Abstract
Low or declining plant diversity is a persistent problem in tallgrass prairie restoration. Most efforts to increase diversity involve seed additions and ecosystem manipulation to increase establishment. However, the species added often decline and are lost after the initial planting. An alternative would be to alter composition using species commonly found in remnants. Regeneration traits (clonal and non-clonal) may affect the ability of species to reproduce in established vegetation, affecting persistence and diversity. This study compares the composition and trait dominance of tallgrass remnants to restorations in the Midwestern U.S. The relative frequency of species found in nine high-quality remnants was compared to eight degraded remnants and 18 restorations at both site and regional scales by regeneration traits. Recommended species lists from three tallgrass restoration handbooks were compared using clustering analysis. At a regional scale, ten (71%) species greater than 25% relative frequency were clonal. At an individual site scale, sixty-two (52%) species greater than 25% relative frequency were clonal. Recommended seed mix composition is, on average, 68% different from remnant composition, with guerrilla species underrepresented and non-clonal species over-represented. At the 0.25m2 quadrat scale, richness was positively correlated with all growth forms. Where properties of natural or remnant ecosystems is a goal, a knowledge of common species or traits in a reference ecosystem should be represented in the seed mix. The regeneration niche of clonal species appears to be more advantageous than reproduction by seed in established vegetation, resulting in a majority of common species being clonal.
Restoration Recap
Increasing clonal species that may be more competitive with warm-season grasses in restorations may improve results by providing stability in composition and diversity.
Including seldom used but common species in seed mixes may provide a better understanding of their role in plant communities. Other species are included in seed mixes that are not found in remnant grasslands and may contribute to the instability of restorations.
When setting goals for an ecosystem restoration project, knowledge of all native species, and their commonness, found in that ecosystem is an important step in designing a local native reference ecosystem model.
One of the most fundamental steps in any restoration or reconstruction (hereafter both concepts are included in the term “restoration”) is defining the target ecosystem, a key attribute of which is species composition (McDonald et al. 2016). Climate change dictates that species diversity and persistence will become increasingly important in restoration ecology (Wilsey 2020). Temperate grasslands are recognized for their high small-scale species richness (Wilson et al. 2012), which has been demonstrated in a mesic prairie target reference ecosystem in the eastern region of tallgrass prairie in NE Illinois, United States, based on remnant grasslands (Sluis et al. 2017). However, a widely recognized shortcoming in many grassland restorations is a decline of forb richness and corresponding increase of warm-season grasses (Norland et al. 2015, Baer et al. 2015). Management practices such as mowing and burning aim to reduce this widely recognized effect. The magnitude of studies employing a range of techniques attempting to increase species richness in grasslands testifies to the recognition, importance, and persistence of this shortcoming (Hellstrom et al. 2009, McCain et al. 2010, Koziol and Bever 2016, Jaksetic et al. 2017, Klaus et al. 2017, Young et al. 2017, Zamin et al. 2017).
Nearly as prevalent, though less studied and experimentally manipulated, are compositional differences between restorations and their target ecosystems (Barr et al. 2017, Sluis et al. 2017). Compositional differences are likely due to seed mix composition, which may be significantly different than target ecosystem composition (Weber 1999, Morgan 2005). More importantly, if biotic filtering such as competition is a strong determinant of composition, the absence of more-competitive species may result in the dominance by a few species. Attempts have been made to alter competitive outcomes by differential timing of planting (Young et al. 2017). However, these priority effects are not differences in competitive ability. Given enough time, a more-competitive species may still outcompete a less-competitive one (Myers and Harms 2009, Kelemen et al. 2014, Alstad et al. 2016), including rare plants of special conservation interest (Albrecht et al. 2018). Longterm studies are few, but most show continued loss of species richness, (McLachlan and Knispel 2005, Heslinga and Grese 2010, Rinella et al. 2012) or fluctuating results (Velbert et al. 2017). These compositional differences and, therefore absent species, may play important roles in richness and dominance differences between remnant reference sites and restorations.
