Background information and definitions
Important wild vertebrates in marshes include mammals (e.g. deer, rabbits, hares, kangaroos, feral horses, feral pigs), birds (e.g. ducks, geese, swans), reptiles (e.g. turtles) and fish (e.g. carp). These animals can damage vegetation directly by eating it. They can also affect vegetation indirectly, for example by trampling, creating trails, digging, burrowing or defecation (Fuller 1985). Wild vertebrates could be physically excluded from pristine sites to prevent damage, or from degraded sites to let them recover.
Caution: Disturbance from animals may be desirable in some habitats. It can help to control undesirable vegetation and maintain species richness or open water patches (e.g. Smith et al. 2012). Fences or cages can be expensive and require ongoing maintenance.
Although studies often intend to exclude a particular problematic species, other animals of a similar size will incidentally be excluded. Smaller animals such as insects can usually still access vertebrate exclusion plots. The benefits of this action may be highly dependent on the type/size of fencing used, and the abundance of problematic animals in the study site.
Related actions: Use barriers to keep livestock off ungrazed marshes; Exclude or remove livestock from historically grazed marshes; Exclude wild invertebrates using physical barriers; Use fences or barriers to protect planted areas.
Fuller D.A., Sasser C.E., Johnson W.B. & Gosselink J.G. (1985) The effects of herbivory on vegetation on islands in Atchafalaya Bay, Louisiana. Wetlands, 4, 105–114.
Smith A.N., Vernes K.A. & Ford H.A. (2012) Grazing effects of black swans Cygnus atratus (Latham) on a seasonally flooded coastal wetland of eastern Australia. Hydrobiologia, 697, 45–57.
Supporting evidence from individual studies
A replicated, controlled study in 1980–1981 in a freshwater marsh in Louisiana, USA (Fuller et al. 1985) found that plots fenced to exclude wild rodents contained more vegetation biomass than plots that remained open to grazing, but had a similar vegetation density. After both one and two growing seasons, fenced plots contained more live, above-ground vegetation biomass (242–312 g/ m2) than open plots (117–187 g/m2). The same was true for live biomass of the two most common species in the marsh: broadleaf arrowhead Sagittaria latifolia (fenced: 172; open: 106 g/m2) and valley redstem Ammania coccinea (fenced: 43; open: 10 g/m2) and for dead biomass (fenced: 281–348; open: 145–204 g/m2). However, the density of live plant stems did not significantly differ between fenced (291–396 stems/m2) and open plots (265–481 stems/m2). Methods: In March 1980, ten 1-m2 plots in a freshwater marsh were fenced (2 x 4 cm wire mesh) to exclude nutria Myocastor coypus and muskrat Ondatra zibethicus. Waterbirds could access all plots. In October 1980 and 1981, vegetation was sampled (cut, counted, dried and weighed) in each plot and 10 adjacent plots that were left open to herbivores.Study and other actions tested
A replicated, randomized, paired, controlled study in 1991 in a freshwater marsh in Louisiana, USA (Taylor & Grace 1995) reported that plots fenced to exclude nutria Myocastor coypus contained more overall vegetation biomass than plots that remained open to grazing, but had similar plant species richness. Statistical significance was not assessed. After two growing seasons, above-ground vegetation biomass was 1,600 g/m2 in fenced plots, compared to 1,270 g/m2 in open plots. However, fenced and open plots contained a statistically similar biomass of the dominant plant species: switchgrass Panicum virgatum (fenced: 771; open: 517 g/m2) and big cordgrass Spartina cynosuroides (fenced: 381; open: 355 g/m2). Fenced plots contained 12.3 plant species/m2, compared to 12.7 plant species/m2 in open plots. Methods: In March 1990, twelve 4-m2 plots were established (in three sets of four) in a freshwater marsh. Six of the plots (two random plots/set) were fenced (2.5 cm plastic-coated mesh) to exclude nutria (and other large mammals). The other six plots were left open. In September 1991, all vegetation was cut from one 1-m2 quadrat/plot. Plant species were identified, then the vegetation was dried and weighed. This study was in the same area as (4), but used a different experimental set-up.Study and other actions tested
