Incorporating Phylogenetic Information to Guide Tallgrass Prairie Restoration

NORTHWESTERN UNIVERSITY

Using Trees to Seed Prairies: Incorporating Phylogenetic Information to Guide Tallgrass Prairie Restoration

A DISSERTATION

SUBMITTED TO THE GRADUATE SCHOOL

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS

for the degree

DOCTOR OF PHILOSOPHY

Field of Plant Biology and Conservation

Rebecca Samantha Barak

EVANSTON, ILLINOIS

June 2017 2

ABSTRACT

Ecological restoration is vital to the conservation of biodiversity and provision of ecosystem services in a changing world. Biodiversity is often a goal of restoration, and species to be planted for restoration are often selected based on diversity objectives. But species are not independent; they are related to one another through the evolutionary tree of life. Species’ positions on the phylogenetic tree can reflect their traits. How broadly species in a community are drawn from across the tree can predict ecosystem function. I studied the role of evolutionary history and diversity in restoration of the tallgrass prairie. Prairies are one of the most endangered ecosystems on earth, and they have been an intensive focus of ecological restoration throughout the history of the field.

In chapter 1, I describe how restoration can be informed by historical ecological information like recorded, archeological, paleoecological, and evolutionary data. These “long view” persectives can provide context for better understanding contemporary ecosystems, and can contribute to goal setting, management, and monitoring for restoration. Phylogenetic diversity specifically can inform restoration because it is a strong predictor of ecosystem functions that are also key restoration objectives, like stability, productivity, support of higher trophic levels, and invasion resistance.

Phylogenetic information can also be useful for understanding how restored sites compare to remnant habitats that serve as reference sites for restoration. In chapter 2, I found that restored prairies have lower phylogenetic diversity than remnant prairies, in addition to differing in species richness and community composition. These differences may occur because restored 3 prairies are subject to higher levels of disturbance than remnants, and because species seeded to establish restored prairies are more closely related than expected by chance. I identified “missing branches” – clades found in remnant prairies, but absent from restoration seed mixes – that could be planted in restorations to increase their compositional and functional equivalency to reference systems.

Increasing biodiversity of restored prairies depends on an understanding of how seeds germinate and establish to build restored plant communities. Seed traits, which are understudied relative to vegetative plant traits, are critical for understanding assembly of restored communities. In chapter 3 I tested the effects of seed traits, phylogenetic position, and germination pre-treatment on germination response in species commonly used in prairie restoration. I found that seed traits, particularly shape variables, predicted germination response. Phylogenetic position was also an important predictor of germination, indicating that the phylogeny may supply information that is integrative over many traits, both measured and unmeasured.

Seeds come together to form seed mixes, the raw materials of prairie restoration. Seed mix design is motivated by objectives related to biodiversity and ecosystem function, but also by economic constraints. In chapter 4 I studied biodiversity of commercially available seed mixes, in terms of species richness, conservatism, and phylogenetic diversity, and compared commercial mixes to restored and remnant prairies. I also tested whether price was predictive of biodiversity in commercial mixes. I found that commercial mixes were generally less diverse than remnant prairies, but similar in diversity to extant restored prairie communities. Lastly, I found that seed mix price was predictive of multiple measures of biodiversity. 4

ACKNOWLEDGEMENTS

Most sincere thanks to my dissertation advisor, Dan Larkin. Your mentorship in all things is helping me become the scientist I aspire to be. Thank you to my doctoral committee, Kay

Havens, Andrew Hipp, Andrea Kramer, and Jack Pizzo, for your insight. I am not saying goodbye, because I look forward to working with all of you in the future.

Thank you to the fantastic research assistants I have had the pleasure of working with during my doctoral studies: Gabriella Carr, Meghan Kramer, Taran Lichtenberger, Jessica Riebkes, Robert

Sherman, and Alyssa Wellman-Houde. Extra special thanks to Taran for measuring literally thousands of seeds. Thanks to Evelyn Williams for being my co-head-counselor for Camp Prairie

Phun. Thanks also to the Larkin lab, the Kramer-Havens lab, and the ladies of community ecology for their constant support and insight.

Thank you to the following people for access to and information about restored prairies and species used in this dissertation: Kyle Banas, Dawn Banks, Trish Burns, Jenny Clauson, Pat

Hayes, Erick Huck, Keith Guimon, Jack Pizzo, Nagulapalli Rao, Cassi Saari, Sue Swithin, Byron

Tsang and Lauren Umek.

Thank you to the Program in Plant Biology and Conservation, the Illinois Association for

Environmental Professionals, the Society for Ecological Restoration, Midwest-Great Lakes

Chapter, and The Dr. John N. Nicholson fellowship from Northwestern University, and National

Science Foundation awards DEB-1354426, DEB-1354551 and DBI-1461007 for financially supporting this research. 5

Thank you to my family for their support throughout graduate school. Yuval, you are the best partner I could ever ask for. Boaz and Oren, you are the absolute joys of my life. To my parents,

Karen and Hanoch Barak, you have always believed that I could do anything. Your massive, constant support (and hours and hours of babysitting) made this work possible. Thank you to my network of family and friends for all the love, and for indulging me when I just want to talk to you about plants.

DEDICATION

This dissertation is dedicated to my grandmother, Shirley Oppenheim, and my grandmother-in- law, Janice Cohn. They sow seeds of wisdom and generosity that will bear fruit for generations to come.

And

To the land stewards and managers that restore and conserve the tallgrass prairie ecosystem.

TABLE OF CONTENTS

1 Title page

2 – 3 Abstract

4 – 5 Acknowledgements

6 Dedication

7 Table of contents

8-9 List of tables and figures

10 – 36 Chapter 1- Taking the long view: Integrating recorded, archeological, paleoecological, and evolutionary data into ecological restoration

37 – 56 Chapter 2 - Restored tallgrass prairies have reduced phylogenetic diversity compared with remnants

57 – 72 Chapter 3 - Cracking the case: seed traits, phylogeny, and seed pre-treatments drive germination in tallgrass prairie species used for ecological restoration

73 – 85 Chapter 4 - Shopping for a prairie: species richness, conservatism, and phylogenetic diversity of commercially available seed mixes for restoration

86 – 91 Dissertation summary and implications for management

92 – 114 Tables and figures

115 – 142 Literature cited

143 – 166 Appendix 1 - Supplementary material from chapter 2

167 – 170 Appendix 2 – Supplementary material from chapter 3

LIST OF TABLES AND FIGURES

Chapter 1

92 – 93 Table 1.1 - Restoration questions that could be addressed by data from different temporal scales

94 Figure 1.1 - Historical timescales and their potential contributions to ecological restoration

95 Figure 1.2 - Vegetation changes at 100-yr intervals over the past 1200 years in Wisconsin sand plain

96 Figure 1.3 - Conceptual framework for the placement of traits according to their ecological and evolutionary structure

Chapter 2

97 Figure 2.1 - Site and plot diversity metrics for species richness, species richness with non-native species excluded, and standard effect sizes for mean pairwise distance (MPD), and mean nearest taxon distance (MNTD)

98 Figure 2.2 - Non-metric dimensional scaling ordinations of remnant and restored prairies and seed mixes, based on community taxonomic and phylogenetic ordinations: mean pairwise distance (MPD), and mean nearest taxon distance (MNTD)

99 Figure 2.3 - Taxonomic and phylogenetic diversity of seed mixes and resultant prairie communities for only species planted at each site, for species richness, standard effect sizes of mean pairwise distance, and mean nearest taxon distance

100 Figure 2.4 - Planted (green) and unplanted (grey) species are indicated on the phylogeny containing all species found at remnant and restored sites and seed mixes (n = 589)

Chapter 3

101 – 102 Table 3.1 - Species included in the experiment, and final percent germination under three pre-treatments: cold stratification, gibberellic acid and an untreated control 9

103 Table 3.2 - Best models of time-to-germination ranked by Akaike information criterion (AIC) for 32 prairie species

104 Table 3.3 - Model-averaged estimate, standard error, and 95% confidence interval (CRI) for all parameters in best fitting models (∆AIC ≤ 4) for 32 prairie species.

105 Table 3.4 - Phylogenetic signal of measured traits

106 Figure 3.1 - Ordination of phylogenetic distance matrix for 32 species in the study

107 Figure 3.2 - Time-to-event curves for all species by germination treatment

108 – 109 Figure 3.3 - Phylogenetic tree of species used in the experiment and phylogenetic distribution of trait values representing seed size (mass), shape (width) and embryo traits (ESlength)

110 Figure 3.4 - Estimates from averaged models (table 3.2), for treatment (blue), trait (purple) and phylogenetic (green) model terms.

Chapter 4

111 Table 4.1 - Companies included in this study that sell seed mixes for prairie restoration

112 Table 4.2 - Price is predictive of some biodiversity components in commercially available seed mixes for prairie restoration

113 Figure 4.1 - Biodiversity of commercially available seed mixes and remnant and restored prairies with respect to species richness, native species richness (i.e., non-native species excluded from remnant and restored prairies), mean C, and MPD

114 Figure 4.2 - Price significantly predicted biodiversity with respect to species richness, Shannon Wiener diversity, Mean C (both presence-absence and abundance-weighted), and abundance-weighted MPD, but not presence-absence MPD.

CHAPTER ONE

Taking the long view: Integrating recorded, archeological, paleoecological, and evolutionary data into ecological restoration

Chapter one has been published in International Journal of Plant Science (IJPS) with the same title (Barak et al. 2016), with coauthors A.L. Hipp, J. Cavender-Bares, W.D. Pearse, S.C.

Hotchkiss, E.A. Lynch, J.C. Callaway, R. Calcote, and D.J. Larkin. I conducted the literature search, and organized and wrote the manuscript. In addition, I created Table 1, and Figure 1, while Figure 2 was submitted by Calcote, and Figure 3 was submitted by Pearse. I received input from all couathors on drafts of the manuscript, prioir to and following submission to IJPS as an invited persectives paper. The manuscript was based on a symposium at the Society for

Ecological Restoration’s World Conference in Madison, Wisconsin, in October 2013. I was a speaker in the symposium, along with Pearse, Hotchkiss, and Callaway. Hipp and Larkin were co-organizers of the session, Cavender-Bares was a coauthor of Pearse’s presentation, and Lynch and Calcote were coauthors of Hotchkiss’ presentation. Therefore, all were included as coauthors of the manuscript. This manuscript is included as chapter one with permision of all coauthors and the journal publisher.

Abstract

Historical information spanning different temporal scales (from tens to millions of years) can inﬂuence restoration practice by providing ecological context for better understanding of contemporary ecosystems. Ecological history provides clues about the assembly, structure, and dynamic nature of ecosystems, and this information can improve forecasting of how restored 11 systems will respond to changes in climate, disturbance regimes, and other factors. History recorded by humans can be used to generate baselines for assessing changes in ecosystems, communities, and populations over time. Paleoecology pushes these baselines back hundreds, thousands, or even millions of years, offering insights into how past species assemblages have responded to changing disturbance regimes and climate. Furthermore, archeology can be used to reconstruct interactions between humans and their environment for which no documentary records exist. Going back further, phylogenies reveal patterns that emerged from coupled evolutionary-ecological processes over very long timescales. Increasingly, this information can be used to predict the stability, resilience, and functioning of assemblages into the future. We review examples in which recorded, archeological, paleoecological, and evolutionary information has been or could be used to inform goal setting, management, and monitoring for restoration. While we argue that long view historical ecology has much to offer restoration, there are few examples of restoration projects explicitly incorporating such information or of research that has evaluated the utility of such perspectives in applied management contexts. For these ideas to move from theory into practice, tests performed through research-management partnerships are needed to determine to what degree taking the long view can support achievement of restoration objectives.

Keywords: climate change, ecosystem function, phylogeny, resilience.

Introduction

Ecological restoration is a discipline that is both past and future oriented. Restoration practitioners aim to mitigate past environmental degradation while creating ecosystems that will 12 be stable, resilient, and self-sustaining in the future (Clewell et al. 2004). Even projects that focus on restoring ecosystem function—rather than closely replicating historical communities— can beneﬁt from ecological history that improves understanding of systems’ functioning (Millar and Brubaker 2006; Higgs et al. 2014). Furthermore, the utility of history for informing contemporary restoration ecology does not end only centuries ago (e.g., pre-European settlement ecosystems as a New World restoration target) or even with the last glaciation before the

Holocene (Egan and Howell 2005). Perspectives that take an even longer view, delving thousands or even millions of years into the past, can be useful in guiding contemporary restoration.

Studying how ecosystems, communities, and populations have responded to past disturbances provides the largest source of information on how they will respond to future changes. In this way, understanding the makeup of past ecosystems and their responses to disturbance can reveal potential future trajectories (history as revealing the future, sensu Higgs et al. 2014; see also

Dietl and Flessa 2011; Dietl et al. 2015). Some disturbances impacting modern ecosystems have analogs in the historical ecological record that provide useful direction to contemporary restoration (Swetnam et al. 1999; Millar and Brubaker 2006). Moreover, disturbances that have the potential to incite the largest-magnitude changes—such as phenotypic or distributional changes in response to climate change—unfolded at deeper timescales than the tens to hundreds of years commonly considered in identifying restoration targets (Egan and Howell 2005).

Working backward from data on community turnover at long timescales and from large disturbances, scientists can identify the factors that conferred resilience to past communities. The 13 composition of these communities and the processes that shaped them can potentially help to design more resilient restored communities today.

We propose that long view temporal perspectives have a role in restoration planning and design.

Reference sites are invaluable for deﬁning restoration goals, but there is additional information to be gleaned from recorded history, archeology, the paleoecological record, and evolutionary history. In some cases, historical ecological data can be a useful complement to contemporary data. In other cases, historical ecological data offer previously unconsidered opportunities for restoration or might even guide restoration in directions that would not be considered on the basis of contemporary data alone. Longer term historical ecological data can also provide options for restoration objectives when restoration of habitats to a more recent pre-degradation state is impossible (Cavender-Bares and Cavender 2011; Balaguer et al. 2014). Use of historical data in restoration is not without constraints, such as limits to data availability, the fading historical record, challenges of matching contemporary management activities with targets inﬂuenced by ecological history, and previously unseen changes to ecosystems in the Anthropocene.

Nonetheless, perspectives from the past may help advance the restoration of functional, resilient systems in the near future.

Historical timescales and their relevance to restoration

Historical information from different timescales can inﬂuence all steps of ecological restoration, from developing restoration goals and garnering public support for restoration to informing ongoing site management and prioritizing restoration efforts. In the following sections, we consider ﬁve streams of information: the contemporary landscape, recorded history, archeology, 14 paleoecology, and evolutionary ecology. We describe the types of ecological information that can be garnered from each perspective and how they can inform ecological restoration (Table

1.1; Figure 1.1). We focus on critical steps in the restoration process, particularly developing restoration objectives, implementing management, performing monitoring, and planning for resilience. In all cases, we stress the role of history as a guide rather than as a stable end point

(Higgs et al. 2014). We follow Rick and Lockwood (2013) in using the term “historical ecology” to encompass recorded history, archeology, and paleoecology and also include under that term evolutionary history determined through phylogenies. This view of historical ecology comprises both natural environmental ﬂuctuations and the effects of humans on ecosystems (Rick and

Lockwood 2013).

Both historical ecologists and restoration ecologists note the value of integrating historical ecological data in restoration. Many of the 50 priority research questions in paleoecology are directly or indirectly related to restoration ecology, including issues such as invasive species, novel ecosystems, resilience, and climate change (Seddon et al. 2014). Similarly, restoration ecologists discuss the multiple ways history inﬂuences restoration ecology (Higgs et al. 2014).

Management stakeholders value long-term ecological data, but barriers—such as lack of awareness of historical ecological information and insufﬁcient testing of its applied usefulness— have prevented its adoption in restoration (Davies et al. 2014).

Many reviews stress the importance of ecological history in guiding conservation and restoration. Nonetheless, it appears that few restoration studies and projects use historical ecological data in this way. To illustrate this, we used Web of Science (Thomson Reuters 2015) to query titles, abstracts, and keywords for articles in the applied restoration journal Restoration 15

Ecology (from 1998 to 2014) for terms relating to the temporal scales reviewed here. A search for “histor*” returned 216 articles, but “anthropol*” and “phylo*” returned none, “paleo*” four, and “evol*” 29. We thus extend our review to include examples from the historical ecological literature that do not address restoration speciﬁcally but have applied implications that could be used to guide restoration projects. The examples mostly, though not exclusively, deal with plant communities and are disproportionately from North America.

We argue that historical ecology is underutilized as a guide for restoration. Few published studies use historical ecological data to inform restoration, and those that do tend to be relatively recent in scope, for example, tracking tens or hundreds of years into the past (Egan and Howell

2005). However, it may be possible to use knowledge of events that occurred much deeper in time—even millions of years ago—to inﬂuence contemporary restoration practice.

Sources of historical information

Information from a variety of contemporary, recorded historical, archaeological, and paleoecological sources can be useful in determining goals for restoration, managing restored ecosystems, and predicting ecosystem resilience (Figure 1.1). Sources of information about past ecosystems include analysis of contemporary sites, historical and archaeological artifacts left by humans, plant and animal remains, and physical and chemical data (Table 1.1).

Contemporary sites contain clues to their own histories in terms of soil properties, species composition, and other factors. Comparisons among multiple sites show the range of variability over space and can be used to infer factors driving community structure and change. Historical sources add another dimension of information, allowing for comparisons of conditions over 16 multiple time points. Using paleoecological and evolutionary information can reveal the imprint of historical processes on modern ecosystems. Delving into the history of how contemporary sites developed can enhance the utility of these sites for guiding restoration.

Contemporary sites

Contemporary ecosystems contain evidence of the short and long-term historical processes that inﬂuenced them, but signs of these processes are not always clear or easy to disentangle.

Information of use to restoration managers can be gathered from the speciﬁc site(s) to be restored as well as contemporary reference sites (i.e., relatively undisturbed extant sites; White and

Walker 1997; Schaefer and Tillmanns 2015). Data from the site to be restored can be used to identify disturbances or constraints that might interfere with meeting restoration objectives.

Clues in the site to be restored can also reveal past history. For example, decayed stumps have been used to recreate logging history (Marks and Gardescu 2005), which can inﬂuence restoration decisions, for example, focusing on reintroducing species sensitive to logging.

Contemporary reference sites provide information on environmental conditions, community composition, and ecosystem functions at sites similar to those being restored that have not been subjected to the same degradation (White and Walker 1997). Restoration practitioners can also evaluate what management actions may be necessary to maintain sites in the target (reference) state and expect similar management to be necessary at the restored site, even at later stages of the restoration. Further, reference sites can continue to inﬂuence restoration decisions by serving as baselines of comparison for monitoring the development of restored sites. Despite the 17 importance of reference sites and large literature advocating their use, less than half of contemporary restoration projects use reference sites as a guide (Wortley et al. 2013).

A limitation of using contemporary reference sites is that they may be only a small or unrepresentative remnant of the original system because of the effects of habitat loss, fragmentation, and other disturbances (White and Walker 1997). High quality reference sites may no longer exist; those that do are subject to shifting baselines, that is, reference systems themselves may have undergone signiﬁcant changes relative to the past (Pauly 1995; Rick and

Lockwood 2013). Furthermore, dynamics of long-lived species may not be detectable when using only contemporary information (Davies and Bunting 2010). When available, other sources of historical ecological information can be used to place extant reference sites in a broader context.

Recorded history

Recorded history is information directly documented by humans, such as species lists, written descriptions from journals or travelogues, oral histories, drawings, photographs, and maps.

Historical records can provide information on species assemblages and ecosystem structure (e.g., canopy openness). These sources of information provide baselines for assessing community change over time and have frequently been used to establish restoration objectives (Radeloff et al. 1999; Swetnam et al. 1999; Bolliger et al. 2004).

Archaeology

Archaeology can clarify restoration objectives and inform management activities by shedding light on past human impacts to communities. Humans have managed ecosystems for thousands 18 of years through actions such as prescribed burning, agriculture, and harvesting and transport of plant and animal species—even in areas earlier thought to be pristine and free of human impact

(Hayashida 2005; Alagona et al. 2012; Rick and Lockwood 2013). Past human settlement can have legacy impacts on ecosystem properties that last for centuries (Hejcman et al. 2013). Past human activity can also constrain the possible trajectories of ecosystems, even with restoration interventions (Higgs et al. 2014). Therefore, knowledge of past human impacts is important for developing feasible restoration goals (Foster et al. 2003) and separating anthropogenic effects from longer-term natural cycles (Swetnam et al. 1999; Millar and Brubaker 2006).

Paleoecology

Paleoecological data can complement perspectives from recorded history and archeology and extend beyond those sources of information into the distant past, to a time before both human records and humans. Analysis of dendrochronology (tree rings), preserved biological remains, and isotopic/biochemical compounds (Hayashida 2005; Dietl et al. 2015) can reveal past community trajectories, species invasions, extinctions, and community responses to changes in climate and disturbance regimes.

Evolutionary history

A phylogenetically informed approach to restoration would use the evolutionary history of species to shape management of contemporary communities. Greater phylogenetic diversity of plant communities—species being drawn more broadly from across the tree of life—is associated with increased ecosystem function (Srivastava et al. 2012). The link between phylogenetic diversity and ecosystem function is based on niche conservatism, the idea that closely related 19 species share similar functional traits and thus similar ecological niches (Wiens and Graham

2005). In this way, maximizing the evolutionary distance between co-occurring species can increase niche breadth. In some cases, phylogenetic diversity is even more strongly related to ecosystem function than functional diversity, because it accounts for ecologically important but unmeasured latent traits not captured by measured traits (Cadotte et al. 2009; Pearse and Hipp

2009; Díaz et al. 2013). Because of the close relationship between phylogenetic diversity and ecosystem function, phylogenetics can be a useful tool in restoration, inﬂuencing objectives, management, and monitoring (Hipp et al. 2015).

Limitations

All of these approaches are subject to data limitations, including a fading record through time, poor spatial and/or temporal resolution, limited taxonomic resolution, and inconsistent preservation. Recorded data can be subjective, ambiguous, or inconsistent because of social norms and values dictating what types of information on which species were recorded and preserved (Edmonds 2005; Lucia et al. 2008; Alagona et al. 2012). Records from the Public Land

Survey (PLS) have become a standard, valuable resource for restoration ecologists in the United

States, but even this rich data set represents only a snapshot of past communities (Shea et al.

2014) and is subject to surveyor bias and taxonomic uncertainty (Schulte and Mladenoff 2005).

Similarly, pollen data are valuable for reconstructing past plant communities but are limited in taxonomic resolution and constrained by variability in pollen output and preservation (Peters

2010). For example, grasses, which are extremely important in many restorations, cannot be identiﬁed beyond family using pollen (Davis 2005). There are also several limitations to the use of phylogenetic data in restoration ecology. The relevance of phylogenetics for restoration has 20 rarely been tested (but see Cavender-Bares and Cavender 2011; Verdú et al. 2012; Whitﬁeld et al. 2014), and concepts and tools of phylogenetic ecology are relatively new and not widely known (Hipp et al. 2015).

Understanding methodological limitations and integrating data from multiple sources can help researchers interpret past conditions (Lucia et al. 2008; Whipple et al. 2011). Of course, historical data are relevant to restoration practitioners only when they are available and accessible (Davies and Bunting 2010; Brewer et al. 2012; Davies et al. 2014; Gillson and

Marchant 2014). Collaborations between restoration scientists/ practitioners and researchers in historical ecology can help address these limitations. For example, paleoecological data are becoming more widely available to nonexperts through the development of databases (Brewer et al. 2012).

Uses of historical information in restoration

Where it is available, historical ecological information is commonly used to establish reference points for restoration; this is particularly true for determining pre-European settlement conditions in North America. For projects where explicit reference sites are lacking, historical information can still be used to select native species and habitat types. More recently, paleoecological and archaeological studies have been used to elucidate a longer-term context for pre-European conditions. These studies have allowed restoration ecologists to consider historic ranges of variation rather than static time points when developing restoration goals and management strategies (Asbjornsen et al. 2005; Keane et al. 2009). Finally, as restoration ecologists 21 increasingly plan for novel climatic conditions and human inﬂuences, they can look to the past to understand factors that impart resilience to communities and ecosystems.

Setting restoration objectives

Historical ecological information can be used to develop support for restoration activities by documenting habitat destruction. In prairies, recorded history tracks the rapid decline in habitat area. Studying maps, Iverson (1988) documented loss of prairie area from 59% of Illinois in

1820 to approximately 0.01% in 1980, primarily as a result of agriculture. Such clear documentation of drastic declines in habitat area can help the public understand the need for restoration and be used to prioritize locations for restoration. Hessburg et al. (1999) used historic maps from the Cascade Mountains in Washington from 1938 to 1956, along with remote sensing, to identify forest patches in need of restoration and create a tool for restoration mangers to prioritize their efforts.

In contrast, historical ecological data can also be used to identify issues that are not of immediate concern for restoration. For example, Watson et al. (2011) used a combination of cultural and paleoecological sources to document habitat changes in the Elkhorn Slough estuary in California.

Aerial photographs and historical maps indicated major declines in marsh area leading to concerns about marsh degradation. However, when the authors analyzed paleoecological data to reconstruct changes in salinity, sedimentation rates, and plant communities over the past 5000 years, they found that some of the recently degraded marshes were products of anthropogenic sedimentation and that there is actually now more marsh area than there had been for most of the past. Given this and the fact that newer marshes could be difﬁcult to sustain, they concluded that 22 it may be wise to direct restoration efforts to other concerns rather than continue to protect marshes in this location.

In addition to prioritizing restoration effort, historical ecological information can also be used to identify constraints to restoration and modify project objectives accordingly. If it is impossible to restore habitats to a recent pre-disturbance state, there may be earlier conditions that can be identiﬁed as suitable alternatives. For example, it would have been impossible to restore Spanish sand quarries to the hills they were before mining, and such landscapes would have been unstable. However, Balaguer et al. (2014) identiﬁed a geomorphological reference of a cultural habitat from 1000 years ago as a feasible target for restoration activity.

Species selection

Restoration managers may be interested in restoring a particular past community, understanding the range of variability of past communities, or identifying stable communities to inform species selection for restoration. General Land Ofﬁce/ PLS records, which date to the late 1700s in the eastern United States, provide spatially broad community data for much of the country (Whipple et al. 2011; Shea et al. 2014). Shea et al. (2014) used PLS records from the mid-1800s to reconstruct tree communities for the Driftless Area of the Midwestern United States. Using data representing more than 100,000 trees, the authors found that oaks were dominant and savanna was the most common habitat type. The authors created a map of past tree communities paired with environmental data, such as topography and soil characteristics, to inform restoration.

Data from multiple time points can be used to show the range of conditions over long time periods and provide a historical context for more recent conditions. For example, pollen analysis 23 of sediments from 13 lakes on a 450-km2 sand plain in northwestern Wisconsin was used to place the vegetation patterns recorded by the PLS during the 1850s and 1860s (Radeloff et al.

1999) into a historical context (Hotchkiss et al. 2007; Tweiten et al. 2015). Maps showing current and reconstructed vegetation communities at 100-year intervals over the past 1200 years

(Figure 2) revealed that the mixed pine and oak communities recorded by the PLS developed relatively recently. White pine pollen became more common and the inﬂux of charcoal from forest ﬁres decreased with the onset of Little Ice Age climatic conditions (Hotchikss et al. 2007;

Tweiten et al. 2015). These results demonstrate the transient nature of plant communities and suggest that conditions indicated in the PLS may not be an ideal restoration target (Hotchkiss et al. 2007), particularly given that Little Ice Age climatic conditions are not representative of current or future climatic conditions in this region. The longer paleoecological record provides an important perspective on the natural range of variability in this landscape that can be used to deﬁne more sustainable restoration goals.

In ecosystems that were rapidly transformed by humans before natural communities could be described, evidence from the paleoecological record can be particularly valuable. For example, on San Cristóbal Island in the Galápagos, cattle destroyed the vegetation in the 1930s, while the

ﬁrst formal descriptions of the vegetation were not completed until the 1960s. Several native plant taxa were not included in restoration efforts because they had largely been extirpated before being documented (Bush et al. 2014). Managers have been removing exotic plant species and planting Miconia robinsiniana, an endemic shrub species. But pollen and sediment records document several other taxa that were abundant during the past 10,000 yr. Bush et al. (2014) 24 suggest that these species should be included in restoration efforts to better reﬂect historic composition and possibly increase the resilience of restored vegetation.

Phylogenetic ecology can also be used to guide species selection for restoration. In ongoing work, we have found initial evidence that restored prairies in northeastern Illinois are less phylogenetically diverse than remnant prairies (W. Sluis, M. L. Bowles, M. D. Jones, and R. S.

Barak, unpublished data). This appears to be driven by higher relative abundance of species from certain families in restored sites and the absence of species representing families that are rare but present in reference sites. Restoration seed mixes could be adjusted accordingly to approximate the phylogenetic diversity of prairie remnants, perhaps helping to increase their functional equivalency with reference sites (sensu Zedler and Callaway 2000).

Niche evolution may also inﬂuence the design of communities for restoration. Species with contrasting alpha niches (local niches at the scale at which species interact with one another) would be appropriate to plant together in a restoration. These species may have reduced competition due to differences in traits (Silvertown et al. 2006; Chesson 2014). For example, shallow and deep-rooting species may be better able to coexist. On the other hand, it is likely to be beneﬁcial for species used in restoration to have similar beta and gamma niches (habitat and geographical-range niches, respectively), since they would overlap in edaphic and/or climatic requirements. For example, wetland species share suites of traits that enable them to survive in waterlogged soils. Cavender-Bares and Cavender (2011) describe oak communities in Florida where closely related oak species did not co-occur at the site (alpha) level, though there were many closely related oak species present within habitats (beta) over the region (gamma). 25

Replicating such patterns when they occur in reference communities may be important in selecting species for restoration.

Monitoring and management

Invasive species

While it is not difﬁcult to observe and document the introduction and spread of recent invasive species, it is not always clear when a species invaded a site or what its impact has been

(Hotchkiss and Juvik 1999; Lynch and Saltonstall 2002; Coffey et al. 2010). Fossil pollen data were used to support the exotic status of Phalaris arundinacea on Vancouver Island, British

Columbia, Canada. Grass pollen was found in high concentrations only in upper (recent) layers of sediment in wetlands, not in the high-diversity native wetlands captured earlier in the record.

