Examining host specificity of spp. (Coleoptera: Chrysomelidae) to inform management of invasive dog strangling vine Vincetoxicum rossicum ()

by

Rhoda Bernadette deJonge

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Faculty of Forestr

University of Torontoy

© Copyright by Rhoda B. deJonge 2018 Examining host specificity of Chrysochus spp. (Coleoptera: Chrysomelidae) to inform management of invasive

dog strangling vine Vincetoxicum rossicum (Apocynaceae)

Rhoda Bernadette deJonge

Doctor of Philosophy

Faculty of Forestry University of Toronto 2018

Abstract

Native have the capacity to form novel associations with invasive , which in some cases may allow them to exploit the invaders as hosts. To date, little work has been done to predict these novel associations and their outcomes before they occur. By conducting host- specificity tests on native North American Chrysochus spp. ( C. auratus , C. cobaltinus , and their hybrid) with the European invasive vine, Vincetoxicum rossicum and North American

Apocynaceae plants, this thesis: (1) addresses the question as to whether native Chrysochus spp. will form a novel association with the vine; (2) predicts the outcomes of such an association; (3) suggests ways to improve the potential biotic resistance of these to reduce the spread of this vine; (4) compares host specificity results of these North American beetles with their

European congener C. asclepiadeus , in order to enhance predictions of its ecological host range.

Feeding, survival, oviposition, and development tests with North American Chrysochus at all life stages demonstrate: (1) V. rossicum may act as an ovipositional sink for eastern C. auratus ;

ii (2) the vine could be a source of food for C. cobaltinus when the expands its range westward; and (3) Chrysochus hybridization will not increase use of this invasive vine. In addition, host-specificity tests conducted here show that C. auratus , which does not feed or develop on spp . in its natural ecological host range , will complete development on plants in this genus when exposed under lab conditions. This false positive for host plant specificity suggests that, in this genus, the predicted ecological host range of the European congener (C. asclepiadeus) of the two North American may have been overestimated.

Overall, my findings highlight: (1) the importance of predicting novel associations before they occur in the field; (2) demonstrate how the testing of native relatives of biological control candidates can enhance ecological host-range predictions for classical agents; and (3) increase our knowledge about the potential outcomes of novel associations between introduced V. rossicum and North American Chrysochus spp.

iii Acknowledgments

I learned early on in my graduate career that a PhD thesis is something that will never be finished without the help and support of an excellent supervisor, committee, family, friends and colleagues.

First I want to thank my supervisor, Sandy Smith. She accepted me as a graduate student and has continually supported me financially, intellectually and with helpful insights and advice.

Without her positivity, guidance, and encouragement, the prospect of tackling a PhD thesis would have seemed insurmountable. Acting as a co-supervisor, I also have Rob Bourchier so much to thank. For financial support, helpful critiques, advice in test design, and always finding the practical angle in the research. Both Sandy and Rob’s constant availability and helpful feedback has improved this thesis immeasurably. I also want to thank my committee members,

Peter Kotanen and Marc Cadotte. Both gave up much of their time to offer insightful suggestions for the thesis and had time to meet with me one-on-one to instruct me in areas where I needed more help. In particular, Marc’s class in how to use the statistical software R has been invaluable. I also appreciated the time, comments and critique from my external reviewers Danijela Puric-Mladenovic (Faculty of Forestry) and Steve Murphy (University of

Waterloo).

Staff at the Faculty of Forestry have been incredibly helpful and patient over the years, particularly Tony Ung, Deborah Paes, and Ian Kennedy (minus the patience for this last fella). I want to thank Jay Malcolm and Marie-Josée Fortin for their help with statistics, as well as Terry

Carleton, Andy Kenney, Danijela Puric-Mladenovic, Kathleen Ryan, Sandy Smith, and Sean

iv Thomas for teaching such interesting and helpful classes that have all grounded my research. I also want to thank René Sforza at EBCL- and André Gassman at CABI- for introducing me to this fascinating study system back in 2010. As well, Merrill Peterson, Janis

Dickinson, and Aaron Weed are owed my great thanks for help in assisting in my learning ‘all things Chrysochus’.

I am indebted to all the property owners who allowed myself and research assistants to collect insects and plant material on their land: The City of Toronto’s High Park, Koffler Scientific

Reserve, Royal Botanical Gardens, Riley Wilderness Park, Yosemite National Park, Hastings

Scientific Reserve, La Primavera Farms, as well as the Harrop and Loewith families. This research was funded by the Invasive Species Centre, Agriculture and Agri-Food Canada,

Faculty of Forestry, Ontario Ministry of Natural Resources and Forestry, as well as the Ontario

Graduate Scholarship and Jeanne F. Goulding Fellowship.

There are a number of undergraduate students who have helped with the project in varying capacities over the years: Elpidio Chavez, Atheena Dy, Jeremy Gautherot, Iris Hu, Rosemary

Martin, Felicity Ni, Ariane Pouet, Nina Sokolov, Alex Stepniak, and particularly, Frank

Oukhouia, who did so much to improve the testing methods early on in the project.

I owe mountains of thanks to all my friends and colleagues at the Faculty of Forestry including

Melissa Apostoli, Kira Borden and Anjali Karve for all our great chats at the GSU pub in those early days and for continuing to be good friends (when we actually get to see each other!). My lab and office mates over the years: Sadia Butt, Amy Choi, Eric Davies, Yicheng Du, Alexa

Feldberg, Susan Frye, Justin Gaudon, Nurul Islam, Amany Mansour, Bo Patankar, Paul Piascik,

v Lucas Roscoe, Tara Sackett, Lukas Seehausen, Janani Sivarajah, Tim Skuse, Graham Watt, and in particular Richard Dickinson to whom I owe so much for all the chats over noodles and long days in the field where we solved all the world’s invasive species problems. Without these friends and colleagues, I would have felt completely lost navigating the foreign world that is graduate school.

Last, I want to express my love and gratitude to my family and friends. Many thanks to Joanne

Feddes and Angela Reitsma for all their encouragement over the years. Thanks particularly to

Joanne for her help along with little Margot to collect beetles on hot summer days. Thanks to all my many siblings, in particular Sara, Dave and Jess who spurred me on by making me feel like I was accomplishing this thesis not just for myself, but for them as well. To my in-laws, Henk and

Anna, I owe special thanks for all the much-needed childcare and taking an entire weekend to attend a conference in Montreal with me so I could present and still take care of Willa when she was yet so small. I owe so much gratitude to my amazing parents; first to my mom, who encouraged my love of the outdoors and showed me how to sew malaise traps before I even finished high school. I think she would have gotten a big kick out of this whole ‘PhD thing’.

Thanks to my Dad and Linda for giving up weeks upon weeks to collect and measure insects all over California, and in particular Dad for introducing me to insects in the swamps of

Southwestern Ontario when I was little. We make such a great team. The support from both Dad and Linda has been invaluable over the years including their help with childcare and all the random tasks I’ve assigned.

My greatest thanks I save for my amazing husband Johan. He has given up every summer for the last few years to be my number one field assistant. His love and support for me undertaking

vi this grand adventure has been without equal. Many thanks to my daughter Willa for her help in the field (her tiny stature gives her a great angle to find all the beetles hiding under leaves!) and for her extreme interest in all things larvae. Both Johan and Willa have slogged through hours in the field, the humid greenhouse, and have put up with getting kicked out of the house so I can write in peace. Thank you all.

vii Table of Contents

Acknowledgments ...... iv Table of Contents ...... viii List of Tables ...... ix List of Figures ...... xi List of Appendices ...... xiii Chapter 1: An Introduction to Novel Associations and Biological Control ...... 1 Novel Associations ...... 2 Biological Control for Managing Invasive Plants ...... 9 Thesis Problem Definition and Approach...... 16 Study System ...... 20 Dog strangling vine (DSV) ( Vincetoxicum rossicum) ...... 20 Chrysochus species in North America ...... 24 Thesis Objectives and Overview ...... 27 Chapter 2: Initial Response by a Native , (Coleoptera: Chrysomelidae), to a Novel Introduced Host-plant, Vincetoxicum rossicum (: Apocynaceae) ...... 31 Abstract ...... 31 Introduction ...... 32 Materials and Methods ...... 35 Results ...... 42 Discussion ...... 46 Chapter 3: Western North American Chrysochus Beetles: Weak Candidates for Conservation Biological Control of the Invasive Vine, Vincetoxicum rossicum (Gentianales: Apocynaceae) 55 Abstract ...... 55 Introduction ...... 56 Methods ...... 61 Results ...... 67 Discussion ...... 72 Chapter 4: Host Specificity Testing of North American Chryoshus to Enhance the Ecological Host Range Prediction for European (Col: Chrysomelidae) ...... 83 Abstract ...... 83 Introduction ...... 84 Methods ...... 88 Results ...... 93 Discussion ...... 96 Chapter 5: Discussion ...... 109 Predictions for Novel Associations ...... 110 Recommendations for Biological Control ...... 118 Management Applications of Novel Associations ...... 120 Conclusion ...... 128 References ...... 130

viii List of Tables

Chapter 2:

