (from elsewhere in the UK). Plants were monitored every month after sowing until April 1995, then in April and September 1996.


Two trials on Monte Generoso, Switzerland (Stampfli & Zeiter 1999) investigated the effects of different plant establishment techniques on an abandoned grassland. The controlled trial found that introducing mowing to an abandoned grassland increased the number of plant species after ten years, but the response was slow. The replicated controlled trial found that when 12 native grassland species were sown, fewer seedlings emerged in burned plots (average 5.6 seedlings/species/plot after one year) than in plots mown with litter removal before sowing (average 8 seedlings/species/plot). Plots mown without litter removal were intermediate (6.1 seedlings/species/plot). Mown grassland plots were still dominated by tor grass Brachypodium pinnatum at the end of the experiment. Mown plots increased from a total of 46 plant species in 1989 to 68 species in total by 1996-1997. The total frequencies of all species other than tor grass stayed low for six years, and increased strongly in the final four years of the experiment. The replicated controlled study also found common rockrose Helianthemum nummularium had the lowest seedling emergence (0-0.8 seedlings/plot after one year) and rough hawkbit Leontodon hispidus had the highest (17-24 seedlings/plot after one year). After two years, ten of the sown species survived, with the grasses having the highest survival rates (27-36% of seeds germinated). No bulbous buttercup Ranunculus bulbosus or meadow clary Salvia pratensis seedlings survived. In the controlled study, mowing was resumed in the abandoned grassland in two 100 m2 plots. From 1988, one plot was mown twice a year (July and September-October), one was mown in early July only. A control plot was not mown. Plant species composition was monitored in September-October at 456 point quadrat samples/plot over ten years, from 1988 to 1997. In July 1989 and 1995, percentage cover across each whole plot was estimated for plant species. In the replicated controlled study, forty-five seeds of each species, hand-collected from a nearby meadow were sown into twenty-four 60 x 60 cm subplots in October 1995. Each plot was subject to one of three treatments before sowing (six replicates of each): mowing with removal of mown vegetation, burning or mowing with complete removal of litter. For each treatment, a control 60 x 60 cm plot was treated the same but not sown. Seedling emergence was recorded every one to five months for the following two years. Plots were mown twice in 1996 and 1997.



A replicated, controlled study in 1988-1997 in temperate grassland and seasonal freshwater marshes in Strathrory Glen, northeast UK (Humphrey & Patterson 2000) found that cattle grazing at a stocking rate of 2.2-2.5/ha prevented further decline in plant species richness, but did not promote recovery. In grazed plots, overall species richness remained static and moss/liverwort richness increased, whilst in control plots species richness declined. However, cattle grazing had no effect on species cover and plants of conservation importance showed no increase. The study used four plots in each of three different vegetation types: acid grassland, rush pasture and vegetation associated with calcareous springs (seepage flush). Each plot had two 10 x 10 m subplots, one of which was fenced to exclude cattle. Sampling was carried out in 1988 before the start of grazing, and again in 1991 and 1997, using four permanent 2 x 1 m quadrats within a central 7 x 7 m area in each plot.



A replicated trial from 1990 to 1998 of combined management treatments on an agriculturally-improved meadow in the Pennine Dales Environmentally Sensitive Area, North Yorkshire, England (Smith et al. 2000) (same study as (Smith et al. 2002)) found that the highest increase in plant species diversity was achieved with a combination of autumn and spring grazing, 21 July hay cut date and sowing native plant species. The study took place in 6 x 6 m plots on a 2.75 ha meadow within the Ingleborough National Nature Reserve. Plots were either sown with many native grassland species (including yellow rattle Rhinanthus minor) or not. The experiment also included three different grazing treatments (sheep and cattle), plots with or without fertilizer and three earliest dates for hay cut. Yellow rattle spread to most plots after its introduction as a constituent of the seed addition treatment. By 1996 it was particularly abundant in treatment combinations that included autumn grazing, no mineral fertilizer and a July hay cut. Populations of over 40 plants/m² were associated lowest hay yields, presumably as it suppressed grass growth.



A replicated study in 1995-2000 in 11 sites with newly created flower-rich meadows on set-aside land in central Switzerland (Bosshard & Kuster 2001) found generally higher butterfly (Lepidoptera) densities on the created meadows (e.g. 170 individuals/ha in Riedikon) than on intensively managed arable land. However, the highest densities were found in flower-rich field margins (e.g. 440 individuals/ha in Seewadel). Adult butterfly abundance was positively correlated to the number of flower units, and up to 98% of flower visits were recorded on only five plant species. Note that no statistical analyses are presented in this study. On three main sites (Berg, Riedikon and Seewadel), flower-rich meadows were established on 0.5-0.6 ha experimental plots. Butterflies were recorded during seven visits from May-September 1999 along fixed transects. In an additional eight sites, butterflies and grasshoppers (Orthoptera) were monitored in 1995-2000 approximately twice a year.



A randomized, replicated controlled trial at Hill Farm, Little Wittenham, Oxfordshire, England (Coulson et al. 2001) found that both yellow rattle Rhinanthus minor and oxeye daisy Leucanthemum vulgare could be effectively established on a pasture field by ‘slot seeding’. Different management treatments, cutting, grazing or both, did not affect the survival or establishment of either species, but yellow rattle seeds were spread a greater distance when hay was cut in July than without a hay cut. Seeds were sown in strips previously sown with herbicide by a tractor-mounted slot seeder, in October 1995. Four management treatments were replicated five times in 20 x 10 m plots. The treatments were cut once (July), cut twice (July and September), cut July and autumn grazed. Monitoring of plant dispersal was carried out using seed traps at the soil surface, from June to October 1997.



A study in eastern England of the pollinator community on a species-rich grassland restoration experiment compared to native grassland of the same plant community (Dicks 2002) found a greater diversity of pollinating insects on the restored hay meadow site than on the ancient meadow. Six common species of bumblebee Bombus spp. were recorded at both sites, and the most abundant insect visitor was a bumblebee on both meadows: white-tailed bumblebees Bombus terrestris/ lucorum at the restored site, red-tailed bumblebees B. lapidarius at the ancient meadow site. Seven and five species of solitary bee were recorded at restored and ancient sites respectively.



A randomized, replicated study in 1994-1998 in arable fields in five lowland areas in the UK (Pywell et al. 2002) found that ploughing to 30-40 cm depth and sowing with a species-rich seed mixture created a community similar to the target community on neutral soils. This was significantly more successful than natural regeneration or sowing with a species-poor mix. Sites on acidic or calcareous soils were less similar to their specific target communities. Sowing a nurse crop had no beneficial effects. All treatments reduced nutrient levels. The five sites had four replicate blocks each containing seven experimental plots with different treatments. Vegetation was cut and removed each year in June or July, and sheep were grazed between October and December at 25-40 sheep/ha for six to eight weeks. Vegetation sampling used three 40 x 40 cm quadrats randomly placed within each plot in June each year. Nutrient sampling used ten soil samples per plot in September 1994 and 1998.



