Convergent and Contingent Community Responses to Grass Source and Dominance During Prairie Restoration Across a Longitudinal Gradient

Convergent and Contingent Community Responses to GrassSource and Dominance During Prairie Restoration Acrossa Longitudinal Gradient

Ryan P. Klopf • Sara G. Baer • David J. Gibson

Received: 23 February 2013 / Accepted: 17 November 2013

� Springer Science+Business Media New York 2013

Abstract Restoring prairie on formerly cultivated land

begins by selecting propagule seed sources and the diversity

of species to reintroduce. This study examined the effects of

dominant grass propagule source (cultivar vs. non-cultivar)

and sown propagule diversity (grass:forb sowing ratio) on

plant community structure. Two field experiments were

established in Kansas and Illinois consisting of identical split

plot designs. Dominant grass source was assigned as the

whole-plot factor, and sown dominance of grasses (five

levels of seeded grass dominance) as the subplot factor.

Species density, cover, and diversity were quantified for

5 years. The effect of dominant grass source on the cover of

focal grasses, sown species, and volunteer species was

contingent upon location, with variation between dominant

grass sources observed exclusively in Kansas. Species den-

sity and diversity showed regionally convergent patterns in

response to dominant grass source. Contrary to our hypoth-

eses, total species density and diversity were not lower in the

presence of grass cultivars, the grass source we had predicted

would be more competitive. Sown grass dominance effects

on the cover of the focal grass species were contingent upon

location resulting from establishment corresponding better

to the assigned treatments in Illinois. All other cover groups

showed regionally convergent patterns, with lower cover of

volunteers and higher cover of sown forbs, diversity, and

species density in the lowest sown grass dominance treat-

ment in both sites. Thus, decisions regarding the diversity of

propagules to reintroduce had more consequence for plant

community structure than cultivar or non-cultivar source of

dominant grasses.

Keywords Restoration � Grassland � Ecotype � Seed

source

Introduction

Reversing ecosystem degradation to improve the structure,

function, and services provided by ecosystems is a com-

mon goal of ecological restoration (Jackson and Hobbs

2009; Doherty et al. 2011). Human decisions represent a

deterministic influence on the prairie community reas-

sembly process through restoration, starting with the

selection of species to reintroduce followed by the source

of propagules (i.e., local ecotypes, cultivars, or genotypic

mixtures) and the relative abundance of each species to add

(i.e., diversity). The composition of a restored community

will be determined by each species tolerance to abiotic

conditions and interactions with other species (Hobbs and

Norton 2004). Community composition in a restoration

may also be influenced by the genetic composition of

source populations as affected by the size of population

sources and propagule collection methods (Broadhurst

et al. 2008). Reassembled community structure may also be

influenced by stochastic factors that can modulate the

strength of biotic and abiotic filters during restoration

(Fattorini and Halle 2004). Empirically derived informa-

tion on the ecological consequences of these socio-eco-

nomic filters (i.e., propagule source and diversity selection)

Electronic supplementary material The online version of thisarticle (doi:10.1007/s00267-013-0209-3) contains supplementarymaterial, which is available to authorized users.

R. P. Klopf (&) � S. G. Baer � D. J. Gibson

Department of Plant Biology and Center for Ecology, Southern

Illinois University, Carbondale, IL 62901-6509, USA

e-mail: [email protected]

R. P. Klopf

5162 Valleypointe Parkway, Roanoke, VA 24019, USA

123

Environmental Management

DOI 10.1007/s00267-013-0209-3

and whether the community consequences are generaliz-

able or contingent upon local factors is needed to under-

stand the degree to which these decisions influence

richness and diversity in restored communities.

Propagule source selection criteria have been proposed

based on genetic and ecological principles (Lesica and

Allendorf 1999; Broadhurst et al. 2008). Local ecotypes are

assumed to possess adaptations to the local environment

and a ‘‘genetic memory’’ shaped by past stochastic events

important to the long-term success of a species (Falk 1990;

Montalvo et al. 1997). While a species can contain eco-

types adapted to local environmental conditions (Clausen

and Hiesey 1958), other species have wide ecological

amplitudes made possible by phenotypically plastic geno-

types (Bradshaw 1965; Lesica and Allendorf 1999).

Development of plant cultivars generally involves selection

for one or a combination of specific traits (e.g., drought

tolerance, disease resistance, and high reproductive output)

(Fehr 1987). Cultivars may establish successfully, but there

is concern that populations restored with cultivars will not

be able to persist over the long term if they lack alleles to

survive extreme selective episodes (Montalvo et al. 1997;

Lesica and Allendorf 1999; McKay et al. 2005).

Cultivars of the dominant prairie grasses have been

developed by the United States Department of Agriculture

(USDA 1995). Genetic, physiological, and functional dif-

ferences have been documented between cultivar and non-

cultivar sources of the dominant prairie grasses. For

example, Gustafson et al. (2004) documented cultivars of

Andropogon gerardii Vitman (big bluestem) and Sorgha-

strum nutans (L.) Nash (Indiangrass) were genetically

distinct from populations in remnant and restored prairies,

as determined by assays of neutral genetic markers. Cul-

tivars of A. gerardii, S. nutans, and Schizachyrium scopa-

rium (Michx.) Nash (little bluestem) have been shown to

have equivalent or higher rates of net photosynthesis than

non-cultivars of these species (Skeel and Gibson 1996;

Lambert et al. 2011). Further, Klopf and Baer (2011)

demonstrated greater belowground net primary productiv-

ity and lower moisture and plant available nitrate in the

rhizosphere of prairie grass cultivars compared to non-

cultivars during prairie restoration. No studies have com-

pared the community effects of dominant grass source

across multiple sites to elucidate whether community

composition, if affected by grass propagule source, is a

general phenomenon or contingent upon the local climate

or other site-specific property.

The diversity of propagules introduced to restore a plant

community represents a second decision and potential filter

(affected by collection effort and cost) on the community

assembly process in ecological restoration. Attaining a

community composition representative of remnant prairie

is a common goal of prairie restoration (Betz 1986; Sluis

2002; Hansen and Gibson 2013). However, plant diversity

is often lower in restored than remnant prairies due to the

abundance of dominant warm season grasses that suppress

establishment of less common species (Baer et al. 2004;

Polley et al. 2005; Taft et al. 2006; McCain et al. 2010;

Wilsey 2010). Although plant species richness in restored

prairie increases with the number of species sown (Piper

et al. 2007; Carter and Blair 2012a), prairies seeded and

over seeded with a high richness can lose species over time

(Hansen and Gibson 2013). No studies have evaluated

whether the loss of species and diversity varies across

regional conditions.