It was thus inferred that nonnative P. arundinacea colonized only recently and should be controlled to manage for high plant diversity (Townsend and Hebda 2013). In contrast, paleoecological and genetic analyses from sediments of a Lake Superior wetland provide evidence that native genotypes of Phragmites australis have undergone recent, rapid expansion

(Lynch and Saltonstall 2002).

Looking further back into the paleoecological record in Hawaii, it was found that native palms in the genus Pritchardia disappeared from the record around the time of rat introductions. This led to the recommendation that contemporary practitioners control rats as part of palm restoration efforts (Burney and Burney 2007). In contrast, fossil pollen data were used to demonstrate that several species thought to be invasive—on the basis of modern observations of their adaptation to disturbance (the fern Dicranopteris linearis) or widespread distribution and aggressive 26 behavior (the moss Sphagnum palustre)— were in fact native to Hawaii (Hotchkiss and Juvik

1999; Karlin et al. 2012).

Disturbance regime

In many ecosystems, the structure and composition of vegetation as well as nutrient cycling are inﬂuenced by the frequency, intensity, and magnitude of ﬁres, wind storms, extreme droughts,

floods, or insect outbreaks. To be effective, restoration projects may need to replicate or mimic natural disturbance regimes (McLauchlan et al. 2014). Where there are long-lived trees, fire-scar records have provided valuable information about the nature of fire regimes and their variation over time (Heinselman 1973; Swetnam et al. 1999; Guyette et al. 2006). Such information is relevant to restoration managers, since reinstating a historical fire regime could aid recovery of biological diversity and ecosystem function (Bergeron et al 2004).

Where dendroecological records are lacking, charcoal preserved in lake sediments can provide information about past ﬁre regimes (Gavin et al. 2007; Higuera et al. 2009; Lynch et al. 2010).

While these records typically lack the ﬁne temporal and spatial resolution of tree-ring records, they offer the potential to examine how disturbance regimes were affected by major changes in climate and/or vegetation that occurred beyond the range of tree ring records. This long-term perspective is useful in predicting how resilient communities will be to future climatic changes

(Lynch et al. 2014).

In the Garry Oak savanna of Vancouver Island, British Columbia, Canada, anthropological and paleoecological evidence (dendrochronological, fossil pollen, and charcoal data) was used to determine that the savanna’s open structure was maintained by burning by indigenous peoples 27

(McCune et al. 2013). The savanna persisted even in climatic periods that would favor closed woodlands (McCune et al. 2013). This ﬁnding provides an interesting decision point for modern restoration: should savanna structure be maintained as a cultural landscape, or should forest closure be allowed? The decision on whether to maintain past anthropogenic levels of disturbance may depend on the resources needed to maintain them and the values placed on these cultural landscapes by stakeholders (Motzkin and Foster 2002; Dunwiddie 2005).

Grazing is another source of past and contemporary disturbance that may be relevant to restoration managers. Campbell et al. (2010) found that arid rangeland grazed earlier in the season (i.e., winter-spring grazing) was phylogenetically and functionally more similar to ungrazed sites than those grazed later in the season. Grazing was also found to affect plant biodiversity over a 400-yr time span in Scottish upland sites. Hanley et al. (2008) used fossil pollen data to uncover past plant communities and livestock prices to determine past grazing pressures. Findings such as these can guide the intensity and timing of grazing and other disturbances, when they are compatible with management objectives at restored sites.

Monitoring ecosystem changes

Phylogenetic ecology could be used to monitor ecological change in restored sites and help predict future trajectories. In vulnerable communities, it may be important to assess whether phylogenetic diversity is declining and whether species’ vulnerability to extirpation is phylogenetically autocorrelated. For example, in fragmented tropical forests of Mexico, while species richness declined, phylogenetic diversity did not, indicating low phylogenetic conservatism of traits associated with vulnerability to forest fragmentation (Arroyo-Rodríguez et 28 al. 2012). In this case, ecosystem function and stability may thus be maintained, despite the loss of tree species. In contrast, if vulnerability is a phylogenetically conserved emergent property of species, then certain disturbances could cleave entire branches from communities’ evolutionary trees, likely reducing ecosystem function (Díaz et al. 2013). If continued monitoring reveals a decrease in phylogenetic diversity over time, additional management may be needed to restore phylogenetic diversity.

Further, restoration ecologists may be able to monitor restored sites using the relationship between a community’s evolutionary structure and its trait similarity, linking species’ ecology today with their evolutionary history (Pearse et al. 2015). Figure 3 shows hypothetical relationships between similarity of traits among co-occurring species (horizontal axis, ecologically similar vs. dissimilar) and the mode of evolution of those traits (vertical axis). In this case, traits that are similar in cooccurring species show evidence of convergent evolution

(bottom left), while traits that are dissimilar show evidence of constrained (conserved) evolution

(top left). These relationships (termed ﬁngerprint regressions; Pearse et al. 2015) could be used in restoration planning, for instance, in seed mix design and setting compositional targets. Such relationships could also be used to monitor changes in restored communities over time.

Managers could assess whether the relationship between evolutionary and ecological processes found in functional reference systems is preserved in restored systems. Perhaps by matching these patterns, managers could increase the likelihood of the restored system being functionally equivalent to reference sites and resilient to future changes.

Using the past to restore for future resilience: climate change and ecosystem function

Restorations can be explicitly planned for the future while also being informed by ecological history. Alagona et al. (2012, p. 65) suggested that “the past may be imperfect as a model for the future, but it is an indispensable guide for understanding a world in ﬂux.” Perspectives from historical ecology can inform the restoration of communities and ecosystems that will be resilient and functional in the face of future change.

Climate change

Looking at the past is the best way to predict the effects of climate change on communities and ecosystems (Jackson 2007). Many North American plant species originated 20– 40 million years ago and have thus been exposed to numerous periods of warming and cooling over that time period (Millar and Brubaker 2006). Understanding the trajectories of species and communities over past climate changes can help inform design and implementation of modern restorations.

Paleoecological and evolutionary data—combined with modeling—allow for the reconstruction of past responses to climate change and can help contemporary restorationists plan for the future.

Paleoclimate reconstructions paired with paleoecological data expand the range of conditions that supply perspective to restoration efforts. The paleoecological record contains examples of community stability over thousands of years, despite climate change (Brubaker 1975; Minckley et al. 2011), as well as sometimes dramatic and rapid community changes in response to climate change (Grimm 1983; Umbanhowar 2004). There are also no-analog pollen records from the past during the late glacial periods of the Quaternary, from 17,000 to 12,000 years ago (Williams and

Jackson 2007). These communities contained species that still exist today but are no longer 30 found together in ecological communities (Williams and Jackson 2007). Such communities also existed much more distantly in the past, for example, during the Paleocene-Eocene Thermal

Maximum (PETM), a period of intense climatic change ca. 55.8 million years ago that is used as an analog for today’s anthropogenic climate change, since warming during the PETM was also caused by elevated carbon dioxide emissions (Dietl and Flessa 2011; McInerney and Wing

2011). Studying plant macrofossils from before, during, and after the PETM, Wing and Currano

(2013) determined that plant community composition during the PETM is distinct from that before or after. This reﬂects migration rather than extinction, since missing species reappeared in the fossil record following the PETM. That there was little evidence of mass extinctions during the warming of the PETM may provide some comfort to restoration ecologists today. However, it is unclear to what extent current warming will mirror that of the PETM, especially as contemporary rises in carbon dioxide emissions are occurring at much faster rates than during the

PETM (McInereny and Wing 2011).

Most pollen-based vegetation reconstructions do not provide the spatial resolution necessary to reconstruct the heterogeneity of vegetation at the scale of landscapes. However, as more data from closely spaced sites with similar climate but differences in soils and topography are collected, it is becoming possible to reconstruct landscape-scale vegetation patterns. In North

America, there are several regions with a dense grid of sites, including the upper Midwest

(Umbanhowar 2004; Nelson and Hu 2008; Lynch et al. 2014) and New England (Foster et al.

2006; Oswald et al. 2007, 2011). Data from networks of sites are particularly useful for understanding which parts of a landscape are most resilient and which are more likely to undergo 31 state shifts in response to climatic changes (Ireland et al. 2012; Lynch et al. 2014; Tweiten et al.

2015).

Paleoecological and phylogeographic data, along with species distribution modeling, are being used to determine the locations of past climate refugia—areas where species survived periods of intense climate change—and to predict the locations of future refugia (Gavin et al. 2014).

Management can be prioritized to conserve and/or restore these areas in preparation for further change (Millar et al. 2007; Shoo et al. 2013). Similarly, phylogenetic data can be used to determine species that are likely to be vulnerable to climate change. Willis et al. (2008) studied the phylogenetic signal of changes in species’ abundance and ﬂowering time after 150 years, using data initially collected by Henry David Thoreau in Concord, Massachusetts. They found that lineages with ﬂowering times that did not track with climate change were declining and in danger of local extirpation. Thus, phylogeny, along with historical data, could be used to identify vulnerable species that would be unlikely to adapt (through evolution or plasticity) to changing climates and prioritize those species for interventions, such as assisted migration (Vitt et al.

2010).

Historical and modeling data can also be used to identify locations for establishing neonative communities, deﬁned as restoring species to an area where they were found in the past but do not currently occur (Millar et al. 2007). On a shorter timescale, dendrochronology in combination with climate projections can be used to identify the tree species and communities most vulnerable to changing climates (Williams et al. 2010). Fulé (2008) recommends focusing management on forest habitats that are likely to persist through climate change—such as higher- 32 latitude, higher-elevation sites—and using both historical and predicted climate data to engineer forests in areas where they are likely to persist in future climates.

Restoration practitioners may be able to use evolutionary theory to develop restored communities with greater potential to adapt in the face of future change (Sgrò et al. 2011). Both inter- and intraspecific genetic variation are thought to maximize evolutionary potential in restoration seed mixes (Broadhurst et al. 2008; Kettenring et al. 2014). Furthermore, increasing evolutionary potential may be accomplished by maximizing phylogenetic diversity of restored sites (Forest et al. 2007; Rosauer and Mooers 2013). Building corridors between fragmented sites is also thought to increase evolutionary potential by allowing for increased gene flow between populations (Sgrò et al. 2011; Haddad et al. 2014). In addition, conserving centers of endemism—areas of high historical evolutionary diversification—may be important for preserving evolutionary potential in the face of an uncertain future (Jetz et al. 2004).

Several of these ideas, including climate refugia and corridors, are tied into the concept of conserving nature’s stage. This strategy focuses on conserving geological diversity

(geodiversity) as a surrogate for biological diversity (Beier et al. 2015). Geodiversity is strongly tied to biological diversity, and conservation of geodiversity may help to mitigate species losses due to climate change (Gill et al. 2015; Lawler et al. 2015). Ensuring that restoration areas include geomorphic heterogeneity may be one way to prepare for a changing climate.

Appropriate species (the actors on the stage) may be added as climates change (Comer et al.

2015). Conserving the stage will create diverse habitats for evolution in future climate regimes

(Lawler et al. 2015). 33

Ecosystem function

Sediment records are well suited for measuring changes in ecosystem processes in aquatic environments (Willard and Cronin 2007). Multiproxy approaches—including diatoms, magnetic susceptibility, biogenic silica, organic content, and nutrient ﬂuxes to sediments—have been used to measure modern impacts of agriculture and the eutrophication of lakes and rivers (Edlund et al. 2009a; Engstrom et al. 2009) and then been applied to restoration goal setting (Edlund et al.

2009b).

Isotopic data are also widely used. For instance, Callaway et al. (2007, 2012) used radioisotopes to infer sedimentation rates in reference California coastal wetlands and determined that sediment accretion is currently keeping pace with sea level rise. This information can be used to set restoration targets for sediment accretion in restored sites so that restored wetlands can remain functional (tidal) in the face of predicted change. Also, these radioisotope data can be used to estimate the rate of carbon sequestration in wetlands, which is useful for restoration planning pertaining to climate change mitigation goals (Callaway et al. 2012).

Stable isotopes can be used to elucidate other ecosystem processes and potential restoration impacts as well. For example, stable isotopes from archeological middens have been used to identify changes in productivity and trophic status of various marine species in the Paciﬁc

Northwest over 4500 years (Misarti et al. 2009), and ratios of nitrogen (N) stable isotopes in lake sediments have been used to estimate the abundance of migrating salmon in Alaska over the past

300 years and changes imparted by commercial ﬁshing and dam building (Finney et al. 2000).

These data document the large-scale movement of N from the ocean to inland, oligotrophic 34 lakes. In addition to providing baseline data to help establish restoration goals, it helps modern restoration workers to predict ecosystem effects of salmon restoration.

It is relatively more difﬁcult to infer how terrestrial ecosystems functioned in the past. However, as collaborations between restoration practitioners and historical ecologists become more common, approaches for addressing past terrestrial ecosystem functioning are emerging

(Dunnette et al. 2014; McLauchlan et al. 2014). Efforts are underway to integrate paleoecological data with ecosystem modeling to facilitate modeling of future ecosystems

(PalEON 2015). These results will also be relevant to restoration ecologists.

The connections between key ecosystem functions and phylogenetic diversity may be the strongest argument for the use of phylogenetic information in restoration ecology. For example, more phylogenetically diverse plant communities are more productive and stable (Cadotte et al.

2008, 2012). In a plot experiment, Cadotte (2013) found an increase in primary productivity of

12 g/m2 of biomass for every additional 5 million years of evolutionary history encompassed by an assemblage. Greater phylogenetic diversity of resident communities is also associated with reduced invasion by nonnative species (Davies et al. 2011; Li et al. 2015). In addition, greater facilitation was found between more distantly related co-occurring species used in arid lands restoration (Verdú et al. 2012). Plant phylogenetic diversity was also associated with greater productivity of soil microbes in gypsum ecosystems (Navarro-Cano et al. 2014). Productivity, stability, facilitation, invasion resistance— these are all factors that may enhance the achievement of restoration objectives (Rowe 2010; Wortley et al. 2013). 35

Phylogenetic diversity of plant communities is also associated with greater biodiversity support at higher trophic levels. In strip-mined lands restored to prairie in southern Ohio, butterﬂy species diversity increased with plant phylogenetic diversity (Cavender-Bares and Cavender

2011). Similar effects on higher trophic levels were seen in experimental manipulations of plant phylogenetic diversity, with phylogenetic diversity of plants predicting arthropod richness and abundance (Dinnage et al. 2012) and phylogenetic diversity (Lind et al. 2015). Conversely, phylogenetic diversity of insects may beneﬁt plants. Increased phylogenetic diversity in pollinator communities reduced rates of self-pollination in the plant Plectritis congesta

(Adderley and Vamosi 2015). This could help limit reductions in plant genetic diversity associated with habitat fragmentation. Each of these approaches to planning resilient and functional restorations in the face of rapidly changing climates combines the ancient with the novel. A novel ecosystem is deﬁned as one that cannot be returned to its historical trajectory and therefore can contribute better to landscape conservation and restoration if it is managed to provide ecosystem services (Hobbs et al. 2014). Using the approaches described above (e.g., designing species mixes with high phylogenetic diversity), restoration can be guided by ecological history, even in novel situations where objectives center on provision of ecosystem services (Willis et al. 2010; Cavender-Bares and Cavender 2011).

Just as it is worthwhile to use a range of spatial references in developing restoration targets and conceptual models for how restored ecosystems develop and respond to change, it is also worthwhile to use a range of historical conditions, especially given an uncertain future. Jackson and Hobbs (2009, p. 568) suggest that “restoration efforts might aim for mosaics of historic and engineered ecosystems, ensuring that if some ecosystems collapse, other functioning ecosystems 36 will remain to build on.” The historic ecosystems may also be a patchwork, with different restorations being informed by different slices of history and different approaches to historically informed restoration. This approach would support ongoing learning and adaptability of restoration as restoration approaches that are informed by ecological history are tested in a changing future (Millar et al. 2007).

Acknowledgments

This article was based on a symposium convened at the Society for Ecological Restoration’s

World Conference in Madison, Wisconsin, in October 2013. We thank two anonymous reviewers and Patrick S. Herendeen for insightful comments on a previous draft. This article is based in part on work supported by grants from the National Science Foundation (DEB1354551,

DEB-1354426, DEB-1118644, DEB-0816991, DEB-0816557, DEB-0816762, DEB-808466,

DEB-0321563), the Paciﬁc Islands Climate Change Cooperative, the Pittman Robertson program of the Wisconsin Department of Natural Resources, and the Program in Plant Biology and

Conservation of Northwestern University and the Chicago Botanic Garden.

CHAPTER TWO

Restored tallgrass prairies have reduced phylogenetic diversity compared with remnants

Chapter two has been published in Journal of Applied Ecology (Barak et al. 2017), with coauthors E.W. Williams, A.L. Hipp, M.L. Bowles, G.M. Carr, R. Sherman, and D.J. Larkin. I conceived the ideas for the study of restored prairies and seed mixes, analyzed the data, and wrote the manuscript. I designed the methodology with Williams, Hipp, Bowles and Larkin. I collected data from restored prairies along with Williams, Carr and Sherman and Bowles collected data from remnant prairies in 2001 that I used for the analysis. Hipp helped with construction of the phylogenetic trees used in this and other chapters. Larkin, my dissertation advisor, provided valuable guidance in all stages of the study. All coauthors provided input to the final manuscript. This manuscript is included as chapter two with permission of all coauthors and the journal publisher.

Summary 1. Ecological restoration is critical for mitigating habitat loss and providing ecosystem

services. However, restorations often have lower diversity than remnant, reference sites.

Phylogenetic diversity is an important component of biodiversity and ecosystem function

that has only recently been used to evaluate restoration outcomes. To move towards

prediction in the restoration of biodiversity, it is necessary to understand how

phylogenetic diversity of restorations compares with that of reference sites, and where

deficits are found, to evaluate factors constraining phylogenetic diversity. 38

2. We quantified plant taxonomic and phylogenetic diversity in eastern tallgrass prairie, one

of the most endangered ecosystems on earth. We measured diversity at large (site) and

small (plot) scales in 19 restored prairies and compared patterns with those from 41

remnant prairies. To evaluate how environmental conditions and management actions

influence outcomes, we tested the effects of soil properties and seed mix composition on

diversity of restorations.

3. Restored prairies were less phylogenetically diverse than remnants at both spatial scales.

On the other hand, the total species richness of remnant and restored prairies did not

significantly differ, but remnants had higher native richness. Restored communities were

taxonomically and phylogenetically distinct from remnants.

4. Soil properties (moisture and pH) influenced phylogenetic diversity and composition.

There were positive relationships between the taxonomic and phylogenetic diversity of

seed mixes and resulting diversity of planted assemblages (excluding volunteer species).

Species in seed mixes were more closely related than expected by chance, and several

clades found in remnant prairies were missing from seed mixes.

5. Synthesis and applications. Restored tallgrass prairies had lower phylogenetic diversity

than remnant prairies, which may contribute to the widely observed phenomenon of

restorations not being functionally equivalent to reference sites. It is encouraging for

restoration efforts that seed mix phylogenetic diversity predicted phylogenetic diversity

of planted assemblages. This indicates that designing phylogenetically diverse seed mixes

for restoration is beneficial. In addition, clades found in reference sites that are missing

from restoration seed mixes could be added to new or existing restorations to reduce gaps 39

in phylogenetic diversity. Further work on the effects of management on phylogenetic

diversity is needed to advance restoration of biodiversity.

Key-words: biodiversity, composition, establishment, grassland, phylogenetic diversity, phylogeny, reference site, seed mix, soil, tallgrass prairie

Introduction

Ecological restoration is essential for biodiversity conservation and ecosystem service provision in a changing world (Rey Benayas et al. 2009; Possingham, Bode and Klein 2015). Phylogenetic measures of biodiversity – metrics that reflect evolutionary distance among species in a community – are predictive of ecosystem functions and properties that are important restoration objectives (Srivastava et al. 2012; Hipp et al. 2015). Plant communities with higher phylogenetic diversity have been shown to be more productive, stable, diverse at higher trophic levels and resistant to invasion than those with lower phylogenetic diversity (Cadotte, Cardinale and Oakley

2008; Davies, Cavender-Bares and Deacon 2011; Cadotte, Dinnage and Tilman 2012; Dinnage et al. 2012; Li et al. 2015).

Further, phylogenetically informed measures of biodiversity can be more predictive of ecosystem function than other diversity metrics, such as species richness (SR) and functional diversity (e.g.,

Cadotte et al. 2009). The relationship between phylogenetic diversity and ecosystem function is partly driven by phylogenetic information providing a proxy for functional trait information via phylogenetic niche conservatism (Wiens and Graham 2005). Phylogenetic information can also account for ecologically relevant variation not captured by measured traits alone (Cadotte et al.

2009; Pearse and Hipp 2009; Larkin et al. 2015a). Because phylogenetic diversity is often 40 closely related to ecosystem function, phylogenetic analyses can inform ecological restoration

(Verdú, Gomez-Aparicio and Valiente-Banuet 2012; Hipp et al. 2015; Barber et al. 2016).

Beyond their use in predicting ecosystem function of restored sites, phylogenetic measures can enrich understanding of remnant communities, relatively undegraded sites that can serve as reference sites for restoration (Larkin et al. 2015a). Phylogenetic measures of biodiversity can combine aspects of SR, composition and evolutionary history and help to provide a deeper understanding of reference communities. Comparing phylogenetic diversity and composition of reference and restored sites can pinpoint deficits that may be correctable through management actions (e.g., adding species from clades that are represented in remnants, but absent from restorations). Understanding discrepancies between restored and reference systems with respect to phylogenetic diversity and composition can also aid in identifying historical constraints to restoration and shaping future restoration objectives (Barak et al. 2016; Turley and Brudvig

2016). Whilst there are many metrics for quantifying phylogenetic diversity, mean pairwise distance (MPD) and mean nearest taxon distance (MNTD) are commonly used in community ecology and have shown utility for community-level comparisons (Tucker et al. 2016). MPD and

MNTD are less confounded by SR than other phylogenetic diversity metrics, and therefore are useful in comparative studies with SR. We henceforth refer to MPD and MNTD measures jointly as ‘phylogenetic diversity.’

Phylogenetic diversity is a potentially useful criterion for designing functional restorations, particularly if restoration inputs (e.g., seed mixes) are reliable predictors of outcomes. Plant community inputs comprise the potential restored community, but relationships between the composition of seed mixes and properties of resulting plant communities are not always strong. 41

Seeded species need to pass through multiple filters to be represented in restorations: emerging, establishing and persisting only with the right confluence of species biology, management actions and site conditions (Hillhouse and Zedler 2011; Grman and Brudvig 2014; Grman et al.

2015; Barber et al. 2016). Because of this, many seeded species fail to become part of realized plant communities (Foster et al. 2007; Hillhouse and Zedler 2011; Grman et al. 2015). Soil properties can also constrain the restoration of biodiversity and ecosystem function (Baer,

Heneghan and Eviner 2012; Grman and Brudvig 2014; Sollenberger et al. 2016). To advance restoration of phylogenetic diversity, it is critical to understand how seed mixes and site conditions shape restoration outcomes.

We evaluated taxonomic and phylogenetic components of plant diversity in restored tallgrass prairies and compared patterns with those found in remnant prairies. We had several goals: First, we tested the hypotheses that restorations would have lower taxonomic and phylogenetic diversity than remnants, and that restorations and remnants would differ in taxonomic and phylogenetic community composition. Since differences between remnant and restored sites are often scale-dependent, we assessed these relationships at two spatial scales (plot and site)

(Allison 2002; Martin, Moloney and Wilsey 2005). Second, we tested factors that may influence restorations’ phylogenetic diversity outcomes, evaluating the influence of soil properties and seed mix composition on resulting assemblages. Lastly, we sought to identify phylogenetic deficits in restored prairies that could be remedied through management actions, by identifying

‘missing branches’ – clades that were either absent from seed mixes, or that were present in mixes, but that failed to establish or persist in restored prairies.

Materials and methods

Site descriptions

We surveyed 19 restored prairies within 11 preserves in northeastern Illinois, USA. Restorations were initiated between 1998 and 2012 (mean: 8.3 ± 3.5 SD years restored) and ranged in size from 0.7 to 300 ha (mean: 27.0 ± 66.3 SD, See Table S1, Appendix 1). All restorations were initiated or at some point managed by the same restoration contractor (Pizzo and Associates,

Leland, IL, USA). We obtained seeding lists for all sites from restoration managers. We were unable to consistently obtain records for other management actions that might be influential, e.g., burning and/or herbicide treatment and seeding rate. Such factors are thus not included in this analysis.

We used plant community data from 41 remnant prairies as a reference dataset for evaluating restorations. These sites include some of the highest quality remnant prairies in the region; they exhibit no evidence of soil disruption (e.g., due to past tillage agriculture) but are embedded within a largely developed/agricultural landscape (Bowles and Jones 2013). While all remnants and restorations are in northeastern Illinois, they are not within the same preserves and therefore did not share substrates or management regimes. Remnant prairies were surveyed by Bowles and

Jones (2013). Detailed phylogenetic and functional trait analyses of these remnants are reported in Larkin et al. (2015a). Though Bowles and Jones (2013) and Larkin et al. (2015a) report data from both 1976 and 2001, we used only 2001 data in our comparisons with restorations. We surveyed restored prairies using the same methods as Bowles and Jones to facilitate comparisons

Vegetation surveys

At each study site, we placed two 50-m transects. Study plots (0.25 m2 round quadrats) were positioned every 5 m along each transect, for a total of 20 plots per site. Each plot was randomly placed either left or right of the transect line and offset by a randomly selected distance of 2.0–

7.0 m (at 1.0-m intervals). We identified all taxa within each plot to species, or genus if we could not reliably identify to species. Samples that could not be identified in the field were collected and pressed for identification at the Chicago Botanic Garden. Vegetation surveys were conducted between 16 June and 4 August 2015. Later in the season (September 2015), we returned to each site to conduct a directed search for species present at the site, but not captured in our transect survey data. Two researchers (EWW and RSB) walked through each site for at least 60 min, identifying and recording species. For comparisons between restored and remnant prairies, we define ‘site scale’ diversity as the additive richness across all 20 plots at each site.

For comparisons of restored sites and their seed mixes, ‘site scale’ also included additional species found in our directed search (which was not performed during sampling of remnants).

Taxonomic names for all species found in remnants, restorations and seed mixes were standardized using the Taxonomic Name Resolution Service (Boyle et al. 2013). Species were assigned native status according to Swink and Wilhem (1994) and the USDA PLANTS database

(USDA 2016).

Soil data

For 18 of the 19 restored sites, we collected and composited two 15-cm deep soil cores from within each surveyed plot (we were not permitted to collect soil from the remaining site). For 44 each sample, we measured gravimetric soil moisture (GSM), soil pH, organic matter by mass loss on ignition (LOI) and electrical conductivity (EC). Samples were sifted through a metal sieve with 0.5-cm openings to remove rocks, roots and other debris. Subsamples (4–10 g) were weighed, dried in an oven to constant mass (≥24 h) at a temperature of 105 °C and weighed again to calculate GSM. To measure LOI, dried samples were placed in a muffle furnace at 300

°C for 4 h, moved to a 105 °C oven for 20 min and then to a desiccator for 30 min, and weighed.

We tested pH and EC using a handheld probe (Multi-parameter PCSTestr 35; Eutech/Oakton

Instruments, Vernon Hills, IL, USA) on dried soil rehydrated with approximately 15 mL of deionized water. We calibrated pH values using standard solutions of pH 4.01, 7.00 and 10.01 and EC values using 1413 and 12880 µs/cm solutions. As GSM and LOI values were highly collinear, we used only GSM values in our analyses. GSM and EC values were natural-log transformed to better approximate normality for statistical analyses.

Phylogenetic tree

We constructed a community phylogeny of 589 species found in restored and reference sites and seeding lists by modifying a published tree of 32 223 plant taxa (Zanne et al. 2014). We grafted taxa not present in the Zanne et al. tree (131 identified to species, 63 identified to genus), and pruned taxa absent from our dataset in R version 3.1.3 (R Development Core Team 2013), using the weldTaxa and make.matAndTree functions in the Morton R project (A. Hipp, https://github.com/andrew-hipp/morton). The Zanne et al. tree was constructed using GenBank sequences for seven gene regions (18S rDNA, 26S rDNA, ITS, matK, rbcL, atpB and trnL-F) using maximum likelihood for tree estimation. A single non-angiosperm species found in a restored site (Sceptridium multifidum) was excluded from analysis. 45

Data analysis

All analyses were performed using R version 3.1.3. We calculated three measures of biodiversity

– SR and two measures of phylogenetic diversity, MPD and MNTD – at both plot and site scales.

MPD is the mean phylogenetic distance between all pairs of species in a community; MNTD is the mean phylogenetic distance between each species and its closest relative in the community.

Metrics were calculated based on the presence-absence of species in plots and sites, i.e., were not abundance-weighted (Webb et al. 2002; Kembel et al. 2010).

To remove the influence of SR on MPD and MNTD, we calculated standard effect sizes:

(observed value – expected value)/ standard deviation of the expected value. We calculated expected values under a ‘frequency’ null model with 999 permutations using the functions

SES.MPD and SES.MNTD in the picante package (Kembel et al. 2010). The frequency null randomizes the presence/absence of species across samples while controlling for species’ frequency of occurrence. Positive values of SES.MPD/ SES.MNTD indicate that species present in the community are more distantly related than expected by chance (phylogenetically overdispersed), while negative values indicate that species are more closely related than expected by chance (clustered). Consequently, higher values of SES.MPD/SES.MNTD indicate higher phylogenetic diversity.

We tested for differences in diversity (SR, SES.MPD and SES.MNTD) between remnant and restored sites. At the site scale, we tested for significant differences using ANOVA; at the plot- scale we used linear mixed-effects models, with site as a random effect, to account for non- independence of plots within sites. We repeated analyses with all species included and with only 46 native species included (to reflect plant assemblages targeted in restoration). Mixed-effects models were fit using the lme function in the nlme package (Pinheiro et al. 2016).

We used linear models to test the effects of prairie size and years since restoration was initiated

(age) on each diversity metric at the site scale, and mean plot scale. We natural-log transformed area to better approximate a normal distribution. To test for effects of soil properties (GSM, pH and EC) on diversity measures, we used linear models at the site scale and linear mixed-effects models at the plot scale, with site as a source of random error. For the linear mixed-effects models we report both marginal and conditional R2 as described in Nakagawa and Schielzeth

(2013).

We also investigated the relationships between restored prairies and the seed mixes used to establish them. To determine whether seed mix diversity was predictive of site diversity, we constructed linear models with seed mix diversity as the predictor variable and site diversity as the response variable. We separately tested the effects of seed mix diversity on site diversity of all species in communities and of only the planted community, i.e., the subset of species seeded into each site, excluding colonizing ‘volunteer’ species.

