THE EFFECT OF SEED MIX DESIGN ON PLANT REPRODUCTIVE OUTPUT IN AN
EXPERIMENTAL PRAIRIE SYSTEM

By

Brandon Edward Latorre

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

Plant Biology – Master of Science

2024

ABSTRACT

Ecological restoration reverses habitat loss and destruction by promoting the

establishment of native communities. Despite this, most restoration outcomes are

notoriously unpredictable. Restoration practitioners need to make important decisions

while restoring ecological communities and these decisions can have lasting impacts on

the outcome of the restoration. In particular, managerial decisions regarding the

richness and location sourcing of a seed mix are especially important in seed based

restoration, such as prairie ecosystems. These decisions can influence reproductive

output. For example, seeds sourced from a distant location might produce individual

plants that are maladapted to the local environment and produce fewer seeds.

Additionally, higher initial seed mix richness may correlate with a decrease in seeding

density, resulting in a plant population that is too small, sparse, or clumped for optimal

reproduction to occur. In this study, I sought to answer the following question: How does

initial seed mix richness and location sourcing affect plant reproductive output across

different biological scales? To answer this question, the Brudvig Lab restored twelve

experimental prairies in the Kellogg Biological Station, where we manipulated initial

seed mix richness and the number of sources. We collected total plant and floral

abundance data by counting the number of plants and flowers across various transects.

I estimated reproductive output by counting the total number of seeds of nearly 400

individuals of two dominant prairie forbs, Ratibida pinnata and Echinacea purpurea. I

derived values that allowed the scaling of our data, such as the number of

inflorescences per plant, plant count per area, developed seed count per seed head,

and more. Our results suggest that initial seed mix design is important for the

reproductive output of target plant populations in restoration. Namely, I found that

restorations sown with three sources and 71 species at lower seeding rates decrease

reproductive output for Echinacea across all biological scales, from individual

infructescences to the entire population, but only for Ratibida at the population scale.

There are predictable lasting impacts of seed mix design on reproductive output. Our

study suggests prioritizing seeding rate per species in seed mixes, ideally with a high

species richness from fewer sources.

ACKNOWLEDGEMENTS

I am very grateful for the help that I received at Michigan State University. The

Kellogg Biological Station was a wonderful place to conduct research. I am grateful for

all the financial assistance provided by the Kellogg Biological Station and MSU that

made everything possible. I would not have been able to conduct this research without

their help.

The AGEP Community was instrumental to my success at MSU. Thank you so

much for providing a safe space where I found my community. I am grateful for all the

opportunities that I had to practice research presentations, mentor undergraduate and

graduate students, travel and more.

A warm thank you to Mark Scriber and the Entomology Department, for believing

in me and helping me restore prairie habitat in my local neighborhood. Without the

Scriber Scholar Award for Butterfly Conservation, the establishment of the public Wild

Wonder Community Prairie Garden would not have been possible.

Thank you Bruce Martin and Christoper Warneke for all the help with statistical

analysis in R. I learned so much from both of you and feel confident working with R.

Thank you Lars Brudvig, for your limitless patience and guidance through the

tribulations of graduate school.

iii

TABLE OF CONTENTS

INTRODUCTION.............................................................................................................. 1

METHODS........................................................................................................................ 6

RESULTS........................................................................................................................11

DISCUSSION................................................................................................................. 27

CONCLUSION................................................................................................................30

BIBLIOGRAPHY............................................................................................................ 31

APPENDIX..................................................................................................................... 37

iv

INTRODUCTION

Habitat restoration promotes biodiversity and ecosystem functioning in degraded

ecosystems (Suding, 2011). A cornerstone of any successful restoration is the

promotion of viable populations of plant species characteristic of the target ecosystem

(i.e., target species) under restoration (Lesica and Allendorf, 1999). Most assessment

criteria focus on plant species richness and diversity, yet these metrics provide coarse

information about target plant populations (  Ruiz-Jaen and Aide, 2005). More precise

restoration can be informed by demographic responses of target species that can help

inform management and in turn, improve the predictability of success in restoration

practice (Ballare et al., 2017). It is critical to understand how management decisions

affect reproductive output (production of viable seeds), since species persistence is

dependent on the replacement of individuals (Huang et al., 2013; Wilsey, 2021).

Reproductive output influences the trajectory of a population’s size (Menges,

1995; Molano-Flores and Hendrix, 1999). Thus, practitioners must make decisions and

engage in activities that increase reproductive output of target plant populations.

Recruitment via seed propagation may only take place when environmental conditions

are right. In one case study, soil disturbance and burning increased seedling recruitment

threefold in a rare prairie forb (Menges, 1995); thus, promoting reproductive output can

help to ensure recruitment during correct environmental conditions. Moreover, although

new recruits may be rare in the short term, consistent yearly contributions to a seed

bank is an important process in the demographic cycle of many species and is critical to

the persistence of many plant populations (Bullock, 2000).

New efforts are needed to test effects of alternative restoration strategies on

reproductive output of target species during the restoration process. Many factors can

affect reproductive output at various biological scales: from individual reproductive

structures (infructescences), to individual plants, to an entire plant population. For

example, the infructescences of individual plants may produce varying numbers of

seeds, due to their number of florets, damage, production of nectar/pollen that may

attract pollinators, or other factors related to environmental conditions within a

restoration (Klinkhamer and Jong, 1990; Cresswell, 1999; Cardel and Koptur, 2010).

These differences in the infructescence may scale up to entire plants; however,

1

additional variation in reproductive allocation across plants within the same population

may result in differences in reproductive output: some plants might invest heavily in

flowering and, later, seeds, while others, especially new recruits, might invest

disproportionately in structures related to establishment like stem elongation and root

development (Wenk and Falster, 2015). Further, at the scale of a population, if the

population size of a specific plant species is too small, then it will be less likely to attract

pollinators and may thus have low reproductive output due to low rates of pollination

(Wilcock and Nieland, 2002; Kwak and Bekker, 2006). Large populations may also have

more reproductive output simply because they contain more individuals. Despite this,

Mustajarvi et al. (2001) found that plants in sparser and lower density populations

invested more in their flowers, which led to a stronger pollinator response and thus

enhanced reproductive output.

Initial seed mix design may influence the configuration, size and identity of target

plant species populations and new experiments that alter seed mix design are

warranted to more clearly understand the effects on target plant species reproductive

output. During seed based restoration, a core consideration is the number of species

sown. Budgetary constraints can lead practitioners to use species poor seed mixes

(Barr et al., 2017). This is problematic, as high richness seed mixes increase plant

species richness (Leps et al., 2007; Carter and Blair, 2012; Kirmer et al., 2012; Schmidt

et al., 2022). Since plant diversity and pollinator diversity are positively correlated

(Haddad et al., 2009), high richness plant assemblages may attract diverse pollinator

communities that promote plant reproduction (Frund et al., 2010; Albrecht et al., 2012,

Wei et al., 2021). Despite the potential benefits of restoring ecosystems using a high

richness seed mix, a necessary consequence may be lowering of seeding rate. The

total seeding rate (total number of seeds sown per area) is often held constant due to

budgetary constraints (Barak et al. 2022), so when seed mixes become more diverse,

each species is sown with fewer seeds per area.

