Soil heterogeneity increases plant diversity after 20 years of manipulation during grassland restoration

Abstract

The “environmental heterogeneity hypothesis” predicts that variability in resources promotes species coexistence, but few experiments support this hypothesis in plant communities. A previous 15‐yr test of this hypothesis in a prairie restoration experiment demonstrated a weak effect of manipulated soil resource heterogeneity on plant diversity. This response was attributed to a transient increase in richness following a post‐restoration supplemental propagule addition, occasionally higher diversity under nutrient enrichment, and reduced cover of a dominant species in a subset of soil treatments. Here, we report community dynamics under continuous propagule addition in the same experiment, corresponding to 16–20 yr of restoration, in response to altered availability and heterogeneity of soil resources. We also quantified traits of newly added species to determine if heterogeneity increases the amount and variety of niches available for new species to exploit. The heterogeneous treatment contained a factorial combination of altered nutrient availability and soil depth; control plots had no manipulations. Total diversity and richness were higher in the heterogeneous treatment during this 5‐yr study due to higher cover, diversity, and richness of previously established forbs, particularly in the N‐enriched subplots. All new species added to the experiment exhibited unique trait spaces, but there was no evidence that heterogeneous plots contained a greater variety of new species representing a wider range of trait spaces relative to the control treatment. The richness and cover of new species was higher in N‐enriched soil, but the magnitude of this response was small. Communities assembling under long‐term N addition were dominated by different species among subplots receiving added N, leading to greater dispersion of communities among the heterogeneous relative to control plots. Contrary to the deterministic mechanism by which heterogeneity was expected to increase diversity (greater variability in resources for new species to exploit), higher diversity in the heterogeneous plots resulted from destabilization of formerly grass‐dominated communities in N‐enriched subplots. While we do not advocate increasing available soil N at large scales, we conclude that the positive effect of environmental heterogeneity on diversity can take decades to materialize and depend on development of stochastic processes in communities with strong establishment limitation.

Full text

Soil heterogeneity increases plant diversity after 20 years of
manipulation during grassland restoration
SARA G. BAER,
1,5
TIANJIAO ADAMS,
2
DREW A. SCOTT,
2
JOHN M. BLAIR,
3
AND SCOTT L. COLLINS
4
1
Kansas Biological Survey and Department of Ecology & Evolutionary Biology, University of Kansas, Lawrence, Kansas 66047 USA
2
Department of Plant Biology and Center for Ecology, Southern Illinois University, Carbondale, Illinois 62901 USA
3
Division of Biology, Kansas State University, Manhattan, Kansas 66506 USA
4
Department of Biology, University of New Mexico, Albuquerque, New Mexico 87131 USA
Citation: Baer, S. G., T. Adams, D. A. Scott, J. M. Blair, and S. L. Collins. 2019. Soil heterogeneity
increases plant diversity after 20 years of manipulation during grassland restoration. Ecological Applica-
tions 00(00):e02014. 10.1002/eap.2014
Abstract. The “environmental heterogeneity hypothesis”predicts that variability in
resources promotes species coexistence, but few experiments support this hypothesis in plant
communities. A previous 15-yr test of this hypothesis in a prairie restoration experiment
demonstrated a weak effect of manipulated soil resource heterogeneity on plant diversity. This
response was attributed to a transient increase in richness following a post-restoration supple-
mental propagule addition, occasionally higher diversity under nutrient enrichment, and
reduced cover of a dominant species in a subset of soil treatments. Here, we report community
dynamics under continuous propagule addition in the same experiment, corresponding to
16–20 yr of restoration, in response to altered availability and heterogeneity of soil resources.
We also quantified traits of newly added species to determine if heterogeneity increases the
amount and variety of niches available for new species to exploit. The heterogeneous treatment
contained a factorial combination of altered nutrient availability and soil depth; control plots
had no manipulations. Total diversity and richness were higher in the heterogeneous treatment
during this 5-yr study due to higher cover, diversity, and richness of previously established
forbs, particularly in the N-enriched subplots. All new species added to the experiment exhib-
ited unique trait spaces, but there was no evidence that heterogeneous plots contained a greater
variety of new species representing a wider range of trait spaces relative to the control treat-
ment. The richness and cover of new species was higher in N-enriched soil, but the magnitude
of this response was small. Communities assembling under long-term N addition were domi-
nated by different species among subplots receiving added N, leading to greater dispersion of
communities among the heterogeneous relative to control plots. Contrary to the deterministic
mechanism by which heterogeneity was expected to increase diversity (greater variability in
resources for new species to exploit), higher diversity in the heterogeneous plots resulted from
destabilization of formerly grass-dominated communities in N-enriched subplots. While we do
not advocate increasing available soil N at large scales, we conclude that the positive effect
of environmental heterogeneity on diversity can take decades to materialize and depend on
development of stochastic processes in communities with strong establishment limitation.
Key words: assembly; community; functional traits; niche; nitrogen; richness; tallgrass prairie.
INTRODUCTION
There is an escalating need to manage processes that
maintain and promote species richness as human-driven
environmental change continues to degrade ecosystems
and reduce biodiversity (Vitousek et al. 1997, MacDougall
et al. 2013). Global conversion of grasslands to agricul-
ture reduces native species richness, landscape hetero-
geneity (Ellis and Ramankutty 2008), and propagule
supply for community regeneration (Willand et al.
2013). Temperate grassland ranks highest in the ratio of
area converted to protected lands, and as such, this
biome is considered to be “in crisis”and a high conser-
vation priority (Hoekstra et al. 2005). Human interven-
tion is generally needed to restore grasslands from
agricultural conditions. Grassland restoration has been
practiced for over a century, with the first effort aimed at
reconstructing North American tallgrass prairie (Mlot
1990, Sperry 1994). Despite the long practice of prairie
restoration, many restored prairies (particularly on for-
merly cultivated land) suffer from diversity that declines
over time to levels less than extant prairies (Kindscher
and Tieszen 1998, Camill et al. 2004, Martin et al. 2005,
Polley et al. 2005, Willand et al. 2013, Hansen and
Gibson 2014, Barak et al. 2017, Bauer et al. 2017).
Manuscript received 28 November 2018; revised 9 July 2019;
accepted 30 August 2019. Corresponding Editor: Elisabeth
Huber-Sannwald.
5
E-mail: sgbaer@ku.edu
Article e02014; page 1
Ecological Applications, 0(0), 2019, e02014
©2019 by the Ecological Society of America

