THE ROLE OF BIODIVERSITY IN PRAIRIE RESTORATION:
TESTS OF THEORY AND IMPLICATIONS FOR MANAGEMENT
By
Tyler Jacob Bassett

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Plant Biology – Doctor of Philosophy
Ecology, Evolutionary Biology and Behavior – Dual Major
2017

ABSTRACT
THE ROLE OF BIODIVERSITY IN PRAIRIE RESTORATION:
TESTS OF THEORY AND IMPLICATIONS FOR MANAGEMENT
By
Tyler Jacob Bassett
Biodiversity is a primary focus of conservation and restoration, because it has intrinsic
value, and because it supports the ecosystem functioning that human well being ultimately
depends upon. Theory and experiments support the hypothesis that greater diversity in plant
communities supports greater primary productivity, nutrient cycling, invasion resistance and a
range of other processes linked to the healthy functioning of ecosystems. However, most of the
evidence for diversity-function relationships is from manipulations of diversity, and a limited
number of environmental variables, in small-scale plots. As a result, it is unclear how diversityfunction relationships will scale up to dynamic, “real-world” ecosystems, which limits the
capacity to effectively manage both biodiversity and ecosystem functioning. I examined
diversity-function relationships in prairie restorations, which provide an ideal scenario for
bridging the gap between experimental and natural ecosystems because diversity is manipulated
at large scales and across complex biotic and abiotic gradients. It is clear from experimental
evidence that diversity plays a role in supporting ecosystem functioning. My findings elucidate
how important diversity is at the scale of natural ecosystems, relative to both abiotic (e.g., soil
properties) and biotic (e.g., dominant species) factors that are likely to covary with diversity at
large scales. I also contribute directly to the practice of restoration by working in real
restorations, linking variation in management actions, such as seed sowing and prescribed fire, to
outcomes of immediate concern to managers, such as the relationship between native and exotic
species.

	

For Zak and Fen, the future.

iii

	
ACKNOWLEDGEMENTS

First and foremost, I cannot underestimate what a great advisor Dr. Jennifer Lau has
been. She always knew when to push me a little harder, and when to leave me alone to struggle
on my own. Her passion for science always inspired me. Meetings with Jen somehow always
managed to transform despair into confidence. I faced intellectual and emotional challenges I
could not have predicted over the past 7.5 years and Jen was simultaneously a constant source of
compassion and motivation.
The rest of my advisory committee – Kay Gross, Doug Schemske, and Lars Brudvig provided indispensable guidance. I truly appreciate the attention they paid, not only to the
academic rigor of my dissertation, but also to my career goals. As I bandied about the in the sea
of exciting ideas, they helped me steer my boat to productive waters. The Lau Lab was amazing!
I love you all for the people that you are, and deeply appreciate each and every one of you for
your readiness to offer advice on experimental design… or the designs of life; for your time
getting an experiment going… or helping each other get to the bottom of a glass of beer. Kane
Keller – I knew you were good stuff as soon I saw your ‘Of Montreal’ t-shirt, and when you
assured me that nobody would judge me for my ridiculously long goatee. Tomomi Suwa – you
will always be the little ‘t’ to my big ‘T’. Liz Schultheis – even though you pronounce ‘orange’
and ‘forest’ wrong… I’ve got to admit, I’ll always appreciate your sass. Susan Magnoli – it was
great having someone around to temper all the unbridled positivity. Meredith Zettlemoyer –I’m
so glad you picked up the mantle of Hanes’ work, and I look forward to helping with that in the
future! And of course, “ol’ steady” Mark Hammond – I’m glad we keep crossing paths in life.
I’m looking forward to the next intersection.

iv

	
The Kellogg Biological Station is a wonderful community. As a family man, I didn’t
integrate myself into the community as much as most grad students, but I always felt welcome,
and part of the club. There is not enough space here to name everybody, but I count each and
every one of you among my friends. And I’d imagine I received assistance from each and every
one of you at some point, if only a smile or kind gesture when I needed it. I especially want to
thank Colin Kremer, Beth Miller, Kyle Edwards, Dylan Weese, Casey TerHorst, Rachel Prunier,
and Nash Turley for helping me out when I was in the weeds of statistics, experimental design,
and theory. And the KBS K-12 crew - Michael Kuszynski (a true Behemoth), Jake Nalley (sorry
I embarrassed you by being better at workshop organizing than you), Sara Garnett, Cara Krieg,
Robin Tinghitella, Sara Bodbyl, Kara Hass, Misty Klotz, Tom Getty - for keeping my life
interesting, hilarious, and deeply fulfilling. I was never wanting for resources to carry out my
research. No matter what I was trying to do, there was always somebody at KBS who knew how
to do it and graciously connected me with the tools to get it done – folks like Andy Fogiel, Mark
Williams, Mark Manuszak, Stacey VanderWulp, Kevin Kahmark, Joe Simmons, and Cathy
McMinn. A huge thank you to KBS Director Kay Gross for always figuring out creative ways to
connect grad students with research funding, as well as tuition and graduate stipends. Thank you
to George Lauff for funding so much graduate research over the years. And I am grateful for the
MSU Plant Sciences fellowship and USDA-NIFA program for each funding two years of my
research.
This dissertation represents a subset of the output from a large collaborative project
initiated by Lars Brudvig, Emily Grman, and myself in 2011; we were joined by Chad Zirbel in
2013. I have really enjoyed learning the craft of collaborative science alongside Lars, Emily, and
Chad. And we had SO MUCH HELP! I had the pleasure of mentoring several undergraduate

v

	
students - always positive, incredibly smart - during this dissertation: Michael Gonczar, Michael
Havern, Gabriel Stewart (future MMA Champ), Skye Greenler, Kent Connell, and Daniel Xie!
Several other hardworking individuals contributed – Jonathon Landis, Mike Epperley, Alischa
Fisher, to name a few all-stars.
There are a slew of former mentors, bosses, friends, and coworkers that have inspired and
guided me in my career that led me to this point. I appreciate the opportunities and insights
gained from association with Woody Ehrle, Todd Barkman, Steve Malcolm, Steve Keto,
Raymond Adams, Jr., Michael Kost, Bradford “Old Bradd” Slaughter, Jerry Stewart, Nate Fuller,
Chad Hughson, Scott Namestnik, Ken Hiser, and again, way too many to mention. Most
importantly, without the guidance, patience, and wisdom of Tom and Nancy Small, I would
never have developed the passion for botany that led me to this place. I am forever grateful.
A big nod to the music of Tortoise, Stereolab, and Brian Eno, my consistent companions
during hours of analysis, reading, and writing. A bigger nod to my fellow music makers (the
larger Nwe Spryghts and Double Phelix family), especially Bryan Heany, MW, Anger Watson,
Jay Gavan, and the late, great Pat Carroll… my consistent companions and sources of sanity
when not analyzing, reading, writing, or parenting. A huge nod to the old guard of friends that
kept me grounded at various times when I needed it: Benn & Dave; Seb, Ryan, Alex, and that
whole crew; and Jenny – thanks for trying so hard.
And to Zak and Fen, who were children of 10 and 7 when I began this journey, and are
now practically adults. I love you more and more every day and I’m incredibly proud of the
people you have become and are in the process of becoming. I’m sorry about the messed up
world you are inheriting, but I trust that your ingenuity, creativity, and open hearts will do better
things with it than my and previous generations ever could.

vi

	
TABLE OF CONTENTS

LIST OF TABLES ......................................................................................................................... ix
LIST OF FIGURES ....................................................................................................................... xi
CHAPTER ONE ............................................................................................................................. 1
INTRODUCTION .......................................................................................................................... 1
Introduction ......................................................................................................................... 1
Main results and significance.............................................................................................. 4
CHAPTER TWO ............................................................................................................................ 6
DIVERSITY REDUCES STABILITY IN GRASSLAND RESTORATIONS ............................. 6
Abstract ............................................................................................................................... 6
Introduction ......................................................................................................................... 7
Methods............................................................................................................................. 10
Study sites ............................................................................................................. 10
Drought ................................................................................................................. 11
Field sampling ....................................................................................................... 11
Environmental factors ........................................................................................... 11
Data analysis ......................................................................................................... 12
Overview ................................................................................................... 12
Stability, diversity, and dominant species metrics .................................... 13
Question 1 - Diversity-stability relationships ........................................... 14
Question 2 - Effects of dominant species abundances and
environmental factors on diversity-stability relationships ........................ 14
Question 3 - Relative importance of diversity compared to other
putative drivers for stability ...................................................................... 16
Results ............................................................................................................................... 17
Drought effects on annual productivity ................................................................ 17
Question 1 - Diversity-stability relationships ....................................................... 18
Question 2 - Effects of dominant species abundances and environmental
factors on diversity-stability relationships ............................................................ 19
Question 3 - Relative importance of diversity compared to other putative
drivers for stability ................................................................................................ 20
Discussion ......................................................................................................................... 25
Diverse sites were less stable ................................................................................ 25
Diversity effects on resistance and resilience were strong relative
to environmental drivers ....................................................................................... 28
Conclusions ........................................................................................................... 29
Acknowledgements ........................................................................................................... 30
APPENDIX ....................................................................................................................... 31

vii

	
CHAPTER THREE ...................................................................................................................... 34
MORE IS BETTER – SOWING MORE SPECIES AT HIGHER DENSITIES
SUPPORTS GREATER NATIVE SPECIES RICHNESS IN PRAIRIE
RESTORATIONS THAN PREDICTED BY NATIVE-EXOTIC RELATIONSHIP ................. 34
Abstract ............................................................................................................................. 34
Introduction ....................................................................................................................... 35
Methods............................................................................................................................. 39
Study system ......................................................................................................... 39
Plant community sampling and species origin...................................................... 40
Management and environmental factors ............................................................... 40
Management - seed addition and prescribed fire ...................................... 42
Environmental factors related to immigration rates.................................. 42
Environmental factors related to resource availability ............................. 43
Data analysis ......................................................................................................... 44
Results ............................................................................................................................... 47
Discussion ......................................................................................................................... 51
Management favors native richness ...................................................................... 54
The impact of seed sowing may fade over time ................................................... 56
Conclusions and management implications .......................................................... 57
Acknowledgements ........................................................................................................... 59
APPENDIX ....................................................................................................................... 60
CHAPTER FOUR ......................................................................................................................... 76
RESTORING GRASSLANDS TO RESIST INVASION – RESTORED DIVERSITY
CAN MEDIATE HISTORICAL LAND-USE EFFECTS ............................................................ 76
Abstract ............................................................................................................................. 76
Introduction ....................................................................................................................... 77
Methods............................................................................................................................. 80
Prairie establishment ............................................................................................. 80
Field sampling ....................................................................................................... 80
Classifying invasive species ................................................................................. 81
Classifying propagule pressure, abiotic site conditions, restored community
and management ................................................................................................... 82
Data analysis ......................................................................................................... 85
Results ............................................................................................................................... 86
Discussion ......................................................................................................................... 89
Impacts of propagule pressure – landscape context and land use history............. 92
Impacts of site conditions – soil moisture............................................................. 94
Impacts of community structure – Andropogon gerardii abundance and
sown richness ........................................................................................................ 94
Impacts of management – seed mix richness and prescribed fire ......................... 95
Summary and management implications .............................................................. 96
Acknowledgements ........................................................................................................... 99
APPENDIX ..................................................................................................................... 100
LITERATURE CITED ............................................................................................................... 111

viii

	
LIST OF TABLES

Table 2.1. Model selection results of environmental interaction, nested additive, and diversity
models (dAICc ≤ 4.0) for a) resistance, b) resilience, and c) temporal stability .............. 19
Table 2.2. Model selection results of dominant species interaction, nested additive, and diversity
models (dAICc ≤ 4.0) for a) resistance, b) resilience, and c) temporal stability .............. 21
Table 2.3. Results of model averaging from confidence sets (all models ΔAICc ≤ 4) for each
measure of stability ........................................................................................................... 23
Table S2.1. Percent cover of most abundant species .................................................................... 32
Table S2.2. Pearson correlation coefficients between all predictors used in “global model” for
model averaging ................................................................................................................ 33
Table 3.1. Summary of variables used in models to predict native richness relative to exotic
richness ............................................................................................................................. 41
	
Table 3.2. Top management and environmental models predicting native species richness after
controlling for association with exotic species richness (residuals of linear regression
model: native species richness ~ exotic species richness) selected by AICc.................... 48	
	
Table 3.3. Seed composition models predicting native species richness after controlling for
association with exotic species richness (residuals of linear regression model: native
speices richness ~ exotic species richness) selected by AICc .......................................... 53	
	
Table S3.1. Origin and abundance of species in dataset ............................................................... 61
Table S3.2. Full model set predicting residuals of native richness~exotic richness regression ... 67
Table S3.3. Correlation coefficients among factors used to predict native richness .................... 68	
	
Table S3.4. Variation in functional group composition of seed mixes ......................................... 69
Table S3.5. Correlation coefficients among seed mix composition factors used to predict native
richness ............................................................................................................................. 69
Table 4.1. Potential drivers that are hypothesized to determine the degree of invasion............... 83
Table 4.2. Effects of each exogenous variable each endogenous variable in SEMs .................... 92
	
Table S4.1 List of invasive species in dataset, and summary information on invasive and sown
richness and cover ........................................................................................................... 101

ix

	

Table S4.2. Goodness-of-fit metrics for structural equation models from main text and
supplementary information ............................................................................................. 106	

x

	
LIST OF FIGURES

Figure 2.1. Across-site mean biomass (+ SE) in 2011, 2012, and 2013 ....................................... 17
Figure 2.2. Effect of two components of diversity (richness and evenness) on three metrics of
stability ............................................................................................................................ 18
Figure 2.3. Effect of soil moisture and texture (soil PC1) on drought resilience ......................... 24
Figure 2.4. Timing of prescribed fire impacts stability ................................................................ 24
Figure 3.1. Native and exotic species richness was positively correlated across 29 prairie
restorations ...................................................................................................................... 47
Figure 3.2. Native species richness was higher than predicted from the native-exotic richness
relationship, in sites sown with many species at high rates ............................................ 49
Figure 3.3. Native-exotic richness relationship is not modified by fire frequency....................... 50
Figure 3.4. In older sites, native species richness is lower than predicted from native-exotic
richness relationship ........................................................................................................ 51
Figure 3.5. Interaction between seed mix richness and seeding rate is driven largely by forbs in
seed mix, and to a lesser extent by C3 grasses................................................................. 52
Figure S3.1. Change in richness over time differs by species origin ............................................ 70
Figure S3.2. Interaction between seed mix richness and seeding rate not driven by legumes or
C4 grasses in seed mix ..................................................................................................... 71
Figure S3.3. Dominant species effects differ across components of richness .............................. 72
Figure S3.4. The restoration of native richness relative to exotic richness is linked to reduced
exotic dominance ............................................................................................................. 73
Figure S3.5. Random (null) vs. observed correlation coefficient and slope of native-exotic
richness relationship ........................................................................................................ 74
Figure S3.6. Observed native-exotic richness relationship does not differ from null .................. 75
Figure 4.1. Structural equation models for invasive richness (a) and percent cover (b) .............. 87
Figure 4.2. Total standardized effects on invasive richness (a) and cover (b).............................. 88

xi

	
Figure 4.3. Structural equation models for percent cover of invasive species groups.................. 90
Figure 4.4. Total standardized effects on Poa pratensis (a), Trifolium spp. (b), and all invasive
forbs (c) ........................................................................................................................... 91
Figure 4.5. Invasive species cover is lower in sites restored from row crops than from perennial
grasslands ........................................................................................................................ 93
	
Figure 4.6. Decision tree for reducing degree of invasion in prairie restoration .......................... 97
Figure S4.1. Meta-model visualizing predictions shown in Table 4.1 ....................................... 107
Figure S4.2. Abundance of Andropogon gerardii is correlated with A. gerardii seeding rate ... 108
Figure S4.3. Structural equation model showing effects on the richness (a) and percent cover
(b) of ALL exotic species .............................................................................................. 109
Figure S4.4. The direct and indirect effects of management (fire frequency, seed mix richness)
on invasive cover differ by land-use history ................................................................. 110

xii

	
CHAPTER ONE
INTRODUCTION
Introduction
Anthropogenic forces increasingly threaten the functioning of Earth’s ecosystems and the
biodiversity they support (Vitousek et al. 1997, Foley et al. 2005). Initially motivated by
biodiversity conservation for its intrinsic value, ecologists have increasingly shifted their focus to
how biodiversity supports the way ecosystems function, and ultimately how diversity-function
relationships sustain human well-being (Hooper et al. 2005, Millennium Ecosystem Assessment
2005). Experimental evidence, bolstered by a rich body of theory (e.g., MacArthur 1955, Elton
1958, Lehman and Tilman 2000, Diaz and Cabido 2001), demonstrates that diversity plays a role
in supporting many specific functions, including primary productivity (Hector et al. 1999,
Tilman et al. 2001), invasion resistance (Fargione and Tilman 2005), nutrient cycling (Spehn et
al. 2005), and many others. Diversity-function relationships strengthen the case for both the
conservation of existing biodiversity and the restoration of biodiversity in degraded habitats
(Young et al. 2005, Cardinale 2012).
However, questions remain about how diversity-function relationships will scale up from
small-scale experimental communities (1-400m2) to dynamic, “real-world” ecosystems
(Cardinale 2012, Tilman et al. 2014). More studies are needed to understand how theoretical
predictions play out at large scales, in naturally assembled ecosystems and across realistic biotic
and abiotic gradients, in order to understand how to effectively manage both biodiversity and
ecosystem functioning. Experimental evidence suggests that, the importance of diversity may
surpass the importance of abiotic conditions for driving some functions (e.g., primary
production; Hooper et al. 2012, Tilman et al. 2012), but strongly depend on abiotic conditions for

1	

	
driving other functions (e.g., stability, Hautier et al. 2014). Experiments such as BIODEPTH,
which is replicated across sites spanning 25° of latitude from Greece to Sweden, often show
variation in the strength and direction of diversity-function relationships, due to variation in
factors like temperature and precipitation (Hector et al. 1999, Spehn et al. 2005). Diversityfunction relationships may be even less predictable in natural and managed ecosystems, due to
broader or more complex biotic and abiotic gradients. Aspects of community structure diversity, dominance, and species composition – may vary both spatially and temporally, and
may covary with abiotic factor that also drive function (Collins 2000, Ricklefs 2004).
It is also important to address diversity-function relationships explicitly in a management
context, to better understand how to translate these results into practice. Restorations, where
diversity is at least partially manipulated at large scales and across environmental gradients,
provide a unique opportunity to test theory at large scales by bridging the gap between
experiments and natural ecosystems (Young et al. 2005). Examining the causes and
consequences of biodiversity in restorations provides an opportunity to bring ecological theory to
bear on issues of immediate to concern to managers. For example, exotic species invasions are a
primary challenge to the primary goals of restoration - diverse communities of native species and
ecosystem functioning (Parker et al. 1999, Matthews and Spyreas 2010, Suding 2011, Vila et al.
2011). Examining the way exotic species invasions vary with the restoration of native
biodiversity provides a test of diversity-function theory and at the same time leads to
recommendations with direct applicability for managers (Rowe 2010). My dissertation uses a
large dataset from 29 tallgrass prairie restorations in southwestern Michigan (Grman et al. 2014)
to test basic ecological theory at large scales and in a real-world context, and to harness that
theory to inform management of restored ecosystems. Specifically, I ask:

2	

	
Chapter TWO) What is the relative importance of diversity, dominant species
abundances, and environmental factors for ecosystem stability?
Chapter THREE) Can restoration decouple predicted positive correlations between
native and exotic species richness?
Chapter FOUR) Can restoration manipulate the diversity and dominance of plant
communities to resist exotic species invasions?

The tallgrass prairie grassland of North America, as with other grasslands worldwide, is
one of Earth’s most endangered ecosystems, the majority of historical prairie having been
converted to productive farmland (Samson and Knopf 1994, Packard and Mutel 1997, Hoekstra
et al. 2005). The continued erosion of biodiversity and functioning associated with tallgrass
prairie is a consequence of extremely small patch size in fragmented habitats, and conversion to
to shrubland due to the absence of natural disturbances (especially fire) (Alstad et al. 2016). As a
result, prairie is now a major target of restoration across its historical range, where patches of
former agricultural land are increasingly sown with native prairie seeds and managed with
prescribed fire. As with any ecosystem, prairie restoration faces many challenges, including the
restoration of native diversity and control of exotic invasive species (Ruiz-Jaen and Aide 2005,
Rowe 2010, Suding 2011). Due to its widespread adoption as a land management practice across
a wide geographic area, prairie restoration provides an opportunity to address a range of basic
and applied questions in ecology (Grman et al. 2014).

3	

	
Main results and significance
I combined field observations of plant community composition and environmental
variation with data on management history to elucidate the links between the restoration of
biodiversity and the functioning of restored prairies. My research shows that, while managers
can strongly influence the restoration of diverse communities of native species, the role of
biodiversity in restored prairies differs among functions, components of the plant community,
and relative to a few key environmental variables. First, biodiversity was positively correlated
with invasion resistance but negatively correlated with ecosystem stability. Furthermore, while
diversity played a roughly equivalent role in limiting invasive species as dominant species and
environmental factors, the negative effect of diversity on stability was stronger than these other
drivers. The consensus of experimental evidence points toward positive diversity-stability
relationships, so my finding of a negative diversity-stability relationship is especially significant.
Second, the effects of environmental factors differed between the questions my research
addressed, which emphasizes how scaling up predictions from ecological theory can lead to
complex outcomes. For example, I found that soil moisture weakly limited both ecosystem
stability and the richness of invasive species. In contrast, land use history had strong effects only
on invasive species, while fire did not influence invasive species and only weakly affected
stability.
My findings also have implications for land managers. Components of seed mixes were
important in assembling communities of native species that resisted invasion, but the other aspect
of management we assessed, prescribed fire, had at most a weak influence on any outcome.
However, my research underscores the conventional wisdom that the relationship between native
and exotic species is dynamic and difficult to disentangle. Native and exotic species richness

4	

	
were positively correlated, suggesting that efforts to restore native diversity while resisting
exotic species invasions may be challenging. However, sites sown with more species at higher
rates, particularly of forb species, had higher native richness than predicted from this correlation.
The richness of sown species (a subset of native species, originating from seed mixes and not
natural colonization) was also higher in sites sown with more species, which in turn led to
reductions in invasive species (a subset of exotic species, most likely to have negative ecosystem
impacts) abundances. Therefore, despite certain limitations (i.e., the inevitability of a certain
richness and abundance of exotic species), the restoration of diversity, primarily via seed sowing,
was linked to invasion resistance, a primary goal of managers.

5	

	
CHAPTER TWO
DIVERSITY REDUCES STABILITY IN GRASSLAND RESTORATIONS
Abstract
Experimental evidence strongly suggests that local-scale biodiversity plays an important
role for stabilizing ecosystem functioning. Yet, many questions remain about how diversitystability relationships scale up to natural and managed ecosystems, where broader environmental
gradients and variable community structure may modify or weaken the importance of diversity
for stability. In 28 grassland restorations, we tested the relationship between plant species
diversity and three measures of stability in primary production - resistance and resilience to
drought, and temporal stability. We also examined the importance of diversity relative to other
putative stability drivers, including dominant species abundances, abiotic conditions (prescribed
fire and soil properties), and restoration age. Diversity was the strongest predictor of resistance
and resilience, but contrary to expectations, diversity was negatively correlated with all three
measures of stability. Diversity-stability relationships generally did not depend on other putative
drivers, as they do in many experiments. These findings illustrate that the commonly-accepted
benefits of biodiversity to stability may not consistently scale up from small-scale experiments to
natural and managed ecological systems, illustrating the critical need to evaluate the
relationships between biodiversity and aspects of ecosystem functioning, including stability,
under real world conditions,. Increasing the predictability of these relationships will require an
improved understanding of the range of conditions that lead to positive vs. negative diversitystability relationships.

6	

	
Introduction
There is a growing need to strengthen the capacity for natural habitats to sustain
ecosystem functioning, given increases in both persistent (e.g., rising mean temperature) and
periodic (e.g., drought) stresses (Millennium Ecosystem Assessment 2005, Smith et al. 2009,
Ibanez et al. 2013). Decades of theory (e.g., MacArthur 1955, Lehman and Tilman 2000) and
experiments (e.g., Tilman et al. 2006, Hector et al. 2010) suggest that the conservation or
restoration of biodiversity reinforces ecosystem stability, the constancy of ecosystem functioning
through time and in response to discrete perturbations (Pimm 1984, Griffin et al. 2009). Yet, it
remains uncertain how diversity-stability relationships scale up from small (1-400m2)
experimental plots to natural and managed ecosystems which span hundreds or thousands of
hectares and encompass broader abiotic gradients and more trophic complexity than existing
experiments (Romanuk et al. 2009, Cardinale 2012, Tilman et al. 2014). Examining diversitystability relationships in restored habitats, where diversity is manipulated at the ecosystem scale,
provides an opportunity to bridge the gap in understanding from experiments to natural
ecosystems (Young 2005, Suding 2011). Here, we examine the relationship between plant
species diversity and the stability of primary production, a common measure of ecosystem
stability, in response to drought in restored grasslands.
Plant species diversity, in particular richness, may affect several components of stability
(Pimm 1984, Ives and Carpenter 2007). Here, we focus on three common measures of stability temporal stability, resistance, and resilience of productivity (defined here as peak above-ground
biomass) (Griffin et al. 2009). Richness may reduce variation in productivity over time (temporal
stability), through both ecological and statistical mechanisms that average out the fluctuations of
individual species (Lehman and Tilman 2000). Compensatory dynamics, where an increase in

7	

	
some species is offset by decreases in others as communities respond to environmental
fluctuations or interspecific competition, should be stronger in species rich communities (Tilman
et al. 1998, Lehman and Tilman 2000). The portfolio effect, where the sum of individual species’
variances decreases with richness, is largely the product of statistical probability and also results
in positive richness-stability relationships (Doak et al. 1998, Tilman et al. 1998). Richness may
also stabilize productivity in response to discrete perturbations (resistance) and assist recovery
from discrete perturbations (resilience), although theoretical and empirical support is more
equivocal than for temporal stability (Loreau and Behera 1999, Griffin et al. 2009). The
mechanisms may be similar, however, as species rich communities are likely to include more
species tolerant of a specific perturbation (Yachi and Loreau 1999).
While relationships between plant species richness and stability have been widely
explored, the relative importance of other biotic and abiotic factors for controlling stability is less
certain (Hooper et al. 2005, Hillebrand et al. 2008, Tilman et al. 2014). Among potential biotic
drivers, components of diversity other than richness are likely to be important. Species
abundances vary in natural ecosystems (Preston 1948, Whittaker 1965, McGill et al. 2007).
Greater evenness should strengthen the stabilizing effect of richness via compensatory dynamics
and portfolio effects (Doak et al. 1998, Cottingham et al. 2001, Thibault and Connolly 2013),
and may directly underpin stability (Hillebrand et al. 2008). At low evenness, stability may
depend on how dominant species’ traits align with environmental fluctuations, especially in the
case of resistance and resilience to discrete perturbations (e.g., water use efficiency during a
drought) (Leps et al. 1982, Polley et al. 2013, Hoover et al. 2014). As such, increasing the
abundance of stable dominant species, rather than diversity, may enhance stability (Leps 2004,
Polley et al. 2007, Wilsey et al. 2014). It is also necessary to understand how the importance of

8	

	
diversity for stability varies across abiotic gradients in natural ecosystems (Cardinale 2012,
Tilman et al. 2014). For example, resource availability influences both stability and diversity
(Collins 2000, Grman et al. 2010), and may modify diversity-stability relationships (Hautier et
al. 2014, Xu et al. 2015, Zhang et al. 2016). Despite evidence that aspects of biotic communities
- richness, evenness, and dominant species – and several environmental factors control stability,
the relative importance of each remains less clear.
Ultimately, it is important to understand how diversity-stability relationships will scale
from controlled, small-scale experiments to large-scale natural and managed landscapes
(Cardinale 2012, Tilman et al. 2014). First, environmental factors may have stronger effects on
stability and diversity-stability relationships due to broader abiotic gradients at larger spatial
scales (Symstad et al. 2003, Hooper et al. 2005, Romanuk et al. 2009). Second, community
composition varies temporally and spatially at large scales, in response to abiotic variation and
multi-trophic interactions (Collins 2000, Ricklefs 2004), and the consequences for diversitystability relationships are not certain (Tilman et al. 2014). As a result, the importance of diversity
in determining stability is inconsistent across observational studies (e.g., Grman et al. 2010,
Hallett et al. 2014) and in experiments where community composition is not maintained
(Pfisterer et al. 2004, Bezemer and van der Putten 2007, Roscher et al. 2013), with some studies
finding positive, negative or neutral diversity-stability relationships. In contrast, in experiments
where richness is manipulated and composition is determined randomly, strong positive
diversity-stability relationships are observed and relatively few species are needed to maximize
stability (ca. 12 species; Tilman et al. 2006, Roscher et al. 2011). Against a backdrop of biotic
and abiotic variation, it is hard to predict whether more or less diversity is required to support
stability at large scales, or whether variation in diversity has consistent impacts on stability.