Manipulating the species-level seeding rate (the number of seeds sown per

species) has implications for the reproductive output of restored populations. The higher

seeding rate typical of low richness seed mixes can enhance viability of target

populations (Kunin, 1997). Higher seeding rates promote establishment and can help

2

eliminate weedy species more effectively in the short term (Stevenson et al., 1995;

Harker et al., 2003). High density seeding treatments can also result in greater species

richness and cover of native seeded species, which can also increase seeds per fruit

and seeds per flower (Feldman, 2006; Carter and Blair, 2012). Enhanced establishment

of target species will result in greater floral abundance and an abundance of nectar

resources, which has shown to initiate a stronger pollinator response (Johnson et al.,

2004; Fowler et al., 2016; Li et al., 2021).

Promoting a stronger pollinator response is critical to ensuring plant reproduction.

Pollinator behavior is important for the success of sown species. Pollinators are

expected to adapt their foraging behavior in accordance to surrounding resource

scarcity (Kunin and Iwasa, 1996). When floral resources are scarce (i.e. plant

populations are small), pollinators are likely to adopt a more generalist strategy and

tend to visit flowers indiscriminately (Kunin and Iwasa, 1996). When floral resources are

plentiful, it is more cost-effective for pollinators to specialize in one type of flower or

similar suite of flowers (Kunin and Iwasa, 1996). Specialization in one type of species

increases efficiency of resource collection by decreasing pollinator foraging costs

(Armbruster, 2017). Higher conspecific visitation will lead to enhanced conspecific

pollen deposition, which is expected to increase plant reproductive output at the

infructescence scale (Waites and Ågren, 2004). With larger population sizes, there may

be greater reproductive output at the population scale and thus restoration success.

The initial seed mix design may have a lasting impact on the restoration. In

addition to species richness and seeding rate, the location where practitioners source

seed may also be critical to restoration success. There is a lack of rigorous empirical

research on how they affect plant reproduction during restoration. Examples of sourcing

methods include: 1) local provenancing uses locally-sourced seeds; 2) predictive

provenancing uses southerly-sourced seeds, attempting to “keep up” with climate

change and use genotypes that are more suited for predicted future climates; 3)

admixture provenancing uses a combination of many different sources and aims to

maximize evolutionary potential by promoting genetic diversity while ignoring gene flow

dynamics; and 4) Composite provenancing incorporates genotypes that are more

distant, but attempts to imitate natural gene flow (Breed et al., 2013). Large plant

3

populations have adapted to climatic and environmental clines specific to their

respective regions (Joshi et al., 2001; Leimu and Fischer, 2008; Woods, 2012).

Because of this, the most common recommendation is local provenancing, that is, as

close to the restoration site as possible (Lesica and Allendorf, 1999; Hufford and Mazer,

2003). Locally sourced seeds are more likely to harbor locally adapted genotypes, and

may therefore establish plant populations that persist and reproduce better than plant

populations from non-locally sourced seeds (Lesica and Allendorf, 1999). This

technique may be obstructive to the restoration process, however, because there are

often few locally adapted populations for any given site (Mortlock, 2000; Borders et al.,

2011). Using seeds from only one or a few local source(s) may introduce founder effects

via population bottlenecks in the restoration; this may lead to decreased overall plant

reproductive output due to lack of genetic diversity and adaptive potential (Whiteley et

al., 2015). On the other hand, admixture provenancing increases evolutionary novelty,

which is expected to enhance population resilience in the face of unpredictable

challenges that future environmental change may posit (Choi, 2004; Harris et al., 2006;

Broadhurst et al., 2008; Havens et al., 2015). In an experiment with wildflower strips

restored from local and nonlocal seeds, Bischoff et al. (2010) found that the plants with

high genotypic diversity (i.e., from multiple sources) increased productivity,

establishment, and seed production for only one of the species surveyed. Yet,

Bucharova et al (2017) found that species fitness decreased with increasing

geographical distance of seed origin, supporting the use of local seed sources.

Disentangling the effects of seed sourcing can be instructive and provide necessary

guidance to restore viable populations.

Experiments are needed to determine the effects of seed mix design on restored

species reproductive output. Experiments overcome uncontrollable differences in

management and site history that may influence restoration outcomes. For example,

Grman and Brudvig (2013) found that the composition, diversity, and density of seed

mixes applied, in addition to site age, were critical in explaining differences among

restoration sites. Any one of these factors might obscure the effects of seeded richness

and seed source locality, the latter of which is poorly explored (though see Pizza et al.

2023). Moreover, I know of no set of ongoing restoration projects that allow for the study

4

of varying numbers of seed sources, much less a set of projects that allow for the

independent and interactive study of seed source number and sown species richness.

There is a need for controlled experiments that manipulate initial seed mix design and

provenancing to shed more light on how these decisions influence plant reproductive

output and viability by exploring seed sets directly.

The Brudvig Lab created a field-scale experiment to test the effects of the

number of sown species (12 vs. 71) and seed sources per species (1 vs. 3), on

reproductive output at three different scales: 1) the individual infructescence, within a

plant ; 2) all of the infructescences in an individual plant; and 3) an entire population. To

do this, the Brudvig Lab restored an experimental prairie system and explored the

reproductive output of two prairie forbs that are commonly re-established during prairie

restoration: Echinacea purpurea (purple coneflower) and Ratibida pinnata (gray-head

coneflower).

Research Questions:

1. How does the number of sown plant species affect plant reproductive output?

2. How does the number of sown seed sources affect plant reproductive output?

3. What is the role of pollination in driving effects?

5

Experimental System

METHODS

Prairies are an ideal ecosystem to explore how differences in management

influence reproductive output at various scales. Prairies take just a few years to

establish from seed because they are dominated by herbaceous plants such as forbs

and grasses. Additionally, they are the most endangered ecosystem in North America,

yet provide valuable resources for pollinators (Samson and Knopf, 1996; Gilgert and

Vaughan, 2011).

The Conservation Lands Experiment (CLE) Prairies at the W.K. Kellogg

Biological Station (KBS) in Michigan (Figure 1) consists of twelve prairie restorations

established on former agricultural land. The experiment was prepared by removing all

non-prairie old field vegetation through the application of glyphosate herbicide twice in

the summer of 2015, with subsequent seeding in the same year in late November and

early December. In 2019 and 2021, the prairies were managed with prescribed burns.

The size of the restorations varies between 0.5-3.5 ha. Each site consists of two

half-sites, where the number of sown sources was manipulated. In low richness

half-sites, twelve focal species were sown, which includes four forbs, three C4 grasses,

three legumes, and two C3 grasses (Catano et al., 2021). In high richness half-sites,

seventy-one species were sown, consisting of the twelve focal species from the low

richness half-sites and an additional fifty-nine species. Total seeding rate was held

constant (9.87 kg/ha), resulting in the twelve focal species sown with more seeds in low

richness half-sites and fewer seeds in the high richness half-sites (see Table 1 in

Appendix).