Table 2.1 Field collection sites for Chrysochus auratus adults in Ontario, Canada and eastern Washington State, USA…………………………………………………………………………50

Chapter 3:

Table 3.1. Collection locations of North American Chrysochus spp. (eastern C. auratus, western C. cobaltinus and their hybrids found in Washington State), their native field host plants, and Asclepias spp. found within this region. To determine host plant acceptance by beetles in this genus, adult beetles collected from these sites in the province of Ontario (ON), Canada, and the USA states of California (CA) and Washington (WA) were tested for feeding on cut leaves of invasive V. rossicum , their native field host plants, and nearby Asclepias spp. (all in Apocynaceae). Beetle collection locations are ordered first by beetle species and then geographically from north to south…………………………………………………………...…76

Chapter 4:

Table 4.1. Survey locations for North American Chrysochus species conducted in 2012 and 2013. Each site was surveyed once. The site boundaries were determined by known ecological host plant presence ( Asclepias spp. and/or spp.). Ecological hosts of C. auratus are limited to Apocynum spp., whereas hosts for C. cobaltinus are both Asclepias spp. and Apocynum spp. Sites were searched thoroughly to count adults and egg masses. Each site was surveyed to record vegetative cover. At sites over 100 m 2, a transect was set lengthwise with five transects evenly spaced perpendicular to the main transect. Four 1 m 2 quadrats were evenly spaced along each perpendicular transect to determine % coverage (up to 100%) of the available host plants at the site ( Asclepias spp. or Apocynum spp.). Sites less than 100 m2 were divided into 1m 2 quadrats and surveyed…………………………………………………………..…....103

Table 4.2. Replicates (no. tested) and mean (±SE) amount of leaf surface removed from feeding (mm2) and % of adults that fed on leaf species by both North American Chrysochus spp. and their hybrid in no-choice tests conducted with cut leaves in petri dishes in the lab 2012-2015. Beetles were collected from sites in Ontario and British Columbia in Canada as well as California and Washington in the USA (Table 1.) Dashes (-) indicate no testing ...... 104

Table 4.3. Results of no-choice larval development tests conducted with Chrysochus asclepiadeus in Switzerland and with C. auratu s in Canada. Recently hatched larvae were placed at the base of screened potted plants. Pots were dissected 85 days after initial larval placement to determine larval survival on plant roots. Dashes (-) indicate no testing……..….105

ix Table 4.4. Summary of fundamental host ranges of North American Chrysochus species and European Chrysochus asclepiadeus compiled from both no-choice and choice tests on leaves and plants and available survey data (deJonge et al. 2017a & 2017b, Gassmann et al. 2009 & 2010, and Sforza 2009, 2011). Dashes (-) refer to lack of use by beetle at this life stage. Blank spaces refer to no testing done. Scores in bold are plant species only in beetles' fundamental host range. Scores in grey are plant species in beetles' ecological host range………………...106

x

List of Figures

Fig. 2.1. Mean (±SE) number of days adult Chrysochus auratus lived when caged on potted plants of either Apocynum androsaemifolium, Ap. cannabinum, Vincetoxicum nigrum or V. rossicum in greenhouse experiments, 2011. Bars with the same letter are not significantly different from each other. Numbers within bars reflect sample size (number of potted plants). Test was stopped one week after the last beetle on Vincetoxicum spp. died, at which time only beetles on Apocynum spp. remained alive...... 52

Fig. 2.2. Mean (±SE) number of Chrysochus auratus egg masses laid by beetle pairs on each plant species in a common garden experiment, 2012. Bars with the same letter are not significantly different from each other. Numbers within bars reflect sample size (number of potted plants) for each species. Although 20 tested pairs were used for each plant treatment, only beetle pairs later confirmed through dissection to be male-female were used for statistical tests…………………………………………………………………………………….………..53

Fig. 2.3. Percentage of Chrysochus auratus larvae found alive each week on excised root segments of either (native host) or Vincetoxicum rossicum (non-native host) in lab experiments, 2015 (means ±95% confidence intervals) (n=245)…………………..54

Chapter 3:

Fig. 3.1. North American adult Chrysochus beetles can be distinguished by colour (left to right): a) Chrysochus auratus : both dorsal and ventral sides exhibit iridescent greenish gold coloration, often with red tones (ovipositing female), b) Chrysochus hybrid (dorsal): lacking iridescence, typically purple, though some exhibit coloration intermediate of parents (Peterson et al. 2001) c) Chrysochus hybrid (ventral): abdominal sterna are typically dull brown, d) Chrysochus cobaltinus : both dorsal and ventral sides exhibit iridescent metallic blue coloration …………………………………………………………………….…………………..77

Fig. 3.2. Mean (±SE) leaf area of Apocynum cannabinum and Vincetoxicum rossicum removed by Chrysochus cobaltinus beetles collected from sites in California and Washington State in 2013. Beetles are grouped into California (4 sites), inside the Hybrid zone in WA (2 sites) and Washington outside of the Hybrid Zone (1 site). Numbers below the bars are the number of beetles that fed versus the number of beetles tested for each site. Sites with the same letter are not significantly different from each other according to a post-hoc Tukey’s HSD test (α = 0.05). Bars are ordered south to north………………………………………………………………….78

Fig. 3.3. Mean (±SE) leaf surface area removed by adult C. cobaltinus beetles fed V. rossicum and Solidago canadensis at Yosemite National Park Site, CA during 2013. Solidago canadensis , a common weed found at this site, and on which C. cobaltinus has been observed) was included to estimate the base level of adult exploratory feeding on a non-host plant by C. cobaltinus

xi beetles. Numbers at the bottom of the bars are the number of beetles that fed versus the number of beetles tested for each site. Bars with the same letter are not significantly different from each other, according to a PERMANOVA (P<0.050)…………………….………………………….79

Fig. 3.4. Mean (± SE) feeding by adult C. auratus , C. cobaltinus and their hybrid on cut leaves of test species. All beetles were collected in the center of the hybrid zone (Mabton, WA) in 2013. Bars with the same letter are not significantly different from each other. Numbers below the bars are the number of beetles that fed versus the number of beetles tested for each site. Bars with the same letter are not significantly different from each other, according to ANOVA (Ap. cannabinum and As. speciosa ) and PERMANOVA ( V. rossicum ) (Bonferonni corrected α = 0.016)…………………………………………………………………………………….…....80

Fig. 3.5. Mean (±SE) feeding on Asclepias speciosa by adult C. cobaltinus collected from outside of C. auratus x C. cobaltinus hybrid zone (Ellensburg, WA) and in the central hybrid zone (Mabton, WA). Bars with the same letter are not significantly different from each other. Numbers at the bottom of the bars are the number of beetles tested for each site (all beetles fed). Bars with the same letter are not significantly different from each other, according to a t-test (P<0.001)…………………………………………………………………………….……….…81

Fig. 3.6. Mean (±SE) dry weight (g) of Apocynum cannabinum potted plant roots following treatment of either 0 (control), 11, or 16 C. cobaltinus larvae per plant. Numbers within the bars are the number of plant roots tested. Bars with the same letter are not significantly different from each other according to ANOVA (Bonferonni corrected α = 0.025)…………………….82

Chapter 4:

Fig. 4.1. Mean (±95% CI) head capsule width of Chrysochus auratus larvae on potted plants of either Apocynum androsaemifolium (n = 15) or Asclepias incarnata (n = 15) in 2012, or on Apocynum cannabinum (n = 12) or Asclepias syriaca (n = 5) in 2015 in the greenhouse. Numbers below bars reflect no. of larvae found on roots of each plant species when pots were dissected 85 days after initial placement. In both 2012 ( t90.358 =2.252, P<0.050) and 2015 (t15.847 =3.9301, P<0.010), head capsules were larger on Apocynum spp. (ecological hosts) than on Asclepias spp. (not ecological hosts of C. auratus )………………………………………..107

xii List of Appendices

Appendix A

Table 1A Chrysochus beetle populations observed but not collected or surveyed due to the small number of beetles present or because a larger population was available nearby (<250m distant). Number of adults reflects number seen on last date the population was observed. Not all sites were visited each year, and therefore only the date of last observance is listed...... 108

xiii Chapter 1: An Introduction to Novel Associations and Biological Control

Introduction

Invasive species are one of the greatest threats to native (Pimentel et al. 2005) reducing the fitness, diversity and abundance of . Invasive plants in particular reduce both abundance and diversity of herbivorous insects (Schirmel et al. 2016). As not all non-native species become invasive upon arrival in novel environments, it is important to understand the possible outcomes and impacts of species introductions. When non-native species first arrive in a new habitat, a number of outcomes can occur: (1) the non-native species does not reproduce or thrive in the new environment, due to abiotic factors (such as temperature or water availability) and/or biotic resistance from native species (Richardson et al. 2000, Kraft et al. 2015); (2) the non-native species may be able to reproduce and grow in the new habitat, yet remain localized

(i.e. non-invasive colonizers or ‘naturalized’ species) (Richardson et al. 2000); or (3) the non- native species may establish, reproduce and spread beyond the area of introduction, becoming invasive in the new habitat (Richardson et al. 2000). Scenario three may occur immediately upon introduction, as with the well-known example of the brown tree snake’s invasion of Guam

(Fritts and Rodda 1998), or after a lag phase or ‘latency period’ (Richardson and Pyšek 2006).

