Effects of Sowing Time and RElative Prairie Quality on First Year Establishment of 23 Native Prairie Species

Item

Title

Eng Effects of Sowing Time and RElative Prairie Quality on First Year Establishment of 23 Native Prairie Species

Date

2016

Creator

Eng Krock, Sarah

Subject

Eng Environmental Studies

extracted text

EFFECTS OF SOWING TIME AND RELATIVE PRAIRIE QUALITY ON FIRST YEAR
ESTABLISHMENT OF 23 NATIVE PRAIRIE SPECIES

by
Sarah Krock

A Thesis
Submitted in partial fulfillment
of the requirements for the degree
Master of Environmental Studies
The Evergreen State College
June 2016

©2016 by Sarah Krock. All rights reserved.

This Thesis for the Master of Environmental Studies Degree
by
Sarah Krock

has been approved for
The Evergreen State College
by

________________________
Kevin Francis, Ph. D.
Member of the Faculty

________________________
Date

ABSTRACT
Effects of sowing time and relative prairie quality on first year establishment of
23 native prairie species.
Sarah Krock
The restoration of south Puget Sound prairie ecosystems requires a considerable input of
time and resources. This project aims to identify if sowing seeds at different times of year and/or
in different quality sites could increase the first year establishment rates of 23 selected native
species. A full-factorial randomized block design was used to test the success of direct seed
sowing efforts in September, October, December, and March against an unsown control across
three prairies that were designated High, Medium, and Low quality relative to each other. A
Shannon-Wiener diversity index of 16 sown species suggests richness and abundance increase
when sowing occurs in September or October, regardless of prairie quality. Thirteen species were
analyzed independently using generalized regression, while nine were excluded from this
analysis, due to extremely low germination. Four species were significantly influenced by sowing
time: Collinsia spp. (includes C. grandiflora and C. parviflora), Lupinus albicaulis, Lupinus
bicolor, and Plectritis congesta. Four species were significantly influenced by relative prairie
quality: Achillea millefolium, Danthonia californica Eriophyllum lanatum, and Ranunculus
occidentalis. No species responded significantly to both sowing time and relative prairie quality
at α=0.05. Five species did not show a significant response to any treatment: Cerastium arvense,
Festuca roemeri, Koeleria macrantha, Microseris laciniata, and Sericocarpus rigidus. Five
species were excluded from generalized linear model analysis due to low abundances:
Balsamorhiza deltoidea, Clarkia amoena, Lomatium utriculatum, Potentilla gracilis, and
Sisyrinchium idahoense. Four species were excluded from analysis because they were not found
in any treatment or quality combination: Armeria maritima, Erigeron speciosus, Solidago
simplex, and Viola adunca. Overall, first year establishment rates were very low, but these results
suggest that fall sowing times and higher quality prairie sites result in higher establishment rates
for some native prairie species.

Table of Contents
INTRODUCTION ........................................................................................................................... 1
Significance ................................................................................................................................. 1
Practical and Theoretical Application .......................................................................................... 2
Research questions ....................................................................................................................... 3
Hypotheses ................................................................................................................................... 5
Roadmap of Thesis ...................................................................................................................... 6
LITERATURE REVIEW ................................................................................................................ 7
Part 1: Current SPS restoration practices ..................................................................................... 7
Current establishment rates ...................................................................................................... 9
Natural seed dispersal times in SPS ....................................................................................... 10
Importance ............................................................................................................................. 11
Part 2: Practical Implications ..................................................................................................... 11
Sowing time ........................................................................................................................... 11
Relative prairie quality ........................................................................................................... 14
Establishment ......................................................................................................................... 15
Part 3: Theoretical Implications ................................................................................................. 23
Introduction to community assembly theory ......................................................................... 23
Roadmap ................................................................................................................................ 24
Niche theory- Ecological Filtering......................................................................................... 25
Neutral Theory- Priority Effects ............................................................................................ 25
Overlap of Niche and Neutral Theories ................................................................................. 28
Conclusion of Literature Review ........................................................................................... 29
METHODS .................................................................................................................................... 31
Site Description.......................................................................................................................... 31
Experimental Design.................................................................................................................. 37
Species selection ........................................................................................................................ 38
Monitoring Methods .................................................................................................................. 42
Statistical Analysis ..................................................................................................................... 42
RESULTS ...................................................................................................................................... 43
Community Response ................................................................................................................ 46
Individual Species Response...................................................................................................... 47
Species significantly influenced by sowing time ................................................................... 48
Species significantly influenced by site quality ..................................................................... 54
Species not significantly influenced by either sowing time or site quality ............................ 60

iv

Species excluded from analysis due to low abundance ......................................................... 66
Species not encountered ......................................................................................................... 71
DISCUSSION ................................................................................................................................ 71
Community response ................................................................................................................. 72
Individual species response........................................................................................................ 72
Sowing time ............................................................................................................................... 73
Relative Quality ......................................................................................................................... 74
Suggestions for future research .................................................................................................. 76
CONCLUSION .............................................................................................................................. 77
PRACTICAL RECOMMENDATIONS ........................................................................................ 79
Bibliography .................................................................................................................................. 81

v

List of Figures
Figure 1: Map of replicate locations .............................................................................................. 32
Figure 2: Percent cover of native and exotic plants for each study site ......................................... 34
Figure 3: Experimental design ....................................................................................................... 38
Figure 4: Community diversity ...................................................................................................... 47
Figure 5: Collinsia spp. results ...................................................................................................... 49
Figure 6: Lupinus bicolor results ................................................................................................... 50
Figure 7: Plectritis congesta results ............................................................................................... 51
Figure 8: Lupinus albicaulis results 1 ............................................................................................ 53
Figure 9: Lupinus albicaulis results 2 ............................................................................................ 54
Figure 10: Achillea millefolium results .......................................................................................... 55
Figure 11: Danthonia californica results ....................................................................................... 56
Figure 12: Eriophyllum lanatum results ........................................................................................ 58
Figure 13: Ranunculus occidentalis results ................................................................................... 59
Figure 14: Koeleria macrantha results .......................................................................................... 61
Figure 15: Cerastium arvense results ............................................................................................ 62
Figure 16: Festuca idahoense subsp. roemeri results .................................................................... 63
Figure 17: Microseris laciniata results .......................................................................................... 64
Figure 18: Sericocarpus rigidus results ......................................................................................... 65
Figure 19: Balsamorhiza deltoidea results..................................................................................... 66
Figure 20: Clarkia amoena results ................................................................................................. 67
Figure 21: Lomatium utriculatum results ....................................................................................... 68
Figure 22: Potentilla gracilis results.............................................................................................. 69
Figure 23: Sisyrinchium idahoense results..................................................................................... 70

vi

List of Tables

Table 1. Burn history of study sites ............................................................................................... 33
Table 2: Presence of seeded species in control plots ..................................................................... 37
Table 3: Species selected ............................................................................................................... 40
Table 4: Seed mix calculations ...................................................................................................... 42
Table 5: Overall establishment rates .............................................................................................. 45

vii

Acknowledgements
This research would not have been possible without the help of many people. I would
especially like to thank my better half, Tel Vaughn, and friends and family that have been
supportive throughout this adventure. I am grateful for my colleagues from the Center for Natural
Lands Management (CNLM) and Joint Base Lewis-McChord (JBLM) Fish and Wildlife Program
staff, many of whom have been instrumental in one or several stages of this research from initial
development to implementation to analysis. Thanks to Sarah Hamman, CNLM, for the idea,
design, set up, analysis, and interpretation of results; Cara Applestein, CNLM, for help with data
analysis; and Sierra Smith, CNLM, for the seed mix calculations and materials. Thanks, as well,
to JBLM staff: Paula Cracknell, Kharli Rose, Fiona Edwards, and Amber Martens for helping to
identify and count over 13,000 seedlings; Jim Lynch for initial development and set up; Emily
Phillips, John Richardson, and Nick Miller for help raking plots, and other co-workers, interns,
and volunteers for all of their support with this on-going project. Finally, thank you to my
Evergreen MES faculty, especially to my reader, Kevin Francis, for all of his help on various
drafts and for encouraging me to explore the ecological theory concepts.

viii

INTRODUCTION
Restoration projects, such as those occurring in the south Puget Sound (SPS) prairies,
often take years to carry out and a significant investment of time and resources in the process
(Sinclair, Alverson, Dunn, Dunwiddie, & Gray, 2006). Unfortunately, this investment sometimes
goes to waste when the restoration doesn’t go as planned: sometimes invasive species persist, and
sometimes native species fail to grow. Increasing the first year establishment of native plants
could go a long way toward jumpstarting a successful restoration process. The restoration
process increasingly includes seed sowing to reintroduce or augment native plant populations
(Stanley et al., 2011). Unfortunately, seed sowing has often resulted in incredibly low
establishment rates, for often unknown reasons. First year establishment rates of 0-33% have
been observed for a suite of 24 native prairie species using two different sowing methods and
across two different sites (Hamman, Bakker, & Smith, 2015). Establishment rates are even lower
for subsequent years post seeding. The second year establishment rates drop to 0-6% for the same
suite of 24 native prairie species (Hamman et al., 2015). Low establishment rates are particularly
challenging for land managers wishing to establish resilient prairies with limited time and
resources. While some resources exist for growers of native plant materials in greenhouses or
seed production fields (Native Plant Network, 2016) these protocols are often limited in their
applicability to restoration in the field.
Significance
Restoration has endless unknown variables, many of which are out of our immediate
control. Will the weather cooperate? Will next June be too hot? Perhaps tweaking one of the
variables that we can control (the timing of seed sowing) and understanding the influence of
general site conditions (relative site quality) can put the odds of restoration success in our favor.
Improved restoration success, however big or small, has the potential to save time and resources
in the long term. Long-term persistence of the SPS prairie habitat is necessary for the many plants
and animals that depend on them, including four ESA listed species: golden paintbrush (Castilleja
levisecta), Taylor’s checkerspot butterfly (Euphydryas editha taylori), streaked horned lark

1

(Eremophila alpestris strigata), and Mazama pocket gopher (Thomomys mazama). Additionally,
approximately 350 plant species and subspecies are restricted to the prairie-oak ecosystems of the
WPG Ecoregion (Sinclair et al., 2006). The phenomenal biodiversity of the SPS prairies make
this ecosystem a high priority for conservation and restoration efforts.
Glacial outwash prairies were formed about 8,000-12,000 years ago as a warmer, drier
climate caused the Vashon Ice Sheet to retreat northward (Pielou 1991). Prairies all across the
Pacific Northwest were historically maintained in an early seral state by Native Americans
through intentional burning and other management techniques (Walsh, Pearl, Whitlock, Bartlein,
& Worona, 2010; Weiser & Lepofsky, 2009). These types of actions increased production of
edible and fibrous plant materials as well as attracted large game. Burning largely ceased in this
area when European American settlers began occupying the territory in the mid 1850’s (Hamman,
Dunwiddie, Nuckols, & McKinley, 2011). This cultural history of the SPS prairies is key to
understanding the ecosystem’s structure, function, and processes. Today, the SPS prairies face
future threats in four main areas: population growth, encroachment of trees and shrubs, invasive
non-natives, and climate change (Dunwiddie & Bakker, 2011). Managing the tree/shrub
encroachment and invasion of non-native species while simultaneously increasing the diversity
and abundance of native species is key to maintaining and restoring the structure and function of
diverse prairie ecosystems.

Practical and Theoretical Application
Restoration ecology is uniquely situated between practical and theoretical worlds. On the
one hand, research in this field is carried out for very practical reasons, and the results have value
in real-world restoration processes. For example the questions: “when should I sow seeds to get
the highest plant establishment?” and “which species do better in more or less degraded sites?”
are very practically oriented to get the most restoration value from limited resources. On the other
hand, research in restoration ecology can also illuminate theoretical questions. For example, the

2

question: “what mechanisms are driving how communities of plants re-assemble after
disturbance?” is a question that has been plaguing ecologists such as Frederic Clements (1916),
Henry Gleason (1927), and other thinkers for over one hundred years. To what extent are
communities assembled randomly through stochastic processes versus deterministically with
predictable patterns of succession? These debates of relative influence persist to this day as the
neutral versus niche theory of community assemblage.
The practical idea that altering the timing of seed sowing may positively benefit the
species that arrive at a site before their competition is parallel to the term “priority effects” in
theoretical ecological research. Priority effects represent one mechanism of a larger theory that
communities are assembled through stochastic, or neutral, means. Likewise, the practical
assumption that species will respond more positively to higher quality sites is analogous to the
idea of ecological filtering in theoretical ecological research. Ecological filtering, a process
identifying the biotic and abiotic conditions required for establishment, is the basic premise of a
larger theory that communities are assembled through deterministic means, often thought of as
filling available niches.
Today, neutral and niche theories are not as mutually exclusive as they were back in
Clements’ and Gleason’s time. The relative importance of these mechanisms in SPS prairie
restoration has not yet been explicitly investigated, to my knowledge. It is not uncommon for
research in restoration ecology to only consider practical implications. However, research in
restoration ecology may be able to answer both practical and theoretical questions
simultaneously. This experiment aims to take advantage of this overlap in order to both solve
practical problem and provide insight into a larger theoretical question.
Research questions
This thesis aims to answer two different, yet overlapping questions. The first question is a
very practical one, and the second is based in ecological theory. The practical question is: Does
temporal variation of seed sowing affect the first year establishment of 23 native prairie species

3

across a gradient of prairie qualities? The theoretical question is: To what degree is plant
community assembly driven by neutral or niche processes?
The species tested here are found in ecologically and culturally important landscapes in
the south Puget Sound prairies and elsewhere across the Willamette Valley-Puget TroughGeorgia Basin (WPG) Ecoregion, and sometimes in other areas such as the east side of
Washington. This suite of species was chosen to represent a broad spectrum of functional groups,
including perennial grasses, perennial forbs, and annual forbs. Some of these species are common
in SPS prairies, while others are rare and were likely extirpated from many prairie sites. Sowing
seeds at the right times and places has the potential to result in more efficient restoration of SPS
prairies.
Establishment rates are not often quantified, but research from other studies using the
same or similar species in south Puget Sound prairies has shown very low establishment rates.
The establishment of all sown species never reached over 10%, and was often less than 5% of
total bulk seed (Hamman, Smith, & Bakker, 2015). While seed viability varies greatly among
species, generally seeds are healthy and should be capable of becoming established plants. This
research aims to examine the mechanisms behind the discrepancy between the expected and
observed plant establishment in SPS prairies, and offer suggestions for improving establishment
by altering seed sowing time and/or tailoring species selection to appropriate site conditions.
Currently, seed sowing typically occurs during fall and winter. However, most of the
seeds ripens and falls in spring or summer. Altering sowing times to more closely match natural
dispersal times could improve establishment. Seed sowing typically occurs on medium quality
prairie, but all species do not occur equally or evenly across all prairies, thus there is a mismatch
between site condition and species selection. This mismatch between site condition and species
selection is potentially wasting plant materials. Identifying which species, if any, respond
positively to certain site conditions could also improve establishment rates.

4

Restoration strategies in the SPS prairies can be made more efficient by maximizing the
amount of established plants that are produced from a given amount of native seed. Temporal and
spatial variability can constrain restoration options, and make it difficult to foresee how these
ecosystems will respond to management actions (Dunwiddie & Bakker, 2011). Sowing time is
one aspect of the restoration process that is relatively easy to control, unlike the weather or many
other abiotic and biotic limitations to recruitment.
Hypotheses
In response to the practical question (Does temporal variation of seed sowing affect the
first year establishment of 23 native prairie species across a gradient of prairie qualities?), I
hypothesized that sowing time and relative prairie quality would affect the establishment of these
species. The null hypothesis for the practical question was that the mean count of plants in the
seeded plots will not be different than the mean count of plants in the control plots. Of course,
there are many reasons that plants fail to establish from seeds (see establishment section in part
two of the literature review) but the scope of this research only covered two (sowing time and
relative prairie quality) of many possible mechanisms. These two mechanisms address
germination and establishment limitations. It was expected that seed sowing will generally
overcome the dispersal/seed limitation which others have previously documented in SPS prairies
(Stanley et al., 2008, 2011). Thus sowing seeds will likely result in some quantity of established
native plants. The real question is whether some species positively respond to differences in
sowing time or sowing location. I predicted that some combination of sowing times and prairie
qualities will result in higher establishment rates, and could lead to more efficient use of
restoration materials.
In response to the theoretical question (To what degree is plant community assembly
driven by neutral or niche processes?) I hypothesized that both neutral and niche processes are
acting upon these plants, driving community assembly. In order to study community assembly
theory I used seed sowing time as a proxy for priority effects, which is a mechanism of neutral

5

theory; and relative prairie quality as a proxy for ecological filtering, which is a mechanism of
niche theory. Generally, I hypothesize that both neutral and niche processes are at work on these
seeds, therefore I expect that some species will respond positively to both earlier sowing times
and higher quality prairie. The null hypothesis is that none of the species respond to either the
sowing time treatments or the relative site qualities.
If priority effects were the most influential mechanism driving community assemblage,
one would expect species to respond more positively to sowing time, rather than to any of the
relative site qualities. Specifically, I hypothesize that the earlier the arrival time, the better a
species’ chances of survival would be. Based on this neutral theory of community assemblage,
September should have the highest count of plants, followed by October, December, March, and
finally the control. The null hypothesis would be that there is no difference in the mean counts of
plants in any sowing time treatment, thus the count of plants found in any of the treatments would
not be different from one another.
If ecological filtering is the most influential mechanism driving community assemblage,
one would expect species to respond more positively to the relative prairie quality, rather than to
any of the sowing times. The high, medium, and low relative prairie qualities are proxies for
biotic and abiotic site conditions. Specifically, I hypothesize that the higher quality the site, the
higher the native species establishment will be. Based on this niche theory of community
assemblage, the high quality site should have the highest count of plants, followed by the medium
quality site, and finally the low quality site. The null hypothesis would be that there is no
difference in the mean counts of plants in any relative prairie quality, thus the count of plants
found in any of site conditions would not be different from another.
Roadmap of Thesis
This thesis investigates whether temporal variation of seed sowing time and relative site
quality affects the establishment of 23 native prairie species in the South Puget Sound. The
literature review is broken into three parts: 1) current restoration practices; 2) practical

6

implications of sowing time, site conditions, and challenges to establishment; 3) theoretical
arguments of neutral and niche community assembly. The second half of this thesis describes the
original research undertaken to test the hypotheses. Methods, results, discussion and conclusion
sections describe this seed addition experiment. Finally, this thesis concludes with a few practical
recommendations for land managers.

