RESEARCH ARTICLE Forb Species Establishment Increases with Decreased Grass Seeding Density and with Increased Forb Seeding Density in a Northeast Kansas, U.S.A., Experimental Prairie Restoration Timothy L. Dickson1,2,3 and William H. Busby4 Abstract dominant grasses. The results also suggest that seed avail- Most prairie restorations fail to produce the diversity of ability limits the establishment of some forbs. Forb diver- species found in unplowed remnants. This lack of restored sity can therefore be increased by decreasing grass diversity is hypothesized to be partly due to the inhibition seeding density, by increasing forb seeding density, or of forb species by high seeding densities of dominant both. However, forb seeds are generally expensive, and grasses and partly due to the low seeding densities of forbs increasing forb seeding density across the entire area of used in many restorations. We tested this hypothesis by a restoration may be prohibitively expensive. We there- sowing various densities of forb and warm-season grass fore recommend a low seeding density of dominant seeds into a restoration begun on bare soil. This is the first grasses, and we recommend spatially separating forbs replicated restoration experiment we are aware of that from dominant grasses by adding most forb seeds to areas varies grass seeding densities to examine the effects on with little to no dominant grasses and by adding the rest forbs. Four years after seeding, we found that higher den- of the forb seeds to areas with a low density of dominant sities of grass seeds decreased forb cover, biomass, and grasses. richness, and higher densities of forb seeds increased forb richness. These results suggest that dominant grasses Key words: conservative forb species richness, Conserva- compete strongly with native forb species and that many tion Reserve Program, grassland, optimal seed mix, weedy forb species thrive when they are spatially separated from species. Introduction sity of forbs (Seabloom et al. 2003; Sheley & Half 2006) or Prairie restorations generally fail to produce the diversity to the inhibitory effects of seeding high densities of domi- of plant species found in prairie remnants (Weber 1999; nant grasses (Weber 1999; Kindscher & Fraser 2000). Polley et al. 2005; but see Pfeiffer 1999; Whitney 1999). ‘‘Conservative’’ species are also generally absent from Restorations in other parts of the world also generally restorations. Conservative species are rarely found outside fail to produce plant diversity similar to remnant levels undisturbed remnant plant communities and can be (Lockwood & Pimm 1999; Walker et al. 2004). This lack thought of as late successional (Swink & Wilhelm 1994). of restored diversity is largely due to a lack of establish- These species are hypothesized to be inhibited by high ment from forbs (broadleaf herbs). Forbs are the most densities of dominant grasses and other aggressive species diverse group of plant species in the tallgrass/mixed-grass (Packard 1997). prairie. They comprise at least 60% of the plant species in Although it is recognized that high densities of domi- remnant areas of Konza Prairie in northeastern Kansas nant grasses may hinder the establishment of forbs (Collins et al. 1998) and the proportion increases to the (Kindscher & Fraser 2000) and conservative species east of Konza. Comparatively low forb establishment in (Weber 1999), there are several reasons many prairie restorations may be due to low seeding richness and den- restorationists continue to seed dominant grasses at high densities. Restorationists have been influenced by the agricultural practice of planting high densities of domin- 1 Department of Ecology and Evolutionary Biology, The University of Kansas, Lawrence, KS 66045, U.S.A. ant grasses as a way to reduce soil erosion (Boyd 1942; 2 Present address: Department of Ecology, Evolution, and Organismal Biology, Jelinski & Kulakow 1996). Planting high densities of dom- Iowa State University, Ames, IA 50011, U.S.A. 3 Address correspondence to T. L. Dickson, email [email protected] inant grasses is championed in restoration by the ‘‘prairie 4 Kansas Biological Survey, The University of Kansas, Lawrence, KS 66047, matrix’’ concept (Betz 1986; Betz et al. 1999) and other U.S.A. planting strategies (Wilson 1970; Schramm 1992) as Ó 2007 Society for Ecological Restoration International a method to control weedy species and improve soil char- doi: 10.1111/j.1526-100X.2008.00427.x acteristics. A high density of grass seed also costs very SEPTEMBER 2009 Restoration Ecology Vol. 17, No. 5, pp. 597–605 597 little because grass seed is generally much less expensive Program (CRP; Jelinski & Kulakow 1996) and seeded to than forb seed. native species in 2001, several days before the experimen- Few explicit tests of the effects of dominant prairie tal area received any seeds. Management during the