In the same replicated study described in (Smith et al. 2000) in North Yorkshire, UK (Smith et al. 2002), plots sown with grassland species had more plant species in the vegetation, and plots cut in July had more plant species in the seed bank, ten years after the restoration began. There were 22 species/m2 in sown plots compared to 18 species/m2 on unsown plots. Of three species sown in 1999, only bird’s-foot trefoil Lotus corniculatus was found frequently in 2000. A single plant each of quaking grass Briza media and bulbous buttercup Ranunculus bulbosus (both also sown in 1999) were found. The upper 5 cm of soil held viable seeds of on average 13 species in plots always cut in July, <12 species in plots previously cut in September and 10 species in plots previously cut in June. From September 1990-1998, the experiment combined grazing, cutting date and fertilizer level treatments in 6 x 3 m plots (three replicates of each). From 1998 onwards, all plots were cut on 21 July, with both autumn and spring grazing. From 1999, plots were either treated with farmyard manure at 12 tonnes/ha/yr in April, or not. The soil seed bank was monitored from soil cores in July 1998, and the vegetation surveyed in June/July 2000.


A replicated trial from 1994 to 1990 in Aberdeenshire, Scotland (Warren et al. 2002), found that sowing meadow plants followed by summer cattle grazing achieved the best results in terms of recreating a species-rich grassland on a former arable field last ploughed in 1993. After six years, sown plots on average contained more sown species (4.9 species/m²) and had greater cover of sown species (97%) than non-sown plots (1.8/m², 43%). Cattle grazed plots contained more sown species (4.8/m²) than sheep grazed plots (2.2/m²), but sheep grazed plots had greater sown species cover (91% vs 46%). August cut plots with cuttings removed had more sown species but lower cover than plots with cuttings left (3.4 vs 2.3/m², 63% vs 76%, respectively). Twenty-four 20 x 40 m fenced plots were either sown with ten herb and four grass species from a crested dog’s tail Cynosurus cristatus-lesser knapweed Centaurea nigra (UK National Vegetation Classification MG5) grazed hay meadow community or not sown after ploughing in April 1993. Within these treatments, six management treatments were tested: sheep or cattle summer grazing, cut in June with aftermath grazing by sheep or cattle, or cut in August with cuttings either left or removed. No plots were cut in August with aftermath grazing.



A replicated site comparison study in 1999 and 2000 in southwest Finland (Pykala 2003) (the same study as (Poyry et al. 2004, Poyry et al. 2005, Pykala 2005)) found that resuming grazing on abandoned species-rich grasslands began to enhance the number of plant species after around five years. The number of plant species was higher on restored pastures with resumed grazing than on old abandoned pastures (for example, 16.4 species/m2 on average, compared to 11.2 species/m2) but the difference was not statistically significant at any scale. Old grazed pastures had significantly more plant species than restored or abandoned pastures at the 1 m2 scale, and significantly more species than abandoned pastures but not significantly more than restored pastures at the whole site scale (0.25-0.8 ha). However, the number of rare native plant species had not increased in response to resumed grazing. Plants were monitored in 1999 or 2000 on 11 old grazed pastures, 12 abandoned pastures (no grazing for more than 10 years) and 10 restored pastures abandoned for more than 10 years with grazing re-started three to eight years (average five years) before the study.



A replicated controlled study of former agricultural land in four European countries (Gormsen et al. 2004) (same study as (Leps et al. 2007)) found that the abundance of earthworms (Lumbricidae) was higher in sown and naturally colonized grassland than in an agricultural rotation in two of four European countries. Numbers and biomass of earthworms were significantly higher in the restoration plots (Netherlands: high plant diversity plots 43 individuals/m², low diversity plots 52, naturally colonized plots 95; Sweden: high 254, low 289, natural 169) than in the agricultural rotation treatment (Netherlands: 5, Sweden: 15 individuals/m²). In the Netherlands, numbers were significantly higher in the naturally colonized plots than in the sown treatments. Differences between treatments did not differ at the UK or Czech Republic sites. In Sweden, species diversity was lower in the agricultural plots (2 vs 3-5 species) and worm biomass increased with legume biomass. In the UK worm biomass increased with grass biomass. In each country there were five blocks each with four treatment plots (10 x 10 m): seed sown to give high (15 species) and low diversity (four species), natural colonization, and a continued agricultural rotation treatment. A target plant community was also sampled. Earthworms were collected within 30 minutes from 4 to 5 squares in each plot (July or August 1998), identified to species or genus and wet weight recorded. Vegetation was sampled within 12 permanent subplots in each plot.



A replicated controlled trial near Göttingen Germany (Hofmann & Isselstein 2004) found that harrowing before sowing enhanced the survival of most grassland plants sown into existing grassland. For six of eight species tested, seedling emergence was highest if seeds were sown after harrowing. All species, except red clover Trifolium pratense, were more likely to have survived after one year if sown after harrowing than on control plots without harrowing. Complete removal of vegetation with herbicide before sowing, or frequent cutting before or after, did not enhance seedling emergence any further. For red clover, cutting and pre-sowing disturbance made no difference to seedling emergence. Autumn hawkbit Leontodon autumnalis, had the highest emergence with the highest disturbance (cutting every week, or complete removal of vegetation before planting). All tested species survived better in plots cut every one or three weeks after sowing than in those cut every nine weeks. Eight plant species were sown in May 1998, in 0.5 x 0.5 m plots on a previously intensively managed, species-poor grassland. Plots were cut every one, three or nine weeks in the nine weeks before sowing and every one, three or nine weeks after sowing. They were either left undisturbed, harrowed or treated with herbicide before sowing. There were three replicates of each treatment combination. Seedling survival was recorded until July 1999.



A replicated, controlled, site-comparison study of 26 restored semi-natural grasslands in south-eastern Sweden (Lindborg & Eriksson 2004) found that continuously grazed control sites had higher plant species diversity and a higher proportion of typical grassland species in the community than restored grasslands. Plant species diversity at restored sites was 16-20 species/m² compared to 24-30 species/m² at continuously grazed control sites. Total species richness was positively associated with time since restoration (1-7 years) and the abundance of trees and shrubs. Overall species composition differed between restored and control sites, with control sites having a higher proportion of typical grassland species than restored sites. However within grassland types (dry, dry to damp (mesic) or damp to wet), species composition was similar between each pair of restored and control sites. Restored damp to wet grassland was dissimilar in species composition to all other plots. Abundance of 10 grazing-indicator species tended to be lower at restored sites. Restored site area (3-35 ha), time between abandonment and restoration, time since restoration and abundance of trees and shrubs were not related to species composition among restored sites or the 10 grazing-indicator species. Restored sites were grazed before abandonment and after restoration, control sites had been grazed continuously. The six control sites were compared to restored sites in the same region. Plants were sampled within 10 randomly distributed plots (1 m²) in July-August 2001. Trees and shrubs were counted within a 40 m diameter circle at each site.