In addition, four different sourcing treatments were established at the different

restoration sites to test the effects of admixture provenancing: seeds were sourced from

local, non-local northern, and southern seed distributors, as well as all three sources

mixed. The twelve sites were sown with seed either sourced exclusively from one region

or from all regions, with two sites for each single region (in total six single-sourced

sites), and the other six sites included a mix of all three regions. I focused our analyses

on comparing single (1) to multiple (3) sources to test our focal questions about seed

6

source number and due to the limited replication when looking at each single sourced

site.

Figure 1. Conservation Lands Experiment
Illustration of the Conservation Lands Experiment Design. The Kellogg Biological
Station is located in Southwest Michigan. Seeds were sourced from three different
locales: local (L), non-local northern (M) and southern (S). Red sites harbor prairie
communities with initial seeding treatment from all three sources while blue sites harbor
prairie communities with initial seeding treatment from a single source.

Study Species

I conducted all analyses using the information collected from two different

species in the Asteraceae family: Echinacea purpurea and Ratibida pinnata, hereafter

referred to only by their genus name, Echinacea and Ratibida. These two species are

part of the focal set of twelve species that were sown in all twenty-four half-sites, and

are commonly included in many prairie seed mixes. I focused on these two species

because they had high establishment rates across all CLE treatments, permitting tests

of our focal questions. Echinacea and Ratibida reproduce via clonal propagation to form

7

ramets, that is, an independent clonal colony. Ramets were only considered different if

they were at least 1 meter away from other ramets.

Seed Procurement

I collected seeds across the CLE from two plants: Echinacea and Ratibida. In the

beginning of Summer 2019, I tagged eight ramets for each species, distributed

throughout each half-site. Then, I randomly selected a seed head within each ramet.

After the inflorescence began to develop into an infructescence, I installed pollinator

exclusion bags to prevent seed dispersal and deter predation. Once seeds were fully

developed, I clipped the tagged seed heads with scissors and carefully placed them in

dry paper bags. I then extracted seeds using forceps.

Proxies for Pollinator Success

Fruit development was variable and some seeds were not fully developed. I used

various properties of individual seeds to differentiate developed from underdeveloped

seeds. Underdeveloped seeds were often discolored, disfigured, smaller, and more

brittle. I counted the number of developed and underdeveloped seeds separately in

each seed head to determine seed production as a measure of reproductive output. I

considered only developed seeds for measurements of reproductive output; however, I

counted both developed and undeveloped seeds. Because flowers of species in the

Asteraceae can form either developed seeds when pollinated or undeveloped seeds

when unpollinated, the ratio of developed to developed and undeveloped seeds can

serve as a measure of pollination rate (Brudvig et al. 2015).

Flower-to-fruit conversion ratio= [Developed seed count / (Developed seed +

Underdeveloped seed count)]

To more clearly understand the role of pollination in our target species, I

conducted a pollinator exclusion experiment and quantified seed production in bagged

and open infructescences across the CLE Prairies. In each of the 24 half-sites, I bagged

a single reproductive structure before bloom to deter pollinator visitation and thus

animal-mediated pollen deposition. After reproductive structures senesced into

infructescences, I removed seeds using forceps and counted all developed seeds.

Infructescences that were not visited by pollinations (i.e, infructescences that had

exclusion bags installed) produced considerably fewer seeds compared to open

8

infructescences, usually approaching zero (Latorre, 2019). Echinacea purpurea and

Ratibida pinnata highly depend on outcrossing events by pollinators for seed

production.

Inflorescence and Population Surveys

To estimate plant-level reproductive output, population size, and population-scale

reproductive output, I combined the infructescence-level seed count data with surveys

of the two focal species in fixed-area plots, within each half site. Jenn Lau counted the

total number of ramets and inflorescences in five 25m^2 plots (5mX5m) within each half

site during September 2019. If floral density was too high to count, she subsampled a

1m^2 area (1mX1m) within a given plot.

Reproductive output throughout different scales

Using reproductive output data (developed seed count), inflorescence and

population surveys, I determined differences among treatment groups at three scales:

1) infructescence scale was the average number of developed seeds per infructescence

in a half site; 2) ramet scale was the average number of developed seeds per ramet in a

half site; 3) population scale was the estimated total number of developed seeds per

125 m^2 in a half site.

To evaluate reproductive output at the infructescence scale, I counted all

developed seeds in each infructescence and derived the average value per

infructescence for each half site. To evaluate reproductive output at the ramet scale, I

multiplied average seed count per infructescence by the average number of

infructescences per ramet, for each half site. To evaluate reproductive output at the

population scale, I multiplied reproductive output at the ramet scale (see above) by the

total number of plants counted in each half site. In instances where 5x5m plots were

subsampled, I estimated the number of plants per 5x5m plot (e.g., I multiplied a 1x1m

subsample by 25).

Statistical Methods

I performed all analyses in R (Version 4.1.2, 2021). Response variables included

Echinacea and Ratibida reproductive output for infructescences, ramets, and population

scales, in addition to flower-to-fruit conversion ratios, analyzed as separate models. For

each model, fixed effects included the number of sown species (twelve and seventy-one

9

species), number of sown sources (1 and 3 sources), and the interaction between

number of species and number of sources sown. Site was used as a random effect.

Models were made using the lme4 package (Bates et al., 2015). For the seed count

data, I initially fit a Poisson distribution, but after confirming overdispersion, I fit these

models with a negative binomial distribution. The resulting models were not

overdispersed and had uniform residual plots. To test the model factors, I performed

ANOVAs using the car package (Fox, 2019). Pairwise comparisons were assessed to

look at differences between specific combinations of treatments. I plotted results using

negative binomial model predictions and 95% confidence intervals using the ggplot2

package (Wickham, 2016).

10

Echinacea infructescence scale

RESULTS

I found evidence for effects of source number, sown richness, and an interaction

between source number and sown richness on Echinacea seed count at the

infructescence scale (Table 1; Figure 2). With this significant interaction, I interpret

results in context of the interaction. Half-sites sown with three sources and 71 species

had 39-46% lower infructescence-scale seed counts than the other three treatments,

whereas none of the other treatments differed (Table 2).

Figure 2. Echinacea infructescence scale
Seed count for individual Echinacea purpurea infructescences was influenced by an
interaction between the number species sown and number of Echinacea sources sown
into a prairie restoration experiment. Seed counts were lower when prairie plantings

11

Figure 2 (cont’d)
were sown with 71 species and three Echinacea sources, relative to 12 species and
three Echinacea sources and single Echinacea source plantings. Hollow circles
represent observed seed counts, solid circles and lines represent predicted seed counts
and 95% confidence intervals.

Table 1. Results of an ANOVA testing the effects of seed source number (1 vs. 3),
seeded richness (12 vs. 71) and their interaction on infructescence-level reproductive
output for Echinacea purpurea during prairie restoration.

