Interactions of Erysimum Capitatum with Pollinators and Herbivores Over Space and Time Claire Ruth Lay University of Colorado at Boulder, Erysimumcap@Gmail.Com

Home , Erysimum, Erysimum capitatum

University of Colorado, Boulder CU Scholar Ecology & Evolutionary Biology Graduate Theses & Ecology & Evolutionary Biology Dissertations

Spring 1-1-2011 The good, the bad and the flexible: interactions of Erysimum capitatum with pollinators and herbivores over space and time Claire Ruth Lay University of Colorado at Boulder, [email protected]

Follow this and additional works at: http://scholar.colorado.edu/ebio_gradetds Part of the Biology Commons, and the Ecology and Evolutionary Biology Commons

Recommended Citation Lay, Claire Ruth, "The oodg , the bad and the flexible: interactions of Erysimum capitatum with pollinators and herbivores over space and time" (2011). Ecology & Evolutionary Biology Graduate Theses & Dissertations. Paper 19.

This Dissertation is brought to you for free and open access by Ecology & Evolutionary Biology at CU Scholar. It has been accepted for inclusion in Ecology & Evolutionary Biology Graduate Theses & Dissertations by an authorized administrator of CU Scholar. For more information, please contact [email protected].

THE GOOD, THE BAD AND THE FLEXIBLE: INTERACTIONS OF ERYSIMUM CAPITATUM WITH POLLINATORS AND HERBIVORES OVER SPACE AND TIME by CLAIRE RUTH LAY B.S., Truman State University, 2002

A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirement for the degree of Doctor of Philosophy Department of Ecology and Evolutionary Biology 2011

This thesis entitled: The good, the bad and the flexible: interactions of Erysimum capitatum with pollinators and herbivores over space and time written by Claire Ruth Lay has been approved for the Department of Ecology and Evolutionary Biology

Dr. Pamela K. Diggle

Dr. Yan B. Linhart

Dr. Deane Bowers

Dr. Carol A. Kearns

Dr. Gregory Carey

Date

The final copy of this thesis has been examined by the signatories, and we find that both the content and the form meet acceptable presentation standards of scholarly work in the above mentioned discipline.

iii

Lay, Claire Ruth (Ph.D., Ecology and Evolutionary Biology) The good, the bad and the flexible: interactions of Erysimum capitatum with pollinators and herbivores over space and time Thesis directed by Professors Pamela K. Diggle and Yan B. Linhart

The diversity of flower structure and flowering characteristics in angiosperms has long inspired researchers to study evolutionary mechanisms that have produced it. Darwin’s recognition of many apparent flower adaptations to pollinators led to a focus on animal pollinators as selective forces, but recently, the importance of simultaneously considering other selective agents has been recognized. Particularly, herbivores can select for plant traits, and can interact with or limit selection by pollinators in complex ways. Assemblages of pollinators and herbivores can vary greatly among plant populations. In plants with generalized pollination systems and multiple herbivores, the potentially complex effects of these biotic selective forces may lead to local adaptation. Conversely, conflicting selection by multiple herbivores and pollinators may limit local adaptation and maintain variation within populations. Erysimum capitatum (Brassicaceae) is a widespread and variable plant species with generalized pollination that is attacked by a number of herbivores. I studied interactions of E. capitatum with pollinators and herbivores in four populations over an elevational gradient in the Rocky

Mountains, and tested for local adaptation to suites of pollinators and herbivores. I used observational and experimental methods in natural populations, and performed a common garden iv

experiment using plants grown from two of the original four source populations. Plant traits varied among natural populations, and individuals of E. capitatum were visited by diverse groups of pollinators and herbivores that shifted in abundance and importance in time and space. Both pollinators and herbivores preferentially visited plants with more flowers, and herbivory reduced pollinator visitation in some populations and years. Pollinators did not select on plant traits in any year or population, but herbivores sometimes selected for plant traits. These results suggest that plant plasticity may mediate interactions with both pollinators and herbivores. Pollination and herbivory both changed plant fitness, but I found no signs of local adaptation to pollinators and herbivores in the common garden experiment. However, non-local plants did have lower reproductive success than local ones, and may be locally adapted to other factors.

ACKNOWLEDGEMENTS

Many people have helped me to get through graduate school and finish my thesis.

Thanks go first of all to Pamela Diggle for taking me on as a student and for keeping me on once she realized I wouldn’t stay out of the field. Her great insights into big ideas behind evolution and plant development, and her unique way of looking at results have helped me to consider many perspectives I wouldn’t have found on my own. In addition, her excellent teaching skills have helped me to improve my own. I’m also very grateful to my second adviser, Yan Linhart, who took me on as a student. He has shepherded me through some of the challenges of field work and graduate school and his constant enthusiasm for the natural world is inspiring. Deane

Bowers has been an impressive source of ideas, knowledge and excitement about the biological and chemical world. She also gave me an opportunity to work in the entomology collection at the museum. Carol Kearns generously lent me a her property for field work, and although the pocket gophers that shared it with me thwarted the inclusion of the data here, I very much thank her for the space. She also provided data that helped me to know what to expect when doing observations on my plants. Thanks to Greg Carey for the statistical mentorship. Greg’s skill and easy manner helped me love multivariate statistics and get this project done.

Thanks to the other members of my two labs. Rob Baker has been a good friend and collaborator. In addition to hours of field and greenhouse labor, and comments on drafts and talks, he has generously given me data to play with, problems to solve, and professional inspiration. Chi-Chih Wu has also given me interesting things to work on, and listened to talks.

Erin Bissell helped me in the field and led the way to the light at the end of the thesis-writing tunnel. April Randle was a great help in organizing my thoughts for the third chapter. Bianca

Breland, Ken Keefover-Ring and Kailen Mooney were good companionship as the last students vi of the Linhart lab, and Bianca sat out watching for flies on one memorably drizzly and cold day.

Ken snapped my favorite picture of my study species.

Additional thanks go to the members of the Plant Animal Interactions reading group, the members of the Friedman and Bowers labs and some other great graduate students I’ve interacted with in my time here. Jennifer Winther always made me soup when I was sick and entertained me in the lab when we were there early or staying late. Jonathan Krieger, Sarah

Wise and Heidi Schutz were good for scientific talk and indoor climbing.

Ned Friedman has an enthusiasm for the history of biology and the beauty of ideas and life that have always made me smile, and it was a privilege to see his inspiring and entertaining teaching lectures when working as his teaching assistant. I have deep gratitude to John Basey for mentoring me through my first semesters as a general biology teaching assistant—his support helped me to develop skills and confidence in front of a classroom. Thanks also to Stephanie

Mayer for all the time we spent preparing labs, and for the plant-centered teaching assistantships

I’ve been fortunate enough to have with her. Thom Lemieux has helped me with the details of planting and greenhouse work, and has been fun to know. Virginia Scott, Boris Kondratieff,

Lubo Masmir, Leticia Sanchez, Evan Lampert and Andy Hicks helped me identify insects.

I could not have completed my fieldwork or lab work without the dedicated help of some amazing student employees and volunteers. Special thanks go to John Murgel and Mark Gorman for their hard work and skill in field logistics. I also give special thanks also to Carli Busse, who spent hundreds of hours counting seeds and measuring fruits for no pay and no credit just because she wanted to help. Every time I found her counting and sorting seeds, I was surprised and grateful. Thanks also to Sharanya Thurman, Stephanie Miller, Rachel Merriweather, Chris vii

Chalstrom, Thomas Gabel, Lisa Friedman, Jamie Buckner, Tegan McGillivray, and Alex

Steinberg.

This work was supported by grants from the EEB department, the John Marr Fund, the

CU Museum Walker Van Riper Fund, the Colorado Mountain Club, and a Beverly Sears

Graduate student grant. Many thanks go to our departmental staff, Bill Franz, Linda Bowden, Jill

Skarstad, Melanie Green, and Melissa Keller for helping with administrative details, computers and office necessities.

Boulder City and County Open Space, and the CU Mountain Research Station all permitted me to use land for field studies, and to do collections of seeds and insects. Thanks to

Lynn Riedel for the help.

Lastly, thanks to my fantastic family of friends and relatives who helped me in the field and in life while I was picking up bits of information and translating them into something coherent. My exceedingly bright and energetic stepdaughters, Sophia and Charissa, helped me plant a garden, recapture marked beetles, water and record data. Both of them had a remarkable command of scientific thought and attention to detail for people under age 13. In addition to being great field help, they are way cooler than I ever was and I am glad to have them in my life.

Thanks to all my beautifully bright, fun, creative and hard-working Burning Man family.

They’ve provided quiet places to stay when I needed to get work done, noisy places to dance when I needed a break, and a creative community of deep friendship the rest of the time. My little brother, Aaron, taught me basic relational database design and MySQL, without which I would still be futzing with large datasets in Excel. My parents, Mary Jo and Kent, have sent me love and money and fresh tomatoes from Missouri, and help with the work whenever they are around. I’m very grateful for such a wonderful bunch of people. viii

CONTENTS

CHAPTER I INTRODUCTION ...... 1

CHAPTER II THE IMPORTANCE OF LOCALE: DEFINING DIFFERENCES AMONG STUDY POPULATIONS OF THE WESTERN WALLFLOWER, ERYSIMUM CAPITATUM ...... 6

ABSTRACT ...... 6 INTRODUCTION ...... 7 MATERIALS AND METHODS ...... 8 RESULTS ...... 14 DISCUSSION ...... 23 APPENDICES ...... 29

CHAPTER III THE GOOD, THE BAD AND THE FLEXIBLE: PLANT INTERACTIONS WITH POLLINATORS AND HERBIVORES OVER SPACE AND TIME ARE MODERATED BY PLANT COMPENSATORY RESPONSES ...... 30

ABSTRACT ...... 30 INTRODUCTION ...... 31 MATERIALS AND METHODS ...... 33 RESULTS ...... 42 DISCUSSION ...... 56 APPENDICES ...... 66

CHAPTER IV ON COMMON GROUND: ERYSIMUM CAPITATUM AND LOCAL ADAPTATION TO POLLINATORS AND HERBIVORES ...... 73

ABSTRACT ...... 73 INTRODUCTION ...... 74 MATERIALS AND METHODS ...... 77 RESULTS ...... 84 DISCUSSION ...... 95 APPENDICES ...... 101

CHAPTER V CONCLUSIONS ...... 106

LITERATURE CITED ...... 112

LIST OF TABLES

TABLE 2-1. MEANS FOR PHENOTYPIC VALUES IN EACH POPULATION ...... 17

TABLE 2-2. LIST OF SPECIES IDENTIFIED FROM SAMPLES COLLECTED AFTER POLLINATOR OBSERVATIONS HAD CONCLUDED AT EACH SITE ...... 18-19

TABLE 2-3. MORISITA-HORN SIMILARITY INDICES FOR POLLINATING VISITOR FUNCTIONAL GROUPS IN PAIRS OF POPULATIONS ...... 20

TABLE 3-1 MODEL FIT STATISTICS FOR NULL, MOST-INCLUSIVE AND BEST-FIT MODELS FOR OBSERVATIONAL STUDY ...... 43

TABLE 3-2. STEPDOWN RESULTS FOR FITNESS COMPONENTS IN GBT2006 ...... 51

TABLE 3-3. STEPDOWN RESULTS FOR FITNESS COMPONENTS IN WAK2005 ...... 52

TABLE 3-4. STEPDOWN RESULTS FOR FITNESS COMPONENTS IN WAK2006 ...... 53

TABLE 3-5. STEPDOWN RESULTS FOR FITNESS COMPONENTS IN NED2005 ...... 54

TABLE 3-6. STEPDOWN RESULTS FOR FITNESS COMPONENTS IN NED2006 ...... 55

TABLE 3-7. STEPDOWN RESULTS FOR FITNESS COMPONENTS IN ELK2006 ...... 55

TABLE 4-1. MODEL FIT STATISTICS FOR RANDOM EFFECTS OF PLANT, MATERNAL FAMILY AND DATE ON VISITORS AND PETAL DAMAGE ...... 86

TABLE 4-2. EFFECT SIZES AND ASSOCIATED Z-VALUES AND P-VALUES OF SOURCE POPULATIONS, INSECTICIDE TREATMENT, FLOWER CHARACTERISTICS, FLORAL DISPLAY SIZE, HEIGHT AND INTERSPICEIS INTERACTIONS ON POLLINATOR VISITS AND INSECT-CAUSED PETAL DAMAGE ...... 87

LIST OF FIGURES

FIGURE 1-1. SOME OF THE RANGE OF FLOWER COLOR VARIATION IN E. CAPITATUM ...... 4

FIGURE 1-2. ANTS REMOVING NECTAR FROM THE BASES OF FLOWERS ON A PLANT WITH SEVERE PETAL DAMAGE (A) AND PLANT WITH LESS SEVERE PETAL DAMAGE (B) ...... 4

FIGURE 1-3. GALLS ON A PLANT (A) CECIDOMYIID GALL MIDGES IN A FLOWER BUD OF E. CAPITATUM (B)...... 5

FIGURE 1-4. MELIGETHES SP. EMERGING FROM A COROLLA TUBE COVERED IN POLLEN . 5

FIGURE 2-1. AVERAGE COROLLA COLOR IN EACH SITE ...... 17

FIGURE 2-2. RELATIVE FREQUENCEIS OF POLLINATORS IN EACH SITE ...... 17

FIGURE 3-1. BEST-FIT MODELS FROM STRUCTURAL EQUATIONS ANALYSIS OF GBT2005(A), GBT2006(B), AND WAK2005(C) ...... 47

FIGURE 3-2. BEST-FIT MODELS FROM STRUCTURAL EQUATIONS ANALYSIS OF WAK2006(A), NED2005(B), ELK2005(C), AND ELK2006(D) ...... 48

FIGURE 3-3. SUMMARY DIAGRAM OF BEST-FIT MODELS SHOWING TRAITS THAT ATTRACTED BOTH POLLINATORS AND HERBIVORES ...... 56

FIGURE 3-4. SUMMARY DIAGRAM OF BEST-FIT MODELS THAT SHOWED EFFECTS OF HERBIVORY ON POLLINATOR VISITATION ...... 57

FIGURE 3-5. SUMMARY DIAGRAM OF BEST-FIT MODELS THAT SHOWED EFFECTS OF PLANT TRAITS AND INSECTS ON FITNESS ...... 58

FIGURE 3-6. RELATIONSHIP BETWEEN ABORTED AND TOTAL FLOWERS ...... 59

FIGURE 3-7. RELATIONSHIP BETWEEN GALLS AND TOTAL FLOWERS ...... 62

FIGURE 4-1. DESIGN OF (A) TWO SAMPLE CIRCULAR PLOTS AND (B) OVERALL EXPERIMENTAL GARDEN ...... 78

FIGURE 4-2. MEANS OF ESTIMATED TOTAL SEED NUMBER ...... 89 FIGURE 4-3. MEANS OF VIABLE SEEDS PER FRUIT ...... 90

FIGURE 4-4. MEANS OF INVIABLE SEEDS PER FRUIT ...... 91

FIGURE 4-5. MEANS OF FRUITS ...... 92

FIGURE 4-6. MEANS OF ABORTED FRUITS ...... 93

FIGURE 4-6. MEANS OF SEED WEIGHTS ...... 94

APPENDICES

APPENDIX 2-1. VISITS PER FLOWER PER HOUR DIFFERENCES AMONG POPULATIONS ..... 29

APPENDIX 3-1. POLLINATOR FREQUENCY ...... 66

APPENDIX 3-2. CORRELATIONS AMONG COMPONENTS OF FITNESS ...... 66

APPENDIX 3-3. MODEL FIT STATISTICS FOR COMPLETE SETS OF MODELS TESTED IN THE ANALYSIS OF INSECT PREFERENCE, FITNESS EFFECTS AND SELECTION FOR GBT 2005, GBT 2006, WAK 2005 AND WAK 2006 ...... 67

APPENDIX 3-4. MODEL FIT STATISTICS FOR COMPLETE SETS OF MODELS TESTED IN THE ANALYSES OF INSECT PREFERENCE, FITNESS EFFECTS, AND SELECTION FOR NED 2005, ELK 2005 AND ELK 2006 ...... 68

APPENDIX 3-5. MOST-INCLUSIVE MODELS FOR GBT 2005, GBT 2006, AND WAK 2005 ...... 69

APPENDIX 3-6. MOST-INCLUSIVE MODELS FOR GBT 2005, GBT 2006, AND WAK 2005 ...... 70

APPENDIX 3-7. EFFECTS OF HEIGHT AT HARVEST, EXPERIMENTAL TREATMENTS AND GALLS ON ESTIMATED FEMALE FITNESS ...... 71

APPENDIX 3-8. COMPLETE MANCOVA RESULTS FOR ANALYSIS OF FITNESS COMPONENTS ...... 70

APPENDIX 4-1. FIXED EFFECTS OF LINEAR MIXED EFFECTS MODEL FOR TOTAL ESTIMATED SEEDS ...... 102

APPENDIX 4-2. FIXED EFFECTS OF LINEAR MIXED EFFECTS MODEL FOR VIABLE SEEDS PER FRUIT ...... 103

APPENDIX 4-3. FIXED EFFECTS OF LINEAR MIXED EFFECTS MODEL FOR INVIABLE SEEDS PER FRUIT ...... 103

APPENDIX 4-4. FIXED EFFECTS OF LINEAR MIXED EFFECTS MODEL FOR FRUITS WITH SEEDS ...... 104

APPENDIX 4-5. FIXED EFFECTS OF LINEAR MIXED EFFECTS FOF ABORTED FLOWERS .... 104

APPENDIX 4-6. EFFECTS OF SOURCE POPULATION, LOCATION (FIELD OR GARDEN), INSECTICIDE AND POLLINATION TREATMENTS ON THE WEIGHT OF 20 SEEDS PER PLANT...... 105

APPENDIX 4-7. FLOWER TRAIT MEANS FOR PLANTS IN THE COMMON GARDEN FROM EACH OF THE TWO SOURCE POPULATIONS FOR WHICH ALL FLOWER TRAITS WERE MEASURED ...... 105 1

CHAPTER I

INTRODUCTION

The great diversity of flower structure and flowering characteristics in angiosperms has inspired generations of researchers to examine the evolutionary mechanisms which have produced this diversity. Darwin proposed pollinators as agents of natural selection on flowers (Darwin 1859,

1862, 1877). Recently, the long-standing focus on the role of pollinators in selecting for flower and plant reproductive traits (e.g. Fenster 2004) has turned to a simultaneous consideration of other factors, particularly herbivores. This new focus on multiple selective agents for plant flower and reproductive traits is largely due to evidence that conflicting and interactive selection by herbivores influences the outcome of selection by pollinators. Conflicting selection occurs when pollinators and herbivores act on the same traits in opposite directions, and this is the most likely outcome of selection by both for the same traits (Krupnick and Weis 1999). Interactive selection is also possible, though, as either pollinators or herbivores may alter the attractiveness of plants to other organisms. Because the outcome of selection can be changed by either conflicting or interactive effects, the influences of either pollinators or herbivores upon plant evolution may be misunderstood when they are not studied simultaneously (Strauss 1997, Gomez 2003, Irwin et al. 2003, Gomez

2005).

Biotic selective forces such as pollinators and herbivores are important drivers of local adaptation for a number of plant traits (Haloin and Strauss 2008, Urban 2011). Generalized pollination systems may select for the maintenance of relationships with a diversity of pollinators in multiple sites, or may make phenotypic divergence among locations more likely when some 2 pollinators are consistently more common or effective in some locations than others. The additional influence of herbivores, which also vary spatially and temporally, can affect pollinator choices (eg.

Adler 2000, Sanchez-Lafuente 2007) and result in adaptive tradeoffs or reduce the strength of selection by pollinators (Irwin and Adler 2006, Lay et al. 2011).

To better understand the various selective scenarios possible in complex systems, and to identify potential drivers of local adaptation, I chose to perform my studies of interactions among a focal plant species, and its herbivores and pollinators in several sites. Many plant species vary in floral morphology, phenology and floral resources among sites, and this variation is often related to abiotic factors (Linhart and Grant 1996). However, communities of biotic interacting partners also play a frequent role in evolutionary divergence (Herrera 1988, Gomez 1993, Herrera 1993, Benkman

1999, Gomez and Zamora 2000, Haloin and Strauss 2008, Gomez et al. 2009, Urban 2011).

I found the idea of a generalized plant-pollinator-herbivore system with the potential for diffuse selection by a number of species to be particularly compelling from the standpoint of local adaptation, as many studies of local adaptation to biotic interactions are of tightly co-adapted systems. Diffuse selection can occur when the selective outcome of one biotic selective agent is altered by the presence of another (reviewed in Haloin and Strauss 2008). In my thesis work, I have considered the relative selective contributions of pollinators and herbivores and their potential interactions on floral and flowering traits, using a geographical perspective. I examined phenotypic variation and interactions with pollinators and herbivores in the Western Wallflower, Erysimum capitatum in sites along an elevational gradient in the Rocky Mountains. A geographical perspective allowed me to consider the relative importance of pollinators and herbivores on selection in a more general way than would have focusing on a particular set of pollinators and herbivores in a particular 3 population. Biotic partners and interactions change over space and time, and viewing them in multiple populations allowed me to observe a variety of interacting partners in a variety of situations.

Erysimum capitatum provides an ideal focal plant species to answer questions about the joint and separate selective effects of pollinators and herbivores, as well as about local adaptation to different communities of these insects. Floral traits of E. capitatum vary throughout its range (Price

1984, Fig. 1-1), individuals are visited by a wide array of pollinating insects and reproductive tissues are attacked by a number of herbivores (Figs. 1-2 to 1-4). Small bees are the most common pollinating visitors to E. capitatum, but larger bees, flies, nitidulid beetles (Fig. 1-4), butterflies and moths also visit (personal observation). Some herbivores common on E. capitatum could impact pollinator choice. Ants remove nectar from the base of flowers, and occasionally attack bees (Fig. 1-

2). Other herbivores, particularly beetle larvae, damage petals, potentially making plants less attractive to pollinators (Fig. 1-2). Also, gall midge larvae form galls from floral buds and typically prevent flowers from opening, which decreases floral display size (Fig. 1-4). The prevalence of each of these insect groups differs among sites, and it is thus possible that local adaptation to insect communities has affected the evolution of traits in E. capitatum. I used observational studies, insecticide treatments, hand-pollination, and common garden experiments to examine the effects of pollinators and floral herbivores on the reproductive success of E. capitatum.

In Chapter 2, I describe differences in phenotype, pollinator community composition and herbivory among four populations over an elevational gradient in order to provide a basis for further work on local differentiation of these populations. Chapter 3 focuses on testing for natural selection by pollinators and herbivores in those four populations and reveals the potential role of plant plasticity in mediating interactions with both pollinators and herbivores. Chapter 4 4 describes the results of a common garden experiment performed with two of the four populations.

FIGURE 1-1. Some of the range of flower color variation in E. capitatum.

FIGURE 1-2. A) Ants removing nectar from the bases of flowers on a plant with severe petal damage. B) A plant with less severe petal damage.

FIGURE 1-3. A) Galls on a plant B) Cecidomyiid gall midges in a flower bud of E. capitatum. Galls typically contain between 6 and 20 larvae.

FIGURE 1-4. Meligethes sp. emerging from a corolla tube covered in pollen.

CHAPTER II

THE IMPORTANCE OF LOCALE: VARIATION AMONG POPULATIONS OF ERYSIMUM CAPITATUM IN PHENOTYPE, POLLINATION AND HERBIVORY

ABSTRACT

Local adaptation of plant flower and life-history characters to biotic interactions has often been studied in the context of specialized pollinators that vary in frequency or behavior among sites. Recently, the potential importance of herbivores and the interactions of pollinators and herbivores has been recognized. I studied pollinators, herbivores and plant phenotypes in four populations of the western wallflower (Erysimum capitatum, Brassicaceae) along an elevational gradient in the Front Range of the Eastern Slope of the Colorado Rocky Mountains in 2005 and

2006. Several phenotypic characters varied among populations over a relatively small spatial scale (<50km), and plants were visited by a wide variety of pollinators and herbivores. E. capitatum plants in these populations were primarily pollinated by bees, pollen beetles, and flies.

The most common types of pollinating visitors differed among populations. Ants were the most common visitors in all populations, and removed nectar from the base of flowers, occasionally attacked bees, and rarely came into contact with stigmas. Many other herbivores fed on flowers and buds of E. capitatum. Some of these damaged petals, potentially making plants less attractive to pollinators. Also, gall midge larvae (Cecidomyiidae) formed galls from flower buds and typically prevented flowers from opening, which decreased floral display size. Both herbivores and pollinators were different from year to year and from population to population. I argue that the potential for local adaptation in Erysimum capitatum exists.

INTRODUCTION

Floral morphology, phenology and floral resources can vary within species over distances as short as 1m (reviewed in Linhart and Grant 1996). Such patterns of geographic variation have frequently been related to physical features of the environment (Linhart and Grant 1996), but communities of both pollinators and herbivores may also play a role in evolutionary divergence of plant characteristics (Herrera 1988, Gomez 1993, Herrera 1993, Gomez and Zamora 2000, Haloin and Strauss 2008, Gomez et al. 2009, Urban 2011). Interactions between plants, their pollinators and herbivores can be highly site- and time-specific (Thompson 1988), resulting in different selection regimes (Rand 2002, Bradley et al. 2003), and phenotypic divergence among sites. Site-specific comparisons of plant phenotype, herbivory and pollinator community composition or behavior allow for investigation of local divergence.

Plants with generalized pollination systems have only rarely been studied in the context of local adaptation to particular communities of pollinators and herbivores (but see Aigner 2005,

Gomez et al. 2007, Gomez et al. 2009). Local adaptation may be unlikely if generalized pollination morphology results from selection to maintain relationships with a diverse group of pollinators

(Waser et al. 1996, Dilley et al. 2000, Ellis and Johnson 2009). Conversely, plants with generalized pollination systems may show population-level phenotypic divergence if pollinator communities are very different among sites. Consistent differences in the most common or effective pollinators could result in local adaptation. Moreover, herbivores, which also vary spatially and temporally, can affect pollinator choices (Adler 2000) and result in adaptive tradeoffs or alter the strength of selection by pollinators (Irwin and Adler 2006, Gomez 2008). Addressing conflicting predictions raised by these issues requires comparative study of multiple populations of plants likely to have different pollinators and herbivores. 8

Erysimum capitatum is an exceptional plant species in which to examine geographic variation in flower morphology, pollinators and herbivores. E. capitatum plants are quite variable in morphology and life history characters across their broad geographic range and elevational distribution (Price 1984). Individuals are visited by a wide array of pollinating insects, and reproductive tissues are attacked by a number of herbivores (C. Lay, personal observation). Plant species, such as E. capitatum, that occupy montane regions may be particularly likely to show local adaptation because these areas are very heterogeneous over short distances in both micro-climate and co-occurring species (reviewed in Lomolino 2001).