The absence of species that could potentially reach and inhabit a community is referred to as dark diversity (Pärtel et al. 2011) and may be related to long-term and successional community stability, landscape properties, and biotic interactions (Pärtel et al. 2013). Restorations missing species may be less stable and become dominated by a few highly competitive species. This is especially true if species missing from restorations are common in target communities (Gaston 2010), demonstrating the ability to persist or thrive in remnants. Common species have repeatedly been found to be much better correlated with overall spatial patterns of species richness than rare species (Šizling et al. 2009). Species may be common locally in remnants due to factors such as soil or hydrology, or be regionally common due to factors affecting geographic range. One method of determining the role of common species in achieving high levels of diversity involves breaking down spatial patterns of species richness into component groups and determining how the groups affect the overall richness patterns (Marquet et al. 2004), including classes of commonness/rarity (Lennon et al. 2004) and functional groups (Keil et al. 2008). This can be done by examining areas that have high levels of diversity and to see which species or types of species grow there, and which species characteristics or traits may contribute to their persistence and the high levels of diversity. For restorations, some of the most important traits and processes would involve the regeneration niche, the adaptations by which species survive, grow, and reproduce. These would include seed production and dispersal, and germination and establishment, which are often overlooked in trait analysis (Larson and Funk 2016). Many restoration implementation and management activities involve manipulation of the regeneration niche of species. Overcoming low rates of natural seed dispersal using any of the different seed planting techniques is one of the most common restoration activities. Planting timing and seed treatment, such as scarification, affect germination and establishment rates. Soil preparation and weed control facilitate growth and reproduction, which are essential for the persistence of desired species. Burning, mowing, grazing and herbicide applications select for the reproduction of some species and the reduction or death of others. However, once vegetation is established, changes in successful regeneration processes occur.
While much establishment in restorations initially occurs from planting seed, clonal reproduction is common in grasslands (Tamm et al. 2001, Lauenroth and Adler 2008, Gough et al. 2012, Willand et al. 2013), and is responsible for most reproduction in established vegetation (Benson and Harnett 2006, Ozinga et al. 2007, Herben et al. 2014). Clonal species are often divided into spatially non-mobile “matrix-forming (phalanx)” species, such as the warmseason grasses Andropogon gerardii (Big Bluestem) and Sorghastrum nutans (Indian Grass), and mobile “intermatrix (guerrilla)” species (Lovett-Doust 1981, Zobel et al. 2010) such as the rhizomatous species Solidago altissima (Tall Goldenrod) and Symphyotrichum ericoides (Heath Aster). However, the extent, commonness, and effects of clonal species in mesic grasslands and their use in restorations is poorly documented.
To determine types of species or traits that may be important to maintain or promote diversity in restorations, I examined the prevalence of three types of regenerative niche growth forms: phalanx, guerrilla, and non-clonal, in remnant grasslands used for a mesic grassland target ecosystem in NE Illinois, United States. My goals were: 1) To compare the species composition of restored grasslands to those of remnant grasslands, especially with respect to the prevalence of three different regenerative niche growth forms (phalanx, guerilla, non-clonal). If restorations are missing species that are the most common or reproduce the most, they are likely to be unstable, making them susceptible to invasion by undesirable species or loss of species richness. 2) To compare the species composition of recommended seed mixes to that of restored and remnant grasslands. Using general seed mixes may contribute to similarity among restorations, reducing diversity among restoration projects. Using more local or site-specific species composition may contribute more to larger-scale species diversity. 3) To determine the extent to which the composition of the seed mixes can explain the observed differences between restored and remnant grasslands. Changes to seed mixes may be needed to include missing species to create more species-rich and persistent restorations. Seed mixes lacking common or persistent species may be contributing factors in declines in species richness and dominance by warm-season grasses.