LITERATURE REVIEW
This literature review is split into three sections. The first section focuses on current
restoration practices and the resulting seedling establishment rates. This section also addresses
natural seed dispersal times and the importance of this research.
The second section focuses on practical implications. This section examines practical,
restoration-focused studies involving sowing time and relative site quality. Sowing time and
relative prairie quality are the two independent variables of this study. Lastly, a thorough
examination of the dependent variable, establishment, wraps up this section.
The third section focuses on theoretical implications. This section begins with a
description of the current status of the niche versus neutral debate of community assembly. Next,
the mechanism of each theory is examined. Niche theory is best supported by ecological filters
created by biotic and abiotic conditions. In order to study this mechanism, relative prairie quality
is used as a proxy. Neutral theory is best supported by the history of arrival, or priority effects. In
order to study this mechanism, sowing time is used as a proxy. The relative contribution of each
of these theories is considered, as well as some concluding thoughts about the overlap of practical
and theoretical work, and applicability to prairie restoration in the SPS.
Part 1: Current SPS restoration practices
Seed addition is one of several techniques currently being used to restore SPS prairies
that are fragmented and degraded. Remnant prairies have experienced significant losses of native
plant taxa, and increasing overall species diversity is among the challenges facing prairie
restoration throughout the ecoregion (Dunwiddie and Bakker, 2011). Introduction of target

7

species is necessary for many prairie restorations because the seeds no longer persist in the soil
seed bank (Stanley et al. 2011). Biodiversity is increasingly recognized as a major determinant of
ecosystem functionality (Hooper et al., 2005; Tilman, Isbell, & Cowles, 2014; Zavaleta, Pasari,
Hulvey, & Tilman, 2010). Restoring SPS to species rich ecosystems will require reintroduction
and augmentation of native plant populations.
Seed addition is not the only technique used in restoration. Planting plugs, small plants
grown in cone-shaped containers in the greenhouse, can also be a useful technique for prairie
restoration. Plugs have shown to be useful in certain situations (Agee, 1996; Ewing, 2002,
Sinclair et al., 2006) but there are advantages to using direct seed addition methods rather than
planting plugs. The cost of seed and plug production varies by species and by year, but seeds are
generally much cheaper. For example, Castilleja levisecta plugs produced by CNLM’s nurseries
cost about $3.00 each, while C. levisecta seeds cost approximately $0.30 per 1,000 (Dunwiddie &
Martin, 2016). Additionally, seed sowing is much more cost effective than plug planting,
however establishment rates in the field are generally <1% for this species (Dunwiddie & Martin,
2016). The decision to use plugs or seeds in restoration is driven by many factors including cost
and availability of materials.
In a review of species introductions for grassland restoration Hedberg and Kotowski
(2010) found that direct seeding or hay spreading, rather than plug planting or other methods, is
preferable to achieve desired results; though they do note that the method is dependent on the
species, size of target area, and funding availability. Both methods are currently employed in SPS
prairie restoration, and both are useful techniques. I chose to focus on direct seeding rather than
plug production because fine-tuning seed addition techniques has the potential to restore a larger
area in a shorter amount of time.
Extensive amounts of native seed are being produced and applied in SPS prairies in order
to restore biodiversity and overcome dispersal limitations. Currently, native seed is produced on a
large scale at one of several farming facilities designed specifically for that purpose. In 2014 the

8

Center for Natural Lands Management (CNLM) in partnership with the Sustainability in Prisons
Project (SPP) produced 2,109 pounds of seed and 344,000 plugs of 104 prairie species (Smith &
Elliot, 2015). Now that dispersal limitation is less of an issue, the focus transitions to the
establishment limitations of the native prairie species themselves.
Even as native seed for restoration becomes more available, the timing of seed sowing
remains dependent on several factors. After seed is grown, it is collected, cleaned to remove
debris, and distributed to partners to be sown into the prairies they manage. Seed sowing is often
done with drop seeding implement pulled behind a tractor or utility task vehicle (UTV) to cover
many acres of prairie efficiently. For smaller areas, or smaller amounts of seed, hand sowing is
usually sufficient. The time of year that seeds are typically sown is in the fall (October or
November), due to logistical constraints of harvesting, cleaning, and dispersal, in addition to site
preparation and personnel availability. A survey of 38 prairie restorationists across 11 U.S. states
showed that most prefer to sow in dormant season (fall or winter) to take advantage of a natural
cold stratification, as well as freeze-thaw cycles to bury the seeds, and spring moisture (Rowe,
2010). The timing of seed sowing for restoration purposes depends heavily upon logistical
considerations, while underlying assumptions are that mimicking the natural seed dispersal times
is beneficial.
Current establishment rates
In order to maintain a stable population, every adult plant must be replaced by an average
of one successfully established offspring. Therefore, one would expect short-lived plants to have
higher seedling recruitment than longer lived plants (Eriksson & Ehrlén, 2008). This also
suggests that perennial plants do not necessarily have to produce new recruits every year in order
to maintain some baseline population level. Results from a seed addition field experiment (seed
was sown in mid-November) in the SPS, including eight of the species tested here, show first year
establishment rates ranging 0-33% across two prairie restoration sites (Hamman et al. 2015). In a
seed addition field experiment (seed was sown in late September) in a western Oregon prairie,

9

native prairie seeds established at a rate of <1-66% depending on the species, litter removal
treatment, and study location (Maret & Wilson, 2005). This western Oregon study included three
of the same species used in this research: Festuca idahoensis subsp. roemeri establishment rates
ranged from 3-32%, Eriophyllum lanatum establishment rates ranged from <1-23%, and
Potentilla gracilis establishment rates ranged from 5-15% (Maret & Wilson, 2005). These low
and variable establishment rates indicate that seeds must overcome incredible challenges to
become an established plant in the field. When calculating seed mix proportions, if the field
establishment of native species is not known from previous seed addition experiments,
greenhouse trials, or found in the literature, a standard survival rate of 5% of pure live seed is
used (Sierra Smith, CNLM, personal communications, September 2015). Findings from coastal
California prairies of 1-2% first year establishment (Holl et al., 2014) suggest that low
establishment rates are common in other grassland ecosystems as well. Learning more about
native species’ natural and artificial establishment rates could provide insights about how to make
restoration practices more efficient.
Natural seed dispersal times in SPS
Native species found in SPS prairies typically disperse their seeds in the spring and
summer (May-August), though some species may hold a portion of their seeds in seed pods until
fall (September-October). After dispersal, seeds experience a natural cold-moist stratification,
which is a period of time in cold temperatures that often breaks certain types of dormancy in
seeds (Drake, Ewing, & Dunn, 1998; Russell, 2011; Krock et al., 2016). Additionally, cold
stratification may involve periods of time when the ground freezes and thaws again, sometimes
resulting in frost heave, which could create appropriate microsites for seed germination. In the
SPS annual species typically germinate in the fall, while perennial species typically germinate in
the spring. Some perennial plants do not reproduce for several years after germinating. Annual
species must establish as adult plants in the first year to produce the following year’s generation,

10

while it may take more than one growing season for a perennial plant to become a reproductive
adult.
Importance
Restoration and conservation of south Puget Sound (SPS) prairies is necessary to ensure
the survival of four Endangered Species Act (ESA) listed threatened and endangered species
(U.S. Fish and Wildlife, 2015). Additionally, SPS prairies, like other natural areas, provide
various ecosystem services such as water filtration, climate regulation, and cultural, spiritual, and
recreational services (World Health Organization, 2005). Temperate grasslands and savannahs are
one of two biomes at greatest risk due to habitat loss and lack of protection (Hoekstra, Boucher,
Ricketts, & Roberts, 2005). Effective and efficient restoration, and in some places re-creation of
prairie ecosystems, requires an adaptive management approach (Delvin, 2013; Sinclair et al.,
2006). Prairies across the Willamette Valley-Puget Trough-Georgia Basin (WPG) ecoregion have
been shown to be limited by availability of native seed, and even relatively high quality remnant
prairies are strongly seed-limited (Stanley, Dunwiddie, & Kaye, 2011). Seed limitation, also
known as dispersal limitation, is a key variable preventing the successful restoration of many
prairies. In fact, many species are dispersal limited due to habitat degradation and fragmentation
(Hedberg & Kotowski, 2010). Establishing native flora is also a challenge in the tallgrass prairies
of the Midwest (Carter & Blair, 2012), coastal prairies of California (Holl et al., 2014), and
elsewhere, thus it is necessary to assess the efficiency of seed sowing in these ecosystems as well.
Part 2: Practical Implications
Sowing time
Currently, in the SPS prairies, timing of seed sowing is typically driven by logistical
considerations, such as seed availability and personnel time, rather than an attempt to mimic
species biology. Many of these native prairie species naturally disperse their seeds in the spring
and summer, leading to a potential mismatch of natural dispersal and sowing times. The status
quo (fall-winter sowing) has resulted in extremely low establishment rates, however, adjusting the

11

timing of seed sowing to match the biology of the species could result in higher establishment
rates, and more efficient prairie restoration.
Sowing time and establishment have not been well documented in all ecosystems, but
there is emerging evidence that sowing time does have an effect on many prairie species.
Anecdotal evidence from native seed farms shows that earlier fall sowing of certain species,
particularly annual forbs, results in higher establishment and overall vigor of the plants (Kathryn
Donovan, CNLM, personal communications, May 2016). It is difficult to completely separate
sowing time (i.e. seed arrival time) from seasonality, especially in the field, so no attempt was
made to do so in this study.
Studies showing a significant effect of sowing time on establishment rates
Several studies support the idea that sowing time has some amount of influence on
establishment rates of plants. Frischie and Rowe (2012) found that of seven tested tallgrass prairie
species that naturally dispersed their seeds in early season, five of them had higher establishment
rates if the sowing was done in the summer (growing season) rather than the subsequent winter
(dormant season). They attributed these findings to abiotic conditions and arrival time- plants
were simply germinating and growing earlier because they were seeded earlier. This agrees with
my expectation of SPS prairie restoration, and hypotheses for this experiment.
Another study found that a June seeding time resulted in a higher total plant density and
lower total biomass than the December seeding treatment for three native tallgrass prairie species,
though seeding time did not affect all three species in the same way (Sullivan and Howe, 2009).
In this study the effects of seeding time and seeding density could be seen seven years after
seeding, with larger, more productive Echinacea purpurea plants, and more abundant, larger
Desmanthus illinoensis plants in the June seeding treatment than in the December seeding
treatment (Sullivan & Howe, 2009). The authors suggest that seeding time is important not only
for the plant size in subsequent years, but also vole herbivory patterns.

12

Sowing time may not have a significant effect on establishment when there is generally
low establishment of sown species, or a complete dominance of the site by one sown species
(Lawson, Ford, & Mitchley, 2004). However at one site that was not dominated by the native
grass Holcus lanatus, Lawson et al. (2004) found earlier (spring) sowing resulted in higher
species richness and abundance of sown species than the following (fall) sowing time. The
authors of this study inferred that competition was the main mechanisms driving these results.
In one study aimed at establishing North American prairie species in an urban park in
northern England, Hitchmough, de la Fleur, and Findlay (2004) found the emergence across all
species tested was significantly higher for winter rather than summer sown seed, though
individual species exhibited variable responses. These results are counter to this study’s
hypothesis. One potential reason for these results are the confounding factors of seed stratification
between the summer and winter sowing treatments, and increased moisture stress for the summer
sown seeds.
Studies showing multiple factors influencing establishment
Another study has found that sowing time is one factor in combination with other factors
that influences plant establishment rates. Establishment of native species is not only affected by
the timing of seed sowing, but also the seeding method used. Larson et al. (2011) found that seed
drill application was preferable during the growing season, while broadcast application was
preferable in the dormant season; and the timing had differential impacts on establishment of
functional groups (forbs vs. warm season grasses). They attributed the success of the dormant
season broadcast seeding over the subsequent growing season broadcast seeding to arrival time,
among other biotic and abiotic factors. These findings are counter to the hypothesis in this study,
but this study only uses one sowing method- broadcast application.
Studies showing sowing time as a non-significant factor on establishment
Other studies have shown a marginal or non-significant effect of sowing time on
establishment rates. For example, earlier sowing time (March, beginning of fall in southern

13

hemisphere) had a marginally higher maximum percentage emergence than later sowing time
(May, end of fall in southern hemisphere), on 21 species sown in a South African grassland field
experiment (Sayuti and Hitchmough, 2013). Most of the 21 species tested had higher emergence
rates when sown in March rather than May, but statistically significant results were not reported
for these findings in the study.
In another study, Doll, Haubensak, Bouressa, and Jackson (2011) found that seeding
time, November or subsequent May, was not a significant factor in native grass recruitment in a
Wisconsin prairie, however disturbance and nitrogen availability were significant factors. These
two studies highlight the need for further investigation of the effects of sowing time on field
establishment rates in a wide range of ecosystems.
Conclusion
Several studies outlined here have shown significant effects of sowing time on
establishment, but the lack of standardization of time periods, species/functional groups, and life
histories makes it difficult to compare studies to each other. To my knowledge, the effects of
sowing time on establishment of native plants has not been tested on SPS prairie species. Based
on other studies in prairie ecosystems, it is likely that temporal variation of seed sowing in SPS
prairies will show a significant effect on establishment rates of at least some of the species tested.
Relative prairie quality
The practical and theoretical implications of site quality is well recognized in restoration
ecology. However, within SPS the application of a highly diverse seed mix has not been
experimentally tested across a gradient of relative prairie qualities, all occurring on the same soil
type and with a relatively similar site history, in order to determine the influence of site
conditions on species selection. Knowing which species, if any, respond to different site
conditions could make prairie restoration more effective and efficient. Unfortunately, high and
low quality sites in the SPS have often been lower priority for restoration due to limited
resources. Meanwhile, the medium quality prairie receives the bulk of restoration efforts in order

14

to make it more similar to high quality reference sites. Some native species may establish and
persist more readily in high quality prairie rather than low quality, or vice versa, though these
species have only been anecdotally identified. Identifying which species perform well under
certain site conditions may help tailor species selection and increase efficiency of the restoration
process.
Establishment
In order for plants to become successfully established1 they must overcome many
challenges. Every life stage of a plant is met with challenges that are in some way influenced by
environmental conditions (Nathan & Muller-Landau, 2000). Environmental conditions provide
the basis for relative site quality in this study. In this study, relative site quality has been
categorized into three groups (high, medium, and low), reflecting the native species richness at
each site (see methods section). However, actual environmental conditions of these three sites are
along more of a spectrum, and much less clear-cut. The complex environmental conditions at
each site may help or hinder the establishment of native species. Species establishment may be
positively or negatively correlated with site quality. Some species are more apt at colonizing or
persisting in degraded habitats than other species. There are a suite of reasons that plants are
unable to establish in a given site with a unique set of environmental conditions and limitations,
which I will examine below.
The literature broaches the topic of establishment limitations in a variety of ways. One
study states that “[r]ecruitment limitation may result from any or all of source limitation (recruits

1

Establishment can be defined as “the process that results in a population that persists for many
generations” (Liebhold, 2000). The term establishment is typically used for longer term >1 year survival
studies, though it can be used as a surrogate term to describe abundance, density, emergence, and other
measurable terms that indicate living plants. In contrast, “recruitment refers to the process by which new
individuals found a population or are added to an existing population” (Eriksson & Ehrlén, 2008).
Recruitment is thus a pre-requisite for establishment; the seedling phase before the plant becomes an
established reproductive adult. Eriksson and Ehrlén (2008) also note that “population studies often define
seedlings somewhat arbitrarily, implying also that population processes such as seedling recruitment may
be assessed arbitrarily.” The terminology is a bit confusing, especially concerning annual species,
regardless, recruitment/establishment here refers to the presence of living plants.

