Use prescribed fire to maintain or restore disturbance: freshwater marshes
Overall effectiveness category Trade-off between benefit and harms
Number of studies: 15
Background information and definitions
Disturbance can clear dominant plants, maintain light availability and control nutrient levels – and may maintain vegetation in a desirable and/or species-rich state (Hall et al. 2008; Middleton 2013). Therefore, conservationists sometimes want to actively restore disturbance where it has ceased, or maintain disturbance at a site where it would otherwise be lost. Prescribed burns are one way to do this.
Fire itself may be the historic or traditional disturbance that maintains wetlands in a desirable state. Some wetlands, especially ones that dry out in summer, burn naturally every few years (Sutter & Kral 1994). In other wetlands, prescribed burns have been used by humans to manage vegetation (Middleton 2013). Reduced disturbance from fire in these systems – whether through abandonment or deliberate control of fire (e.g. via fire breaks or legislation) – can be detrimental to vegetation diversity, composition and structure (e.g. Clark & Wilson 2001).
Caution: Disturbance, and fire in particular, is not a natural feature of all wetlands. For example, even within the southeast USA, the natural fire frequency can vary from once per year to once per century (Sutter & Kral 1994). It can be difficult to control the intensity, duration and area of prescribed burns. Burns in winter or wet season, might be easier to control than burns in the summer or dry season. Smoke from prescribed burns could be detrimental to human health, especially near urban areas (Agee 1996). Also note potential impacts on animals within wetlands – but that some taxa might be unaffected or be able to avoid fire (e.g. Ditlhogo et al. 1992).
The timing and duration of monitoring might be particularly important when evaluating the effects of this action. Burning might produce apparently desirable changes in vegetation over the short term, followed by a rapid return to a degraded state.
Related actions: Use prescribed fire to control problematic plants, whose success is not linked to a change in disturbance regime; Reduce frequency of prescribed burning; Reduce intensity of prescribed burning; Change season/timing of prescribed burning.
Agee J. (1996) Achieving conservation biology objectives with fire in the Pacific Northwest. Weed Technology, 10, 417–421.
Clark D.L. & Wilson M.V. (2001) Fire, mowing, and hand-removal of woody species in restoring a native wetland prairie in the Willamette Valley of Oregon. Wetlands, 21, 135–144.
Ditlhogo M.K.M., James R., Laurance B.R. & Sutherland W.J. (1992) The effects of conservation management of reed beds. I. The invertebrates." Journal of Applied Ecology, 29, 265–276.
Hall S.J., Lindig-Cisneros R. & Zedler J.B. (2008) Does harvesting sustain plant diversity in Central Mexican wetlands? Wetlands, 28, 776–792.
Middleton B.A. (2013) Rediscovering traditional vegetation management in preserves: trading experiences between cultures and continents. Biological Conservation, 158, 750–760.
Sutter R.D. & Kral R. (1994) The ecology, status, and conservation of two non-alluvial wetland communities in the South Atlantic and Eastern Gulf coastal plain, USA. Biological Conservation, 235–243.