9	

	
We evaluated the relative importance of richness, evenness, dominant species and
environmental conditions for controlling three components of stability over three years in 28
grassland restorations. As severe drought conditions occurred in the second year of the study, we
analyzed both resistance and resilience to drought, in addition to temporal stability. We asked 1)
are species richness and evenness associated with stability? 2) How do diversity-stability
relationships vary with the abundance of dominant species and across environments? And, 3)
what is the relative importance of each putative driver of stability: diversity, dominant species
abundances, and environmental factors?

Methods
Study sites
We sampled 28 grassland restorations across 1300 km2 in southwest Michigan, USA
during 2011-2013 (Grman et al. 2014). Between 2003 and 2008, former agricultural sites,
ranging in size from 0.3 to 38.9 hectares (mean=5.3), were herbicided and seeded once with
native tallgrass prairie grasses and forbs. Tallgrass prairie is commonly restored using these
methods and, as a fire-dependent ecosystem, is managed with periodic prescribed fire to reduce
woody species encroachment and promote native prairie species (Packard and Mutel 1997).
Restored sites were sown with between 8 and 71 species (mean=35); observed richness was
correlated with this gradient in seed mix richness (r=0.38, p=0.04). Averaged across all sites,
sown species composed 42% of richness and 67% of cover. Non-sown species were both native
and exotic in origin. Soils were primarily sandy loams or loams (USDA-NRCS 2014), but vary
in physical (e.g, % sand) and chemical (e.g., nutrients) properties.

10	

	
Drought
Southwest Michigan experienced a record heat wave and moderate to severe drought in
summer 2012. Drought was most severe in July and early August, the peak growth period for
tallgrass prairie plants (National Drought Mitigation Center et al. 2016). Mean temperature for
June-August 2012 was 23.1 °C (8% above 1981-2010 normals), while precipitation ranged
between 0.5 and 4 cm in June (5-25% of normals), and between 3 and 12 cm in July and August
(25-75% of normals) (National Oceanic and Atmospheric Association 2016a,b).

Field sampling
We established 10 evenly spaced 1-m2 plots along one 45 m transect in the center of each
site. During peak productivity (July-September 2011), we harvested all live aboveground
biomass from each 1-m2 plot. We resampled each transect in October-November 2012 and
September–October 2013, offsetting each successive transect by 5m to avoid previously
harvested areas. Each year, we dried biomass for 72 hours at 65° C prior to weighing. In JulySeptember 2011, prior to harvesting biomass, we recorded percent cover of all (sown and nonsown) vascular plant species in each 1-m2 plot. All plants were identified to species when
possible, and were otherwise included in diversity metrics only when they clearly represented
unique taxa (Voss and Reznicek 2012).

Environmental factors
In grasslands, soil resources (moisture, nutrients) and natural disturbance regimes
(primarily fire) affect diversity and productivity (Collins 2000, Knapp and Seastedt 1986, Baer et
al. 2003), and are likely to influence diversity-stability relationships during drought

11	

	
(Dimitrakopoulos et al. 2006, Koerner and Collins 2014). For example, soil fertility and moisture
may disrupt or strengthen diversity-stability relationships (Hautier et al. 2014, Xu et al. 2015,
Zhang et al. 2016), or increase or decrease stability directly without altering diversity-stability
relationships (Leps 2004, Grman et al. 2010, Yang et al. 2012). We collected eight 20 × 3 cm2
2

soil cores at each 1-m plot in 2011 and analyzed pooled samples from each site for soil organic
matter, soil texture (sand, silt, and clay content), Mehlich-III phosphorus, Bray-II phosphorus,
and pH (Brookside Laboratories, New Knoxville, OH, USA). To assess the capacity of soil to
retain moisture and alleviate drought impacts, we calculated the proportion of oven dried soil
weight to saturated wet weight at the plot level (hereafter, soil water holding capacity; Brudvig
and Damschen 2011) and used site means for analysis. To simplify site-level soils data, we
conducted a principal components analysis. The first PCA axis (hereafter, soil PC1) explained
58% of the variance; higher values of soil PC1 represent sandier sites with lower water holding
capacity and higher phosphorus. Land managers provided information on the history of
prescribed fire at each site. We included two binomial indicator variables in our analysis: sites
burned in 2011 (n=10) and 2013 (n=4). No burns occurred in 2012, the year of the drought.
Finally, both productivity and the dependence of productivity on diversity increase over time in
experiments (Tilman et al. 2012, Tilman et al. 2014). As stability and diversity-stability
relationships may likewise be stronger in older restorations, we calculated the age (years since
sowing) of each restoration in 2011.

Data analysis
Overview: To assess the relationship between diversity and stability (Question 1), we used
multiple linear regressions testing whether richness and evenness were associated with each

12	

	
component of stability (temporal stability, resistance, and resilience). We used two different
information-theoretic approaches to address Questions 2 and 3. Information-theoretic methods
are ideal for evaluating evidence for multiple competing hypotheses, using observational data
with limited replication (Stephens et al. 2005, Aho et al. 2014). First, we used model selection
with AICc to assess whether diversity-stability relationships depend on the abundance of
dominant species and environmental conditions (Question 2). Then, we performed model
averaging to assess the relative importance of all predictors in determining stability (Question 3).
All analyses were conducted using R 3.0.3 (R Core Team 2014).

Stability, diversity, and dominant species metrics: We calculated three measures of stability
using the mean above-ground biomass/m2 for each site in each year. Temporal stability, the
inverse coefficient of variation in productivity over the years 2011-2013 (𝜇/𝜎), is a measure of
the constancy of productivity over time. Resistance to drought {ln(biomass2012/biomass2011)}
indicates the capacity to maintain productivity during drought conditions. Resilience to drought
{ln(biomass2013/biomass2011)} indicates the capacity for recovery of productivity to pre-drought
levels. For all stability metrics, higher values correspond with greater stability.
We partitioned diversity into richness and evenness. We calculated plot richness as the
number of species observed and Simpson’s evenness as E = (1/

!
!
!!! 𝑝!

) × 1/S, where 𝑝! is the

relative abundance of each 1 to S species, then averaged across plots within a site. Richness
ranged from 3.7-28.2 species/m2, and evenness from 0.15-0.38.
We classified dominant species as species with the highest percent cover in at least three
sites. Species meeting these criteria are therefore common enough to be important in our system,
and if they are particularly stable species, may contribute strongly to stability in sites where they

13	

	
are abundant. Three sown C4 grasses (Andropogon gerardii, Schizachyrium scoparium, and
Sorghastrum nutans), one non-sown exotic C3 grass (Poa pratensis), and one non-sown native
clonal forb (Solidago canadensis) were dominant in at least three sites, although abundances of
these species varied considerably in abundance across all sites (Table S2.1). These five species
also had the highest mean cover across sites. Because evenness also varied across sites, these
species ranged in abundance between 14.1% and 53.3% in sites they dominated.

Question 1 - Diversity-stability relationships: We performed linear regressions to test whether
diversity predicted each measure of stability, including both richness and evenness as
independent variables. We also evaluated whether diversity-stability relationships were nonlinear (by adding quadratic terms and performing likelihood ratio tests), but they were never
2
!
significant (LRT, 𝜒!"!!
!" ! , all p>0.10). We used partial-R to assess the relative importance of

richness and evenness.

Question 2 - Effects of dominant species abundances and environmental factors on diversitystability relationships: We used model selection to examine whether the effect of diversity on
each stability metric varied by 1) dominant species abundances or 2) environmental factors
(soilPC1, site age, 2011 burn, 2013 burn). For each stability metric, we built separate model sets
for dominant species and environmental factors. Each model set included four types of simple
models. We used a regression model testing the effect of richness and evenness on stability (first
type; “diversity model”). We also used regression models adding a single environmental factor
or dominant species to the diversity model, either with interactions with richness or evenness
(second type; “interaction model”) or without interactions (third type; “nested additive model”).

14	

	
As a baseline of no effect, we included a model with only an intercept term (fourth type;
“intercept-only model”). This approach allowed us to simultaneously evaluate support for each
“interaction hypothesis” that the effect of diversity on stability depends on each environmental
driver or dominant species. By including nested additive models, we compared support for
interaction hypotheses to more parsimonious “additive hypotheses” that each dominant species
or environmental factor independently effects stability without influencing diversity-stability
relationships.
We performed model selection with AICc, Aikake’s information criterion corrected for
small sample size (Burnham and Anderson 2002), with the ICtab function in the bbmle package
(Bolker and R Core Team 2014). To assess support for each model within each set, we calculated
the difference in AICc relative to the top-ranked model (ΔAICc) and Aikake weights (𝑤! ).
Aikake weights indicate model uncertainty, or the relative likelihood that each model is the bestfitting model, given the data and the set of models under consideration. Among models with
good to moderate support overall (ΔAICc≤7), ‘competitive’ models were those that fit as well or
better than (ΔAICc≤2) all models in which they were nested. Therefore, interaction models were
only competitive when they fit better than nested additive, diversity, and intercept-only models.
Models that fit poorly relative to the intercept-only model predicted no variation in stability.
Interaction models that fit poorly relative to the diversity or nested additive models indicate
insufficient evidence for an interactive effect, relative to the main effects of richness, evenness,
environmental factors, or dominant species abundances. For all competitive interaction models,
we examined the form of interactions graphically and calculated partial-R2 to assess the relative
importance of interactive and main effects. We also compared the log-likelihood of competitive
interaction and nested additive models to rule out models with ‘uninformative’ interaction terms

15	

	
(Arnold 2010). We standardized all continuous dependent variables to ¾=0 and σ=0.5, the
standard deviation for balanced binary predictors (Gelman 2008, Schielzeth 2010). Effect size of
2011 burns (10 of 28 sites, σ = 0.49) was therefore in the same scale as continuous predictors.
Because 2013 burns were considerably less balanced (4 of 28 sites, σ = 0.36), estimates were
biased slightly upwards.

Question 3 - Relative importance of diversity compared to other putative drivers for stability:
We used model-averaged parameter estimates and relative importance values to compare the
strength and magnitude of richness and evenness effects on stability to the effects of each
dominant species and each environmental factor. Model-averaged parameter estimates (𝜃! ) are
calculated as the sum of estimates from each model i in a model set containing that predictor,
weighted by each model’s Aikake weight ( 𝑤! 𝜃! ) (‘natural average’ method; Burnham and
Anderson 2002). When using standardized variables, model-averaging produces reliable
estimates of the relative effect size of each predictor (Grueber et al. 2011). Relative importance
values (𝑤!! =

! !"# !! 𝑤!

, where j=all models containing predictor x) indicate the importance

of each predictor in determining the response on a scale of 0 (predictor in no models) to 1
(predictor in all models).
We conducted model-averaging with the ‘dredge’ and ‘model.avg’ functions in the
MuMIn R library (Barton 2014). We generated a set of all possible models for each component
of stability, including up to four terms to avoid overfitting models, and calculated ΔAICc and 𝑤!
for each model. Variance inflation factors for all three global models (containing all diversity,
dominant species and environmental variables) were <6, and variables were generally not
strongly correlated (Table S2.2). We did not include interaction models, which can bias both

16	

	
model-averaged estimates and importance values (Galipaud et al. 2014), and because evidence
for the importance of interactions was weak (see Results). In large model sets, estimates from
models with low weights may be spurious, so we restricted further analysis to all models with
ΔAICc ≤ 4 (Burnham and Anderson 2002). For each predictor retained in each restricted model
set, we calculated 𝜃! , unconditional error estimates, and 𝑤!! (Grueber et al. 2011) and compared
the effect size of predictors where 90% unconditional confidence intervals did not include zero.
We used 90% confidence intervals to avoid rejecting potentially important drivers, especially
given the relatively small sample size (Arnold 2010).

Results
Drought effects on annual productivity
The 2012 drought reduced productivity by 46% (range: 22%-67%) between 2011 and
2012 (t27= -11.36, p < 0.001) (Figure 2.1). Productivity returned to 96% (range: 46-179%) of
pre-drought levels in 2013 (t27 = -0.67, p = 0.51).

Figure 2.1. Across-site mean biomass (+ SE) in 2011, 2012, and 2013.

17	

	
Question 1 - Diversity-stability relationships
Contrary to predictions, stability was higher in low diversity sites. Both richness and evenness
were strongly associated with lower resistance (model adj-R2 = 0.25) and resilience (model adjR2 = 0.15) to drought (Figure 2.2). Richness (but not evenness) was weakly associated with
lower temporal stability (model adj-R2 = 0.07) (Figure 2.2).

Figure 2.2. Effect of two components of diversity (richness and evenness) on three metrics
of stability. Both plant species richness and evenness predict lower resistance (model R2 = 0.25)
(a,b) and resilience (model R2 = 0.15) (c,d). Richness also predicts lower temporal stability (e),
which is not correlated with evenness (f) (model R2 = 0.07). Component residual plots showing
effect of richness controlling for effect of evenness, and vice versa.

a)

c)

e)

-0.2

partial-R2=0.29

-0.8

0.0
-0.2
-0.4

-1.0

Temporal Stability

Resilience

Resistance

partial-R2=0.16

0.2

-0.6

15

20

25

5

10

15

20

3.0
2.5

25

Plant Species Richness

d)

b)

3.5

1.5

-0.8
10

4.0

2.0

-0.6

5

partial-R2=0.14

4.5

0.4

-0.4

5

10

15

20

25

f)

0.6
0.4

4.0

Temporal Stability

-0.4

Resilience

Resistance

0.2
-0.6

-0.8

0.0
-0.2
-0.4

-1.0

3.5

3.0

2.5

2

partial-R =0.16
0.15

0.20

0.25

-0.6

0.30

0.35

2.0

partial-R2 =0.17
0.15

0.20

0.25

0.30

Plant Species Evenness

18	

0.35

partial-R2 =0.04
0.15

0.20

0.25

0.30

0.35

	
Question 2 - Effects of dominant species abundances and environmental factors on diversitystability relationships
There was only weak evidence that diversity-stability relationships varied with dominant
species abundances or the timing of prescribed fire. At least one interaction model occurred in
most sets of competitive models (Tables 2.1, 2.2). However, competitive interaction models
predicted stability no better than (within 2 AICc) diversity or nested additive models, and either
interaction terms explained little variation in stability (small partial-R2) or support for interaction

Table 2.1. Model selection results of environmental interaction, nested additive, and
diversity models (dAICc ≤ 4.0) for a) resistance, b) resilience, and c) temporal stability.
Competitive models in bold. Env=Environmental factor. Standardized effect size shown for each
predictor in model (**p<0.01, *p<0.05, +p<0.10), with partial-R2 in parentheses. Terms not in
model indicated with a dash (-). For each model: logLik=log-likelihood, dAICc=delta AICc.
a) Resistance
Model type

Factor

Richness

Evenness

Env

Env*R

Env*E

diversity

-

-0.24** (0.29)

-0.16* (0.16)

-

-

-

nested additive

soilPC1

-0.24** (0.29)

-0.17* (0.16)

-0.03 (0.01)

-

-

nested additive

burn2011

-0.22* (0.18)

-0.16+ (0.15)

-0.03 (0.01)

-

-

nested additive

age

-0.24** (0.25)

-0.16* (0.16)

0.00 (0.00)

-

-

	

	

	

	

	

	

Model type

Factor

Richness

Evenness

Env

Env*R

Env*E

b) Resilience
interaction

burn2013

-0.32** (0.27)

-0.12 (0.21)

-0.23** (0.26)

-

-0.09

nested additive

burn2013

-0.36** (0.28)

-0.25* (0.19)

0.40* (0.24)

-

-

nested additive

soilPC1

-0.28* (0.20)

-0.28* (0.20)

-0.22* (0.16)

-

-

interaction

burn2013

-0.26* (0.18)

-0.26* (0.18)

-0.20* (0.24)

0.01 (0.00)

-

Model type

Factor

Richness

Evenness

Env

Env*R

Env*E

intercept

-

-

-

-

-

-

interaction

burn2011

-0.27 (0.03)

-0.03 (0.03)

0.25 (0.10)

-

0.53+ (0.15)

c) Temporal Stability

diversity

-

-0.59+ (0.14)

-0.30 (0.04)

-

-

-

nested additive

burn2011

-0.27 (0.02)

-0.23 (0.03)

-0.51 (0.09)

-

-

interaction

burn2011

-0.20 (0.02)

-0.15 (0.01)

0.31+ (0.09)

-0.40 (0.06)

-

nested additive

soilPC1

-0.58+ (0.13)

-0.29 (0.04)

0.19 (0.02)

-

-

nested additive

age

-0.51 (0.09)

-0.29 (0.04)

0.15 (0.01)

-

-

19	

	
Table 2.1. (cont’d)
a) Resistance
Model type

Factor

logLik

dAICc

df

weight

R2

diversity

-

10.5

0

4

0.46

0.25

nested additive

soilPC1

10.6

2.7

5

0.12

0.22

nested additive

burn2011

10.6

2.8

5

0.12

0.22

nested additive

age

10.5

3

5

0.1

0.22

b) Resilience

	

	

	

	

	

Model type

Factor

logLik

dAICc

df

weight

R2

interaction

burn2013

2.6

0

6

0.39

0.38

nested additive

burn2013

0.8

0.3

5

0.33

0.32

nested additive

soilPC1

-0.5

2.9

5

0.09

0.26

interaction

burn2013

0.8

3.6

6

0.07

0.3

Factor

logLik

dAICc

df

weight

R2

intercept

-

-29.8

0

2

0.26

0

interaction

burn2011

-24.2

0.3

6

0.22

0.21

c) Temporal Stability
Model type

diversity

-

-27.7

1.1

4

0.15

0.07

nested additive

burn2011

-26.4

1.5

5

0.12

0.12

interaction

burn2011

-25.5

3

6

0.06

0.13

nested additive

soilPC1

-27.4

3.5

5

0.04

0.05

nested additive

age

-27.6

3.8

5

0.04

0.04

models was similar to the intercept-only model (Table 2.1, 2.2). This suggests that, even among
competitive interaction models, interactions were weak overall (Figure S2.1), and never more
important than the main effects of diversity, environmental factors, and dominant species.

Question 3 - Relative importance of diversity compared to other putative drivers for stability
Diversity and environmental factors (especially prescribed fire) were important predictors
of stability, while dominant species played a more limited role (Table 2.3). Effects on temporal
stability were weak overall, while several factors predicted both resistance and resilience well.
To compare the effect size of putative stability drivers, we present only model-averaged
parameter estimates here (𝜃! ); see Table 2.3 for full summary of statistics, including
unconditional error estimates and relative importance values. Greater abundances of S.

20	

	
canadensis reduced resistance (𝜃! = -0.14) (Figure S2.2a), but it did so more weakly than
richness (-0.26) and evenness (-0.17) (Table 2.3). Richness (-0.35) and evenness (-0.26) were
also strongly associated with lower resilience to drought, when compared to sites where soil is
Table 2.2. Model selection results of dominant species interaction, nested additive, and
diversity models (dAICc ≤ 4.0) for a) resistance, b) resilience, and c) temporal stability.
Competitive models in bold. Dom=Dominant species. Standardized effect size shown for each
predictor in model (**p<0.01, *p<0.05, +p<0.10), with partial-R2 in parentheses. Terms not in
model indicated with a dash (-). For each model: logLik=log-likelihood, dAICc=delta AICc.
Andger = Andropogon gerardii, Poapra = Poa pratensis, Schsco = Schizachyrium scoparium,
Solcan = Solidago canadensis, Sornut = Sorghastrum nutans.
a) Resistance
Model type

Factor

Richness

Evenness

Dom

Dom*R

Dom*E

nested additive

Solcan

-0.27*** (0.37)

-0.17* (0.21)

-0.15* (0.18)

-

-

interaction

Solcan

-0.26** (0.39)

-0.20* (0.26)

-0.14* (0.20)

0.22 (0.08)

-

diversity

-

-0.24** (0.29)

-0.16* (0.16)

-

-

-

interaction

Solcan

-0.27** (0.34)

-0.17* (0.21)

-0.15* (0.18)

-

0.01 (0.00)

b) Resilience
Model type

Factor

Richness

Evenness

Dom

Dom*R

Dom*E

diversity

-

-0.26* (0.16)

-0.28* (0.17)

-

-

-

intercept-only

-

-

-

-

-

-

nested additive

Schsco

-0.26* (0.15)

-0.29* (0.18)

-0.07 (0.02)

-

-

nested additive

Sornut

-0.30* (0.16)

-0.27* (0.17)

0.08 (0.01)

-

-

nested additive

Andger

-0.29* (0.16)

-0.28* (0.17)

-0.06 (0.01)

-

-

nested additive

Solcan

-0.27* (0.16)

-0.28* (0.17)

-0.03 (0.00)

-

-

nested additive

Poapra

-0.26* (0.15)

-0.28* (0.17)

0.00 (0.00)

-

-

c) Temporal Stability
Model type

Factor

Richness

Evenness

Dom

Dom*R

Dom*E

nested additive

Schsco

-0.55+ (0.14)

-0.40 (0.08)

-0.53+ (0.15)

-

-

intercept-only

-

-

-

-

-

-

diversity

-

-0.59+ (0.14)

-0.30 (0.04)

-

-

-

interaction

Schsco

-0.55+ (0.15)

-0.38 (0.08)

-0.60* (0.16)

-

-0.52 (0.05)

interaction

Schsco

-0.43 (0.14)

-0.32 (0.05)

-0.53* (0.16)

0.62 (0.05)

-

nested additive

Solcan

-0.65* (0.17)

-0.32 (0.05)

-0.39 (0.08)

-

-

nested additive

Sornut

-0.79* (0.18)

-0.27 (0.03)

0.36 (0.05)

-

-

nested additive

Andger

-0.46 (0.07)

-0.29 (0.04)

0.25 (0.03)

-

-

interaction

Solcan

-0.62* (0.18)

-0.41 (0.08)

-0.37 (0.09)

-0.71 (0.05)

-

21	

	
Table 2.2. (cont’d)
a) Resistance
Model type

Factor

logLik

dAICc

df

weight

R2

nested additive

Solcan

13.3

0

5

0.40

0.36

interaction

Solcan

14.5

1

6

0.25

0.38

diversity

-

10.5

2.7

4

0.11

0.25

interaction

Solcan

13.3

3.3

6

0.08

0.33

Model type

Factor

logLik

dAICc

df

weight

R2

diversity

-

-3

0

4

0.28

0.15

intercept-only

-

-6.3

1.5

2

0.14

0

nested additive

Schsco

-2.7

2.6

5

0.08

0.13

nested additive

Sornut

-2.8

2.6

5

0.08

0.13

nested additive

Andger

-2.8

2.7

5

0.07

0.12

nested additive

Solcan

-2.9

2.9

5

0.07

0.12

nested additive

Poapra

-3

3

5

0.06

0.12

Model type

Factor

logLik

dAICc

df

weight

R2

nested additive

Schsco

-25.4

0

5

0.23

0.18

intercept-only

-

-29.8

0.4

2

0.19

0

diversity

-

-27.7

1.6

4

0.11

0.07

interaction

Schsco

-24.7

1.8

6

0.10

0.18

interaction

Schsco

-24.8

1.9

6

0.09

0.18

nested additive

Solcan

-26.5

2.1

5

0.08

0.11

nested additive

Sornut

-27

3.2

5

0.05

0.08

nested additive

Andger

-27.3

3.8

5

0.03

0.05

interaction

Solcan

-25.8

4

6

0.03

0.11

b) Resilience

c) Temporal
Stability

sandy with low water holding capacity (-0.22) (Figure 2.3), and the abundance of A. gerardii is
high (-0.21) (Figure S2.2b) (Table 3). Sites burned in 2013 were more resilient (0.40) (Figure
2.4a), but this effect size may be biased upward, as only 4 sites were burned in 2013. Temporal
stability was lower in sites burned in 2011 (-0.66) (Figure 2.4b). 2011 burns reduced temporal
stability more strongly than Schizachyrium scoparium abundance (-0.54) (Figure 2.S2c, Table

22	

	
2.3). Richness and evenness did not strongly predict temporal stability when controlling for other
factors.

Table 2.3. Results of model averaging from confidence sets (all models ΔAICc ≤ 4) for each
measure of stability. Model-averaged estimate (𝜃! ), unconditional (adjusted) SE, 90% CI, and
relative importance value (Imp) shown for all predictors retained in each confidence set. Values
in bold indicate estimates where 90% CI did not include zero; predictors in bold are important
for at least one stability metric. Effects of 2013 burns (Burned 2013) on resistance not modeled;
S. scoparium, S. nutans, P. pratensis, S. canadensis, Burned 2011, and Restoration age not
retained as predictors in confidence set for resilience.