A replicated site comparison study in 1999 and 2000 in southwest Finland (Poyry et al. 2004) (same study as (Pykala 2003, Poyry et al. 2005, Pykala 2005)) found that abandoned species-rich grasslands restored with cattle grazing had similar butterfly and day-flying moth (Lepidoptera) numbers to old grazed pastures. The abundance of butterflies and moths, and the number of species, did not differ between grazed and restored pastures. There were 22-26 species and 126 individuals/site in restored and old grazed pastures. Restored pastures varied more in the identities of species found than old grazed pastures. Some restored pastures had less ‘diverse’ butterfly and moth communities than old pastures because they were more likely to be dominated by abundant, common species. Some species only occurred in old pastures. Butterflies and moths were monitored in 1999 or 2000 on 11 old grazed pastures and 10 restored pastures abandoned for more than 10 years, with grazing re-started three to eight years before the study. All restored pastures received support under the Finish agri-environment support scheme for managing semi-natural grassland. Insects were counted along walked transects between four and seven times between May and August. Either transect length (2000) or searching time (1999) were standardized across sites.



A 2004 review of published and unpublished literature from the UK (Walker et al. 2004) found that introducing plant species and removing nutrients are important to effective grassland restoration. The review identified eight studies that tested effects of reinstating cutting and grazing management on grassland, and ceasing fertilizer use and concluded that this could enhance the number of plant species, but it was slow and did not always work. Just one study, in west Wales, found a marked increase in the number of plant species over eight years (Hayes & Sackville Hamilton 2001). Five found a slight or gradual increase, over 4-14 years ((Bullock et al. 1994), Oomes 1990, Olff & Bakker 1991, Hayes et al. 2000, (Smith et al. 2000)). Two studies found no change or a decrease (Oomes 1990, Mountford et al. 1994). Increases in the number of species were modest and slow. Cutting and grazing together were generally more effective than cutting or grazing alone. On existing grassland, the review found nine studies that tested various methods of adding plant species: adding seed (over-sowing), drilling seed (slot-seeding) and planting small plants (plug-planting), with effects monitored over two to eight years. Three studies found that over-sowing was most effective when combined with either cutting and grazing or de-turfing (Hopkins et al. 1999, Jones & Hayes 1999, Smith et al. 2000). Five studies found that slot-seeding was not very effective (Wells et al. 1989, (Hopkins et al. 1999, Coulson et al. 2001)). Of five studies that tested plug-planting, three found it was initially effective, but survival of the introduced plants fell after two to five years (Wells et al. 1989, (Hopkins et al. 1999), Barratt et al. 2000). Two found 60-70% of plants established (over two or five years) (Boyce 1995, (Hopkins et al. 1999)). On ex-arable land, ten studies tested sowing grassland species, monitoring effects for between one and 20 years. All found increased plant species diversity and enhanced similarity to the target plant community, which was either upland, chalk or neutral (mesotrophic) grassland. Similarity to the target community was quantified for six of these studies and fell between 50% and 81%, usually after two to five years (20 years in one case) (McDonald 1992, Wells et al. 1994, (Stevenson et al. 1995), Pywell et al. 2000, (Pywell et al. 2002)). Cultivation, followed by a relatively high seeding rate, seemed the most effective approach. On upland grassland, adding sulphur to acidify the soil prior to seeding led to effective establishment of sown species in two studies.

Additional references:

Wells T.C.E., Cox R. & Frost A. (1989) Diversifying grasslands by introducing seed and transplants into existing vegetation. Pages 283-298 in: G.P. Buckley (ed.) Biological Habitat Reconstruction. Belhaven, London.

Oomes M.J.M. (1990) Changes in dry matter and nutrient yields during the restoration of species-rich grasslands. Journal of Vegetation Science, 1, 333-338.

Olff H. & Bakker J.P. (1991) Long-term dynamics of standing crop and species composition after the cessation of fertiliser application to mown grassland. Journal of Applied Ecology, 28, 1040-1052.

McDonald A.W. (1992) Succession in a 3-year-old flood-meadow near Oxford. Aspects of Applied Biology, 29, 345-352.

Mountford J.O., Tallowin J.R.B., Kirkham F.W. & Lakhani K.H. (1994) Effects of inorganic fertilisers in flower-rich hay-meadows in the Somerset Levels. Pages 74-85 in: R.J. Haggar, S. Peel (eds.) Grassland Management and Nature Conservation. British Grassland Society Occasional Symposium, 28. British Grassland Society, Reading.

Wells T.C.E., Pywell R.F. & Welch R.C. (1994) Management and restoration of species-rich grassland. Report to the Ministry of Agriculture, Fisheries and Food (BD0306), Institute of Terrestrial Ecology, Monks Wood.

Boyce D.V.M. (1995) Survival and spread of wildflowers planted into ex-agricultural grassland. Journal of Practical Ecology and Conservation, 1, 38-44.

Barratt D.R., Mountford J.O., Sparks T.H., Walker K.J., Warman E.A. & Garbutt A. (2000) The effects of field elevation and water levels on the establishment of plug-plants in an ex-arable grassland. Aspects of Applied Biology, 58, 425-430.

Hayes M.J., Sackville Hamilton N.R., Tallowin J.R.B., Buse A. & Davies O. (2000) Methods of enhancing diversity in upland environmentally sensitive area swards. Report to the Ministry of Agriculture, Fisheries and Food (BD1424), Institute of Grassland and Environmental Research, Aberystwyth.

Pywell R.F., Warman L., Walker K.J. & Sparks T.H. (2000) Reversion of intensive arable land to grass heath and Calluna heath: vegetation aspects. Report to the Ministry of Agriculture, Fisheries and Food (BD1502), Institute of Terrestrial Ecology, Monks Wood.

Hayes M.J. & Sackville Hamilton N.R. (2001) The effect of sward management on the restoration of species-rich grassland: a reassessment of IGER’s grassland restoration experiment, Trawsgoed. Countryside Council for Wales Contract Science Report No. 438, Bangor.


A small site comparison study of two restored hay meadows with two ancient hay meadows in the Bristol area, UK (Forup & Memmott 2005) found no consistent differences in the abundance or diversity of pollinating insects (dominated by bees (Hymenoptera) and flies (Diptera)) between ancient and restored sites, and considered the pollinator community to be effectively restored. The four hay meadows were 1-9 ha and managed with an annual hay cut, no grazing and no artificial fertilizer. One restored meadow was previously a golf course, restored in the early 1990s using a traditional annual hay cut. The other restored meadow was previously part of an urban park, restored in 1981 by translocating turf from an ancient meadow, planting and sowing wildflowers and using traditional meadow management. Flower-visiting insects were sampled on two 50 x 2 m transects/meadow every two weeks from early May to the end of July 2000.



A 2005 review of six studies exploring the best management for restoring upland hay meadow vegetation on semi-improved grassland in the UK (Jefferson 2005) suggested that the highest plant species richness is produced by spring and autumn grazing, a mid-July hay cut and no inorganic fertilizer. Addition of seed from outside the site (either from natural dispersal or sowing) is also likely to be necessary. Three studies found that adding Rhinanthus minor hay rattle seed can help the colonization of other sown species (Smith et al. 2003, Pywell et al. 2004, Smith 2005). One study in North Yorkshire (Smith 2005) found that adding farmyard manure had a generally harmful effect on restoration of upland hay meadow communities, and recommended that this should be avoided, at least in the early stages of restoration. However, results were based on using larger quantities of manure than under traditional management.

