Restore/create freshwater marshes or swamps (multiple actions)
Overall effectiveness category Likely to be beneficial
Number of studies: 17
Background information and definitions
This section includes studies of marsh or swamp restoration/creation that test more than three separate actions at once, such that it is difficult to attribute outcomes to any single specific action. Where three or fewer actions have been used together in a study, results are reported elsewhere on this site: under each action (but noting the influence of the others, where appropriate) or sometimes as a combined action (e.g. Deposit soil/sediment and introduce vegetation). When multiple actions have been used but not clearly described, studies are summarized under Restore/create marshes or swamps (specific action unclear).
This section does not include studies that simply report the area or number of sites “restored” or “created”, without quantifying the vegetation in those sites.
Supporting evidence from individual studies
A site comparison study in 1978–1980 of three freshwater marshes in Florida, USA (Swanson & Shuey 1980) reported that a created marsh contained fewer plant species than natural marshes. After two years, the created marsh contained 70 vascular plant species (vs 76–88 in natural marshes; statistical significance not assessed). Vegetation cover in the created marsh was dominated by broadleaf cattail Typha latifolia, but this was not quantified. Methods: In summer 1978, one 0.16-ha depression was excavated in rangeland. Initially, 1.8 m of topsoil was removed but 0.6 m was backfilled to reach the final depth. The depression and surrounding site were also seeded with two pioneer herb species (to prevent erosion), limed (2,245 kg/ha dolomite) and fertilized (450 kg/ha). In summer 1980, plant species were recorded in the created marsh and two natural marshes (along a transect extending from the centre to the edge of each).Study and other actions tested
A replicated, paired, site comparison study in 1988–1990 involving six created freshwater wetlands in eastern Massachusetts, USA (Jarman et al. 1991) reported that they all developed vegetation cover within 1–2 years, with a similar plant species richness to adjacent natural wetlands but a distinct species composition. Statistical significance was not assessed. Created wetlands contained 9–20 plant species (vs natural: 13–23). The created and natural wetlands had 35–79% of plant species in common (average: 52%). All six created wetlands had >75% cover of native wetland vegetation. Woody plant cover was only 5–30%, suggesting herbaceous species were dominant (data not reported). Most adjacent natural wetlands, representing the target state, had a tree canopy with shrub and herb understory layers (not quantified). Red maple Acer rubrum seedlings were found in three of six created wetlands, at a density of 3–18/m2. Methods: In late 1990, vegetation was surveyed in six created wetlands (88–800 m2) and an unspecified number of adjacent natural wetlands. Surveys covered all woody vegetation and herbaceous vegetation in five 1-m2 quadrats/wetland. Wetland creation, in summer 1988 or 1989, involved excavating, creating “hydrological connections” to adjacent wetlands, adding wetland soil, planting wetland shrubs (six sites), herbs (five sites) or sods of vegetation (three sites), and sowing seeds of wetland herbs (one site).Study and other actions tested
A study in 1992–1993 on a former sand mine in New Jersey, USA (Vivian-Smith & Handel 1996) reported that following multiple restoration interventions the site developed vegetation cover, including some wetland-characteristic species. After 10 months, 82 plant species were recorded in quadrats across the site (area surveyed: 26.75 m2) with 6.3 species/0.25-m2 quadrat. There were at least 20 wetland-characteristic species across the site. The most abundant taxa were panicgrasses Panicum spp. (33 plants/m2), tapertip rush Juncus acuminatus (15 plants/m2) and toad rush Juncus bufonius (8 plants/m2). The only common woody taxa were willows Salix spp. (two species; 0.3–0.4 plants/m2). Methods: In October–November 1992, a former sand mine (last mined in the early 1990s) was subjected to multiple interventions: reprofiling, adding soil from another wetland (5–10 cm layer over the site), planting herbs and woody plants (species not reported), sowing a grass cover crop (including a panicgrass species), mulching with straw and adding lime. The aim was to restore a shrubby freshwater wetland. Vegetation was surveyed in August 1993, in 107 quadrats (each 0.25 m2) spread across the site.Study and other actions tested