Methods
To compare patterns of species richness in remnants and restorations, and determine what kinds of species persist there, I used two classes of commonness/rarity (Lennon et al. 2004) and regeneration niche functional groups (Keil et al. 2008). The first class of common species were those found with an average of at least one individual in every square meter at nine grade A (described below) sites sampled, a frequency of 25% of quadrat samples or higher. These species would be expected to play an important role in the structure and diversity of sites because they are common and have shown the ability to persist despite other invading species attempting to outcompete or displace them. They may also be the most important to establish in restorations that have an objective to mimic the composition or diversity of remnants sites, which may be important in the conservation of rare species that exist at such sites. The second class of common species were those that have the same frequency of 25% of quadrat samples, or higher, at one or a few sites, but are not common across all sites. These species are likely to be locally common due to some site-specific characteristic such as soil composition or hydrology that may be more important for conserving rarer species.
I used data from a study of 17 remnant sites and 18 restoration sites in Northeast Illinois, USA, (Sluis et al. 2017), in which spatial patterns of species richness were determined at alpha, beta, and gamma scales using presence data from twenty 0.25 m2 samples from each site, randomly located within 5m sections, usually along two parallel 50m transects. Remnant data sets comprised vegetation sampled by the Illinois Natural Areas Inventory (INAI, White 1978) or by Bowles and Jones (2013) from INAI sites. Restored sites included some of the oldest restorations and others considered some of the best restorations in the region, including the R.F. Betz prairie at Fermilab, planted in 1977, the Schulenberg Prairie at The Morton Arboretum, planted in 1962, and The Doris Westfall Prairie, the first restoration to be designated as an Illinois Nature Preserve, planted in 1972. Sites were then assigned to one of four categories based on site history of disturbance, mostly cattle grazing, which was reflected by sample richness. The compositional differences of the categories were also supported by hierarchical cluster analysis (Sluis et al. 2017): Grade A (9 less-disturbed sites) or C (8 more-disturbed sites) remnant or Grade A (9 sites) or C (9 sites) restoration (Table 1). I used the same 35 sites to assess the three regeneration growth forms: phalanx, guerilla, and non-clonal. Sites range from 42°21′12″ N to 40°03′14″ N and 89°22′35″ W to 87°38′44″ W. Taxonomy follows Swink and Wilhelm (1994).
The 35 sites used in analysis separated by remnant/restoration and quality based on Sluis et al. (2017). McLaughlin is the same prairie sampled 30 years apart. Sante Fe RR is two different sections of the same prairie, one with a history of disturbance. College of DuPage are separate areas, one planted with seed only, the other plants only. Schulenberg 1, 2, & 3 are three different areas of the same restoration. Fermilab 1, 2 & 15 are three separate restorations based on the years apart they were planted.
Species of two genera, Poa and Carex, were not always identifiable to species due to lack of reproductive structures. As a result, the two Poa species were combined. Carex spp. were also combined unless identified to species, but the Carex genus contains both phalanx and guerrilla forms. Most identified Carex spp. were guerrilla, so all unidentified Carex species were combined and categorized as guerrilla.
Categories of commonness/rarity were examined two ways. First, I calculated the relative frequency of each species for each site sample. Species with a relative frequency greater than 25%, or one individual per square meter, in at least one site (transect) I refer to as locally common. I then totaled the number of locally common species with each growth form, phalanx, guerilla, and non-clonal. Second, the nine Grade A remnants serve as a model for mesic grassland restorations in the region. I therefore used commonness classes of 1–5, 5–10, 10–25, and > 25% relative frequency averaged across all Grade A remnants. Species with a relative frequency greater than 25% average across all nine Grade A remnants are referred to as regionally common, and were used for comparison to seed mixes, which are recommended for tallgrass prairies of the region. Also, for comparison to recommended seed mixes, cumulative commonness classes of > 10% and > 5% were used. Seed mix recommendation are list only, with no recommended quantities, so the presence or absence of a species on a list was used for comparison. The recommended species lists were compiled for mesic prairie restorations from three tallgrass restoration books: 1. The Tallgrass Restoration Handbook for prairies, savannas, and woodlands (Morgan 2005), 2. The Tallgrass Prairie Center Guide to Prairie Restoration in the Upper Midwest (Smith et al. 2010), and 3. A Practical Guide to Prairie Reconstruction (Kurtz 2013). The three recommended species lists contained an average of 58 species, so I made comparisons to the 58 most frequent species from the list for Grade A remnants, ranked by relative frequency. To rank each species on a recommended list not found in the first 58 species of the Grade A remnant list, I included ranks and relative abundance of the 150 most frequent species found in the Grade A remnants and their inclusion on the recommended species lists from each book. I also compared the percent of species for each mix with the Grade A remnant list in each of the phalanx, guerilla, and non-clonal regenerative niche types.