Geographic replication of field experiments, to include

variation in environmental conditions, can reveal whether

ecological patterns are convergent or contingent upon some

condition (Huxman et al. 2004), and if convergent, to what

extent experimental treatments are deterministic factors on

community reassembly (Fattorini and Halle 2004; Mac-

Dougall et al. 2008). The overall objective of this study

was to quantify the effects of grass propagule source (i.e.,

cultivar vs. non-cultivar) and sown diversity (i.e., grass

dominance) on the cover of planted and volunteer species

in developing prairie. This experiment tested the hypoth-

esis that human decisions regarding source and composi-

tion of propagules is a deterministic process that modulates

the biotic filter to affect richness and diversity in restored

prairie. Specifically, we predicted that prairie restored with

cultivars of the dominant grasses (selected for vigor and

reproductive output in degraded environments) or sown

with high grass dominance would attain greater grass cover

to adversely affect richness and diversity. Two experi-

mental restorations were established 620 km apart at the

same latitude (corresponding to contrasting locations

across a natural precipitation gradient) to elucidate whether

changes in plant cover, richness, and diversity over time

and in response to the manipulated biotic filters was con-

vergent and generalizable, or contingent upon site

conditions.

Methods

Study Sites

In 2005, two identical experiments were established in

former agricultural fields in Kansas and Illinois. The wes-

tern restoration experiment was conducted at the Konza

Prairie Biological Station (KPBS) and Long Term Eco-

logical Research site, in the Flint Hills of eastern Kansas

(39�050N, 96�350W). Elevation at KPBS is 340 m above sea

level. While interannual precipitation and temperature

variability is substantial, the long-term mean annual pre-

cipitation has been 835 mm, of which 75 % has occurred

Environmental Management

123

during the growing season (Hayden 1998). During the

course of this study (2006, 2007, 2008, 2009, and 2010),

total precipitation was 808, 1139, 1099, 983, and 847 mm,

of which 629, 772, 865, 706, and 672 mm fell during the

growing season (April through September). The coefficient

of variation (cv) for monthly precipitation was 109, 84, 85,

80, and 77 in 2006, 2007, 2008, 2009, and 2010, respec-

tively. Mean monthly temperature during the 2006, 2007,

2008, 2009, and 2010 growing seasons was 23, 22, 20, 19,

and 22 �C, respectively (NOAA National Climatic Data

Center 2012). The experimental area contained a Reading

silt loam (fine silt, mixed, superactive, and mesic Pachic

Argiudoll) soil formed by colluvial and alluvial deposits.

Historically, the study site would have been a native prairie

community, dominated by A. gerardii, S. nutans, and many

prairie forbs that contribute most to floristic diversity

(Freeman 1998). Prior to restoration, the field had been in

cultivation since the early twentieth century. Since 1976 the

field had been exclusively in Triticum aestivum L. (wheat),

Zea mays L. (maize), or Glycine max (L.) Merr. (soybean)

production. In the final year of cultivation, T. aestivum was

planted and harvested.

The same restoration experiment was established in a

more mesic climate, 620 km east-southeast of KPBS at the

Southern Illinois University Belleville Research Center

(BRC) (38�310N, 89�500W). Belleville, IL lies within a

longitudinal band that has received an average 200 mm

more annual precipitation relative to eastern KS (Lauren-

roth et al. 1999). During the course of this study, however,

the IL site received 142 mm more average annual precipi-

tation than Kansas. In 2006, 2007, 2008, 2009, and 2010

total precipitation at the BRC was 847, 827, 1444, 1367, and

1102 mm, of which 384, 450, 826, 693, and 762 mm fell

during the growing season (April through September). The

coefficient of variation (cv) for monthly precipitation was

45, 47, 60, 65, and 68 in 2006, 2007, 2008, 2009, and 2010,

respectively. Mean monthly temperature during the 2006,

2007, 2008, 2009, and 2010 growing seasons was 22, 22,

21, 21, and 23 �C (NOAA National Climatic Data Cen-

ter 2012). Soil at the Belleville site was a Cowden silt loam

(fine, smectitic, mesic, and Vertic Albaqualf). Prior to res-

toration, the field was recently in G. max, Z. mays, and T.

aestivum rotation. In the growing season just prior to

seeding of the experimental plots, Z. mays was planted and

harvested.

Experimental Design

Both experiments were established according to a split plot

design (Online Resource 1). Whole plots were assigned to

source of dominant grasses (cultivar or non-cultivar)

according to a randomized complete block design in

Kansas and according to a completely randomized design

in Illinois. Different whole plot treatment designs were

necessary due to different field dimensions in each loca-

tion. Sown diversity was the subplot treatment at both sites.

Cultivars of the dominant grasses were obtained from the

USDA, and we used those specifically recommended for

each region (USDA 1995). Cultivars of the grasses sown at

KPBS were A. gerardii ‘Kaw’, S. nutans ‘Osage’, and S.

scoparium ‘Camper’; non-cultivar seed sources of these

species were hand collected from remnant prairie popula-

tions at KPBS. The cultivars of A. gerardii ‘Rountree’, S.

nutans ‘Rumsey’, and S. scoparium ‘Aldous’ were used at

BRC. Due to limited remnant prairie near the Illinois site,

non-cultivar ‘‘Missouri ecotype’’ seed for each of these

three grass species was purchased from Hamilton Seed

Company (Elk Creek, MO, USA).

Each 25 m 9 5 m whole plot [KS: n = 12 (n = 6 per

block); IL: n = 10] was sown with a mixture of three

dominant grasses and fifteen other native species that occur

in tallgrass prairie. The same non-dominant species were

seeded at each site (Online Resource 2), with the exception

of Baptisia australis (sown in KS) and B. leucantha (sown

in IL) based on the respective distribution of these species.

Each whole plot contained five 5 m 9 5 m subplots sown

with consecutively lower densities of three dominant

grasses and increasing densities of fifteen other prairie

plants (2 non-dominant grasses and 13 forbs) (Online

Resource 2). The ratio of focal grass (equal amounts of A.

gerardii, S. nutans, and S. scoparium) to forb seed sown

varied among five subplots within each plot to create a

gradient of sown grass dominance and diversity. Percent

live seed (PLS) of cultivars and non-cultivars of A. ger-

ardii, S. nutans, and S. scoparium was determined by

Hulsey Seed Laboratory, Inc. (Decatur, GA, USA); PLS

was not determined for non-dominant species, but the same

source of seeds were used for each location so the live seed

of each species was constant. Subplot treatments of 97, 87,

60, 40, and 20 % sown grass dominance corresponded to a

total live focal grass:forb seeding rate of 585:15, 525:75,

450:150, 300:300, and 150:450 seeds/m2, and seed mix

Shannon diversity (H0) of 0.28, 0.59, 0.86, 1.40, 2.02,

respectively.