Elevational gradients have been associated with population differentiation in many species

(Linhart and Grant 1996).

I studied four montane populations of E. capitatum for two growing seasons to answer the following questions: Is there differentiation in phenotype among populations? What are the common pollinators and herbivores of this relatively unstudied species? Do communities of pollinators and herbivores differ among populations?

MATERIALS AND METHODS

Study species:

Erysimum capitatum (Brassicaceae) occurs throughout most of the United States (USDA

2004) and northern Mexico (Turner 2006). It is often a biennial monocarp, though its life history is somewhat plastic and can differ among individuals and populations (Kim and Donohue 2011).

Plants flower from early spring to mid-summer (USDA 2004), and populations of E. capitatum are characterized by considerable variation in flower color and other traits throughout its range

(Price 1984, Weber 1990). Flowers range from yellow and white to red and lavender (Price,

1984). Although I assumed that all populations and individuals studied belonged to a single 9 species, the taxonomy of E. capitatum is currently undecided. Its great phenotypic variation may be the result of multiple hybridization events between subspecies (Turner 2006). Other plants in the genus Erysimum contain glucosinolates and cardiac glycosides (Renwick et al. 1989, Stadler et al. 1995), and though secondary chemistry has not been studied in E. capitatum, it most likely also contains these compounds.

Populations:

I used four study populations located in the Front Range on the Eastern slope of the

Rocky Mountains of Colorado. The Elk Meadow population (ELK) grows at 2900 m (40°01’N,

105°32’W) in a subalpine dry meadow at the University of Colorado Mountain Research station.

The population at Nederland (NED) is at 2590 m (39°58’N 105°30’W) in open lower montane

Ponderosa pine forest. The Walker Ranch population (WAK) is at 2209 m (39°56’N 105°20’W) in open lower montane Ponderosa pine forest. The Greenbelt Plateau population (GBT) grows at

1980 m (39°55’N 105°14’W) in a short grass prairie. Marr (1967) gives complete descriptions of these vegetation types. Flowering begins as early as March and continues through early June in GBT, the lowest elevation population. In the highest elevation population, ELK, flowering begins in late June and continues through mid-July, and a very few plants have open flowers through the fall. We assessed plant phenotype, herbivory, and pollinator visitation for 50-200 haphazardly selected flowering plants in each of these populations in 2005 and 2006.

Phenotypic characters:

I measured corolla width at the widest point, corolla tube depth, corolla color, number of open flowers, height of tallest inflorescence, height to first flower on the tallest inflorescence and stem diameter in 2005 and 2006 in ELK, NED, WAK and GBT. 10

For all populations, the sampling schemes differed slightly between 2005 and 2006. In

2005, we measured larger plants that were likely to have open flowers throughout the planned period of pollinator observations. In 2006, sample sizes were increased at all sites, necessitating the inclusion of smaller plants. Also, the timing of rainfall and flowering differed between years, and may have resulted in differences in plant size. I therefore included height as a covariate in analyses of plant phenotype to control for differences in plant size between years that are unlikely to be meaningful to a description of consistent population-level differences.

Measurements of open flowers, and flower width and depth were repeated on up to four days, or on multiple flowers per day, though it was not possible to obtain equal replications on all plants due to weather and flower availability. In 2005, I measured characteristics of three flowers per plant whenever possibly on each measurement day. Preliminary analyses suggested that this was unnecessary, as the difference in flower characters among plants was much greater than differences in flower measurements on the same plant, and in 2006, corolla width and depth were measured only once for one flower per plant. Because of the difficulty of achieving a balanced repeated measures design in the field, I collapsed measurements for phenotypic characters as follows. For characters such as corolla width, where differences among measurements due to intraplant variation or measurement error were likely to be the biggest contributors to within-plant measurement differences, I used the average of all measurements as the character values. For characters such as inflorescence height, for which the biggest contributor to differences among measurements was likely to be growth, I used the maximum recorded value for each plant as the character values.

Corolla color was quantified by comparing flowers to color chips (Behr paint chips of colors yellow S-G-390 to red S-G-230) in the field. The accuracy of this method was confirmed 11 using digital photographs and standardized RGB color values. Flowers of some plants were old or otherwise deteriorated on days that color was measured, and so I was unable to obtain color values for all plants. Corolla width and depth were measured with calipers on flowers that had at least three undamaged petals. Corolla width was the diameter of the flower as seen from the front, and corolla depth was the distance from the base of the calyx to the angle between the claw and the limb. Flowers were counted as “open” if they retained at least 3 petals (flowers with limbs damaged by herbivory were scored as “open” if they had three or more attached claws).

Inflorescence height was measured from the ground to the highest point on the tallest inflorescence, and the height to the first flower was also measured from the ground on the tallest inflorescence. Stem diameter was measured 30 mm above the ground on the tallest inflorescence.

I tested for significant phenotypic differences among plants from different sites using nested two-way multivariate analysis of covariance (MANCOVA) followed by univariate analyses of variance (ANOVA) with height as a covariate for size-related characters (all characters but color). This analysis was followed with univariate two-way analysis of covariance

(ANCOVA) for each trait. Correlation analysis (proc CORR, SAS 2003) showed that color was not related to size or other characters so it was analyzed separately with a two-way ANOVA. In both analyses, site was nested in year. Inflorescence height of the tallest inflorescence was included as a covariate for characters correlated to it to control for differences between years in plant size that may have been caused by differences in rainfall or sample selection. Including inflorescence height as a covariate rather than as a dependent phenotypic variable also allowed me to test whether the effect of plant size on a given character is the same in all populations.

Pollination and insect visitation:

To observe insect visitation and pollinator community composition, individual observers watched groups of 1 to 10 plants for 10-min periods during peak visitation times. Observations were repeated twice per day for as many days as possible given weather, availability of observers, and availability of flowering plants. Observers minimized their own shadows on plants, and avoided disturbing visitors. I divided insects seen during observation periods into those that contacted reproductive parts and those that did not. Visitors were identified to functional groups, and insects were not collected until observations had been finished for the season to avoid influencing visitation on subsequent days.

I grouped likely pollinators (those that contacted reproductive parts) into functional groups based on size and taxonomic order. After observations were completed for a site, I collected samples of insects that were found visiting plants for identification. Sampling effort differed among sites and years due to differences in the number of available plants and phenology.

In order to determine the level of similarity in pollinator visitation among populations, I calculated Morisita-Horn similarity indices in EstimateS (Colwell 2005), using pooled pollinator functional group counts from each population. This index is robust to differences in sampling effort, but is strongly affected by the most common visitor (Magurran 2004). Differences in sample size between years were large (with as few as 7 observed pollinators in one population in one year). The unequal sample sizes greatly reduced statistical power for analysis of similarity

(ANOSIM), a standard non-parametric test for community differences. Therefore, I instead pooled pollinator counts from both years and show only qualitative Morista-Horn similarity indices. 13

Herbivory:

Three general types of herbivore damage were common on flowers of Erysimum capitatum: petal damage, flower bud galls, and nectar robbing. A variety of petal herbivores created visually similar damage, so there was no reliable way to determine what herbivore had damaged a flower. I was able to identify some key petal herbivores by carefully observing for their presence, but could not test for site differences in herbivore assemblage.

Damage levels were recorded for each damaged flower on a plant. Petal limb damage was scored 1 to 4, with an additional category 5 indicating complete destruction of anthers and stigma. “1” corresponded to 10% damage or less, “2” to damage between

10% and 25%, “3” to 25-50%, “4” to 50% damage to damage of all petal limbs. I summed all damage scores from each plant on each day for a measure of overall damage, and used the maximum value of summed petal damage as the dependent variable in an analysis of covariance (ANCOVA) with site and year as main effects and the total number of flowers on the day of maximum petal damage as a covariate. This analysis was performed with aov from the stats package in R (R Development Core Team 2010a).

Gall midge larvae (Cecidomyiidae) form galls from flower buds and typically prevent flowers from opening. I counted flower bud galls, which were readily identifiable as large buds with very swollen calyces, on each plant on each observation day. I then log-transformed the maximum count for galls found on a single day over the observation period to improve normality. This log-transformed value for galls was the dependent variable in an analysis of covariance (ANCOVA) in which site and year were main effects and height, a good predictor of the total number of available buds, was a 14 covariate. This analysis was performed using the aov function of the stats package in R

(R Development Core Team 2010a).

Finally, ants are frequent visitors to flowers, and remove nectar from between the free sepals. I observed ant visitation during pollinator observations, and counted ants present on plants directly after pollinator observation periods.

RESULTS

Plant size:

Size-related characters differed among sites, although there were no consistent patterns related to elevation (Table 2-1). Inflorescence height varied with site

(F(1,913)=49.6, p<0.0001, Table 2-1), and plants were significantly shorter in 2006 than in

2005 in all sites (F(1,913)=189.2, p<0.001). Height differences may be due to reduced rainfall during the flowering period in 2006, or to sample selection differences among years or to other factors. When controlling for inflorescence height (F(1,799)=566.2, p<0.0001), stem diameter differed among sites (F(3,799)=6.9, p=0.003, Table 2-1), but not between years (F(4,799)=1.4, p=0.23). In addition, there was a significant interaction between site and inflorescence height, indicating that the relationship between inflorescence height and stem diameter differed among populations (F(3,799)=16.0, p<0.0001). Height to first flower differed among sites (F(3,799)=6.6, p<0.001) and between years (F(4,799)=2.9, p=0.02) even controlling for the large effect of inflorescence height (F(1,799)=863.7, p<0.0001). Inflorescences of the same height in different sites had different heights to the first flower (F(3,799)=16.6, p<0.0001).

Flower characters:

Site means of flower characters are shown in Fig. 2-1 and Table 2-1. Corolla color varied greatly from yellow to red with site (F(3,514)=42.01, p<0.0001, Fig. 2-1), but did not differ between years (F(1,514)=0.09, p=0.75). Both corolla width and depth increased with inflorescence height (FFW(1,799)=151.89, pFW<0.0001; FFD(1,799)=132.31, pFD<0.0001). Corolla width varied among sites (F(3,799)=10.85, p<0.0001) and between years (F(4,799)=5.45, p=0.04) when controlling for the relationship with inflorescence height. Also, there was a significant interaction between inflorescence height and site meaning that relationship between inflorescence height and corolla width differed among sites (F(3,799)=8.15, p<0.0001). Corolla depth did not differ among sites (F(3,799)=0.71 p=0.55), though there was a significant interaction between site and inflorescence height

(F(3,799)=4.17, p<0.01). Corolla depth also differed between years within sites

(F(4,799)=5.01, p=0.02). Corolla width and depth were correlated in all sites and years, even when the effect of inflorescence height on flower size was removed (r=0.6, p<0.001).

Fig. 2-1. Average corolla color in each site. “n” is the number of plants included from each site. The small circles are approximations of the color one standard deviation away from the mean.

Correlations among size-related characters were not constant among sites (Wilk’s lambda=0.90, p<0.0001) or between years (Wilk’s lambda=0.86, p<0.001), and varied 16 with plant height (Wilk’s lambda=0.34, p<0.0001). This may indicate some plasticity in plant traits, or may be due to differences in sampling between years.

Pollinator community composition:

I considered visitors to be pollinators if they came into contact with the reproductive parts of a flower. Pollinators in the studied populations were assigned to functional groups that included large bees, medium bees, small bees, flies, bee-flies, butterflies, and moths (Table 2-2). “Large bees” include bees over 2 cm in body length

(Table 2-2). “Medium” bees include bees with body lengths between 1 and 2 cm (Table

2-2). “Small bees” include bees under 1 cm in body length (Table 2-2). “Flies” included flies that were above 0.25 cm in body length (Table 2-2). Smaller flies rarely contacted reproductive parts and were nearly impossible to see during observations. Pollen beetles

(Meligethes sp., Nitidulidae) interact differently with flowers than other beetles; they enter corolla tubes, are coated in pollen, and spend long periods of time in flowers.

Therefore, I treated them as a separate functional group and grouped the small number of other beetle visitors that contacted reproductive parts together (Table 2-2). These included beetles of the families Coccinellidae and Buprestidae, as well as other unidentified beetles. All bees and pollen beetles were visibly coated in pollen after visiting flowers (Table 2-2).

Populations had qualitatively different pollinator community compositions, with

GBT and ELK being the most different (Table 2-3). The most similar pollinator communities were at NED and WAK (Table 2-3). Flies were more common at ELK, and pollen beetles and bees were more common at GBT (Fig. 2-2).

Inflorescence Height of Stem Open Corolla Corolla height (mm) First Flower Diameter flowers width (mm) Depth (mm) (mm) (mm) GBT 185.11±55.2c 135.18±35.73c 2.30±0.70a 9.63±6.71a 15.05±2.46b 10.61±12.3 n=141 n=141 n=141 n=141 n=135 n=135 WAK 163.03±62.4d 131.22±58.23c 1.80±0.62c 5.43±4.02bc 14.19±2.19c 9.85±1.91 n=244 n=243 n=282 n=254 n=254 n=255 NED 241.16±79.17a 197.81±76.67a 2.14±0.66a 6.89±7.20b 15.83±2.56a 10.23±1.81 n=245 n=235 n=254 n=238 n=207 n=208 ELK 207.74±78.46b 176.30±74.95b 1.90±0.52b 5.33±3.38c 15.07±3.08b 10.99±2.26 n=283 n=272 n=273 n=280 n=263 n=263 TABLE 2-1. Means ± SE for phenotypic values in each population. Means different by Tukey’s HSD post hoc tests are indicated by different superscripts. Sample size (n) for each trait and site is listed below means. Sample size differed for traits due to availability of flowers and human error. 18

TABLE 2-2. List of potential pollinator species identified from samples collected after pollinator observations had concluded at each site. Some unidentified samples were lost when a freezer malfunctioned, so this list is not comprehensive.

Site Order Family ID Count Functional group GBT Coleoptera Unidentified Coleoptera a. 1 na GBT Diptera Canopidae Zodion fulvifrons (Say) 1 Fly GBT Hemiptera Rhopalidae Rhopalidae sp. 2 na GBT Hymenoptera Formicidae Forelius foetidus 6 Ant GBT Hymenoptera Formicidae Tapinoma sessile 10 Ant GBT Hymenoptera Halictidae Andrenidae sp e. 1 Ant GBT Hymenoptera Halictidae Lasioglossum evylaeus 1 Ant GBT Lepidoptera Unidentified Lepidoptera a. 1 Lepidoptera GBT Hymenoptera Halictidae Halictus rubicundus 1 Medium Bee WAK Coleoptera Coccinellidae Coccinellidae sp. 1 Other Beetles WAK Coleoptera Curculionidae Curculionidae sp. 1 Other Beetles WAK Coleoptera Unidentified Coleoptera b. 1 Other Beetles WAK Diptera Empididae Empis sp. 1 Fly WAK Diptera Muscidae Drymeia sp. 1 Fly WAK Hemiptera Cicadellidae Cicadellidae sp. 1 na WAK Hemiptera Lygaeidae Lygaeus kalmii 2 na WAK Hymenoptera Formicidae Tapinoma sessile 9 Ant WAK Hymenoptera Andrenidae Andrenidae sp. 1 Medium Bee WAK Hymenoptera Halictidae Halictus sp. c. 1 Medium Bee WAK Hymenoptera Halictidae Lasioglossum sp a 1 Small Bee WAK Hymenoptera Halictidae Lasioglossum sp. b. 1 Small Bee WAK Lepidoptera Unidentified Lepidoptera b. 1 Lepidoptera NED Coleoptera Coccinellidae Coccinellidae 1 Other Beetles NED Coleoptera Nitidulidae Meligethes sp. 7 Nitidulidae NED Coleoptera Unidentified Coleoptera c. 1 Other Beetles NED Diptera Chironomidae Chironomidae 1 Fly NED Hemiptera Cicadellidae Cicadellidae 2 na NED Hemiptera Cicadellidae Cicadellidae 1 na NED Hemiptera Lygaeidae Lygaeus kalmii 3 Ant NED Hymenoptera Formicidae Formica argentea 4 Ant NED Hymenoptera Formicidae Formica obscuriventris elivia 5 Ant NED Hymenoptera Formicidae Lasius niger 3 Ant NED Hymenoptera Formicidae Tapinoma sessile 4 Ant NED Hymenoptera Formicidae Temnothorax rugatulus 1 Ant NED Hymenoptera Formicidae Temnothorax sessile 1 Ant NED Hymenoptera Halictidae Lasioglossum sp. c. 1 Small Bee NED Hymenoptera Unidentified Unidentified (tiny wasp) 1 na ELK Coleoptera Coccinellidae Coccinellidae sp. 1 na ELK Coleoptera Nitidulidae Meligethes sp. 11 Nitidulidae ELK Coleoptera Unidentified Unidentified a 1 Other Beetles ELK Coleoptera Unidentified Unidentified b. 1 Other Beetles 19

ELK Coleoptera Unidentified Unidentified c. 1 Other Beetles ELK Diptera Diptera Diptera sp. 1 Fly ELK Diptera Muscidae Muscidae sp. A 1 Fly ELK Diptera Muscidae Muscidae sp. B 1 Fly ELK Diptera Muscidae Muscidae sp. C 1 Fly ELK Diptera Muscidae Muscidae sp. D 1 Fly ELK Diptera Muscidae Thricops sp. 1 Fly Sphaerophoria contigua ELK Diptera Syrphidae (Macquart) 1 Fly ELK Hemiptera Aphididae Aphididae sp. 1 na ELK Hemiptera Cicadellidae Cicadellidae sp. A 1 na ELK Hemiptera Cicadellidae Cicadellidae sp. B 1 na ELK Hemiptera Cicadellidae Cicadellidae sp. c 2 na ELK Hemiptera Lygaeidae Lygaeus kalmii 4 na ELK Hemiptera Miridae Miridae sp. 1 na ELK Hemiptera Rhopalidae Rhopalidae sp. 1 na ELK Hymenoptera Chrysididae Chrysididae 1 Medium Bee ELK Hymenoptera Formicidae Formica argentea 3 Ant ELK Hymenoptera Formicidae Tapinoma sessile 2 Ant ELK Hymenoptera Formicidae Tetramorium hispidum 3 Ant ELK Hymenoptera Halictidae Andrenidae sp. a 1 Small Bee ELK Hymenoptera Halictidae Andrenidae sp. c 1 Small Bee ELK Hymenoptera Halictidae Andrenidae sp. d 1 Small Bee ELK Hymenoptera Halictidae Andrenidae sp. f 1 Small Bee ELK Hymenoptera Halictidae Bombus flavifrons 1 Large Bee ELK Hymenoptera Halictidae Bombus mixtus 1 Large Bee ELK Hymenoptera Halictidae Celioxys 1 Medium “Bee” ELK Hymenoptera Halictidae Colletes 1 Small Bee ELK Hymenoptera Halictidae Duforea maura 1 Medium Bee ELK Hymenoptera Halictidae Halictus sp. a. 1 Small Bee ELK Hymenoptera Halictidae Hoplitus fulgida 1 Medium Bee ELK Hymenoptera Halictidae Hylaeus basalis 1 Medium Bee ELK Hymenoptera Halictidae Lasioglossum sp. a. 1 Small Bee ELK Hymenoptera Halictidae Lasioglossum sp. e 1 Small Bee ELK Hymenoptera Halictidae Osmia sp. 1 Medium Bee ELK Hymenoptera Unidentified Unidentified (large wasp) 1 na ELK Hymenoptera Unidentified Unidentified (large wasp) 1 na ELK Hymenoptera Unidentified Unidentified (tiny wasp) 1 na ELK Hymenoptera Unidentified Unidentified (tiny wasp) 1 na ELK Hymenoptera Unidentified Unidentified (tiny wasp) 1 na 20

Site comparison

GBT and WAK 0.723

GBT and NED 0.752

GBT and ELK 0.516

WAK and NED 0.849

WAK and ELK 0.834

NED and ELK 0.848 TABLE 2-3. Morisita-Horn similarity indices for pollinating visitor functional groups in pairs of populations. “1” indicates complete similarity. The least similar populations are GBT and the ELK

FIGURE 2-2. Relative frequencies of pollinators in each site.

Herbivory:

Petal herbivores included grasshoppers, lepidopteran larvae, and nitidulid beetle larvae that lived inside of flowers (most frequently observed). Megachilid bees were never observed, but the rounded edge wounds that are characteristic of these bees occurred frequently. I was unable to rear the any of the larvae to maturity, and they are yet to be identified.

On average, plants received about 10% damage to their corolla limbs for each flower, though there was a good deal of variation in damage to individual flowers. On any given plant, roughly 30% of flowers were damaged. Levels of petal damage did not differ among sites

(F(3,780)=1.9, p=0.19). Because there was a significant interaction between the number of flowers and site (F(3,780)=5.38, p=0.001), there was a difference among sites in the relationship between flowers and petal herbivory. There was no overall difference between years in petal damage

(F(1,780)=2.8, p=0.09). However, there was a significant interaction between year and site

(F(3,780)=3.8, p=0.01).

Galls were formed by gall midges (Cecidomyiidae). The incidence of galls ranged from

2% of 50 plants at WAK in 2005 to 50% of 197 plants at WAK in 2006. Numbers of galls averaged between 1 and 5 per plant in all sites. Gall number did not vary among sites

(F(3,874)=2.32 p=0.07), but was greater in 2006 than in 2005 (F(1,874)=27.66, p<0.0001) and there was an interaction between site and year (F(3,874)=17.69, p<0.0001). The effect of inflorescence height on gall number was strong (F(1,874)=16.06, p<0.0001), likely because taller plants have more buds to attack. There was no significant interaction between the effect of inflorescences height and site on gall number (F(3,874)=18.86, p=0.23), suggesting that plants in different sites of the same height receive similar numbers of galls.

Ants removed nectar from flowers, and occasionally attacked bees. Unlike many ant nectar-robbing systems, ants do not damage flower parts of E. capitatum, but remove nectar from openings between sepals. Ants comprised 39% of all insect visitors at GBT, 50% at WAK, 57% at NED and 62% at the ELK.

Other arthropods observed on inflorescences:

A number of arthropods that did not contact reproductive parts or visibly remove nectar were observed on inflorescences during observation periods. These included crab spiders, hemipterans, homopterans, and ladybird beetles. In WAK, stem boring beetle larvae clipped entire inflorescences.

DISCUSSION

Phenotypic differentiation:

Populations of E. capitatum in this study region differed in phenotype and pollinator community composition, and levels of herbivory changed from year to year. The traits that varied among the study populations may affect pollinator and herbivore preferences. In particular, corolla color (Fig. 2-1), varied among populations. Corolla color is known to affect pollinator choice, and color variation is associated with variation in pollinator types (reviewed in

Willmer 2011). In addition, variation in flower color is linked to variation in secondary metabolites, some of which, like anthocyanins, can deter feeding by insects (Fineblum and

Rausher 1997, Irwin et al. 2003, Strauss et al. 2004). Flower color may be selected upon independently of other traits and has the potential to be selected by both pollinators and herbivores because it is tied to both attractiveness and plant secondary chemistry (Frey 2004,

Strauss et al. 2004). GBT, at the lowest elevation, and ELK, at the highest elevation, were the

24 most different from one another in flower color and also were most divergent in pollinator community composition (Table 2-1, Fig. 2-2) and ant visitation.

Not surprisingly, taller inflorescences bore more flowers, and flowers on these inflorescences tended to be larger (Table 2-1). However, both flower size and number varied among populations in ways that were similar between years (Table 2-1). The strong relationship with inflorescence height and flower number suggests that flower size is somewhat plastic

(Frazee and Marquis 1994, Conner and Sterling 1996). Corolla width and depth were correlated in all populations and years, which suggests that the shape of flowers may be less variable than their size. Increased flower size and number can increase pollinator visitation, but this effect may differ between types of pollinators (Conner and Rush 1996).

Common pollinators and pollinator community differences:

At the four sites studied, E. capitatum has a generalized pollination system. It is visited by a wide variety of insects that come into contact with reproductive parts, and eight of the nine functional groups identified contain multiple species. Whether most pollination systems are generalized or specialized is still debated (Waser et al. 1996, Johnson and Steiner 2000, Vazquez et al. 2009), but studies that examine plants in different populations and in the same population over time find that multiple insects act as pollinators to some extent (Herrera 1988, Brunet

2009), and other Brassicaceous species have been found to have generalized pollination systems

(Gomez and Zamora 1999, Gomez et al. 2009). Generalized pollination has been proposed as a likely outcome of pollinator communities that vary in time and space (Waser et al. 1996, Dilley et al. 2000, Ellis and Johnson 2009). However, generalized pollination can select for divergent floral morphologies when populations have somewhat different pollinator community compositions, or when pollinator behavior differs among populations (Dilley et al. 2000).

Differences in pollinator community composition among populations suggest that divergence as a result of differing pollinator preferences or effectiveness is a possibility in E. capitatum.

Pollinator community composition differed qualitatively among sites (Table 2-3, Fig. 2-

2). Flies are more common visitors at high-elevation sites, and bees and pollen beetles are more common visitors at low-elevation sites. Although it was not possible to test differences quantitatively because of limited sample size in some sites and some years, the differences in pollinator community are consistent with the results of other studies. For example, Kearns

(1992) found that in the Rocky Mountains, flies are more frequent visitors than bees at higher elevations, and flies are frequently found to be common visitors in high elevation sites in other mountain ranges, as well (Arroyo et al. 1982, Elberling and Olesen 1999, Devoto et al. 2005).

The frequency of pollen beetle visits differed greatly among populations (Fig. 2-2), but this difference may reflect a behavioral difference between high and low elevation beetle populations; beetles are often hidden inside of corolla tubes and hard to observe. Their increased frequency at lower elevations may reflect higher rates of movement from corolla tubes rather than higher abundance. Such differences are possible in other functional groups as well, but are likely to be more pronounced in these beetles because they spend time in corolla tubes and may be more affected by differences in temperature.

Other than corolla color, trait differences among populations did not appear to be associated with elevation (Table 2-1). Similarities in communities of pollinating visitors are related to elevation, with bees decreasing and flies increasing as a proportion of the total pollinators (Table 2-3). The lack of congruence in variation between plant traits and pollinator types suggests that pollinator communities are not shaping flower size or number in these

26 populations and raises the possibility that other selective agents, possibly herbivores, are more likely to be driving selection for flower number and size than pollinator communities.