A replicated, controlled study in 1990–1991 in a freshwater marsh in the Netherlands (Clevering & van Gulik 1997) reported that fencing to exclude waterfowl maintained the density and biomass of lakeshore bulrush Scirpus lacustris ssp. lacustris over one growing season, but did not affect vegetation recovery over a second growing season. Statistical significance was not assessed. Over the first growing season, fenced plots contained more bulrush (density: 165–360 shoots/m2; above-ground biomass: 800–1,150 g/m2) than plots open to summer grazing (density: 70–225 shoots/m2; above-ground biomass: 200–510 g/m2). Over the second growing season, all plots were fenced and recovered to have similar bulrush density (315–450 g/m2) and above-ground biomass (1,210–1,480 g/m2) by late summer. Methods: The study used twelve 6-m2, tidal, lakeshore plots with 3-year-old bulrush stands. Four plots were fenced (12 cm wire mesh) in spring 1990 to protect them from further waterfowl grazing. The other eight plots were left open to one or two grazing events in summer 1990. All plots were fenced from autumn 1990. Bulrush shoots were counted and measured throughout the 1990 and 1991 growing seasons. Above-ground dry biomass was estimated from length-mass relationships.Study and other actions tested
A replicated, randomized, paired, controlled, before-and-after study in 1992–1994 in a freshwater marsh in Louisiana, USA (Ford & Grace 1998) found that fencing to exclude wild mammals increased overall vegetation biomass, but had mixed effects on the cover of dominant plant species and no significant effect on plant species richness. After two years, above-ground vegetation biomass was higher in fenced plots (960–2,080 g/m2) than in plots that remained open to grazing (780–920 g/m2). Fenced plots also had greater cover of switchgrass Panicum virgatum than open plots (fenced: 54–68%; open: 30–51%), but statistically similar cover of saltmeadow cordgrass Spartina patens (fenced: 31–78%; open: 56–71%). Fencing had no significant effect on plant species richness, with statistically similar changes in fenced plots (increase of 0–2.4 species/m2 over two years) and open plots (increase of 1.8 species/m2 over two years). Methods: In autumn 1992, ten pairs of 4-m2 plots were established in a freshwater marsh. Ten plots (one random plot/pair) were fenced (5 cm wire mesh with hooks to prevent burrowing) to exclude nutria Myocastor coypus and wild boar Sus scrofa (and other large mammals). The other 10 plots were not fenced. Half of the plots under each treatment were also burned in autumn 1992 and 1993. Plant species and cover were recorded in autumn 1992 (before intervention) and 1994. Vegetation was cut from one 0.25-m2 quadrat/plot, then dried and weighed, in autumn 1994. This study was in the same area as (2), but used a different experimental set-up.Study and other actions tested
A replicated, paired, controlled study in 1996–1997 in four freshwater and brackish marshes in Delaware, USA (Sherfy & Kirkpatrick 2003) reported that plots fenced to exclude snow geese Chen caerulescens had greater vegetation cover (15–57%) than plots grazed by geese (<1–11%). Statistical significance was not assessed. Methods: In September–October 1996, sixteen goose exclosures were established across four impounded marshes with fresh or “slightly” brackish water. The study does not separate results for each marsh type. There were four exclosures/marsh. Exclosures were 1.2 x 1.2 m, fenced with 1.5 x 1.5 cm plastic mesh and topped with bright plastic strips to prevent snow geese from landing. Over winter 1996/1997, total vegetation cover was estimated in the 16 exclosures and 16 adjacent plots open to, and grazed by, snow geese.Study and other actions tested
A replicated, controlled study in 1995–1998 on the shore of a freshwater lake in the Netherlands (Coops et al. 2004) reported that areas fenced to exclude waterbirds contained more emergent vegetation biomass, over three years, than plots left open to grazing. In three of three years, the total above-ground biomass of tall emergent vegetation was greater in fenced than open areas (data reported graphically). This was driven by herbivory in shallow water: the maximum biomass in deeper water was actually slightly lower in fenced plots than open plots (e.g. after three years, fenced: 520 g/m2; open: 630 g/m2). Methods: The study used a 3-ha area of the shoreline of Lake Volkerak-Zoommeer, where tall emergent vegetation was developing following experimental drawdowns and floods (beginning in spring 1995). Parts of the study area were fenced to exclude waterbirds (2-m-high fence with ropes above) and parts were left open. Each summer between 1995 and 1998, above-ground biomass was sampled in transects in the fenced and open areas. The study does not report further details of the experimental set-up or sampling methods.Study and other actions tested
