Sow grass seeds

Supporting evidence from individual studies

1. A replicated, randomized, controlled study in 1972–1976 in a sagebrush grassland affected by wildfire in Nevada, USA (Evans & Young 1978) found that sowing grass seeds did not alter the density of native grass or forbs but led to a decrease in the number of shrubs. After four years, the density of crested wheatgrass Agropyron desertorum and native forbs did not differ significantly between plots sown with grass seeds (wheatgrass: 0.35–0.70 plants/m²; forbs: 0.46–0.48 plants/m²) and unsown plots (wheatgrass: 0.35–1.04 plants/m²; forbs: 0.85–0.88 plants/m²). Total shrub density was lower in plots sown with grass seeds (74–84 shrubs/1,000 m²) than in unsown plots (116–119 shrubs/1,000 m²). In October 1972, eight 12 x 12 m plots were established. Four of the plots were sown with seeds of crested wheatgrass and intermediate wheatgrass Agropyron intermedium, while four plots were left unsown. The number of grass, forb and shrub plants was counted within each plot in 1973, 1974 and 1976.

   Study and other actions tested

   Referenced paper

   Evans R.A. & Young J.A. (1978) Effectiveness of rehabilitation practices following wildfire in a degraded big sagebrush-downy brome community. Journal of Range Management, 31, 185-188.

2. A replicated, controlled study in 1973–1977 in a grassland where powerlines had been constructed in Arizona, USA (Hessing & Johnson 1982) found that sowing a mixture of native and non-native grass seeds did not alter vegetation cover in most cases. In two of three years, there was no significant difference in vegetation cover between seeded (8–10%) and unseeded areas (10–12%). In one year, seeded areas had significantly higher vegetation cover than unseeded areas in the spring (53% vs 34%) but not in the summer (18% vs 16%). After construction was completed in summer 1973, two areas were sown with a seed mixture of eight grass species at a rate of 15.3 kg/ha, while two areas were not sown with seeds. Vegetation cover was surveyed using eight 6.1-m long line transects in each area in summer 1975, spring and summer 1976, and spring 1977.

   Study and other actions tested

   Referenced paper

   Hessing M.B. & Johnson C.D. (1982) Early secondary succession following restoration and reseeding treatments in Northern Arizona. Journal of Range Management, 35, 667-669.

3. A site comparison study in 1992 in a former arable field site in Kansas, USA (Kindscher & Tieszen 1998) found that sowing grass seeds resulted in species richness that was lower than that found in intact prairie sites. Plant species richness in the two sites where local seeds were sown was lower (9.2–12.0 species/quadrat) than in a nearby intact prairie (15.0 species/quadrat). In 1957, soil in the arable field was disturbed by disking and sown with Andropogon gerardii, Andropogon scoparius, Sorghastrum nutans, and Panicum virgatum seeds. In June 1992, vegetation cover was surveyed in ninety 1-m² quadrats in the site sown with seed, and 30 quadrats in an adjacent intact prairie.

   Study and other actions tested

   Referenced paper

   Kindscher K. & Tieszen L.L. (1998) Floristic and soil organic matter changes after five and thirty‐five years of native tallgrass prairie restoration. Restoration Ecology, 6, 181-169.

4. A replicated, randomized, paired, controlled study in 1993–1996 in ex-arable land in the UK (Manchester et al. 1999) found that sowing commercial grass seeds did not alter plant species richness. Plant species richness was the same in sown and unsown areas (28 vs 28 species). No statistical analyses were carried out in this study. A commercial seed mix containing four grass species was sown in 10 plots (size of plots unclear), while no seed was sown in 10 plots. The survey methods used to assess species richness in this study were not clear.

   Study and other actions tested

   Referenced paper

   Manchester S.J., McNally S., Treweek J.R., Sparks T.H. & Mountford J.O. (1999) The cost and practicality of techniques for the reversion of arable land to lowland wet grassland - an experimental study and review. Journal of Environmental Management, 55, 91-109.

5. A paired, site comparison study in 2004 in 40 restored and 40 ancient grasslands in southern England, UK (Fagan et al. 2008) found that sowing grass seeds resulted in plant communities that were different from those of ancient grasslands or sites where natural regeneration was allowed. The plant community of restoration sites where grass seeds were sown was different from the plant communities of paired ancient grasslands, as well as the plant communities of restoration sites where both grass and forb seeds were sown, or where natural regeneration was allowed (results presented as graphical analysis). Between one and >20 years prior to the study, 40 ex-arable sites were seeded with grass seeds only (12 sites), grass and forb seeds (12 sites) or no seeds (natural regeneration; 16 sites). Each site was paired with the closest ancient grassland (0–9 km away; aged >200 years). All sites were grazed, and occasionally mown. In June–August 2004, the cover of plant species was estimated within 50 x 50 cm quadrats placed at 10 m intervals along a 100-m transect at each site.