15

fail to arrive at a site due to reduced fecundity of adults or reduced dispersal of propagules),
germination limitation (site conditions prevent or reduce seed germination or increase seed
mortality), and establishment limitation (seedlings fail to mature)” (Henry, Stevens, Bunker,
Schnitzer, & Carson, 2004). Many other studies list a few key challenges that the author(s)
assume are responsible for the loss of establishment. For example, Frischie and Rowe (2012) list
low quality seed, predation, inappropriate storage, and unknown germination requirements as
potential reasons for a reduction in germination and establishment. Other sources, such as (Martin
& Wilsey, 2006, and citations therein) often choose to focus solely on two elements of
recruitment limitation: seed limitation and microsite limitation. These two ideas have been well
explored (Clark, Poulsen, Levey, & Osenberg, 2007; Turnbull et al., 2000; Zobel et al., 2000).
These two elements are important, but they encompass many variables, and completely disregard
the establishment limitation (seedlings do not survive). It is useful to consider challenges that are
unique to each life stage in order gain a more holistic overview of the causes of establishment or
mortality, rather than just two possible explanations (seed limitation or microsite limitation).
While there are many types of challenges or limits to establishment, I will examine three
in detail. These three types of limitations are source, germination, and establishment. Below, I
outline each limitation, its significance, and how it may influence my study of SPS prairie plant
establishment
A. Source limitation
Source limitation, or the lack of healthy seeds in a given area, is the first key recruitment
limitation. Source limitation can be broken into two parts: fecundity, or reproductive rate, of
existing adult plants, and a reduction in dispersal of seeds.
Fecundity
The fecundity of native seeds is important because without healthy, viable seeds
populations cannot sustain themselves. Fecundity could be reduced by a lack of sufficient
pollination. At least one species the SPS prairies is known to be pollinator limited, Balsamorhiza

16

deltoidea (Fazzino, Kirkpatrick, & Fimbel, 2011), although there may be others as well. In
restoration seeds, however, the purity and viability of each species is typically known, and most
seeds are of high quality. As such, fecundity is not likely to be a key limiting factor for this
experiment.
Dispersal/seed limitations
Dispersal limitation, also known as seed limitation, has been a known limiting factor in
SPS prairie restoration (Stanley et al., 2011). However, natural levels of seed rain are not known
and probably highly variable due to patchy distributions of species. In seed addition experiments
such as this one, seed limitations are overcome by adding seed to a given amount of space, rather
than letting seed naturally disperse to that space. For this experiment we were both re-introducing
species that were likely found in these areas historically, as well as augmenting existing
populations. Some seed naturally dispersed into the study area in the same year as the seeding,
but the control plots can help differentiate natural levels from our experimentally augmented
ones. Higher seeding rates resulted in higher plant densities in another seed addition experiment
in the SPS (Hamman et al., 2015). Generally speaking, the more seeds that are added to an area,
the higher the plant establishment rate will be.
Dispersal limitation is an important and well-documented phenomenon. Tallgrass prairies
in Iowa were found to be severely seed limited, and even the enhanced availability of microsites
through ungulate grazing did not increase seedling emergence, likely because there was no seed
of rare species available to germinate (Martin & Wilsey, 2006). Additionally, seed limitation,
rather than lack of microsite availability, influenced species richness in a calcareous grassland
(Zobel et al., 2000). Turnbull, Crawley, & Rees (2000) also found that seed limitation decreased
from sand dune to woodland, to grassland ecosystems, which they correlated with an increase in
bare ground, and thus competition-free microsites. In this review, Turnbull, Crawley, & Rees
(2000) found that new ploughed and early/mid successional prairies had roughly equivalent
proportion of seed limitation. These habitats were more seed limited than intact arid and intact

17

mesic grasslands. The SPS prairies that I used for this study could be classified as early/mid
successional rather than intact. However, even intact, high diversity prairies have been found to
be seed limited, and in one case, the effect of seed addition increased species richness for eight
growing seasons (Foster and Tilman, 2003). In reality, these theoretical arguments agree with
Stanley et al. (2011) that even relatively high quality areas across the WPG ecoregion, including
sites in the SPS, are limited by native seed availability.
Seed limitation is a principle component of community assemblage. Increasing native
seed density increases the number of established plants, though not always in a linear fashion.
There can be a decrease in establishment at extremely high seeding densities due to density
dependent mortality (Burton, Burton, Hebda, & Turner, 2006). It is important to find the balance
between overcoming seed limitation and wasting seeds by over sowing. Higher densities of seed
may increase establishment of seeded species, though there may be a point at which additional
seed does not have additional restoration value (Goldblum, Glaves, Rigg, & Kleiman, 2013).
Native seeds are expensive, and in low supply, so it is important to maximize efficiency of seed
use in restoration projects. In fact, “sites restored to native prairie and wetlands are frequently
seed limited for several reasons: 1) the minimal extent of remnant habitat and vegetation
remaining; 2) the substantial effort needed to harvest a diverse and large quantity of local
genotype seed; and 3) degraded soil seed banks. In combination, these factors make it difficult to
restore a diverse native plant community” (Goldblum et al., 2013). Community assemblage can
also be altered through overcoming seed limitation. Higher seeding densities resulted in a rapid
decrease in weeds, though even low seeding densities had similar results after two years
(Stevenson, Bullock, & Ward, 1995). However, other findings suggest that while native seed
addition may reduce the proportion of exotic species, the community dominance was not shifted
from exotic to native (Martin & Wilsey, 2014). While this project’s findings may support the
findings of Stanley et al., (2011) and others, the theory of seed limitation is tangential to the
questions that I am attempting to answer with this study.

18

B. Germination limitation
Germination limitation, or failure of seed to become a seedling, is the second key
recruitment limitation. Clark & Wilson (2003) state that seeds can either germinate to become
seedlings, remain as seeds (enter dormancy), or die; thus germination limitation can be broken
into three parts: inappropriate site conditions for germination, dormancy, and seed death. The first
part, inappropriate site conditions, is best examined in the context of microsite availability.
Microsite
The microsite, or the small area where a seed lands, is a key factor involved in plant
establishment. For example, a seed that lands upon rocks would have an inappropriate abiotic
environment, and a seed that lands in an area occupied by other plants will likely have an
inappropriate biotic environment. A recent study by Dunwiddie and Martin (2016) highlights the
need for appropriate microsite placement of Castilleja levisecta plugs. Castilleja levisecta is a
hemiparasitic plant endemic to the WPG ecoregion. Higher survival and flowering was correlated
with higher native forb richness in the area around the plugs. Additionally, topography influenced
survival of plugs, with those planted on mounds or in swales having an increased survival rate
five years after planting (Dunwiddie & Martin, 2016). These findings highlight the need for
suitable microsites not only for seed sowing, but plug planting as well.
Native seed addition following prescribed fire has been found to be more successful than
additions after other site pre-treatments (Maret & Wilson, 2000, 2005; Stanley et al., 2011),
presumably due to the increased availability of appropriate microsites. Simply burning an area for
restoration or sowing seeds alone would be counterproductive in an area without sufficient
microsites or native seed bank (Maret & Wilson, 2005). Currently, prescribed fire is the
restoration practice most preferred in the SPS, often in combination with other site preparation
techniques.
In the SPS prairies buildup of moss, lichen, and thatch creates unfavorable microsites in
the absence of burning (Hamman et al., 2011). Resumption of periodic burning may be especially
beneficial for annual forbs that rely on these microsites between perennial bunch grasses and

19

forbs (Dunwiddie et al., 2014). In other ecosystems, such as the tallgrass prairie of central Texas,
the rates of germination and establishment depend on microsites that protect seeds from
desiccation, in the form of either litter or rocks, compared to protections-free bare ground
(Fowler, 1986). Other sources cite disturbances such as tilling, grazing, digging by fossorial
animals, or other actions which create bare ground as the mechanisms responsible for the creation
of appropriate microsites (Dunwiddie et al., 2014; Pfeifer-Meister et al., 2012; Bakker et al.,
2003). Note that disturbances assist the establishment of species as long as they coincide with
favorable abiotic conditions (Knappová, Knapp, & Münzbergová, 2013). Appropriate microsites
are one important factor that may be limiting the germination of native species in the SPS
prairies.
Dormancy
The second germination limitation, seed dormancy, is the only non-permanent stage of
the three possible fates for a seed: germination, dormancy, or death. Most seeds have the ability
to persist in this life stage until conditions become more favorable for germination, and thus
growth and development. The lifespan of seeds varies greatly, even under favorable storage
conditions such as low relative humidity and temperature (Justice and Bass, 1978). There are
several different types of dormancy, and most species have unique requirements for breaking
their dormancy (see Baskin and Baskin, 1998). Cold moist stratification has been shown to
overcome dormancy in several WPG eco-regional species (Drake et al., 1998; Russell, 2011;
Krock et al. 2016) and physical scarification such as heat shocking may or may not be necessary
for some species (Elliott, Fischer, & LeRoy, 2011). Dormancy is not likely one of the main
limiting factors in this experiment, or in general prairie restoration.
Seed death
The third and final type of germination limitation is increased seed death. Seed death is
best examined at the scale of the individual seed which is undoubtedly difficult and time

20

consuming. It makes sense then, that few scientists have tackled this problem. However, seed
death by predation is widely acknowledged, and is investigated here in more detail.
Although a mysterious part of their life history, the seed stage of a plant’s lifecycle can
only have a few general outcomes. Clark & Wilson (2003) found that for four western
Washington prairie species, that seed death was the most common fate. I would assume that this
phenomenon is not limited to merely four SPS species. This is likely at least one of the reasons
that establishment rates have been so low.
Seed death is often driven by predation of some sort. Seed predation is difficult to study
because predation can happen in several ways. For example, seeds can be eaten by an herbivore
while still on the plant, by a bird or rodent once they have dispersed, or by pathogen such as fungi
or bacteria at any stage of development. In some studies predation has not shown to influence
seedling density (Henry et al., 2004), however models have shown that herbivores may have a
larger impact on plant population dynamics than previously expected (Maron & Gardner, 2000).
Seed herbivory by rodents has even shown impacts on community succession in prairies of
southwestern Wisconsin (Howe & Brown, 2000). Predators may selectively consume seeds as
well. Only one of four species was substantially preyed upon (21% reduction) in a seed-fate study
of western Oregon (Clark and Wilson, 2003). Anecdotal evidence from CNLM’s seed production
facilities suggests that mice preferentially prey upon Viola adunca seed, which results in low
yields for this species (Angela Winter, CNLM, personal communications, May 2016).
Unfortunately, little is known about the effect of predation on SPS prairie species in the field.
C. Establishment limitation
Establishment limitation, or seedling death, is the third and final key recruitment
limitation. The three biggest challenges to seedling survival are drought, herbivory, and
pathogens (Moles and Westoby, 2004). Another challenge to seedling survival is competition
with other seedlings, but is generally less problematic than the first three challenges (Moles and

21

Westoby 2004). These four issues affecting seedling survival can be divided into biotic and
abiotic factors.
Biotic
Biotic factors include herbivory and pathogens (see discussion above of seed predation),
as well as competition with other seedlings and adult plants. While some would assume adult
plants would act as nurse plant to small seedlings, this is not necessarily true. In a greenhouse
experiment of three grasses in Central Texas, adult plants negatively impacted grass seedlings at
least six cm away, and most of all natural microsites were found within six cm of an existing
plant (Fowler, 1986). Litter from adult plants can either act in a positive way (retaining soil
moisture) or a negative way (decreasing sunlight availability, and/or acting as physical barrier).
Additionally, a large amount of litter has shown a negative impact on seedling survivorship of the
three grass species of central Texas (Fowler, 1986). In western Oregon prairies, litter has been
shown to inhibit seedling establishment (Maret & Wilson, 2005). Litter likely has the same
negative effect in our SPS prairies, but this is not an issue addressed by this research because
prescribed burning removed most above ground biomass before the experiment began.
Abiotic
Abiotic factors such as temperature, water and nutrient availability are very important for
the survivorship of seedlings. Of these, it appears that nutrient availability is not as dynamic as
the other two factors. It has been shown that most properties of prairie soils are static, or are
difficult to measure over short time frames, such as less than five years (Brye, Norman, & Gower,
2002). Grassland restorations are complex, and it is important to consider treatment effects over
time. For example, soil amendments to alter the nutrient availability have been shown to have
different effects on plants during different years (Doll, Brink, Cates, & Jackson, 2009). There is a
negative relationship between nitrogen availability and invasibility, the ability of non-native
plants to invade an area (Tilman, 1997) which is counter-intuitive. Other studies have found the
nitrogen fixing shrub, Cytisus scoparius, which invades SPS prairies elevates nitrogen levels in

22

the soil and may facilitate the invasion by other non-native species (Kirkpatrick and Lubetkin,
2011). Today, remnant prairie is found most often on sites with the poorest quality soils, thus the
least appealing for agriculture (Dunwiddie & Bakker, 2011). Low nutrient prairie soils may be
one of the key abiotic factors influencing plant establishment in SPS prairies.
Annual variation can also affect the plant community, due to dynamic changes in
temperature, moisture, and competition. The timing of precipitation and temperature fluctuation
can influence which species may be successful in different years due to variability in germination
time (Fowler, 1986). Unsurprisingly, lack of water may be one of the biggest enemies of
seedlings. Establishment of native and non-native species increases with amount of precipitation
they receive (Bakker et al., 2003). While abiotic factors are largely out of the control of
restoration ecologists, these factors can be used in a predictive manner. Establishment may be
increased if the timing of seeding is based on abiotic factors such as temperature and precipitation
(Bakker et al., 2003; Westoby, Walker, & Noy-Meir, 1989). Maintaining flexibility in the seeding
time could overcome unfavorable abiotic factors to increase establishment of native species. This
also highlights the need to experimentally test seed sowing between years to understand the
temporal variation as well as the spatial variation (Vaughn & Young, 2010)
Part 3: Theoretical Implications
Introduction to community assembly theory
It is common knowledge that plants can, and will, colonize an area after most types of
disturbances. New communities of plants are assembled, sometimes quickly or sometimes more
slowly, and often change over time. This phenomena is easily observed in cracks in the sidewalk,
abandoned agricultural fields, or after a wild fire. Are the new colonizing plants products of
random chance, or are they filling a specific niche created environmental conditions? Community
assembly provides a framework for understanding which species, and how many, can coexist in a
given area (Chase, 2003). Whether community assembly is more predetermined by biotic and
abiotic filters or more stochastically developed is yet unknown and remains a controversial topic

23

(Fukami, Martijn Bezemer, Mortimer, & Putten, 2005; Weiher et al., 2011). One can trace the
roots of the argument back to the beginning of the field of ecology with Clements’ (1916)
analysis of deterministic climax communities versus Gleason’s (1927) view of random species
assemblages influenced by historical factors. Today the argument continues in terms of niche
versus neutral assembly theory. Niche theory focuses on the biotic and abiotic filters that create a
local community out of a subset of a regional species pool, while neutral theory focuses on how
stochastic forces such as historical inertia, dispersal, and ecological drift influence community
assembly (Weiher et al., 2011). Today the core arguments for each theory have been thoroughly
investigated and slightly rebranded (see Chase and Leibold (2003) for niche theory, and Hubbell
(2001) for neutral theory). Although the vocabulary has shifted, the questions remain unresolved.
Recent work has shown that in coral reef community dynamics are non-neutral, overturning a
decade of assumptions that stochastic processes drive biodiversity (Connolly et al., 2014). The
ongoing debate is not likely to be resolved in the near future. It is likely that both processes are
acting simultaneously, and thus the degree to which they are occurring may be more useful to
investigate (Weiher et al., 2011). Understanding these drivers of community composition may
have practical implications in addition to theoretical ones.
Roadmap
First I will investigate niche theory, especially how this widely accepted concept of biotic
and abiotic environmental conditions is prevalent in the literature, and yet often overlooked in
restoration. Then I will describe the counter-argument of neutral theory in more detail and
pinpoint the specific mechanism, priority effects, which is most applicable to this study. Priority
effects have been well studied in laboratory and natural settings, including grasslands. These
findings have potentially useful application in the restoration of SPS prairies. Finally, I will
conclude this section with a discussion of the relative significance of these competing theories,
and the overlap between practical and theoretical work.

24

Niche theory- Ecological Filtering
In contrast to the neutral theory, niche theory suggests that biotic and abiotic conditions
influence how communities re-establish after a disturbance. Community assemblage proceeds in
predictable, deterministic fashion due to ecological filtering. Ecological filters are the
environmental conditions such as soil type, water availability, or interactions with other species.
Ecological filters create a niche which can only be filled by a subset of species from a larger
regional species pool (Weiher et al., 2011). Plants that can colonize and persist in a given niche
create a somewhat predictable community. Some have even gone so far as to propose a search for
“assembly rules” (Diamond, 1975). While generally, the concept of niches has gone out of favor,
a semi-recent re-investigation of the niche theory by Chase and Liebold (2003) has renewed
interest in this concept as a unifying theory in ecology. Additionally, recent work in coral reef
community dynamics overturns a decade of assumptions that stochastic processes drive
biodiversity (Connolly et al., 2014) and exemplifies the need for further research for this and
other ecological theories. Generally, the idea that site conditions are appropriate or inappropriate
for species establishment is well recognized in the literature. Fewer studies delve into the reasons
why plants do not establish in certain areas. These challenges to establishment are discussed in
the practical section under site conditions.
Neutral Theory- Priority Effects
There are several proposed mechanisms of stochastic, or neutral community assemblage
processes, including site history, competitive asymmetry, and priority effects (Ejrnaes, Bruun, &
Graae, 2006). I will be focusing only on priority effects because I think it can be easily
incorporated into restoration planning. Priority effects were originally defined as the competitive
boost that an early-establishing species gets from reaching a large size before its competitors
arrive (Wilbur & Alford, 1985). Alternative definitions state that priority effects could have a
positive or negative effect on later occurring species, thus early arriving species may either
impede or facilitate the establishment of later arriving species (Ejrnaes et al., 2006; Vannette &
Fukami, 2014). Regardless, evidence of priority effects can be considered a historical and

25

stochastic process driving community assemblages, supporting the neutral assembly theory
(Weiher et al., 2011). The timing of sowing and priority effects have been shown to be persistent,
and important for developing a stable native state rather an exotic state (Martin & Wilsey, 2014),
which is among the ultimate goals of prairie restoration.
Priority effects have been demonstrated in a variety of different ecosystems with many
different species. Priority effects have been well tested in plant communities in natural settings
such as grasslands and prairies (Dickson, Hopwood, & Wilsey, 2012; Fukami et al., 2005; Grman
& Suding, 2010; Helsen et al., 2012; Hooper & Dukes, 2010; Plückers et al., 2013; von
Gillhaussen et al., 2014; Larson et al., 2011; Martin & Wilsey, 2012) wetlands (Pfeifer-Meister et
al., 2012) and shrub steppe (Schantz et al., 2014) as well as in microcosm experiments (Ejrnaes et
al., 2006; Körner, Stöcklin, Reuther-Thiébaud, & Pelaez-Riedl, 2008; Ross & Harper, 1972).
Other species that have been tested include frogs (Wilbur & Alford, 1985), fungi (Fukami et al.,
2010), phytoplankton (Robinson & Edgemon, 1988) multi-trophic aquatic systems (Drake, 1991)
dragon flies (Amoroso & Chalcraft, 2015) and yeasts (Vannette & Fukami, 2014). While priority
effects are not a new or understudied idea, their practical application to prairie restoration in the
south Puget Sound has yet to be determined.