A replicated, site comparison study in 1990–1993 of eight dune slacks in California, USA (Parikh & Gale 1998) reported that created slacks contained different plant communities to mature slacks, with more plant species and greater cover of annual herbs (but similar cover of other plant groups). Results summarized for this study are not based on assessments of statistical significance. Over their first three years, created slacks contained plant communities distinct from those in mature slacks – although this distinctiveness declined over time (data reported as a graphical analysis). Created slacks contained 69–86 plant species/year (vs mature: 28–59 species/year). After three years, created slacks had 70–95% cover of annual herbs (vs mature: <1–37%). For all other plant groups, cover did not clearly differ between created and mature slacks: perennial herbs (created: 69–135%; mature: 82–127%), shrubs (created: 27%; mature: 13–86%), trees (created: 7–11%; mature: 0–43%) and floating aquatic plants (created: 3–5%; mature: 0–8%). For data on the abundance of individual plant species, see original paper. Methods: In winter 1990/1991, multiple interventions were used to create wetlands within dune slacks (low-lying areas amongst dunes): removing existing vegetation and topsoil, relandscaping, adding wetland soil and sowing seeds. Afterwards, non-native plants were controlled by cutting, physical removal or herbicide application. Between 1991 and 1993, vegetation was surveyed in the two created slacks (twenty 5-m2 plots/site) and six nearby mature slacks (six to nine 30-m2 plots/site). Surveys included wetlands and immediately adjacent uplands.Study and other actions tested
A before-and-after study in 1993–1999 of a freshwater marsh in Ontario, Canada (Smith et al. 2001) reported that following multiple restoration interventions, there were increases in emergent vegetation coverage and species richness. Statistical significance was not assessed. In the year before intervention began, the marsh contained 33 ha of emergent vegetation and 17 emergent plant species. After 5–6 years of intervention, there were 51 ha of emergent vegetation and 23 emergent plant species. Four species present in historical records (1950) were not found in the more recent surveys, either before or after intervention. Methods: Multiple interventions were carried out from 1993: planting emergent vegetation, containing sewage overflow, encouraging sustainable land management practices in the watershed and, in 1997, installing a barrier to keep large carp Cyprinus carpio out of the marsh. Emergent vegetation surveys and mapping were carried out before intervention (1993) and after 5–6 years (1998–1999). The study notes likely differences in survey effort and emergent plant definitions between surveys, and that low water levels in 1998–1999 could have contributed to expansion of emergent vegetation.Study and other actions tested
A study in 1993–1998 of a created, tidal, freshwater marsh in New Jersey, USA (Leck 2003) reported that vegetation was present on the site, but that species richness and cover were highly variable across space and time. Over all samples taken 2–5 years after wetland creation began, 92 plant species were recorded in the marsh. This included 59 wetland-characteristic species, and 14 of 14 planted species. The number of species per sampling quadrat (5–30 species/0.25 m2) and cover of individual plant species (see original paper for full data) were highly variable: depending on the area of the marsh, water level, and year. For example, two years after wetland creation began, jewelweed Impatiens capensis cover was 0–4% in all areas and water levels. After five years, jewelweed cover was 23–58% at the driest points but 0–2% at the wettest points. Methods: In 1993, an area of deposited dredge spoil was cleared of vegetation, excavated to form islands and channels, and graded. Throughout 1994, dams were removed to open the channels. In 1994 and 1995, fourteen herb species were planted across the site. Each August between 1995 and 1998, plant species and their cover were surveyed in 108 quadrats in wetland areas (water depth at high tide approximately 5–50 cm).Study and other actions tested
A study in 1995–2000 of an ephemeral wetland restoration site on farmland in Minnesota, USA (Bohnen & Galatowitsch 2005) reported that following multiple interventions, the site developed vegetation cover, including some naturally colonizing species and some wetland species. Approximately five years after restoration began, 256 plant species were recorded in the site. Of these, 112 had been introduced as plants or seeds whilst 144 had colonized completely on their own. There were 147 wetland species. Amongst the planted/sown species, establishment success and abundance varied between wetland zones. The most abundant species in each zone were broadleaf arrowhead Sagittaria latifolia (wettest, emergent marsh zone; 25–49% average cover), bluejoint Calamagrostis canadensis and two forbs (sedge meadow zone; each 5–24% average cover) and black-eyed Susan Rudbeckia hirta (wet grassland zone; 25–49% average cover). Methods: From August 1995, multiple interventions were carried out to restore a zoned wetland on former agricultural land: applying herbicide, prescribed burning and physical removal to control invasive plants, cutting and applying herbicide to woody plants, breaking drainage systems to rewet the site, and sowing (autumn 1996) and planting (summer 1997) >100 plant species found in local wetlands. In summer 2000, vegetation was surveyed across the wetland in 28 monitoring units.Study and other actions tested