Supporting evidence from individual studies
A replicated, paired, controlled, before-and-after study in 1972–1975 in a created reedbed in the Netherlands (Mook & van der Toorn 1982) found that spring-burned plots typically contained more common reed Phragmites australis biomass and thicker reed shoots than unburned plots. In late April/early May, burned plots contained thicker reed shoots than unburned plots in three of four comparisons (all burned mid-April; burned: 6.2–7.2 mm; unburned: 4.5–5.7 mm; other comparison no significant difference). In August/September, burned plots contained more above-ground reed biomass than unburned plots in seven of eight comparisons (for which burned: 1,200–1,760 g/m2; unburned: 530–1,270 g/m2; other comparison no significant difference). In the autumn before intervention, reed biomass was similar under both treatments (burned: 1,010–1,040 g/m2; unburned: 1,080–1,120 g/m2; statistical significance not assessed). Methods: In 1971, two pairs of plots were established in a young reedbed (sown in 1968). One pair was in a wetter area (flooded spring to autumn). One pair was in a drier area (water table 30–100 cm below surface). Between 1972 and 1975, one plot/pair was burned each spring (as early as possible). The other plots were not burned. Between April and October each year, all standing reed stems were cut from quadrats (0.25–0.50 m2; 2–6 quadrats/plot/sampling date), then measured, dried and weighed. This study used the same experimental set-up as (2).Study and other actions tested
A replicated, paired, controlled study in 1972–1975 in a created reedbed in the Netherlands (van der Toorn & Mook 1982) reported that the effect of spring burning on the subsequent density of common reed Phragmites australis depended on how wet the plots were. Statistical significance was not assessed. In a wetter area of the reedbed, the maximum annual reed density was typically lower in burned than unburned plots (three of four years, for which burned: 220–440 stems/m2; unburned: 310–920 stems/m2). In a drier area of the reedbed, the maximum annual reed density was typically higher in burned than unburned plots (three of four years, for which burned: 530–630 stems/m2; unburned: 230–270 stems/m2). Methods: In 1971, two pairs of plots were established in a young reedbed (sown in 1968). One pair was in a wetter area (flooded spring to autumn). One pair was in a drier area (water table 30–100 cm below surface). Between 1972 and 1975, one plot/pair was burned each spring (as early as possible). The other plots were not burned. All standing reed stems were counted between April and October each year (0.25–0.50 m2 quadrats; 4–6 quadrats/plot/sampling date). This study used the same experimental set-up as (1).Study and other actions tested
A study in 1985 in an ephemeral freshwater marsh in Florida, USA (Herndon et al. 1991) reported that following prescribed burning, the height of the dominant plant species increased. Over 50 days following burning, the average height of sawgrass Cladium jamaicense increased by 36 cm (or 0.76 cm/day). Methods: In August 1985, an area of sawgrass-dominated marsh was deliberately burned. Lighting fires at the end of the dry season, when there is no standing water, are a common natural disturbance. Remnant shoots after the burn were 13 cm tall on average, and remained “well above water immediately post-fire”. Sawgrass was monitored for 50 days, by measuring the distance between the soil surface and the tip of the tallest leaf on each of 30 random plants (culms).Study and other actions tested
A replicated, randomized, paired, controlled study in 1988 in a reedbed in England, UK (Cowie et al. 1992) reported that burned plots contained more, shorter and thicker reed stems than unburned plots after one growing season, and had higher plant richness but not diversity. After 3–5 months, burned plots contained a higher density of reed stems (736 stems/m2) than unburned plots (484 stems/m2). On average, reed stems were shorter but thicker in burned plots (143 cm tall; 3.7 mm diameter) than unburned plots (177 cm tall; 3.5 mm diameter). Burned plots also had higher plant species richness than unburned plots (data reported but units not clear) but statistically similar diversity (data reported as a diversity index). Of 17 common plant species for which data were reported, 16 were more frequent in burned plots than in unburned plots (statistical significance not assessed; see original paper for data). Methods: Five pairs of 30 x 40 m plots were established in a reedbed that had not been managed for ≥10 years. In March 1988, one random plot/pair was burned. The other plots were left unmanaged. All plots were flooded from April 1988. Vegetation was surveyed in July and August 1988. Live and dead reed stems were counted, and plant species were recorded, in 0.25-m2 quadrats (number not clear). Forty live reed stems/plot were measured. This summary takes some contextual and methodological details from Ditlhogo et al. (1992).