Herbivory and other flower visitors:

Ants were not considered to be part of the “pollinator community” in analyses of community similarity. Nonetheless, they varied as a proportion of the total number of visitors from site to site, increasing as a proportion of the total visitors in higher elevation sites. They were the most frequent visitors to inflorescences in all populations and removed visible nectar quickly in the morning before other visitors were active. This may not affect pollinator preferences: most bees collect pollen rather than nectar from flowers, and most flies did not appear to be collecting nectar during visits as they neither entered corrolla tubes nor approached flowers from the back. Although nectar removal by ants may not affect attractiveness to other visitors, other ant behavior may. In ELK and NED, aggressive ants were observed attacking approaching bees on several occasions. Bees and ants have only been observed on the same plant on a few occassions, so the frequency of this behavior is unknown.

For other taxa where ants constitute the most common or constant visitors, and where plants produce dense flowers low to the ground, ants may be effective pollinators (Gomez 2000,

Ashman and King 2005). Ants can kill pollen with chemical secretions on their bodies (Beattie et al. 1985), however, viability tests of E. capitatum pollen suggest that it may be less susceptible than pollen of other plants (data not shown). This may be of little importance, as ants only rarely contacted reproductive parts, but they are such frequent visitors that they may perform some pollination service. In addition to possible effects as pollinators, ants may benefit plants by deterring some herbivores, though such interactions were never observed in the field.

The incidence of galls differed among years, but was not different among sites. Height was a significant predictor of the number of galls, perhaps because height is strongly related to the number of flowers produced (ie. sites for galls). However, WAK and GBT had greatly increased numbers of galls in 2006 compared to 2005, despite having much shorter plants, on average. NED and ELK had roughly equivalent numbers of galls from year to year, despite lower average height in 2006. Although the effects of non-pollinating gall-forming insects that interact with flowers have been studied in fig-fig wasp mutualisms (eg. Kerdelhue and Rasplus

1996, Wang et al. 2010), the effects of flower bud galls on pollination in other systems are relatively unstudied. They have the potential to both affect fitness directly, through decreased ovules for fertilization, and indirectly, through reduced pollinator attractiveness.

Unlike galls, which are formed by gall midges (Cecidiomyiidae), petal damage is caused by a wide variety of insects. Beetle larvae (Nitidulidae), which spend most of their time in flowers, are common sources of damage. However, other insects, including megachilid bees, lepidopteran larvae and grasshoppers, damage petals of E. capitatum. The types of damage caused by all of the herbivores that feed on petals are difficult to distinguish, and petal damage thus reflects the preference and abundance of a number of insects. In general, the study populations had relatively equal amounts of petal damage. Because the insects responsible for damaging petals were rarely observed, differences among populations in assemblages of petal- feeding herbivores could not be evaluated.

With the exception of an increase in ant visits as a proportion of total visits, herbivory on flowers differed more between years than among populations. Several types of herbivory are likely to have significant impacts on plant fitness. Galls prevent flowers from opening and setting fruit, so investment in those flowers is lost, and reduced floral display size could have impacts on the

28 attractiveness of plants to pollinators. Both petal herbivores and ants may impact fitness directly through loss of resources and indirectly through reduction in pollinators (Burkle et al. 2007, McCall

2010).

Conclusions:

Populations of E. capitatum in this study region differed in phenotype and pollinator community composition, with bees being more common visitors at lower elevations and flies being more common at higher elevations. Populations experienced different incidences of galling insects in different years, and the relative frequency of ant visitors increased with elevation, but otherwise, differences in herbivory were more apparent between years than among populations. Trait differences other than color were patchy, rather than continuous with elevation. It is possible that differences among populations in pollinator visitation and temporal differences in herbivory have acted to cause differentiation in plant traits.

2005 GBT WAK NED ELK Ant 0.24 0.46 0.45 0.85 Flies 0.02 0.12 0.04 0.23 Bees 0.13 0.07 0.06 0.04 Meligethes NA 0.04 0.00 0.00 Total Pollinator 0.15 0.23 0.10 0.27

2006 GBT WAK NED ELK Ant 0.16 2.39 0.04 0.50 Flies 0.09 0.61 0.01 0.19 Bees 0.02 0.52 0.01 0.18 Meligethes 0.07 0.10 0.00 0.02 Total Pollinator 0.18 1.34 0.01 0.47

APPENDIX 2-1. Visits / flower / hour, calculated using the average number of flowers per plant × the number of plants. Ant visits increased as a proportion of total visits in higher elevation populations, and this did not seem to decrease the overall visitation rate by pollinators.

CHAPTER III

THE GOOD, THE BAD AND THE FLEXIBLE: PLANT INTERACTIONS WITH POLLINATORS AND HERBIVORES OVER SPACE AND TIME ARE MODERATED BY PLANT COMPENSATORY RESPONSES

ABSTRACT

Plants are sessile organisms that face selection by both herbivores and pollinators.

Herbivores and pollinators may select on the same traits and/or mediate each others’ effects.

Erysimum capitatum (Brassicaceae) is a widespread and variable plant species with generalized pollination that is attacked by a number of herbivores. I addressed the following questions: (1)

Are pollinators and herbivores attracted by similar plant traits? (2) Does herbivory affect pollinator preferences? (3) Do pollinators and/or herbivores affect fitness and select on plant traits? (4) Do plant compensatory responses affect the outcome of interactions among plants, pollinators and herbivores? (5) Do interactions among E. capitatum and its pollinators and herbivores differ among sites and years? In 2005 and 2006, I combined observational and experimental studies in four populations at different elevations to examine selection by pollinators and herbivores on floral traits of E. capitatum. Pollinator and herbivore assemblages varied spatially and temporally, as did their effects on plant fitness and selection. Both pollinators and herbivores preferred plants with more flowers, and herbivory sometimes reduced pollinator visitation. Pollinators did not select on plant traits in any year or population and E. capitatum was not pollen limited, however, supplemental pollen resulted in altered plant resource allocation. Herbivores reduced fitness and selected for plant traits in some populations, and these effects were mediated by plant compensatory responses. Individuals are visited by diverse

31 groups of pollinators and herbivores that shift in abundance and importance in time and space.

Compensatory reproductive mechanisms mediate interactions with both pollinators and herbivores and may allow E. capitatum to succeed in this complex selective environment.

INTRODUCTION

The evolution of reproductive traits in animal-pollinated plants has been increasingly viewed in light of natural selection by both pollinators and herbivores (Louda and Potvin 1995, e.g. Armbruster 1997, Herrera et al. 2002, Irwin et al. 2004, Armbruster et al. 2009, Gomez et al.

2009, Whittall and Carlson 2009). Pollinators and herbivores can interact in several ways to simultaneously influence the evolution of plant traits. First, selection by herbivores can be so strong that it precludes the possibility of selection by pollinators (e.g. Herrera 2000). For instance, selection by ungulates feeding on flowering stalks overrides selection by pollinators in both Erysimum mediohispanicum and Hormathophyllum spinosum (Gomez and Zamora 2000,

Herrera et al. 2002, Gomez et al. 2009). Second, the same or correlated traits can be under selection by both pollinators and herbivores, producing adaptive compromises and exaptations

(Galen and Cuba 2001, Ehrlen 2002, Irwin et al. 2003, Adler and Bronstein 2004, Armbruster et al. 2009). For example, ants which reduced fitness and bumblebees which increased fitness in

Polemonium viscosum preferred similar floral characteristics, producing counteractive selective effects (Galen and Cuba 2001, Galen and Butchart 2003). Third, herbivory can alter plant attractiveness to pollinators directly, by damaging attractive structures, or indirectly, through plant responses to damage (e.g. Juenger and Bergelson 1997, Krupnick et al. 1999, McCall and

Irwin 2006, McCall 2008, 2010). Pollinators often prefer flowers that are undamaged, and in

32 many cases, any form of floral damage may reduce pollinator visitation (Adler 2000, McCall

2010).

Plant responses to herbivory may either limit or exacerbate the effects of damage.

Herbivory may result in compensatory responses such as increased flower production (Paige and

Whitham 1987, Agrawal 2000, Juenger and Bergelson 2000), reducing any negative effects of damage on plant-pollinator interactions. However, plants also may reduce flower production when damaged (Krupnick et al. 1999, Sharaf and Price 2004). Krupnick, et al. (1999), for example, found that floral herbivory caused plants to produce fewer flowers and receive fewer pollinator visits. Such developmental responses are critical components of the complex interplay among plants, pollinators, and herbivores, but rarely have been addressed in multiple populations in the context of both pollinator visitation and herbivory.

Pollinator and herbivore assemblages often vary among plant populations (Herrera 1988,

Kearns 1992, Dilley et al. 2000, Rand 2002, Bradley et al. 2003, Aigner 2005, Gomez et al.

2010) and can create geographically distinct selection regimes that result in phenotypic differentiation among plant populations (Thompson 1988, Rey et al. 2006, Gomez et al. 2009).

This may be particularly true for plant species that occupy montane regions, which are very heterogeneous over short distances in both micro-climate and co-occurring species (reviewed in

Lomolino 2001). In addition, elevational gradients often have been associated with population differentiation in plants (Linhart and Grant 1996). For this reason, mountainous regions provide excellent opportunities to examine the effects of different pollinator and herbivore communities on plant traits.

I combined observational and experimental studies across an elevational gradient in the

Colorado Rocky Mountains to examine natural selection by pollinators and herbivores on traits

33 of Erysimum capitatum. E. capitatum is a widespread native species that varies in flower color, flower number, size, and other characteristics throughout its range (Price 1984). Preliminary observations of this species suggested it was visited by many different pollinators and herbivores, and that the composition of pollinator and herbivore communities varied among sites.

I addressed the following questions: (1) Are pollinators and herbivores attracted by similar plant traits? (2) Does herbivory affect pollinator preferences? (3) Do pollinators and/or herbivores affect fitness and select on plant traits? (4) Do plant compensatory responses affect the outcome of interactions among plants, pollinators and herbivores? (5) Do interactions among E. capitatum and its pollinators and herbivores differ among sites and years?

MATERIALS AND METHODS

Study species: Erysimum capitatum (Brassicaceae) is native throughout most of the United

States (USDA 2004) and northern Mexico in lowlands and montane areas (Turner 2006). It is often a biennial, although life history varies among individuals and populations, and requires insect visitation for full fruit set (Price 1984). Plants flower from early spring to mid-summer

(USDA 2004), and populations of E. capitatum are characterized by variation in flower color and other traits across the species range (Price 1984, Weber 1990). Flowers may be white, lavender, or range from yellow to red (Price 1984). All populations and individuals studied are recognized as a single species. The taxonomy and identification of E. capitatum is currently under debate, however, and the great phenotypic variation within this species may be the result of hybridization events between subspecies (Turner 2006).

Populations: Four study populations were located in the Front Range of the Eastern slope of the

Rocky Mountains of Colorado. The Elk Meadow population (ELK) is in a subalpine dry

34 meadow at 2900 m. The population at Nederland (NED) is in open lower montane Ponderosa pine forest at 2590 m. The Walker Ranch population (WAK) is in open lower montane

Ponderosa pine forest at 2209 m. The Greenbelt Plateau population (GBT) is in a short-grass prairie at 1980 m. Complete descriptions of these vegetation types are in Marr (1967). In the lowest-elevation population, GBT, flowering begins as early as March, and continues through early June. In the highest-elevation population, flowering begins in late June and continues through mid-July, though some plants have flowers as late as August. At each location, I performed a combination of observational and experimental studies.

Observational study and selection analysis

To examine the nature of pollinator and herbivore preferences, and their relationship with fitness, I measured plant traits, pollinator and herbivore visitation, and plant fitness in the four study populations on 50-200 haphazardly selected flowering plants in 2005 and 2006. Sample sizes differed among population and years for several reasons. First population size and density differed both among populations and between years. In addition, I attempted to choose plants before they had bolted in 2006 in NED and WAK, and some of these plants did not bolt.

Phenotypic characters: I measured corolla width, corolla tube depth, petal color, number of open flowers, stem diameter, and the total number of flowers that had opened by harvest. The count variables (open and total flowers) were square-root transformed to improve normality.

I quantified flower color by comparing flowers to color chips (Behr paint chips of colors yellow S-G-390 to red S-G-230) in the field, and confirmed the accuracy of this method using digital photographs and standardized RGB color values. Yellower color was quantified as a larger number and redder color as a smaller number. Flower width and flower depth were measured in the field with digital calipers. Width was the diameter of the flower from the tip of

35 one corolla lobe to the tip of the opposite lobe, and depth was the distance from the base of the calyx to the angle between the claw and the limb of the corolla. Flowers were considered to be

“open” if they retained at least 3 petals (flowers with damaged limbs damaged were counted as

“open” if they retained at least 3 attached claws). Inflorescence height was measured from the ground to the apex of the tallest inflorescence. Heights were measured at harvest so that all plants would be at the same developmental stage. Stem diameter was measured 30 mm above the ground on the tallest inflorescence.

Measurements of plant characteristics were repeated on up to four days, or on multiple flowers per day, though it was not possible to obtain equal repetitions of all measurements on all plants due to weather and flower availability. In 2005, I measured corolla width and depth for the oldest and newest open flowers, and one haphazardly chosen flower in between. When plants had 3 or fewer open flowers, all were measured. Preliminary analyses of 2005 data suggested that these multiple measures were unnecessary, as the difference in floral characters among plants was much greater than differences in floral measurements on the same plant, so in

2006, I measured corolla width and depth on one flower per plant.

I collapsed multiple measurements for all phenotypic characters into a single measurement per character per plant. For characters such as corolla width, where differences among measurements due to intraplant variation or measurement error were likely to be the biggest contributors to within-plant measurement differences, I used the averages of all measurements per plant as the character values. For characters such as inflorescence height, for which the greatest contributor to differences among measurements was likely to be plant growth, I used the maximum recorded value for each plant as the character values.

Sampling schemes differed slightly between years. In 2005, I attempted to choose plants that would have open flowers throughout the planned period of pollinator observations. In 2006,

I increased sample sizes in many sites. Because of this change, plants were larger, on average, in

2005 than in 2006. This may have changed estimated effects of plant size variables if there were non-linear effects of size characters, which I did not test for.

Insect visitation: To quantify pollinator preferences and the relationships between visitation and fitness, individual observers watched groups of 1-10 plants for10-minute periods between 9

AM and 3 PM, which covered peak visitation times. Observations of each plant were repeated twice per day for as many days as was possible given weather, availability of observers, and availability of flowering plants (total number of minutes per population are in Appendix 1).

Visitors were identified to functional groups (described below) to reduce training time for observers, and insects were not collected until all observations had been completed to avoid influencing visitation on subsequent days. Ants were included regardless of where on the plant they were observed, because tracked individual ants always collected nectar, but rarely contacted reproductive parts.

Visitors were assigned to the following functional groups: large bees, medium-sized bees, small bees, flies, pollen beetles, and ants. Because there were very few visitors from some bee size classes, I added all bee visitors together in the final analysis.

The number of observations differed among plants due to weather, and individual plant flowering times. To incorporate the number of observations into a measure of visitation by each functional group, I added 0.10 to the total count of visitors from a particular functional group and divided by the number of observations carried out, then log-transformed this measure of visitation to improve normality. Any functional groups for which 26 or fewer visits were

37 observed in a given year and site were excluded from analyses because I determined that this sample size was too small to make meaningful estimates of parameters.

Herbivory: I studied associations between herbivore damage, plant traits, and fitness. Three general types of herbivory were common at all sites: petal damage, flower bud galls, and nectar robbing by ants. A fourth type of herbivore, coleopteran stem borers, was common only at

WAK.

A variety of petal herbivores created visually similar damage, so there was no reliable way to determine what herbivore had damaged a flower and petal damage from all consumers was measured together. Damage levels for individual flowers were recorded. Petal limb damage was scored 1 to 4, and 5 indicated complete destruction of anthers and stigma. Level “1” corresponded to 10% damage or less, “2” to damage between 10% and 25%, “3” to 25-50%, “4” to 50% or more. I then log-transformed the sum of all damage scores from all flowers for each plant for the day of the season with the most petal damage.

Gall midge larvae (Cecidomyiidae) form galls from flower buds and typically prevent flowers from opening. I counted flower bud galls on each observation day and used the log- transformed maximum count for galls found on a single day over the observation period in analyses because galls remained on the plant throughout observations.

Nectar removal by ants was measured as the number of ant visitors observed, as explained in the previous section.

Stem borers were unidentified beetle larvae that typically killed whole inflorescences.

When this type of herbivory could be distinguished from mammalian herbivory, it was included in analyses.

Mammalian herbivores, chiefly pocket gophers and ruminants, damaged standing inflorescences in the three highest elevation locations (WAK, NED and ELK). Because both ruminants and gophers often removed tags or rendered them illegible, it was difficult to tell whether plants were still alive but had lost identifying information or if they had been killed by mammalian herbivores. Therefore, study plants were removed from analysis whether they were clearly affected by mammalian herbivores or if their tags could not be found. Overall, between

10 and 25 plants were removed from analyses for WAK, NED and ELK in each year.

Fitness: Plants were collected after senescence. The total number of flowers produced by plants was obtained by counting fruits and scars of flowers that reached anthesis but aborted. To estimate female fitness, the lengths of each silique were measured and added together. The number of seeds per silique is positively correlated (R2=0.69, P<0.01) to the length of the siliques (C Lay, CU Boulder, unpubl. res.), and using this measure allowed me to measure the fitness contribution of dehisced siliques. Estimated fitness was total silique length; this value was log-transformed to improve normality.

Analysis of visitation, herbivory, and fitness: For each population and year where there were sufficient visitation data, structural equations modeling (SEM) was used to examine the relationship between visitation, herbivory, and plant traits. I created an inclusive model and compared nested models to more inclusive ones. Because GBT 2005 was heavily damaged by cattle and was missing all fitness data, it was impossible to run multigroup analysis (Grace 2006,

Rey et al. 2006) to determine whether the same model would apply to all populations.

Preliminary analysis suggested that populations and years were too dissimilar in the types of visitors and herbivores present to use the same model for all populations and years, and the saturated model was unidentified in some populations and years. Because of these issues, I

39 created separate inclusive models for each population and year. The most-inclusive models were intended to be as similar to one another as possible, given population differences. The most- inclusive models for each population are shown next to the best-fit models in the results. All structural equations modeling (SEM) was performed in the sem package in R (Fox et al. 2010), which uses full-information maximum likelihood to estimate path strengths. I calculated covariance matrices using pairwise complete observations, and interpreted standardized path coefficients. Independent variables were centered (by subtracting the mean) to allow comparison of parameter estimates among populations, and some were transformed to improve normality.

For each population, I chose the simplest model that had a significantly lower Chi-square value than simpler models. Only significant paths are included in Figs 1-5. Models with lower Chi- square values are a better fit to the data than models with higher Chi-square values. Although transformations improved normality of the variables on which they were used, the transformed variables are not normal, and this can affect the validity of p-values (Tabachnick and Fidell

1996b). I therefore show the best-fit models even when the Chi-square value is significant and include corrected values of Akaike’s Information Criterion (AICc).

Experimental Manipulations

To test the importance of pollinator visitation for plant fitness, the effect of herbivory on plant fitness, and any joint effects of herbivores and pollinators, I added pollen and insecticide on

50-200 haphazardly selected plants in 2005 and 2006. Where possible, I used four treatments in a fully factorial two-way design: control, hand pollination, insecticide application, and insecticide application plus hand pollination. In GBT and ELK, insecticide use was not possible due to environmental concerns, and only hand pollination and control groups were included.

Plants were haphazardly assigned to treatments, and some were lost throughout the season. To

40 add pollen, I collected anthers from 10-15 nearby plants, mixed them in a glass vial, then applied the pollen to receptive stigmas with a paintbrush. Vials were replenished as needed from nearby plants. Insecticide was added to plants before observations and pollinations began. In 2005, pollen was added every other day, but greenhouse experiments showed that pollinating every third day was sufficient for pollinating all the flowers on a plant, so pollen was added every third day in 2006.

In 2005 and 2006, different insecticides were used. In 2005, I used a topical insecticide spray consisting of 0.5% fenbutatin and 8% acephate, and sprayed all other plants with water to simulate insecticide addition. I found this treatment was not very effective in deterring herbivory and switched to a systemic insecticide (Marathon 1% granular imidacloprid) insecticide in 2006 which was added and watered in to the bases of plants at the beginning of flowering (some plants had already bolted). In GBT and ELK, it was not possible to use insecticide due to environmental and/or permitting considerations, and only pollen addition was performed.

Estimated female fitness was measured as in the observational study. For experimental plants, I also calculated the average length of siliques produced by each plant, the total number of flowers, and the number of aborted flowers. Counts of total and aborted flowers were square- root transformed to improve normality in all populations and years, and average silique length was square-root transformed where it was variable enough to create non-normal distributions of residuals.

Analysis of experimental results: ANCOVA with height of the tallest inflorescence at harvest a covariate was used to test the effects of treatment, and, in some populations, galls, on total fitness.

Analysis of compensatory response: To determine whether plants arrived at similar fitness outcomes through variation among the components of total silique length (total flowers, aborted flowers, and average silique length), I included these as dependent variables in a multivariate analysis of covariance (MANCOVA), with height as a covariate. In populations where the number of galls present on experimental plants was measured, the count was included as a covariate and square-root transformed when it was variable enough to create non-normal distributions of residuals. Inflorescence height at harvest was centered by subtracting the mean from all values. The MANCOVA allowed us to identify independent variables that affected any of the fitness components so I could examine the nature of those changes in further univariate stepdown analyses.

Because the purpose of the MANCOVA was to determine which variables to include in further stepdown analysis, I interpreted the MANCOVA significance tests of type II sums of squares. Significance tests of type II sums of squares are valid for all effects in models that have no significant interaction terms, and for interaction terms in models with significant interactions.

They are sometimes preferable because main effects are not calculated while controlling for interactions (Fox 2008). I identified potential multivariate outliers using the mvoutlier package in R (Gschwandtner and Filzmoser 2009), and univariate outliers using plot from the base package (R Development Core Team 2010b). Although there were several apparent outliers, they did not affect the outcome of significance tests so no observations were removed.

When there were significant effects of independent variables on the dependent variables, those independent variables were included in univariate ANCOVA stepdown analyses for each dependent variable. I based this procedure on Roy-Bargmann stepdown analysis (Tabachnick and Fidell 1996a). Stepdown analysis is used to determine how independent variables affect

42 dependent variables. Dependent variables of most interest or importance (in this case aborted flowers) are regressed on independent variables, and are then included as independent variables in analyses of dependent variables of lesser interest. I analyzed aborted flowers first, then total flowers, then average silique length. I included observations even if it was not possible to measure one of the dependent variables, so in some cases, the sample size is different for different variables. I used significance values from type III sums of squares in the stepdown analysis to improve interpretation main effects. Stepdown analyses were performed with glm in

R (R Development Core Team 2010b). Because Bonferonni adjustments did not change the interpretation of significance levels for total flowers and aborted flowers, and average silique length was not strongly correlated with the other two dependent variables (Appendix 3-2), I interpreted non-adjusted p-values.

The details of sampling and analysis differed slightly among populations for both observational and experimental studies. These differences are explained in the results in figure and table captions for each year and population.

RESULTS

Summary of results for pollinator and herbivore preferences for each population and year:

Pollinator and herbivore preferences (Observation): I used the best-fit models produced by observational study to determine pollinator and herbivore preferences and relationship between herbivory and pollinator visitation. Comparisons of most-inclusive and best-fit models for each site and year are shown in Table 3-1.

Fitness effects of pollinators and herbivores (Observation and Experiment): Best-fit models of data from the observational study show selection (paths that link traits to fitness, sometimes

43 TABLE 3-1. Model fit statistics for null, most-inclusive and best-fit models for observational study. The null model contains all error variances and can be thought of as the least-inclusive model. Lower X2indicate better-fitting models.

Model X2 DF P AICc Model X2 DF P AICc GBT GBT 2005 2006 Null 102.69 28 Null 317.63 55 Inclusive 11.08 6 0.08 167.75 Inclusive 29.08 17 .03 217.54 Best-fit 25.35 19 0.15 76.89 Best-fit 54.03 40 .07 123.36 WAK WAK 2005 2006 Null 90 45 Null 412.35 66 Inclusive 33.96 15 <0.01 155.28 Inclusive 102.64 28 <0.001 252.13 Best-fit 47.3 30 0.03 109.64 Best-fit 122.99 39 <0.001 227.94 NED NED 2005 2006 Null 291.13 45 -- Inclusive 25.08 14 0.03 162.67 -- Best-fit 36.52 31 0.23 99.05 -- ELK ELK 2005 2006 Null 170.55 45 Null 374.33 55 Inclusive 33.87 14 <0.01 190.41 Inclusive 38.44 17 <0.001 230.57 Best-fit 52.35 34 0.02 107.48 Best-fit 61.76 36 0.004 146.98

44 through pollinator visitation and herbivory). Pollinators and herbivores can affect fitness (total silique length) without selecting on measured plant traits if there are no links between traits and visitation. Paths that go from traits to pollinator visitation and herbivory to fitness are taken as evidence of selection by visitors on particular traits. Fit statistics, including corrected AIC values, for complete sets of tested models are given in Appendices 3-3 and 3-4 as are the most- inclusive models for each population and year (Appendices 3-5 to 3-6). Experimental additions of pollen and insecticide also identified fitness effects of pollinators and herbivores. Main effects of pollen, insecticide and plant size, as well as significant interaction terms are detailed below, and full tables of the ANCOVA analyses are in Appendix 3-7.

GBT 2005 (Fig 3-1A)

Pollinator and herbivore preferences: Plants with more flowers at the beginning of the study received more bee visits, ant visits and had sustained more petal damage before observations began. Plants with wider, shallower flowers received more bee visits, but these characters did not affect herbivory.

Fitness effects of pollinators and herbivores: Plants were trampled by cattle and neither the fitness of observational plants nor the effect of experimental pollen manipulations could be assessed.

GBT 2006 (Fig 3-1B)

Pollinator and herbivore preferences: Bees preferred plants with wider flowers. Plants with more flowers produced over the course of the season received more beetle visits but also had more galls. Ants preferred plants with more open flowers, and plants with more open flowers had more petal damage, but the number of open flowers did not affect pollinator visitation.