We tested for differences in community composition between remnants and restorations and between restorations and seed mixes, using site-level presence-absence data. To compare the taxonomic composition between groups, we performed non-metric multidimensional scaling ordination using the vegan package (Oksanen et al. 2016). We also performed community phylogenetic ordinations wherein dissimilarity was based on phylogenetic, rather than taxonomic, distance between samples, using the functions comdist (for MPD-based calculations) and comdistnt (for MNTD) in picante. We tested for significant differences between community 47 types (remnants vs. restorations, seed mixes vs. sites) using permutational multivariate analysis of variance (PERMANOVA, Anderson 2001) with the adonis function in vegan. We evaluated which species were driving differences between groups by quantifying the contributions of each species to overall Bray-Curtis dissimilarity using the similarity percentage (SIMPER, Clarke

1993) function in vegan.

We also evaluated the effects of environmental factors on community composition with redundancy analysis (rda, Legendre and Legendre 2012) for continuous predictors using the capscale function in vegan. We used site-level mean values of GSM, EC and pH as predictor variables and site-level community composition as response variables. We tested for significance using the ANOVA function, with 999 permutations, and calculated R2 values using the function

RSquareAdj in vegan.

Lastly, we tested whether species included in restoration seed mixes, and species that emerged and persisted after planting, were more closely related to one another than expected under a null model. We first tested for phylogenetic signal in planted species, relative to the full phylogeny of all species present in remnants and restorations. The characteristic evaluated was binary, i.e., whether or not a given species was planted. To test this, we calculated the D statistic (Fritz and

Purvis 2010) using the phylo.d function in caper (Orme et al. 2013). We estimated D and tested whether it was phylogenetically random (D = 1), non-random (D < 1), or structured as would be expected under a Brownian motion model of evolution (D = 0). We also tested for phylogenetic signal in two continuous characteristics: planting frequency (proportion of seed mixes in which a particular species occurred) and persistence (proportion of sites a given planted species was found in during sampling). We evaluated these by calculating the K statistic using the 48 phylosignal function in picante. K = 1 indicates the degree of phylogenetic signal expected under

Brownian motion, while K < 1 and K > 1 indicate lower and greater than expected phylogenetic signal respectively (Blomberg, Garland and Ives 2003). For both D and K, significance was evaluated by comparing observed values with results from 999 permutations of tip-shuffling randomizations. Phylogenetic signal of persistence was separately tested for all seeded species and for more commonly seeded species (included in seed mixes for ≥3 sites).

Results

Remnant vs. restored prairies

We found a total of 341 species in restored sites representing 54 families. Of these, 252 species from 42 families were found during transect surveys. An additional 89 species, and 12 families were found through the directed search. In contrast, there were 353 species from 64 families found at remnant sites (Table S2).

Restored prairies had lower phylogenetic diversity than remnant prairies by all metrics and at both spatial scales (Figure 2.1, Appendix S1). On average, MPD of remnant prairies was 17 million years higher (raw MPD, 261.5 ± 1.5 vs. 244.5 ± 2.5, 1 SE), and MNTD was 19 million years higher in remnants (71.6 ± 1.6 vs. 52.6 ± 1.5) at the site scale (Appendix S1). SES.MPD and SES.MNTD values indicated that restorations, but not remnants, were phylogenetically clustered (Figure 2.1). The same patterns persisted when non-native species were excluded from analyses (Appendix S1). SR did not significantly differ between remnant and restored prairies at site or plot scales when all species were included. However, when non-native species were excluded, mean SR was higher in remnant prairies at both scales. 49

Remnant and restored communities were distinct from one another in taxonomic and phylogenetic composition (Figure 2.2). SIMPER analyses indicated that taxonomic differences were driven primarily by more frequent occurrence of regionally common native species occurring as volunteers in restored prairies (e.g., Ambrosia artemisiifolia, Oxalis stricta,

Symphyotrichum pilosum, Oenothera biennis, Table S3). Secondarily, these differences were associated with higher frequency of common restoration species in restored sites (e.g.,

Penstemon digitalis and Bouteloua curtipendula) and habitat specialist, conservative species in remnant sites that were not found in restorations (e.g., the hemiparasite Comandra umbellata,

Swink and Wilhem 1994).

Factors influencing diversity of restored sites

Restoration size and age were not significant predictors of plot or site scale diversity (Table S4).

At the site scale, soil moisture (GSM) was a significant predictor of SES.MNTD (F1,16 = 2.243,

R2 = 0.234, P = 0.042). However, this trend was strongly influenced by a single site with very high soil moisture and was not significant with that site excluded. In plot-scale analyses, soil

2 moisture was a significant predictor of SES.MNTD (F1,339 = 8.00, conditional R = 0.177,

2 marginal R = 0.042, P = 0.005), even when the outlier site was excluded (F1,320 = 4.72, conditional R2 = 0.157, marginal R2= 0.024, P = 0.030). In multivariate analyses (rda), pH was a significant predictor of taxonomic community composition, and pH and soil moisture were significant predictors of phylogenetic community composition (MNTD, Table S5).

Restoration seed mixes included a total of 209 species. Seed mix richness ranged from 29 to 125 species (mean ± 1 SE = 63.3 ± 6.8). The proportion of planted species observed at each site

(including both transect surveys and directed searches) ranged from 25% to 77% (44.6 ± 3.4%). 50

Seed mix diversity was not a significant predictor of overall plant community diversity but was significant when volunteer species were excluded (Figure 2.3). Equations from raw MPD and

MNTD data are as follows, SR: y = 0.28x + 10.35, MPD: y = 1.35x—98.38, MNTD: y = 0.86x +

12.77). Equations for raw values are reported for ease of interpretation (raw and SES values were highly correlated, R2 = 0.999 for both MPD and MNTD).

Restored plant communities were taxonomically and phylogenetically distinct from seed mixes when all species were included and, to a lesser degree, when only planted species were included

(Figure 2.2). When all species were included, seed mix and site differences were driven by the presence of non-native species or volunteer native species present in many restorations, but few or no seed mixes (Figure 2.2d). When only planted species were included, restored communities showed much more overlap with seed mixes, despite loss of seeded species that failed to establish (Figure 2.2g).

We found significant phylogenetic signal with respect to which species were seeded (D = 0.65,

P < 0.001 relative to random null, Figure 2.4). Several families well-represented in remnant prairies were not present in any seed mixes; these included the Liliaceae, Orchidaceae,

Convolvulaceae and Polygalaceae. The entire order Caryophyllales (Amaranthaceae, Cactaceae,

Caryophyllaceae, Phytolaccacea, Polemoniaceae, Polygonaceae, Portulacaceae and Primulaceae families) was absent from seed mixes. Of the species that were planted, there was no phylogenetic signal in the proportion of seed mixes each species appeared in (K = 0.003, P =

0.82) or in their persistence (K = 0.005, P = 0.42). When we repeated these tests using only species seeded in at least three sites, the same pattern emerged (K = 0.005, P = 0.49 and K =

0.005, P = 0.42). 51

Discussion

Phylogenetic diversity of restored vs. remnant prairies

Restorations exhibited consistently lower phylogenetic diversity than remnants regardless of the metric used or whether all or only native species were included. In addition, restorations’ native

SR and phylogenetic diversity was lower than that of remnants at both site and plot scales; this contrasts with previous findings using nonphylogenetic metrics that restorations may reach reference levels of diversity for one but not multiple spatial scales (Allison 2002; Martin,

Moloney and Wilsey 2005; Middleton, Bever and Schultz 2010). Understanding the processes that mediate diversity at different scales is an important step towards restoring ecosystems that are predictably more similar to remnants (Polley, Derner and Wilsey 2005).

The finding that restored prairies had lower phylogenic diversity than remnants is in agreement with a small number of studies that have tested for differences between relatively intact natural habitats and degraded or disturbed habitats. For example, successional old fields disturbed by mowing showed lower than expected phylogenetic diversity (Dinnage 2009). Turley and Brudvig

(2016) found persistent negative effects of agricultural legacies on phylogenetic diversity in longleaf pine savannas. Loss of historical disturbance regimes has also been linked to lower phylogenetic diversity, as in the case of unburned remnant prairies (Larkin et al. 2015a). The finding that disturbed sites tend to harbor communities that are more closely related than expected by chance may reflect harsh environmental conditions acting as a filter on community membership (Dinnage 2009). In the restored prairies we studied, such effects would be compounded by the restoration inputs themselves (i.e., seed mixes) having been phylogenetically clustered. 52

Factors driving phylogenetic diversity in restored prairies

In order to advance prediction in the restoration of biodiversity, it is necessary to not only compare restorations to reference communities, but also to interpret variation between existing restorations (Brudvig et al., 2017). Among the restored prairies, soil moisture and pH were drivers of diversity and composition. Soil properties have been linked to plant phylogenetic diversity, and soil moisture to establishment of species from seed mixes (Grman and Brudvig

2014; Sollenberger et al. 2016). However, effects of anthropogenic disturbance on contemporary habitats may overwhelm underlying environmental gradients, such as those imparted by soil heterogeneity (Alstad et al. 2016).

In sampled restorations, seed mix diversity was predictive of planted assemblages’ SR and phylogenetic diversity. Use of more phylogenetically diverse seed mixes yielded greater phylogenetic diversity despite, on average, <50% persistence of seeded species. Given the relationship between phylogenetic diversity and ecosystem function, these results may indicate that there are functional benefits to planting phylogenetically diverse seed mixes, even though many planted species may not emerge or persist (Arroyo-Rodríguez et al. 2012; Srivastava et al.

2012).

That seed mix diversity did not predict total plant community diversity (similarly to other studies, e.g., Barber et al. 2016), indicates that colonization by volunteer species can increase or decrease measures of phylogenetic diversity depending on the phylogenetic position of volunteers relative to the planted community. In addition to non-native species, we identified native volunteers in restored communities. These species were rare in seed mixes but found at numerous restorations, e.g., Eupatorium altissimum and Asclepias syriaca. It is unknown 53 whether volunteer species colonized restored sites through natural dispersal, or were deliberately introduced after initial seed mix determination. Additional study of establishment dynamics at restorations would help to clarify these patterns.

Colonization by volunteer species may explain why plant communities at restored sites tended to converge regardless of the seed mix originally used. Taxonomic and MNTD-based community results showed similar patterns, with little overlap between seed mixes and sites. In contrast,

MPD-based analyses showed greater overlap between seed mixes and sites. We interpret this difference to mean that seed mixes and their resultant restored prairies are composed primarily of species from major families of the prairie (i.e., Asteraceae, Poaceae, Fabaceae and Lamiaceae), but that there is turnover in the particular species representing these families. This is because

MPD measures reflect patterns of branching deep within a phylogeny (e.g., within families), while MNTD measures reflect more recent branching, towards the tips of the phylogeny (e.g., within genera) (Tucker et al. 2016). We found that seed mixes and sites showed considerable overlap with respect to MPD, but diverged in terms of MNTD and SR, which is neutral with respect to evolutionary relatedness.

Our finding that clades found in remnant prairies were absent from seed mixes highlights opportunities to enrich the palette of plant materials used for restoration. There are multiple explanations for why species from these clades may have been missing, including lack of commercial availability, high cost or inadequate source populations (Peppin et al. 2010; Rowe

2010). Notable examples of missing families include Lilaceae and Orchidaceae, both of which contain species of regional conservation concern in remnant prairies. These species may have 54 more precise requirements for germination and growth, making them difficult to source as seed and establish in restorations.

Seed and/or habitat modifications may be needed to facilitate their use in restoration (e.g.,

Bowles et al. 2005; Ault and Siqueira 2008). Hemiparasites can be important for ecosystem function and restoration of biodiversity (Pywell et al. 2004; Decleer, Bonte and van Diggelen

2013); several species from missing clades are hemiparasites, including Comandra umbellata

(Santalaceae) and multiple native Cuscuta (Convolvulaceae) species. Focusing on representatives of missing clades could be a means to increase phylogenetic diversity and associated function, while also supporting species-specific conservation goals.

Implications for management and restoration

Incorporating phylogenetic diversity considerations into ecological restoration could advance management in at least two significant ways. First, increasing phylogenetic diversity of restorations may increase performance of key ecosystem functions (Cavender-Bares and

Cavender 2011; Hipp et al. 2015). Second, phylogenetic ecology enriches understanding of differences between communities, offering new opportunities to monitor and evaluate performance of restored sites.

Our finding that restorations have lower phylogenetic diversity than remnants at multiple spatial scales and by multiple metrics indicates that understanding the factors that drive community phylogenetic patterns of restorations is important for building restorations that are predictably more similar to remnants. We found that seed mix diversity predicts established diversity for a range of diversity metrics. More detailed study of seeding techniques and outcomes would refine 55 understanding of relationships between seed mixes and restored plant communities. For example, restorations established through multiple, rather than single, seeding events, are taxonomically more similar to high-quality reference prairies (W. Sluis, M. Bowles, and M. Jones, unpublished data). Similar patterns may emerge with respect to restoration of phylogenetic diversity.

Identifying underrepresented or missing clades in restoration inputs through phylogenetic comparison with reference sites can inform species selection for restoration. Increasing equivalence of restorations’ phylogenetic diversity to that of remnants requires redesigning seed mixes to include species from clades not typically incorporated into restoration, and could include emphasis on planting species of conservation concern.

Furthermore, phylogenetic diversity can be a useful component of ongoing monitoring for restoration (Montoya, Rogers and Memmott 2012; Barak et al. 2016). Species from missing clades could be added in later stages of restoration. There are potential constraints to including these species in restoration, including low availability and/ or low establishment. However, in the short term, planting plugs, rather than seeds, of desired species from underrepresented clades may help increase phylogenetic diversity, and ecosystem function, at restored sites (Middleton,

Bever and Schultz 2010; Gallagher and Wagenius 2016).

Authors’ contributions

R.B., E.W., A.H., M.B. and D.L. conceived the ideas and designed methodology; R.B., E.W.,

G.C. and R.S. collected the data; R.B. analyzed the data; R.B. led the writing of the manuscript.

All authors contributed critically to the drafts and gave final approval for publication.

Acknowledgements

This study was supported by NSF awards DEB-1354426, DEB-1354551 and DBI-1461007.

R.S.B was supported by the Graduate Program in Plant Biology and Conservation, the Illinois

Association of Environmental Professionals and a Dr. John N. Nicholson Fellowship. We thank

Taran Lichtenberger, Meghan Kramer and Alyssa Wellman-Houde for help with data collection from restored sites, Michael Jones for vegetation sampling from remnant sites, Rachel Goad,

David Sollenberger and Michael Jones for assistance with plant identification and Bryant

Scharenbroch for help with soil analyses. We thank the following for access to and information about restored sites: Dawn Banks, Trish Burns, Jenny Clauson, Pat Hayes, Erick Huck, Keith

Guimon, Jack Pizzo, Nagulapalli Rao, Cassi Saari, Sue Swithin, Byron Tsang and Lauren Umek.

We thank two anonymous reviewers for helpful comments on an earlier draft of this article.

Data accessibility

Plant community data, seed mix data, soil data and the phylogenetic tree are archived in Dryad, https://doi.org/10.5061/dryad.90124 (Barak et al. 2017). Remnant data are already published and accessible at: https://doi. org/10.5061/dryad.763v6 (Larkin et al. 2015b).

CHAPTER THREE

Cracking the case: seed traits, phylogeny, and seed pre-treatments drive germination in tallgrass prairie species used for ecological restoration

Summary

1. Functional traits are important for understanding how communities assemble and

function. However, the majority of studies relating functional traits to community

assembly rely upon vegetative traits of mature plants. Seed traits, which are understudied

relative to vegetative plant traits, are critical for understanding assembly of plant

communities. This is particularly true for restored prairie communities, which are

typically started from seed. Rapid germination of target species is an important goal for

restoring diverse, high functioning plant communities.

2. I tested the effects of seed traits (mass, shape, and embryo to seed size ratio),

phylogenetic position, and germination pre-treatment (cold stratification or gibberellic

acid application) on germination response in 32 species commonly used in tallgrass

prairie restoration using time-to-event analysis.

3. Seed traits, phylogenetic position, and germination pre-treatment all contributed

significantly to best-fit models of the germination data. Out of all traits tested, seed shape

variables (length, width, and height) were most predictive of germination response, with

longer, narrower seeds germinating faster. Phylogenetic position was an important

predictor of germination response despite my finding that all measured seed traits showed 58

phylogenetic signal. This indicates that the phylogeny supplies residual or integrative

information that is important for understanding germination response.

4. Seed traits and phylogenetic position are important predictors of germination response for

a suite of species commonly used in prairie restoration. This information can be used to

guide restoration planning and seed mix design. In addition, practitioners can increase

germination rates through relatively simple, inexpensive seed treatments.

Introduction

Functional traits are important predictors of how plant communities will assemble and function, influencing the ecosystem services these communities provide (Dı́az and Cabido 2001; Díaz et al. 2013; Laughlin 2014; Zirbel et al. 2017). The vast majority of studies linking functional traits to community assembly use vegetative plant traits of mature life stages, like plant height and specific leaf area, to predict community outcomes. Regenerative traits that govern propagule production and dispersal, dormancy, germination, and establishment, are vital to understanding assembly and persistence of plant communities; however, they are surprisingly understudied relative to traits of mature plants (Huang et al. 2015; Jiménez-Alfaro et al. 2016; Larson and

Funk 2016). The role of seed traits may be an especially underappreciated aspect of the assembly of restored plant communities. In restorations that involve active approaches to revegetation, seeding is the most commonly employed tool. Unlike most extant natural plant communities, restorations are largely started from scratch by seed. Thus seed traits may be as or more important than vegetative plant traits for understanding assembly of restored communities

(Hoyle et al. 2015; Jiménez-Alfaro et al. 2016; Larson and Funk 2016). 59

Seed germination is a critical life-stage that drives assembly of diverse restored plant communities (Larson et al. 2015). Germination is irreversible, and therefore early establishment is more sensitive to environmental variation than plant growth and survival in later life stages. A seed that germinates at an inappropriate time may not survive to maturity, but ungerminated seeds face death by predation or disease (Clark and Wilson 2003). Because of this, understanding germination responses of species used in restoration is vital to establishing diverse restored plant communities. Rapid germination, high overall germination, and the ability to germinate without cold stratification have been shown to impact establishment of species in restorations (Pywell et al. 2003). Early-germinating species can interfere with the establishment, growth or persistence of later-germinating species, granting “priority” to the early germinators. These priority effects can operate on very short timescales, and be persistent over many growing seasons (Young et al.

2016). Priority effects can not only favor early-germinating native species over later-germinating natives but also, and importantly for restoration, impede establishment of non-native species

(Grman and Suding 2010; Young et al. 2016). Rapid germination and establishment of native species are desired outcomes for pre-empting invasive species that are common in disturbed habitats and tend to have early germination phenology (McGlone, Sieg and Kolb 2011; Martin and Wilsey 2012). Understanding seed traits that impact germination response may provide guidance for establishing diverse restorations.

Seed mass is the most commonly studied (and frequently only) seed trait in functional ecology research, likely due to the wide availability of seed mass data in trait databases (Jiménez-Alfaro et al. 2016). Seed mass is related to plant functions such as seed dispersal, establishment, competition, and plant growth rate (Turnbull, Rees and Crawley 1999; Weiher et al. 1999; 60

Westoby et al. 2002; Kleyer et al. 2008). Seed mass can be both positively or negatively predictive of germination (e.g., Kahmen and Poschlod 2008; Norden et al. 2009), or not predictive at all (Shipley and Parent 1991). Though seed mass has an important influence on assembly, seed mass alone provides an insufficient basis for predicting differences in germination, establishment, and persistence (Larson and Funk 2016).

External morphological traits like seed shape and internal traits of the seed embryo may be important for understanding germination, and ultimately emergence and persistence. Seed shape is predictive of persistence in soil seed banks, with rounder seeds lasting longer than flat or pointed seeds (Thompson, Brand and Hodgson 1993). Seed shape has also been linked to germination, with elongated seeds germinating more rapidly than rounded seeds (Grime et al.

1981, Bu et al. 2017). Seed shape has been a stronger predictor of germination than mass in some cases (Yu et al. 2007).

Embryo to seed size (E:S) ratio, a measure relating the size of the embryo to that of the whole seed, is a morphological measure that has also been related to seed germination and establishment. A high E:S ratio indicates that endosperm is depleted over the course of seed development, and nutrients are incorporated into the embryo’s cotyledons (Finch-Savage and

Leubner-Metzger 2006). E:S is a phylogenetically conserved trait, and E:S ratio tended to increase over evolutionary time, with smaller E:S ratios found in basal angiosperms (Forbis,

Floyd and de Queiroz 2002). Ecologically, E:S ratio was found to govern species’ establishment in multiple European habitats. Low E:S genera tended to be present in moist areas, and high E:S genera dominated dry habitats, likely because seeds with a high E:S ratio can germinate rapidly 61 after imbibing water, an advantage in arid areas (Linkies et al. 2010; Vandelook, Verdú and

Honnay 2012).

The relationship between traits and germination is likely to be mediated by phylogeny. That is, closely related species have more similar values for many traits due to phylogenetic conservatism, and legacies of shared ancestry would be expected to be evident in germination and seed traits, as has been widely observed for seed mass (Moles et al. 2005; Norden et al.

2009). Therefore, a simple regression between traits and germination response may be inappropriate. In order to better understand the effects of traits themselves on germination, phylogenetic comparative methods can be used to “correct” for the role of phylogeny on distribution of trait values, i.e., phylogenetic autocorrelation (Pagel 1999). On the other hand, rather than correcting for the effects of phylogeny, phylogeny can be explicitly tested as a predictor variable. Testing the role of phylogeny in such a way can allow researchers to quantify residual information that is contained within the phylogeny, but not captured by measured traits

(Pearse and Hipp 2009; Larkin et al. 2015). Phylogenetic position can also summarize key information about species in a way that integrates over many traits (Cadotte et al. 2009; Burns and Strauss 2011; Srivastava et al. 2012). Phylogenetic conservatism has been found to play a role in both seed traits and germination responses, and phylogeny can be used to understand variation in germination response that is not accounted for by measured seed traits alone (Wang et al. 2009; Hoyle et al. 2015; Bu et al. 2017). In seeking to understand the relationship between seed traits and germination response, it is important to explicitly consider the role of phylogeny.

I conducted laboratory investigations to test whether measured traits and phylogeny were predictive of seed germination in a diverse set of plant species commonly used in ecological 62 restoration of the tallgrass prairie. I tracked germination of individual measured seeds and analyzed germination response using statistical time-to-event (survival) analysis with time to germination as the response variable and seed traits and phylogenetic position as predictor variables (McNair, Sunkara and Frobish 2012). To further disentangle the effects of traits and phylogeny on seed germination, I tested whether the traits I measured showed significant phylogenetic signal (Blomberg and Ives 2003). In addition, since I suspected that seed dormancy would mediate the effects of seed traits and phylogeny on germination, I tested these relationships in seeds that were or were not subjected to two treatments intended to break dormancy (cold stratification and gibberellic acid application). Taken together, I tested the effects of seed traits, phylogenetic position, and germination pre-treatment on time to germination of prairie plant species.

Methods

Seed traits

I obtained seeds of 32 species (representing 26 genera and 14 families) that are commonly used in prairie restoration (Table 3.1) from Pizzo Native Plant Nursery (Leland, IL, USA). Seeds were refrigerated until I initiated measurements. I measured seed traits for each of 96 individual seeds per species. Traits fell into three broad categories: (1) seed mass, (2) seed shape, and (3) E:S ratio. I measured seed mass by weighing individual seeds using a precision balance. I collected shape data by measuring three planes of the seed (length, width, and height) using an ocular ruler on a dissecting microscope and by calculating variance between the three planes (Kleyer et al.

2008). Lastly I measured E:S ratio using x-ray analysis (Faxitron, Model MX-W, Tucson, AZ, 63

USA) to quickly and non-invasively measure the embryo relative to the whole seed. X-ray technology has been used to measure seed embryos in crop species like cucumber (Cucumis sativus, Cucurbitaceae, Gomes-Junior et al. 2013) and sunflower (Helianthus annus, Asteraceae,

Rocha et al. 2014). I captured x-ray images and calculated the E:S ratio of each seed using imageJ (Schneider, Rasband and Eliceiri 2012). I calculated E:S ratio in three ways: linear measures of embryo length and width relative to seed length and width, respectively, and embryo area relative to whole seed area. I used contrast measures to estimate embryo area and whole seed area. The three E:S ratio measures are hereafter referred to as ESlength, ESwidth, and ESarea.

Germination

After collecting trait data, I assessed seed germination. All 96 measured seeds of each species, as well as 48 unmeasured seeds (to account for possible effects of measurements on germination), were randomly assigned to three germination treatments: control, gibberellic acid, and cold stratification. I used locations within 96-well plates to track individual seeds for pre-treatment and germination. Using this method, I was able to obtain germination data for each seed for which I had previously collected all trait data, i.e., mass, length, width, height, shape variance,

ESlength, ESwidth, and ESarea.

I prepared 96-well plates for germination by pouring a 2% agar solution into each well. Seeds were randomly placed in individual wells for germination. I included 12 unmeasured control seeds of each species in each treatment. Separate 96-well plates were used for each of the three treatments. Control seeds were plated directly into the wells containing agar without any pre- treatment. Before being plated into agar, seeds in the gibberellic acid group were placed into 64 individual wells that did not contain agar, and soaked in 500-ppm gibberellic acid solution overnight (16-18 hours). Seeds in the cold stratification treatment were placed in wells containing agar, covered with brown paper, placed in a cardboard box to keep out light, and refrigerated (at 3 °C) for 14 weeks to mimic overwintering conditions.

For germination assays, the 96-well plates containing seeds were randomly arranged in an incubator set to a 12-hour photoperiod with day/night temperatures of 20/10 °C. Seeds were checked for germination (radical emergence of ≥1 mm) three times each week for a total of 4 weeks. All germination tests and data collection took place between June 23rd, 2016 and January

11th, 2017.

Phylogenetic tree

I constructed a phylogeny of the 32 species in this study by pruning a larger tree of 589 prairie plant species (Barak et al. 2017), which was modified from a published tree of 32,223 plant taxa

(Zanne et al. 2014). The Zanne et al. (2014) tree was constructed using GenBank sequences for seven gene regions (18S rDNA, 26S rDNA, ITS, matK, rbcL, atpB, and trnL-F) using maximum likelihood for tree estimation. The Barak et al. (2017) tree was made by grafting species not present in the Zanne et al. and pruning non-focal species using the weldTaxa and make.matandtree functions in the Morton R project (A. Hipp, https://github.com/andrew- hipp/morton) in R version 3.1.3 (R. Core Team 2013).

Data analysis

All analyses were performed using R version 3.1.3 (R. Core Team 2013). The germination response variables were 1) a binary measure of whether or not a seed germinated, and 2) the 65 experimental day a seed germinated with day 1 being the day it was placed in the incubator, and day 29 being the last day of the experiment. Predictor variables tested included seed traits, phylogenetic position, and germination pre-treatment. Seed traits comprised eight continuous measurements: mass, length, width, height, shape, ESlength, ESwidth, and ESarea. Phylogenetic position was represented by quantitative, multivariate axes characterizing phylogenetic position for each species. To obtain these axes, I used the phylogenetic distance matrix (pairwise phylogenetic distance between each species) for the tree of the 32 focal species. I square-root transformed the distance matrix (De Vienne, Aguileta and Ollier 2011), performed non-metric multidimensional scaling (NMDS) ordination of the matrix using the isoMDS function in vegan

(Oksanen et al. 2016) and extracted the position of each species along 4 NMDS axes.

Germination pre-treatment was a categorical factor with three levels: cold stratification, gibberellic acid, and a control group with no pre-treatment.

I tested the effects of seed traits, phylogenetic position, and germination pre-treatment on time to germination over the course of the experiment using time-to-event (or survival) analyses using the survival package in R (Therneau and Grambsch 2000). Time-to-event analysis incorporates not just whether or not an event like germination occurs (a binary response) but also the amount of time it takes for the event to occur (a continuous response). I built survival models using a cox proportional hazards model, which allows for both categorical and continuous predictors

(McNair, Sunkara and Frobish 2012), using the coxph function in the survival package. The response variable in survival analyses was time to germination (in experiment days). Predictors were seed traits (8), phylogenetic position (4 NMDS axes), and germination pre-treatment

(categorical predictor with 3 factors). Seed mass data were log-transformed to better approximate 66 a normal distribution. All continuous predictor variables were standardized prior to analysis (to mean = 0 and S.D. = 1) to produce standardized coefficients that could be readily compared among variables as indicators of effect sizes. Candidate models comprising different combinations of predictor variables were constructed, and AIC-based model selection was performed using the StepAIC function (MASS package, Venables and Ripley). I performed stepwise model modification in both forward and backward directions. I performed this analysis twice, once using all species in the experiment (n = 32), and a second time excluding species with very low germination rates (< 5 % germination in any treatment), so that they would not have undue influence on interpretation of results. I also tested for the effects of seed traits on intraspecific variation in time-to-germination, using cox models for each individual species. I first tested for the effects of germination pre-treatment on time-to-germination. In species that did not have a significant effect of pre-treatment, I pooled treatment groups to test for the effects of seed traits on germination. In species with significant differences between treatment groups, I analyzed groups separately. Lastly, I used time-to-event analysis to test for differences in germination response between measured seeds and unmeasured controls.

I tested for phylogenetic signal in the 8 measured seed traits, i.e., phylogenetic autocorrelation in species’ trait values indicative of phylogenetic trait conservatism. Phylogenetic signal was evaluated with the K statistic using the phylosignal function in picante (Kembel et al. 2010). K =

1 indicates the degree of phylogenetic signal in a trait that would be expected under a Brownian motion model of evolution, while K < 1 and K > 1 indicate lower and greater phylogenetic signal, respectively (Blomberg, Garland and Ives 2003). Significance was assessed by comparing observed values of K to results from 1,000 permutations of tip-shuffling randomizations. 67

Results

Final percent germination ranged from 0 – 94 % depending on species and germination treatment

(Table 3.1). Seeds of one species, Maianthemum racemosum (Asparagaceae), did not germinate under any germination treatments. Only a single seed of the species Sisyrinchium angustifolium

(Iridaceae) germinated. In contrast, three species (Dalea candida, Monarda bradburiana, and

Thalictrum dasycarpum) reached 94 % germination under gibberellic acid (D. candida and M. bradburiana) and cold stratification (T. dasycarpum) pre-treatments.