Restoration Approach

Seeds of each species were weighed out in the laboratory

prior to sowing and legume seeds were coated with genus

specific Rhizobia inoculum. Baptisia australis and B. leu-

cantha seeds were treated with Bradyrhizobium strains

UMR 7101 and 7102. Dalea candida was inoculated with

Saintfoin type F rhizobium. Desmanthus illinoensis was

inoculated with Rhizobium giardinii strain UMR 6029.

Lespedeza capitata was inoculated with Bradyrhizobium

Environmental Management

123

strain UMR 6564. Plots were sown in the winter to ensure

adequate cold stratification.

In the winter of 2005, the study sites were disked and

each subplot was sown by hand broadcasting seeds mixed

with damp sand. Immediately after sowing the dominant

grasses, plots were covered with large weave 7 ounce

burlap to maximize seed-soil contact, and minimize inter-

subplot dispersion of the light-weight grass seed. Burlap

was removed a month after sowing, prior to germination of

sown seed. Both sites were managed with annual pre-

scribed burns in the late winter or early spring.

Plant Community Response

In the center of each 5 m 9 5 m subplot, percent cover of

each species was estimated in two permanent 1 m 9 1 m

quadrats in the first (2006), second (2007), third (2008),

and fifth (2010) year of restoration. Cover was estimated in

the late spring (late May or early June) and late summer

(late August or September) each year. The maximum cover

value of each species recorded was used to calculate

community indices including Shannon–Wiener diversity

(H0), species density (number of species m-2), and cover of

specific groups within the community (i.e., dominant

grasses, sown forbs, and volunteer species).

Statistical Analyses

We analyzed the effects of dominant grass seed source and

sown grass dominance on total plant diversity, sown spe-

cies density, volunteer species density, sown forb density,

and cover of sown forbs, dominant grasses, and volunteer

species. Community responses to the fully factorial com-

bination of the sown grass:forb ratio (sown diversity: five

levels) and source (two levels) were analyzed according to

a split plot design with repeated measures using the mixed

model procedure (SAS Inc. 2003). Source and sown

diversity were the two fixed effects. Block was a random

effect to identify the appropriate error for the whole plot

factor in the split plot design. The mixed model procedure

was selected to analyze the data due to the split plot design,

which results in different error structures for whole-plot

and subplot factors. The most appropriate covariance

structure was selected based on the Akaike’s Information

Criterion, AIC (Littell et al. 2006). When appropriate,

community data were log or square root transformed prior

to statistical analysis in SAS to increase normality. Because

factorial designs with repeated measures can result in sig-

nificant interactions that might not be of interest (Milliken

and Johnson 1992), we used ‘‘contrast’’ and ‘‘estimate’’

statements in SAS to perform a priori comparisons of

interest. There were no significant three way interactions.

Significance was assigned at a = 0.05.

Results

Plant Cover

Focal Dominant Grass Species

Cover of the focal grass species increased over time in both

restoration sites (Fig. 1, Online Resource 2). After the first

year of restoration in Kansas, S. scoparium consistently had

less cover than A. gerardii or S. nutans. Sorghastrum nutans

was the dominant grass species in the non-cultivar plots, as

A. gerardii cover was\3 % throughout the first 5 years of

restoration. In Illinois, all three species of dominant grasses

were present in all years, but the site was co-dominated by

A. gerardii and S. nutans.

In Kansas, cover of the dominant grasses was affected by

an interaction between source and year (F3, 50 = 4.84;

P = 0.005) resulting from similar cover of the dominant

grasses during the first growing season (Contrast,

P = 0.097), but increasingly higher cover of the cultivar

source relative to the non-cultivar source in subsequent years

(Contrasts, P \ 0.008) (Fig. 1a). There was no effect of the

sown grass dominance treatment on the cover of dominant

grasses (F4, 44.6 = 0.67; P = 0.619) (Fig. 1c).

In Illinois, in contrast to Kansas, the cover of the dominant

grasses increased rapidly and stabilized between the third

and fifth year of restoration (Fig. 1b). There was a strong

main effect of sown grass dominance (F4, 38.7 = 9.81;

P \ 0.001), such that in all the years of this experiment,

dominant grass cover was higher in 97 % grass dominance

subplots than in the 20 % grass dominance subplots

(P \ 0.008) (Fig. 1d).

Sown Forb Cover

Sown forb cover exhibited different temporal patterns across

the longitudinal gradient (Fig. 2). In Kansas, sown forb

cover was affected by an interaction between source and year

(F3, 50 = 7.19; P \ 0.001) resulting from a gradual increase

in both sources during the first 3 years of restoration and a

relatively larger increase in forb cover within the non-cul-

tivar plots than the cultivar plots from 2008 to 2010 (Fig. 2a).

This difference in cover was attributed to an increase in three

species (Rudbeckia hirta, Monarda fistulosa, and Aster ob-

longifolius) in non-cultivar plots. Sown forb cover was also

affected by an interaction between sown grass dominance

and year (F12, 50 = 3.37; P = 0.001) (Fig. 2c). In all years,

sown forb cover was similar in the 20 and 40 % grass

dominance treatments and higher in these treatments than

any of the higher grass dominance treatments (P \ 0.05).

Over time, the absolute difference in sown forb cover

between subplots with the highest and lowest sown grass

dominance treatments increased. By 2010, sown forb cover

Environmental Management

123

in the 97 % sown grass dominance treatment was primarily

composed of A. oblongifolius (5 %) and R. hirta (3 %),

whereas sown forb cover in the 20 % sown grass dominance

treatment contained R. hirta (25 %), A. oblongifolius

(13 %), Baptisia australis (8 %), Lespedeza capitata (4 %),

and M. fistulosa (3 %).

In Illinois, sown forb cover only exhibited main effects

of sown grass dominance (F4, 50.6 = 62.60; P \ 0.001) and

year (F3, 116 = 63.15; P \ 0.001) (Fig. 2). Across both

sources, sown forb cover increased over time, but not

incrementally, as occurred in Kansas, to stabilize around

20 % of total plant cover (Fig. 2b). Averaged across all

years and sources, sown forb cover was higher in the 20

and 40 % subplots (34 ± 3 %) than all other sown grass

dominance treatments (Fig. 2d).

Volunteer Cover

Total cover of volunteers from the regional species pool

declined over time in both experiments, but the initial

cover of these species in Kansas was approximately twice

that in Illinois (Fig. 3). In Kansas, common volunteer

species varied temporally (Online Resource 2). Total vol-

unteer cover was affected by an interaction between source

and year (F3, 50 = 3.85; P = 0.015) resulting from a

steeper decline in the cover of volunteers over time in the

cultivar plots relative to the non-cultivar plots (Fig. 3a).

Over all years and sources of dominant grasses, volunteer

cover was affected by the subplot sown grass dominance

treatment (F4, 38.8 = 3.12; P = 0.025), with increasing

volunteer cover corresponding to increasing sown grass

dominance (Fig. 3c).