Fitness effects of pollinators and herbivores: Estimated fitness was directly increased by total number of flowers (Fig 3-1B). Plants with more total flowers had more galls, and plants with more galls were less fit than plants with fewer galls (Fig 3-1B). There were no effects of ants, petal damage or pollinator visitation on fitness (Fig 3-1B). Experimental treatments (data from

47 supplemental pollen and 45 control plants were used) showed that pollination did not change fitness as measured by total silique length (β=0.29, F1,88=0.10, P=0.37) when controlling for height (β=0.11, F1,88=41.99, P<0.0001). Insecticide was not applied at this site; treatments included only control and supplemental pollen addition.

WAK 2005 (Fig 3-1C)

Pollinator and herbivore preferences: Fly visits and petal damage were positively related to the total number of flowers. Galls were not common enough to measure and preferences of borers were not tested.

Fitness effects of pollinators and herbivores: Plants with more total flowers had greater fitness

(Fig 3-1C). Stem borers reduced fitness (Fig 3-1C), but their preferences were not measured in this year, so there is no evidence for selection by borers on any measured trait. I found no effects of ants, petal damage or pollinator visitation on fitness (Fig 3-1C), and galls were not common enough to include in this year.

Experimental treatments (data from 26 control, 29 supplemental pollen, 24 insecticide and

25 supplemental pollen and insecticide plants were used) showed that fitness was not limited by pollen availability (β=-0.06 F1,97=0.11, P=0.97). Fitness increased with height (β=0.06,

F1,97=35.76, P<0.0001) but did not change with insecticide application (β=0.18, F1,97=1.51,

P=0.22).

WAK 2006 (Fig 3-2A):

Pollinator and herbivore preferences: Plants with more open flowers had more bee visits, ant visits, and damage. Bees preferred redder, narrower, deeper flowers, while ants preferred wider, shallower flowers. Plants with more total flowers had more galls. Borers tended to attack plants with thicker stems.

Fitness effects of pollinators and herbivores: Total flowers increased fitness (Fig 3-2A). Stem borers preferred thicker stems and reduced fitness. Therefore, borers select for plants with narrower stems (Fig 3-2A). There was no selection by pollinators, ants, or petal herbivores (Fig

3-2A). Experimental treatments included 51 control, 22 supplemental pollen, 53 insecticide and

23 supplemental pollen and insecticide plants. Galls were included as a predictor variable in analyses of treatment effects. In general, female reproduction was not pollen limited (β=0.43,

F1,140=0.90, P=0.35), although supplemental pollen improved fitness for plants with galls (β=

0.45, F1,140=9.07, P<0.01). Fitness decreased with galls (β= -0.18, F1,140=18.93, P<0.0001).

There was no main effect of insecticide (β=0.06, F1,140=0.08, P=0.77), but fitness improvements associated with height at harvest (β=0.12, F1,140=55.06, P<0.0001) increased on plants treated with insecticide (β=0.12, F1,140=12.67, P<0.001).

NED 2005 (Fig 3-2B)

Pollinator and herbivore preferences: Bees and ants preferred plants with more flowers, and damage was more common on those plants. In addition, ants visited flowers that were wider and redder more frequently. Petal damage also was higher on plants with wider, yellower flowers.

Galls were more common on plants that produced more flowers overall. Although both bee visits and petal damage were associated with greater numbers of open flowers, bees were less likely to visit plants with petal damage.

FIGURE 3-1. Best-fit models from structural equations analysis of GBT 2005 (A), GBT 2006 (B), and WAK 2005 (C). (A) In GBT 2005, the number of open flowers (*) and petal damage were measured on the day before beginning observations. Data on galls and insect visitation were collected between 17 May and 26 May on 75 plants not included in experimental treatments. Neither total silique length nor total flowers could be measured because all plants were trampled by cattle. (B) In GBT 2006, measurements were taken for 100 plants not included in experimental between 15 May and 23 May. (C) In WAK 2005 data were taken for 100 plants not included in experimental treatments between 1 Jun and 23 Jun; petal damage and open flowers were measured on 50 of these plants.

FIGURE 3-2 (A) In WAK 2006, data were collected for 270 plants, including plants treated with supplemental pollen when it was shown not to affect fitness, were taken between 3 Jun and 20 Jun. (B) In NED 2005, measurements were taken on 99 plants not included in experimental treatments between 18 Jun and 30 Jun. (C) In ELK 2005 data were collected on 100 control plants between 1 Jul and 20 Jul. (D) In ELK 2006, data were collected for 187 plants, including plants treated with supplemental pollen between 20 Jun and 30 Jul. For 98 plants on which pollen was not added, I observed insect visitation.

Fitness effects of pollinators and herbivores: Fitness increased with total flowers but was not otherwise affected by variables included in structural equations models (Fig 3-2B).

Experimental treatments included 27 control plants, 26 supplemental pollen plants, 27 insecticide plant and 24 plants to which both pollen and insecticide were added. In these plants, fitness increased with height at harvest (β=0.04, F1,96=47.47, P<0.0001). There were no effects of pollination (β= 0.11, F1,96=0.45, P=0.50) or insecticide (β= -0.01, F1,96=0.08, P=0.92).

NED 2006

Pollinator and herbivore preferences: Observations of visitors were precluded by wet and cool weather in this year and site, so preferences were not modeled.

Fitness effects of pollinators and herbivores: Preferences of pollinators and herbivores were not measured in this year, and observation plants were used as controls in experimental due to poor weather conditions for insect observations. Experimental treatments included 22 control plants,

20 supplemental pollen plants, 22 insecticide plants and 21 plants to which both pollen and insecticide were added. Galls were included as a predictor variable in analyses of treatment effects. Experimental results showed that fitness increased with height at harvest (β=0.11

F1,75=57.76, P<0.0001) and decreased with galls (β= -0.21, F1,75=22.52, P< 0.0001), but was not affected by supplemental pollination (β= -0.09, F1,75=0.24, P=0.63), or insecticide (β= -0.23,

F1,75=1.37, P=0.25).

ELK 2005 (Fig 3-2C)

Pollinator and herbivore preferences: Too few bees visited to be included in this year. Flies and ants visited plants with more open flowers, and these plants had more petal damage. Flies visited plants with yellower flowers and ants visited plants with wider flowers more often. Galls were positively associated with total flowers.

Fitness effects of pollinators and herbivores: Fitness was affected only by the total number of flowers (Fig 3-2C). Weather and phenology precluded pollination treatments in this site and year.

ELK 2006 (Fig 3-2D)

Pollinator and herbivore preferences: Flies, bees and ants preferred plants that had more flowers on observation days. Ants visited plants with wider flowers more frequently. More

50 damage occurred on plants that had narrower flowers. Flies were less likely to visit plants with galls, and bees less likely to visit plants with substantial petal damage.

Fitness effects of pollinators and herbivores: Petal damage was associated with reduced fitness

(Fig 3-2D). Plants with more total flowers had more galls and galls reduced fitness (Fig 3-2D).

Experimental treatments included 81 control plants and 76 plants to which supplemental pollen were added. Insecticide was not used in this site. Galls were included as a predictor variable and were variable enough to require square-root transformation. For experimental plants, fitness decreased with galls (β= -0.33, F1,151=10.53, P<0.01). Pollination did not improve fitness (β=

0.51, F1,151=3.09, P>0.05). The positive effect of height (β=0.08, F1,151=53.14, P<0.0001) was reduced by added pollen (β= -0.09, F1,151=6.45, P<0.05), suggesting shorter plants may have been pollen-limited although taller plants were not.

Analysis of plant compensatory responses to herbivory and pollination (Experiment): To determine how plant compensatory responses contributed to fitness in the manipulative experiments, I tested how inflorescence height at harvest, galls, supplemental pollination and insecticide changed at least one of the fitness components including flower abortion, flower production and allocation to individual fruits (aborted flowers, total flowers and average silique length). This was done using MANCOVA and stepdown analysis for three components of fitness (total flowers, aborted flowers, and average silique length) from experimental plants.

Fruit number was not included because the numbers of fruit and aborted flowers together determine total flower number. Full tables of MANCOVA results are in Appendix 8.

GBT 2005: Plants were trampled by cattle and no fitness components were measured.

TABLE 3-2. Stepdown results for fitness components in GBT 2006. All fitness components increase with inflorescence height at harvest. Total flowers increase with aborted flowers, and pollination reduces the number of aborted flowers on tall plants. β is the unstandardized regression coefficient, or effect size, for each relationship.

Height Pollination Pollination* Aborted Total Height flowers flowers Aborted β=0.10 β=0.02 β= -0.14 -- -- flowers F1,91=20.29 F1, 91<0.01 F1, 91=5.01 P<0.0001 P=0.97 P=0.03 Total β=0.11 β=0.32 β= -0.06 β=0.39 -- flowers F1,90=50.10 F1,90=1.46 F1,90=1.72 F1,90=31.81 P<0.0001 P=0.23 P=0.19 P<0.0001

Average β=1.04 β= -0.62 β= -0.71 β= -2.43 β=1.81 silique F1,89=10.76 F1,189=0.02 F1,189=1.12 F1,89=3.76 F1,89=1.17 length P<0.01 P=0.88 P=0.29 P>0.05 P=0.28

GBT 2006: When controlling for inflorescence height at harvest (Wilk’s λ=0.78, P<0.0001), there was no effect of pollination (Wilk’s λ=0.98, P=0.69) on total flowers, aborted flowers, or average silique length. Even though the interaction between pollination and height did not significantly affect dependent variables (Wilk’s λ=0.91, P>0.05), I tested it in the post-hoc stepdown analysis. Inflorescence height at harvest was associated with an increase in the number of aborted flowers, total flowers, and average silique length (Table 3-2). Main effects of pollination on flower abortion were not significant, but supplemental pollen decreased the positive association between height at harvest and aborted flowers (Table 3-2). The number of total flowers increased with aborted flowers (Table 3-2).

WAK 2005: Height at harvest (Wilk’s λ=0.47, P<0.0001) and insecticide (Wilk’s λ=0.88,

P<0.01) significantly affected one or more of the three fitness components, but neither pollination (Wilk’s λ=0.99, P=0.98) nor the interaction between pollination and insecticide

(Wilk’s λ=0.98, P=0.72) had an effect. Height at harvest was associated with an increase in the number of aborted flowers, total flowers, and average silique length (Table 3-3). Flower

52 production was higher on plants with more aborted flowers and on plants treated with insecticide

(Table 3-3).

TABLE 3-3. Stepdown results for fitness components in WAK 2005. All fitness components increased with inflorescence height at harvest. Total flowers increased with insecticide and aborted flowers. Height Insecticide Aborted Total flowers flowers Aborted β=0.03 β=0.27 -- -- flowers F1,102=8.41 F1,102=2.69 P<0.01 P=0.10

Total β=0.08 β=0.48 β=0.40 -- flowers F1,101=70.22 F1,101=10.60 F1,101=20.30 P<0.0001 P<0.01 P<0.0001

Average β=0.69 β= -1.57 β= -1.07 β= -1.02 silique F1,100=15.05 F1,100=0.53 F1,100=0.64 F1,100=0.55 length P<0.001 P=0.47 P=0.43 P=0.46

WAK 2006: I square-root transformed average silique length of these data to improve normality. Galls (Wilk’s λ=0.81, P<0.0001) and the interaction between insecticide and height at harvest (Wilk’s λ=0.90, P<0.01) affected the dependent variables. I tested the interaction between pollination and galls in the post-hoc stepdown analysis, even though it did not significantly influence variables (Wilk’s λ=0.94, P=0.08), because it affected total silique length.

Height at harvest was associated with an increase in total flowers and average silique length

(Table 3-4). Galls increased the number of aborted flowers, and decreased average silique length, but pollination attenuated the negative effects of galls on silique length (Table 3-4).

Total flowers increased with the number of aborted flowers (Table 3-4). Insecticide increased the positive effects of height on silique length.

53 TABLE 3-4. Stepdown results for WAK 2006. Height at harvest increased total flowers and silique length but not aborted flowers. Some siliques were broken at harvest, so sample sizes were smaller for measures of silique length. Supplemental pollen attenuated the negative relationship between galls and silique length.

Height Insecticide Pollination Galls Insecticide Pollination Aborted Total *Height *Galls flowers flowers Aborted β= -1.88x10-3 β=0.22 β= -0.13 β=0.10 β= -5.8x10-3 β=0.12 -- --

flowers F1,158=0.02 F1,158=1.97 F1,158=0.09 F1,158=7.12 F1,158=0.046 F1,158=0.70 P=0.89 P=0.16 P=0.76 P<0.01 P=0.83 P=0.40

Total β=0.10 β=0.03 β=0.09 β= -0.01 β=0.03 β= -0.02 β=0.61 --

flowers F1,157=7.12 F1,157=0.10 F1,157=0.11 F1,157=0.36 F1,157=2.43 F1,157=0.04 F1,157=144.89 P<0.01 P=0.74 P=0.75 P=0.55 P=0.12 P=0.84 P<0.0001

Average β=0.09 β=0.07 β= -1.39 β= -0.31 β=0.12 β= 0.72 β= -5.8x10-3 β= -0.09 -4 silique F1, 122=6.59 F1,122=0.05 F1,122=2.64 F1,122=14.75 F1,122=4.33 F1,122=7.82 F1,122=7x10 F1,122=0.11 length P=0.01 P=0.82 P=0.11 P<0.001 P=0.03 P<0.01 P=0.98 P=0.74

NED 2005: Insecticide (Wilk’s λ=0.87, P<0.01) and height at harvest (Wilk’s λ=0.51, P<0.001) significantly affected the three dependent variables, but pollination did not (Wilk’s λ=0.71,

P=0.55). Insecticide decreased aborted flowers (Table 3-5). Height at harvest increased total flowers (Table 3-5). Total flowers increased with aborted flowers (Table 3-5). Silique length increased with aborted flowers and height at harvest (Table 3-5).

TABLE 3-5. Stepdown results for NED 2005. Total flowers and silique length both increased with height at harvest. Plants to which insecticide was added aborted fewer flowers. Total flowers increased with the number of aborted flowers. Flower abortions were associated with longer siliques.

Height Insecticide Aborted Total flowers flowers Aborted β=0.01 β= -0.44 -- -- flowers F1,101=3.18 F1,101=9.44 P=0.08 P<0.01

Total β=0.06 β= -0.29 β=0.35 -- flowers F1,100=83.34 F1,100=2.75 F1,100=143.73 P<0.0001 P=0.10 P<0.0001

Average β=0.68 β=2.74 β=4.46 β= -2.37 silique F1,99=1.10 F1,99=6.26 F1,99=6.27 F1,99=2.52 length P<0.0001 P=0.30 P=0.0139 P=0.12

NED 2006: Height at harvest (Wilk’s λ=0.47, P<0.001) significantly affected the three dependent variables, as did the interaction between pollination and galls (Wilk’s λ=0.85,

P=0.02). In stepdown analysis, galls increased aborted flowers, and pollination also increased aborted flowers on plants with no galls (Table 3-6). Pollination reduced flower abortions on plants with galls, and the effect increased with the number of galls (Table 3-6). Total flowers increased with aborted flowers and height at harvest, but decreased with galls (Table 3-6).

Average silique length increased with height at harvest, but decreased with galls (Table 3-6).

TABLE 3-6. Stepdown results for NED 2006. All fitness components increased on taller plants. The number of aborted flowers increased with galls, and galls were related to a significant reduction in total flowers when controlling for the number of aborted flowers. Pollination had complex effects. Some siliques were broken at harvest, so sample sizes were smaller for measures of silique length.

Height Pollination Galls Pollination Aborted Total *Galls flowers flowers Aborted β=0.03 β=0.66 β=0.22 β= -0.18 -- -- flowers F1,81=4.07 F1,81=11.04 F1,81=32.59 F1.81=7.06 P<0.05 P<0.01 P<0.0001 P<0.01

Total β=0.05 β= -0.32 β= -0.09 β=0.09 β=0.64 -- flowers F1,80=83.34 F1,80=3.86 F1,80=6.87 F1,80=2.55 F1,80=56.66 P<0.0001 P=0.053 P=0.0105 P=0.11 P<0.0001

Average β= 1.76 β= -2.99 β= -2.29 β= -0.03 β=0.01 β= -0.55 -4 -5 silique F1,76=40.58 F1,76=0.63 F1,76=7.09, F1,76=4.0x10 F1,76=4.0x10 F1,76=0.04 length P<0.0001 P=0.43 P<0.01 P=0.98 P=0.99 P=0.83

ELK 2005: Extremely dry weather precluded pollination experiments in ELK 2005, so I did not examine components of fitness.

ELK 2006: Galls (Wilk’s λ=0.65, P<0.001) affected the three dependent variables, as did the interaction between pollination and height at harvest (Wilk’s λ=0.91, P<0.001). Stepdown analysis showed that galls increased aborted flowers (Table 3-7). Height at harvest increased the number of total flowers as did the number of aborted flowers (Table 3-7). Silique length increased with height at harvest, but this effect decreased with pollination (Table 3-7).

TABLE 3-7. Stepdown results for ELK 2006. Height at harvest was associated with more total flowers and longer siliques, but not with aborted flowers. The number of aborted flowers increased with galls silique length decreased with galls. Supplemental pollen increased fitness for some plants.

Height Pollination GALLS Pollination Aborted Total flowers *Height flowers Aborted β=4.3x10-3 β=0.05 β=1.02 β =0.03 -- -- flowers F1,149<0.01 F1,149<0.01 F1,149=76.44 F1,149=3.24 P=0.95 P=0.98 P<0.0001 P=0.07

Total β=0.004 β= 0.05 β=0.05 β=1.99x10-4 β=0.50 -- flowers F1,148=40.69 F1,148=0.20 F1,148= 2.12 F1,148=0.03 F1,148=82.33 P<0.0001 P=0.62 P=0.15 P=0.88 P<0.0001

Average β=0.10 β=8.34 β= -6.48 β= -0.14 β= -3.40 β= -3.44 silique F1,147=0.25 F1,147=5.80 F1,147=4.56 F1,147=11.52 F1,147=2.48 F1,147=1.79 length P<0.0001 P=0.02 P=0.03 P<0.001 P=0.12 P=0.18

DISCUSSION

Pollinator and herbivore preferences:

Pollinators and herbivores may be attracted by similar traits (Strauss 1997, Irwin et al. 2004,

McCall and Irwin 2006), particularly large flower displays (Brody and Mitchell 1997, Collin et al. 2002, Ehrlen et al. 2002). In the current study, E. capitatum plants that produced more flowers over the course of the season and that had more open flowers per day experienced more visitation by both pollinators (flies, bees, beetles) and most herbivores (ants, gall midges, and petal consumers; Fig 3-3). Other than flower production, no measured traits simultaneously attracted both pollinators and herbivores (Figs 3-1 & 3-2). For instance, in GBT, wider flowers attracted more pollinators but not more herbivores, whereas in other populations, wider flowers attracted more ants but had no effect on pollinators. Divergent pollinator and herbivore preferences should allow E. capitatum to escape potential tradeoffs between attracting pollinators and herbivores (Adler 2000, Herrera et al. 2002, Irwin et al. 2004).

FIGURE 3-3. Summary diagram of best-fit models showing traits that attracted both pollinators and herbivores. Arrow thickness indicates the relative strength of an effect. See Fig 1 for explanation of symbols.

Herbivory reduced pollinator visits to E. capitatum in some populations. Bees in NED and

ELK preferred plants with less petal damage, and flies in ELK preferred plants with fewer galls

(Fig 3-4). In WAK, plants with galls were sometimes pollen limited while plants without galls were not, suggesting that galls may have reduced pollinator service to individual flowers even though whole-plant visitation was not affected. Reduced attractiveness to pollinators is a commonly observed outcome of floral herbivory (e.g. Kudoh and Whigham 1998, Krupnick et al. 1999, Adler 2000, Sanchez-Lafuente 2007, Danderson and Molano-Flores 2010), and is one of the primary reasons to measure both pollination and herbivory when examining selection on floral traits.

FIGURE 3-4. Summary diagram of best-fit models that showed effects of herbivory on pollinator visitation. See Fig 3-1 for explanation of symbols.

Pollinator and herbivore effects on fitness, selection, and plant compensatory responses:

Pollinators:

While selection by pollinators can occur when plants are not pollen limited (Parachnowitsch and Kessler 2010), pollen limitation makes pollinator-mediated selection more likely to occur.

The study populations of E. capitatum were not pollen limited during 2005 and 2006. E. capitatum requires out-crossed pollen to set fruit (Price 1984) and greenhouse-grown plants from the four study populations did not set fruit unless hand-pollinated (C Lay, CU Boulder, unpubl. res.); thus, pollinator visitation is required for reproduction. Differences in pollinator visitation, however, were not related to variation in estimated female fitness in any site or year (Fig 3-5), and experimental pollen addition did not result in increased female fitness. The lack of pollen

58 limitation suggests that plant reproductive success, as measured by total silique length, was primarily limited by resource availability, though I did not explicitly test this by adding nutrients or water. Observed levels of visitation were quite low in all populations, yet appear to have been adequate given the level of resources available for fruit set. In contrast, populations of Erysimum mediohispanicum in Spain have similar levels of visitation but are pollen limited (Gomez et al.

2010). Pollinators in the Rocky Mountain sites may be more effective, or E. capitatum may be more capable of avoiding pollen limitation than E. mediohispanicum.

FIGURE 3-5. Summary diagram of individual path analyses that showed effects of plant traits and insects on fitness. There was no best-fit model that included fitness for either NED 2006 or GBT 2005. See Fig 1 for explanation of symbols.

Two separate mechanisms can explain the absence of pollen limitation despite self- incompatibility and low visitation in all populations. First, E. capitatum has a long period of stigmatic receptivity, up to 7 days when measured by peroxidase assay (C Lay, CU Boulder, unpubl. res.). Prolonged stigmatic receptivity allows plants to tolerate low frequencies of pollinator visitation without loss of fitness (Ashman and Schoen 1994, Bingham and Ort 1998,

Navarro et al. 2007). Second, variation in pollination may be compensated by changes in components of fitness (aborted flowers, flower production, and silique length). In the experimental study, the number of aborted flowers (flowers that reach anthesis but fail to produce fruit) was associated with a greater number of total flowers (after controlling for plant height; Fig 3-6). This relationship suggests that flowers that do not produce fruit are partially replaced by the subsequent production of additional flowers on these indeterminate inflorescences. Pollination also may cause changes in the components of fitness. For instance, in ELK 2006, supplemental pollination led to increased allocation to individual fruits as measured by average silique length (Table 3-7). Also, in GBT 2006, taller plants aborted more flowers, but supplemental pollination reduced the number of aborted flowers in taller plants

(Table 3-2). These data suggest that E. capitatum may shift resources between flower production and developing fruits in response to changes in pollen quality or quantity.

FIGURE 3-6. Relationship between aborted and total flowers. Total flowers increased with the number of aborted flowers in all populations and years. Lines show predicted values of total flowers for values of aborted flowers (both are square-root transformed) when accounting for other independent variables. 95% confidence intervals of effect sizes are too small to be visible on graph.

Burd (Burd 2008, Burd et al. 2009) suggested that the general prevalence of pollen limitation could be the result of stochastic pollination environments that have selected many plants to overproduce ovules. However, the iterative and modular nature of plant reproduction was only briefly addressed in Burd’s analysis, and the inclusion of plant response strategy in

60 studying pollen limitation is important (Wesselingh 2007). Whereas large numbers of ovules per flower allow plants to take advantage of high levels of pollen delivered by single pollinator visits, the number of flowers (and the ovules they contain) produced by indeterminate inflorescences is free to change over time with varying resource conditions as the flowering period unfolds (Lloyd 1980, Stephenson 1992, Diggle and Miller 2004). Sequential, indeterminate flowering could be favored over initial overproduction of ovules in some cases

(Wesselingh 2007). Mechanisms such as prolonged stigmatic receptivity, indeterminate flowering, and reallocation of reproductive resources may be particularly important for E. capitatum, which is visited by a diverse group of pollinators and multiple herbivores that may differ in time and space. The reproductive success of individuals with a generalized pollination system typically depends more on the frequency and effectiveness of individual visitors than the diversity of visitors (Perfectti et al. 2009). Variation among the study sites in common visitors of

E. capitatum (C Lay, CU Boulder, unpubl. res.), and low visitation generally, suggests that not all populations have dependable and constant pollinators. The reallocation of resources to production of new flowers and to surviving fruits may reduce the impact of spatiotemporal variation in pollinator assemblages as selective agents.

Herbivores:

Unlike pollinator visitation, herbivory influenced total fitness in several populations and years (Fig 3-5). Moreover, well-supported paths between herbivore preferences, traits, and fitness provide evidence of selection. For instance, stem borers were most common in WAK and preferred plants with thicker stems when their preferences were measured in 2006. Plants at that site had the narrowest stems of the four populations (C Lay, CU Boulder, unpubl. res.),

61 suggesting that selection by borers in WAK may have resulted in narrower stems compared to plants in other populations.

Petal damage was greater on plants with narrower flowers and more open flowers, and because petal damage reduced fitness, petal herbivores selected for fewer open flowers and wider flowers in this population and year. However, the reason for reduced fitness of plants with damaged petals was ambiguous. Petal damage commonly affects plant fitness through reduced pollinator visitation (McCall and Irwin 2006). Bees were less likely to visit E. capitatum plants with more petal damage, but the best-fit model for ELK 2006 did not show effects of pollinators on fitness.

Ants never affected plant fitness despite their prevalence as floral visitors, removal of nectar, and occasional agonistic interactions with potential pollinators (Fig 3-5), and ant exclusion had no effect on fitness (C Lay, CU Boulder, unpubl. res.). Therefore, while ants preferred larger display sizes and wider flowers (Figs 3-1, 3-2, & 3-3), they were not effective agents of selection on these characters in the populations studied. Species such as E. capitatum that are not pollen- limited are less likely to experience a reduction in fitness when nectar is removed without pollination service (Burkle et al. 2007).

Galls reduced fitness in all populations in 2006, though not in 2005 (Fig 3-5), and plants with more total flowers had more galls. Total flowers were related to galls only through floral abortion (Figs 3-6 & 3-7), however, so it is unlikely that gall midges prefer plants with more flowers (and hence, select for fewer flowers). Instead, this association can be explained by plant compensatory response. Galled flowers were almost always aborted, and plants offset this loss by increasing flower production (Figs 3-6 & 3-7). Plants did not fully compensate for galling; galled flowers were only partially replaced by new flowers, and plant investment in individual

62 fruits was reduced on plants with galls, so overall fitness effects of galls were negative. In addition, galls appeared to reduce pollination success, even without affecting visitation. Added pollen increased the length of surviving siliques on plants with galls in WAK 2006, suggesting that plants with galls experienced pollen limitation (that is, flowers on galled plants received insufficient or inferior pollen) while plants without galls did not.