𝜃!

Resistance
(90%
SE
CI)

Imp

𝜃!

(-0.39,
-0.13)
(-0.30,
-0.05)

1.00

Diversity
Richness

-0.26

0.08

Evenness

-0.17

0.08

Dominant spp.
A. gerardii

0.00

0.02

S. scoparium

-0.01

0.03

S. nutans

0.00

0.02

P. pratensis

0.00

0.02

S. canadensis

-0.14

0.08

Env. Factors
Soil PC1

0.00

0.03

Burned 2011

0.00

0.02

Burned 2013

-

-

0.00

0.02

Restoration
age

Resilience
(90%
SE
CI)

Imp

-0.35

0.11

1.00

-0.43

0.33

0.94

-0.26

0.11

(-0.54,
-0.17)
(-0.43,
-0.09)

0.88

-0.28

0.32

(-0.16,
0.10)
(-0.17,
0.05)
(-0.18,
0.11)
(-0.13,
0.09)
(-0.26,
-0.04)

0.07

-0.21

0.12

0.12

0.25

0.31

0.11

-

-

(-0.41,
-0.01)
-

-

-0.54

0.32

0.07

-

-

-

-

0.54

0.39

0.07

-

-

-

-

0.01

0.28

0.91

-

-

-

-

-0.34

0.28

(-0.16,
0.06)
(-0.15,
0.13)
-

0.09

-0.22

0.10

0.86

0.40

0.29

0.06

-

-

(-0.39,
-0.07)
-

-

-0.66

0.34

-

0.40

0.14

1.00

-0.05

0.46

(-0.12,
0.13)

0.06

-

-

(0.17,
0.64)
-

-

0.39

0.41

23	

Temporal Stability
(90%
𝜃!
SE
CI)
Imp
(-0.96,
0.11)
(-0.80,
0.24)

0.26

(-0.26,
0.76)
(-1.07,
-0.02)
(-0.09,
1.17)
(-0.44,
0.46)
(-0.79,
0.11)

0.08

(-0.08,
0.87)
(-1.22,
-0.10)
(-0.80,
0.70)
(-0.27,
1.05)

0.31

0.11

0.46
0.27
0.04
0.22

0.59
0.04
0.13

	
Figure 2.3. Effect of soil moisture and texture (soil PC1) on drought resilience. Sites with
higher clay and water holding capacity (WHC) are more resilient. Residual plot controlling for
effects of other variables identified as important from model averaging (see Table 3.3).

0.4

Resilience

0.2
0.0

-0.2
-0.4
-0.6
-4

-2

0

2

4

clay/high WHC [soilPCA] sand/low WHC

Figure 2.4. Timing of prescribed fire impacts stability. a) Post-drought fire supports resilience
while b) pre-drought fire reduces temporal stability. Residual plots controlling for effects of
other variables identified as important from model averaging (see Table 2.3).
a)

b)

0.6

4.5

Temporal Stability

Resilience

0.4
0.2
0.0

-0.2
-0.4

4.0
3.5
3.0
2.5
2.0

Not Burned 2013

Burned 2013

Not Burned 2011

24	

Burned 2011

	
Discussion
Theory (Yachi and Loreau 1999, Lehman and Tilman 2000) and experiments (Gross et
al. 2014, Isbell et al. 2015) demonstrate that diversity can stabilize community productivity, but
the importance of diversity relative to biotic and abiotic environmental variation at the scale of
natural communities is poorly understood (Cardinale 2012, Tilman et al. 2014). Here, we
analyzed the relationship between plant species diversity and stability in 28 grassland
restorations, revealing two important findings. First, both components of diversity, richness and
evenness, were strongly associated with lower resistance and resilience to drought, and richness
was weakly associated with lower temporal stability. Second, the negative effects of richness and
evenness on resistance and resilience were at least as strong as environmental factors and
dominant species abundances. The most stable sites had lower species richness and evenness,
were burned after and not before drought; had soils high in clay and moisture retention capacity,
but low in phosphorus; and tended to have low abundances of some dominant species
(Andropogon gerardii, Schizachyrium scoparium, Solidago canadensis).

Diverse sites were less stable
What distinguishes our system, where stability was lower in diverse sites, from the
majority of studies that find positive diversity-stability relationships? Diversity may be more
likely to reduce stability: 1) in certain communities or under certain perturbations, such as
grassland plant communities under drought, 2) in certain community structures (i.e.,
composition), 3) when community composition is free to vary through natural colonization and
extinction dynamics, particularly during early succession, or 4) under certain environmental
conditions. We discuss each of these hypotheses below.

25	

	
First, studies of other grassland plant communities have detected negative diversitystability relationships in response to drought (Pfisterer and Schmid 2002, Van Peer et al. 2004,
van Ruijven and Berendse 2010, Vogel et al. 2012). However, diversity in algal and microbial
communities has also been associated with reduced stability (Petchey et al. 2002, Allison 2004,
Gonzalez and Descamps-Julien 2004, Zhang and Zhang 2006), so negative relationships are not
unique to grassland plant communities. Furthermore, negative diversity-stability relationships
occur in the same community types, across similar richness gradients, and in response to the
same perturbations where positive and neutral relationships are more frequently observed
(Griffin et al. 2009). Therefore, it seems unlikely that negative diversity-stability relationships
are a distinctive property of grasslands plant communities under drought.
Second, variation in community composition, rather than diversity per se, may drive
stability. Other grassland studies show that the stability of dominant species often determines
community stability (Smith et al. 2004, Polley et al. 2007). Low diversity sites in our system may
be more stable because a large proportion of biomass in these sites is composed of stable
dominant species that are less abundant in high diversity sites. A single dominant species,
Andropogon gerardii, was more abundant in low-richness sites, while other dominant species
were either less abundant or not correlated with richness (Table S2.2, S2.4). Regardless, if
particular dominant species contribute disproportionately to community stability, sites in which
they are most abundant should be more stable. Instead, dominant species abundances were poor
predictors of stability. The abundance of three dominant species, A. gerardii, Schizachyrium
scoparium, and Solidago canadensis, was associated with lower stability. While S. canadensis is
typically associated with drought-sensitivity (Hoover et al. 2014, Smith et al. 2016), the weak
and negative effect among C4 grasses (A. gerardii, S. nutans, S. scoparium) is surprising as they

26	

	
frequently support stability in both remnant and restored grasslands (e.g., Tilman and Downing
1994, Smith et al. 2004, Polley et al. 2007). Finally, when controlling for the effects of
dominance and dominant species abundances, species richness was still strongly associated with
lower stability. Even in regression models where higher dominance (low evenness) predicted
greater stability, the independent negative effect of richness was at least as strong as the effect of
evenness in the same models. Controlling for other components of dominance also did not alter
how richness and evenness influenced stability (Table S2.3).
Third, communities experiencing successional changes may respond differently to
perturbations than communities that are closer to equilibrium. The way composition changes
through time, via colonization and extinction of both sown and non-sown species, may alter
diversity-stability relationships. Diversity-stability relationships are typically weak in grassland
experiments allowing natural colonization, compared to plots where composition is maintained
(Pfisterer et al. 2004, Bezemer and van der Putten 2007, Roscher et al. 2013). Composition, as
well as diversity and abundance, shifts in grassland restorations over at least the first decade of
establishment (Sluis 2002, Heslinga and Grese 2010, Grman et al. 2013). In young (9 and 20 yr)
grassland restorations in Texas, stability was negatively correlated with species richness as in our
study, and not correlated with richness in nearby remnant grasslands (Polley et al. 2007).
Additionally, root systems and soil microbial communities likely mediate aboveground drought
responses and may be poorly developed in young communities restored in former agricultural
land (Antoninka et al. 2009, Bauer et al. 2015).

27	

	
Diversity effects on resistance and resilience were strong relative to environmental drivers
Although negatively correlated, diversity was a strong predictor of stability in our system,
relative to putative environmental drivers. Notably, diversity was the strongest predictor of
resistance and resilience, even when controlling for covariation between diversity and several
factors known to drive both diversity and stability in grasslands. Only one environmental driver
(fire) in our study was stronger than diversity for one measure of stability (temporal stability),
although neither diversity nor fire predicted temporal stability well. Experimental manipulations
of precipitation (Xu et al. 2014), disturbance (mowing; Vogel et al. 2012) and fertilization
(Hautier et al. 2014, Xu et al. 2015), on the other hand, frequently weaken or overwhelm
diversity-stability relationships (but see Tilman et al. 2006, Xu et al. 2014). Experimental
treatments (nitrogen and water addition, mowing) likely have stronger effects because they are
applied at much greater levels than the analogous environmental effects we modeled (soil
nutrients and water holding capacity, and prescribed fire), and vary independently of other biotic
and abiotic conditions. Our results contrasted with experimental results in other ways, as well,
particularly how the importance of diversity and abiotic treatments vary across components of
stability. In experiments, both are likely to drive temporal stability and resistance, but not
resilience (e.g., Tilman et al. 2006, Isbell et al. 2015). Diversity in our system, in contrast, had
particularly strong (negative) effects on both resistance and resilience, but not temporal stability.
Environmental drivers were linked to temporal stability and resilience, but not resistance.
Diversity-stability relationships were also never negative in experiments explicitly comparing
diversity and environmental factors. When diversity is destabilizing, as in our system, it may
have stronger impacts on both resistance and resilience than other drivers. Therefore, the relative

28	

	
importance of diversity and environmental factors likely depends on both the direction of
diversity effects and the intensity of environmental gradients.
The timing of fire, and to a lesser extent the influence of soil properties, had important
consequences for stability. Pre-drought fire reduced temporal stability, while post-drought fire
supported resilience. Pre-drought fires may have increased drought susceptibility by stimulating
a pulse in productivity and reducing soil resources (Blair 1997, Turner et al. 1997, Craine and
Nippert 2014). This suggests, however, that sites burned in 2011 should also be less resistant and
possibly less resilient, which we did not find. Moisture and texture defined the strongest
gradients in soil resources (soil PC1). Unsurprisingly, less sandy sites with greater soil water
holding capacity were better able to rebound from drought. Sites burned in 2013, in contrast,
rebounded strongly. Together, this suggests that stability was driven in part by the fact that fire
stimulated productivity in non-drought years, but productivity at all sites was limited by water
holding capacity during drought.

Conclusions
In contrast to the extensive experimental evidence that plant species richness stabilizes
community productivity (Gross et al. 2014, Isbell et al. 2015), we found that both plant species
richness and evenness in grassland restorations was associated with lower stability. This negative
association of diversity with resistance and resilience to drought was 1.2-1.9 times stronger than
other putative stability drivers, including prescribed fire, soil properties, site age, and dominant
species’ abundance. Diversity-stability relationships vary widely at large scales, ranging from
weakly negative to strongly positive in both naturally-assembled plant communities (e.g., Grman
et al. 2010, Hallett et al. 2014) and other grassland restorations (Biondini 2007, Polley et al.

29	

	
2007, Carter and Blair 2012), likely because a complex suite of biotic and abiotic factors drive
temporal and spatial variation in diversity-stability relationships. Better predictions of when
large-scale diversity-stability relationships are positive vs. negative or neutral, will likely require
large-scale experiments that replicate species diversity manipulations across realistic
environmental gradients.

Acknowledgements
I would like to thank the legions of worker bees who helped clip, sort, dry, and weigh
biomass over the three-year period of this project. Insights from Kane Keller, Liz Schultheis,
Tomomi Suwa, Susan Magnoli, Mark Hammond, were invaluable to the development of this
paper; the analysis was improved by advice by Colin Kremer and Beth Miller. Portions of this
work was funded by USDA-NIFA grant 2013-67011-21281.

30	

	

APPENDIX

31	

	
Table S2.1. Percent cover of most abundant species. Species dominant at site in bold. Andger
= Andropogon gerardii, Broine = Bromus inermis, Hiespp = Hieracium species, Junten = Juncus
tenuis, Monfis = Monarda fistulosa, Panvir = Panicum virgatum, Poapra = Poa pratensis,
Rubocc = Rubus occidentalis, Schsco = Schizachyrium scoparium, Solcan = Solidago
canadensis, Solspe = Solidago speciosa, Sornut = Sorghastrum nutans, Tripra = Trifolium
pratense.
site

Andger

Broine

Hiespp

Junten

Monfis

Panvir

Poapra

Schsco

Solcan

Solspe

Sornut

Tripra

1

37.5

0

0

0

0

0

3

0.3

0

0

10.6

0

2

29

0

0

0

0

0

0

3.25

3.1

0

0.73

0

3

7.7

0.04

0.01

0.04

0

1.41

0.08

27.9

5.78

0

6.24

0

4

4.6

0

0

0.23

0

0.31

0.02

11.3

30.5

0

10.2

0

5

8.9

0.2

0

0.77

3.2

3.15

11.9

0.2

0.1

0

41.5

2.26

6

39

0

0

0

6.7

2.6

0

7

0.1

0

10.7

0

7

10.4

0

0

0.01

0.2

1.15

2.05

7.81

0.4

0

28.5

3.06

8

17

0

0

17.2

3.8

0

14

8.2

5.6

0.2

4.1

0

9

21.6

0

0.03

0.3

2.95

0

38.5

7.3

1.38

0.2

4.6

0.07

10

2.5

1.25

0

0

4.04

0.2

0

27.5

1.2

0

12.7

0

11

33

0

0

0

3.3

2.7

7.32

5.35

0

0

3.72

0

12

18.3

0

0

0.45

0.141

0.65

0

4.74

0

0

16

0

13

11.1

0

0.08

0

0.12

16.83

0.13

0

11.5

0

4

0

14

25.8

0

0

0

0.46

2.6

34.5

2.46

0

0

12.6

0.01

15

3

2.55

0.1

0

0.9

0.05

53.3

8.5

23.7

0

0.33

0.01

16

1

5.6

0

0

4.6

0

31.3

2

10.9

0.33

8

0.01

17

2.7

0.02

0

4.8

0.03

0.4

2.2

12.7

0.2

0.6

46

0

18

8

0

0

0

0

0

2.65

30.5

3.04

0

5.3

0

19

3.2

0

0

5.07

0

0

0.03

8.25

4.9

14.5

15.3

0

20

0

0

0

0

1.35

0.4

4.7

8.5

0

0

5.3

33.5

21

45

0

0

0

0.4

0.05

0

2.82

14.3

0

3.04

0

22

10.2

0

29.5

0

0

0.1

0.03

8.1

0.04

0

16.6

0

23

25.2

0

0

0

2.05

0.2

0.1

0.2

15.24

0

1.75

0

24

7.8

14

0

0

3.64

0.9

6.4

3.2

14.2

0

8.5

0

25

4.7

0.07

0

0

2.6

0.5

21.3

36

3

0

4.2

0

26

13.7

0

0

0.1

3.78

1.4

0

9.3

0.42

2.8

25

0

27

14.1

0

0

0

3.98

0.001

0

2.87

0.5

0

2

0

28

16.6

0

0

0

13.92

0.011

0

3.56

16.8

0

1.7

0

# sites
dominant

8

0

1

1

0

1

4

4

3

0

5

1

mean
%
cover

15.06

0.85

1.06

1.03

2.22

1.27

8.34

8.92

5.96

0.67

11.04

1.39

32	

	
Table S2.2. Pearson correlation coefficients between all predictors used in “global model”
for model averaging. ***p < 0.001, **p < 0.01, *p < 0.05, +p < 0.10. 	

Soil
PCA

Age

2011
Burn

2013
Burn

Evenness

Evenness

Richness
-0.44*

Soil PCA

-0.03

-0.01

Age
2011
Burn
2013
Burn
Andger

-0.45*

0.16

-0.07

0.59**

-0.15

0.04

-0.57**

0.33+

-0.21

-0.02

-0.45*

0.33+

-0.49**

0.20

-0.27

-0.48*

-0.05

Sornut

0.57**

-0.31

0.01

0.53**

0.62***

-0.26

Schsco

0.16

-0.23

0.37*

0.47*
0.75***
-0.15

0.45*

-0.08

-0.42*

-0.04

Solcan

-0.15

0.01

-0.13

0.11

0.03

-0.24

-0.15

-0.37*

-0.12

Poapra

-0.11

-0.08

-0.09

0.09

-0.20

-0.14

-0.15

-0.17

-0.02

33	

Andger

Sornut

Schsco

Solcan

0.16

	
CHAPTER THREE
MORE IS BETTER – SOWING MORE SPECIES AT HIGHER DENSITIES SUPPORTS
GREATER NATIVE SPECIES RICHNESS IN PRAIRIE RESTORATIONS THAN
PREDICTED BY NATIVE-EXOTIC RELATIONSHIP
Abstract
Given dramatic losses of native species due to human activities, the restoration of native
biodiversity in degraded habitats is a primary aim of land managers. A major challenge to
restoration is to design strategies that simultaneously benefit the establishment of native species
and discourage the establishment of exotic species. A better understanding of positive
correlations between native and exotic species richness, a pattern that is nearly ubiquitous at
large scales in plant communities, may help managers modify these correlations to favor native
plant species during restoration. Across 29 tallgrass prairies restored through seed sowing onto
former agricultural lands, we examined: whether the relationship between native and exotic
richness is 1) altered by management, such as seed additions and prescribed fire, 2) controlled
instead by environmental conditions and successional processes, or 3) whether management is
effective at altering native-exotic richness relationships in certain environments and limited in
others. As is commonly found, native and exotic richness were positively correlated at large
scales (e.g., across sites). However, management actions explained much of the remaining
variation in native richness, while environmental conditions explained very little. Sites sown
with more species at higher seeding rates, especially forb species, had higher native richness than
predicted by the native-exotic relationship. In contrast, older sites had proportionally lower
native richness, because native richness, and not exotic richness, declined with restoration age.
We show that management actions, such as seed sowing, can modify the native-exotic richness

34	

	
relationship to favor native species during restoration. The development of management actions
that proportionally favor native species over time will further benefit native species restoration.

Introduction
Anthropogenic habitat degradation leads to native plant species loss (Vitousek et al.
1997, Foley et al. 2005, Rockstrom et al. 2009), so a primary goal of restoring habitats is to
reassemble a diverse native community (Ruiz-Jaen and Aide 2005, Brudvig 2011). This goal is
addressed directly (e.g., plant or seed additions), as well as indirectly through management
actions designed to increase native diversity and resist invasion by exotic species (e.g.,
prescribed fire, carbon addition) (Suding 2011). Yet, efforts to restore native diversity are
frequently limited by exotic species invasions (Didham et al. 2007, Matthews and Spyreas 2010,
Kettenring and Adams 2011). Previously, we have compared a range of management and
environmental factors to explain variation in both native and exotic species (see Chapter 4;
Grman et al. 2013). Here, we address the challenge of simultaneously promoting native species
and reducing exotic species in the context of native-exotic richness relationships (Fridley et al.
2007).
The correlation between native and exotic species richness has been well explored in both
experimental systems and natural habitats, and provides a useful theoretical framework for
understanding how the restoration of diverse native-dominated communities is connected to
limiting exotic species invasions (Suding and Gross 2006, Fridley et al. 2007, Hill and Fischer
2014). Early theory predicted negative native-exotic richness relationships, because resources in
diverse native communities should be more fully exploited, leaving less niche-space for exotic
species to invade (Elton 1958, Levine et al. 2004). Therefore, restorations that increase native

35	

	
richness should also reduce exotic richness. Results from many small-scale observational and
experimental communities support this theory and find that exotic richness is negatively
correlated with native richness (e.g., Kennedy et al. 2002, Brown and Peet 2003. However,
several decades of observations and experiments across a range of habitat types reveal that
correlations can range from negative to positive to neutral, and are particularly variable at small
scales (Fridley et al. 2007, Guo 2015).
At large scales, on the other hand, native and exotic richness correlations are consistently
positive (Fridley et al. 2007). In other words, at scales most relevant to restoration, natural
habitats with more native species also support more exotic species. Positive correlations between
native and exotic species richness may result from similar responses of these groups to
environmental gradients of increasing resource availability (Stohlgren et al. 1999, Davis et al.
2000) and increasing heterogeneity of resources (Davies et al. 2005, Melbourne et al. 2007), the
frequency and intensity of disturbance (Taylor and Irwin 2004, Belote et al. 2008), and
immigration rates (Tilman 1997). Given the range of potential causes, positive correlations may
also result if native and exotic richness respond to different drivers operating simultaneously
(Carboni et al. 2010, Parker et al. 2010, Jauni and Hyvonen 2012, Brummer et al. 2016). Nativeexotic richness relationships may also strengthen or weaken over time, depending on whether
competition-colonization tradeoffs or other successional processes lead to disproportionate losses
of either native or exotic species (Tilman 1994, Molina-Montenegro 2012). Null modeling
approaches also predict positive correlations at large scales, in the absence of these niche-based
factors, if communities are assembled randomly from a regional species pool with a fixed ratio of
native to exotic species (Fridley et al. 2004, Herben et al. 2004). Regardless of the mechanism,

36	

	
restoration of native richness at meaningfully large spatial scales may also facilitate or encourage
increases in exotic richness.
A better understanding of native-exotic richness relationships may improve the ability of
managers to design restoration strategies to proportionally favor native species. Importantly, if
multiple drivers underlie positive native-exotic correlations but affect native and exotic species
differently, managers may be able to identify which drivers more strongly support native species,
but inhibit or minimally affect exotic species, and tailor their approaches accordingly. Here, we
assess the relationship between native and exotic plant species richness in former agricultural
land restored to tallgrass prairie grasslands through seed sowing. We also explore whether
management and a range of environmental factors shift native-exotic richness relationships to
favor native species.
Both seed addition and fire are considered critical for restoring native plant diversity in
tallgrass prairie (Foster et al. 2007). Increases in native richness with seed additions indicate
chronic seed limitation in this system (Carter and Blair 2012, Grman et al. 2013, Foster et al.
2015), due to the contemporary rarity of many tallgrass prairie species across their historical
range (Samson and Knopf 1994). However, the effectiveness of native seed additions to restore
species composition and diversity in prairies may still be limited, particularly under some abiotic
conditions, for example those related to soil moisture and fertility (Foster et al. 2004, Myers and
Harms 2011). The necessity of fire is indicated in part by the decline in plant species diversity in
remnant prairie habitats over decadal time scales as a result of fire suppression (Leach and
Givnish 1996, Alstad et al. 2016). However, whether burning increases or decreases native
richness relative to exotic richness may depend on other factors, such as fire frequency and
edaphic conditions (Howe 1994, Collins 2000, Howe 2011, Bowles and Jones 2013). Few

37	

	
studies have explicitly considered whether particular restoration actions like seed sowing and fire
can increase native richness by explicitly modifying native-exotic richness relationships (though
see Suding and Gross 2006, Hill and Fischer 2014) and many contingencies need to be addressed
in order to provide concrete recommendations on how managers can increase native relative to
exotic richness. For example, virtually nothing is known about how seeding density and richness
influence native-exotic relationship, or how strongly these effects depend on the identity of
species that are sown. A better understanding is also needed of how the effect of seed additions
and fire on native-exotic richness relationships varies across the environmental gradients thought
to structure these relationships, such as environmental variation in resources, disturbance, and
landscape characteristics influencing immigration rates (Tilman 1997, Stohlgren et al. 1999).
Previous studies exploring how management modifies existing native-exotic richness
relationships have generally been conducted in sites containing intact native vegetation, and at
small spatial scales (< 500 hectares) and across narrow environmental gradients that limit the
ability to make general predictions (Suding and Gross 2006, Hill and Fischer 2014). In contrast,
we examined native-exotic species richness relationships across 29 sites restored with native
prairie seed from bare soil in former agricultural land, within landscapes supporting few native
prairie species. These sites were therefore intrinsically seed-limited at the time of restoration and
open to colonization by both native and exotic species from seed banks and the surrounding
landscape, in addition to the native species included in seed mixes. Importantly, the richness and
seeding rate of seed mixes explains the composition and richness of sown species (a subset of
native species) across these sites (Grman et al. 2013, 2015). Therefore variation in seed mixes
among these sites is likely to impact native-exotic richness relationships, especially if the
richness of native and exotic colonizers tends to be correlated in the absence of seed addition

38	

	
(Meiners et al. 2002). We also examined these relationships at a regional spatial extent (130,000
hectares), similar to many observational studies of natural habitats, but greater than other
restoration studies (~50-500 hectares; Foster et al 2002, Suding and Gross 2006, Hill and Fischer
2014). Examining native-exotic richness relationships across a large spatial extent allowed us to
more accurately assess the utility of this framework for guiding restoration by spanning more
natural variation in edaphic and landscape factors. First, we asked whether native and exotic
species richness were correlated across this landscape. Then, we tested whether management,
including prescribed fire and seed mix design (richness, density, composition), environmental
factors (soil characteristics, landscape context, productivity, light availability, restoration size
and age), or interactions among these factors shifted this relationship to favor native species.

Methods
Study system
Our study occurred across 29 grassland restorations in southwestern Michigan (Grman et
al. 2014). Between 2003 and 2008, former agricultural land was treated with herbicide and sown
with seed of native tallgrass prairie species. Tallgrass prairies are commonly restored using these
methods in the Midwestern USA, and are generally managed with periodic prescribed fires to
limit growth of woody species, reduce cover and diversity of invasive exotic species, and
increase cover and diversity of native prairie species (Packard and Mutel 1997). Sites ranged in
size from 0.22 to 39.0 ha (median = 8 ha).

39	

	
Plant community sampling and species origin
We sampled plant community composition in July-September 2011. At the center of each
site in 2011, we established a 45-meter transect and visually estimated the percent cover of all
vascular plant species in each of 10 equally spaced 1-m2 plots. See Grman et al. (2014) for
detailed sampling methods.
We calculated native and exotic species richness for each site as the cumulative richness
per transect. To do this, we classified each species as native if it was present in Michigan prior to
European settlement, and otherwise as exotic. Four sown species are considered exotic in
Michigan, but are native in adjacent states and are members of the target community, so were
classified as native here (see Table S3.1). Most species in our dataset classified as exotic in
Michigan are also considered exotic throughout North America (Table S3.1), and reclassifying
based on continental origin gave equivalent results.

Management and environmental factors
We collected data on management and environmental factors that are likely to influence
native-exotic relationships (see Table 3.1 for a summary of these factors and predictions).
Management manipulated immigration rates of native species through seed additions, and natural
disturbance regimes through prescribed fire. We examined environmental factors related to the
immigration rates of native and exotic species, as well as the availability and heterogeneity of
resources. Additionally, species declines are common early on in establishing prairie
restorations, and either native or exotic species may be lost more rapidly (Sluis et al. 2002,
Carter and Blair 2012, Grman et al. 2013). We used restoration age to model potential changes in
richness over time.