A study in 2000–2003 of a freshwater swamp restoration site on former cropland in North Carolina, USA (Bruland & Richardson 2005) reported that following multiple interventions, nineteen plant species colonized the created hummocks, hollows and flats within two years. Of these, 15 were wetland species and six were wetland-characteristic species. Eighteen species occurred on only one landform: either on raised hummocks, low hollows or the flats in-between. Flats had a higher plant species richness (5.0 species/m2) than hollows (3.5 species/m2) or hummocks (1.9 species/m2). Hollows supported a higher plant biomass (1,390 g/m2) than flats (900 g/m2) or hummocks (290 g/m2). Methods: In winter 2000/2001, a 37-ha agricultural field was subjected to multiple restoration interventions: stripping the topsoil; reprofiling the surface into hummocks (1 m tall; 1.5 m diameter), hollows (30 cm deep; 20–40 m2) and flats; blocking ditches to raise the water table; replacing the topsoil; and planting tree seedlings in the hollows and flats (1,680 seedlings/ha). In October 2003, herbaceous vegetation was surveyed in eighteen 5-m2 plots (six plots/landform). Above-ground biomass was cut from one 50-cm diameter subplot/plot, then dried and weighed.Study and other actions tested
A replicated, before-and-after study in 1953–2002 of two freshwater wetland restoration sites in England, UK (Akers & Alcorn 2006) reported that following multiple interventions, nine plant species (re)appeared. Five years after restoration, seven wetland plant species had recolonized the sites (i.e. present before degradation in 1953, absent after degradation in 1983, then present after restoration in 2002). Two new plant species colonized the wetlands (i.e. not present in 1953 or 1983). Eight locally rare plant species that were present in 1983 were absent in 2002. The study also reported reduced cover of Cladonia lichens following intervention (but this was not quantified). Methods: In 1997, multiple interventions were applied in two degraded wetland depressions (overgrown by willow Salix spp. after grazing stopped in the 1950s and water levels dropped in the 1970s). Willow trees and common reed Phragmites australis were cut and removed, willow stumps were treated with herbicide, and late-summer grazing by sheep and goats was reintroduced (further details not reported). Vegetation surveyed in 2002 was compared to previously published records from 1953 and 1983.Study and other actions tested
A replicated, before-and-after, site comparison study in 2000–2003 involving 15 ephemeral freshwater wetland restoration sites in South Carolina, USA (De Steven et al. 2006) reported that multiple interventions changed the vegetation type, cover and species richness. Before intervention, the sites were dominated by facultative wetland trees (see original paper for data on individual species abundance). They contained 22 plant species on average (including 5 wetland-characteristic) and had 143% vegetation cover (herbaceous: 6%; wetland-characteristic: 29%). One and two years after intervention, during a dry spell, restored wetlands were dominated by facultative and wetland-characteristic herbs. They contained 36–44 plant species (including 14–20 wetland-characteristic) and had 65–78% vegetation cover (herbaceous: 40–60%; wetland-characteristic: 24–37%). In the third, wetter year, the vegetation in restored wetlands was dominated by facultative trees (as saplings or resprouts) with some submerged, floating and emergent herbs. There were now only 17 plant species/wetland (including 8 wetland-characteristic) and vegetation cover was only 23% (herbaceous: 13%; wetland-characteristic: 13%). Three unrestored wetlands retained similar vegetation to pre-restoration conditions throughout the study (e.g. dominated by woody vegetation; 18–27 plant species). Methods: In 2000–2001, fifteen degraded wetlands (≤2 ha; drained and overgrown but with actively flowing remnant ditches) were subjected to multiple restoration interventions: plugging drainage ditches, cutting and removing existing trees, and applying herbicide to resprouting stumps. Eight of the wetlands were also sparsely planted with seedlings of wetland-characteristic trees; see Barton et al. (2004) and (12). Vegetation was sampled in August before intervention (2000) and for three years after (2001–2003). Three unrestored wetlands were also monitored for comparison. Some of the restored wetlands in this study were used in (11), and all were used in (12).