Additional references:

Smith R.S., Shiel R.S., Bardgett R.D., Millward D., Corkhill P., Rolph G., Hobbs P.J. & Peacock S. (2003) Soil microbial community, fertility, vegetation and diversity as targets in the restoration management of meadow grassland. Journal of Applied Ecology, 40, 51–64.

Pywell R.F., Bullock J.M., Walker K.J., Coulson S.J., Gregory S.J. & Stevenson M.J. (2004) Facilitating grassland diversification using the hemiparasitic plant Rhinanthus minor. Journal of Applied Ecology, 41, 880–887.

Smith R.S. (2005) Ecological mechanisms affecting the restoration of diversity in agriculturally improved meadow grassland. Defra Project BD1439.


A replicated, controlled study of semi-natural grasslands at two sites in southeast Sweden (Lindborg et al. 2005) found that the frequency of harebell Campanula rotundifolia, cowslip Primula veris and yellow rattle Rhinanthus minor was lower in restored than traditionally managed, continuously grazed (from 1945 to 1998) grasslands and in one case lower than abandoned grasslands. Cowslip had significantly lower frequency in restored (6% plots contained cowslip) than continuously grazed (22%) and abandoned grasslands (20%) at one site. At the other site cowslip frequency did not differ significantly (restored 24%, grazed 38%, abandoned 16%). Yellow rattle was absent or less frequent in restored (0-6%) than continuously grazed grasslands (6-27%) and was absent from abandoned grasslands. Harebell had a lower frequency in restored (28-54%) and abandoned grasslands (28-62%) than continuously grazed sites (42-80%). There was no significant difference in seedling emergence for cowslip (0.2-2.0/plot) or harebell (5-8/plot) within continuously grazed grasslands or those restored three or 40 years ago or abandoned, however harebell did not emerge in abandoned grasslands. At each site, 100 randomly distributed 1 m² plots were sampled for presence/absence of the three species within each grassland type. At one site, 50 seeds of each species (collected in the area in autumn 2002) were sown (within two weeks) in each of eight randomly distributed 1 dm² plots in each grassland. Emergence was recorded in 2003.



A further report (Poyry et al. 2005) from the same replicated site comparison study in southwest Finland as (Pykala 2003, Poyry et al. 2004, Pykala 2005) looking at the responses of individual butterfly and moth (Lepidoptera) species showed that three species were most abundant in old pastures and did not recover in pastures where grazing had been reintroduced following abandonment. These were the purple-edged copper Lycaena hippothoe, the common blue Polyommatus icarus and the yellow shell moth Camptogramma bilineatum. Three moth species were more abundant in restored pastures than old pastures: Epirrhoe hastulata (no common name), the silver-ground carpet Xanthorhoe montanata and the latticed heath Chiasmia clathrata. Two species, the scarce copper butterfly Lycaena virgaureae and the black-veined moth Siona lineata were less abundant in restored pastures than in old grazed pastures or abandoned pastures (12 pastures abandoned for more than 10 years were monitored for comparison), implying that they were negatively affected by the reintroduction of grazing.


A further report (Pykala 2005) from the same replicated site comparison study in southwest Finland as (Pykala 2003, Poyry et al. 2004, Poyry et al. 2005) found that the frequency of 31 of 76 plant species increased or recovered following the reintroduction of grazing on abandoned species-rich grasslands. Twenty-nine species seemed to have increased in response to grazing, having higher frequencies on restored pastures than on abandoned pastures but lower than old pastures. Species included common bent grass Agrostis capillaris and harebell Campanula rotundifolia. Just two species, red fescue Festuca rubra and Goldilocks buttercup Ranunculus auricomus had similar frequencies in old and restored pastures. Eight species, such as meridian fennel Carum carvi, cowslip Primula veris and self-heal Prunella vulgaris, showed no recovery in response to resumed grazing, having similar frequencies in restored and abandoned pastures. Thirty one species did not differ between three types of grassland.



A randomized, replicated, controlled trial from 1998 to 2003 on a meadow in northern Finland (Hellström et al. 2006) (same study site as (Hellström et al. 2009)) found that the number of plant species did not increase in response to conservation-oriented mowing regimes. There were 14 plant species/plot on average in 1998 (not including mosses and lichens) and 12 species/plot in 2003, and no difference between mowing treatments. Some species responded to treatments. For example, late mowing plus disturbance favoured harebell Campanula rotundifolia but reduced the cover of common bent grass Agrostis capillaris. The Kiiminki Haaraoja Meadow in northern Finland was traditionally grazed, but abandoned in 1985. In 1993, the meadow was divided into forty 50 x 50 cm study plots, each at least 2 m apart, and annually mown in August, without grazing. From 1998, ten plots were mown in June, ten in August, and ten mown in August with bare soil exposed in 25% of the plot area, using a spade. Ten control plots were not mown. The percentage cover of all plant species (including mosses and lichens) in the treatment plots was monitored in June every year from 1998 to 2003.



A replicated controlled trial from 1992 to 2003 in southeast Germany (Kiehl & Wagner 2006) (same study as (Kiehl & Pfadenhauer 2007)) found that hay spreading enhanced plant species richness but not grasshopper (Orthoptera) species richness on hay meadows and that topsoil removal enhanced some grasshopper and plant species but not the total number of plant species. For example, in 2001 there were 20-25 plant species/plot with hay transfer and 10-20 species/plot without hay transfer. The number of plant species from the target plant community and the number of Red-listed plant species were also higher on sites with hay transfer. Hay transfer had no effect on the number of grasshopper species (5-6 species/100 m on all plots), or grasshopper species associated with dry grassland, or Red-listed grasshopper species. Topsoil removal enhanced the number of dry grassland grasshopper species, plant species from the target plant community, Red-listed species of plant and grasshopper, and the total number of grasshopper species, but not the total number of plant species. Four ex-arable fields were half spread with hay from a nearby nature reserve between July and September 1993. The other half had no hay added. One field had the topsoil removed (to 40 cm depth) in 1993. Plant species were monitored every year from 1993 to 2002 on thirty 4 m2 plots/field. Grasshoppers were counted four times between July and September 2001 on 18 transects at the restoration experiment.



A site comparison study of semi-natural grasslands near Lund, Sweden (Öckinger et al. 2006) found no significant difference in plant species richness or abundance in recently restored, abandoned or continuously grazed grasslands. There was a decrease of management-dependent plant species with increasing tree and shrub cover at abandoned sites. Present management significantly affected butterflies (Lepidoptera) and plants, their species richness increased with increasing vegetation height, but this differed between sites depending on whether they were grazed by cattle, horses or sheep. Sheep grazing negatively impacted species richness compared to cattle or horses. There were 12 grasslands of each type and current management comprised 12 sites cattle grazed, six horse grazed, eight sheep grazed and 10 with no grazing. Butterflies and burnet moths (Zygaenidae) were sampled using a transect count method (150 m/ha) six to seven times in May-August 2003 or June-August 2004. Plant presence was sampled in ten 0.25 m² quadrats (divided into twenty-five 10 x 10 cm squares) at each site in June-August 2004. Vegetation height was also measured.