40	

	
Table 3.1. Summary of variables used in models to predict native richness relative to exotic
richness. Predicted effects, and mechanisms contributing to predicted effects (see Results:
Management and environmental factors in text for references).
Measured
variable
Management
Seed mix
richness
Seeding rate
(g/m2)
Fire frequency
(fires/year)

Mean
(range)

Description

Mechanism

Number of native species
dispersing via seed sowing
1.0 (0.68- Density of native species
1.30)
dispersing via seed sowing
0.24 (0 Number of burns per year
0.67)

Prediction

Increases native richness, over native
positive
species that naturally colonize
Increase establishment probability of
positive
sown native species
Native prairie species are more likely
positive
than exotic species to be adapted to fire

35 (8 - 71)

Environmental Factors
Non-sown species decline over time;
5.28 (3-8) Number of years since sowing exotics may decline because all exotic
species are non-sown
Environmental factors related to immigration rates
Restoration age
(years)

positive

Principal components axis;
higher values indicate greater
forest and grassland cover,
lower agricultural

Perennial habitats (forest/grassland)
occur in disturbed landscapes so they
Landscape PCA NA
negative
should contribute more exotic than
native species
Seedbanks from former perennial
Legacy effects from prior land histories (oldfields, hay fields, pastures)
Land-use history NA
negative
use (e.g. seedbanks)
should contribute more exotic than
native species
Edge:area ratio 0.04 (0.01- Ratio of site perimeter:site
Edge effects increase immigration rates
negative
(meters/hectares) 0.11)
area
of exotic species
Environmental factors related to resource availability
Principal components axis;
higher values indicate lower Exotic species more successful in
Soils PCA
NA
water holding capacity, higher resource-rich sites
phosphorus and sand
Mean biomass
541 (229 Exotic species more successful in
Productivity
(g/m2)
872)
resource-rich sites
Total %
99 (69 - Percent cover of all dead and Exotic species more successful in
vegetative cover 116)
standing biomass
resource-rich sites

positive

negative
negative

Environmental factors related to resource heterogeneity
CV of water
0.10 (0.02 Heterogeneity of soil moisture Exotic species more successful in highnegative
holding capacity - 0.25)
retention (along 45 m transect) resource patches in heterogenous sites
CV of biomass

Heterogenity of site
0.28 (0.13
productivity (along 45 m
- 0.60)
transect)

Exotic species more successful in highnegative
resource patches in heterogenous sites

41	

	
Management - seed addition and prescribed fire: Native species richness should increase in
restorations, relative to exotic species, with the richness and seeding rate of prairie species in
seed mixes, and may depend on which species are sown. We collected data on seed mix
composition from the seed mix provider (Native Connections, Three Rivers, MI). Sites were
sown with between 8 and 71 native prairie species, the majority of which are unlikely to disperse
into sites due to regional rarity, and at densities of 0.68-1.30 g seeds/m2. See Grman et al. (2014)
for additional info on composition and seeding rates. Fire is a key natural disturbance in tallgrass
prairie, providing pulses of resources such as soil nutrients and light availability, which alter the
availability of microsites for colonization (Briggs and Knapp 1995). We consulted landowners
for prescribed fire records. To model the effects of prescribed fire, we calculated fire frequency
as the number of fires divided by restoration age.

Environmental factors related to immigration rates: Both surrounding landscape composition
and land-use history can affect immigration rates of non-sown native and exotic species (Hobbs
and Huenneke 1992, Meiners et al. 2004), and therefore may alter the ratio of native to exotic
species richness (Lonsdale 1999, Simberloff 2009). To assess how different landscapes
influenced immigration rates, we identified the area of forest, grassland, wetland, row crops, and
developed land within 500 meters of the center of each site in ArcGIS (Grman et al. 2013). We
simplified data on landscape composition through principal components analysis (PCA). On the
first axis (landPC1), which explained 59% of the variation in landscape composition, higher
values represented greater forest and grassland cover, and lower cover of agricultural uses. We
predicted that because restorations occur in disturbed post-agricultural landscapes, the species
pool in these landscapes would be dominated by exotic species (Parker et al. 2010, Jauni and

42	

	
HyvĂśnen 2012). As a result, landscapes with greater cover of established habitats (forest and
grassland) would support larger exotic populations and greater propagule pressure, and
restorations in these landscapes would therefore have proportionally higher exotic (and lower
native species) richness. Propagules of non-sown species also originate in the seed- or budbank
of restoration sites, which may vary depending on land-use history. We consulted landowners to
determine the immediate pre-restoration land use of each site, which we classified as row-crops
(generally tillage agriculture, some no-till) or perennial grassland (hay, pasture, and old fields).
Similarly to our landscape prediction, we predicted that sites restored from perennial grasslands
(hay, pasture, old fields) would have a lower proportion of native species richness. Finally, we
also calculated the perimeter:area ratio of each site in ArcGIS. We predicted that edge effects in
sites with higher perimeter:area ratios would enhance invasion from surrounding habitats, and
lead to a lower proportion of native species richness (With 2002 Conservation Biology, Vila and
Ibanez 2011).

Environmental factors related to resource availability: Native and exotic species may respond
differently along resource gradients, or to spatial variation in resources (Davies et al. 2005,
Carboni et al. 2010, Brummer et al. 2016). Grassland plant communities are most strongly
influenced by soil resources, productivity, and light availability (Briggs and Knapp 1995, Borer
et al. 2014). Exotic species may be more successful than native species in high resource
environments (Seabloom et al. 2015), so we predicted that native species richness would be
higher relative to exotic species richness in low resource environments. To assess soil resources,
we sampled eight 20 cm × 3 cm2 soil cores at each plot (10 plots per site). We pooled samples
for each plot and measured water-holding capacity in the lab as the proportional difference of

43	

	
saturated to oven-dried weight (after Brudvig and Damschen 2011). We calculated three
measures of site-level resource availability – soil resources, productivity, and light availability.
To assess site-level resource availability, we pooled samples for each site and analyzed soil
organic matter, soil texture (sand, silt, and clay content), Mehlich-III phosphorus, Bray-II
phosphorus, and pH (Brookside Laboratories, New Knoxville, OH, USA). We simplified sitelevel soils data by conducting a principal components analysis (PCA). On the first axis (soil
PC1), which explained 56% of the variation in soils data, higher values indicated sites with
sandier soils, low water holding capacity, and higher phosphorus. To assess productivity, we
clipped, dried at 155° C, and weighed all above-ground biomass in each 1-m2 plot. We used
mean biomass/m2 as a measure of site-level productivity. We used the sum of total percent
vegetative cover and litter cover, which were assessed visually in each plot and together ranged
from 69-116%, as an inverse measure of light availability. Finally, heterogeneous environments
have high-resource patches that may favor exotic species (Shea and Chesson 2002). As in low
resource environments, we predicted that native species richness would be higher relative to
exotic species richness in more homogeneous environments. We used the coefficient of variation
in water holding capacity and biomass among plots to model heterogeneity soil resources and
productivity, respectively.

Data analysis
To test whether native and exotic species richness were correlated across sites, we
calculated Pearson’s r. While not central to the goals of this study, null models are commonly
applied in similar studies; therefore, we compared observed native-exotic correlations to a null
distribution based on random assembly from a species pool with a fixed ratio of native to exotic

44	

	
species (Fridley et al. 2004, Herben et al. 2004), and include these results in the Appendix
(Figures S3.5, S3.6).
We examined whether management and environmental factors shifted the native-exotic
species richness relationship to favor native species. First, we regressed native richness on exotic
richness and calculated least-squared residuals. Then, we conducted model selection with AICc
(Burnham and Anderson 2002) on a set of models with these residuals as the response variable,
and in which management, environmental factors, and their interactions predicted these residuals
(full model set in Table S3.2). This model set contained three types of additive models: 1)
models that included at least one management variable as a predictor (Table 3.1; seed mix
richness, seeding rate, fire frequency); 2) models that included a single environmental variable
(Table 3.1; restoration age, perimeter:area ratio, landscape PCA, land-use history, soils PCA,
aboveground biomass, % vegetative cover, CV of water holding capacity, and CV of biomass) as
a predictor; and 3) models that included at least one management variable plus at least one
environmental variable. We also included two types of interaction models to test whether 1) the
effect of management varies across environments (e.g., fire x productivity, 2) or the effect of
individual management variables depends on other management variables (e.g., seed mix
richness x seeding rate). Predictors in these models were not strongly correlated (r < 0.55, mostly
< 0.30; Table S3.3), and variance inflation factors were all < 4 .We also included an interceptonly model as a baseline of no effect. We examined the direction and strength of standardized
parameter estimates in all models with ‘considerable support’ (dAIC ≤ 4.0), which we refer to as
‘competitive models’ and use for inference, but also report models with ‘considerably less
support’ (4.0 < dAIC ≤ 7.0) (Burnham and Anderson 2002) in Table 3.2.

45	

	
Because we found strong support for a model in which the effect of seed mix richness on
native richness depended on seeding rates (see Table 3.2), we conducted an additional model
selection analysis to test whether specific aspects of seed mix composition related to nativeexotic richness relationships. Specifically, we focused on variation in the abundance of plant
functional groups, or species grouped by shared functional traits, because they are known to have
different establishment probabilities (Tilman 1997, Symstad 2000). Additionally, managers can
adjust the seeding rates and richness of functional groups that are most likely to increase native
richness. We sorted seed mix composition into four functional groups: C3 graminoids (including
Cyperaceae and Juncaceae), C4 graminoids (all Poaceae), nitrogen-fixers (forbs and two subshrubs, mostly Fabaceae but including Rhamnaceae; hereafter “legumes”), and non-nitrogenfixing forbs (hereafter “forbs”) (Table S3.4). Two sites lacked detailed seed mix composition
data, so we conducted this analysis on a restricted dataset of 27 sites. We constructed a model
set, as above, where each model predicted the residuals of regressing native on exotic species
richness. First, we built four individual models, testing whether the effect of the species richness
of each functional group on these residuals depended on the total seeding rate of species within
that same functional group (e.g., richness of forbs x rate of forbs). We also modeled the
interaction of total seed mix richness with the seeding rate of each functional group (e.g., total
seed mix richness x forb seeding rate), and the total seeding rate with the richness of each
functional group (e.g., forb seed mix richness x total seeding rate). Finally, because dominance
of C4 grasses may reduce diversity by limiting the establishment of other sown species (Howe
1994), we also modeled the interaction between C4 grass seeding rate and the richness of each
functional group (Dickson and Busby 2009). Then, we compared this more detailed set of seed
mix composition models, as well as the original seed mix model, with AICc, as described above.

46	

	
Finally, because native richness was lower in older sites (see Table 3.2), we controlled for
restoration age in all models. In summary, each seed composition model had four terms: richness
of either the total seed mix or a signal functional group in the seed mix, the seeding rate of the
total seed mix or a single functional group in the seed mix, the interaction between rate and
richness, and restoration age.

Results
Native species richness was positively correlated with exotic species richness across our
sites (Pearson’s r = 0.54, p < 0.01) (Figure 3.1). However, management variables influenced the
residuals of the native-exotic regression. Seed mix richness increased native richness most
strongly at high seeding rates; at low seeding rates, seed mix richness had no effect on native
richness (Model 1 in Table 3.2, seed mix richness x seeding rate, p < 0.01; Figure 3.2). While

Figure 3.1. Native and exotic species richness was positively correlated across 29 prairie

30
20
10
0

Native Richness

40

restorations. Pearson’s r = 0.54, p <0.01.

0

5

10

15

Exotic Richness

47	

	
native richness was strongly influenced by seed mixes, we found that more frequent fire
increased native species richness only weakly (Table S3.2, all models including a term for fire,
ΔAICc ≥ 8.5; Figure 3.3). In addition, older restorations had proportionally lower native richness
than younger restorations in a competitive model (ΔAICc < 4.0), which included the main effects
of seed mix richness and seeding rates (Model 2 in Table 3.2, ΔAICc = 2.6, adj-R2 = 0.43; Figure
3.4). This pattern was apparently driven by loss of native, and not exotic species over time
(Figure S3.1a). There was little evidence that abiotic environmental conditions modified the
native-exotic relationship (Tables 3.2, S3.1), suggesting that modifying native-exotic richness
relationships are largely under the control of managers.

Table 3.2. Top management and environmental models predicting native species richness
after controlling for association with exotic species richness (residuals of linear regression
model: native species richness ~ exotic species richness) selected by AICc. Only models with
dAICc < 7.0 are shown. Standardized estimates are shown (***p<0.001, **p<0.01, *p<0.05).
Terms not in model indicated with a dash (-). Mix SR = seed mix richness; Mix Rate = seeding
rate (g/m2); see Table 2.1 for Covariate definitions.

Model

Covariate

Mix SR

Mix
Rate

Mix SR
x Rate

Covariate

logLik

dAICc

df

weight

R2

1

-

5.91***

-3.36*

3.65**

-

-88.7

0

5

0.59

0.48

2

4.14**

-3.80**

-

-2.72*

-90

2.6

5

0.16

0.43

3

Age
CV
Biomass

4.30**

-5.17**

-

-2.38

-91.5

5.6

5

0.04

0.37

4

Biomass

4.60**

-4.57**

-

1.99

-91.5

5.7

5

0.03

0.37

5

-

4.83**

-4.13**

-

-

-93

5.8

4

0.03

0.33

6

Land PCA

5.25***

-3.99**

-

-1.85

-91.7

6.1

5

0.03

0.36

7

Edge:Area

5.05***

-4.27**

-

1.78

-91.8

6.1

5

0.03

0.36

48	

	
Figure 3.2. Native species richness was higher than predicted from the native-exotic
richness relationship, in sites sown with many species at high rates. Heat map showing
response of residuals from native SR~exotic SR regression to interactive effect: seed mix
richness x seeding rate, p < 0.01 (see Model 1 in Table 2). Darker shading = higher proportional
native richness (i.e., positive residuals); lighter shading = lower proportional native richness (i.e.,
negative residuals). Points in gray show coverage of data over predicted relationship.

The effects of seed mix richness and seeding rates on the relative number of native
species compared to exotics were driven primarily by forb species (Table 3.3, Figures 3.5a-c).
Among seed mix composition models, modeling the interaction between forb richness and total
seeding rate predicted native species richness as well modeling the interaction between total
richness and total seeding rate (Table 3.3; Forb SR-Mix Rate: ΔAICc = 0.4, adj-R2 = 0.52; Mix
SR-Mix Rate: ΔAICc = 0.0, adj-R2 = 0.53). As with total seeding richness and rate (Figure 3.2),
forb seeding richness also increased native richness only at high total seeding rates (Figure 3.5c),

49	

	
although interactions between richness and rates in this restricted dataset were only marginally
significant (Table 3.3; total seed mix richness x total seeding rate, p = 0.052; forb seed mix
richness x total seeding rate, p = 0.054). Forb seeding rates did not predict native richness as well
as forb richness, based on AICc. The next two highest-ranked models included interactions of
forb seeding rates with total seed mix richness (Figure 3.5b, Table 3.3; Mix SR-Forb Rate:
ΔAICc = 3.6, adj-R2 = 0.46) and forb richness (Figure 3.5a, Table 3.3; Forb SR-Forb Rate:
ΔAICc = 4.6, adj-R2 = 0.44). Among other functional groups, only C3 grass models predicted
native richness moderately well (Table 3.3; C3 SR-Mix Rate: ΔAICc = 6.0, adj-R2 = 0.41, Mix
SR-C3 Rate: ΔAICc = 7.2, adj-R2 = 0.38; Figures 3.5e,f). Contrary to expectations, the effect of
functional group richness did not depend on C4 grass seeding rate (Table 3.3; all models
including a term for C4 grass seeding rate, ΔAICc >10).

Figure 3.3. Native-exotic richness relationship is not modified by fire frequency. Native
richness increases with fire frequency, but effect is only marginally significant (a); fire
frequency, p = 0.09. The effect of fire frequency is not significant when controlling for seed mix
richness and seeding rates (b); fire frequency, p = 0.70.
b)

10

Residual of NativeSR~ExoticSR

Residual of NativeSR~ExoticSR

a)
5
0
-5
-10

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.0

0.1

Fire Frequency

0.2

0.3

0.4

Fire Frequency

50	

0.5

0.6

	
Figure 3.4. In older sites, native species richness is lower than predicted from native-exotic
richness relationship. Partial residual plot; restoration age, p = 0.02; Model 2 in Table 3.2.

Residual of NativeSR~ExoticSR

10

5

0

-5

-10
3

4

5

6

7

8

Restoration Age

Discussion
One of the greatest restoration challenges is designing strategies to increase native
species diversity while also discouraging exotic species (Matthews and Spyreas 2010, Brudvig
2011, Suding 2011). It is a challenge in part because native and exotic species may respond to
the same environmental drivers in both established (Stohlgren et al. 1999) and successional
(Meiners et al. 2004) natural habitats, and may respond similarly during restoration (Suding and
Gross 2006). We found that, while native and exotic species richness was positively correlated
across 29 prairie restorations, sites sown with more species at higher rates had higher native
species richness than predicted from this underlying correlation. In contrast, sites sown with high
densities of few species had lower proportions of native species than predicted by native-exotic
correlations. This pattern was primarily driven by the forb component of seed mixes, and to a
lesser extent by the C3 graminoid component.

51	

	
Figure 3.5. Interaction between seed mix richness and seeding rate is driven largely by
forbs in seed mix, and to a lesser extent by C3 grasses. Heat map showing response of
residuals from native SR~exotic SR regression to interactive effect of of the seed mix richness
and seeding rate of important functional groups, and between these functional groups and the
total richness and rate of seed mixes (see models in Table 3.3). Darker shading = higher
proportional native richness (i.e., positive residuals from native SR ~ exotic SR regression);
lighter shading = lower proportional native richness (i.e., negative residuals). Interactions
between richness and seeding rate of forbs (a), and C3 grasses (d); interactions between total seed
mix richness and seeding rate of forbs (b) and C3 grasses (e); interactions between total seeding
rate and seed mix richness of forbs (c), C3 grasses (f). See Figure S2 for interactions with
legumes and C4 grasses.

52	

	
Table 3.3. Seed composition models predicting native species richness after controlling for
association with exotic species richness (residuals of linear regression model: native speices
richness ~ exotic species richness) selected by AICc. All models shown, but models with
dAIC>10.0 have little or no support, while those <4.0 all have reasonable support. Standardized
estimates are shown (***p<0.001, **p<0.01, *p<0.05, +p<0.10). Terms not in model indicated
with a dash (-). Mix SR = seed mix richness; Mix Rate = seeding rate (g/m2); Func=functional
group. See Figures 3.5 and S3.2 for form of interactions.
Model
Mix SR x Mix
Rate
Forb SR x
Mix Rate
Mix SR x
Forb Rate
Forb SR x
Rate
C3 SR x Mix
Rate
Mix SR x C3
Rate
Legume SR x
Mix Rate
Mix SR x
Legume Rate
C4 SR x Mix
Rate
Legume SR x
Rate
Forb SR x C4
Rate
Mix SR x C4
Rate

Mix
SR

Mix
Rate

5.00*** -3.77**
-

Func
SR
-

-3.84** 5.12***
-

Func Func SR x Mix SR x Func SR x Mix SR x
delta
Rate Func Rate Func Rate Mix Rate Mix Rate Age AICc df weight R2
-

-

-

-

2.53+

-2.48* 0.0

6

0.45

0.53

-

-

-

2.68+

-

-2.64* 0.4

6

0.36

0.52

1.53

-

-

-4.17* 3.6

6

0.07

0.46

1.61

-

-

-

-4.24* 4.6

6

0.05

0.44

-1.67

6.0

6

0.02

0.41

4.83**

-

-4.84**

-

-

-

-2.37+

3.30*

-

-

-

2.27

-

-0.48

-

-

3.29

-

4.27*

-

-

-3.42** 7.2

6

0.01

0.38

-

-4.05*

3.82*

-

-

-

1.48

-

-3.58* 7.5

6

0.01

0.37

4.58*

-

-

-2.70+

-

2.88

-

-

-1.52

9.1

6

0.005 0.33

-

-3.25*

3.24*

-

-

-

1.36

-

-4.67** 9.2

6

0.005 0.33

-

-

5.00* -5.01**

2.23

-

-

-

-3.26* 9.7

6

0.004 0.32

-

-

1.72

-3.29*

-2.67

-

-

-

0.01

10.4 6

0.003 0.30

1.65

-

-

-3.28+

-

-2.41

-

-

-0.08 11.1 6

0.002 0.28

INTER-CEPT
C3 SR x C4
Rate

-

-

-

-

-

-

-

-

-

-

1.19

-2.89

-1.56

-

-

-

-0.20 13.0 6 <0.001 0.23

C3 SR x Rate
Legume SR x
C4 Rate

-

-

2.20

1.32

1.79

-

-

-

-1.48 13.6 6 <0.001 0.21

-

-

0.59

-3.65+

-1.52

-

-

-

-1.27 14.4 6 <0.001 0.19

C4 SR x Rate

-

-

1.39

-2.50

-0.57

-

-

-

-2.17 14.9 6 <0.001 0.18

4.74** -4.72**

53	

12.9 2 <0.001

0

	
Management favors native richness
Consistent with large-scale observational studies in natural habitats, native and exotic
species richness were positively correlated across our sites. Yet, our results suggest that
managers can modify the native-exotic richness relationship to favor native species by using seed
mixes that increase the immigration rates of native species. Proportional increases in native
species required not only the immigration of many native species (e.g, > 50 species sown), but
also sowing at high densities (e.g, > 1 g seed/m2). These high seeding rates in species-rich mixes
may be necessary to overcome demographic barriers to population establishment, especially for
forbs. While the overall seeding rate of forbs increased with forb richness (Table S3.5), the
seeding rates of individual species are often low due to the high cost of many forb species. At
low overall seeding rates, the number of colonizing individuals of many forb species is probably
too low to establish viable populations, especially if further limited by abiotic conditions (Grman
et al. 2015). The importance of forb species for driving native richness may also be in part
because forbs account for the majority of the variation in seed mix richness and seeding rates
(Grman et al. 2015; Table S3.4). The influence of C3 graminoids and legumes on native richness
was similar to the influence of forbs, but variation (or maximum amounts) in the richness and
seeding rate of C3 graminoids and legumes may not be great enough to allow for detectable
effects on overall native richness (forbs: 4-51 species, 0.003-0.473 g/m2; C3 graminoids: 0-9
species, 0-0.252 g/m2; legumes: 1-8 species, 0.001-0.126 g/m2; Table S3.4).
In contrast to sowing greater densities of many species, sowing greater densities of fewer
species was associated with proportionally lower native richness. Highly competitive C4 grasses
dominate low-richness seed mixes (Table S3.4). C4 grasses can reduce the richness of other
native and exotic species when sown at high rates (Dickson and Busby 2009). We tested for

54	

	
overall effects of C4 grass richness and rate in seed mixes, and whether the effect of functional
group richness depended on C4 grass seeding rate. Because C4 grass richness is consistently low
(3-6 species) and seeding rates consistently high (0.392-0.785 g/m2), there was likely insufficient
variation at biologically meaningful ranges (e.g., no low C4 seeding rates or high C4 richness) to
explain differences in native richness. From a management perspective, this suggests that
investing in high seeding rates of many forb species benefits native diversity. When the level of
investment required for both higher seeding rates and more species is not feasible, however,
appreciable increases in native richness may be limited. In fact, when sowing mixes with low to
intermediate richness (< ~ 40 species), our results suggest that sowing at low rates (< ~ 0.8
grams seed/m2) may lead at most to modest proportional increases in native richness, but
increasing seeding rates beyond 1.0 grams seed/m2 is likely to lead to decreased native richness.
Surprisingly, we detected little evidence that fire, one of the most common management
tools, influences the relative proportion of natives. This result contrasts with similar studies,
where both seed additions and fire have increased native, relative to exotic richness (Suding and
Gross 2006, Hill and Fischer 2014). These studies, however, occurred in later successional
grasslands. For example, Suding and Gross (2006) imposed burning and seed addition treatments
across a degraded prairie landscape in southwestern Michigan in which native and exotic
richness was positively correlated. Burning alone broke the correlation, seed addition alone
increased native richness only at intermediate exotic richness, while the combination of seed
addition and burning increased native richness but maintained the positive correlation (Suding
and Gross 2006). Colonization rates of exotic species may be higher in earlier-successional
systems such as ours, overwhelming any negative impact of fire on exotic species.

55	

	
The impact of seed sowing may fade over time
While our results showed how seed sowing can modify the native-exotic richness
relationship to favor native species, we also found that native-exotic richness relationships
shifted to favor exotic species over time. This shift was due to declining native richness, not
increasing exotic richness, with restoration age. Species richness often declines over time during
prairie restoration (Sluis 2002, Middleton et al. 2010, Grman et al. 2013), although other studies
have pointed to reductions in exotic and not native richness as the cause of this pattern (Heslinga
and Grese 2010, Carter and Blair 2012). In our system, greater loss of native species may be due
to variation in the rates, richness, and composition of seed mixes with restoration age, or because
successional processes influenced native vs. exotic species differently. We find some support for
this first idea, as seed mixes in older restorations contain fewer forbs and C3 grasses, and more
C4 grasses (Table S3.5) and these differences may influence the development and maintenance of
native richness over time. However, independent of these differences, restoration age predicted
lower native richness in models (Table 3.3). This suggests that restoration age and seed mix
properties have independent effects on native richness.
As further evidence that loss of native richness is not simply a function of seed mix
properties, the richness of non-sown native species appears to be more strongly correlated with
restoration age than the richness of sown native species (Grman et al. 2013, Figure S3.1b). In the
landscapes where these restorations occur, native species may colonize at low rates because
propagule sources are rare. As a result, non-sown native species may be prone to localized
extinctions over time. In contrast, exotic species occur at higher densities in the landscape and
local populations may be augmented or periodically rescued through repeat dispersal events
(Gotelli 1991). Alternately, dominant community members, such as the C4 prairie grass

56	

	
Andropogon gerardii, may exclude native species more strongly than exotics (Howe 1994,
Heslinga and Grese 2010, Carter and Blair 2012), but we found no evidence for this effect
(Figure S3.3). We therefore suggest that native species declines are a function of low recurring
propagule inputs.

