The Evergreen State College Institutional Repository

Taylor's Checkerspot (EUPHYDRYAS EDITHA TAYLORI) Oviposition Habitat Selection and Larval Hostplant Use in Washington State

Item

Title (dcterms:title)

Eng Taylor's Checkerspot (EUPHYDRYAS EDITHA TAYLORI) Oviposition Habitat Selection and Larval Hostplant Use in Washington State

Date (dcterms:date)

2011

Creator (dcterms:creator)

Eng Grosboll, Daniel N

Subject (dcterms:subject)

Eng Environmental Studies

extracted text (extracttext:extracted_text)

TAYLOR'S CHECKERSPOT (EUPHYDRYAS EDITHA TAYLORI)
OVIPOSITION HABITAT SELECTION AND LARVAL HOSTPLANT USE IN
WASHINGTON STATE

By
Daniel Nelson Grosboll

A Thesis
Submitted in partial fulfillment
of the requirements for the degree
Master of Environmental Studies
The Evergreen State College
June 2011

© 2011 by Daniel Nelson Grosboll. All rights reserved.

This Thesis for the Master of Environmental Study Degree
by
Daniel Nelson Grosboll

has been approved for
The Evergreen State College
by

________________________
Judy Cushing, Ph.D.
Member of the Faculty

______________
Date

Abstract
Taylor's checkerspot (Euphydryas editha taylori) oviposition habitat
selection and larval hostplant use in Washington State
Taylor’s checkerspot (Euphydryas editha taylori (W.H. Edwards 1888)), a
Federal Endangered Species Act candidate species, is found in remnant colonies
between extreme southwestern British Columbia and the southern Willamette
Valley in Oregon. This butterfly and its habitat have declined precipitously largely
due to anthropogenic impacts. However, this butterfly appears to benefit from
some land management activities and some populations are dependent on an
exotic hostplant.
Oviposition sites determine what resources are available for larvae after
they hatch. Larval survival and growth on three reported hostplants (Castilleja
hispida, Plantago lanceolata, and P. major) were measured in captivity to
determine the suitability of hostplant species and to develop captive rearing
methods. Larvae successfully developed on C. hispida and P. lanceolata.
Parameters of oviposition sites were measured within occupied habitat at
four sites in Western Washington. Sampling occurred at two spatial scales with
either complete site censuses or stratified systematic sampling on larger sites.
Within the sampled or censused areas, oviposition sites were randomly selected
for paired oviposition/adjacent non-oviposition microhabitat measurements.
Taylor’s checkerspot oviposited on all reported perennial hosts available.
Oviposition was significantly associated with high hostplant density at three of
four sites. All but one of 31 oviposition locations selected for microhabitat
measurement contained >10,000cm3 hostplant volume within 1m2 surrounding
oviposition sites. Other habitat factors examined were suggestive but not
statistically significant as influencers of oviposition perhaps because only
occupied habitats were measured and thus they were de facto broadly suitable
for checkerspot use.
The consistent preference for high density hostplant patches indicates that
assessments of habitat and restoration objectives should weight high density
hostplant areas more heavily than mean site level host plant abundance or cover
and that, within areas of broadly suitable grassland vegetation structure, sites
containing high densities of hostplants can be found and used for oviposition.

Table of Contents

Table of Contents ........................................................................................... v
List of Figures ............................................................................................... vii
List of Tables ............................................................................................... viii
List of Appendices ........................................................................................ ix
Acknowledgements ........................................................................................ x
I. Introduction.................................................................................................. 1
Conservation Status .............................................................................................. 2
Biology and Description ......................................................................................... 5
Anthropogenic Habitats ....................................................................................... 11
Taylor’s checkerspot ecology............................................................................... 12

II. Developing captive rearing methods and measuring larval
performance on Castilleja hispida and Plantago lanceolata .................... 17
Methods .............................................................................................................. 18
Results ................................................................................................................ 22
Discussion ........................................................................................................... 23

III. Oviposition in the field: Where Taylor’s checkerspot chooses to
lay eggs ......................................................................................................... 28
Methods .............................................................................................................. 30
Sampling scheme ........................................................................................... 30
Finding oviposition sites .................................................................................. 32
Hemispherical photos ..................................................................................... 34
Analysis methods ........................................................................................... 35
Site descriptions and methods ............................................................................. 36
Eden Valley .................................................................................................... 36

iv

Eden Valley sampling methods .................................................................. 37
Eden Valley Results ................................................................................... 39
Dan Kelley Ridge ............................................................................................ 42
Dan Kelley Ridge sampling methods ......................................................... 42
Dan Kelley Ridge results............................................................................ 44
―Near Sequim‖ ................................................................................................ 46
“Near Sequim” sampling methods .............................................................. 47
“Near Sequim” results ................................................................................ 48
91st Division Prairie ......................................................................................... 51
91st Division Prairie sampling methods ...................................................... 51
91st Division Prairie results ......................................................................... 54

V. Summary, synthesis, and recommendations ........................................ 57
Hostplant density drives oviposition ..................................................................... 57
Slope, aspect, vegetation, and canopy shading ................................................... 60
Climate change impacts oviposition and larval survival........................................ 60
Future directions for research and lessons learned ............................................. 63
Literature Cited ......................................................................................................... 65
Appendices .............................................................................................................. 71

v

List of Figures

Figure 1: Adult male Euphydryas editha taylori ........................................................... 6
Figure 2: Taylor’s checkerspot eggs on Castilleja hispida ........................................... 8
Figure 3: Map of historic range and study locations .................................................. 10
Figure 4: Captive larvae feeding on Castilleja hispida............................................... 24
Figure 5: Larval webbing on Plantago lanceolata ..................................................... 29
Figure 6: Search plot at Eden Valley ........................................................................ 31
Figure 7: Search plot layout ...................................................................................... 32
Figure 8: Map of search plot and oviposition locations at Eden Valley ...................... 39
Figure 9: Search plot at Eden Valley showing dense oviposition .............................. 40
Figure 10: Paired oviposition/non-oviposition microhabitat plots at Eden Valley
and Dan Kelley Ridge ............................................................................................... 41
Figure 11: Taylor’s checkerspot ovipositing on Plantago lanceolata next to road
at Dan Kelley Ridge .................................................................................................. 45
Figure 12: Larval masses per search plot by habitat block at ―Near Sequim‖ ............ 48
Figure 13: Paired oviposition/non-oviposition plots at ―Near Sequim‖ ....................... 49
Figure 14: Linear regression of the count of larval masses versus number of
Plantago lanceolata at ―Near Sequim‖ ...................................................................... 50
Figure 15: Castilleja hispida used for oviposition on 91st Division Prairie .................. 53
Figure 16: Paired microhabitat plots at 91st Division Prairie ...................................... 55
Figure 17: Example of hemispherical photograph at Eden Valley ............................. 72
Figure 18: Thresholded hemispherical photo from Eden Valley ................................ 73

vi

List of Tables

Table 1: Life history of Euphydryas editha taylori........................................................ 7
Table 2: Care of captive reared Taylor’s checkerspot ............................................... 20
Table 3: Variables measured at oviposition sites ...................................................... 29

vii

List of Appendices

Appendix A: Hemispherical photography .................................................................. 72
Appendix B: List of plants identified at Eden Valley .................................................. 74
Appendix C: List of plant identified at Dan Kelley Ridge ........................................... 75
Appendix D: List of plants identified at ―Near Sequim‖ .............................................. 76
Appendix E: List of plants identified at 91st Division Prairie ....................................... 77

viii

Acknowledgements

Many have been instrumental in helping me through the twists and turns of this
thesis. In particular the help I received from Dave Hays, Mike Singer, Patrick Dunn, Ann
Potter, Mary Linders, Mary Jo Anderson, Erin Sullivan, Kelly McAllister, Dave Nunnallee,
John Pelham, Dave Clouse, Rod Gilbert, Jim Lynch, Anita McMillan, and Gordon Pratt
was invaluable.
Spending many long days in the field with Paul Severns, my research partner for
the 2010 oviposition fieldwork, was a pleasure and our discussions and debates on field
and analytical methodologies were instrumental in bringing this work to fruition. Paul
was equally responsible for study design and fieldwork for Chapter III.
My advisors at The Evergreen State College have been essential to the
completion of this thesis. John Longino was there at the beginning to guide the original
ideas and as an influence in rekindling my interest in the world of insects which had
been partially dormant since the infamous morning when the grasshopper with a pin
through its thorax wandered onto the breakfast table when I was a child. Judy Cushing
influenced the data analysis approach, helped me place the work in a broader context,
and did the difficult work of helping me bash my prose into shape.
Access to the four study sites was facilitated by Joint Base Lewis-McChord,
Washington Department of Natural Resources, and a private landowner who can’t be
named. Some of this work was funded by the Washington Department of Fish and
Wildlife, and some by the Army Compatible Use Buffer program.
My wife, Penny Kelley, has done her best to keep me sane during this work. It
has not been easy and I am forever in her debt.

ix

I.

Introduction
Taylor’s checkerspot (Euphydryas editha taylori) is a modestly sized yet

charismatically colored butterfly endemic to Northwest grasslands between the Cascade
Range and the Pacific Ocean. Historically found from southern Vancouver Island, British
Columbia, Canada, to the southern Willamette Valley, Oregon, in the United States, its
range has been dramatically fragmented by habitat loss. The few populations that
remain are threatened with extinction. This study was designed to support the
conservation and recovery of this butterfly by identifying critical aspects of oviposition
and larval habitats important for the survival of individual checkerspots.
The survival of Taylor’s checkerspot individuals is in large part dependent on its
surrounding habitat. The amount of habitat available is restricted by the ability and
motivation of the individual to move. In the life cycle of Taylor’s checkerspot, mobility
increases from zero at the egg stage, to a few centimeters in early instars, up to tens of
meters in later instars, and up to thousands of meters in adults though most adults stay
within a small area relative to their potential dispersal distance. Because of this
relationship between movement ability and habitat, and because previous work has
indicated prediapause larval survival is a major population driver, this study focused on
habitat selected by adult female butterflies for oviposition and larval survival on two
oviposition host plant species. By raising larvae in captivity this study also developed
methods for larger scale captive rearing that can generate individuals for population
reintroduction or augmentation.
The need for a study of Taylor’s checkerspot was driven by its endangered
conservation status coupled with a near absence of information about certain aspects of
its life cycle and habitat needs. This introduction will clarify this checkerspot’s

1

conservation status and outline ways in which the studies presented herein address
information gaps in our understanding of this butterfly.
This thesis has taken a number of unusual turns. It began in 2004 with a study of
Taylor’s checkerspot oviposition at Bald Hill in Thurston County, Washington, and
captive rearing to develop methods and determine larval performance on various
hostplants. Taylor’s checkerspot captive rearing and larval performance is presented in
Part II. The Bald Hill population with hundreds of adults spread across numerous small
balds was, at the time, considered the most secure population. That population crashed
for unknown reasons in 2005 which put thesis work on hold until the population
recovered. Unfortunately, Taylor’s checkerspot has not been seen in the Bald Hill Area
since 2007 despite substantial survey efforts and the population is likely extirpated. In
2010, I restarted work on the thesis by gathering data on Taylor’s checkerspot
oviposition at four western Washington sites; that work is found in Part III.

Conservation status
Native grasslands throughout the Willamette Valley-Puget Trough-Georgia Basin
Ecoregion, the habitat of Taylor’s checkerspot, have declined in extent to about 3% of
the area they occupied in 1850 (Crawford and Hall 1997). This decline is primarily due to
development for agriculture, urbanization, gravel mining, succession to forest, and nonnative invasive species. Many species of prairie dependent vertebrate animals have
declined dramatically and some are threatened with extinction or have already been
extirpated from prairies (Leonard and Hallock 1997; Rogers, Norman, and Rolph 1997;
Rogers 2000; Ryan 1997; Stinson 2005). Plants of the prairie have declined as well,
with golden paintbrush (Castilleja levisecta), a Taylor’s checkerspot host plant, federally
listed as threatened under the Endangered Species Act (Caplow 2004). Several other
plants are listed as sensitive under the Washington State Natural Heritage Program.
2

Prairie dependent butterflies show a similar pattern of population declines and
extirpations. Four butterflies of the Puget Prairies are listed as ―endangered‖ or ―species
of concern‖ by the state of Washington. Thirteen butterflies that use grasslands in the
Willamette Valley-Puget Trough-Georgia Basin Ecoregion are listed as endangered,
threatened, candidate, or extirpated (or equivalents) by national or state/provincial
governments. Taylor’s checkerspot is listed as ―endangered‖ by the State of Washington
(Stinson 2005) and considered ―critically imperiled‖ by state and provincial natural
heritage programs. The species has been petitioned to the U.S. Fish and Wildlife
Service and is a candidate for emergency listing under the Endangered Species Act
(Stinson 2005). This means there is likely enough evidence to warrant listing but the US
Fish and Wildlife Service has yet to list the species. Agencies in British Columbia have
developed draft recovery plans that call for the reestablishment of three robust
metapopulations within the area formerly occupied by the butterfly (Guppy, Kondla, and
Schaeffer 2003) and investigation of the one remaining Canadian population of Taylor’s
checkerspot continues (Page et al. 2009)
Taylor’s checkerspot was abundant in 60 documented, but probably thousands of
localities in 1800, but certainly many of these were gone by the early 1900’s by which
time much of the habitat had been converted to agriculture. Now however the butterfly is
almost gone with one population in Canada, approximately 10 small populations in
Washington, and one small population in Oregon. It was known to be abundant in some
colonies in fairly recent times in Oregon (Dornfeld 1980) and Washington (Stinson 2005)
and, as recently as 1996, several populations in the South Puget Sound numbered in the
thousands. By 2000 only a few populations were found and even in the largest of these,
few checkerspots were found (Fleckenstein and Potter 1999; Remsberg 2000; Stinson
2005). A likely metapopulation spread across several miles of bald habitat in

3

Washington went from hundreds (maybe thousands) of adults down to zero between
2003 and 2006.
There are no clear answers as to why Taylor’s checkerspot populations have
crashed recently even in protected areas. Habitat loss likely played the major part in
earlier declines and may be driving recent extirpations as well by disrupting
metapopulations and confining butterflies to marginal sites. Many other populations of E.
editha act as metapopulations and exhibit tremendous variability in abundance and local
distribution (Thomas, Singer, and Boughton 1996; Harrison 1989; Boughton 1999;
McLaughlin et al. 2002; Ehrlich and Murphy 1987). The fragmentation of prairies and
the low vagility of Taylor’s checkerspot indicates the isolated populations can no longer
act as metapopulations (Char and Boersma 1995). Isolation increases the chances that
individual populations will become permanently extirpated by making natural
recolonization unlikely. Climate change may also be a cause of population extirpations
and this is likely to increase in the future (Hellmann 2004).
Butterflies are often considered to be both good indicator (Black, Shepard, and
Allen 2001; Pollard and Yates 1993; Samways 1994) and umbrella (Launer and Murphy
1994; Dunn 1998) taxa. Thus the imperiled nature of prairie endemic butterflies along
with massive habitat loss may indicate the prairie system is in crisis.
A number of research and conservation efforts targeting E.e. taylori habitat are
planned or ongoing. Efforts by the Washington Department of Fish and Wildlife and
others are underway to survey historical locations in Washington. So far, all these efforts
have failed to find taylori populations except for a few small populations in the
northeastern Olympic Peninsula lowlands (Potter pers. com.). Invasive plant control has
been a focus for several years on many E.e. taylori sites including three of the four sites
used in this study.

4

Because Taylor’s checkerspot has been extirpated from numerous sites
containing suitable habitat and sites that may contain suitable habitat in the future with
restoration and enhancement, it is necessary to develop methods for reintroduction.
Most of these sites are beyond dispersal distance for butterflies in extant colonies.
Captive rearing is an essential part of a successful reintroduction strategy.
Captive rearing is necessary for propagating enough individuals for reintroduction
into areas from which this species has been extirpated and for detailed study of
biological and ecological attributes impossible to study in the field. The reproductive
potential of founder adults can produce at least a 100 fold increase in the captive
population for release into the wild once techniques are refined. Reintroduction success
is positively correlated with the number of released individuals (Oates and Warren
1990). E. editha had been thought to be difficult to rear though stimulation of oviposition
in captivity is straightforward (Singer et al. 1992). Protection, restoration and
enhancement of habitat is far more important than captive rearing in the long run but
captive rearing can help bridge the gaps in fragmented habitats and the gaps in
biological knowledge.

