Research Projects
Are we close enough? Comparing prairie reconstruction chronosequences to remnants following two site preparation methods in Missouri, U.S.A.
Period: March 1, 2020 - Present
Contact: Prairie Fork
Organization: University of Missouri Prairie Fork Trust, Missouri Department of Conservation, US Forest Service
Objectives: Restoration Ecology -March 2020
RESEARCH ARTICLE
Are we close enough? Comparing prairie reconstruction chronosequences to remnants following two site preparation methods in Missouri, U.S.A.
Chris Newbold1, Benjamin O. Knapp2,3 , Lauren S. Pile4
Prairie reconstruction has become a common method for reestablishing tallgrass prairie communities in the central United States. With the objective of creating plant communities that approximate remnant (reference) prairies, managers are interested in identifying: (1) best methods for reconstructing reference community conditions; (2) the rate of change in plant communities through time following reconstruction; and (3) species present in remnant communities but missing from reconstructed communities. This information is important in the development of adaptive management strategies during active reconstruction. We used a chronosequence approach to assess the success of two reconstruction methods in emulating local, reference remnant prairie plant communities. We compared broadcast dormant seeding following two types of site preparation, agricultural cropping (Crop) or herbicide control in existing grass assemblages (Grass), and remnant communities. The Crop site preparation method resulted in a rapid increase in richness shortly following seeding. Although more similar to remnant assemblages initially, the Grass method took longer for mean coefficient of conservatism and floristic quality index to approach conditions of the reference communities. However, neither method resulted in plant community compositions that converged with the reference through time. Further, indicator species analysis identified a diverse assemblage of species lacking from the reconstructed prairies. These results suggest the need to develop management strategies for establishing the “missing” species during reconstruction and provide further support for protection and conservation of existing remnant prairies.
Key words: nonmetric multidimensional scaling, prairie reconstruction, remnant plant community, restoration success, site preparation
range (Samson et al. 2004; Hoekstra et al. 2005). In Missouri, this decline is even more acute, with less than 1% of the original 6 million ha of tallgrass prairie remaining (Schroeder 1981; Samson & Knopf 1994). As economic forces continue to encourage conversion of prairie to agricultural uses (Lark et al. 2015), the loss of the biodiversity and ecosystem func- tions that prairies provide will continue. To combat this loss, prairie reconstructions have recently become more common as a management practice in the Midwest (Packard & Mutel 2005; Anderson & Benda 2016; Rothrock et al. 2016).
Prairie restorations (i.e. management to rehabilitate degraded prairie communities that still hold some relict species) and reconstructions (i.e. reestablishment of prairie communities on
Author contributions: CN conceived the project; CN collected and organized the data; BOK, LSP analyzed the data; CN, BOK, LSP wrote and edited the manuscript.
1 Missouri Department of Conservation, 3500 E. Gans Road, Columbia, MO 65201, U.S.A.
2 School of Natural Resources, University of Missouri — Columbia, 203-S ABNR Building, Columbia, MO 65211, U.S.A.
3 Address correspondence to B. O. Knapp, email knappb@missouri.edu
4 USDA Forest Service — Northern Research Station, 202 ABNR Building, Columbia, MO 65211, U.S.A.
Introduction
The tallgrass prairie is one of the most endangered ecosystems in North America, with over 85% decline across its historic
© 2019 Society for Ecological Restoration
This article has been contributed to by US Government employees and their work is in the public domain in the USA
doi: 10.1111/rec.13078 Supporting information at:
http://onlinelibrary.wiley.com/doi/10.1111/rec.13078/suppinfo
Are we close enough? Prairie reconstruction
previously converted agricultural lands where prairie species no longer exist) are two land management practices that can be used to mitigate the loss of prairie biodiversity and ecosystem function (Smith et al. 2010). Prairie reconstruction practitioners attempt to emulate reference prairie communities by seeding diverse plant assemblages often collected locally from remnant prairies (Dickson & Busby 2009; Goldblum et al. 2013). Studies evaluating the success of restorations and reconstructions have not always used reference sites as comparative benchmarks (Wortley et al. 2013), despite the International Standards for the Practice of Ecological Restoration recommending to do so (McDonald et al. 2016).
Prior to seeding, site preparation methods are commonly implemented to reduce undesirable plant species (such as non- native [NN] species) or improve the likelihood of establishment success of the seed mixture. Site preparation methods may have a lasting effect on reconstructed plant communities (Millikin et al. 2016). Most prairie reconstruction managers prefer plant- ing on sites that have had existing vegetation removed (Rowe 2010). Removal of existing vegetation can be easily accom- plished by using commercial agricultural practices to prepare a site. However, due to a variety of reasons (lack of equipment or nearby producers, nonarable lands, etc.) commercial agricul- tural practices as a site preparation method may not be feasible in all prairie reconstruction efforts. In these cases, other forms of vegetation control to reduce competition with native remnant plants are used, although practitioners have indicated that these methods may be less effective than cropping (Rowe 2010).
Although prairie reconstructions may never fully support
the biodiversity and ecosystem function provided by remnant prairie communities (Polley et al. 2005; Bullock et al. 2011; Barak et al. 2017), they can provide conservation benefits, espe- cially when compared to degraded and NN grassland communi- ties (Rey Benayas et al. 2009; Tonietto et al. 2016; Trowbridge et al. 2016). Studies have shown that increased floristic diversity through prairie reconstruction can result in increased diversity of other taxa (Rowe & Holland 2012; Tonietto et al. 2016; Port & Schottler 2017). With time, reconstructed prairies can approx- imate the soil characteristics of undisturbed prairies (Rosen- zweig et al. 2016). Further, reconstructed prairies can provide resistance against invasion from NN plants (Blumenthal et al. 2005; Foster et al. 2015).