Declining diversity during grassland restoration is gener-
ally attributed to environmental filtering, increasing
dominance of grasses, and limited niche space for new
species to exploit (Polley et al. 2007, McCain et al. 2010,
Wilsey 2010, Klopf et al. 2014, Baer et al. 2016, Scott
and Baer 2018).
Numerous studies that show species richness
increases with environmental heterogeneity (reviewed
by Wilson 2014) and heterogeneity is one of many fac-
tors (and processes) operating simultaneously to con-
nect richness and productivity (Grace et al. 2016).
Available niche space acts as an environmental filter in
community assembly and is influenced by deterministic
drivers and stochastic processes acting at multiple
scales. For example, fine- and intermediate-scale spatial
heterogeneity have been shown to increase species den-
sity and diversity in plant communities (Richardson
et al. 2012, Williams and Houseman 2014). In tallgrass
prairie, varied ecological and geographical drivers pro-
mote large-scale heterogeneity in plant communities
(Tilman 1984, Briggs and Knapp 1995, Turner et al.
1997, Collins et al. 2018). At local scales, plant diver-
sity responds negatively to drivers that promote grass
dominance and cause local extinction of subordinate
species. For example, plant diversity in native prairie is
generally higher in shallow soils with limited rooting
depth, lower water availability, and lower grass domi-
nance (Gibson and Hulbert 1987, Bush and Van
Auken 2010, Collins and Calabrese 2012). Alterna-
tively, nitrogen enrichment initially increases grass
dominance and reduces plant diversity (Turner et al.
1997, Collins et al. 1998). Therefore, increasing hetero-
geneity of soil resources (i.e., soil depth and N avail-
ability) could be key to reconstructing more diverse
and heterogeneous communities under initially homo-
geneous agricultural conditions, provided established
or colonizing species contain sufficient functional vari-
ation to exploit the range of niches that environmental
heterogeneity creates.
Plant functional traits can indicate resource availabil-
ity and heterogeneity under various environmental con-
ditions (Lavorel and Garnier 2002, Harpole and Tilman
2007). The variation in plant functional traits in the pool
of dispersing species is a regional process influencing
community assembly and local diversity (Schellberg and
Pontes 2012). Following arrival, functional traits deter-
mine the ability of a species to pass through multiple
interacting environmental filters (e.g., soil nutrient sta-
tus, moisture conditions, competitors, mutualists, etc.)
that influence the establishment, survival, and reproduc-
tion of species in a specific environment (Reich et al.
2003). For example, species with thin leaves, tall stature,
and fast growth rates establish better under conditions
of high resource availability (Dyer et al. 2001). Seed
traits can also influence species establishment, preda-
tion, and germination success (Westoby 1998, Franzen
2004, Larson et al. 2015). Ultimately, there must be suf-
ficient resource availability in the environment and
sufficient functional variation in the pool of dispersing
species for new species to become recruited into estab-
lished communities (Conradi and Kollmann 2016).
In 1997, an experimental prairie restoration was estab-
lished to test the applicability of the “environmental
heterogeneity hypothesis”(EHH; Ricklefs 1977, Huston
1979). The field experiment contains replicated plots of
prairie restored with no soil manipulations and prairie
restored under a factorial combination of altered nutri-
ent availability and soil depth (Baer et al. 1999, 2003,
2016). Plant community dynamics over time showed no
effect of manipulated soil heterogeneity on species diver-
sity during the first five years of the experiment (Baer
et al. 2004). A weak effect of heterogeneity developed
over the longer term (15 yr) and was attributed to a tran-
sient increase in richness following a one-time propagule
addition of new species 8 yr post-restoration, which
resulted in a lower rate of species loss over time in more
heterogeneous plots with manipulated N availability and
soil depth (Baer et al. 2016). These results led us to
hypothesize that spatial heterogeneity in belowground
resources promotes greater community heterogeneity
that will reduce local extirpation of previously estab-
lished species and increase openness for new species to
colonize over the longer term. More succinctly, we
hypothesized that heterogeneity begets heterogeneity.
This hypothesis can be tested by adding propagules of
new species to communities, a prerequisite to increasing
richness (Eriksson and Ehrlen 1992a,b, Tilman 1993,
1997, Zobel 1997, Zobel et al. 2000, Myers and Harms
2009). Because we were not testing dispersal limitation
per se, we added propagules of the same suite of new
species to all of the experimental plots. Plant community
composition was quantified for 5 yr (corresponding to
16–20 yr post-restoration) under continuous propagule
addition to (1) reveal whether heterogeneous soil condi-
tions increase diversity through its effect on the estab-
lished local community or openness to the recruitment
of new species; and (2) identify which soil treatments are
most influential on the diversity response to heterogene-
ity, if any we also quantified traits of new species sup-
plied as propagules to reveal whether heterogeneous
plots contain more niches for new species to exploit (the
key mechanism underlying the EHH), indicated by
establishment of species containing unique trait spaces
(the collection of traits that help describe a species’
niche).
METHODS
Study site
The experiment was established in a former agricul-
tural field at the Konza Prairie Biological Station
(KPBS), a 3,457-ha Long-Term Ecological Research
(LTER) site in the Flint Hills region of northeastern
Kansas (39°05°N, 96°35°W), USA. Average annual
precipitation at KPBS is 835 mm. Annual precipitation
Article e02014; page 2 SARA G. BAER ET AL. Ecological Applications
Vol. 0, No. 0

received from 2013 through 2017 was 783, 701, 998, 984,
and 721 mm, respectively (precipitation data available
online).
6
The native vegetation at Konza Prairie is domi-
nated by C
4
grass species (Andropogon gerardii,Sorghas-
trum nutans,Panicum virgatum, and Schizachyrium
scoparium), but >300 subordinate forb species occur
throughout the site and contribute most to diversity. Soil
where the experiment was conducted was classified as
Reading silt loam (mesic Typic Arguidoll).
Experimental design and restoration approach
The experiment was a randomized complete block
design containing four blocks. Each block contained two
698 m plots randomly assigned to maximum hetero-
geneity and control treatments (Baer et al. 2016). The
maximum heterogeneity treatment contained a 2 93
factorial combination of soil depth and nutrient avail-
ability arranged according to a split block design. Soil
depth (two levels: reduced and deep) was assigned to
two of four alternating 2 96 m strips. Soil nutrient
availability (three levels: reduced-N, ambient-N and
enriched-N) was assigned to one of three 2 98 m strips
(perpendicular to the depth treatment). Reduced soil
depth was achieved by excavating the soil to a depth of
20–25 cm prior to the initial planting and burying pieces
of rough-cut limestone slabs (Baer et al. 1999). The aver-
age depth to limestone was 24 3 cm. All plots were
excavated to control for this disturbance. Sawdust was
initially added to reduce N availability to plants and
effectively immobilized N (Baer et al. 2003). Starting in
2005, the reduced N treatment has been maintained by
applying 84.2 g C/m
2
(sucrose-C) three times each grow-
ing season, which sustained reduced-N conditions (Baer
and Blair 2008, Baer et al. 2016). Elevated-N strips have
received 5 g N/m
2
in the form of NH
4
-NO
3
in early June
every year since 1998. The factorial combination of N
and soil depth treatments resulted in 12 subplots that
varied in soil N level and soil depth in each maximum
heterogeneity plot. The control plots contained no sub-
plot treatments.
In 1998, all plots were sown with 42 native prairie spe-
cies using a log-normal distribution of dominant grasses
and subordinate forbs to resemble never-cultivated tall-
grass prairie (Baer et al. 1999, 2003). In 2005, seeds of
15 additional forb species were sown into all plots at a
rate of 25 live seeds/m
2
(Baer et al. 2016). Beginning in
2013, we added 17 additional species (14 forbs, 2 grasses,
and 1 sedge) that were never previously recorded in
this experiment, but occur at KPBS, to all plots
(Appendix S1: Table S1). Seeds were purchased from
Prairie Moon Nursery (Winona, Wisconsin, USA). Per-
cent live seed was either provided by the seed supplier or
determined by the Illinois Seed Testing Lab. Seeds were
hand-broadcasted in spring each year at a rate of 20 live
seeds/m
2
per species for a total seeding rate of 300 live
seeds/m
2
. The experimental area was fenced in 1998 to
prevent deer browsing and has been burned almost
annually in the early spring (a common management
practice in this region) since 1998, with the exception of
2000 and 2003.
Plant community measurements
Each plot was divided into 12 2 92 m subplots based
on the orthogonally crossed assignment of treatments in
the maximum heterogeneity plots. The percent cover of
each species was visually estimated in two permanent 0.5
⨉0.5 m quadrats in each subplot in late spring (late
May–early June) and late summer (late August–early
September) from 2013 through 2017. The maximum
cover of each species across the two seasonal samplings
was used as the cover value for each species in a quadrat.
Species composition surveys conducted prior to 2013
contained few seedlings, so their presence and frequency
was attributed to the most recent propagule addition.
Maximum cover of each species was then averaged
across the two quadrats in each subplot. Plot-level rich-
ness (S) was calculated by summing the total number of
species from all subplots within a plot. Plot-level diver-
sity (Shannon’s diversity index, H0) and evenness (Pie-
lou’s evenness, J=H0/ln[S]) were calculated from the
average cover of each species across all 12 subplots
within a plot.
Resource heterogeneity
Nitrate availability (NO
3
-N) was quantified in all sub-
plots using buried ion exchange resin bags (Binkley and
Matson 1983). The collection of ions onto exchange
resins occurs passively through movement of soil water.
We chose to measure nitrate because it is more mobile in
soil with a high cation exchange capacity relative to
ammonium. Bags were constructed with nylon and con-
tained 5 g of anion exchange resin (Dowex 1X8-50;
Dow Chemical, Midland, Michigan, USA) preloaded
with Cl

. Two bags were buried ~10 cm deep in each
subplot (opposite of species composition quadrats) in
June and retrieved in September each year (Baer et al.
2003, Baer and Blair 2008, Baer et al. 2016). In the labo-
ratory, resin bags were rinsed with deionized water then
extracted using a 5:1 ratio of 2 mol/L KCl: resin on an
orbital shaker (200 rpm) for one hour. Solutions were
filtered through 0.4-lm polycarbonate membrane filters
and extracts were analyzed for NO
3
-N using an OI Ana-
lytical Flow Solution IV autoanalyzer (OI Analytical,
College Station, Texas, USA).
Light availability at the soil surface was determined by
measuring photosynthetically active radiation (PAR)
above and below the canopy with a 50-cm ceptometer
(Decagon Devices, Pullman, Washington, USA) at the
same time species composition was collected. Measure-
ments (n=5 above and below the canopy) were taken
6
http://nadp.slh.wisc.edu/data/sites/siteDetails.aspx?net=NTN&
id=KS31
Xxxxx 2019 HETEROGENEITY INCREASES DIVERSITY Article e02014; page 3