Biology and description
Euphydryas butterflies have been extensively investigated as model organisms
for metapopulation studies, genetics, and conservation and population biology (Britten et
al. 1995; Ehrlich 1992; Ehrlich and Hanski 2004; Ehrlich and Murphy 1987; McLaughlin
et al. 2002). Another subspecies, Euphydryas editha bayensis is one of the most
thoroughly studied non domesticated organisms on earth (Ehrlich and Hanski 2004; New
1997). Euphydryas editha is a highly variable species with many subspecies though
there are differences of opinion on exactly how many subspecies are valid (Austin and
Murphy 1998; Baughman and Murphy 1998; Guppy and Shepard 2001; Hodges et al.
5

1983; Scott 1986) but the most recent authoritative taxonomic catalog lists 26 (Pelham
2008). The case for subspecific rank in taylori is fairly strong—that is to say, it likely
designates real and properly delimited (monophyletic) nearly discrete variation within
Euphydryas editha resting on three attributes: phenotype, ecology, and distribution but
molecular evidence is as yet unavailable.
Like other members of its genus, E.e. taylori is quite variable in the proportions of
colors expressed on wings, but wings are usually spectacularly checkered with white
and brick orange-red spots on a deep brown/black background (Figure 1). The variation
is striking, from specimens that are dorsally mostly black to those that are quite light with
narrower black lines between the orange and white ―cells.‖ The ventral side of the hind
wing has an ―editha line‖ that reliably separates editha from other Euphydryas (Guppy
and Shepard 2001; Pyle 2002; Scott 1986). Antennae are ringed with alternating bands
of black and orange on basal segments while the terminal club segments are banded

Figure 1: Adult male Euphydryas editha taylori perching on unopened Camassia quamash
adjacent to Plectritis congesta, a post diapause host. Adult from the now apparently extirpated
Bald Hill (Thurston County, Washington) area population.
6

in orange. Males are smaller than females overall with a wingspread averaging 3.8cm.
Female wingspread averages 4.1cm with a proportionately longer and wider abdomen
than the male.
The life cycle of Taylor’s checkerspot is similar to most other Lepidoptera with the
exception of a long larval diapause between early summer and late winter (Table 1).
First instar larvae are tan with small dark spines along the dorsal and lateral surfaces.
Third instar larvae are brown with branched spines. The larvae gradually become darker
with each instar change so that by the fourth and later instars the larvae have black to
dark brown skin splotched with pearlescent lighter spots and a row of orange tubercles
topped with complex black spines along the dorsal surface. Larvae feed in groups within
a loose silken web during the first through third instars. During the third instar larvae will
often leave the web and by the fourth instar larvae begin a mostly solitary existence.
Table 1: Life history of Euphydryas editha taylori in Washington. Note that there is some variation in the
number of instars each larva goes through. Temperature, insolation, and the condition of host plants
influence speed of development and larval movement.
Life stage

Behavior

Egg

Cluster of 15-100 laid on or near the base of the pre-diapause host, Castilleja hispida (Figure 2), or
Plantago lanceolata. Egg laying begins inbetween the beginning of April and ends in late June
depending on elevation, latitude and seasonal weather variations. Most laying occurs within three
weeks in each population.

1

Eggs hatch about 7-14 days after laying and larvae begin feeding communally in a web they create
on host plant. Instar lasts about one week.

2

Larvae continue feeding communally in a web on host plant. Instar lasts about one week.

3

Larvae continue feeding communally in a web on host plant. Larvae start to wander more outside the
web but stay near host plants unless searching for a new host. Instar lasts about ten days.

4

Larvae continue feeding on host plant quickly becoming much more solitary. After feeding for about
ten days larvae begin to disperse and diapause. Larvae diapause until late winter. After emergence
from diapause, larvae feed on a wider range of host plants. Hosts may be spatially separate from
prediapause hosts so they may have to move to find food.

5

Larvae continue feeding on host plant until shortly before pupation. They stop eating and wander
until choosing a location for pupation.

Pupae

The pupal stage generally lasts about three weeks. Pupae have been found close to the ground
hanging from the underside of plants and litter.

Adult

Adults begin eclosing in mid-March (males eclose first) or later depending on weather conditions.
Adults probably live from a few days to about two weeks based on data from other subspecies.

7

Pupae have irregular splotches of muted white, reddish yellow and black, similar
to those found on the underside of the adult wing but with grey overtones. They have
been found suspended from low growing vegetation within a few centimeters of the
ground during pupation (D. Ross pers. com.).

Figure 2: Taylor’s checkerspot eggs on the base of Castilleja hispida at 91st Division Prairie.
Usually the eggs are on the underside of one of the lowest leaves or on the stem near the base of
the plant. Each egg is approximately 1mm in diameter.
Euphydryas editha has one of the most complex taxonomic histories of any
butterfly in North America. There are at least 26 currently valid (non-synonymous)
subspecies (Pelham 2008) with 21 in California alone (Murphy et al. 2004). Edwards
originally named E.e. taylori in 1888 and subsequently Gunder named both victoriae and
barnesi which are considered synonymous with taylori. The phenotype of E.e. taylori is,
on average, darker than any other E. editha subspecies with a higher proportion of dark
8

ground color scalation and it has a distinctively rounded forewing apex and outer margin
(Warren 2005). Some of the distinctions are subtle and there is some overlap with other
subspecies. Taken as a whole it can be reliably distinguished from other subspecies
though the most reliable characteristic is the location the butterfly was found (Figure 3).
Euphydryas is derived from the Greek euphys (a goodly shape) and dryas (a
dryad or wood nymph). It has been assumed that the specific epithet editha came from a
woman named Edith (Guppy and Shepard 2001) though this is not clear in the original
description (Boisduval 1852). Subspecies taylori is named after the Reverend George
W. Taylor, an early British Columbia lepidopterist (Guppy and Shepard 2001). There
have been two widely used common names. Taylor’s checkerspot is obviously based on
the subspecific designation and probably the most ―proper‖ common name since it was
first applied in print by Holland (Holland 1898). It has also been referred to as the
Whulge checkerspot (Pyle 1989) after the Salish word for what later became known as
Puget Sound.

9

10

Anthropogenic habitats
The story of human interactions with the land is not a sideline to the main
narrative of ecological processes; rather it is at the core of what created and maintains
the unique habitat of Taylor’s checkerspot. Fire was a tool used by native people
throughout North America and it had tremendous influences on the grassland habitats
used by Taylor’s checkerspot. Both fire and the harvest of camas (Camassia quamash,
C. leichtlinii) and other foods occurred in what is now Taylor’s checkerspot habitat
(Carpenter 2002). Native people in what is now western British Columbia, western
Washington, and western Oregon developed in synergy with the prairies and the
surrounding habitats that together created a rich mosaic of resources for people and
wildlife (Boyd 1999b; Leopold and Boyd 1999; Turner 1999). In some areas these native
practices and ecological processes created some of the most productive agricultural
land in the western United States, but that productivity did not last long under the
European model of agriculture (White 1980).
Taylor’s checkerspot may have already existed in the ecoregion when the first
humans arrived after the end of the Vashon glaciation. The historical distribution of the
Taylor’s checkerspot is Willamette Valley-Puget Trough-Georgia Straight grasslands and
grassy balds from southern Vancouver Island through the Willamette valley in Oregon
(Figure 3). It likely co-evolved with the ―natural‖ communities that came into being as a
result of Native American cultural burning practices (Stinson 2005).
Many of the cultural practices of the European settlers were profoundly different
than those of the Native American inhabitants. Domesticated cattle and sheep, exotic
plants from Europe, plowing, disruption of the fire regime, logging, growth of cities,
introduction of exotic insects by organic gardeners, suburban development, hobby
farming, chemical and concentrated dairy waste fertilizers, military training, and finally
anthropogenic climate change have all profoundly altered the available habitat.
11

The relationship between anthropogenic habitat modifications and Taylor's
checkerspot is complex and apparently contradictory because, while human activity has
led to the extirpation of populations and climate change may be a range wide threat,
human activities both past and present may be essential for the persistence of this
butterfly where it currently exists. All remaining populations occupy areas heavily
modified by humans including road edges, an artillery impact area, and habitats modified
by ecological restoration actions including shrub and tree removal and herbicide use.
Exotic invasive plants have reduced habitat quality in most areas but one exotic plant,
Plantago lanceolata has been adopted as a host with some populations apparently
dependent on this exotic host for completion of their life cycle (Severns and Warren
2008). The patterns of vegetation and host plant abundance can be driven in
unexpected ways by complex factors including grazing and anthropogenic nitrogen
deposition (Weiss 1999).

Taylor’s checkerspot ecology
Euphydryas editha is perhaps the best known butterfly species in North America
since it has been very intensively studied by a large number of workers (Ehrlich and
Hanski 2004), and it is likely that many aspects of that information can be, with caution,
applied to taylori. Other Euphydryas species provide additional pertinent information on
study methodologies (Konvicka, Hula, and Fric 2003). The larvae need non-senescent
host plants to eat and sun to increase their metabolism (Hellmann 2002; Kuussaari et al.
2004; Weiss, Murphy, and White 1988). Females are picky about where, on which plant
species, on which individuals, and under what conditions they oviposit. There are also
heritable differences in these oviposition preferences between even nearby populations
(Rausher, Mackay, and Singer 1981; Parmesan, Singer, and Harris 1995; Singer, Ng,
and Morre 1991; Singer, Ng, and Thomas 1988; Singer 2004). Adult taylori usually
12

avoid tall dense non-native vegetation, and all extant colonies are in native grasslands
and balds (Hays et al. 2000; Stinson 2005).
Extant Euphydryas editha taylori populations in Washington are known to use
Castilleja hispida and Plantago lanceolata as host plants for oviposition and prediapause
feeding (Guppy and Shepard 2001; Pyle 2002; Severns and Grosboll 2011). Canadian
populations also use Veronica scutellata (Page et al. 2009) while Oregon populations
appear to be restricted to P. lanceolata. Recent work has shown post-diapause feeding
on the oviposition hosts as well as Orthocarpus pusillus, Collinsia parviflora, C.
grandiflora, and Plectritis congesta and (apparently rare) use of other species (e.g.,
Symphoracarpos alba, (unpublished data)) and Veronica scutellata, V. serpyllifolia,
Plantago major, and Centaurium erythraea (Page et al. 2009) by post-diapause larvae.
Other populations of E. editha are known to use various Scrophulariaceae and a few
Valerianaceae as hosts and some populations require more than one host for
development—one or more prediapause and one or more postdiapause (Cushman et al.
1994; Hellmann 2002; Rausher 1982; Singer 2004; Singer, Ng, and Thomas 1988;
Singer, Thomas, and Parmesan 1993; White and Singer 1974). The distribution of
Castilleja hispida in prairies is highly patchy with many existing prairies containing few
plants. The non-native Plantago lanceolata is widespread but locations where E.e.
taylori is dependent on this host appear to have not only widespread P. lanceolata but
high density areas dominated by this plant. The other potential oviosition hosts are
either absent or nearly absent from lowland prairie sites in Washington. Euphydryas
editha bayensis is less able to persist in areas with only Plantago erecta (a native
annual) hostplants than it is in areas with Castilleja (Hellmann 2002), and taylori may
depend primarily on perennial hosts prediapause and annual hosts post diapauses.
In many E. editha populations, most larvae starve before diapause (Kuussaari et
al. 2004; White 1974), but this is strongly influenced by weather. This appears to be the
13

case for E.e. taylori as well (unpublished observations). All extant populations appear to
exclusively use perennial hosts for oviposition which increases larval survival odds
because most individuals of their perennial hosts senesce later than their annual hosts,
but many hostplants still senesce prior to larval diapause. Unless they can move to a
non-senescent host, larvae on those plants lose the race to diapause and do not survive.
Slope and aspect influence many aspects of habitat for many species as soil
development, vegetation, temperature, insolation, plant phenology, and rainfall can
change dramatically depending on slope and aspect. In some populations Euphydryas
editha larval survival is largely determined by the slope and aspect where larvae occur
(Weiss 1997). Shading influences larval development times and hostplant senescence
and known hostplants are all associated with open unshaded habitats.
Nectar plants used by the E.e. taylori include Camassia quamash, Lomatium
triternatum (Hays et al. 2000 and Dunn, unpublished data), L. utriculatum (Guppy and
Shepard 2001), Balsamorhiza deltoidea (Hardwick and Potter pers. com.), Plectritis
congesta, Marah oraganum (pers. obs.), Fragaria virginiana, Callicortis spp. and Malus
spp. (Severns and Warren 2008; Ross 2003). Like most butterflies they can, to some
extent, adapt to different nectar sources but in most cases use native plant sources that
are particularly abundant and in bloom when they are in flight. Extant populations are
found in areas with a high proportion of native grass cover. In areas heavily invaded by
non-native vegetation, E.e. bayensis is less successful (Weiss 1999) and this is probably
the case for the E.e. taylori as well (Severns and Warren 2008). Most other western
Washington prairie specialist butterflies avoid dense, tall non-native grasses (Hays et al.
2000) and anecdotal accounts suggest that E.e. taylori from open prairies often avoid
flying into dense vegetation over about 0.75m in height while flight behavior of
populations in balds is less predictable (pers. observation). Restricted nectar sources
can severely limit egg production in most butterfly species though this may be of slightly
14

less importance in E. editha than some other species (Boggs 1997). In any case, nectar
distribution likely influences oviposition site choice by increasing the probability that
females will be near nectar sources.
Many populations of E. editha act as metapopulations and exhibit tremendous
variability in abundance and local distribution (Baughman 1999; Baughman, Murphy,
and Ehrlich 1988; Ehrlich 1961; Harrison, Murphy, and Ehrlich 1988; McLaughlin et al.
2002; Singer and Ehrlich 1979). The low vagility of E.e. taylori combined with habitat
fragmentation, makes it likely that most of the now isolated populations can no longer act
as metapopulations (Char and Boersma 1995). Recolonization of suitable formerly
occupied habitat or colonization of new habitat is unlikely because distances between
most suitable habitat sites are beyond the ability of the butterfly to disperse and most
matrix habitat between sites is poor quality even for dispersal. In order for populations to
survive, habitat occupied by the butterfly has to be suitable for population persistence
under all conditions; given the complexities of this species’ population biology, and its
interactions with anthropogenic habitat and climate alteration, recolonization of otherwise
suitable habitat is unlikely.
As articulated above, Taylor’s checkerspot requires healthy larval hostplants
located in a microclimate and vegetation structure that makes these hostplants available
to ovipositing females and hospitable for larval feeding. Thus I contend that populations
of Taylor’s checkerspot butterfly may be partially limited by the lack of suitable habitat
preferred for oviposition. In this thesis I will show the characteristics of preferred
oviposition habitat essential for the recovery of this endangered species. These
characteristics can be used in combination with information developed about larval
survival on different hostplant species, different hostplant populations, oviposition
preference in captivity, and hostplant transplant methods to optimize restoration of
habitat for this endangered butterfly within the human dominated/degraded Willamette
15

Valley-Puget Trough–Georgia Basin Ecoregion. The characteristics of the selected
oviposition habitats indicate that the long history of anthropogenic habitat modification
continues and that this butterfly is likely bound to human alteration of the landscape in
unexpected ways.