The documented benefits of prairie reconstruction offer promise for conserving tallgrass prairie ecosystems, yet eval- uating restoration success has remained a challenge for practi- tioners, due in part to different perspectives regarding success criteria (Higgs 1997; Zedler 2007; Wortley et al. 2013). Studies commonly use measures of plant diversity or vegetation struc- ture to evaluate success, but few prairie reconstruction projects have quantified the outcomes of different reconstruction meth- ods relative to reference conditions to provide feedback to an adaptive management process (Ruiz-Jaen & Aide 2005; Larson et al. 2018). Explicitly, an adaptive management framework is a two-stage iterative process whereby learning through time influ- ences subsequent management through adaptation (Williams 2011). Based on this approach, it is important to identify (learn) how close the reconstruction approaches approximate remnant
community composition and what species are missing to target for future management efforts (adapt). Using a chronosequence approach, we present plant community dynamics through 14 years after reconstruction in comparison to remnant tallgrass prairies in central Missouri. Our objectives are to: (1) describe how plant communities of prairie reconstructions change through time relative to nearby remnant reference communities;
(2) compare response patterns for two prairie reconstruction site preparation methods (cropping vs. herbicide control with no cropping); and (3) compare plant community composition between reconstructions and remnant reference communities to identify species or species groups missing from reconstructions.
Methods
Study Sites
Prairie Fork Conservation Area (PFCA) is a 367 ha property located in the southern portion of Missouri’s Claypan Till Plains Subsection (Nigh & Schroeder 2002) (Fig. 1) and managed cooperatively between the Missouri Department of Conserva- tion (MDC), Prairie Fork Trust (PFT), and the Missouri Prairie Foundation (MPF). For most of the twentieth century, PFCA had a history of agricultural use that included conversion of most of the area’s original natural communities to cropland and NN grasslands. In 2004, MDC, PFT, and MPF started a project with the goal of reconstructing PFCA’s pre-European tallgrass prairie. Remnant tallgrass prairies identified as ref- erence communities for the project included Tucker Prairie Natural Area (TPNA) and Marshall Diggs Conservation Area (MDCA), each within 23 km of PFCA (Fig. 1). TPNA is a 59 ha original claypan prairie, with Mexico and Armstrong soil series (Soil Survey Staff 2019), that has never been plowed and represents the largest remaining known tract of intact claypan prairie in this ecological subsection of Missouri. MDCA is a 410 ha area that contains a mix of woodland, savanna, and prairie communities with several small (≤4 ha) scattered clay- pan prairie remnants comprised of Calwood and Keswick soil series. The remnant prairie at TPNA is generally described as mesic, whereas remnants at MDCA are considered more xeric. PFCA is comprised of both mesic and xeric site types and soil series found at MDCA and TPNA.
Prairie Reconstruction Management
The steps taken for prairie reconstruction at PFCA include: (1) site preparation; (2) seeding native plants collected from ref- erence communities; and (3) mowing and prescribed burning for establishment and maintenance of the reconstructed prairies. Prior to reconstruction activities, Grabner and Grabner (1999) reported that vegetation at PFCA was dominated by undesir- able NN and native ruderal species (RD) (top 10 recorded species in order of coverage abundance: Festuca arundinacea [Tall fescue] Schreb. [NN], Solidago altissima [Tall goldenrod]
L. [RD], Lespedeza cuneata [Silky bush clover] [Dum. Cours.]
G. Don [NN], Kummerowia stipulacea [Korean bushclover] [Maxim.] Makino [NN], Setaria faberi [Giant foxtail] Herrm., Desmodium perplexum [Confusing trefoil] B.G. Schub. [RD],
Are we close enough? Prairie reconstruction
Figure 1. Location of Claypan Till Plains Subsection of Missouri and (A) Prairie Fork Conservation Area, (B) Tucker Prairie Natural Area, and (C) Marshall Diggs Conservation Area.
Trifolium pratense [Red clover] L. [NN], Ambrosia artemisiifo- lia [Common ragweed] L. [RD], Vernonia baldwinii [Western ironweed] Torr. [RD], and Eupatorium altissimum [Tall bone- set] L. [RD]).
For this study, two methods of site preparation for prairie reconstruction were compared (see Table S1 for schedule of all management actions). The first method (referred to as “Crop”) consisted of planting and harvesting glyphosate-resistant crops (e.g. soybeans and corn) by privately contracted agricultural producers for ≥3 years prior to seeding prairie species. This method removed most of the established undesirable species from the site, prepared a seed bed for good soil– seed contact, and was generally cost effective because the commercially harvested crops covered the cost of herbicide treatments. The second method (referred to as “Grass”) consisted of broadcast broadleaf herbicide (triclopyr) application (June– August) for at least 1 – 2 years, followed by 2 – 3 applications of glyphosate (May– June, August– September, and if needed, October– November) in the growing season prior to planting. All herbicide rates followed label directions and applications were determined based on manager assessment of current condi- tions. The Grass site preparation method also removed residual vegetation and prepared the seed bed for good soil– seed contact but was generally conducted over a shorter time period when compared to the Crop method.
Following both methods of site preparation, all reconstruc-
tions were seeded with local ecotype native collections from TPNA, MDCA, other local (within 75 km distance from PFCA) remnant natural communities, and mature reconstruction plant- ings from PFCA. Seed was harvested throughout the year using a combination of hand and machine (Woodward Flail-Vac Seed Stripper, Ag-Renewal, Inc., Weatherford, OK, U.S.A.) col- lection methods. From 2004 to 2016, 268 species of native grasses, sedges, and forbs were collected for reconstruction plantings, with an average of 179 species collected per year (Table S2). Although seed collections contained over 75% of the
same species each year, specific species and quantities varied based upon annual seed production of the plants in the pop- ulation. Both common and uncommon native prairie species were targeted for seed collection each year. After collecting seed throughout the growing season, reconstruction plantings occurred the following dormant season (January– February) by broadcast seeding on prepared ground (Crop or Grass). Seed viability tests were not conducted on collected seed and there- fore Pure Live Seed seeding rates are not available. Seeding rates of bulk material varied across years (ranged from 13.4 and
18.2 kg/ha) but did not vary across Crop and Grass prepared sites within planting years.