Priority effects in greenhouse studies
Priority effects have been observed in greenhouse studies to have long lasting
implications for plant community development, which is often the target of restoration ecology.
These effects are found both within the arrival times of seeded species versus the existing
community into which they are sown, and the equal start given to faster or slower developing
species (Körner, Stöcklin, Reuther-Thiébaud, & Pelaez-Riedl, 2008). In a greenhouse study, a
three week earlier arrival time of certain plant functional groups resulted in lasting community
composition and biomass effects even after several harvests and two growing seasons (Körner et
al., 2008). Priority effects act quickly, even a few weeks delay in sowing time substantially alters

26

the resulting community. Certain plant functional groups can dominate over others when given a
six week head start, but more later sown functional groups are represented when there is only a
three week gap (von Gillhaussen et al., 2014). Likewise, in a greenhouse study, native Australian
grasses were able to suppress growth of a competitive non-native grass when given a three week
head start (Firn, MacDougall, Schmidt, & Buckley, 2010). In an outdoor microcosm experiment
the species composition was determined more by priority effects than by traits and preferred
habitat type of the species tested (Ejrnaes et al., 2006). The relative arrival times of species and/or
functional groups can influence their successional trajectories.
Priority effects in field studies
Priority effects may have long-term implications for restoration. Species sown first may
have future implications for community diversity and composition (Bullock, Pywell, & Walker,
2007; Hoelzle, Jonas, & Paschke, 2012). A sowing experiment in the dry, acidic grassland of
Germany showed that after four years priority effects were still evident in the above ground
productivity, community cover, and functional group composition, though not on total species
number, nor target species number (Plückers et al., 2013). Priority effects are a complex and
important part of many restoration processes, the practical application of which has been largely
overlooked in SPS prairie restoration.
Some studies have found that other mechanisms are more responsible for community
assembly than priority effects. A study by (Amoroso & Chalcraft, 2015) found that the duration
(length of time) that available habitat was open for colonization was more indicative of patch
biodiversity than variation (when) the window of time (early vs late season) was open for
colonization in a study of ephemeral ponds and dragonflies. Others have found that propagule
pressure (amount of seeds) may be more indicative of community assemblage than priority
effects (Schantz et al., 2014). The species richness of phytoplankton was caused by invasion rate
(i.e. propagule pressure), invasion timing, and invasion order, with invasion order explaining the
least of the variation (Robinson & Edgemon, 1988). In another example, the convergence of

27

community assembly in a wetland restoration was attributed to both inhibitory priority effects and
competitive dominance by perennial grass species (Pfeifer-Meister et al., 2012). Fukami, Martijn
Bezemer, Mortimer, & Putten (2005) found that priority effects and “trait-based assembly rules”
were simultaneously responsible for community assembly in abandoned agricultural fields. These
studies indicate that further testing is needed to determine the relative importance of priority
effects within the larger context of community assembly theory.
Overlap of Niche and Neutral Theories
Of course, there is considerable overlap between these two theoretical frameworks. For
example, the strength of priority effects may be dependent upon niche components (Vannette &
Fukami, 2014) or ecological filters, such as drought (Chase, 2007). This suggests that priority
effects are not acting alone as the sole mechanism of community assembly, but they are one piece
of a larger puzzle. This highlights the idea that there is a relative influence between priority
effects, and ecological filtering, just as there is a middle ground between neutral and niche theory.
Some researchers have found that community assembly is generally deterministic (niche theory),
but that the actual species composition is historically influenced (neutral theory) (Fukami et al.,
2005; Helsen et al., 2012). Thus, while all prairies will look structurally similar to each other,
they may be made up of different species. These findings have implications or theoretical insights
and for the practical application prairie restoration- whichever species occupies a niche first may
persist so long as there is an appropriate niche available for it. Predictability would be a useful
tool for the restoration ecologist, but to date standard ‘assembly rules’ have been elusive despite
patterns in some ecosystems (Weiher, Clarke, & Keddy, 1998). Hopefully this experiment will
shed some light on whether community assembly is driven more by neutral or niche processes, as
well as the applicability of these theories to on the ground restoration practices in the SPS
prairies.

28

Application to SPS prairies
My theoretical research question is: To what degree is plant community assembly driven
by neutral or niche processes? The sown seeds are arriving in an environment where some native
and non-native species already occur, and may be arriving before or after the propagules of these
and other species. Presumably, the earlier the sown seeds arrive the better chance they will have
of competing for resources and achieving establishment, demonstrating priority effects, a neutral
process. Simultaneously, these seeds are affected by the biotic and abiotic conditions of three
different locations. The relative quality of each location is a rough proxy describing ecological
filtering processes. Presumably ecological filters only allowing a subset of the seeds or species to
establish in each location, thus demonstrating niche processes. I hypothesized that both neutral
and niche processes are likely acting on this restoration experiment. Better understanding
practical research questions in the context of theoretical frameworks potentially broadens the
usefulness of such research.
Sufficiently testing any community assembly theory in the field setting with native plants
will likely require more than one growing season. The first growing season’s data presented here
only offers a preliminary snapshot into what will be influencing the establishment and long term
community composition. A longer study period for this experiment may show population
dynamics and trends within plant community assembly and coexistence, thus shedding more light
on priority effects, ecological filtering, and overall community assembly theory. Understanding
community assembly theory and its drivers may determine the success of restoration projects
(Chase, 2003). Testing ecological theory while simultaneously doing prairie restoration seems
like a mutually beneficial relationship between theory and practice, the results of which not only
benefit local ecosystems, but broader audiences as well.
Conclusion of Literature Review
This study has many potential benefits to both science and the ecosystem. In the scientific
community, restoration ecology has been viewed as the testing grounds of ecological theory
(Cairns and Heckman, 1996), thus a natural fit for both theoretical and practical learning. This

29

study provides benefits to the ecosystem in multiple ways, as it increases our understanding, but
as a field experiment, it actually is a small-scale restoration project. Seed addition experiments
often increase the native species richness. This study both re-introduces species that were likely
found in these areas historically but have since been extirpated, and augments other populations
that are currently found in and near the study areas but at a lower densities. Increasing native
biodiversity through re-introduction and augmentation can have an impact on ecosystem health
and resiliency to invasion. Restoration of vegetative communities may have positive implications
for arthropod community assemblage (Déri et al., 2011; Krauss, Steffan-Dewenter, & Tscharntke,
2003; Summerville, 2008). Overall, this study is currently benefitting the ecosystem in the small
scale, but the lessons that can be learned are applicable to large-scale restorations.
Although prairie restoration can happen much more quickly than restoration of an old
growth forest, it still takes a considerable amount of time. Without active restoration efforts it
takes a long time, perhaps 50 years or more, for an ex-arable field to regain the species richness
of a semi-natural grassland even in areas where seeds naturally disperse from adjacent seminatural grasslands (Öster, Ask, Cousins, & Eriksson, 2009). Even with active restoration effort
such as seed addition, it may take over a decade to reach target objectives. Unfortunately, some
studies have shown that seed addition may not result in long-term persistent populations (Rinella,
Mangold, Espeland, Sheley, & Jacobs, 2012). Other long term studies have shown unanticipated
results such as the dominance of seeded native grass species to the exclusion of seeded native
forbs (Sluis, 2002). Regardless, seed addition and other types of experiments can improve
decision making on the amount of seed needed to reach target populations, as well as species
selection (Sayuti & Hitchmough, 2013). Some studies have shown successful restoration, and
persistent communities eight years after a one time seed addition (Foster & Tilman, 2003). There
is much to be learned about the long term, short term, and transient coexistence of native and
exotic species in grassland ecosystems.

30

METHODS
Site Description
The experiment was conducted from September 2014 until July 2015 at Joint Base
Lewis-McChord’s Rainier Training Area (RTA) in western Washington (46°54’ N, -122°43’W).
The RTA lies within the Willamette Valley-Puget Trough- Georgia Basin Ecoregion which is a
long, narrow stretch of lowlands between the Cascades and coastal mountain ranges. More
specifically, the RTA is part of the south Puget Sound (SPS) prairies, which were formed when
glaciers retreated across this landscape approximately 12,000 years ago. The soils where study
plots were located are Spanaway gravelly sandy loam, 0-3% slopes, created from parent materials
of volcanic ash over gravelly outwash (Natural Resources Conservation Service, 2015). These
soils are 10% organic matter and “somewhat excessively drained,” though it is classified as
“prime farmland if irrigated” (Natural Resources Conservation Service, 2015). The mean annual
precipitation for the study site is 40-60 inches, and the mean annual air temperature is 48-52
degrees F (Natural Resources Conservation Service, 2015). The climate is temperate maritime,
with mild wet winters, and warm dry summers (LandScope America, 2015). Frost action for soils
of study site are ranked low. The frost-free period is 200-240 days (Natural Resources
Conservation Service, 2015).
In 1943 Fort Lewis (now Joint Base Lewis-McChord) acquired the 18,000 acres south of
the Nisqually River, which became the RTA (Lewis History Museum, 2014). Currently, the land
is used for military training exercises and public recreation. Mowing has been used at this site to
reduce tree and shrub encroachment and maintain native prairie species. Prescribed fire has been
used at this size for at least the last six to eight years. No records could be found describing the
burn history prior to 2008.
Locations on Upper Weir, Lower Weir, and South Weir prairie were selected (Figure 1).
These prairies have relatively high levels of access for setup and monitoring, and they are at little
risk of being impacted by military training. Each of the three prairie represents a different quality
category.

31

Figure 1: Map of replicate locations
Map shows locations of Upper Weir prairie (high quality), Lower Weir prairie (medium
quality) and South Weir prairie (low quality) replicates in relation to one another. Each prairie
contains three replicates (n=3). Each replicate contains one array of plots.

32

Each of the three prairies has a unique burn history (Table 1), though all three locations
were burned in summer 2014 before experiment began. During site selection an effort was made
to choose only areas with relatively homogeneous burn effects. Future studies should include a
pre-burn survey, minimally for species richness, to help tease apart the confounding factors of
naturally occurring (extant populations) and seeded species.
Location of Replicates

Recent Burn History

Upper Weir

8/12/2008, 8/30/2011, 9/8/2014

Lower Weir

9/14/2009, 9/21/2012, 7/17/2014, 8/5/2015*

South Weir

7/14/2010, 9/17/2012, 9/4/2014

Table 1. Burn history of study sites
Replicates locations on Upper, Lower, and South Weir prairies have had unique, but generally
similar management in the recent past. Replicates on each prairie where positioned within the
same burn unit so that they would have likely received a similar burn history. Burning before
2008 could have occurred, but there are no written records of such treatments. There was
likely a large gap in burn history due to the cessation of Native American burning following
European American settlement in the 1850’s, and the initiation of ecological prescribed fire in
the 2000’s.
*8/5/2015 burn date occurred after the completion of data collection for this project. However,
this may impact the results of continued monitoring efforts of this research.

The quality of prairies can be rather subjective. The structure and function of a prairie can
be evaluated by looking at the types and abundances of native and non-native vegetation, the
amount of woody vegetation present, and the ground cover types, among other ecological cues.
Comparing these ecological cues to high quality remnant prairies offer a rough comparison
between sites. Personal observation and data from this experiment show the three locations
selected in this experiment are different from one another despite some similarities. Data

33

collected during the monitoring of this experiment suggest that the three locations are comprised
of different quantities of native and non-native species (Figure 2).

Figure 2: Percent cover of native and exotic plants for each study site
Cover of native, exotic (non-native) and no plant is shown as an average of each replicate
(n=3). UW, SW, and LW are Upper Weir, South Weir, and Lower Weir, respectively.
Coverage over 1.00 (100%) is possible due to overlapping of plants.
Upper Weir had 83-124% native cover, hereafter “high quality,” South Weir had 32-53%
hereafter “low quality,” and Lower Weir had 72-94% native cover, hereafter “medium quality.”
Percent cover is often over 100% due to overlapping plants. Native perennial species may take
several years to attain a large size that would skew the results of coverage calculations. Likewise,
the annual species selected for this experiment are typically very small in these growing
conditions, and were not likely to skew the results coverage calculations. With these

34

considerations in mind, the circularity of the argument “which came first the seed addition or
relative site quality?” is partially avoided. Unfortunately, due to the techniques used in this study
it is impossible to entirely separate the pre-existing species and their abundance, with those that
established as a result of the experimental treatments. An additional step, preliminary vegetation
monitoring prior to site preparation (i.e. burning) and seed addition, would have been extremely
useful in this situation.
Upper Weir prairie replicates were in the highest quality prairie of the three locations,
and the control plots contained 11 of 23 seeded species (Table 2). Lower Weir prairie replicates
were in medium quality prairie which contained nine of 23 seeded species. South Weir prairie
replicates were located in the lowest quality prairie, relative to the other two sites, but still nine of
23 seeded species persisted in the control plots. Currently, the dominant vegetation on the site is
exotic colonial bentgrass (Agrostis capillaris), and native Roemer’s fescue (Festuca idahoensis
subsp. roemeri), as well as other exotic and native graminoids, forbs, and occasionally shrubs
such as exotic Scot’s broom (Cytisus scoparius). South Weir and Upper Weir have had targeted
restoration efforts for Mazama pocket gopher. South Weir has had some habitat enhancement in
preparation for proposed Taylor’s checkerspot butterfly release on adjacent CNLM owned land
(Prairie Habitat Enhancement Report on JBLM, Annual Report, 2014). These restoration efforts
do not include the areas where the plots are located, but may contribute to the overall health of
the prairies.

35

Species

Low Quality
(South Weir)

Medium Quality
(Lower Weir)

High Quality
(Upper Weir)

Achillea millefolium

Y

Y

Y

Armeria maritima
Balsamorhiza deltoidea
Cerastium arvense

Y

Clarkia amoena
Collinsia grandiflora
Collinsia parviflora
Danthonia californica

Y

Y

Y

Eriophyllum lanatum

Y

Y

Y

Festuca idahoensis subsp.
roemeri

Y

Y

Y

Koeleria macrantha

Y

Y

Lomatium utriculatum

Y

Y

Lupinus albicaulis

Y

Erigeron speciosus

Y

Lupinus bicolor

Y
Y

Microseris laciniata

Y

Y

Plectritis congesta
Potentilla gracilis
Ranunculus occidentalis

Y

Y

Y

Sericocarpus rigidus

Y

Y

Y

9/23

9/23

11/23

Sisyrinchium idahoense
Solidago simplex
Viola adunca
Number of sown species
present in control plots

36

Table 2: Presence of seeded species in control plots
Seeded species are listed in alphabetical order, and the presence of the species in at least one
of the control plots per prairie indicated by “Y.” This gives the reader a glimpse of the species
found in the prairie prior to the initiation of this experiment. While the control plots do not
explain all of the variation found within the replicates on each prairie, it provides an overview
of the most common species.
Experimental Design
In September 2014, a systematic block array in which each set of replicates was spatially
segregated from the others was established in the RTA prairies. Within each of the relative
qualities of prairie (high, medium, and low) there were three replicates for each sowing time
(n=3). Each replicate was placed on a roughly homogeneous 5x5m area with a 5m buffer zone
between each seeded treatment. Each plot was raked by hand with standard garden rakes prior to
sowing in order to mimic a tractor-pulled harrow’s soil and moss disturbance. This pre-sowing
raking created more available microsites that seeds require for germination, and increased
likelihood of seed-soil contact.
At each of four treatment time points, a seed mix of 23 native species was mixed with
vermiculite to enable more even distribution, and sown via drop seeder into each of nine plots.
Sowing time treatments were September 29, 2014; October 29, 2014; December 17, 2014, and
March 16, 2015. Control plots were raked and vermiculite was added, while seeds were withheld.
Additionally, un-manipulated control plots were established within the plot array, and these were
not raked, and vermiculite and seeds were withheld. The un-manipulated control was added to
make sure that the soil disturbance (micro-site preparation) did not influence the establishment of
seeds already present in the soil-seed bank. Array set up is shown in Figure 3. Each treatment plot
is 5 x 5 meters with a 5-meter buffer zone around each, with the exception of the un-manipulated
controls, which were set up in a representative, untouched block within array. These unmanipulated controls were added after the arrays were set up when further questions were raised
about the effects of disturbance, and potentially activating the soil seed bank. There are a total of
54 plots, which were monitored in spring 2015.

37

5m

5m

September

October

December

Unmanipulated

Control

March

Control

25m

Figure 3: Experimental design
Typical array set up is shown. Each treatment plot is 5 x 5 meters. Sowing time treatments
were September 29, 2014; October 29, 2014; December 17, 2014, and March 16, 2015.
Control plots were raked and vermiculite was added, while seeds were withheld.
Unmanipulated control plots were established within the plot array, and these were not raked,
and vermiculite and seeds were withheld. Three arrays were set up per prairie.

Species selection
There are 190 past and present native herbaceous species (Dunwiddie, Alverson, Martin,
& Gilbert, 2014) it would be logistically challenging to test all of them. A subset consisting of 23
species was selected based on availability of a sufficient quantity of seed, time, and resources. An
effort was made to select a broad spectrum of plants including perennial grasses, perennial forbs,
and annual forbs (Table 3). Some species are resources for threatened and endangered species
(e.g., Plectritis congesta is a Taylor’s checkerspot butterfly nectar resource) but most are

38

relatively common and widespread. A broad range of plant families are represented and this may
allow some transferability to other species not tested here. Seeds were obtained from the Center
for Natural Lands Management (Olympia, Washington, USA) and were not stratified or otherwise
pre-treated before sowing. Seeds were grown at Webster’s Nursery and were of “A” grade
(highest quality) clean seed. A restoration seed mix for upland prairie was calculated using Center
for Natural Lands Management’s 2014 Seed Mix Calculator. The seed mix calculator calculates
the weight of bulk seed need to meet specific restoration targets using the latest seed purity and
viability information and the best available science for field establishment rates. See Table 4 for
more information about seed mix calculations.

Scientific Name
Achillea millefolium L.
Armeria maritima (Mill.)
Willd.
Balsamorhiza deltoidea Nutt.
Cerastium arvense L.
Clarkia amoena (Lehm.) A.
Nelson & J.F. Macbr.
Collinsia grandiflora (Lindl.)
Collinsia parviflora (Lindl.)
Danthonia californica Bol.
Eriophyllum lanatum (Pursh)
Forbes
Erigeron speciosus (Lindl.)
DC.
Festuca idahoensis Elmer
subsp. roemeri (Pavlick) S.
Aiken
Koeleria macrantha (Ledeb.)
Schult.
Lomatium utriculatum (Nutt.
Ex Torr. & A. Gray)

Common
Name
common
yarrow

Family

Growth
Habit

Duration

Asteraceae

F

P

thrift seapink
deltoid
balsamroot

Plumbaginaceae

F

P

Asteraceae

F

P

field chickweed
farewell to
spring
giant blue eyed
Mary

Caryophyllaceae

F

P

Onagraceae

F

A

Scrophulariaceae

F

A

Blue eyed Mary
California
oatgrass
common
woolly
sunflower

Scrophulariaceae

F

A

Poaceae

G

P

Asteraceae

F

P

aspen fleabane

Asteraceae

F

P

Poaceae

G

P

Poaceae

G

P

Apiaceae

F

P

Roemer's
fescue
prairie
Junegrass
common
lomatium

39

Lupinus albicaulis Douglas
Lupinus bicolor Lindl.
Microseris laciniata (Hook.)
Sch. Bip.
Plectritis congesta (Lindl.) DC.
Potentilla gracilis Douglas ex
Hook.
Ranunculus occidentalis Nutt.
Sericocarpus rigidus (Lindl.)
Sisyrinchium idahoense E.P.
Bicknell
Solidago simplex Kunth
Viola adunca Sm.

sicklekeel
lupine
miniature
lupine
cutleaf
silverpuffs
shortspur
seablush
slender
cinquefoil
western
buttercup
Columbia
whitetop aster
Idaho blue-eyed
grass
Mt. Albert
goldenrod
Hookedspur
violet

Fabaceae

F

P*

Fabaceae

F

A

Asteraceae

F

P

Valerianaceae

F

A

Rosaceae

F

P

Ranunculaceae

F

P

Asteraceae

F

P

Iridaceae

F

P

Asteraceae

F

P

Violaceae

F

P

Table 3: Species selected
Species are listed with their common name, family, functional group where F is forb and G is
grass, and life history, where A is annual and P is perennial. Information from USDA, NRCS
Plants Database (2016). P*= perennial plant in SPS prairies, but is listed as an annual plant in
USDA, NRCS Plants Database (2016).