16

II. Captive rearing methods and a comparison of prediapause
larval survival using Castilleja hispida and Plantago lanceolata
as hostplants for Taylor’s checkerspot (Euphydryas editha
taylori)1
Pre-diapause Taylor’s checkerspot larval hostplants are largely synonymous with
oviposition hostplants because larvae (particularly in the first two instars) have very
limited abilities to move given their minute size, slow locomotion, and limited energy
reserves. Because of the larvae’s dependence on oviposition hosts it is essential to
understand whether larvae survive and grow at different rates on different hostplant
species. To investigate their affects on larval survival and growth, two pre-diapause
hosts, Castilleja hispida and Plantago lanceolata were fed to different groups of
prediapause larvae.
Butterfly captive rearing for reintroduction is an expanding initiative in North
America (Lipman et al. 1999; Anderson et al. 2001) that has a long history in the United
Kingdom (Duffy 1977; Oates and Warren 1990; Pullin 1997). The potential for captive
rearing to contribute to the conservation of wild populations has been recognized for
some time in the United States (Pyle 1988), but only a few species have been reared for
reintroduction thus far. Rearing butterflies is an effective method for investigating
aspects of their behavior and biology as well.
In contrast to butterflies and other invertebrates, captive rearing of vertebrates is
well established as an emergency tool in conservation biology. Vertebrates are more
forgiving in captive situations because they tend to have significantly longer reproductive
lives than butterflies but, when captive rearing methods are well developed, butterflies
have the capacity to produce larger numbers of individuals quickly. In the case of these
butterflies there is only one opportunity, about two weeks long, for an individual animal to

1

Portions of this work were previously reported in Grosboll (2004).

17

successfully reproduce. Mistakes at any point in the butterfly life cycle commonly are
catastrophic for the captive population and mistakes are likely during initial rearing.
Because of their short life cycle, probable negative genetic consequences associated
with captive breeding (Brook et al. 2002; Bryant et al. 1999; Nieminen et al. 2001;
Saccheri et al. 1999) and substantial long term augmentation of wild populations with
captive bred animals (Heath et al. 2003), captive rearing using limited numbers of wild
mated females to produce eggs primarily for reintroduction is more conservative and
more appropriate for butterflies than transferring of large numbers of individuals from
depressed wild populations.

Methods
Captive rearing method development remains an iterative process that proceeds
through experiment and experience. A relatively large number of eggs are needed for
experimental rearing in order to test various treatments but clearly removal of individuals
from the wild should be minimized. In this study, one female Taylor’s checkerspot
provided enough (126) eggs for initial testing of three different host plants, different
group sizes, and feasibility testing of outdoor rearing.
This rearing project used a combination of methods developed by Gordon Pratt
(pers. com.) for Quino checkerspots (Euphydryas editha quino), and large scale rearing
techniques developed by the Oregon Zoo for Oregon silverspots (Speyeria zerene
hippolyta) (Anderson et al. 2001). Techniques developed at the Zoo are particularly
applicable to large scale rearing. The standard techniques include control of cross
contamination by pathogens by keeping larvae segregated, treatment of all surfaces that
will come into contact with larvae by autoclave, alcohol, or chlorine disinfection, frequent
hand washing by workers, searching host plant material for predators, and host plants

18

unsprayed with either pesticides or biologically active liquids (i.e., compost tea) (Sullivan
pers. com.; Anderson pers. com.).
Immediately after hatching larvae were moved to Petri dishes and given
appropriate food and space using larger dishes as they grew (Table 2). Larvae were
moved by picking them up with a cleaned water moistened size 0 paintbrush during the
first instar until midway through the third instar under a stereomicroscope at 8x – 16x
magnification. In later instars larvae were moved by scooping them with a small flat
spatula. The larvae were segregated into groups as they hatched to minimize the
transmission of disease and to allow the testing of different methods. Larvae were never
allowed to come in contact with larvae outside their group after hatching. All surfaces
that touched the larvae were disinfected to minimize pathogen transmission among
groups. Petri dishes were washed and soaked in a 2% chlorine bleach solution between
uses. The paintbrushes and spatulas used for handling larvae were washed in 95%
ethanol then washed twice in distilled water between handling each group of larvae.
Day length and average temperature were similar to those under outdoor
conditions at a lowland South Puget Sound grassland. The rearing area was located on
the Thurston County Glacial Heritage Preserve in an unheated room with large windows
that were constantly open to the outdoors during the pre-diapause larval feeding season.
The amplitude of temperature change throughout the day was likely less than outdoors
because of the sheltered location. Diapausing Taylor’s checkerspot larvae continued to
be segregated and were placed into either an outdoor ―humidor‖ chamber in an insulated
cooler that contained 4L of water in the bottom or a ―humidor‖ chamber in a non frostfree refrigerator at 4C. Larvae were individually checked weekly throughout diapause
for mortality and illness. When mold, mildew, or bacterial growth were found in
containers or in humidifying water the larvae were moved to new containers and the
humidor chamber was cleaned.
19

Table 2: Care of captive laboratory reared Taylor’s checkerspot larvae fed Castilleja hispida and Plantago
lanceolata. There was some instar overlap with individuals at varying life stages.
Life Stage

Action

Egg

Eggs laid on Castilleja hispida in captivity. Ninety-two of 94 eggs hatched.

Instar 1

Placed in 5.5-cm petri dishes with moist, not wet, filter paper in bottom. Two groups of five, two groups
of ten were placed on P. lanceolata. One group of three, five groups of five and two groups of ten were
placed on C. hispida. One group of five was provided with both hosts. Moved to new dish/fresh plants
every other day. Two medium (2.5 cm) C. hispida or one medium (6 cm) P. lanceolata leaf per five
larvae

Instar 2

Treatments and dishes as above. One medium C. hispida or half a P. lanceolata leaf per larvae

Instar 3

8.5-cm dish, otherwise as above. Two medium (2.5 cm) C. hispida or one medium (6 cm) P. lanceolata
leaf per larvae

Instar 4

13.5-cm dish, otherwise as above. Four medium (2.5 cm) C. hispida or one medium (6 cm) P.
lanceolata leaf per larvae. Began to treat as diapausing when they stopped feeding despite access to
fresh host material.

Diapause

After all larvae began diapause, they were weighed individually on an analytical balance. Petri dishes
were maintained in a humidified chamber at outdoor ambient temperatures until October 6 at which
point larvae were moved to well plates and Toledo jars. Two groups of larvae were kept in refrigerator
at 4C through early March, one group in Toledo jars initially then transferred back to petri dishes on 12
November; the other group was maintained in well plates. One group was kept in an insulated cooler
hydration chamber outdoors. In each case larvae were checked at least weekly.

Instar 4

Returned to feeding in early March. Dish treatment as in pre-diapause instar 4. Extra fresh plants
seemed to stimulate feeding. Experimental feeding groups were abandoned shortly after diapause
ended because not enough C. hispida plant material was available. Some larvae re-entered diapause.

Instar 5

Treatments and dishes as above; moved daily to new clean dish with fresh plants daily

Pupae

Pupae allowed to eclose in cages kept in sheltered area outdoors.

A single adult female was captured and destined to be a voucher specimen in a
newly discovered colony of Taylor’s checkerspot on a grassy bald near Port Angeles,
Washington. She was kept alive in captivity and laid 126 eggs on a C. hispida plant in
the one gallon cage in which she was kept. Thirty two of the eggs were given to David
Nunnalee, a local volunteer captive rearing expert. Those larvae subsequently died in
the second instar after feeding on Plantago major, apparently not an acceptable
prediapause host though postdiapause feeding has been seen in other populations
(Page et al. 2009). Those larvae will not be considered further. The remaining 94 eggs
20

were used for this captive rearing study (Table 1). There were 15 groups overall with
nine small groups (3-5 larvae) and five large groups (10 larvae). One group was raised
outdoors on a fine mesh enclosed and potted C. hispida plant.
The plant material provided to the caterpillars was cut from plants and
immediately washed. Washing was accomplished by putting whole stems or leaves of
host plant into a salad spinner strainer and washing plant material with a forceful, but not
plant bruising, stream of water for at least 30 seconds while moving plant material to
ensure thorough coverage. After washing, the plant material was spun in a salad
spinner to remove most of the water. The plants were then examined under 5x
magnification under a stereomicroscope to remove all other insects and eggs. Whole
stems of C. hispida were cut from the plants and larvae were provided a piece of stem
with attached leaves. When leaves were removed from stems of C. hispida, the leaves
would dry out quickly and become unpalatable to larvae. Whole leaves of P. lanceolata
were cut from plants growing in an area that received some water from a nearby lawn
sprinkler in an area 200m from a recently extirpated P. lanceolata feeding Taylor’s
checkerspot colony. Only leaves that appeared to be fairly new and tender without
significant herbivore damage were selected.
After they began to diapause, all groups were kept in a humidified environment in
the same room where they were reared until October 1 except Outdoor Groups 1 that
was reared outdoors on a potted plant. On October 1 three groups were placed into a
refrigerator at 4C: Indoor Group 1 was kept in Petri dishes throughout; Indoor Groups 2
was placed into well plates in which the area between the cells was filled with distilled
water; Indoor Groups 3 was placed into Toledo jars (modified narrow mouth one pint
mason style canning jars) in which caterpillars are suspended above water in a silk
organza cell but these Indoor Groups 3 larvae were moved back into Petri dishes
(similar to Indoor Group 1) in mid November because of problems with condensation in
21

the jars. The final group (In/Out Group) group was placed into an insulated outdoor
enclosure and exposed to outdoor conditions.
Statistical analysis were performed using Microsoft Excel and MINITAB statistical
software (MINITAB Statistical Software 2000).

Results
Larvae were successfully reared through diapause using any of the methods
when fed either Plantago lanceolata or Castilleja hispida. Survival for all hand reared
Taylor’s checkerspot larvae was 84% from egg through to the end of diapause. The
difference between survival on the two hosts P. lanceolata and C. hispida was not
statistically significant (85% for C. hispida, 90% for P. lanceolata, two proportions test P=
0.517). Weight gain also seemed to be unaffected by host plant (mean weight 2.62mg
when fed C. hispida; 2.59mg when fed P. lanceolata, Student’s t-test P = 0.809). One
group of five was fed both P. lanceolata and C. hispida and they fed readily on both
hosts generally choosing whichever they encountered. Their weight (2.42mg) was not
significantly different than the others though the tiny sample size gives very little power
to detect differences with this group.
There was no difference in survival or weight between 5 or 10 larvae per
container (five larvae reared mean wt. 2.53mg, ten larvae reared 2.60mg, Student’s ttest, P=0.602). There were no significant interactions between group size, host plant,
and weight.
Hand raised larval mortality was associated with molting. Often one of the group
would fail to molt, become lethargic and die. Larvae died more often in early instars and
less often as the larvae entered the fourth instar and diapause. Of the nine larvae
placed on a potted C. hispida plant, four made it through the winter and began feeding
again on warmer days in late February. They moved within the enclosure during the
22

winter, down into a gap between the pot and the potting soil during colder weather, and
up onto the wooden supports or leaf surfaces during warmer weather but none of them
successfully pupated. The pot was exposed to temperatures as low as -11C but was
slightly protected under the eaves of a building.
Most of the C. hispida plants used for rearing survived the regular cutting they
received but plants in larger pots that included another plant (either Festuca roemeri or
Eriophyllum lanatum) they could parasitize recovered more quickly than those grown
alone. As a facultative hemiparasitic species, C. hispida was likely able to replace some
of the production lost from its own leaves by that produced by its hosts.

Discussion
This work demonstrated that it is possible to captive rear Taylor’s checkerspot on
either Castilleja hispida or Plantago lanceolata. Methods used for Taylor’s checkerspot
worked remarkably well to get caterpillars from egg through diapauses and some larvae
successfully pupated and eclosed to successfully complete their life cycle.
The Petri dish rearing system was effective (Figure 4). The low volume and slow
but adequate ventilation, combined with a large bottom for dispersing frass keeps the
larvae moist enough while at the same time reducing mold formation. There have not
been any escapes and the dishes are easy to handle. At a larger scale, Petri dish racks
could be used to stack the containers which would ease handling and minimize the
space required.

23

Figure 4: Second instar larvae feeding on Castilleja hispida in captivity. Note medium tip ball point
pen on left for scale. The larvae in this photo have only been on the plant for a few minutes but
frass is already visible both on the leaves and filter paper substrate.
P. lanceolata and C. hispida, at least in captivity, appeared to work equally well
as host plants in this study and larvae will eat either and readily switch hosts, both in
captivity and in the wild (pers. obs.). Subsequent rearing efforts with larger numbers of
larvae at the Oregon Zoo indicate that mean weight of caterpillars is less when fed C.
hispida but survival through diapause was unimpaired (Linders 2007). These somewhat
ambiguous results may be the result of factors other than hostplant species such as
differences in the nutritional content of hostplant material caused by growing conditions.
For example, C. hispida is a facultative hemiparisitic plant that is affected by the
condition or presence of a host from which it draws nutrients.
The survival of larvae on both hosts is encouraging because, even though there
are significant efforts underway to plant C. hispida in proposed reintroduction areas, P.
lanceolata is widespread. Whether there are any tertiary effects, such as reduced levels
of compounds protecting caterpillars against predation, is unknown.
24

The effectiveness of P. lanceolata, an exotic species of European origin, as a
host creates a conundrum for habitat restoration and enhancement. Generally
ecological restoration attempts to create a functional ecological community or ecosystem
using native species and ecological processes. Several populations of Taylor’s
checkerspot appear to be completely dependent on P. lanceolata as a host, laboratory
measurements indicate that it works well for the butterfly, and other populations of
Euphydryas editha have exhibited heritable evolutionary shifts away from native hosts to
P. lanceolata (Singer, Thomas, and Parmesan 1993).
Group size did not affect rearing weight and therefore, because it is less labor
intensive to rear larvae in large groups, captive rearing should use larger group sizes. It
might be possible to increase the group size even more in hand rearing; this would be
more similar to wild larval groups which are usually much larger. Rearing in larger group
sizes would need to occur in a larger sized rearing dish to accommodate the additional
larvae. The space per larvae available for ten larvae in a 13.5cm petri dish is 13cm2,
and should probably not be reduced for fourth or fifth instars because of waste
generation, unless the petri dish is to be changed once a day or more.
Outdoor enclosure rearing of Taylor’s checkerspot was promising. Low
temperatures were the likely cause of mortality for five of the nine larvae as there were
at least seven alive prior to the cold temperatures. An insulated retreat area would likely
reduce this mortality. Not all larvae can be always seen so there is a loss of information
on exact numbers of larvae but enough can be seen to indicate whether there has been
a catastrophic change. Potentially a raised bed of C. hispida, Collinsia parviflora,
Plantago lanceolata, and Plectritis congesta could be constructed and netted to raise
larvae in an inexpensive and more natural situation. This could both reduce costs and
decrease selection for captivity. Excluding predators from such an enclosure is difficult
but it should be possible to eliminate many of them.
25

Potted C. hispida plants required regular attention and perhaps in the ground
plants would be more effective in a larger rearing operation. Fifty fairly well established
C. hispida plants were just barely enough for the fifty caterpillars feeding on this host. A
ratio of two plants for each larva would be better and reduce the stress on plants from
excessive cutting.
Scaling up rearing gradually provides opportunities to learn additional information
about the species’ natural history and minimizes the loss of animals. During this project,
with 92 larvae in captive rearing, a few inferences about different methods could be
drawn. With more larvae in the captive rearing program, investigation of rearing using
adaptive management techniques has become more powerful and rigorous. Current
captive rearing at the Oregon Zoo is producing many hundreds of larvae each year for
reintroduction (Anderson, pers. com) and this has generated additional biological
information.
Because larval mortality is cumulative and higher in early instars, reintroduction
can be done with larvae either at the beginning or end of diapause, or as adults.
Releasing larvae rather than adults eliminates the unique problems of handling pupae
and freshly eclosed adults but it is difficult to know what happens to the cryptic larvae
once they are released (Fimbel 2008). It also exposes larvae to natural conditions which
may increase their propensity to exit diapause in the spring. Current rearing efforts use
a mix of larval and adults in experimental reintroduction releases (Linders 2007).
Success in getting butterflies through diapause and to pupate and eclose at the same
time as their wild counterparts is essential for scaling up for successful adult
reintroduction releases into suitable habitat.
Initial captive rearing results were promising but details of winter diapause
emergence, pupation and eclosion timing still needed additional effort after this project.
After this mid scale exploratory attempt, current attempts have built on the natural history
26

information generated and allow increasingly rigorous comparisons between different
rearing protocols. Subsequent projects proved the effectiveness of these methods
(Linders 2007) and refined them further leading to experimental reintroduction attempts
at three sites.