Seed traits, phylogenetic position, and germination treatment were all retained in the top-ranked models (Table 3.2). Seed shape measurements (seed length, width, and height), all phylogenetic axes and germination pre-treatment were retained in all top models, and were significant contributors to the averaged model (Table 3.3). Measuring seeds did not influence germination response, as measured and unmeasured seeds did not differ in time to germination (Z = 0.71, P =

0.48).

Length was a positive predictor of germination, while height were negative predictors. Taken together, these patterns suggest that long, narrow seeds should have higher germination rates, though my measure of shape variance was not a strong predictor of germination in either averaged model. As described above, I performed model selection using data from all species and with Maianthemum racemosum and Sisyrinchium angustifolium removed due to low germination rates. Results based on all species are reported in the main text and those of the second analysis are provided in Appendix 2. Interpretation of results was generally consistent between these two analyses, with the exception of patterns related to seed mass, which was a 68 significant negative predictor only in the averaged model based on all species. Maianthemum racemosum had the heaviest seed of all species, but never germinated. When this species (and S. angustifolium) was excluded, mass was no longer a significant predictor of germination in the averaged model. In addition, ESwidth was the only significant E:S measure in the first analysis, while ESlength was the only significant E:S measure in the second analysis.

The phylogenetic ordination produced four axes describing phylogenetic position (stress =

18.25). The first ordination axis was strongly associated with the separation between monocots and dicots, and dicots (higher values on axis 1) tended to have more rapid germination. The second axis moved across the phylogeny from Asteraceae to Fabaceae, with seeds at the

Fabaceae (lower) part of the axis having higher germination. The third and fourth axes appeared to further differentiate further within the dicots and monocots, respectively (Figure 3.1). All phylogenetic axes used in the model were significant predictors of germination. In general, germination rates were higher under cold stratification, which is necessary for dormancy break of many prairie species (Figure 3.2).

In intraspecific analyses I found that seeds subjected to different germination treatments differed in germination response in 22 species out of 30 species tested. In addition, seed traits were predictive of germination in 17 species. Results are summarized in the Appendix, Figure 1X.

All seed traits showed significant phylogenetic signal, with K values ranging from 0.025 to 0.219

(Table 3.4, Figure 3.3). These values indicate lower phylogenetic signal than would be expected relative to a Brownian motion model of evolution.

Discussion

Seed traits and phylogenetic position were necessary to explain differences in time to germination. Despite my findings that seed traits were phylogenetically structured, and were key to explaining germination patterns, phylogenetic position was still a relevant predictor of germination response. Phylogenetic position likely provided complementary information to the measured traits, as a proxy for unmeasured trait variation, or a summation of many relevant traits.

Mass is the seed trait most commonly used in functional ecology (Larson and Funk 2016). That mass was not one of the traits that best explained germination indicates that incorporating additional seed traits would be beneficial for understanding links between traits, community assembly and ecosystem function. Shape-based measures, including length, width, and height, were retained in all of the top models in my analyses. Like other studies, I found that shape predicted germination, with longer and narrower seeds having greater germination rates.

The three embryo measurements were retained in some, but not all the best models, and contributed to the models less than the shape-based traits. In averaged models, ESwidth positively predicted germination. This finding, that E:S measurements had relatively modest explanatory power, can be explained in several ways. First, E:S ratio is known to be a phylogenetically structured trait, for instance, basal angiosperms have a smaller E:S ratio than more derived species. Therefore, it is possible that the measure of E:S ratio does not have much “added value” over the variation explained by the axes of phylogenetic position (Finch-Savage and Leubner-

Metzger 2006; Linkies et al. 2010). However, in my experiment, while E:S traits showed 70 significant phylogenetic structure, they were less phylogenetically structured than other traits – like seed width – that were still retained in the final models. A second explanation is that using x-ray methods I may have been unable to completely distinguish between embryo and endosperm in my E:S measurements. Therefore, for species with underdeveloped embryos, and copious endosperm, I may have overestimated the true size of the embryo and the E:S ratio.

However, use of an x-ray method was needed for my experimental design, as embryo excision would have been destructive. Furthermore, a similar method has been used in studying embryos and germination rates in crop species (Gomes-Junior, Chiquito and Marcos-Filho 2013; Rocha,

Silva and Cicero 2014). While E:S measurements were less predictive of germination in my overall analysis than other traits, I found that they were the traits most often predictive of germination in analyses of germination of individual seeds (Appendix 2).

Phylogenetic information was necessary for understanding differences in germination even though the traits studied showed phylogenetic structure. Phylogenetic information likely served as a proxy for trait information that remained unmeasured, but that was important for understanding germination response (e.g., seed coat thickness or water content). Furthermore, phylogenetic information is integrative over evolutionary history and can often be a stronger predictor of ecologically relevant information than traits alone (Pearse and Hipp 2009; Srivastava et al. 2012; Hipp et al. 2015). Seed traits, dormancy patterns, and germination responses have ancient origins, and therefore phylogenetic relationships remain an important part of understanding how they vary (Forbis, Floyd and de Queiroz 2002; Donohue et al. 2010; Linkies et al. 2010). 71

While I uncovered effects of seed traits, phylogeny, and germination pre-treatment on germination response, there are opportunities to broaden this analysis to include other relevant considerations. For example, I did not vary germination temperatures, cold stratification lengths or gibberellic acid concentrations. Varying these pre-treatments would allow us to better understand dormancy status and dormancy-break requirements for the tested species. There are also opportunities for understanding how traits and phylogeny impact the range of possible germination responses (e.g., germination tolerance range), which may have implications for ecological restoration and predicting plant regeneration under climate change (Barak et al. 2015;

Jiménez-Alfaro et al. 2016, Seglias 2017). Furthermore, my experimental design accounted for individual differences, but I used only one seed source per species, all of which came from a commercial nursery. There are known population-level, intraspecific differences in seed traits

(e.g., Völler et al. 2012) and timing of seed germination (e.g., Meyer, Kitchen and Carlson

1995). While I detected intraspecific differences in seed traits and germination response (Figure

3.3), I know that explicitly addressing population-level differences would provide additional nuance to my questions (Violle et al. 2009; Völler et al. 2012, Seglias 2017)

Larson et al (2015) advocated for a trait-based framework for understanding community assembly that can inform decision making for restoration. However, these authors note that a constraint to a trait-based approach is that only a fraction of traits that impact establishment are known and understood by researchers and managers. Here I demonstrate that seed traits – beyond seed mass – are predictors of germination response for a suite of species commonly seeded to establish restored prairie plant communities, and furthermore that phylogeny is also necessary for understanding germination response. That phylogeny is also an important predictor 72 of germination indicates that there is variation in germination that was not explained by the measured traits. my findings support both the need to incorporate additional traits into the study of germination, and the use of phylogeny as a tool for understanding seed germination. In future work, I will uncover the effects of these traits past the germination stage, into plant emergence and persistence.

Acknowledgements

This work was supported by NSF awards DEB-1354426 DEB-1354551 and DBI-1461007, the

Society for Ecological Restoration Midwest Great Lakes Chapter and the Illinois Association of

Environmental Professionals. I thank Kyle Banas and Jack Pizzo from Pizzo Native Plant

Nursery for assistance, information and seeds.

CHAPTER FOUR

Shopping for a prairie: species richness, conservatism, and phylogenetic diversity of commercially available seed mixes for restoration

Summary 1. Biodiversity is an important driver of ecosystem function, and a key restoration objective.

Different facets of diversity can be linked to different components of ecosystem function.

Therefore, restoration projects may be guided by multiple, possible competing,

biodiversity objectives, as well as constraints such a cost. Seed mixes provide the raw

materials for restored plant communities, and diverse seed mixes can be used to establish

diverse communities.

2. I analyzed commercially available seed mixes for prairie restoration in terms of three

measures of biodiversity: species richness, conservatism, and phylogenetic diversity. I

compared biodiversity of commercial mixes to plant communities at extent remnant and

restored prairies. Lastly, as cost is a major constraint for restoration, I tested whether seed

mix price predicted biodiversity.

3. Commercial seed mixes were less diverse than remnant prairies in terms of species

richness and phylogenetic diversity, but were similar in these diversity metrics to extant

restored prairie communities. On the other hand, commercial mixes had higher mean

conservatism than remnant or restored prairies. Seed mix price was predictive of

biodiversity for most of the metrics I studied, that is, more expensive mixes had higher

measures of diversity. 74

4. While commercial seed mixes do not have the potential to restore communities that are as

diverse as prairie remnants, they are similar in diversity to restored prairies, and their

components of diversity track with seed mix price. Adding species from “missing

branches” – clades that are found in remnant prairies but not in seed mixes for restoration

would increase the species richness, conservatism, and phylogenetic diversity of mixes

and restored prairies. In addition, computational methods like machine learning have the

potential to design seed mixes that meet multiple biodiversity simultaneously while

operating under realistic cost constraints.

Introduction

Biological diversity is linked to many valuable ecosystem functions and services (Rey

Benayas et al. 2009). For this reason, biodiversity is a frequent objective of ecological restoration

(Wortley, Hero and Howes 2013). Plant communities that are more diverse are more productive and stable, more resistant to invasive weeds, and support a greater diversity of insects and other animals (Balvanera et al. 2006). But biodiversity can be described in many ways, from counts of species to intensive measurements of functional traits (Pavoine and Bonsall 2011). Different facets of biodiversity may be drivers of different components of ecosystem function (Díaz et al.

2013). Individual restoration projects may therefore have multiple objectives related to different facets of biodiversity and ecosystem function. Seed mixes are the raw ingredients that build restored plant communities, so designing effective mixes is a necessary step for building diverse, functional communities that will meet restoration objectives. 75

There have been calls to increase biodiversity of seed mixes for restoration, as seed mixes, and the restored plant communities they help to build, are frequently less diverse than reference communities (Polley et al. 2005; Hansen and Gibson 2014; Harmon-Threatt and Hendrix 2015;

Barak et al. 2017). Plant species richness is a common goal of restoration (Wortley, Hero and

Howes 2013). Richness is desirable both to mimic species numbers at remnant sites, and because of the link between species richness and ecosystem functions, including community stability, nutrient cycling, and support for animal communities (Balvanera et al. 2006; Brudvig 2011;

Rowe and Holland 2013; Wortley, Hero and Howes 2013).

Beyond species richness, managers may target species of special interest, such as those with a high coefficient of conservatism (C value), a measure that reflects fidelity to high quality habitats and sensitivity to disturbance (Swink and Wilhelm 1994; Middleton et al. 2010). Mean

C values are assigned based on expert opinion (Swink and Wilhelm 1994). Because of this, their objectivity and value as ecological metrics have been questioned (e.g., Bowles and Jones 2006).

However, C values have been shown to be indicative of plant functional traits (Bauer et al. in review) and plant assemblage composition (i.e., species with similar C values are more likely to co-occur, Matthews et al. 2015). Mean C is used as a measure of habitat quality, to evaluate restoration outcomes, and to monitor changes in plant communities over time (Taft, Hauser and

Robertson 2006; Hansen and Gibson 2014)

Phylogenetic diversity, a measure that reflects evolutionary distance among species in a community, has also been called for in restoration to increase ecosystem multifunctionality, and similarity to remnant habitats (Hipp et al. 2015; Larkin et al. 2015; Barak et al. 2016).

Phylogenetic diversity is also an important driver of ecosystem functions that are themselves key 76 restoration objectives, like plant productivity, community stability, diversity at higher trophic levels, and invasion resistance (Cadotte, Cardinale and Oakley 2008; Davies, Cavender-Bares and Deacon 2011; Cadotte, Dinnage and Tilman 2012; Dinnage et al. 2012; Li et al. 2015; Lind et al. 2015). Phylogenetic diversity is often a stronger predictor of ecosystem function than species richness alone (Srivastava et al. 2012). Based on recent work, it appears that restored communities may consistently have lower phylogenetic diversity than reference ecosystems, perhaps due to greater influence of disturbance in restorations (Dinnage 2009; Turley and

Brudvig 2016; Barak et al. 2017).

Trade-offs can exist in seed mix design between multiple biodiversity and restoration goals, and between seed costs and ecological goals (Rowe and Holland 2013; Wilkerson et al. 2014).

Economic considerations are strong drivers of restoration decision making, and these considerations may impact the likelihood of meeting restoration objectives, and can have long lasting effects on restored communities (Rowe 2010).

Here, I assess the range of diversity objectives met by commercially available seed mixes using tallgrass prairie restoration as the model system. Grasslands are a major focus of global restoration efforts, and the North American tallgrass prairie is a highly endangered ecosystem

(Hoekstra et al. 2005). Attempts to re-create these habitats over the past 75+ years were the origin of the science of ecological restoration (Cottam and Wilson 1966). I assessed commercially available seed mixes (representing typical restoration inputs) in terms of three restoration objectives: species richness, conservatism, and phylogenetic diversity. I then compared the commercially available seed mixes with plant community metrics for restored prairies (typical restoration outputs) and remnant prairies (reference states targeted by 77 restoration). As cost is a major constraint for restoration managers, I also assessed how each diversity component was predicted by the cost of the seed mix (Rowe 2010). Finally, when a goal of seed mix design is to meet multiple, possibly competing biodiversity objectives, computational tools can aid in providing solutions to these complex restoration problems

(M’Gonigle et al. 2016). I discuss future work that will use computational approaches to attempt to optimize seed mixes to meet multiple biodiversity objectives while operating under realistic cost constraints. Such an approach may provide an opportunity to improve upon currently available seed mixes for restoration efforts.

Methods

Commercially available mixes

I queried Google, using the terms, “tallgrass prairie seed mix,” “tallgrass prairie mix,” and

“native prairie seed mix,” to identify companies selling prairie seed mixes. For each company I collected the following data (where available): name, location, year established, and seed sources

(i.e., the state where seed is grown). I sought to select five seed mixes for each company; however, some companies only had four prairie mixes available. For each seed mix, I recorded: name and description, species included, species composition by weight, cost (per pound), seeding rate (pounds per acre), and the targeted soil type for the mix. If these data were not available on the company’s website, I contacted representatives from the company, making at least two attempts to contact each company via email. Following these attempts, I excluded a company and its seed mixes from analysis if I could not obtain the necessary data. I selected only seed mixes that included both grasses and forbs. In cases where companies sold grass and forb seeds in separate, co-marketed mixes, I composited these mixes into single prairie mixes. I 78 resolved taxonomy across seed mixes by first translating from common to scientific names (if necessary) and then standardizing scientific names using the Taxonomic Name Resolution

Service (Boyle et al. 2013). All seed mix and price information is from the year 2015.

Remnant and restored prairies

I compared the commercially available seed mixes to existing plant communities from remnant and restored prairies in Illinois, for which I had robust field data. I used plant community data from 41 remnants that include some of the highest quality reference sites in the tallgrass prairie region. These reference sites were sampled in 2001 (Bowles and Jones 2013). I also used data from 19 restored prairies in the same region; restorations were initiated between 1998 and 2012, and sampled in 2015 (Barak et al. 2017).

Phylogenetic tree

I constructed a community phylogeny of 632 species, which comprised the complete species pool encompassing all commercial seed mixes, restored prairies, and reference prairies. The tree was developed by modifying a published tree of 32,223 plant taxa (Zanne et al. 2014). The

Zanne et al. tree was constructed using GenBank sequences for seven gene regions (18S rDNA,

26S rDNA, ITS, matK, rbcL, atpB, and trnL-F) using maximum likelihood for tree estimation. I grafted taxa not present in the Zanne et al. tree (272), and pruned taxa absent from the dataset, in

R version 3.1.3 (R Development Core Team 2013) using the weldTaxa and make.matandtree functions in the Morton R project (A. Hipp, https://github.com/andrew-hipp/morton).

Biodiversity measures

I calculated three biodiversity measures for plant assemblages represented in seed mixes and observed in restored and remnant prairies: (1) Taxonomic diversity was indicated using species richness. (2) Mean C value is the average coefficient of conservatism for all species in a community. Each species was assigned a C value from Swink and Wilhem (1994). For species that were not found in Swink and Wilhelm, I obtained a C value from the universal FQA calculator (Freyman, Masters and Packard 2016), using values from the 2014 Chicago region database. C values are expert opinion scores intended to reflect habitat fidelity and sensitivity to disturbance, and range from 0 – 10. Low-fidelity, disturbance-tolerant species have low C values, while more conservative species have higher values. Nonnative species are not assigned

C values. A plant community with a value for mean C of 3.5 or above is thought to have

“sufficient floristic quality to be at least of marginal natural area quality,” while a mean C of 4.5 or above indicates, “almost certain…natural area potential” (Swink and Wilhelm 1994). This baseline has been used to evaluate restoration outcomes (Hansen and Gibson 2014). (3)

Phylogenetic diversity was defined as mean pairwise phylogenetic distance (MPD) – the mean phylogenetic distance between every pair of species co-occurring in a mix or sampled site. MPD was calculated from a phylogenetic distance matrix of all species included in the analyses, using the Picante package in R (Kembel et al. 2010), Version 3.3.1 (R development core team 2013).

In comparing commercial seed mixes to remnant and restored prairies, I used metrics based on presence/absence for species richness, mean C, and MPD. This is because there is not a clear means to compare species’ relative abundances based on seed weight (the relevant measure for mixes) vs. frequency of occurrence (the data available for sampled prairies). However, analyses 80 that focused on price and diversity included only commercial mixes, thus I was able to calculate abundance-weighted measures (using seed weight) for taxonomic diversity (Shannon Wiener

Index), mean C (abundance-weighted mean C), and phylogenetic diversity (abundance-weighted

MPD).

Statistical analyses were conducted using R version 3.1.3 (R Development Core Team 2013). I compared the commercially available mixes to plant community data from remnant and restored prairies using linear mixed effects models with community type (i.e., remnant, restored, commercial mix) as the predictor variable and diversity metrics (SR, mean C, and MPD) as response variables using the nlme package in R (Pinheiro et al. 2016). The random error term company seed company for commercial mixes and site ID for prairies. I tested for differences between commercial mixes and remnant and restored prairies by performing Tukey’s honest significant difference (HSD) tests. Using only the seed mix data, I used linear mixed effects models (with seed company as a random effect) to test whether diversity metrics significantly differed with seed mix price. To evaluate these models I calculated both marginal and conditional R2 values as in Nakagawa and Schielzeth (2013). Marginal R2 is a measure of variance explained by the fixed effects (here, price), while conditional R2 is a measure of variance explained by both fixed factors (price) and random factors (seed company). Price per pound data were log-transformed to better approximate a normal distribution.

Results

I obtained data for 68 seed mixes from 14 native seed companies in five states: Illinois, Indiana,

Iowa, Minnesota, and Wisconsin (Table 4.1). I excluded an additional 4 companies because I could not obtain all necessary data. In total, the 68 mixes contained 215 species from 36 families. 81

Mean species richness of the commercially available seed mixes was 34.3 ± 12.6 species (± 1

S.D.), with a range of 14 – 91 species. Species richness significantly varied by “community” type (F2,71 = 18.23, P < 0.0001, Figure 4.1), with commercial mixes having lower richness than both remnant (P < 0.0001) and restored (P = 0.0002) prairies (Tukey’s HSD). When non-native species were excluded from the remnant and restored communities, species richness still varied by community type (F2,71 = 9.03, P = 0.0001) and commercial seed mixes still had lower richness than remnant prairies (P = 0.0003), but richness of restored prairies and commercial seed mixes did not differ (P = 0.94).

Mean C also differed by community type (F = 45.162, P < 0.0001). In this case, commercial seed mixes had higher mean C than both remnants (P = 0.018) and restorations (P < 0.001).

Phylogenetic diversity (MPD) was influenced by site type (F2,71 = 16.85, P < 0.0001).

Phylogenetic diversity of commercial mixes was significantly lower than remnant prairies (P <

0.0001), but did not differ from restored prairies (P = 0.97). The same patterns held when nonnative species were excluded from analysis.

Seed mix prices spanned an order of magnitude, from $22 to $245 USD per pound (mean ± S.D.:

$91 ± 48). Price per pound positively predicted seed mix species richness, Shannon Wiener diversity, and mean C. Also, while price was not predictive of MPD based on presence/absence measures, abundance-weighted MPD was positively correlated with price (Table 4.2, Figure 4.2).

There were seed mixes that appeared to be outliers for species richness and abundance-weighted

MPD (1 high and 2 low outliers, respectively). I repeated price-diversity analyses with these mixes removed; results did not qualitatively differ, and are not shown. 82

Discussion

The commercial seed mixes I found lacked the raw ingredients for creating restored plant communities that match the levels of biodiversity seen in remnant prairies, i.e., the seed mixes themselves were less diverse than remnants by all metrics evaluated. These differences are compounded by the fact that not all species included in seed mixes will successfully germinate, establish, and persist at a given restoration site (Hillhouse and Zedler 2011; Grman et al. 2015;

Barak et al. 2017). Therefore, the “effective” biodiversity of any restoration seed mix is likely to be considerably lower than the potential diversity it represents; results should be interpreted in light of this caveat.

That said, matching the diversity found in remnant prairies is a very high standard. Commercial seed mixes can be less stringently evaluated by comparing their attributes to the levels of diversity typically seen in extant restorations. By this standard, commercial mixes were similar to restored communities by several diversity measures, including native species richness and

MPD. In terms of taxonomic composition of the seed mixes, some mixes contained species that are not considered native, like Medicago sativa (alfalfa), and Achillea millefolium (common yarrow) – likely as forage additions or to encourage rapid establishment of cover. These findings underscore the necessity for restoration mangers to assess composition of commercially available mixes with consideration for their own biodiversity goals.

Commercial mixes had higher mean C values than either remnant or restored prairies. This could reflect a loss of conservative species during establishment. I found that C values of restored plant communities were consistently lower than their seed mixes (R. Barak et al. 2017, unpublished data), and mean C tends to decrease with the age of a restoration (Hansen and Gibson 2014). 83

This discrepancy could also be explained by the finding that dry-prairie communities tend to have higher mean C values than mesic prairies (Bowles and Jones 2006), and I only surveyed mesic restored prairies. However, when I excluded dry prairie seed mixes, mean C was still higher overall in seed mixes than in restorations. Regardless, all commercial seed mixes, and all remnant prairies (though not all restorations), met the benchmark set by Swink and Wilhelm

(1994) of mean C > 3.5.

Phylogenetic diversity did not differ between restored prairies and commercial seed mixes, but both were less phylogenetically diverse than remnant prairies. In previous work I noted “missing branches” – clades that are present in remnant prairies, but missing from restoration seed mixes

(Barak et al. 2017). I noted several species and families missing from restoration seed mixes that could be targeted to increase the richness and phylogenetic diversity of restored prairies while also meeting species-specific conservation goals. These missing branches were not found in the commercial seed mixes I identified. However, some of these species were commercially available, including multiple Lilium spp. and Comandra umbellata. Other species, like native orchid species, were not readily commercially available. Sourcing these species through specialty growers or by hand-collecting wild seeds would considerably increase seed costs, and would need to be weighed against other restoration objectives (Frischie and Rowe 2012).

Harmon-Threatt and Hendrix (2015) demonstrated that by adding just four additional species to commercial mixes, managers could support higher diversity and abundance of pollinators.

Similarly, I suggest that by supplementing plantings with species from missing branches, managers could simultaneously increase species richness, conservatism, and phylogenetic diversity of commercial mixes. 84

Price was predictive of seed mix biodiversity for most of the metrics I studied, that is, more expensive mixes had higher measures of diversity. However, for phylogenetic diversity, price was only predictive of MPD when it was abundance-weighted, but not when calculated based on presence/absence. This pattern is consistent with less-expensive mixes including species that represent distinct evolutionary lineages, but containing relatively small amounts of these species.

An important issue to consider is the effective biodiversity imparted by species sown at very low densities. Are these species likely to persist in restored plant communities? Previous work has shown that species seeded at low densities do not establish in restorations (Goldblum et al. 2013;

Grman et al. 2015). Thus, though present in seed mixes, some species may be at insufficient densities to become established. However, it is difficult to predict establishment of particular species based on diversity and composition of the seed mix. Establishment of species sown at low densities is likely to depend on several factors, including their time to maturity, dispersal ability, and other species-specific traits (Pakeman, Pywell and Wells 2002).

Future directions

Cost is a major constraint in the decision-making process for restoration managers, and obtaining seed for restoration is one of the most expensive components of the entire restoration process

(Frischie and Rowe 2012). Managers must face tradeoffs that involve balancing multiple biodiversity considerations with cost constraints. Meeting biodiversity goals at lower costs could allow managers to increase the land area to be restored (Rowe 2010; Wilkerson et al. 2014).

Expert opinion is an extremely important component of seed mix design. For example, restoration mangers tend to more frequently plant species with higher probabilities of establishment, increasing the likelihood of building a diverse plant community and meeting 85 restoration goals (Grman et al. 2015). However, as restoration problems become more complex and involve multiple objectives, computational tools can complement expert opinion to aid decision making (M’Gonigle et al. 2016).

In future work, I will use a machine-learning approach to develop optimal mixes from a pool of commercially available prairie plant species. These mixes will be created using a genetic algorithm, a machine-learning optimization based on the principles of natural selection (Olden,

Lawler and Poff 2008; M’Gonigle et al. 2016). The genetic algorithm will be used to generate seed mixes based on an objective function optimizing across four biodiversity components: species richness, phylogenetic diversity, coefficient of conservatism, and phenological diversity

(bloom time variance). I will include bloom time diversity as it is important for supporting pollinator communities and the pollination services they provide. Commercial seed mixes are lacking in phenological diversity. In particular, early-blooming forb species tend to be missing from commercial seed mixes (Havens and Vitt 2016).

I will use the genetic algorithm to generate seed mixes at multiple price points, ranging from

~$200 to ~$2,000 per 10-pound seed mix, reflecting the range of prices observed for commercial mixes. Candidate species for the optimal mixes will consist of the 215 plant species found in the commercial seed mixes. I will consider this approach useful if I can use it to design mixes that outperform currently commercially available seed mixes with respect to biodiversity objectives and cost.

DISSERTATION SUMMARY AND IMPLICATIONS FOR MANAGEMENT

Tallgrass prairie is one of the most endangered habitats on earth. Thankfully, prairie restoration can be used to gain back lost prairie acreage. Managers restore prairie by planting seeds of native prairie species, using the small patches of existing remnant prairie as a source of inspiration (and sometimes seeds). In addition to using extant remnants as baselines for developing restoration objectives, managers can use historical ecological data – history recorded by humans – as well as archeological, paleoecological, and evolutionary information.

Historical ecology involves finding clues in modern ecosystems to reconstruct their past. For instance, scars on long-lived trees can be used to reconstruct hundreds of years of fire history.

Fossilized pollen in lake sediment can be used to study changes in plant communities over tens of thousands of years, and track the arrival and trajectory of colonizing species. In habitats that have been managed by humans for many years (like prairies, which have been burned by humans for thousands of years!), archeological and paleoecological evidence can be used to uncover the role of human influence on plant communities. Historical ecological data can be used to develop restoration objectives, direct monitoring and management programs, and develop restored communities that are more resilient to future change. A limitation of this approach is that recorded history is not available for many restoration sites, and collecting additional data based on paleoecology can be costly. To address these limitations, managers and researchers could collaborate to reconstruct site history. Also, historical ecological data is being collected into accessible databases (see Brewer et al. 2012 for examples of databases). 87

Researchers and managers can also use evolutionary history to understand contemporary species and communities, without digging deep into lakes. All species that have ever lived on earth are related to one another though the evolutionary tree of life. The position of species on this tree of life can inform restoration by providing clues as to how that species might “behave” – germinate, emerge, establish, grow, compete, reproduce, and persist in a restored community. In addition, the degree to which co-occurring plant species span the evolutionary tree of life can predict key ecosystem functions. In this dissertation, I used several aspects of phylogenetic ecology to understand germination and persistence of species commonly seeded in prairie restorations, compare remnant and restored prairies, and assess the diversity of commercially available seed mixes for restoration.

A persistent challenge of restoration science is being able to predict restoration outcomes (e.g., the resulting plant community) from restoration inputs (e.g., seeds used). Plant species’ functional traits are often used to attempt to understand how communities assemble. However, vegetative plant traits of mature plants are typically used to make these predictions. This is a limitation in the case of restored plant communities, most of which are started from seed. Seed traits may be more important than vegetative traits for understanding assembly of restored communities. I studied germination in the lab to assess how predictive traits of seeds, as well as species’ position on the evolutionary tree of life (i.e., phylogenetic position), are of germination timing. I found that long, flat, and pointy seeds germinated faster than rounder seeds. I also found that phylogenetic position was an important predictor of germination timing –even after accounting for measured traits. In future work, I will test whether these patterns hold true by studying emergence and persistence of these species in an ongoing field restoration experiment. 88

The fact that phylogenetic position was important for understanding germination supports the idea that phylogenetic position can integrate across many traits – known and unknown, measured and unmeasured. There are numerous traits that interact to determine germination timing, only some of which I directly measured. For example, seed coat thickness and seed water content were not measured, but are known to influence germination. Knowing the phylogenetic position of a species may provide a shorthand for predicting its biology and ecological performance in restored communities.

Just as phylogenetic position can provide a shorthand for predicting how individual plant species will perform in restoration, the phylogenetic diversity of plant assemblages may do the same for predicting performance and function of whole communities. Phylogenetic diversity refers to evolutionary relatedness among co-occurring species. A plant community with high phylogenetic diversity is characterized by distantly related species living together, for example, species from many different plant families, while a community with low phylogenetic diversity might comprise species from only a few families. It has been experimentally demonstrated that communities with higher phylogenetic diversity are more productive and stable, support more pollinators and other insects, and are more resistant to invasion than those with lower phylogenetic diversity. These ecosystem functions are also important restoration objectives. I compared plant taxonomic and phylogenetic diversity of restored prairies to high-quality remnant prairies to determine if there were diversity deficits in restored prairies that might be remedied by management activities. I found that the numbers of species did not significantly differ between remnants and restorations. However, remnant prairies were richer in native 89 species. Remnant prairies also had higher phylogenetic diversity (based on multiple measures) at both small (neighborhood/plot) and large (site) scales.