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 burning reduced overall vegetation biomass, but had mixed effects on cover of the dominant plant species and no significant effect on plant species richness. One year after the latest burn, above-ground vegetation biomass was lower in burned areas (780–960 g/m2) than in unburned areas (920–2,080 g/m2). Burned and unburned areas had statistically similar cover of the dominant plant species in three of four comparisons. In the other comparison, amongst subplots fenced to exclude wild mammals, burned areas had lower cover of saltmeadow cordgrass Spartina patens (31%) than unburned areas (78%). Although plant species richness significantly increased in burned areas over two years of burning (by 1.8–2.4 species/m2), this change was not significantly different from the change in unburned areas (where species richness increased by 0–1.8 species/m2). Methods: Five pairs of 100-m2 plots were established in a freshwater marsh (regularly burned for at least 100 years). One random plot in each pair was burned in autumn 1992 and 1993. Each plot contained two 4-m2 subplots, one of which was fenced. Plant species and their cover were recorded in each subplot in autumn 1992 (before intervention) and 1994. Vegetation was cut from one 0.25-m2 quadrat/subplot, then dried and weighed, in autumn 1994.Study and other actions tested
A replicated, paired, site comparison study in 1994 of eight sedge meadows in Wisconsin, USA (Kost & De Steven 2000) reported that burning reduced plant species richness in the short term and increased it in the long term, but found that burning typically had no significant effect on vegetation abundance after 1–8 growing seasons. Burned meadows had lower species richness than unburned meadows in three of five comparisons (burned ≤3 growing seasons previously: 27–30; unburned: 32–39 species/7.5 m2) but higher species richness in two of five comparisons (burned ≥4 growing seasons previously: 32–42; unburned: 26–39 species/7.5 m2). Statistical significance of richness results was not assessed. In two of five comparisons, burned meadows had higher cover of grasses (burned: 13–24%; unburned: 4%) and sedges/rushes (burned: 100–110%; unburned: 75%). Otherwise, vegetation abundance (grass cover, sedge/rush cover, forb cover, total live above-ground biomass) did not significantly or consistently differ between burned and unburned meadows. For data on these outcomes and on the cover of individual plant species, see original paper. Methods: In summer 1994, vegetation was surveyed in eight sedge meadows: five last burned, in spring, 1–8 growing seasons previously; three not burned for >30 years. Plant species and cover were recorded along three 100-m-long transects/meadow. Live vegetation was cut from five 0.1-m2 plots/meadow, then dried and weighed.Study and other actions tested
A site comparison study in 1996–1997 of two reedbeds in Romania (Rolletschek et al. 2000) found that a burned reedbed contained fewer, shorter, thinner common reed common reed Phragmites australis shoots than an unburned reedbed, and a lower reed biomass. In the spring after intervention, the burned reedbed contained fewer reed shoots (40 shoots/m2) than the unburned reedbed (105 shoots/m2). Reed shoots in the burned reedbed were also shorter (burned: 150 cm; unburned: 194 cm) and thinner (burned: 9.9 mm; unburned: 12.9 mm). Accordingly, the peak above-ground biomass was lower in the burned reedbed (burned: 2,738 g/m2; unburned: 3,468 g/m2; statistical significance not assessed). Methods: In September 1996 (biomass) and May 1997 (all other metrics), vegetation was surveyed in two reedbeds with comparable nutrient levels: one burned in the previous winter, and one that had not been burned. The reedbeds were not flooded between burning and measurement. Surveys included measurements of 25 shoots/reedbed, and counts of shoots in five 1-m2 quadrats/reedbed.Study and other actions tested
A replicated, randomized, paired, controlled, before-and-after study in 1994–1997 in an ephemeral wet prairie in Oregon, USA (Clark & Wilson 2001) found that burning woody plants reduced their survival and cover and increased native forb abundance, but had no significant effect on overall vegetation or herb cover. After one year, the survival rate of woody plants was lower in burned (33%) than unburned plots (83%). Over three years, woody plant cover decreased in burned plots (by 63%) but increased in unburned plots (by 20%). Native forb cover increased in burned plots at the expense of non-native forbs (natives: 8% increase; non-natives: 77% decrease). The opposite was true in unburned plots (natives: 30% decrease; non-natives: 28% increase). However, burned and unburned plots experienced statistically similar changes in overall vegetation cover (increase; burned: 41%; unburned: 31%) and cover of the dominant herb species, tussock grass Deschampsia cespitosa (increase; burned: 31%; unburned: 31%; see original paper for data on other individual plant species). Methods: In 1994, five pairs of plots (each 64–160 m2) were established in a degraded, seasonally flooded prairie. Woody plants had grown over 200 years of fire suppression. In each pair, one random plot was burned in autumn 1994 and 1996. Vegetation was surveyed before (summer 1994) and after burning. Survival of six tagged woody plants/plot was recorded in summer 1995. Cover of selected herb species was recorded in three 0.5-m2 quadrats/plot in summer 1997.Study and other actions tested