Volunteers from the regional species pool were also

temporally dynamic in Illinois, but species with the highest

cover were different than in Kansas (Online Resource 2).

As in Kansas, the cover of volunteers was affected by an

interaction between source and year (F3, 113 = 3.58;

P = 0.016), but the relative decline over time and differ-

ence in cover of volunteers between sources was less

pronounced relative to the Kansas experiment (Fig. 3b).

Fig. 1 Focal grass cover response to dominant grass source and sown

dominance in a and c Kansas and b and d Illinois. An asterisk

indicates an effect of source within a year (contrast P \ 0.05). Letters

a–d indicate differences among years within a source; means

accompanied by the same letter were not significantly different

(P [ 0.05). Means encompassed by an ellipse were significantly

different from means within a different ellipse within a year

(P \ 0.05)

Environmental Management

123

Volunteer cover was also affected by sown grass domi-

nance (F4, 46.6 = 5.32; P = 0.001). Averaged across all

years and both sources, volunteer cover was higher in the

highest grass dominance treatment than in any other sub-

plot treatments (Fig. 3d).

Plant Species Density

Total Species Density

Although nearly twice as many species initially colonized

the restoration experiment in Illinois compared to Kansas,

total species density was nearly identical between the two

sites following 5 years of restoration (Fig. 4). A total of 53,

46, 58, and 59 species were observed in the Kansas

experiment from 2006 through 2010, respectively. Of these

species, *30 % were sown and the remainder established

from the volunteer (regional) species pool. In Kansas, total

species density was affected by an interaction between

source and year (F3, 150 = 4.4; P = 0.005) (Fig. 4a). Total

species density was also affected by an interaction between

source and sown grass dominance (F4, 40 = 6.04;

P \ 0.001). In cultivar and non-cultivar plots, total species

density was higher in subplots 20, 40, and 60 than in the

97 % grass dominance subplots. Total plant species density

was higher in cultivar than non-cultivar treatments of 20

and 87 % sown grass dominance (Contrast, P \ 0.010),

although species density was higher in the non-cultivar

than cultivar 97 % sown grass dominance treatment

(Contrast, P = 0.049). An interaction between sown

grass dominance and year also affected total species den-

sity (F12, 150 = 2.25; P = 0.012). In all years except 2010,

total species density in Kansas was higher in the 20 %

sown grass dominance subplot than in the 97 % sown

grass dominance subplot (Contrast, P \ 0.005). Addition-

ally, as total species density declined with time, the number

of species in each subplot treatment converged, becoming

more similar. In 2010, total species density was higher in

Fig. 2 Sown forb cover response to dominant grass source and sown

dominance in a and c Kansas and b and d Illinois. An asterisk

indicates an effect of source within a year (contrast P \ 0.05). Letters

a–d indicate differences among either year within a source or

dominance treatments across all years; means accompanied by the

same letter were not significantly different (P [ 0.05). Means

encompassed by an ellipse were significantly different from means

within a different ellipse within a year (P \ 0.05)

Environmental Management

123

the 20 % sown grass dominance subplot than in the 87 %

sown grass dominance subplot, but was otherwise similar

among all subplot treatments (Fig. 4c).

In Illinois, a total of 66, 55, 49, and 38 species were

observed, and 22, 26, 28, and 33 % of these were planted

species in 2006, 2007, 2008, and 2010, respectively. Total

species density at this site declined with time (F3, 40 =

507.72; P \ 0.001) similarly in cultivar and non-cultivar plots

(source 9 year F3, 40 = 1.80; P = 0.162) (Fig. 4b) and

averaged across all years was lowest in the 97 % sown

grass dominance subplot treatment (P \ 0.001) (Fig. 4d).

Volunteer Species Density

Despite initial differences, both field sites contained fewer

than six volunteer species after 5 years of restoration. In

Kansas, volunteer species density was affected by an

interaction between source and year (F3, 50 = 3.48;

P = 0.023) resulting from similar volunteer species den-

sity among the grass sources during the first 3 years of

restoration, until 2010, when there were more volunteer

species in the non-cultivar than cultivar plots (Online

Resource 3A). In Illinois, volunteer species density

declined each year (F3, 40 = 447.52; P \ 0.001), similarly

among sources (source 9 year F3, 40 = 1.44; P = 0.245)

(Online Resource 3B).

Sown Species Density

Overall, more species that were sown in the Illinois resto-

ration experiment established and persisted over time rela-

tive to Kansas (Fig. 5). The density of sown species in

Kansas was affected by source, sown grass dominance, and

time. A source by year interaction for sown species density

(F3, 143 = 2.97; P = 0.034) resulted from similar species

density in year 1 between the sources, and a decline in spe-

cies density in the non-cultivar plots over time that did not

occur in the cultivar plots to result in higher sown species

density within cultivar plots (Fig. 5a). An interaction

between source and sown grass dominance also affected

Fig. 3 Total volunteer cover response to dominant grass source and sown dominance in a and c Kansas and b and d Illinois. Means with a

different letter were different (P \ 0.05), and an asterisk indicates an effect of source (P \ 0.05)

Environmental Management

123

sown species density (F4, 54.5 = 2.94; P = 0.028) (Fig. 5c).

Sown species density was higher in subplots sown with the

lowest grass dominance than in subplots sown with the

highest grass dominance treatment. In subplots sown with 20

and 87 % grass, sown species density was higher in treat-

ments containing cultivars of the dominant grasses.

In Illinois, sown species density declined similarly in

cultivar and non-cultivar treatments (source F1, 44.2 = 0.00;

P = 0.990) over time (F3, 110 = 31.08; P \ 0.001) for both

dominant grass sources (F3, 110 = 0.85; P = 0.471)

(Fig. 5b). Sown species density in Illinois was affected by

an interaction between sown grass dominance and year

(F12, 110 = 2.80; P = 0.002) (Fig. 5d). In the 20, 40, 60, and

87 % sown grass dominance treatments, sown species den-

sity declined over time, whereas sown species density

remained steady in the 97 % sown grass dominance subplot

treatment across time. In all years, sown species density was

consistently higher in the 20 and 40 % grass dominance

subplots than in all other sown grass dominance treatments.