After seeding, reconstruction management included two mowing treatments (to residual heights of 15 – 20 cm) in the first growing season and one mowing treatment in the second growing season. Mowing treatments were designed to reduce dominance of fast-growing annual species that competed with new germinants for light and soil resources. Prescribed fire was introduced to the reconstructions during the dormant season prior to the second or third growing season (Table S1). Additionally, annual growing season spot foliar herbicide applications were conducted to control NN invasive species, particularly L. cuneata. Following the first prescribed burn, the reconstruction plantings entered a 2 – 4 year adaptive man- agement prescribed fire return interval using both growing and dormant season burns. For example, once the seeded vegetation was well established, growing season burns were applied to diminish the dominance of competitive C4 grasses.
Similarly, the remnant sites were managed using spot herbicide
applications to control invasive species and a combination of both growing and dormant season burns with an average fire return interval of 2.5 years (i.e. a late winter/early spring burn followed 2.5 years later with a late summer/early fall burn; Table S1).
These prairie reconstruction methods have occurred at PFCA since 2004, with 16 – 25 ha typically planted each year. Both Crop and Grass site preparation treatments were used during this period, with treatment designation dictated by management con- siderations rather than from an experimental design. When the monitoring plots were installed, the treatments were assigned opportunistically in that some areas were already in cropping (Crop) and others in pasture (Grass). As the reconstructions continued through time, decisions on Crop or Grass areas were often made based on compatibility with other research and edu- cational activities, with some consideration of accessibility for permit farmers to implement the cropping treatment. As a result, PFCA included a patchwork of reconstructions that varied in age across the two site preparation methods. Although these treatments were not randomly assigned at the beginning of the project, the site factors and land use legacies are similar across the range of treatments.
Vegetation Sampling
In 2000, prior to reconstruction at PFCA, permanent vegetation sampling sites were established to provide feedback to area management activities. These sites were stratified across soil
Are we close enough? Prairie reconstruction
types. Each site was marked with a point location that then had four sampling plots established in cardinal directions at random distances (between 15 and 90 m). This sampling design resulted in 23 sampling sites, totaling 92 sampling plots, located in the areas designated for prairie reconstruction. Each sampling plot had four 1-m2 quadrats located 3 m from center at bearings
45∘, 135∘, 225∘, and 315∘. In 2017, similar sampling plots were
randomly located within remnant prairies at TPNA (24 sampling plots) and MDCA (20 sampling plots).
Vegetation sampling occurred from 20 June to 25 July of 2017 at PFCA, TPNA, and MDCA. Within each 1-m2 quadrat, all rooted vascular plants were identified to species or genera and assigned a cover class based on their aerial coverage: (1) 0 – 5%; (2) 5 – 15%; (3) 15 – 25%; (4) 25 – 50%; (5) 50 – 75%; (6)
75 – 95%; and (7) 95 – 100%. When possible, plants were identi-
fied to species following Yatskievych (1999, 2006, 2013), with
>95% of species encounters recorded at the species level. How- ever, when specimen maturity or condition did not allow positive identification to species, plants were recorded to genera. Of the 19 genera for which this occurred, 14 occurred in less than 3% of the sampling plots (≤4 of 135; Aristida (Threeawn) L., Carya (Hickory) Nutt., Convolvulus (Bindweed) L., Cuscuta (Dodder) L., Cyperus (Flat sedge) L., Galium (Bedstraw) L., Helianthus (Sunflower) L., Lactuca (Lettuce) L., Plantago (Plantain) L., Prunus (Plum or Cherry) L., Rubus (Blackberry) L., Symphy- otrichum (Aster) Nees, Trifolium (Clover) L., and Vitis (Grape) L.). The other unidentified genera lumped were Carex (Sedge) L., Crataegus (Hawthorn) L., Ulmus (Elm) L., Juncus (Rush) L., and Melilotus (Sweet clover) (L.) Mill., which occurred in 51, 8, 6, 4, and 4% of the sampling plots, respectively. Uniden- tified Carex species made up the bulk of plants identified only to genera and summed to 1.6% of the total vegetation coverage observed across the sample plots. Nomenclature was based on Ladd and Thomas (2015).
Data Analysis
For each species, cover was converted to the cover class mid-point and averaged at the sample plot level (total of four 1-m2 sampling quadrats) for analyses. For each sample plot, we calculated total species richness, total native species rich- ness, mean species density (species richness per m2), and mean native species density. Following the Floristic Quality Assess- ment (FQA) system (Swink & Wilhelm 1994; Matthews et al. 2015), we calculated mean coefficient of conservatism (mean C) and floristic quality index (FQI) for each sample plot. Coeffi- cients of conservatism (C values) for each species were assigned by Ladd and Thomas (2015), and we assigned a C value of zero to each NN species. Occurrences recorded at the genera level were excluded from these analyses to provide a conservative estimate of species richness and to avoid introducing uncertainty into the FQA calculations.
We analyzed changes in the reconstructed plant commu- nities using a chronosequence approach. For some analyses, we grouped age since reconstruction into classes, including pre-reconstruction seeding (PS; Crop, n = 10; Grass, n = 21), 1 – 3 years since reconstruction (A1 – 3; Crop, n = 20; Grass,
n = 4), 4 – 7 years since reconstruction (A4 – 7; Crop, n = 9; Grass, n = 7), >8 years since reconstruction (A8+; Crop, n = 13; Grass, n = 7), and remnant communities (RM; n = 44). Best-fit bivariate models, using linear or nonlinear regression, determined relationships between age since reconstruction and plant community metrics for each of the site prepara- tion methods (Crop and Grass). Model fit was assessed by comparing statistical significance, mean square error, and r2 values. For each model, we calculated the 95% confidence interval. Similarly, we calculated the mean values and 95% confidence intervals from the remnant community plots (both TPNA and MDCA combined). We considered overlap in the confidence intervals to indicate no statistical difference in the plant community metric between site preparation methods or compared to the reference communities. Regression analyses were conducted using SAS 9.4 Software (SAS Institute Inc., Cary, NC, U.S.A.).