27

III. Oviposition in the field: Where Taylor’s checkerspot chooses
to lay eggs2
Several aspects of Taylor’s checkerspot oviposition were of particular interest in
this study. First, the spatial distribution of oviposition at each site has habitat
management, as well as ecological, implications. Second, understanding which aspects
of habitat are most important to an ovipositing checkerspot is critical for undertaking
habitat restoration and selecting suitable sites for reintroduction of checkerspots. The
remaining two questions are of primarily scientific rather than management significance:
1) are there physical characteristics of hostplants, hostplant distribution, or nectar plants
that drive the choice of oviposition; and 2) are there microhabitat parameters for
oviposition that differ from broad scale habitat preferences.
Oviposition habitat was measured in four populations (three in Clallam County,
Washington: Dan Kelley Ridge, Eden Valley, and ―Near Sequim‖, and one in Pierce
County, Washington: 91st Division Prairie). The exact locations of some of these sites
cannot be given in this thesis because of concerns that populations will be damaged by
either malicious individuals or unscrupulous butterfly collectors; thus map locations of
study sites are deliberately imprecise (Figure 3 map of locations/range). The ―Near
Sequim‖ population is on privately owned habitat and access to the site was given with
the stipulation that its exact location not be publicized.
Stratified random sampling was used to subsample each site and some smaller
areas were censused. Measurements for hostplants not chosen for oviposition (no eggs
or larvae) were compared to hostplants chosen for oviposition (with eggs). The variables
in Table 3 were measured for each plot. Differences between plants or microsites

2

Paul Severns, Ph.D., was jointly responsible with the author for field work and study
design; the author was responsible for field work, study design, analysis, and writing.
28

Table 3: Variables measured to determine oviposition site preference. Note that vegetation, number of
host plants, and vegetation height was also recorded for search plots.
Measure
Aspect/slope
Larval clusters
Plant species
Nearest plant
Plant height
Plant # stems
Plant volume
Plant phenology
Canopy shade
Vegetation

Explanation
GPS coordinates on digital elevation model determined aspect and slope of each sample plot.
Count of egg clusters/larval masses on plant.
Species on which oviposition occurred. Based on earlier observations and accounts from other
subspecies, only likely hosts were searched.
Nearest conspecific hostplant.
Maximum height of plant.
Number of stems on the plant.
Plant area as measured by distance between furthest points horizontally on the plant multiplied
by the distance across the plant perpendicular to that axis multiplied by the height.
Categorical data: flowering, senescent, etc.
Shade experienced by the plant using the sunpath method on hemispherical photographs.
Percent cover for 1m2 quadrat centered on the oviposition site, and adjacent randomly selected
non-oviposition plot.

selected for oviposition should indicate which variables are most important in how
butterflies chose where they oviposit. Searching for larval masses (groups of larvae
sharing a webbed shelter) was effective, and damaged habitat less than searching for
eggs because larval masses are much more visible (Figure 5) than the inconspicuous
eggs and required far less trampling and disturbance of hostplants (Figure 2). Moving

Figure 5: Larval webbing on Plantago lanceolata. This webbing is visible within two to three days
after eggs hatch. These larvae are likely at least two weeks old and in the second and third instar.
Photo by Paul Severns.
29

leaves to examine the underside often detaches leaves or dislodges eggs or larvae
either of which can lead to the death of eggs or larvae.

Methods
Sampling scheme
A stratified random scheme to sample for oviposition sites and habitat was used
in most areas. Some habitat patches were small enough that a total census was
performed by searching the entire area. Randomization, dispersion, minimizing
disturbance to habitat, sampling efficiency, and safety (two sites contained steep cliffs
and one contained unexploded artillery shells) were considered in developing a sample
scheme for each site. In addition, landowners restricted access to two of the sites which
limited the time available for sampling. The biology of the butterfly also influenced the
available search time because larval groups begin to disperse in search of food as
hostplants are denuded or become senescent. Because we could control the sampling
scheme but not the other factors we attempted to maximize the available time at each
site to generate as many samples as possible.
The sampling was nested in order to determine both broad site preferences and
microsite preferences. Search plots were 5m x 5m which was an area that could be
searched in a reasonable amount of time and in which the vegetation could be
characterized in a straightforward manner. Plots were enclosed in a 5m x 5m rope
quadrat prior to search (Figure 6). Within plots, all potential host plants were
systematically counted and searched for the presence of larvae. Larval masses were
flagged as they were found. Vegetation and other cover within the 5m x 5m plot was
characterized using visual estimation of cover by species within two 1m x 2.5m sample
quadrats randomly placed on each side of the 5m x 5m quadrat and placed across

30

Figure 6: Search plot at Eden Valley. Note distinct edge of bald habitat behind author and
moderately steep south facing slope. The edges of bald habitat were generally clear with a
minimum of ecotone habitat between the bald and surrounding forest. Photo by Paul Severns.
(lengthwise) to any environmental variation (e.g., perpendicular to a road, across an
ecotone). This was done to measure broad scale habitat preferences within the site. If
multiple oviposition sites were found in a search plot, one was randomly selected for
microsite measurement. A 1m x 1m quadrat was centered and oriented square to the
cardinal directions on the selected larval mass (when present) and vegetation and
hostplants were measured at that scale as well as in a randomly selected adjacent and
larvae free 1m x 1m quadrat to measure finer scale habitat selection (Figure 7).
Categories measured included bare ground, cryptogam, shell, rock, gravel, sand, wood,
thatch, and plastic. Ocular estimation was also used to measure cover in the 1m x 1m
paired oviposition quadrats. Cover values were recorded to the nearest 1% for
species/categories with > 1% cover and to the nearest 0.1% for species/categories with
<1% cover. For known and potential host plants (Castilleja hispida, Collinsia parviflora,
Plantago lanceolata, Plectritis congesta, and Triphysaria pusilla were found in plots) we
recorded cover to the nearest 0.1%.
31

In addition, within the paired microplots all
host plant heights, lengths, and widths were
measured and the number of stems of
Castilleja hispida were counted. To obtain a
coarse estimate of vegetation height within
the 5m x 5m oviposition search plots, we
measured the height of the tallest vegetation
occurring within each of the 25 1-m2 in the
search plot. For plants taller than 2m we

Figure 7: Search plot layout showing paired
plots. Randomly selected oviposition site
(Ovi) paired with adjacent randomly selected
non-oviposition site (No ovi).

estimated to the nearest decimeter and to the
nearest 0.5m for plants taller than 4m.

Finding oviposition sites
To study the interaction between where Taylor’s checkerspot lays eggs and its
habitat it is essential to find oviposition sites; this is difficult for several reasons. First,
butterfly eggs are small, in the case of Taylor’s checkerspot, less than 1mm in diameter.
Second, the butterfly nearly always lays eggs in a clutch of 6-50 eggs which helps in
spotting them but reduces the number of total oviposition sites (compared to most
species which lay single eggs) and thus the probability of detection on an individual
plant. Third, the eggs are nearly always laid in a hidden location near the base of a host
plant, which can be incredibly abundant, either on the stem or under a leaf. Fourth,
Taylor’s checkerspot is an endangered species living in fragile, rare, and declining
habitats so trampling and disturbance of host plants (which sometimes dislodges eggs)
needs to be minimized. Finally, the timing of the flight period and thus egg laying
generally lasts two to four weeks which means that some eggs are still being laid when
others have already hatched.
32

In order to overcome the challenge of finding oviposition sites we used larval
groups as a proxy for oviposition sites, and we only searched known host plants. When
larvae hatch from a clutch they immediately begin feeding and forming a larval web
starting at the oviposition site that serves to protect them from some predators and
parasitoids and perhaps to assist in thermoregulation. Larvae stay within and near the
web as long as the host plant remains palatable. Early instar larvae have very limited
dispersal range and will likely die when the host they are on is either consumed or
senesces if they do not find a new host plant very quickly; early instar larvae will not
successfully feed on non-hosts. Feeding webs become visible from several feet away
within a few days thus making the larval aggregation visible without disturbing the host
plant or larvae (Figure 5). At some point, often in the third or fourth instar, the larvae
usually begin to disperse to find other host plants as the original plant has been largely
consumed. Because Euphydryas editha is known to have high fidelity to its host plants,
the assumption was made that taylori would follow the same pattern. All observations of
taylori actually ovipositing have been on known host plants and even if there were
oviposition on other plants, larval survival is unlikely and thus feeding webs would never
be formed. We assumed that egg laying on non hosts was negligible. Even if some
eggs were laid on non-hosts it is unclear that useful information would be generated.
Since the purpose of this study was to develop information useful for the conservation of
this species, we found no practical value for restoring habitat or recovery of Taylor’s
checkerspot in the finding that some of the butterflies lay eggs where larvae would
invariably starve though finding such an apparently maladaptive behavior would be of
evolutionary interest. Based on these factors and the impossibility of searching every
plant thoroughly for eggs, short of an inefficient and habitat damaging search using large
numbers of people, searching for larval webbing was used as a proxy for oviposition site.

33

Euphydryas editha uses plants in a few families in the order Lamiales and
Valerianaceae for oviposition. All members of these families, which included Castilleja
hispida, Collinsia parviflora, Mimulus spp., Plantago lanceolata, Plectritis congesta, and
Triphysaria pusilla, were searched within the search plots and search areas.

Hemispherical photos
At two sites (Eden Valley and Dan Kelley Ridge, which will be introduced below)
hemispherical photographs of the sky above each oviposition site and each search plot
were taken using a Nikon Coolpix 4500 coupled with a Nikon FC-E8 0.21x fisheye
converter. The camera was placed on a tripod in the center of each plot directly above
the vegetation, oriented to point north, and leveled. If an oviposition site was present
and greater than 1m from the center of the plot an additional photo was taken and that
location. Exposure was adjusted to achieve maximum contrast using methods intended
to maximize contrast between sky and everything else (Zhang, Chen, and Miller 2005),
and photos were taken prior to sunrise, after sunset, and under cloudy conditions. All
hemispherical photographs were taken July 17-23, 2010. No hemispherical photographs
were taken at ―Near Sequim‖ or 91st Division Prairie because neither has substantial
canopy or terrain that would be captured using this method. The resulting digital
hemispherical photographs were manually thresholded (see Appendix A for examples of
original and thresholded images) and analyzed using Gap Light Analysis V2.0 software
to estimate the amount of solar radiation at each plot.

34

Analysis methods
Because this study measured variables at four different sites each with their own
characteristics and at different spatial scales within sites, the results were analyzed first
within sites and then, when appropriate, across sites. This study generated a large
amount of data and sifting through the data to find the information that is important at
each site and to then zoom out to consider common aspects of Taylor’s checkerspot
oviposition across all sites was challenging. For purposes of the oviposition study,
approximately random sampling as well as census data at individual sites was treated as
blocks for analysis among sites. Paired oviposition-non-oviposition plots were sampled
at every site while hemispherical photos were only taken at two sites. In each case the
sites used in multi-site analysis are noted. Statistically significant findings that are
unlikely to be biologically or ecologically significant are reported and placed in a
biological context.
GIS analysis was limited to developing information for other analysis. GPS
locations for oviposition search habitats were projected onto a digital elevation model of
the area to determine the slope and aspect of each search plot. Those results were
used as an input for Gap Light Analysis software and were analyzed statistically to
determine their influence on oviposition site selection.
Previous vegetation analysis using Multi-Rank Permutation Procedure (MRPP)
(Mielke 1984) and Indicator Species Analysis (ISA) (Dufrene and Legendre 1997)
indicated that most vegetation measurements were not predictive of oviposition at the
scale of the study at any site (Severns and Grosboll 2011). Perhaps this is because the
selected sites were already known as occupied habitat that contained a small subset of
the available habitats and vegetation within the range of Taylor’s checkerspot. By some
measures, species richness for example, the vegetation sampled at every site is
dramatically different between sites and between areas within sites. By other measures,
35

such as vegetation structure, the sampled areas were very similar across and within
sites. Rather than re-analyze plant community and vegetation data as a whole, analyses
of vegetation were focused on selected attributes of vegetation data (particularly host
plants) that were associated with oviposition.
Data were transformed to improve normality where appropriate and
transformations used followed the recommendations of Zar (1999). Transformation
methods are noted where used. Initial data entry and manipulation was done in
Microsoft Excel. Statistical analyses were performed in Minitab and Analyze-it where
noted.

Site descriptions and methods
The four sites, Eden Valley, Dan Kelley Ridge, ―Near Sequim‖, and 91st Division
Prairie, are spectacularly different in many ways but all contain primarily low growing
herbaceous vegetation with either very thin, excessively well drained, or somewhat
excessively well drained soils that dry out during the usual summer drought. Sampling
methods were adapted to each site as necessary because of site differences and
constraints. Additional site information is available for researchers with approval from
the Washington Department of Wildlife.

Eden Valley
Eden Valley is a grassy bald complex west of Port Angeles topping out at
approximately 400m. It contains steep bald and cliff habitats with shallow soil over
bedrock in a matrix dominated by second or third growth mixed conifer/deciduous forest.
Portions of this site are inaccessible due to the steep terrain but most of the area
occupied by host plants is accessible with caution. Plantago lanceolata, Castilleja

36

hispida, Plectritis congesta, Mimulus sp. and Collinsia parviflora occur on the site.
Rough estimates of adult Taylor’s checkerspot range between 100 and 300 in years
when they are most abundant but monitoring has been sporadic and not designed to
give a true population estimate (Ann Potter pers. com.). Our goals at Eden Valley were
to sample, detect, and describe the habitat and species used for oviposition in the large
bald (thought to contain a large proportion of the population) and to census the smaller
balds to determine if they are used for oviposition.

Eden Valley sampling methods
Eden Valley vegetation was sampled June 19-29, 2010 though earlier visits to
the site were used to lay out the sampling scheme and provide anecdotal information
about the butterfly flight period (in 2010 adults were observed beginning April 21 through
June 27, much longer than the average 3-4 week flight period). The vast majority of
adult butterflies were gone by the beginning of June, two weeks prior to the initiation of
oviposition sampling, but three were spotted during sampling; thus adults, eggs, and
larvae were all present at the same time.
The main bald (Figure 8) was defined as our sample universe for this site but
unfortunately not all of it could be randomly sampled because of vertical cliffs and some
of it needed to be excluded because it was not bald habitat. Even with the use of
climbing gear sampling the cliffs would have been impractical and because they were
covered in a fragile and poorly anchored layer of cryptogams, would have caused
substantial damage for little gain. Arguably these cliffs are not bald habitat but cliff
habitat and therefore could be removed from the sampling scheme for that reason as
well. Fortunately many cliff areas could be seen from other vantage points and no
Castilleja hispida was seen. We developed rules for excluding sample sites that
included avoiding areas that had poor footing and exposure to life threatening falls. We
37

also excluded forest and large shrub dominated inclusions in the bald (>95% tall shrub
cover or tree canopy cover >75% directly over plot) and used these same parameters to
identify the bald edges. These areas were excluded because, while they were present
within and adjacent to the bald, they are not bald habitat and they clearly were not
habitat used by Taylor’s checkerspot.
The bald was divided into a 5m x 5m grid and north-south transects every 20
meters, with a random start, were established. For each transect an origin of 0m, 5m, or
10m from the northern edge of the bald was randomly selected using a random number
generator. The remainder of the transect was sampled every 15m with the 5m x 5m
rope quadrat until the edge of the sample bald was reached either because of tree/shrub
cover or encountering cliffs. This created a randomized but well dispersed sampling
scheme across the large bald (Figure 8) and yielded 51 search plot locations four of
which were excluded as not bald habitat (patches of dense forest) though they were
within the boundaries of the bald.
The smaller balds to the north of the main bald were systematically censused for
both host plants and larval masses. The searchers walked transects across the balds 25m apart (depending on vegetation density) searching for and counting host plants and
larval masses. The balds are largely surrounded by continuous forest. A small portion of
two of the balds was searched in part with binoculars because of poor footing and
exposure to cliffs but this did not likely substantially degrade the search effort in those
areas because vegetation was very sparse and individual plants were easy to see and
search.