A controlled, before-and-after study in 1994–1998 in a freshwater marsh in Florida, USA (Ponzio et al. 2004) reported that prescribed burning increased plant species richness and temporarily increased the density of one of two dominant species, but had no clear effect on the frequency of these two species. Unless specified, statistical significance was not assessed. Burned plots contained 6–9 plant species before burning, then 8–11 species over the four years after burning. Burning significantly but temporarily increased the density of southern cattail Typha domingensis (before: 2–3 stems/m2; after one to two years: 4–6 stems/m2; after three to four years: 2 stems/m2). Sawgrass Cladium jamaicense density was statistically similar before and after burning in seven of eight comparisons (for which before: 6–13 stems/m2; after: 6–15 stems/m2). Burning had no clear effect the frequency of southern cattail (before: in 93–100% of quadrats; after: 83–100%) or sawgrass (before: in 87–100% of quadrats; after: 80–100%). The frequency of eight other common plant species showed mixed responses to burning (see original paper). In unburned plots, metrics were generally stable (before they were affected by wildfire): 9 species/plot, 1–2 cattail stems/m2, 10–13 sawgrass stems/m2, cattail in 73–80% quadrats and sawgrass in 97–100% of quadrats. Methods: In June 1994, a 265-ha area of marsh was deliberately burned. Lighting fires are a common natural disturbance in similar marshes, but the study marsh had not burned for ≥5 years. The marsh was flooded when burned and for most of the time after burning. Vegetation was surveyed before burning (1994) and for up to four years after (burned: 1995–1998; unburned: 1995–1996), in two plots within the burned area and one adjacent unburned plot (thirty 2-m2 quadrats/plot).Study and other actions tested
A replicated, randomized, paired, controlled, before-and-after study in 1998–2000 of a range of marsh and wet meadow habitats around one lake in Idaho, USA (Austin et al. 2007) found that prescribed burning typically had no clear or significant effect on plant community composition or biomass. Over two years, the overall plant community composition within freshwater habitats remained similar in burned and unburned plots (data presented as graphical analyses; statistical significance of differences not assessed). In six of eight comparisons, changes in live, above-ground plant biomass (from before to after grazing) were not significantly different in burned plots (decrease of 200 g/m2 to non-significant increase of 20 g/m2) and unburned plots (decrease of 170 g/m2 to non-significant increase of 130 g/m2). Methods: Three pairs of fields with similar neighbouring vegetation were studied. Each field contained a range of freshwater habitats, from permanently flooded marshes to ephemeral wet meadows. All fields had been historically grazed and cut, but were undisturbed from 1996. In October 1998 (when vegetation was dormant) one random field in each pair was burned. Vegetation was surveyed in June–July before intervention (1998) and for two years after (1999, 2000).Study and other actions tested
A replicated, paired, controlled, before-and-after study in 1989–1991 in a riparian wet meadow in California, USA (McWilliams et al. 2007) found that prescribed burning changed the overall plant community composition and increased plant diversity, but had no significant effect on plant species richness or the proportion of native species. Over two years, burning had a significant effect on the overall plant community composition (reported as statistical model results). The effect of burning on the relative abundance of individual species depended on the year and community type, but burning generally reduced the relative abundance of the most common species (see original paper for data). Accordingly, plant diversity increased more in burned than unburned plots (data reported as a diversity index). However, plant species richness increased by a similar amount in burned plots (from 5–6 species/plot to 5–11 species/plot) and unburned plots (from 5–7 species/plot to 5–11 species/plot). Burning has no significant effect on the proportion of native plant species (data not reported). Methods: Eight plots (approximately 50 x 460 m) were established in a seasonally flooded wet meadow. The meadow was grazed by livestock until 1988, then managed for waterfowl without grazing. Four plots were burned in November 1990 and December 1991. Vegetation was surveyed along two transects/plot before (August–September 1990) and approximately nine months after each burn (September 1991 and 1992). Data were split by plant community type for analysis.Study and other actions tested
A replicated, randomized, controlled, before-and-after study in 2000–2005 aiming to restore ephemeral freshwater marshes within pine forest in Georgia, USA (Martin & Kirkman 2009) found that prescribed burning (along with killing trees by cutting and applying herbicide) altered the overall plant community composition, favouring herbaceous and wetland-characteristic species. Over five years, the community composition of managed wetlands diverged significantly from that of unmanaged wetlands (data reported as a graphical analysis). This effect was stronger in the core of the wetlands than on the wetland-upland boundary. Of 26 plant taxa whose frequency increased in managed wetlands (statistical significance not assessed), 25 were herbs and 15 were obligate wetland taxa. Methods: Between 2000 and 2005, five depressional wetlands were burned three times (once every two years, matching the historical fire regime). In summer 2000, mature stands of fire-resistant trees (oak Quercus spp.) had been removed by cutting and/or applying herbicide. The study does not distinguish between the effects of burning and tree removal. Five additional wetlands were not managed (no burning and trees not removed). Plant species presence/absence was recorded before (2000) and after (2005) intervention, in three to seven 100-m2 plots/wetland.Study and other actions tested