We used nonmetric multidimensional scaling (NMDS) to assess similarity among plant communities through time for Crop, Grass, and the remnant communities. For the ordination, we used the Bray-Curtis distance measure with six initial dimensions, random starting coordinates, and a maximum of 500 iterations. The dataset used mean coverage per m2 for each species and for those grouped by genera for all sample plots. The secondary matrix included age since reconstruction for each sample plot and the method of site preparation. For visual display, we indicated the age since reconstruction group and both reference sites (MDCA and TPNA). To determine the distance of the reconstruction plots to the remnant plots, we first determined the centroid of the plots within the remnant communities. Then to determine community dynamics through time, we calculated the Euclidean distance of each reconstruc- tion sampling plot from the remnant centroid in ordination space (Rydgren et al. 2019). We used bivariate regression to model the rate of community recovery through time and plotted regression lines with 95% confidence intervals for the Crop and Grass site preparation treatments, as well as the mean and 95% confidence intervals for the remnant sample plots.
To determine species missing in the reconstruction treat-
ments, indicator species analysis was used to identify species associated with each age since reconstruction group (i.e. PS, A1 – 3, A4 – 7, A8+, and RM) using the same data matrix as used for the NMDS ordination. The indicator species analy- sis calculates an indicator value (IV) based on the proportional abundance of a species in a group relative to the abundance of that species in all groups and the proportional frequency of the species in each group (McCune et al. 2002). For each species, the age since reconstruction group with the greatest IV was determined, and 4,999 Monte Carlo permutations were calcu- lated to determine if the probability that the observed IV was greater than randomly generated mean IV. For the indicator species analysis, we combined the Crop and Grass site prepara- tion datasets because we were primarily interested in identifying species in remnant communities that were not present in recon- structed communities, and we presented only significance for plants identified to species. We calculated mean C for indicator species of each age since reconstruction group and summarized
Are we close enough? Prairie reconstruction
the distribution of C values to the relative Importance Value for all species by age since reconstruction group. Importance Values were calculated for each species by summing their rel- ative coverage and relative frequency across each sample plot. The NMDS ordination and indicator species analysis were con- ducted using PC-ORD 6.08 Software (MjM Software Design, Gleneden Beach, OR, U.S.A.). For all statistical analyses, we considered significance for p < 0.05.
Results
In total, 241 taxa were identified across all plots, including 222 unique species and 19 additional genera groups (Table S3). A total of 183 taxa were recorded at PFCA, including 134 across all age classes of reconstruction plantings (A1 – 3, A4 – 7, and A8+). Prior to the completion of site preparation and broadcast seeding (PS), there were 16 species identified in the Crop plots and 124 species identified in the Grass plots. Across the remnant plots (MDCA and TPNA), there was a total of 139 taxa identified.
For the Crop site preparation, the reconstructed prairies showed patterns of rapid increase of each plant community met- ric in the first few years following reconstruction, reaching max- imum by around year 4 (Fig. 2). Based on the bivariate models, age since reconstruction had a significant, positive relationship with each plant community metric for Crop site preparation. For Grass site preparation, however, age since reconstruction had a significant positive relationship for only native density, mean C, and FQI (Fig. 2, Table S4). While native density for Grass site preparation was best fit with an exponential rise to maximum model form, mean C and FQI were best fit with linear models (Table S4). For the Crop site preparation, the 95% confidence intervals overlapped with the reference community, indicating no statistical difference, for total species density, total species richness, mean C, and FQI (Fig. 2). Although the Grass site preparation initially had greater values than Crop for all metrics, by around 2 years after treatment each metric was higher for the Crop site preparation. However, for mean C and FQI, the Grass site preparation reached levels similar to the Crop site prepa- ration and the remnant communities at around 10 years after treatment.
The NMDS was fit with a two-dimensional ordination, with a final stress of 17.52 and 77 iterations for the final solution. Axis 1 accounted for 51.6% of the variation in the ordination and Axis 2 accounted for an additional 12.4% of the variation (cumulative r2 of 0.64). The ordination indicated separation of Crop and Grass site preparation prior to the prairie reconstruction (PS), with communities moving closer to the remnant communities through time (Fig. 3). The two remnant sites used as reference communities (MDCA and TPNA) separated along Axis 2 of the ordination. The plant communities in the Crop site preparation differed more from the remnant communities than those in the Grass site preparation at the start of the reconstruction. By 6 years after reconstruction the Crop and Grass communities were not different (Fig. 3B). Age since reconstruction explained 93.1% of the variation in the distance to the remnant centroid
for the Crop site preparation and 78.1% of the variation for the Grass site preparation (Table S4). Neither the Crop nor Grass communities overlapped with the 95% confidence interval of the remnant communities at any point in time.
The indicator species analysis identified 20 species asso- ciated with the PS condition, of which 8 were NN species (Table S5). There were 12, 7, and 17 species identified as indi- cator species for A1 – 3, A4 – 7, and A8+ age groups, respec- tively. The remnant sampling plots had 33 species identified as
indicator species, none of which was NN. The mean C of indi- cator species increased with age since reconstruction, with the greatest mean C for the remnant communities (Fig. 4A). Simi- larly, the relative Importance Value for the PS group was domi- nated by NN or low C value species; as age since reconstruction increased, the contribution of more conservative species (C val- ues 4 – 8) generally increased (Fig. 4B).
Discussion
Our results show clear differences in plant community dynamics between the Crop and Grass site preparations used in recon- struction. The Crop sites were initially limited in species but quickly developed after seeding, with most community metrics approaching the remnant prairie sites through time. In contrast, sites that did not receive agricultural site preparation (Grass) took longer to develop remnant levels of floristic quality (mean C and FQI) and maintained lower levels of species richness. These patterns might be explained by reduced competition, available growing space, and favorable conditions for germi- nation and establishment in Crop versus Grass treatments. Ini- tially, Crop site preparation was much more effective at reducing or eliminating competing residual vegetation than noncropped sites (16 vs. 124 species observed prior to planting in the Crop and Grass sites, respectively). With less competition, germina- tion and initial establishment of new seedlings from broadcast seed may have been more successful in the cropped sites.