Conclusions and management implications
Additional work in our system has compared a wide range of management and
environmental factors to explain variation in both sown native (Grman et al. 2013, 2014, 2015)
and invasive exotic species (see Chapter 4). The restoration of diverse sown (native)
communities is limited somewhat by soil moisture (Grman et al. 2014), but largely influenced by
sowing more species, and greater densities of forbs (Grman et al. 2013, 2015). In turn, exotic
invasive species are limited by diversity and dominance of sown species, but also strongly
influenced by land-use history legacies (see Chapter 4). Together, this suggests that managers
can influence the dynamics between native and exotic species through seed sowing, but also that
the effectiveness of the sown community for resisting invasion may be limited under certain
conditions (e.g., sites with low soil moisture, or certain land-use histories). Here we contribute to
our understanding of how restored grassland communities assemble by explicitly considering the
relationship between native and exotic species richness. Across a wide range of hypothesized
drivers, including many that are difficult for managers to manipulate (e.g., soil properties), our
results suggest that modifying native-exotic richness relationships to favor native species is
largely determined by seed mix design, which managers control. However, practical constraints
(i.e., financial costs) may limit this control. Obtaining a large amount of seed of many forb
species, which our results suggest is necessary to increase native richness, generally requires a

57	

	
significant investment of time or money. This investment is clearly warranted when increasing
native richness is the goal. When managers lack financial or labor resources to sufficiently invest
in the forb component of seed mixes, however, it may be difficult to increase native richness,
relative to exotic richness, and restoration goals may need to be adjusted to simply aim for native
dominance, instead of a diverse native community. Restorations with low to intermediate native
plant diversity still provide many important ecosystem services and may provide habitat for
important wildlife species (e.g., grassland birds) (Werling et al. 2014).
Positive native-exotic richness relationships are nearly ubiquitous in both managed and
unmanaged plant communities, but this does not necessarily represent a limitation for restoration.
First, native species richness was generally high at our sites, between 1.2 and 12.9 times higher
than exotic species richness (median = 2.8 times higher), and the relative abundance of native
species was also high (median = 84%). Therefore, managers generally restored communities
largely composed of native species. Second, the correlation between native and exotic richness
does not necessarily predict negative impacts of exotic species on community and ecosystem
processes (Seabloom et al. 2013). The ability for managers to increase native relative to exotic
richness is especially impactful if those increases translate into reductions in the dominance of
exotic species. In these 29 sites, exotic dominance was lowest in sites where native richness was
higher than predicted by the native-exotic richness relationship (Figure S3.4). Therefore, while
some exotic species will inevitably be a component of both managed and unmanaged plant
communities, the presence of exotic species, by itself, need not be a major concern for managers
(D’Antonio and Myerson 2002, Hobbs et al. 2009). In fact, our results here and elsewhere
(Grman et al. 2013) suggest that managers have a great deal of influence in restoring plant

58	

	
communities with largely desirable qualities (e.g., are dominated by a diversity of native
species).

Acknowledgements
Insights from Jeff Conner improved the interpretation of results. Thank you to Beth Miller for
help coding the null model distribution. Portions of this work was funded by USDA-NIFA grant
2013-67011-21281.

59	

	

APPENDIX

60	

	
Table S3.1. Origin and abundance of species in dataset. *In seed mix but not native to
Michigan. Y*, in seed mix but frequently naturally colonizes sites.
Species
Abutilon theophrasti
Agrostis gigantea
Alliaria petiolata
Arctium minus
Barbarea vulgaris
Bromus inermis
Centaurea stoebe
Cerastium fontanum
Chenopodium album
Chenopodium sp.
Cirsium arvense
Crepis capillaris
Crepis sp.
Dactylis glomerata
Daucus carota
Dianthus armeria
Digitaria sp.
Elymus repens
Frangula alnus
Hieracium sp.
Hypericum perforatum
Hypochaeris radicata
Leucanthemum vulgare
Lonicera sp.
Medicago lupulina
Medicago sativa
Melilotus sp.
Mollugo verticillata
Morus alba
Phleum pratense
Plantago lanceolata
Plantago major
Plantago sp.
Poa pratensis
Polygonum aviculare
Polygonum persicaria
Potentilla argentea
Potentilla recta

Origin
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic

Mean Cover
0.02%
0.08%
0.05%
0.03%
0.09%
2.38%
0.58%
0.04%
0.00%
0.00%
1.00%
0.01%
0.00%
0.77%
1.44%
0.02%
0.26%
3.31%
0.01%
2.98%
0.22%
1.42%
0.01%
0.00%
0.01%
0.01%
0.03%
0.03%
0.14%
0.02%
0.97%
0.01%
0.02%
24.10%
0.00%
0.01%
0.01%
0.20%
61	

In seed mix?
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N

	
Table S3.1 (cont’d)
Species
Potentilla sp.
Rosa multiflora
Rumex acetosella
Rumex crispus
Saponaria officinalis
Setaria sp.
Silene latifolia
Solanum carolinense
Stellaria media
Taraxacum officinale
Trifolium arvense
Trifolium pratense
Trifolium repens
Trifolium sp.
Verbascum thapsus
Veronica arvensis
Veronica serpyllifolia
Veronica sp.
Vicia villosa
Acalypha rhomboidea
Acer sp.
Achillea millefolium
Agastache foeniculum*
Agastache nepetoides
Ambrosia artemisiifolia
Andropogon gerardii
Anemone cylindrica
Anenome sp.
Antennaria parlinii
Apocynum cannabinum
Aquilegia canadensis
Arnoglossum atriplicifolium
Artemesia ludoviciana*
Asclepias syriaca
Asclepias tuberosa
Asplenium platyneuron
Botrychium dissectum
Bouteloua curtipendula
Brickellia eupatorioides
Carex bicknellii

Origin
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
exotic
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native

Mean Cover
0.03%
0.00%
1.20%
0.31%
0.02%
0.05%
0.28%
0.95%
0.05%
0.34%
0.09%
3.89%
0.05%
0.64%
0.08%
0.00%
0.03%
0.00%
0.01%
0.00%
0.01%
1.36%
0.01%
0.03%
0.24%
45.31%
0.02%
0.13%
0.00%
0.05%
0.01%
0.66%
0.06%
0.37%
0.14%
0.00%
0.00%
1.73%
0.07%
1.08%
62	

In seed mix?
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
Y*
Y
Y
N
Y
Y
N
N
N
Y
Y
N
Y*
Y
N
N
Y
Y
Y

	
Table S3.1 (cont’d)
Species
Carex brevior
Carex sparganioides
Carex swanii
Celtis occidentalis
Chamaecrista fasciculata
Conyza canadensis
Coreopsis lanceolata
Coreopsis palmata
Coreopsis tripteris
Crataegus sp.
Cyperus erythrorhizos
Cyperus lupulinus
Dalea purpurea
Desmodium canadense
Desmodium ciliare
Desmodium obtusum
Desmodium paniculatum
Desmoidum hybrid2
Dichanthelium clandestinum
Dichanthelium lindheimeri
Dichanthelium meridionale
Dichanthelium oligosanthes
Digitaria cognatum
Echinacea pallida
Echinacea purpurea
Elymus canadensis
Elymus virginicus
Epilobium ciliatum
Equisetum sp.
Erigeron sp.
Euphorbia corollata
Euthamia graminifolia
Fragaria virginiana
Fraxinus sp.
Galium aparine
Galium pilosum
Geum canadense
Geum sp.
Gnaphalium obtusifolium
Habenaria lacera

Origin
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native

Mean Cover
0.27%
0.02%
0.11%
0.01%
0.01%
0.03%
0.14%
0.10%
0.88%
0.01%
0.00%
0.01%
0.03%
1.00%
0.10%
0.02%
0.03%
0.08%
0.82%
0.00%
0.01%
0.01%
0.10%
0.06%
3.30%
2.39%
0.07%
0.01%
0.02%
0.46%
0.02%
1.61%
0.00%
0.00%
0.00%
0.01%
0.05%
0.05%
0.10%
0.00%
63	

In seed mix?
Y
Y
N
N
Y
N
Y
Y
Y
N
N
N
Y
Y
N
N
N
N
N
N
N
N
N
Y
Y
Y
Y
N
N
N
N
Y*
N
N
N
N
N
N
N
N

	
Table S3.1 (cont’d)
Species
Helenium autumnale
Helianthus grosseserratus
Helianthus maximilianii
Helianthus mollis
Helianthus occidentalis
Helianthus strumosus
Heliopsis helianthoides
Heuchera sp.
Hypericum ascyron
Hypericum punctatum
Hypericum sp.
Juglans sp.
Juncus tenuis
Koeleria macrantha
Lactuca canadensis
Lactuca sp.
Lespedeza capitata
Lespedeza hirta
Liriodendron tulipifera
Lobelia inflata
Lupinus perennis
Lycopus sp.
Monarda fistulosa
Monarda punctata
Muhlenbergia frondosa
Muhlenbergia schreberi
Oenothera biennis
Oenothera sp.
Onoclea sensibilis
Oxalis sp.
Panicum dichotomiflorum
Panicum virgatum
Parthenium integrifolium*
Parthenocissus quinquefolia
Penstemon digitalis
Phalaris arundinacea
Phytolacca americana
Plantago rugelii
Polygala polygama
Polygonum pensylvanicum

Origin
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native

Mean Cover
0.00%
0.02%
0.01%
0.15%
0.06%
0.08%
1.65%
0.01%
0.00%
0.00%
0.02%
0.02%
2.90%
0.01%
0.25%
0.00%
0.06%
0.03%
0.03%
0.02%
0.26%
0.07%
6.25%
0.02%
0.06%
0.00%
0.19%
0.01%
0.12%
0.13%
0.00%
3.56%
0.18%
0.08%
0.56%
0.01%
0.02%
0.01%
0.00%
0.00%
64	

In seed mix?
Y
Y
Y
Y
Y
Y
Y
Y
Y
N
N
N
N
Y
N
N
Y*
N
N
N
Y
Y
Y*
Y
N
N
Y*
N
N
N
N
Y
Y
N
Y
N
N
N
N
N

	
Table S3.1 (cont’d)
Species
Populus deltoides
Populus grandidentata
Populus tremuloides
Potentilla simplex
Prunus serotina
Pycnanthemum virginianum
Ranunculus abortivus
Ratibida pinnata
Rhynchospora capitellata
Ribes sp.
Rubus allegheniensis
Rubus flagellaris
Rubus occidentalis
Rudbeckia hirta
Rudbeckia subtomentosa
Rudbeckia triloba
Salix sp.
Sassafras albidum
Schizachyrium scoparium
Senna hebecarpa
Silphium integrifolium
Silphium perfoliatum
Solanum ptychanthum
Solidago canadensis
Solidago rigida
Solidago rugosa
Solidago speciosa
Sorghastrum nutans
Symphyotrichum lateriflorum
Symphyotrichum novaeangliae
Symphyotrichum oolentangiense
Symphyotrichum pilosum
Symphyotrichum puniceum
Symphyotrichum urophyllum
Toxicodendron radicans
Tradescantia ohiensis
Tridens flavus
Triodanis perfoliata
Ulmus sp.
Verbena hastata

Origin
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native
native

Mean Cover
0.08%
0.02%
0.01%
0.11%
0.00%
0.06%
0.00%
4.12%
0.00%
0.00%
0.48%
2.04%
1.46%
4.05%
0.06%
0.74%
0.00%
0.05%
25.63%
0.09%
0.12%
0.01%
0.01%
16.83%
0.99%
0.10%
1.86%
31.82%
0.18%
0.45%
0.08%
0.69%
0.01%
0.07%
0.22%
0.11%
0.36%
0.00%
0.05%
0.00%
65	

In seed mix?
N
N
N
N
N
Y
N
Y
N
N
N
N
N
Y
Y
Y
N
N
Y
Y
Y
Y
N
N
Y
N
Y
N
N
Y
Y
Y*
N
Y
N
Y
N
N
N
Y

	
Table S3.1 (cont’d)
Species
Verbena stricta*
Veronicastrum virginicum
Viola striata
Vitis aestivalis
Vitis riparia
Zizia aurea

Origin
native
native
native
native
native
native

Mean Cover
0.03%
0.02%
0.01%
0.01%
0.14%
0.16%

66	

In seed mix?
N
Y
N
N
N
Y

	
Table S3.2. Full model set predicting residuals of native richness~exotic richness
regression. See Table 3.2 for details. Cov = Covariate.
Cov

Mix SR

Mix
Rate

Fire

Cov

SR x
Cov

Rate
x Cov

Fire x
Cov

Delta
AICc

df

weight

R2

None

5.91***

-3.36*

-

-

3.65**

-

-

0

5

0.59

0.48

Age

4.14**

-3.80**

-

-2.72*

-

-

-

2.6

5

0.16

0.43

CV Biomass

4.30**

-5.17**

-

-2.38

-

-

-

5.6

5

0.04

0.37

Biomass

4.60**

-4.57**

-

1.99

-

-

-

5.7

5

0.03

0.37

None

4.83**

-4.13**

-

-

-

-

-

5.8

4

0.03

0.33

Land PCA

5.25***

-3.99**

-

-1.85

-

-

-

6.1

5

0.03

0.36

Edge:Area

5.05***

-4.27**

-

1.78

-

-

-

6.1

5

0.03

0.36

Land Use

7.96***

-4.12*

-

2.96

-6.48*

2.69

-

7.6

7

0.01

0.42

Age

4.41**

-4.33**

-

-2.43+

-0.76

1.22

-

7.9

7

0.01

0.41

Land Use

4.73**

-3.42+

-

2.16

-

-

-

8.1

5

0.01

0.31

CV WHC

4.61**

-3.94**

-

-1.58

-

-

-

8.3

5

0.01

0.31

Veg Cover

4.81**

-4.07**

-

-5.39

-

-

-

8.5

5

0.01

0.30

Fire

4.59**

-3.93*

5.32

-

-

-

-

8.5

5

0.01

0.30

Soil PCA

4.84**

-4.16**

-

-1.65

-

-

-

8.7

5

0.01

0.30

Fire

4.83**

-4.46**

1.22

-

-2.60*

1.22

-

9.3

7

0.01

0.38

Age

-

-

-

-3.56*

-

-

-

9.7

3

0.005

0.18

Biomass

4.58**

-4.27**

-

2.36+

0.65

1.52

-

10.4

7

0.003

0.36

Edge:Area

5.86**

-5.24**

-

2.40+

2.55

-0.92

-

11.3

7

0.002

0.34

Land PCA

5.54**

-3.15+

-

-1.68

0.95

0.55

-

11.7

7

0.002

0.33

Veg Cover

4.11**

-3.29*

-

-1.23

2.68

0.78

-

11.7

7

0.002

0.33

CV Biomass

4.34**

-5.10**

-

-2.21

-0.59

0.30

-

12

7

0.001

0.32

None

-

-

2.47+

-

-

-

-

13.5

3

<0.001

0.07

CV WHC

-

-

2.50+

-0.16

-

-

5.71+

13.8

5

<0.001

0.16

CV WHC

4.65**

-3.88*

-

-0.83

2.04

-1.64

-

14

7

<0.001

0.27

Edge:Area

-

-

2.97*

2.00

-

-

-

14.1

4

<0.001

0.10

NULL

-

-

-

-

-

-

-

14.2

2

<0.001

0

Land Use

-

-

-

4.35

-

-

-

14.2

3

<0.001

0.05

Land PCA

-

-

2.85+

-1.80

-

-

-

14.5

4

<0.001

0.09

CV WHC

-

-

-

-4.06

-

-

-

14.6

3

<0.001

0.04

Soil PCA

4.66**

-4.20**

-

-0.01

-0.99

-0.23

-

14.9

7

<0.001

0.25

Biomass

-

-

-

1.73

-

-

-

15.2

3

<0.001

0.02

CV WHC

-

-

2.05

-2.88

-

-

-

15.2

4

<0.001

0.07

Land Use

-

-

1.89

2.87

-

-

-

15.3

4

<0.001

0.07

Biomass

-

-

2.15

1.13

-

-

-

15.6

4

<0.001

0.05

CV Biomass

-

-

-

-1.42

-

-

-

15.7

3

<0.001

0

Edge:Area

-

-

-

1.25

-

-

-

15.9

3

<0.001

0

Land PCA

-

-

-

-1.19

-

-

-

16

3

<0.001

0

67	

	
Table S3.2. (cont’d)

-

Rate
x
Cov
-

Fire
x
Cov
-

2.46

-

-

2.39

-0.36

-

2.45+

0.2

-

-

-0.85

2.26

-

-

-

CV Biomass

-

Biomass
Soil PCA

Mix
SR

Mix
Rate

Fire

Cov

SR x
Cov

CV Biomass

-

-

2.24

-0.73

Edge:Area

-

-

3.52*

Veg Cover

-

-

Soil PCA

-

-

Veg Cover

-

-

Land PCA

-

-

Soil PCA

-

Veg Cover

-

Land Use

Cov

Delta
AICc

df

weight

R2

16

4

<0.001

0.04

1.83

16.1

5

<0.001

0.1

-

-

16.2

4

<0.001

0.04

-

-

16.2

4

<0.001

0.03

-

-

-

16.3

3

<0.001

0

-1.24

-

-

1.68

16.6

5

<0.001

0.08

-

0.36

-

-

-

16.6

3

<0.001

0

2.16

-0.8

-

-

2.19

17.8

5

<0.001

0.04

2.52

3.01

-

-

-1.46

17.9

5

<0.001

0.04

-

2.31

-0.35

-

-

0.94

18.5

5

<0.001

0.02

-

-

2.18

1.11

-

-

-0.11

18.5

5

<0.001

0.02

-

-

2.37

0.12

-

-

0.29

19.1

5

<0.001

0

Table S3.3. Correlation coefficients among factors used to predict native richness. Pearson’s
r, except for land use, which is result of one-way ANOVA; *p<0.05, +p<0.10.
Mix
SR
0.46*
0.27

Mix
Rate
-0.18

Fire

Edge:
Area

Land
PCA

Biomass

CV
Biomass

Veg
Cover

Soil
PCA

-0.12
0.36+
0.24

-0.43*
0.06

0.21

-

Mix Rate
Fire
Edge:
Area
Land PCA
Biomass
CV
Biomass
Veg Cover
Soil PCA
CV whc

-0.09
0.27
0.21

0.02
0.18
0.27

-0.25
0.21
0.28

-0.29
0.26

0.24

-0.42*
0.03
0.04
-0.17

-0.54*
0.11
-0.08
0.12

-0.30
-0.20
0.07
-0.29

0.07
-0.04
0.16
0.04

-0.18
0.20
-0.15
0.05

Age

-0.20

0.01
< in
row
crop

-0.04
> in
row
crop

-0.12

0.02

0.46*
0.18
-0.14
-0.05
0.36+

-

-

-

Land Use

-

68	

0.21

0.05
< in
row
crop

CV
whc

-0.06

0.06

-

-

Age

-

	
Table S3.4. Variation in functional group composition of seed mixes.
Non-leguminous
Forbs

Leguminous
Forbs

C3
Graminoids

C4
Graminoids

Minimum Richness

4

1

0

3

Maximum Richness

51

8

9

6

Median Richness

22

5

2

5

0.62

0.15

0.07

0.16

Minimum mass of Seeds Sown (g/m )

0.003

0.001

0

0.392

Maximum mass of Seeds Sown (g/m2)

0.473

0.126

0.252

0.785

Median mass of Seeds Sown (g/m )

0.286

0.061

0.123

0.549

Proportion of Seeds Sown (by mass)

0.25

0.06

0.12

0.57

Proportion of Richness
2

2

Table S3.5. Correlation coefficients among seed mix composition factors used to predict
native richness. Pearson’s r, *p<0.05, +p<0.10. SR=species richness. Coefficients between SR
and rate of same functional group in bold.
Total
Rate

C3
Rate

C4
Rate

Forb
Rate

Legume
Rate

Total
SR

C3
SR

C4
SR

Forb
SR

C3 Rate

0.43*

C4 Rate

0.07

-0.48*

Forb Rate
Legume
Rate

0.82*

0.37+

-0.40*

0.69*

0.35+

-0.33+

0.70*

Total SR

0.48*

0.73*

-0.38+

0.51*

0.20

C3 SR

0.26

0.53*

-0.63*

0.55*

0.14

0.76*

C4 SR

0.22

0.60*

-0.23

0.14

0.20

0.66*

0.33+

Forb SR
Legume
SR

0.48*

0.70*

-0.30

0.48*

0.13

0.99*

0.71*

0.65*

0.53*

0.73*

-0.36+

0.48*

0.51*

0.81*

0.51*

0.55*

0.77*

Age

-0.25

-0.08

0.48*

-0.61*

-0.23

-0.26

-0.61*

0.20

-0.24

69	

Legume
SR

-0.04

	
Figure S3.1. Change in richness over time differs by species origin. a) Native species richness
declines with restoration age (r = -0.50, p < 0.01), while exotic species richness does not (r = 0.21, p = 0.28); b) declines in native species richness are not driven strongly by either the sown
(r = -0.31, p = 0.11) or non-sown (r = -0.34, p = 0.07) component.

a)

b)

20

30

40

All; r = -0.50, p < 0.01
Non-sown; r = -0.34, p = 0.07
Sown; r = -0.31, p = 0.11

0

10

Native Species Richness

30
20
10
0

Species Richness

40

Native, r = -0.50, p<0.01
Exotic, r = -0.21, p=0.28

3

4

5

6

7

8

3

4

5

6

Restoration Age

Restoration Age

70	

7

	
Figure S3.2. Interaction between seed mix richness and seeding rate not driven by legumes
or C4 grasses in seed mix. Heat map showing response of residuals from native SR~exotic SR
regression to interactive effect of the seed mix richness and seeding rate of functional groups not
shown in Figure 5, and between each functional group and the total richness and rate of seed
mixes (see models in Table 3.3). Darker shading = higher proportional native richness (i.e.,
positive residuals from native SR ~ exotic SR regression); lighter shading = lower proportional
native richness (i.e., negative residuals). Interactions between richness and seeding rate of
legumes (a), and C4 grasses (d); interactions between total seed mix richness and seeding rate of
legumes (b) and C4 grasses (e); interactions between total seeding rate and seed mix richness of
legumes (c), C4 grasses (f).

71	

	
Figure S3.3. Dominant species effects differ across components of richness. The richness of
exotic and non-sown native species is lower when Andropogon gerardii is abundant (a,d), while
the richness of sown native species is associated instead with Sorghastrum nutans abundance (h).
When controlling for the abundance of A. gerardii and S. nutans, the richness of exotic (c), nonsown native (f), and sown native species is not correlated with restoration age. Partial-residual
plots for the model: richness~A. gerardii + S. nutans + age.

a)

b)
Exotic Richness

Exotic Richness

20

10

5

p = 0.16
Exotic Richness

p = 0.02

15

c)

15
10
5

15

p = 0.28

10

5

0
30

40

p = 0.08

30

20

10

0

Sown Native Richness

g)

0

10

20

30

Non-sown Native Richness

% Cover Andropogon gerardii

20

15

10

p = 0.58
0

10

20

30

40

% Cover Andropogon gerardii

10

20

30

0

40

% Cover Sorghastrum nutans

35

p = 0.36

30
25
20
15
10

f)

3

4

5

6

7

8

7

8

7

8

Restoration Age

p = 0.91

20

15

10

5

5

40

% Cover Andropogon gerardii

5

e)

0

Sown Native Richness

20

0

h)
30

10

20

30

40

% Cover Sorghastrum nutans

p = 0.07

i)

3

4

5

6

Restoration Age

20

Sown Native Richness

10

Sown Native Richness

Non-sown Native Richness

d)

0

25
20
15
10
5

15

10

5

p = 0.99
0

10

20

30

40

% Cover Sorghastrum nutans

72	

3

4

5

6

Restoration Age

	
Figure S3.4. The restoration of native richness relative to exotic richness is linked to
reduced exotic dominance. The relative abundance of exotic species was lower in sites where
native richness was higher than predicted by the native-exotic richness relationship. Curve fit by
logistic regression with a quasi-binomial error distribution; pseudo-R2 = 0.31, p<0.01.

1

Exotic Dominance

0.8

0.6

0.4

0.2

0
-10

0

10

Residuals of NativeSR~ExoticSR

73	

	
Figure S3.5. Random (null) vs. observed correlation coefficient and slope of native-exotic
richness relationship.	We used a randomization procedure to examine whether the observed
native-exotic richness relationship differed from the null expectation. Here, the null expectation
is the native-exotic richness relationship in communities randomly assembled from a regional
species pool; without respect to species origin, and not shaped by native and exotic species
responding similarly to factors like resource availability or heterogeneity (Stohlgren et al. 1999,
Fridley et al. 2004). Holding the proportion of native and exotic species in the observed species
pool constant, as well as the total site-level species richness, we generated a null distribution
(with replacement) of both correlation coefficients (r) and slopes (s). We tested whether the
observed r and s differed from the null distribution with one-tailed p-values. The observed
pattern, however, did not differ from the null model based on communities simulated under the
assumption of random assembly (correlation coefficient (r), one-tailed p = 0.28; slope, one-tailed
p = 0.20).

a)

b)

null r = 0.59 (0.41,0.77)

solid=observed slope, dashed=mean(null slope)

null slope = 1.30 (0.77,1.82)

80

50

solid=observed r, dashed=mean(null r)

obs slope = 1.07, p=0.20

40

Frequency

30

20

20
10

0

0

Frequency

60

40

obs r = 0.54, p=0.28

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0.5

Null distribution of r, 2011 Native~Exotic Transect SR

1.0

1.5

2.0

Null distribution of slope, 2011 Native~Exotic Transect SR

74	

	
Figure S3.6. Observed native-exotic richness relationship does not differ from null.
Observed = solid; null = dashed.

30
20
10
0

Native Richness

40

observed, y=1.07x + 15.65; r = 0.54
null, y=1.30x + 12.68; r = 0.59

0

5

10

15

Exotic Richness

75	

	
CHAPTER FOUR
RESTORING GRASSLANDS TO RESIST INVASION – RESTORED DIVERSITY CAN
MEDIATE HISTORICAL LAND-USE EFFECTS
Abstract
While exotic species invasions are a primary threat to native ecosystems, the extent to
which communities are invaded varies widely. This variation is due to processes occurring in the
landscape surrounding invaded communities, as well as intrinsic community and ecosystem
properties. However, the relative influence of these processes for controlling invasion remains
poorly resolved. Importantly, this limits the capacity for developing management actions to
promote invasion-resistant ecosystems. We used structural equation models to evaluate how
factors influencing propagule pressure (landscape context and land-use history), abiotic site
conditions (soil moisture availability), and the restored native community (richness of sown
species, abundance of the community dominant Andropogon gerardii) related to the richness and
abundance of invasive species in 29 tallgrass prairie grassland drestorations. We also tested
whether management factors (seed mix richness and fire frequency) supported invasion
resistance directly by reducing the richness or abundance of invasive species, or indirectly by
modifying aspects of the restored community. We found that sandy, dry soils and A. gerardii
abundance limited invasive richness, but propagule pressure and management did not. In
contrast, invasive abundance was associated with management, propagule pressure, site
conditions, and the restored native community. Invasive species abundances were dramatically
lower in sites restored from row crops (1.6% cover) than in sites restored from hay fields or old
fields (23.8% cover). Seed mix richness limited invasive species abundances indirectly by
increasing the richness of sown species, which in turn was generally associated with lower

76	

	
abundances of invasive species. The abundance of A. gerardii also generally reduced invasive
abundance. In contrast, fire frequency had virtually no direct or indirect effect on invasive
richness or abundance. Our results illustrate how a suite of non-independent processes, including
propagule pressure, resource availability, and community structure, can together affect the
degree of invasion. From a practical perspective, land-use legacies may require management to
reduce invasive species abundance (e.g., in seed banks) prior to restoration. However, managers
can also mediate these effects and resist invasion by altering the restored community through
seed additions.