   Study and other actions tested

   Referenced paper

   Fagan K.C., Pywell R.F., Bullock J.M. & Marrs R.H. (2008) Do restored calcareous grasslands on former arable fields resemble ancient targets? The effect of time, methods and environment on outcomes. Journal of Applied Ecology, 45, 1293-1303.

6. A replicated, site comparison study in 2004 in 14 sites disturbed by pipeline construction in Alberta, Canada (Desserud et al. 2010) found that areas where grass seeds had been sown had lower cover of native plants, shrubs, grasses and forbs than nearby natural grasslands, but had higher cover of non-native plants. Areas where seeds had been sown after disturbance by pipeline construction had lower cover of native plants (39%) than nearby natural grasslands (71%). There was a similar pattern for native shrubs (seeded: 2%, natural: 7%), native grasses (seeded: 27%, natural: 35%) and native forbs (seeded: 11%, natural: 29%). However, non-native species cover was higher in seeded areas (32%) than in natural grasslands (21%). Fourteen sites where pipelines had been constructed 7–40 years prior were selected for the study. All sites were sown with native grass seeds, including rough fescue Festuca campestris, after construction. In June–August 2004, plant cover was surveyed using 10 quadrats along one 30 x 1 m transect in the reseeded area of each site and one transect in adjacent intact grasslands.

   Study and other actions tested

   Referenced paper

   Desserud P., Gates C.C., Adams B. & Revel R.D. (2010) Restoration of foothills rough fescue grassland following pipeline disturbance in southwestern Alberta. Journal of Environmental Management, 91, 2763-2770.

7. A replicated, controlled study in 2004–2006 in a former arable field in Gauteng province, South Africa (Oudtshoorn et al. 2011) found that sowing grass seeds increased the abundance of sown species and reduced the abundance of unsown species. After two years, in two comparisons the density of sown species was higher in areas where soil was disturbed and grass seeds were sown (35–80 plants/m²) than in areas where soil was disturbed but no grass seeds were sown (2 plants/m²). The opposite was true for plants whose seed was not sown (disturbed and sown: 51–122 plants/m²; disturbed and unsown 250 plants/m²). In 2004, eight 20 × 10 m plots were ploughed and sown with seeds of five grass species. Eight plots were disturbed with a single toothed ripper and sown with seeds of five grass species, and four plots were ploughed or disturbed with a ripper and not sown with seeds. In March 2005 and 2006, eight 1-m² quadrats were placed in each plot and the number of plants counted.

   Study and other actions tested

   Referenced paper

   Van Oudtshoorn F., Brown L. & Kellner K. (2011) The effect of reseeding methods on secondary succession during cropland restoration in the Highveld region of South Africa. African Journal of Range & Forage Science, 28, 1-8.

8. A replicated, controlled study in 1999–2005 in a formerly overgrazed area in Iceland (Petursdottir et al. 2013) found that sowing grass seeds increased vegetation cover, but not plant species richness. Vegetation cover was higher in areas where seeds were sown (72–91%) than in areas where they were not (6%). Plant species richness did not differ significantly between areas that were seeded and areas that were not seeded (no data reported). In 1999, some areas of the site were sown with red fescue Festuca rubra and Kentucky bluegrass Poa pratensis seeds, while other areas were not sown with seeds (levels of replication were unclear from the study). In August–September 2005, five 10 × 10 m plots were established in seeded areas and five plots were established in unseeded areas. Vegetation cover in ten 0.5 x 0.5 m quadrats within each plot was surveyed.

   Study and other actions tested

   Referenced paper

   Petursdottir T., Aradottir A.L. & Benediktsson K. (2013) Evaluation of the short‐term progress of restoration combining ecological assessment and public perception. Restoration Ecology, 21, 75-85.

9. A replicated, randomized, controlled study in 2006–2011 in arid rangelands in Arizona, USA (Bernstein et al. 2014) found that sowing grass seeds did not alter the density of the sown grasses Indian ricegrass Achnatherum hymenoides and needle-and-thread grass Hesperostipa comata. After five years, there was no significant difference in the average density of Indian ricegrass and needle-and-thread grass between plots where their seeds were sown (0.03 plants/m²) and plots where no seeds were sown (0.01 plants/m²). In November 2006, twenty 3 × 3 m plots were sown with native C3 grass seeds and ten plots were not. Counts of grass species were made in all plots in May 2007, 2010 and 2011.

   Study and other actions tested

   Referenced paper

   Bernstein E.J., Albano C.M., Sisk T.D., Crews T.E. & Rosenstock S. (2014) Establishing cool-season grasses on a degraded arid rangeland of the Colorado plateau. Restoration Ecology, 22, 57-64.