Total Plant Diversity

Despite different temporal dynamics during the first

3 years of restoration, total plant diversity in both sites

converged to similar levels by the fifth growing season

(Fig. 6). Further, diversity in cultivar plots was either

greater than or equivalent to diversity in non-cultivar plots

over the study period in both locations (Fig. 6). In Kansas,

total plant diversity was affected by an interaction between

source and year (F3, 50 = 3.70; P = 0.018) resulting from

similar diversity between the sources in all, but one year

(Fig. 6a). In the second year of restoration, diversity was

higher in cultivar than non-cultivar treatment. There was

no interaction between sown grass dominance and year for

diversity (F12, 50 = 1.66; P = 0.104) resulting from con-

sistent change in the sown grass dominance treatments

over time (Fig. 6c). Diversity was affected by an interac-

tion between source and sown grass dominance (F4, 39.8 =

2.69; P = 0.045). In both cultivar and non-cultivar plots,

Fig. 4 Total plant species density in Kansas as affected by a grass

source and c sown dominance treatments. Total plant species density

in Illinois as affected by b grass source and d sown grass dominance

treatments. In a, b an asterisk indicates an effect of source, and letters

a–d indicate an effect of time (P \ 0.05). In c, means encompassed

by an ellipse were significantly different from means within a

different ellipse within a year (P \ 0.05). In d, means with a different

letter are different (P \ 0.05)

Environmental Management

123

diversity declined with increased sown grass dominance

treatments. However, in both the 20 % grass dominance

subplot treatment and the 87 % grass dominance subplot

treatment, diversity was higher in plots sown with cultivars

(Online Resource 4A).

In Illinois, total plant diversity declined over time in both

sources, but an interaction between source and year resulted

from higher diversity in cultivar plots than non-cultivar

plots in 2007 and 2008 (source 9 year: F3, 113 = 1.78;

P = 0.154) (Fig. 6b). Diversity was also affected by

an interaction between sown grass dominance and year

(F12, 113 = 2.30; P = 0.012) (Fig. 6d). Diversity was con-

sistently lower in subplots sown with 97 % grass than in

subplots sown with 20, 40, or 60 % grass (Contrast,

P \ 0.040). Unlike the Kansas restoration, diversity was

not affected by an interaction between source and sown

grass dominance (F4, 57.7 = 0.47; P = 0.760) in Illinois

(Online Resource 4B).

Discussion

Using two field experiments, this study aimed to elucidate

the generality or contingency of potential deterministic

biological filters including dominant grass source (cultivar

vs. non-cultivar), and sown grass dominance (grass: forb

sowing ratio) on the assembly of plant communities in

response to ecological restoration (Table 1). The effect of

dominant grass source on many cover groups was contin-

gent upon location. This filter operated more on cover

groups in the western region of tallgrass prairie relative to

the east, where climate was less variable and annual rain-

fall was more abundant. Plant species density and diversity

response to grass source exhibited similar patterns across

sites, indicating a convergent pattern in the effect of cul-

tivars on these community metrics. Results from this study

did not support our hypothesis that species richness (den-

sity) and diversity would be lower in the presence of

Fig. 5 Sown species density in response to dominant grass source

and sown dominance in a and c Kansas and b and d Illinois. An

asterisk indicates an effect of source within a year (contrast

P \ 0.05). Letters a–c indicate differences among years within a

source or between grass dominance treatments of the same grass

source; means accompanied by the same letter were not significantly

different (P [ 0.05). Means encompassed by an ellipse were signif-

icantly different from means within a different ellipse within a year

(P \ 0.05)

Environmental Management

123

cultivars of prairie grasses, which we had surmised would

be more competitive than non-cultivar sources of the

grasses. In fact, diversity and species density in the cultivar

source treatment was greater than or equal to diversity and

species density in non-cultivar sources. We recognize that

the limited effects of dominant grass source are restricted

to these community metrics, as potential differences in

genetic structure observed between cultivar and non-culti-

var populations of prairie grasses (Gustafson et al. 2004)

could have implications for adaptive potential of restored

populations and do not discount concerns about outbreed-

ing depression from the use of cultivars in restoration

(Hufford and Mazer 2003). The effect of sown grass

dominance on cover of the focal grass species was also

contingent upon location resulting from establishment

inconsistent with the sown dominance treatments in Kansas

and establishment corresponding better to the assigned

dominance treatments in Illinois. In both sites, the diversity

and species density were higher when sown diversity was

highest, a restoration practice often employed by

practitioners, but not empirically well documented (but see

Carter and Blair 2012b).

The dynamic filter model of community assembly

(Keddy 1992) provides a valuable framework to concep-

tualize and test the role of environmental conditions and

biological interactions on the development of communities

(Hobbs and Norton 1996, 2004). This model predicts that

community membership (species present) is a subset of a

larger regional species pool resulting from interspecific

variation in species tolerances to conditions that affect the

ability of each species to pass through the abiotic filter, and

the strength of interspecific interactions that influences

persistence of species over time (Gibson et al. 2012).

Regional variation in environmental conditions may pre-

scribe contrasting communities to develop despite using

similar practices (Hilderbrand et al. 2005; Paradeis et al.

2010) due to variation in the strength of ecological filters

(Fattorini and Halle 2004; Hobbs and Norton 2004). Strong

regional variation in the strength of ecological filters would

likely result in highly contingent reassembled community

Fig. 6 Shannon’s diversity response to dominant grass source in

a Kansas, and b Illinois. The response of diversity to sown grass

dominance in c Kansas, and d Illinois. An asterisk indicates an effect

of source, and in a–d letters a–d indicate an effect of time (P \ 0.05).

In c–d, data within an ellipse is different from data within another

ellipse of the same year (P \ 0.05)

Environmental Management

123

structure. If community assembly was a process entirely

contingent upon regional variation and stochastic events,

community ecology would not be a predictive science

(Lawton 1999; Bradford et al. 2012). The two restoration

experiments conducted using the same approaches in

Kansas and Illinois demonstrates the importance of abiotic

and biotic filters on modulating community assembly and

the value of multi-site restoration studies to elucidate both

convergent and contingent assembly patterns (Table 1).

Despite documented differences in traits of cultivar and

non-cultivar sources of the dominant grasses used in prairie

restoration (Klopf and Baer 2011; Lambert et al. 2011),

there is no evidence that trait variation affects ecosystem

functioning (Wilsey 2010; Baer et al. 2013) or community

structure (Gibson et al. 2013) in restored prairie. In this

study the effects of the dominant grass population source on

the cover of these dominant grasses, sown species, and

volunteer species were contingent upon location and for all

cover metrics. Differences in community metrics between

dominant grass sources were only observed in Kansas.

Environmental conditions at the time of establishment may

be especially critical filters on grassland community reas-

sembly (MacDougall et al. 2008). A potentially harsher

abiotic filter in Kansas during the establishment year may

have exacerbated site based differences between dominance

grass sources, as cultivars may be more capable of

successful establishment under harsh conditions because

they were developed to stabilize degraded soil and are often

sought for drought tolerance (USDA 1995). In June of 2006,

precipitation at Konza Prairie totaled 3.66 cm (6 cm below

the 1971–2000 average), the average maximum tempera-

ture was 33 �C, and temperatures exceeded 32 �C for

17 days (NOAA National Climatic Data Center 2012).