Treatment

Chisq

Df

p

600.1

6.3

12.8

9.5

1

1

1

1

<2.2e-16

0.0117

0.00035

0.00201

Source

Richness

Source X

Richness

Table 2. Pairwise comparisons among treatment groups with varying combinations of
seed source number (1 vs. 3) and seeded richness (12 vs. 71) were conducted to test
for differences in infructescence-level reproductive output for Echinacea purpurea
during prairie restoration.
Treatment comparison

contrast

p

3 sources, 71 species vs. 3 sources, 12 species

-0.49

<0.001

3 sources, 71 species vs. 1 source, 71 species

-0.62

0.012

3 sources, 71 species vs. 1 source, 12 species

-0.53

0.033

3 sources, 12 species vs. 1 source, 71 species

-0.13

0.592

3 sources, 12 species vs. 1 source, 12 species

-0.04

0.881

1 source, 71 species vs. 1 source, 12 species

0.09

0.47

12

Echinacea ramet scale

I found evidence for effects of source number, sown richness, and an interaction

between source number and sown richness on Echinacea seed count at the scale of

individual ramets (Table 3; Figure 3). With this significant interaction, I interpret results in

context of the interaction. Half-sites sown with three sources and 71 species had

59-62% fewer ramet-scale total seed counts than treatments with three sources and 12

species, as well as a single source and 71 species (Table 4). None of the other

treatments differed (Table 4).

Figure 3. Echinacea ramet scale
Seed count for Echinacea at the ramet scale was influenced by an interaction between
the number species sown and number of Echinacea sources sown into a prairie
restoration experiment. Seed counts were lower when prairie plantings were sown with

13

Figure 3 (cont’d)
71 species and three Echinacea sources, relative to 12 species and three Echinacea
sources and 71 species and a single Echinacea source planting. Hollow circles
represent observed seed counts, solid circles and lines represent predicted seed counts
and 95% confidence intervals.

Table 3. Results of an ANOVA testing the effects of seed source number (1 vs. 3),
seeded richness (12 vs. 71) and their interaction on ramet-level reproductive output for
Echinacea purpurea during prairie restoration.

Treatment

Chisq

Df

p

(Intercept)

312.4

Source

Richness

Source X

Richness

5.9

4.9

4.8

1

1

1

1

<2e-16

0.015

0.026

0.029

Table 4. Pairwise comparisons among treatment groups with varying combinations of
seed source number (1 vs. 3) and seeded richness (12 vs. 71) were conducted to test
for differences in ramet-level reproductive output for Echinacea purpurea during prairie
restoration.
Treatment comparison

contrast

p

3 sources, 71 species vs. 3 sources, 12 species -0.88

0.026

3 sources, 71 species vs. 1 source, 71 species

-0.96

0.015

3 sources, 71 species vs. 1 source, 12 species

-0.62

0.12

3 sources, 12 species vs. 1 source, 71 species

-0.08

0.837

3 sources, 12 species vs. 1 source, 12 species

1 source, 71 species vs. 1 source, 12 species

0.26

0.34

0.507

0.385

14

Echinacea population scale

I found evidence for effects of source number, sown richness, and an interaction

between source number and sown richness on Echinacea seed count at the scale of the

entire population (Table 5; Figure 4). With this significant interaction, I interpret results in

context of the interaction. Half-sites sown with three sources and 71 species had

85-96% fewer population -scale total seed counts than the other three treatments; other

treatments did not differ (Table 6).

Figure 4. Echinacea population scale
Seed count for all of the Echinacea purpurea individuals at the population scale was
influenced by an interaction between the number species sown and number of
Echinacea sources sown into a prairie restoration experiment. Seed counts were lower
when prairie plantings were sown with 71 species and three Echinacea sources, relative
to 12 species and three Echinacea sources and single Echinacea source plantings.

15

Figure 4 (cont’d)
Hollow circles represent observed seed counts, solid circles and lines represent
predicted seed counts and 95% confidence intervals.

Table 5. Results of an ANOVA testing the effects of seed source number (1 vs. 3),
seeded richness (12 vs. 71) and their interaction on population-level reproductive output
for Echinacea purpurea during prairie restoration.

Treatment

Chisq

Df

p

(Intercept)

200.1

Source

Richness

Source X

Richness

5.4

16.6

3.9

1

1

1

1

<2.2e-16

0.0196

0.00005

0.049

Table 6. Pairwise comparisons among treatment groups with varying combinations of
seed source number (1 vs. 3) and seeded richness (12 vs. 71) were conducted to test
for differences in population-level reproductive output for Echinacea purpurea during
prairie restoration.
Treatment comparison

contrast p

3 sources, 71 species vs. 3 sources, 12 species -3.26

<0.001

3 sources, 71 species vs. 1 source, 71 species

-1.89

0.02

3 sources, 71 species vs. 1 source, 12 species

-2.91

<0.001

3 sources, 12 species vs. 1 source, 71 species

1.37

0.102

3 sources, 12 species vs. 1 source, 12 species

0.35

0.67

1 source, 71 species vs. 1 source, 12 species

-1.01

0.226

16

Ratibida infructescence scale

I did not find any significant effects of the number of species sown or number of

Ratibida sources sown on Ratibida infructescences, nor was there evidence for an

interaction between these two factors (Figure 5, Table 7). None of the four treatment

groups differed (Table 8.)

Figure 5. Ratibida infructescence scale
There were no significant effects of the number of species sown and number of Ratibida
sources sown on Ratibida infructescence reproductive output. Hollow circles represent
observed seed counts, solid circles and lines represent predicted seed counts and 95%
confidence intervals.

17

Table 7. Results of an ANOVA testing the effects of seed source number (1 vs. 3),
seeded richness (12 vs. 71) and their interaction on infructescence-level reproductive
output for Ratibida pinnata during prairie restoration.

Treatment

Chisq

Df

p

(Intercept)

7119.6

Source

Richness

Source X

Richness

2.6

0.3

1.06

1

1

1

1

<2e-16

0.104

0.581

0.303

Table 8. Pairwise comparisons among treatment groups with varying combinations of
seed source number (1 vs. 3) and seeded richness (12 vs. 71) were conducted to test
for differences in infructescence-level reproductive output for Ratibida pinnata during
prairie restoration.
Treatment comparison

contrast p

3 sources, 71 species vs. 3 sources, 12 species

-0.04

0.581

3 sources, 71 species vs. 1 source, 71 species

-0.16

0.104

3 sources, 71 species vs. 1 source, 12 species

-0.09

0.346

3 sources, 12 species vs. 1 source, 71 species

-0.12

0.249

3 sources, 12 species vs. 1 source, 12 species

-0.05

0.614

1 source, 71 species vs. 1 source, 12 species

0.07

0.364

Ratibida ramet scale

I did not find evidence for effects on source number and sown richness on

Ratibida ramet-scale seed count; however, there was a significant interaction between

source number and richness (Table 9, Figure 6). Half-sites sown with a single source

18

and 71 species had 25% more ramet-scale seed counts than half-sites sown with a

single source and 12 species (Table 10). All other treatments did not differ (Table 10).

Figure 6. Ratibida ramet scale
Seed count for Ratibida at the ramet scale was influenced by an interaction between the
number species sown and number of Ratibida sources sown into a prairie restoration
experiment. Seed counts were higher when prairie plantings were sown with 71 species
and a single Ratibida source, relative to 12 species and single Ratibida source. Hollow
circles represent observed seed counts, solid circles and lines represent predicted seed
counts and 95% confidence intervals.