and averaged in each species composition sampling
quadrat in two perpendicular directions. Percent
available PAR at the soil surface was calculated as
(PAR
[soil surface]
/PAR
[above canopy]
)9100.
Plant functional traits (PFTs)
The same species added to the field experiment start-
ing in 2013 were grown in Conviron CMP 6050 growth
chambers (Conviron, North Branch, Minnesota, USA)
in the Southern Illinois University Horticulture
Research Greenhouse (Southern Illinois University, Car-
bondale, Illinois, USA). We quantified plant functional
traits following the methods used by Tucker et al.
(2011). Growth chamber conditions were 16 h day
length from 06:00 to 22:00 at 25°C with 1,200 lmol light
intensity and 20°C at night from 22:00 to 06:00. Plants
were watered twice daily, once at 07:30 and once at 16:30
and treated biweekly with commercial fertilizer (Scotts-
Miracle-Gro 24-8-16 [N-P-K] All Purpose Fertilizer;
ScottsMiracle-Gro, Marysville, Ohio, USA) to eliminate
nutrient stress. We used soil collected from the restora-
tion site, but outside of the experimental plots. Twelve
replicates of each species were planted in plastic cone-
tainers (D-40; Stuewe and Sons, Corvallis, Oregon,
USA). Several seeds of the same species were planted in
each cone-tainer, and the first emerging seedling was
used for PFT measurements (the rest were discarded).
We used 8 of the 12 replicates of each species and PFTs
were measured 8–9 weeks after the seedlings emerged.
We measured a suite of plant traits indicative of
growth rate and resource use (Weiher et al. 1999, Funk
et al. 2008, Tucker et al. 2011). Traits measured included
stem mass fraction (stem dry mass per module dry
mass), shoot length, longest internode length, maximum
leaf length, and average specific leaf area (SLA), and
nitrogen acquisition indicated by leaf nitrogen concen-
tration and nitrogen use efficiency (NUE; Gubsch et al.
2011). Gas exchange was measured on the youngest fully
expanded leaf using a Li-Cor LI-6400 Portable Photo-
synthesis System (LICOR Biosciences, Lincoln,
Nebraska, USA) in the growth chamber. Plant height
was used for shoot length. The upper three internodes
were measured to determine the longest internode length
on plants with measurable internodes. Specific leaf area
(SLA) was measured by scanning fresh fully expanded
leaves (n=3 leaves per individual) and leaf area was
then calculated using leaf area measurement software
v.1.3 (A.P. Askew 2003, The University of Sheffield,
Sheffield, UK). Leaf area was divided by the dry mass of
the leaf to determine SLA. Leaf nitrogen (N) concentra-
tion was measured on dried ground leaves using a
Thermo Scientific Flash CNHOS Elemental Analyzer
(Thermo Fisher Scientific, Waltham, Massachusetts,
USA). Percent N was then multiplied by the total dry
leaf mass to obtain the leaf N mass. Nitrogen use effi-
ciency (NUE) was obtained for healthy completely
expanded leaves by dividing rate of gas exchange
measured using the Li-COR 6400 by leaf N. Above-
ground biomass from each cone-tainer was clipped and
dried at 60°C for 5 d before weighing.
We also included seed traits of dry seed mass and seed
moisture content because they are important for germi-
nation and survival. Seed moisture content was deter-
mined by drying 30 crushed seeds at 105°C for 24 h,
cooling in a desiccator for 40 min, weighing, and apply-
ing the International Seed Testing Association formula
(ISTA 2006).
Statistical analysis
Plot-level heterogeneity effects (HETTRT) on varia-
tion in resources (coefficients of variation [CV] in NO
3
-
N and light availability), Shannon’s diversity, species
richness, evenness, grass and forb cover, forb richness,
and forb diversity from 2013 to 2017 were analyzed
according to a randomized complete block design with
repeated measures using the mixed model procedure in
SAS 9.4 (SAS Institute, Cary, North Carolina, USA).
Block was assigned as a random effect and year as the
repeated measure. The least squares means procedure
was used to compare main effect means and contrast
statements were used to test for differences between the
heterogeneity and control treatments within each year
(a=0.05).
Separate (subplot-level) mixed model analyses were
performed in SAS (SAS 2014) using only the maximum
heterogeneity plots to examine the main effects and
interaction between nitrogen (NUT) and soil depth
(DEPTH) on relative NO
3
-N availability, percent light
availability at the soil surface, total diversity, total rich-
ness, and cover and richness of new species added from
2013 to 2017. These data were analyzed according to a
split-block design with year as a repeated measure. Sub-
plots within each NUT and DEPTH level were assigned
to vertical (VS) and horizontal strips (HS), respectively
(Baer et al. 2003, 2016). Block, VS(NUT), and HS
(DEPTH) were assigned as random effects in the model.
Because there were no significant three-way interactions
(DEPTH 9NUT 9YEAR), or interactions between
DEPTH and YEAR, we used contrast and estimate
statements to compare nutrient means within a year if
there was an interaction between NUT and YEAR. The
least squares means procedure was used to compare
main effect means (a=0.05). Light availability at the
soil surface was analyzed by year.
For all repeated-measures analyses, we used the Ken-
wood-Rogers method to estimate degrees of freedom.
We ran each analysis with compound symmetry (CS),
autoregressive (AR), and unstructured (UN) covariance
structures, and selected the analysis with the lowest
Akaike information criterion (AIC; Littell et al. 2006).
The covariance structure accompanies each Fvalue pre-
sented in the results, followed by numerator and denomi-
nator degrees of freedom. For many response variables,
there was a significant main effect of year across plot or
Article e02014; page 4 SARA G. BAER ET AL. Ecological Applications
Vol. 0, No. 0

subplot treatments. Year main effects are only described
in the results if they showed a clear pattern (increasing
or decreasing over time).
We used PERMDISP (Anderson et al. 2006) on com-
position data to determine if spatial heterogeneity of
vegetation varied between the control and maximum
heterogeneity treatments during the 5-yr study. PERM-
DISP uses Bray-Curtis dissimilarity to measure the spa-
tial dispersion of samples around the group centroid.
Larger values reflect increasing dissimilarity (higher
heterogeneity) among samples. We conducted PERM-
DISP analyses at two levels of resolution. At the first
level, we compared dissimilarity among all subplots
across all replicate control or maximum heterogeneity
plots to determine an overall treatment effect each year
using a ttest. At the second level, we calculated dissimi-
larity among all subplots within each replicate of each
treatment and compared differences between control
and maximum heterogeneity plots (n=4) using
ANOVA. PERMDISP analyses were run using PRI-
MER-6 (Clarke and Gorley 2006).
Comparison of added species that colonized
between the heterogeneity treatments was performed
with the model-based multivariate approach of the
mvabund package (Wang et al. 2012) in R (R Core
Team 2018). A presence–absence (binomial) response
to heterogeneity (whole-plot) treatment was fit for
each species individually and all species jointly. Signif-
icance was tested with likelihood ratio tests with a
null model (100 bootstraps).
Plant functional and seed traits were used to create a
trait space, with each species assigned to a functional
group (C
3
grass, C
4
grass, forb, and sedge). The degree
of overlap of trait space among species was assessed
using nonmetric multidimensional scaling (NMDS)
based on Bray-Curtis dissimilarity measures. NMDS is a
robust ordination technique that provides a visual repre-
sentation of the similarities of the variables being com-
pared (Minchin 1989). Trait spaces were compared using
analysis of similarity (ANOSIM). The NMDS and
ANOSIM analyses were conducted using DECODA
3.01 software (Minchin 1989) on transformed data (log
[x+1]). We used a=0.004 to determine significance for
multiple comparisons accounting for 12 species; if spe-
cies were significantly different, they were considered to
occupy different trait spaces. Fitted vectors represent
Pearson correlation coefficients between the measured
plant functional trait variables and the NMDS axes.
RESULTS
Treatment effects on resource variability and availability
The soil treatments increased heterogeneity in N avail-
ability (Table 1). The CV of resin-collected NO
3
-N was
one to three orders of magnitude higher in the maximum
heterogeneity than control plots, with some years
exhibiting more disparity between the treatments than
others, as reflected by an interaction between HETTRT
and YEAR (F
4,24 (AR1)
=15.5, P<0.001). Higher vari-
ability in NO
3
-N availability in the maximum hetero-
geneity plots was due to lower NO
3
-N availability in the
reduced-N (C-amended) treatment and highest NO
3
-N
availability in the enriched-N treatment in all years, with
the exception of similar NO
3
-N availability between
ambient-N and reduced-N soil in 2015 (NUT 9YEAR
interaction: F
8,139 (AR1)
=12.5, P<0.001). There was no
effect of the soil depth treatment on nitrate availability.
The heterogeneity treatment effect on the CV of light
availability was marginally significant (F
1,10.1
=3.79;
P=0.080), and the difference between the CVs over all
years (26.4 vs 21.8 in the maximum heterogeneity and
control plots, respectively) was small (Table 1). Despite
this, the depth and nutrient treatments interacted to
affect light availability in 2013 (F
2,19.6
=7.8; P=0.003)
and 2014 (F
2,18.7
=5.14; P=0.017), and there was a
main effect of nutrient treatment on light availability in
2015 (F
2,24.9
=7.4; P=0.003) and 2017 (F
2,4.7
=11.2;
P=0.016). The only consistent pattern among all years
when the subplot treatments significantly affected light
availability was higher light availability in deep soil
under reduced-N conditions compared to deep soil
under enriched-N conditions (Table 1).
Heterogeneity effects on plant community structure
Following 20 yr of community assembly, we recorded
a total of 49 species in the maximum heterogeneity and
control plots, with 40 and 34 species occurring in each
treatment, respectively. There were 15 species found only
in the maximum heterogeneity plots and 9 species found
only in the control plots, with 25 species occurring in
both treatments.
Total diversity and richness began to diverge between
the maximum heterogeneity and control treatments 16
to 20 yr post-restoration (Fig. 1), concurrent with
propagule addition. Despite variation in diversity among
years during this period (YEAR main effect: F
4,23.1
(AR1)
=7.57, P<0.001), there was no interaction
between HETTRT and YEAR (F
4,23.1 (AR1)
=0.63,
P=0.649). Diversity was higher in the maximum
heterogeneity plots across years 16 through 20
(HETTRT: F
1,5.38 (AR1)
=8.16, P=0.033) (Fig. 1A,B).
Richness showed a similar response, with significant
variation among years (YEAR: F
4,21.8 (AR1)
=7.94,
P<0.001), no interaction between HETTRT and
YEAR (F
4,21.8 (AR1)
=0.42, P=0.792), and higher rich-
ness in the maximum heterogeneity treatment relative to
the control over the last 5 yr (HETTRT: F
1,11.8
(AR1)
=4.87, P=0.048; Fig. 1B,C). Evenness also var-
ied among years (YEAR: F
4,24 (CS)
=12.67, P<0.001),
ranging from 0.41 in 2013 to 0.48 in 2017 and oscillating
between these years (data not presented). There was
no effect of heterogeneity on evenness (HETTRT:
F
1,6 (CS)
=2.15, P=0.193) or interaction between
HETTRT and YEAR (F
4,24 (CS)
=2.22, P=0.100).
Xxxxx 2019 HETEROGENEITY INCREASES DIVERSITY Article e02014; page 5