This lag phase can be brief ( i.e. rapid adaption during colonization/invasion phase (Mcevoy et al. 2012)), or last decades to centuries after initial establishment (Crooks 2011). As the number of species introductions continues to increase (Keller et al. 2011), it is important to understand the reasons why some introduced species become invasive, whether immediatly or after a lag- phase, while others remain benign.

1 The biology of the introduced species and the ecology of the native habitat into which it arrives affect whether it will become invasive. Introduced species may carry a number of biological traits that may allow them to successfully establish and invade, including their high genetic diversity (often due to multiple introductions) (Novak and Mack 2005), evolutionary potential

(Whitney and Gabler 2008), and their unique combination of life history traits, which may include high seed output, vigorous vegetative reproduction and low germination requirements

(Baker 1974, Daehler 1998, Sutherland 2004). When looking at how likely a habitat is to be invaded, key factors include the phylogenetic diversity of the habitat (with greater diversity often correlated with more stable environments) (Cadotte et al. 2012) and disturbance level of the region ( i.e. , areas disturbed by human factors more liable for invasion than undisturbed)

(Elton 1958, Alpert et al. 2000). However, even when all these characteristics are met, a species may still not establish or become invasive. Another key factor determining the invasability of an introduced species is the ecology of the area of introduction (with high species diversity, low fragmentation and high levels of environmental stress less liable for invasion (Alpert et al.

2000)) and, specifically, the novel association’s that may form with native species.

Novel Associations

Native insects have the abililty to shift onto non-native plants; acting as predators, pollinators, herbivores, or seed-eaters on the introduced species (Janzen 1985, Carroll 2007). If this occurs without the requriment of a co-evolutionary relationship, circumventing the adaptive process, it is typically referred to as ‘ecological fitting’ (Janzen 1985, Agosta 2006, Agosta and Klemens

2008). When referring to a host shift of a native onto a non-native plant (or vice-versa) without the requirement of ecological fitting, the term ‘novel association’ has been used (Agosta

2 2006, Sunny et. al 2015). This thesis uses the term ‘novel association’ in reference to a native herbivorous insect’s host shift onto an introduced plant species that: (1) forms immediately following initial interaction or after an adaptive period, (2) permits feeding, oviposition and/or development by the native insect on the introduced plant, and (3) persists to the extent that a coevolutionary relationship may occur. Although the majority of interactions between native and introduced species are negative, (and this is particularly true for herbivorous insects

(Schirmel et al. 2016)), novel associations can benefit, be detrimental or have a neutral effect on both the native and introduced species involved. For example, mutualistic novel associations can aid an introduced species in establishing or invading a new habitat (Richardson et al. 2000); i.e. , ingestion of invasive plant seeds by North American birds has been implicated in their spread (Ewel 1986, White and Stiles 1992).

Novel associations can be to the benefit of native species by allowing them to increase their geographic range or fitness. For example, the North American weevil, Euhrychiopsis lecontei

(Dietz) (Coleoptera: Curculionidae) has formed a novel association with invasive Eurasian

Milfoil ( Myriophyllum spicatum L. (Haloragaceae)) in northern lakes (Creed and Sheldon

1995). Throughout most of its range, the native weevil demonstrates higher fitness on the

European milfoil compared to native milfoil (Solarz and Newman 2001) (although in some areas of their range there is no difference in fitness outcomes (Tamayo and Grue 2004)). In another example, Pieris oleracea Harris, a native butterfly that had been nearly extirpated from the state of Massachusetts, was found thriving in high densities on a novel host, cuckoo flower,

Cardamine pratensis [L.] var. pratensis , which prevented them from being parasitized by the introduced Cotesia glomerata (Hymenoptera: Braconidae) (Herlihy et al. 2014).

3 Some novel associations can appear to be without negative or positive impact on the native and non-native species involved. However, often with further study, the novel association is determined to have an indirect effect on either the plant or the insect. For example, one research team determined invasive honeysuckle to have a neutral or positive effect on generalist North

American nesting birds’ nest building and fledgling growth by providing habitat and fruit as an added food source (Gleditsch and Carlo 2011, 2014). However, others have determined songbird nests laid in honeysuckle are predated at a higher rate than those built in native shrubs likely due to increased visability to predators (Rodewald 2012, Rodewald et al. 2010).

Novel associations with invasive species can also result in harm to the native species. By providing resources similar to the native host plant but with a lower degree of fitness

(Schlaepfer et al. 2005), non-native plants can act as oviposition sinks (if the non-native host is less preferred) or evolutionary traps (if the non-native host is more preferred) (Battin 2004,

Sunny et al. 2015). For example, the ( Danaus plexippus (L.) (Lepidoptera:

Nymphalidae) recognizes the invasive vine Vincetoxicum rossicum (Kleopow) Barbar. as an ovipositional site (DiTommaso and Losey 2003, Casagrande and Dacey 2007). Unfortunately, this host recognition is to the detriment of D. plexippus , as its larvae are unable to complete development on V. rossicum , making the novel plant an ‘oviposition sink’ for the butterfly

(DiTommaso and Losey 2003, Casagrande and Dacey 2007). In a similar example, a rare univoltine butterfly, the West Virginia white ( Pieris virginiensis Edwards), oviposits on invasive garlic mustard ( Alliaria petiolata). The larvae of this butterfly species are unable to develop on garlic mustard, and most do not live for more than a few days (Davis and Cipollini

2014). This novel association results in an ‘evolutionary trap’, as the females of this species lay

4 eggs on garlic mustard (as with an ovipositon sink), preferentially to their native hosts (Keeler and Chew 2008).

Novel associations may also cause harm to the introduced species, reducing their growth and spread. Native insects have the capacity to help control the spread of invasive plants (Carroll

2007) by replacing the role played by the plant’s native ‘enemies’ in its home range (sensu

Strauss et al. 2006). This is a key component of the ‘Biotic Resistance’ invasion theory. Biotic

Resistance infers that native species in an introduced species’ new geographic range will be better-suited to their home environment, and therefore can out-compete the non-native, limiting its initial growth and establishment (Elton 1958, Maron and Vila 2001), or assist in limiting the spread of a now-established or invading species (Seabloom et al. 2003). Indeed, some of the most serious disease problems in plantations of introduced trees in the southern hemisphere are caused by native pathogens that have the ability to infect non-native tree hosts (Slippers et al.

2005, Wingfield et al. 2008).

Beneficial ‘biotic resistance’ novel associations may occur immediately or after an adaptive lag period. One example of an association occurring without an extended lag period is the use of the introduced plant Bunias orientalis , a crucifer invasive throughout northern and central , by the native cabbage moth, Mamestra brassicae . The larvae of this species have not undergone any evolutionary change in order to feed on B. orientalis ; however the larvae of this species exhibit greater survival on the introduced plant compared to their native hosts (Harvey et al.

2010). The formation of a novel association after an adaptive lag period is demonstrated in the oft-cited case of two species of Leptocoris (Hemiptera: Rhopalidae) or soapberry bugs. These insects demonstrated rapid evolution of host use by shifting onto introduced sapind hosts

5 (Carroll 2007) to the point of reducing their fitness on their original native hosts (Carroll and

Loye 2012). Such an adaptation in feeding preference on a noval host is often thought to be due to frequency-dependent selection. In similar examples, the selected phenotype does not necessarily have to offer greater growth or survival as long as it produces a greater number of offspring than other phenotypes. As the phenotype increases in the population, so does its fitness. For example, Pieris oleracea Harris (Lep.: Pieridae) has adapted to invasive garlic mustard, Alliaria petiolata (Bieb.) Cavara and Grande (Brassicaceae) (Morton et al. 2015) and

Papilio zelicaon Lucas (Lep.: Papilionidae) has adapted to Ammi visnaga (Apiaceae) (L.)

Lam. (Graves and Shapiro 2003, Strauss et al. 2006), even though in both cases the introduced plants and were initially unable to support complete insect development and provided a lower level of fitness.

The formation of novel associations is not always certain. Native insects use non-native species as hosts less often than native plant species (Keane and Crawley 2002), in part due to the complexity involved in taking the initial steps to form the novel association (Janzen 1985,

Araujo et al. 2015, Sunny et al. 2015). Failure to form a novel association with endemic species can often be detrimental to an introduced species (e.g. lack of pollinators to aid in sexual reproduction), and are often associated with their lack of success in the new habitat. For example, recent research shows the importance of endosymbionts in expanding the host ranges of introduced insects (Sudakaran et al. 2015) and poor associations with specific endosymbionts may limit the invasibility of newly introduced species (Goryacheva et al. 2017). In contrast, an introduced plant may benefit when a novel association fails to develop, and this absence of novel associations with ‘enemies’ in an introduced species’ new geographic range is the basis for one of the key theories regarding species invasions. Termed the ‘Enemy Release Hypothesis

6 (ERH)’, it posits that native enemies in an introduced species’ new habitat may avoid, limit or be unable to feed on an unrecognized or unpalatable introduced species, and therefore fail to cause sufficient damage to limit its spread (Maron and Vila 2001). The introduced species may now be able to invade as it is ‘released’ from enemies and can spread unchecked in its new environment (Keane and Crawley 2002). There is some evidence to suggest that the escape from natural enemies may allow non-native plants to better invest resources into growth instead of defense, thus compounding their success against neighbouring plants (the ‘Evolution of

Increased Competitive Ability (EICA)’) (Blossey and Notzold 1995, Colautti et al. 2004), although other studies (Richardson and Pyšek 2006) and meta-analysis (Felker-Quinn et al.