Introduction
Exotic species invasions are one of the principal threats to populations and communities
of native organisms, and often disrupt vital ecosystem functions (Parker et al. 1999, Vila et al.
2011). However, the extent to which communities are invaded varies widely, suggesting that
some communities are better able to resist invasion than others. Theory attributes this variation
to many factors (Catford et al. 2009), including propagule pressure of invaders (Simberloff
2009), resource availability and heterogeneity (Davis et al. 2000, Melbourne et al. 2007), and
resident species diversity (Elton 1958, Levine et al. 2004). While there is empirical support for
each of these hypotheses in some contexts, and most authors recognize that multiple factors may
simultaneously drive invasion, the relative role of each process is unclear in most systems
(Catford et al. 2009, Gurevitch et al. 2011).
The extent or degree of invasion, defined here as the abundance or richness of invasive
species, is the outcome of processes occurring at multiple scales - both in the invaded
community, and in the surrounding landscape (Lonsdale 1999, Richardson and Pysek 2006,

77	

	
Chytry et al. 2008, Catford et al. 2012, Guo et al. 2015). First, the propagule pressure of invaders
from surrounding landscape - the rate, frequency, and diversity at which they arrive – drives
initial population growth in the invaded community, and land use and landscape composition
influence propagule pressure (Colautti et al. 2006, Simberloff 2009, Hufbauer et al. 2013).
Landscapes with more intense anthropogenic disturbance (e.g., agricultural and urban
development) support a greater density of species with traits linked to invasiveness, so
anthropogenic disturbance is frequently correlated with propagule pressure (Stohlgren et al.
2006, Vila and Ibanez 2011, Gifford and Otfinowski 2013). Propagule pressure may also have a
temporal component – variation in anthropogenic disturbances through time may determine the
accumulation of invasive propagules in seed- and budbanks, especially on sites with agricultural
histories (Gioria et al. 2014, Zylka et al. 2016). Second, both abiotic site conditions and aspects
of the restored community determine the susceptibility of a community to invasion (Davis et al.
2000, Melbourne et al. 2007). Sites with low or homogenous resource availability are predicted
to be more resistant to invasion, because fewer niches are available to be exploited (Davis et al.
2000, Shea and Chesson 2002, Melbourne et al. 2007). Similarly, communities with high
diversity or where highly competitive species occur at high abundances are predicted to be more
resistant to invasion (in part because less resources are available for invaders to exploit) (Davis
et al. 2000, Kennedy et al. 2002, Smith et al. 2004). Third, additional factors may influence the
degree of invasion indirectly, particularly by modifying community structure. Disturbances in
particular can influence invasion resistance by modifying resource availability, altering
community structure, or both (Burke and Grime 1996, Clark and Johnston 2011). For example,
fire may have direct negative effects on some invasive species, or the effect of fire may be
mediated via effects on native plant abundance (Copeland 2002, Howe 2011).

78	

	
In this paper, we focus on how invasion ecology theory can be tested in restored
ecosystems that vary in characteristics that are potentially important to invasion, including
landscape context, land-use history, soil moisture, and diversity and abundance of the restored
community. Furthermore, by testing this theory in real restorations, this theory can be translated
into management guidelines. First, although it may be difficult to control propagule pressure,
identifying how historical and surrounding land-uses influence invasion can inform site selection
for restoration projects or allow managers to budget for more intensive invasive species control
or for setting manager’s expectations. Second, an understanding of how resource availability is
linked to the degree of invasion, as with determinants of propagule pressure, can also inform site
selection and planning (Zefferman et al. 2015). Third, managers can build invasion resistance
into restored communities in multiple ways. For example, sowing more species at higher rates
may resist invasion indirectly by restoring diverse communities of native species that will
compete more efficiently for resources (Biondini 2007, Piper et al. 2007, Middleton et al. 2010).
On the other hand, managers can restore natural disturbance regimes like fire that may directly
reduce the degree of invasion (Howe 1994). Alternately, managers can manipulate the
dominance of strong resource competitors (Smith et al. 2004) through natural disturbance
regimes such as fire (Copeland 2002, Howe 2011). We use structural equation modeling to test
how propagule pressure, site conditions, and community structure were linked to the degree of
invasion in 29 tallgrass prairie grassland restorations, and to better understand how management
can strengthen invasion resistance. Our purpose was twofold: provide a real-world test of theory,
and assess the utility of this theory for guiding the management of invasive species in prairie
restorations. In order to strengthen the applied value of our work, we classified invasive species
narrowly, as the introduced species that have large negative impacts on native plants and

79	

	
communities and that are prioritized for control by managers. Specifically, we asked how
management factors, seed mix richness and the frequency of prescribed fire, influenced the
richness and abundance of invasive species and compared the relative influence of management
to other factors that theory has linked to invasion success – landscape context, land-use history,
soil moisture, and diversity and dominance in the sown community - as well as how management
may influence the degree of invasion indirectly by modifying diversity and dominance.

Methods
Prairie establishment
We conducted our study in 29 tallgrass prairies in southwest Michigan, restored from
former agricultural land, using methods typical for prairie restoration in the upper Midwest
(Packard and Mutel 1997). Between 2003 and 2008, each site was cleared of existing vegetation
with herbicides, and then sown with a mix of native prairie species. Following restoration, all
sites were initially managed with mowing to minimize the fecundity of weeds and increase light
availability for establishing native species. Thereafter, most restorations were managed with
prescribed fire, to discourage woody species and exotic C3 grasses and encourage sown native
species. Additional site details are found below (Classifying propagule pressure, abiotic site
conditions, restored community and management), and in Grman et al. (2014).

Field sampling
To characterize both the richness and abundance of invasive species (i.e., degree of
invasion) and sown species (e.g. elements of community structure predicted to resist invasion),
we sampled plant community composition in July-September 2011. We recorded the percent

80	

	
cover of all vascular plant species in 10 evenly spaced 1-m2 plots along a 45-meter transect,
located at the center of each site. The cover of all species could sum to greater than 100%. To
quantify resources as a determinant of community invasibility, we sampled 80 soil cores per site
(8 equidistant cores, 20 cm x 3 cm2, around each 1-m2 plot). We pooled soil samples by site, and
analyzed soil organic matter, soil texture (sand, silt, and clay content), Mehlich-III phosphorus,
Bray-II phosphorus, and pH (Brookside Laboratories, New Knoxville, OH, USA). We also
determined plot-scale soil water holding capacity in the lab as the proportion of saturated wet to
oven dry weight (after Brudvig and Damschen 2011).

Classifying invasive species
We classified all introduced plant species in our dataset as invasive or non-invasive
with the U.S. Invasive Species Impact Rank (I-Rank, Morse et al. 2004). Here, invasive species
are those introduced species that have large negative impacts on native plants and communities,
and are also the species that are prioritized for control by managers. Because introduced species
impacts may increase with time since introduction (Ahern et al. 2010), we conducted a parallel
analysis with all introduced species, but the results were qualitatively similar (see Figure S4.3).
The impacts of most species are not consistent in all habitats and regions of the United States, so
we also crosschecked several regional published lists. We identified a core list of fourteen
invasive species, which had I-Ranks associated with “Medium” or “High” impacts and occurred
on at least two regional lists (Category A in Table S4.1). Finally, we consulted regional
restoration professionals on the impacts of the remainder of introduced species. In so doing, we
identified five additional species that are considered invasive, specifically in prairie restoration
(Category B in Table S4.1), resulting in 19 invasive species included in our dataset. Next, we

81	

	
characterized the degree of invasion in two ways – richness as the cumulative number of
invasive species per transect, and abundance as the mean percent cover of invasive species per
m2 along each transect. We interpreted invasive richness as an indicator of colonization success
and an increased probability of future impacts. The percent cover of invasive species is a better
determinant of current impacts. Finally, the response of individual species or groups of species
may differ, and management can be tailored to address these species, so we also modeled the
percent cover of the C3 clonal grass Poa pratensis, all invasive Trifolium spp. (T. hybridum, T.
pratense, and T. repens), and the summed cover of all invasive forbs. These species groups
represent three different functional groups (C3 grasses, legumes, forbs) that may respond
differently to environmental conditions and management actions (Symstad 2000), and for which
there was sufficient variation in abundances to satisfy the parametric assumptions of SEM.
Percent cover values of invasive species groups were log-transformed to conform to assumptions
of normality.

Classifying propagule pressure, abiotic site conditions, restored community and management
We characterized attributes of each site related to propagule pressure, including the
landscape context and land use history of each restoration (Table 4.1). In ArcGIS, we calculated
the area of forest, grassland, wetlands, agriculture, and development within 500 meters of the
center of each site. To make inferences on the effects of landscape context using these highly
correlated data, we conducted a principal components analysis on landscape data and used the
first axis for analysis (hereafter, landscape context). Explaining 59% of the variation, high values
on the landscape PC axis correspond to high cover of agriculture and low cover of forest and
grassland. We consulted land managers on the land use prior to restoration. While all sites were

82	

	
Table 4.1. Potential drivers that are hypothesized to determine the degree of invasion. Each
hypothesized driver is linked to restoration based on theory (How is theory applied to
restoration?), and based on how managers can manipulate each driver to reduce the degree of
invasion (Implications for management). See Figure S4.1 for SEM meta-model.
Hypothesis

How is theory applied in restoration?

Implications for management

Propagule Pressure
Landscape
Context
(Landscape PC
Axis)

Propagule pressure
(Lonsdale 1999,
Colautti et al. 2006,
Simberloff 2009)

Some land cover types (e.g., agriculture) may
support smaller invasive species populations than
others (e.g., forest, grassland) and therefore
contribute invasive propagules at lower rates.

Site selection. Managers can select sites in
landscapes with lower propagule pressure,
or plan/budget to engage in more intensive
management, post-restoration.

Land-use History

Propagule pressure
(Lonsdale 1999,
Colautti et al. 2006,
Simberloff 2009)

Differences in historical land management (row
crops vs. perennial grasslands) select for different
types and abundances of invasive species in the
seedbank and budbank.

Site selection/Pre-restoration
management. Managers can select sites
with histories that contribute lower
propagule pressure; or adjust management
prior to restoration to reduce propagule
pressure.

Sites with lower resource availability (here,
primarily soil moisture, a key driver of community
dynamics in grassland) are more difficult to
invade.

Site selection/Post-restoration
management. Managers can select sites
with lower resource availability; or
plan/budget to engage in more intensive
management, post-restoration.

Sown Richness
(Established)

Diversityinvasibility (Elton
1958, Levine et al.
2004)

The establishment of a diverse community of
sown native species confers biotic resistance;
diverse communities of resident species occupy
available resources more efficiently (e.g.,
complementarity).

Seed additions. Managers can sow more
diverse seed mixes.

% cover of
dominant species
(A. gerardii)

Selection effect
(Crawley et al.
1999, Smith et al.
2004)

Sites dominated by a single highly competitive
species may be more resistant to invasion; A.
gerardii may draw down resources more
effectively than a diverse community.

Seed additions/Prescribed fire. Managers
can sow A. gerardii at higher rates, or
adjust timing of burns to encourage A.
gerardii.

Seed Mix
Richness (Sown)

Diversityinvasibility (Elton
1958, Levine et al.
2004)

Sowing more species should establish a more
diverse "resident community" and increase biotic
resistance to invasive species.

Seed additions. Managers can sow more
diverse seed mixes.

Fire Frequency

Disturbance (Hobbs
and Huenneke
1992)

Fire increases light availability but decreases soil
nutrient availability, so may increase or decrease
invasibility. The response of individual species to
fire also varies - so the response of specific
invaders may vary.

Prescribed fire. Managers can adjust the
timing of burns.

Resource Availability

Soil Moisture
(Soil PC Axis)

Fluctuating
resources (Davis et
al. 2000); Invasion
windows (Shea and
Chesson 2002)

Community Structure

Management

historically in row crops, sites occupied different land uses immediately prior to restoration –
perennial grasslands (hay fields, old fields, pastures; n=16) and row crops (n=13). Landscapes
and land uses dominated by non-agricultural land uses, especially perennial grasslands, are more
likely to accumulate larger populations of invasive species (Gifford and Otfinowski 2013). We
83	

	
predict that these landscapes and land uses will be associated with increased propagule pressure
(Lonsdale 1999, Simberloff 2009) (Table 4.1).
We also characterized abiotic site conditions and the community structure of sown
species (Table 4.1). First, we conducted a principal components analysis on site-level soils data.
The first axis, which we used as a measure of resource availability, explained 56% of the
variation in soil resources. High values on this axis (hereafter, soil moisture) correspond to sandy
sites with low water holding capacity. Sandy, dry sites represent low overall resource availability
and provide shorter windows of opportunity for invasive species establishment by retaining
moisture for shorter periods of time (Shea and Chesson 2002) (Table 4.1). Second, we used the
richness of sown species and the percent cover of the dominant sown species, Andropogon
gerardii as aspects of community structure that may confer invasion resistance. We used sown
richness, calculated as the cumulative richness of sown species that were observed along each
transect, as a measure of resident community diversity (Levine and D’Antonio 2004) (Table 4.1).
Andropogon gerardii, the most abundant C4 grass at our sites, is a widespread dominant across a
wide variety of soils in both remnant and restored prairies (Weaver and Fitzpatrick 1932, Carter
and Blair 2012). Dominance of C4 grasses, rather than species richness, often drives invasion
resistance by competing more efficiently for resources (Tilman 1999, Smith et al. 2004,
Mahaney et al. 2015) (Table 4.1).
Additionally, we used seed mix richness and prescribed fire frequency to test for the
effects of management on the degree of invasion (Table 4.1). We obtained data on seed mix
richness from the restoration contractor (Native Connections, Three Rivers, MI). Sites were
sown with 8-71 species (mean = 35 species). Increasing seed mix richness should establish a
more diverse resident (e.g. sown) community (Grman et al. 2013), which may more effectively

84	

	
resist invasion (Table 4.1). Fire is a primary natural disturbance in our system, and more frequent
fire may directly limit the richness and abundance of invasive species (Emery and Gross 2005,
Brudvig et al. 2007, Li et al. 2013). We consulted land managers on the prescribed fire history of
each site. Sites were burned between 0 and 4 times since being restored, and we calculated fire
frequency as the number of burns divided by years since restoration. While more frequent fire
may reduce invasive species such as C3 grasses without impacting C4 grasses (Li et al. 2013),
more frequent fire may also resist invasion indirectly by increasing the cover of A. gerardii
(Copeland 2002, Howe 2011) (Table 4.1).

Data analysis
We constructed structural equation models (SEM; Grace 2006) to test how propagule
pressure, site conditions, and community structure influence invasion and how management
might influence these processes. We used a SEM approach to better understand both direct and
indirect causal relationships that lead to observed levels of invasion, as a test of theory and in
order to better inform management (Grace et al. 2010). The exact metric used to characterize the
degree of invasion can have different interpretations (Guo et al. 2015), and we created five
separate models with different invasion-related endogenous variable – invasive richness, the
summed percent cover of all invasive species, and the percent cover of specific invasive species
or groups (Poa pratensis, Trifolium spp., and all invasive forbs).
We started by fitting the same simple SEM for each invasion metric. This initial
model included the direct effects of invasive propagule pressure from the surrounding landscape
(landscape PC axis) and via site history (land use history), abiotic site conditions (soil moisture),
community structure (richness of established sown species, abundance of Andropogon gerardii),

85	

	
and management (prescribed fire frequency) (Figure S4.1). We also modeled the indirect effect
of management on invasion through its effects on community structure, by including an effect of
seed mix richness on the sown species richness, and the effect of fire frequency on the
abundance of A. gerardii. Then, we assessed change in model fit (primarily by likelihood ratio
tests), after adding paths suggested by modification indices. When we determined a final model,
we calculated standardized path coefficients (r) for each predictor. Finally, to estimate the
influence of each variable on invasive richness and abundance, we calculated the total
standardized effect (TSE) of each predictor: the sum of all indirect and direct standardized path
coefficients.

Results
Modification indices suggested that soil moisture also influenced the richness of sown
species. We therefore included this path in our models, and this improved the fit of models for
both richness (model χ 2 = 12.53, df=9, p = 0.19) and cover (model χ 2 = 11.16, df=9, p = 0.27).
These models explained 46% of the variation in the richness, and 78% of the variation in the
cover of invasive species. Models with the same structure predicted 45% of the variation in the
cover of the invasive C3 grass Poa pratensis (χ 2 = 11.14, df=9, p = 0.27), 68% of Trifolium spp.
cover (χ2 = 11.50, df=9, p = 0.24), and 32% of summed invasive forb cover (χ 2 = 17.76, df=9, p
= 0.23). In all cases, the models fit the data well (p > 0.05 indicates good fit; see Table S4.2 for
additional fit measures).
Invasive species richness was strongly associated with abiotic site conditions and
community structure (Figures 4.1a, 4.2a), while propagule pressure, site conditions, community
structure and management all influenced invasive species cover (Figures 4.1b, 4.2b). Invasive

86	

	
Figure 4.1. Structural equation models for invasive richness (a) and percent cover (b). Solid
arrows = positive effect, dashed = negative; black arrows = significant effect (p<0.05); gray =
non-significant (p>0.05). Width of arrows roughly proportional to effect size: predictions for
each path in Table 4.1; standardized path coefficients in Table 4.2.

87	

	
Figure 4.2. Total standardized effects on invasive richness (a) and cover (b). Effects of
landscape context (landscape), land use history (land use), soil moisture (soil mstr.), abundance
of Andropogon gerardii (A. gerardii), richness of sown speices (sown SR), seed mix richness
(seed mix), and fire frequency (fire). See Table 4.2 for parameter estimates.

richness was lower in in sites dominated by Andropogon gerardii (r = -0.43, p < 0.001; Table
4.2), and in sandy sites with lower soil moisture retention capacity (r = -0.42, p < 0.001; TSE = 0.38; Table 4.2). The effects of both seed mix richness (TSE = -0.14) and fire frequency (TSE =
-0.13) on invasive species richness were weak (Table 4.2), suggesting a limited role for these
management practices in determining invasive richness. Invasive species cover was reduced by
sown species richness (r = -0.45, p < 0.001; Table 4.2) and A. gerardii abundance (r = -0.41, p <
0.001; Table 4.2), and was lowest in sites that were in row crops immediately prior to restoration
(r = -0.60, p < 0.001; Table 4.2). Soil moisture only had a marginally significant direct effect on
invasive cover (r = -0.15, p = 0.099; Table 4.2), but also indirectly increased invasive cover by
88	

	
reducing sown species richness (r = -0.22, p = 0.02; Table 4.2), resulting in no overall effect of
soil moisture on invasive cover (TSE = -0.05; Table 4.2). Fire frequency had no direct or overall
effect on invasive cover (TSE = 0.01) (Table 4.2). Seed mix richness, however, had a negative
effect on invasive cover (TSE = -0.26) by increasing sown species richness (r = 0.84, p < 0.001;
Table 4.2). The response of invasive species groups to land-use history and A. gerardii mirrored
the response of total invasive cover, but otherwise varied among species. As we found for
summed invasive species cover, sites recently in row crops had or trended toward lower cover
for Poa pratensis (r = -0.40, p < 0.001), Trifolium (r = -0.22, p = 0.07), and invasive forbs (r = 0.39, p = 0.03) (Table 4.2; Figures 4.3, 4.5); and A. gerardii abundance limited or tended to limit
the cover of Poa pratensis (r = -0.27, p = 0.06) and Trifolium (r = -0.50, p < 0.0001) (Table 4.2;
Figures 4.2, 4.4). Poa pratensis was also less abundant in landscapes dominated by agriculture (r
= -0.40, p < 0.01), while Trifolium spp. were limited by soil moisture (r = -0.44, p < 0.0001; TSE
= -0.50). Sown species richness was associated with reduced invasive forb cover (r = -0.36, p =
0.03), and increased Trifolium spp. cover (r = 0.27, p = 0.02), but did not influence P. pratensis
(r = -0.23, p = 0.15). Because seed mix richness increased sown species richness (r = 0.84, p <
0.001; Table 2), seed mix richness was associated with higher cover of Trifolium spp. (TSE =
0.22), and lower cover of invasive forbs (TSE = -0.31) and P. pratensis (TSE = -0.20).

Discussion
We tested the relative roles of multiple factors hypothesized to influence invasion, and
used that theory to ask how managers might manipulate community structure to better resist
exotic species invasions in prairie restorations. As predicted, each class of drivers we considered

89	

	
Figure 4.3. Structural equation models for percent cover of invasive species groups. Poa
pratensis (a), Trifolium spp. (b), and all invasive forbs (c). Solid arrows = positive effect, dashed
= negative; black arrows = significant effect (p<0.05); gray = non-significant (p>0.05). Width of
arrows roughly proportional to effect size: predictions for each path in Table 4.1; standardized
path coefficients in Table 4.2.

90	

	
Figure 4.4. Total standardized effects on Poa pratensis (a), Trifolium spp. (b), and all
invasive forbs (c). Landscape context (landscape), land use history (land use), soil moisture (soil
mstr.), abundance of Andropogon gerardii (A. gerardii), richness of sown speices (sown SR),
seed mix richness (seed mix), and fire frequency (fire). See Table 4.2 for parameter estimates.

– propagule pressure, abiotic site conditions, community structure, and management – explained
variation in the degree of invasion. Invasive species richness was limited most strongly by site
conditions and community structure and not management. In contrast, management limited
invasive species abundances by increasing invasion resistance of the sown community. However
propagule pressure also strongly influenced invasive abundance, particularly via land-use
history. Among factors predicted to drive the degree of invasion, community structure
(specifically, the abundance of A. gerardii) was consistently associated with invasion resistance,
while site conditions (soil moisture) and propagule pressure (land-use history) appeared to play
particularly strong roles in driving the richness and abundance of invasive species, respectively.
91	

	
Below, we discuss how the different invasion hypotheses contributed to invasion in our study,
beginning with factors that are difficult to control (propagule pressure, abiotic site conditions),
and ending with factors that can be addressed through management.

Impacts of propagule pressure - landscape context and land use history
Invasive species abundances were higher in sites restored from perennial grasslands, and
to a lesser degree, in landscapes dominated by perennial grassland and forest instead of
agriculture. Propagule pressure drives invasion by overcoming demographic barriers to
colonization by increasing the density, frequency, and richness of propagules immigrating into a
community (Simberloff 2009, Hufbauer et al. 2013). Invasive species develop larger populations
over time in perennial habitats (i.e., in the absence of frequent or recent disturbance such as
tillage) such as hay- and old fields. Despite attempts to eliminate them prior to restoration, large

Table 4.2. Effects of each exogenous variable each endogenous variable in SEMs. Effects on:
sown SR (species richness), Andropogon gerardii, inv (invasive) SR, inv cover, Poa pratensis
cover, Trifolium spp. cover, and inv forb cover. Including direct standardized path coefficients
(***p < 0.001, **p < 0.01, *p < 0.05, +p < 0.10), and total standardized effect (TSE); from
structural equation models in Figures 4.1 and 4.3.
Sown
A.
SR
gerardii Inv SR TSE

Inv
Cover

P.
Inv
TSE pratensis TSE Trifolium TSE Forbs TSE

Land PCA

-

-

0.07

0.07 -0.37*** -0.37

-0.40**

-0.40

-0.01

-0.01 -0.02 -0.02

Land Use

-

-

-0.17

-0.17 -0.60*** -0.60

-0.37*

-0.37

-0.22+

-0.22 -0.39* -0.39

Soil PCA

-0.22*

-

-0.42** -0.38

-0.05

-0.21

-0.16 -0.44*** -0.50 -0.02

0.06

A. gerardii

-

-

-0.43** -0.43 -0.41*** -0.41

-0.27+

-0.27 -0.50*** -0.50

0.01

Sown SR

-

-

-0.16

-0.16 -0.45*** -0.45

-0.23

-0.23

0.27*

0.27 -0.36* -0.36

Fire Freq.

-

-0.14

-0.19

-0.13

-0.16

-0.10

-0.08

-0.04

-0.11

-0.04 -0.16 -0.16

0.84***

-

-

-0.14

-

-0.38

-

-0.19

-

Mix SR

-0.15+

92	

0.22

0.01

-

-0.31

	
Figure 4.5. Invasive species cover is lower in sites restored from row crops than from
perennial grasslands. (a) all invasive species, (b) the invasive C3 invasive grass Poa pratensis,
(c) Trifolium spp., and (d) all invasive forbs.

seed- and budbanks remain in sites restored from perennial grasslands, from which these species
quickly establish populations (Gioria et al. 2014, Zylka et al 2016). Invasive populations also
appear to be augmented by dispersal from perennial habitats in the surrounding landscape,
especially for the clonal C3 grass Poa pratensis, which is likely to be generally abundant in postagricultural landscapes (Toledo et al. 2014, Zylka et al 2016). Drivers related to propagule
pressure did not influence invasive richness, suggesting that propagule pressure generally
augmented existing invasive populations, as opposed to introducing additional invasive species.

93	

	
Impacts of site conditions – soil moisture
Low soil moisture limited invasive richness and the abundance of Trifolium spp. Sandy
soils with low water holding capacity represent lower overall resource levels and a harsher
environment, provide shorter windows of opportunity for invasive species establishment by
retaining moisture for shorter periods of time, and may constrain which species are able to persist
at these sites (Shea and Chesson 2002, Zefferman et al. 2015). Interestingly, low soil moisture
also limited sown richness, likely for similar reasons (Grman and Brudvig 2014). When sown
richness reduced invasive cover (summed invasive cover and cover of invasive forbs), the
indirect effect of soil moisture on invasive cover was largely neutral (or more weakly negative)
because soil moisture reduced sown richness, thereby reducing the negative effects of sown
richness on invasive cover. In contrast, the abundance of the invasive legume Trifolium spp. was
positively correlated with sown richness. By reducing the richness of sown species, soil moisture
therefore had an indirect negative effect on Trifolium spp. abundance.