38

Figure 8: Search plot distribution in main bald at Eden Valley. Plots in blue show larval masses. Note the
two plots in the upper left area that appear to be in a shrub or tree habitat are actually in an area that was
cleared of shrubs after this orthophoto was taken.

Eden Valley Results
Eden Valley yielded 51 5m x 5m search plot samples, 11 of which contained
checkerspot larvae with a peak of 6 oviposition sites in one plot (Figure 9). A total of 83
plant species (Appendix B) were recorded in these plots along with the cover of bare
soil, gravel, rock, wood, thatch, and qualitative data on previous restoration treatments.
Vegetation as a whole proved to be a poor predictor of oviposition habitat with
multivariate examination leading to statistically insignificant results (MRPP: A-statistic =
0.009, p = 0.13) when host plants were excluded from the analysis (Severns and
Grosboll 2011). Host plant density, in this case Castilleja hispida, is strongly associated
with oviposition. Oviposition was also observed in five of the six small censused balds
outside the main bald. With the exception of one larval mass found on Plantago
lanceolata in one of the small balds, all oviposition sites were on C. hispida.

39

Figure 9: Search plot at Eden Valley. Blue flags mark larval masses found in this dense patch of
Castilleja hispida (flowering orange/red in photo). The density of oviposition appeared to be similar
throughout areas with high cover of this plant.
The correlation between C. hispida cover and oviposition is strong when all n-51
plots are considered (untransformed data, Pearson’s Correlation Coefficient=0.622,
p<0.000) but this is biologically trivial because it includes plots that include no host
plants and therefore will not contain larvae by definition. Considering only plots that
contained C. hispida the relationship weakens but is still significant (data transformed
Log (x+1), linear regression: adjusted r2=27.4%, p=0.006). The results of binomial
regression, to test oviposition versus no oviposition (C. hispida cover transformed Log
(x+1)), binomial logistic regression p=0.032) versus C. hispida cover. The count of C.
hispida was less powerful as a predictor of oviposition (untransformed data, Pearson’s
Correlation Coefficient=0.272, p=0.054). The level of Castilleja hispida cover is
significantly associated with Taylor’s checkerspot oviposition compared to C. hispida
cover plots without oviposition (data transformed Log (x+1), Students t-test, p = 0.013).

40

Results from the paired 1m^2 plots echo the findings at the broader scale 5m x
5m plots. Removing from analysis all paired non-oviposition plots without Castilleja
hispida leaves n=9 pairs of plots. Using a paired t-test on untransformed measurements
of total C. hispida volume (Figure 10) oviposition plots contained a mean of 44580 cm^3
C. hispida compared to 3621 cm^3 in the paired non-oviposition plots (p = 0.006).

Volume of Castilleja hispida per paired 1m^2 plot (log
scale)

100000

10000

1000

100

Eden Valley
Dan Kelley

10

Oviposition plot

Non-oviposition plot

Figure 10: Paired plots at Eden Valley and Dan Kelley. Volume of Castilleja hispida estimated by
measuring the height x length x width of each plant in plots. Non-oviposition plots without C. hispida (n=7)
were excluded. The oviposition plot was higher in C. hispida volume than randomly selected adjacent plot
in 18 of 23 pairs (Wilcoxon paired 2-tailed t-test: Eden Valley p=0.0039; Dan Kelley Ridge p = 0.0353).

41

Hemispherical photographs of each plot illuminate differences between plots.
The range of transmitted light oviposition plots received was not significantly different
than those not used for oviposition. Despite this, ovipositon sites received more light
than non-ovipostion sites.

Dan Kelley Ridge
This is another steep grassy bald complex and the highest elevation site studied
topping out over 600m in elevation. Much of the site burned in a wildfire in 1987 and
was apparently salvage logged thereafter. Many of the balds have been succeeding to
shrubs (Holodiscus discolor is particularly abundant) and young Pseudotsuga menziesii.
Anecdotal evidence indicates that the dominance of shrubs has substantially increased
in recent years thus likely reducing the abundance and availability of Castilleja hispida.
Substantial portions of this site are inaccessible because of the steep terrain and much
of the habitat is seriously degraded by invasive species and logging activities. Taylor’s
checkerspots were discovered on the site in 2003 but their numbers and distribution on
the site are largely unknown. Anecdotes about oviposition are concentrated on the road
edge but this perhaps is more indicative of the distribution of biologists than the
distribution of checkerspot oviposition. The goals at Dan Kelley Ridge were to determine
what habitats and species were being used for oviposition in particular within previously
identified and restored habitats and in general across the accessible portions of the site.
Plantago lanceolata, Castilleja hispida, and Plectritis congesta occur on the site.

Dan Kelley Ridge sampling methods
Dan Kelley Ridge vegetation was sampled July 13-20, 2010. Site visits in June
and early July indicated that the flight period for Taylor’s checkerspot extended to early
July and this was the earliest that we could be sure that most eggs had hatched. By mid
42

July many host plants were stressed by a lack of water and had begun to senesce thus
likely leading the starvation of many larvae.
Steep cliffs, large site size, and limited information about the distribution of
checkerspots led us to focus our search area on three particular locations: 1) a small
natural grassy bald that was outside the burn and recent logging area, 2) an adjacent
area, recently treated to control shrubs, that contained substantial Castilleja hispida, 3)
the road edge, 4) old log landings/pull outs where oviposition had been previously
observed, and 5) a search of accessible areas along the crest of the ridge.
The natural grassy bald and the adjacent treated area were each census
searched and larvae were flagged. Both balds are steep south facing slopes. In 2009 a
former bald area that contained large numbers of suppressed Castilleja hispida was
treated to control shrubs (Dave Hays pers. com.). The census in the treated area was
more haphazard than would be ideal because the terrain is steep, with thousands of
short stumps from shrub removal and old logging debris on the treated area and very
loose soils throughout most of the site. The resulting census likely undercounted the
number of larval masses. We attempted to count as many as possible as systematically
as possible using approximate 2m transect spacing without completely destroying either
the habitat or ourselves. Butterflies have occasionally been seen in these balds in very
low numbers in the past but most have been observed along the ridge top road several
hundred meters to the east (Ann Potter pers. com.). Nineteen of the masses were
randomly selected for paired microplot vegetation sampling and hemispherical
photographs were taken of each.
The road along the ridge was subsampled and the old log landings/pull outs were
censused. The road is too wide for the 5m x 5m quadrat to straddle and there are few
host plants in the road along much of its length. Therefore the sides of the road were
sampled with one edge along the furthest vegetation into the road and the remainder of
43

the quadrat perpendicular to the road. A random start point in the first 50m of the road
starting on the eastern end at the Washington Department of Natural Resources
property line was used as the origin, and search plots were located every 50m
alternating between the north and south sides of the road giving n=22 search plots.
Measurements were stopped at the beginning of each of the old log landing/pull-outs
and restarted on the other side in order to exclude this separately searched area from
the sample. These areas were searched and host plants were counted but vegetation
was not quantified. Several hours of earlier haphazard searches in what appeared to be
the most likely locations did not turn up any larvae, so characterizing the vegetation was
not especially useful. Systematic overlapping transect searches of host plants in the
four old log landing/pull-out areas provided a thorough census of larval masses in those
areas. The old log landing/pull-outs are anthropogenically flattened wide areas through
which the road passes, now used sporadically as parking spaces and illicit ATV spinout
areas. They have generally sparse vegetation growing up through the gravel. In
addition, we searched and counted all host plants, in this case Plantago lanceolata,
growing in the center of the road.

Dan Kelley Ridge results
Dan Kelley samples contained 57 plant species (Appendix C). Vegetation as a
whole proved to be a poor predictor of oviposition habitat in paired plots with multivariate
examination leading to statistically significant but perhaps biologically trivial results
(MRPP: A-statistic = 0.045, p = 0.007) (Severns and Grosboll 2011). ISA showed
Castilleja hispida and bare ground significantly associated with oviposition (Severns and
Grosboll 2011). Given the large scale shrub control and associated ground disturbance
done to open up habitat specifically in areas with C. hispida, bare ground may be an

44

artifact of workers’ boots disturbing the soil rather than a true indication of oviposition
preference.
The two searched balds yielded 57 oviposition sites of which half were randomly
selected for sampling. When the randomly selected sites were less than 2m apart, one
of them was randomly selected to avoid overlapping samples resulting in n=19 paired
1m x 1m samples.
The road sampling (n = 22) found no larvae but both Castilleja hispida (mean of
5.3 plants per plot) and Plantago lanceolata (mean 80.5 plants per plot) were scattered
along and in the road (n = 487 P. lanceolata in the road were searched). Much of the
road edge is shrub dominated and some sections are shaded making the habitat less
suitable. Censuses of the old log landings were more productive of oviposition sites
sampled and censused respectively. The road edge is used even if rarely (Figure 11).

Figure 11: Taylor’s checkerspot ovipositing on Plantago lanceolata at Dan Kelley Ridge. Her
abdomen is curled under the leaf on which she is perched and methodically laying a clutch of eggs.
This plant was found in an area with a high cover of P. lanceolata in the road verge.

45

“Near Sequim”
―Near Sequim‖ is a privately owned stabilized saltwater shoreline sand dune east
of Port Angeles with a maximum elevation of approximately 5m. The landowners
allowed access to the site for this study subject to certain restrictions. It contains a
typical dune plant community with salt, wind, and summer drought tolerant species. The
site is bordered by natural salt marsh which is surrounded by agricultural fields (primarily
berries), and an occasionally used two track gravel road runs through the middle of the
site. Sampling at this site was stratified to measure the road, the dune, and a mesic
ecotone along a brackish slough bordering the site. This Taylor’s checkerspot
population is unique in its ocean proximity, stabilized dune habitat, and very early flight
period. In the past Taylor’s checkerspot may have been more widespread in this habitat
(Pyle 1989). Most of the dune habitat in Washington and Oregon has been severely
altered by invasive species and development (Wiedemann, Dennis, and Smith 1999)
and other dune and shoreline grassland associated butterflies have declined
precipitously on the Northwestern United States coast (e.g., Oregon silverspot (Speyeria
zerene hippolyta) and sand verbena moth (Copablepharon fuscum)) (Hammond and
McCorkle 1983).
Taylor’s checkerspot is known to use Plantago lanceolata as an oviposition host
on this site but there are other potential hosts present including Triphysaria pusilla and
Collinsia parviflora. Post-diapause larvae have been observed feeding on those hosts
as well (Ann Potter pers. com.). Taylor’s checkerspots have been monitored at this site
for several years but the data have not yet been analyzed. In some years there are at
least several hundred individuals in the adult population.
The goals of the study at ―Near Sequim‖ were to determine the distribution of
oviposition across three identified potential habitat areas: road, dune, and the mesic
ecotone between the dune and a bordering slough.
46

“Near Sequim” sampling methods
The ―Near Sequim‖ site was sampled May 23-25 and June 1-3, 2010. Site
access was limited by the landowner to Tuesday through Thursday 0800-1700. Earlier
site visits indicated that this time frame would likely coincide with full larval webbing
development for most clutches. Adults regularly fly in early April, though during the
study a lone adult female was seen on May 25 but that was clearly an outlier as most
larvae were well into their second instar by then.
Oviposition had in the past been observed along the road that bisects the site but
it was unknown whether the anecdotally observed oviposition was representative of the
distribution of butterflies or the distribution of biologists. The road was sampled by using
the 5m x 5m search plot quadrat centered on the road and located randomly within each
of 16 segments along the road. Between the dune and the deflation plain wetland
behind it is a narrow linear mesic ecotone a bit more than 450m in length. The ecotone
vegetation is clearly distinct from both the dune and wetland and it contained a much
higher density of Plantago lanceolata than the dune. This was sampled (as with the
road) by dividing it up into 50m segments and randomly locating the search plot, with
one side on what appeared to be the edge of the wetland in each segment (n=9). The
dune habitat was sampled by randomly locating a search plot in 50m x 50m grid areas
(n=13) across the site. If a randomly selected location fell within road or ecotone habitat
we skipped that plot. The road and ecotone were intentionally oversampled relative to
the dune habitat because, while the vast majority of the area of the site is dune, the vast
majority of the Plantago lanceolata is in the road and ecotone areas. Collinsia parviflora
was common in the dune habitat but it had nearly all senesced making it unusable as a
pre-diapause host plant. We decided it was more important to quantify the habitat they
were using for oviposition rather than the habitat they were less likely to use. If we had
more time on the site, we would have sampled the dune habitat more intensively.
47

“Near Sequim” Results
―Near Sequim‖ is much different from the previous two sites with obviously lower
levels of biomass and increased bare ground. 48 species were found (Appendix D).
Vegetation as a whole proved a poor predictor of oviposition habitat in paired plots.
A total of 48 larval masses were found in the road habitat and the total number of
larval masses along the road was estimated to be between 275 and 1504 (95% CI). The
number of larvae per mass averaged 38.1 thus giving a total larval population estimate
in the road habitat alone of 10500 to 57300 larvae.
There were more larval masses per search plot in the road habitat than in the
ecotone and more in the ecotone than in the dune habitat (Figure 12). The road
contained areas of dense Plantago lanceolata as did some areas of the ecotone. Most
of the dune habitat was devoid of P. lanceolata but low density (and senescent) Collinsia
parviflora was scattered through many of the search plots

Figure 12: Larval masses per search plot at “Near Sequim” (error bars 95% CI). The road habitat was used
unevenly with some parts receiving high density oviposition and other areas <50m distant receiving none.

48

Volume of Plantago lanceolata per paired 1m^2 plot
(log scale)

100000

10000

1000

100

Oviposition plot

Non-oviposition plot

Figure 13: Paired plots at “Near Sequim”. Volume of Plantago lanceolata estimated by measuring the
height x length x width of each plant in plots. Excludes paired plots without host plants in non-oviposition
plots. Volume of P. lanceolata in the road habitat oviposition plots (mean = 2639 cm3) was not significantly
different (n=11, paired t-test, p=0.303) than in paired adjacent non-oviposition plots (mean = 2220 cm3).
Small sample sizes containing both larvae and host plants prevented paired analysis in Dune (n=1) and
Ecotone (n=3) habitats.

Paired plot results from ―Near Sequim‖ were dramatically different than results
from the three other sites. Oviposition occurred in microsites surrounded by less
hostplant volume and only one 1m2 oviposition plot contained >10,000cm3 P. lanceolata.

49

There was not a significant difference in hostplant volume between oviposition and nonoviposition paired plots (Figure 13)
At the site scale oviposition was associated with higher counts of P lanceolata in
5m x 5m search plots (Figure 14). Within the densest hostplant areas oviposition
density was as high as 26 in one plot.
Because larval masses were both abundant and exposed relative to other sites
where vegetation was more dense, larvae in the masses were counted. Larval masses
(n = 48) along the road were sampled by carefully prodding and unfolding plants and
webbing to count larvae. Assuming each mass webbing was a single mass the mean of
38.1 larvae (95% CI 30.1 - 45.5) is on the high end of reported Euphydryas editha
average clutch size but the data distribution indicated the likelihood that some masses
contained two or three clutches.