FIGURE 3-7. Relationship between galls and total number of flowers produced over the season. When controlling for the effect of aborted flowers on total flowers, galls and total flowers are not positively related on experimental plants in any of the three populations. This suggests that the positive associations between galls and total flowers in best-fit models of observational data are due to plant compensatory responses. Lines show predicted values of total flowers for gall counts in each population. 95% confidence intervals of effect sizes were too small to be seen on graphs. The negative relationship is significant in NED 2006, but not in the other two populations.

Plant compensatory responses moderate fitness consequences of both pollinators and herbivores:

Multiple herbivores and damage to multiple tissues are a perennial problem for plants

(Linhart 1991, Strauss et al. 2005), as is pollinator unpredictability (Ashman et al. 2004, Knight et al. 2005, Burd et al. 2009) and E. capitatum is no exception. Several herbivores strongly reduced female fitness, although their selective influences on plant traits were varied and relatively independent of one another. Surprisingly, although pollinator visitation was rare and pollinator communities were variable, pollinators did not affect fitness or select on plant traits in

63 the study populations (Fig 3-5). The effects of herbivores (e.g., gall midges) and infrequent pollinators appeared to be moderated by compensatory responses in E. capitatum. Plants produce flowers sequentially and indeterminately, allowing individuals to replace damaged and unpollinated flowers over the course of the season. The development of new flowers in response to herbivory is hardly unique to E. capitatum. Removal of flower buds increases the production of new flowers in Brassica napus (Williams and Free 1979), a close relative of E. capitatum, and many plants are able to respond to herbivory and uneven pollen receipt among flowers by changing patterns of fruit and seed set (Marshall et al. 1985, Stephenson 1992, Knight et al.

2005, Wise and Cummins 2006, Nunez-Farfan et al. 2007, Wesselingh 2007, Wise and

Abrahamson 2007). The heritability of plant resource allocation strategies is difficult to measure

(but see Juenger and Bergelson 2000, Wise et al. 2008). However, plant compensatory strategies are likely key traits that allow plants to succeed in unpredictable environments with selection pressures from multiple interacting partners.

Spatiotemporal variation in the selective landscape:

A recent meta-analysis by Harder and Johnson (2009) found that directional selection by pollinators is often weak or non-existent and tends to be inconsistent over time and space. In addition, they found that herbivores often produce stronger selection on plant traits than pollinators, and positive directional selection for flower number occurs three times more often than other commonly measured traits (Harder and Johnson 2009). My results are consistent with these general patterns (see also Strauss and Whittall 2006).

I did not identify selective effects of pollinators on measured traits of E. capitatum in any population or year, and elevational gradients were not associated with any gradual biotic gradients (Figs 3-1 to 3-5). Herbivores selected on specific plant traits, but their prevalence and

64 preferences varied over time and space. For instance, stem-boring beetle larvae selected for narrower stems but were only common in WAK (Fig 3-5). Insects that caused petal damage selected for fewer, wider flowers in ELK 2006, but in no other population or year. Although the prevalence of galls differed somewhat among populations, their effects were similar among populations; gall midges decreased fitness in 2006 in all populations and did not affect fitness in any population in 2005 (Fig 3-5). Some of this temporal variation may have been driven by water availability or other abiotic factors. 2005 was very dry and many plants at the three highest elevation populations in 2005 wilted during observation periods. Under such conditions, water availability was likely the primary determinant of fitness, with pollinators and herbivores playing a more minor role.

As in the meta-analysis (Harder and Johnson 2009), total flower number was the only trait that was under selection in all populations and years. Flower number was always directly and positively associated with plant female fitness (Fig 3-5), so overall, selection favored plants with more flowers. The relationship between flower number, insects, and fitness was not always straightforward, however. Structural equations models of observational data suggested that the selective benefit of increased flower production was attenuated by gall midges in 2006, as galls reduced fitness and were found in greater numbers on plants with more flowers (Fig 3-5).

Furthermore, experimental evidence suggests that increased flower production actually may have been a compensatory response to gall midge attack (Figs 3-6 & 3-7). If increased flower production in response to herbivory is common, negative effects of herbivores may obscure the positive effects of flower number on fitness. This in turn suggests that selection for increased flower number may be even more frequent than was suggested by Harder and Johnson’s meta- analysis (2009). Compensatory responses such as flower production may vary among

65 individuals and result in increased phenotypic variation. Such variation can mediate selection by pollinators and herbivores in both time and space (Fordyce 2006).

Conclusions:

Individuals of Erysimum capitatum face an adaptive landscape that is shaped by diverse generalist interacting partners that shift in abundance and importance in time and space.

Evidence from both experimentation and observation was used to determine how herbivores and pollinators alter this landscape. Neither pollen limitation nor selection by pollinators was evident in E. capitatum, though selection by herbivores was occasionally important. Compensatory reproductive mechanisms may allow E. capitatum to succeed in this complex selective environment. In addition to flower longevity, plants respond to both galls and low pollinator visitation by altering flower abortion, new flower production, and allocation to individual fruits.

Developmental mechanisms that allow plants to make iterative decisions about reproductive resource allocation over the course of the flowering season may be under selection in E. capitatum.

GBT WAK NED MRS GBT WAK NED MRS 2005 2005 2005 2005 2006 2006 2006 2006 Total minutes of 350 560 1020 1500 1690 640 240 730 observation Number of plants × 1440 2480 3590 3890 8140 1280 7780 3220 minutes of observations Total pollinators observed 52 77 67 108 190 134 7 123 APPENDIX 3-1. Pollinator frequency. Pollinators included bees, flies, butterflies, beetles, and all other visitors which contacted reproductive parts.

Aborted Total Aborted Total

flowers flowers flowers flowers

GBT 06 NED 05

Total Total 0.63 0.21 flowers flowers Average silique Average silique 0.09 0.40 0.25 0.40 length length

WAK 05 NED 06

Total Total 0.80 0.49 flowers flowers Average silique Average silique -0.01 0.15 0.14 0.25 length length

WAK 06 ELK 06

Total Total 0.66 0.68 flowers flowers Average silique Average sillique 6.90E-04 0.17 -0.15 0.09 length length APPENDIX 3-2. Correlations among components of fitness (square-root transformed aborted flowers and total flowers and average silique length).

Akaike Akaike MODEL X2 DF AICC weights MODEL X2 DF AICC weights

GBT05 1 11.08 6 167.75 0 GBT06 1 29.08 17 217.54 0 2 12.23 9 133.09 0 2 35.47 21 196.18 0 3 14.86 11 117.03 0 3 36.2 23 184.47 0 4 22.99 12 116.99 0 4 44.39 25 181.05 0 5 20.39 13 106.87 0 5 55.39 26 186.54 0 6 29.46 13 115.94 0 6 47.82 29 163.44 0 7 20.62 13 119.73 0 7 49.6 34 142.35 0 8 32.99 16 100.14 0 8 53.93 38 130.64 0.02 9 20.62 16 87.76 0.004 9 54.03 40 123.36 0.64 10 25.34 19 76.89 0.99 10 62.61 42 124.95 0.34 11 175.16 46 136.67 0

12 84.72 49 215.63 0

WAK05 1 33.96 15 155.28 0 WAK06 1 97.65 28 247.15 0 2 56.15 17 176.78 0 2 252.75 43 343.63 0 3 56.6 19 167.37 0 3 106.21 32 249.9 0 4 60.61 20 137.13 0 4 117.7 38 226.33 0.12 5 44.38 23 166.67 0 5 118.12 39 223.07 0.63 6 44.53 25 129.04 0 6 134.23 43 225.11 0.22 7 47.3 31 109.64 0.17 7 150.99 46 231.95 0 8 53.96 33 109.66 0.16 8 141.87 44 225.39 0.02 9 57.46 35 106.84 0.7 9 160.28 46 241.23 0 10 75.93 37 119.3 0.001 10 175.46 50 243.96 0 11 117 38 158.03 0 APPENDIX 3-3. Model fit statistics for complete sets of models tested in the analyses of insect preference, fitness effects, and selection for GBT 2005, GBT 2006 and WAK 2005 and WAK 2006.

Akaike Akaike MODEL X2 DF AICC weights MODEL X2 DF AICC weights NED05 1 25.78 15 157.76 0 2 32.46 23 134.37 0 3 32.46 24 105.33 0 4 35.29 25 121.41 0 5 35.76 29 120.15 0.04 6 32.92 27 109.92 0.004 7 36.52 31 99.05 0.89 8 115.29 34 167.92 0 9 48.22 33 104.06 0.07 10 128.47 35 117.97 0 11 141.25 40 176.19 0 ELK05 1 33.87 14 190.41 0 ELK06 1 38.44 17 230.57 0 2 37.78 19 161.21 0 2 50.32 23 200.82 0 3 41.7 22 170.93 0 3 53.69 25 212.52 0 4 42.73 23 144.17 0 4 55.62 27 183.02 0 6 44.96 26 131.96 0 5 55.7 28 177.79 0 5 42.28 24 138.72 0 6 73.75 29 190.71 0 7 45.25 26 132.25 0 7 57.81 30 169.81 0 8 48.31 28 126.51 0 8 59.94 34 153.55 0.02 9 48.49 29 122.53 0 9 61.76 36 146.98 0.63 10 50.48 32 112.8 0.05 10 71.98 38 149.27 0.2 11 50.49 33 109.16 0.28 11 93.11 42 155.83 0.007 12 52.35 34 107.48 0.66 12 90.67 43 149.99 0.14 13 59.92 34 115.04 0.15 APPENDIX 3-4. Model fit statistics for complete sets of models tested in the analyses of insect preference, fitness effects, and selection for NED 2005, ELK 2005 and ELK 2006.

APPENDIX 3-5. Most-inclusive models for GBT 2005, GBT 2006, and WAK 2005.

APPENDIX 3-6. Most-inclusive models for WAK 2006, NED 2005, ELK 2005, and ELK 2006.

β F P DF β F P

71 DF GBT 06 INTERCEPT 5.77 2611.21 < 0.0001 88 NED 05 INTERCEPT 6.73 12494.78 < 0.0001 96 HEIGHT 0.11 41.99 < 0.0001 HEIGHT 0.04 47.47 < 0.0001 POLLINATE 0.29 0.10 0.37 POLLINATE 0.11 0.45 0.50 POLLINATE*HEIGHT -0.06 1.40 0.24 INSECTICIDE -0.01 0.08 0.92 WAK 05 INTERCEPT 6.22 7113.25 < 0.0001 97 POLLINATE*HEIGHT -0.003 0.04 0.85 HEIGHT 0.06 35.76 < 0.0001 INSECTICIDE*HEIGHT 0.009 1.11 0.29 POLLINATE -0.06 0.11 0.97 POLLINATE*INSECTICIDE -4.79 1.96 0.17 INSECTICIDE 0.18 1.51 0.22 POLLINATE*INSECTICIDE -0.005 0.08 0.78 POLLINATE*INSECTICIDE 0.20 0.25 0.62 *HEIGHT INSECTICIDE*HEIGHT -0.003 0.25 0.87 NED 06 INTERCEPT 5.61 1415.26 < 0.0001 75 POLLINATE*HEIGHT 0.01 0.16 0.69 HEIGHT 0.11 57.76 <0.0001 POLLINATE*INSECTICIDE 0.006 0.01 0.91 POLLINATE -0.09 0.24 0.63 *HEIGHT INSECTICIDE -0.23 1.37 0.25 WAK 06 INTERCEPT 4.68 1306.96 < 0.0001 140 GALLS -0.21 22.52 < 0.0001 HEIGHT 0.12 55.06 < 0.0001 POLLINATE*HEIGHT -0.02 0.46 0.50 GALLS -0.18 18.93 < 0.0001 INSECTICIDE*HEIGHT -0.002 0.002 0.96 INSECTICIDE 0.06 0.08 0.77 POLLINATE*INSECTICIDE 0.1 0.06 0.80 POLLINATE 0.43 0.90 0.35 POLLINATE*GALLS 0.16 1.70 0.09 POLLINATE*HEIGHT 0.02 0.09 0.76 POLLINATE*INSECTICIDE 0.11 2.66 0.11 INSECTICIDE*HEIGHT 0.12 12.67 < 0.001 *HEIGHT POLLINATE*INSECTICIDE -1.10 2.11 0.15 ELK 06 INTERCEPT 5.90 2774.13 < 0.0001 151 POLLINATE *GALLS 0.45 9.07 <0.01 HEIGHT 0.08 58.52 < 0.0001 POLLINATE*INSECTICIDE -0.04 0.09 0.75 POLLINATE 0.51 3.09 0.08 *HEIGHT GALLS 1/2 -0.33 10.53 <0.01 POLLINATE *GALLS1/2 -0.35 0.71 0.40 POLLINATE*HEIGHT -0.09 6.45 <0.05 APPENDIX 3-7. Effects of height at harvest, experimental treatments, and galls on estimated female fitness.

Wilk’s λ F Wilk’s λ F 72 POLLINATE GBT 06 0.98 F3,86 =0.54 HEIGHT 0.78 F3,86 =43.56*** POLLINATE*HEIGHT 0.91 F3,86 =2.60 .

POLLINATE WAK 05 0.99 F3,95=0.05 WAK 06 1.0 F3,117=0.01 INSECTICIDE 0.88 F3,95=4.28** 0.97 F3,117=1.18

HEIGHT 0.47 F3,95=34.44*** 0.70 F3,117=16.06***

GALLS -- -- 0.81 F3,117=8.79***

POLLINATE*INSECTICIDE 0.98 F3,95=0.44 0.98 F3,117=0.70

POLLINATE*HEIGHT 0.98 F3,95=0.51 0.98 F3,117=0.42

POLLINATE*GALLS -- -- 0.94 F3,117=2.32 .

INSECTICIDE*HEIGHT 0.97 F3,95=0.88 0.90 F3,117=4.47**

INSECTICIDE*GALLS -- -- 0.98 F3,117=0.88

POLLINATE*INSECTICIDE*HEIGHT 1.0 F3,95=0.05 0.97 F3,117=1.24

POLLINATE*INSECTICIDE*GALLS -- -- 0.07 F3,117=1.29

POLLINATE NED 05 0.98 F3,94=0.71 NED 06 0.88 F3,64 =2.98*

INSECTICIDE 0.87 F3,94=4.32** 0.98 F3,64 =0.30

HEIGHT 0.51 F3,94=29.48*** 0.47 F3,64 =23.67***

GALLS -- -- 0.66 F3,64 =10.88***

POLLINATE*HEIGHT 0.99 F3,94=0.19 1.0 F3,64 =0.10

POLLINATE*GALLS -- -- 0.85 F3,64 =3.60*

POLLINATE*INSECTICIDE 0.99 F3,94=0.33 0.92 F3,64 =1.93

INSECTICIDE*HEIGHT 0.98 F3,94=0.57 0.94 F3,64 =1.38

INSECTICIDE*GALLS -- -- 0.97 F3,64 =0.63

POLLINATE 0.99 F3,94=0.38 0.97 F3,64 =0.59 *INSECTICIDE*HEIGHT POLLINATE *INSECTICIDE*GALLS -- -- 0.97 F3,64 =0.57

POLLINATE ELK 06 0.96 F3,=2.19 . HEIGHT 0.67 24.17*** GALLS 0.65 26.59*** POLLINATE*HEIGHT 0.91 4.62*** POLLINATE*GALLS 0.98 0.91 APPENDIX 3-8. Complete MANCOVA results (Wilk’s λ, F and p-values) for the three fitness components (total flowers, aborted flowers and average silique length) for all experiments. P<0.05*,P<0.01**,P<0.001***

CHAPTER IV

ON COMMON GROUND: ERYSIMUM CAPITATUM AND LOCAL ADAPTATION TO POLLINATORS AND HERBIVORES

ABSTRACT

Communities of pollinators and herbivores can vary considerably among plant populations in their prevalence, importance, and interactions. In species with generalized pollination systems and multiple herbivores, these local differences in biotic communities may drive local adaptation for flower and flowering traits. Conversely, conflicting selection by multiple herbivores and pollinators may limit local adaptation and maintain variation within populations.

To determine whether there is evidence of local adaptation or maladaptation to pollinators and herbivores in Erysimum capitatum (Brassicaceae), a species that shows considerable phenotypic variation and interacts with multiple species of pollinators and herbivores, I used plants from two populations from separate elevations in a common garden experiment located near the lower-elevation population to address the following questions: 1) Do pollinators and/or herbivores prefer particular traits of local or non-local plants? 2) Do pollinators and herbivores have additive and/or interactive effects on fitness in E. capitatum? 3) Do fitness consequences of pollinator and/or herbivore activity differ for local versus non-local plants?

In the common garden, E. capitatum plants from the two study populations do not show evidence of local adaptation to pollinators and herbivores, though non-local plants produced fewer seeds. Plants from both sites were pollen limited in the common garden, although they were not in natural populations. Seed production also increased with the addition of insecticide. These effects

74 were similar for plants from both sites, so pollinators and herbivores were not responsible for the differences in reproduction between the two source sites. However, non-local plants did produce fewer seeds than local ones, and received a greater benefit in mean seed mass from the addition of insecticide. Plants from the two populations differed in seed number and seed mass suggesting different patterns of investment.

INTRODUCTION

Local adaptation of plants to particular pollinators can be a strong force in shaping flower morphology and reproductive characteristics such as flower production and timing (Campbell et al. 1997, Caruso et al. 2003, Anderson and Johnson 2009, Schlumpberger et al. 2009, Gomez and

Perfectti 2010, Nattero et al. 2010). Recently, simultaneous selection by other agents, such as herbivores, has been shown to be important in the evolution of plant reproductive features

(reviewed in Strauss and Whittall 2006). Despite the many instances of traits that appear to be adaptations to particular pollinators, most studies of phenotypic selection have found that selection by pollinators tends to be weak and inconsistent, while that by other agents is often strong and consistent (Harder and Johnson 2009). Even so, pollinators and herbivores can interact in complex ways to effect selection for plant traits (Louda and Potvin 1995, e.g.

Armbruster 1997, Herrera et al. 2002, Irwin et al. 2004, Pohl et al. 2006, Armbruster et al. 2009,

Gomez et al. 2009b, Whittall and Carlson 2009). Selection by herbivores can overwhelm that by pollinators (e.g. Gomez and Zamora 2000, Herrera 2000, Herrera et al. 2002, Gomez et al.

2009b). In other cases, conflicting pressures may produce adaptive compromises (Brody 1997,

Galen and Cuba 2001, Ehrlen 2002, Irwin et al. 2003, Adler and Bronstein 2004, Gomez 2008,

Armbruster et al. 2009). For instance, taller plants of Erysimum mediohispanicum are better

75 pollinated, but are attacked more frequently by predispersal seed predators (Gomez 2008). In addition to such conflicting additive effects in which herbivores and pollinators favor different values of the same traits, interactive effects of herbivores and pollinators may also occur. These interactive effects occur because herbivory can affect the attractiveness of plants to pollinators and pollinators may affect the attractiveness of plants to herbivores (e.g. Juenger and Bergelson

1997, Krupnick et al. 1999, Cariveau et al. 2004, McCall and Irwin 2006, McCall 2008, 2010).

For example, Nemophila menziesii flowers damaged by lepidopteran larvae are less attractive to bees (McCall 2008), and such reduced attractiveness could increase pollen limitation (McCall

2010).

Such additive and interactive effects may lead to complex patterns of adaptation of plants to pollinator and herbivore groups. Communities of pollinators and herbivores can vary considerably among plant populations in prevalence, importance, and their interactions (Herrera

1988, Kearns 1992, Dilley et al. 2000, Gomez and Zamora 2000, Rand 2002, Bradley et al. 2003,

Aigner 2005, Gomez et al. 2010), creating differences in selection regimes that may result in phenotypic divergence and local adaptation among populations (Thompson 1988, Rey et al.

2006, Gomez et al. 2009b, Lay et al. 2011). Conversely, conflicting selection by multiple herbivores and pollinators may limit the potential for local adaptation and maintain variation within populations. Therefore, studies of the evolution of flowers and flowering traits designed to measure the contribution of these selective agents should include multiple sites to account for the existence of selection mosaics.

Populations in montane regions provide excellent opportunities to examine the effects of different pollinator and herbivore communities on population differentiation (Linhart and Grant

1996, Lomolino 2001) because of the great variation in abiotic and biotic conditions that occurs 76 over very short geographical distances in these regions. The present study focuses on Erysimum capitatum (Brassicaceae), which occurs over a broad geographical range including montane regions, and shows considerable phenotypic variation among populations for traits that may affect pollinator and herbivore choices (Price 1984, Chapter 2). A previous study of four populations of this species over an elevational gradient showed that herbivore and pollinator communities differed among populations (Chapter 2). In this study, two of these populations are used in a common garden experiment.

Although a variety of studies have examined the effects of selection by pollinators and herbivores in different locations (Abdala-Roberts et al. 2009, Gomez et al. 2009b, Kolb and

Ehrlen 2010, Parra-Tabla and Herrera 2011), a common garden approach with plants from different populations is critical for determining whether differences in phenotype among populations reflect local adaptations to pollinators and herbivores. Such studies are lacking.

Local selection by pollinators and herbivores may result in adaptation to local conditions or maladaptation (Thompson 1999). To determine whether there is evidence of either of these outcomes in E. capitatum, I grew plants from local and non-local populations in a common garden experiment. In this garden, I treated plants with added pollen or insecticide in a fully- factorial design and observed pollinator and herbivore visitation to address the following questions: 1) Do pollinators and/or herbivores prefer particular traits of local or non-local plants? 2) Do pollinators and herbivores have additive and/or interactive effects on fitness in E. capitatum? 3) Do fitness consequences of pollinator and/or herbivore activity differ for local versus non-local plants? 77

MATERIALS AND METHODS

Study species: Erysimum capitatum (Brassicaceae) is native throughout most of the United

States (USDA 2004, Chapter 2) and northern Mexico in lowlands and montane areas (Turner

2006). It is often a biennial, although its life history varies among individuals and populations

(Price 1984). Plants flower from early spring to mid-summer (USDA 2004), and populations are characterized by variation in flower color and other traits across the species range (Price 1984,

Weber 1990). Flowers may be white, lavender, or range from yellow to red (Price 1984). All populations and individuals in the present study are recognized as a single species. The taxonomy and identification of E. capitatum is currently under debate, however, and the great phenotypic variation within this species may be the result of past hybridization events between subspecies (Turner 2006).

Sites: I chose two source populations for which I had studied pollination and herbivory in previous years: Greenbelt Plateau Open Space (GBT), a short-grass prairie at 1980 m, and a site near

Nederland, CO (NED) in open lower montane Ponderosa pine forest at 2590 m. I planted 400 plants grown from seed collected in 2004 from these two source sites in a common garden about 3km from

GBT on the South Campus of the University of Colorado at Boulder. These source sites had plants that were phenotypically distinct and subject to different levels of pollination and floral herbivory

(Chapters 2 & 3). A second common garden of 188 plants near NED experienced heavy damage from pocket gophers. After flowering, but before senescence, pocket gophers clipped tops of inflorescences from the bases with identifying tags and often dragged inflorescences a meter or more from the base. This made fitness outcomes for these plants impossible to determine, as some fruits contained mature seeds at the point of clipping. Results from this population will not be reported.

Greenhouse and Garden: Plants were grown in the greenhouse from seed collected from 21 maternal GBT plants and 19 maternal NED plants in 10cm pots beginning in mid-March of 2007.

Plants from both source populations began flowering at the same time in February of 2008. To prolong the flowering period through the time that they would be installed in common gardens

(May) and to make sure that all fruits present at harvest resulted from flowers that had reached anthesis in the field, I removed flowers and inflorescences once a week until all plants were installed in the common garden on 22 May. Study plants were considerably larger than most plants in natural populations; they tended to have more inflorescences overall, and more basal inflorescences, particularly when primary inflorescences were clipped (unpubl. data).

I transplanted 400 plants to a common garden near GBT. Eight plants in four treatments designed to test the effects of pollinators and herbivores (control, added pollen, added insecticide, added both pollen and insecticide) were arranged evenly around the edges of 40 circular plots 1.5m in diameter (Fig 4-1A). These circular plots were placed in eight staggered rows with five groups per row, and were 1m apart at the edges of each plot (Fig 4-1B). Each plot had only one representative of any given family and contained an individual from every treatment*source population combination. This resulted in evenly spaced plants that could be easily watched for insect visitors during observation periods.

Plants from each treatment and source population were placed haphazardly within plots. To prevent edge effects from influencing pollinator visitation or plant fitness, I added 63 plants, not included in the study, around the edge of the garden. I added supplemental water as needed throughout the season (from every other day to every day during the driest weeks) and fertilized once a week with Scotts Peter’s Excel solution (Scotts-Sierra Horticultural Products

Company, Marysville, Ohio) with a ratio of 15-5-15 N-P-K mixed to provide 150-200 ppm N. This fertilizer also provides trace amounts of micronutrients.

FIGURE 4-1. Design of (A) two sample circular plots and (B) overall garden. (A) There were 8 plants in each circular plot, 1 from each pollination*insecticide*source population combination. “I” indicates insecticide treatment, “P” hand pollination, “PI” both hand pollination and insecticide ,and “C” controls. Grey and white indicate different source populations. The position of treatment and source population was randomized within each plot, although all plots contained all 8 treatment*source population combinations, and no plot contained more than 1 member of the same maternal family. (B) 40 plots were arranged in staggered rows of 10 plots for a garden of 320 plants, with 80 edge plants (in grey) from which no data were taken, but were sometimes used as pollen donors. The dark grey line indicates the path taken during 30-s observations.

Pollinator and herbivore preferences: Multiple reproductive traits that have the potential to affect both pollinator and herbivore choice were measured: corolla width, corolla depth, corolla color, petal width, height of the tallest inflorescence at the beginning of flowering, and the number of open flowers on each observation day. To measure visitation by pollinators, three people watched each circular plot for 10 minutes starting at 0900hr and recorded the identity of each visitor and the length of visit, on one day per plot. In addition, I performed as many 30-second observation periods per plot as possible on 13 other days. On most days, it was possible to perform four 30-second observations per plot. 30-second observation periods were used to allow a single observer to observe all plants at least once within a 20-minute period. This method reduced the amount of variation in pollinator visits due to time of day. I pooled the number of visits of bees, flies and beetles for each plant on each day. There was no difference between the day with 10 min observation and days with 30-second observations that could not be explained by the number of

80 observations, so I included this day with the others. Because a variety of petal herbivores created visually similar damage, there was no reliable way to determine what herbivore had damaged a flower. Therefore, flower damage from all consumers was measured together. Damage levels for individual flowers were recorded on each day, and I did not distinguish between new and old damage. Flower damage was scored “1” to “5” for each flower. Level “1” corresponded to petal limb damage10% damage or less, “2” to damage between 10% and 25%, “3” to 25-50%, “4” to petal damage of 50% or more, while “5” indicated complete destruction of anthers and stigma.