40

Species

Species
Achillea
millefolium
Armeria
maritima
Balsamorhiza
deltoidea
Cerastium
arvense
Clarkia
amoena
Collinsia
grandiflora
Collinsia
parviflora
Danthonia
californica
Eriophyllum
lanatum
Erigeron
speciosus
Festuca
idahoensis
subsp.
roemeri
Koeleria
macrantha
Lomatium
utriculatum
Lupinus
albicaulis
Lupinus
bicolor
Microseris
laciniata
Plectritis
congesta
Potentilla
gracilis
Ranunculus
occidentalis
Sericocarpus
rigidus
Sisyrinchium
idahoense
Solidago
simplex

Target
density
(plants/
m2)

Target
cover

Approx.
number of
seeds/gram

Grams/
replicate
(5mx5m
plot )

Seeds/
replicate
(5mx5m
plot)

Seeds/
m2
sowing
rate

Seeds/
3m2
sowing
rate

Target
density
(plants/
m2)

Target
cover

Approx.
number of
seeds/gram

Grams/
replicate
(5mx5m
plot)

Seeds/
replicate
(5mx5m
plot)

Seeds/
m2
sowing
rate

Seeds/
3m2
sowing
rate

2

1%

6455

0.05

322.75

1 2.91

38.73

2

1%

465

0.38

176.7

7.07

21.20

0.5

10%

138

1.81

249.78

9.99

29.97

4

2%

5704

0.04

228.16

9.13

27.38

4

2%

2152

0.64

1377.28

55.09

165.27

3

0.50%

268

1.12

300.16

12.01

36.02

4

1%

653

0.61

398.33

15.93

47.80

3

2%

250

1

250

10.00

30.00

1

2%

2710

0.03

81.3

3.25

9.76

2

2%

5240

0.07

366.8

14.67

44.02

5

40%

1123

1.25

1403.75

56.15

168.45

2

1%

8132

0.2

1626.4

65.06

195.17

4

2%

631

0.71

448.01

17.92

53.76

1

2%

33

8.24

271.92

10.88

32.63

2

2%

169

5.32

899.08

35.96

107.89

3

1%

924

0.49

452.76

18.11

54.33

4

2%

1699

0.24

407.76

16.31

48.93

2

2%

2901

0.18

522.18

20.89

62.66

3

2%

419

3.46

1449.74

57.99

173.97

7

3%

2941

0.54

1588.14

63.53

190.58

2

0.50%

763

0.79

602.77

24.11

72.33

4

2%

3446

0.49

1688.54

67.54

202.62

41

Viola adunca

6

3%

1293

2.51

3245.43

129.82

389.45

Total:

70.5

86%

n/a

30.17

18357.74

734.31

2202.93

Table 4: Seed Mix Calculations
Each species is listed with restoration target densities and target cover, approximate number of
seeds per gram of bulk seed weight, the quantity of seed that was weighed out for each 5x5m
replicate, the average seeding rate per square meter, and finally the average seeding rate per 3
square meters. 3 square meters is amount of area monitored for this study, for easy
comparison. These numbers were calculated using Center for Natural Lands Management’s
2014 Seed Mix Calculator which takes into account the most recent restoration targets and
seed viability information.

Monitoring Methods
Monitoring within each replicate was done using two different methods. Point intercept
method was done every half meter along three randomly selected transects per replicate. The
identity of any species (native or exotic) was recorded in order to calculate percent cover. These
point intercept data were used to verify the relative quality of each of the three prairies. The
second method, density of plants per 1m2 quadrats was done to determine count of individual
species per unit area. These count data were collected for seeded species only within three
randomly selected samples per replicate, these samples were later summed to the plot level. These
density count data were used to determine the effects of sowing time and relative prairie quality
on the first year establishment of the seeded native species. All data were recorded by hand on
data sheets and later entered into Excel.
Statistical Analysis
To compare the raked control with the un-manipulated control a two tail t-test with
unequal variances was performed in Excel. The mean density between these two treatments were
not significantly different (α=0.05) for each species, so only the control was used in further
analysis.
To test the effects of sowing time and relative prairie quality on the overall diversity of
sown native species in the plots, a Shannon-Wiener diversity index was calculated in Excel. This

42

diversity index takes into consideration both the species richness (defined as the number of
different species) and abundance (defined here as the count of individuals in each species)
(Spellerberg & Fedor, 2003). Count data were summed for each sowing time and prairie quality
combination. A total of 15 different Shannon’s H’ values were calculated.
To test the effects of sowing time and site quality on each species individually, a
generalized regression model, a type of general linear model, was developed in JMP. A number
of considerations went into making this decision because these data were over-dispersed meaning
that variance was greater than mean for each species, and most closely fit a negative binomial
distribution. A factorial cross of sowing time and relative prairie quality was used as the predictor
variables, and count data were the response variable. Maximum likelihood was the estimation
method and the mean model link was logarithmic. General linear models are often used for
analyzing non-parametric data, the product of random variation in ecology and evolution studies
(Bolker et al., 2008). General linear models have been used in experiments similar to this one (see
Bakker, Colasurdo, & Evans, 2012; Dunwiddie & Martin, 2016; Zeiter, Stampfli, & Newbery,
2006) though other analyses such as hurdle models, may be more appropriate in certain
circumstances (Potts & Elith, 2006; Zuur, Ieno, Walker, Saveliev, & Smith, 2009). If the
generalized regression demonstrated an effect (p<0.05) from either sowing time or relative prairie
quality on each species then a post-hoc Steel-Dwass all pairs multiple comparisons test was
performed. The Steel-Dwass test is similar to a Tukey’s honestly significantly different test, but is
for non-parametric data, and is more conservative.

RESULTS
Overall establishment rates were low for most species, with 13 of 22 species showing less
than 5% establishment (Table 5). Some species (five of 22) had modest establishment rates of 630%. Two species had greater than 30% establishment, though this was likely due to
experimental design flaw, rather than as a result of the treatments. One species, Festuca
idahoensis subsp. roemeri, had a relatively high rate of establishment, 55%, presumably due to

43

large number of already established plants and/or large number of small seedlings. The second
species, Eriophyllum lanatum, had an unrealistic establishment rate of 887% primarily due to the
way plants were counted and the background presence of several adult plants. Eriophyllum
lanatum is rhizomatous so stems were counted because it was impossible to positively identify
individual plants without digging them up. See “suggestions for future research” section for ideas
of ways to better test the establishment rates of these species.

Species

Achillea
millefolium
Armeria
maritima
Balsamorhiza
deltoidea
Cerastium
arvense
Clarkia amoena
Collinsia
grandiflora
Collinsia
parviflora
Danthonia
californica
Eriophyllum
lanatum
Erigeron
speciosus
Festuca
idahoensis subsp.
roemeri
Koeleria
macrantha
Lomatium
utriculatum
Lupinus
albicaulis
Lupinus bicolor
Microseris
laciniata

Target
density
(plants/
m2)

Target
cover
(%m2)

Sowing
rate
(seeds/ m2)

Observed
Average
density
(plants/m2)

Difference
between
observed
and target
densities
(plants/m2)

Establishment rate
(plants/seeds)

2

1%

12.91

1.33

-0.67

10%

2

1%

7.07

0.00

-2.00

0%

0.5

10%

9.99

0.09

-0.41

1%

4
4

2%
2%

9.13
55.09

0.39
0.06

-3.61
-3.94

4%
0%

3

0.5%

12.01

5.56

-1.44

20%

4

1%

15.93

3

2%

10

0.09

-2.91

1%

1

2%

3.25

28.84

27.84

887%

2

2%

14.67

0.00

-2.00

0%

5

40%

56.15

30.86

25.86

55%

2

1%

65.06

1.10

-0.90

2%

4

2%

17.92

0.15

-3.85

1%

1
2

2%
2%

10.88
35.96

2.69
1.06

1.69
-0.94

25%
3%

3

1%

18.11

1.40

-1.60

8%

44

Plectritis
congesta
Potentilla
gracilis
Ranunculus
occidentalis
Sericocarpus
rigidus
Sisyrinchium
idahoense
Solidago simplex
Viola adunca

4

2%

16.31

0.84

-3.16

5%

2

2%

20.89

0.19

-1.81

1%

3

2%

57.99

4.96

1.96

9%

7

3%

63.53

19.34

12.34

30%

2
4
6

0.5%
2%
3%

24.11
67.54
129.82

0.17
0.00
0.00

-1.83
-4.00
-6.00

1%
0%
0%

Table 5: Overall establishment rates
Species are listed with target density, target percent cover, calculated sowing rate, observed
average density of plants, difference between observed and target counts of plants, with
negative numbers in red, and establishment rate. Observed average density (plants/m2) was
calculated by averaging counts of all sowing times except the control plots. Establishment
rates highlighted in green are those species not found in any control plot.
All species showed a unique response to the treatments in the study. For various reasons
not all species were analyzed. Three species which were not encountered at all during monitoring:
Armeria maritima, Solidago simplex, and Viola adunca. A fourth species, Erigeron speciosus is
notoriously difficult to identify as a seedling, so this species was also considered “not
encountered.” These four species were excluded from all analysis, and further research should be
done to determine their unique establishment requirements. Two congeneric species--Collinsia
grandiflora and Collinsia parviflora--were often indistinguishable from each other during
monitoring and were combined under the label Collinsia spp.. Collinsia spp. was treated as a
single species for analysis.
The Shannon-Weiner diversity index included sixteen sown native species: Achillea
millefolium, Balsamorhiza deltoidea, Cerastium arvense, Clarkia amoena, Collinsia spp.,
Danthonia californica, Koeleria macrantha, Lomatium utriculatum, Lupinus albicaulis, Lupinus
bicolor, Microseris laciniata, Plectritis congesta, Potentilla gracilis, Ranunculus occidentalis,
Sericocarpus rigidus, and Sisyrinchium idahoense. Meanwhile, six total species were excluded
from this analysis. Of these, two species, Festuca idahoensis subsp. roemeri and Eriophyllum

45

lanatum were excluded from this analysis because the majority of the plants counted were not a
result of the seed addition. Additionally, the four species not encountered were excluded from
this analysis: Armeria maritima, Erigeron speciosus, Solidago simplex, and Viola adunca.
The general linear model analysis was run on thirteen species independently. The thirteen
species were: Achillea millefolium, Cerastium arvense, Collinsia spp., Danthonia californica,
Eriophyllum lanatum, Festuca idahoensis subsp. roemeri, Koeleria macrantha, Lupinus
albicaulis, Lupinus bicolor, Microseris laciniata, Plectritis congesta, Ranunculus occidentalis,
and Sericocarpus rigidus. A total of nine species were excluded from this analysis. The four
species not encountered were obviously excluded. Additionally, five species were excluded
because they were found in too low abundance, typically fewer than 30 individual plants were
found across all treatment combinations. The plants found in too low of abundance were:
Balsamorhiza deltoidea, Clarkia amoena, Lomatium utriculatum, Potentilla gracilis, and
Sisyrinchium idahoense.
Community Response
The Shannon-Weiner diversity index takes into account both richness and abundance of
all seeded species (Figure 4). The higher the H’ value, the more types and abundances of sown
species. The H’ value was highest for the September treatment in the high quality prairie with
H’=2.05, followed by October treatments in medium and high quality prairies with H’=2.03 and
H’=1.93 respectively. For reference, highly diverse, unharvested tropical forests in Papua New
Guinea showed a Shannon’s diversity index of 4.9 ± 0.21 standard deviation (Yosi, Keenan, &
Fox, 2011), meanwhile forest stands in Oklahoma ranged from 0.72-1.48 and had a maximum of
32 species (Risser & Rice, 1971). Only the native, sown species were included in this analysis,
with the exception of the two most common species that were found in high numbers. A
Shannon’s diversity index that included other native species and non-native species would have
resulted in a much higher H’ values.

46

1.38

2.05

1.93

1.47

1.70

1.18

1.75

2.03

1.18

0.18

0.75

1.63

1.69

1.05

0.78

Figure 4: Community diversity
Shannon Weiner Diversity Index calculated on only seeded species, and summed to the plot
level. Festuca idahoensis subsp. roemeri and Eriophyllum lanatum were excluded from this
analysis due to high numbers and relatively large plants which suggests that most of the plants
counted were already established before seeding took place.
Individual Species Response
Individual species responded uniquely to both sowing time treatments and varying prairie
qualities. Three species, Collinsia spp., Lupinus bicolor, and Plectritis congesta were statistically
significantly affected by treatment time (α=0.05). Initially, five species: Achillea millefolium,
Danthonia californica, Eriophyllum lanatum, Koeleria macrantha, and Ranunculus occidentalis,

47

were statistically significantly affected by prairie quality (α=0.05). However, upon further
analysis (post hoc Steel-Dwass multiple comparisons test, α=0.05) Koeleria macrantha showed
that it was not actually significantly influenced by quality. One species, Lupinus albicaulis, was
initially affected by both treatment and quality, but upon further analysis (post hoc Steel-Dwass
multiple comparisons test, α=0.05) was found to not actually be significantly influence by quality.
The other four analyzed species, Cerastium arvense, Festuca idahoensis subsp. roemeri,
Microseris laciniata, and Sericocarpus rigidus were not significantly affected by treatment or
prairie quality.
Each species is grouped by type of response, and results are described in more detail
below. Where appropriate, generalized regression and Steel-Dwass all pairs multiple comparisons
test results are considered and general trends are also discussed.
Species significantly influenced by sowing time
Collinsia spp., which includes both C. parviflora and C. grandiflora, commonly known
as blue-eyed Mary and giant blue-eyed Mary respectively, were strongly influenced by sowing
time (p<0.0001), but not by site quality (p<0.39). These annual forbs responded more positively
to September and October sowing times than December sowing times (Figure 5). Of the total 600
individual plants found across all relative prairie qualities, there were 324 in September, 223 in
October, and 53 in December. No plants were found in the March or Control plots. Plants were
found across all three site conditions. Of the 600 total plants found, 273 were in medium quality,
197 were found in low quality, and 130 in high quality. These plants are rare in SPS prairies, and
were likely extirpated from the area. In order to accurately distinguish these congeneric species
one must monitor multiple times during the year, since they can only reliably be distinguished by
their flower sizes.

48

A

A

B

Figure 5: Collinsia spp. results
Different letter (A, B, C, etc.) indicate that each sowing time is significantly different from
one another (α = 0.05). Letters that are shared between two sowing times indicate that they are
not significantly different from one another. Error bars represent standard error.

Lupinus bicolor, commonly known bicolor lupine or miniature lupine, was influenced by
seed sowing time (p<0.0001) but not by relative site quality (p<0.67). This species responded
positively to all sowing times equally, except the control (Figure 6). Of the total 115 individual
plants found across all relative prairie qualities, the highest counts were found in September (47)
followed by October (32) and December (28) and very few were found in March (7) and in
Control (1). This annual forb is fairly common across restored SPS prairies, and was observed in
medium quality (48), high quality (40), and low quality (27) prairies.

49

A

B

B

B

AB

Figure 6: Lupinus bicolor results
Different letters (A, B, C, etc.) indicate that each sowing time is significantly different from
one another (α = 0.05). Letters that are shared between two sowing times indicate that they are
not significantly different from one another. Error bars represent standard error.

Plectritis congesta, commonly known as shortspur seablush, was influenced by seed
sowing time (p<0.0001) but not relative site quality (0.052). It was found in three of five sowing
times across all three site conditions. Plectritis congesta responded positively to earlier sowing
times than later sowing times (Figure 7). Of the total 91 individual plants found across all relative
prairie qualities, the highest count of was found in September (64), followed by October (23), and
finally December (4). No plants were found in March or Control plots. The highest count of

50

plants was found in high quality (45) followed by medium (28) and low quality (18). This annual
forb is rare in SPS prairies, and was likely extirpated from some areas.

A

B

C

Figure 7: Plectritis congesta results
Different letters (A, B, C, etc.) indicate that each sowing time is significantly different from
one another (α = 0.05). Letters that are shared between two sowing times indicate that they are
not significantly different from one another. Error bars represent standard error.

Lupinus albicaulis, commonly known as sicklekeel lupine, was initially influenced by
both seed sowing time (p<0.0001) and relative site quality (p<0.0001). It responded positively to
fall sowing times rather than spring or the control (Figure 8). However, upon further
investigation, a Steel-Dwass test showed that none of the site conditions were statistically

51

significantly different from each other at α=0.05 (Error! Reference source not found.). While
igh quality and medium quality had similar means (p=1.00), the low quality mean was not
statistically significantly different from medium quality (p=0.12). Likewise, the low quality mean
was not statistically significantly different from the high quality mean (p=0.07). While these
results are not statistically significant they also show a pattern that this species prefers or is found
more often in high and medium quality prairie rather than low quality prairie. These results
should be further investigated. Of 319 total individual plants found across all relative qualities,
the highest abundances were found in October (112) and September (99) with fewer plants found
in December (67), Control (28) and March (13). Of 319 total plants across all treatment times
more were observed in medium quality (139) and high quality (130) than in low quality (50). This
perennial forb is fairly common in SPS prairies, thus found in the control plots, but seed sowing
likely greatly augmented this population.