Impacts of community structure – Andropogon gerardii abundance and sown richness
The richness and abundance of invasive species was lower in sites with high sown
species richness, greater abundance of Andropogon gerardii, or both. Among invasive species
groups, A. gerardii more strongly resisted the mat-forming P. pratensis and Trifolium spp., likely
by limiting light and soil resource availability for invasive species to exploit during colonization
and establishment (Tilman and Wedin 1991, Silletti et al. 2004, McCain et al. 2010). In contrast,
sown richness had particularly strong negative effects on invasive forbs, presumably because the
additional species found in sites with high sown richness occupied the same niche-space that
invasive forbs would otherwise occupy (Funk et al. 2008). Greater sown species richness was

94	

	
likely associated with more complete utilization of multiple resources, which were subsequently
unavailable for invaders to exploit (Kennedy et al. 2002, Levine and D’Antonio 2004). The
direction of causality, that of sown richness limiting invasive species abundances, is modelimplied and supported by theory and experiments (Levine and D’Antonio 2004). However, it is
also possible that invasive species abundances instead limit sown richness, or the two are
collinear along an unmeasured gradient.

Impacts of management – seed mix richness and prescribed fire
Sowing seed mixes with a greater number of prairie species encouraged the establishment
of a more diverse resident community, which in turn conferred greater invasion resistance in the
form of reduced cover of most invasive species. Therefore, managers can effectively manipulate
the propagule pressure of native prairie species to partially resist the colonization and
establishment of invasive species (Melbourne et al. 2007, Funk et al. 2008). In fact, dispersal via
seed mixes is the strongest determinant of sown species richness (Grman et al. 2013), although
dry, sandy soils limit sown species richness in our models here, and in previous work (Grman
and Brudvig 2014). Prescribed fires, another major management tool, typically occur during
early-spring green-up in order to target invasive C3 grasses such as P. pratensis, which often
serves to promote native C4 grasses such as A. gerardii (Howe 1994). However, fire frequency
had negligible effects on invasive richness and abundance in our models, as well as A. gerardii
abundance. The impact of fire on invasive species may not be consistent across space, as fire
effects often vary with seasonality (Howe 1994), soil type, and moisture availability (Collins
2000), and other factors that vary geographically (Bowles and Jones 2013).

95	

	
Summary and management implications
By considering a suite of processes believed to influence the extent exotic species
invasions, we were able to explain a considerable amount of the variation (R2 ≥ 0.45 in all but
one model) in both the richness and abundance of invasive species. As predicted, propagule
pressure, abiotic site conditions, and community structure all played a role in determining the
degree of invasion. Our study supports many theoretical predictions, including: 1) communities
with lower resource availability (Davis et al. 2000), greater resident species diversity (Levine et
al. 2004), and higher abundances of superior competitors (Smith et al. 2004) are often less
invasible; and 2) communities likely exposed to higher propagule pressure (particularly through
seedbanks resulting from prior land uses) are more heavily invaded (Simberloff et al. 2009).
With SEM, we too important steps toward clarifying the relative importance of each
hypothesized invasion driver in our system, and how their effects varied among metrics used to
define the degree of invasion. Community structure was a consistently important driver – the
abundance of A. gerardii was strongly linked to both lower invasive richness and abundances.
Soil moisture, on the other hand, was particularly important for limiting invasive richness, but
not abundance; while land-use history had large effects on invasive abundance and not richness.
The SEM approach also allowed us to connect management actions (seed sowing, fire) to
measurable invasion outcomes in sites undergoing restoration, in the context of realistic
environmental variation. Direct management of invasive species, through manual and chemical
means, is effective but time-consuming, and it is more efficient to build invasion resistance into
restored communities (Suding 2011). In order to reduce the richness and abundance of invasive
species in prairie restoration, our study suggests specific management actions during at least
three phases of the restoration process – site selection, site preparation, and establishment of the

96	

	
Figure 4.6. Decision tree for reducing degree of invasion in prairie restoration. Invasion
may be limited in sites managed in row crops prior to restoration (a) and sites with low soil
moisture (b). In other sites, the degree of invasion may be reduced by (c) managing with row
crops for a year or two prior to restoration or by (d) increasing the richness of sown species or
abundance of Andropogon gerardii via augmented seed additions.

restored community. Managers can directly control some factors that dictate the degree of
invasion, while others may present inherent limitations to managing invasive species. First, sites
with lower soil moisture availability are likely to have fewer invasive species overall, and lower
abundances of some species in particular (e.g., Trifolium), although it can also be more
challenging to establish desirable sown species in low moisture sites as well. Second, sites
formerly in row crops, and in some cases sites in landscapes dominated by agriculture, are likely
to have lower abundances of invasive species, especially clonal C3 grasses (e.g., Poa pratensis).

97	

	
For these reasons, selecting sites with lower soil moisture and row-crop land-use history may
benefit restoration by resulting in less invasion and need for subsequent management to target
invasive species (Figure 4.6a,b). Therefore, when restoring from row crops, and in drier sites,
standard site preparation may be sufficient for reducing the degree of invasion. When restoring
from sites in perennial vegetation and in clayey soils, it is advisable to take a more aggressive
approach toward reducing existing vegetation prior to planting. In practical terms, that may
translate into several herbicide treatments during a single year prior to planting, or multiple years
of herbicide and other treatments (disking, planting to row crops such as soybeans) (Figure 4.6c).
Finally, establishing a native plant community that naturally resists invasion, i.e. a resilient
community, is a primary goal of restoration (Suding 2011), which may be especially beneficial in
wetter sites restored from perennial grasslands (Figure 4.6d). Our analyses suggest that the
degree of invasion is low in sown communities that are either high in species richness, or where
Andropogon gerardii is abundant. Increasing the richness of the sown community is easily
achieved by simply increasing the number of species sown (Grman et al. 2013). Increasing the
abundance of A. gerardii, although typically not a challenge (or goal) in prairie restoration can
similarly be achieved by sowing A. gerardii at high rates (Figure S4.2). However, abundance of
A. gerardii, which had stronger and more consistent effects on invasion resistance, is generally
achieved at the expense of diversity in general, including of native species (Dickson and Busby
2009, Carter and Blair 2012). Therefore, managers may need to balance between the priorities of
stronger invasion resistance, via A. gerardii dominance; or establishing diverse sown
communities with weaker invasion resistance. Some managers advocate a third, staged approach,
whereby early dominance of A. gerardii is followed by auxiliary seed-sowing or other
management to establish diverse communities of native prairie species.

98	

	

Acknowledgements
I thank Don Schoolmaster, Jim Grace, and Sandra Marquart-Pyatt for advice on SEM. Nate
Fuller, Jerry Stewart, Chad Hughson, Bill Schneider, and Esther Durnwald offered useful
insights on defining invasive species. Portions of this work was funded by USDA-NIFA grant
2013-67011-21281.

99	

	

APPENDIX

100
	

	
Table S4.1 List of invasive species in dataset, and summary information on invasive and
sown richness and cover. Lists: State/Province lists including or (- not including) species.
Criteria: A=Core list; B= Identified by managers (see main text).

1

Agropyron
repens
0

Alliaria
petiolata
0

Bromus
inermis
0

Centaurea
maculosa
0

Cirsium
arvense
0

2

0

0

0.101

1.901

0

3

0

0

0

0

0

4

0

0

0.041

0.021

0

5

0

0

0

0

0

6

0

0

0.2

0

0

7

0

0

0

0

0

8

0

0

0

0

0

9

0

0

0

0

0

10

0.11

0

0

0

0

11

4.18

0

1.25

0

0

12

0.33

0

0

0

0

13

0

0

0

0

0

14

0

0.301

0

0

0

15

0.001

0

0

0

0

16

13.4

0

2.55

0

0

17

11.1

0

5.6

0

10

18

0

0

0.02

0

0

19

0

0

0

0

0

20

0

0

0

0

0

21

0.001

0

0

0

0

22

0

0

0

0

0

23

0

0

0

0

0

24

0.05

0

0

0

0

25

1.7

0.22

14

0

0.02

26

0

0

0.07

3.9

0

27

2.22

0

0

0

0

28

0

0

0

0

0

29

0

0

0

0

0

C3 clonal grass

Biennial Forb

C3 clonal grass

Biennial Forb

Perennial Forb

High

High

High

High

High

MI,WI,MN,

ALL

ALL (-IN, ON)

ALL (-ON)

ALL (-ON)

A

A

A

A

A

mean

1.141

0.018

0.822

0.201

0.346

max

13.4

0.301

14

3.9

10

Site

Function
I-Rank
Lists
Criteria

101
	

	
Table S4.1. (cont’d)

1

Dactylis
glomerata
0

Hieracium
spp.
0

Hypericum
perforatum
1.1

Lonicera
spp.
0

2

0

0.041

0.251

0

0

3

0

0

0

0

0

4

0

0.011

0

0

0

5

0

0

0

0

0

6

0

0

0

0.02

0

7

0

0

0

0

0

8

0.1

0

0

0

0.01

9

0

0

0.56

0

0

10

0

0.025

0

0

0

11

0

0

0

0

0

12

0

0

0.3

0

0

13

0

0

0

0

0

14

0

0.08

0

0

0

15

0

0

0

0

0

16

0

0.1

0

0

0

17

0

0

0

0

0

18

0

0

0

0

0

19

0

0

0

0

0

20

0

0

0

0

0

21

0

0

0

0

0.14

22

0

0

0

0

0

23

0

29.5

0

0

0

24

0

0

0

0

0

25

7.6

0

0

0

0

26

0

0

0

0

0

27

0

0

0

0

0

28

0

0

0

0

0

29

0

0

0

0

C3 bunch grass

Perennial Forb

Perennial Forb

Shrub/Tree

Medium

Medium

High

High

0
Leguminous
Forb
Medium

NONE

MI, WI, MN

MI, WI, IN, IL

ALL

ALL (-ON)

B

A

A

A

A

mean

0.266

1.026

0.076

0.001

0.005

max

7.6

29.5

1.1

0.02

0.14

Site

Function
I-Rank
Lists
Criteria

102
	

Melilotus spp.
0

	
Table S4.1. (cont’d)

1

Morus
alba
0

Phalaris
arundinacea
0

Phleum
pratense
0

3

Rhamnus
frangula
0

2

0.44

0

0

7.501

0.06

3

0

0

0

0

0

4
5

0

0.082

0.001

0.083

0

0

0

0.16

0.02

0

6

0

0

0

11.9

0

7

0.2

0

0

0

0

8

0

0

0

2.05

0

9

0

0

0

14

0

10

0.4

0

0

38.5

0

11

0

0

0

0

0

12

0

0

0

7.321

0

13

0

0

0

0

0

14

0.3

0

0

0.13

0

15

0.01

0

0

34.5

0

16

0

0

0

53.3

0

17

0

0

0

31.3

0

18

0

0

0

2.2

0

19

0

0

0

2.65

0

20

0

0

0

0.03

0

21

0

0

0

4.7

0

22

0

0

0

0

0

23

0

0

0

0.03

0

24

0

0

0

0.1

0

25

0.04

0

0

6.4

0

26

0

0

0

21.3

0

27

0

0

0

0

0

28

0

0

0

0

0

29

0

0

0

0

Shrub/Tree

C3 clonal grass

C3 bunch grass

Shrub/Tree

High

High

Medium

0
C3 clonal
grass
Medium

WI, IN, IL

ALL (-ON)

MI

MI, IL

ALL

A

A

B

A

A

mean

0.048

0.003

0.006

8.311

0.002

max

0.44

0.082

0.16

53.3

0.06

Site

Function
I-Rank
Lists
Criteria

103
	

Poa pratensis

High

	
Table S4.1. (cont’d)

1

Rosa
multiflora
0

Trifolium
hybridum
0.001

Trifolium
pratense
0

2

0

0

0

0

3

0

0

0

0

4

0

0

0

0.011

5

0

0.012

0

0.002

6

0

0

2.26

0

7

0

0

0

0

8

0.001

0

3.06

0

9

0

0.32

0

0

10

0

0

0.07

0.171

11

0

0.08

0

0.01

12

0

0

0

0

13

0

0

0

0.001

14

0

0

0

0

15

0

0

0.01

0

16

0

0

0.013

0

17

0

0

0.01

0

18

0

0

0

0

19

0

0

0

0

20

0

0

0

0.102

21

0

0

33.5

0.16

22

0

0

0

0

23

0

0

0

0

24

0

0

0

0

25

0

5.92

0

0

26

0

0.06

0

0

27

0

0.02

0

0

28

0

0

0

0

29

0

0

Shrub/Tree

Leguminous Forb

Medium

Medium

0
Leguminous
Forb
Low

0
Leguminous
Forb
Medium

ALL (-ON)

NONE

MI

MI

Criteria

A

B

B

B

mean

0

0.221

1.342

0.016

max

0.001

5.92

33.5

0.171

Site

Function
I-Rank
Lists

104
	

Trifolium repens
0

	
Table S4.1. (cont’d)

1

Invasive
Cover
4.1

Sown
Cover
51.15

Invasive
SR
3

Sown
SR
6

2

10.3

49.33

7

8

3

0

34.46

0

7

4

0.25

41.84

7

3

5

0.19

26.1

4

3

6

14.38

74.92

4

29

7

0.2

79.71

1

17

8

5.22

57.53

5

26

9

14.88

50.78

3

17

10

39.28

45.88

6

15

11

5.52

53.36

4

15

12

7.95

55.44

3

18

13

0.001

59.31

1

17

14

0.81

36.87

4

12

15

34.52

49.97

4

13

16

69.36

13.55

5

7

17

58.01

25.71

5

12

18

2.22

70.3

2

15

19

2.65

48.02

1

9

20

0.13

61.36

2

14

21

38.5

36.18

5

12

22

0

52.5

0

12

23

29.53

35.36

2

5

24

0.15

35.32

2

18

25

35.9

45.5

8

17

26

25.33

59.13

4

15

27

2.24

86.52

2

22

28

0

31.52

0

18

29

0

46.84

0

15

mean

13.85

48.77

3.24

13.69

max

69.363

86.52

8

29

Site

105
	

	
Table S4.2. Goodness-of-fit metrics for structural equation models from main text and
supplementary information. Definition of metrics (with recommended cutoffs in parentheses):
GFI=goodness-of-fit index (≥0.90), NNFI=nonnormed fit index (≥0.95), CFI=comparative fit
index (≥0.95), SRMR=standardized root mean square residual (≤0.08), RMSEA=root mean square
error of approximation (≤0.06), CHISQ=Chi-square (N/A), p=p-value for Chi-square test
(>0.05), df=degrees of freedom for Chi-square test (N/A). SEM=figure associated with model.

nnfi
0.86
0.94
0.91
0.92
0.87

gfi
0.99
0.99
0.98
0.98
0.98

cfi
0.93
0.97
0.95
0.96
0.94

srmr
0.06
0.07
0.06
0.06
0.06

rmsea
0.12
0.09
0.09
0.10
0.10

Exotic richness
Exotic cover

0.99 0.89 0.94
0.99 0.94 0.97

0.06
0.06

0.10
0.10

11.67 0.23
11.45 0.25

9
9

0.38 Fig. S4.3a
0.78 Fig. S4.3b

Row crops
Grasslands

1.00 1.54 1.00
1.00 1.18 1.00

0.04
0.02

0.00
0.00

0.93
0.55

4
3

0.30 Fig. S4.4a
0.56 Fig. S4.4b

106
	

chisq
12.53
11.16
11.15
11.50
11.76

p
df R2 SEM
0.19 9 0.46 Fig. 4.1a
0.27 9 0.78 Fig. 4.1b
0.27 9 0.45 Fig. 4.3a
0.24 9 0.68 Fig. 4.3b
0.23 9 0.32 Fig. 4.3c

Response
Invasive richness
Invasive cover
P. pratensis
Trifolium spp.
Invasive forbs

0.92
0.91

	
Figure S4.1. Meta-model visualizing predictions shown in Table 4.1. * path suggested by
modification indices.

A. Some land cover types (e.g., agriculture) may support smaller invasive species populations
than others (e.g., forest, grassland) and therefore contribute invasive propagules at lower rates
(Gifford and Otfinowski 2013).
B. Differences in historical land management (row crops vs. perennial grasslands) select for
different types and abundances of invasive species in the seedbank and budbank (Zylka et al.
2016).
C. Fire increases light availability but decreases soil nutrient availability, so may increase or
decreasing invasibility. The response of individual species to fire also varies - so the response of
specific invaders may vary (Howe et al. 1994).
D. Sites with lower resource availability (here, primarily soil moisture, a key driver of
community dynamics in grassland) are more difficult to invade (Davis et al. 2000, Melbourne et
al. 2007).
E. Sites dominated by a single highly competitive species may be more resistant to invasion; A.
gerardii may draw down resources more effectively than a diverse community (Smith et al.
2004).
F. The establishment of a diverse community of sown native species confers biotic resistance;
diverse communities of resident species occupy available resources more efficiently (e.g.,
complementarity) (Kennedy et al. 2002, Levine et al. 2004).
G. More frequent fire, particularly spring fires, may increase dominance by A. gerardii (Howe et
al. 2011), which [G > E] may increase the invasion resistance of the sown community..
H. Sowing more species increases the richness of the sown community (Grman et al. 2013),
which [H > F] may increase the invasion resistance of the sown community.
I. Sites with lower resource availability may also limit invasion by sown species (Grman and
Brudvig 2014), which [I > F] may reduce the invasion resistance of the sown community.
107
	

	
Figure S4.2. Abundance of Andropogon gerardii is correlated with A. gerardii seeding rate.

30
20
10
0

Percent Cover of A. gerardii

40

R2 = 0.29, p < 0.01

0.00

	

0.05
0.10
0.15
0.20
2
Grams of A. gerardii sown/m

	

108
	

	
Figure S4.3. Structural equation model showing effects on the richness (a) and percent
cover (b) of ALL exotic species. Management (fire frequency, seed mix richness), propagule
pressure (landscape context, land-use history), resource availability (soil moisture), and sown
community structure (A. gerardii abundance, sown species richness) Compare to Figure 4.2, the
SEM for only the invasive exotic species.

109
	

	
Figure S4.4. The direct and indirect effects of management (fire frequency, seed mix
richness) on invasive cover differ by land-use history. Management explained little variation
(R2 = 0.30) in cover of invasive species in sites restored from row crops (a). Cover of the sown
dominant A. gerardii increased invasive cover, but fire frequency indirectly reduced invasive
cover by reducing the cover of A. gerardii. Management explained more variation (R2 = 0.56) in
sites restored from perennial grasslands (b). Both fire frequency and seed mix richness reduced
invasive cover by increasing sown species richness.
	

	

110
	

	

LITERATURE CITED

111
	

	
LITERATURE CITED

Aho, K., Derryberry, D. & Peterson, T. (2014). Model selection for ecologists: the worldviews of
AIC and BIC. Ecology 95: 631-636.
Allan, E., and M. J. Crawley. (2011). Contrasting effects of insect and molluscan herbivores on
plant diversity in a long-term field experiment. Ecology Letters 14:1246–1253.
Allison, G. (2004). The influence of species diversity and stress intensity on community
resistance and resilience. Ecological Monographs, 74, 117-134.
Alstad, A. O., E. I. Damschen, T. J. Givnish, J. A. Harrington, M. K. Leach, D. A. Rogers, and
D. M. Waller. 2016. The pace of plant community change is accelerating in remnant
prairies. Science advances 2:e1500975-e1500975.
Antoninka, A., J. E. Wolf, M. Bowker, A. T. Classen, and N. C. Johnson. 2009. Linking aboveand belowground responses to global change at community and ecosystem scales. Global
Change Biology 15:914-929.
Arnold, T.W. (2010). Uninformative parameters and model selection using Akaike's Information
Criterion. Journal of Wildlife Management, 74, 1175-1178.
Bartoń, K. 2014. MuMIn: Multi-Model Inference. R package version 1.12.1. http://CRAN.Rproject.org/package=MuMIn.
Bauer, J. T., K. M. L. Mack, and J. D. Bever. 2015. Plant-soil feedbacks as drivers of succession:
evidence from remnant and restored tallgrass prairies. Ecosphere 6:158.
Belote, R. T., R. H. Jones, S. M. Hood, and B. W. Wender. 2008. Diversity-invasibility across an
experimental disturbance gradient in Appalachian forests. Ecology 89:183-192.
Bezemer, T.M. & van der Putten, W.H. 2007. Ecology - Diversity and stability in plant
communities. Nature, 446, E6-E7.
Biondini, M. 2007. Plant diversity, production, stability, and susceptibility to invasion in restored
northern tall grass prairies (United states). Restoration Ecology 15:77-87.
Blair, J.M. 1997. Fire, N availability, and plant response in grasslands: a test of the transient
maxima hypothesis. Ecology 78: 2359-2368.
Bolker, B. and R Core Team 2014. bbmle: Tools for general maximum likelihood estimation. R
package version 1.0.16. http://CRAN.R-project.org/package=bbmle.

112
	

	
Borer, E. T., E. W. Seabloom, D. S. Gruner, W. S. Harpole, H. Hillebrand, E. M. Lind, et al.
2014. Herbivores and nutrients control grassland plant diversity via light limitation.
Nature 508:517-+.
Bowles, M. L., and M. D. Jones. 2013. Repeated burning of eastern tallgrass prairie increases
richness and diversity, stabilizing late successional vegetation. Ecological Applications
23:464-478.
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.
Brown, R. L., and R. K. Peet. 2003. Diversity and invasibility of southern Appalachian plant
communities. Ecology 84:32-39.
Brudvig, L. A. 2011. The restoration of biodiversity – where has research been and where does it
need to go? American Journal of Botany 98:549-558.
Brudvig, L. A., and E. I. Damschen. 2011. Land-use history, historical connectivity, and land
management interact to determine longleaf pine woodland understory richness and
composition. Ecography 34:257-266.
Brudvig, L. A., C. M. Mabry, J. R. Miller, and T. A. Walker. 2007. Evaluation of central north
American prairie management based on species diversity, life form, and individual
species metrics. Conservation Biology 21:864-874.
Brummer, T. J., A. E. Byrom, J. J. Sullivan, and P. E. Hulme. 2016. Alien and native plant
richness and abundance respond to different environmental drivers across multiple gravel
floodplain ecosystems. Diversity and Distributions 22:823-835.
Burke, M. J. W., and J. P. Grime. 1996. An experimental study of plant community invasibility.
Ecology 77:776-790.
Carboni, M., W. Thuiller, F. Izzi, and A. Acosta. 2010. Disentangling the relative effects of
environmental versus human factors on the abundance of native and alien plant species in
Mediterranean sandy shores. Diversity and Distributions 16:537-546.
Cardinale, B. J. 2012. Impacts of biodiversity loss. Science 336: 552.
Carter, D. L., and J. M. Blair. 2012. Recovery of native plant community characteristics on a
chronosequence of restored prairies seeded into pastures in west-central Iowa.
Restoration Ecology 20:170-179.
Catford, J. A., R. Jansson, and C. Nilsson. 2009. Reducing redundancy in invasion ecology by
integrating hypotheses into a single theoretical framework. Diversity and Distributions
15:22-40.

113
	

	
Catford, J. A., P. A. Vesk, D. M. Richardson, and P. Pysek. 2012. Quantifying levels of
biological invasion: towards the objective classification of invaded and invasible
ecosystems. Global Change Biology 18:44-62.
Chytry, M., V. Jarosik, P. Pysek, O. Hajek, I. Knollova, L. Tichy, and J. Danihelka. 2008.
Separating habitat invasibility by alien plants from the actual level of invasion. Ecology
89:1541-1553.
Clark, G. F., and E. L. Johnston. 2011. Temporal change in the diversity-invasibility relationship
in the presence of a disturbance regime. Ecology Letters 14:52-57.
Colautti, R. I., I. A. Grigorovich, and H. J. MacIsaac. 2006. Propagule pressure: A null model for
biological invasions. Biological Invasions 8:1023-1037.
Collins, S. L. 2000. Disturbance frequency and community stability in native tallgrass prairie.
American Naturalist 155:311-325.
Copeland, T. E., W. Sluis, and H. F. Howe. 2002. Fire season and dominance in an Illinois
tallgrass prairie restoration. Restoration Ecology 10:315-323.
Cottingham, K.L., Brown, B.L. & Lennon, J.T. 2001. Biodiversity may regulate the temporal
variability of ecological systems. Ecology Letters 4: 72-85.
Craine, J.M. & Nippert, J.B. 2014. Cessation of burning dries soils long term in a tallgrass
prairie. Ecosystems 17: 54-65.
D'Antonio, C., and L. A. Meyerson. 2002. Exotic plant species as problems and solutions in
ecological restoration: A synthesis. Restoration Ecology 10:703-713.
Davies, K. F., P. Chesson, S. Harrison, B. D. Inouye, B. A. Melbourne, and K. J. Rice. 2005.
Spatial heterogeneity explains the scale dependence of the native-exotic diversity
relationship. Ecology 86:1602-1610.
Davis, M. A., J. P. Grime, and K. Thompson. 2000. Fluctuating resources in plant communities:
a general theory of invasibility. Journal of Ecology 88:528-534.
Diaz, S., and M. Cabido. 2001. Vive la difference: plant functional diversity matters to
ecosystem processes. Trends in Ecology & Evolution 16:646-655.
Dickson, T. L., and W. H. Busby. 2009. Forb Species Establishment Increases with Decreased
Grass Seeding Density and with Increased Forb Seeding Density in a Northeast Kansas,
USA, Experimental Prairie Restoration. Restoration Ecology 17:597-605.
Didham, R. K., J. M. Tylianakis, N. J. Gemmell, T. A. Rand, and R. M. Ewers. 2007. Interactive
effects of habitat modification and species invasion on native species decline. Trends in
Ecology & Evolution 22:489-496.