A replicated, before-and-after study of two degraded freshwater marshes in South Africa (Luvuno 2013) reported that prescribed burning reduced total plant diversity, but had mixed effects across sites on tree density and height. Unless specified, statistical significance was not assessed. In both sites, plant species diversity was lower five months after burning than just before (data reported as a diversity index). In the drier site (Z34), the overall tree density was significantly lower after burning than before. Density declined for 8 of 10 species (before: 1–23; after: 0–18 trees/species/0.25 ha). The average height of trees was statistically similar before (0–4 m) and after (0–5 m) burning. In the wetter site (Z49), burning had no significant effect on the overall tree density. Although density declined for 5 of 8 species (before: 2–8; after: 0–3 trees/species/0.25 ha), this was compensated for by increases in 2 of 8 species (before: 0–2; after: 1–12 trees/species/0.25 ha). The average height of trees was significantly lower after burning (0–1 m) than before (0–3 m). Mature trees (>2 m tall) were more likely to be killed in the wetter site, where ferns created taller flames. Methods: The two studied wetlands were within a forest plantation where natural fire was suppressed. As a result, woody vegetation was colonizing the wetlands. Vegetation was surveyed along four 50-m transects/wetland, before and five months after a prescribed burn (dates and methods not reported). Tree measurements included seedlings, saplings and mature trees.Study and other actions tested
A replicated, paired, controlled study in 2006–2009 in 40 freshwater marshes within a ranch in Florida, USA (Boughton et al. 2016) found that prescribed burning typically had no significant effect on the overall plant community composition, richness and diversity, but had mixed effects on vegetation quality. Statistical significance was assessed for all results, but data were generally not reported. After one and two summers, burned and unburned marshes contained similar overall plant communities (based on the species present and their abundance; data not reported). In four of six cases, burned and unburned marshes supported a similar relative abundance of forbs, grass-like plants and shrubs (with mixed effects depending on the group, year and grazing in the other two cases). Burned and unburned marshes also had similar overall plant species diversity and richness, and similar native plant species richness. After two summers, species in burned marshes were less characteristic of pristine Florida marshes, on average, than were the species in unburned marshes (data reported as a conservatism score). The effect of burning on this outcome after one summer was more complicated, differing between marshes and depending on whether they were grazed or not. Methods: The study used forty 0.5–1.5 ha marshes, grouped into five blocks of eight, within a 4,000-ha ranch that was historically managed with sporadic prescribed burns. In February 2008, twenty marshes (four marshes/block) were deliberately burned. The other 20 marshes (four marshes/block) were left unburned. In each block, two burned and two unburned marshes were also fenced to exclude cattle. Plant species presence/absence was recorded in October before (2006) and after (2008, 2009) burning, in fifteen 1-m2 quadrats/marsh.Study and other actions tested
A replicated, controlled study in 2010 in a permanent freshwater marsh in Florida, USA (Venne et al. 2016) found that burned plots had similar overall vegetation cover to unburned plots, but contained greater cover and biomass of surface-encrusting algae and contained shorter vegetation. Over 72 days following intervention, burned plots had statistically similar overall vegetation cover (25%) to unburned plots (41%). However, burned plots contained a greater abundance of surface-encrusting algae, both in terms of cover (burned: 27%; unburned: 21%) and biomass (burned: 51 g/m2; unburned: 42 g/m2). Finally, burned plots contained shorter vegetation, both in terms of average height (burned: 89 cm; unburned: 165 cm) and maximum height (burned: 104 cm; unburned: 200 cm). Methods: In early April 2010, a 690 ha area of marsh (dominated by sawgrass Cladium mariscus ssp. jamaicense) was burned. Standing water was present during the burn. Historically, this type of marsh was frequently disturbed by lightning fires. Vegetation and algae were surveyed every 10 days between 2 and 72 days after burning, in four 100-m2 plots in the burned area and four 100-m2 plots in a nearby unburned area. Algae were dried before weighing.Study and other actions tested