To examine the effect on species richness of each regenerative niche type, total species richness for each quadrat was correlated with the phalanx, guerrilla, and non-clonal species abundance. Pearson correlation coefficients were tested for significance, adjusted using Bonferroni’s correction. To determine the extent to which the composition of the recommended seed mixes can explain the observed differences between restored and remnant grasslands, I conducted a hierarchcal cluster analysis of all 35 sites, classified by remnant and restoration grade, and included the Grade A relative frequency list and the three recommended seed mixes. I used Ward’s linkage method and Euclidean distances in PCord (McCune and Mefford 2006, Version 5.10, MjM Software, Gleneden Beach, Oregon, U.S.A.). Other analyses were conducted using SYSTAT (Version 8.0, SPSS, Inc, San Jose, Ca.).
Results
Among the 35 sites, 253 species were sampled. A total of 119 species had a relative frequency of 25% or greater within at least one site transect. Of those locally common species, sixty-two (52%) were clonal: 17 (14%) phalanx and 45 (38%) guerrilla, seven were non-native. Some species were locally common in most sites, others in only one site. Of the 57 (48%) non-clonal species, four were non-native. A general decline in species richness from grade A remnants to grade C restorations is reflected in several metrics (Figure 1): 1. total species per site (one-way ANOVA; F3,31 = 36.196, p < 0.001), 2. locally common species per site (one-way ANOVA; F3,31 = 7.662, p = 0.001), 3. clonal species per site (one-way ANOVA; F3,31 = 8.34, p = 0.003), and 4. guerrilla species per site (one-way ANOVA; F3,31 = 9.007, p < 0.001). Although grade A remnants had more phalanx species than grade C restorations, with grade C remnants and grade A restorations intermediate, differences were not statistically significant. The differences in the number of guerrilla species per transect is much more abrupt between remnants and restorations, with both remnants and both restorations similar to each other.
Trends in species abundance (mean ± S.D.) by quality types (Sluis et al. 2017): grade A remnants (1 Rem A), grade C remnants (2 Rem C), grade A restorations (3 Rest A), and grade C restorations (4 Rest C). Different letters indicate statistically significant differences after one-way ANOVA and Bonferroni’s post hoc test: A. total species per site (F3,31 = 36.196, p < 0.001), B. locally common species per site (F3,31 = 7.662, p = 0.001), C. clonal species per site (F3,31 = 8.34, p = 0.003), and D. guerrilla species per site (F3,31 = 9.007, p < 0.001) and, E. no significant differences among phalanx species per transect after Bonferroni’s correction (F3,31 = 3.095, p = 0.041).
Averaged across all nine Grade A remnant sites, 14 species had relative frequencies higher than 25%. Ten of these regionally common species (71%) were clonal: two (14%) phalanx and eight (57%) guerrilla, four (29%) were non-clonal (Table 2). All were native. The two phalanx species were the dominant grasses A. gerardii and S. nutans. Percent guerilla species decreased and non-clonal species increased as relative frequency went from > 25% to 1–5%. Forty-one species had a relative frequency higher than 10%, including those in the > 25% class, making them comparable to the 41 species in the seed mix recommended by Smith (2010) which had 27, 7, and 66% for phalanx, guerilla, and non-clonal species, respectively, compared to 15, 34, and 51% for Grade A remnants. Sixty-seven species had a relative frequency greater than 5%, including those in the > 10 and > 25% classes, making them comparable to the seed lists recommended by Morgan (70 species) and Kurtz (57 species). For phalanx, guerilla, and non-clonal species, respectively, Grade A remnants had 12, 29, and 59%, Morgan (2005) 10, 18, and 72% and Kurtz (2013) 14, 18, and 68%. Among the regionally common species, there were no non-clonal or phalanx species in grade C remnants and no guerrilla species in grade A restorations (Table 3).
