Abstract
Precision Prairie Reconstruction (PPR) is a site-specific application for enhancing native species richness (specifically native forbs) in old fields dominated by exotic grass species. The method consists of establishing native grasses and native forbs in micro-communities in an existing grass matrix. The micro-communities are constructed by broadcast seeding small-scale disturbances (8 m2 in size) installed over a given percentage of the site (5%, 25%, or 50%). Here we compared results from PPR treatments with two standard restoration techniques: herbicide application/drill-seeding (H&D) and rototill/broadcast seeding (R&B). In year 15 of the experiment, the 25% PPR had higher native forb density (27 plants m−2) and seeded forb density (16 plants m−2) than the other treatments with their corresponding averages of 10 plants m−2 and 4 plants m−2 respectively. Implementation cost of the 25% PPR treatment was approximately 1/3 of the traditional treatment costs in terms of seed purchase, cultivation, and seeding. Average results across 15 years showed the 25% PPR and R&B treatments were statistically similar in terms of total species richness (22 species), seeded forb richness (8 species), and seeded-forb-richness stability, measured by coefficient of variation (CV of 0.4, where high CV equals low stability). The other treatments were significantly different with corresponding averages: total species richness of 16, seeded forb richness of 3, CV of 0.97. After a 15-year field experiment, we are confident that the PPR technique offers an innovative, cost effective approach for increasing native forb diversity and stability in old fields dominated by exotic grass species.
- exotic grass matrix
- forb diversity
- precision prairie reconstruction (PPR)
- seed islands
- small-scale disturbances
Restoration Recap
Precision Prairie Reconstruction (PPR) consists of broadcast seeding native grasses and forbs in small-scale disturbances (8 m2) installed over a specified percentage (5%, 25%, or 50%) of the vegetation matrix.
The hypothesis was that the seeded patches would remain stable sources of propagules for dispersal, colonizing the surrounding vegetation matrix.
Fifteen years of data has shown that the 25% treatment was as effective as rototilling and broadcast seeding (R…B) the entire site for increasing total species richness, native seeded forb richness, and native seeded grass frequency in a site dominated by Bromus inermis Leyss.(smooth brome) and Poa pratensis L. (Kentucky bluegrass).
In year 15, however, the total density of native forbs and native seeded forbs was significantly higher in the 25% PPR treatment (27 and 16 plants m-2) than the average of the other treatments (10 and 4 plants m−2). The 25% PPR treatment was shown to be superior to the 5% and 50% PPR treatments and superior to the herbicide/drill (H…D) treatment.
The expense associated with implementing the 25% PPR treatment, including seed purchase, cultivation and seeding costs, was approximately 1/3 the cost of the R…B treatment.
In 2000, we installed an experiment based on field observation and data gathered on the work of a prairie ecosystem engineer, the plains pocket gopher (Geomys bursarius [Grygiel et al. 2009, C. Grygiel, personal observation and unpublished data]). Ecosystem engineers have been identified as organisms that modify or create habitats (Jones et al 1994, Wright et al. 2004, Byers et al. 2006). One pocket gopher can overturn prairie soil at the rate of one to three mounds per day, bringing 2.3 metric tons per ha of sub-soil to the surface annually. Most of their activity is in the spring and autumn (Andelt and Case 2007). The disturbances caused by pocket gophers and their impacts on vegetation diversity have been extensively studied with results showing positive, negative, and neutral effects (Huntly and Reichman 1994, Steuter et al. 1995, Olff and Ritchie 1998, Rogers and Hartnett 2001).
Our objective in this experiment was to determine whether seeded small-scale disturbances could be used to increase native species richness and forb density on sites dominated by introduced perennial grasses. The working hypothesis was that by installing and seeding specific quantities of small-scale disturbances with native species, a self-sustaining community would be established that would generate a persistent source of propagules which would, in time, colonize the surrounding vegetation matrix (by taking advantage of favorable conditions that can occur over the years), and thus increase native species diversity. Data for the first five years (Grygiel et al. 2009) showed the following: 1) Strategically placed, broadcast-seeded, small-scale disturbances (2.84 m x 2.84 m in size) covering 25% of an area were as effective as rototilling and broadcast seeding for increasing total species richness, native seeded forb richness, native seeded forb density, and native seeded grass frequency in a field dominated by the introduced grasses, Bromus inermis and Poa pratensis; 2) The 25% treatment was superior to the 5% and 50% small-scale disturbance treatments as well as the herbicide/drill treatment of the entire field; 3) The cost of implementing (seed, cultivation, and seeding) the 25% treatment was approximately 1/3 of the R&B treatment; and 4) Longer term results that include a prescribed burn cycle were needed to fully evaluate the results.