Table 9. Results of an ANOVA testing the effects of seed source number (1 vs. 3),
seeded richness (12 vs. 71) and their interaction on ramet-level reproductive output for
Ratibida pinnata during prairie restoration.

Treatment

Chisq

(Intercept)

969.45

Df

1

p

<2e-16

19

Table 9 (cont’d)

Source

Richness

Source X

Richness

0.13

0.28

90.24

1

1

1

0.715

0.599

<2e-16

Table 10. Pairwise comparisons among treatment groups with varying combinations of
seed source number (1 vs. 3) and seeded richness (12 vs. 71) were conducted to test
for differences in ramet-level reproductive output for Ratibida pinnata during prairie
restoration.
Treatment comparison

contrast p

3 sources, 71 species vs. 3 sources, 12 species -0.00805

0.599

3 sources, 71 species vs. 1 source, 71 species

-0.13

0.715

3 sources, 71 species vs. 1 source, 12 species

0.09

0.808

3 sources, 12 species vs. 1 source, 71 species

-0.12

0.732

3 sources, 12 species vs. 1 source, 12 species

0.09

0.79

1 source, 71 species vs. 1 source, 12 species

0.22

<0.001

Ratibida population scale

I found evidence suggesting an effect of sown richness and an interaction

between source number and sown richness on Ratibida population-scale seed counts

(Table 11, Figure 7). With this significant interaction, I interpret results in context of the

interaction. Half-sites sown with three sources and 71 species had 36% less

population-scale seed counts than half-sites sown with three sources and 12 species

(Table 12). Additionally, half-sites sown with a single source and 71 species had 2%

more population-scale seed counts than half-sites with a single source and 12 species

(Table 12).

20

Figure 7. Ratibida population scale
Seed count for all of the individuals at the population scale was influenced by an
interaction between the number species sown and number of Ratibida sources sown
into a prairie restoration experiment. Seed counts were lower when prairie plantings
were sown with 71 species and three Ratibida sources, relative to 12 species and three
Ratibida sources. Additionally, seed counts were higher when prairie plantings were
sown with 71 species and a single Ratibida source, relative to 12 species and single
Ratibida source. Hollow circles represent observed seed counts, solid circles and lines
represent predicted seed counts and 95% confidence intervals.

Table 11. Results of an ANOVA testing the effects of seed source number (1 vs. 3),
seeded richness (12 vs. 71) and their interaction on population-level reproductive output
for Ratibida pinnata during prairie restoration.

Treatment

Chisq

(Intercept)

1590.2

Df

1

p

<2e-16

21

Table 11 (cont’d)

Source

0.017

Richness

56257.2

Source X

Richness

30387.6

1

1

1

0.897

<2e-16

<2e-16

Table 12. Pairwise comparisons among treatment groups with varying combinations of
seed source number (1 vs. 3) and seeded richness (12 vs. 71) were conducted to test
for differences in population-level reproductive output for Ratibida pinnata during prairie
restoration.
Treatment comparison

contrast

p

3 sources, 71 species vs. 3 sources, 12 species

-0.45

<0.001

3 sources, 71 species vs. 1 source, 71 species

-0.08

0.897

3 sources, 71 species vs. 1 source, 12 species

-0.06

0.927

3 sources, 12 species vs. 1 source, 71 species

3 sources, 12 species vs. 1 source, 12 species

1 source, 71 species vs. 1 source, 12 species

0.37

0.39

0.02

0.573

0.548

<0.001

Echinacea flower-to-fruit conversion rate

I did not find any significant effects of the number of species sown and number of

Echinacea sources sown on Echinacea flower-to-fruit conversion rate (Figure 8, Table

13) None of the four treatment groups differed (Table 14.)

22

Figure 8. Echinacea flower-to-fruit conversion rate
There were no significant effects of the number of species sown and number of
Echinacea sources sown on Echinacea flower-to-fruit conversion rate. Hollow circles
represent observed flower-to-fruit conversion rates, solid circles and lines represent
predicted rates and 95% confidence intervals.

Table 13. Results of an ANOVA testing the effects of seed source number (1 vs. 3),
seeded richness (12 vs. 71) and their interaction on flower-to-fruit conversion rate for
Echinacea purpurea during prairie restoration.

Treatment

Chisq

Df

p

(Intercept)

52.3

Source

Richness

Source X

Richness

1.9

3.3

0.9

1

1

1

1

4.70E-13

0.167

0.07

0.355

23

Table 14. Pairwise comparisons among treatment groups with varying combinations of
seed source number (1 vs. 3) and seeded richness (12 vs. 71) were conducted to test
for differences in flower-to-fruit conversion rate for Echinacea purpurea during prairie
restoration.
Treatment comparison

contrast p

3 sources, 71 species vs. 3 sources, 12 species -0.15

0.323

3 sources, 71 species vs. 1 source, 71 species

-0.17

0.537

3 sources, 71 species vs. 1 source, 12 species

-0.21

0.359

3 sources, 12 species vs. 1 source, 71 species

-0.02

>0.999

3 sources, 12 species vs. 1 source, 12 species

-0.06

0.962

1 source, 71 species vs. 1 source, 12 species

-0.04

0.956

Ratibida flower-to-fruit conversion rate

I found evidence for effects of source number and sown richness, but no

interaction between these factors on Ratibida flower-to-fruit conversion rate (Table 15,

Figure 9). Seed counts were 5% lower when prairie plantings were sown with 71

species and three Ratibida sources, relative to 12 species and three Ratibida sources

and single Ratibida source plantings. There was no significant interaction between both

source number and sown richness (Table 15).

24

Figure 9. Ratibida flower-to-fruit conversion
Rates of flower-to-fruit conversion for Ratibida pinnata was influenced by the number of
species sown and number of Ratibida sources, but the interaction between the two was
not significant. The ratio of flower-to-fruit conversion was lower when prairie plantings
were sown with 71 species and three Ratibida sources, relative to 12 species and three
Ratibida sources and single Ratibida source plantings. Hollow circles represent
observed flower-to-fruit conversion rates, solid circles and lines represent predicted
rates and 95% confidence intervals.

Table 15. Results of an ANOVA testing the effects of seed source number (1 vs. 3),
seeded richness (12 vs. 71) and their interaction on flower-to-fruit conversion rate for
Ratibida pinnata during prairie restoration.

Treatment

Chisq

(Intercept)

2547.1

Df

1

p

<2.2e-16

25

Table 15 (cont’d)

Source

Richness

Source X

Richness

4.5

6.9

2.2

1

1

1

0.0337

0.00830

0.136

Table 16. Pairwise comparisons among treatment groups with varying combinations of
seed source number (1 vs. 3) and seeded richness (12 vs. 71) were conducted to test
for differences in flower-to-fruit conversion rate for Ratibida pinnata during prairie
restoration.
Treatment comparison

contrast

p

3 sources, 71 species vs. 3 sources, 12 species -0.05

0.025

3 sources, 71 species vs. 1 source, 71 species

3 sources, 71 species vs. 1 source, 12 species

-0.05

-0.06

0.06

0.03

3 sources, 12 species vs. 1 source, 71 species

-0.00156

0.952

3 sources, 12 species vs. 1 source, 12 species

-0.00947

0.713

1 source, 71 species vs. 1 source, 12 species

-0.01

0.606

26

DISCUSSION

Early management decisions may have implications on the viability of target plant

populations if they affect reproductive output. I provide evidence that the richness and

number of sources in seed mixes influence plant reproductive output, especially for

Echinacea purpurea and less so for Ratibida pinnata.