Cover of established grasses and forbs varied between
the heterogeneity treatments. From 2013 to 2017, grass
cover was 12.5% lower in the maximum heterogeneity
treatment (56.9% 1.2%) relative to the control
(65.1% 2.9%) treatment across all years (HETTRT:
F
1,3 (CS)
=15.94, P=0.023). Forb cover exhibited
an interaction between HETTRT and YEAR (F
4,24
(CS)
=2.82, P=0.047) resulting from a slow develop-
ment of higher forb cover in the maximum heterogeneity
treatment relative to the control over time (Pvalues
corresponding to maximum heterogeneity vs. control
contrasts: 2013 =0.541; 2014 =0.404; 2015 =0.248;
2016 =0.209; and 2017 =0.023 (Fig. 2A). Forb richness
and diversity also began to respond consistently to the
heterogeneity treatment. Over the last 5 yr, forb richness
and diversity were 17% and 26% higher in the maximum
heterogeneity treatment compared to the control
(HETTRT forb richness: F
1,9.2 (AR1)
=9.96, P=0.011;
HETTRT forb diversity: F
1,6 (CS)
=6.76, P=0.041;
Fig. 2B–E).
The maximum heterogeneity plots became more com-
positionally different from one another over time relative
to the control plots (Appendix S1: Table S2). Greater
dispersion of the community among the maximum
heterogeneity plots was due to less dominance by Andro-
pogon gerardii Vitman (big bluestem) and more variation
in dominant and codominant species from various
sources (sown in 1998, added in 2005, and natural colo-
nization from the regional species pool) in the N-
enriched subplots. In 2017, for example, different species
(Ambrosia psilostachya DC., Eupatorium altissimum L.,
and Asclepias verticillata L.) dominated or codominated
cover with A. gerardii in three of the four N-enriched
subplots in Block 1. In Block 2, the codominant species,
based on average percent cover in the N-enriched strip,
were Salvia azurea Michx. ex Lam. (26.5% 7.4%) and
A. gerardii (19.4% 5.0%). In Block 3, the N-enriched
strip was codominated by Teucrium canadense L.
(30.4% 9.7%) and A. gerardii (26.9% 8.1%). In
Block 4, different species (Ambrosia psilostachyia DC.
TABLE 1. Coefficients of variation (CV) in resin-collected NO
3
-N and light availability in the maximum and control heterogeneity
treatments each year and availability of NO
3
-N and light in response to the soil treatments in the maximum heterogeneity plots
each year.
CV and NO
3
-N
and light
availability
Plot-level treatment responses Subplot-level treatment responses
Heterogeneous Control Reduced N Ambient N Enriched N
CV in resin-collected NO
3
-N
2013 1,247.0%
a
250.8% 28.3%
b
1.5%
2014 1,499.0%
a
473.0% 99.8%
b
33.0%
2015 4,027.7%
a
1,052.2% 4.9%
b
0.7%
2016 3,537.3%
a
910.1% 13.1%
b
2.3%
2017 2,109.5%
a
722.5% 9.8%
b
1.7%
CV of light availability†
2013 34.0% 4.38% 27.3% 3.26%
2014 20.7% 1.58% 20.6% 5.81%
2015 35.0% 3.78% 26.9% 3.88%
2016 14.5% 1.72% 14.0% 2.07%
2017 27.9%
a
2.70% 20.3%
b
0.87%
Concentration of resin-collected NO
3
-N (lg/bag)
2013 7.40
a
3.21 53.5
b
11.2 3,431
c
754
2014 6.41
a
1.90 130.5
b
33.0 4,360
c
1417
2015 0.97
a
0.67 4.4
a
1.5 12,738
b
2614
2016 0.67
a
0.58 10.2
b
0.7 11,268
c
2,230
2017 0.93
a
0.29 6.8
b
0.8 6,457
c
2,041
Light availability†
2013
Deep 27.5%
b
6.82% 19.6%
ab
3.75% 14.7%
a
2.95%
Shallow 17.3%
a
3.63% 21.0%
b
5.70% 18.2%
ab
3.81%
2014
Deep 41.2%
b
5.15% 31.5%
a
3.53% 37.9%
ab
5.70%
Shallow 39.5%
ab
2.80% 37.9%
ab
4.19% 30.9%
b
3.51%
2015 36.1%
b
3.07% 35.1%
b
3.58% 27.3%
a
7.98%
2016 38.5% 1.92% 37.5% 1.54% 35.0% 1.49%
2017 31.3%
b
1.46% 25.3%
a
1.16% 22.1%
a
3.79%
Notes: Within a year, means accompanied by the same letter were not significantly different (a=0.05). All values are mean 
SE.
†At the soil surface.
Article e02014; page 6 SARA G. BAER ET AL. Ecological Applications
Vol. 0, No. 0

and Panicum virgatum L.) were equivalent or second in
maximum cover to A. gerardii, respectively. In contrast,
A. gerardii was the dominant species in 100% of the sub-
plots within the control plots.
Soil treatment effects on the plant community
Diversity was affected by an interaction between
DEPTH and NUT (F
2,18.9 (UN)
=6.23, P=0.008)
resulting from higher diversity in shallow soil relative to
deep soil under ambient N conditions and higher diver-
sity in enriched-N soil relative to ambient and reduced-
N conditions in deep soil (Fig. 3A). Over the last 5 yr of
study, A. gerardii had 31–40% less cover in deep N-
amended soil relative to all other treatments (NUT 9
DEPTH interaction: F
2,84 (AR1)
=29.5, P=0.001) and
total richness was 20% and 51% higher in the enriched-
N soil relative to the ambient-N and reduced-N soil
treatments, respectively (NUT: F
2,27.1(CS)
=4.90,
P<0.015; Fig. 3B).
Colonization of new species
Twelve of the 17 species added starting in 2013 were
recorded during the 5-yr period. Not counting unidenti-
fiable seedlings, eight new species were recorded from
control plots and seven new species were recorded from
the maximum heterogeneity plots. Of the eight species
that established from the 2005 propagule addition, four
occurred only in the maximum heterogeneity plots and
the maximum heterogeneity plots contained twice as
many species from this supplemental seed addition than
FIG. 1. Plot-level Shannon’s diversity (A) all years measured since 1998 and (B) averaged over the last 5 yr (2013 to 2017), and
plot-level species richness (C) all years measured since 1998 and (D) averaged over the last 5 yr (2013 to 2017) in the maximum
heterogeneity and control plots. Arrows indicate supplemental propagule additions. All values are mean and SE.
Xxxxx 2019 HETEROGENEITY INCREASES DIVERSITY Article e02014; page 7