A before-and-after study in 1934–1999 of a freshwater marsh in Ontario, Canada (Chow-Fraser 2005) found that excluding common carp Cyprinus carpio had no significant effect on the area covered by emergent vegetation. Two years after carp exclusion, the area covered by emergent vegetation (19% of the marsh) did not significantly differ from expected coverage given the water level at the time (23–28%). Methods: From spring 1997, a barrier system was used to prevent large carp (>40 cm long) from migrating into the marsh from the adjacent lake. The area of emergent vegetation across the marsh before (1934–1990) and after (1999) carp exclusion was obtained from previously published data (based on aerial photographs or field surveys). Carp had been introduced in 1908. The relationship between emergent vegetation coverage and water level before exclusion was used to determine the expected coverage based on the water level after exclusion. Note that other restoration interventions had been carried out since 1992 (sewage management, watershed land management, planting vegetation; see Smith et al. 2001).
Additional reference: Smith T., Lundholm J. & Simser L. (2001) Wetland vegetation monitoring in Cootes Paradise: measuring the response to a fishway/carp barrier. Ecological Restoration, 19, 145–154.Study and other actions tested
A replicated, paired, controlled study in 1999 in a tidal freshwater marsh in Maryland, USA (Haramis & Kearns 2007) found that plots from which large vertebrates were excluded developed a greater density of wild rice Zizania aquatica than exposed plots. After one growing season, exclusion plots contained more wild rice plants on average (97 plants/m2; 105 flowering stalks/m2) than adjacent open plots (3 plants/m2; 0 flowering stalks/m2). The mesh size of exclosures had no significant effect on the density of wild rice plants or flowering stalks, plots enclosed by a smaller mesh supported taller and thicker wild rice shoots (see original paper for data; height and stem diameter only reported for exclosures). Methods: In April 1999, twenty-four 1-m2 plots were established (in six sets of four) in a marsh with naturally germinating wild rice. Eighteen of the plots (three plots/set) were fenced with 1.5-m-tall wire mesh to exclude vertebrates (birds, mammals, large turtles and fish). Six fences (one plot/set) had each of three mesh sizes: small (1.3 x 1.3 cm), medium (2.5 x 2.5 cm) or large (5.1 x 10.2 cm). The other six plots (one plot/set) were left open to all animals. The study reported intense grazing by Canada geese Branta canadensis in these open plots, and that sediment was trapped by the fences (especially those with small mesh). After one growing season, all rice plants were counted in each plot and 10 rice plants/plot were measured.Study and other actions tested
A replicated, controlled study in 2007 in a freshwater marsh in New South Wales, Australia (Smith et al. 2010) found that plots fenced to exclude black swans Cygnus atratus contained a greater biomass and density of dominant spikesedge Eleocharis equisetina, than plots left open to swans. After 20 weeks, fenced plots contained more spikesedge biomass (above-water: 540 g/m2; above-sediment: 1,200 g/m2) than open plots (above-water: 3 g/m2; above-sediment: 580 g/m2). Fenced plots contained 64–130 spikesedge stems/m2 compared to 1–15 spikesedge stems/m2 in open plots (statistical significance not assessed). Methods: In February 2007, ten 4-m2 plots in a freshwater marsh (occasionally brackish) were fenced (5 cm wire mesh) to exclude black swans. The whole study area was also fenced to exclude cattle. Vegetation was sampled in the 10 swan exclosures, and five nearby plots grazed by swans, until July 2007. Emergent spikesedge stems were counted. All spikesedge material above the sediment was cut, dried and weighed. This summary does not include data (a) for five open plots that were not grazed by swans in 2007, and (b) for exclosures after July, because some exclosures were corroded and grazed by swans.Study and other actions tested