A replicated trial from 1987 to 2004 at Somerford Mead, Oxfordshire, UK (Woodcock et al. 2006) found that both plant and beetle (Coleoptera) communities on an experimentally restored meadow were closest to the flood meadow restoration target under a regime of hay cutting and aftermath grazing. For plants, sheep grazing was slightly better, but for beetles, cattle grazing was better. There were fewer beetles and beetle species on plots cut for hay but without aftermath grazing. After 18 years, neither the plant nor the beetle communities were fully restored to floodplain meadow species assemblages. The site was characterized by a high percentage cover of red fescue Festuca rubra. A former arable field was sown with seed harvested from a local floodplain meadow in 1985. From 1987 it was cut in July and aftermath grazed. From 1989, three aftermath grazing treatments were tested: sheep, cattle or no grazing, on three 0.4 ha plots each. Plants and invertebrates were monitored in 2004 and compared with communities on two nearby floodplain meadows.



A replicated controlled seed addition trial in 2000-2001 in four grasslands subject to different management practices in southeastern Sweden (Lindborg 2006 ) found that seedling emergence was higher in grazed grasslands than in an abandoned grassland. Seedling emergence was similar for six plant species favoured by grazing (target species) and six plant species favoured by no grazing (generalist species) in all four grasslands. The proportion of sown seeds emerging differed among species (range: 1.2-12.6%). The highest proportion of seeds germinated at an intermediate sowing density (20-50 seeds/dm²). Target species recruited well in the former arable fields (grazed for 10 or 30 years) and generalists also recruited well at grazed sites. All sown species performed poorly in the abandoned (40 years ago) grassland. The two grasslands with the longest grazing history (continuously grazed grassland and former arable grazed for 30 years) were positively associated with emergence of target species. There were four grassland types: former arable field grazed for 30 years, former arable field grazed for 10 years, continuously grazed (since 17th century) semi-natural grassland, abandoned semi-natural grassland (previously grazed, abandoned 40 years ago). Seed was collected locally in autumn 2000 and sown within two weeks. Each species was sown in ten 10 x 10 cm plots/grassland at four densities: 10, 21, 46, and 100 seeds. Ten plots were left unsown as controls. Seedling emergence was recorded in June 2001 in randomly placed 1 m² quadrats.


A replicated study in 1995-1998 and 2002 of former arable fields at two sites in England (Bullock et al. 2007) found that after eight years, plots sown with species-rich mixtures resembled target grassland community types. Plots sown with a species-poor mix, although colonized by some additional species, had fewer grass, legume and other broadleaved species. Hay yield increased in the species-rich plots in the first years of the experiment and the increased yield was still apparent after eight years (43% higher yield than species-poor plots). This was largely due to differences in numbers of non-leguminous broadleaved plants. There were four replicate blocks of plots (6 x 4 m). The species-rich mixture comprised 11 grasses and 28 broadleaved species, to resemble species-rich hay meadows. The species-poor mixture comprised seven grasses to establish moderately diverse grassland. Vegetation was sampled in early June in three quadrats (0.4 x 0.4 m) per plot in 1995-1998 and 2002. During the hay cut in July, a 6 x 1.2 m sample of hay was removed from each plot and weighed and a 500 g sub-sample was dried to calculate hay yield.


A 2007 review of experimental evidence on how to restore species-rich grassland on old arable fields (Diggelen 2007) found that removing excess nutrients is very slow if done simply by grazing and cutting hay (two studies), with only 3-5% of the soil nutrient pool removed each year. Removing topsoil can effectively remove nitrogen but not phosphorus (one study; Verhagen et al. 2001). The authors argued it is necessary to introduce plants by sowing because rare grassland species are under-represented in the seed bank. They found one review (Pywell et al. 2003) showing that plants that were good colonizers and competitors, associated with fertile soils were most likely to establish in restoration experiments. Two sets of experiments demonstrated that seedlings of grassland or wet grassland plants survive less well in low light conditions (as in a dense productive grassland).

Additional references:

Verhagen R., Klooker J., Bakker J.P. & van Diggelen R. (2001) Restoration success of low-production plant communities on former agricultural soils after topsoil removal. Applied Vegetation Science, 4, 75-82.

Pywell R. F., Bullock J. M., Roy D. B., Warman E. A., Walker K. J. & Rothery P. (2003) Plant traits as predictors of performance in ecological restoration. Journal of Applied Ecology, 40, 65-77.



A randomized, replicated, controlled trial from 2000 to 2004 at a farm in East Sussex, UK (Edwards et al. 2007) found that hay spreading was the most effective technique for restoring a hay meadow plant community similar to the seed donor site. Both hay spreading and the addition of brush-harvested seed increased plant species richness, and harrowing increased the effectiveness of the seed addition treatments. Hay spreading was thought more effective because it captured seeds from a greater range of heights in the sward, and allowed for seeds to mature on the restored site after the restoration activity. Eight different combinations of harrowing and the two methods of applying seed were tested, on land that had been improved agricultural grassland, with two different rates of hay application. There were four replicates of each combination of treatments. Plants were monitored before treatment (July 2000) in two random plots from each block, and every June from 2001 to 2004, in ten 50 x 50 cm quadrats in each plot.



A replicated, randomized study of sown grassland species on an intensive cattle farm in Ireland (Geijzendorffer 2007) found that for individual species, the maximum growth rate was either higher at a fertilization rate of 225 kg nitrogen/ha than at 90 kg N/ha, or did not differ. On dry soils, species either experienced slightly more competition with members of the same species at 90 kg N/ha than at 225 kg N/ha, or there was no difference between treatments. Twelve plant species were sown in mixtures of one, two or three species in equal amounts, comprising one species plus perennial rye grass Lolium perenne plus a production/weed grass or legume. Sowing was in April 2004 at a rate of 1.5 or 3 g/m² in plots of 1.5 x 1.5 m at a dry and wet site. High (225 kg N/ha) or low (90 kg N/ha) fertilization rates were applied with three replicates for each seed mixture and soil type in a randomized block design. Biomass was recorded weekly for four to five weeks following cuts in alternate months. Botanical composition and weekly changes in green plant area and relative growth rates were also recorded.



A randomized, replicated, controlled trial at two experimental farms in Wales (Hayes & Tallowin 2007) (results from the two farms are presented in (Jones & Hayes 1999) and (Morgan et al. 2008)) found that plant species richness on grasslands increased over 10-13 years in response to imposing traditional management practices, providing the management involved both hay cutting and aftermath (autumn-winter) grazing. In the final year, these sites had over 13 plant species/quadrat (over 15 at the upland site) and over 40 plant species on each plot (over 50 at the upland site), compared to 8-9 species/quadrat and 24 species/plot in control plots. They were colonized by desirable plant species, such as yellow rattle Rhinanthus minor and heath spotted orchid Dactylorhiza maculata. Plots managed with just hay cutting, or just grazing, did not show strong increases in plant species richness. The experiments took place in Ceredigion (lowland) from 1992 to 2005 and the Cambrian Mountains (upland fringe) from 1995 to 2005. Six or seven treatments were each replicated three times on 0.15 ha plots. The control plot was species poor pasture sheep-grazed from April to November, fertilized (with nitrogen, phosphorous, potassium NPK fertilizer) at a rate of 150 kg nitrogen/ha, and limed once, in the second or third experimental year. Other plots were not fertilized at all, but had various combinations of cutting, grazing and liming. Adding lime slightly enhanced plant species richness in summer grazed treatments at the upland site. Plants were monitored in summers of 1992-1997 and 2000, 2003 and 2005 in ten quadrats in each plot.