FIG. 2. Plot-level (A) forb cover each year, (B) forb richness each year, (C) forb richness over the last 5 yr, (D) forb diversity
each year, and (E) forb diversity the last 5 yr in the maximum heterogeneity and control treatments. Asterisks indicate significant
differences (P≤0.05) between control and maximum. All values are mean and SE. *,**, and *** for P≤0.05. 0.01, and 0.001.
Article e02014; page 8 SARA G. BAER ET AL. Ecological Applications
Vol. 0, No. 0

the control plots (6 vs. 3, respectively). There was a low
frequency of occurrence (<2.2% of the subplots recorded
over all 5 yr) of all but one species, Mirabilis nyctaginea
(Michx.) MacMill. (15% of subplots). Colonization of
any or all new species was not influenced by heterogene-
ity (bootstrap likelihood ratio test, P>0.05), but the
number of plots colonized varied by species.
Although new species comprised <2% of total plant
cover, the cover of species sown (starting in 2013 through
2017) in the plot-level heterogeneity treatments exhibited
an interaction with time (HETTRT 9YEAR: F
4,17.7
(AR1)
=2.95, P=0.049). There was no difference in
cover of new species between the heterogeneity treat-
ments from 2013 through 2015, but their cover became
higher in the maximum heterogeneity treatment com-
pared to control in 2016 and 2017. Although the cover
of new species added in the enriched-N soil was more
than two times higher than the ambient-N and reduced-
N treatments, the number of species/subplot averaged
across years was <1 in all nutrient treatments.
Plant functional traits
Of the 17 sown species grown in the growth chamber,
only 12 grew to maturity. A two-dimensional ordination
using plant functional traits was generated with NMDS
(2D stress =0.13; Fig. 4). All species occupied a
different trait space in the ordination (ANOSIM,
P<0.002). There was no evidence that species occupy-
ing a wider range of trait spaces colonized the maximum
heterogeneity plots (data not presented). In other words,
seedlings that matured enough to be confidently identi-
fied were not clustered or more dispersed in the control
and maximum heterogeneity treatments, respectively.
DISCUSSION
Ecological theory should robustly inform efforts to
restore biodiversity (Torok and Helm 2017). Under-
standing what constrains the colonization and persis-
tence of species in a community is at the heart of
community assembly theory (Diamond 1975, Keddy
1992, Lockwood et al. 1997, Belyea and Lancaster 1999,
Temperton et al. 2004, Weiher and Keddy 2004,
D’Amen et al. 2017) and essential for conserving biodi-
versity. Membership in a local community is a conse-
quence of colonization influenced by regional
(stochastic) processes and deterministic abiotic and bio-
tic filters that result in local extinction (Keddy 1992,
Marquez et al. 2016, Ulrich et al. 2016, Torok et al.
2018). If local richness is limited by colonization from a
regional species pool and communities are not saturated
with species (Eriksson 1993, Foster 2001, Zobel 2016),
then richness is expected to respond positively to seed
FIG. 3. Diversity and richness response to the soil treatments within the maximum heterogeneity plots. (A) Interactive effect of
soil depth and soil nitrogen on mean (and SE) subplot diversity over all years; shaded bars indicate shallow soil treatment. (B) Main
effect of nutrient treatment on subplot species richness over all years. Means accompanied by the same letter were not significantly
different (a=0.05).
Xxxxx 2019 HETEROGENEITY INCREASES DIVERSITY Article e02014; page 9

addition (Eriksson and Ehrlen 1992a,b, Eriksson 1993,
Tilman 1993, 1997, Zobel 1997, 2001, Zobel et al. 2000).
If niche availability limits recruitment and richness, then
increasing environmental heterogeneity may be a key
mechanism for promoting and/or maintaining species
coexistence (Levin 1974, Ricklefs 1977, Grime 1979,
Huston 1979).
Because landscape heterogeneity corresponds with
high floristic diversity of tallgrass prairie (Seastedt et al.
1991, Collins et al. 1998, Collins and Calabrese 2012),
we predicted that communities established under greater
variation in soil resource availability would develop
more heterogeneous communities that, in turn, would
lead to divergence in species richness and diversity over
time and create more niche space for new species to
exploit. Although Williams and Houseman (2014)
showed a positive short-term effect of soil heterogeneity
on species richness in restored prairie, many experimen-
tal tests of the EHH in plant communities, including this
experiment (Baer et al. 2005, 2016), provide under-
whelming support for the EHH (reviewed by Lundholm
2009). After 20 yr of soil manipulation in this experi-
ment, plant diversity and richness began to diverge, with
higher richness and diversity emerging consistently in
the maximum heterogeneity treatment. Higher diversity
and richness in the heterogeneous treatment resulted
from (1) higher cover, richness, and diversity of estab-
lished forbs averaged across all subplots; (2) higher rich-
ness in the N-enrichment subplots within the maximum
heterogeneity plots; (3) higher diversity in shallow soil
relative to deep soil under ambient N and reduced-N
conditions; and (4) lower cover of a dominant grass spe-
cies in deep fertilized soil.
Nutrient enrichment and soil depth are known to
increase and reduce the cover of C
4
grasses in tallgrass
prairie, respectively, and in this way indirectly determine
diversity via more and less competition with subordinate
species. Nitrogen is a limiting nutrient in tallgrass prairie
(Blair 1997), demonstrated by positive productivity
responses to added N (Tilman 1987, Seastedt et al.
1991) that typically correspond initially with a reduction
in species richness and diversity (Jacquemyn et al. 2003,
Baer et al. 2004, Clark and Tilman 2008, Hautier et al.
2009) owing to a reduced variety of niches (niche dimen-
sionality; Harpole and Tilman 2007). Species loss in
response to nutrient enrichment is a common phe-
nomenon (Suding et al. 2005) that may be difficult to
reverse if alternative stable states develop (Isbell et al.
FIG. 4. Nonmetric multidimensional scaling ordination of 12 prairie species added to the field experiment starting in 2013. Dif-
ferent ordination (trait) spaces were determined using ANOSIM, all species had different trait spaces (a=0.004). Vector lengths
correspond to variance in axis scores explained by each environmental variable.
Article e02014; page 10 SARA G. BAER ET AL. Ecological Applications
Vol. 0, No. 0

2013). Our multi-decadal study of community response
to nutrient enrichment thus far demonstrates dynamic
transient community state changes. For example, during
the first 3 yr of restoration, there was a precipitous
decline in species richness and an increase in grass domi-
nance while richness remained higher and grass cover
lower in reduced-N soil (Baer et al. 2003). Over the next
10 yr, a strong negative correlation between cover of A.
gerardii and species richness developed (Baer et al.
2016). After 15 yr of community assembly, higher rich-
ness in reduced-N soil proved to be short lived and spe-
cies richness was not consistently lower in the N-
enriched soil relative to the other soil treatments in later
years because grass cover declined with persistent N
addition (Baer et al. 2016).
Diversity became consistently higher under nutrient
enrichment in this developing prairie only after 15 yr of
N addition. This response was due to higher forb cover
and richness under nutrient addition as grass cover
declined. Avolio et al. (2014) also found that long-term
N addition promoted fast-growing forbs with low N-use
efficiency that replaced the dominant C
4
grasses under
high N availability over the long-term in tallgrass prairie.
Other studies have shown long-term N addition has
potential to decrease temporal stability of grassland
communities as dominance decreases (Hautier et al.
2014, Zhang et al. 2016). In our experiment, nutrient
enrichment produced strong shifts in community struc-
ture over time, starting with early dominance by Pan-
icum virgatum L. (Baer et al. 2003), followed by
increasing dominance of Andropogon gerardii (Baer et al.
2016), and eventually different codominant species
among N-enriched subplots as grass cover declined. This
suggests that community assembly processes shift from
strongly deterministic to stochastic (more randomly dri-
ven by the regional species pool) over the long-term
under nutrient enrichment, supported by greater varia-
tion (dispersion) in community composition among the
maximum heterogeneity plots relative to each other than
composition in the controls plots relative to each other.
One possible explanation for dynamic community
change and increase in species richness under long-term
nutrient enrichment may be a corresponding change in
the community composition and richness of arbuscular
mycorrhizae fungi (AMF). The rationale for this pro-
posed mechanism is based on a previous study at Konza
Prairie that showed higher richness of AMF in long-
term fertilized vs. unfertilized native grassland soil
(Egerton-Warburton et al. 2007) coupled with the gener-
ally positive relationship between AMF and plant biodi-
versity (van der Heijden et al. 1998) and the key role
AMF play in promoting plant species coexistence
(Crawford et al. 2019).
High species richness in native tallgrass prairie is pri-
marily a function of forb richness and cover, and as
such, ecological drivers that reduce grass dominance in
native prairie support higher plant diversity (Gibson
and Hulbert 1987, Collins et al. 1998, Collins and
Calabrese 2012, Manning et al. 2017). Diversity and
forb richness are also inversely related to the cover of
dominant grasses in restored prairie (McCain et al.
2010, Baer et al. 2016). Because fire and grazing are dif-
ficult to manipulate on small scales, we imparted the
shallow soil treatment, based on the observation that
there is less cover of C
4
grasses and higher richness in
shallow upland soils in surrounding native prairie (Gib-
son and Hulbert 1987). We expected the community
response to this manipulation to develop over many
years. The interaction that occurred between nutrient
and soil depth (i.e., higher diversity in shallow soil in
only one nutrient treatment) during the first 15 yr of
community assembly (Baer et al. 2016) persisted for five
additional years. Interestingly, diversity was similar in
shallow soil under all nutrient regimes, as was the cover
of A. gerardii. Baer et al. (2016) previously documented
that differences in the cover of A. gerardii among nutri-
ent treatments only occurred in deep soil. Thus, less
effect of N-availability on species composition in shallow
soil appears to be persistent and suggests that stochastic
processes have less influence on community assembly in
shallow soil under a potentially different limiting
resource (e.g., soil moisture).
In the absence of continuous propagule supply,
restored communities can develop distorted species com-
position relative to target assemblages (Howe 1999,
Maina and Howe 2000). Limited propagules or bud
banks of subordinate species can constrain diversity and
richness in grassland (Foster et al. 2004, Dalgleish and
Hartnett 2009, Willand et al. 2013). Further, several
studies indicate that plant diversity response to hetero-
geneity interacts with dispersal, demonstrated by higher
diversity in more heterogeneous environments when
propagules are supplied (Coulson et al. 2001, Foster
et al. 2004, Questad and Foster 2008, Baer et al. 2016).
As such, we added propagules of new species to all treat-
ments, not to test dispersal limitation, but to determine
if resource heterogeneity increased niche availability for
new species to exploit.
We quantified trait spaces of newly added species to
reveal community assembly processes (van der Plas et al.
2015). More specifically, we used trait analyses to eluci-
date the relative strength of environmental filtering in
homogenous and heterogeneous conditions. We expected
greater clustering of species recruited into more homoge-
nous communities established under more homogeneous
environmental conditions in the trait ordination space,
indicative of niche availability for a narrower range of
traits relative to heterogeneous conditions. Despite
higher recruitment of newly sown species in the maxi-
mum heterogeneity plots, the magnitude of this response
was low and these species did not collectively occupy a
larger trait space, suggesting a strong filtering process
(Foster et al. 2004, Ackerly and Cornwell 2007, Dickson
and Foster 2008, Grman et al. 2015). The most frequent
establishment by Mirabilis nyctaginea, suggests traits
associated with light capture (e.g., longest internode,
Xxxxx 2019 HETEROGENEITY INCREASES DIVERSITY Article e02014; page 11