Seeding rate has implications for species establishment and reproduction. In our

study design, the Brudvig Lab kept the total seed sown constant throughout all

half-sites. A consequence of this is that low richness half-sites sown with 12 species

had a higher seeding rate per species than high richness half-sites sown with 71

species. Higher seeding rate in half-sites with 12 species led to increased establishment

relative to half-sites with 71 sown species, which in turn resulted in higher

population-scale reproductive output. Similar seeding rate effects have been

documented elsewhere (Applestein et al, 2018; Dickson and Busby, 2009; Pizza et al,

2023). These patterns were especially pronounced at the population scale for both

Echinacea and Ratibida. Half-sites sown with a high density produced larger

populations of focal species; this may have attracted a stronger pollinator response,

which in turn promoted focal plant populations. Indeed, ninety percent of angiosperms

at least benefit from pollinator functioning and services (Ollerton et al., 2011). I see

evidence of greater reproductive output in the Echinacea infructescence-scale, but not

for Ratibida. Additionally, the strong pollinator response is expected to increase

flower-to-fruit conversion ratio; I do not see this pattern in Echinacea, but sowing few

species (and thus a higher species-level rate) led to greater flower-to-fruit conversion

ratio for Ratibida. The stronger pollinator response may be a consequence of pollinators

adapting their foraging behavior in accordance to surrounding resource scarcity (Kunin

and Iwasa, 1996; Armbruster, 2017). These density-dependent effects are congruent

with other studies; low density plant populations displayed significantly reduced

visitation rates and seed production, and pollinators visiting sparser plant populations

had higher energetic flying costs (Kunin, 1993; Cartar, 2009; Redhead et al., 2016).

There is also evidence for a significant interaction between source and richness,

often with lower reproductive output in half-sites that were sown with three sources and

71 species. Incorporation of more genetic diversity by sowing seed from three different

27

provenances may have introduced variability in genetically-determined traits, such as

flowering phenology (Vigo et al., 2013). Phenology is a phenotype of particular

importance because it determines when the reproductive organs of a plant are

accessible to pollinating insects and for how long, which may have direct implications on

reproductive output. Half-sites sown with three sources may have had a higher degree

of phenological mismatch, if individuals from different sources did not open flowers

synchronously. Phenological mismatches develop when flowering phenology does not

overlap with a pollinator’s active season and can have deleterious effects on plant

reproduction (Forrest, 2015). The smaller population sizes of plants in half-sites sown

with 71 species combined with asynchronous flowering phenology may have collectively

lowered reproductive output together. However, this conjecture is not well supported via

our flower-to-fruit conversion analyses. Although there is some evidence that plants in

half-sites sown with 71 species and three sources had lower developed seed count in

Ratibida pinnata, I did not find any differences in Echinacea. Despite our results,

Bolmgren and Cowan (2008) found correlations between flowering phenology and seed

mass: perennial herbs produced larger seeds when they flowered earlier, while annual

plants produced larger seeds when they flowered later. Manipulation of flowering

phenology may have consequences to reproductive output.

Our results suggest some implications for restoration practice. Restoring prairie

with fewer species will produce a few but larger plant populations that reproduce more

effectively. Additionally, incorporating fewer sources may help synchronize flowering

phenology in a species, enhancing pollinator functioning. Together, these larger

populations with synchronized phenology will stimulate a more robust pollinator

response and promote reproduction. These results may be specific to the taxa in this

study, and there may be additional benefits to sowing more diverse mixes at the

community level and multiple sources over evolutionary time. For example, the number

of species sown positively influenced native plant richness and/or diversity (Leps et al.,

2007; Carter and Blair, 2012; Kirmer et al., 2012; Catano et al., 2021; Schmidt et al.,

2022). The positive effects of seeding rate and number of species sown may collectively

help establish a diverse set of plant populations that harbors many individuals. This

suggests that when engaging in seed based restoration, practitioners should restore

28

ecosystems with seed mixes that are more species rich and at a high seeding rate.

Breaking away from the current tradeoff many practitioners need to make, of more

species at the reduction of seeding rate (Barak et al. 2022), would require additional

resources for restoration practice or compromises elsewhere. If more restorations

succeed with this technique, it may be more fruitful to focus restoration efforts within a

smaller area.

29

CONCLUSION

Our results suggest that initial seed mix design is important for the reproductive

output of target plant populations in restoration. Namely, I found that restorations sown

with three sources and 71 species at lower seeding rates decrease reproductive output

for Echinacea across all biological scales, from individual infructescences to the entire

population, but only for Ratibida at the population scale.

In order to increase

predictability in restoration outcomes, practitioners should prioritize high seeding rates

in initial seed mix designs. Establishing larger populations will bolster plant reproductive

output. Ideally, seed mixes should be more diverse, but not at the cost of lowering

seeding rate per species. Our study also suggests that a fewer number of sourcing

locations increases plant phenology matching, which enhances plant reproductive

output. There are predictable lasting impacts of seed mix design on reproductive output,

and our study suggests careful consideration when establishing an initial seed mix for

seed based restoration.

30

BIBLIOGRAPHY

Albrecht, M., Schmid, B., Hautier, Y., & Müller, C. B. (2012). Diverse pollinator

communities enhance plant reproductive success. Proceedings of the Royal Society B:
Biological Sciences, 279(1748), 4845-4852.

Armbruster, W. S. (2017). The specialization continuum in pollination systems:

diversity of concepts and implications for ecology, evolution and conservation.
Functional ecology, 31(1), 88-100.

Applestein, C., Bakker, J. D., Delvin, E. G., & Hamman, S. T. (2018). Evaluating

seeding methods and rates for prairie restoration. Natural Areas Journal, 38(5),
347-355.

Ballare, K. M., Stevens, L., & Jha, S. (2017). Reproductive biology of three

wildflower species: implications of floral compatibility systems and pollen limitation for
prairie restoration. Rhodora, 119(979), 212-223.

Barak, R. S., Ma, Z., Brudvig, L. A., & Havens, K. (2022). Factors influencing

seed mix design for prairie restoration. Restoration Ecology, 30(5), e13581.

Barr, S., Jonas, J. L., & Paschke, M. W. (2017). Optimizing seed mixture diversity

and seeding rates for grassland restoration. Restoration Ecology, 25(3), 396-404.

Bates D, Mächler M, Bolker B, Walker S (2015). “Fitting Linear Mixed-Effects

Models Using lme4.” Journal of Statistical Software, 67(1), 1–48.

Bischoff, A., Steinger, T., & Müller‐Schärer, H. (2010). The importance of plant

provenance and genotypic diversity of seed material used for ecological restoration.
Restoration ecology, 18(3), 338-348.

Bolmgren, K., & D. Cowan, P. (2008). Time–size tradeoffs: A phylogenetic
comparative study of flowering time, plant height and seed mass in a north‐temperate
flora. Oikos, 117(3), 424-429.