Furthermore, longer periods of reduced competition in Crop versus Grass sites, particularly from C4 grasses, may allow other perennial species to mature more quickly. To reduce com-
petition during the establishment phase, seed mixes used for reconstruction at PFCA purposefully contain low proportions of Andropogon gerardii (Big bluestem) Vitman and Sorghastrum nutans (Indiangrass) Nash, two dominant C4 prairie grasses. These two species are not specifically targeted in seed collec- tions and are only gathered incidentally as other species are collected and cleaned. Both grass species occur within the plant- ings, presumably from secondary seed collections, colonization from nearby sites, persistence in the seed bank, and as surviv- ing plants particularly in noncropped sites. The abundance of these species immediately after seeding was greater in the Grass sites than in the Crop sites (28.8% and 13.3% cover within 2 years, respectively), with these grasses becoming dominant (i.e. > 50% cover) in plantings by around age 3 in Grass sites and age 6 in Crop sites. Previous studies have suggested that domi- nance of C4 grasses may cause reductions in species richness of forbs (Camill et al. 2004; McCain et al. 2010; Pfeifer-Meister
Are we close enough? Prairie reconstruction
Figure 2. Relationships between age since reconstruction and plant community metrics: species density (number of species per m2) for native species (A) and total species (B); species richness (number of species per sampling plot, 4 m2) for native species (C) and total species (D); coefficient of conservatism (E); and floristic quality index (F). Vegetation data were collected in 2017 across all plots, with Crop (red circles) and Grass (blue triangles) site preparation methods fit with equations provided in Table S4. Shading indicates 95% confidence intervals, and data from remnant plots are shown for comparison to treatments.
Are we close enough? Prairie reconstruction
Figure 3. NMDS ordination of each sampling plot (A), showing age since reconstruction group, site preparation treatment, and remnant prairie communities and centroid. Regression of distance to remnant centroid by age since reconstruction (B) for Crop (red circles) and Grass (blue triangles). Shading indicates 95% confidence intervals, and data from remnant plots are shown for comparison to treatments.
et al. 2012; Baer et al. 2016). Thus, the additional period of lower competition in the Crop sites may benefit many prairie forb species in establishment and early persistence. From an operational standpoint, the suggestion that cropping for site preparation encourages greater native prairie planting establish- ment is fortunate given that commercial agricultural operators can conduct site preparation and costs can be recovered through harvested crops.
Our reconstruction plantings have not shown declines in floristic quality or species richness across older reconstructions, as reported by others (Sluis 2002; Camill et al. 2004; Hansen & Gibson 2013). The oldest reconstruction planting at PFCA was 14 years old in 2017, and it is possible that declines in richness and floristic quality may occur as the plantings continue
to age. However, PFCA management in mature (>7-year-old)
plantings is often directed at limiting the dominance of C4 grasses, specifically by periodically conducting late growing season (August– October) prescribed fires. Late summer-early
fall prescribed fires have been shown to reduce the relative cover of C4 grasses (Weir & Scasta 2017) and increase prairie forb biomass and diversity when compared to late winter-early spring burns (Towne & Craine 2014). Although not tested in this study, targeting C4 grasses with maintenance burns
through time may contribute to the sustained diversity in PFCA reconstructions.
Based on results from the ordination, the reconstruction plant communities moved toward but did not converge with the rem- nant reference communities over time. In fact, reconstruction communities seem to mature by 6 years of age, with little addi- tional movement toward remnant prairie communities. The dis- tributions of Importance Values by C value provide insight into general compositional differences. Remnant prairies had a unimodal distribution centered on species with C values of 4 and 5, which are typical of matrix prairie species (Ladd & Thomas 2015). In contrast, all reconstruction plantings exhib- ited bimodal distributions, with the primary mode centered on species with C values of 4 and 5 and a secondary mode centered on RD with C values of 0 and 1 (Ladd & Thomas 2015). Although the contribution of higher C value species increased with age since reconstruction, the reconstructions maintained a higher proportion of disturbance-adapted species than the remnants. Future monitoring will be needed to see if disturbance-adapted species continue to persist through time in the reconstruction plantings and to provide feedback to the adap- tive management cycle.
The indicator species analysis identified several species that are generally missing from the prairie reconstructions. Species significantly associated with remnant prairies occurred in three general categories: (1) native ruderal and woody species that are not typically collected for PFCA reconstructions; (2) spring ephemerals and other prairie “understory” species that produce seed that is difficult to collect in the late growing season; and
(3) species that are typically collected and seeded but inexpli- cably do not seem to successfully establish across the PFCA reconstructions. Species in the first category include Potentilla simplex (Common cinquefoil) Michx., Vernonia baldwinii, Cor- nus (Dogwood) L. spp., and Rhus (Sumac) L. spp., with the latter two documented to encroach upon prairies and degrade plant biodiversity (Ratajzac et al. 2011, 2012). Inclusion of these often-overlooked species in seed collections might be war- ranted to better emulate remnant prairie communities. Native shrub species play an important role in providing structural diversity on remnant prairies and their establishment in recon- structions may provide benefits to some prairie fauna (i.e. shrub nesting birds, small mammals, etc.). For example, the lack of availability of grassland-shrub breeding habitat for Bell’s Vireo (Vireo bellii) throughout the central United States is a contribut- ing factor in its decline (Budnik et al. 2000). Similarly, other passerine Midwestern shrub obligates are experiencing popula- tion declines due to a lack of available food or nest substrates, or both (Mabry 2013). Managers would need to weigh the struc- tural and compositional diversity benefits native woody shrubs would provide in prairie reconstructions against their long-term prairie encroachment threat.