2013) find little support for this theory. Although there is some evidence demonstrating that lack of enemies is not the mechanism behind the invasion of all species (Agrawal and Kotanen

2003), literature review supports this theory as a key factor in many invasions (Colautti et al.

2004, Jeschke et al. 2012), and it is the key theory behind the use of biological control to limit the spread of invasions.

Predicting whether novel associations will form and whether they may be positive, negative or neutral to the species involved can be difficult for many study systems. However, there are a number of biological characteristics that suggest the likelihood of a species forming a novel association. Native insects that form novel associations with invasive plants are more likely to be generalist herbivores (Bertheau et al. 2010), even though specialist ectophagous herbivores often cause the greatest amount of damage (Lawton and Schroder 1977, Bertheau et al. 2010).

Regardless of their specificity, insects closely related to herbivores of the introduced species in the country of origin (Futuyma and Mitter 1996) and/or herbivores that feed on native plants carrying similar traits or genetics as the introduced plant have an increased likelihood of

7 forming a novel association than those more distantly related (Futuyma and Mitter 1996, Jobin et al. 1996, Agrawal and Kotanen 2003, Dalin and Bjorkman 2006, Pearse et al. 2013). As higher genetic diversity increases adaptive ability, introduced species with high genetic variability (often due to multiple introductions) may express greater ability to form novel associations than those with low genetic variability (Bossdorf et al. 2005). The same may be suggested for native insects, following the general evolutionary rule that greater genetic diversity leads to greater adaptability (Frankham 2005). However, even with all of the knowledge of these traits, it can still be extremely difficult to predict whether a novel association will form (Pearse et al. 2013). Predictive models have been proposed to forecast novel associations (Peterson et al. 2011, Pearse et al. 2013). Some models provide species lists showing species with a higher likelihood for novel associations to form, based on phylogenetic and trait similarities (similar to how the centrifugal phylogenetic method is used as a guideline for determining the host range of biological control candidates (Wapshere 1974, Briese 2005); although they do not suggest the direction of the novel association. Current predictive models are reliant on extensive data sets of previously identified novel associations, which in many cases may be highly incomplete or non-existent. Although the current predictive models can assist in creating species lists for further testing, they are still liable to false positives (Pearse and Altermatt 2013). A false-positive occurs when an insect feeds on a plant in lab/greenhouse tests that it would not otherwise attack in the field (Marohasy 1998). The authors of these predictive models also realize that they do not adequately predict novel associations at finer phylogenetic scales including predicting interactions between specific species (Pearse et al.

2013).

8 The difficulty in predicting novel associations is demonstrated in the extensive host-range testing required to identify host-specific classical biological control candidates (Mason et al.

2017). Further, without knowledge of the direction of the novel association, it is unclear whether any association formed will be to the benefit or detriment of either species. However, outside of host specificity testing for biological control purposes (to be described in greater detail in the next section), there are very few ecological studies investigating the potential for a novel association to form before it is observed in the field (although see Chupp and Battaglia 2014,

Dalosto et al. 2015, Pfammatter et al. 2015). Current methods of predicting novel associations can provide general guidelines but are insufficient at forecasting associations at the species level and do not attempt to suggest the directionality of future associations.

Biological Control for Managing Invasive Plants

Biological control is the use of living organisms to suppress an unwanted pest or weed (Van

Drieshce and Bellows 1996, Hajek 2004). This management approach takes many forms depending on the biology of the pest/weed and the environment in which it has spread. Classical biological control is the deliberate release of a pest species’ host-specific natural enemies with the purpose to have them establish and assist in limiting the spread of the invasive in its new geographic range (Schaffner 2001). While this form of biological control has had widespread success (Howarth 1983, Messing and Wright 2006, Hajek et al. 2016), its use can also be controversial over concerns of unintended impacts to native species (Louda et al. 2003,

Hufbauer and Roderick 2005), some of which will be addressed in this thesis. Native insects can also be used as ‘enemies’ to assist in reducing the spread of invasive species. Termed

‘conservation biological control’, conditions are improved within the surrounding ecosystem to

9 increase the population of native insects that may assist as control agents for pest or weed over the long term (Cullen et al. 2008). This form of biological control is not as common in use as the classical approach for control of invasive plants, likely due to the rarity of positive novel association forming between native enemies and introduced species (Schirmel et al. 2016). Both imported and native species can be used in augmentative biological control, which is purposeful increase in abundance of established biocontrol agents for release in locations containing high densities of the weedy or invasive species (Hajek et al. 2016). Regardless of the approach, biological control is often recommended as the best tool for restoring ecosystems from invasive damage, as specialists insect agents need not be applied continually as with mechanical control

(Schlaepfer et al. 2005) and can prevent the widespread use of chemicals for control (Van

Driesche 2011).

A lack of novel associations can be both a driving factor for and a goal of biological control.

The lack of novel associations formed with native species that reduce the spread of an introduced species can often be the reason why former associations with ‘natural enemies’ of the now-invasive species are sought out to assist in reducing their spread. At the same time, novel associations between imported biological control agent and non-target native species are to be avoided to reduce the risk of harming indigenous species. Therefore, biological control practitioners test candidate agents thoroughly to determine potential novel associations before they can occur in the field.

When considering a natural ‘enemy’ to introduce in order to assist as a biological control agent of an invasive species, biological control researchers conduct thorough host specificity testing to identify candidates that will not form novel associations with native species and put them at

10 undue risk. The goal of host specificity testing is to determine the fundamental host range of the candidate which, when interpreted through the lens of pertinent biological and behavioural information, is used to predict its potential ecological host range, including whether novel associations may form with native species. The fundamental host range includes all the species that support development and sustain feeding by a candidate agent as determined by no-choice lab and field tests (Schaffner 2001). The ecological or ‘realized’ host range includes all the species that support development and sustain feeding by the candidate agent once/if it is released into the target weed’s novel environment (Schaffner 2001). Put another way, the ecological host range is how the fundamental host range is actually expressed in the field

(Nechols 1992). With proper application and interpretation of an insects’ fundamental host range testing results, researchers can identify which candidates may be appropriate host-specific biological control agents.

Host specificity testing of classical agents is often a multi-year, research-intensive process. The process is different depending on whether the target invasive is an insect, fungus, or plant and whether candidate agents are parasitoids, herbivores, etc..The search for potential biological control agents begins in the invasive plant’s new geographic range in order to determine whether any of its ‘enemies’ have traveled to the new range along with it. These surveys can also determine whether indigenous insects have shifted onto the novel host and may be able to assist in conservation biological control. If no damaging enemies are found in the invasive plant’s new geographic range, a search for classical biological control agents in the region of origin is initiated. Qualitative surveys on the target plant and surrounding vegetation in its native environment supplemented by host collection records can indicate whether insects found feeding on the plant are highly host-specific and worth further consideration or instead are

11 generalists that should be dropped from further investigation (Mason et al. 2017). Insects that have multiple generations a year or attack multiple plant parts most important for the growth and spread of the target plant are often chosen above other specialist insects in order to improve efficacy (McClay and Balciunas 2005). The best candidate agents are then subjected to no- choice tests from a thorough plant list approved by governing bodies (Mason et al. 2017,

Sheppard et al. 2005). This list includes plant species in the proposed area of introduction that are closely-related phylogenetically to the target weed (Wapshere 1974b, Briese 2005) or are of special economic or cultural importance (Sheppard et al. 2005, Mason et al. 2017). Conservative

‘no-choice’ tests, where the insect has the option of feeding on the test plant species or starving, delineate the insect’s broadest range of plant acceptance. Choice tests, on the other hand, offer the candidate insect a choice between potential host plants in a shared pot and/or cage, this can either be a single choice between two plant species or a multiple choice between more than two species, often with a number of replicates of each species. No-choice tests help determine the range of hosts biologically accepted by the insect, whereas choice tests determine which of these hosts are preferred, and thereby are at greatest risk for damage (van Klinken 2000, Sheppard et al. 2005). In some cases, field trials may be set up in the region of the insect’s origin. The goal of these trials is to mimic natural processes as much as possible, to obtain a clearer understanding of host specificity (Mason et al. 2017). Once all the above tests have been concluded and the risk of the agent forming novel associations with non-target native plants is deemed acceptably low, a host-specific agent is permitted for release and the final step in host specificity testing occurs. Field-release studies in the country of import are conducted to further confirm ecological host range predictions and efficacy (Sheppard et al. 2005) before the agent is distributed.