A replicated trial in the White Carpathians Protected Landscape Area, in the eastern Czech Republic (Jongepierova et al. 2007), found that sowing a regional seed mixture over the entire plot was the most effective treatment for establishing hay meadow vegetation. Four restoration treatments were tested, each in four 55 x 20 m plots, replicated within a single 3 ha arable field. The experimental treatments were sowing seven grasses and 20 herb species throughout the plot, or sowing 2.5 m-wide strips of just the herb species with or without a commercial grass mix. Control plots were left to naturally regenerate. In the fully sown plots, 19 of the 20 herb species, and all seven grass species had established by 2004, providing 30% and 55% cover on average. The cover of sown herb and grass species in the strip-sown or unsown treatments were less than 5% and 2-9% respectively. Plots were sown in spring 1999, and vegetation monitored in June 2002-2004. All plots were cut once in July and the hay removed, following restoration.



A replicated controlled trial near Munich, southeast Germany from 1993 to 2002 (Kiehl & Pfadenhauer 2007) (same study as (Kiehl & Wagner 2006)), found that spreading hay from a nearby nature reserve rapidly increased the number of plant species, and the number of target hay meadow species in ex-arable fields managed to restore hay meadow vegetation. The removal of topsoil combined with hay spreading increased the proportion of target species and the persistence of species, but led to a very low hay crop even after nine years. Mowing (once or twice) also increased plant species richness, and the number of target plant species. Nine years after restoration, the best plots in this experiment (mown, with hay spreading) still had a different plant community from species-rich grassland on a nearby nature reserve (Garchinger Heide). Restoration was tested on four ex-arable fields, 1.3-3.2 ha in size, beginning in 1993. Half of each field had hay added between July and September 1993 (once only) and the other half did not. Experimental plots within these treatments were either mown once, mown twice, mown with cuttings left as mulch, or grazed through spring and summer. One field had the upper 40 cm of topsoil removed. This field was either mown once in July or left unmanaged. Plant species were monitored every year on thirty 4 m2 plots per field.



A replicated controlled experiment in five European countries (Leps et al. 2007) from 1996 to 2003 (the same study as (Gormsen et al. 2004)) found that most hay meadow species sown on plots of abandoned arable land established well at four locations (all except Sweden, where less than half of the sown species established). In the UK, the Czech Republic and the Netherlands, more than 70% of the sown species were established after eight years. Grasses, bird’s-foot trefoil Lotus corniculatus and red clover Trifolium pratense established well at almost all sites, but small legumes, black medick Medicago lupulina and lesser hop trefoil Trifolium dubium disappeared quickly at all except the UK site. The success of other plant species varied between sites. Plots sown with 15 plant species always established some of the sown species, while some plots sown with just four plant species failed to establish any. Plots (10 x 10 m) were either left to naturally regenerate, or sown with four or 15 plant species in autumn 1995. The experiment was repeated in five countries: the Czech Republic, the Netherlands, Spain, Sweden and the UK, with five replicates of each treatment. Plant species naturally occurring in local grassland systems were sown. Red fescue Festuca rubra, Timothy grass Phleum pratense, bird’s-foot trefoil, red clover, and ribwort plantain Plantago lanceolata were sown at all sites. All plots were mown once or twice a year and separated by 2 m borders. Plant cover was measured from 1996 to 1998 and 2002 to 2003, in ten 1 m2 quadrats/plot.



A randomized, replicated and controlled trial in 1999-2003 of restoration methods at two sites in the UK (Pywell et al. 2007) found that turf removal followed by seed addition was the most effective means of increasing plant diversity. Multiple harrowing was moderately effective and was enhanced by applying snail/slug pesticide and sowing yellow rattle Rhinanthus minor (which reduced competition from grasses). Grazing, slot-seeding and inoculation with soil microbial communities from species-rich grasslands did not increase botanical diversity, and different grazing management regimes had little impact. Thirteen treatments were applied to 15 x 15 m plots at sites in Devon and Buckinghamshire, with eight replicates of each treatment. All treatments were managed with a single July hay cut.



A controlled study in 2001-2005 of two sown meadows in the Kedainiai District, Lithuania (Sendzĭkaite et al. 2007) found that an extensively managed meadow was restored faster than an intensively managed meadow. The total number of plant species was higher in the extensively managed meadow compared to the intensively managed meadow (79 species, annual range 27-40 species in the extensively managed meadow vs 39 species, annual range 22-30 species in the intensive meadow). The same trend was seen for biomass (850-1,480 vs 720-1,340 g/m²), moss/liverwort/hornwort (bryophyte) content (6-26% vs <1.4%) and dead plant matter (2-22% vs 1-12%). A grassland/clover Trifolium spp. mixture was sown (27 kg/ha) in two arable fields in 1991. One received intensive management: fertilization, hay making and grazing (June, August, September). The other received extensive management: no fertilization, annual hay cutting and occasional grazing (once in July 2001-2005). Botanical composition was sampled in three permanent plots (100 m²) in each field in June, July and August from 2001 to 2005. Above ground biomass was sampled in three 1 m² quadrats in representative areas (by composition and cover) in each plot.



A site comparison study in 2006 on a 35 ha area of an ancient hay field in Lääne, western Estonia (Aavik et al. 2008) found that recovery of the plant community was slow after restoring regular annual mowing. Even though sites were located alongside one another, there were differences in the plant community between plots where annual mowing had been reinstated for five or thirteen years, and sites with continuous annual mowing since the 1960s. Areas without continuous management had fewer plant species. Management history, not soil conditions, was the most important factor determining the number and identity of plant species. Plants were counted in five 1 m² quadrats at 30 sites of known management history in 2006. Three management histories were identified: continuously managed for 200 years and annually mown since the 1960s, irregularly mown every two or three years from the early 1980s to 1993 then annually since 1993 (regular mowing restored 13 years before the study), or unmanaged from the early 1980s until 2000 or 2001 when annual mowing was restored (five or six years before the study). The latter group of sites had become overgrown with trees. Mowing was in late June or early July in each case.



A randomized, paired site comparison in five areas of southern England (Fagan et al. 2008) found distinct differences in vegetation between restored and ancient chalk/limestone (calcareous) grasslands, even after 60 years. Sites seeded with just grasses remained dominated by a few grass species. Sites allowed to regenerate naturally moved towards the target plant community over time, although success was limited by proximity to ancient grasslands. Some features of restored grassland (such as the proportion of perennial plants) became more like ancient grasslands with increasing age. High soil phosphorus concentration (due to former fertilizer application) was detrimental to restoration. Forty restored grassland sites were randomly selected from all those available, to give equal representation in four age classes and the five areas (North Downs, South Downs, South Wessex Downs, Chilterns, Cotswolds). Sites were one to 103 ha in size. They were restored either by natural regeneration, seeding with grasses, or seeding with a flower-rich seed mix. All sites were grazed and some occasionally mown. Each was paired with an ancient grassland no more than 9.3 km away. Plants were surveyed in ten 0.25 m2 quadrats at each site, in summer 2004, and soil analysed in September 2004.