10. A replicated, randomized, paired, controlled study in 2005–2012 in two cleared sites in Oregon, USA (Busby & Southworth 2014) found that sowing grass seeds initially increased grass cover, but after seven years there was no difference in grass cover between sown and unsown areas. After two years, grass cover was higher in plots where grass seeds were sown (9.7%) than in plots where they were not sown (0.2%). However, after seven years there was no significant difference in the grass cover in sown (1.9%) and unsown plots (0.4%). In 2001–2002, woody vegetation was removed using a masticator, and in 2005, prescribed burning was carried out. Two weeks after burning, native grass seeds of four species were sown in 15 randomly located 1-m² plots at each site, while no seeds were sown in another 15 plots. Grass cover was estimated in each plot in 2006, 2007 and 2012.

    Study and other actions tested

    Referenced paper

    Busby L. & Southworth D. (2014) Minimal persistence of native bunchgrasses seven years after seeding, following mastication and prescribed fire in southwestern Oregon, USA. Fire Ecology, 10, 63-71.

11. A site comparison study in 2010 on 10 road verges in the Dovre Mountains, Norway (Hagen et al. 2014) found that sowing grass seeds increased vegetation cover overall but reduced native vegetation cover. Total vegetation cover was on average higher in areas where a commercial grass seed mixture was sown (85–96%) than in areas where no seeds were sown (65–90%). The opposite was true for native vegetation cover (seeded: 48–72%; unseeded: 65–74%). In 1989, ten road verges were sown with a commercial seed mixture containing four non-native grass species at a rate of 7 kg/1,000 m². Commercial fertilizer was also added at a rate of 50 kg/1,000 m². Ten other areas on each of the 10 road verges were not seeded. All sites were grazed occasionally. In July and August 2010, the abundance of plant species was recorded in five 0.5 × 0.5 m plots in each of the seeded and unseeded areas.

    Study and other actions tested

    Referenced paper

    Hagen D., Hansen T.I., Graae B.J. & Rydgren K. (2014) To seed or not to seed in alpine restoration: introduced grass species outcompete rather than facilitate native species. Ecological Engineering, 64, 255-261.

12. A replicated, controlled study in 2012–2013 in a serpentine grassland in California, USA (Funk et al. 2015) found that sowing grass seeds did not alter native plant species richness or the cover of native or non-native invasive plant species. Average numbers of native plant species did not differ significantly between plots sown with grass seeds and plots not sown with seeds (both 9–10 species/plot). The same was true for the cover of native plants (sown: 63–77%; unsown: 54–71%) and the cover of non-native invasive plants (sown: 20–29%; unsown: 18–26%). In November 2012, twenty 1 x 1 m plots were sown with 20 seeds of three native grass species (collected onsite 2–3 months prior), while 20 plots were left unsown. All plots were irrigated for 21 days in August 2012. Half of the plots for each treatment were grazed by cattle. Vegetation cover was estimated in March and April 2013 using a 0.25 × 0.25 m quadrat placed in each plot.

    Study and other actions tested

    Referenced paper

    Funk J.L., Hoffacker M.K. & Matzek V. (2015) Summer irrigation, grazing and seed addition differentially influence community composition in an invaded serpentine grassland. Restoration Ecology, 23, 122-130.

13. A replicated, randomized, paired, controlled study in 2008–2014 in a pine plantation burnt in a wildfire in Hungary (Szitár et al. 2016) found that sowing with grass seeds reduced plant species richness and cover of specialist grassland species, but the plant communities of both seeded and unseeded areas became more similar to that of intact grasslands over time. After six years, plant species richness of areas where grass seeds were sown was lower (3.4 species/m²) than that in areas where no seeds were sown (7.0 species/m²). Cover of grassland species showed a similar pattern (seeded: 5%, unseeded: 16%). Over six years, the plant community both in areas where seeds were sown and areas where seeds were not sown became more similar to that of intact grasslands (data presented as graphical analysis). These results are not based on tests of statistical significance. In autumn 2008, seeds of two grass species Festuca vaginata and Stipa borysthenica were sown, at a rate of 1500 seeds/m² and 100 seeds/m² respectively, in twenty 1 x 1 m plots, while in 20 plots no seeds were sown. Cover of all plants was estimated yearly in June between 2008 and 2014. In June 2014, ten intact grassland areas that had also been burnt were surveyed.

    Study and other actions tested

    Referenced paper

    Szitár K., Ónodi G., Somay L., Pándi I., Kucs P. & Kröel-Dulay G. (2016) Contrasting effects of land use legacies on grassland restoration in burnt pine plantations. Biological Conservation, 201, 356-362.