Since our original study (Grygiel et al. 2009), and subsequent complementary ones (Grygiel et al. 2012, 2014), a series of more recent papers have explored some of the aspects of maintaining species diversity and the role of disturbances, but none have produced long-term data sets. Corbin and Holl (2012) have shown that planting patches to restore deforested habitats into heterogeneous canopies with a diverse community composition appear to be highly effective and less expensive than plantation designs. Valko et al. (2016) showed in a two-year study that creating gaps and sowing diverse seed mixtures were able to overcome microsite and propagule limitation, successfully introducing target species into the species-poor grasslands. Seahra et al. (2016) found that structured seeding approaches may be used over traditional ones to control species dominance and preserve seeded species diversity within grassland systems. Rinella et al. (2012) found that since short-term data are not predictive of long-term seeded species performance, additional long-term data are needed to identify effective practices, traits, and species for revegetating grasslands invaded by introduced species.
This paper presents results after 15 years of the ongoing experiment (at five-year intervals) that included a standard prescribed burn. We present: 1) data on total species richness (native and non-native), native forb richness, and native forb density across the entire field; and 2) a comparison of the stability over multiple years of the richness of the seeded native forb community.
Methods
Study Site
The study was conducted in an old field (previously farmed and then planted with B. inermis and grazed) on The Nature Conservancy’s (TNC) Bluestem Prairie Scientific and Natural Area near Glyndon, Minnesota (46°49’49.096” N, 96°27’21.017” W). The regional native vegetation is northern tallgrass prairie (Barker and Whitman 1988), but the site was dominated by B. inermis and P. pratensis and a few native and exotic forbs: Lithospermum incisum (narrowleaf stoneseed), Melilotus officinalis (sweet clover), Physalis longifolia (longleaf groundcherry), Rosa arkansana (prairie rose), and Symphyotrichum ericoides (white heath aster). For details regarding the site history, climate, soil characteristics, vegetation, and prior site management please see Grygiel et al. (2009). The study blocks were installed in 2000 (Grygiel et al. 2009). In 2007 the site was subjected to a spring burn following the prescribed burning regime for the area.
Treatments
The study was organized as a randomized block design with five treatments, a control, and five replicates of each treatment (Figure 1). Each block consisted of six 44-m x 44-m plots with a 6-m buffer zone surrounding each plot, and a 6-m buffer zone between blocks. Treatments and controls were randomly assigned to plots within each block. Treatments consisted of three levels of small-scale disturbances that were individually rototilled/broadcast seeded, comprising 5% PPR, 25% PPR, and 50% PPR of the plots; a conventional restoration treatment consisting of rototilling/broadcast seeding (R&B) of the entire plot; a conventional restoration treatment consisting of herbicide application/drill seeding (H&D) of the entire plot; the untreated control (C).
Experiment layout of the Precision Prairie Reconstruction (PPR) experimental site on The Nature Conservancy’s Bluestem Prairie Scientific and Natural Area (Glyndon, Minnesota). R&B = rototilling/broadcast seeding of the entire plot. H&D = a conventional restoration treatment consisting of herbicide application/drill seeding of the entire plot.
Individual small-scale disturbances (SSDs) were 2.84 m x 2.84 m in size. The 5% PPR treatment consisted of three equidistant transects (11 m apart) and four equidistant SSDs per transect located at 8.8 m from center to center for a total of 12 SSDs. The 25% PPR treatment consisted of10 equidistant transects (4 m apart) and six equidistant SSDs per transect located at 6.3 m from center to center for a total of 60 SSDs. The 50% PPR treatment consisted of 15 equidistant transects (2.84 m apart) with eight SSDs per transect (4.9 m apart) for a total of 120 SSDs. In the 50% PPR treatment the SSDs were arranged in a checkerboard pattern along the transects so as not to have side-by-side SSDs from adjacent transects. The objective was to facilitate the SSD installation by a tractor-mounted rototiller and to make the method applicable to larger installations where patterns can be established using GPS coordinates. For details regarding the on-site implementation of all the treatments please see Grygiel et al. (2009).