Based on this research, I concluded that restored prairies are phylogenetically depauperate compared to remnants. There may be opportunities to increase phylogenetic diversity of restored prairies by seeding phylogenetically diverse seed mixes. To evaluate this, I tested whether seed mix diversity was predictive of plant community diversity. I found that seed mix diversity was predictive of the diversity of intentionally planted species, but not of the community as a whole

(that is, colonization by volunteer native species and non-native species reduced the effects of seed mix diversity on plant community diversity). I also found significant phylogenetic signal in terms of which species were included in restoration seed mixes. That is, the species in the mixes were more closely related to one another than would be expected by chance. I identified “missing branches” – clades of plants that were found in remnant prairies but were not included in restoration seed mixes. Key missing branches comprised entire families: Liliaceae, Orchidaceae,

Convolvulaceae, and Polygalaceae. These families include species that are of special concern in

Illinois prairies and hemiparasitic species (like Comandra umbellatea, bastard toadflax, and

Cuscuta glomeratea, rope dodder), which have been shown to be beneficial in restoration by reducing the abundance of dominant species.

Planting species from missing branches in restored prairies could advance multiple biodiversity goals. However, this is not an easily implementable recommendation given that missing branches represent species that may be difficult to obtain, afford, grow, or establish. In other words, there are reasons – both biological and practical – why these branches are missing. In the short term, these hurdles can be addressed by planting missing species as plugs, rather than seeds. Adding 90 missing branches (especially from clades distantly related to those already present) would be an efficient means to increase restored communities’ phylogenetic diversity. Planting species that are missing from restoration seed mixes but found in remnants would also increase similarity of restored prairies to remnants in terms of species richness, phylogenetic diversity, and overall composition. When missing species are of special concern, their addition can further advance species-specific conservation goals. Thus, identifying and trying to fill these phylogenetic gaps could advance species conservation goals while also increasing multiple measures of community biodiversity, and building restored prairies that are more similar to remnants.

Lastly, knowing that seed mix diversity is at least partially predictive of plant community diversity, I studied commercially available seed mixes for prairie restoration. I wanted to know whether seed mixes currently commercially available have the potential to establish diverse prairies, so I compared them to plant communities at remnant and restored prairies. I found that seed mixes had lower species richness and phylogenetic diversity than remnant prairies, but were similar in these measures to restored prairies. Interestingly, I found that commercial mixes actually had higher mean coefficients of conservatism (mean C) than both remnant and restored prairies. This may indicate that the most conservative species included in seed mixes fail to establish in restored sites. Analysis of seed mix cost and diversity metrics showed that there are greater ecological returns with increasing financial investment: more-expensive mixes had higher taxonomic diversity, conservatism, and phylogenetic diversity.

In future work, I will extend this study of commercial seed mixes by developing seed mix design tools using machine-learning algorithms, with the goal of developing seed mixes that optimize multiple biodiversity components while minimizing cost. These mixes will be created using a 91 genetic algorithm, a machine-learning optimization based on the principles of natural selection.

The genetic algorithm will be used to generate seed mixes that optimize four biodiversity components: species richness, phylogenetic diversity, conservatism, and phenological diversity

(bloom time variance, which is important for supporting pollinator communities and the pollination services they provide). I will consider this approach useful if it can design mixes that outperform currently available commercial mixes with respect to biodiversity objectives and cost.

Implications for management

 Historical ecological data can be useful for informing contemporary restoration efforts. Integrating historic ecology into restoration will require partnerships between researchers and managers.  Managers can increase the phylogenetic diversity of restored prairies by planting phylogenetically diverse seed mixes, and seeding or planting species from missing branches. This may make restored prairies more similar to remnants in terms of their composition, diversity, and functioning.  Planting species from missing branches that are also of conservation concern in restored prairies can support species-specific conservation goals while also meeting overall biodiversity goals.  Seed mix diversity is predictive of plant community diversity of planted species. This is important when restoration objectives include specific diversity goals for planted species.  Commercially available seed mixes do not have the potential to establish restored prairies that have comparable biodiversity to remnant prairies. However, they have similar diversity to existing restored prairies.  In general, commercial seed mixes show a return on investment. That is, increased cost is linked to increased biodiversity.

TABLES AND FIGURES

Table 1.1. Restoration questions that could be addressed by data from different temporal scales.

Restoration Temporal scale Stage Contemporary sites Recorded history Archeological/ Evolutionary Paleoecological

Goal Setting

Habitat type What is the habitat What was the What was the range like in a reference habitat like in the of past habitat types site? What is the recent past? in the area? What successional stage factors drove habitat relative to shifts? Was human restoration goals? activity influential?

Species What species are What was the What taxa and/or What is the selection found in reference species communities were desired level of sites? Is it feasible composition in the most common and/or phylogenetic to replicate these recent past? Is it most stable in the diversity for assemblages in feasible to past? restored restored sites? replicate these communities? assemblages in restored sites?

Identifying What factors limit Does past human Where do current Can evolutionary constraints restoration use and/or conditions fit within theory guide to effectiveness? Are disturbance limit the historic range of restoration in the restoration they present in options for variability? absence of reference sites? restoration? complete ecological information?

Management

Invasive Are invasive When did invasive When did invasive How closely species species present in species arrive? species arrive in the related is the reference sites? Was their arrival invader to the associated with 93

How should they changes in area? Did they alter resident be managed? community ecological dynamics? community? composition?

Disturbance What disturbance What was the What were historic How does regime is required level of levels of disturbance? disturbance alter to maintain disturbance in the How variable was the community reference sites? Is recent past? How disturbance regime phylogenetic the reference did humans over time? Did structure? Is there disturbance regime influence the humans influence the a relationship feasible to disturbance frequency or intensity between implement in regime? of disturbance? phylogenetic and restored sites? trait diversity? Is disturbance limiting the types of species that can persist in restored sites?

Monitoring

Are conditions at Are conditions in Does the restored Is community the restored site the restored site site have analogs in phylogenetic similar to those at similar to those of past communities structure the reference site the recent past? and habitats? changing over in terms of time and with diversity, management? Is composition, and contemporary functioning? evolution influencing ecological outcomes?

Figure 1.1. Historical timescales and their potential contributions to ecological restoration. Thick lines indicate the main use of each type of data currently. Thin lines indicate additional uses (or potential uses) of historical information in informing restoration.

Figure 1.2. Vegetation changes at 100-yr intervals over the past 1200 years in Wisconsin sand plain. Inset map (top left) shows the location of the northwestern Wisconsin sand plain. Maps show vegetation changes at 100-yr intervals over the past 1200 years (BP = years before present, where present is 1950 AD). Symbols indicate the vegetation community represented by pollen at each site over time. Dashed lines in 1100 BP map indicate sites surrounded by coarse sandy soil and few ﬁre breaks (northern) and abundant ﬁre breaks (south). Shading indicates three vegetation regions described from Public Land Survey data (Radeloff et al. 1999); white indicates jack pine forests and barrens, light gray indicates closed-canopy mixed pine forests, and dark gray indicates oak and pine savannas. Updated from Hotchkiss et al. (2007) using sites from Tweiten et al. (2015).

Figure 1.3. Conceptual framework for the placement of traits according to their ecological and evolutionary structure. Based on ﬁgure 3B of Cavender-Bares et al. (2006) and termed a ﬁngerprint regression by Pearse et al. 2015. In each quadrant of the space, a phylogeny is plotted with a trait (represented by the size of the circles) and likelihood of species coexisting represented by the color of the circles (one community shown in red and another in black). Fingerprint regressions could be used to develop restoration goals based on reference communities from the present or the past as well as to monitor changes in the biodiversity of restored communities over time.

Figure 2.1. Site and plot diversity metrics for species richness (a, b), species richness with non- native species excluded (c, d) and standard effect sizes for mean pairwise distance (MPD) (e, f) and mean nearest taxon distance (MNTD) (g, h). Boxplot centers represent median values and hinges the 25th and 75th percentiles. F values (all d.f. 1,58): (a) 0.332, (b) 0.431, (c) 4.385, (d) 7.728, (e) 22.556, (f) 15.106, (g) 39.194, (h) 10.827. 98

Figure 2.2. Non-metric dimensional scaling ordinations of remnant and restored prairies and seed mixes, based on community taxonomic (a, d, g) and phylogenetic ordinations: mean pairwise distance (MPD) (b, e, h) and mean nearest taxon distance (MNTD)(c, f, i). In d-i lines connect each restored community to their planted seed mix. Number of dimensions K = 2 unless indicated by ‘*’, then K = 3. Ellipses represent one standard deviation.

Figure 2.3. Taxonomic and phylogenetic diversity of seed mixes and resultant prairie communities for only species planted at each site, for species richness (a), standard effect sizes of mean pairwise distance (b) and mean nearest taxon distance (c). Results from linear regression. Values for all species (not shown, species richness (SR): R2 = - 0.055, P = 0.817, SES.MPD: R2 = - 0.040, P = 0.590, SES.MNTD: R2 = 0.087, P = 0.118).

100

Figure 2.4. Planted (green) and unplanted (grey) species are indicated on the phylogeny containing all species found at remnant and restored sites and seed mixes (n = 589). Planted species are phylogenetically clustered relative to a random null model (D = 0.65, P < 0.0001). Crass./Sax. = Crassulaceae/Saxifragaceae.

101

Table 3.1. Species included in the experiment, and final percent germination under three pre- treatments: cold stratification, gibberellic acid and an untreated control

Species Family Cold Gibberellic Control stratified acid

Andropogon gerardii Poaceae 18.75 56.25 78.13

Anemone cylindrica Ranunculaceae 84.38 87.50 87.50

Asclepias syriaca Apocynaceae 43.75 15.63 15.63

Asclepias verticillata Apocynaceae 59.38 21.88 18.75

Bromus kalmii Poaceae 87.50 75.00 90.63

Carex bicknellii Cyperaceae 56.25 0 0

Carex brevior Cyperaceae 37.50 0 0

Dalea candida Fabaceae 62.50 93.75 78.13

Dalea purpurea Fabaceae 75.00 43.75 75.00

Desmodium canadense Fabaceae 84.38 68.75 59.38

Desmodium illinoense Fabaceae 31.25 40.63 28.13

Eryngium yuccifolium Apiaceae 84.38 62.50 40.63

Euphorbia corollata Euphorbiaceae 84.38 0.00 3.13

Liatris scariosa Asteraceae 62.50 15.63 3.13

Liatris spicata Asteraceae 90.63 21.88 6.25

Maianthemum racemosum Asparagaceae 0 0 0

Monarda bradburiana Lamiaceae 40.63 93.75 12.50

Monarda fistulosa Lamiaceae 59.38 90.63 59.38

Panicum virgatum Poaceae 68.75 18.75 12.50

Penstemon digitalis Plantaginaceae 78.13 0 0

Polemonium reptans Polemoniaceae 3.13 25.00 6.25 102

Rudbeckia hirta Asteraceae 59.38 84.38 75.00

Schizachyrium scoparium Poaceae 56.25 71.88 71.88

Sisyrinchium angustifolium Iridaceae 3.13 0 0

Solidago rigida Asteraceae 40.63 3.13 12.50

Sporobolus heterolepis Poaceae 15.63 21.88 21.88

Symphyotrichum laeve Asteraceae 59.38 59.38 62.50

Symphyotrichum novae-angliae Asteraceae 65.63 71.88 59.38

Thalictrum dasycarpum Ranunculaceae 93.75 56.25 0

Tradescantia ohiensis Commelinaceae 68.75 0 12.50

Vernonia gigantea Asteraceae 75.00 25.00 9.38

Zizia aptera Apiaceae 50.00 0 0

103

Table 3.2. Best models of time-to-germination ranked by Akaike information criterion (AIC) for 32 prairie species. K is the number of factors in the model, ∆AIC is the difference in AIC between each model and the model with the lowest AIC, w is the model weight and Cw is the cumulative model weight. Shown are all models with ∆AIC ≤ 4. Treat. = treatment, V1 – V4 = st th the 1 – 4 phylogenetic axes. ESwidth = E:S ratio measured by width, ESwidth = E:S ratio measured by length, ESarea = E:S ratio measured by area, L = length, W = width, H = height and VS = shape, measured as the variance between L, W and H.

Model factors K AIC ∆AIC W Cw R2

Treat. + V1 + V2 + V3 + V4 + 12 19356.02 0.00 0.37 0.37 0.197

ESwidth + ESarea + L + W + H + Mass

Treat. + V1 + V2 + V3 + V4 + 11 19357.06 1.04 0.22 0.60 0.197

ESwidth + L + W + H + Mass

Treat. + V1 + V2 + V3 + V4 + 13 19357.38 1.36 0.19 0.78 0.198

ESwidth + ESarea + L + W + H +

Mass + ESlength

Treat. + V1 + V2 + V3 + V4 + 13 19357.92 1.90 0.14 0.93 0.197

ESwidth + ESarea + L + W + H + Mass + VS

Treat. + V1 + V2 + V3 + V4 + 14 19359.33 3.30 0.07 1.00 0.198

ESwidth + ESarea + L + W + H +

Mass + VS + ESlength

~1 (Intercept-only model) 0 20007.51 651.48 0.00 1.00

104

Table 3.3. Model-averaged estimate, standard error, and 95% confidence interval (CRI) for all parameters in best fitting models (∆AIC ≤ 4) for 32 prairie species. An asterisk (*) indicates a variable that was a significant contributor to the averaged model based on 95% CRI. GA = gibberellic acid treatment, CON = untreated control.

Model term Estimate SE 95% CRI

Treatment – GA* -0.5 0.07 -0.63, -0.37

Treatment – CON* -0.83 0.07 -0.97, -0.69

V1* 0.22 0.03 0.16, 0.28

V2* -0.36 0.04 -0.44, -0.27

V3* -0.22 0.03 -0.28, -0.15

V4* 0.30 0.04 0.21, 0.38

Length* 0.35 0.07 0.21, 0.49

Width* -0.33 0.09 -0.51, -0.15

Height* -0.38 0.06 -0.49, -0.27

Variance (shape) -0.08 0.1 -0.27, 0.11

ESlength 0.01 0.04 -0.07, 0.1

ESwidth* 0.11 0.04 0.03, 0.18

ESarea 0.06 0.04 -0.01, 0.13

Mass* -0.21 0.09 -0.38, -0.05

105

Table 3.4. Phylogenetic signal of measured traits. K is the observed value of phylogenetic signal relative to a Brownian motion model of evolution. P is significance of phylogenetic signal based on a randomization test with 1,000 permutations.

Trait K P

Length 0.073 0.004

Width 0.219 0.001

Height 0.076 0.005

Variance 0.118 0.001

ESlength 0.034 0.047

ESwidth 0.036 0.028

ESarea 0.030 0.042

Mass (log) 0.115 0.001

106

Figure 3.1. Ordination of phylogenetic distance matrix for 32 species in the study. Monocots are shown as green squares and dicots as purple triangles.

107

Figure 3.2. Time-to-event curves for all species by germination treatment. 108

109

Figure 3.3. Phylogenetic tree of species used in the experiment and phylogenetic distribution of trait values representing seed size (mass), shape (width) and embryo traits (ESlength). All measured traits showed significant phylogenetic signal (see Table 3.4). 110

Figure 3.4 Estimates from averaged models (table 3.2), for treatment (blue), trait (purple) and phylogenetic (green) model terms. Error bars represent 95 % confidence intervals (CRI). An asterisk (*) indicates a variable that was a significant contributor to the averaged model based on 95% CRI.

111

Table 4.1. Companies included in this study that sell seed mixes for prairie restoration.

Company State

Agrecol WI

Cardno Native Plant Nursery IN

Earth Source Inc, Heartland Restoration Services IN

EC3 Environmental Consulting Group, Inc. WI

Iowa Pheasants Forever IA

Minnesota Native Landscapes MN

Morning Sky Greenery MN

National Seed IL

Pizzo Native Plant Nursery IL

Prairie Moon MN

Prairie Nursery WI

Prairie Restorations, Inc. MN

Shooting Star Native Seeds MN

Spence Restoration Nursery IN

112

Table 4.2. Price is predictive of some biodiversity components in commercially available seed mixes for prairie restoration. Data from linear mixed-effects models with seed company as a random factor.

Diversity measure F R2 marginal R2 conditional P

Species richness 14.858 0.096 0.690 0.0003

Shannon-Wiener 24.583 0.202 0.647 <0.0001

MPD, presence-absence 0.824 0.009 0.546 0.368

MPD, abundance-weighted 60.708 0.483 0.735 <0.0001

Mean C, presence-absence 24.675 0.177 0.660 <0.0001

Mean C, abundance- 15.050 0.136 0.679 0.0003 weighted

113

Figure 4.1. Biodiversity of commercially available seed mixes and remnant and restored prairies with respect to species richness, native species richness (i.e., non-native species excluded from remnant and restored prairies), mean C, and MPD. Different letters indicate significant differences based on Tukey’s HSD tests. Boxplot centers represent median values and hinges the 25th and 75th percentiles, whiskers represent the range of the data without outliers, and outliers are shown as points.

114

Figure 4.2. Price significantly predicted biodiversity with respect to species richness, Shannon Wiener diversity, Mean C (both presence-absence and abundance-weighted), and abundance- weighted MPD, but not presence-absence MPD. Statistical results in Table 2.

115

LITERATURE CITED

Adderley, L.J., & J.C. Vamosi (2015) Species and phylogenetic heterogeneity in visitation

affects reproductive success in an island system. International Journal of Plant Science,

176, 186–196.

Alagona, P.S., J. Sandlos, & Y.F. Wiersma (2012) Past imperfect. Environmental Philosophy 9,

49–70.

Allison, S.K. (2002) When is a restoration successful? Results from a 45-year-old tallgrass

prairie restoration. Ecological Restoration, 20, 10–17.

Alstad, A.O., Damschen, E.I., Givnish, T.J., Harrington, J.A, Leach, M.K., Rogers, D.A. &

Waller, D.M. (2016) The pace of plant community change is accelerating in remnant

prairies. Science Advances, 2, e1500975.

Anderson, M.J. (2001) A new method for non-parametric multivariate analysis of variance.

Austral Ecology, 26, 32–46.

Arroyo-Rodríguez, V., J. Cavender-Bares, F. Escobar, F.P.L. Melo, M. Tabarelli, & B.A. Santos

(2012) Maintenance of tree phylogenetic diversity in a highly fragmented rain forest.

Journal of Ecology, 100, 702–711.

Asbjornsen, H., L.A. Brudvig, C.M. Mabry, C.W. Evans, & H.M. Karnitz (2005) Deﬁning

reference information for restoring ecologically rare tallgrass oak savannas in the

Midwestern United States. Journal of Forestry, 103, 345–350.

Ault, J.R. & Siqueira, S.S. (2008) Morphogenetic response of Lilium michiganense to four

auxin-type plant growth regulators in vitro. HortScience, 43, 1922–1924. 116

Baer, S.G., Heneghan, L. & Eviner, V. (2012) Applying soil ecological knowledge to restore

ecosystem services. Soil Ecology and Ecosystem Services (eds D.H. Wall, R.D. Bardgett,

V. Behan-Pelletier, J.E. Herrick, H. Jones, K. Ritz, J. Six, D.R. Strong & W.H. van der

Putten), pp. 377–393. Oxford University Press, Oxford, UK.

Balaguer, L., A. Escudero, J.F. Martín-Duque, I. Mola, & J. Aronson (2014) The historical

reference in restoration ecology: re-deﬁning a cornerstone concept. Biological

Conservation, 176, 12–20.

Balvanera, P., Pfisterer, A.B., Buchmann, N., He, J.S., Nakashizuka, T., Raffaelli, D. & Schmid,

B. (2006) Quantifying the evidence for biodiversity effects on ecosystem functioning and

services. Ecology Letters, 9, 1146–1156.

Barak, R.S., Fant, J.B., Kramer, A.T. & Skogen, K.A. (2015) Assessing the value of potential

‘native winners’ for restoration of cheatgrass-invaded habitat. Western North American

Naturalist, 75, 58–69.

Barak, R.S., Hipp, A.L., Cavender-Bares, J., Pearse, W.D., Hotchkiss, S.C., Lynch, E.A.,

Callaway, J.C., Calcote, R. & Larkin, D.J. (2016) Taking the long view: integrating

recorded, archeological, paleoecological, and evolutionary data into ecological

restoration. International Journal of Plant Sciences, 177, 90–102.

Barak, R.S., Williams, E.W., Hipp, A.L., Bowles, M.L., Carr, G.M., Sherman, R. & Larkin, D.J.

(2017) Restored tallgrass prairies have reduced phylogenetic diversity compared with

remnants. Journal of Applied Ecology.

Barber, N.A., Jones, H.P., Duvall, M.R., Wysocki, W.P., Hansen, M.J. & Gibson, D.J. (2016)

Phylogenetic diversity is maintained despite richness losses over time in restored tallgrass

prairie plant communities. Journal of Applied Ecology, 54, 137–144. 117

Beier, P., M.L. Hunter, & M. Anderson (2015) Special section: conserving nature’s stage.

Conservation Biology, 29, 613–617.

Bergeron, Y., M. Flannigan, S. Gauthier, A. Leduc, & P. Lefort (2004) Past, current and future

ﬁre frequency in the Canadian boreal forest: implications for sustainable forest

management. Ambio, 33, 356– 360.

Blomberg, S.P., Jr, T.G. & Ives, A.R. (2003) Testing for phylogenetic signal in comparative

data: behavioral traits are more labile. Evolution, 57, 717–745.

Bolliger, J.L., L.A. Schulte, S.N. Burrows, T.A. Sickley, & D.J. Mladenoff (2004) Assessing

ecological restoration potential of Wisconsin (USA) using historical landscape

reconstructions. Restoration Ecology, 12, 124–142.

Bowles, M. & Jones, M. (2006) Testing the efficacy of species richness and floristic quality

assessment of quality, temporal change, and fire effects in tallgrass prairie natural areas.

Natural Areas Journal, 26, 17–30.

Bowles, M., Zettler, L., Bell, T. & Kelsey, P. (2005) Relationships between soil characteristics,

distribution and restoration potential of Platanthera leucophaea (Nutt.) Lindl. American

Midland Naturalist, 154, 273–285.

Bowles, M.L. & Jones, M.D. (2013) Repeated burning of eastern tallgrass prairie increases

richness and diversity, stabilizing late successional vegetation. Ecological Applications,

23, 464–478.

Boyle, B., Hopkins, N., Lu, Z. et al. (2013) The taxonomic name resolution service: an online

tool for automated standardization of plant names. BMC Bioinformatics, 14, 16.

Brewer, S., S.T. Jackson, & J.W. Williams (2012) Paleoecoinformatics: applying geohistorical

data to ecological questions. Trends in Ecology and Evolution, 27, 104–112. 118

Broadhurst, L.M., A. Lowe, D.J. Coates, S.A. Cunningham, M. McDonald, P.A. Vesk, & C.

Yates (2008) Seed supply for broadscale restoration: maximizing evolutionary potential.

Ecological Applications, 1, 587–597.

Brubaker, L.B. (1975) Postglacial forest patterns associated with till and outwash in northcentral

Upper Michigan. Quaternary Research, 5, 499– 527.

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., Barak, R.S., Bauer, J.T. et al. (2017) Interpreting variation to advance predictive

restoration science. Journal of Applied Ecology, in press.

Bu, H.-Y., Wang, X.-J., Zhou, X.-H., Qi, W., Liu, K., Ge, W.-J., Xu, D.-H. & Zhang, S.-T.

(2017) The ecological and evolutionary significance of seed shape and volume for the

germination of 383 species on the eastern Qinghai-Tibet plateau Hai-Yan. Folia

Geobotanica.

Burney, D.A., & L.P. Burney (2007) Paleoecology and “inter-situ” restoration on Kaua’i,

Hawai’i. Frontiers in Ecology and the Environment, 5, 483–490.

Burns, J.H. & Strauss, S.Y. (2011) More closely related species are more ecologically similar in

an experimental test. Proceedings of the National Academy of Sciences of the United

States of America, 108, 5302–5307.

Bush, M.B., A. Restrepo, & A.F. Collins (2014) Galapagos history, restoration, and a shifted

baseline. Restoration Ecology, 22, 296–298.

Cadotte, M.W. (2013) Experimental evidence that evolutionarily diverse assemblages result in

higher productivity. Proceedings of the National Academy of Sciences of the United

States of America, 110, 8996–9000. 119

Cadotte, M.W., Cardinale, B.J. & Oakley, T.H. (2008) Evolutionary history and the effect of

biodiversity on plant productivity. Proceedings of the National Academy of Sciences of

the United States of America, 105, 17012–17017.

Cadotte, M.W., Cavender-Bares, J., Tilman, D. & Oakley, T.H. (2009) Using phylogenetic,

functional and trait diversity to understand patterns of plant community productivity.

PLoS ONE, 4, e5695.

Cadotte, M.W., Dinnage, R. & Tilman, D. (2012) Phylogenetic diversity promotes ecosystem

stability. Ecology, 93, 223–233.

Callaway, J.C., E.L. Borgnis, R.E. Turner, & C.S. Milan (2012) Carbon sequestration and

sediment accretion in San Francisco Bay tidal wetlands. Estuaries and Coasts, 35, 1163–

1181.

Callaway, J.C., V.T. Parker, M.C. Vasey, & L.M. Schile (2007) Emerging issues for the

restoration of tidal marsh ecosystems in the context of predicted climate change.

Madroño, 54, 234–248.

Campbell, W.B., D.C. Freeman, J.M. Emlen, & S.L. Ortiz (2010) Correlations between plant

phylogenetic and functional diversity in a high altitude cold salt desert depend on sheep

grazing season: implications for range recovery. Ecological Indicators, 10, 676–686.

Cavender-Bares, J. & Cavender, N. (2011) Phylogenetic structure of plant communities provides

guidelines for restoration. Restoration Ecology (ed. S. Greipsson), pp. 119–129. Jones

and Bartlett Learning, Sudsbury, MA, USA.

Cavender-Bares, J., A. Keen, & B. Miles (2006) Phylogenetic structure of Floridian plant

communities depends on taxonomic and spatial scale. Ecology, 87, 109–122. 120

Chesson, P. (2014) Mechanisms of maintenance of species diversity. Annual Review of

Ecological Systems, 31, 343–358.

Clark, D.L. & Wilson, M. V. (2003) Post-dispersal seed fates of four prairie species. American

Journal of Botany, 90, 730–735.

Clarke, K.R. (1993) Non-parametric multivariate analyses of changes in community structure.

Australian Journal of Ecology, 18, 117143.

Clewell, A., J. Aronson, & K. Winterhalder (2004) The SER international primer on ecological

restoration. Ecological Restoration, 2, 206–207.

Coffey, E.E.D., C.A. Froyd, & K.J. Willis (2010) When is an invasive not an invasive?

macrofossil evidence of doubtful native plant species in the Galapagos Islands. Ecology,

92, 805–812.

Comer, P.J., R.L. Pressey, M.L. Hunter, C.A. Schloss, S.C. Buttrick, N.E. Heller, J.M. Tirpak,

D.P. Faith, M.S. Cross, & M. Shaffer (2015) Incorporating geodiversity into conservation

decisions. Conservation Biology, 29, 692–701.

Cottam, G. & Wilson, H.C. (1966) Community Dynamics on An Artificial Prairie Author.

Ecology, 47, 88–96.

Davies, A.L., & M.J. Bunting (2010) Applications of paleoecology in conservation. The Open

Ecology Journal, 3, 54–67.

Davies, A.L., S. Colombo, & N. Hanley (2014) Improving the application of long-term ecology

in conservation and land management. Journal of Applied Ecology, 51, 63–70.

Davies, K.F., Cavender-Bares, J. & Deacon, N. (2011) Native communities determine the

identity of exotic invaders even at scales at which communities are unsaturated. Diversity

and Distributions, 17, 35–42. 121

Davis, O.K. 2005 Palynology: an important tool for discovering historic ecosystems. Pages 229–

256 in D Egan, EA Howell, eds. The historical ecology handbook: a restorationist’s guide

to reference ecosystems. 2nd ed. Island, Washington, DC.

De Vienne, D.M., Aguileta, G. & Ollier, S. (2011) Euclidean nature of phylogenetic distance

matrices. Systematic Biology, 60, 826–832.

Decleer, K., Bonte, D. & van Diggelen, R. (2013) The hemiparasite Pedicularis palustris:

‘Ecosystem engineer’ for fen-meadow restoration. Journal for Nature Conservation, 21,

65–71.

Dı́az, S. & Cabido, M. (2001) Vive la différence: plant functional diversity matters to ecosystem

processes. Trends in Ecology & Evolution, 16, 646–655.

Díaz, S., Purvis, A., Cornelissen, J.H.C., Mace, G.M., Donoghue, M.J., Ewers, R.M., Jordano, P.

& Pearse, W.D. (2013) Functional traits, the phylogeny of function, and ecosystem

service vulnerability. Ecology and Evolution, 3, 2958–2975.

Dietl, G.P., & K.W. Flessa (2011) Conservation paleobiology: putting the dead to work. Trends

in Ecology and Evolution, 26, 30–37.

Dietl, G.P., S.M. Kidwell, M. Brenner, DA Burney, KW Flessa, ST Jackson, & PL Koch (2015)

Conservation paleobiology: leveraging knowledge of the past to inform conservation and

restoration. Annual Review of Earth and Planetary Sciences, 43, 79–103.

Dinnage, R. (2009) Disturbance alters the phylogenetic composition and structure of plant

communities in an old field system. PLoS ONE, 4, e7071.

Dinnage, R., Cadotte, M.W., Haddad, N.M., Crutsinger, G.M. & Tilman, D. (2012) Diversity of

plant evolutionary lineages promotes arthropod diversity. Ecology Letters, 15, 1308–

1317. 122

Dinnage, R., M.W. Cadotte, N.M. Haddad, G.M. Crutsinger, & D. Tilman (2012) Diversity of

plant evolutionary lineages promotes arthropod diversity. Ecology Letters, 15, 1308–

1317.

Donohue, K., Rubio de Casas, R., Burghardt, L., Kovach, K. & Willis, C.G. (2010) Germination,

postgermination adaptation, and species ecological ranges. Annual Review of Ecology,

Evolution, and Systematics, 41, 293–319.

Dunnette, P.V., P.E. Higuera, K.K. McLauchlan, K.M. Derr, C.E. Briles, & M.H. Keefe (2014)

Biogeochemical impacts of wildﬁres over four millennia in a Rocky Mountain subalpine

watershed. New Phytologist, 203, 900–912.

Dunwiddie, P.W. (2005) Using historical data in ecological restoration: a case study from

Nantucket. Pages 367–390 in D Egan, EA Howell, eds. The historical ecology handbook:

a restorationist’s guide to reference ecosystems. 2nd ed. Island, Washington, DC.