Additional Reference: Barton C.D., De Steven D. & Kilgo J.C. (2004) Mitigation bank promotes research on restoring coastal plain depression wetlands. Ecological Restoration, 22, 291–292.Study and other actions tested
A replicated study in 2000–2004 of 12 ephemeral freshwater wetland restoration sites in South Carolina, USA (De Steven & Sharitz 2007) reported that following multiple interventions, the sites developed vegetation cover including wetland-characteristic species. Approximately one year after intervention, overall vegetation cover was 48% (wetland-characteristic species: 23%) and there were 11.4 plant species/4 m2 (wetland-characteristic: 4.9). Approximately three years after intervention, overall vegetation cover was 90% (wetland-characteristic: 54%) and there were 8.7 plant species/4 m2 (wetland-characteristic: 4.7 species/4 m2). Methods: In 2000–2001, twelve degraded wetlands (≤2 ha; drained and overgrown by facultative wetland trees) were restored by plugging drainage ditches, cutting and removing existing trees, and applying herbicide to resprouting stumps. Some of the wetlands were also sparsely planted with seedlings of wetland-characteristic trees; see Barton et al. (2004) and (12). In August 2002 and 2004, plant species and cover (excluding resprouting trees) were recorded in one 4-m2 quadrat/wetland. The first survey was during a drought, but the second after normal rainfall. The wetlands in this study were also used in (10) and (12).
Additional Reference: Barton C.D., De Steven D. & Kilgo J.C. (2004) Mitigation bank promotes research on restoring coastal plain depression wetlands. Ecological Restoration, 22, 291–292.Study and other actions tested
A replicated, before-and-after, site comparison study in 2000–2005 of 16 ephemeral freshwater wetland restoration sites in South Carolina, USA (De Steven et al. 2010) reported that multiple interventions changed the vegetation type, cover and species richness. Results summarized for this study are not based on assessments of statistical significance. Before intervention, the sites were dominated by facultative wetland trees (data not reported). They contained 23 plant species on average (including 8 wetland-characteristic) and had 141% overall vegetation cover (woody: 130%; herbaceous: 10%; wetland-characteristic: 48%). Reference wetlands contained 10–33 species. After one year, restored wetlands were dominated by facultative and wetland-characteristic herbs. They contained 43 plant species (including 22 wetland-characteristic) and had 77% vegetation cover (woody: 18%; herbaceous: 59%; wetland-characteristic: 39%). After five years, restored wetlands contained a mixture of herbs and young woody plants. They contained 35 plant species (including 21 wetland-characteristic) and had 102% vegetation cover (woody: 40%; herbaceous: 64%; wetland-characteristic: 63%). At this point, the overall plant community in restored wetlands was 37–41% similar to 29 reference local marsh and swamp communities (vs 36–41% similarity between natural marsh or swamp communities from different sites). For data on the abundance of individual plant species, see original paper. Methods: In 2000–2002, sixteen degraded wetlands (≤2 ha; drained and overgrown) were subjected to multiple restoration interventions: plugging drainage ditches, cutting and removing existing trees, and applying herbicide to resprouting stumps. Eight of the wetlands were also sparsely planted with seedlings of wetland-characteristic trees. Vegetation was sampled in August before restoration (2000) and for five years after (2001–2005). Some of the restored wetlands in this study were also used in (10) and (11).Study and other actions tested
A replicated study in 2004–2010 of four ephemeral marsh restoration sites within farmland in Oregon, USA (Wold et al. 2011) reported that following multiple interventions, vegetation cover developed. Approximately three and a half years after intervention began, there were 55–99 plant species/marsh (native: 42–67; non-native: 13–33) and total vegetation cover was 128–177% (native: 91–174%; non-native: 4–37%). Two marshes were also monitored after five and a half years. There were now 86–112 plant species/marsh (native: 58–74; non-native: 28–38). Total vegetation cover had increased in one of two marshes (to 242%). Native cover increased in both (to 103–240%) and exotic cover had decreased in both (to 2–12%). These results are not based on assessments of statistical significance. Methods: Four areas (3–16 ha) of agricultural land were managed to restore ephemeral marshland. Interventions included mowing, burning, applying herbicide (general or grass-specific), rewetting by removing drainage ditches, removing weeds by hand, seeding herbs and directly planting herbs. In the second and fifth summer after intervention began, vegetation was surveyed at ≥400 points/marsh (≥200 points in each of 2–4 plots/marsh).Study and other actions tested