The seed mixture of native grasses and forbs common to the native vegetation of the area was comprised of seed purchased from Prairie Restoration, Inc. (Hawley, MN) and seed hand-collected by the researchers. All seeds used in this experiment were collected from within 80 km of the study site in accordance with TNC requirements (Table 1). Species were seeded at the rate of 27 kg/ha in the 5% PPR, 25% PPR, 50% PPR, and R&B treatments and 13.5 kg/ha in the H&D treatment which were the standard practices for broadcast seeding and drill seeding rates in the area. The study plots were seeded during the third week of October 2000. Fall planting was used because our prior experience in a large-scale restoration and biodiversity experiment in the Sheyenne National Grasslands had shown that in this area fall planting is, in general, more effective than the commonly used spring and summer planting (Biondini 2007). No herbicide, with the exception of the H&D treatment, was used in this phase of the study.
Species list of native grasses and forbs used in the seeding treatments on the Precision Prairie Reconstruction (PPR) experimental site on The Nature Conservancy’s Bluestem Prairie Scientific and Natural Area (Glyndon, Minnesota). The survivability is the average percent survival over 15 years across all treatments in the colonized vegetation matrix (Control treatment excluded). Species nomenclature follows: USDA, NRCS. 2010.
Vegetation Sampling and Data Analysis
All treatments were sampled yearly from 2001 to 2005. Following the 2007 prescribed spring burn the treatments were sampled in 2010 and again in 2016. The sampling was conducted in mid-July of each year using 20 nested quadrats (0.1 m2 nested inside a 0.25 m2) randomly distributed within the entire treatment plots (4 random transects, 5 random quadrats per transect). This approach was used so over time a representative sample of the experimental plots was collected. The entire 0.25-m2 quadrat was sampled for forb density by species while the 0.1-m2 quadrat was sampled for the presence of grasses also by species. While not recorded, over the 15 years the number of quadrats landing in the matrix was likely inversely proportional to the seeded areas.
Statistical Analysis
Species richness was estimated as the average number of species taxa per treatment. Differences in species richness (total and seeded forbs), seeded forb density, and average seeded grass frequency among treatments were analyzed using a randomized block analysis of variance (RAOV) with a repeated observation model (years being the repeated observation factor [Winer 1962]). In accordance with Tilman (1996), the coefficient of variation (CV) in seeded forb richness was used over the 15-year period of the study (CV = SD/Mean) as a measurement of seeded forb community stability: the higher the CV the lower the stability.
Results
There were significant differences among treatments in total species richness (RAOV; F5,24 = 12.9, p <0.001) but no overall year (F2,48 = 2.2, p >0.12), or treatment*year interactions (RAOV; F10,48 = 1.9, p >0.07). Across all sampling years, the 25% PPR and R&B treatments were statistically similar in terms of total species richness (an average of 22 species) but higher than the 50% PPR, H&D, and 5% PPR treatments, which averaged 18 species (Figure 2A). The control had the lowest species richness with an average of 12 species.
Species richness for years 5, 10, and 15 of all seeded and non-seeded grasses and forbs by treatments. B) Species richness for years 5, 10, and 15 of seeded native forbs by treatments. Vertical bars represent 95% CI. For statistical differences check the results section.
Seeded forb richness showed a more complex pattern with statistical differences for treatment (RAOV; F5,24 = 23.9, p <0.001), year (RAOV; F2,48 = 35, p <0.001), and a treatment*year interaction (RAOV; F10,48 = 6.6, p <0.05). Seeded forb species richness across all sampling years was higher in the 25% PPR and R&B treatments (an average of seven species) than in the 50% PPR, H&D, and 5% PPR treatments, which averaged four species (Figure 2B). The control was the lowest with only one seeded forb species. Overall seeded forb species richness doubled from an average of three species after five years to six species after 15 years (Figure 2B).