52

AC

AB

B

A

C

Figure 8: Lupinus albicaulis results 1
Different letters (A, B, C, etc.) indicate that each sowing time is significantly different from
one another (α = 0.05). Letters that are shared between two sowing times indicate that they are
not significantly different from one another. Error bars represent standard error.

53

A

A

A

Figure 9: Lupinus albicaulis results 2
Different letters (A, B, C, etc.) indicate that each relative quality is significantly different from
one another (α = 0.05). Letters that are shared between two relative qualities indicate that they
are not significantly different from one another. Error bars represent standard error.

Species significantly influenced by site quality
Achillea millefolium, commonly known as yarrow, was not influenced by seed sowing
time (p<0.75), but was influenced by site quality (p<0.04). This perennial forb responded most
positively to high and medium quality site conditions (Error! Reference source not found.).
cross all sowing times, of the total 167 individual plants, 112 of them were found in high relative
quality prairie. Of the total 167 individuals, fewer plants were found in medium quality (37) and
low quality (18). Achillea millefolium is a common SPS prairie species, and can be weedy in
certain circumstances. Achillea millefolium was found in all five sowing time treatments,

54

including the control. Of the total 167 individual plants, most plants were found in September (51
total plants), and the fewest plants were found in the control (23 total plants). Achillea millefolium
spreads by rhizomes as well as reproduces by seed, and monitoring protocols did not discern
between the two forms of reproduction. The count of plants likely overestimates the number of
genetically distinct individuals.

A

AB

B

Figure 10: Achillea millefolium results
Different letters (A, B, C, etc.) indicate that each relative quality is significantly different from
one another (α = 0.05). Letters that are shared between two relative qualities indicate that they
are not significantly different from one another. Error bars represent standard error.

55

Danthonia californica, commonly known as California oatgrass, was not influenced by
seed sowing time (p<0.58), but was influenced by site quality (p<0.02). It responded most
positively to high and medium quality sites than to low quality sites (Figure 11). Of the total 145
individual plants found across all sowing times, 66 were found in medium quality, 58 in high
quality, and 21 in low quality. This common perennial grass was found in all five sowing time
treatments, including the control. The highest count occurred in October (35 total plants) and the
lowest count occurred in Control and December (both had 22 total plants). Like many grasses,
they are difficult to identify, especially when very small or when intermixed with other grass
species, so the actual count of this species may be higher than what was found during monitoring.

A

B

B

Figure 11: Danthonia californica results
Different letters (A, B, C, etc.) indicate that each relative quality is significantly different from
one another (α = 0.05). Letters that are shared between two relative qualities indicate that they
are not significantly different from one another. Error bars represent standard error.

56

Eriophyllum lanatum, commonly known as woolly sunflower or Oregon sunshine, was
not influenced by seed sowing time (p<0.42), but was strongly influenced by site quality
(p<0.0001). This species responded most positively to medium quality rather than high or low
quality (Figure 12). Of 3,923 total stems found across all sowing times, most (3026) were found
in medium quality, while fewer (866) were found in high quality, and very few (21) in low
quality. This perennial forb is very common in SPS prairies and was found in all five sowing time
treatments, including the control. Of the total stems counted across all relative prairie qualities,
the highest count of stems was observed in September (970) and the lowest in October (658) This
plant was counted by stems rather than by plants because it was very difficult to determine
independent plants without destructively sampling. This plant reproduces by rhizomes as well as
by seed, so the count of genetically distinct individual plants may be much lower.

57

A

B

C

Figure 12: Eriophyllum lanatum results
Different letters (A, B, C, etc.) indicate that each relative quality is significantly different from
one another (α = 0.05). Letters that are shared between two relative qualities indicate that they
are not significantly different from one another. Error bars represent standard error.

Ranunculus occidentalis, commonly known as western buttercup, was not influenced by
seed sowing time (p<0.29), but was influenced by relative site quality (p<0.01). This species
responded positively to both high and low quality sites rather than medium quality (Figure 13).
Of the total 793 individual plants found, across all of the sowing times, the greatest number of
plants were found in high quality (455) followed by low quality (258) and medium quality (80).
This perennial forb is very common in SPS prairies and was found in all five sowing time
treatments, including the control. Of the total 793 plants found across all relative prairie qualities,

58

the most plants were found in the control (257) while the lowest count was found in March (85).
The highest count is probably skewed due to a much higher on average count (119 plants) in one
quadrat in one of the control plots within the low quality site. This exceptionally high density of
Ranunculus occidentalis may be due to recording error during monitoring or due to the patchy
distribution of the species.

AB

A

B

Figure 13: Ranunculus occidentalis results
Different letters (A, B, C, etc.) indicate that each relative quality is significantly different from
one another (α = 0.05). Letters that are shared between two relative qualities indicate that they
are not significantly different from one another. Error bars represent standard error.

59

Species not significantly influenced by either sowing time or site quality
Koeleria macrantha, commonly known as prairie Junegrass, was not influenced by seed
sowing time (p<0.85), but was initially influenced by relative site quality (p< 0.02). However,
upon further investigation, Steel-Dwass all pairs test showed that none of the site qualities were
actually different from each other (Figure 14). The means of low and medium qualities were very
similar (p=1.00). The means of medium quality and high quality were not significantly different
(p=0.056). Likewise, the means of low and high quality were not significantly different
(p=0.053). While these results are not statistically significant they show a pattern that this species
prefers or is found in higher quality prairie more than low quality prairie. Of 160 total plants
found across all sowing times, most of them were found in high quality prairie (135), while fewer
were found in medium (16) and low qualities (9). This perennial grass is a fairly common species
in SPS prairies, and it was found across all five sowing time treatments, including the Control.
Of the 160 total plants found across all relative prairie qualities, the most total plants were found
in September (43) and the fewest total plants were found in March (7). Like many grasses, they
are difficult to identify, especially when very small or when intermixed with other grass species,
so the actual count of this species may be higher than what was found during monitoring.

60

A

A

A

Figure 14: Koeleria macrantha results
Different letters (A, B, C, etc.) indicate that each relative quality is significantly different from
one another (α = 0.05). Letters that are shared between two relative qualities indicate that they
are not significantly different from one another. Error bars represent standard error.

Cerastium arvense, commonly known as field chickweed, was not influenced by seed
sowing time (p<0.89) or site quality (p<0.18). However, this fairly common perennial forb was
found in all three site qualities and in all five time treatments, including the control (Figure 15).
Of 47 total plants, 28 were found in medium quality, 16 were found in high quality, and three
were found in low quality. The highest count of plants was found in December with 15 total

61

plants and the lowest in the control with five plants. It reproduces by rhizomes as well as by seed,
so the count of genetically distinct individual plants may be much lower.

Figure 15: Cerastium arvense results
Species did not respond to either sowing time or relative prairie quality. Error bars represent
standard error.
Festuca idahoensis subsp. roemeri, commonly known as Roemer’s fescue, was not
influenced by seed sowing time (p<0.63) or site quality (p<0.21). Of 4183 total plants, the highest
number was found in the medium quality prairie (1782) followed by high quality (1376) and low
quality (1025). Festuca idahoensis subsp. roemeri is the most common plant in SPS prairies, and
was found high abundances in all five time treatments, including the control (Figure 16). The

62

highest count of plants was observed in March (1139), and the lowest in December (616).
Identifying individual plants can be challenging in areas that were recently burned. Often the
plant will re-sprout along the outside edges of the original bunch only, thus it will look like many
small plants when in reality it is sourced from one large plant, which could have affected the
count data for this species.

Figure 16: Festuca idahoense subsp. roemeri results
Species did not respond to either sowing time or relative prairie quality. Error bars represent
standard error.

63

Microseris laciniata, commonly known as cutleaf silverpuffs, was not influenced by seed
sowing time (p<0.77) or relative site quality (p<0.82). This is a fairly common perennial forb in
SPS prairies, and was found in all 5 of the sowing time treatments, and in high and medium
quality but not in low quality prairie (Figure 17). Of 191 total plants more than twice as many
were found in high quality (129) than medium quality (62). The highest total count was found in
September (63) and the lowest in March (19).

Figure 17: Microseris laciniata results
Species did not respond to either sowing time or relative prairie quality. Error bars represent
standard error.

64

Sericocarpus rigidus, commonly known as white-topped aster, was not influenced by
seed sowing time (p<0.71) or relative site quality (p<0.65). This perennial forb is common in SPS
prairies and was found in all five sowing treatments, including the control, as well as across all
three site conditions (Figure 18). Of 2483 total plants, the highest count was found in the medium
quality (1225) followed by high quality (784) and low quality (474). The highest counts were
found in December (854) and the lowest were found in September (150). It reproduces by
rhizomes as well as by seed, so the count of genetically distinct individual plants may be much
lower.

Figure 18: Sericocarpus rigidus results
Species did not respond to either sowing time or relative prairie quality. Error bars represent
standard error.

65

Species excluded from analysis due to low abundance
Balsamorhiza deltoidea, commonly known as deltoid balsamroot, was found in very low
quantities: a total of eight plants were found. This plant is relatively rare in the SPS prairies, but
seedlings were found across all three prairie qualities. Due to low sample size this species was
excluded from analysis, but it is interesting that all eight seedlings encountered were in
September seeding time treatments (Figure 19). No seedlings were found in October, December,
March, or Control plots.

Figure 19: Balsamorhiza deltoidea results
Species was excluded from generalized regression analysis due to low abundance. Error bars
represent standard error.

66

Clarkia amoena, commonly known as farewell to spring, was found in very low
quantities: a total of six individuals were found. Due to low sample size this species was excluded
from analysis. This annual forb is very rare in SPS prairies, and was likely extirpated from some
areas. Plants were found in only medium (1) and low qualities (5). No plants were found in high
quality site conditions. Interestingly, of the six total plants, four were found in September plots,
and two were found in October plots. No plants were found in December, March, or Control plots
(Figure 20Figure 20).

Figure 20: Clarkia amoena results
Species was excluded from generalized regression analysis due to low abundance. Error bars
represent standard error.

67

Lomatium utriculatum, commonly known as spring gold or common lomatium, was
found in very low quantities: a total of 16 individuals were found. Due to low sample size this
species was excluded from analysis. This perennial forb is very common in SPS prairies, but
interestingly was not found in low quality prairie, or in the control plots across any of the three
site conditions (Figure 21). Of the 16 total plants, nine of them were found in high quality, and
seven were found in medium quality. Half of the plants were found in the September sowing time
treatment, while only one plant was found in December treatment plots.

Figure 21: Lomatium utriculatum results
Species was excluded from generalized regression analysis due to low abundance. Error bars
represent standard error.

68

Potentilla gracilis, commonly known as slender cinquefoil, was found in very low
quantities, with a total of 20 individuals. This species was excluded from analysis. This plant is
fairly common in SPS prairies, but it was not found in the control plots, or in September or March
(Figure 22). It was mostly found in October (17) and a few in December (3). It was not found in
the low quality site, but was found in the medium quality (15) and high quality (5).

Figure 22: Potentilla gracilis results
Species was excluded from generalized regression analysis due to low abundance. Error bars
represent standard error.

69

Sisyrinchium idahoense, commonly known as Idaho blue-eyed grass, was found in very
low quantities, with a total of 18 individuals. This perennial forb was excluded from analysis.
This plant is rare in SPS prairies, and was not found in the control plots. Most of them were found
in December (17) and one in October. All 18 plants were found in the high quality site, and none
were found in the medium or low quality sites (Figure 23). This forb it looks very much like a
grass before it flowers, so small individuals or individuals intermixed with grass species may
have been overlooked. The actual count of this species may be higher than results presented here.

Figure 23: Sisyrinchium idahoense results
Species was excluded from generalized regression analysis due to low abundance. Error bars
represent standard error.

70

Species not encountered
Armeria maritima, commonly known as sea thrift or sea pink, was not found in any
treatment plot in any of the site conditions. This plant is very rare in SPS prairies and was likely
extirpated from some areas.
Erigeron speciosus, commonly known as aspen fleabane, is notoriously difficult to
identify, and was excluded from analysis. This plant is very rare in SPS prairies and was likely
extirpated from some areas.
Solidago simplex, commonly known as sticky goldenrod, or Mt. Albert goldenrod, was
not found in any treatment plot in any of the site conditions. This plant is very rare in SPS prairies
and was likely extirpated from some areas.
Viola adunca, commonly known as early blue violet or other names, was not found in
any treatment plot in any of the site conditions. This plant is fairly common in SPS prairies, and
was likely extirpated from some areas. In the past the viability of the seeds of this species has
been highly variable due to a combination of improper drying and storage techniques as well as
some natural variability.

DISCUSSION
While sowing time and relative prairie quality affected a few species, many other factors
may have influenced these results, such as precipitation, seed distribution, seed-soil contact,
dormancy, predation, seedling death prior to monitoring, or other factors. Despite the complexity
of these interactions I feel confident in these experiments results. Restoration ecology is a messy
science, full of unexpected surprises and complicated natural phenomena. The lessons learned in
this experiment are worthwhile, and it demonstrates the kinds of interesting questions and
patterns that can be observed in a relatively simple experiment. This experiment also serves as a
pilot project for future studies on these practical questions of when and where seeds should be
sown in order to achieve the greatest restoration success.

71

Community response
The community of sown species showed some interesting trends. The total plot diversity
was highest for September sowing in high quality prairie, H’=2.05. The lowest Shannon-Weiner
diversity index number was found in March treatment in medium quality prairie at H’= 0.18. Low
diversity in March suggests that raking may be having some negative impacts on the existing
native vegetation, or that the seeds did not germinate. Continued monitoring of this experiment
may be able to determine if December or March sown seeds do better the following year,
resulting in delay in diversity. General trends shown by the Shannon’s diversity index indicate
that fall sowing is better than winter or spring sowing, and the overall diversity of the prairies,
despite their relative qualities, increases with seed addition, compared to the unsown controls.
In response to the practical question (Does temporal variation of seed sowing affect the
first year establishment of 23 native prairie species across a gradient of prairie quality?), I
hypothesized that sowing time and relative prairie quality would affect the establishment of these
species. The Shannon’s diversity index results support this hypothesis, showing that both sowing
time and relative prairie quality increase native community diversity. As expected, this research
shows that seed addition generally overcomes the dispersal/seed limitation. This finding agrees
with the findings of Stanley et al. (2008, 2011). This research suggests that fall sowing and higher
relative quality prairie tends to result in higher richness and abundance of native plants than
winter or spring sowing or medium and low qualities.
Individual species response
When considered individually, the sown species demonstrated an incredible amount of
variability. Some species responded to sowing time, others to site condition, and some were not
affected by either treatment. No plant responded to both sowing time and site conditions (α =
0.05). In response to the theoretical question (To what degree is plant community assembly driven
by neutral or niche processes?), I hypothesized that both neutral and niche processes are acting
upon these plants. In order to study community assembly theory I used seed sowing time as a
proxy for priority effects, which is a mechanism of neutral theory, and relative prairie quality as a

72

proxy for ecological filtering, which is a mechanism of niche theory. The individual species
responses do not support my hypothesis that both neutral and niche processes are influencing
community assembly. Several species were excluded from the generalized regression analysis due
to low abundance. Species that were not influence by any treatment or were excluded due to low
abundance, or were not encountered all support the hypothesis is that none of the species respond
to either the sowing time treatments or the relative site qualities.
Sowing time
If priority effects were the most influential mechanism driving community assemblage,
one would expect species to respond more positively to sowing time, rather than to any of the
relative site qualities. Specifically, I hypothesize that the earlier the arrival time, the better a
species chances of survival would be. Based on this neutral theory of community assemblage
September should have the highest count of plants, followed by October, December, March, and
finally the control. Plectritis congesta was the only tested species that responded to seeding time
entirely as I had hypothesized. Additionally two species which were excluded from analysis due
to low abundance, Balsamorhiza deltoidea and Clarkia amoena showed similar patterns of
positive response to September sowing. Two species, Collinsia spp. and Lupinus albicaulis
showed partial support for my hypothesis that early seed sowing times would result in higher
establishment, as they responded positively to both September and October sowing times. For the
theoretical question the null hypothesis stated no difference in the mean counts of plants in any
sowing time treatment, thus the count of plants found in any of the treatments would not be
different from one another. Lupinus bicolor performed well in every sowing time treatment
except the control, which supports the null hypothesis, although still demonstrates that this
species is dispersal/seed limited.
Generally speaking, the species that responded to variation in sowing time preferred
earlier fall sowings (September and October) over later sowings (December and March). Three of
the four species that responded to the fall sowing were annual forbs (Collinsia spp., Lupinus

73

bicolor, and Plectritis congesta) suggesting that annual prairie species would benefit from fall
sowing in restoration sites. Lupinus albicaulis, a perennial forb, also responded to earlier sowing
times; this result suggests that earlier sowing times may be beneficial for other perennial forbs as
well, though more research is needed. This research also does not take into consideration the
health, longevity, or size of the seedlings found in each of the sowing time treatments, though
typically fall sown plants seemed more robust (author’s personal observation).
Relative Quality
If ecological filtering is the most influential mechanism driving community assemblage,
one would expect species to respond more positively to the relative prairie quality, rather than to
any of the sowing times. The high, medium, and low relative prairie qualities are proxies for
biotic and abiotic site conditions. Specifically, I hypothesized that the higher quality the site, the
higher the native species establishment. Based on this niche theory of community assemblage, the
high quality site should have the highest count of plants, followed by the medium quality site, and
finally the low quality site. Two of the four species that responded to site quality, Achillea
millefolium and Danthonia californica, performed best in high or medium quality prairie. These
two species partially support my hypothesis that higher quality prairie will have higher
establishment. The null hypothesis would be that there is no difference in the mean counts of
plants in any relative prairie quality, thus the count of plants found in any of site conditions would
not be different from another. One of the four species, Eriophyllum lanatum, had the highest
mean count of plants in medium quality. This species does not necessarily support my hypothesis
that the higher quality prairie would have higher first year establishment rates. The cause of this
phenomenon may be partially due to the protocol for counting this species. Since E. lanatum is
rhizomatous, stems were counted as a proxy for plants, while the actual genetically unique
number of plants is likely much lower than represented in this research, though it is difficult to
tell for sure without destructive sampling techniques. The fourth and final species that responded
to relative prairie quality, Ranunculus occidentalis, performed well in both high and low quality

74

prairies but less so in medium quality prairie. This species does not necessarily support my
hypothesis that higher quality prairie would result in higher establishment rates. The causes of
this phenomenon are unknown, but perhaps that microsite conditions were not favorable, or some
influence of pre-existing plants/seeds influence this result. In the future, preliminary site
monitoring could help de-tangle all of the background noise caused by pre-existing species,
though it would still be difficult to know that this common plant was not present in the seedbank.
Generally speaking, the species that responded to site conditions seemed to prefer better
site conditions (medium or high quality) rather than poor site conditions (low quality). No species
responded solely to relatively low quality prairie, which is unfortunate for the restoration of badly
degraded sites. Three perennial forbs (Achillea millefolium, Eriophyllum lanatum, and
Ranunculus occidentalis) and one perennial grass (Danthonia californica) responded to site
conditions. However, it is difficult to attribute these trends entirely to causation and not
correlation. Was the site considered high quality because these species were present there, or
were the seeded species actually responding to some element of the high quality? Further research
could better answer this question.
Some of the five species that were excluded from the generalized regression analysis due
to low abundances showed patterns of response to relative prairie quality. I had hypothesized that
higher quality prairie would result in higher establishment rates. One species, Sisyrinchium
idahoense was only found in high quality prairie, which supports this hypothesis. Two other
species, Lomatium utriculatum and Potentilla gracilis, show partial support for this hypothesis, as
they were found in only high and medium quality prairie. It is unclear if the results for these three
species were due to the seeding treatments or they were already present prior to this experiment.
Regardless, these three species either show preference for high quality prairie, or they are more
persistent in higher quality sites. Another species, Balsamorhiza deltoidea, showed a small
amount of support for my hypothesis: of eight individual seedlings found, five of them were in
medium quality prairie, two in high quality prairie, and one in low quality prairie. While this is

75

much too small of a sample size to say anything definitively about the condition preferences for
Balsamorhiza deltoidea, continued monitoring of these long-lived individuals could shed some
light on the patterns. Another species, Clarkia amoena, supported the null hypothesis. Of six
individual plants found, five were in low quality, one in medium quality, and zero in high quality.
A larger sample size is need to determine if this is an actual effect, or due to random chance.
More research is need on the species that were found in low abundances in order to determine if
they truly support the hypotheses.