Borders, B. D., Cypher, B. L., Ritter, N. P., & Kelly, P. A. (2011). The challenge of

locating seed sources for restoration in the San Joaquin Valley, California. Natural Areas
Journal, 31(2), 190-199.

Breed, M. F., Stead, M. G., Ottewell, K. M., Gardner, M. G., & Lowe, A. J. (2013).
Which provenance and where? Seed sourcing strategies for revegetation in a changing
environment. Conservation Genetics, 14, 1-10.

Broadhurst, L. M., Lowe, A., Coates, D. J., Cunningham, S. A., McDonald, M.,

Vesk, P. A., & Yates, C. (2008). Seed supply for broadscale restoration: maximizing
evolutionary potential. Evolutionary Applications, 1(4), 587-597.

31

Brudvig, L. A., Damschen, E. I., Haddad, N. M., Levey, D. J., & Tewksbury, J. J.

(2015). The influence of habitat fragmentation on multiple plant–animal interactions and
plant reproduction. Ecology, 96(10), 2669-2678.

Bucharova, A., Michalski, S., Hermann, J. M., Heveling, K., Durka, W., Hölzel, N.,

... & Bossdorf, O. (2017). Genetic differentiation and regional adaptation among seed
origins used for grassland restoration: lessons from a multispecies transplant
experiment. Journal of Applied Ecology, 54(1), 127-136.

Bullock, J. M. (2000). Gaps and seedling colonization. Seeds: The ecology of

regeneration in plant communities, 375-395.6

Cardel, Y. J., & Koptur, S. (2010). Effects of florivory on the pollination of flowers:

an experimental field study with a perennial plant. International Journal of Plant
Sciences, 171(3), 283-292.

Cartar, R. V. (2009). Resource-tracking by bumble bees: What explains local
responses to density of bergamot (Monarda fistulosa) flowers?. Ecoscience, 16(4),
470-475.

Carter, D. L., & Blair, J. M. (2012). High richness and dense seeding enhance

grassland restoration establishment but have little effect on drought response.
Ecological Applications, 22(4), 1308-1319.

Catano, C. P., Grman, E., Behrens, E., & Brudvig, L. A. (2021). Species pool size

alters species–area relationships during experimental community assembly.

Choi, Y. D. (2004). Theories for ecological restoration in changing environment:

toward ‘futuristic’restoration. Ecological research, 19, 75-81.

Cresswell, J. E. (1999). The influence of nectar and pollen availability on pollen

transfer by individual flowers of oil‐seed rape (Brassica napus) when pollinated by
bumblebees (Bombus lapidarius). Journal of Ecology, 87(4), 670-677.

Dickson, T. L., & Busby, W. H. (2009). Forb species establishment increases with
decreased grass seeding density and with increased forb seeding density in a Northeast
Kansas, USA, experimental prairie restoration. Restoration Ecology, 17(5), 597-605.

Feldman, S. T. (2006). Pollinator aggregative and functional responses to flower

density: does pollinator response to patches of plants accelerate at low‐densities?.
Oikos, 115(1), 128-140.

Forrest, J. R. (2015). Plant–pollinator interactions and phenological change: what
can we learn about climate impacts from experiments and observations?. Oikos, 124(1),
4-13.

32

Fowler, R. E., Rotheray, E. L., & Goulson, D. (2016). Floral abundance and

resource quality influence pollinator choice. Insect Conservation and Diversity, 9(6),
481-494.

Fox J, Weisberg S (2019). An R Companion to Applied Regression, Third edition.

Sage, Thousand Oaks CA.

Fründ, J., Linsenmair, K. E., & Blüthgen, N. (2010). Pollinator diversity and

specialization in relation to flower diversity. Oikos, 119(10), 1581-1590.

Gilgert, W., & Vaughan, M. (2011). The value of pollinators and pollinator habitat

to rangelands: connections among pollinators, insects, plant communities, fish, and
wildlife. Rangelands, 33(3), 14-19.

Grman, E., Bassett, T. and Brudvig, L.A. (2013), EDITOR'S CHOICE: Confronting

contingency in restoration: management and site history determine outcomes of
assembling prairies, but site characteristics and landscape context have little effect. J
Appl Ecol, 50: 1234-1243.

Harker, K. N., Clayton, G. W., Blackshaw, R. E., O’Donovan, J. T., & Stevenson,

F. C. (2003). Seeding rate, herbicide timing and competitive hybrids contribute to
integrated weed management in canola (Brassica napus). Canadian Journal of Plant
Science, 83(2), 433-440.

Harris, J. A., Hobbs, R. J., Higgs, E., & Aronson, J. (2006). Ecological restoration

and global climate change. Restoration Ecology, 14(2), 170-176.

Havens, K., Vitt, P., Still, S., Kramer, A. T., Fant, J. B., & Schatz, K. (2015). Seed

sourcing for restoration in an era of climate change. Natural Areas Journal, 35(1),
122-133.

Huang, Y., Martin, L. M., Isbell, F. I., & Wilsey, B. J. (2013). Is community
persistence related to diversity? A test with prairie species in a long-term experiment.
Basic and applied ecology, 14(3), 199-207.

Hufford, K. M., & Mazer, S. J. (2003). Plant ecotypes: genetic differentiation in the

age of ecological restoration. Trends in Ecology & Evolution, 18(3), 147-155.

Johnson, S. D., Peter, C. I., & Ågren, J. (2004). The effects of nectar addition on

pollen removal and geitonogamy in the non-rewarding orchid Anacamptis morio.
Proceedings of the Royal Society of London. Series B: Biological Sciences, 271(1541),
803-809.

Joshi, J., Schmid, B., Caldeira, M. C., Dimitrakopoulos, P. G., Good, J., Harris,

R., ... & Lawton, J. H. (2001). Local adaptation enhances performance of common plant
species. Ecology letters, 4(6), 536-544.

33

Kemp, L., Xu, C., Depledge, J., Ebi, K. L., Gibbins, G., Kohler, T. A., ... & Lenton,

T. M. (2022). Climate Endgame: Exploring catastrophic climate change scenarios.
Proceedings of the National Academy of Sciences, 119(34), e2108146119.

Kirmer, A., Baasch, A., & Tischew, S. (2012). Sowing of low and high diversity

seed mixtures in ecological restoration of surface mined‐land. Applied Vegetation
Science, 15(2), 198-207.

Kiss, R., Deák, B., Török, P., Tóthmérész, B., & Valkó, O. (2018). Grassland seed

bank and community resilience in a changing climate. Restoration Ecology, 26,
S141-S150.

Klinkhamer, P. G., & de Jong, T. J. (1990). Effects of plant size, plant density and
sex differential nectar reward on pollinator visitation in the protandrous Echium vulgare
(Boraginaceae). Oikos, 399-405.

Kunin, W. E. (1993). Sex and the single mustard: population density and

pollinator behavior effects on seed‐set. Ecology, 74(7), 2145-2160.

Kunin, W. E. (1997). Population size and density effects in pollination: pollinator

foraging and plant reproductive success in experimental arrays of Brassica kaber.
Journal of Ecology, 225-234.