Comparatively, in June of 2006 at the Belleville, Illinois

field site, precipitation totaled 8.03 cm, the average maxi-

mum temperature was 31 �C, and temperatures exceeded

32 �C for only 9 days (NOAA National Climatic Data

Center 2012). The large disparity in establishment success

of A. gerardii cultivars and non-cultivars in Kansas likely

reflects phenotypic differences between these two source

populations, corroborated by no survival of this species in a

previous study aimed to quantify belowground traits in the

first year of this restoration (Klopf and Baer 2011). The

difference in dominant grass cover between the two popu-

lation sources likely modified available niche space (Gibson

et al. 2012) and variation in community structure (higher

cover of sown forbs and volunteer species in plots sown

with non-cultivars of the dominant grasses) in Kansas, but

not in Illinois.

Unlike plant cover, total species density and diversity

showed convergent patterns in response to dominant grass

source. Contrary to our hypotheses, total species density

Table 1 Summary of whether

the manipulated biotic filter

effects on plant cover, species

density, and diversity were

regionally similar (i.e.,

convergent) or different (i.e.,

contingent) between the two

field experiments in Kansas and

Illinois based on dominant grass

seed source (cultivar or non-

cultivar) and sown grass

dominance

Community metric Response to manipulated biotic filter

Following 5 years of community assembly

Dominant grass source Sown grass dominance

Sown dominant grass

cover

Contingent: variation between sources in

KS; higher cover of cultivar grasses

Contingent: significant variation in

response to sown dominance in IL

Sown forb cover Contingent: variation between sources in

KS; higher with non-cultivar grasses

Convergent: forb cover highest in 20

and 40 % sown grass dominance

Volunteer species

cover

Contingent: Variation between sources in

KS: higher with non-cultivar grasses

Convergent: Higher volunteer cover

with high grass dominance

Species density Convergent: higher species density with

cultivars in one year (KS), but no

difference between sources and similar

species density in KS and IL by year 5

Convergent: higher species density in

the 20 and 40 % sown grass

dominance treatments

Sown species density Contingent: variation between sources in

KS; higher with cultivar grasses

Convergent and contingent: sown

species density higher with lower

grass dominance in KS and IL: sown

grass dominance interacted with

grass source (higher with cultivars in

some dominance treatments) in KS

Volunteer species

density

Contingent: variation between sources in

KS; higher species density with non-

cultivar grasses

Convergent: no significant effect

Diversity Convergent: total plant diversity higher in

some years with cultivar grasses; no

difference between sources in year 5

Convergent and contingent: total plant

diversity higher with lower sown

grass dominance in KS and IL; sown

grass dominance interacted with

grass source (higher with cultivars in

some dominance treatments) in KS

Environmental Management

123

and diversity were not adversely affected by cultivars,

which we surmised would be more competitive, and in

some years total species density and diversity were higher

with cultivars in both locations. There were site-contingent

patterns in sown and volunteer species density that resulted

from no differences in response to dominant grass source in

Illinois, whereas volunteer species density was higher with

non-cultivars and sown species density was higher with

cultivars of the focal grasses in Kansas. Weedy species can

compete with sown restored species and inhibit the estab-

lishment of prairie species (Blumenthal et al. 2003; Dick-

son and Busby 2009). It is possible that the higher cover of

grasses ameliorated harsh environmental conditions to

facilitate the persistence of more species (Smith et al.

2004) in prairie restored with grass cultivars, and lower

cover of non-cultivar grasses increased potential space and

resources for volunteer species. These results suggest cul-

tivars of the dominant grasses used in this experiment may

more effectively meet restoration goals of establishing C4

grasses and low weed cover than non-cultivars without

compromising diversity or species density of target species.

The community similarities between cultivar and non-

cultivar plots in Illinois could be attributed to environ-

mental conditions or genetic factors. Conditions in Illinois

during the first year of restoration (with ample precipita-

tion) could have dampened the expression of intraspecific

differences between sources of the dominant grasses.

Alternatively, the ‘‘Missouri Ecotype’’ used in the Illinois

experiment, produced commercially, may have traits sim-

ilar to the cultivars as a result of unintentional selection

during the seed production process (Montalvo et al. 1997);

but this would impose a selection on phenology rather than

traits of fitness and vigor that are selected during cultivar

development. A separate regional experiment in Carbon-

dale, IL, USA compared the same cultivars used in this

study to non-cultivar sources of the same dominant grasses

using local ecotypes collected from remnant prairies with

documented functional differences between the sources

(Lambert et al. 2011) and found limited effects of dominant

grass source on plant community composition (Gibson

et al. 2013) and ecosystem processes (Baer et al. 2013).

The sown grass dominance treatment effects on the

cover of the focal species were contingent on location

resulting from establishment corresponding better to the

assigned treatments in Illinois relative to Kansas, but the

cover of sown and volunteer species were regionally con-

vergent. In both locations, lower seeding rates of grasses

correspond with higher cover of sown forbs and lower

cover of volunteer species. Multiple studies have attributed

low forb establishment to the inhibitory effects of high

grass dominance (Weber 1999; Kindscher and Frazer 2000;

Dickson and Busby 2009), and this study provides justifi-

cation for enhanced effort to include a greater proportion of

forb propagules in prairie restoration to promote floristic

diversity. It is important to note, however, that the greater

sown forb cover in non-cultivar plots (across all dominance

treatments) in Kansas during the fifth growing season

resulted from a surge in the cover of a single species,

Rudbeckia hirta.

By the 5th year of restoration in Kansas and Illinois,

diversity and species density were greater in the low grass

dominance subplot treatments. However, diversity was

greater in low grass dominance treatments during the entire

study in Illinois, but only in two of the four years in

Kansas. Furthermore in 2010, the magnitude of the dif-

ference between diversity of the highest and lowest

diversity treatments was 2.8 times greater in Illinois than

Kansas likely resulting from the stronger correspondence

between sown grass dominance and the cover of the focal

grass species in Illinois, which could have been related to

contrasting climate conditions during the establishment

year. Thus, the acclaimed potential value of restorations to

test ecological theory (e.g., deterministic role of restored

species density or diversity on the functioning of restored

systems) (Bradshaw 1987), could be complicated by sto-

chastic environmental variation.

Implications

Using locally sourced propagules to restore plant commu-

nities has been recommended to avoid introducing geno-

types that may be maladapted to the local environment or

catalysts for outbreeding depression (Dyer and Rice 1997;

Montalvo et al. 1997; Gordon and Rice 1998; Keller et al.

2000; Hufford and Mazer 2003), which could undermine

restoration objectives (Templeton 1986; Keller et al. 2000;

Hufford and Mazer 2003; McKay et al. 2005). However, not

all populations may be locally adapted (e.g., high gene flow

reduces local adaptation) and the importance of locally

adapted genotypes to the long-term success of restoration

remains largely unknown, particularly the adaptive poten-

tial needed to withstand environmental change (Rice and

Emery 2003; Harris et al. 2006; Broadhurst et al. 2008).