Fifteen years after the experiment was started, the density of native forbs was higher (RAOV; F5,20 = 10.3, p <0.001) in the 25% PPR (27 plants m-2) than in all the other treatments, which averaged 10 plants m-2 (Figure 3A). The non-seeded density component of native forbs, however, was lower (RAOV; Fs,20 = 3,p <0.05) in the H&D treatment (three plants m-2) than in all the other treatments, which averaged eight plants m-2 (Figure 3A).
A) Total native forb density, and total non-seeded native forb density in year 15 by treatments. B) Total density of seeded forbs for years 5, 10, and 15 by treatments. Vertical bars represent 95% CI. For statistical differences check the results section.
A summary of seeded forb density over the years is shown in Figure 3B. There were overall differences in treatment effect (RAOV; F5,24 = 17.3, p <0.001), year of sampling (RAOV; F2,48 = 18.1, p <0.001), and treatment*year interaction (RAOV; Fio,48 = 6.8,p <0.001). Across all years, seeded forb density was higher in the 25% PPR and R&B treatments, averaging nine plants m-2, than in the 50% PPR, 5% PPR, and H&D treatments, which averaged three plants.m-2 (Figure 3B). The Control treatment was the lowest with 1 plant.m-2. Overall, seeded forb density across all treatments tripled from two plants m-2, five years into the experiment, to six plants m-2 in year 15 (Figure 3B). In year 15, however, the total density of seeded forbs was significantly higher in the 25% PPR treatment (16 plants m-2) than the average of the other treatments (four plants m-2 [Figure 3B]). Table 2 shows a distribution of seeded forb density and seeded grass frequency by species at the end of 15 years. Seeded grasses responded more favorably under the R&B treatment (average 0.25) and the H&D treatment (average 0.21) than in the 25% PPR and 50% PPR treatments, which averaged 0.17 (Table 2). The 5% PPR and Control had almost no seeded grasses after 15 years. We found a significant difference in variability in seeded forb richness among treatments as calculated by the CV over the 15 year period (F5,20 = 9.3,p <0.001). The R&B, 25% PPR, and 50% PPR treatments had the lowest variability with an average CV of 0.4 (Figure 4) while the 5% PPR and control had the highest with an average CV of 1.13 (Figure 4).
Average forb density (plants/m2) and grass frequency for the seeded species plus B. inermis and P. pratensis after 15 years for each treatment: small-scale-disturbances (SSD), control (C), herbicide and seed drilling (H&D), and rototilling and broadcast seeding (R&B). Statistics were calculated for total seeded forb density and average grass frequency. Treatments with different letters are different at p <0.05. Note: we list only densities or frequencies greater than 0.05.
Coefficient of variation for seeded forb richness over the 15 years of the study by treatments. Vertical bars represent 95% CI. For statistical differences, check the results section.
Discussion
Our study tested a novel methodology for increasing native forb species richness in an exotic grass matrix. We found, after 15 years, similar patterns to those we reported after the first five years of the experiment (Grygiel et al. 2009). The difference after 15 years, however, was that: 1) the total native forb density was significantly higher in the 25% PPR treatment than in the other treatments (27 vs. an average of 10 plants m-2); and 2) a similar pattern was observed for the seeded forb density (16 vs. an average of four plants m-2). Seeded forb community stability (low CV) remained higher in the 25% PPR and R&B treatments (Figure 4) and is most likely related to higher forb richness (Table 2, Figure 2B), since there is an extensive literature showing an inverse relationship between community level CV and species richness (see for example Tilman 1996, Biondini 2007).