Edlund, M., D.R. Engstrom, L.D. Triplett, B.M. Lafrancois, & P.R. Leavitt (2009a) Twentieth

century eutrophication of the St. Croix River (Minnesota–Wisconsin, USA) reconstructed

from the sediments of its natural impoundment. Journal of Paleolimnology, 41, 641–657.

Edlund, M., L.D. Triplett, M.D. Tomasek, & K. Barilson (2009b) From paleo to policy:

partitioning the historical point and nonpoint phosphorus loads to the St. Croix River,

Minnesota–Wisconsin, USA. Journal of Paleolimnology, 41, 679–689.

Edmonds, M. (2005) The pleasures and pitfalls of written records. Pages 73–100 in D Egan, EA

Howell, eds. The historical ecology handbook: a restorationist’s guide to reference

ecosystems. 2nd ed. Island, Washington, DC.

Egan, D., & E.A. Howell (2005) The historical ecology handbook: a restorationist’s guide to

reference ecosystems. 2nd ed. Island, Washington, DC. 123

Engstrom, D.R., J.E. Almendinger, & J.A. Wolin (2009) Historical changes in sediment and

phosphorus loading to the Upper Mississippi River: mass-balance reconstruction from the

sediments of Lake Pepin. Journal of Paleolimnology, 41, 563–588.

Finch-Savage, W.E. & Leubner-Metzger, G. (2006) Seed dormancy and the control of

germination. New Phytologist, 171, 501–523.

Finney, B.P., I. Gregory-Eaves, J. Sweetman, M.S. Douglas, & J.P. Smol (2000) Impacts of

climatic change and ﬁshing on Paciﬁc salmon abundance over the past 300 years.

Science, 290, 795–799.

Forbis, T.A., Floyd, S.K. & de Queiroz, A. (2002) The evolution of embryo size in angiosperms

and other seed plants: implications for the evolution of seed dormancy. Evolution;

international journal of organic evolution, 56, 2112–25.

Forest, F., R. Grenyer, M. Rouget, T.J. Davies, R.M. Cowling, D.P. Faith, A. Balmford, et al.

(2007) Preserving the evolutionary potential of ﬂoras in biodiversity hotspots. Nature,

445, 757–760.

Foster, B.L., Murphy, C.A., Keller, K.R., Aschenbach, T.A., Questad,E.J. & Kindscher, K.

(2007) Restoration of prairie community structure and ecosystem function in an

abandoned hayfield: a sowing experiment. Restoration Ecology, 15, 652–661.

Foster, D., F. Swanson, J. Aber, I. Burke, N. Brokaw, D. Tilman, & A. Knapp (2003) The

importance of land-use legacies to ecology and conservation. BioScience, 53, 77–88.

Foster, D.R., W.W. Oswald, E.K. Faison, E.D. Doughty, & B.C. Hansen (2006) A climatic

driver for abrupt mid-Holocene vegetation dynamics and the hemlock decline in New

England. Ecology, 87, 2959–2966. 124

Freyman, W.A., Masters, L.A. & Packard, S. (2016) The Universal Floristic Quality Assessment

(FQA) Calculator: An online tool for ecological assessment and monitoring. Methods in

Ecology and Evolution, 7, 380–383.

Frischie, S.L. & Rowe, H.I. (2012) Replicating life cycle of early-maturing species in the timing

of restoration seeding improves establishment and community diversity. Restoration

Ecology, 20, 188–193.

Fritz, S.A. & Purvis, A. (2010) Selectivity in mammalian extinction risk and threat types: a new

measure of phylogenetic signal strength in binary traits. Conservation Biology, 24, 1042–

1051.

Fulé, P.Z. (2008) Does it make sense to restore wildland ﬁre in changing climate? Restoration

Ecology, 16, 526–532.

Gallagher, M.K. & Wagenius, S. (2016) Seed source impacts germination and early

establishment of dominant grasses in prairie restorations. Journal of Applied Ecology, 53,

251–263.

Gavin, D.G., D.J. Hallett, F.S. Hu, K.P. Lertzman, S.J. Prichard, K.J. Brown, J.A. Lynch, P.

Bartlein, & D.L. Peterson (2007) Forest ﬁre and climate change in western North

America: insights from sediment charcoal records. Frontiers in Ecology and the

Environment, 5, 499–506.

Gavin, D.G., M.C. Fitzpatrick, P.F. Gugger, K.D. Heath, F. RodríguezSánchez, S.Z. Dobrowski,

A. Hampe, et al. (2014) Climate refugia: joint inference from fossil records, species

distribution models and phylogeography. New Phytologist, 204, 37–54. 125

Gill, J.L., J.L. Blois, B. Benito, S. Dobrowski, M.L. Hunter, & J.L. McGuire (2015) A 2.5-

million-year perspective on coarse-ﬁlter strategies for conserving nature’s stage.

Conservation Biology, 29, 640–648.

Gillson, L., & R. Marchant (2014) From myopia to clarity: sharpening the focus of ecosystem

management through the lens of palaeoecology. Trends in Ecology and Evolution, 29,

317–325.

Goldblum, D., Glaves, B.P., Rigg, L.S. & Kleiman, B. (2013) The impact of seed mix weight on

diversity and species composition in a tallgrass prairie restoration planting, Nachusa

Grasslands, Illinois, USA. Ecological Restoration, 31, 154–167.

Gomes-Junior, F.G., Chiquito, A.A. & Marcos-Filho, J. (2013) Semi-automated assessment of

the embryonic area of cucumber seeds and its relationship to germination and seedling

length. Journal of Seed Science, 35, 183–189.

Grimm, E.C. (1983) Chronology and dynamics of vegetation change in the prairie-woodland

region of southern Minnesota, U.S.A. New Phytologist, 93, 311–350.

Grman, E. & Brudvig, L.A. (2014) Beta diversity among prairie restorations increases with

species pool size, but not through enhanced species sorting. Journal of Ecology, 102,

1017–1024.

Grman, E. & Suding, K.N. (2010) Within-year soil legacies contribute to strong priority effects

of exotics on native California grassland communities. Restoration Ecology, 18, 664–

670.

Grman, E., Bassett, T. & Brudvig, L.A. (2013) EDITOR’S CHOICE: Confronting contingency

in restoration: Management and site history determine outcomes of assembling prairies, 126

but site characteristics and landscape context have little effect. Journal of Applied

Ecology, 50, 1234–1243.

Grman, E., Bassett, T., Zirbel, C.R. & Brudvig, L.A. (2015) Dispersal and establishment filters

influence the assembly of restored prairie plant communities. Restoration Ecology, 23,

892–899.

Guyette, R.P., D.C. Dey, M.C. Stambaugh, & R.M. Muzika (2006) Fire scars reveal variability

and dynamics of eastern ﬁre regimes. Pages 20–39 in MB Dickinson, ed. Fire in eastern

oak forests: delivering science to land managers. Gen Tech Rep NRS-P-1. USDA Forest

Service, Northern Research Station, Newtown Square, PA.

Haddad, N.M., L.A. Brudvig, E.I. Damschen, D.M. Evans, B.L. Johnson, D.J. Levey, J.L.

Orrock, et al. (2014) Potential negative ecological effects of corridors. Conservation

Biology, 28, 1178–1187.

Hanley, N., A. Davies, K. Angelopoulos, A. Hamilton, A. Ross, D. Tinch, & F. Watson (2008)

Economic determinants of biodiversity change over a 400-year period in the Scottish

uplands. Journal of Applied Ecology, 45, 1557–1565.

Hansen, M.J. & Gibson, D.J. (2014) Use of multiple criteria in an ecological assessment of a

prairie restoration chronosequence. Applied Vegetation Science, 17, 63–73.

Harmon-Threatt, A.N. & Hendrix, S.D. (2015) Prairie restorations and bees: The potential ability

of seed mixes to foster native bee communities. Basic and Applied Ecology, 16, 64–72.

Havens, K. & Vitt, P. (2016) The importance of phenological diversity in seed mixes for

pollinator restoration. Natural Areas Journal, 36, 531–537.

Hayashida, F.M. (2005) Archaeology, ecological history, and conservation. Annual Review of

Anthropology, 34, 43–65. 127

Heinselman, M.L. (1973) Fire in the virgin forests of the Boundary Waters Canoe Area,

Minnesota. Quaternary Research, 3, 329–382.

Hejcman, M., P. Karlík, J. Ondráček, & T. Klír (2013) Short-term medieval settlement activities

irreversibly changed forest soils and vegetation in central Europe. Ecosystems, 16, 652–

663.

Hessburg, P.F., B.G. Smith, & R.B. Salter (1999) Detecting change in forest spatial patterns

from reference conditions. Ecological Applications, 9, 1232–1252.

Higgs, E., D.A. Falk, A. Guerrini, M. Hall, J. Harris, R.J. Hobbs, S.T. Jackson, J.M. Rhemtulla,

& W. Throop (2014) The changing role of history in restoration ecology. Frontiers in

Ecology and the Environment, 12, 499–506.

Higuera, P.E., L.B. Brubaker, P.M. Anderson, F.S. Hu, & T.A. Brown (2009) Vegetation

mediated the impacts of postglacial climate change on ﬁre regimes in the south-central

Brooks Range, Alaska. Ecological Monographs, 79, 201–219.

Hillhouse, H.L. & Zedler, P.H. (2011) Native species establishment in tallgrass prairie plantings.

The American Midland Naturalist, 166, 292–308.

Hipp, A.L., Larkin, D.J., Barak, R.S., Bowles, M.L., Cadotte, M.W., Jacobi, S.K., Lonsdorf, E.,

Scharenbroch, B.C., Williams, E. & Weiher, E. (2015) Phylogeny in the service of

ecological restoration. American Journal of Botany, 102, 1–2.

Hobbs, R.J., E. Higgs, C.M. Hall, P. Bridgewater, F.S. Chapin III, E.C. Ellis, JJ Ewel, et al.

(2014) Managing the whole landscape: historical, hybrid, and novel ecosystems.

Frontiers in Ecology and the Environment, 12, 557–564.

Hoekstra, J.M., Boucher, T.M., Ricketts, T.H. & Roberts, C. (2005) Confronting a biome crisis:

Global disparities of habitat loss and protection. Ecology Letters, 8, 23–29. 128

Hotchkiss, S.C., & J.O. Juvik (1999) A Late-Quaternary pollen record from Ka‘au Crater, O‘ahu,

Hawai‘i. Quaternary Research, 52, 115–128.

Hotchkiss, S.C., R.R. Calcote, & E.A. Lynch (2007) Response of vegetation and ﬁre to Little Ice

Age climate change: regional continuity and landscape heterogeneity. Landscape

Ecology, 22, 25–41.

Hoyle, G.L., Steadman, K.J., Good, R.B., McIntosh, E.J., Galea, L.M.E. & Nicotra, A.B. (2015)

Seed germination strategies: an evolutionary trajectory independent of vegetative

functional traits. Frontiers in Plant Science, 6, 1–13.

Huang, Z., Liu, S., Bradford, K.J., Huxman, T.E. & Venable, D.L. (2015) The contribution of

germination functional traits to population dynamics of a desert plant community.

Ecology, 97, 150821143437005.

Ireland, A.W., R.K. Booth, S.C. Hotchkiss, & J.E. Schmitz (2012) Drought as a trigger for rapid

state shifts in kettle ecosystems: implications for ecosystem responses to climate change.

Wetlands, 32, 989–1000.

Iverson, L.R. (1988) Land-use changes in Illinois, USA: the inﬂuence of landscape attributes on

current and historic land use. Landscape Ecology, 2, 45–61.

Jackson, S.T. (2007) Looking forward from the past: history, ecology and conservation. Frontiers

in Ecology and the Environment, 5, 455.

Jackson, S.T., & R.J. Hobbs (2009) Ecological restoration in the light of ecological history.

Science 325, 567–569.

Jetz, W., C. Rahbek, & R.K. Colwell (2004) The coincidence of rarity and richness and the

potential signature of history in centres of endemism. Ecology Letters, 7, 1180–1191. 129

Jiménez-Alfaro, B., Silveira, F.A.O., Fidelis, A., Poschlod, P. & Commander, L.E. (2016) Seed

germination traits can contribute better to plant community ecology. Journal of

Vegetation Science, 27, 637–645.

Kahmen, S. & Poschlod, P. (2008) Does germination success differ with respect to seed mass

and germination season? Experimental testing of plant functional trait responses to

grassland management. Annals of Botany, 101, 541–548.

Karlin, E.F., S.C. Hotchkiss, S.B. Boles, H.K. Stenøien, K. Hassel, K.I. Flatberg, & A.J. Shaw

(2012) High genetic diversity in a remote island population system: sans sex. New

Phytologist 193, 1088–1097.

Keane, R.E., P.F. Hessburg, P.B. Landres, & F.J. Swanson (2009) The use of historical range

and variability (HRV) in landscape management. Forrest Ecology and Management 258,

1025–1037.

Kembel, S.W., Cowan, P.D., Helmus, M.R., Cornwell, W.K., Morlon, H., Ackerly, D.D.,

Blomberg, S.P. & Webb, C.O. (2010) Picante: R tools for integrating phylogenies and

ecology. Bioinformatics, 26, 1463–1464.

Kettenring, K.M., K.L. Mercer, C. Reinhardt Adams, & J. Hines (2014) Application of genetic

diversity-ecosystem function research to ecological restoration. Journal of Applied

Ecology, 51, 339–348.

Kleyer, M., Bekker, R.M., Knevel, I.C., et al. (2008) The LEDA Traitbase: a database of life-

history traits of the Northwest European flora. Journal of Ecology, 96, 1266–1274.

Larkin, D.J., Hipp, A.L., Kattge, J., Prescott, W., Tonietto, R.K., Jacobi, S.K. & Bowles, M.L.

(2015a) Phylogenetic measures of plant communities show long-term change and impacts 130

of fire management in tallgrass prairie remnants. Journal of Applied Ecology, 52, 1638–

1648.

Larkin, D.J., Hipp, A.L., Kattge, J., Prescott, W., Tonietto, R.K., Jacobi, S.K. & Bowles, M.L.

(2015b) Data from: phylogenetic measures of plant communities show long-term change

and impacts of fire management in tallgrass prairie remnants. Dryad Digital Repository,

https://doi.org/10.5061/dryad.763v6

Larson, J.E. & Funk, J.L. (2016) Regeneration: an overlooked aspect of trait-based plant

community assembly models. Journal of Ecology, 104, 1284–1298.

Larson, J.E., Sheley, R.L., Hardegree, S.P., Doescher, P.S. & James, J.J. (2015) Seed and

seedling traits affecting critical life stage transitions and recruitment outcomes in dryland

grasses. Journal of Applied Ecology, 52, 199–209.

Laughlin, D.C. (2014) Applying trait-based models to achieve functional targets for theory-

driven ecological restoration. Ecology Letters, 17, 771–784.

Lawler, J.J., D.D. Ackerly, C.M. Albano, M.G. Anderson, S.Z. Dobrowski, J.L. Gill, N.E.

Heller, R.L. Pressey, E.W. Sanderson, & S.B. Weiss 2015 The theory behind, and the

challenges of, conserving nature’s stage in a time of rapid change. Conservation Biology,

29, 618–629.

Legendre, P. & Legendre, L. (2012) Numerical Ecology, 3rd English edn. Elsevier, Amsterdam,

The Netherlands.

Li, S., Guo, T., Cadotte, M.W., Chen, Y., Kuang, J., Hua, Z., Zeng, Y., Song, Y., Liu, Z., Shu,

W. & Li, J. (2015) Contrasting effects of phylogenetic relatedness on plant invader

success in experimental grassland communities. Journal of Applied Ecology, 52, 89–99. 131

Lind, E.M., Vincent, J.B., Weiblen, G.D., Cavender-Bares, J. & Borer, E.T. (2015) Trophic

phylogenetics: evolutionary influences on body size, feeding, and species associations in

grassland arthropods. Ecology, 96, 998–1009.

Linkies, A., Graeber, K., Knight, C. & Leubner-Metzger, G. (2010) The evolution of seeds. New

Phytologist, 186, 817–831.

Lucia, S., J. Tibby, M.B. Lane, &P.A. Gell (2008) Local knowledge and environmental

management: a cautionary tale from Lake Ainsworth, New South Wales, Australia.

Environmental Conservation, 34, 334–341.

Lynch, E.A., & K. Saltonstall (2002) Paleoecological and genetic analyses provide evidence for

recent colonization of native Phragmites australis populations in a Lake Superior wetland.

Wetlands, 22, 637– 646.

Lynch, E.A., R. Calcote, S.C. Hotchkiss, & M.A. Tweiten (2014) Presence of lakes and wetlands

decreases resilience of jack pine ecosystems to late-Holocene climatic changes. Canadian

Journal of Forest Research, 44, 1331–1343.

Lynch, E.A., S.C. Hotchkiss, & R. Calcote (2010) Charcoal signatures de ﬁned by multivariate

analysis of charcoal records from 10 lakes in northwest Wisconsin (USA). Quaternary

Research 75, 125–137.

M’Gonigle, L.K., Williams, N.M., Lonsdorf, E. & Kremen, C. (2016) A tool for selecting plants

when restoring habitat for pollinators. Conservation Letters, 00, 1–7.

Marks, P.L., & S. Gardescu (2005) Inferring forest stand history from observational ﬁeld

evidence. Pages 177–198 in D Egan, EA Howell, eds. The historical ecology handbook: a

restorationist’s guide to reference ecosystems. 2nd ed. Island, Washington, DC. 132

Martin, L.M. & Wilsey, B.J. (2012) Assembly history alters alpha and beta diversity, exotic-

native proportions and functioning of restored prairie plant communities ed M. Cadotte.

Journal of Applied Ecology, 49, 1436–1445.

Martin, L.M., Moloney, K.A. & Wilsey, B.J. (2005) An assessment of grassland restoration

success using species diversity components. Journal of Applied Ecology, 42, 327–336.

Matthews, J.W., Spyreas, G. & Long, C.M. (2015) A null model test of Floristic Quality

Assessment: Are plant species’ Coefficients of Conservatism valid? Ecological

Indicators, 52, 1–7.

McCune, J.L., M.G. Pellatt, & M. Vellend (2013) Multidisciplinary synthesis of long-term

human-ecosystem interactions: a perspective from the Garry oak ecosystem of British

Columbia. Biological Conservation, 166, 293–300.

McGlone, C.M., Sieg, C.H. & Kolb, T.E. (2011) Invasion resistance and persistence: established

plants win, even with disturbance and high propagule pressure. Biological Invasions, 13,

291–304.

McInerney, F.A., & S.L. Wing (2011) The Paleocene-Eocene thermal maximum: a perturbation

of carbon cycle, climate, and biosphere with implications for the future. Annual Review

of Earth and Planetary Sciences, 39, 489– 516.

McLauchlan, K.K., P.E. Higuera, D.G. Gavin, S.S. Perakis, M.C. Mack, H. Alexander, J. Battles,

et al. (2014) Reconstructing disturbances and their biogeochemical consequences over

multiple timescales. BioScience, 64, 105–116.

McNair, J.N., Sunkara, A. & Frobish, D. (2012) How to analyse seed germination data using

statistical time-to-event analysis: non-parametric and semi-parametric methods. Seed

Science Research, 22, 77–95. 133

Meyer, S.E., Kitchen, S.G. & Carlson, S.L. (1995) Seed Germination Timing Patterns in

Intermountain Penstemon (Scrophulariaceae). American Journal of Botany, 82, 377–389.

Middleton, E.L., Bever, J.D. & Schultz, P.A. (2010) The effect of restoration methods on the

quality of the restoration and resistance to invasion by exotics. Restoration Ecology, 18,

181–187.

Millar, C.I., & L.B. Brubaker (2006) Climate change and paleoecology: new contexts for

restoration ecology. Pages 315–340 in DA Falk, MA Palmer, JB Zedler, eds. Foundations

of restoration ecology. Island, Washington, DC.

Millar, C.I., N.L. Stephenson, & S.L. Stephens (2007) Climate change and forests of the future.

Ecological Applications, 17, 2145–2151.

Minckley, T.A., A. Brunelle, & S. Blissett (2011) Holocene sedimentary and environmental

history of an in-channel wetland along the ecotone of the Sonoran and Chihuahuan

Desert grasslands. Quaternary International, 235, 40– 47.

Misarti, N., B. Finney, H. Maschner, & M.J. Wooller (2009) Changes in northeast Paciﬁc marine

ecosystems over the last 4500 years: evidence from stable isotope analysis of bone

collagen from archeological middens. Holocene, 19, 1139–1151.

Moles, A.T., Ackerly, D.D., Webb, C.O., Tweddle, J.C., Dickie, J.B. & Westoby, M. (2005) A

Brief History of Seed Size. Science, 307, 576–580.

Montoya, D., Rogers, L. & Memmott, J. (2012) Emerging perspectives in the restoration of

biodiversity-based ecosystem services. Trends in Ecology and Evolution, 27, 666–672.

Motzkin, G. & D.R. Foster (2002) Grasslands, heathlands and shrublands in coastal New

England: historical interpretations and approaches to conservation. Journal of

Biogeography, 29, 1569–1590. 134

Nakagawa, S. & Schielzeth, H. (2013) A general and simple method for obtaining R2 from

generalized linear mixed-effects models. Methods in Ecology and Evolution, 4, 133–142.

Natural Resources Conservation Service. http://plants.usda.gov/ (accessed 12 December 2015).

Navarro-Cano, J.A., M. Goberna, A. Valiente-Banuet, A. MontesinosNavarro, C. García, & M.

Verdú (2014) Plant phylodiversity enhances soil microbial productivity in facilitation-

driven communities. Oecologia, 174, 909–920.

Nelson, D.M., & F.S. Hu (2008) Patterns and drivers of Holocene vegetational change near the

prairie-forest ecotone in Minnesota: revisiting McAndrews’ transect. New Phytologist,

179, 449– 459.

Norden, N., Daws, M.I., Antoine, C., Gonzalez, M.A., Garwood, N.C. & Chave, J. (2009) The

relationship between seed mass and mean time to germination for 1037 tree species

across five tropical forests. Functional Ecology, 23, 203–210.

Oksanen, J.F., Guillaume Blanchet, F., Friendly, M. et al. (2016) vegan: Community Ecology

Package. R package version 2.4-0. Available at: https://CRAN.R-

project.org/package=vegan (accessed 13 June 2016).

Olden, J.D., Lawler, J.J. & Poff, N.L. (2008) Machine learning methods without tears : A primer

for ecologists. The Quarterly Review of Biology, 83, 171–193.

Orme, D., Freckleton, R., Thomas, G., Petzoldt, T., Fritz, S., Isaac, N. & Pearse, W. (2013)

caper: comparative analyses of phylogenetics and evolution in R. R package version

0.5.2. Available at: https://CRAN.R-pro ject.org/package=caper (accessed 16 March

2016)

Oswald, W.W., & D.R. Foster (2011) Middle-Holocene dynamics of Tsuga canadensis (eastern

hemlock) in northern New England, USA. Holocene 22, 71–78. 135

Oswald, W.W., E.K. Faison, D.R. Foster, E.D. Douhty, B.R. Hall, & B.C.S. Hansen (2007) Post-

glacial changes in spatial patterns of vegetation across southern New England. Journal of

Biogeography 34, 900–913.

Pagel, M. (1999) Inferring the historical patterns of biological evolution. Nature, 401, 877–884.

Pakeman, R.J., Pywell, R.F. & Wells, T.C.E. (2002) Species spread and persistence: implications

for experimental design and habitat re-creation. Applied Vegetation Science, 5, 75–86.

PalEON (Paleo-Ecological Observatory Network) (2015) PalEON project.

http://www3.nd.edu/∼paleolab/paleonproject.

Pauly, D. 1995 Anecdotes and the shifting baseline syndrome of ﬁsheries. Trends in Ecology

and Evolution, 10, 430.

Pavoine, S. & Bonsall, M.B. (2011) Measuring biodiversity to explain community assembly: A

unified approach. Biological Reviews, 86, 792–812.

Pearse, I.S. & Hipp, A.L. (2009) Phylogenetic and trait similarity to a native species predict

herbivory on non-native oaks. Proceedings of the National Academy of Sciences of the

United States of America, 106, 18097–18102.

Pearse, W.D., M. Cadotte, J. Cavender-Bares, A.R. Ives, C. Tucker, S. Walker, & M.R. Helmus

(2015) pez: phylogenetics for the environmental sciences. Bioinformatics, 31, 2888–

2890.

Peppin, D.L., Ful´e, P.Z., Lynn, J.C., Mottek-Lucas, A.L. & Hull Sieg, C. (2010) Market

perceptions and opportunities for native plant production on the Southern Colorado

Plateau. Restoration Ecology, 18, 113–124. 136

Peters, D.P.C. (2010) Accessible ecology: synthesis of the long, deep, and broad. Trends in

Ecology and Evolution, 25, 592–601.

Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D. & R Core Team (2016)._nlme: Linear and

Nonlinear Mixed Effects Models_. R package version 3.1-128. Available at:

http://CRAN.R-project.org/package=nlme

Polley, H.W., Derner, J.D. & Wilsey, B.J. (2005) Patterns of plant species diversity in remnant

and restored tallgrass prairies. Restoration Ecology, 13, 480–487.

Possingham, H.P., Bode, M. & Klein, C.J. (2015) Optimal conservation outcomes require both

restoration and protection. PLOS Biology, 13, e1002052.

Pywell, R.F., Bullock, J.M., Roy, D.B., Warman, L., Walker, K.J. & Rothery, P. (2003) Plant

traits as predictors of performance in ecological restoration. Journal of Applied Ecology,

40, 65–77.

Pywell, R.F., Bullock, J.M., Walker, K.J., Coulson, S.J., Gregory, S.J. & Stevenson, M.J. (2004)

Facilitating grassland diversification using the hemiparasitic plant Rhinanthus minor.

Journal of Applied Ecology, 41, 880–887.

R Development Core Team (2013) R: A Language and Environment for Statistical Computing. R

Foundation for Statistical Computing, Vienna, Austria.

Radeloff, V.C., D.J. Mladenoff, H.S. He, & M.S. Boyce (1999) Forest land scape change in the

northwestern Wisconsin Pine Barrens from preEuropean settlement to the present.

Canadian Journal of Forest Research, 29, 649–1659.

Rey Benayas, J.M., Newton, A.C., Diaz, A. & Bullock, J.M. (2009) Enhancement of biodiversity

and ecosystem services by ecological restoration: a meta-analysis. Science, 325, 1121–

1124. 137

Rick, T.C., & R. Lockwood (2013) Integrating paleobiology, archeology, and history to inform

biological conservation. Conservation Biology, 27, 45– 54.

Rocha, C.R.M. da, Silva, V.N. & Cicero, S.M. (2014) Internal morphology and germination of

sunflower seeds. Journal of Seed Science, 36, 48–53.

Rosauer, D.F., & A.O. Mooers (2013) Nurturing the use of evolutionary diversity in nature

conservation. Trends in Ecology and Evolution, 28, 322–323.

Rowe, H.I. & Holland, J.D. (2013) High plant richness in prairie reconstructions support diverse

leafhopper communities. Restoration Ecology, 21, 174–180.

Rowe, H.I. (2010) Tricks of the trade: techniques and opinions from 38 experts in tallgrass

prairie restoration. Restoration Ecology, 18, 253–262.

Schaefer, V., & A. Tillmanns (2015) Listening to ecosystems: ecological restoration and the

uniqueness of a place. Ecological Restoration, 33:3–10.

Schneider, C.A., Rasband, W.S. & Eliceiri, K.W. (2012) NIH Image to ImageJ: 25 years of

image analysis. Nature Methods, 9, 671–675.

Schulte, L.A., & D.J. Mladenoff (2005) Severe wind and ﬁre regimes in northern forests:

historical variability at the regional scale. Ecology, 86, 431–445.

Seddon, A.W.R., A.W. Mackay, A.G. Baker, H.J.B. Birks, E. Breman, C.E. Buck, E.C. Ellis, et

al. (2014) Looking forward through the past: identiﬁcation of 50 priority research

questions in palaeoecology. Journal of Ecology, 102, 256–267.

Sgrò, C.M., A.J. Lowe, & A.A. Hoffmann (2011) Building evolutionary resilience for

conserving biodiversity under climate change. Ecological Applications, 4, 326–337. 138

Shea, M.E., L.A. Schulte, & B.J. Palik (2014) Reconstructing vegetation past: pre-Euro-

American vegetation for the midwest Driftless Area, USA. Ecological Restoration, 32,

417–433.

Shipley, B. & Parent, M. (1991) Germination responses of 64 wetland species in relation to seed

size, minimum time to reproduction and seedling relative growth rate. Functional

Ecology, 5, 111–118.

Shoo, L.P., A.A. Hoffmann, S. Garnett, R.L. Pressey, Y.M. Williams, M. Taylor, L. Falconi, et

al. (2013) Making decisions to conserve species under climate change. Climate Change,

119, 239–246.

Silvertown, J., M. Dodd, D. Gowing, C. Lawson, & K. McConway (2006) Phylogeny and the

hierarchical organization of plant diversity. Ecology, 87, S39–S49.

Sollenberger, D., Kadlec, C., O’Shaughnessy, J. & Egerton-Warburton, L. (2016) Environmental

filtering mediates grassland community assembly following restoration with soil carbon

additions. Restoration Ecology, 24, 892–899.

Srivastava, D.S., Cadotte, M.W., Macdonald, A.A.M., Marushia, R.G. & Mirotchnick, N. (2012)

Phylogenetic diversity and the functioning of ecosystems. Ecology Letters, 15, 637–648.

Swetnam, T.W., C.D. Allen, & J.L. Betancourt (1999) Applied historical ecology: using the past

to manage for the future. Ecological Applications, 9, 1189–1206.

Swink, F. & Wilhem, G. (1994) Plants of the Chicago Region, 4th edn. Indiana Academy of

Science, Indianapolis, IN, USA.

Taft, J.B., Hauser, C. & Robertson, K.R. (2006) Estimating floristic integrity in tallgrass prairie.

Biological Conservation, 131, 42–51. 139

Therneau, T.M. & Grambsch, P.M. (2000) Modeling Survival Data: Extending the Cox Model.

Springer, New York.

Thompson, K., Brand, S.R. & Hodgson, J.G. (1993) Seed size and shape predict persistence in

soil. Functional Ecology, 7, 236–241.

Townsend, L., & R.J. Hebda (2013) Pollen and macro-fossil assemblages in disturbed urban

wetlands on South Vancouver Island reveal recent invasion of reed canarygrass (Phalaris

arundinacea) and guide restoration. Restoration Ecology, 21, 114–123.