Cultivars have generally been discouraged and recom-

mended for use only in extremely degraded sites that are

small in spatial extent (Lesica and Allendorf 1999). The

high availability and low cost of prairie grass cultivars,

present a difficult decision at the onset of restoration,

exacerbated by the extremely limited area for collection of

local ecotypes in most of the historic extent of this eco-

system (e.g., Illinois), which has largely been converted to

agriculture (Manning 1995; Samson et al. 2004). Despite

the limited empirical information on the ecological conse-

quences of using grass cultivars to re-establish prairie, many

restoration practitioners perceive dominant grass cultivars

as competitively superior, and thus prefer forb-rich seed

Environmental Management

123

mixes of local ecotypes (Erickson and Navarrete-Tindall

2004; Rowe 2010). This study demonstrates that dominant

grass source effects on target (sown) and non-target (vol-

unteer) species cover and density are contingent by loca-

tion, but broader community metrics of total plant diversity

and richness were generally not adversely affected by cul-

tivars of the most common prairie grasses.

In closing, successful restoration of diverse prairie can

be extremely challenging and unexpected outcomes can

arise (Cottam 1987; Weber 1999). An important lesson

from this research is that regional replication is needed to

elucidate the generality of potential biotic filters on

developing communities. Ecosystems encompass large,

heterogeneous landscapes. The two field experiments were

located at similar latitude, on silt loam soils with a similar

history of agricultural degradation, and restored using the

same species and field practices. Regional environmental

variation in climate, soil microflora, composition of vol-

unteer species, and plant pests were expected. Though not

all quantifiable, some combination of these variables

inherent to this variation influenced the relative abundance

(cover) of species, but total species density and diversity

converged across sites. Regional replication is needed to

develop broadly applicable guidelines for restoration.

Acknowledgments Field and laboratory assistance were provided

by Elizabeth Bach, Ryan Campbell, Rachel Goad, Allison Lambert,

Lewis Reed, Jason Willand, Ben Wodika, Bryan Young, staff at both

the Konza Prairie Biological Station and the Southern Illinois Uni-

versity Carbondale Belleville Center, and many undergraduate stu-

dents. This research was funded by the Konza Prairie Long Term

Ecological Research program and the National Science Foundation

(DEB 0516429).

References

Baer SG, Gibson DJ, Benscoter AM, Reed LK, Campbell RE, Klopf

RP, Willand JE, Wodika BR. (2013) No effect of seed source on

multiple aspects of ecosystem functioning during ecological

restoration: cultivars compared to local ecotypes of dominant

grasses. Evol Appl. doi:10.1111/eva.12124

Baer SG, Blair JM, Collins SL, Knapp AK (2004) Plant community

responses to resource availability and heterogeneity during

restoration. Oecologia 139:617–629

Betz RF (1986) One decade of research in prairie restoration at the

Fermi National Accelerator Laboratory (Fermilab), Batavia,

Illinois. In Clambey GK and Pemble RH (eds) The prairie: past,

present, and future. Proceedings of Ninth North American Prairie

Conference, Moorhead State University, Morehead, MN. Tri-

College University Center for Environmental Studies, Fargo,

North Dakota, pp 179–185

Blumenthal DM, Jordan NR, Russelle MP (2003) Soil carbon addition

controls weeds and facilitates prairie restoration. Ecol Appl

13:605–615

Bradford MA, Wood SA, Maestre FT, Reynolds JF, Warren RJ II (2012)

Contingency in ecosystem but not plant community response to

multiple global change factors. N Phytol 196:462–471

Bradshaw AD (1965) Evolutionary significance of phenotypic

plasticity in plants. Adv Genet 13:115–156

Bradshaw AD (1987) Restoration: an acid test for ecology. In: Jordan

W, Gilpin M, Aber J (eds) Restoration ecology: a synthetic

approach to ecological research. Cambridge University Press,

Cambridge, pp 23–30

Broadhurst LM, Lowe A, Coates DJ, Cunningham SA, McDonald M,

Vesk PA, Yates C (2008) Seed supply for broadscale restoration:

maximizing evolutionary potential. Evol Appl 1:587–597

Carter DL, Blair JM (2012a) Recovery of native plant community

characteristics on a chronosequence of restored prairies seeded

into pastures in west-central Iowa. Restor Ecol 20:170–179

Carter DL, Blair JM (2012b) High richness and dense seeding

enhance grassland restoration establishment, but have little effect

on drought response. Ecol Appl 22:1308–1319

Clausen J and Hiesey WM (1958) Experimental studies on the nature

of species. IV. Genetic structure of ecological races. Publication

#615, Carnegie Institution of Washington, Washington, DC

Cottam G (1987) Community dynamics on an artificial prairie. In:

Jordan WR III, Gilpin ME, Aber JD (eds) Restoration ecology: a

synthetic approach to ecological research. Cambridge University

Press, Cambridge, pp 257–270

Dickson TL, Busby WH (2009) Forb species establishment increases

with decreased grass seeding density and with increased forb

seeding density in a Northeast Kansas, USA, experimental

prairie restoration. Restor Ecol 17:597–605

Doherty JM, Callaway JC, Zedler JB (2011) Diversity-function

relationships changed in a long-term restoration experiment.

Ecol Appl 21:2143–2155

Dyer AR, Rice KJ (1997) Evidence of spatial genetic structure in a

California bunchgrass population. Oecologia 112:333–339

Erickson B, Navarrete-Tindall NE (2004) Missouri native ecotype

program: increasing local-source native seed. Nat Areas J

24:15–22

Falk DA (1990) Integrated strategies for conserving plant genetic

diversity. Ann Mo Bot Gard 77:38–47

Fattorini M, Halle S (2004) The dynamic environmental filter model:

how do filtering effects change in assembling communities after

disturbance? In: Temperton VM, Hobbs RJ, Nuttle T, Halle S

(eds) Assembly rules and restoration ecology. Island Press,

Washington, pp 96–114

Fehr WR (1987) Principles of cultivar development: theory and

technique. Macmillan Publishing Company, New York

Freeman CC (1998) The flora of Konza Prairie: a historical review

and contemporary patterns. In: Knapp AK, Briggs JM, Hartnett

DC, Collins SL (eds) Grassland dynamics: long-term ecological

research in tallgrass prairie. Oxford Press, New York, pp 69–80

Gibson DJ, Allstadt AJ, Baer SG, Geisler M (2012) Effects of

foundation species genotypic diversity on subordinate species

richness in an assembling community. Oikos 121:496–507

Gibson DJ, Baer SG, Klopf RP, Reed LK, Wodika BR, Willand JE

(2013) Limited effects of dominant species population source on

community composition during community assembly. J Veg Sci

24:429–440

Gordon DR, Rice KJ (1998) Patterns of differentiation in wiregrass

(Aristida beyrichiana): implications for restoration efforts.