12 Each life stage of the candidate insect may have a different host range (van Klinken 2000). For example, in lab tests the adults of a candidate agent, Chrysolina aurichalcea asclepiadis (Villa) (Coleoptera: Chrysomelidae), demonstrated the ability to sustain feeding on

13 host species, whereas the larvae were only able to complete development on six of the tested species (Weed and Casagrande 2011). Knowing which life-stage needs to be host-specific is important (van Klinken 2000). For example, if the larval stage of the insect causes damage, the host specificity of this life stage would be of greater importance over other less or non- damaging life-stages. Alternatively, if adults can harm a non-target plant even through simple exploratory feeding due to being virus vectors (Briese 1989), strict specialization by these adults would be of the utmost importance. Therefore, biological control practitioners will often test all life stages and/or prioritize host specificity testing of an insect’s most damaging life stage(s), in order to ensure efficacy on the target invasive host plant and avoidance of harm to non-targets.

Predicting the ecological host range of a biological control agent requires extensive knowledge of the candidate insect as well as the target species and potential non-target plants. Knowledge of life-history traits, genetics, and behaviours, among other biological factors of both the agent and target species all contribute to better predictions of the ecological host range and efficacy of the biological control candidate (Schaffner 2001). One concrete example where the knowledge of the behaviour of a candidate led to the correct prediction of its ecological host range is in the leaf-feeding beetle, Gratiana boliviana Spaeth, used to control invasive Tropical Soda Apple,

Solanum viarum Dunal, in Florida. During lab tests in Argentina, larvae of G. boliviana wer able to complete development on the important crop eggplant Solanum melongena L. in no- choice tests, even though this had never been observed in similar tests conducted in Florida

(Medal et al. 2002). Field tests were conducted demonstrating no use of the non-target crop

13 (Hinz et al. 2014). The knowledge that, when given a choice between hosts in field conditions

G. boliviana would not attack S. melongena, caused researchers to predict this non-target effect to be highly unlikely if the agent was released (Gandolfo et al. 2007). Following post-release monitoring and a stark reduction in the target plant Solanum viarum, G. boliviana has never been observed feeding on non-target Solanum species, even w ithin close proximity to thousands of acres of S. melongena grown in the region (Diaz et al. 2014, Medal et al. 2008). Therefore, the knowledge of the behaviour of G. boliviana in the field permitted its successful release and accurate prediction of its highly-host-specific ecological host range. A wealth of knowledge regarding the candidate’s biology and behaviour, whether gleaned from choice-tests, observations in the field, or other means, is exceptionally important for an accurate prediction of their ecological host range.

Even with the knowledge that fundamental host use does not always equate to ecological host use, there are cases when a candidate agent has not been permitted for release, due to its perceived risk of forming a novel association with non-target species following inclusion in the candidate agent’s fundamental host range. The results of fundamental host range tests are still weighted too heavily when predicting risk to non-target species (Hinz et al. 2014) and are not, by themselves, a direct predictor of the novel associations that will form between a candidate agent and non-target species. In one example where it appears the fundamental host range was relied on too heavily, the European weevil Ceratapion basicorne , (Illiger) (Coleoptera:

Apionidae), a candidate agent of target weed yellow starthistle (Centaurea solstitialis L.), was found to feed on the important crop plant, safflower ( Carthamus tinctorius L.), during in-lab host specificity tests. This was a surprising result as safflower is also grown in C. basicorne ’s native range and yet the weevil has never been reported feeding on this crop (Cristofaro et al.

14 2013). Three years of field trials in demonstrated that the use of safflower by C. basicorne was merely an effect of the lab and that safflower is not part of the insect’s ecological host range (Smith et al. 1999). However, even with the results of these field trials, C. basicorne was not permitted for release due to concerns over the risk to safflower crops. The use of a non- target plant in an insect’s fundamental host range need not, by itself, be cause to eliminate the candidacy of an otherwise promising agent. For example, initial host specificity testing on two

European chrysomelid species ( Chrysolina hyperici and C. quadrigemina ) revealed they could oviposit and develop on native Hypericum species in New Zealand (Groenteman et al. 2011).

The native plants were deemed to be low risk for attack, and both Chrysolina spp. were released to control St. John’s Wort ( Hypericum perforatum ) (Groenteman et al. 2011). Seventy years after their initial release, the beetles have successfully reduced the invasive plant and have caused no damage to the native non-targets included in their fundamental host range

(Groenteman et al. 2011). Clearly, more information beyond a particular insect’s fundamental host range is needed to assist biological control researchers to accurately predict the risk of potential novel associations that may form following the proposed release of a candidate biological control agent.

Since the already rigorous and thorough process of host specificity testing can still over-estimate risks to non-targets species, further information is needed to more accurately predict whether novel associations will form between candidate agents and non-targets (Messing and Wright

2006). For example, when an herbivorous classical biocontrol agent feeds upon a non-target plant species in host-range tests in the lab, the assumption is that this herbivore-host relationship is due to the insect-specific traits allowing it to demonstrate a broader host range in the lab. This assumption leads to the prediction of a broader ecological host range, which in many cases, may

15 be correct. However, in other cases, the plant, and not the insect, may be the cause of the herbivory observed in the lab. Traits the plant species carries may allow it to broaden its palatability in the lab due to this unique and artificial environment. For example, the fertilization alone of plants in lab experiments can increase their palatability (Van Hezewijk et al. 2008).

The protection from weather stress and natural enemies often experienced in a greenhouse or lab environment can also make plants more palatable to insects and liable for attack (Karban et al.

1997, Bernays and Graham 1998). Given that lab conditions may induce a high level of palatability in contrast to conspecific plants in the field, the use of this host in the lab requires further investigation. More information is needed to determine whether the herbivore, host or both, are responsible for the novel association observed in the fundamental host range.

Thesis Problem Definition and Approach

An approach to solving the issues mentioned in the above sections, which can broadly be referred to as: (1) the need for better predictions of novel associations between native herbivorous insects and invasive plants and their outcomes and (2) the improvement of classical biological control agent non-target risk assessment, can be addressed by conducting host specificity tests with native relatives of classical biological control agents. In order to improve predictions of novel associations, whether for ecological or biological control purposes, host specificity testing can occur in the lab before interactions are observed in the field. As closely- related species often demonstrate similar ecological niches (Freckleton et al. 2002, Wiens and

Graham 2005) and feed on similar species (Futuyma and Mitter 1996), the results of host-use by phylogenetically-similar native insects make ideal comparisons to the host-use exhibited by

16 classical biological control candidates. Although predictive models may not be accurate forecasters of novel associations at a fine scale, they, alongside species identified through the centrifugal phylogenetic method, can be used to create lists of native species with the highest likelihood of forming a novel association with an introduced or invasive plant. As mentioned earlier, native insects that share similar life-history traits and feed on the same host family/tribe/genus as phylogenetically closely-related classical biological agents may be most likely to form novel associations with the invasive species the agent is being tested upon. These native insects could be subjected to the same host specificity tests as the candidate agent, including the target invasive weed and native non-target plants. The results of these tests could provide important information for both ecological and biological control purposes.

As stated, one major aim of this thesis is to improve predictions of novel associations. These novel associations may be important in both an ecological context (e.g. novel associations that form following non-native plant invasions) and in helping to identify host-specific agents for biological control purposes. Although the predictions here may have ecological implications, they aren’t formal predictive models for ecological systems (as laid out recently in Dietz et. al

2017) and are instead focused specifically on forecasting the possibility of specific host- herbivore interactions.

Host specificity testing with native species predicted to form a novel association with an introduced species can provide key baseline data to study the potential evolution of novel associations and has important implications for how the introduced species should be managed.

Specifically, identifying such novel associations before they are observed in the field allows us to: (1) capture information on the association before it is altered through longer-term interaction,

17 creating a baseline from which to compare any future association between these species (Gandhi and Herms 2010); (2) determine whether the novel association occurred due to adaptation or whether it can be classified as an ‘ecological fit’ (Agosta 2006, Harvey et al. 2010): (3) identify the extent of the lag period, if such exists, before the association occurred in the field; and (4) identify species of concern that may be affected directly when an invasive arrives and becomes an ovipositional sink or ecological trap. Dalosto et al. (2015) used this strategy to protect native species at risk by making recommendations for the management and control of an introduced crayfish in . One final benefit of conducting host specificity testing with natives likely to form a novel association with an introduced plant is to add information that could be used to create datasets for better predictive modeling of novel associations. Predictive models would be vastly improved by utilizing the wealth of knowledge gleaned from fundamental and ecological host ranges coupled with key biological traits and phylogenies in order to identify trends and forecast future novel associations. Overall, the identification of potential novel associations through host specificity testing of insects with a higher likelihood to form such an association before they occur in the field can add valuable information on how and which novel associations will form and improve our understanding as to the potential threat to specific native species following invasions.