Percent and number of species (N) for relative frequency classes found in nine grade A remnant grasslands for each growth form: phalanx (P), guerrilla (G), and non-clonal (NC). Morgan = Tallgrass Restoration Handbook (Packard and Patel 2005), Smith = The Tallgrass Prairie Center Guide to Prairie Restoration in the Upper Midwest (Smith et al. 2010), Kurtz = A Practical Guide to Prairie Reconstruction, 2nd ed. (Kurtz 2013).
Number of regionally abundant species, mean relative frequency > 0.25, for each growth form (G = guerrilla, P = phalanx, NC = non-clonal) for each grassland quality type. rf = relative frequency.
At the 0.25m2 quadrat scale, overall richness was positively correlated with all growth forms, but most weakly with phalanx species and most strongly with non-clonal species. Results were similar among quality types, with non-clonal species most strongly correlated with richness (Table 4). There were two exceptions: phalanx species in grade C remnants and guerrilla species in grade C restorations were not significantly correlated with richness.
Pearson correlations of species richness at the 0.25 m2 quadrat scale with the number of species of each growth form by site quality. Significance adjusted using Bonferroni’s correction. *p < 0.001.
Comparing recommended seed mixes, which averaged 58 species, to Grade A remnants using the 58 most frequent species, 24 (41%) were not found in any seed mixes, 14 (24%) were found in all three mixes (Table 5). Among the three seed mixes, 27% of species were not in the 58 most frequent species in grade A remnants and seed mix species had an average relative frequency rank of 58 among the 150 most frequent remnant species. Combining the 41% missing species and 27% species in seed mixes that were not found in remnants adds up to approximately a 68% difference between remnant composition and average seed mix composition. The seed mixes also clustered with Grade A restorations and more distantly with grade C restorations (Figure 2). Unsurprisingly, the Grade A relative frequency list clustered closely with Grade A remnants presence list but was very distant from the seed mixes.
Hierarchical cluster analysis of remnants, restorations, and seed mixes. Seed mixes are more similar to four grade A restorations and one grade A remnant, than other grade A and C restorations. Seed mixes are furthest (most different) from the Grade A relative frequency list and Grade A remnants. Categories are as follows: A remn = grade A remnant, C remn = grade C remnant, A rest = grade A restoration, C rest = grade C restoration, A list = 85 most frequent species from list in Table 5, seed mixes labeled as in Table 5.
A comparison of recommended seed mixes for mesic tallgrass prairies of the Midwestern United States. The recommended species lists were compiled for mesic prairie restorations from three tallgrass restoration books: 1. The Tallgrass Restoration Handbook for prairies, savannas, and woodlands (Packard and Mutel 2005), 2. The Tallgrass Prairie Center Guide to Prairie Restoration in the Upper Midwest (Smith et al. 2010), and 3. A Practical Guide to Prairie Reconstruction (Kurtz 2013). Among the three seed mixes, 27% of species were not in the 58 most frequent species in remnants and seed mix species had an average relative frequency rank of 58 among the 150 most frequent remnant species. Growth forms: p = phalanx, g = guerrilla, nc = non-clonal. RF = relative frequency. Rank is the rank position among the 150 most frequent species from the nine grade A remnant grasslands. Seed mix recommendation are lists only, with no quantities, so the presence or absence of a species on a list was used for comparison, including the Grade A list.