114
	

	
Dimitrakopoulos, P.G., Siamantziouras, A.S.D., Galanidis, A., Mprezetou, I. & Troumbis, A.Y.
2006. The interactive effects of fire and diversity on short-term responses of ecosystem
processes in experimental Mediterranean grasslands. Environmental Management
37:826-839.
Doak, D. F., D. Bigger, E. K. Harding, M. A. Marvier, R. E. O'Malley, and D. Thomson. 1998.
The statistical inevitability of stability-diversity relationships in community ecology.
American Naturalist 151:264-276.
Elton, C. 1958. The ecology of invasions by animals and plants. The ecology of invasions by
animals and plants.:181 pp.-181 pp.
Emery, S. M., and K. L. Gross. 2005. Effects of timing of prescribed fire on the demography of
an invasive plant, spotted knapweed Centaurea maculosa. Journal of Applied Ecology
42:60-69.
Fargione, J. E., and D. Tilman. 2005. Diversity decreases invasion via both sampling and
complementarity effects. Ecology Letters 8:604-611.
Foley, J. A., R. DeFries, G. P. Asner, C. Barford, G. Bonan, S. R. Carpenter, et al.. 2005. Global
consequences of land use. Science 309:570-574.
Foster, B. L., T. L. Dickson, C. A. Murphy, I. S. Karel, and V. H. Smith. 2004. Propagule pools
mediate community assembly and diversity-ecosystem regulation along a grassland
productivity gradient. Journal of Ecology 92:435-449.
Foster, B. L., G. R. Houseman, D. R. Hall, and S. E. Hinman. 2015. Does tallgrass prairie
restoration enhance the invasion resistance of post-agricultural lands? Biological
Invasions 17:3579-3590.
Foster, B. L., C. A. Murphy, K. R. Keller, T. A. Aschenbach, E. J. Questad, and K. Kindscher.
2007. Restoration of prairie community structure and ecosystem function in an
abandoned hayfield: A sowing experiment. Restoration Ecology 15:652-661.
Foster, B. L., V. H. Smith, T. L. Dickson, and T. Hildebrand. 2002. Invasibility and
compositional stability in a grassland community: relationships to diversity and extrinsic
factors. Oikos 99:300-307.
Fridley, J. D., R. L. Brown, and J. E. Bruno. 2004. Null models of exotic invasion and scaledependent patterns of native and exotic species richness. Ecology 85:3215-3222.
Fridley, J. D., J. J. Stachowicz, S. Naeem, D. F. Sax, E. W. Seabloom, M. D. Smith, T. J.
Stohlgren, D. Tilman, and B. Von Holle. 2007. The invasion paradox: Reconciling
pattern and process in species invasions. Ecology 88:3-17.

115
	

	
Funk, J. L., E. E. Cleland, K. N. Suding, and E. S. Zavaleta. 2008. Restoration through
reassembly: plant traits and invasion resistance. Trends in Ecology & Evolution 23:695703.
Gifford, M., and R. Otfinowski. 2013. Landscape Disturbances Impact Affect the Distribution
and Abundance of Exotic Grasses in Northern Fescue Prairies. Invasive Plant Science
and Management 6:577-584.
Gioria, M., and B. A. Osborne. 2014. Resource competition in plant invasions: emerging patterns
and research needs. Frontiers in Plant Science 5.
Gelman, A. 2008. Scaling regression inputs by dividing by two standard deviations. Statistics in
Medicine, 27, 2865-2873.
Griffin, J.n., O'Gorman, E.J., Emmerson, M.C., Jenkins, S.R., Klein, A.-M., Loreau, M. et al.
2009. Biodiversity and the stability of ecosystem functioning.
Grueber, C.E., Nakagawa, S., Laws, R.J. & Jamieson, I.G. 2011. Multimodel inference in
ecology and evolution: challenges and solutions. Journal of Evolutionary Biology
24:699-711.
Gotelli, N. J. 1991. Metapopulation models – the rescue effect, the propagule rain, and the coresatellite hypothesis. American Naturalist 138:768-776.
Grace, J. B. 2006. Structural equation modeling and natural systems. Structural equation
modeling and natural systems.:i-xii, 1-365.
Grace, J. B., T. M. Anderson, H. Olff, and S. M. Scheiner. 2010. On the specification of
structural equation models for ecological systems. Ecological Monographs 80:67-87.
Grman, E., T. Bassett, and L. A. Brudvig. 2013. Confronting contingency in restoration:
management and site history determine outcomes of assembling prairies, but site
characteristics and landscape context have little effect. Journal of Applied Ecology
50:1234-1243.
Grman, E., T. Bassett, and L. A. Brudvig. 2014. A prairie plant community data set for
addressing questions in community assembly and restoration. Ecology 95:2363-2363.
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.
Grman, E., and L. A. Brudvig. 2014. Beta diversity among prairie restorations increases with
species pool size, but not through enhanced species sorting. Journal of Ecology
102:1017-1024.

116
	

	
Grman, E., J. A. Lau, D. R. Schoolmaster, and K. L. Gross. 2010. Mechanisms contributing to
stability in ecosystem function depend on the environmental context. Ecology Letters
13:1400-1410.
Gross, K., B. J. Cardinale, J. W. Fox, A. Gonzalez, M. Loreau, H. W. Polley, P. B. Reich, and J.
van Ruijven. 2014. Species Richness and the Temporal Stability of Biomass Production:
A new analysis of recent biodiversity experiments. American Naturalist 183:1-12.
Guo, Q. 2015. No consistent small-scale native-exotic relationships. Plant Ecology 216:12251230.
Guo, Q., S. Fei, J. S. Dukes, C. M. Oswalt, B. V. Iannone, and K. M. Potter. 2015. A unified
approach for quantifying invasibility and degree of invasion. Ecology 96: 2613–2621.
Gurevitch, J., G. A. Fox, G. M. Wardle, Inderjit, and D. Taub. 2011. Emergent insights from the
synthesis of conceptual frameworks for biological invasions. Ecology Letters 14:407418.
Hallett, L. M., J. S. Hsu, E. E. Cleland, S. L. Collins, T. L. Dickson, E. C. Farrer, et al. 2014.
Biotic mechanisms of community stability shift along a precipitation gradient. Ecology
95:1693-1700.
Hautier, Y., Seabloom, E.W., Borer, E.T., Adler, P.B., Harpole, W.S., Hillebrand, H. et al. 2014.
Eutrophication weakens stabilizing effects of diversity in natural grasslands. Nature
508:521-+.
Hector, A., B. Schmid, C. Beierkuhnlein, M. C. Caldeira, M. Diemer, P. G. Dimitrakopoulos, et
al. 1999. Plant diversity and productivity experiments in European grasslands. Science
286:1123-1127.
Herben, T., B. Mandak, K. Bimova, and Z. Munzbergova. 2004. Invasibility and species richness
of a community: A neutral model and a survey of published data. Ecology 85:3223-3233.
Heslinga, J. L., and R. E. Grese. 2010. Assessing Plant Community Changes Over Sixteen Years
of Restoration in a Remnant Michigan Tallgrass Prairie. American Midland Naturalist
164:322-336.
Hill, K. C., and D. G. Fischer. 2014. Native-Exotic Species Richness Relationships Across
Spatial Scales in a Prairie Restoration Matrix. Restoration Ecology 22:204-213.
Hillebrand, H., D. M. Bennett, M. W. Cadotte. 2008. Consequences of dominance – a review of
evenness effects on local and regional ecosystem processes. Ecology, 89:1510–1520.
Hobbs, R. J., E. Higgs, and J. A. Harris. 2009. Novel ecosystems: implications for conservation
and restoration. Trends in Ecology & Evolution 24:599-605.

117
	

	
Hobbs, R. J., and L. F. Huenneke. 1992. Disturbance, diversity, and invasion – implications for
conservation. Conservation Biology 6:324-337.
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.
Hooper, D.U., Chapin, F.S., Ewel, J.J., Hector, A., Inchausti, P., Lavorel, S. et al. 2005. Effects
of biodiversity on ecosystem functioning: a consensus of current knowledge. Ecological
Monographs 75:3-35.
Hooper, D. U., E. C. Adair, B. J. Cardinale, J. E. K. Byrnes, B. A. Hungate, K. L. Matulich,
A. Gonzalez, J. E. Duffy, L. Gamfeldt and M. I. O’Connor. 2012. A global synthesis
reveals biodiversity loss as a major driver of ecosystem change. Nature 486: 105–108
Hoover, D.L., Knapp, A.K. & Smith, M.D. 2014. Resistance and resilience of a grassland
ecosystem to climate extremes. Ecology 95:2646-2656.
Howe, H. F. 1994. Managing species diversity in tallgrass prairie – assumptions and
implications. Conservation Biology 8:691-704.
Howe, H. F. 2011. Fire season and prairie forb richness in a 21-y experiment. Ecoscience
18:317-328.
Hufbauer, R. A., A. Rutschmann, B. Serrate, H. V. de Conchard, and B. Facon. 2013. Role of
propagule pressure in colonization success: disentangling the relative importance of
demographic, genetic and habitat effects. Journal of Evolutionary Biology 26:1691-1699.
Isbell, F., D.Craven. J. Connolly, M. Loreau, B. Schmid, C. Beierkuhnlein, et al. Biodiversity
increases the resistance of ecosystem productivity to climate extremes. Nature 526:574577+.
Ives, A. R., S. R. Carpenter. 2007. Stability and diversity of ecosystems. Science 317:58-62.
Jauni, M., and T. Hyvonen. 2012. Positive diversity-invasibility relationships across multiple
scales in Finnish agricultural habitats. Biological Invasions 14:1379-1391.
Kennedy, T. A., S. Naeem, K. M. Howe, J. M. H. Knops, D. Tilman, and P. Reich. 2002.
Biodiversity as a barrier to ecological invasion. Nature 417:636-638.
Kettenring, K. M., and C. R. Adams. 2011. Lessons learned from invasive plant control
experiments: a systematic review and meta-analysis. Journal of Applied Ecology 48:970979.
Knapp, A.K. & Seastedt, T.R. 1986. Detritus accumulation limits productivity of tallgrass
prairie. Bioscience, 36, 662-668.

118
	

	
Koerner, S.E. & Collins, S.L. 2014. Interactive effects of grazing, drought, and fire on grassland
plant communities in North America and South Africa. Ecology 95:98-109.
Leach, M. K., and T. J. Givnish. 1996. Ecological determinants of species loss in remnant
prairies. Science 273:1555-1558.
Lehman, C. L., and D. Tilman. 2000. Biodiversity, stability, and productivity in competitive
communities. American Naturalist 156:534-552.
Leps, J., J. Osbornovakosinova, and M. Rejmanek. 1982. Community stability, complexity and
species life history strategies. Vegetatio 50:53-63.
Leps, J. 2004. Variability in population and community biomass in a grassland community
affected by environmental productivity and diversity. Oikos 107:64-71.
Levine, J. M., P. B. Adler, and S. G. Yelenik. 2004. A meta-analysis of biotic resistance to exotic
plant invasions. Ecology Letters 7:975-989.
Li, W., X. Zuo, and J. M. H. Knops. 2013. Different Fire Frequency Impacts Over 27 Years on
Vegetation Succession in an Infertile Old-Field Grassland. Rangeland Ecology &
Management 66:267-273.
Lonsdale, W. M. 1999. Global patterns of plant invasions and the concept of invasibility.
Ecology 80:1522-1536.
Loreau, M., and N. Behera. 1999. Phenotypic diversity and stability of ecosystem processes.
Theoretical Population Biology 56:29-47.
Macarthur, R. 1955. Fluctuations of animal populations, and a measure of community stability.
Ecology 36:533-536.
Mahaney, W. M., K. L. Gross, C. B. Blackwood, and K. A. Smemo. 2015. Impacts of prairie
grass species restoration on plant community invasibility and soil processes in abandoned
agricultural fields. Applied Vegetation Science 18:99-109.
Matthews, J. W., and G. Spyreas. 2010. Convergence and divergence in plant community
trajectories as a framework for monitoring wetland restoration progress. Journal of
Applied Ecology 47:1128-1136.
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.
McGill, B.J., Etienne, R.S., Gray, J.S., Alonso, D., Anderson, M.J., Benecha, H.K. et al. 2007.
Species abundance distributions: moving beyond single prediction theories to integration
within an ecological framework. Ecology Letters 10:995-1015.

119
	

	
Meiners, S. J., M. L. Cadenasso, and S. T. A. Pickett. 2004. Beyond biodiversity: individualistic
controls of invasion in a self-assembled community. Ecology Letters 7:121-126.
Meiners, S. J., S. T. A. Pickett, and M. L. Cadenasso. 2002. Exotic plant invasions over 40 years
of old field successions: community patterns and associations. Ecography 25:215-223.
Melbourne, B. A., H. V. Cornell, K. F. Davies, C. J. Dugaw, S. Elmendorf, A. L. Freestone, et al.
2007. Invasion in a heterogeneous world: resistance, coexistence or hostile takeover?
Ecology Letters 10:77-94.
Middleton, E. L., J. D. Bever, and P. A. Schultz. 2010. The Effect of Restoration Methods on the
Quality of the Restoration and Resistance to Invasion by Exotics. Restoration Ecology
18:181-187.
Millennium Ecosystem Assessment. 2005. Ecosystems and human well-being. Washington, D.C:
Island Press. 	
Molina-Montenegro, M. A., E. E. Cleland, S. M. Watts, and B. R. Broitman. 2012. Can a
breakdown in competition-colonization tradeoffs help explain the success of exotic
species in the California flora? Oikos 121:389-395.
Myers, J. A., and K. E. Harms. 2011. Seed arrival and ecological filters interact to assemble
high-diversity plant communities. Ecology 92:676-686.
National Drought Mitigation Center (NDMC), the U.S. Department of Agriculture (USDA) and
the National Oceanic and Atmospheric Association (NOAA). (2016). U.S. drought
monitor map archive. Available at:
http://droughtmonitor.unl.edu/MapsAndData/MapArchive.aspx. Last accessed 01 JULY
2016.
National Oceanic and Atmospheric Association 2016a. AHPS Precipitation Analysis. Available
at: http://water.weather.gov/precip/. Last accessed 01 JULY 2016.
National Oceanic and Atmospheric Association 2016b. National weather service climate.
Available at: w2.weather.gov/climate/index.php?wfo=grr. Last accessed 01 JULY 2016.
Packard, S. and C.F. Mutel (eds.). 1997. The tallgrass restoration handbook for prairies,
savannas, and woodlands. Island Press, Washington, D.C. xxxii+ 463 pp.
Parker, I. M., D. Simberloff, W. M. Lonsdale, K. Goodell, M. Wonham, P. M. Kareiva, M. H.
Williamson, B. Von Holle, P. B. Moyle, J. E. Byers, and L. Goldwasser. 1999. Impact:
toward a framework for understanding the ecological effects of invaders. Biological
Invasions 1:3-19.

120
	

	
Parker, J. D., L. J. Richie, E. M. Lind, and K. O. Maloney. 2010. Land use history alters the
relationship between native and exotic plants: the rich don't always get richer. Biological
Invasions 12:1557-1571.
Petchey, O.L., Casey, T., Jiang, L., McPhearson, P.T. & Price, J. 2002. Species richness,
environmental fluctuations, and temporal change in total community biomass. Oikos
99:231-240.
Pfisterer, A.B., Joshi, J., Schmid, B. & Fischer, M. 2004. Rapid decay of diversity-productivity
relationships after invasion of experimental plant communities. Basic and Applied
Ecology, 5, 5-14.
Pfisterer, A.B. & Schmid, B. 2002. Diversity-dependent production can decrease the stability of
ecosystem functioning. Nature 416:84-86.
Pimm, S. L. 1984. The complexity and stability of ecosystems. Nature 307:321-326.
Piper, J. K., E. S. Schmidt, and A. J. Janzen. 2007. Effects of species richness on resident and
target species components in a prairie restoration. Restoration Ecology 15:189-198.
Polley, H. W., B. J. Wilsey, and J. D. Derner. 2007. Dominant species constrain effects of
species diversity on temporal variability in biomass production of tallgrass prairie. Oikos
116: 2044=2052.
Polley, H.W., Isbell, F.I. & Wilsey, B.J. 2013. Plant functional traits improve diversity-based
predictions of temporal stability of grassland productivity. Oikos 122:1275-1282.
Preston, F.W. 1948. The commonness, and rarity, of species. Ecology 29:254-283.
R Core Team. 2014. R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria. URL http://www.R-project.org/.
Richardson, D. M., and P. Pysek. 2006. Plant invasions: merging the concepts of species
invasiveness and community invasibility. Progress in Physical Geography 30:409-431.
Ricklefs, R. 2004. A comprehensive framework for global patterns in biodiversity.	Ecology
Letters, 7: 1-15.
Rockstrom, J., W. Steffen, K. Noone, A. Persson, F. S. Chapin, III, E. F. Lambin, et al. 2009. A
safe operating space for humanity. Nature 461:472-475.
Romanuk, T. N., R. J. Vogt, and J. Kolasa. 2009. Ecological realism and mechanisms by which
diversity begets stability. Oikos 118:819-828.

121
	

	
Roscher, C., Fergus, A.J.F., Petermann, J.S., Buchmann, N., Schmid, B. & Schulze, E.-D. 2013.
What happens to the sown species if a biodiversity experiment is not weeded? Basic and
Applied Ecology 14:187-198.
Roscher, C., A. Weigelt, R. Proulx, E. Marquard, J. Schumacher, W. W. Weisser, and B.
Schmid. 2011. Identifying population- and community-level mechanisms of diversitystability relationships in experimental grasslands. Journal of Ecology 99:1460-1469.
Rowe, H. I. 2010. Tricks of the Trade: Techniques and Opinions from 38 Experts in Tallgrass
Prairie Restoration. Restoration Ecology 18:253-262.
Ruiz-Jaen, M. C., and T. M. Aide. 2005. Restoration success: How is it being measured?
Restoration Ecology 13:569-577.
Samson, F., and F. Knopf. 1994. PRAIRIE CONSERVATION IN NORTH-AMERICA.
Bioscience 44:418-421.
Schielzeth, H. 2010. Simple means to improve the interpretability of regression coefficients.
Methods in Ecology and Evolution 1:103-113.
Seabloom, E. W., E. T. Borer, Y. Buckley, E. E. Cleland, K. Davies, J. Firn, et al. 2013.
Predicting invasion in grassland ecosystems: is exotic dominance the real embarrassment
of richness? Global Change Biology 19:3677-3687.
Seabloom, E. W., E. T. Borer, Y. M. Buckley, E. E. Cleland, K. F. Davies, J. Firn, et al. 2015.
Plant species' origin predicts dominance and response to nutrient enrichment and
herbivores in global grasslands. Nature Communications 6.
Shea, K., and P. Chesson. 2002. Community ecology theory as a framework for biological
invasions. Trends in Ecology & Evolution 17:170-176.
Silletti, A. M., A. K. Knapp, and J. M. Blair. 2004. Competition and coexistence in grassland
codominants: responses to neighbour removal and resource availability. Canadian Journal
of Botany-Revue Canadienne De Botanique 82:450-460.
Simberloff, D. 2009. The Role of Propagule Pressure in Biological Invasions. Annual Review of
Ecology Evolution and Systematics 40:81-102.
Sluis, W. J. 2002. Patterns of species richness and composition in re-created grassland.
Restoration Ecology 10:677-684.
Smith, M. D., J. C. Wilcox, T. Kelly, and A. K. Knapp. 2004. Dominance not richness
determines invasibility of tallgrass prairie. Oikos 106:253-262.

122
	

	
Smith, N. G.; Schuster, M. J.; Dukes, J. S. 2016. Rainfall variability and nitrogen addition
synergistically reduce plant diversity in a restored tallgrass prairie.
Journal of Applied Ecology 53:579-586.
Spehn, E. M., A. Hector, J. Joshi, M. Scherer-Lorenzen, B. Schmid, E. Bazeley-White, et al.
2005. Ecosystem effects of biodiversity manipulations in European grasslands.
Ecological Monographs 75:37-63.
Stephens, P.A., Buskirk, S.W., Hayward, G.D. & Del Rio, C.M. 2005. Information theory and
hypothesis testing: a call for pluralism. Journal of Applied Ecology 42:4-12.
Stohlgren, T. J., D. Barnett, C. Flather, P. Fuller, B. Peterjohn, J. Kartesz, and L. L. Master.
2006. Species richness and patterns of invasion in plants, birds, and fishes in the United
States. Biological Invasions 8:427-447.
Stohlgren, T. J., D. Binkley, G. W. Chong, M. A. Kalkhan, L. D. Schell, K. A. Bull, Y. Otsuki,
G. Newman, M. Bashkin, and Y. Son. 1999. Exotic plant species invade hot spots of
native plant diversity. Ecological Monographs 69:25-46.
Suding, K., and K. L. Gross. 2006. Modifying native and exotic species richness correlations:
The influence of fire and seed addition. Ecological Applications 16:1319-1326.
Suding, K. N. 2011. Toward an Era of Restoration in Ecology: Successes, Failures, and
Opportunities Ahead. Pages 465-487 in D. J. Futuyma, H. B. Shaffer, and D. Simberloff,
editors. Annual Review of Ecology, Evolution, and Systematics, Vol 42.
Symstad, A. J., F. S. Chapin, D. H. Wall, K. L. Gross, L. F. Huenneke, G. G. Mittelbach, D. P.
C. Peters. D. Tilman. 2003. Perspectives on the relationship between biodiversity and
ecosystem functioning. Bioscience 53: 89-98.
Symstad, A. J. 2000. A test of the effects of functional group richness and composition on
grassland invasibility. Ecology 81:99-109.
Taylor, B. W., and R. E. Irwin. 2004. Linking economic activities to the distribution of exotic
plants. Proceedings of the National Academy of Sciences of the United States of America
101:17725-17730.
Thibaut, L. M., and S. R. Connolly. 2013. Understanding diversity-stability relationships:
towards a unified model of portfolio effects. Ecology Letters 16:140-150.
Tilman, D. 1994. Competition and biodiversity in spatially structured habitats. Ecology 75:2-16.
Tilman, D. 1996. Biodiversity: Population versus ecosystem stability. Ecology 77:350-363.
Tilman, D. 1997. Community invasibility, recruitment limitation, and grassland biodiversity.
Ecology 78:81-92.

123
	

	
Tilman, D. 1999. The ecological consequences of changes in biodiversity: A search for general
principles. Ecology 80:1455-1474.
Tilman, D. and J. A. Downing. 1994. Biodiversity and stability in grasslands. Nature 367: 363365.
Tilman, D., P B. Reich, J. Knops, D. Wedin, T. Mielke, C. Lehman. 2001. Diversity and
productivity in a long-term grassland experiment. Science 294: 843-845
Tilman, D., C. L. Lehman, and C. E. Bristow. 1998. Diversity-stability relationships: Statistical
inevitability or ecological consequence? American Naturalist 151:277-282.
Tilman, D., and D. Wedin. 1991. Plant traits and resource reduction for 5 grasses growing on a
nitrogen gradient. Ecology 72:685-700.
Toledo, D., M. Sanderson, K. Spaeth, J. Hendrickson, and J. Printz. 2014. Extent of Kentucky
Bluegrass and Its Effect on Native Plant Species Diversity and Ecosystem Services in the
Northern Great Plains of the United States. Invasive Plant Science and Management
7:543-552.
Turner, C.L., Blair, J.M., Schartz, R.J. & Neel, J.C. 1997. Soil N and plant responses to fire,
topography, and supplemental N in tallgrass prairie. Ecology 78:1832-1843.
Van Peer, L., Nijs, I., Reheul, D. & De Cauwer, B. 2004. Species richness and susceptibility to
heat and drought extremes in synthesized grassland ecosystems: compositional vs
physiological effects. Functional Ecology 18:769-778.
van Ruijven, J., and F. Berendse. 2010. Diversity enhances community recovery, but not
resistance, after drought. Journal of Ecology 98:81-86.
Vila, M., J. L. Espinar, M. Hejda, P. E. Hulme, V. Jarosik, J. L. Maron, J. Pergl, U. Schaffner, Y.
Sun, and P. Pysek. 2011. Ecological impacts of invasive alien plants: a meta-analysis of
their effects on species, communities and ecosystems. Ecology Letters 14:702-708.
Vila, M., and I. Ibanez. 2011. Plant invasions in the landscape. Landscape Ecology 26:461-472.
Vitousek, P. M., H. A. Mooney, J. Lubchenco, and J. M. Melillo. 1997. Human domination of
Earth's ecosystems. Science 277:494-499.
Vogel, A., Scherer-Lorenzen, M. & Weigelt, A. 2012. Grassland resistance and resilience after
drought depends on management intensity and species richness. Plos One, 7.
Von Holle, B., and D. Simberloff. 2005. Ecological resistance to biological invasion
overwhelmed by propagule pressure. Ecology 86:3212-3218.

124
	

	
Voss, E. G. and A. R. Reznicek. 2012. Field manual of Michigan flora. The University of
Michigan Press, Ann Arbor, MI. 1008 pp.
Weaver, J. E., and T.J. Fitzpatrick. 1932. Ecology and relative importance of the dominants of
tall-grass prairie. Botanical Gazette 93:113-150.
Werling, B. P., T. L. Dickson, R. Isaacs, H. Gaines, C. Gratton, K. L. Gross, et al. 2014.
Perennial grasslands enhance biodiversity and multiple ecosystem services in bioenergy
landscapes. Proceedings of the National Academy of Sciences of the United States of
America 111:1652-1657.
Whittaker, R. H. 1965. Dominance and diversity in land plant communities: numerical relations
of species express importance of competition in community function and evolution.
Science, 147, 250-&.
Wilsey, B. J., P. P. Daneshgar, K. Hofmockel, and H. W. Polley. 2014. Invaded grassland
communities have altered stability-maintenance mechanisms but equal stability compared
to native communities. Ecology Letters 17:92-100.
With, K. A. 2002. The landscape ecology of invasive spread. Conservation Biology 16:11921203.
Xu, Z. W., H. Y. Ren, J. Cai, R. Wang, M. Li, S. Wan, X. Han, B. J. Lewis, Y. Jiang. 2014.
Effects of experimentally-enhanced precipitation and nitrogen on resistance, recovery and
resilience of a semi-arid grassland after drought. Oecologia 176:1187–1197
Xu, Z. W., H. Y. Ren, M. H. Li, J. van Ruijven, X. G. Han, S. Q. Wan, H. Li, Q. Yu, Y. Jiang,
and L. Jiang. 2015. Environmental changes drive the temporal stability of semi-arid
natural grasslands through altering species asynchrony. Journal of Ecology 103:13081316.
Yachi, S. & Loreau, M. 1999. Biodiversity and ecosystem productivity in a fluctuating
environment: The insurance hypothesis. Proceedings of the National Academy of
Sciences of the United States of America, 96, 1463-1468.
Yang, H. J., L. Jiang, L. H. Li, A. Li, M. Y. Wu, and S. Q. Wan. 2012. Diversity-dependent
stability under mowing and nutrient addition: evidence from a 7-year grassland
experiment. Ecology Letters 15:619-626.
Young, T. P., D. A. Petersen, and J. J. Clary. 2005. The ecology of restoration: historical links,
emerging issues and unexplored realms. Ecology Letters 8:662-673.
Zefferman, E., J. T. Stevens, G. K. Charles, M. Dunbar-Irwin, T. Emam, S. Fick, L. V. Morales,
K. M. Wolf, D. J. N. Young, and T. P. Young. 2015. Plant communities in harsh sites are
less invaded: a summary of observations and proposed explanations. Aob Plants 7.

125
	

	
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
population stability in a temperate grassland. Global Change Biology 22:1445-1455.
Zhang, Q. and D. Zhang. 2006. Species richness destabilizes ecosystem functioning in
experimental
aquatic microcosms. Oikos 112: 218-226.
Zylka, J. J., C. J. Whelan, and B. Molano-Flores. 2016. Restoration implications of land
management legacy on aboveground and seed bank composition of North American
grasslands. American Midland Naturalist 176:36-59.

126