study grasses on forb species have been completed, but period consisted of mowing the entire field (including the Kindscher and Fraser (2000) report that areas sown with- experiment) to approximately 17.8 cm height in July 2001 out grass seeds contain 84.6% forb cover (both sown and and June 2002 and conducting a controlled burn in March nonsown species) and higher species richness than areas 2003. Soil at the site was Sibleyville Loam with 3–7° slopes sown with grass seeds (18.2% forb cover). Also, plant spe- (Dickey et al. 1977). The soil was derived from sandstone cies diversity appears to stay the same or slightly decline and loamy shale and has experienced some erosion. The over time in sites initially sown with high densities of grass climate of the region was humid continental, with mean and dominant prairie species (Sluis 1999). In other studies, annual temperature of 12.7°C and mean annual precipita- forbs do not establish well when seeded into intact stands tion of 952 mm. Seasonal temperatures can be extreme, of grass (Williams et al. 2007), even though transplants have with minimum temperatures in winter as low as 229°C moderate establishment success (Warkins & Howell 1983; and maximum temperatures in summer as high as 43°C. Brown & Bugg 2001). Previous studies do not examine The seasonal distribution of precipitation was unimodal, whether low densities of grass allow greater forb establish- peaking in June (Kettle & Whittemore 1991). According ment than high densities of grass. Also, in the study most to the U.S. drought monitor, there were no drought condi- similar to the experiment presented here (Kindscher & tions during the growing seasons of 2001 and 2004, but Fraser 2000), the no grass treatment is not truly replicated there were abnormally dry to severe drought conditions because it is on a strip of land that accidentally does not from July to October 2002 and abnormally dry to extreme receive grass seeds. drought conditions from July to October 2003 (NDMC, Low seeding densities of forb seeds may also decrease USDA, and NOAA 2008). forb abundance and richness (Seabloom et al. 2003; Sheley & Half 2006). Many of the most difficult forb spe- cies to establish from seed also have very expensive seed. Field Methods As a result, these species are often seeded at low density or All seeds in the experiment were sown at densities relative not at all. Increasing the seeding density of these forb spe- to the densities of purchased seeds in the surrounding cies may therefore increase their abundance and richness. CRP planting (ambient density). Grass seeds (0.53 ambi- Although we focus on forb species establishment, we ent density; 23 ambient density) and forb seeds (23 ambi- also examine whether seeding density affects weed abun- ent density; 83 ambient density) were sown on bare soil dance (Betz 1986; Schramm 1992) and whether seeding in a factorial design. We also added one treatment with no density affects bare ground. Competition from certain grass and 83 ambient forb seed density. We hereafter weeds can dramatically reduce prairie establishment refer to 0.53 grass density and 23 forb density as low den- (Blumenthal et al. 2003), and increased bare ground can sity, and we refer to 23 grass density and 83 forb density increase soil erosion (Ellison 1950) and the spread of as high density. Low density of grass seed was 0.08 g/m2 of weeds (Bergelson et al. 1993) while also aiding the estab- Tripsacum dactyloides and 0.42 g/m2 of all other grasses lishment of some seeded species (Gross 1984). combined (Table 1; 1 g/m2 ¼ 8.92 pounds per acre). Low We manipulate both grass and forb seeding density, but density of forb seed was 0.18 g/m2 of Dalea purpurea and not seeding richness, in our experiment. The objective for 0.42 g/m2 of all other forbs combined. The low and high the study is to find the level of grass and forb seeding densities bracket Diboll’s (1997) recommended seeding density that will maximize forb abundance and richness. rates for tall-prairie plantings. Low densities were approx- Our hypothesis is that higher grass seeding density will imately half the recommended weight of grasses and of decrease forb establishment, and higher forb seeding den- forbs, whereas high densities were approximately twice sity will increase forb establishment only at low grass sow- the recommended weight. The number of seeds added at ing density (forb 3 grass sowing density interaction). We low density was lower for grasses than for forbs, with ap- also examine whether abundant forb species are decreas- proximately 1 T. dactyloides seed/m2 and 150 seeds/m2 of ing the percentage cover of other forb species. all other grasses combined, and 96 D. purpurea seeds/m2 and 184 seeds/m2 of all other forbs combined. The number of seeds was estimated from the seed weights listed in Methods the 2007 Prairie Moon Nursery catalog (Winona, MN, U.S.A.).