Host-range testing of native congeners can play a key role in filling in some of the knowledge gaps for estimating risk to non-target plants by candidate agents. The host-range testing of close relatives native to the proposed area of introduction can assist in determining whether in-lab host-use is due to the candidate agent’s ability to feed broadly or the plant’s broad lab-induced- palatability. The results of host specificity testing of native congeners of biological control candidates can help inform whether use of a native non-target species in lab tests will equate to

18 a novel association in the field. For example, if the closely-related insects feed on plant species in lab tests that they do not typically use when in the field, these plants can be identified as

‘highly palatable’ and their use by the candidate agent in the lab should be further investigated

(i.e. potentially a false positive). Alternatively, if closely-related insects to the candidate agent avoid plants in the lab that they typically use in the field, the lack of use of this same plant species by candidate agents in lab tests may be suspect ( i.e. potentially a false negative). Others have suggested the use of species closely related to the candidate agents that are native or naturalized to the area of introduction be used to improve biological control. However, they instead suggest related native species be used as surrogates for efficacy testing (Puliafico et al.

2008). To date, there have been no previous recommendations to use native insects related to agents for purpose of interpreting fundamental host range results.

Another benefit of studying closely related species is the potential for identification of native biocontrol agents that were not found feeding on the invasive species during initial searches for damaging native insects but have adapted to it in the intervening years. If the host specificity testing of natives likely to form a novel association with specific introduced species begins early in the invasion process, native agents may be found in time to reduce widespread invasions. For example, the native weevil Euhrychiopsis lecontei (Dietz), was first observed feeding on

Eurasian watermilfoil in 1991 (Creed and Sheldon 1993). This weevil is currently being used in augmentative biological control across the wide geographic range Eurasian watermilfoil has now invaded (Sheldon and Creed 2003, Winston et al. 2014). If researchers had tested the potential of E. lectontei to host switch onto the invasive watermilfoil early in its invasion process (1950s), its current use in augmentative biological control could have begun much sooner. There may be other native insects that have recently adapted to invasive plants yet only

19 exist in isolated populations and therefore have not been observed using the invader in the field.

By conducting host specificity testing with native insects most likely to form a novel association with the invasive plants, biological control practitioners may identify native insects available to assist in biological control, whether through conservation or augmentative biological control, without waiting to first observe the interaction in the field.

Study System

I apply this approach of conducing host specificity testing on native congeners to the unique system of a European biological candidate, Chrysochus asclepiadeus , of the invasive vine,

Vincetoxicum rossicum and its North American Chrysochus congeners. This system is ideal to determine whether host specificity testing of native congeners can answer the above questions in part due to the unique biological traits these species carry. All insects in this system are specialists within the same family. One North American relative is monophagous and genetically uniform, whereas the other is more genetically heterogeneous and feeds broadly within the family. In addition, the two North American species hybridize, making an interesting study as to the effect of hybridization and impact of gene introgression on novel host use.

Dog strangling vine (DSV) ( Vincetoxicum rossicum)

Vincetoxicum rossicum (Kleopow) Barbar. (Apocynaceae syn. rossicum (Kleopow)

Borhidi), commonly known as pale swallow-wort or dog-strangling vine (DSV), is a perennial herbaceous vine and a successful invader of terrestrial habitats throughout eastern North

20 America. The first records of V. rossicum in North America date from the late 1800 s, although it was not until 1973 that it was first described as ‘weedy’ in Ontario, Canada (Pringle 1973,

Sheeley and Raynal 1996). Field abundance and distribution pattersn suggest greater dispersal and competitive ability when compared with other grasses and herbaceous plants in their new geographic range (DiTommaso, et al. 2005). The growth and reproductive strategies of V. rossicum may be responsible in part for its ability to invade. The vine has substantial ability to spread via seed, with over 54,000 seedlings produced per m 2 annually when in full sun (Smith et al. 2006). Perennating buds sprouting from roots produce additional stems following clipping, mowing or burning without the requirement of sexual reproduction (DiTommaso et al. 2005).

Recent work demonstrates that the allelochemical antofine produced by V. rossicum

(Cappuccino and Arnason 2006) harms soil microbes that are otherwise symbionts with native plants, further increasing its competitive ability (R. Dickinson, unpublished). Dense stands can smother surrounding vegetation (Douglass et al. 2009). Thick stands of dog strangling vine support a lower number of , (particularly with the leaf-chewing guild with Asclepias syriaca demonstrating a five-fold increase in chewers), which in turn may impact insectivorous birds and mammals by reducing food availability from insect sources (Ernst and Cappucino

2005). Vincetoxicum rossicum is invasive in open pastures, no-till fields, tree plantations and forest understories throughout eastern North America (Sheeley and Raynal 1996, Lawlor 2000,

DiTommaso et al. 2005, Weston et al. 2005, Milbrath 2010). Of particular concern are rare plants including Jesup’s milkvetch (Astragalus robbinsii ) and Hart’s tongue fern ( Phyllitis scolopendrium ), that can be displaced by V. rossicum ’s invasion of rare alvar communities

(Lawlor 2000, DiTommaso et al. 2005).

21 Conventional management strategies, including chemical and mechanical means, have not been effective in controlling dog-strangling vine (Lawlor and Raynal 2002, Averill et al. 2010,

Douglass et al. 2011). Biological control is seen by some as the best long-term strategy for controlling this vine due to the expensive and labour-intensive alternatives (DiTomassso 2005,

Averill et al. 2008, Milbrath 2010). Initial surveys in North America revealed few insect herbivores have been found feeding on this plant and, of those that do, none cause sufficient damage to limit its growth or spread (Sheeley and Raynal 1996, Lawlor 2000, Ernst and

Cappuccino 2005, Milbrath 2010, Milbrath and Biazzo 2012). The lack of observed novel associations with native insects may be due in part to V. rossicum ’s plant defenses. A member of the well-defended family, Apocynaceae, this plant contains a number of secondary toxic plant compounds (Weston et al. 2005, Douglass et al. 2011) as well as latex (Liede 1996). No native

Vincetoxicum species are known from North America (Tewksbury et al. 2002), thus removing the possibility that natural enemies can move from native plants in the same genus.

Surveys for an appropriate biological control agent for dog-strangling vine began overseas in

2006 (Gassmann et. al. 2011). The leaf-feeding moth, Hypena opulenta (Christoph)

(Lepidoptera: Erebidae), was identified as one potential agent for V. rossicum (Weed and

Casgrande 2010). Following host specificity testing of H. opulenta, the insect was approved for release in Canada in 2013 and is now established at least one location (R. S. Bourchier, personal communication). Lab studies have shown the potential for H. opulenta to successfully reduce above-ground biomass of the vine and seed set, with as few as two larvae per V. rossicum plant

(Weed and Casagrande 2010, Milbrath and Biazzo 2016). To date H. opulenta is still at low densities in the field and is not causing visible biomass reductions (R. S. Bourchier, personal communication). Based on previous lab studies, the expected biomass reductions rarely kill the

22 plants (requiring over 50% defoliation to occur multiple subsequent years) (Weed and

Casagrande 2010, Milbrath et al. 2016), and thus identifying additional biocontrol agents with the potential to damage roots of V. rossicum is desirable.

Concurrent studies revealed Chrysochus asclepiadeus (Pallas) (Coleoptera: Chrysomelidae; syn.

Atymius asclepiadeus Gistel, syn. asclepiadeus Illiger), a leaf-feeding beetle found throughout southern Europe (Jolivet and Verma 2008, Schmitt 2011), to be the most effective herbivore on V. rossicum in lab and common garden tests, with as few as twenty larvae able to reuce the entire biomass of a plant (Weed et al. 2011a, Weed, et al. 2011b). Female beetles of this species oviposit egg clutches held in place with a ring of frass on host plants, typically at the base of the plants stems (personal observation). Adults feed on the leaves, and larvae feed, develop and overwinter on plant roots (Weed 2010). Although the beetle’s ecological host-range in Europe appears limited to V. hirundinaria , a more wide-spread species in this genus, researchers at both CABI (Centre for Agricultural Bioscience International) and EBCL

(European Biological Control Laboratory) have been able to easily demonstrate that the fundamental host-range of this beetle includes the invasive relatives ( V. rossicum and V. nigrum ), often performing better on these plant species (Weed et al. 2011b). Even though it was the most destructive feeder on roots of V. rossicum , C. asclepiadeus ’ candidacy was terminated out of concern for non-target impacts to native North American species. Specifically, in no- choice larvae tests, C. asclepiadeus were able to complete development on the following North

American plants in the Apocynaceae: Asclepias fascicularis Decne. , Asclepias incarnata L.,

Asclepias speciosa Torr . Asclepias syriaca L. Asclepias tuberosa L. (Gassmann et al. 2010,

Sforza 2011), and Apocynum cannabinum L. (all Apocynaceae) as well as Cephalonthus occidentalis L. (Rubiaceae) (Gassmann et al. 2011). Larvae were also able to develop to the 3 rd

23 instar on Asclepias viridiflora Raf. and Cynanchum angustifolium Pers. (both Apocynaceae)

(Gassmann et al. 2010). Adults were able to sustain feeding on As. tuberosa in no-choice tests, surviving over two weeks, although little feeding was observed on the species in the presence of

Vincetoxicum spp. during choice tests (Sforza 2011). Minor feeding by adults was also observed on non-target species Cynanchum laeve (Michx.) Pers. (Apocynaceae) and As. syriaca during multiple choice tests (Gassmann et al. 2010, Sforza 2011). With this beetle’s candidacy terminated, research continues to identify root-feeding insects capable of assisting in the control of V. rossicum .