Three days were discarded from analyses because flowers and damage were not measured, and some plants were discarded because they began flowering after measurements were taken. I removed observations of any plant with 0 flowers on a particular day from analyses, so the numbers of observations per plant, per family and per population are unequal. This left 12 days and 193 plants for analyses, with 1493 unique day*plant observations (694 of these were of plants from NED, 799 were of plants from GBT). Although all families were included, some plants were not included at all, and some plants and families were only observed once.

Analyses of pollinator and herbivore preferences: Source site, pollination and insecticide were crossed in a three-way factorial design with repeated measures of pollinator visitation and herbivory on several times per day on multiple days for each plant nested in maternal family. To examine the effects of source population and plant characters on pollinator and herbivore preferences while controlling for random variation caused by plant identity, maternal family and date, I tested generalized linear mixed models (GLMM, Pinheiro and Bates 2000, Gelman and Hill 2007) with

Poisson error distribution for each dependent variable (BEE, FLY, BEETLE and DAMAGE).

GLMM is often better for such designs than repeated measures ANOVA as it can be applied to data with missing values and imbalance in random (grouping) factors, such as were caused by plants from

81 particular maternal families failing to flower on particular days (Pinheiro and Bates 2000, Bolker et al. 2009). I used lmer in the R package lme4 (Bates and Maechler, 2010) for GLMM analyses of pollinator and herbivore preferences. To find significance estimates for parameters, I compared likelihood ratios for random effects (plant nested in maternal family crossed with date) and interpreted the results of Wald tests for fixed effects (Bolker et al. 2009). I interpreted the significance of fixed effects in the model including all random effects to avoid pseudoreplication.

Random effects included individual plant nested in maternal family and crossed with date.

By design, both maternal family and plant were nested within source population (a fixed effect).

Including these random effects in the model allowed control for any effects of family or plant identity or date that were unrelated to measured characteristics and source population. Data were unbalanced, due mainly to differences in flowering time among individuals and families. Most observations did not contain pollinator visits, and data were zero-inflated compared to a typical

Poisson distribution. However, errors were under- rather than over-dispersed for each dependent variable, and further correction was unnecessary. Fixed effects included number of 30-second observation periods, corolla width, corolla depth, corolla color, height, open flowers, damage, insecticide treatment, and source population. Observation periods were the same for all plants on a given day. Flower traits, height and insecticide treatment did not change across days but were different for each plant. Open flowers and damage differed across both plants and days. For bees, all fixed effects were tested. Inclusion of damage in analyses of fly visitation caused a problem with model fitting, likely because there were no instances in which flies visited damaged plants and comparatively few flies were seen, so I removed damage from the analysis of fly preference.

Shriveled and dry petals, a type of damage associated with nitidulid beetles on rapeseed plants, were common in the gardens, although this type of damage seemed less common in the natural 82 populations I observed. Because beetles were likely the cause of some damage, I used beetles as a fixed effect in modeling damage, and removed damage from the fixed effects model for beetle visitation.

Tests for fitness effects of pollinators and herbivores and local adaptation: To test for additive and interactive effects of pollinators and herbivores, I crossed hand-pollination and insecticide treatments such that there were 4 treatment groups of 40 plants per source site: 1) control, 2), hand- pollinated only 3) insecticide only and 4) hand-pollinated and insecticide added. The 19 maternal families from NED and 21 maternal families from GBT were crossed with treatment such that there were no more than 3 and no fewer than 1 individual per family in each treatment, with most families represented by 2 individuals per treatment. Slight imbalance in the design was caused by plant death before the garden was planted.

Erysimum capitatum requires pollination to set seed (Price 1984), and there are no obvious or clear-cut differences in self-incompatibility between populations (personal observation). I tested for reproductive limitation by pollinators, by hand-pollinating all flowers on 80 plants per source site

(Kearns and Inouye 1993). Pollen was collected in two ways: first, by collecting anthers from flowers that had been removed before the start of the experiment and storing them in the freezer for later use (Larden and Triboi-Blondel 1994). Second, I collected anthers from edge plants in the garden. Pollen from 6-10 donors from both populations was mixed in a glass vial by shaking collected anthers and stirring, then applied with a small paint brush every third day throughout the experiment.

I tested for the effects of deterring herbivores by treating 80 seedlings from each source site

(21 maternal families from GBT, 19 from NED) with the systemic imidacloprid insecticide

(Marathon, 1% granular) in January and in April before planting the garden. Marathon deters a broad

83 variety of herbivores for 4 months or more, and is unlikely to harm pollinators when applied as a systemic treatment (Gels et al. 2002, Rogers and Kemp 2003, but see Decourtye and Devillers

2010).

Fitness measurements: Plants were collected after fruits had set but prior to dehiscence so that seeds could be counted. Estimated seed set was used as a measure of female fitness. I measured fruit length, and counted the number of apparently viable seeds from a subset of fruits on each plant and calculated seeds/mm fruit, and counted the total number of fruits. To estimate seed set,

I multiplied the average number of seeds per mm of fruit by the total number of fruits and the average length of measured fruits. In addition, I counted scars from flowers that had reached anthesis but not set fruit (aborted flowers) for each inflorescence of each plant, and noted the number of flowers that had been cut prior to beginning the experiment that were evident on each stem (these had also been recorded at the beginning of the study). I also attempted to distinguish increased ovule number per flower from decreased ovule abortion by counting ovules that appeared to contain embryos but did not appear viable and ovules that appeared unfertilized

(unfertilized). Unfertilized ovules were often very tiny and tended to break or blow away during counting, and these counts may be less reliable than counts of unfilled or viable seeds.

To determine whether differences in seed number between populations were due to innate differences in ovule packaging and seed provisioning, I weighed groups of 20 viable seeds per plant from a subset of control plants in the experimental gardens and untreated plants in intact field populations. I also weighed groups of 20 viable seeds from a subset of plants in the pollination and insecticide treatment groups in the garden plants to better understand how pollination and insecticide treatments altered seed production and mass.

Analyses of effects of treatments and source population on fitness: Source site, pollination and insecticide were crossed in a three-way design with repeated measures of fitness for each maternal family (repeated measure=multiple plants per treatment per family) and plant (repeated measure=flowers per plant). I used two different general linear mixed models (GLMM) to test for the effects of source site and treatment on reproductive success. To test for the effects of source site and treatment on the entire reproductive output of plants, I used the estimated number of seeds for a plant (total length of all fruits*average number of seeds/mm of fruit) as a dependent variable, maternal family as a random effect and source site, pollination, insecticide, and the interactions of these crossed factors as fixed effects. The estimated number of seeds for each plant was square-root transformed to improve the normality of residuals. I also examined the effect of treatment, source population, and intra-plant variation on the number of viable and inviable seeds found in each fruit, as well as the total number of fruits produced and the number of fruits that were aborted. All variables except the number of viable seeds per fruit were square-root transformed to improve the normality of residuals. Residuals from the analysis of viable seeds per fruit were normal, and transformation made them less so. These GLMM included treatment and source population as fixed effects and maternal family and plant as random effects. Square-root transformation was used for these count variables. I did GLMM analyses using lmer in the lme4 package in R (Pinheiro and

Bates 2000, Crawley 2007, Bates and Maechler, 2010)

RESULTS

Pollinator and herbivore preferences: During 1220 minutes of observation over 15 days, the following flower visitors were observed on experimental garden plants: 311 bees (Bombus spp.,

Lasioglossum (Dialictus) spp., Halictus spp.), 71 flies (Syrphidae, Bombyllidae) and 35 beetles 85

(Meligethes, Nitidulidae). Small bees (mainly Lasioglossum and Halictus) were the most common visitors. The low number of flies and beetles observed relative to the number of observations lowered the power of the analyses to detect preferences of these visitors, as a considerable number of visits by these two groups were on days during which flower counts were unavailable.

Bee visitation depended mainly on date and the number of flowers on a plant on a given day.

36.92% of the variation in bee visitation was explained by date, 9.13% by plant nested in family, and

4.91% by family. For bees, removing either date or the nested term containing both plant and family from the model resulted in significant reductions in model fit compared to the model containing all terms, but removing either family or plant alone did not (Table 4-1). Bees preferred plants that had more flowers. There were no significant effects of flower size, color or plant height on bee visitation after controlling for date, plant and maternal family (Table 4-2).

Fly visitation was dependent on individual plant and flower number per plant, though part of the dependence on plant may be due to the small number of fly visits (71) included in analyses.

65.64% of the variation in fly visitation was explained by plant and less than 0.01% was explained by either maternal family or date, and plant was the only random factor with a significant effect

(Table 4-1). The association of fly visits with individual plants may be an artifact of observing only one visit per plant during an observation period. Flies visited plants with more flowers more frequently. Otherwise, flies had no preferences after controlling for plant, maternal family and date

(Table 4-2).

86 TABLE 4-1. Model fit statistics (Aikaike’s information criterion (AIC) and log-likelihood) for random effects of plant, maternal family and date on visitors and petal damage. Asterisks indicate models that are significantly worse than the most inclusive model. The best random effects model is in bold. By design, maternal family and plant were nested within source site in all models they were included in. All fixed effects were included in all tested models.

Bee Fly (model not including Nitidulid beetle (fixed Petal damage (fixed effects damage) effects do not include include beetle, but not obs) damage) Plant nested in family AIC=707.35 AIC=253.77 AIC=166.11 AIC=2069.90 crossed with date Log-likelihood=-339.68 Log - likelihood= -113.88 Log-likelihood=-70.055 Log-likelihood=-1021.55

Plant nested in family AIC=803.04 AIC=251.76 AIC=164.12 AIC=3149.8 Log-likelihood= -388.52*** Log - likelihood= -113.88 Log-likelihood=-70.062 Log-likelihood= -1563.9***

Date AIC=712.03 AIC=252.87 AIC=175.40 AIC=2828.3 Log-likelihood= -344.02* Log - likelihood= - 116.44 . Log-likelihood= -76.700** Log-likelihood= -1404.2 ***

Family crossed with AIC=707.95 AIC=255.09 AIC=171.45 AIC=2593.5 date Log-likelihood= -340.97 Log - likelihood= - 116.54 Log-likelihood= -73.726** Log-likelihood= -1285.8***

Plant crossed with AIC=707.24 AIC=250.54 AIC=164.11 AIC= 2067.9 date Log-likelihood= -340.62 Log - likelihood= - 114.27 * Log-likelihood= -70.055 Log-likelihood= -1023.0

Family AIC=802.55 AIC=252.79 AIC=169.46 AIC=3806.4 Log-likelihood= -389.28*** Log - likel ihood= -116.39 * Log - likelihood= - 73.729** Log-likelihood= -1893.2***

Plant AIC=803.33 AIC=248.52 AIC=162.12 AIC=3148.0 Log-likelihood=-389.67*** Log-likelihood= -114.26 Log - likelihood= -70.062 Log-likelihood= -1564.0***

TABLE 4-2. Effect sizes (B) and associated z-values (z) and P-values (P) of source population, insecticide treatment, flower characteristics, floral display size, height and interspecies interactions on pollinator visits and insect-caused petal damage. Significant effects are in bold.

Bee visits Fly visits Beetle Insect visits petal damage (Intercept) B =-0.07 B=-0.27 B=-5.67 B=4.49 z=-0.02 z=-0.03 z=-0.2 z=0.88 P=0.98 P=0.98 P=0.85 P=0.38 Source population B=-0.18 B=-0.63 B=-0.90 B=-0.81 z=-0.62 z=-0.73 z=-0.32 z=-1.68 P=0.54 P=0.46 P=0.75 P=0.09 Insecticide B=-0.25 B=0.29 B=0.39 B=-0.78 treatment z=-1.4 z=0.51 z=0.24 z=-2.55 P=0.16 P=0.61 P=0.81 P=0.01

30-second B=0.04 B=0.03 B=-0.04 observations z=0.75 z=0.86 z=-0.61 na P=0.45 P=0.39 P=0.54 Height B=9.92*10-3 B=2.80*10-3 B=-0.01 B=1.45*10-3 z=0.49 z=0.76 z=-0.57 z=0.66 P=0.62 P=0.45 P=0.57 P=0.51

Corolla color B=-0.01 B=-0.02 B=-4.90*10-3 B=-0.02 z=-1.19 z=-0.77 z=-0.06 z=-1.18 P=0.23 P=0.44 P=0.95 P=0.24 Corolla depth B=6.46*10-4 B=0.02 B=0.15 B=-0.16 z=0.05 z=0.12 z=0.25 z=-1.54 P=0.96 P=0.9 P=0.80 P=0.12 Corolla width B=-0.05 B=0.03 B=8.50*10-4 B=-0.06 z=0.75 z=1.06 z=0.003 z=-0.84 P=0.27 P=0.29 P=1.00 P=0.40 Petal width B=0.11 B=0.09 B=-0.15 B=0.15 z=1.32 z=0.37 z=-0.21 z=1.01 P=0.19 P=0.71 P=0.84 P=0.31 Number open B=0.05 B=0.05 B=0.06 B=0.04 flowers z=5.63 z=2.11 z=1.15 z=4.69 P<0.001 P=0.04 P=0.25 P<0.01 Damage (all B=-0.02 sources) z=-0.49 na na na P=0.62 Beetle visits B=0.05 na na na z=0.22 P=0.83

Nitidulid beetles visited some plants more than others. 35.32% of the variation in beetle visits was explained by plant individual and 0.18% by date, while less than 0.01% was explained by maternal family, and plant was the only random factor with a significant effect (Table 4-1). Even fewer beetles than flies were observed, and there were observation days when I saw no beetles.

With so few visits, power is likely low, but there were no significant effects of any plant trait other than plant identity on beetle visits (Table 4-2).

Most variation in insect-caused petal damage (51.98%) can be explained by observation day: damage became more common later in the season. 35.32% of the variation in insect-related petal damage can be explained by individual plant. Plant and date are both included in the best model

(Table 4-1). Damage persisted for several days after it occurred, so the same damage may have been recorded multiple times, resulting in the relatively large effect of individual plant. Insect-caused petal damage increased with the number of flowers present on a plant and decreased with insecticide treatment (Table 4-2). There is a trend (P=0.09) for plants from the non-local source population

NED to receive less damage than plants from GBT (Table 4-2).

In sum, all visitors preferred plants with more flowers, and damage caused by insects was also greater on these plants (Table 4-2).

Fitness: Estimated total fitness (estimated number of total viable seeds) did not differ among maternal families within populations (<1% of the variation in estimated seed number was related to maternal family). Both supplemental hand pollination and insecticide treatment improved fitness of plants from both source populations (Fig 4-2). Plants from NED produced fewer seeds than plants from GBT in all treatments, but pollen and insecticide increased fitness in similar ways for plants from both source populations (Fig 4-2). If the fitness differential between source populations were caused by differences in pollinator or herbivore visitation in the common garden, adding pollen and

89 insecticide should have had a greater effect on NED plants than GBT plants and treatment means should be similar between the two populations. Therefore, differences in seed production between

NED and GBT plants were not due to pollinator visitation or herbivory, but to other factors. Most of the differences between source populations and treatment groups in overall fitness (estimated total seed production) were related to the number of viable seeds produced per fruit (Fig 4-3), rather than differences in fruit production (Fig 4-4), although the source population, insecticide treatment and pollination affected the number of viable seeds per fruit in different ways (Figs 4-4 & 4-5).

The number of inviable seeds (seeds that had embryos but were shriveled with little endosperm) was reduced by insecticide, but not by pollination (Fig 4-4), and pollination likely increased seed production by increasing the number of fertilized ovules. The number of mature fruits produced did not differ by population or treatment (Fig 4-5), although insecticide did reduce the number of aborted flowers (Fig 4-6). Plants from NED and GBT produced similar numbers of both aborted and mature fruits (Figs 4-5 & 4-6), suggesting that there was little difference in flower production between the two populations. In contrast, plants growing in the GBT source population produced more flowers than plants in the NED source population on average (Chapter 2). Overall, the lower production of seeds per plant by NED in comparison to GBT appears to have been caused by differences in the number of viable seeds per fruit rather than by differences in flower production or fruit set (Fig 4-3 to 4-6). NED plants produced fewer viable seeds, but these seeds were heavier, in both field and garden plants (Fig 4-7). Additionally, insecticide treatment increased mean seed mass in both NED and GBT plants, although this effect was stronger for plants from NED (Fig 4-7).

FIGURE 4-2. Back-transformed means with 95% confidence intervals of square-root transformed estimated total seed number per plant. Family describes less than 1% in the total variation in estimated seed number. Plants from NED produced significantly fewer seeds than plants from GBT. (B=-6.80, F(1,38)=11.57, P<0.01), and both pollination (B=4.56, F(1,196)=5.21, P<0.05) and insecticide (B=5.76, F(1,196)=8.30, P<0.01) increased seed production. Damage per flower does not affect total estimated seed number (B=-0.91, F(1,196)=2.42, P=0.12). Plants that produced no flowers during the garden planting and plants for which there were no damage measurements were not included in analyses. 91

FIGURE 4-3. Raw means of viable seeds per fruit. Bars represent 95% confidence limits. Sample sizes show the number of total fruits for which seeds were counted. Individual plant (nested in family) from which fruits were taken explains 39.12% of the total variance in seeds per fruit, while the family alone describes less than 1%. Plants from NED produced fewer seeds than plants from GBT (B=-6.33, F(1,38)=19.98, P<0.001), and plants that were either treated with pollen (B=4.60, F(1,200)=11.02, P<0.01) or insecticide (B=6.66, F(1,200)=23.12, P<0.0001) produced more seeds per fruit. Treating plants with both pollen and insecticide did not increase seeds per fruit over treating them with insecticide alone (B=-5.99, F(1,200)=4.69, P<0.05). Petal damage also reduced viable seeds per fruit (B=-1.25, F(1,200)=9.66, P<0.01), even when controlling for the increase caused by insecticide. A complete table of results is included in the appendix. 92

FIGURE 4-4. Back-transformed means of inviable seeds per fruit. Bars represent 95% confidence limits. Sample sizes show the number of total fruits for which seeds were counted. Individual plant (nested in family) from which fruits were taken explains 40.14% of the total variance in inviable seeds per fruit, while the family alone describes less than 1%. Damage per flower increased the number of inviable seeds per fruit (B=0.08, F(1,200)=8.03, P<0.01), and some of the effect of insecticide treatment appears to be through a reduction in petal damage (B=-0.14, F(1,200)=5.87, P<0.05). In addition to this effect, plants treated with insecticide had significantly fewer inviable seeds than plants in any other group (B=-0.75, F(1,200)=54.61, P<0.0001). 93

FIGURE 4-5. Back-transformed means of square-root transformed counts of fruits that contained seeds. Bars represent 95% confidence limits. There are no significant differences among populations or treatments, and family explains less than 1% of the variation in fruit production.

FIGURE 4-6. Back-transformed means of square-root transformed counts of aborted flowers. 26% of the variation in aborted flowers was explained by family. Damage per flower increased the number of aborted flowers (B=0.23, F(1,196)=7.92, P<0.01), and insecticide reduced it (B= - 0.75, F(1,196)=7.29, P<0.01), but these effects appear to be separate.

FIGURE 4-7. Mass of 20 seeds per plant for plants in the common garden and in the field. Seeds from NED plants were heavier than seeds from GBT plants (B=3.50, F(1,151)=38.46, P<0.001), and this effect was smaller in the garden than in field populations (B=- 2.19, F(1,151)=7.46, P<0.01). Seed mass of plants in the garden did not differ in seed mass from those found in the original field populations (B=0.20, F(1,151)=0.24, P=0.62). Insecticide increased seed mass (B=2.14, F(1, 151)=30.24, P<0.001), and this effect was stronger for plants from NED than from GBT (B=2.92, F(1, 151)=14.06, P<0.001). 95

DISCUSSION

Local adaptation to herbivores and pollinators is common (reviewed in Laine 2009), but maladaptation and lack of local adaptation also occur frequently (reviewed in Hoeksema and

Forde 2008). Coevolutionary processes can drive patterns of local divergence in phenotype

(Thompson 1999, Gomulkiewicz et al. 2000, Thompson 2009), and while systems with tightly- paired specialists are most frequently studied in this context (eg. Miller 1981, Anderson and

Johnson 2008, Anderson et al. 2010, Nattero et al. 2010, Nattero et al. 2011), local adaptation may also occur in systems with many interacting partners (Cane et al. 2005, Gomez et al. 2009a,

Gomez et al. 2009b, Schlumpberger et al. 2009). Erysimum capitatum plants in natural populations are visited by a wide variety of pollinators and damaged by a number of herbivores, some of which prefer the same traits as pollinators (Chapter 3). The present study detected no evidence for local adaptation to either pollinators or herbivores in E. capitatum. In the common garden, plants from both sites were pollen-limited, although they were not in natural populations.

Seed production for plants from both sites also increased with the addition of insecticide, as they did in natural populations. However, these effects were similar for plants from both sites, so pollinators and herbivores were not responsible for the differences in reproduction between the two source sites. Non-local plants did produce fewer seeds than local plants, and received a greater benefit in mean seed mass from the addition of insecticide. Plants from the two populations differed in seed number and seed mass suggesting different patterns of investment.

Pollinator and herbivore preferences:

Both pollinators and herbivores preferred plants with more flowers, but showed no preferences for any other measured trait or for either source population (Table 4-2). Similarly, in native field populations, preferences of pollinators and herbivores for flower number were

96 consistent, but preferences for other traits were weak and variable (Chapter 3). The attractiveness of greater flower number could result in conflicting selection by herbivores and pollinators (e.g. Juenger and Bergelson 1997, Krupnick et al. 1999, Adler 2000, McCall and

Irwin 2006, McCall 2008, 2010). Flower number is the plant reproductive character most commonly found to be under selection in plants (Harder and Johnson 2009), but its measurement is complex because it can be quite plastic and change in response to many environmental factors, including visitation by pollinators and herbivores (Marshall et al. 1986, Stanton et al. 1987,

Harder and Johnson 2005, Wolfe and Mazer 2005). Plants from the two source populations differed in characteristics of individual flowers, with plants from NED having wider, deeper corollas with wider, redder petals (Appendix 4-7). Still, local pollinators and herbivores did not discriminate between the local (GBT) and non-local (NED) plants (Table 4-2), probably because plants in the experimental gardens produced similar numbers of flowers (Figs 4-5 & 4-6).

Effects of pollinators on fitness:

Generalized pollination may arise from shifting pollinator community composition in time and space (Waser et al. 1996, Dilley et al. 2000, Ellis and Johnson 2009), and may select for divergent flower and reproductive characters when populations have different pollinator community compositions, or when pollinator behavior differs among populations (Dilley et al. 2000). Local adaptation to pollinators would likely result in greater pollen-limitation for non-local (NED) plants, but both populations showed similar increases in seed production with added pollen (Figs 4-2 & 4-

3). Viable seed increased through the number of ovules that were fertilized per fruit in plants with added pollen (Figs 4-3 & 4-4) rather than through increased fruit production (Figs 4-4 & 4-6).

Garden plants received both water and fertilizer and were larger and had more flowers than plants in natural populations. They were also separated from a natural source of insect pollinators specific to

97 this plant. The larger size, and distance from the source of pollinators may have increased the likelihood of pollen limitation (Lawrence 1993, Burd 1994, Ashman et al. 2004, Burd 2008, Burd et al. 2009), which did not occur in natural populations. Alternatively, increases in seed production with hand-pollination may have resulted from pollen quality rather than quantity as plants may mature more seeds when pollen comes from a more diverse group of donors (Aizen and Harder

2007).

Effects of flower herbivores on fitness:

If there is local adaptation to herbivores, excluding them should benefit non-local plants disproportionately, while plant maladaptation would cause the opposite result. For both source populations, insecticide treatment produced similar increases in estimated total seed production per plant (Fig 4-2), seed production per fruit (Fig 4-3), reduced the numbers of aborted flowers (Fig 4-

6), and increased mean seed mass (Fig 4-7). Seed production per fruit increased through a reduction in the number of inviable seeds (Fig 4-5), and insecticide-treated plants were taller at harvest

(F(1,248)=6.86, p<0.01), suggesting there was an increase in the resources available to mature seeds.

Reduced herbivory could have resulted in greater vigor of insecticide-treated plants (Table 4-2), but petal damage did not decrease total seed production per plant, and the effects of insecticide only interacted with those of petal damage for the number of inviable seeds. Additionally, vegetative herbivory was negligible in the common garden (personal observation), and is unlikely to be responsible for increases in vigor in insecticide-treated plants. It is possible that nitrite additions to the soil through the breakdown of the herbicide, imidacloprid, (Zheng and Liu 1999, Liu et al. 2006) could account for the apparent increase in resource availability to insecticide-treated plants.

Nitrogen levels are extremely low at the garden site (Cherwin et al. 2009) and supplemental fertilizer

98 may not have eliminated differences in nutrient availability between plants treated with insecticide and controls (Tim Seastedt, personal communication September 23, 2011).

Interactive effects of pollination and herbivory:

In natural populations, pollinators sometimes preferred plants with less herbivore damage

(Chapter 3). In contrast, herbivory did not decrease pollinator visitation in the common garden

(Table 4-1), and pollen limitation was of similar magnitude for insecticide-treated and untreated plants (Fig 4-2). At the level of individual flowers, addition of supplemental pollen to insecticide- treated plants did not change seed maturation per fruit compared to plants treated only with insecticide, which could indicate plants treated with insecticide received more pollinator visits per flower. However, the effect of supplemental pollination on seed and fruit production did not differ between damaged and undamaged plants (Figs 4-2 to 4-5), and there was a slight, non-significant, decrease in bee visits on insecticide-treated plants. Increased resource availability can reduce pollen-limitation associated with pollen quality rather than pollinator visitation, as resource-rich conditions allow more fertilized ovules to be provisioned and reduces maternal selectivity (Marshall and Diggle 2001). The lack of pollen-limitation in insecticide -treated plants is likely due to increased resource levels, given the reduction in inviable seeds.