shoot length, and stem mass fraction) are important
traits for new species to recruit into established commu-
nities.The very low occurrence of all but one species
demonstrates establishment limitation for a wide range
of species and niches regardless of environmental hetero-
geneity. Recruitment of new species could also take
longer than 5 yr, as it took >10 yr for many species sup-
plied as propagules in 2005 to occupy more than 5% of
the total cover (see Data Availability).
CONCLUSION
Although the “environmental heterogeneity hypothe-
sis”is a widely accepted mechanism for species coexis-
tence (Kolasa and Pickett 1991), this study
demonstrated that long-term manipulation may be
required for environmental variation in resources to
increase plant diversity, particularly during ecosystem
recovery following long-term disturbance. We attribute
the slow-to-emerge diversity response to environmental
heterogeneity to (1) initial dominance of clonal grasses
in this experiment (Baer et al. 2016), (2) time required
for community destabilization to occur in nutrient
enriched patches, and (3) time required for roots and
competition for belowground resources to develop in
the shallow soil treatment. Long-term community
response to N-addition in this experiment runs counter
to the general phenomenon of lower species diversity
with nutrient enrichment, particularly in grasslands
(Bobbink et al. 2010, Harpole et al. 2016), but supports
the proposition that global change drivers, such as N
enrichment, can alter spatial heterogeneity in ecological
communities (Avolio et al. 2015). This agrees with spa-
tially variable change in community structure in nutri-
ent-amended native grassland (Koerner et al. 2016).
Change from deterministic to stochastic community
assembly processes has also been demonstrated in suc-
cessional sequences along a multi-century arable-to-
grassland chronosequence (Purschke et al. 2013). Our
multi-decadal study suggests that heterogeneity that
includes nutrient enrichment might hasten this natural
phenomenon, but we do not advocate adding nutrients
to restorations at large scales. Means to increase patchy
resource heterogeneity at large scales and consequences
for achieving the composition of species practitioners
strive to restore deserves further investigation.
ACKNOWLEDGMENTS
Funding for this research was provided by the National
Science Foundation (IBN 9603118; DEB-1922915) with support
from the Konza Prairie Long-Term Ecological Research Pro-
gram. Laboratory and field assistance were provided by G.
Manning, S. Black, J. Willand, P. Harris, J. Fitzgibbon, and A.
Rothert.
LITERATURE CITED
Ackerly, D. D., and W. K. Cornwell. 2007. A trait-based
approach to community assembly: partitioning of species
trait values into within- and among-community components.
Ecology Letters 10:135–145.
Anderson, M. J., K. E. Ellingsen, and B. H. McArdle. 2006.
Multivariate dispersion as a measure of beta diversity. Ecol-
ogy Letters 9:683–693.
Avolio, M. L., S. E. Koerner, K. J. La Pierre, K. R. Wilcox, G.
W. T. Wilson, M. D. Smith, and S. L. Collins. 2014. Changes
in plant community composition, not diversity, during a dec-
ade of nitrogen and phosphorus additions drive above-
ground productivity in a tallgrass prairie. Journal of Ecology
102:1649–1660.
Avolio, M. L., K. J. La Pierre, G. R. Houseman, S. E. Koerner,
E. Grman, F. Isbell, D. S. Johnson, and K. R. Wilcox. 2015.
A framework for quantifying the magnitude and variability
of community responses to global change drivers. Ecosphere
6:1649–1660.
Baer, S. G., and J. M. Blair. 2008. Grassland establishment
under varying resource availability: A test of positive and
negative feedback. Ecology 89:1859–1871.
Baer, S. G., J. M. Blair, and A. K. Knapp. 1999. Manipulation
of soil resource heterogeneity in a tallgrass prairie restoration.
Proceedings of the North American Prairie Conference
16:78–87.
Baer, S. G., J. M. Blair, S. L. Collins, and A. K. Knapp. 2003.
Soil resources regulate productivity and diversity in newly
established tallgrass prairie. Ecology 84:724–735.
Baer, S. G., J. M. Blair, S. L. Collins, and A. K. Knapp. 2004.
Plant community responses to resource availability and
heterogeneity during restoration. Oecologia 139:617–629.
Baer, S. G., S. L. Collins, J. M. Blair, A. K. Knapp, and A. K.
Fiedler. 2005. Soil heterogeneity effects on tallgrass prairie
community heterogeneity: An application of ecological the-
ory to restoration ecology. Restoration Ecology 13:413–424.
Baer, S. G., J. M. Blair, and S. L. Collins. 2016. Environmental
heterogeneity has a weak effect on diversity during commu-
nity assembly in tallgrass prairie. Ecological Monographs
86:94–106.
Barak, R. S., et al. 2017. Restored tallgrass prairies have
reduced phylogenetic diversity compared with remnants.
Journal of Applied Ecology 54:1080–1090.
Bauer, J. T., N. Blumenthal, A. J. Miller, J. K. Ferguson, and H.
L. Reynolds. 2017. Effects of between-site variation in soil
microbial communities and plant-soil feedbacks on the pro-
ductivity and composition of plant communities. Journal of
Applied Ecology 54:1028–1039.
Belyea, L. R., and J. Lancaster. 1999. Assembly rules within a
contingent ecology. Oikos 86:402–416.
Binkley, D., and P. Matson. 1983. Ion-exchange resin bag
method for assessing forest soil-nitrogen availability. Soil
Science Society of America Journal 47:1050–1052.
Blair, J. M. 1997. Fire, N availability, and plant response in
grasslands: A test of the transient maxima hypothesis. Ecol-
ogy 78:2359–2368.
Bobbink, R., et al. 2010. Global assessment of nitrogen deposi-
tion effects on terrestrial plant diversity: a synthesis. Ecologi-
cal Applications 20:30–59.
Briggs, J. M., and A. K. Knapp. 1995. Interannual variability
in primary production in tallgrass prairie—climate, soil-
moisture, topographic position, and fire as determinants of
aboveground biomass. American Journal of Botany
82:1024–1030.
Bush, J. K., and O. W. Van Auken. 2010. Competition between
Schizachyrium scoparium and Buchloe dactyloides: The role of
soil nutrients. Journal of Arid Environments 74:49–53.
Camill, P., et al. 2004. Community- and ecosystem-level
changes in a species-rich tallgrass prairie restoration. Ecologi-
cal Applications 14:1680–1694.
Article e02014; page 12 SARA G. BAER ET AL. Ecological Applications
Vol. 0, No. 0