Tucker, C.M., Cadotte, M.W., Carvalho, S.B., et al. (2016) A guide to phylogenetic metrics for

conservation, community ecology and macroecology. Biological Reviews. [Epub ahead

of print]. doi: 10.1111/brv. 12252.

Turley, N.E. & Brudvig, L.A. (2016) Agricultural land-use history causes persistent loss of plant

phylogenetic diversity. Ecology, 97, 2240–2247. USDA (2016) Plants Database. United

States Department of Agriculture,

Turnbull, L.A., Rees, M. & Crawley, M.J. (1999) Seed mass and the competition/Colonization

trade-off: A sowing experiment. Jounal of Ecology, 87, 899–912.

Tweiten, M.A., R.R. Calcote, E.A. Lynch, S.C. Hotchkiss, & G.W. Schuurman (2015)

Geophysical features inﬂuence the climate change sensitivity of northern Wisconsin pine

and oak forests. Ecological Applications, 25,1984– 1996.

Umbanhowar, C.E. (2004) Interaction of ﬁre, climate and vegetation change at a large landscape

scale in the Big Woods of Minnesota, USA. Holocene, 14, 661–676.

Vandelook, F., Verdú, M. & Honnay, O. (2012) The role of seed traits in determining the

phylogenetic structure of temperate plant communities. Annals of Botany, 110, 629–636.

Venables, W.N. & Ripley, B.D. (2002) Modern Applied Statistics with S. Springer, New York. 140

Verdu, M., Gomez-Aparicio, L. & Valiente-Banuet, A. (2012) Phylogenetic relatedness as a tool

in restoration ecology: a meta-analysis. Proceedings of the Royal Society B: Biological

Sciences, 279, 1761–1767.

Violle, C., Castro, H., Richarte, J. & Navas, M.L. (2009) Intraspecific seed trait variations and

competition: Passive or adaptive response? Functional Ecology, 23, 612–620.

Vitt, P., K. Havens, A.T. Kramer, D. Sollenberger, & E. Yates (2010) Assisted migration of

plants: changes in latitudes, changes in attitudes. Biological Conservation, 143, 18–27.

Völler, E., Auge, H., Prati, D., Fischer, M., Hemp, A. & Bossdorf, O. (2012) Geographical and

land-use effects on seed-mass variation in common grassland plants. Basic and Applied

Ecology, 13, 395–404.

Wang, J.H., Baskin, C.C., Cui, X.L. & Du, G.Z. (2009) Effect of phylogeny, life history and

habitat correlates on seed germination of 69 arid and semi-arid zone species from

northwest China. Evolutionary Ecology, 23, 827–846.

Watson, E.B., K. Wasson, G.B. Pasternack, A. Woolfolk, E. Van Dyke, A.B. Gray, A.

Pakenham, & R.A. Wheatcroft (2011) Applications from paleoecology to environmental

management and restoration in a dynamic coastal environment. Restoration Ecology, 19,

765–775.

Webb, C., Ackerly, D.D., Mcpeek, M.A. & Donoghue, M.J. (2002) Phylogenies and community

ecology. Annual Review of Ecology and Systematics, 33, 475–505.

Weiher, E., Werf, A. Van Der, Thompson, K., Roderick, M., Garnier, E. & Werf, V. Der. (1999)

Challenging Theophrastus : A common core list of plant traits for functional ecology.

Journal of Vegetation Science, 10, 609–620. 141

Westoby, M., Falster, D.S., Moles, A.T., Vesk, P.A. & Wright, I.J. (2002) Plant ecological

strategies: Some leading dimensions of variation between species. Annual Review of

Ecology and Systematics, 33, 125–159.

Whipple, A.A., R.M. Grossinger, & F.W. Davis (2011) Shifting baselines in a California oak

savanna: nineteenth century data to inform restoration scenarios. Restoration Ecology,

19, 88–101.

White, P.S., & J.L. Walker (1997) Approximating nature’s variation: selecting and using

reference information in restoration ecology. Restoration Ecology, 5, 338–349.

Whitfeld, T.J.S., A.G. Lodge, A.M. Roth, & P.B. Reich (2014) Community phylogenetic

diversity and abiotic site characteristics inﬂuence abundance of the invasive plant

Rhamnus cathartica L. Journal of Plant Ecology, 7, 202–209.

Wiens, J.J., & C.H. Graham (2005) Niche conservatism: integrating evolution, ecology, and

conservation biology. Annual Review of Ecology, Evolution, and Systematics, 36, 519–

539.

Wilkerson, M.L., Ward, K.L., Williams, N.M., Ullmann, K.S. & Young, T.P. (2014)

Diminishing returns from higher density restoration seedings suggest trade-offs in

polinator seed mixes. Restoration Ecology, 22, 782–789.

Willard, D.A., & T.M. Cronin (2007) Paleoecology and ecosystem restoration: case studies from

Chesapeake Bay and the Florida Everglades. Frontiers in Ecology and the Environment,

5, 491– 498.

Williams, A.P., C.D. Allen, C.I. Millar, T.W. Swetnam, J. Michaelsen, C.J. Still, & S.W. Leavitt

(2010) Forest responses to increasing aridity and warmth in the southwestern United 142

States. Proceedings of the National Academy of Sciences of the United States of

America, 107:21289–21294.

Williams, J.W., & S.T. Jackson (2007) Novel climates, no-analog communities, and ecological

surprises. Frontiers in Ecology and the Environment, 5, 475–482.

Willis, C.G., B. Ruhfel, R.B. Primack, A.J. Miller-Rushing, & C.C. Davis (2008) Phylogenetic

patterns of species loss in Thoreau’s woods are driven by climate change. Proceedings of

the National Academy of Sciences of the United States of America, 105, 17029–17033.

Willis, K.J., R.M. Bailey, S.A. Bhagwat, & H.J.B. Birks (2010) Biodiversity baselines,

thresholds and resilience: testing predictions and assumptions using palaeoecological

data. Trends in Ecology and Evolution, 25, 583–591.

Wing, S.L., & E.D. Currano (2013) Plant response to a global greenhouse event 56 million years

ago. American Journal of Botany, 100, 1234–1254.

Wortley, L., Hero, J.M. & Howes, M. (2013) Evaluating ecological restoration success: A review

of the literature. Restoration Ecology, 21, 537–543.

Young, T.P., Stuble, K.L., Balachowski, J.A. & Werner, C.M. (2016) Using priority effects to

manipulate competitive relationships in restoration. Restoration Ecology.

Zanne, A.E., Tank, D.C., Cornwell, W.K. et al. (2014) Three keys to the radiation of

angiosperms into freezing environments. Nature, 506, 89–92.

Zedler, J.B., & J.C. Callaway (2000) Evaluating the progress of engineered tidal wetlands.

Ecological Engineering, 15, 211–225.

Zirbel, C.R., Bassett, T., Grman, E. & Brudvig, L.A. (2017) Plant functional traits and

environmental conditions shape community assembly and ecosystem functioning during

restoration. Journal of Applied Ecology. 143

APPENDIX ONE

Supplementary material from chapter two

Supplement 1. Additional diversity metrics

Table 1: Diversity metrics for remnant and restored study sites and restoration seed mixes. “Restored (trans.)” includes only those species found in the transect-based plot surveys of restored prairies, while “Restored (all)” also includes species found via directed search throughout study sites.

Species richness Raw MPD SES.MPD Raw MNTD SES.MNTD

Remnant 50.4 ± 2.2 261.5 ± 1.5 0.37 ± 0.15 71.6 ± 1.6 0.37 ± 0.16

Restored, trans. 48.3 ± 1.6 244.5 ± 2.5 -1.21 ± 0.24 52.6 ± 1.5 -1.44 ± 0.15

Restored, all 83.2 ± 2.7 242.8 ± 2.5 -0.19 ± 0.27 42.8 ± 1.2 -0.56 ± 0.17

Seed mixes 54.6 ± 4.7 234.8 ± 4.8 -1.07 ± 0.53 41.7 ± 2.0 -0.73 ± 0.28

Table 2. Results of phylogenetic diversity tests with non-native species excluded show similar patterns to those observed with all species included (Compare to Figure 1 in manuscript).

DF F P

SES.MPD, site 1,58 20.145 <0.0001

SES.MPD, plot 1,58 23.606 <0.0001

SES.MNTD, site 1,58 26.342 <0.0001

SES.MNTD, plot 1,58 5.355 0.0242

144

Table S1. Site information for restored prairies.

The Ball Ecological Restoration Ball Horticultural Dupage 2003 0.9 Project Bambrick Park Bambrick Park Cook 2009 1.6 Chicago Park District - Forty-First Street Bioretention Cook 2006 0.7 41st Chicago Park District - McCormick Bird Sanctuary Cook 2002 2.4 BBS Chicago Park District - Burnham Nature Sanctuary Cook 1998 13.1 BNS Chicago Park District - Burnham Centennial Prairie, N Cook 2009 3.4 BCP Chicago Park District - Burnham Centennial Prairie, SE Cook 2009 2.5 BCP Chicago Park District - Burnham Centennial Prairie, SW Cook 2009 12.3 BCP Harvey Creek Conservation Area Harvey Creek Park DeKalb 2003 6.9 Orland Grassland, North Orland Grassland Cook 2009 8.6 Orland Grassland, Phoenix Orland Grassland Cook 2009 3.4 Orland Grassland, South Orland Grassland Cook 2009 69.5 Orland Grassland, Henslow's Peak Orland Grassland Cook 2009 299.8 Peck Farm, Miller Parcel Peck Farm Park Kane 2009 9.4 Peck Farm, B Site Peck Farm Park Kane 2003 28.7 Peck Farm, F Parcel Peck Farm Park Kane 2010 19.2 Peck Farm, C Parcel Peck Farm Park Kane 2012 7.3 Swami Temple Sri Venkateswara S.T. Kane 2008 23.4 Wildwood Nature Center Wildwood Cook 2002 0.4

145

Table S2. Plant taxa found at remnant and restored study sites and in restoration seed mixes.

Restored (trans.)” includes only those species found in the transect-based plot surveys of restored prairies, while “Restored (all)” also includes species found via directed search throughout study sites. Ones and zeros indicate true/present and false/absent, respectively.

Species Family Native Remnant Restored Restored Seed sites sites sites mixes (trans.) (all) Ruellia humilis Acanthaceae 1 0 0 0 1 Acorus calamus Acoraceae 1 0 0 0 1 Alisma subcordatum Alismataceae 1 0 0 0 1 Sagittaria latifolia Alismataceae 1 0 0 0 1 Salsola kali Amaranthaceae 0 1 0 0 0 Amaranthus albus Amaranthaceae 0 0 1 1 0 Chenopodium album Amaranthaceae 0 0 1 1 0 Allium stellatum Amaryllidaceae 1 0 0 1 0 Allium sp. Amaryllidaceae 0 0 1 1 0 Allium canadense Amaryllidaceae 1 1 0 0 1 Allium cernuum Amaryllidaceae 1 1 1 1 1 Toxicodendron radicans Anacardiaceae 1 1 0 0 0 Rhus glabra Anacardiaceae 1 1 0 1 0 Rhus typhina Anacardiaceae 1 0 0 1 1 Sanicula gregaria Apiaceae 1 1 0 0 0 Cicuta maculata Apiaceae 1 0 0 0 1 Oxypolis rigidior Apiaceae 1 1 0 0 0 Pastinaca sativa Apiaceae 0 1 0 0 0 Angelica atropurpurea Apiaceae 1 0 0 0 1 Conioselinum chinense Apiaceae 1 0 0 1 0 Sium suave Apiaceae 1 1 0 0 1 Zizia sp. Apiaceae 0 0 1 1 0 Zizia aptera Apiaceae 1 1 0 0 1 Daucus carota Apiaceae 0 1 1 1 0 Eryngium yuccifolium Apiaceae 1 1 1 1 1 Zizia aurea Apiaceae 1 1 1 1 1 Apocynum Apocynaceae 1 0 0 1 0 androsaemifolium Asclepias sullivantii Apocynaceae 1 0 0 0 1 Asclepias longifolia Apocynaceae 1 0 0 0 1 Asclepias sp. Apocynaceae 0 0 1 1 0 Asclepias viridiflora Apocynaceae 1 1 0 0 1 Apocynum cannabinum Apocynaceae 1 1 1 1 0 Asclepias incarnata Apocynaceae 1 0 1 1 1 146

Asclepias verticillata Apocynaceae 1 1 1 1 1 Asclepias tuberosa Apocynaceae 1 1 1 1 1 Asclepias syriaca Apocynaceae 1 1 1 1 1 Camassia scilloides Asparagaceae 1 0 0 0 1 Asparagus officinalis Asparagaceae 0 1 0 0 0 Maianthemum Asparagaceae 1 1 0 0 0 canadense Maianthemum Asparagaceae 1 1 0 0 0 racemosum Maianthemum stellatum Asparagaceae 1 1 0 0 1 Cichorium intybus Asteraceae 0 0 0 1 0 Krigia biflora Asteraceae 1 1 0 0 0 Cirsium muticum Asteraceae 1 1 0 0 0 Arctium minus Asteraceae 0 0 0 1 0 Vernonia gigantea Asteraceae 1 0 0 0 1 Pilosella caespitosa Asteraceae 0 1 0 0 0 Hieracium longipilum Asteraceae 1 0 0 1 0 Arnoglossum Asteraceae 1 1 0 0 0 plantagineum Senecio aureus Asteraceae 1 1 0 0 0 Doellingeria umbellata Asteraceae 1 1 0 0 0 Aster linariifolius Asteraceae 1 1 0 0 0 Symphyotrichum Asteraceae 1 1 0 0 0 praealtum Symphyotrichum firmum Asteraceae 1 1 0 0 0 Symphyotrichum Asteraceae 1 0 0 1 0 drummondii Symphyotrichum shortii Asteraceae 1 0 0 0 1 Rudbeckia speciosa Asteraceae 1 0 0 0 1 Rudbeckia fulgida Asteraceae 1 0 0 0 1 Solidago uliginosa Asteraceae 1 1 0 0 0 Solidago ohioensis Asteraceae 1 1 0 0 0 Solidago sempervirens Asteraceae 0 0 0 1 0 Euthamia Asteraceae 1 1 0 0 0 gymnospermoides Artemisia campestris Asteraceae 1 0 0 0 1 Antennaria Asteraceae 1 1 0 0 0 plantaginifolia Bidens vulgata Asteraceae 1 0 0 1 0 Cosmos sulphureus Asteraceae 0 0 0 0 1 Bidens frondosa Asteraceae 1 1 0 0 0 Bidens comosa Asteraceae 1 0 0 0 1 Gaillardia pulchella Asteraceae 0 0 0 0 1 Liatris scariosa Asteraceae 1 0 0 0 1 Liatris ligulistylis Asteraceae 1 0 0 1 0 147

Eupatorium maculatum Asteraceae 1 0 0 0 1 Helianthus petiolaris Asteraceae 0 0 0 0 1 Helianthus sp. Asteraceae 0 0 0 1 0 Helianthus maximiliani Asteraceae 1 0 0 1 0 Cirsium altissimum Asteraceae 1 0 1 1 0 Cirsium arvense Asteraceae 0 0 1 1 0 Cirsium vulgare Asteraceae 0 0 1 1 0 Hieracium sp. Asteraceae 0 0 1 1 0 Prenanthes sp. Asteraceae 0 0 1 1 0 Lactuca serriola Asteraceae 0 0 1 1 0 Lactuca canadensis Asteraceae 1 1 0 1 0 Nabalus asper Asteraceae 1 1 0 0 1 Sonchus arvensis Asteraceae 0 0 1 1 0 Sonchus asper Asteraceae 0 0 1 1 0 Sonchus oleraceus Asteraceae 0 0 1 1 0 Boltonia asteroides Asteraceae 1 0 0 1 1 Erigeron sp. Asteraceae 0 0 1 1 0 Rudbeckia sp. Asteraceae 0 0 1 1 0 Rudbeckia laciniata Asteraceae 1 0 0 1 1 Solidago sp. Asteraceae 0 0 1 1 0 Solidago riddellii Asteraceae 1 1 0 0 1 Solidago ptarmicoides Asteraceae 1 1 0 0 1 Artemisia vulgaris Asteraceae 0 0 1 1 0 Bidens cernua Asteraceae 1 0 0 1 1 Bidens trichosperma Asteraceae 1 0 1 1 0 Ageratina altissima Asteraceae 1 1 0 1 0 Brickellia eupatorioides Asteraceae 1 1 0 0 1 Liatris sp. Asteraceae 0 0 0 1 1 Liatris cylindracea Asteraceae 1 1 0 0 1 Helianthus hirsutus Asteraceae 1 0 1 1 0 Helianthus mollis Asteraceae 1 1 0 0 1 Helianthus occidentalis Asteraceae 1 1 0 0 1 Ambrosia trifida Asteraceae 1 0 1 1 0 Silphium perfoliatum Asteraceae 1 0 1 1 1 Verbesina alternifolia Asteraceae 1 0 1 1 1 Cirsium sp. Asteraceae 0 1 1 1 0 Taraxacum campylodes Asteraceae 0 1 1 1 0 Arnoglossum Asteraceae 1 0 1 1 1 atriplicifolium Erechtites hieraciifolius Asteraceae 1 1 1 1 0 Symphyotrichum sp. Asteraceae 0 1 1 1 0 Symphyotrichum Asteraceae 1 1 0 1 1 sericeum 148

Symphyotrichum Asteraceae 1 1 1 1 0 pilosum Symphyotrichum Asteraceae 1 0 1 1 1 cordifolium Conyza canadensis Asteraceae 1 1 1 1 0 Erigeron philadelphicus Asteraceae 1 1 1 1 0 Erigeron annuus Asteraceae 1 1 1 1 0 Erigeron strigosus Asteraceae 1 1 1 1 0 Rudbeckia triloba Asteraceae 1 0 1 1 1 Solidago missouriensis Asteraceae 1 1 1 1 0 Solidago nemoralis Asteraceae 1 1 0 1 1 Solidago canadensis Asteraceae 1 1 1 1 0 Solidago gigantea Asteraceae 1 1 1 1 0 Liatris spicata Asteraceae 1 1 0 1 1 Eupatorium serotinum Asteraceae 1 1 1 1 0 Eupatorium perfoliatum Asteraceae 1 1 0 1 1 Heliopsis helianthoides Asteraceae 1 0 1 1 1 Echinacea purpurea Asteraceae 1 0 1 1 1 Helianthus divaricatus Asteraceae 1 1 0 1 1 Ambrosia artemisiifolia Asteraceae 1 1 1 1 0 Ratibida pinnata Asteraceae 1 1 1 1 1 Silphium Asteraceae 1 1 1 1 1 terebinthinaceum Silphium laciniatum Asteraceae 1 1 1 1 1 Silphium integrifolium Asteraceae 1 1 1 1 1 Cirsium discolor Asteraceae 1 1 1 1 1 Vernonia fasciculata Asteraceae 1 1 1 1 1 Lactuca sp. Asteraceae 0 1 1 1 1 Packera paupercula Asteraceae 1 1 1 1 1 Symphyotrichum Asteraceae 1 1 1 1 1 oolentangiense Symphyotrichum laeve Asteraceae 1 1 1 1 1 Symphyotrichum Asteraceae 1 1 1 1 1 novaeangliae Symphyotrichum Asteraceae 1 1 1 1 1 ericoides Symphyotrichum Asteraceae 1 1 1 1 1 lanceolatum Rudbeckia Asteraceae 1 1 1 1 1 subtomentosa Rudbeckia hirta Asteraceae 1 1 1 1 1 Solidago juncea Asteraceae 1 1 1 1 1 Solidago speciosa Asteraceae 1 1 1 1 1 Solidago rigida Asteraceae 1 1 1 1 1 Euthamia graminifolia Asteraceae 1 1 1 1 1 Leucanthemum vulgare Asteraceae 0 1 1 1 1 149

Achillea millefolium Asteraceae 0 1 1 1 1 Antennaria neglecta Asteraceae 1 1 1 1 1 Coreopsis tripteris Asteraceae 1 1 1 1 1 Coreopsis palmata Asteraceae 1 1 1 1 1 Coreopsis lanceolata Asteraceae 1 1 1 1 1 Helenium autumnale Asteraceae 1 1 1 1 1 Liatris pycnostachya Asteraceae 1 1 1 1 1 Liatris aspera Asteraceae 1 1 1 1 1 Eupatorium altissimum Asteraceae 1 1 1 1 1 Echinacea pallida Asteraceae 1 1 1 1 1 Helianthus pauciflorus Asteraceae 1 1 1 1 1 Helianthus Asteraceae 1 1 1 1 1 grosseserratus Parthenium integrifolium Asteraceae 1 1 1 1 1 Impatiens capensis Balsaminaceae 1 1 1 1 1 Corylus americana Betulaceae 1 1 0 0 0 Corylus sp. Betulaceae 0 0 1 1 1 Hackelia virginiana Boraginaceae 1 1 0 0 0 Lithospermum incisum Boraginaceae 1 1 0 0 0 Cordia fasciculata Boraginaceae 0 0 0 0 1 Lithospermum Boraginaceae 1 1 0 0 1 canescens Arabidopsis lyrata Brassicaceae 1 1 0 0 0 Brassica nigra Brassicaceae 0 0 0 1 0 Thlaspi arvense Brassicaceae 0 0 1 1 0 Alliaria petiolata Brassicaceae 0 0 1 1 0 Lepidium campestre Brassicaceae 0 0 1 1 0 Lepidium virginicum Brassicaceae 1 1 0 1 0 Lepidium densiflorum Brassicaceae 0 0 1 1 0 Barbarea vulgaris Brassicaceae 0 0 1 1 0 Opuntia humifusa Cactaceae 1 1 0 0 0 Campanula americana Campanulaceae 1 0 0 0 1 Lobelia kalmii Campanulaceae 1 1 0 0 0 Campanula aparinoides Campanulaceae 1 1 0 0 0 Lobelia cardinalis Campanulaceae 1 0 0 1 1 Lobelia spicata Campanulaceae 1 1 0 1 1 Lobelia siphilitica Campanulaceae 1 0 1 1 1 Viburnum sp. Caprifoliaceae 0 0 0 1 0 Viburnum prunifolium Caprifoliaceae 1 1 0 0 0 Lonicera x Caprifoliaceae 0 1 0 0 0 Dipsacus fullonum Caprifoliaceae 0 1 0 0 0 Valeriana ciliata Caprifoliaceae 1 1 0 0 0 Viburnum recognitum Caprifoliaceae 1 0 1 1 0 150

Viburnum lentago Caprifoliaceae 1 0 1 1 0 Symphoricarpos sp. Caprifoliaceae 0 0 1 1 0 Dipsacus laciniatus Caprifoliaceae 0 1 0 1 0 Cerastium glomeratum Caryophyllaceae 0 1 0 0 0 Minuartia stricta Caryophyllaceae 1 1 0 0 0 Silene stellata Caryophyllaceae 1 1 0 0 0 Stellaria media Caryophyllaceae 0 0 1 1 0 Stellaria longifolia Caryophyllaceae 1 0 1 1 0 Cerastium sp. Caryophyllaceae 0 0 1 1 0 Cerastium fontanum Caryophyllaceae 0 0 1 1 0 Cerastium pumilum Caryophyllaceae 0 0 1 1 0 Dianthus armeria Caryophyllaceae 0 0 1 1 0 Silene latifolia Caryophyllaceae 0 0 1 1 0 Silene vulgaris Caryophyllaceae 0 0 1 1 0 Moehringia lateriflora Caryophyllaceae 1 1 1 1 0 Celastrus scandens Celastraceae 1 1 0 0 0 Helianthemum bicknellii Cistaceae 1 0 0 0 1 Tradescantia sp. Commelinaceae 0 0 0 1 0 Tradescantia ohiensis Commelinaceae 1 1 1 1 1 Cuscuta sp. Convolvulaceae 0 1 0 0 0 Cuscuta glomerata Convolvulaceae 1 1 0 0 0 Cuscuta pentagona Convolvulaceae 1 0 0 1 0 Calystegia catesbeiana Convolvulaceae 1 1 0 0 0 Convolvulus arvensis Convolvulaceae 0 0 1 1 0 Ipomoea purpurea Convolvulaceae 0 0 1 1 0 Calystegia sepium Convolvulaceae 1 1 1 1 0 Cornus amomum Cornaceae 1 1 0 0 0 Cornus racemosa Cornaceae 1 1 0 0 0 Cornus sericea Cornaceae 1 0 1 1 0 Penthorum sedoides Crassulaceae 1 0 0 1 1 Cladium mariscoides Cyperaceae 1 1 0 0 0 Scleria triglomerata Cyperaceae 1 1 0 0 0 Rhynchospora alba Cyperaceae 1 1 0 0 0 Eleocharis sp. Cyperaceae 0 1 0 0 0 Eleocharis compressa Cyperaceae 1 1 0 0 0 Eleocharis tenuis Cyperaceae 1 1 0 0 0 Eleocharis elliptica Cyperaceae 1 1 0 0 0 Bolboschoenus fluviatilis Cyperaceae 1 0 0 0 1 Cyperus schweinitzii Cyperaceae 1 1 0 0 0 Cyperus sp. Cyperaceae 0 0 0 1 0 Cyperus esculentus Cyperaceae 1 0 0 1 0 Scirpus sp. Cyperaceae 0 1 0 0 0 151

Scirpus cyperinus Cyperaceae 1 0 0 0 1 Carex sartwellii Cyperaceae 1 0 0 0 1 Carex brevior Cyperaceae 1 0 0 0 1 Carex scoparia Cyperaceae 1 0 0 0 1 Carex tribuloides Cyperaceae 1 0 0 0 1 Carex normalis Cyperaceae 1 0 0 0 1 Carex stipata Cyperaceae 1 0 0 0 1 Carex stricta Cyperaceae 1 0 0 0 1 Carex emoryi Cyperaceae 1 0 0 0 1 Carex buxbaumii Cyperaceae 1 1 0 0 0 Carex sprengelii Cyperaceae 1 0 0 0 1 Carex pellita Cyperaceae 1 1 0 0 0 Carex lacustris Cyperaceae 1 0 0 0 1 Carex frankii Cyperaceae 1 0 0 0 1 Carex comosa Cyperaceae 1 0 0 0 1 Carex hystericina Cyperaceae 1 0 0 0 1 Carex lupulina Cyperaceae 1 0 0 0 1 Carex lurida Cyperaceae 1 0 0 0 1 Carex grayi Cyperaceae 1 0 0 0 1 Eleocharis acuta Cyperaceae 0 1 0 0 1 Schoenoplectus Cyperaceae 1 0 0 1 1 tabernaemontani Scirpus atrovirens Cyperaceae 1 0 0 1 1 Scirpus pendulus Cyperaceae 1 0 1 1 0 Carex bebbii Cyperaceae 1 0 1 1 0 Carex molesta Cyperaceae 1 0 1 1 0 Carex bushii Cyperaceae 0 0 1 1 0 Carex blanda Cyperaceae 1 0 1 1 0 Eleocharis erythropoda Cyperaceae 1 1 1 1 0 Carex sp. Cyperaceae 0 1 1 1 0 Carex cristatella Cyperaceae 1 0 1 1 1 Carex vulpinoidea Cyperaceae 1 0 1 1 1 Carex annectens Cyperaceae 1 0 1 1 1 Carex bicknellii Cyperaceae 1 1 1 1 1 Vaccinium angustifolium Ericaceae 1 1 0 0 0 Euphorbia geyeri Euphorbiaceae 0 1 0 0 0 Euphorbia cyparissias Euphorbiaceae 0 1 0 0 0 Euphorbia maculata Euphorbiaceae 1 1 0 0 0 Euphorbia nutans Euphorbiaceae 1 0 0 1 0 Acalypha virginica Euphorbiaceae 0 0 1 1 0 Euphorbia corollata Euphorbiaceae 1 1 1 1 1 Senna hebecarpa Fabaceae 1 0 0 0 1 Gleditsia triacanthos Fabaceae 1 0 0 1 0 152

Apios americana Fabaceae 1 1 0 0 0 Amphicarpaea bracteata Fabaceae 1 1 0 0 0 Robinia pseudoacacia Fabaceae 0 1 0 0 0 Vicia americana Fabaceae 1 1 0 0 0 Lathyrus palustris Fabaceae 1 1 0 0 0 Melilotus officinalis Fabaceae 0 0 0 1 0 Baptisia bracteata Fabaceae 1 0 0 1 1 Psoralea tenuiflora Fabaceae 1 1 0 0 1 Securigera varia Fabaceae 0 0 1 1 0 Lotus corniculatus Fabaceae 0 0 1 1 0 Astragalus canadensis Fabaceae 1 0 0 1 1 Vicia sativa Fabaceae 0 0 1 1 0 Lathyrus venosus Fabaceae 1 1 0 0 1 Melilotus sp. Fabaceae 0 1 0 1 0 Melilotus alba Fabaceae 0 0 1 1 0 Medicago lupulina Fabaceae 0 0 1 1 0 Trifolium hybridum Fabaceae 0 0 1 1 0 Trifolium repens Fabaceae 0 0 1 1 0 Desmanthus illinoensis Fabaceae 1 0 1 1 1 Baptisia alba Fabaceae 1 0 1 1 1 Dalea purpurea Fabaceae 1 0 1 1 1 Trifolium sp. Fabaceae 0 1 1 1 0 Trifolium pratense Fabaceae 0 1 1 1 0 Chamaecrista Fabaceae 1 1 1 1 1 fasciculata Amorpha canescens Fabaceae 1 1 1 1 1 Dalea candida Fabaceae 1 1 1 1 1 Desmodium illinoense Fabaceae 1 1 1 1 1 Desmodium canadense Fabaceae 1 1 1 1 1 Lespedeza capitata Fabaceae 1 1 1 1 1 Quercus sp. Fagaceae 0 0 0 1 0 Quercus velutina Fagaceae 1 1 0 0 0 Quercus rubra Fagaceae 1 1 0 0 0 Gentiana sp. Gentianaceae 0 0 0 1 0 Gentianopsis crinita Gentianaceae 1 1 0 0 0 Bartonia virginica Gentianaceae 1 1 0 0 0 Gentianella quinquefolia Gentianaceae 1 1 0 0 1 Gentiana puberulenta Gentianaceae 1 1 0 1 1 Gentiana andrewsii Gentianaceae 1 1 1 1 1 Gentiana alba Gentianaceae 1 1 1 1 1 Geranium maculatum Geraniaceae 1 1 0 0 0 Hypericum sp. Hypericaceae 0 1 0 0 0 153