A replicated controlled trial at the Pwllpeiran Research Centre, in the Cambrian Mountains, west Wales (Morgan et al. 2008) (partly the same study as (Hayes & Tallowin 2007)) found that plant species richness increased and rye grass Lolium perenne cover declined on improved upland grassland after ten years of management with hay cutting and/or grazing but no fertilizer addition. In the restoration plots, rye grass cover declined from 58% to just under 10% on average. All treatments enhanced plant species richness, but the hay cut and grazing combined treatments were the most effective. These plots had an average of 51 species/plot by 2005, compared to 24 species in control plots. They also had almost 50% cover by non-grass, desirable herbaceous species (forbs). Treatments with hay cut but no grazing had 29-30 species on average in 2005, and those with grazing only had 31-35 plant species in 2005. Both had an increase in weedy, undesirable species. Seven management treatments were set up in 1994 on 0.15 ha plots with three replicates of each treatment. Control plots had standard intensive management, fertilized with nitrogen, phosphorous, potassium (NPK) fertilizer, limed and grazed by sheep. Six restoration treatments were either grazed from April to November, cut for hay in July/August without grazing, or hay cut and grazed from September to November, each with or without lime added in 1998. Plots with an application of lime had more desirable species by 2005 than those without lime.



A replicated, controlled study of chalk grassland restoration on land taken out of arable production in Oxfordshire, UK (Woodcock et al. 2008) found that plant and beetle (Coleoptera) species richness tended to be higher when seeds from the local area were applied by brush harvesting or hay spreading. Sowing of a grass-only seed mix, reduced the overall number of plant (36 vs 40 species) and beetle species (5-7 vs 6-8 species), independent of whether or not it was used in combination with the addition of local seeds. Plant species richness tended to be higher in plots receiving brush-harvested seeds or hay than controls, although there was limited change in those plots from 2002 to 2004 (at low or high seed application rates). The highest beetle species richness was found on plots with high rates of hay application without grass-only seed mix. Changes in beetle community structure were significant for control, high rate hay spreading, grass-only and low rate brush harvesting with grass-only mix. Plots receiving local seeds tended to become more similar to the donor plant community over time. The similarity was greatest in plots without the grass-only mix and with brush harvesting at a high rate for plants, and high rate hay spreading for beetles. Changes in beetle assemblage were much greater than for plants. Forty plots of 10 x 10 m were established in four blocks. Grass-only seed mixture was sown in half the plots in August. Seeds from the local area were obtained from an adjacent unimproved chalk grassland and were applied by brush harvesting or hay at high or low rates. Plots were grazed by sheep but were not cut. Plants were sampled in ten 0.5 x 0.5 m randomly located quadrats/plot in August 2002-2004, the donor site was sampled in 2004. Plant-eating beetles were sampled using a Vortis suction sampler (15 positions/plot) in May, July and September in 2002-2004, the donor site was sampled in 2001.



A randomized, replicated, controlled trial from 2003 to 2006 on a meadow in northern Finland (Hellström et al. 2009) (same study as (Hellström et al. 2006)) found that of eight plant species sown, only two had established themselves after three years. The two species, maiden pink Dianthus deltoides and self-heal Prunella vulgaris, were only growing well on plots mown in August, with soil disturbance. Four plant species, small mousetail Myosurus minimus, water avens Geum rivale, northern dock Rumex longifolius and tansy Tanacetum vulgare did not establish, with 0-3 seedlings observed in the entire experiment, and none on any plots by 2006. Longleaf speedwell Veronica longifolia and sticky catchfly Lychnis viscaria grew well on the August-mown disturbed plots (10-18 seedlings in total) in the first year, but not in subsequent years. No species grew well on the other treatments (0-6 seedlings in total of each species/year). The meadow was divided into forty 50 x 50 cm study plots. In September 2003, half of each plot was sown with 30 locally-collected seeds of eight plant species. Seedlings were counted in June-July 2004-2006. Ten plots were mown in June, ten in August, and ten mown in August with bare soil exposed in 25% of the plot area, using a spade. Ten control plots were not mown.



A replicated, site comparison study in Scotland (Lye et al. 2009) found that Rural Stewardship scheme species-rich grassland attracted more nest-searching queen bumblebees Bombus spp. but fewer foraging queens than areas of naturally regenerated, largely unmanaged grasslands. Five Rural Stewardship Scheme farms participating in the species-rich grassland management or restoration option were paired with five conventional farms. Across all farms, unmanaged grassland on conventional farms attracted the highest abundance of foraging queen bumblebees (over 4 queens/100 m transect on unmanaged grassland vs less than 3 on species-rich grassland), also in comparison with hedgerow and field margin transects. Unmanaged grassland transects had more nectar and pollen-providing flowers than species-rich grassland in April and May, when queen bumblebees are on the wing. Bees were surveyed once a week for five weeks April-May 2008, using a transect walk method.



A 2009 literature review of agri-environment schemes in England (Natural England 2009) found evidence that plant diversity was higher on Countryside Stewardship scheme plots sown with a chalk grassland mix than on Environmentally Sensitive Area sites sown with a basic grass mix (CABI 2003). However the same study also reported that few of the sown sites were classed as Biodiversity Action Plan habitats. One study found that few sites that had undergone arable reversion for at least five years could be classed as lowland chalk/limestone (calcareous) grassland or lowland meadow under Biodiversity Action Plan definitions (Kirkham et al. 2006). Instead many of the sites were comparable to semi-improved grassland.

Additional references:

CABI (2003) Chalk Grassland: Enhancement of plant and invertebrate diversity through the use of Environmental Land Management Schemes. Defra project report BD1414, London.

Kirkham F.W., Davis D., Fowbert J.A., Hooke D., Parkin A.B. & Sherwood A.J. (2006) Evaluation of arable reversion agreements in the Countryside Stewardship and Environmentally Sensitive Areas Scheme. Defra project report MA0105/RMP 1982, London.


A replicated, controlled study of ten nitrogen-rich grasslands heavily populated with white hellebore Veratrum album in Switzerland (Spiegelberger et al. 2009) found that sawdust addition had limited effects on mountain grassland communities. Sawdust addition reduced grass cover slightly (grazed control 51%, grazed plus sawdust 48%, ungrazed control 48%, ungrazed plus sawdust 42%) but plant diversity and species richness were unaffected, with species richness generally increasing with decreasing productivity in grazed and ungrazed areas. Above-ground grass and broadleaved plant biomass (excluding white hellebore) was 20-25% lower in sawdust plots and biomass of white hellebore slightly higher, compared to controls. Number of shoots was not influenced by sawdust (control 21 shoots, sawdust 27 shoots). Paired plots (6 x 3 m) were established, one cattle-grazed, the other not. From 2002 to 2004, sawdust (from local beech trees Fagus sylvatica) was hand spread over half of each plot (0.5 kg/m2/month) over three months. Above ground (1 cm) biomass samples were taken in autumn 2004 (white hellebore shoots also counted) and spring 2005. Cover of each plant species in central 2 x 2 m quadrats was also recorded in summer 2002 and spring 2005.