Clark, C. M., and D. Tilman. 2008. Loss of plant species after
chronic low-level nitrogen deposition to prairie grasslands.
Nature 451:712–715.
Clarke, K. R., and R. N. Gorley. 2006. PRIMER v6: user man-
ual/tutorial. PRIMER-E, Plymouth, UK.
Collins, S. L., and L. B. Calabrese. 2012. Effects of fire, grazing
and topographic variation on vegetation structure in tallgrass
prairie. Journal of Vegetation Science 23:563–575.
Collins, S. L., A. K. Knapp, J. M. Briggs, J. M. Blair, and E. M.
Steinauer. 1998. Modulation of diversity by grazing and
mowing in native tallgrass prairie. Science 280:745–747.
Collins, S. L., et al. 2018. Temporal heterogeneity increases with
spatial heterogeneity in ecological communities. Ecology
99:858–865.
Conradi, T., and J. Kollmann. 2016. Species pools and environ-
mental sorting control different aspects of plant diversity and
functional trait composition in recovering grasslands. Journal
of Ecology 104:1314–1325.
Coulson, S. J., J. M. Bullock, M. J. Stevenson, and R. F. Pywell.
2001. Colonization of grassland by sown species: dispersal
versus microsite limitation in responses to management. Jour-
nal of Applied Ecology 38:204–216.
Crawford, K. M., et al. 2019. When and where plant-soil feed-
back may promote plant coexistence: a meta-analysis. Ecol-
ogy Letters 22:1274–1284.
Dalgleish, H. J., and D. C. Hartnett. 2009. The effects of fire
frequency and grazing on tallgrass prairie productivity and
plant composition are mediated through bud bank demogra-
phy. Plant Ecology 201:411–420.
D’Amen, M., C. Rahbek, N. E. Zimmermann, and A. Guisan.
2017. Spatial predictions at the community level: from cur-
rent approaches to future frameworks. Biological Reviews
92:169–187.
Diamond, J. M. 1975. Assembly of species communities. Pages
342–344 in J. M. Diamond, and M. L. Cody, editors. Ecology
and evolution of communities. Harvard University Press,
Boston, Massachusetts, USA.
Dickson, T. L., and B. L. Foster. 2008. The relative importance
of the species pool, productivity and disturbance in regulating
grassland plant species richness: a field experiment. Journal
of Ecology 96:937–946.
Dyer, A. R., D. E. Goldberg, R. Turkington, and C. Sayre.
2001. Effects of growing conditions and source habitat on
plant traits and functional group definition. Functional Ecol-
ogy 15:85–95.
Egerton-Warburton, L. M., N. Collins Johnson, and E. B.
Allen. 2007. Mycorrhizal community dynamics following
nitrogen fertilization: a cross-site test in five grasslands. Eco-
logical Monographs 77:527–544.
Ellis, E. C., and N. Ramankutty. 2008. Putting people in the
map: anthropogenic biomes of the world. Frontiers in Ecol-
ogy and the Environment 6:439–447.
Eriksson, O. 1993. The species-pool hypothesis and plant com-
munity diversity. Oikos 68:371–374.
Eriksson, O., and J. Ehrlen. 1992a. Seed and microsite limitation
of recruitment in plant-populations. Oecologia 91:360–364.
Eriksson, O., and J. Ehrlen. 1992b. Seed and microsite limitation
of recruitment in plant-populations. Oecologia 91:360–364.
Foster, B. L., T. L. Dickson, C. A. Murphy, I. S. Karel, and V.
H. Smith. 2004. Propagule pools mediate community assem-
bly and diversity-ecosystem regulation along a grassland pro-
ductivity gradient. Journal of Ecology 92:435–449.
Franzen, D. 2004. Plant species coexistence and dispersion of
seed traits in a grassland. Ecography 27:218–224.
Funk, J. L., E. E. Cleland, K. N. Suding, and E. S. Zavaleta.
2008. Restoration through reassembly: plant traits and inva-
sion resistance. Trends in Ecology & Evolution 23:695–703.
Gibson, D. J., and L. C. Hulbert. 1987. Effects of fire, topogra-
phy, and year-to-year climatic variation on species composi-
tion in tallgrass prairie. Vegetatio 72:175–185.
Grace, J. B., et al. 2016. Integrative modelling reveals mecha-
nisms linking productivity and plant species richness. Nature
529:390–393.
Grime, J. P. 1979. Plant strategies and vegetation processes.
John Wiley & Sons, New York, New York, USA.
Grman, E., T. Bassett, C. R. Zirbel, and L. A. Brudvig. 2015.
Dispersal and establishment filters influence the assembly of
restored prairie plant communities. Restoration Ecology
23:892–899.
Gubsch, M., N. Buchmann, B. Schmid, E. D. Schulze, A.
Lipowsky, and C. Roscher. 2011. Differential effects of plant
diversity on functional trait variation of grass species. Annals
of Botany 107:157–169.
Hansen, M. J., and D. J. Gibson. 2014. Use of multiple criteria
in an ecological assessment of a prairie restoration chronose-
quence. Applied Vegetation Science 17:63–73.
Harpole, W. S., and D. Tilman. 2007. Grassland species loss
resulting from reduced niche dimension. Nature 446:791–
793.
Harpole, W. S., et al. 2016. Addition of multiple limiting
resources reduces grassland diversity. Nature 537:93–96.
Hautier, Y., P. A. Niklaus, and A. Hector. 2009. Competition
for light causes plant biodiversity loss after eutrophication.
Science 324:636–638.
Hautier, Y., et al. 2014. Eutrophication weakens stabilizing
effects of diversity in natural grasslands. Nature 508:521–525.
Hoekstra, J. M., T. M. Boucher, T. H. Ricketts, and C. Roberts.
2005. Confronting a biome crisis: global disparities of habitat
loss and protection. Ecology Letters 8:23–29.
Howe, H. F. 1999. Dominance, diversity and grazing in tallgrass
restoration. Ecological Restoration North America 17:59–66.
Huston, M. 1979. General hypothesis of species diversity.
American Naturalist 113:81–101.
Isbell, F., P. B. Reich, D. Tilman, S. E. Hobbie, S. Polasky, and
S. Binder. 2013. Nutrient enrichment, biodiversity loss, and
consequent declines in ecosystem productivity. Proceedings of
the National Academy of Sciences USA 110:11911–11916.
ISTA. 2006. International rules for seed testing. International
Seed Testing Association, Zurich, Switzerland.
Jacquemyn, H., R. Brys, and M. Hermy. 2003. Short-term
effects of different management regimes on the response of
calcareous grassland vegetation to increased nitrogen. Biolog-
ical Conservation 111:137–147.
Keddy, P. A. 1992. Assembly and response rules—2 goals for
predictive community ecology. Journal of Vegetation Science
3:157–164.
Kindscher, K., and L. L. Tieszen. 1998. Floristic and soil
organic matter changes after five and thirty-five years of
native tallgrass prairie restoration. Restoration Ecology
6:181–196.
Klopf, R. P., S. G. Baer, and D. J. Gibson. 2014. Convergent
and contingent community responses to grass source and
dominance during prairie restoration across a longitudinal
gradient. Environmental Management 53:252–265.
Koerner, S. E., M. L. Avolio, K. J. La Pierre, K. R. Wilcox, M.
D. Smith, and S. L. Collins. 2016. Nutrient additions cause
divergence of tallgrass prairie plant communities resulting in
loss of ecosystem stability. Journal of Ecology 104:1478–1487.
Kolasa, J., and S. Pickett. 1991. Ecological heterogeneity.
Springer-Verlag, New York, New York, USA.
Larson, J. E., R. L. Sheley, S. P. Hardegree, P. S. Doescher, and
J. J. James. 2015. Seed and seedling traits affecting critical life
stage transitions and recruitment outcomes in dryland
grasses. Journal of Applied Ecology 52:199–209.
Xxxxx 2019 HETEROGENEITY INCREASES DIVERSITY Article e02014; page 13