Restor Ecol 6:166–174

Gustafson DJ, Gibson DJ, Nickrent DL (2004) Conservation genetics

of two co-dominant grass species in an endangered grassland

ecosystem. J Appl Ecol 41:389–397

Hansen MJ and Gibson DJ (2013) Use of multiple criteria in an

ecological assessment of a prairie restoration chronosequence.

Appl Veg Sci, in press doi:10.1111/avsc.12051

Harris JA, Hobbs RJ, Higgs E, Aronson J (2006) Ecological

restoration and global climate change. Restor Ecol 14:170–176

Environmental Management

123

Hayden BP (1998) Regional climate and the distribution of tallgrass

prairie. In: Knapp AK, Briggs JM, Hartnett DC, Collins SL (eds)

Grassland dynamics: long-term ecological research in tallgrass

prairie. Oxford Press, New York, pp 19–34

Hilderbrand RH, Watts AC, Randle AM (2005) The myths of

restoration ecology. Ecol Soc 10:19

Hobbs RJ, Norton DA (1996) Towards a conceptual framework for

restoration ecology. Restor Ecol 4:93–110

Hobbs RJ, Norton DA (2004) Ecological filters, thresholds, and

gradients in resistance to ecosystem reassembly. In: Temperton

VM, Hobbs RJ, Nuttle T, Halle S (eds) Assembly rules and

restoration ecology. Island Press, Washington, pp 72–95

Hufford KM, Mazer SJ (2003) Plant ecotypes: genetic differentiation

in the age of ecological restoration. Trends Ecol Evol 18:147–155

Huxman TE, Smith MD, Fay PA et al (2004) Convergence across

biomes to a common rain-use efficiency. Science 429:651–654

Jackson ST, Hobbs RJ (2009) Ecological restoration in the light of

ecological history. Science 325:567–569

Keddy PA (1992) Assembly response and rules—2 goals for

predictive community ecology. J Veg Sci 3:157–164

Keller M, Kollman J, Edwards PJ (2000) Genetic introgression from

distant provenances reduces fitness in local weed populations.

J Appl Ecol 37:647–659

Kindscher K, Frazer A (2000) Planting forbs first provides greater

species diversity in tallgrass prairie restorations (Kansas). Ecol

Restor 18:115–116

Klopf RK, Baer SG (2011) Root dynamics of cultivar and non-

cultivar population sources of two dominant grasses during

initial establishment of tallgrass prairie. Restor Ecol 19:112–117

Lambert AM, Baer SG, Gibson DJ (2011) Intraspecific variation in

ecophysiology of three dominant prairie grasses used in resto-

ration: cultivar versus non-cultivar population sources. Restor

Ecol 19:43–52

Laurenroth WK, Burke IC, Gutmann MP (1999) The structure and

function of ecosystems in the central North American grassland

region. Gt Plains Res 9:223–259

Lawton JH (1999) Are there general laws in ecology? Oikos

84:177–192

Lesica P, Allendorf FW (1999) Ecological genetics and the restora-

tion of plant communities: mix or match? Restor Ecol 7:42–50

Littell RC, Milliken GA, Stroup WW, Wolfinger RD, Schabenberger

O (2006) SAS for mixed models, 2nd edn. SAS Institute Inc.,

Cary

MacDougall AS, Wilson SD, Bakker JD (2008) Climatic variability

alters the outcome of long-term community assembly. J Ecol

96:346–354

Manning R (1995) Grassland: the history, biology, politics, and

promise of the American Prairie. Penguin Books, New York

McCain KNS, Baer SG, Blair JM, Wilson GWT (2010) Dominant

grasses suppress local diversity in restored tallgrass prairie.

Restor Ecol 18:40–49

McKay JK, Christian CE, Harrison S, Rice KJ (2005) ‘‘How local is

local?—a review of practical and conceptual issues in the

genetics of restoration. Restor Ecol 13:432–440

Milliken, GA and Johnson DE (1992) Analysis of messy data. Vol 1:

designed experiments. Chapman & Hall, London

Montalvo AM, Williams SL, Rice KJ, Buchmann SL, Cory C, Handel

SN, Nabhan GP, Primack R, Robichaux RH (1997) Restoration

biology: a population biology perspective. Restor Ecol

5:277–290

National Oceanic and Atmospheric Administration. National Climatic

Data Center. Asheville, NC, USA. http://www.ncdc.noaa.gov/.

Accessed 21 Oct 2012

Paradeis BL, DeKeyser ES, Kirby DR (2010) Evaluation of restored

and native pothole region plant communities following an

environmental gradient. Nat Areas J 30:294–304

Piper JK, Schmidt ES, Janzen AJ (2007) Effects of species richness

on resident and target species components in a prairie restora-

tion. Restor Ecol 15:189–198

Polley HW, Derner JD, Wilsey BJ (2005) Patterns of plant species

diversity in remnant and restored tallgrass prairies. Restor Ecol

13:480–487

Rice KJ, Emery NC (2003) Managing microevolution: restoration in

the face of global change. Front Ecol Environ 1:469–478

Rowe HI (2010) Tricks of the trade: techniques and opinions from 38

experts in tallgrass prairie restoration. Restor Ecol 18:253–262

Samson FB, Knopf FL, Ostlie WR (2004) Great plains ecosystems:

past, present, and future. Wildl Soc Bull 32:6–15

Skeel VA, Gibson DJ (1996) Physiological performance of Andro-

pogon gerardii, Panicum virgatum, and Sorghastrum nutans on

reclaimed mine spoil. Restor Ecol 4:355–367

Sluis WJ (2002) Patterns of species richness and composition in re-

created grassland. Restor Ecol 10:677–684

Smith MD, Wilcox JC, Kelly T, Knapp AK (2004) Dominance not

richness determines invasibility of tallgrass prairie. Oikos

106:253–262

Taft JB, Hauser C, Robertson KR (2006) Estimating floristic integrity

in tallgrass prairie. Biol Conserv 131:42–51

Templeton AR (1986) Coadaptation and outbreeding depression. In:

Soule M (ed) Conservation biology: the science of scarcity and

diversity. Sinauer Associates Inc., Sunderland, pp 105–116

United States Department of Agriculture (USDA) (1995) Grass

varieties in the United States. CRC Press, Boca Raton

Weber S (1999) Designing seed mixes for prairie restorations:

revisiting the formula. Ecol Restor 17:196–201

Wilsey BJ (2010) Productivity and subordinate species response to

dominant grass species and seed source during restoration.

Restor Ecol 18:628–637

Environmental Management

123