Local adaptation:

Although pollinator communities and some types of herbivores differed between the source populations of E. capitatum (Chapter 3), differences in fitness between NED and GBT plants in the common garden are likely due to factors other than adaptation to local suites of pollinators and herbivores. Seed production of the non-local (NED) individuals was significantly lower than that of local (GBT) individuals at both the whole-plant and fruit levels

(Figs 4-2 & 4-3), but neither pollinator nor herbivore visitation differed between plants of the

99 two populations (Table 4-2) and supplemental pollination and insecticide treatment increased seed production in plants from both source populations equally (Figs 4-2 to 4-6). The lack of local adaptation observed in this experiment is consistent with the prediction that vistiation by multiple pollinators and a variety of herbivores limits the potential for population divergene.

However, stem boring beetle larvae and flower bud galls were common in natural populations

(Chapter 3) but were not present in the common garden, and this may have reduced the likelihood of detecting local adaptation.

The increase in seed number in local plants in comparison to non-local plants (Fig 4-3) may have been due to differences between plants from the two source populations in ovule number per fruit, fertilization success, or abortion of fertilized ovules. Ovule number varies among Spergularia marina plants from different source populations when grown in common gardens (Mazer and Delesalle 1996), but is quite plastic in some species (Mazer and Delesalle

1996, Vogler et al. 1999) and plasticity of ovule number can vary among genotypes (Vogler et al. 1999). I did not measure the total number of ovules per fruit, however, differences strategies in ovule packaging and provisioning can be inferred. Tradeoffs between seed size and seed number are common (Linhart 1988, Agren 1989, Venable 1992, Jakobsson and Eriksson 2000,

Westoby et al. 2002), and the fact that seeds produced by plants from NED were larger than those produced by plants from GBT (Fig 4-7) suggests that flowers of NED plants may produce fewer ovules per ovary. Yet, differences in seed size are small in comparison to differences in seed number (Figs 4-3 & 4-7), and do not account for the large differences in seed production.

Greater seed production by GBT plants also may be due to a difference in fertilization success and selective post-fertilization abortion of ovules. A mixture of pollen from equal numbers of GBT and NED pollen donors was used for supplemental pollinations, and pollinators visiting

100 plants would likely have moved roughly equal quantities of NED and GBT pollen. Nevertheless, plants from the two source populations may have responded differently to the same mix of pollen.

Pollen from more distant populations can reduce seed production for multiple reasons, including outbreeding depression (e.g. Price and Waser 1979, Bermingham and Brody 2011), inbreeding effects, and other consequences of specific combinations of maternal and paternal genotypes

(Schemske and Pautler 1984, Sork and Schemske 1992, Marshall and Diggle 2001). Additionally, non-local (NED) plants may have aborted more fertilized ovules than local (GBT) plants. Selective abortion of fertilized ovules can occur more frequently in resource-limited or stressed plants (Lee and Bazzaz 1986, Diggle et al. 2010). The addition of insecticide increased seed size more for NED plants than for GBT plants (Fig 4-7), but there was no other evidence that resources were more limiting for NED plants than for GBT plants. In sum, differences in seed production between the two populations likely has less to do with local adaptation to biotic factors or resource availability than to innate, and potentially adaptive, differences in the production of ovules per fruit.

Conclusions: This study examined the effects of supplemental pollination and herbivore deterrence on plant fitness and tested for local adaptation to groups of pollinators and herbivores in Erysimum capitatum. Plants from both sites were pollen limited in the common garden, although they were not in natural populations (Chapter 3), perhaps because garden plants had more flowers than plants in natural populations. In addition, insecticide increased seed production, and this influence may have occurred through both reductions in herbivory and nutrient addition. Increases in seed production with both pollen and insecticide addition were similar for plants from both source populations, indicating that pollinators and herbivores were not responsible for differences in reproduction between the two source sites. The two study populations of E. capitatum do not show signs of local adaptation to pollinators and herbivores, but show different patterns of seed production. NED plants

101 produced fewer, but heavier seeds than GBT plants, and mean seed mass increased more with the addition of insecticide. The differences between populations in seed provisioning and production suggest adaptation to environmental factors other than pollinators and/or herbivores and warrant further investigation.

102

B F df P (Intercept) 29.94 897.85 196 <0.0001 Damage -0.91 2.43 196 0.12 Population -6.80 11.57 38 <0.01 Pollinate 4.56 5.21 196 <0.05 Insecticide 5.76 8.30 196 <0.01 Damage×Population -0.42 0.13 196 0.72 Damage×Pollinate 1.81 2.42 196 0.12 Population×Pollinate -1.03 0.07 196 0.80 Damage×Insecticide 1.05 0.81 196 0.37 Population×Insecticide 2.68 0.45 196 0.50 Pollinate×Insecticide -5.97 2.23 196 0.14 Damage ×Population×Pollinate 0.72 0.10 196 0.76 Damage×Population×Insecticide 1.94 0.69 196 0.41 Damage×Pollinate×Insecticide 0.65 0.08 196 0.78 Population×Pollinate×Insecticide -6.32 0.63 196 0.43 Damage×Population×Pollinate ×Insecticide 0.01 0.00 196 0.99 APPENDIX 4-1. Fixed effects of petal damage, source population, pollination and insecticide treatment on total estimated seeds for linear mixed effects model tested with plant nested in family as random effects.

103

B F DF P (Intercept) 20.89 870.67 4081 <0.0001 Damage -1.25 9.66 200 <0.01 Population -6.33 19.99 38 <0.001 Pollinate 4.60 11.03 200 <0.01 Insecticide 6.66 23.12 200 <0.0001 Damage×Population 0.23 0.08 200 0.99 Damage×Pollinate 0.36 0.20 200 0.78 Damage×Insecticide 0.17 0.05 200 0.83 Population×Pollinate -1.78 0.41 200 0.52 Population×Insecticide 0.18 0.00 200 0.95 Pollinate×Insecticide -6.00 4.69 200 <0.05 Damage ×Population×Pollinate 1.56 0.94 200 0.33 Damage×Population×Insecticide 0.93 0.33 200 0.56 Damage×Pollinate×Insecticide 1.96 1.48 200 0.22 Population×Pollinate×Insecticide 0.63 0.01 200 0.91 Damage×Population×Pollinate ×Insecticide -2.14 0.44 200 0.51

APPENDIX 4-2. Fixed effects of petal damage, source population, pollination and insecticide treatment on viable seeds per fruit for linear mixed effects model tested with plant nested in family as random effects.

B F DF P (Intercept) 0.65 166.95 4081 0 Damage 0.08 8.03 200 <0.01 Population -0.05 0.23 38 0.06 Pollinate -0.07 0.49 200 0.49 Insecticide -0.75 54.61 200 0.00 Damage×Population 0.03 0.31 200 0.58 Damage×Pollinate 0.03 0.28 200 0.60 Damage×Insecticide 0.08 0.16 200 0.69 Population×Pollinate -0.14 5.87 200 0.02 Population×Insecticide 0.15 0.57 200 0.45 Pollinate×Insecticide 0.15 0.58 200 0.45 Damage ×Population×Pollinate -0.05 0.19 200 0.66 Damage×Population×Insecticide -0.03 0.06 200 0.81 Damage×Pollinate×Insecticide 0.04 0.11 200 0.74 Population×Pollinate×Insecticide 0.12 0.09 200 0.76 Damage×Population×Pollinate -0.31 1.71 200 0.19 ×Insecticide APPENDIX 4-3. Fixed effects of petal damage, source population, pollination and insecticide treatment on inviable seeds per fruit for linear mixed effects model tested with plant nested in family as random effects.

104

B F DF P (Intercept) 7.31 1595.69 196 0.00 Damage -0.02 0.04 196 0.84 Population -0.47 1.63 38 0.21 Pollinate 0.37 1.01 196 0.32 Insecticide 0.05 0.02 196 0.88 Damage×Population -0.13 0.36 196 0.55 Damage×Pollinate 0.29 1.82 196 0.18 Damage×Insecticide -0.21 0.08 196 0.78 Population×Pollinate 0.15 0.50 196 0.48 Population×Insecticide 0.78 1.14 196 0.29 Pollinate×Insecticide -0.50 0.47 196 0.50 Damage ×Population×Pollinate 0.06 0.02 196 0.88 Damage×Population×Insecticide 0.14 0.11 196 0.75 Damage×Pollinate×Insecticide 0.12 0.08 196 0.77 Population×Pollinate×Insecticide -0.52 0.13 196 0.72 Damage×Population×Pollinate ×Insecticide 0.13 0.02 196 0.88

APPENDIX 4-4. Fixed effects of petal damage, source population, pollination and insecticide treatment on fruits with seeds for linear mixed effects model tested with plant nested in family as random effects.

B F DF P (Intercept) 4.09 508.26 196 0.00 Damage 0.23 7.92 196 0.01 Population 0.09 0.06 38 0.81 Pollinate -0.22 0.63 196 0.43 Insecticide -0.75 7.29 196 0.01 Damage×Population 0.11 0.46 196 0.50 Damage×Pollinate 0.00 0.00 196 0.98 Damage×Insecticide -0.42 0.58 196 0.45 Population×Pollinate -0.18 1.20 196 0.27 Population×Insecticide -0.07 0.01 196 0.91 Pollinate×Insecticide 0.43 0.58 196 0.45 Damage ×Population×Pollinate -0.38 1.35 196 0.25 Damage×Population×Insecticide -0.21 0.40 196 0.53 Damage×Pollinate×Insecticide -0.15 0.21 196 0.64 Population×Pollinate×Insecticide 1.00 0.81 196 0.37 Damage×Population×Pollinate ×Insecticide 0.72 1.18 196 0.28 APPENDIX 4-5. Fixed effects of petal damage, source population, pollination and insecticide treatment on aborted flowers for linear mixed effects model tested with plant nested in family as random effects.

105

B F P Intercept 6.93 602.85 < 0.0001 Population 3.50 38.46 < 0.0001 Location (field or 0.20 0.24 0.62 garden) Insecticide 2.14 30.24 <0.0001 Pollinate 0.06 0.02 0.89 Population×Location -2.19 7.46 <0.01 Population×Insecticide 2.92 14.06 <0.001 Population×Pollinate 0.17 0.05 0.83 APPENDIX 4-6. Effects of source population, location (field or garden), insecticide and pollination treatments on the mass of 20 seeds per plant.

APPENDIX 4-7. Flower trait means for plants in the common garden from each of the two source populations for which all flower traits were measured (N=103 plants from GBT, N=90 plants from NED). Bars represent 95% confidence intervals. 106

CHAPTER V

CONCLUSIONS

My dissertation dealt with complex interactions among plants, pollinators and herbivores in different populations along an elevational gradient in the Rocky Mountains. I examined the potential for tradeoffs between pollinator attraction and herbivore deterrence, as well as for local adaptation to these challenges in Erysimum capitatum. I also examined the potential for plant developmental plasticity to change the outcome of interactions with pollinators and herbivores and the potential for population differences in resource allocation. The important contributions of my work are as follows:

• I found phenotypic differences among four natural populations of Erysimum capitatum

along an elevational gradient in the Rocky Mountains (Chapter 2).

• I described and quantified differences among the communities of pollinators and

herbivores in those populations (Chapters 2 & 3).

• I found different patterns of selection by pollinators and herbivores in these multiple

natural populations over a two-year period (Chapter 3).

• I showed plant responses by E. capitatum mediate interactions with both pollinators and

herbivores and buffer against changes in communities of insects in time and over

geographical space (Chapter 3).

• I conclusively showed there was no evidence of local adaptation to pollinators and

herbivores in a common garden experiment carried out with plants from two of the four

source populations (Chapter 4).

107

• I showed there was the possibility for local adaptation to other factors in that common

garden experiment (Chapter 4).

• I identified the potential for different strategies of resource allocation in plants from the

two populations used in the common garden (Chapter 4).

Implications:

Although Erysimum capitatum is a widespread species, its interactions with animals are not well-documented. I identified pollinators and herbivores of this species in several locations along an elevational gradient, and examined phenotypic selection in these populations in two years. I found that it is visited by a diversity of pollinators and herbivores. E. capitatum is known to be quite phenotypically variable both within and among populations, and I documented variation in traits likely to interact with pollinators and herbivores among populations to lay the groundwork for further study the causes of this variation. I found that for some traits, variation among populations was consistent between years, although there was also some variation within populations between years (Chapter 2).

I found that Erysimum capitatum is quite similar to congeneric species, including Erysimum mediohispanicum, that have been the focus of investigation into the habits of pollinators and herbivores by the lab of J.M. Gomez-Reyes in Spain and Portugal. E. capitatum and the European

Erysimum species are visited by similar types of herbivores and pollinators, and share many of the same obvious variations in phenotypic characters (personal observation). Like the populations of E. capitatum I studied in the Rocky Mountains, E. mediohispanicum shows variation across short spatial scales over an elevational range in flower number, inflorescence height, and flower color. In addition, E. mediohispanicum varies in flower shape among sites, and this variation appears to be the result of local adaptation to particular types of pollinators. Local adaptation was not evident in the

108 two populations of E. capitatum I examined in Colorado, despite the prevalence of different pollinators in different areas, perhaps because plants did not appear to be pollen-limited in the source populations.

Phenotypic selection studies of E. mediohispanicum have mainly focused on the effects of pollinators and vertebrate herbivores (Gomez 2005, Gomez et al. 2006, Gomez et al. 2007, Gomez

2008, Gomez et al. 2008, Gomez et al. 2009b), although this species is also visited by a suite of insect herbivores similar to those that attack E. capitatum, including gall-makers (Gomez 2008), ants and petal herbivores (personal observation, personal communication, M. Gomez). My work on E. capitatum shows that these insect herbivores can have an impact on pollinator visitation.

Unlike the populations of E. capitatum I studied, E. mediohispanicum is pollen limited in most sites (Gomez et al. 2010), and selection for flower shape by pollinators is common and variable among multiple sites (Gomez et al. 2006, Gomez et al. 2009b, Gomez and Perfectti 2010).

Zygomorphy and petal width are the main floral shape traits Gomez et al. have found to vary and to be under selection (Gomez et al. 2006, Gomez et al. 2008, Gomez et al. 2009b, Gomez and Perfectti

2010). I observed, but did not measure, shape variation in flowers. There was variation in ratios of petal width to length, and in zygomorphy similar to that observed in E. mediohispanicum and other

Erysimum species by Gomez et al. in Europe. This raises the possibility that the variation in flower color and shape that contributes to the maintenance of a generalized pollination system in this genus are ancient and have been maintained through multiple speciation events. Regardless of whether or not the variation in the genus Erysimum is ancient or has arisen independently in multiple species, such variation is useful for study. A group of closely related species with similar patterns of phenotypic variation and multiple potential selective agents that is found on multiple continents can provide information about the interplay of phenotypic variation, selection by biotic forces and the

109 maintenance of variation in particular traits across time and space even when selection occasionally favors particular trait values.

My experimental manipulations of pollination and herbivory in multiple populations showed that plant responses could act to limit selection by both pollinators and herbivores by reducing their impacts on fitness. Specifically, prolonged stigmatic receptivity, flower abortion, new flower production, and allocation to individual fruits proved to be key aspects of plant response that may influence interactions with pollinators and herbivores to the benefit of plants.

Plasticity has long been known to be an important contributor to phenotypic variation among and within environments. Plastic responses that affect characters likely to interact with other organisms and affect selective outcomes are widespread in plant species (reviewed in Agrawal

2001). In fact, some of the characters that are most frequently shown to be under selection by pollinators and herbivores, such as plant height, flower production and defensive chemistry, are also quite plastic (Bradshaw 1965, Marshall et al. 1986, Stanton et al. 1987, DeWitt et al. 1998,

Meagher and Delph 2001, Harder and Johnson 2005, Wolfe and Mazer 2005). Although plasticity may represent the best evolutionary option for plants as they are rooted in place and unable to move to different conditions, there are costs to maintaining developmental flexibility for some characters (DeWitt et al. 1998, Dechaine et al. 2007, van Kleunen and Fischer 2007), and plasticity itself may be under selection in some cases (de Jong 2005). Plasticity may evolve frequently in response to variable conditions in either time or space, and can reduce ecological divergence by allowing individuals to disperse more readily among populations (Thibert-Plante and Hendry 2011).

Few studies have explicitly tested the effects of herbivory and pollinator visitation jointly on a species with a generalized pollination system and many herbivores. My studies in natural

110 populations of E. capitatum showed that pollinator and herbivore preferences and communities varied among years and sites and that herbivores, at least, select for plant traits. I therefore designed a reciprocal transplant experiment to explicitly test for local adaptation to either or both groups. I incorporated treatments to exclude herbivores and the effects of pollinators with insecticide and supplemental pollen, and applied them to plants from both populations.

Unfortunately, one of the common gardens in the reciprocal transplant experiment was badly damaged by vertebrate herbivores, and fitness results were unobtainable. Neither pollen limitation nor selection by pollinators was evident in natural populations of E. capitatum. Yet, pollen limitation was present in the common garden, raising the possibilities that measurements of fitness may not have been accurate enough to identify pollen limitation in the field populations or that pollen limitation was caused by resource additions in the common garden.

Selection by herbivores was occasionally important in natural populations for the flower and reproductive traits measured. I found that compensatory reproductive mechanisms such as flower production may be activated by both pollination and herbivory, and that these may allow

E. capitatum to succeed in its complex environment. Many plants may respond to herbivory and pollen receipt by changing patterns of fruit and seed set (Marshall et al. 1985, Stephenson 1992,

Knight et al. 2005, Wise and Cummins 2006, Nunez-Farfan et al. 2007, Wesselingh 2007, Wise and Abrahamson 2007). The heritability of such compensatory strategies can be difficult to measure (but see Juenger and Bergelson 2000, Wise et al. 2008). Nonetheless, mechanisms that allow plants to make iterative decisions about reproductive resource allocation over the course of the flowering season may be under selection in E. capitatum, and this will be an interesting area of future study.

111

Erysimum capitatum inhabits an adaptive landscape shaped by diverse interacting biotic partners that shift in importance in time and space. Geographic variation in plant morphology can be caused by a number of factors including selection by abiotic forces, biotic forces, genetic drift and phenotypic plasticity. Drift and limited gene flow may act to create different average phenotypes in small populations, but local adaptation may explain at least some of the divergent characters found in different populations of the same species, and environmental factors may affect the plastic expression of traits. Additionally, hybridization events may increase the amount of variation present both within and among populations. The variability of E. capitatum among populations has caused considerable taxonomic confusion (Price 1984, Turner 2006). It is possible that some of the variability found among populations of E. capitatum is a result of hybridization between historically distinct subspecies, and is not caused by current drivers of divergence. In addition, plasticity might cause some of the divergence in traits such as size and perennation among populations. However, the results of this dissertation research show that there is considerable variation in both pollinator and herbivore insect communities on E. capitatum not only in space, but in time. Thus episodes of selection could result in great phenotypic variation within and among the populations.

112

LITERATURE CITED

Abdala-Roberts, L., V. Parra-Tabla, L. Salinas-Peba, and C. M. Herrera. 2009. Noncorrelated effects of seed predation and pollination on the perennial herb Ruellia nudiflora remain spatially consistent. Biological Journal of the Linnean Society 96:800-807.

Adler, L. S. 2000. Alkaloid uptake increases fitness in a hemiparasitic plant via reduced herbivory and increased pollination. American Naturalist 156:92-99.

Adler, L. S. and J. L. Bronstein. 2004. Attracting antagonists: does floral nectar increase leaf herbivory? Ecology 85:1519-1526.

Agrawal, A. A. 2000. Overcompensation of plants in response to herbivory and the by- product benefits of mutualism. Trends in Plant Science 5:309-313.

Agrawal, A. A. 2001. Phenotypic plasticity in the interactions and evolution of species. Science 294:321-326.

Agren, J. 1989. Seed size and number in Rubus chamaemorus: between-habitat variation, and effects of defoliation and supplemental pollination. Journal of Ecology 77:1080-1092.

Aigner, P.A. 2005. Variation in pollination performance gradients in a Dudleya species complex: can generalization promote floral divergence? Functional Ecology 19:681-689.

Aizen, M. A. and L. D. Harder. 2007. Expanding the limits of the pollen-limitation concept: Effects of pollen quantity and quality. Ecology 88:271-281.

Anderson, B., R. Alexandersson, and S. D. Johnson. 2010. Evolution and coexistence of pollination ecotypes in an African Gladiolus (Iridaceae). Evolution 64:960-972.

Anderson, B. and S. D. Johnson. 2008. The geographical mosaic of coevolution in a plant-pollinator mutualism. Evolution 62:220-225.

Anderson, B. and S. D. Johnson. 2009. Geographical covariation and local convergence of flower depth in a guild of fly-pollinated plants. New Phytologist 182:533-540.

Armbruster, W. S. 1997. Exaptations link evolution of plant-herbivore and plant- pollinator interactions: a phylogenetic inquiry. Ecology 78:1661-1672.

Armbruster, W. S., J. Lee, and B. G. Baldwin. 2009. Macroevolutionary patterns of defense and pollination in Dalechampia vines: adaptation, exaptation, and evolutionary novelty. Proceedings of the National Academy of Sciences of Sciences of the United States of America 106:18085-18090. 113

Arroyo, M. T. K., R. Primack, and J. Armesto. 1982. Community studies in pollination ecology in the high temperate andes of central Chile: pollination mechanisms and altitudinal variation. American Journal of Botany 69:82-97.

Ashman, T. L. and E. A. King. 2005. Are flower-visiting ants mutualists or antagonists? A study in a gynodioecious wild strawberry. American Journal of Botany 92:891- 895.

Ashman, T. L., T. M. Knight, J. A. Steets, P. Amarasekare, M. Burd, D. R. Campbell, M. R. Dudash, M. O. Johnston, S. J. Mazer, R. J. Mitchell, M. T. Morgan, and W. G. Wilson. 2004. Pollen limitation of plant reproduction: ecological and evolutionary causes and consequences. Ecology 85:2408-2421.

Ashman, T. L. and D. J. Schoen. 1994. How long should flowers live? Nature 371:788- 791.

Bates, D. and M. Maechler. 2010. lme4: mixed-effects models using S4 classes. R package version 0.999375-34. http://CRAN.R-project.org/package=lme4

Beattie, A. J., S. Jobson, and R. B. Knox. 1985. The vulnerability of pollen and fungal spores to ant secretions: evidence and some evolutionary implications. American Journal of Botany 72:606-614.

Benkman, C. W. 1999. The selection mosaic and diversifying coevolution between crossbills and lodgepole pine. American Naturalist 153:S75-S91.

Bermingham, L. H. and A. K. Brody. 2011. Pollen source affects female reproductive success and early offspring traits in the rare endemic plant Polemonium vanbruntiae (Polemoniaceae). Plant Species Biology 26:244-253.

Bingham, R. A. and A. R. Ort. 1998. Efficient pollination of alpine plants. Nature 391:238-239.

Bolker, B. M., M. E. Brooks, C. J. Clark, S. W. Geange, J. R. Poulsen, M. H. H. Stevens, and J. S. S. White. 2009. Generalized linear mixed models: a practical guide for ecology and evolution. Trends in Ecology & Evolution 24:127-135.

Bradley, K. L., E. I. Damschen, L. M. Young, D. Kuefler, S. Went, G. Wray, N. M. Haddad, J. M. H. Knops, and S. M. Louda. 2003. Spatial heterogeneity, not visitation bias, dominates variation in herbivory. Ecology 84:2214-2221.

Bradshaw, A. D. 1965. Evolutionary significance of phenotypic plasticity in plants. Advances in Genetics 13:115-155.

Brody, A. K. 1997. Effects of pollinators, herbivores and seed predators on flowering phenology. Ecology Letters 78:1624-1631. 114

Brody, A. K. and R. J. Mitchell. 1997. Effects of experimental manipulation of inflorescence size on pollination and pre-dispersal seed predation in the hummingbird-pollinated plant Ipomopsis aggregata. Oecologia 110:86-93.

Brunet, J. 2009. Pollinators of the Rocky Mountain columbine: temporal variation, functional groups and associations with floral traits. Annals of Botany 103:1567- 1578.

Burd, M. 1994. Bateman's principle and plant reproduction: the role of pollen-limitation in fruit and seed set. The Botanical Review 60:83-139.

Burd, M. 2008. The Haig-Westoby model revisited. American Naturalist 171:400-404.

Burd, M., T. L. Ashman, D. R. Campbell, M. R. Dudash, M. O. Johnston, T. M. Knight, S. J. Mazer, R. J. Mitchell, J. A. Steets, and J. C. Vamosi. 2009. Ovule number per flower in a world of unpredictable pollination. American Journal of Botany 96:1159-1167.

Burkle, L. A., R. E. Irwin, and D. A. Newman. 2007. Predicting the effects of nectar robbing on plant reproduction: Implications of pollen limitation and plant mating system. American Journal of Botany 94:1935-1943.

Campbell, D. R., N. M. Waser, and E. J. Melendez-Ackerman. 1997. Analyzing pollinator-mediated selection in a plant hybrid zone: Hummingbird visitation patterns on three spatial scales. American Naturalist 149:295-315.

Cane, J. H., R. Minckley, L. Kervin, and T. Roulston. 2005. Temporally persistent patterns of incidence and abundance in a pollinator guild at annual and decadal scales: the bees of Larrea tridentata. Biological Journal of the Linnean Society 85:319-329.

Cariveau, D., R. E. Irwin, A. K. Brody, L. S. Garcia-Mayeya, and A. von der Ohe. 2004. Direct and indirect effects of pollinators and seed predators to selection on plant and floral traits. Oikos 104:15-26.

Caruso, C. M., S. B. Peterson, and C. E. Ridley. 2003. Natural selection on floral traits of Lobelia (Lobeliaceae): spatial and temporal variation. American Journal of Botany 90:1333-1340.

Cherwin, K. L., T. R. Seastedt, and K. N. Suding. 2009. Effects of nutrient manipulations and gass removal on cover, species composition, and invasibility of a novel grassland in Colorado. Restoration Ecology 17:818-826.

115

Collin, C. L., P. S. Pennings, C. Rueffler, A. Widmer, and J. A. Shykoff. 2002. Natural enemies and sex: how seed predators and pathogens contribute to sex-differential reproductive success in a gynodioecious plant. Oecologia 131:94-102.

Colwell, R. K. 2005. EstimateS: Statistical estimation of species richness and shared species from samples. Version 7.5.

Conner, J. K. and S. Rush. 1996. Effects of flower size and number on pollinator visitation to wild radish, Raphanus raphanistrum. Oecologia 105:509-516.

Conner, J. K. and A. Sterling. 1996. Selection for independence of floral and vegetative traits: evidence from correlation patterns in five species. Canadian Journal of Botany-Revue Canadienne De Botanique 74:642-644.