Chrysochus species in North America

In North America, there are two native congeners of Chrysochus asclepiadeus that are also known specialists on plants within the Apocynaceae. Both North American Chrysochus demonstrate similar life-cycles (Weiss and West 1921, Arnett 1968, Dobler and Farrell 1999,

Peterson et al. 2001, Jolivet and Verma 2008) and share a similar form of self-defense using chemical secretions with their European relative. (Dobler et al. 1998). North American

Chrysochus adults are easy to observe in the field due to their low mobility, iridescent colouration and aposematic behaviour. Female beetles lay frass-covered egg masses on the underside of leaves, along stems and on nearby vegetation (personal observation, Dickinson

1995, Weiss and West 1921). Larvae hatch from their eggs, chew an exit hole through the frass, then drop to the ground to feed on the roots, and overwinter as pupae or pre-pupae in the soil

(Weiss and West 1921).

24 Chrysochus auratus Fabricius (Coleoptera: Chrysomelidae) is found throughout eastern North

America, with populations stretching as far west as southern British Columbia through eastern

Washington State, Oregon (Hatch 1953) and down to Utah and Arizona (Dobler and Farrell

1999). This greenish-gold iridescent beetle, commonly called the dogbane beetle, has been observed to feed exclusively on Apocynum spp. (Apocynaceae) in the field (Arnett 1968,

Doussourd and Eisner 1987, Williams 1991, Dobler and Farrell 1999). There is errant mention of Asclepias spp. use in some of the earliest descriptions of C. auratus (Weiss and West 1921), although more recent publications describe C. auratus as specific to Apocynum spp. (Arnett

1968, Doussourd and Eisner 1987, Williams 1992, Dobler and Farrell 1999). Currently, C. auratus occupies many of the same habitats (Williams 1992, Dobler and Farrell 1999) where dog strangling vine exists (Ditommaso et al. 2005, Sheeley and Raynal 1996), but the beetle has not been observed directly interacting with the plant.

Chrysochus cobaltinus LeConte, commonly called the cobalt beetle, is a blue-iridescent beetle found in western North America specializing on plants in both Apocynum and Asclepias genera

(Apocynaceae: Asclepiadeae) (Dobler and Farrell 1999, deJonge et al. 2017). The geographic range of this beetle does not extend further east than western Montana and Utah (Labeyrie and

Dobler 2004). Previous studies have identified that the beetle is genetically heterogeneous throughout its range and suggest that individual C. cobaltinus populations may be specialized to their locally-available host plants (Dobler and Farrell 1999). The two North American

Chrysochus appear to have speciated from a common ancestor during the last glaciation in the

Rocky Mountain region (Dobler and Farrell 1999).

25 The two North American representatives of this genus hybridize in at least one region: a 25-km section of the Yakima River Valley in eastern Washington State (Peterson et al. 2001, Peterson et al. 2005a). The hybrids are typically dull purple with brown ventral regions (Peterson et al.

2001) and are nearly infertile making it very rare to find any F 2 offspring (Peterson et al.

2005b). Little is known about the host use of the Chrysochus hybrid, except that it has been collected from Apocynum cannabinum in the field (Peterson et al. 2001).

The North American Chrysochus beetles make an excellent study system for investigating novel associations. The likelihood of a novel association forming between North American

Chrysochus and V. rossicum is high due to factors including: (1) the beetles are specialists on plants similar to V. rossicum , all of which are in the same plant family (Apocynaceae), and include the Asclepias genus which is in the same tribe (Asclepiadeae) as Vincetoxicum ; (2) the

North American beetles are able to feed on a variety of toxic compounds found in Vincetoxicum spp. and share similar detoxifying enzymes (Labeyrie and Dobler 2004) and defense strategies as Vincetoxicum -specialists (Dobler et al. 1998); (3) the shared phylogeny with V. rossicum ’s most damaging European herbivore, C. asclepiadeus ; (4) Chrysochus auratus ’ shared temporal and geographical environment with V. rossicum ; (5) the high genetic heterogeneity and broad host use within Apocynaceae demonstrated by C. cobaltinus ; and (6) the hybridization of the beetles, which in some cases can lead to an increase in host range (Schwarz et al. 2005). With such a confluence of biological and ecological factors, the prospect of a novel association forming between these insects and the invasive vine is high and therefore worthy of study. In addition, although a novel association between Chrysochus spp. and V. rossicum could be predicted, whether such an association would result in a positive or negative outcome for the beetles cannot be forecasted without further testing. Only by conducting host specificity tests

26 with native beetles (including tests of longevity, oviposition and larval development), can we more accurately predict the direction of any future novel associations.

The study of the North American Chrysochus spp. can also assist in better interpretation of the fundamental host range of the European congener and candidate agent C. asclepiadeus . By conducting host range tests on native Chrysochus spp ., we may be able to draw conclusions about the predictive ability of C. asclepiadeus’ demonstrated fundamental host range by studying how the North American Chrysochus spp.’s fundamental host ranges translate into their confirmed ecological host ranges. This Chrysochus -Apocynaceae system gives us a unique case study from which to observe the potential for novel associations before they occur in the field, thus allowing us to predict the likelihood and directionality of a novel association with the invasive vine while also improving our interpretation of European Chrysochus host range testing results.

Thesis Objectives and Overview

In this introductory chapter, I have outlined the importance of novel associations to the establishment success, invasive potential, and possibility of future control of introduced species.

As the prediction of novel associations and their potential effects informs conservation efforts, their study is one of great importance. Current predictive models are general in nature and are unable to forecast directionality of future associations. Even though predicting future novel associations and their effects can be extremely difficult, it is ecologically important to do so.

However, outside of the practical purposes of host specificity testing for biological control candidates, the prediction of future potential novel associations at a fine scale is rarely done.

27 Host specificity testing of natives related to classical biological control agents can improve risk assessment to native non-target plants and potentially identifying native insects to use in conservation or augmentative biological control. To date, the results of host specificity testing on native agents has not been used to enhance the ecological host range predictions of classical biological control candidates, even though this concept was suggested in general terms by

Driesche et al. (2004), and similar tests have been used for the express purpose of efficacy testing of classical agents (Puliafico et al. 2008).

The broad goal of this thesis is to inform management aimed at reducing the negative impacts to native species caused by invasive weeds such as V. rossicum . The objective is to use the unique approach of host specificity testing on native insects in order to predict future novel associations between V. rossicum and North American Chrysochus while also improving the interpretation of fundamental host range results of the classical biological control agent C. asclepiadeus . If novel associations between native insects and invasive plants can be identified early within the invasion process, we may be able to increase the biotic resistance and undergo informed management to reduce the spread and harm of an invasive plant. Additionally, current host specificity testing protocols are rigorous and conservative to future ecological risks. However, more information is needed to accurately predict the ecological host range of classical agents to avoid over-emphasizing non-target risks. Without understanding whether a plant is liable to creat false positives in a lab environment, it can be unclear whether a species an insect uses in lab tests will form a novel association with this same species in its ecological host range. The case study system examined here can be used for future work where the prediction of novel associations or the need for additional information to inform risk assessment of novel associations occurring with biological control agents is high. Each chapter has been written as a

28 self-contained manuscript and therefore some overlap, albeit minor, does occur especially in the introduction sections.

Chapter Two answers questions as to whether a novel association has formed between the eastern North American beetle, C. auratus and V. rossicum during its larval or adult stages. The factors of age, sex, and exposure to V. rossicum in both the lab and field are examined to determine the potential for adult feeding on this novel host. This chapter also answers the question as to whether a future novel association with C. auratus and V. rossicum will be positive, negative or neutral for the native beetle. This manuscript has been published in the journal Environmental Entomology (deJonge et al. 2017).

Chapter Three answers the question as to whether a novel association could form between V. rossicum and the western C. cobaltinus or the North American Chrysochus hybrid. It investigates whether the broader-feeding and genetically heterogeneous western Chrysochus beetles can feed, survive, and complete development on V. rossicum . As the geographic ranges of the vine and insects do not currently intersect, any novel association observed in the lab would be an example of an ecological fit. The effect of hybridization and gene introgression on host use is also explored. Land managers can respond appropriately in preparation for the spread of V. rossicum in western North American dependent on the potential impact of novel associations predicted in this chapter. This manuscript has been formatted for the journal

Ecological Entomology.

Chapter Four answers the question as to whether the testing of native congeners of classical agents can enhance proper interpretation of host range test results. This chapter shows how the

29 results of fundamental host range testing of North American Chrysochus translate into their ecological host ranges. It then compares the fundamental host ranges of North American and

European Chrysochus in order to better predict the ecological host range of the latter. This chapter discusses how the knowledge gained from host range testing native congeners of classical agents can be applied when making forest management decisions and to enhance risk assessment in biological control. This manuscript has been formatted for submission to the journal Biological Control.