A majority of the remnant prairie indicator species fit into the category of having difficult seed to collect (e.g. Antennaria [Pussy toes] Gaertn. spp., Crotalaria sagittalis [Rattlebox] L., Fragaria virginiana [Wild strawberry] Duchesne, Lysimachia lanceolata [Lance-leaved loosestrife Michx.], Hieracium
Are we close enough? Prairie reconstruction
Figure 4. Mean coefficient of conservatism (C) of indicator species for each age since reconstruction group (A), with nonnative species assigned “0” for the summary, and the relative Importance Value by C value for each age since reconstruction group (B– F), calculated from all sampling plots within a group. Relative Importance Value sums to 1 for each age group.
longipilum [Long-bearded hawkweed] Torr., and Viola sagit- tata [Arrow-leaved violet] Aiton) and include many spring flora species. Many of these species are important components of prairie communities. For example, V. sagittata is an important larval food for the Regal fritillary (Speyeria idalia Drury) and thus provides known conservation value. Sluis et al. (2017)
reported that spring flowering species (flowering before 1 July) are often poorly represented in prairie reconstructions, likely due to difficulty collecting the seed of spring flowering species that are often small and hidden by later growing species. Fur- thermore, due to the difficulty of collecting these species they are also not widely available for purchase from commercial
Are we close enough? Prairie reconstruction
vendors. Improved collection strategies such as increasing col- lection efforts or establishing production plots for these species would be needed to increase seed availability for reconstruction plantings. Planting small populations of propagated seedlings may be an alternative establishment practice for hard to collect and rare species. Propagation methods using plugs have been documented as increasing germination probability by 3 – 12 times that of direct seeding of C4 grasses (Gallagher & Wage- nius 2016), with establishment twice as successful for perennial herbs (Wallin et al. 2009). However, limited long-term survival without maintenance, proper species by site selection, and additional costs associated with plug production may limit the feasibility of this viable method for establishment (Morgan 1999; Wallin et al. 2009).
Finally, some species that were associated with remnant prairies (e.g. Amorpha canescens [Leadplant] Pursh, Baptisia bracteata [Cream wild indigo] Muhl. ex Elliott, Liatris aspera [Rough blazing star] Michx., and Rosa carolina [Pasture rose] L.] often have fair amounts of seed collected and seeded within the PFCA reconstructions yet are not commonly observed. For early maturing species, establishment may be improved by seed- ing during the growing season (Frischie & Rowe 2011). How- ever, establishment limitation for many species is not clear and may be complex. Generally, seedling recruitment under field conditions is low (Clark et al. 2007) and variable by species and site condition (Grman et al. 2015). Possible contributing factors include viability of collected seed, modified soil microbial com- munities (Tipton et al. 2019), soil structure and porosity, water stress following sowing, litter buildup, sowing grass density, or a host of other unknown factors (Clark et al. 2007; Grman et al. 2015). Repeated overseeding of these, and other missing species associated with remnants, may eventually help establish them in reconstruction plantings (Sluis et al. 2017). Spatially patterned seeding or propagule planting techniques (Grygiel et al. 2009; Rayburn & Laca 2013) might assist with establishment of missing species into older, established plantings. Our results suggest that improved understanding of species-level responses would help identify and overcome limitations to establishment success of key species associated with remnant communities.
Our results provide evidence that operational prairie recon-
structions can move plant communities toward remnant con- ditions, despite some limitations associated with the sampling design of this study. For example, total species richness differ- ences between remnants and reconstructions should be inter- preted with caution. Remnant sites had the greatest number of sampling sites, and therefore could have the highest total species richness due to expected species-area relationships (Gotelli & Colwell 2001). Additionally, the preseeded Grass sites had the greatest number of plots (n = 21) across all reconstruction age classes and this relatively high sampling effort could have con- tributed to the relatively high total species richness (124 total taxa) observed in this treatment. Our density (species/m2) and floristic quality measures, however, should not be impacted by this uneven sampling effort across sites. In addition, our sam- pling design would not have detected all species present at any site, may not have detected rare species, and was not designed
to track small changes in the plant community. For example,
scattered populations of Delphinium carolinianum (Carolina larkspur) Walter (C value = 6), Asclepias sullivantii (Prairie milkweed) Engelm. ex A. Gray (C value = 8), and Melanthium virginicum (Bunchflower) L. (C value = 9) have been observed in both remnant and the older reconstructions at PFCA but were too localized to be detected by our sampling design. A more detailed sampling design, like the use of modified Whittaker plots (Barnett & Stohlgren 2003), would likely be able to better detect smaller changes over time and differences among prairie reconstruction ages and remnants.
Although the reconstruction plantings at PFCA have estab- lished diverse prairie communities, they have not yet matched the remnant reference sites after which they are modeled. This fact provides further evidence as to the value of protecting and conserving the last remaining tallgrass prairie remnants in Mis- souri. Our results suggest that additional research is warranted to understand (1) limitations to establishment or persistence for those species commonly included in the seed mix but not present in reconstruction plantings and (2) methods for improv- ing seed collection of other species. Moreover, identification of native ruderal/woody species as indicators of remnant prairie suggests the need to consider these species in prairie reconstruc- tions when evaluating success criteria. Our work demonstrates that site preparation prior to seeding can quickly promote the development of a diverse prairie planting during reconstruction. Site preparation methodology affects the rate of development and may have lasting impacts on species richness and den- sity. Maintaining and increasing an established reconstruction planting’s diversity will require ongoing monitoring and adap- tive management and may include management practices such as control of aggressive C4 grasses and NN invasive species, repeated overseeding of missing species, and establishment of species colonies through propagule plantings.
Acknowledgments
We would like to thank Prairie Fork Trust and Pat Jones for funding collection of the field data. E. Horner and G. Greumad collected much of the vegetation quadrat data. Additionally, we would like to thank Dr T. P. Young and M. Leahy for their thoughtful comments and suggestions on earlier versions of the manuscript.