Scatter Plot with Fit

30
Linear fit (-5.377
+0.08245x)

25

Larval masses

20

15

10

5

0

-5
40

90

140
# Plantago lanceolata

190

240

Figure 14: Linear regression of the count of larval masses versus number of Plantago lanceolata per 5m x
5m search plot within the road habitat at “Near Sequim” (R2 = 0.36; adjusted R2 = 0.31; slope p = 0.0155).
50

91st Division Prairie
The Pierce County site, 91st Division Prairie, in part the Artillery Impact Area for
Joint Base Lewis-McChord, is a typical South Puget Sound prairie remnant with
Spanaway-Nisqually Gravelly Sandy Loam soils and grassland vegetation. This site is
used for high intensity training by the United States Army and contains perhaps the
highest quality remaining prairie habitat west of the Cascade Range in either
Washington or Oregon along with what is likely the largest population of Taylor’s
checkerspot in the world at Range 76. Portions of the site are concrete fighting
positions, gravel vehicle pads, a mix of wide graded gravel roads, and two track roads.
Sampling at this site was stratified to sample road edges and open prairie. Access to
the site is limited because of military training exercises and, in some areas, the risk of
unexploded ordinance. Castilleja hispida, Collinsia grandiflora, C. parviflora, Plantago
lanceolata, Plectritis congesta, and Triphysaria pusilla occur on the site and prior to this
study only Plantago lanceolata was known as an oviposition host at Range 76 while
Castilleja hispida had been observed as oviposition hosts at Range 51 6km to the west
on the opposite end of 91st Division Prairie.
Taylor’s checkerspot may occur elsewhere in 91st Division Prairie but some areas
are impossible to access because of high densities of unexploded ordinance. A
checkerspot population that occurred at Range 50/51 on the opposite side of this 9,000
acre prairie appears to have been extirpated thus the Range 76 population is the center
of the known Taylor’s checkerspot distribution and perhaps the only currently occupied
portion of 91st Division Prairie.

91st Division Prairie sampling methods
We sampled in Range 76 in the northeast corner of 91st Division Prairie on June
4 and 7-11, 2010. The ideal sampling period may have been 10-15 days later than the
51

available access. Access to this area is often a challenge because it receives heavy
training use by soldiers stationed at Joint Base Lewis-McChord though site availability
was above average for 2010 because many units were deployed overseas. Training
plans for the site are set months in advance which precludes the possibility of adapting
access for research to the weather in a particular year. Checkerspots were still flying
during the sampling period (albeit in low numbers) because the flight period was
extended by historically cold wet weather and many of the eggs had yet to hatch.
Sampling at Range 76 was stratified by habitat. As at ―Near Sequim‖ and Dan
Kelley Ridge, oviposition had been observed in road edges at Range 76. Open prairie,
that also contains of hostplants, surrounds the roads. These two habitats were sampled
as blocks. There are other sub-types of habitat that could have been sampled with
additional time on the range. Of particular interest would be disturbance areas of
various types that often are colonized by Plantago lanceolata.
The two track road was sampled along 400m of its length by delineating it into
50m segments and randomly locating a search plot within each segment. The road was
narrow enough to center search plots on the road and capture both edge and middle
strip vegetation.
Given the limited time available we sampled n=32 search plots in open prairie by
randomly selecting among the 700m long distance monitoring transects used for Taylor’s
checkerspot. We stratified our random selection to choose one transect from each of
three blocks in the sample grid. These transects form a regular grid across the known
checkerspot occupied area but some of them were truncated by the presence of
unexploded ordinance at the time of our sampling. Therefore we had n=14 50m long
segments in transect #12, but only n=10 and n=6 in transects #7 and #3. We sampled
one randomly located search plot along each segment alternating between placing the
search plot north and south of the transect line.
52

In addition to randomized sampling, all Castilleja hispida within Range 76 were
searched for both eggs and larvae. This hostplant is highly visible when flowering
(Figure 15) and rare within Range 76 though it is common at Range 50/51. Most of the
plants are the result of a small direct seeding experiment by the author in 2004.
Results from this site are burdened with the most caveats of any in this study.
Because of access restrictions, the sample period was prior to the peak of observable
larval masses and portions of the site were unavailable for sampling because of access
restrictions caused by unexploded munitions. Together this resulted in fewer samples
collected from a non-random subset of the available habitat. For this reason, results
from this site must be seen as preliminary at best. Approximately 50% of the eggs on
thoroughly searched host plants had not hatched as of the end of the survey period
therefore many potential larval masses were not seen during the sample period.

Figure 15: Castilleja hispida used for oviposition on 91st Division Prairie. This is one of a very few
C. hispida plants at the site. None occurred in search plots but all C. hispida plants at the site were
searched and eggs or larvae were found on 9 of 24 plants.

53

91st Division Results
Sampling at 91st Division Prairie identified 64 plant species (Appendix E) and 10
categorical cover variables in 38 search plots, 30 in open prairie and 8 along a gravel
road. The vast majority of oviposition occurred in road edge habitat (7 of 8 samples) not
in the open prairie (considering only plots containing host plants: Road: 7 of n = 8;
Prairie: 1 of n = 13, two proportions test p = 0.000). This was likely due in part to the
association between P. lanceolata and roads at the site (mean # P. lanceolata in search
plots containing that species: road = 242; open prairie = 63). The single open prairie
oviposition microhabitat paired plot contained more host plant volume than any other
oviposition site detected, indicating that checkerspots can find high density hostplant
sites in open prairie as well. An MRPP test revealed that search plots containing
oviposition were different than search plots not containing oviposition sites (A-statistic =
0.072, p = 0.0046) (Severns and Grosboll 2011) but this was a spurious finding based
on lumping road samples with open prairie samples; when treated separately the
association disappears. MRPP did not indicate a significant difference between paired
microhabitat oviposition versus non-oviposition plots.

54

Volume of Plantago lanceolata per paired
1m^2 plot (log scale)

100000

10000

1000

Oviposition Plot

Non-oviposition plot

Figure 16: Paired plots at 91st Division Prairie. Volume of Plantago lanceolata estimated by measuring the
height x length x width of each plant in plots. Excluding n=1 paired plot sampled outside the road block the
differences between the volume of P. lanceolata in the oviposition plots (mean = 25178 cm^3) was not
significantly different (n=7, paired t-test, p=0.303) than in paired adjacent non-oviposition plots (mean =
21984). Note that no oviposition sites samples contained less than 10000 cm3 P. lanceolata and that when
oviposition occurred in plots with less host plant volume, both plots had >10000 cm3 P. lanceolata.

Among all road edge search plots that contained Plantago lanceolata (n = 8) at
91st Division Prairie, oviposition was very strongly and significantly (Spearman rs = 0.7, p
= 0.026) correlated with the number of P. lanceolata plants but not significant at p < 0.05
within the smaller paired oviposition plots although the association was positive. The
power to detect significant differences was low with n=7 but all oviposition sites
contained >10,000cm3 hostplant volume (Figure 16).

55

The complete census of C. hispida plants included a thorough search for eggs as
well as a search for larvae. These plants have an unusual distribution because most are
growing from seed planted in a line approximately 50m long though a few are scattered
over several acres (Figure 15). Of the 24 plants searched, nine had eggs or larvae
(37%). In contrast, of the 2757 P. lanceolata searched, only 15 had larval masses
(0.54%). While this is not conclusive because the search methods used were different,
and the distribution of C. hispida is unusual, it does strongly suggest the population of
Taylor’s checkerspot at this site may prefer C. hispida to P. lanceolata.

56

V. Summary, synthesis, and recommendations
Taylor’s checkerspots like the rest of Euphydryas editha are maddeningly
complex to understand. Why is this butterfly Endangered – even though it can depend
on a host present on nearly every lawn and roadside in the Northwest, does not require
nectar for reproduction, and is part of a species complex that ranges widely across
western North America? The findings of this thesis suggest that perhaps oviposition site
selection (and thus prediapause larval habitat) is a key, though perhaps a key in a
haystack.
Taylor’s checkerspot populations at the four sites studied are intimately
intertwined with their prediapause hostplants, Castilleja hispida and Plantago lanceolata.
In captivity larvae appear to develop equally well on either host, and in the wild
populations studied used both hostplants when they were available. Other factors play a
role with respect to where checkerspots oviposit but those factors are overwhelmed by
hostplant density. This indicates that the future of this butterfly is utterly dependent on
having hostplants in high enough densities within otherwise suitable habitat.

Hostplant density drives oviposition
Eden Valley and Dan Kelley Ridge are similar habitats and located 1.5 miles
apart. Paired plots at the two sites were analyzed together to determine whether the
biomass density of Castilleja hispida selected for oviposition was similar across the two
sites. The patterns and densities at the two sites were remarkably similar (Figure 8).
Only one of 23 paired oviposition sites (C. hispida plant volume minimum 6624,
maximum 118838; 95% CI +/-11538) occurred in a plot with less than 10,000 cm^3
hostplant volume. The distribution of paired non-oviposition plots overlapped
substantially with oviposition plots overall (C. hispida volume minimum 48, maximum

57

54825; 95% CI +/-7233), but in only five of 23 paired plots was the volume of C. hispida
greater in the non-oviposition plot, and in all samples where oviposition occurred in a
plot with lower C. hispida volume than in the paired non-oviposition plot, both plots had
greater than 10,000cm3 volume of hostplant. Once hostplant volume reached
10,000cm3 only half of eight pairs received oviposition in the higher density plot, which is
consistent with results from 91st Division Prairie where two of five plots where each
member of a pair had >10000cm3 contained oviposition.
Perhaps 10,000 cm3 is a threshold that indicates ―enough‖ hostplant to an
ovipositing female. Ovipositing females may not be able to perceive additional density
beyond 10,000 cm3 per m2 or, as the preference-performance hypothesis (also known as
the ―mother knows best‖ principle) (Jaenike 1978; Gripenberg et al. 2010) would predict,
resources beyond that level may be irrelevant for larval survival because they are in
excess of those needed for optimal development so females will not seek higher
densities. The one site that did not fit this pattern was ―Near Sequim‖ where overall
biomass was obviously lower and only one set of paired plots contained plots with
greater than 10,000cm3 C. hispida volume. With little or no nectar, an exceptionally
early flight period contending with cold, wind, and limited sunlight, oviposition search
flight capacity may be limited and eggs may of necessity be laid on whatever host plants
are encountered. The number of potential oviposition sites containing greater than
10,000 cm3 C. hispida volume on the site is likely very small even if butterflies searched
extensively.
Taylor’s checkerspot and road interactions are strong at ―Near Sequim‖ and 91st
Division Prairie. Even at Eden Valley and Dan Kelley, where roads may be less
important overall, oviposition on P. lanceolata was associated with roads. What this
means for the butterfly is more difficult to ascertain. Taylor’s checkerspot requires a
match between oviposition preferences and larval survival on selected hosts for
58

populations to persist from egg to diapause. Measurements in the field demonstrated
that while oviposition may be driven by multiple factors, hostplant density is by far the
strongest. Captive rearing proved that larvae can survive from hatching through
diapause on both C. hispida and P. lanceolata. While there is much more involved in
recovery of Taylor’s checkerspot than this, clarification of these aspects of the butterfly’s
biology is essential for effective habitat restoration and enhancement.
One overriding factor appears to drive Taylor’s checkerspot oviposition at the
study sites: density of host plants for oviposition. Because the sample areas were all
within extant populations of Taylor’s checkerspot, this finding must be interpreted with
caution. There may very well be significant drivers of oviposition that were not captured
in this study because they do not occur within the currently occupied population areas.
Additionally, enough was known about the habitat requirements of this butterfly that
areas that were known to be unsuitable were excluded from sampling. While each site
is unique and results across sites should be used with caution, some aspects of this
study lend themselves to cross site analysis. The nearly consistent lack of statistically or
biologically meaningful associations between oviposition and most measured habitat
variables was surprising. In part that was due to using study locations that were
obviously, due to the presence of Taylor’s checkerspot populations, broadly suitable
habitat. At all sites additional statistically significant associations might be found with
higher sampling intensity but these associations are likely to be somewhat biologically
weak. An association between host plant density and oviposition is hardly surprising, but
the strength of that association at both the broad site scale and at paired oviposition
microsite sampling was unexpected.
One of the most striking findings of this study was that across three sites with a
similar biomass of herbaceous vegetation, Eden Valley, Dan Kelley Ridge, and 91st
Division Prairie, only one of 31 oviposition events sampled occurred in a plot with less
59

than 10,000 cm3 of host plant volume, but 18 paired non-oviposition plots contained less
than 10,000 cm3. This is particularly interesting from a restoration prescription
perspective because it may indicate a preferred ―threshold level‖ for oviposition.

Slope, aspect, vegetation, and canopy shading
In other populations of Euphydryas editha slope and aspect appear to be
significant drivers of populations but that may not be true for E. e. taylori. The most
intensely studied populations of E. editha occur in central and southern California where,
regardless of slope and aspect, the habitat around checkerspot colonies is grassland
(Weiss 1997; Weiss et al. 1993; Weiss, Murphy, and White 1988). At the two sites with
substantial topographic relief (Dan Kelley Ridge and Eden Valley) all open habitat is flat
or south facing. Because the climate in Western Washington even in the rain shadow of
the Olympic Mountains, where the Clallam County sites are located, is wetter and cooler
than the California sites, slopes that are not south facing invariably become forested and
thus do not provide habitat for Taylor’s checkerspot.
Canopy shading would likely be a controlling factor for oviposition in some
situations, but at the two sites where there were large numbers of trees and shrubs the
results were not statistically significant. The data were suggestive of a positive
correlation between sun exposure and oviposition which would fit with a pattern of
increased activity when adults are in the sun during the typically marginal flight weather
in the Pacific Northwest spring. Because weather varies from year to year, the impact of
shade, slope, and aspect could be highly variable among years.

60

Climate change impacts oviposition and larval survival
Global climate change may have complex effects on E. e. taylori populations.
Parmesan (Parmesan 1996) indicated that global warming was causing a net northward
shift in the distribution of E. editha. This seems manifestly untrue for E. e. taylori, which
has nearly vanished from the northern end of its range, though evidence from other
butterfly species supports this general pattern (Warren et al. 2001). There is evidence
that increases in climate variability can negatively affect E. editha (McLaughlin et al.
2002) and that this is particularly true in homogenous habitats (McLaughlin et al. 2002).
Local populations of a species may be less fit throughout its range in some cases rather
than showing a positive response to warming at the northern limits and a negative
response at the southern limits (Hellmann 2004). Climate change may make extirpations
more likely (McLaughlin et al. 2002) and extirpations are even more likely when the
population depends on only one host (Hellmann 2002).
Data on regional climate change is coarse relative to the scale of available
Taylor’s checkerspot habitat. Even fine scale regional climate change models are
generally in the 15km grid range (Salathe Jr. et al. 2008). All currently occupied Taylor’s
checkerspot habitat in British Columbia, Washington, and Oregon adds up to less than
the 225km2 in one grid and in Washington alone occurs in three rare specialized habitat
―types‖ (stabilized dune, grassy bald, and Puget Sound grassland) so evaluating the
effects of climate change is somewhat subjective and qualitative. Nonetheless some
aspects of climate change can be considered for its impacts on oviposition sites. All
sites reported in this thesis will be impacted by changes to temperature and precipitation
averages and patterns and one site may be inundated by sea level rise.
Changes in mean precipitation and temperature would impact Taylor’s
checkerspot oviposition and larval development. The CCSM3-WRF and ECHAM5-WRF
models indicate relatively little change in mean precipitation between 1970-1999
61

measurements and 2030-2059 projections with the exception of an agreed increase of a
1-2cm/month decrease in spring precipitation under CCSM3-WRF and an increase in
spring precipitation under ECHAM5-WRF (Salathe Jr. et al. 2008). If there is a decrease
in spring precipitation as modeled in CCSM3-WRF, and as both models predict, and
average temperatures increase throughout the year, then hostplant senescence could
arrive earlier in the butterfly life cycle which would decrease survival between egg and
diapause. Even though Taylor’s checkerspot would likely emerge from diapause,
develop, and pupate earlier in response to increased winter temperatures this may or
may not be early enough to oviposit with enough time for larvae to develop (Hellmann
2002).
Globally, climate changes are projected to increase variability of both
temperature and precipitation; however, within the range of Taylor’s checkerspot,
models show limited change in variability of either. Models agree on increasing
variability in both temperature and precipitation in much of the Pacific Northwest but
Taylor’s checkerspots are located in the Puget Trough and in the rain shadow of the
Olympic Mountains where little change is predicted (Salathe Jr. et al. 2008). This is
encouraging for conservation because increases in variability can drive populations to
extinction (McLaughlin et al. 2002).
The most recent Assessment Report of the Intergovernmental Panel on Climate
Change (IPCC) projected global sea level rise in this century will range between 18cm
and 59cm depending on the emissions scenario used, and sea level in Washington is
expected to closely follow this trend (Mote et al. 2008). ―Apparent sea level change‖ is
the difference between change in land level and change in sea level. Deformation of the
earth’s crust is causing the area in which ―Near Sequim‖ is located to rise approximately
1.5mm/year (Verdonck 2006) but it is possible that local factors (e.g., subsidence due to
sediment compaction or dune movement) could change that on the site scale. At the
62

lowest projected rate of sea level rise under the lowest emissions scenario, local crustal
deformation (15cm/100years) nearly equals the rate of sea level rise (18cm/100 years)
and the apparent rise would be relatively trivial. Even at the lowest level (26cm) of the
high emissions scenario projections, additional erosion may impact this narrow strip of
habitat while at the higher end of high emission projections the ―Near Sequim‖ site will
likely be periodically overtopped by storm waves with catastrophic consequences for the
Taylor’s checkerspots on the site.