Kunin, W., & Iwasa, Y. (1996). Pollinator foraging strategies in mixed floral arrays:

density effects and floral constancy. Theoretical Population Biology, 49(2), 232-263.

Kwak, M. M., & Bekker, R. M. (2006). Ecology of plant reproduction: extinction
risks and restoration perspectives of rare plant species. Plant-pollinator interactions:
from specialization to generalization. University of Chicago Press, Chicago, 362-386.

Leimu, R., & Fischer, M. (2008). A meta-analysis of local adaptation in plants.

PloS one, 3(12), e4010.

Lepŝ, J., Doleẑal, J., Bezemer, T. M., Brown, V. K., Hedlund, K., Igual Arroyo, M.,

... & van der Putten, W. H. (2007). Long‐term effectiveness of sowing high and low
diversity seed mixtures to enhance plant community development on ex‐arable fields.
Applied Vegetation Science, 10(1), 97-110.

Lesica, P., & Allendorf, F. W. (1999). Ecological genetics and the restoration of

plant communities: mix or match?. Restoration ecology, 7(1), 42-50.

Li, D. F., Yan, X. C., Lin, Y., Wang, L., & Wang, Q. (2021). Do flowers removed of

either nectar or pollen attract fewer bumblebee pollinators? An experimental test in
Impatiens oxyanthera. AoB Plants, 13(4), plab029.

34

McKay, J. K., Christian, C. E., Harrison, S., & Rice, K. J. (2005). “How local is

local?”—a review of practical and conceptual issues in the genetics of restoration.
Restoration Ecology, 13(3), 432-440.

Méndez‐Vigo, B., Gomaa, N. H., Alonso‐Blanco, C., & Xavier Picó, F. (2013).

Among‐and within‐population variation in flowering time of Iberian Arabidopsis thaliana
estimated in field and glasshouse conditions. New Phytologist, 197(4), 1332-1343.

Menges, E. S. (1995). Factors limiting fecundity and germination in small

populations of Silene regia (Caryophyllaceae), a rare hummingbird-pollinated prairie
forb. American Midland Naturalist, 242-255.

Molano-Flores, B., & Hendrix, S. D. (1999). The effects of population size and

density on the reproductive output of Anemone canadensis L.(Ranunculaceae).
International Journal of Plant Sciences, 160(4), 759-766.

Mortlock, B. W. (2000). Local seed for revegetation: where will all that seed come

from?. Ecological Management & Restoration, 1(2), 93-101.

Mustajärvi, K., Siikamäki, P., Rytkönen, S., & Lammi, A. (2001). Consequences

of plant population size and density for plant-pollinator interactions and plant
performance. Journal of Ecology, 80-87

Ollerton, J., Winfree, R., & Tarrant, S. (2011). How many flowering plants are

pollinated by animals?. Oikos, 120(3), 321-326.

Ooi, Mark KJ. "Seed bank persistence and climate change." Seed Science

Research 22.S1 (2012): S53-S60.\

Pizza, R. B., Foster, J., & Brudvig, L. A. (2023). Where should they come from?

Where should they go? Several measures of seed source locality fail to predict plant
establishment in early prairie restorations. Ecological Solutions and Evidence, 4(2),
e12223.

R Core Team (2021). R: A language and environment for statistical computing. R

Foundation for Statistical Computing, Vienna, Austria.

Redhead, J. W., Dreier, S., Bourke, A. F., Heard, M. S., Jordan, W. C., Sumner,

S., ... & Carvell, C. (2016). Effects of habitat composition and landscape structure on
worker foraging distances of five bumble bee species. Ecological Applications, 26(3),
726-739.

Ruiz‐Jaen, M. C., & Mitchell Aide, T. (2005). Restoration success: how is it being

measured?. Restoration ecology, 13(3), 569-577.

35

Samson, F. B., & Knopf, F. L. (Eds.). (1996). Prairie conservation: preserving

North America's most endangered ecosystem. Island Press.

Schmidt, A., Kirmer, A., Hellwig, N., Kiehl, K., & Tischew, S. (2022). Evaluating
CAP wildflower strips: High‐quality seed mixtures significantly improve plant diversity
and related pollen and nectar resources. Journal of Applied Ecology, 59(3), 860-871.

Stevenson, M. J., Bullock, J. M., & Ward, L. K. (1995). Re‐creating semi‐natural

communities: effect of sowing rate on establishment of calcareous grassland.
Restoration Ecology, 3(4), 279-289.

Suding, K. N. (2011). Toward an era of restoration in ecology: successes,
failures, and opportunities ahead. Annual review of ecology, evolution, and systematics,
42, 465-487.

Waites, A. R., & Ågren, J. O. N. (2004). Pollinator visitation, stigmatic pollen

loads and among‐population variation in seed set in Lythrum salicaria. Journal of
ecology, 92(3), 512-526.

Weeks, A. R., Sgro, C. M., Young, A. G., Frankham, R., Mitchell, N. J., Miller, K.

A., ... & Hoffmann, A. A. (2011). Assessing the benefits and risks of translocations in
changing environments: a genetic perspective. Evolutionary Applications, 4(6), 709-725.

Wei, N., Kaczorowski, R. L., Arceo-Gómez, G., O’Neill, E. M., Hayes, R. A., &

Ashman, T. L. (2021). Pollinators contribute to the maintenance of flowering plant
diversity. Nature, 597(7878), 688-692.

Wenk, E. H., & Falster, D. S. (2015). Quantifying and understanding reproductive

allocation schedules in plants. Ecology and evolution, 5(23), 5521-5538.

Whiteley, A. R., Fitzpatrick, S. W., Funk, W. C., & Tallmon, D. A. (2015). Genetic

rescue to the rescue. Trends in ecology & evolution, 30(1), 42-49.

Wickham H (2016). ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag

New York. ISBN 978-3-319-24277-4

Wilcock, C., & Neiland, R. (2002). Pollination failure in plants: why it happens and

when it matters. Trends in plant science, 7(6), 270-277.

Wilsey, B. (2021). Restoration in the face of changing climate: importance of
persistence, priority effects, and species diversity. Restoration Ecology, 29, e13132.

Woods, E. C., Hastings, A. P., Turley, N. E., Heard, S. B., & Agrawal, A. A.
(2012). Adaptive geographical clines in the growth and defense of a native plant.
Ecological Monographs, 82(2), 149-168.

36

Table A.1. Seed mix information including seeding rate and seed origin of Echinacea
purpurea and Ratibida pinnata seeds used to restore the CLE prairies.

APPENDIX

Colu1

Echinacea

Ratibida

Low richness kg/ha

0.71

Low richness seeds/m2

16.36

High richness kg/ha

0.15

High richness seeds/m2

3.23

0.16

16.36

0.034

3.23

Local nursery

Naturally Native

Naturally Native

Local source origin

Lucas County, OH Northwest Indiana

Non-local northern

Agrecol

Agrecol

nursery

Non-local northern source

Madison, IA

Madison, IA

origin

Southern nursery

Hamilton

Hamilton

Southern non-local source

Putnam County,

Greene and Hickory

origin

MO

Counties, MO

37