Discussion
There has been much debate recently about the current and future direction of ecological restoration. In 2016, the Society for Ecological Restoration’s International Standards for the Practice of Ecological Restoration (McDonald et al. 2016) emphasized using a reference ecosystem as a model, or target, for the local native ecosystems being restored. This issue was extensively debated (Higgs et al. 2018a, Gann et al. 2018, Higgs et al. 2018b, Hobbs 2018, Higgs et al. 2019). In 2019, the second edition reiterated using a reference ecosystem (Gann et al. 2019). Also, the United Nations declared 2021–2030 as the “Decade on Ecosystem Restoration” (UNEA, 2019). In response, Cooke et al. (2019) contend that the current state of knowledge is not sufficient to risk large-scale restorations given success rates. Young and Schwartz (2019) agreed with the lack of knowledge, but countered that restoration efforts should proceed anyways, but smarter monitoring and better accountability, further research and analysis into restoration effectiveness should receive greater emphasis. Jones et al. (2018) conclude that active restoration is largely ineffective and unnecessary, but that complete recovery is rare, so clear goals and objectives are essential. Clearly, setting goals and objectives and measuring the success of restoration efforts is at the core of the current state of restoration practice, and knowing which species (e.g., common species) or functional groups (e.g., guerilla species) comprise remnant ecosystems and which of those are missing from restorations designed to restore those ecosystems is an essential component.
Dark diversity is highly relevant for restoration since most projects are limited in the number of species that are available, planted, and established. The large differences between seed mix and remnant composition and the lack of guerrilla species in exchange for non-clonal species that have been shown not to persist makes it unsurprising that many restoration projects are unstable and decline in diversity. This is especially true for species that are common in remnants but not planted in restorations. Remnants, by definition, contain many of the most persistent species for mostly undisturbed conditions that are generally of conservation interest, in part because remnants usually contain the rarest species as well as common species. Since the role of many species, including their traits and functions in plant communities, is incompletely understood, the effects of species absences are also incompletely understood. Even those restorations considered exceptional lack many of the species found in remnants and many of those present in restorations are in lower abundances than in remnants (Sluis et al. 2017). While it is not necessary to create an exact replica of remnants, the continued grass dominance, and resulting low richness of many restorations, demonstrates an insufficient understanding of the ecology of grasslands necessary to achieve restorations that meet the goals of creating areas with species richness or composition comparable to remnants (Cooke et al. 2018).
The most definitive result of this study is the abundance of guerrilla species in remnants and their scarcity in restorations. Their high abundance indicates they are persistent under remnant conditions, including sites that contain the grasses that become dominant in restorations. This supports findings of other studies about the superior ability of clonal species to reproduce vegetatively in established vegetation compared to species that reproduce only by seed and suggests that clonal species should increase with time rather than decrease (Benson and Harnett 2006, Ozinga et al. 2007, Wilsey 2010, Zobel et al. 2010, Herben et al. 2014). Given that many grassland restorations lose many of the initially planted species (McCain et al. 2010, Jones et al. 2013, Baer et al. 2015), such persistence may be an important factor in maintaining or increasing diversity. In addition, studies of grassland restorations that contain few guerrilla species or have composition dissimilar to remnants rather than studies conducted in remnants or using species common in remnants may be missing important interactions of grassland species. Composition similar to target sites is often recommended (Reid 2015, McDonald et al. 2016), but seldom accomplished (Sluis et al. 2017), and not reflected in most seed mixes (Weber 1999). The composition of recommended seed mixes and restoration results found here also support these differences (Table 5, Figure 2). Focusing on making restoration composition closer to remnant composition by including more missing species rather than continuing to plant species not common or found at all in remnants should improve the stability and potentially increase the diversity of restorations.