Lavorel, S., and E. Garnier. 2002. Predicting changes in com-
munity composition and ecosystem functioning from plant
traits: revisiting the Holy Grail. Functional Ecology 16:545–
556.
Levin, S. A. 1974. Dispersion and population interactions.
American Naturalist 108:207–228.
Littell, R. C., G. G. Milliken, W. W. Stroup, and R. D. Wolfin-
ger. 2006. SAS system for mixed models. SAS Institute, Cary,
North Carolina, USA.
Lockwood, J. L., R. D. Powell, M. P. Nott, and S. L. Pimm.
1997. Assembling ecological communities in time and space.
Oikos 80:549–553.
Lundholm, J. T. 2009. Plant species diversity and environmental
heterogeneity: spatial scale and competing hypotheses. Jour-
nal of Vegetation Science 20:377–391.
MacDougall, A. S., K. S. McCann, G. Gellner, and R. Turking-
ton. 2013. Diversity loss with persistent human disturbance
increases vulnerability to ecosystem collapse. Nature 494:86–
89.
Maina, G. G., and H. F. Howe. 2000. Inherent rarity in commu-
nity restoration. Conservation Biology 14:1335–1340.
Manning, G. C., S. G. Baer, and J. M. Blair. 2017. Effects of
grazing and fire frequency on floristic quality and its relation-
ship to indicators of soil quality in tallgrass prairie. Environ-
mental Management 60:1062–1075.
Marquez, J., J. Kolasa, and L. Sciullo. 2016. Local versus regio-
nal processes and the control of community structure. Com-
munity Ecology 17:1–7.
Martin, L. M., K. A. Moloney, and B. J. Wilsey. 2005. An
assessment of grassland restoration success using species
diversity components. Journal of Applied Ecology 42:327–
336.
McCain, K. N. S., S. G. Baer, J. M. Blair, and G. W. T. Wilson.
2010. Dominant Grasses Suppress Local Diversity in
Restored Tallgrass Prairie. Restoration Ecology 18:40–49.
Minchin, P. R. 1989. DECODA user’s manual: Research School
of Pacific Studies. Australian National University, Canberra,
Australian Capital Territory, Australia.
Mlot, C. 1990. Restoring the prairie. BioScience 40:804–809.
Myers, J. A., and K. E. Harms. 2009. Seed arrival, ecological fil-
ters, and plant species richness: a meta-analysis. Ecology Let-
ters 12:1250–1260.
Polley, H. W., J. D. Derner, and B. J. Wilsey. 2005. Patterns of
plant species diversity in remnant and restored tallgrass prai-
ries. Restoration Ecology 13:480–487.
Polley, H. W., B. J. Wilsey, and J. D. Derner. 2007. Dominant
species constrain effects of species diversity on temporal vari-
ability in biomass production of tallgrass prairie. Oikos
116:2044–2052.
Purschke, O., B. C. Schmid, M. T. Sykes, P. Poschlod, S. G.
Michalski, W. Durka, I. Kühn, M. Winter, and H. C. Prentice.
2013. Contrasting changes in taxonomic, phylogenetic and
functional diversity during a long-term succession: insights
into assembly processes. Journal of Ecology 101:857–866.
Questad, E. J., and B. L. Foster. 2008. Coexistence through spa-
tio-temporal heterogeneity and species sorting in grassland
plant communities. Ecology Letters 11:717–726.
R Core Team. 2018. R: A language and environment for statisti-
cal computing. R Foundation for Statistical Computing,
Vienna, Austria. www.R-project.org
Reich, P. B., I. J. Wright, J. Cavender-Bares, J. M. Craine, J.
Oleksyn, M. Westoby, and M. B. Walters. 2003. The evolu-
tion of plant functional variation: Traits, spectra, and
strategies. International Journal of Plant Sciences 164:
S143–S164.
Richardson, P. J., A. S. MacDougall, and D. W. Larson. 2012.
Fine-scale spatial heterogeneity and incoming seed diversity
additively determine plant establishment. Journal of Ecology
100:939–949.
Ricklefs, R. E. 1977. Environmental heterogeneity and plant
species-diversity—hypothesis. American Naturalist 111:376–
381.
SAS. 2014. SAS. SAS Institute Inc., Cary, North Carolina,
USA.
Schellberg, J., and L. D. Pontes. 2012. Plant functional traits
and nutrient gradients on grassland. Grass and Forage
Science 67:305–319.
Scott, D. A., and S. G. Baer. 2018. Degraded soil increases the
performance of a dominant grass, Andropogon gerardii (Big
bluestem). Plant Ecology 219:901–911.
Seastedt, T. R., J. M. Briggs, and D. J. Gibson. 1991. Controls
of nitrogen limitation in tallgrass prairie. Oecologia 87:72–79.
Sperry, T. M. 1994. The Curtis Prairie restoration, using the sin-
gle species planting method. Natural Areas Journal 14:124–
127.
Suding, K. N., S. L. Collins, L. Gough, C. Clark, E. E. Cleland,
K. L. Gross, D. G. Milchunas, and S. Pennings. 2005. Func-
tional- and abundance-based mechanisms explain diversity
loss due to N fertilization. Proceedings of the National Acad-
emy of Sciences USA 102:4387–4392.
Temperton, V. M., R. J. Hobbs, T. Nuttle, and S. Halle. 2004.
Assembly rules and restoration ecology. Island Press, Wash-
ington, District of Columbia, USA.
Tilman, G. D. 1984. Plant dominance along an experimental
nutrient gradient. Ecology 65:1445–1453.
Tilman, D. 1987. Secondary succession and the pattern of plant
dominance along experimental nitrogen gradients. Ecological
Monographs 57:189–214.
Tilman, D. 1993. Species richness of experimental productivity
gradients—how important is colonization limitation. Ecology
74:2179–2191.
Tilman, D. 1997. Community invasibility, recruitment limita-
tion, and grassland biodiversity. Ecology 78:81–92.
Torok, P., and A. Helm. 2017. Ecological theory provides strong
support for habitat restoration. Biological Conservation
206:85–91.
Torok, P., G. Matus, E. Toth, M. Papp, A. Kelemen, J. Sonkoly,
and B. T
othm
er
esz . 2018. Both trait-neutrality and filtering
effects are validated by the vegetation patterns detected in the
functional recovery of sand grasslands. Scientific Reports
8:10.
Tucker, S. S., J. M. Craine, and J. B. Nippert. 2011. Physiologi-
cal drought tolerance and the structuring of tallgrass prairie
assemblages. Ecosphere 2:Art. 48.
Turner, C. L., J. M. Blair, R. J. Schartz, and J. C. Neel. 1997.
Soil N and plant responses to fire, topography, and supple-
mental N in tallgrass prairie. Ecology 78:1832–1843.
Ulrich, W., M. K. Zaplata, S. Winter, W. Schaaf, A. Fischer, S.
Soliveres, and N. J. Gotelli. 2016. Species interactions and
random dispersal rather than habitat filtering drive commu-
nity assembly during early plant succession. Oikos 125:698–
707.
van der Heijden, M. G. A., J. N. Klironomos, M. Ursic, P.
Moutoglis, R. Streitwolf-Engel, T. Boller, A. Wiemken, and I.
R. Sanders. 1998. Mycorrhizal fungal diversity determines
plant biodiversity, ecosystem variability and productivity.
Nature 396:69–72.
van der Plas, F., T. Janzen, A. Ordonez, W. Fokkema, J. Rein-
ders, R. S. Etienne, and H. Olff. 2015. A new modeling
approach estimates the relative importance of different com-
munity assembly processes. Ecology 96:1502–1515.
Vitousek, P. M., H. A. Mooney, J. Lubchenco, and J. M.
Melillo. 1997. Human domination of Earth’s ecosystems.
Science 277:494–499.
Article e02014; page 14 SARA G. BAER ET AL. Ecological Applications
Vol. 0, No. 0

Wang, Y., U. Naumann, S. Wright, and D. I. Warton. 2012. mv-
abund: an R package for model-based analysis of multivari-
ate data. Methods in Ecology & Evolution 3:471–474.
Weiher, E., and P. Keddy. 2004. Ecological assembly rules: per-
spectives, advances, retreats. Cambridge University Press,
Cambridge, UK.
Weiher, E., A. van der Werf, K. Thompson, M. Roderick, E.
Garnier, and O. Eriksson. 1999. Challenging Theophrastus: a
common core list of plant traits for functional ecology. Jour-
nal of Vegetation Science 10:609–620.
Westoby, M. 1998. A leaf-height-seed (LHS) plant ecology
strategy scheme. Plant and Soil 199:213–227.
Willand, J. E., S. G. Baer, D. J. Gibson, and R. P. Klopf. 2013.
Temporal dynamics of plant community regeneration sources
during tallgrass prairie restoration. Plant Ecology 214:1169–
1180.
Williams, B. M., and G. R. Houseman. 2014. Experimental evi-
dence that soil heterogeneity enhances plant diversity during
community assembly. Journal of Plant Ecology 7:461–469.
Wilsey, B. J. 2010. Productivity and subordinate species
response to dominant grass species and seed source during
restoration. Restoration Ecology 18:628–637.
Wilson, S. 2014. Heterogeneity, diversity, and scale in plant
communities. Pages 53–69 in M. J. Hutchings, E. A. John,
and A. J. A. Stewart, editors. The ecological consequences
of environmental heterogeneity. Blackwell Science, Oxford,
UK.
Zhang, Y. H., M. Loreau, X. T. Lu, N. P. He, G. M. Zhang, and
X. G. Han. 2016. Nitrogen enrichment weakens ecosystem
stability through decreased species asynchrony and popula-
tion stability in a temperate grassland. Global Change Biol-
ogy 22:1445–1455.
Zobel, M. 1997. The relative role of species pools in determin-
ing plant species richness. An alternative explanation of
species coexistence? Trends in Ecology & Evolution 12:266–
269.
Zobel, M. 2016. The species pool concept as a framework for
studying patterns of plant diversity. Journal of Vegetation
Science 27:8–18.
Zobel, M., M. Otsus, J. Liira, M. Moora, and T. Mols. 2000. Is
small-scale species richness limited by seed availability or
microsite availability? Ecology 81:3274–3282.
SUPPORTING INFORMATION
Additional supporting information may be found online at: http://onlinelibrary.wiley.com/doi/10.1002/eap.2014/full
DATA AVAILABILITY
Data sets generated and analyzed for development of this manuscript are available in the Dryad Digital Repository at https://doi.
org/10.5061/dryad.63n980b and the Konza Prairie LTER data portal at https://doi.org/10.6073/pasta/28ce07278347b504fbbc956a
9011ac70.
Xxxxx 2019 HETEROGENEITY INCREASES DIVERSITY Article e02014; page 15