Suggestions for future research
Some of the species used in this research are common in SPS prairies, or are patchily
distributed. This makes interpretation of results difficult, especially when species are found in the
control plots or in higher abundances that are possible due to seeding rate. Increasing the seeding
rate and selecting sites with few native species could be potential strategies for separating
causation from correlation. Alternatively, monitoring the plot areas a year prior to sowing,
selecting species that don’t occur in or nearby the plot area, and adding more replicates could be
ways to improve an experiment similar to this one. In addition, increasing the monitoring effort to
include more samples, or monitoring at several times throughout the growing season could help
with identification of seedlings, and identifying any species that sprout and then die/senesce early
in the season. Removal of dominant grasses has been shown to increase forb productivity, cover,
and species richness (McCain, Baer, Blair, & Wilson, 2010). Use of grass-specific herbicide to
control non-native grasses should be considered for future experiments.
Native species may have unique germination and establishment limitations that were not
examined here. These limitations could explain low abundance or complete absence of some
species. Seed predation is another factor that this experiment did not take into consideration,
which could be a very important variable. For example, rodent predation of Viola adunca seeds
impacted production at native seed nurseries (Angela Winter, CNLM, personal communications,

76

May 2016), though impacts of rodent predation in the field is unknown. Greenhouse studies with
these or similar species could be useful to tease apart these variables, though controlled field
studies may be more immediately practicable.
Future research on these plots could help us understand if first-year establishment of
these species results in successful prairie restoration. Additionally, future research could shed
light on the persistence of multiple generations of annual species. Many annual species have been
largely extirpated from the SPS prairies (Dunwiddie et al., 2014), so understanding where and
how to re-introduce them should be a high priority of restoration in this area. Presumably, the
seed addition in this experiment could have contributed to the soil seedbank, which could
increase establishment in subsequent years, under the right conditions. For instance the seeds
sown in March could positively respond after a year of natural stratification, and establish in the
spring of the second year.

CONCLUSION
Of the four species that responded positively to sowing time, three support the idea of
priority effects, and therefore, the theory of neutral assemblage. These three species are Collinsia
spp., Lupinus albicaulis, and Plectritis congesta. For these three species, the earlier sowing times
result in higher establishment rates. On the other hand, three of the four species that responded
positively to relative prairie quality support the idea of ecological filtering. These three species
are Achillea millefolium, Danthonia californica, and Ranunculus occidentalis. For these three
species the higher quality biotic and abiotic conditions of site influenced the establishment rates.
Other species that responded positively to site conditions support the idea of ecological filtering,
and therefore the theory of niche assemblage. While six species showed some amount of support
for either neutral or niche theories, no species supported both theories. The remaining sixteen
species did not support these theories, had no measurable response, were found in too low of
numbers, or were not found at all. Therefore, it is difficult to draw conclusions about the relative
influence of neutral and niche processes for the native plant community as a whole. It is

77

interesting that only one species (Lupinus albicaulis) initially showed support for both theories,
but upon further investigation only supported priority effects theory. Intuitively, one would
expect that both neutral and niche theory are both acting upon community assembly and are not
mutually exclusive. However, this study’s results indicate that these two theories do not overlap
as much as previously assumed, at least for the few species tested. Further testing of the relative
importance of these theories in restoration ecology is warranted.
This research demonstrates some of the challenges in ecological studies, especially those
due to variability in species responses. There is still a lot to learn about restoration, and how to
harness natural processes to help meet restoration goals. It remains difficult to get native prairie
plants to establish through direct seeding. While a cost-benefit analysis is outside the scope of this
project, future studies could investigate the long-term cost of direct seed sowing, plug planting, or
other methods. This research suggests that SPS prairies are native seed limited, corroborating the
findings of Stanley et al. (2011). Adding seeds of native species increased the diversity in all
three relative qualities of prairie, relative to the unseeded control plots. Interestingly, even the
Upper Weir site, which was the highest quality tested in this study, showed an increase in
diversity after seed sowing. Other studies have found that the most species rich locations are more
resistant to invasion (Naeem et al., 2000; Tilman, 1997; Tilman et al., 2014), thus seed addition of
native or non-native species to a species-rich site should not show an increase in diversity. On a
positive note, many species were found in the control plots, suggesting higher than anticipated
native richness, despite obvious non-native cover.
A one size fits all approach to restoration is more likely to fail than a carefully planned
restoration project that is tailored to fit the site conditions. That is to say, sowing seed at the right
time in the right place will have a positive effect on the species. It is difficult to pinpoint the
optimal time and conditions for each unique species. In total, there were over 13,000 individual
plants counted in this research. It is difficult to unpack all of the patterns of 23 different species
amongst the complexity of the natural world. There is still a lot to be learned about how to do

78

prairie restoration well, and how to do it more efficiently. This research demonstrates that, among
other things, first year establishment rates of native prairie plants are typically very low, and that
there is a lot of room for improvement. However, the patterns seen here for several species can
provide guidance for managers when deciding when and where to sow particular species to meet
restoration goals. This research also attempts to bridge the gap between practical and theoretical
research in restoration ecology. These results indicate a few species that support either niche
theory or neutral theory, but not necessarily both theories, though more research is needed. There
is a synergistic relationship between practical and theoretical fields of study that could lead to
important scientific discoveries in the future.

PRACTICAL RECOMMENDATIONS
1. Sow seeds of native prairie species (especially annuals forbs) as soon as possible
after sites have been prepared in the summer or fall. More closely aligning restoration
sowing with natural seed dispersal times could improve native plant establishment.
Priority effects can be used to favor native species, and help them compete with nonnative species. Four out of five tested annual forb species responded positively to
sowing time, so for these species sowing time is especially important. One perennial
forb, Lupinus albicaulis, also responded positively to earlier sowing times. If sowing
seeds is not logistically possible until late fall or winter, consider storing seeds until
the following summer or fall, and then seeding into newly prepared sites.
2. Sow species in sites where they are known to occur in low abundance, or were
historically known to occur. Persistence of a species in a given site suggests that the
biotic and abiotic site conditions are favorable for growth and persistence. The
relative prairie quality had a significant influence on three perennial forbs and one
perennial grass, though it is difficult to determine if the seed addition was the true
cause of these species’ establishment.

79

3. Try sowing common species, or less expensive seeds, in low quality sites, and/or be
prepared for more intensive follow up work after the initial seeding. Consider using
herbicides to help control non-native pasture grasses and other invasive species, and
mitigate low establishment rates by increasing the seeding rates, if possible.
4. Sow rare species, which are typically more expensive, in medium and high quality
sites only. Consider targeting areas of bare ground or lightly disturbed soil to
improve seed-soil contact, which could help increase establishment rates. Even the
relatively high quality prairie showed an increase in overall species diversity after fall
sowing, suggesting that there is seed limitation present in relatively healthy prairie.
5. This research suggests sowing seeds at a higher rate than currently prescribed is
warranted, though it is difficult to predict long-term restoration success from first
year results. Generally speaking, over half of the species used in this study had 0-5%
establishment rates. Consider using both plugs and direct seeding in restoration
efforts. Repeated monitoring of this and other restoration projects will shed some
light on the long-term success of restoration using direct seeding techniques.

80

Bibliography
Agee, J. K. (1996). Achieving conservation biology objectives with fire in the Pacific Northwest.
Weed Technology, 10(2), 417–421.
Amoroso, N., & Chalcraft, D. R. (2015). Duration of colonization and interactions between early
and late colonists determine the effects of patch colonization history on patch biodiversity.
Oikos, 124(10), 1317–1326. http://doi.org/10.1111/oik.01922
Bakker, J. D., Colasurdo, L. B., & Evans, J. R. (2012). Enhancing Garry oak seedling
performance in a semiarid environment. Northwest Science, 86(4), 300–309.
http://doi.org/10.3955/046.086.0406
Bakker, J. D., Wilson, S. D., Christian, J. M., Li, X., Ambrose, L. G., & Waddington, J. (2003).
Contingency of grassland restoration on year, site, and competition from introduced grasses.
Ecological Applications, 13(1), 137–153. http://doi.org/10.1890/10510761(2003)013[0137:COGROY]2.0.CO;2
Baskin, C. C. & Baskin, J. M. (1998). Seeds: ecology biogeography, and evolution of dormancy
and germination. 1st Edition. San Diego (CA): Academic Press. 666p.
Bolker, B. M., Brooks, M. E., Clark, C. J., Geange, S. W., Poulsen, J. R., Stevens, M. H. H., &
White, J. S. (2008). Generalized linear mixed models: a practical guide for ecology and
evolution, (Table 1). http://doi.org/10.1016/j.tree.2008.10.008
Brye, K. R., Norman, J. M., & Gower, S. T. (2002). Assessing the progress of a tallgrass prairie
restoration in Southern Wisconsin. The American Midland Naturalist, 148(2), 218–235.
http://doi.org/10.1674/0003-0031(2002)148[0218:ATPOAT]2.0.CO;2
Bullock, J. M., Pywell, R. F., & Walker, K. J. (2007). Long-term enhancement of agricultural
production by restoration of biodiversity. Journal of Applied Ecology, 44(1), 6–12.
http://doi.org/10.1111/j.1365-2664.2006.01252.x
Burton, C. M., Burton, P. J., Hebda, R., & Turner, N. J. (2006). Determining the optimal sowing
density for a mixture of native plants used to revegetate degraded ecosystems. Restoration
Ecology, 14(3), 379–390. http://doi.org/10.1111/j.1526-100X.2006.00146.x
Cairns, J., & Heckman, J. R. (1996). Restoration ecology: the state of an emerging field. Annual
Review of Energy and the Environment, 21(1), 167–189.
http://doi.org/10.1146/annurev.energy.21.1.167
Carter, D. L., & Blair, J.M. (2011). Recovery of native plant community characteristics on a
chronosequence of restored prairies seeded into pasture in west-central Iowa. Restoration
Ecology, 20(2), 170 – 179.
Chase, J. M. (2003). Community assembly: When should history matter? Oecologia, 136(4),
489–498. http://doi.org/10.1007/s00442-003-1311-7
Chase, J. M. (2007). Drought mediates the importance of stochastic community assembly.
Proceedings of the National Academy of Sciences of the United States of America, 104(44),
17430–17434. http://doi.org/10.1073/pnas.0704350104
Chase, J. M., & Leibold, M. A. (2003). Ecological niches: Linking classical and contemporary
approaches. Chicago, IL: University of Chicago Press.
Clark, A. C. J., Poulsen, J. R., Levey, D. J., & Osenberg, C. W. (2007). Are plant populations
seed limited? A critique and meta-analysis of seed addition experiments. The American
Midland Naturalist, 170(1), 128–142. http://doi.org/10.1086/518565
Clark, D. L., & Wilson, M. V. (2003). Post-dispersal seed fates of four prairie species. American
Journal of Botany, 90(5), 730–735.
Clements, F. E. (1916). Plant Succession. An Analysis of the Development of Vegetation.
Carnegie Institutue of Washington. http://doi.org/10.1126/science.45.1162.339
Connolly, S. R., MacNeil, M. A., Caley, M. J., Knowlton, N., Cripps, E., Hisano, M., Wilson, R.
S. (2014). Commonness and rarity in the marine biosphere. Proceedings of the National
Academy of Sciences, 111(23), 8524–8529. http://doi.org/10.1073/pnas.1406664111
Delvin, E. G. (2013). Restoring abandoned agricultural lands in Puget lowland prairies: a new

81

approach, 150.
Déri, E., Magura, T., Horváth, R., Kisfali, M., Ruff, G., Lengyel, S., & Tóthmérész, B. (2011).
Measuring the short-term success of grassland restoration: the use of habitat affinity indices
in ecological restoration. Restoration Ecology, 19(4), 520–528.
http://doi.org/10.1111/j.1526-100X.2009.00631.x
Diamond, J.M. (1975). Assembly of Species Communities. In M.L. Cody and J.M. Diamond
(Eds.) Ecology and Evolution of Communities (342-444). Cambridge, MA: Harvard
University Press.
Dickson, T. L., Hopwood, J. L., & Wilsey, B. J. (2012). Do priority effects benefit invasive plants
more than native plants? An experiment with six grassland species. Biological Invasions,
14(12), 2617–2624. http://doi.org/10.1007/s10530-012-0257-2
Doll, J. E., Brink, G. E., Cates, R. L. J., & Jackson, R. D. (2009). Effects of native grass
restoration management on above- and belowground pasture production and forage quality.
Journal of Sustainable Agriculture, 33(5), 512–527.
http://doi.org/10.1080/10440040902997702
Doll, J. E., Haubensak, K. A., Bouressa, E. L., & Jackson, R. D. (2011). Testing disturbance,
seeding time, and soil amendments for establishing native warm-season grasses in nonnative cool-season pasture. Restoration Ecology, 19(101), 1–8.
http://doi.org/10.1111/j.1526-100X.2010.00687.x
Drake, J. A. (1991). Community-Assembly Mechanics and the Structure of an Experimental
Species Ensemble. The American Naturalist, 137(1), 1–26.
Drake, D., Ewing, K., & Dunn, P. (1998). Techniques to promote germination of seed from Puget
Sound prairies. Ecological Restoration, 16(1), 33–40.
Dunwiddie, P., Alverson, E., Martin, R., & Gilbert, R. (2014). Annual species in native prairies of
South Puget Sound, Washington. Northwest Science, 88(2), 94–105.
http://doi.org/10.3955/046.088.0205
Dunwiddie, P. W., & Bakker, J. D. (2011). The future of restoration and management of prairieoak ecosystems in the Pacific Northwest. Northwest Science, 85(2), 83–92.
http://doi.org/10.3955/046.085.0201
Dunwiddie, P. W., & Martin, R. A. (2016). Microsites matter: Improving the success of rare
species reintroductions. Plos One, 11(3), e0150417.
http://doi.org/10.1371/journal.pone.0150417
Ejrnaes, R., Bruun, H. H., & Graae, B. J. (2006). Community assembly in experimental
grasslands: suitable environment or timely arrival? Ecology, 87(5), 1225–1233.
Elliott, C. W., Fischer, D. G., & LeRoy, C. J. (2011). Germination of three native Lupinus species
un response to temperature. Field and Ecosystem Ecology, 85(2), 403–410.
http://doi.org/10.3955/046.085.0223
Eriksson, O. & Ehrlén, J. (2008). Seedling recruitment and population ecology. In: Leck, M. A.,
Parker, V. T., & Simpson, R. L. (eds.) Seedling Ecology and Evolution. pp. 239-254.
[Online]. Cambridge: Cambridge University Press. Available from: Cambridge Books
Online <http://dx.doi.org/10.1017/CBO9780511815133.013>
Ewing, K. (2002). Effects of initial site treatments on early growth and three-year survival of
Idaho fescue. Restoration Ecology, 10(2), 282–288. http://doi.org/10.1046/j.1526100X.2002.02039.x
Fazzino, L., Kirkpatrick, H. E., & Fimbel, C. (2011). Comparison of hand-pollinated and
naturally-pollinated Puget balsamroot (Balsamorhiza deltoidea Nutt.) to determine
pollinator limitations on South Puget Sound lowland prairies. Northwest Science, 85(2),
352–360. http://doi.org/10.3955/046.085.0220
Firn, J., MacDougall, A. S., Schmidt, S., & Buckley, Y. M. (2010). Early emergence and resource
availability can competitively favour natives over a functionally similar invader. Oecologia,
163(3), 775–784. http://doi.org/10.1007/s00442-010-1583-7