Among the four most frequent species in Grade A remnants, all guerilla species, most were not found in any recommended seed mixes. Of the most frequent, Carex spp, only one seed mix contained any species, but they were all non-clonal. Sedges are also commonly found throughout the tallgrass prairie region: 20% cover in Iowa and Missouri (McGranahan et al. 2015), 78% relative frequency in Illinois (Sluis et al. 2017), and up to 25% standing crop biomass in Oklahoma (Coppedge et al. 1998). Although clonal Carex spp. may produce fewer seeds because the trade-offs between seed and ramet production, plugs can easily be produced by division and result in more robust plugs that are frost tolerant, allowing for planting early in the growing season, potentially improving survival rates (Sluis, pers. obs.). The second most frequent, Commandra umbellata (False Toadflax), is known to be parasitic on at least 50 plant species, including many prairie species (Shive 1915). C. umbellata may act as a keystone species by differentially reducing the biomass of some species and not others (Joshi et al. 2000, Heer et al. 2018). It has been proposed as a pseudograzer by Henderson (2003), but its effects remain untested (DiGiovanni et al. 2016), and it was not included in any of the recommended seed mixes, likely because it is challenging to propagate. Smilacina stellata (Starry False Solomon’s Seal), the third most frequent species, is also difficult to propagate, taking up to 15 months (Luna et al. 2008) and was not in any recommended seed mixes. All three species have limited commercial availability. The fourth most frequent species, Rosa Carolina (Pasture Rose), is easier to germinate and was found in two recommended seed mixes. These four species were all more frequent than the dominant grasses A. gerardi and S. nutans but rarely found in recommended seed mixes. Their high frequency also suggests that they are competitive with and persist in the presence of the dominant grasses.
The strong correlations of all growth forms with richness indicate that a full complement of species is needed to achieve a high level of species richness. Sizling et al. (2009) have demonstrated mathematically that common species are more important than rare species to patterns of species richness. This also makes intuitive sense when referencing the regionally common species in this study. Common species, by definition, have higher relative frequencies across all sites, indicating that they occur at most or all sites. Having more of these species at all sites increases richness at the site level. If these common species also intersperse with each other, they contribute to higher small-scale (quadrat) richness. It follows then that species that can disperse to new areas within a site quickly by foraging for open space (clonal species), contribute more to richness at all scales. Species-rich mesic grassland sites with richness at one scale have been shown to have high richness at all scales (Sluis et al. 2017). Common species are also likely to be generalists, and rare species specialists, by the fact that common species are found at many sites while rare species are not. Therefore, not having rapidly dispersing clonal species (guerilla) in a seed mix or at a site may lead to overall declines in richness as phalanx species increase by seed dispersal or initial planting, a common occurrence in many restorations.
The patterns of correlations of richness with common and rare species found in this study is supported by similar results found in a Liaodong oak (Quercus wutaishanica) forest in the Ziwu Mountains of Loess Plateau, northwestern China (Wang et al. 2016). The study concluded that common species had stronger effects on species α diversity and species β diversity compared to rare species. Also, the number of rare species was greater than that of common species, but overall species richness pattern was better predicted by common species than rare species and common species were confirmed to be good indicators of species richness patterns. The results in this study of the correlation of richness with growth forms may be more revealing by examining where richness was not correlated with each growth form (Table 4). This occurred for both phalanx species in Grade C remnants and non-clonal species in Grade A restorations. In both cases, neither class had any species in the most common (> 25%) class (Table 3). Grazing is known to reduce the abundance of phalanx grasses, which are known to exclude non-clonal species in restorations. However, recovery of suppression by grasses may not occur if seeds of the suppressed species are not present.
Finally, studies using restorations missing species or functional groups with important traits may produce inaccurate results because important species or interactions are absent, potentially resulting in faulty assumptions of the completeness of the system being studied. Many studies examine techniques to increase species richness with some type of disturbance or manipulation. However, added species are often common in restoration projects but not remnants (Koziol and Bever 2017), and reproduce mostly by seed only. In addition, some of the species used are easy to establish from seed but are not commonly found in the habitat in which they are being used. Missing species with traits conferring competitive ability or persistence, such as guerrilla species, may be more successful in terms of stability and spread (Wilsey 2010, Mudrák et al. 2018). They may be useful for increasing diversity but are missed because they may be difficult to obtain, establish poorly from seed, or are not recognized as common in remnants. Including species common to remnants in richness-enhancing studies is likely to produce more accurate results.