Hypericum Hypericaceae 1 1 0 0 0 sphaerocarpum Hypericum kalmianum Hypericaceae 1 1 0 0 0 Hypericum gentianoides Hypericaceae 1 1 0 0 1 Hypericum ascyron Hypericaceae 1 0 0 1 1 Hypericum perforatum Hypericaceae 0 1 0 1 0 Hypericum punctatum Hypericaceae 1 1 0 0 1 Hypoxis hirsuta Hypoxidaceae 1 1 0 0 0 Sisyrinchium sp. Iridaceae 0 0 0 1 0 Sisyrinchium albidum Iridaceae 1 1 0 0 0 Iris sp. Iridaceae 0 0 0 1 0 Sisyrinchium campestre Iridaceae 1 0 1 1 0 Iris virginica Iridaceae 1 0 0 1 1 Carya cordiformis Juglandaceae 1 1 0 0 0 Juncus greenei Juncaceae 1 1 0 0 0 Juncus acuminatus Juncaceae 1 1 0 0 0 Luzula multiflora Juncaceae 1 1 0 0 0 Juncus interior Juncaceae 1 1 0 0 0 Juncus balticus Juncaceae 1 1 0 0 0 Juncus brachycephalus Juncaceae 1 1 0 0 0 Juncus torreyi Juncaceae 1 1 0 1 0 Juncus sp. Juncaceae 0 1 0 1 0 Juncus effusus Juncaceae 1 0 0 1 1 Juncus dudleyi Juncaceae 1 1 0 1 1 Juncus tenuis Juncaceae 1 1 1 1 0 Teucrium sp. Lamiaceae 0 1 0 0 0 Scutellaria parvula Lamiaceae 1 1 0 0 0 Stachys sp. Lamiaceae 0 1 0 0 0 Stachys hispida Lamiaceae 1 1 0 0 0 Stachys palustris Lamiaceae 1 0 0 1 0 Mentha arvensis Lamiaceae 1 0 0 1 0 Clinopodium glabrum Lamiaceae 1 1 0 0 0 Lycopus virginicus Lamiaceae 1 1 0 0 0 Lycopus sp. Lamiaceae 0 1 0 0 0 Lycopus uniflorus Lamiaceae 1 1 0 0 0 Teucrium canadense Lamiaceae 1 1 0 0 1 Mentha canadensis Lamiaceae 0 1 0 0 1 Monarda punctata Lamiaceae 1 0 0 1 1 Nepeta cataria Lamiaceae 0 0 1 1 0 Glechoma hederacea Lamiaceae 0 0 1 1 0 Pycnanthemum Lamiaceae 1 0 1 1 1 tenuifolium Prunella vulgaris Lamiaceae 1 1 1 1 0 154

Physostegia virginiana Lamiaceae 1 1 1 1 1 Pycnanthemum Lamiaceae 1 1 1 1 1 virginianum Monarda fistulosa Lamiaceae 1 1 1 1 1 Lycopus americanus Lamiaceae 1 1 1 1 1 Lilium michiganense Liliaceae 1 1 0 0 0 Lilium philadelphicum Liliaceae 1 1 0 0 0 Linum sulcatum Linaceae 1 1 0 0 0 Lythrum salicaria Lythraceae 0 1 0 1 0 Lythrum alatum Lythraceae 1 1 0 1 1 Abutilon theophrasti Malvaceae 0 0 0 1 0 Hibiscus trionum Malvaceae 0 0 1 1 0 Hibiscus palustris Malvaceae 1 0 1 1 1 Morus alba Moraceae 0 1 0 0 0 Comptonia peregrina Myricaceae 1 1 0 0 0 Fraxinus sp. Oleaceae 0 1 0 1 0 Ludwigia alternifolia Onagraceae 1 0 0 0 1 Oenothera pilosella Onagraceae 1 1 0 0 0 Oenothera sp. Onagraceae 0 1 0 0 0 Oenothera perennis Onagraceae 1 0 0 1 0 Epilobium coloratum Onagraceae 1 0 1 1 0 Gaura biennis Onagraceae 1 1 0 1 0 Oenothera biennis Onagraceae 1 1 1 1 0 Cypripedium candidum Orchidaceae 1 1 0 0 0 Spiranthes sp. Orchidaceae 0 1 0 0 0 Spiranthes Orchidaceae 1 1 0 0 0 magnicamporum Spiranthes cernua Orchidaceae 1 1 0 0 0 Platanthera flava Orchidaceae 1 1 0 0 0 Liparis loeselii Orchidaceae 1 1 0 0 0 Liparis liliifolia Orchidaceae 1 1 0 0 0 Pedicularis lanceolata Orobanchaceae 1 1 0 0 0 Castilleja coccinea Orobanchaceae 1 1 0 0 0 Agalinis sp. Orobanchaceae 0 1 0 0 0 Agalinis aspera Orobanchaceae 1 0 0 1 0 Agalinis purpurea Orobanchaceae 1 1 0 0 0 Agalinis tenuifolia Orobanchaceae 1 1 0 0 0 Pedicularis canadensis Orobanchaceae 1 1 1 1 1 Oxalis stricta Oxalidaceae 1 1 1 1 0 Mimulus sp. Phrymaceae 0 0 0 0 1 Mimulus ringens Phrymaceae 1 0 0 1 1 Chelone glabra Plantaginaceae 1 0 0 0 1 Penstemon pallidus Plantaginaceae 1 0 0 0 1 155

Linaria vulgaris Plantaginaceae 0 0 1 1 0 Penstemon sp. Plantaginaceae 0 0 1 1 0 Penstemon grandiflorus Plantaginaceae 1 0 1 1 0 Plantago lanceolata Plantaginaceae 0 0 1 1 0 Plantago major Plantaginaceae 0 0 1 1 0 Plantago rugelii Plantaginaceae 1 0 1 1 0 Veronica sp. Plantaginaceae 0 0 1 1 0 Veronica arvensis Plantaginaceae 0 0 1 1 0 Plantago sp. Plantaginaceae 0 1 1 1 0 Penstemon digitalis Plantaginaceae 1 1 1 1 1 Veronicastrum Plantaginaceae 1 1 1 1 1 virginicum Elymus hystrix Poaceae 1 0 0 0 1 Dactylis glomerata Poaceae 0 1 0 0 0 Lolium multiflorum Poaceae 0 0 0 0 1 Koeleria macrantha Poaceae 1 0 0 0 1 Avena sativa Poaceae 0 0 0 0 1 Calamagrostis Poaceae 1 1 0 0 0 inexpansa Agrostis hyemalis Poaceae 1 1 0 0 0 Cinna arundinacea Poaceae 1 0 0 0 1 Panicum oligosanthes Poaceae 1 1 0 0 0 Panicum leibergii Poaceae 1 1 0 0 0 Panicum boreale Poaceae 1 1 0 0 0 Panicum capillare Poaceae 1 0 0 1 0 Setaria sp. Poaceae 0 0 0 1 0 Setaria faberi Poaceae 0 1 0 0 0 Sporobolus compositus Poaceae 1 1 0 0 0 Sporobolus cryptandrus Poaceae 1 0 0 0 1 Calamovilfa longifolia Poaceae 1 1 0 0 0 Bouteloua sp. Poaceae 0 0 0 1 0 Muhlenbergia sp. Poaceae 0 1 0 0 0 Muhlenbergia glomerata Poaceae 1 1 0 0 0 Muhlenbergia mexicana Poaceae 1 1 0 0 0 Glyceria sp. Poaceae 0 0 1 1 0 Glyceria striata Poaceae 1 1 0 0 1 Stipa spartea Poaceae 1 1 0 0 1 Bromus arvensis Poaceae 0 0 1 1 0 Hordeum jubatum Poaceae 1 0 1 1 0 Elymus trachycaulus Poaceae 1 1 0 0 1 Elymus elymoides Poaceae 1 0 1 1 0 Elymus repens Poaceae 0 0 1 1 0 Festuca sp. Poaceae 0 0 1 1 0 156

Festuca pratensis Poaceae 0 0 1 1 0 Lolium perenne Poaceae 0 0 1 1 0 Deschampsia cespitosa Poaceae 1 1 0 0 1 Rostraria cristata Poaceae 0 1 0 0 1 Calamagrostis Poaceae 1 1 0 0 1 canadensis Poa compressa Poaceae 0 0 1 1 0 Poa nemoralis Poaceae 0 1 0 0 1 Digitaria ischaemum Poaceae 0 0 1 1 0 Panicum sp. Poaceae 0 1 0 0 1 Dichanthelium sp. Poaceae 0 0 1 1 0 Setaria pumila Poaceae 0 0 1 1 0 Muhlenbergia frondosa Poaceae 1 0 1 1 0 Bromus inermis Poaceae 0 1 1 1 0 Elymus virginicus Poaceae 1 0 1 1 1 Festuca arundinacea Poaceae 0 1 1 1 0 Phalaris arundinacea Poaceae 0 1 1 1 0 Agrostis gigantea Poaceae 0 0 1 1 0 Poa sp. Poaceae 0 1 1 1 0 Poa pratensis Poaceae 0 1 1 1 0 Phragmites australis Poaceae 0 1 1 1 0 Echinochloa crusgalli Poaceae 0 1 1 1 0 Panicum acuminatum Poaceae 1 1 1 1 0 Bromus kalmii Poaceae 1 1 1 1 1 Elymus canadensis Poaceae 1 1 1 1 1 Phleum pratense Poaceae 0 1 1 1 1 Leersia oryzoides Poaceae 1 1 1 1 1 Sorghastrum nutans Poaceae 1 1 1 1 1 Andropogon gerardii Poaceae 1 1 1 1 1 Schizachyrium Poaceae 1 1 1 1 1 scoparium Panicum virgatum Poaceae 1 1 1 1 1 Sporobolus heterolepis Poaceae 1 1 1 1 1 Spartina pectinata Poaceae 1 1 1 1 1 Bouteloua curtipendula Poaceae 1 1 1 1 1 Phlox glaberrima Polemoniaceae 1 1 0 0 1 Phlox pilosa Polemoniaceae 1 1 0 1 1 Polygala senega Polygalaceae 1 1 0 0 0 Polygala polygama Polygalaceae 1 1 0 0 0 Polygala verticillata Polygalaceae 1 1 1 1 0 Polygala sanguinea Polygalaceae 1 1 1 1 0 Persicaria longiseta Polygonaceae/Ph 0 0 0 1 0 ytolaccaceae 157

Polygonum tenue Polygonaceae/Ph 1 1 0 0 0 ytolaccaceae Rumex acetosella Polygonaceae/Ph 0 1 0 0 0 ytolaccaceae Persicaria bungeana Polygonaceae/Ph 1 0 1 1 0 ytolaccaceae Persicaria amphibia Polygonaceae/Ph 1 0 1 1 0 ytolaccaceae Persicaria pensylvanica Polygonaceae/Ph 1 0 1 1 0 ytolaccaceae Persicaria lapathifolia Polygonaceae/Ph 1 0 1 1 0 ytolaccaceae Polygonum sp. Polygonaceae/Ph 0 0 1 1 0 ytolaccaceae Polygonum aviculare Polygonaceae/Ph 0 0 1 1 0 ytolaccaceae Rumex crispus Polygonaceae/Ph 0 0 1 1 0 ytolaccaceae Phytolacca americana Polygonaceae/Ph 1 0 1 1 0 ytolaccaceae Persicaria maculosa Polygonaceae/Ph 0 1 1 1 0 ytolaccaceae Polygonum Polygonaceae/Ph 1 1 1 1 0 ramosissimum ytolaccaceae Portulaca oleracea Portulacaceae 0 0 0 1 0 Lysimachia quadriflora Primulaceae 1 1 0 0 0 Lysimachia lanceolata Primulaceae 1 1 0 0 0 Lysimachia ciliata Primulaceae 1 1 0 0 0 Lysimachia thyrsiflora Primulaceae 1 1 0 0 0 Dodecatheon meadia Primulaceae 1 1 1 1 1 Thalictrum revolutum Ranunculaceae 1 1 0 0 0 Anemone sp. Ranunculaceae 0 1 0 0 0 Anemone quinquefolia Ranunculaceae 1 1 0 0 0 Anemone virginiana Ranunculaceae 1 0 1 1 0 Thalictrum dasycarpum Ranunculaceae 1 1 1 1 1 Anemone cylindrica Ranunculaceae 1 1 1 1 1 Anemone canadensis Ranunculaceae 1 1 1 1 1 Frangula alnus Rhamnaceae 0 1 0 0 0 Rhamnus cathartica Rhamnaceae 0 1 0 1 0 Ceanothus americanus Rhamnaceae 1 1 0 0 1 Geum laciniatum Rosaceae 1 1 0 0 0 Rubus idaeus Rosaceae 1 0 0 1 0 Rubus flagellaris Rosaceae 1 1 0 0 0 Rubus setosus Rosaceae 1 1 0 0 0 Agrimonia gryposepala Rosaceae 1 1 0 0 0 Agrimonia parviflora Rosaceae 1 1 0 0 0 Rosa gallica Rosaceae 0 1 0 0 0 158

Rosa setigera Rosaceae 1 0 0 1 0 Rosa multiflora Rosaceae 0 1 0 0 0 Rosa arkansana Rosaceae 1 0 0 1 0 Spiraea alba Rosaceae 1 1 0 0 0 Prunus sp. Rosaceae 0 0 0 1 0 Prunus serotina Rosaceae 1 1 0 0 0 Prunus pumila Rosaceae 1 1 0 0 0 Aronia prunifolia Rosaceae 1 1 0 0 0 Geum macrophyllum Rosaceae 1 0 1 1 0 Geum aleppicum Rosaceae 1 1 0 1 0 Rubus allegheniensis Rosaceae 1 0 1 1 0 Agrimonia pubescens Rosaceae 1 0 1 1 0 Rosa sp. Rosaceae 0 1 0 0 1 Rosa blanda Rosaceae 1 0 1 1 0 Fragaria vesca Rosaceae 1 0 1 1 0 Potentilla norvegica Rosaceae 1 0 1 1 0 Spiraea sp. Rosaceae 0 0 1 1 0 Geum sp. Rosaceae 0 1 1 1 0 Geum canadense Rosaceae 1 1 1 1 0 Rubus occidentalis Rosaceae 1 1 1 1 0 Rubus pensilvanicus Rosaceae 1 1 1 1 0 Fragaria virginiana Rosaceae 1 1 1 1 0 Potentilla simplex Rosaceae 1 1 1 1 0 Crataegus sp. Rosaceae 0 1 1 1 0 Potentilla arguta Rosaceae 1 1 1 1 1 Galium mollugo Rubiaceae 0 0 0 1 0 Galium obtusum Rubiaceae 1 1 0 0 0 Cephalanthus Rubiaceae 1 0 0 1 0 occidentalis Galium boreale Rubiaceae 1 1 0 0 1 Galium concinnum Rubiaceae 1 1 1 1 0 Galium aparine Rubiaceae 1 1 1 1 0 Ptelea trifoliata Rutaceae 1 1 0 0 0 Populus tremuloides Salicaceae 1 1 0 0 0 Salix petiolaris Salicaceae 1 1 0 0 0 Salix sp. Salicaceae 0 0 0 1 0 Salix myricoides Salicaceae 1 1 0 0 0 Salix humilis Salicaceae 1 1 0 0 0 Salix discolor Salicaceae 1 1 0 0 0 Salix amygdaloides Salicaceae 1 1 0 0 0 Salix interior Salicaceae 1 1 0 0 0 Populus deltoides Salicaceae 1 0 1 1 0 Comandra umbellata Santalaceae 1 1 0 0 0 159

Acer negundo Sapindaceae 1 1 0 1 0 Acer saccharum Sapindaceae 1 1 1 1 0 Saxifraga pensylvanica Saxifragaceae 1 1 0 0 0 Heuchera richardsonii Saxifragaceae 1 1 1 1 1 Scrophularia marilandica Scrophulariaceae 1 0 1 1 0 Smilax lasioneura Smilacaceae 1 1 0 0 0 Physalis subglabrata Solanaceae 1 1 0 0 0 Solanum ptychanthum Solanaceae 1 0 0 1 0 Physalis sp. Solanaceae 0 1 0 1 0 Physalis virginiana Solanaceae 1 1 0 1 0 Physalis longifolia Solanaceae 1 0 1 1 0 Solanum dulcamara Solanaceae 0 1 0 1 0 Solanum carolinense Solanaceae 0 0 1 1 0 Physalis heterophylla Solanaceae 1 1 1 1 0 Sparganium eurycarpum Sparganiaceae 1 0 0 0 1 Typha latifolia Typhaceae 1 1 0 0 0 Typha sp. Typhaceae 0 0 1 1 0 Ulmus sp. Ulmaceae 0 1 0 1 0 Verbena urticifolia Verbenaceae 1 0 1 1 0 Verbena stricta Verbenaceae 1 0 1 1 1 Verbena hastata Verbenaceae 1 1 1 1 1 Viola obliqua Violaceae 1 1 0 0 0 Viola macloskeyi Violaceae 1 1 0 0 0 Viola pedata Violaceae 1 1 0 0 0 Viola lanceolata Violaceae 1 1 0 0 0 Viola sororia Violaceae 1 0 1 1 0 Viola pedatifida Violaceae 1 1 0 0 1 Viola sp. Violaceae 0 1 1 1 0 Viola sagittata Violaceae 1 1 1 1 0 Vitis sp. Vitaceae 0 0 0 1 0 Parthenocissus Vitaceae 1 1 0 0 0 quinquefolia Parthenocissus inserta Vitaceae 1 1 0 0 0 Vitis vulpina Vitaceae 1 1 1 1 0

160

Table S3. Contributions of plant taxa to differences in taxonomic community composition (Bray- Curtis dissimilarity) between remnant and restored prairies. Taxa shown are responsible for 50% cumulative contribution to dissimilarity. Mean refers to average contribution of each taxon to overall dissimilarity; Remnant and Restored are average relative abundance of each species (frequency of occurrence) in the respective site types; and Cum. sum is the cumulative contribution to Bray-Curtis dissimilarity between site types.

Mean S.D. Remnant Restored Cum. sum

Poa sp 0.008031 0.004484 0.86842 0.10526 0.009848

Erigeron sp 0.00751 0.004792 0 0.73684 0.019056

Ambrosia artemisiifolia 0.007403 0.004938 0.18421 0.84211 0.028133

Oxalis stricta 0.007239 0.005038 0.15789 0.78947 0.03701

Symphyotrichum pilosum 0.007213 0.005035 0.15789 0.78947 0.045854

Taraxacum campylodes 0.007086 0.005282 0.21053 0.78947 0.054543

Carex sp 0.00704 0.005194 0.97368 0.31579 0.063175

Elymus canadensis 0.007005 0.005109 0.13158 0.73684 0.071764

Medicago lupulina 0.006886 0.004923 0 0.68421 0.080207

Rosa sp 0.006478 0.004859 0.65789 0 0.088151

Penstemon digitalis 0.006429 0.005287 0.05263 0.63158 0.096034

Oenothera biennis 0.006238 0.005358 0.13158 0.63158 0.103683

Euphorbia corollata 0.006141 0.005005 0.63158 0.05263 0.111213

Comandra umbellata 0.006005 0.004702 0.63158 0 0.118576

Poa pratensis 0.00598 0.005389 0.05263 0.57895 0.125909

Trifolium repens 0.005931 0.005278 0 0.57895 0.133181

Daucus carota 0.005865 0.005429 0.39474 0.78947 0.140372

Symphyotrichum oolentangiense 0.005815 0.005246 0.57895 0.05263 0.147502

Sorghastrum nutans 0.005813 0.00539 0.71053 0.36842 0.15463

Spartina pectinata 0.005783 0.005058 0.60526 0.10526 0.161722 161

Pycnanthemum virginianum 0.00578 0.005289 0.73684 0.36842 0.168809

Fragaria virginiana 0.005754 0.005242 0.63158 0.26316 0.175864

Panicum acuminatum 0.005738 0.005145 0.60526 0.15789 0.1829

Symphyotrichum ericoides 0.005701 0.005215 0.65789 0.31579 0.18989

Liatris spicata 0.005647 0.004941 0.57895 0 0.196815

Solidago gigantea 0.005623 0.005282 0.57895 0.21053 0.20371

Cornus racemosa 0.005514 0.005134 0.55263 0 0.210472

Bouteloua curtipendula 0.005307 0.00533 0.15789 0.52632 0.216979

Achillea millefolium 0.005295 0.00521 0.52632 0.10526 0.223472

Juncus tenuis 0.00527 0.005245 0.07895 0.52632 0.229934

Solidago rigida 0.005247 0.005392 0.42105 0.52632 0.236368

Sporobolus heterolepis 0.005205 0.005408 0.5 0.31579 0.24275

Tradescantia ohiensis 0.005165 0.005305 0.52632 0.42105 0.249083

Calamagrostis canadensis 0.00514 0.005273 0.5 0 0.255386

Ratibida pinnata 0.005121 0.005417 0.57895 0.57895 0.261665

Silphium integrifolium 0.005065 0.005202 0.5 0.31579 0.267875

Euthamia graminifolia 0.005056 0.005317 0.44737 0.42105 0.274075

Rudbeckia subtomentosa 0.00504 0.005479 0.02632 0.47368 0.280255

Helianthus grosseserratus 0.004991 0.005157 0.5 0.26316 0.286375

Viola sp 0.00498 0.005103 0.5 0.15789 0.292481

Coreopsis palmata 0.00493 0.005329 0.10526 0.47368 0.298526

Eryngium yuccifolium 0.004905 0.005285 0.36842 0.42105 0.304541

Panicum virgatum 0.004706 0.005307 0.21053 0.42105 0.310311

Sisyrinchium albidum 0.004671 0.005092 0.47368 0 0.316038

Coreopsis tripteris 0.004633 0.0051 0.44737 0.15789 0.321719

Poa nemoralis 0.004626 0.005016 0.47368 0 0.327391 162

Andropogon gerardii 0.004575 0.005267 0.84211 0.57895 0.333001

Maianthemum stellatum 0.004513 0.004833 0.47368 0 0.338534

Calystegia sepium 0.004456 0.005093 0.42105 0.15789 0.343998

Cirsium discolor 0.004456 0.005151 0.34211 0.31579 0.349462

Echinacea purpurea 0.004413 0.005335 0 0.42105 0.354874

Elymus repens 0.004398 0.005283 0 0.42105 0.360266

Veronica arvensis 0.004386 0.005279 0 0.42105 0.365644

Solidago juncea 0.004376 0.004999 0.42105 0.15789 0.37101

Zizia aurea 0.004357 0.005053 0.39474 0.21053 0.376352

Prunella vulgaris 0.004352 0.005079 0.26316 0.36842 0.381689

Trifolium hybridum 0.00431 0.005183 0 0.42105 0.386973

Echinacea pallida 0.004297 0.00507 0.07895 0.42105 0.392242

Lithospermum canescens 0.004282 0.004866 0.44737 0 0.397493

Rudbeckia hirta 0.004275 0.005536 0.73684 0.73684 0.402735

Monarda fistulosa 0.004259 0.005643 0.65789 0.89474 0.407957

Silphium terebinthinaceum 0.004253 0.004762 0.44737 0.05263 0.413173

Symphyotrichum laeve 0.004234 0.005224 0.15789 0.36842 0.418364

Silphium laciniatum 0.004204 0.005224 0.15789 0.36842 0.423519

Heliopsis helianthoides 0.004201 0.005083 0 0.42105 0.42867

Phlox pilosa 0.004174 0.004728 0.44737 0 0.433788

Parthenium integrifolium 0.004127 0.004998 0.39474 0.10526 0.438848

Lycopus americanus 0.004096 0.005062 0.36842 0.15789 0.44387

Erigeron strigosus 0.004085 0.004988 0.26316 0.31579 0.448879

Poa compressa 0.004062 0.004887 0 0.42105 0.45386

Scutellaria parvula 0.004037 0.005226 0.39474 0 0.45881

Phlox glaberrima 0.004024 0.004818 0.42105 0 0.463744 163

Euthamia gymnospermoides 0.004015 0.005094 0.39474 0 0.468667

Potentilla simplex 0.003972 0.004906 0.39474 0.05263 0.473537

Sonchus asper 0.003969 0.005325 0 0.36842 0.478404

Bromus inermis 0.003952 0.00511 0.07895 0.36842 0.48325

Verbena hastata 0.003949 0.00516 0.05263 0.36842 0.488092

Asclepias syriaca 0.003946 0.005236 0.02632 0.36842 0.492931

Galium obtusum 0.003935 0.004728 0.42105 0 0.497757

164

Table S4. Site characteristics and biodiversity. The effects of area and years since restoration was initiated on three measures of biodiversity at the site and plot levels. Effects of area and years since restoration on biodiversity were tested using linear models. Biodiversity response variables at two spatial scales were included. Site scale diversity metrics were calculated as the sum of all plots for each site, and Plot scale diversity metrics were the mean plot-level values for each site. All biodiversity metrics were calculated using presence-absence data as described in the text. Adjusted R2 are reported for each overall model.

Df F P R2 a. Species richness, site

Overall model 2,16 0.158 0.855 0.019

Area 1 0.063 0.805

Years restored 1 0.253 0.622 b. Species richness, plot

Overall model 2,16 0.121 0.356 0.121

Area 1 0.054 0.820

Years restored 1 2.154 0.162 c. SES.MPD, site

Overall model 2,16 1.144 0.343 0.125

Area 1 2.173 0.160

Years restored 1 0.116 0.738 d. SES.MPD, plot

Overall model 2,16 0.544 0.591 -0.005

Area 1 0.349 0.563

Years restored 1 0.740 0.402 e. SES.MNTD, site

Overall model 2,16 1.106 0.355 0.012

Area 1 0.219 0.647

Years restored 1 1.994 0.178 f. SES.MNTD, plot

Overall model 2,16 4.471 0.029 0.359 165

Area 1 5.708 0.087

Years restored 1 3.233 0.091 166

Table S5. Effects of soil properties on multivariate measures of plant community diversity. Results of distance-based redundancy analysis (rda). MPD-based community distance matrix calculated using comdist function, and MNTD-based matrix using comdistnt function in the vegan package in R. Soil properties were plot-level means calculated for each site. Adjusted R2 reported.

Main effects DF Variance F P R2 a. Taxonomic Overall model 3 6.835 1.230 0.017 0.039

GSM 1 2.223 1.200 0.130

pH 1 2.621 1.415 0.013

EC 1 1.991 1.075 0.310

Residuals 14 25.936

b. MPD Overall model 3 5131.9 0.946 0.938 -0.010

GSM 1 1638.7 0.907 0.905

pH 1 1798.7 0.995 0.439

EC 1 1694.4 0.937 0.874

Residuals 14 25309.2

c. MNTD Overall model 3 996.5 2.426 0.001 0.201

GSM 1 364.93 2.665 0.014

pH 1 446.06 3.258 0.003

EC 1 185.51 1.355 0.237

Residuals 14 1916.96 167

APPENDIX TWO

Supplementary material from chapter three Appendix 2A. Results from n = 30 species analysis. Table 1. Best models of time-to-germination ranked by Akaike information criterion (AIC) for 30 prairie species. K is the number of factors in the model, ∆AIC is the difference in AIC between each model and the model with the lowest AIC, w is the model weight and Cw is the cumulative model weight. Shown are all models with ∆AIC ≤ 4. Treat. = treatment, V1 – V4 = st th the 1 – 4 phylogenetic axes. ESwidth = E:S ratio measured by width, ESwidth = E:S ratio measured by length, ESarea = E:S ratio measured by area, L = length, W = width, H = height and VS = shape, measured as the variance between L, W and H.

Model factors K AIC ∆AIC W Cw

Treat. + V1 + V2 + ESlength + ESwidth + ESarea + 10 19330.11 0.00 0.31 0.31 L + W + H

Treat. + V1 + V2 + ESlength + ESarea + L + W + 9 19331.10 1.00 0.19 0.51 H

Treat. + V1 + V2 + ESlength + ESwidth + ESarea + 11 19331.14 1.04 0.19 0.69 L + W + H + Mass

Treat. + V1 + V2 + ESlength + ESwidth + L + W 9 19331.43 1.32 0.16 0.86 + H

Treat. + V1 + V2 + ESlength + ESwidth + ESarea + 12 19331.69 1.58 0.14 1.00 L + W + H + Mass + VS

~1 (Intercept-only model) 0 19772.37 442.26 0.00 1.00

168

Table 2. Model-averaged estimate, standard error, and 95% confidence interval (CRI) for all parameters in best fitting models (∆AIC ≤ 4) for 30 prairie species. An asterisk (*) indicates a variable that was a significant contributor to the averaged model based on 95% CRI. GA = gibberellic acid treatment, CON = untreated control.

Model term Estimate SE 95% CRI

Treatment – GA* -0.52 0.07 -0.65, -0.39

Treatment – CON* -0.86 0.07 -1, -0.73

V1* 0.10 0.03 0.04, 0.16

V2* 0.32 0.05 0.23, 0.41

Length* 0.25 0.07 0.12, 0.38

Width* -0.27 0.07 -0.4, -0.14

Height* -0.48 0.05 -0.58, -0.37

Variance (shape) -0.11 0.09 -0.3, 0.07

ESlength* 0.11 0.04 0.03, 0.19

ESwidth 0.07 0.04 -0.01, 0.15

ESarea 0.06 0.04 -0.01, 0.14

Mass 0.07 0.06 -0.05, 0.19

169

Table 3. Phylogenetic signal of measured traits for 30 prairie species. K is the observed value of phylogenetic signal relative to a Brownian motion model of evolution. P is significance of phylogenetic signal based on a randomization test with 1,000 permutations.

Trait K P

Length 0.071 0.002

Width 0.091 0.002

Height 0.064 0.011

Variance 0.084 0.003

ESlength 0.034 0.039

ESwidth 0.034 0.021

ESarea 0.025 0.060

Mass (log) 0.083 0.001

170

Appendix 2B. Results from intraspecific analyses of the effects of traits on time-to-germination.

Figure 1. Effects of seed traits on individual time to germination. Seeds subjected to different germination treatments differed in germination response in 22 species out of 30 species tested. Mass was a significant predictor of germination response in 6 species (purple). Seed length was significant in 1 species; width, height and shape variance were significant in 2 species each (blue); and ESarea , ESlength and ESwidth were significant in 3, 9 and 7 species, respectively (green). In 13 species tested, no seed traits were significant predictors of intraspecific germination response.