A replicated, randomized, paired, controlled study in 1994–2008 in three floodplain marshes in New South Wales, Australia (Berney et al. 2014) found that plots fenced to exclude wild mammals typically contained more plant biomass than plots that remained open to mammals, but typically had a similar plant community composition and species richness. After four years, fenced plots contained more live, above-ground plant biomass than open plots in two of three marshes (for which fenced: 1,640–2,420 g/m2; open: 930–1,300 g/m2). There was no significant difference in the other marsh (for which fenced: 850 g/m2; open: 630 g/m2). The overall plant community composition was statistically similar in fenced and open plots in at least 33 of 35 comparisons over 14 years (data reported as graphical analyses). In all 35 comparisons, fenced plots had similar plant species richness to open plots (fenced: 2–20 species/m2; open: 3–19 species/m2). The study also reported data on the cover of individual plant species (see original paper). Methods: In early 1994, twelve pairs of 25 x 25 m plots were established across three historically grazed floodplain marshes (four pairs/marsh). In each pair, one random plot was fenced to exclude wild mammals (native marsupials and feral pigs/rabbits). The other plots were left open. Domestic cattle were excluded from all 24 experimental plots. In 1994–1998 (a wetter period) and 2007–2008 (a drier period), plant species and cover were recorded in ten 1-m2 quadrats/plot. In May 1998, live above-ground vegetation was collected from two 0.25-m2 quadrats/plot, then dried and weighed.Study and other actions tested
A replicated, paired, controlled study in 2000–2002 in tidal, freshwater marshes along a river in New Jersey, USA (Nichols 2014) reported that plots fenced to exclude Canada geese Branta canadensis contained more, taller wild rice Zizania aquatica plants than plots exposed to intense goose grazing, although these effects typically disappeared after geese were controlled. In the first year of the study, with a large and uncontrolled goose population, wild rice plants were more abundant and taller in goose exclosures (70 plants/m2, 241 cm tall) than in plots open to geese (15 plants/m2, 200 cm tall). In the following two years, when the goose population was controlled, differences between exclosure and open plots were typically eliminated (and if not, reduced in magnitude). Exclosure and open plots contained a statistically similar density of wild rice in two of two years with goose control (fenced: 60–68 plants/m2; open: 55–58 plants/m2), and contained wild rice of a statistically similar height in one of two years with goose control (for which fenced: 208 cm; open: 212 cm). Methods: Each April between 2000 and 2002, 17–22 pairs of 1-m2 plots were established on tidal freshwater marshes along the lower Maurice River. In each pair, one plot was fenced to exclude geese (5–10 cm wire mesh, 1.5 m high) whilst the other was left open. In 2001 and 2002, the local goose population was reduced by killing and scaring. Wild rice was counted (all plants in each plot) and measured (10 plants in centre of each plot) in autumn each year.Study and other actions tested
A replicated, paired, controlled study in ten freshwater wetlands in the Netherlands (Sarneel et al. 2014) found that plots fenced to exclude wild waterbirds and rodents contained more, and richer but not more diverse, emergent vegetation than plots that remained open to grazing. In both the first and second growing season after intervention, fenced plots had higher emergent vegetation cover (47–62%) than open plots (34–36%). Cover also increased significantly more over time in the fenced plots. In both growing seasons, fenced plots had higher emergent plant species richness than open plots, but statistically similar emergent plant diversity (data not reported). In the first growing season, fenced plots contained more above-ground vegetation biomass, in both permanently flooded areas (fenced: 1,220; open: 790 g/m2) and saturated areas (fenced: 320; open: 180 g/m2). In the second growing season, emergent vegetation extended further into the water in fenced plots (fenced: 490 cm; open: 360 cm). Methods: In March 2011, fifty pairs of 3 x 6 m plots were established at the margins of 10 wetlands. Each plot contained emergent vegetation and open water. One plot in each pair was fenced (chicken wire sides, additional wire on top) to exclude large animals (waterbirds and muskrats Ondatra zibethicus). Plant species and their cover were recorded in a 6-m-long transect crossing each plot in July 2011 and 2012. All vegetation was cut, dried and weighed from two 0.2-m2 quadrats/plot in August 2011.Study and other actions tested