Overcoming dark diversity in restorations will require recognition of missing species, their commonness or importance in target ecosystems, the knowledge of how to establish them, and the effort required to established them. For example, some species, such as the guerilla Potentilla simplex (Common cinquefoil), are thought to belong, and add value to, the plant community, but thrive without help (Packard and Ross 2005). This assumption causes them to be overlooked and not included in seed mixes (Morgan 2005), or restorations in general because they are considered “weedy”, and will eventually be outcompeted (Packard and Ross, 2005). However, P. simplex can be common in natural areas (Deam 1940) including grassland remnants (Swink and Wilhelm 1994, Hilty 2018). In the nine remnants sampled, P. simplex had a relative frequency of 5%, similar to the commonly planted species Amorpha canescens (Leadplant), Silphium laciniatum (Compass plant), and Liatris aspera (Rough Blazing Star). It is ten times more frequent there than some species commonly included in many seed mixes, such as Baptisia leucantha (White Wild Indigo), Desmodium canadense (Showy Tick Trefoil), or Penstemon digitalis (Foxglove Beardtongue).
Some of these species common in remnants but considered undesirable are non-existent in restorations (Sluis et al. 2017), yet their importance remains unknown. P. simplex rarely exceeds 2dm in height and, along with other short guerrilla species that were among the most frequent species found in remnants, some of which are also considered weeds, yet are native, may contribute to remnant shorter stature and higher diversity. P. simplex has a large tuber-like root structure that may confer persistence during stressful events such as droughts. It spreads via stolons that root at the nodes, forming new plantlets (Hilty 2018), thereby avoiding reproduction by seed. It is also spring-flowering, a functional group generally lacking in restorations (Sluis et al. 2017). The fact that P. simplex thrives in disturbed and unnatural areas may have no relevance to its function within a natural area. It may also make it easy to establish in restorations. This may be true of other species as well, some of which are much more frequent.
If restoration is an acid test of our ecological understanding, a thorough understanding should not be assumed until restorations are achieved that closely resemble remnants, even if only experimentally. A starting point should be to include species commonly found in remnants. Many of the most common species found in this study are largely absent from the lists of recommended species for mesic tallgrass restorations (Morgan 2005, Smith et al. 2010, Kurtz 2013). Added to common species should be those either locally common to a particular habitat and those of conservation interest or concern. Also, having species seed in a planting mix does not guarantee establishment. A fundamental step is to determine the importance of common, yet missing, species or traits, such as clonal species. This may require studies or restoration effort to determine how to establish them in restorations, especially if they may compete with dominant grasses and persist despite grasses, potentially reduce grass dominance to levels common in remnants, or change the physical structure of the community to facilitate additional species establishment.
Propagation by seed of many clonal species is difficult, and may be easier by other methods such as root division based on their natural tendency to reproduce vegetatively (Luna 2009). As a result, some common species are currently not economically feasible for commercial nurseries or are cost prohibitive for conservation organizations. Although establishing more clonal species may be more costly initially, if they continue to spread and dominate or alter physical structure allowing greater diversity while species that reproduce mainly by seed decline, the extra initial cost may be more cost-effective in the long run as results are more diverse, more persistent, and less additional management is needed. Alternatively, the conservation value of grass-dominated restorations of low species richness needs to be accurately assessed, clarified, and communicated (McDonald et al. 2016, Cross et al. 2018). This study also emphasizes the importance of having a thorough understanding of an ecosystem targeted for restoration and the goals of the project.
Symphyotricum ericoides. 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
I thank the Illinois Department of Natural Resources for providing original data collected by the Illinois Natural Areas Inventory, as well as the forest preserved districts of Cook, DuPage, Kane, Vermilion and Will counties and Don Gardner for permission to sample prairie restorations. Marlin Bowles and Mike Jones assisted with data collection. Several anonymous reviewers provided comments that improved the manuscript. Data collection at Fermilab was funded by the Fermi National Accelerator Laboratory Environmental Research Park.
This open access article is distributed under the terms of the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0) and is freely available online at:http://jhr.uwpress.org
References
Supplementary Materials
Original data accompanies:
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.