As stated, the hypothesis for this experiment was that installing and seeding specific quantities of small-scale-disturbances with native species will generate, through time, a persistent source of propagules that will colonize the surrounding vegetation matrix. To explore the colonization aspect in further detail we established a separate experiment (Grygiel et al. 2014) to investigate seeded forb colonization over a five-year period on undisturbed areas of the matrix adjacent to the seeded site (1 m) and in disturbed but non-seeded patches located 3 m away. This study (Grygiel et al. 2014) showed that colonization of disturbed patches can occur at least to a distance of 3 m, and the absence of open or disturbed sites can substantially restrict seed-colonization, even at short distances. Seeded native forb density in disturbed plots at a distance of 3 m was found to be higher than in the adjacent undisturbed areas (19 versus five plants m-2). This suggests that competition from the matrix vegetation may be more important in forb colonization than distance from the source. We also found that seed mixtures containing taller forbs, forbs with large seeds, or forbs with rhizomatous growth facilitate forb colonization. The study (Grygiel et al. 2014) indicates that in a site without physical or animal generated soil disturbances (such as occurs on most prairies), disturbance of the vegetation matrix between the seeded patches must be incorporated into the management plan to facilitate propagule colonization from the seeded patches.
A further interest of ours was to test whether soil amendments can improve seedling establishment and persistence in the patches and thus result in higher seed availability for colonization. To test this, we established an additional five-year experiment (Grygiel et al. 2012) in which C and P amendments were applied in various combinations. Results showed P amendments could be a potential tool for enhancing native seeded forb biomass and reducing non-native grass biomass, the caveat being that P amendments can also facilitate the invasion of exotic forbs. Carbon addition results were more complex. The C additions enhanced native and non-native grass biomass but also substantially reduced native forb density and biomass. The mixed results of the Grygiel et al. (2012) study suggest that, for the PPR methods, amendments may not be needed.
The difference in forb density and richness between the 50% PPR and the 25% PPR treatments is intriguing. Several causes may be involved. First, the 25% PPR treatment may have an effect similar to minimum tillage by leaving most of the matrix intact, reducing soil water evaporation, soil erosion, weed invasion and soil temperature fluctuations. In the 50% PPR treatment, half the area is rototilled, thus approaching a conventional cultivation method, losing the advantages of minimum tillage without the benefit of seeding the entire area. The literature has shown that rototilling and broadcast seeding often results in species-poor sites dominated by a few grasses (Kindscher and Tieszen 1998, Allison 2002, Sluis 2002). The H&D was essentially a minimum tillage application, but literature for the Northern Great Plains (Bakker et al. 2003, Wilson et al. 2004) has shown that drill seeding is less effective than broadcast seeding for long-term establishment and survival of native forbs. Second, another factor for these results may involve the fragmentation of the unseeded matrix. As shown in Grygiel et al. (2014), the adjacent 1-m area of the seeded disturbances exhibits very low forb colonization. Accounting for the 1-m distance to the seeded disturbances, the effective matrix area for seeded forb colonization in the 50% PPR treatment is 9%, while the effective matrix area for seeded forb colonization in the 25% PPR treatment is 55%. Future studies using different percentages of seeded small-scale disturbances may prove useful in ascertaining if there is a threshold level of disturbance.
The 25% PPR treatment could prove to be a cost-effective method for increasing native forb richness and density in a standing grass matrix without the need for repeated inter-seeding or a complete re-installation of the site. As we showed in Grygiel et al. (2009), the cost of the 25% PPR treatment was approximately 1/3 the cost of the traditional R&B treatment ($3,196 vs $9,649 in year 2000 USD), with most of the savings resulting from lower seed cost and installation. As we stated in Grygiel et al. (2009) some of the forb species seeds used in this study are expensive and in short supply, thus for a large-scale implementation survival data like the one included in Table 1 could be an effective guide to the most cost-effective species to use in a given area.
Allium stellatum. USDA-NRCS PLANTS Database. Britton, N.L. and A. Brown. 1913. An Illustrated Flora of the Northern United States, Canada and the British Possessions. New York, NY: Charles Scribner’s Sons.
Acknowledgments
This research was funded by The Nature Conservancy Ecosystem Research Program, Contract #HO-CSD-031500ND. We wish to thank Brian Winter, Director of Science and Stewardship and his staff at the Bluestem Prairie Scientific and Natural Area in Glyndon, Minnesota, and all the hardworking Natural Resources Management students at NDSU who comprised field crews over the years. We also want to thank Jim Johansen and his staff at Prairie Restoration, Inc. Bluestem Farm in Hawley, Minnesota for their expertise in managing the field installation of this experiment.