LITERATURE CITED
Anderson R, Benda C (eds) (2016) From Cemetery Prairies to National Tallgrass Prairies: Proceedings of the 24th North American Prairie Conference. North American Prairie Conference Proceedings. Illinois State University, Normal, Illinois
Baer SG, Blair JM, Collins SL (2016) Environmental heterogeneity has a weak effect on diversity during community assembly in tallgrass prairie. Ecological Monographs 86:94 – 106
Barak RS, Williams EW, Hipp AL, Bowles ML, Carr GM, Sherman R, Larkin DJ (2017) Restored tallgrass prairies have reduced phylogenetic diversity compared with remnants. Journal of Applied Ecology 54:1080 – 1090
Barnett DT, Stohlgren TJ (2003) A nested-intensity design for surveying plant diversity. Biodiversity and Conservation 12:255 – 278
Blumenthal DM, Jordan NR, Svenson EL (2005) Effects of prairie restoration on weed invasions. Agriculture, Ecosystems & Environment 107:221 – 230
Are we close enough? Prairie reconstruction
Budnik JM, Ryan MR, Thompson FR III (2000) Demography of Bell’s vireos in Missouri grassland-shrub habitats. The Auk 117:925 – 935
Bullock JM, Aronson J, Newton AC, Pywell RF, Rey-Benayas JM (2011) Restoration of ecosystem services and biodiversity: conflicts and oppor- tunities. Trends in Ecology and Evolution 26:541 – 549
Camill P, Mckone MJ, Sturges ST, Severud WJ, Ellis E, Limmer J, Mar- tin CB, Navratil RT, Purdie AJ, Sandel BS (2004) Community- and ecosystem-level changes in a species-rich tallgrass prairie restoration. Eco- logical Applications 14:1680 – 1694
Clark CJ, Poulsen JR, Levey DJ, Osenberg CW (2007) Are plant populations seed limited? A critique and meta-analysis of seed addition experiments. The American Naturalist 170:128 – 142
Dickson TL, Busby WH (2009) 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. Restoration Ecology 17:597 – 605
Foster BL, Houseman GR, Hall DR, Hinman SE (2015) Does tallgrass prairie restoration enhance invasion resistance of post-agricultural lands? Biolog- ical Invasions 17:3579 – 3590
Frischie SL, Rowe HI (2011) Replicating life cycle of early-maturing species in the timing of restoration seeding improves establishment and community diversity. Restoration Ecology 20:188 – 193
Gallagher MK, Wagenius S (2016) Seed source impacts germination and early establishment of dominant grasses in prairie restorations. Journal of Applied Ecology 53:251 – 263
Goldblum D, Glaves BP, Rigg LS, Kleiman B (2013) The impact of seed mix weight on diversity and species composition in a tallgrass prairie restora- tion planting, Nachusa Grasslands, Illinois, U.S.A. Ecological Restoration 31:154 – 167
Gotelli NJ, Colwell RK (2001) Quantifying biodiversity: procedures and pitfalls in the measurement and comparison of species richness. Ecology Letters 4:379 – 391
Grabner JK, Grabner KW (1999) Prairie Fork Creek conservation area: baseline vegetation monitoring plots— initial establishment, inventory and data summary. Report to Prairie Fork Trust, Williamsburg, Missouri
Grman E, Bassett T, Zirbel CR, Brudvig LA (2015) Dispersal and establish- ment filters influence the assembly of restored prairie plant communities. Restoration Ecology 23:892 – 899
Grygiel CE, Norland JE, Biomdini ME (2009) Precision prairie reconstruction (PPR): a technique for increasing native forb species richness in an estab- lished grass matrix. Ecological Restoration 27:458 – 466
Hansen MJ, Gibson DJ (2013) Use of multiple criteria in an ecological assess- ment of a prairie restoration chronosequence. Applied Vegetation Science 17:63 – 73
Higgs ES (1997) What is good ecological restoration? Conservation Biology 11:338 – 348
Hoekstra JM, Boucher TM, Ricketts TH, Roberts C (2005) Confronting a biome crisis: global disparities of habitat loss and protection. Ecology Letters 8:23 – 29
Ladd D, Thomas JR (2015) Ecological checklist of the Missouri flora for floristic quality assessment. Phyton 12:1 – 274
Lark TJ, Salmon JM, Gibbs HK (2015) Cropland expansion outpaces agricultural and biofuel policies in the United States. Environmental Research Letters 10:11
Larson DL, Ahlering M, Drobney P, Esser R, Larson JL, Viste-Sparkman K (2018) Developing a framework for evaluating tallgrass prairie reconstruc- tion methods and management. Ecological Restoration 36:6 – 18
Mabry C (2013) Optimal shrub density for bird habitat in the midwestern United States. Ecological Restoration 31:63 – 68
Matthews JW, Spyreas G, Long CM (2015) A null model test of floristic quality assessment: are plant species’ coefficients of conservatism valid? Ecological Indicators 52:1 – 7
McCain KNS, Baer SG, Blair JM, Wilson GWT (2010) Dominant grasses suppress local diversity in restored tallgrass prairie. Restoration Ecology 18:40 – 49
McCune B, Grace JB, Urban DL (2002) Analyses of ecological communities.
MjM Software Design, Gleneden Beach, Oregon
McDonald T, Gann GD, Jonson J, Dixon KW (2016) International standards for the practice of ecological restoration— including principles and key concepts. Society for Ecological Restoration, Washington D.C.
Millikin AR, Jarchow ME, Olmstead KL, Krentz RE, Dixon MD (2016) Site preparation drives long-term plant community dynamics in southeastern South Dakota. Environmental Management 58:597 – 605
Morgan JW (1999) Have tubestock plantings successfully established popula- tions of rare grassland species into reintroduction sites in western Victoria? Biological Conservation 89:235 – 243
Nigh TA, Schroeder WA (2002) Atlas of Missouri ecoregions. Missouri Depart- ment of Conservation, Jefferson City, Missouri
Packard S, Mutel CF (2005) The tall grass prairie restoration handbook: for prairies, savannas, and woodlands. 2nd edition. Island Press, Washington D.C.