Future directions for research and lessons learned
This study attempted to measure habitat where eggs were laid to understand
what parameters were critical for egg laying by females and thus where prediapause
larvae would develop. It found that hostplant density was a strong predictor of egg
laying but that many other aspects of the habitat were highly variable, at least within
occupied checkerspot habitat. A substantial hurdle was finding enough oviposition sites
for statistical analysis while using statistically defensible sampling methods. Measuring
habitat only within areas containing hostplants would increase the number of
observations without losing significant information.
Following females through habitats to observe where individuals lay eggs (along
with other spatially explicit behavioral observations) is effective for species that oviposit
frequently (Beyera and Schultz 2010) but less effective when following a butterfly like
Taylor’s checkerspot that may only lay one clutch of eggs per day and is quite cryptic
when laying eggs. Given the impossibility of following butterflies across the steep
landscapes where some Taylor’s checkerspots are found, other methods must be used.
Sequential hostplant testing is effective for determining preferences that individuals have
for particular plants (Singer 2004) but less effective at determining where, across the

63

occupied area, oviposition occurs. Sampling efficiency is critical for detecting enough
oviposition sites for analysis.
Captive rearing methods are being applied concurrently with habitat restoration,
enhancement, and invasive plant management, which will ensure that there are habitats
available for reintroduction. The outlook for restoration is thus promising because,
based on this research, it appears that substantial habitat can be restored simply by
increasing hostplant density within occupied (or potentially occupied) habitats.
Understanding the oviposition preferences, larval survival and performance on different
hostplants, and development of methods for enhancing hostplant resources, are
essential elements in the recovery of the endangered Taylor’s checkerspot butterfly.
Further the context of climate change is critical for understanding future habitat
availability. Without this knowledge, and application of this knowledge in restoration, the
butterfly is likely to go extinct in the near future.
Humans have likely created habitats for this butterfly for thousands of years and
that story—both the science and the culture—is an integral part of who we were and who
we are. Perhaps surprisingly, that story of human habitat modification is also part of the
genome of Taylor’s checkerspot: its use of European introduced hostplants and its
behavioral affinity for grasslands and balds maintained by Native use of fire. The author
hopes the work performed in this thesis will not only increase knowledge about Taylor’s
checkerspot but that it will also lead directly to better management decisions in its
habitat.

64

Literature Cited
Anderson, Mary Jo, Blair Csuti, Nicol Grol, David Shepherdson, and Jan Steele. 2001.
Oregon silverspot butterfly (Speyria zerene hippolyta) Cascade Head population
supplementation project. Portland, OR: Oregon Zoo.
Austin, George T., and Dennis D. Murphy. 1998. Euphydryas editha of the great basin,
with descriptions of three new subspecies (Lepidoptera: Nymphalidae). In
Systematics of western North American butterflies, edited by T. C. Emmel.
Gainsville, FL: Mariposa Press.
Baughman, David A. 1999. Empirical evidence for complex source-sink dynamics with
alternative states in a butterfly metapopulation. Ecology 80 (8):2727-2739.
Baughman, J.F., D.D. Murphy, and P.R. Ehrlich. 1988. Population structure of a
hilltopping butterfly. Oecologia 75:593-600.
Baughman, John F., and Dennis D. Murphy. 1998. Differentiation in a widely distributed,
polytypic butterfly genus: five new subspecies of California Euphydryas
(Lepidoptera: Nymphalidae). In Systematics of western North American
butterflies, edited by T. C. Emmel. Gainsville, FL: Mariposa Press.
Beyera, Loni J., and Cheryl B. Schultz. 2010. Oviposition selection by a rare grass
skipper Polites mardon in montane habitats: Advancing ecological understanding
to develop conservation strategies. Biological Conservation 143 (4):862-872.
Black, Scott Hoffman, Matthew Shepard, and Melody Mackey Allen. 2001. Endangered
invertebrates: the case for greater attention to invertebrate conservation.
Endangered Species UPDATE 18 (2):42-50.
Boggs, Carol, L. 1997. Reproductive allocation from reserves and income in butterfly
species with differing adult diets. Ecology 78 (1):181-191.
Boisduval, J.A. 1852. Lepidopteres de la California. Annales de la Sociaetae
Entomologique de France 10 (2):275-324.
Boughton, David, A. 1999. Empirical evidence for complex source-sink dynamics with
alternative states in a butterfly metapopulation. Ecology 80 (8):2727-2739.
Boyd, Robert. 1999b. Strategies of Indian burning in the Willamette Valley. In Indians,
fire and the land in the Pacific Northwest, edited by R. Boyd. Corvallis, OR:
Oregon State University Press.
Britten, H.B., P.F. Brussard, D.D. Murphy, and P.R. Ehrlich. 1995. A test for isolation-bydistance in central Rocky mountain and Great Basin populations of Edith's
checkerspot butterfly (Euphydryas editha). Journal of Heredity 86:204-210.
Brook, Barry W., David W. Tonkyn, Julian J. O'Grady, and Richard Frankham. 2002.
Contribution of inbreeding to extinction risk in threatened species. Conservation
Ecology [online] 6 (1):16.
Bryant, Edwin H., Vickie L. Backus, Micheal E. Clark, and David H. Reed. 1999.
Experimental tests of captive breeding for endangered species. Conservation
Biology 13 (6):1487-1496.
Caplow, Florence. 2004. Reintroduction Plan for Golden Paintbrush (Castilleja
levisecta). US Fish and Wildlife Service, Western Washington Fish and Wildlife
Office, through Section 6 funding, Region 1; prepared by Washington Natural
Heritage Program, Washington Department of Natural Resources, PO Box
47014, Olympia, WA 98504-7014.
Carpenter, Cecelia Svinth. 2002. The Nisqually, My People: The Traditional and
Transitional History of the Nisqually Indian People. Orting, Washington: Tahoma
Research Service.

65

Char, Patti, and P. Dee Boersma. 1995. The effects of prairie fragmentation on butterfly
species in western Washington. Seattle: University of Washington.
Crawford, Rex C., and Heidi Hall. 1997. Changes in the south Puget prairie landscape.
In Ecology and conservation of the south Puget Sound prairie landscape, edited
by P. V. Dunn and K. Ewing. Seattle, WA: The Nature Conservancy.
Cushman, J. Hall, L. Boggs Carol, B. Weiss Stuart, D. Murphy Dennis, W. Harvey Alan,
and R. Ehrlich Paul. 1994. Estimating female reproductive success of a
threatened butterfly: Influence of emergence time and hostplant phenology.
Oecologia, Berlin 99 (1-2):194-200.
Dornfeld, Ernst J. 1980. Butterflies of Oregon. Forest Grove, Oregon: Timber Press.
Duffy, Eric. 1977. The re-establishment of the large copper butterfly Lycaena dispar
batava Obith. on Woodwalton Fen National Nature Reserve, Cambridgeshire,
England, 1969-73. Biological Conservation 12:143-158.
Dufrene, Marc, and Pierre Legendre. 1997. Species assemblages and indicator
species: the need for a flexible asymmetrical approach. Ecological Monographs
67 (3):345-366.
Dunn, Patrick. 1998. Prairie habitat restoration and maintenance on Fort Lewis and
within the South Puget Sound prairie landscape: Final report and summary of
findings. Seattle, WA: The Nature Conservancy of Washington.
Ehrlich, P.R. 1992. Population biology of checkerspot butterflies and the preservation of
global biodiversity. Oikos 63:6-12.
Ehrlich, Paul R. 1961. Intrinsic barriers to dispersal in checkerspot butterfly. Science
134:108-109.
Ehrlich, Paul R., and Ilkka Hanski, eds. 2004. On the wings of checkerspots: a model
system for population biology. Oxford: Oxford University Press.
Ehrlich, Paul R., and Dennis D. Murphy. 1987. Conservation lessons from long-term
studies of checkerspot butterflies. Conservation Biology 1 (2):122-131.
———. 1987. Monitoring populations on remnants of native vegetation. In Nature
conservation: the role of remnants of native vegetation, edited by D. A.
Saunders, G. W. Arnold, A. A. Burbidge and A. J. M. Hopkins: Surrey Beatty &
Sons Pty Limited.
Fleckenstein, John, and Ann Potter. 1999. 1997, 1998 project summary. Olympia, WA:
Natural Heritage Program Washington Department of Natural Resources and
Wildlife Diversity Division Washington Department of Fish and Wildlife.
Gripenberg, Sofia, Peter J. Mayhew, Mark Parnell, and Tomas Roslin. 2010. A metaanalysis of preference-performance relationships in phytophagous insects.
Ecology Letters 13:383-393.
Grosboll, Daniel N. 2004. Captive rearing the endangered mardon skipper (Polites
mardon) and Taylor's checkerspot (Euphydryas editha taylori) butterflies: initial
results (Lepidoptera, Nymphalidae). Paper read at Species at Risk 2004:
Pathways to Recovery, at Victoria, BC.
Guppy, Crispin S., Norbert G. Kondla, and Lee Schaeffer. 2003. Draft national recovery
strategy for Taylor's checkerspot Euphydryas editha taylori (W.H. Edwards).
Victoria, BC: British Columbia Ministry of Water, Land, and Air Protection.
Guppy, Crispin S., and Jon H. Shepard. 2001. Butterflies of British Columbia: including
western Alberta, southern Yukon, the Alaska panhandle, Washington, northern
Oregon, northern Idaho, and northwestern Montana. Vancouver, BC: University
of British Columbia Press.

66

Hammond, Paul C., and David V. McCorkle. 1983. The decline and extinction of
Speyeria populations resulting from human environmental disturbances
(Nymphalidae: Argynninae). Journal of Research on the Lepidoptera 22 (4):217224.
Harrison, Susan. 1989. Long-distance dispersal and colonization in the bay checkerspot
butterfly, Euphydryas editha bayensis. Ecology 70 (5):1236-1243.
Harrison, Susan, Dennis D. Murphy, and Paul R. Ehrlich. 1988. Distribution of the bay
checkerspot butterfly, Euphydryas editha bayensis: evidence for a
metapopulation model. The American Naturalist 132 (3):360-382.
Hays, David W., Ann E. Potter, Christopher W. Thompson, and Patrick V. Dunn. 2000.
Critical habitat components for four rare south Puget Sound butterflies.
Washington Department of Fish and Wildlife/The Nature Conservancy.
———. 2000. Critical habitat components for four rare south Puget Sound butterflies.
Washington Department of Fish and Wildlife/The Nature Conservancy.
Heath, Daniel D., John W. Heath, Colleen A. Bryden, Rachel M. Johnson, and Charles
W. Fox. 2003. Rapid Evolution of Egg Size in Captive Salmon. Science
299:1738-1740.
Hellmann, Jessica J. 2002. The effect of an environmental change on mobile butterfly
larvae and the nutritional quality of their hosts. Journal of Animal Ecology 71
(6):925-936.
———. 2004. Impacts of climate change on butterfly populations in western North
America. Paper read at Species at Risk 2004: Pathways to Recovery, at Victoria,
BC.
Hodges, Ronald W., Tatiana Dominick, Donald R. Davis, Douglas C. Ferguson, John G.
Franclemont, Eugene G. Munroe, and Jerry A. Powell, eds. 1983. Check list of
the Lepidoptera of America north of Mexico including Greenland. London: E.W.
Classey Limited.
Holland, W. J. 1898. The butterfly book. New York: Doubleday.
Jaenike, J. 1978. On optimal oviposition behaviour in phytophagous insects. Theoretical
Population Biology 14:350-356.
Konvicka, Martin, Vladimir Hula, and Zdenek Fric. 2003. Habitat of pre-hibernating
larvae of the endangered butterfly Euphydryas aurinia (Lepidoptera:
Nymphalidae): What can be learned fom vegetation composition and
architecture? European Journal of Entomology 100:313-322.
Kuussaari, Mikko, Saskya Van Nouhuys, Jessica J. Hellmann, and Micheal C. Singer.
2004. Larval biology of checkerspots. In On the wings of checkerspots: a model
system for population biology, edited by P. R. Ehrlich and I. Hanski. Oxford:
Oxford University Press.
Launer, Alan E., and Dennis D. Murphy. 1994. Umbrella species and the conservation of
habitat fragments: a case of a threatened butterfly and a vanishing grassland
ecosystem. Biological Conservation 69:145-153.
Leonard, William P., and Lisa Hallock. 1997. Herpetofauna of south Puget Sound prairie
landscape. In Ecology and conservation of the south Puget Sound prairie
landscape, edited by P. V. Dunn and K. Ewing. Seattle, WA: The Nature
Conservancy.
Leopold, Estella B., and Robert Boyd. 1999. An ecological history of old prairie areas in
southwestern Washington. In Indians, fire and the land in the Pacific Northwest,
edited by R. Boyd. Corvallis, OR: Oregon State University Press.