82

Foster, B. L., & Tilman, D. (2003). Seed limitation and the regulation of community structure in
oak savanna grassland. Journal of Ecology, 91, 999–1007. http://doi.org/10.1046/j.13652745.2003.00830.x
Fowler, N. L. (1986). Microsite requirements for germination and establishment of three grass
species. The American Midland Naturalist, 115(1), 131–145.
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(2), 188–193. http://doi.org/10.1111/j.1526-100X.2010.00770.x
Fukami, T., Martijn Bezemer, T., Mortimer, S. R., & Putten, W. H. (2005). Species divergence
and trait convergence in experimental plant community assembly. Ecology Letters, 8(12),
1283–1290. http://doi.org/10.1111/j.1461-0248.2005.00829.x
Gleason, H. A. (1927). Further Views on the Succession. Ecology, 8(3), 299–326.
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(2), 154–167.
http://doi.org/10.1353/ecr.2013.0026
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(5), 664–
670. http://doi.org/10.1111/j.1526-100X.2008.00497.x
Hamman, S. T., Dunwiddie, P. W., Nuckols, J. L., & McKinley, M. (2011). Fire as a restoration
tool in Pacific Northwest prairies and oak woodlands: Challenges, successes, and future
directions. Northwest Science, 85(2), 317–328. http://doi.org/10.3955/046.085.0218
Hamman, S. T., Smith, S. & Bakker, J. (2015). Final Report for Center for Natural Lands
Management Budget G1005. Prepared for the U.S. Fish and Wildlife Service. Regional
prairie native seed project. Olympia, WA.
Hedberg, P., & Kotowski, W. (2010). New nature by sowing? The current state of species
introduction in grassland restoration, and the road ahead. Journal for Nature Conservation,
18(4), 304–308. http://doi.org/10.1016/j.jnc.2010.01.003
Helsen, K., Hermy, M., & Honnay, O. (2012). Trait but not species convergence during plant
community assembly in restored semi-natural grasslands. Oikos, 121(12), 2121–2130.
http://doi.org/10.1111/j.1600-0706.2012.20499.x
Henry, M., Stevens, H., Bunker, D. E., Schnitzer, S. a., & Carson, W. P. (2004). Establishment
limitation reduces species recruitment and species richness as soil resources rise. Journal of
Ecology, 92(2), 339–347. http://doi.org/10.1111/j.0022-0477.2004.00866.x
Hitchmough, J., de la Fleur, M., & Findlay, C. (2004). Establishing North American prairie
vegetation in urban parks in northern England. Landscape and Urban Planning, 66(2), 75–
90. http://doi.org/10.1016/S0169-2046(03)00096-3
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(1), 23–29.
http://doi.org/10.1111/j.1461-0248.2004.00686.x
Hoelzle, T. B., Jonas, J. L., & Paschke, M. W. (2012). Twenty-five years of sagebrush steppe
plant community development following seed addition. Journal of Applied Ecology, 49(4),
911–918. http://doi.org/10.1111/j.1365-2664.2012.02154.x
Holl, K. D., Hayes, G. F., Brunet, C., Howard, E. A., Reed, L. K., Tang, M., & Vasey, M. C.
(2014). Constraints on direct seeding of coastal prairie species: Lessons learned from
restoration. Grasslands, 24, 8 – 12.
Hooper, D. U., Chapin, E. S., Ewel, J. J., Hector, A., Inchausti, P., Lavorel, S., & Lawton, J. H.
(2005). Effects of biodiversity on ecosystem functioning: a consensus of current knowledge.
Ecological Monographs, 75(1), 3–35. http://doi.org/10.1890/04-0922
Hooper, D. U., & Dukes, J. S. (2010). Functional composition controls invasion success in a

83

California serpentine grassland. Journal of Ecology, 98(4), 764–777.
http://doi.org/10.1111/j.1365-2745.2010.01673.x
Howe, H. F., & Brown, J. S. (2000). Early effects of rodent granivory on experimental forb
communities. Ecological Applications, 10(3), 917–924.
Hubbell, S. P. (2001). The unified neutral theory of biodiversity and biogeography. Princeton, NJ:
Princeton University Press.
Justice, O. L., & Bass L. N., Science and Education Administration (1978). Principles and
practices of seed storage. (U.S. Department of Agriculture Handbook No. 506). Washington
DC: U.S. Government Printing Office.
Kirkpatrick, H. E., & Lubetkin, K. C. (2011). Responses of native and introduced plant species to
sucrose addition in Puget lowland prairies. Northwest Science, 85(2), 255–268.
http://doi.org/10.3955/046.085.0214
Knappová, J., Knapp, M., & Münzbergová, Z. (2013). Spatio-temporal variation in contrasting
effects of resident vegetation on establishment, growth and reproduction of dry grassland
plants: Implications for seed addition experiments. PLoS ONE, 8(6), e65879.
http://doi.org/10.1371/journal.pone.0065879
Körner, C., Stöcklin, J., Reuther-Thiébaud, L., & Pelaez-Riedl, S. (2008). Small differences in
arrival time influence composition and productivity of plant communities. New Phytologist,
177(3), 698–705. http://doi.org/10.1111/j.1469-8137.2007.02287.x
Krauss, J., Steffan-Dewenter, I., & Tscharntke, T. (2003). Local species immigration, extinction,
and turnover of butterflies in relation to habitat area and habitat isolation. Oecologia,
137(4), 591–602. http://doi.org/10.1007/s00442-003-1353-x
Krock, S., Smith, S., Elliot, C., Kennedy, A., Hamman, S. (2016). Using smoke-water and coldmoist stratification to improve germination of native prairie species. Native Plants Journal
17(1), 19-27.
LandScope America. (2015). Willamette Valley-Puget Trough-Georgia Basin Ecoregion.
http://www.landscope.org/explore/natural_geographies/ecoregions/Puget%20Trough%20%20Willamette%20Valley%20-%20Georgia%20Basin/ accessed 12/27/15.
Larson, D. L., Bright, J., Drobney, P., Larson, J. L., Palaia, N., Rabie, P. A., Wells, D. (2011).
Effects of planting method and seed mix richness on the early stages of tallgrass prairie
restoration. Biological Conservation, 144(12), 3127–3139.
http://doi.org/10.1016/j.biocon.2011.10.018
Lawson, C. S., Ford, M. a, & Mitchley, J. (2004). The influence of seed addition and cutting
regime on the success of grassland restoration on former arable land. Applied Vegetation
Science, 7(2), pp. 259–266. http://doi.org/10.1111/j.1654-109X.2004.tb00618.x
Lewis Army Museum (2014). World War II, 1939-1949. Joint Base Lewis-McChord, DPTMS.
http://www.lewis-mcchord.army.mil/dptms/museum/wwii.htm. (Accessed 6/4/2016).
Maret, M. P., & Wilson, M. V. (2000). Fire and seedling population dynamics in western Oregon
prairies. Journal of Vegetation Science, 11(2), 307–314. http://doi.org/10.2307/3236811
Maret, M. P., & Wilson, M. V. (2005). Fire and litter effects on seedling establishment in western
Oregon upland prairies. Restoration Ecology, 13(3), 562–568. http://doi.org/10.1111/j.1526100X.2005.00071.x
Maron, J. L., & Gardner, S. N. (2000). Consumer pressure, seed versus safe-site limitation, and
plant population dynamics. Oecologia, 124, 260–269.
http://doi.org/10.1007/s004420000382
Martin, L. M., & Wilsey, B. J. (2006). Assessing grassland restoration success: relative roles of
seed additions and native ungulate activities. Journal of Applied Ecology, 43(6), 1098–1109.
http://doi.org/10.1111/j.1365-2664.2006.01211.x
Martin, L. M., & Wilsey, B. J. (2012). Assembly history alters alpha and beta diversity, exoticnative proportions and functioning of restored prairie plant communities. Journal of Applied
Ecology, 49(6), 1436–1445. http://doi.org/10.1111/j.1365-2664.2012.02202.x

84

Martin, L. M., & Wilsey, B. J. (2014). Native-species seed additions do not shift restored prairie
plant communities from exotic to native states. Basic and Applied Ecology, 15(4), 297–304.
http://doi.org/10.1016/j.baae.2014.05.007
McCain, K. N. S., Baer, S. G., Blair, J. M., & Wilson, G. W. T. (2010). Dominant grasses
suppress local diversity in restored tallgrass prairie. Restoration Ecology, 18(SUPPL. 1),
40–49. http://doi.org/10.1111/j.1526-100X.2010.00669.x
Moles, A. T., & Westoby, M. (2004). What do seedlings die from and what are the implications
for evolution of seed size? Oikos, 106(1), 193–199.
Naeem, S., Knops, J. M., Tilman, D., Howe, K. M., Kennedy, T., & Gale, S. (2000). Plant
diversity increases resistance to invasion in the absence of covarying extrinsic factors.
Oikos, 91(1), 97–108. http://doi.org/10.1034/j.1600-0706.2000.910108.x
Nathan, R., & Muller-Landau, H. C. (2000). Spatial patterns of seed dispersal, their determinants
and consequences for recruitment. Trends in Ecology & Evolution, 15(7), 278–285.
http://doi.org/10.1016/S0169-5347(00)01874-7
Native Plant Network, Reforestation, Nurseries, and Genetic Resources Propagation Protocol
Database. (2016). http://npn.rngr.net/npn/propagation accessed 1/11/2016.
Natural Resources Conservation Service. (2015). Web Soil Survey.
http://websoilsurvey.sc.egov.usda.gov/App/WebSoilSurvey.aspx accessed 12/27/2015.
Öster, M., Ask, K., Cousins, S. A. O., & Eriksson, O. (2009). Dispersal and establishment
limitation reduces the potential for successful restoration of semi-natural grassland
communities on former arable fields. Journal of Applied Ecology, 1266–1274.
http://doi.org/10.1111/j.1365-2664.2009.01721.x
Pfeifer-Meister, L., Roy, B. a., Johnson, B. R., Krueger, J., & Bridgham, S. D. (2012).
Dominance of native grasses leads to community convergence in wetland restoration. Plant
Ecology, 213(4), 637–647. http://doi.org/10.1007/s11258-012-0028-2
Pielou, E. C. (1991). After the ice age: The return of life to glaciated North America. Chicago, IL:
University of Chicago Press.
Plückers, C., Rascher, U., Scharr, H., von Gillhaussen, P., Beierkuhnlein, C., & Temperton, V. M.
(2013). Sowing different mixtures in dry acidic grassland produced priority effects of
varying strength. Acta Oecologica, 53, 110–116. http://doi.org/10.1016/j.actao.2013.09.004
Potts, J. M., & Elith, J. (2006). Comparing species abundance models. Ecological Modelling,
199(2), 153–163. http://doi.org/10.1016/j.ecolmodel.2006.05.025
Rinella, M. J., Mangold, J. M., Espeland, E. K., Sheley, R. L., & Jacobs, J. S. (2012). Long-term
population dynamics f seeded plants in invaded grasslands. Ecological Applications, 22(4),
1320–1329.
Robinson, J. V, & Edgemon, M. A. (1988). An experimental evaluation of the effect of invasion
history on community structure. Ecology, 69(5), 1410–1417. http://doi.org/10.2307/1941638
Ross, D. a., & Harper, J. L. (1972). Occupation of biological space during seedling establishment
and growth. Ecology, 60(1), 77–88. http://doi.org/10.2307/2258041
Rowe, H. I. (2010). Tricks of the Trade: Techniques and Opinions from 38 Experts in Tallgrass
Prairie Restoration. Restoration Ecology, 18(November), 253–262.
http://doi.org/10.1111/j.1526-100X.2010.00663.x
Russell, M. (2011). Dormancy and Germination Pre-Treatments in Willamette Valley Native
Plants. Northwest Science, 85(2), 389–402. http://doi.org/10.3955/046.085.0222
Sayuti, Z., & Hitchmough, J. D. (2013). Effect of sowing time on field emergence and growth of
South African grassland species. South African Journal of Botany, 88, 28–35.
http://doi.org/10.1016/j.sajb.2013.04.008
Schantz, M. C., Sheley, R. L., & James, J. J. (2014). Role of propagule pressure and priority
effects on seedlings during invasion and restoration of shrub-steppe. Biological Invasions,
17, 73–85. http://doi.org/10.1007/s10530-014-0705-2
Sinclair, M., Alverson, E., Dunn, P., Dunwiddie, P., & Gray, E. (2006). Bunchgrass prairies. In

85

Apostol, D. & Sinclair, M. (Eds.), Restoring the Pacific Northwest: the art and science of
ecological restoration in Cascadia (29-59).Washington DC: Island Press.
Sluis, W. J. (2002). Patterns of species richness and composition in re-created grassland.
Restoration Ecology, 10(4), 677–684. http://doi.org/10.1046/j.1526-100X.2002.01048.x
Smith, S. & Elliot, C. (2015). South Sound prairie conservation nursery 2014 annual report.
Center for Natural Lands Management and Sustainability in Prisons Project. Olympia, WA.
Spellerberg, I. F., & Fedor, P. J. (2003). A tribute to Claude-Shannon (1916-2001) and a plea for
more rigorous use of species richness, species diversity and the “Shannon-Wiener” Index.
Global Ecology and Biogeography, 12(3), 177–179. http://doi.org/10.1046/j.1466822X.2003.00015.x
Stanley, A. G., Dunwiddie, P. W., & Kaye, T. N. (2011). Restoring invaded Pacific Northwest
prairies: management recommendations from a region-wide experiment. Northwest Science,
85(2), 233–246. http://doi.org/10.3955/046.085.0212
Stanley, A. G., Kaye, T. N., & Dunwiddie, P. W. (2008). Regional strategies for restoring
invaded prairies: Observations from a multisite, collaborative research project. Native
Plants Journal, 9(3), 247–254. http://doi.org/10.2979/NPJ.2008.9.3.247
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. http://doi.org/10.1111/j.1526-100X.1995.tb00095.x
Sullivan, A. T., & Howe, H. F. (2009). Prairie forb response to timing of vole herbivory. Ecology,
90(5), 1346–1355.
Summerville, K. S. (2008). Species diversity and persistence in restored and remnant tallgrass
prairies of North America: a function of species’ life history, habitat type, or sampling bias?
Journal of Animal Ecology, 77(3), 487–494. http://doi.org/10.1111/j.13652656.2008.01356.x
Tilman, D. (1997). Community invasibility, recruitment limitation, and grassland biodiversity.
Ecology, 78(1), 81–92.
Tilman, D., Isbell, F., & Cowles, J. M. (2014). Biodiversity and ecosystem functioning. Annual
Review of Ecology, Evolution, and Systematics, 45(1), 471–493.
http://doi.org/10.1146/annurev-ecolsys-120213-091917
Turnbull, L. a, Crawley, M. J., & Rees, M. (2000). Are plant populations seed-limited? A review
of seed sowing experiments. Oikos, 88(August 1999), 225–238.
http://doi.org/10.2307/3547018
U.S. Fish and Wildlife Service. 2015. Listed species believed to or known to occur in Washington
(updated 2/13/2015). Environmental Conservation Online System, Conserving the Nature of
America. http://ecos.fws.gov/tess_public/reports/species-listed-by-state-report?state=WA
(accessed 10/24/2015)
USDA, NRCS. (2016). The PLANTS Database (http://plants.usda.gov). National Plant Data
Team, Greensboro, NC 27401-4901 USA. (Accessed 5/19/2016).
Vannette, R. L., & Fukami, T. (2014). Historical contingency in species interactions: Towards
niche-based predictions. Ecology Letters, 17(1), 115–124. http://doi.org/10.1111/ele.12204
Vaughn, K. J., & Young, T. P. (2010). Contingent conclusions: Year of initiation influences
ecological field experiments, but temporal replication is rare. Restoration Ecology,
18(SUPPL. 1), 59–64. http://doi.org/10.1111/j.1526-100X.2010.00714.x
von Gillhaussen, P., Rascher, U., Jablonowski, N. D., Plückers, C., Beierkuhnlein, C., &
Temperton, V. M. (2014). Priority effects of time of arrival of plant functional groups
override sowing interval or density effects: A grassland experiment. PLoS ONE, 9(1),
e86906. http://doi.org/10.1371/journal.pone.0086906
Walsh, M. K., Pearl, C. A., Whitlock, C., Bartlein, P. J., & Worona, M. A. (2010). An 11,000 year-long record of fire and vegetation history at Beaver Lake, Oregon, central Willamette
Valley. Quaternary Science Reviews, 29, 1093–1106.

86

Weiher, E., Clarke, P., & Keddy, P. A. (1998). Community assembly rules, morphological
dispersion, of plant species the coexistence. Oikos, 81(2), 309–322.
Weiher, E., Freund, D., Bunton, T., Stefanski, A., Lee, T., & Bentivenga, S. (2011). Advances,
challenges and a developing synthesis of ecological community assembly theory.
Philosophical Transactions of the Royal Society B: Biological Sciences, 366(1576), 2403–
2413. http://doi.org/10.1098/rstb.2011.0056
Weiser, A., & Lepofsky, D. (2009). Ancient land use and management of Ebey’s Prairie,
Whidbey Island, Washington. Journal of Ethnobiology, 29(2), 184–212.
http://doi.org/10.2993/0278-0771-29.2.184
Westoby, M., Walker, B., & Noy-Meir, N. (1989). Opportunistic management for rangelands not
at equilibrium. Journal of Range Management, 42(4), 266–274.
http://doi.org/10.2307/3899492
Wilbur, H. M., & Alford, R. A. (1985). Priority effects in experimental pond communities:
Responses of Hyla to Bufo and Rana. Ecology, 66(4), 1106–1114.
World Health Organization. (2005). Ecosystems and human well-being. Ecosystems, 5(281), 1–
100. http://doi.org/10.1196/annals.1439.003
Zavaleta, E. S., Pasari, J. R., Hulvey, K. B., & Tilman, G. D. (2010). Sustaining multiple
ecosystem functions in grassland communities requires higher biodiversity. Proceedings of
the National Academy of Sciences of the United States of America, 107(4), 1443–1446.
http://doi.org/10.1073/pnas.0906829107
Zeiter, M., Stampfli, A., & Newbery, D. M. (2006). Recruitment limitation constrains local
species richness and productivity in dry grassland. Ecology, 87(4), 942–51.
http://doi.org/10.1890/0012-9658(2006)87[942:RLCLSR]2.0.CO;2
Zobel, M., Otsus, M., Liira, J., Moora, M., & Mols, T. (2000). Is small-scale species richness
limited by seed availability or microsite availability? Ecology, 81(12), 3274.
http://doi.org/10.2307/177492
Zuur, A. F., Ieno, E. N., Walker, N. J., Saveliev, A. A., & Smith, G. M. (2009). Mixed Effects
Models and Extensions with R. New York, NY: Springer.

87

88

Media

Krock_SMESthesis2016.pdf