Pfeifer-Meister L, Roy BA, Johnson BR, Krueger J, Bridgham S (2012) Domi- nance of native grasses leads to community convergence in wetland restora- tion. Plant Ecology 213:637 – 647
Polley HW, Derner JD, Wilsey BJ (2005) Patterns of plant species diversity in remnant and restored tallgrass prairies. Restoration Ecology 13:480 – 487
Port J, Schottler S (2017) The effect of floristic composition on bird communities in a set of four grassland reconstruction types. Ecological Restoration 35:112 – 119
Ratajzac Z, Nippert JP, Hartman JC, Ocheltree TW (2011) Positive feedbacks amplify rates of woody encroachment in mesic tallgrass prairie. Ecosphere 2:1 – 14
Ratajzac Z, Nippert JP, Collins SL (2012) Woody encroachment decreases diver- sity across North American grasslands and savannas. Ecology 93:697 – 703 Rayburn AP, Laca EA (2013) Strip-seeding for grassland restoration: past
successes and future potential. Ecological Restoration 31:147 – 153
Rey Benayas JM, Newton AC, Diaz A, Bullock JM (2009) Enhancement of bio- diversity and ecosystem services by ecological restoration: a meta-analysis. Science 325:1121 – 1124
Rosenzweig ST, Carson MA, Baer SG, Blair JM (2016) Changes in soil properties, microbial biomass, and fluxes of C and N in soil following post-agricultural grassland restoration. Applied Soil Ecology 100:186 – 194
Rothrock PE, Pruitt VB, Reber RT (2016) Prairie reconstruction in Indiana: historical highlights and outcomes. Proceedings of the Indiana Academy of Science 125:114 – 125
Rowe HI (2010) Tricks of the trade: techniques and opinions from 38 experts in tallgrass prairie restoration. Restoration Ecology 18:253 – 262
Rowe HI, Holland JD (2012) High plant richness in prairie reconstructions support diverse leafhopper communities. Restoration Ecology 21:174 – 180
Ruiz-Jaen MC, Aide TM (2005) Restoration success: how is it being measured?
Restoration Ecology 13:569 – 577
Rydgren K, Halvorsen R, Topper JP, Auestad I, Hamre LN, Jongejans E, Sulavik J (2019) Advancing restoration ecology: a new approach to predict time to recovery. Journal of Applied Ecology 56:225 – 234
Samson FB, Knopf FL (1994) Prairie conservation in North America. Bioscience 44:418 – 421
Samson FB, Knopf FL, Ostile W (2004) Great plains ecosystems: past, present and future. Wildlife Society Bulletin 32:6 – 15
Schroeder WA (1981) Presettlement prairie of Missouri. Natural History Series No. 2. Missouri Department of Conservation, Jefferson City, Missouri
Sluis WJ (2002) Patterns of species richness and composition in re-created grassland. Restoration Ecology 10:677 – 684
Sluis WJ, Bowles M, Jones M (2017) Multiscale metrics differentiate among tallgrass prairie restorations and remnant ecosystems along a restorative continuum. Restoration Ecology 26:466 – 475
Smith D, Williams D, Houseal G, Henderson K (2010) The tallgrass prairie guide to prairie restoration in the upper midwest. University of Iowa Press, Iowa City, Iowa
Soil Survey Staff, Natural Resource Conservation Service, United States Depart- ment of Agriculture (2019) Web soil survey. https://websoilsurvey.sc.egov
.usda.gov/ (accessed 08 Jun 2019)
Are we close enough? Prairie reconstruction
Swink F, Wilhelm G (1994) Plants of the Chicago region. 4th edition. Indiana Academy of Science, Indianapolis, Indiana
Tipton AG, Middleton EL, Spollen WG, Galen C (2019) Anthropogenic and soil environmental drivers of arbuscular mycorrhizal community composition differ between grassland ecosystems. Botany 97:85 – 99
Tonietto RK, Ascher JS, Larkin DJ (2016) Bee communities along a prairie restoration chronosequence: similar abundance and diversity, distinct com- position. Ecological Applications 27:705 – 717
Towne EG, Craine JM (2014) Ecological consequences of shifting the timing of burning tallgrass prairie. PLoS One 9:e103243
Trowbridge CC, Stanley A, Kaye TN, Dunwiddie PW, Williams JL (2016) Long-term effects of prairie restoration on plant community structure and native population dynamics. Restoration Ecology 25:559 – 568
Wallin L, Svensson BM, Lonn M (2009) Artificial dispersal as a restoration tool in meadows: sowing or planting? Restoration Ecology 17:270 – 279
Weir JR, Scasta JD (2017) Vegetation responses to season of fire in tallgrass prairie: a 13-year case study. Fire Ecology 13:137 – 142
Williams BK (2011) Adaptive management of natural resources— framework and issues. Journal of Environmental Management 95:1346 – 1353
Wortley L, Hero J, Howes M (2013) Evaluating ecological restoration success: a review of the literature. Restoration Ecology 21:537 – 543
Yatsievych G (1999) Steyermark’s flora of Missouri. Revised edition. Vol 1. Missouri Department of Conservation in Cooperation with the Missouri Botanical Garden Press, St. Louis, Missouri
Yatsievych G (2006) Steyermark’s flora of Missouri. Revised edition. Vol 2. Mis- souri Botanical Garden Press in Cooperation with the Missouri Department of Conservation, Jefferson City, Missouri
Yatsievych G (2013) Steyermark’s flora of Missouri. Revised edition. Vol 3. Mis- souri Botanical Garden Press in Cooperation with the Missouri Department of Conservation, Jefferson City, Missouri
Zedler JB (2007) Success: an unclear, subjective descriptor of restoration out- comes. Ecological Restoration 25:162 – 168
Supporting Information
The following information may be found in the online version of this article:
Table S1. Summary information for each sampling plot.
Table S2. List of species in the seed collections across study years.
Table S3. Mean cover by species for each site preparation treatment.
Table S4. Model form, parameters, and statistical summary for bivariate relationships. Table S5. Indicator species identified for each age since reconstruction group using the combined dataset.
Coordinating Editor: Stuart Allison Received: 18 May, 2019; First decision: 24 June, 2019; Revised: 23 November,
2019; Accepted: 4 November, 2019; First published online: 4 December, 2019