A replicated, site comparison study of six papyrus marshes in East Africa (Kiwango et al. 2013) reported that created marshes developed similar biomass of papyrus Cyperus papyrus to natural marshes, within 18 months. Statistical significance was not assessed. In two created marshes in Tanzania, above-ground papyrus biomass was 3,900 g/m2. This was within the range reported for other natural East African papyrus marshes: 883–8,456 g/m2 (data from the four studies that clearly measured above-ground, rather than total, biomass). Methods: Marsh creation involved multiple interventions: physically removing problematic plants (water hyacinth Eichhornia crassipes and crops), digging the compacted soil, planting papyrus and other wetland reeds/grasses/shrubs, digging channels to rewet the marsh, and fencing to exclude humans and animals. After 18 months, papyrus was cut from the created marshes, then dried and weighed. Previously published biomass data from natural marshes were reported for comparison. The study does not report dates of intervention or monitoring.Study and other actions tested
A replicated, site comparison study in 2009–2010 of six wet prairies in Oregon and Washington, USA (Taylor & Santelmann 2014) found that restored prairies contained a different plant community to remnant semi-natural prairies, but had similar richness and cover. Approximately 3–8 years after restoration began, the overall plant community composition significantly differed between restored and remnant prairies (data reported as a graphical analysis). Other vegetation metrics did not significantly differ between restored and remnant prairies. This was true for overall richness (restored: 18–40; remnant: 13–48 taxa/100 m2), native richness (restored: 11–28; remnant: 9–26 taxa/100 m2), native diversity (data reported as a diversity index), overall vegetation cover (restored: 114–164%; remnant: 95–115%), grass cover (restored: 44–108%; remnant: 72–93%) and forb cover (restored: 20–120%; remnant: 2–43%). For data on the abundance of individual plant species, see original paper. Methods: In summer 2009, plant taxa and their cover were recorded in three restored and three remnant seasonally flooded wet prairies (three 100-m2 plots/site, in areas dominated by tufted hairgrass Deschampsia cespitosa). Taxa were also recorded in spring 2010. Restoration of previously drained and “altered” sites involved prescribed burning, annual herbicide application, annual mowing, sowing cover crops and sowing native species (four of these five interventions/site, over 3–8 years). Remnant sites were the best remaining, but not completely undisturbed, wetland prairies in the area. They were also managed with some of the interventions, plus hand weeding.Study and other actions tested
A study in 2011–2013 of a created freshwater marsh in New York, USA (Pier et al. 2015) reported that it contained 44–46 plant species after approximately two years. There were 17 plant species present in the summer one year after creation began (and before deliberately planting vegetation) and 44–46 species present in the summer two years after creation began (one year after planting). Plant species diversity was also higher in the second summer (data reported as a diversity index; statistical significance not assessed). Note that sampling effort was not the same in both years. In the second summer, the plant communities somewhat differed between pools within the marsh, with an average 59% similarity in composition. Methods: A stormwater treatment marsh was created by (a) demolishing buildings and excavating eight pools in June 2011, (b) clearing established vegetation, reprofiling the pools and planting 85 species of trees, shrubs, emergent herbs and submerged herbs, in September 2012, and (c) mowing in July 2013. Plant species and their cover were surveyed along three 20-m transects in June 2012, then nine 20-m transects (1–2 transects/pool) in June and August 2013. Transects included terrestrial, emergent wetland and aquatic vegetation.Study and other actions tested
A replicated study in 2013 of eight 10-year-old restored/created freshwater wetlands in Maryland, USA (Russell & Beauchamp 2017) reported that they contained a total of 134 plant species, including 65 wetland-characteristic species. There were 45–78 species/wetland in the ground layer (<1 m tall) and 4–10 species/wetland in the tree layer (woody species >1 m tall). The study also noted that several environmental characteristics were related to plant diversity and/or community composition (e.g. wetland size, slope, water regime, soil fertility; see original paper for details). Methods: In June–August 2013, vegetation was surveyed along transects in eight restored/created depressional wetlands (4–6 transects/wetland, extending from the centre to the surrounding upland). The wetlands had been restored (one) or created (seven) on farmland in 2003–2004, by: removing drainage tiles/plugging ditches; adding coarse woody debris; adding wheat/barley straw; and planting trees/shrubs around the margins of the flooded centre and in the surrounding uplands.Study and other actions tested