Crawley, M. J. 2007. Mixed-effects models. Pages 627-660 The R Book. John Wiley and Sons, Ltd.

Danderson, C. A. and B. Molano-Flores. 2010. Effects of herbivory and inflorescence size on insect visitation to Eryngium yuccifolium (Apiaceae) a prairie plant. American Midland Naturalist 163:234-246.

Darwin, C. 1859. The Origin of Species. J. Murray, London.

Darwin, C. D. 1862. On the Various Contrivances by which British and Foreign Orchids are Fertilised by Insects. J. Murray, London.

Darwin, C. D. 1877. The Different Forms of Flowers on Plants of the Same Species. J. Murray, London. de Jong, G. 2005. Evolution of phenotypic plasticity: patterns of plasticity and the emergence of ecotypes. New Phytologist 166:101-117.

Dechaine, J. M., J. A. Johnston, M. T. Brock, and C. Weinig. 2007. Constraints on the evolution of adaptive plasticity: costs of plasticity to density are expressed in segregating progenies. New Phytologist 176:874-882.

Decourtye, A. and J. Devillers. 2010. Ecotoxicity of neonicotinoid insecticides to bees Pages 85-95 in S. H. Thany, editor. Insect Nicotinic Acetylcholine Receptors. Springer-Verlag, Berlin, Germany.

Devoto, M., D. Medan, and N. H. Montaldo. 2005. Patterns of interaction between plants and pollinators along an environmental gradient. Oikos 109:461-472.

DeWitt, T. J., A. Sih, and D. S. Wilson. 1998. Costs and limits of phenotypic plasticity. Trends in Ecology & Evolution 13:77-81.

116

Diggle, P. K., N. J. Abrahamson, R. L. Baker, M. G. Barnes, T. L. Koontz, C. R. Lay, J. S. Medeiros, J. L. Murgel, M. G. M. Shaner, H. L. Simpson, C. C. Wu, and D. L. Marshall. 2010. Dynamics of maternal and paternal effects on embryo and seed development in wild radish (Raphanus sativus). Annals of Botany 106:309-319.

Diggle, P. K. and J. S. Miller. 2004. Architectural effects mimic floral sexual dimorphism in Solanum (Solanaceae). American Journal of Botany 91:2030-2040.

Dilley, J. D., P. Wilson, and M. R. Mesler. 2000. The radiation of Calochortus: generalist flowers moving through a mosaic of potential pollinators. Oikos 89:209-222.

Ehrlen, J. 2002. Assessing the lifetime consequences of plant-animal interactions for the perennial herb Lathyrus vernus (Fabaceae). Perspectives in Plant Ecology Evolution and Systematics 5:145-163.

Ehrlen, J., S. Kack, and J. Agren. 2002. Pollen limitation, seed predation and scape length in Primula farinosa. Oikos 97:45-51.

Elberling, H. and J. M. Olesen. 1999. The structure of a high latitude plant-flower visitor system: the dominance of flies. Ecography 22:314-323.

Ellis, A. G. and S. D. Johnson. 2009. The evolution of floral variation without pollinator shifts in Gorteria diffusa (Asteraceae) American Journal of Botany 96:793-801.

Fineblum, W. L. and M. D. Rausher. 1997. Do floral pigmentation genes also influence resistance to enemies? The W locus in Ipomoea purpurea. Ecology 78:1646-1654.

Fordyce, J. A. 2006. The evolutionary consequences of ecological interactions mediated through phenotypic plasticity. Journal of Experimental Biology 209:2377-2383.

Fox , J. 2008. 8.2 Two-way analysis of variance. Applied Regression Analysis and Generalized Linear Models, Second Edition. Sage Publications.

Fox , J., A. Kramer, and M. Friendly. 2010. sem: Structural equations models. R package version 0.9-20. http://CRAN.R-project.org/package=sem

Frazee, J. E. and R. J. Marquis. 1994. Environmental contribution to floral trait variation in Xhaemaecrista fassiculata (Fabacae Caesalpinoideae) American Journal of Botany 81:206-215.

Frey, F. M. 2004. Opposing natural selection from herbivores and pathogens may maintain floral-color variation in Claytonia virginica (Portulacaceae). Evolution 58:2426-2437.

117

Galen, C. and B. Butchart. 2003. Ants in your plants: effects of nectar-thieves on pollen fertility and seed-siring capacity in the alpine wildflower, Polemonium viscosum. Oikos 101:521-528.

Galen, C. and J. Cuba. 2001. Down the tube: pollinators, predators, and the evolution of flower shape in the alpine skypilot, Polemonium viscosum. Evolution 55:1963- 1971.

Gelman, A. and J. Hill. 2007. Generalized linear models. Pages 109-106.107 Data Analysis Using Regression and Multilevel/Hierarchical Models Cambridge University Press, New York, NY.

Gels, J. A., D. W. Held, and D. A. Potter. 2002. Hazards of insecticides to the bumble bees Bombus impatiens (Hymenoptera : Apidae) foraging on flowering white clover in turf. Journal of Economic Entomology 95:722-728.

Gomez, J. M. 1993. Phenotypic selection on flowering synchrony in a high-mountain plant, Hormathophylla spinosa (Cruciferae). Journal of Ecology 81:605-613.

Gomez, J. M. 2000. Effectiveness of ants as pollinators of Lobularia maritima: effects on main sequential fitness components of the host plant. Oecologia 122:90-97.

Gomez, J. M. 2003. Herbivory reduces the strength of pollinator-mediated selection in the Mediterranean herb Erysimum mediohispanicum: consequences for plant specialization. American Naturalist 162:242-256.

Gomez, J. M. 2005a. Long-term effects of ungulates on performance, abundance, and spatial distribution of two montane herbs. Ecological Monographs 75:231-258.

Gomez, J. M. 2005b. Non-additive effects of herbivores and pollinators on Erysimum mediohispanicum (Cruciferae) fitness. Oecologia 143:412-418.

Gomez, J. M. 2008. Sequential conflicting selection due to multispecific interactions triggers evolutionary trade-offs in a monocarpic herb. Evolution 62:668-679.

Gomez, J. M., M. Abdelaziz, J. P. M. Camacho, A. J. Munoz-Pajares, and F. Perfectti. 2009a. Local adaptation and maladaptation to pollinators in a generalist geographic mosaic. Ecology Letters 12:672-682.

Gomez, J. M., M. Abdelaziz, J. Lorite, A. J. Munoz-Pajares, and F. Perfectti. 2010. Changes in pollinator fauna cause spatial variation in pollen limitation. Journal of Ecology 98:1243-1252.

Gomez, J. M., J. Bosch, F. Perfectti, J. Fernandez, and M. Abdelaziz. 2007. Pollinator diversity affects plant reproduction and recruitment: the tradeoffs of generalization. Oecologia 153:597-605. 118

Gomez, J. M., J. Bosch, F. Perfectti, J. D. Fernandez, M. Abdelaziz, and J. P. M. Camacho. 2008. Spatial variation in selection on corolla shape in a generalist plant is promoted by the preference patterns of its local pollinators. Proceedings of the Royal Society 275:2241-2249.

Gomez, J. M. and F. Perfectti. 2010. Evolution of complex traits: the case of Erysimum corolla shape. International Journal of Plant Sciences 171:987-998.

Gomez, J. M., F. Perfectti, J. Bosch, and J. P. M. Camacho. 2009b. A geographic selection mosaic in a generalized plant-pollinator-herbivore system. Ecological Monographs 79:245-263.

Gomez, J. M., F. Perfectti, and J. P. M. Camacho. 2006. Natural selection on Erysimum mediohispanicum flower shape: Insights into the evolution of zygomorphy. American Naturalist 168:531-545.

Gomez, J. M. and R. Zamora. 1999. Generalization vs. specialization in the pollination system of Hormathophylla spinosa (Cruciferae). Ecology 80:796-805.

Gomez, J. M. and R. Zamora. 2000. Spatial variation in the selective scenarios of Hormathophylla spinosa (Cruciferae). American Naturalist 155:657-668.

Gomulkiewicz, R., J. N. Thompson, R. D. Holt, S. L. Nuismer, and M. E. Hochberg. 2000. Hot spots, cold spots, and the geographic mosaic theory of coevolution. American Naturalist 156:156-174.

Grace, J. B. 2006. Structural equation modeling and natural systems. Cambridge University Press. Cambridge, UK.

Gschwandtner, M. and P. Filzmoser. 2009. mvoutlier: multivariate outlier detection based on robust methods. R package version 1.4. http://www.statistik.tuwien.ac.at/public/filz/

Haloin, J. R. and S. Y. Strauss. 2008. Interplay between ecological communities and evolution review of feedbacks from microevolutionary to macroevolutionary scales. Annals of the New York Academy of Science 1133:87-125.

Harder, L. and S. Johnson. 2005. Adaptive plasticity of floral display size in animal- pollinated plants. Proceedings of the Royal Society 272:2651-2657.

Harder, L. D. and S. D. Johnson. 2009. Darwin's beautiful contrivances: evolutionary and functional evidence for floral adaptation. New Phytologist 183:530-545.

Herrera, C. M. 1988. Variation in mutualisms: the spatio-temporal mosaic of a pollinator assemblage. Biological Journal of the Linnean Society 35:95-125. 119

Herrera, C. M. 1993. Selection on floral morphology and environmental deterrminants of fecundity in a hawk-moth-pollinated violet. Ecological Monographs 63:251-275.

Herrera, C. M. 2000. Measuring the effects of pollinators and herbivores: Evidence for non-additivity in a perennial herb. Ecology 81:2170-2176.

Herrera, C. M., M. Medrano, P. J. Rey, A. M. Sanchez-Lafuente, M. B. Garcia, J. Guitian, and A. J. Manzaneda. 2002. Interaction of pollinators and herbivores on plant fitness suggests a pathway for correlated evolution of mutualism- and antagonism-related traits. Proceedings of the National Academy of Sciences of the United States of America 99:16823-16828.

Hoeksema, J. D. and S. E. Forde. 2008. A meta-analysis of factors affecting local adaptation between interacting species. American Naturalist 171:275-290.

Irwin, R. E. and L. S. Adler. 2006. Correlations among traits associated with herbivore resistance and pollination: implications for pollination and nectar robbing in a distylous plant. American Journal of Botany 93:64-72.

Irwin, R. E., L. S. Adler, and A. K. Brody. 2004. The dual role of floral traits: pollinator attraction and plant defense. Ecology 85:1503-1511.

Irwin, R. E., S. Y. Strauss, S. Storz, A. Emerson, and G. Guibert. 2003. The role of herbivores in the maintenance of a flower color polymorphism in wild radish. Ecology 84:1733-1743.

Jakobsson, A. and O. Eriksson. 2000. A comparative study of seed number, seed size, seedling size and recruitment in grassland plants. Oikos 88:494-502.

Johnson, S. D. and K. E. Steiner. 2000. Generalization versus specialization in plant pollination systems. Trends in Ecology & Evolution 15:140-143.

Juenger, T. and J. Bergelson. 1997. Pollen and resource limitation of compensation to herbivory in scarlet gilia, Ipomopsis aggregata. Ecology 78:1684-1695.

Juenger, T. and J. Bergelson. 2000a. Does early season browsing influence the effect of self-pollination in scarlet gilia? Ecology 81:41-48.

Juenger, T. and J. Bergelson. 2000b. The evolution of compensation to herbivory in scarlet gilia, Ipomopsis aggregata: herbivore-imposed natural selection and the quantitative genetics of tolerance. Evolution 54:764-777.

Kearns, C. A. 1992. Anthophilous fly distribution across an elevation gradient. American Midland Naturalist 127:172-182.

120

Kearns, C. A. and D. S. Inouye. 1993. Techniques for Pollination Biologists. University of Colorado Press, Niwot, Colorado.

Kerdelhue, C. and J. Y. Rasplus. 1996. Non-pollinating Afrotropical fig wasps affect the fig-pollinator mutualism in Ficus within the subgenus Sycomorus. Oikos 75:3-14.

Kim, E. and K. Donohue. 2011. Demographic, developmental and life-history variation across altitude in Erysimum capitatum. Journal of Ecology 99:1237-1249.

Knight, T. M., J. A. Steets, J. C. Vamosi, S. J. Mazer, M. Burd, D. R. Campbell, M. R. Dudash, M. O. Johnston, R. J. Mitchell, and T. L. Ashman. 2005. Pollen limitation of plant reproduction: pattern and process. Annual Review of Ecology Evolution and Systematics 36:467-497.

Kolb, A. and J. Ehrlen. 2010. Environmental context drives seed predator-mediated selection on a floral display trait. Evolutionary Ecology 24:433-445.

Krupnick, G. A. and A. E. Weis. 1999. The effect of floral herbivory on male and female reproductive success in Isomeris arborea. Ecology 80:135-149.

Krupnick, G. A., A. E. Weis, and D. R. Campbell. 1999. The consequences of floral herbivory for pollinator service to Isomeris arborea. Ecology 80:125-134.

Kudoh, H. and D. F. Whigham. 1998. The effect of petal size manipulation on pollinator/seed-predator mediated female reproductive success of Hibiscus moscheutos. Oecologia 117:70-79.

Laine, A. L. 2009. Role of coevolution in generating biological diversity: spatially divergent selection trajectories. Journal of Experimental Botany 60:2957-2970.

Larden, A. and A. M. Triboi-Blondel. 1994. Freezing injury to ovules, pollen and seeds in winter rape. Journal of Experimental Botany 45:1177-1181.

Lawrence, W. S. 1993. Resource and pollen limitation: Plant size-dependent reproductive patterns in Physalis longifolia. American Naturalist 141:296-313.

Lay, C. R., Y. B. Linhart, and P. K. Diggle. 2011. The good, the bad and the flexible: plant interactions with pollinators and herbivores over space and time are moderated by plant compensatory responses. Annals of Botany 108:749-763.

Lee, T. D. and F. A. Bazzaz. 1986. Maternal regulation of fecundity: nonrandom ovule abortion in Cassia fasiculata (Michx). Oecologia 68:459-465.

Linhart, Y. B. 1988. Intrapopulation differentiation in annual plants: the contrasting effects of intraspecific and interspecific competition. Evolution 42:1047-1064.

121

Linhart, Y. B. 1991. Disease, parasitism and herbivory: multidimensional challenges in plant evolution. Trends in Ecology & Evolution 6:392-396.

Linhart, Y. B. and M. C. Grant. 1996. Evolutionary significance of local genetic differentiation in plants. Annual Review of Ecology and Systematics 27:237-277.

Liu, W., W. Zheng, Y. Ma, and K. Liu. 2006. Sorption and degradation of imidacloprid in soil and water. Journal of Environmental Science and Health 41:623-634. Lloyd, D. G. 1980. Sexual strategies in plants: an hypothesis of serieal adjustment of maternal investent during one reproductive session. New Phytologist 86:69-79.

Lomolino, M. V. 2001. Elevation gradients of species-density: historical and prospective views. Global Ecology and Biogeography 10:3-13.

Louda, S. M. and M. A. Potvin. 1995. Effect of inflorescence-feeding insects on the demography and lifetime fitness of a native plant. Ecology 76:229-245.

Magurran, A. E. 2004. Measuring biological diversity. Blackwell Science LtD., Malden, Massachusettes.

Marr, J. W. 1967. Ecosystems of the east slope of the front range in Colorado. University of Colorado Press, Colorado.

Marshall, D. L. and P. K. Diggle. 2001. Mechanisms of differential pollen donor performance in wild radish, Raphanus sativus (Brassicaceae). American Journal of Botany 88:242-257.

Marshall, D. L., D. A. Levin, and N. L. Fowler. 1985. Plasticity in yield components in response to fruit predation and date of fruit initation in three species of Sesbania (Leguminosae). Journal of Ecology 73:71-81.

Marshall, D. L., D. A. Levin, and N. L. Fowler. 1986. Plasticity of yield components in response to stress in Sesbania macrocarpa and Sesbania vesicaria (Leguminosae). American Naturalist 127:508-521.

Mazer, S. J. and V. A. Delesalle. 1996. Floral trait variation in Spergularia marina (Caryophyllaceae): Ontogenetic, maternal family, and population effects. Heredity 77:269-281.

McCall, A. C. 2008. Florivory affects pollinator visitation and female fitness in Nemophila menziesii. Oecologia 155:729-737.

McCall, A. C. 2010. Does dose-dependent petal damage affect pollen limitation in an annual plant? Botany-Botanique 88:601-606.

122

McCall, A. C. and R. E. Irwin. 2006. Florivory: the intersection of pollination and herbivory. Ecology Letters 9:1351-1365.

Meagher, T. R. and L. F. Delph. 2001. Individual flower demography, floral phenology and floral display size in Silene latifolia. Evolutionary Ecology Research 3:845-860.

Miller, R. B. 1981. Hawkmoths and the Geographic patterns of floral variation in Aquilegia caerulea. Evolution 35:763-774.

Nattero, J., A. A. Cocucci, and R. Medel. 2010. Pollinator-mediated selection in a specialized pollination system: matches and mismatches across populations. Journal of Evolutionary Biology 23:1957-1968.

Nattero, J., A. N. Sersic, and A. A. Cocucci. 2011. Geographic variation of floral traits in Nicotiana glauca: Relationships with biotic and abiotic factors. Acta Oecologica- International Journal of Ecology 37:503-511.

Navarro, L., G. Ayensa, and P. Guitian. 2007. Adaptation of floral traits and mating system to pollinator unpredictibility: the case of Disterigma stereophyllum (Ericaceae) in southwestern Colombia. Plant Systematics and Evolution 266:165-174.

Nunez-Farfan, J., J. Fornoni, and P. L. Valverde. 2007. The evolution of resistance and tolerance to herbivores. Annual Review of Ecology Evolution and Systematics 38:541-566.

Paige, K. N. and T. G. Whitham. 1987. Overcompensation in response to mammalian herbivory: the advantage of being eaten. American Naturalist 129:407-416.

Parachnowitsch, A. L. and A. Kessler. 2010. Pollinators exert natural selection on flower size and floral display in Penstemon digitalis. New Phytologist 188:393-402.

Parra-Tabla, V. and C. M. Herrera. 2011. Spatially inconsistent direct and indirect effects of herbivory on floral traits and pollination success in a tropical shrub. Oikos 119:1344- 1354.

Perfectti, F., J. M. Gomez, and J. Bosch. 2009. The functional consequences of diversity in plant- pollinator interactions. Oikos 118:1430-1440.

Pinheiro, J. C. and D. M. Bates. 2000. Mixed-effects models in S and S-plus. Springer Verlag New York, LLC, New York, NY.

Pohl, N., G. Carvallo, C. Botto-Mahan, and R. Medel. 2006. Nonadditive effects of flower damage and hummingbird pollination on the fecundity of Mimulus luteus. Oecologia 149:648-655. 123

Price, M. V. and N. M. Waser. 1979. Pollen dispersal and optimal outcrossing in Delphinium nelsoni. Nature 277:294-297.

Price, R. A. 1984. The systematics of the Erysimum capitatum alliance (Brassicaceae) in North America. PhD. University of California, Davis.

R Development Core Team. 2010. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.

Rand, T. A. 2002. Variation in insect herbivory across a salt marsh tidal gradient influences plant survival and distribution. Oecologia 132:549-558.

Renwick, J. A. A., C. D. Radke, and K. Sachdevgupta. 1989. Chemical constituents of Erysimum cheiranthoides deterring oviposition by the cabbage butterfly, Pieris rapae. Journal of Chemical Ecology 15:2161-2169.

Rey, P. J., C. M. Herrera, J. Guitian, X. Cerda, A. M. Sanchez-Lafuente, M. Medrano, and J. L. Garrido. 2006. The geographic mosaic in predispersal interactions and selection on Helleborus foetidus (Ranunculaceae). Journal of Evolutionary Biology 19:21-34.

Rogers, R. E. L. and J. R. Kemp. 2003. Imidacloprid, potatoes, and honey bees in Atlantic Canada: is there a connection? Bulletin of Insectology 5688:83-88.

Sanchez-Lafuente, A. M. 2007. Corolla herbivory, pollination success and fruit predation in complex flowers: an experimental study with Linaria lilacina (Scrophulariaceae). Annals of Botany 99:355-364.

Schemske, D. W. and L. P. Pautler. 1984. The effects of pollen composition on fitness components in a neotrophical herb. Oecologia 62:31-36.

Schlumpberger, B. O., A. A. Cocucci, M. More, A. N. Sersic, and R. A. Raguso. 2009. Extreme variation in floral characters and its consequences for pollinator attraction among populations of an Andean cactus. Annals of Botany 103:1489-1500.

Sharaf, K. E. and M. V. Price. 2004. Does pollination limit tolerance to browsing in Ipomopsis aggregata? Oecologia 138:396-404.

Sork, V. L. and D. W. Schemske. 1992. Fitness consequences of mixed-donor pollen loads in the annual legume Chamaecrista fasiculata American Journal of Botany 79:508-515.

Stadler, E., J. A. A. Renwick, C. D. Radke, and K. Sachdevgupta. 1995. Tarsal contact chemoreceptor response to glucosinolates and cardenolides mediating oviposition in Pieris rapae. Physiological Entomology 20:175-187.

124

Stanton, M. L., J. K. Bereczky, and H. D. Hasbrouck. 1987. Pollination thoroughness and maternal yield regulation in wild radish, Raphanus raphanistrum (Brassicaceae). Oecologia 74:68-76.

Stephenson, A. G. 1992a. The regulation of maternal investment in plants. Pages 151-171 in C. Marshall and J. Grace, editors. Fruit and seed production: aspects of development, environmental physiology and ecology. Cambridge, Cambridge, U.K.

Stephenson, A. G. 1992b. The regulation of maternal investment in plants. . Pages 151–171 in C.Marshall and J.Grace, editors. Fruit and seed production: aspects of development, environmental physiology and ecology. Cambridge University Press, Cambridge.

Strauss, S. Y. 1997. Floral characters link herbivores, pollinators and plant fitness. Ecology 78:1640-1645.

Strauss, S. Y., R. E. Irwin, and V. M. Lambrix. 2004. Optimal defence theory and flower petal colour predict variation in the secondary chemistry of wild radish. Journal of Ecology 92:132-141.

Strauss, S. Y., H. Sahli, and J. K. Conner. 2005. Toward a more trait-centered approach to diffuse (co)evolution. New Phytologist 165:81-89.

Strauss, S. Y. and J. Whittall. 2006. Non-pollinator agents of selection on floral traits. Pages 120- 138 in L. D. Harder and S. C. H. Barrett, editors. Ecology and Evolution of Flowers. Oxford University Press, Oxford.

Tabachnick, B. G. and L. S. Fidell. 1996a. 9.5.2.2 Roy-Bargmann stepdown analysis. Page 880 Using Multivariate Statistics. HarperCollins Publishers.

Tabachnick, B. G. and L. S. Fidell. 1996b. 14.5.3 Assessing the fit of the model. Page 749 Using Multivariate Statistics. HarperCollins Publishers.

Thibert-Plante, X. and A. P. Hendry. 2011. The consequences of phenotypic plasticity for ecological speciation. Journal of Evolutionary Biology 24:326-342.

Thompson, J. N. 1988. Variation in interspecific interactions. Annual Review of Ecology and Systematics 19:65-87.

Thompson, J. N. 1999. Specific hypotheses on the geographic mosaic of coevolution. American Naturalist 153:S1-S14.

Thompson, J. N. 2009. The Coevolving Web of Life. American Naturalist 173:125-140.

125

Urban, M. C. 2011. The evolution of species interactions across natural landscapes. Ecology Letters 14:723-732.

USDA, N. 2004. The PLANTS Database, Version 3.5, http://plants.usda.gov. National Plant Data Center, Baton Rouge, LA 70874-4490 USA. van Kleunen, M. and M. Fischer. 2007. Progress in the detection of costs of phenotypic plasticity in plants. New Phytologist 176:727-730.

Vazquez, D. P., N. Bluthgen, L. Cagnolo, and N. P. Chacoff. 2009. Uniting pattern and process in plant-animal mutualistic networks: a review. Annals of Botany 103:1445-1457.

Venable, D. L. 1992. Size-number trade-offs and the variation of seed size with plant resource status. American Naturalist 140:287-304.

Vogler, D. W., S. Peretz, and A. G. Stephenson. 1999. Floral plasticity in an iteroparous plant: The interactive effects of genotype, environment, and ontogeny in Campanula rapunculoides (Campanulaceae). American Journal of Botany 86:482-494.

Wang, R. W., B. F. Sun, and Q. Zheng. 2010. Diffusive coevolution and mutualism maintenance mechanisms in a fig-fig wasp system. Ecology 91:1308-1316.

Waser, N. M., L. Chittka, M. V. Price, N. M. Williams, and J. Ollerton. 1996. Generalization in pollination systems, and why it matters. Ecology 77:1043-1060.

Weber, W. A. 1990. Colorado Flora: Eastern Slope. University Press of Colorado.

Wesselingh, R. A. 2007. Pollen limitation meets resource allocation: towards a comprehensive methodology. New Phytologist 174:26-37.

Westoby, M., D. S. Falster, A. T. Moles, P. A. Vesk, and I. J. Wright. 2002. Plant ecological strategies: Some leading dimensions of variation between species. Annual Review of Ecology and Systematics 33:125-159.

Whittall, J. and M. Carlson. 2009. Plant defense: a pre-adaptation for pollinator shifts. New Phytologist 182:5-8.

Williams, I. H. and J. B. Free. 1979. Compensation of oil-seed rape (Brassica napus L.) plants after damage to their buds and pods. Journal of Agricultural Science 92:53- 59.

Willmer, P. 2011. Pollination and Floral Ecology. Pages 105-130. Princeton University Press, Princeton, NJ.

126

Wise, M. J. and W. G. Abrahamson. 2007. Effects of resource availability on tolerance of herbivory: a review and assessment of three opposing models. American Naturalist 169:443-454.

Wise, M. J. and J. J. Cummins. 2006. Strategies of Solanum carolinense for regulating maternal investment in response to foliar and floral herbivory. Journal of Ecology 94:629-636.

Wise, M. J., J. J. Cummins, and C. De Young. 2008. Compensation for floral herbivory in Solanum carolinense: identifying mechanisms of tolerance. Evolutionary Ecology 22:19-37.

Wolfe, L. and S. Mazer. 2005. Patterns of phenotypic plasticity and their fitness consequences in wild radish (Raphanus sativus: Brassicaceae). International Journal of Plant Sciences 166:631-640.

Zheng, W. and W. Liu. 1999. Kinetics and mechanism of the hydrolysis of imidacloprid. Pesticide Science 55:482-485.