67

Linders, Mary J. 2007. Development of captive rearing and translocation methods for
Taylor’s checkerspot (Euphydryas editha taylori) in south Puget Sound,
Washington: 2006-2007 Annual Report. Olympia, WA: Washington Department
of Fish and Wildlife, Wildlife Program.
Lipman, Alison, Travis R. Longcore, Rudi Mattoni, and YinLan Zhang. 1999. Habitat
Evaluation and Reintroduction Planning for the Endangered Palos Verdes Blue
Butterfly. Los Angeles: University of California, Los Angeles- Geography
Department.
McLaughlin, John F., Jessica J. Hellmann, Carol L. Boggs, and Paul R. Ehrlich. 2002.
Climate change hastens population extinctions. Proceedings of the National
Academy of Science 99 (9):6070-6074.
———. 2002. The route to extinction: population dynamics of a threatened butterfly.
Oecologia 132:538-548.
Mielke, Paul W. . 1984. Meteorological applications of permutation techniques based on
distance functions. In Handbook of statistics: Vol. 4. Nonparametric methods,
edited by P. R. Krishnaiah and P. K. Sen. Amsterdam, North-Holland: Elsevier
Science & Technology Books.
MINITAB Statistical Software 13.32. MINITAB Incorporated, State College, PA.
Mote, Philip, Alexander Petersen, Spencer Reeder, Hugh Shipman, and Lara Whitely
Binder. 2008. Sea Level Rise in the Coastal Waters of Washington State.
University of Washington Climate Impacts Group and the Washington
Department of Ecology.
Murphy, Dennis D., Nicklas Wahlberg, Ilkka Hanski, and Paul R. Ehrlich. 2004.
Introducing checkerspots: taxonomy and ecology. In On the wings of
checkerspots: a model system for population biology, edited by P. R. Ehrlich and
I. Hanski. Oxford: Oxford University Press.
New, T. R. 1997. Butterfly conservation. 2nd ed. Oxford: Oxford University Press.
Nieminen, Marko, Michael C. Singer, Wilhelm Fortelius, Katrin Schops, and Ilkka Hanski.
2001. Experimental confirmation that inbreeding depression increases extinction
risk in butterfly populations. The American Naturalist 157 (2):237-244.
Oates, M. R., and M.S. Warren. 1990. A review of butterfly introductions in Britian and
Ireland. Joint Committee for the Conservation of British Insects.
Page, Nick, Patrick Lilley, Jennifer Heron, Nicole Kroeker, and Report prepared for
Agency by Raincoast Appllied Ecology. v + 32 pp. 2009. Distribution and Habitat
Characteristics of Taylor’s Checkerspot on Denman Island and Adjacent Areas of
Vancouver Island (2008). B.C. Ministry of Environment and Parks Canada
Parmesan, Camille. 1996. Climate and species' range. Nature 382:765-766.
Parmesan, Camille, Michael C. Singer, and Ian Harris. 1995. Absence of adaptive
learning from the oviposition foraging behaviour of a checkerspot butterfly.
Animal Behaviour 50 (1):161-175.
Pelham, Johnathan P. 2008. A Catalogue of the Butterflies of the United States and
Canada: with a complete bibliography of the descriptive and systematic literature.
Vol. 40, The Journal of Research on the Lepidoptera. Beverly Hills, CA: The
Lepidoptera Research Foundation, Inc.
Pollard, E., and T. J. Yates. 1993. Monitoring butterflies for ecology and conservation:
the British butterfly monitoring scheme. Edited by F. B. Goldsmith and E. Duffy,
Conservation Biology Series. London: Chapman and Hall.
Pullin, Andrew S. 1997. Habitat requirements of Lycaena dispar batavus and
implications for re-establishment in England. Journal of Insect Conservation
1:177-185.
68

Pyle, R. M. 1988. Captive breeding of butterflies for conservation. Paper read at 5th
World Conference on Breeding Endangered Species in Captivity, at Cincinnati,
OH.
Pyle, Robert Micheal. 1989. Washington Butterfly Conservation Status Report and Plan.
Gray's River, WA: Washington Department of Fish and Wildlife.
———. 2002. The butterflies of Cascadia. Seattle, WA: Seattle Audubon Society.
Rausher, Mark D. 1982. Population differentiation in Euphydryas editha butterflies: larval
adaptation to different hosts. Evolution 36 (3):581-590.
Rausher, Mark D., Duncan A. Mackay, and Micheal C. Singer. 1981. Pre- and postalighting host discrimination by Euphydryas editha butterflies: the behavioral
mechanisms causing clumped distributions of egg clusters. Animal Behavior
29:1220-1228.
Remsberg, Mary. 2000. Summary report of LCTA butterfly surveys conducted on Fort
Lewis, WA. Fort Lewis, WA: United States Army.
Rogers, Russell E. 2000. The status and micro-habitat selection of streaked horned lark,
western bluebird, Oregon vesper sparrow, and western meadowlark in western
Washington. Masters Thesis, Master of Environmental Studies Program, The
Evergreen State College, Olympia.
Rogers, Russell, Donald Norman, and David Rolph. 1997. The status of neotropical
migrant birds in the prairie landscape. In Ecology and Conservation of the south
Puget sound prairie landscape, edited by P. V. Dunn and K. Ewing. Seattle, WA:
The Nature Conservancy.
Ross, Dana N. R. 2003. Report to the Xerces Society: First Year (2003) Census of the
Cardwell Hill Population of Taylor's Checkerspot (Euphydryas editha taylori) in
Benton County, Oregon. Corvallis, OR: Xerces Society.
Ryan, Loreen A. 1997. Ecology of the western gray squirrel in the south Puget Sound. In
Ecology and Conservation of the south Puget Sound prairie landscape, edited by
P. V. Dunn and K. Ewing. Seattle, WA: The Nature Conservancy.
Saccheri, Ilik J., Ian J. Wilson, Richard A. Nichols, Michael W. Bruford, and Paul M.
Brakefield. 1999. Inbreeding of Bottlenecked Butterfly Populations: Estimation
Using the Likelihood of Changes in Marker Allele Frequencies. Genetics 151
(3):1053-1063.
Salathe Jr., Eric P. , Richard Steed, Clifford F. Mass, and Patrick H. Zahn. 2008. A HighResolution Climate Model for the U.S. Pacific Northwest: Mesoscale Feedbacks
and Local Responses to Climate Change. Journal of Climate 21:5708-5726.
Samways, Michael J. 1994. Insect conservation biology. Edited by F. B. Goldsmith and
E. Duffy, Conservation Biology Series. London: Chapman and Hall.
Scott, James A. 1986. The butterflies of North America. Stanford: Stanford University
Press.
Severns, P. M., and A. D. Warren. 2008. Selectively eliminating and conserving exotic
plants to save an endangered butterfly from local extinction. Animal Conservation
(11):476-483.
Severns, Paul M., and Dan Grosboll. 2011. Patterns of reproduction in four Washington
State populations of Taylor’s checkerspot (Euphydryas editha taylori) during the
spring of 2010. The Nature Conservancy.
Singer, M. C., D. Vasco, C. Parmesan, C. D. Thomas, and D. Ng. 1992. Distinguishing
between 'preference' and 'motivation' in food choice: an example from insect
oviposition. Animal Behavior 44:463-471.
Singer, Michael C. 2004. Oviposition preference: its measurement, its correlates and its
importance in the life of the checkerspot. In On the wings of checkerspots, edited
by P. Ehrlich and I. Hanski. Oxford: Oxford University Press.
69

Singer, Micheal C., and Paul R. Ehrlich. 1979. Population dynamics of the checkerspot
butterfly Euphydryas editha. Fortschr. Zool. 25:53-60.
Singer, Micheal C., D. Ng, and R. A. Morre. 1991. Genetic variation in oviposition
preference between butterfly populations. Journal of Insect Behavior 4 (4):531535.
Singer, Micheal C., D. Ng, and C. D. Thomas. 1988. Heritability of oviposition preference
and its relationship to offspring performance within a single insect population.
Evolution 42 (5):977-985.
Singer, Micheal C., Chris D. Thomas, and Camille Parmesan. 1993. Rapid humaninduced evolution of insect-host associations. Nature 366:681-683.
Stinson, Derek W. 2005. Washington State Status Report for the Mazama Pocket
Gopher, Streaked Horned Lark, and Taylor’s Checkerspot. Olympia: Washington
Department of Fish and Wildlife.
Thomas, Chris D., Micheal C. Singer, and David A. Boughton. 1996. Catastrophic
extinction of population sources in a butterfly metapopulation. The American
Naturalist 148 (6):957-975.
Turner, Nancy J. 1999. "Time to burn:" Traditional use of fire to enhance resource
production by aboriginal peoples in British Columbia. In Indians, fire and the land
in the Pacific Northwest, edited by R. Boyd. Corvallis, OR: Oregon State
University Press.
Verdonck, David. 2006. Contemporary vertical crustal deformation in Cascadia.
Technophysics 417:221-230.
Warren, Andrew D. 2005. Butterflies of Oregon: their taxonomy, distribution, and biology.
In Lepidoptera of North America 6. Fort Collins, CO: Colorado State University,
C.P. Gillette Museum.
Warren, M. S., J. K. Hill, J. A. Thomas, J. Asher, R. Fox, B. Huntley, D. B. Roy, M. G.
Telfer, S. Jeffcoate, P. Harding, G. Jeffcoate, S. G. Willis, J. N. GreatorexDavies, D. Moss, and C. D. Thomas. 2001. Rapid responses of British butterflies
to opposing forces of climate and habitat change. Nature 414 (6859):65-69.
Weiss, S. B. 1997. Weather, phenology, and the population ecology of a threatened
checkerspot butterfly in complex terrain. Bulletin of the Ecological Society of
America 78 (4 SUPPL.):205.
Weiss, Stuart B. 1999. Cars, cows, and checkerspot butterflies: nitrogen deposition and
management of nutrient-poor grasslands for a threatened species. Conservation
Biology 13 (6):1476-1486.
Weiss, Stuart B., Dennis D. Murphy, Paul R. Ehrlich, and Charles F. Metzler. 1993. Adult
emergence phenology in checkerspot butterflies: the effects of macroclimate,
topoclimate, and population history. Oecologia 96:261-270.
Weiss, Stuart B., Dennis D. Murphy, and Raymond R. White. 1988. Sun, slope, and
butterflies: topographic determinants of habitat quality for Euphydryas editha.
Ecology 69 (5):1486-1496.
White, R. R. 1974. Food plant defoliation and larval starvation of Euphydryas editha.
Oecologia 14:307-315.
White, Raymond R., and Micheal C. Singer. 1974. Geographical distribution of hostplant
choice in Euphydryas editha (Nymphalidae). Journal of the Lepidopterists'
Society 28 (2):103-107.
White, Richard. 1980. Land use, environment, and social change: the shaping of Island
County, Washington. Seattle, WA: University of Washington Press.
Wiedemann, Alfred M., La Rea J. Dennis, and Frank Smith. 1999. Plants of the Oregon
coastal dunes. 2nd ed. Corvallis: Oregon State University Press.
70

Zar, Jerrold H. 1999. Biostatistical analysis. 4th ed. Upper Saddle River, NJ: Prentice
Hall.
Zhang, Yongqin, Jing M. Chen, and John R. Miller. 2005. Determining digital
hemispherical photograph exposure for leaf area index estimation. Agricultural
and Forest Meteorology 133:166-181.

71

Appendix A: Hemispherical photography.

Figure 17: Hemispherical photo of typical steep south facing slope at Eden Valley. Exposure method
designed to be used with cloud cover or pre-sunrise sky makes the sky look nearly white for high levels of
contrast with surrounding obstructions.

72

Figure 18: Manually thresholded image of Figure 17. When thresholded this becomes a black and white
record of sky obstruction. When the image from this search plot was processed through Gap Light Analysis
60% of the canopy is open but that underestimates the amount of sunlight the plot gets because of its slope,
aspect and where trees block the sun. Total direct sunlight reaches this plot is 90.4% of the time. North is at
the top of photo.

73

Appendix B: List of plants identified at Eden Valley.
Acer macrophyllum
Achillea millefolium
Agrostis sp.
Aira caryophyllea
Aira praecox
Amelanchier alnifolia
Apocynum sp.
Arabis sp.
Arrhenatherum elatius
Balsamorhiza deltoidea
Berberis aquifolium
Berberis nervosa
Brodiaea congesta
Bromus carinatus
Bromus mollis
Castilleja hispida
Carex inops
Cerastium arvense
Cerastium glomeratum
Clarkia amoena
Claytonia perfoliata
Corallorhiza maculata
Cytisus scoparius
Dactylis glomerata
Danthonia californica
Daucus carota
Delphinium menziesii
Elymus glaucus
Epilobium sp.
Eriophyllum lanatum
Festuca occidentalis?
Festuca roemeri
Festuca rubra
Festuca subuliflora
Fragaria virginiana
Fritillaria affinis
Galium aparine
Habenaria sp.
Heuchera spp
Holcus lanatum
Holodiscus discolor
Hypericum perforatum
Hypochaeris radicata
Koehleria macrantha
Lathyrus sp.
Lomatium utriculatum
Lonicera ciliosa
Lotus micranthus

Luzula campestris
Madia sp.
Microseris laciniata
Microsteris gracilis
Moehringia macrophylla
Montia parviflora
Myosotis discolor
Nemophila parviflora
Osmorhiza purpurea
Pentagramma triangularis
Plectritis congesta
Poa bulbosa
Poa spp.
Polystichum imbricans
Polystichum munitum
Prunella vulgaris
Prunus avium
Pseudotsuga menziesii
Ranunculus uncinatus
Rosa nutkana
Rubus ursinus
Rumex acetosella
Satureja douglasii
Sedum spathulifolium
Stellaria media
Symphoricarpos albus
Trifolium spp.
Trifolium repens
Trifolium wormskjoldii
Veronica spp.
Vicia hirsuta
Vicia sativa
Vulpia spp
Whipplea modesta
Zigadenus venenosus

74

Appendix C: List of plants identified at Dan Kelley Ridge.
Acer macrophyllum
Aira caryophyllea
Aira praecox
Anaphalis magaritacea
Arbutus menziesii
Arrhenatherum elatius
Berberis aquifolium
Berberis nervosa
Bromus carinatus
Bromus mollis
Bromus sp.
Castilleja hispida
Carex inops
Cerastium arvense
Cerastium glomeratum
Cirsium edule
Clarkia amoena
Daucus carota
Elymus glaucus
Epilobium sp.
Erodium cicutarium
Festuca rubra
Fragaria virginiana
Fritillaria affinis
Galium aparine
Gaultheria shallon
Geranium spp.
Heuchera spp
Hieracium cynoglossoides

Hieracium sp
Holodiscus discolor
Hypochaeris radicata
Lathyrus sp.
Leucanthemum vulgare
Luzula campestris
Microseris laciniata
Myosotis discolor
Osmorhiza purpurea
Penstemon sp.
Philadelphus lewisii
Plectritis congesta
Polystichum munitum
Prunella vulgaris
Prunus emarginata
Pseudotsuga menziesii
Ranunculus uncinatus
Ribes sanguineum
Rosa nutkana
Rubus discolor
Rubus ursinus
Satureja douglasii
Symphoricarpos albus
Trifolium spp.
Vicia hirsuta
Vulpia spp
Whipplea modesta
Zigadenus venenosus

75

Appendix D: List of vascular plants identified at ―Near Sequim‖.
Abronia latifolia
Achillea millefolium
Agrostis spp
Aira caryophyllea
Aira praecox
Ambrosia chamissonis
Artemesia campestris
Aster spp.
Beach grass
Bromus mollis
Bromus rigidus
Bromus tectorum
Cerastium glomeratum
Cerastium nutans
Cirsium arvense
Cirsium vulgare
Collinsia parviflora
Crataegus sp.
Dactylis glomerata
Epilobium sp
Erodium cicutarium
Festuca rubra
Festuca/Lolium pratensis
Galium aparine

Grindelia
Holcus lanatum
Hordeum spp.
Hypochaeris radicata
Juncus spp.
Lathyrus nevadensis
Lotus spp.
Matricaria chamomilla
Orthocarpus pusillis
Plantago lanceolata
Poa bulbosa
Poa spp.
Polygonum paronychia
Potentilla anserina pacifica
Rosa nutkana
Rumex acetosella
Sagina maxima
Sonchus arvensis
Taraxacum officinale
Trifolium dubium
Trifolium oliganthum
Vicia hirsuta
Vicia sativa
Vulpia spp

76

Appendix E: List of vascular plants identified at 91st Division Prairie.
Achillea millefolium
Agrostis diegoensis
Aira caryophyllea
Aira praecox
Anthoxanthum aristatum
Anthoxanthum odoratum
Apocynum sp.
Arrhenatherum elatius
Balsamorhiza deltoidea
Bromus carinatus
Bromus mollis
Camassia quamash
Campanula rotundifolia
Carex inops
Cerastium spp
Cytisus scoparius
Danthonia spicata
Agropyron canina
Erigeron speciosus
Eriophyllum lanatum
Erodium cicutarum
Festuca roemeri
Filago arvensis
Fragaria virginiana
Fritillaria camschatcensis
Galium aparine
Holcus lanatum
Hieracium cynoglossoides
Hypochaeris radicata
Hypericum perforatum
Koehleria macrantha
Leucanthemum vulgare

Lomatium triternatum
Lomatium utriculatum
Lotus micranthus
Lupinus albicaulis
Lupinus bicolor
Lupinus lepidus
Luzula campestris
Microseris laciniata
Microsteris gracilis
Orthocarpus pusillis
Panicum spp.
Parentucellia viscosa
Plantago lanceolata
Poa spp.
Potentilla gracilis
Potentilla recta
Pteridium aquilinum
Ranunuculus occidentalis
Rubus sp.
Rumex acetosella
Saxifraga integrifolia
Solidago spp.
Spergularia spp.
Spiranthes romanzoffiana
Teesdalia nudicaulis
Tragopogon dubius
Trifolium arvense
Trifolium dubium
Vicia hirsuta
Vicia sativa
Vulpia spp
Zigadenus venenosus

77

Item sets

The Evergreen State College Masters Theses: Environmental Studies

Media

Grosboll_DMESthesis2011.pdf