A replicated, controlled study of five wooded hay meadows on the island of Gotland in Sweden (Wallin et al. 2009) found that plug plants were over twice as effective as sowing for plant establishment. Devil’s-bit scabious Succisa pratensis plugs established in all plots after two growing seasons and seeds in 45% plots. Spotted cat's-ear Hypochoeris maculata plugs established in 81% of plots and seeds in 33%. Germination rate of seeds varied between donor sites, particularly for devil’s-bit scabious (0-11% germination rate), spotted cat's-ear varied only slightly (7-10%). Litter removal did not affect devil’s-bit scabious germination or survival, for spotted cat's-ear the effect of raking on survival depended on donor site. The four donor sites were species-rich, traditionally managed meadows. At the three recipient sites, eight 72 x 72 cm plots were established, each divided into sixteen 18 x 18 cm sub-plots. For each species, six sub-plots had seeds (50 seeds/subplot) and four-month-old plugs (two plugs/subplot) introduced. Four control sub-plots had no seed or plugs. Seeds were sown in October 2003 and plugs planted in May 2004. Litter was removed from half of each plot (randomly selected). Emerging seedlings were recorded in May 2004, survival in October 2005 and plug survival in October 2005.



A replicated, site comparison study of 16 restored and six traditionally managed semi-natural grasslands in southern Sweden (Dahms et al. 2010) found no significant difference in ant (Formicidae) species richness between restored sites and continuously grazed traditional sites. Total species richness, richness of forest species and of open-habitat species did not differ between restored and traditional sites. There were 1-12 ant species per site (average eight species, two forest species, three open habitat species). Total species richness increased with time since restoration (up to 12 years), largely due to increasing open habitat species richness. However, the proportion of rare species was higher at younger restored sites. Vegetation height, size of study site and numbers of trees and shrubs did not affect species richness. Sites were restored from 1994 to 2001, trees and shrubs were removed and regular grazing resumed. Ants were sampled along a transect of 15 pitfall traps (10 m apart) at each site over seven days in June 2006. Vegetation height was measured around randomly selected traps.



A replicated site comparison study in 2008 and 2009 on farms in three regions in England (Field et al. 2010) found that land managed under Higher Level Stewardship grassland creation/restoration options was used significantly more by seed-eating farmland songbirds than improved grassland in two of the three regions. The strongest difference was in the Cotswolds, where almost 4 birds/ha were recorded on restored grassland, compared to around 1 bird/ha on improved grassland. In East Anglia, there were not more birds on Higher Level Stewardship creation/restoration grassland than on improved grassland. Surveys were done in the summers of 2008 and 2009 on 69 farms with Higher Level Stewardship in East Anglia, the West Midlands or the Cotswolds and on 31 farms across all three regions with no environmental stewardship.



A 2010 review of studies of scientific knowledge about how to re-establish plant communities in grasslands by reintroduction (Hedberg & Kotowski 2010) found that direct seeding and hay transfer have been shown to be effective methods. The review found 38 studies, 28 of which provided enough information to evaluate the outcome, 21 of these from European countries (of which some also looked at the effects on wet meadows). Studies were graded as: successful, of limited success, failed introductions, or without the necessary information to evaluate the outcome. Direct seeding had success or limited success in 10 European studies. Hay spreading had success or limited success in seven European studies (four of which on wet meadows) and was not shown to fail ((Smith et al. 2000), Patzelt et al. 2001, Hölzel & Otte 2003, (Kiehl & Pfadenhauer 2007), Rasran et al. 2007, Schmiede et al. 2009, Klimkowska et al. 2010). Plug planting had success/limited success in two European studies (one on wet grassland) ((Fenner & Spellerberg 1988), Tallowin & Smith 2001). Strip seeding did not reintroduce species in two studies (both recorded in (Pywell et al. 2007)).

Additional references:

Patzelt A., Wild U. & Pfadenhauer J. (2001) Restoration of wet meadows by topsoil removal: vegetation development and germination biology of fen species. Restoration Ecology, 9, 127–136.

Tallowin J.R.B. & Smith R.E.N. (2001) Restoration of a Cirsio-Molinietum fen meadow on an agriculturally improved pasture. Restoration Ecology, 9, 167-178.

Hölzel N. & Otte A. (2003) Restoration of a species-rich flood meadow by topsoil removal and diaspora transfer with plant material. Applied Vegetation Science, 6, 131-140.

Rasran L., Vogt K. & Jensen K. (2007) Effects of topsoil removal, seed transfer with plant material and moderate grazing on restoration of riparian fen grasslands. Applied Vegetation Science, 10, 451–460.

Schmiede R., Donath T.W. & Otte A. (2009) Seed bank development after the restoration of alluvial grassland via transfer of seed-containing plant material. Biological Conservation, 142, 404-413.

Klimkowska A., Kotowski W., van Diggelen R., Grootjans A. P., Dzierża P., & Brzezińska K. (2010) Vegetation re-development after fen meadow restoration by topsoil removal and hay transfer. Restoration Ecology, 18, 924-933.


A randomized replicated controlled trial in Berkshire, UK started in 2008 (Pywell et al. 2010) found that grasslands sown with a seed mix containing legumes and other herbaceous plant species attracted significantly more pollinators and pollinator species than those sown with a mix of grasses only in the first year. Between six and eight bee (Apidae), butterfly (Lepidoptera) or hoverfly (Syrphidae) species were recorded/plot, compared to around two species on plots sown with grasses. The abundance of pollinators was strongly related to the cover of legumes and other non-grass plants. The cover of sown grasses and non-leguminous broadleaved plants was higher when sown following deep cultivation and herbicide treatment than after shallow cultivation alone. The cover with sown non-grass species (legumes and other broadleaved plants) was significantly higher in plots that were cut twice or three times for silage than in grazed treatments. There were four replicates of each treatment combination, on plots either 16 x 32 m (those cut for silage) or 25 x 50 m (grazed plots).



A replicated trial from 2005 to 2008 in the Hortobágy National Park, eastern Hungary (Török et al. 2010) found that perennial meadow grass species sown on ex-arable fields established well after two years of management as hay meadows, but created a dense cover that may prevent more specialist meadow species from establishing. In the first year, weedy annual herbs and grasses dominated (63-82% of vegetation). By the second year, sown grasses had increased, accounting for 16 to 86% of the plant cover. Ten fields were sown with two or three grass species in October 2005, at 25 kg/ha (Festuca rupicola, narrow-leaved meadow-grass Poa angustifolia and smooth brome Bromus inermis on six fields, and Festuca pseudovina and narrow-leaved meadow grass on four fields). The fields covered 93 ha in total. They were mown in late June 2007 and 2008 and hay removed. Plants were surveyed in four permanent plots in each field, from 2006 to 2008.


Referenced papers

Please cite as:

Dicks, L.V., Ashpole, J.E., Dänhardt, J., James, K., Jönsson, A., Randall, N., Showler, D.A., Smith, R.K., Turpie, S., Williams D.R. & Sutherland, W.J. (2019) Farmland Conservation Pages 291-330 in: W.J. Sutherland, L.V. Dicks, N. Ockendon, S.O. Petrovan & R.K. Smith (eds) What Works in Conservation 2019. Open Book Publishers, Cambridge, UK.