The final chapter of this thesis discusses the overall findings of the previous chapters. It explores the likelihood of Chrysochus spp. adapting to V. rossicum and suggests novel approaches to encourage biotic resistance. Recommendations include: (1) field manipulations to assist in changing the native and invasive species’ adaptability; (2) methods to accelerate adaptation by selective breeding; and (3) an investigation into the emerging use of endosymbiont microbiota as a management tool to increase the likelihood of novel association formation. The chapter also makes application recommendations to other systems.

30 Chapter 2: Initial Response by a Native Beetle, Chrysochus auratus (Coleoptera: Chrysomelidae), to a Novel Introduced Host-plant, Vincetoxicum rossicum (Gentianales: Apocynaceae) 1

Abstract

Native insects can form novel associations with introduced invasive plants and use them as a food source. The recent introduction into eastern North America of a non-native European vine,

Vincetoxicum rossicum (Kleopow) Barbar., allows us to examine the initial response of a native chrysomelid beetle, Chrysochus auratus F., that feeds on native plants in the same family as V. rossicum (Apocynaceae). We tested C. auratus on V. rossicum and closely related or co- occuring native plants ( Apocynum spp., Asclepias spp., and Solidago canadensis L.) using all life stages of the beetle in lab, garden, and field experiments. Experiments measured feeding

(presence/absence and amount), survival, oviposition, and whether previous exposure to V. rossicum in the lab or field affected adult beetle feeding. Beetles fed significantly less on V. rossicum than on native Apocynum hosts. Adult beetles engaged in exploratory feeding on leaves of V. rossicum and survived up to 10 days. Females oviposited on V. rossicum, eggs hatched, and larvae fed initially on the roots, however no larvae survived beyond second instar.

Beetles collected from Apocynum cannabinum L. field sites intermixed with V. rossicum were less likely to feed on this novel non-native host than those collected from colonies further from and less likely to be exposed to V. rossicum (>5km). Our experimental work indicates that V.

1 This chapter has been published in the journal Environmental Entomology: deJonge, R. B., Bourchier, R. S., and Smith, S. M. (2017) Initial Response by a Native Beetle, Chrysochus auratus (Coleoptera: Chrysomelidae), to a Novel Introduced Host-Plant, Vincetoxicum rossicum (Gentianales:Apocynaceae) 46:617-625 doi: 10.1093/ee/nvx072

31 rossicum may act as an oviposition sink for C. auratus and that this native beetle has not adapted to survive on this recently introduced novel host plant.

Introduction

The invasion of new habitats by non-native plants can lead to dramatic ecological changes including altered nutrient cycling (Vitousek et al. 1997, Simberloff 2011), modification of trophic interactions (Norbury et al. 2013), facilitation of secondary invasions (Simberloff and

Holle 1999, Richardson et al. 2000), and degradation of ecosystem function (Simberloff and

Holle 1999). One mechanism that accounts for these ecological impacts is enemy release (Keane and Crawley 2002), whereby an invasive plant escapes the natural enemies from its source habitat and native herbivorous insects avoid feeding on the unrecognized or unpalatable host plants, thus enabling them to spread unchecked in their new habitat. Such escape however may be temporary as accumulating evidence shows native herbivores will move onto these introduced species and form what is termed a new or novel association (Janzen 1985, Agosta

2006, Carroll 2007, Sunny et al. 2015), often after an adaptive lag period (Janzen 1985). Once formed, these novel associations may serve to regulate the introduced species and lower its populations so that long-term disruption in the new habitat is minimized (Schlaepfer et al. 2005,

Agosta 2006).

Regardless of the mechanism, a number of factors influence whether a novel association will form after an introduced species arrives. Insects that feed on plants that are closely related phylogenetically are more likely to include the novel host in their diet than those that feed on distantly related plants (Futuyma and Mitter 1996, Agrawal and Kotanen 2003, Bertheau et al.

32 2010, Pearse et al. 2013). Closely-related species demonstrate similar ecological niches

(Freckleton et al. 2002, Wiens and Graham 2005); therefore, native insect herbivores in the same genus as the natural enemies of a non-native plant may be best-suited to feed on the novel host. Typically, the length of time insects are exposed to a novel plant (whether in the lab or field) increases their likelihood of using it as a host (Dethier 1982, Papaj and Prokopy 1989,

Lankau et al. 2004, Santana and Zucoloto 2011). Thus, when investigating the potential of a native herbivorous insect to form a novel association with a recently introduced plant, factors such as the phylogenetic relationship between the introduced (potential) host plant and resident native (current) host plants, the insect’s relatedness to natural enemies of the non-native plant in its home-range, and the length of time the native insect has been exposed to the non-native plant should all be considered.

The recognition of a non-native plant as a host can in some cases be to the detriment of a native herbivore’s fitness. By providing resources similar to the native host plant but with a lower degree of fitness (Schlaepfer et al. 2005), non-native plants can act as oviposition sinks (if the host is less preferred) or evolutionary traps (if the host is more preferred) (Battin 2004, Sunny et al. 2015). Vincetoxicum rossicum (Kleopow) Barbar. (Apocynaceae; syn. Cynanchum rossicum

(Kleopow) Borhidi), commonly known as pale swallow-wort or dog-strangling vine (DSV), is a recently introduced perennial herbaceous vine and a successful invader of terrestrial habitats throughout eastern North America. The first records of V. rossicum in North America date from the late 1800’s, although it was not until 1973 that it was first described as ‘weedy’ in Ontario,

Canada (Pringle 1973, Sheeley and Raynal 1996). To date, few insect herbivores have been found to feed on this plant, and of those that do, none cause sufficient damage to limit its growth or spread (Sheeley and Raynal 1996, Lawlor 2000, Ernst and Cappuccino 2005, Milbrath 2010,

33 Milbrath and Biazzo 2012). As a member of the well-defended family, Apocynaceae, this plant contains latex (Liede 1996) and a number of secondary toxic plant compounds (Weston et al.

2005, Douglass et al. 2011). No native Vincetoxicum species are known from North America

(Tewksbury et al. 2002), thus removing the possibility that natural enemies can move from native plants in the same genus. However, insects from other native plants within the same family have responded to Vincetoxicum ; i.e. the monarch butterfly ( Danaus plexippus (L.)

(Lepidoptera: Nymphalidae) recognizes V. rossicum as an oviposition site (DiTommaso and

Losey 2003, Casagrande and Dacey 2007). Unfortunately, this host recognition is to the detriment of D. plexippus , as its larvae are unable to complete development on V. rossicum , making the novel plant an oviposition sink for the butterfly (DiTommaso and Losey 2003,

Casagrande and Dacey 2007). Other native insect Apocynaceae-specialists may similarly recognize non-native V. rossicum as a host, but be unable to complete development on it.

Chrysochus auratus Fabricius (Coleoptera: Chrysomelidae) is a phytophagous insect native to eastern North America found feeding only on Apocynum androsaemifolium L. and Apocynum cannabinum L. , native plants in the family Apocynaceae (Weiss and West 1921, Williams 1992,

Dobler and Farrell 1999). Adults feed on the leaves and female beetles lay eggs on leaves and stems (Hatch 1953, Arnett 1968, Dobler and Farrell 1999), which hatch and larvae drop to the ground to feed on the roots. The insects overwinter as pupae or pre-pupae in the soil (Weiss and

West 1921). We hypothesized that this species has the potential to form a novel host-association with V. rossicum in North America because: a) it is specialized to feed on well-defended lactiferous plants within the same family (Weiss and West 1921, Williams 1992, Dobler and

Farrell 1999); b) it is closely related phylogenetically to Chrysochus asclepiadeus (Pallas), the

34 vine’s most-damaging herbivore in its original habitat in Europe (Weed et al. 2011a); and c) it is found in similar habitats as introduced V. rossicum .

Here we examined the novel interaction of a native insect herbivore, C. auratus, on the introduced vine V. rossicum to assess whether this beetle has the potential to become a biological control agent for this recent invader. Specifically, we measured the response variables of adult feeding, survivorship, and ovipositional preference and success of larval hatching and development of C. auratus on V. rossicum compared with that on its native Apocynum spp. hosts. Key variables of beetle age, sex, and prior exposure to V. rossicum (manipulated in the lab) were used as covariates for the response variables measured. We also assessed the importance of pre-exposure to V. rossicum for acceptance of the novel host by comparing beetle populations from Ontario, Canada (where V. rossicum is present) with colonies from both

Ontario and Washington State, USA where native Chrysochus had never encountered this introduced plant .

Materials and Methods

Beetle and plant collection

Adult C. auratus were collected in the field from Apocynum androsaemifolium or Apocynum cannabinum in southwestern Ontario and eastern Washington (Table 2.1 for collection date and locations). Upon collection, beetles were placed in 4-L (12cm x 12cm x 26cm) clear plastic containers with mesh lids and provided with foliage collected from the same sites. Beetles were held under these conditions at 16:8 L:D and ambient temperature (20-22°C) until tested in the lab, greenhouse, and/or garden.

35 Adult feeding

Experiments to assess the effect of age, sex, and exposure to V. rossicum on the feeding of adult beetles were conducted while controlling for temperature, collection site, and date. Regular small field collections were used for these experiments because C. auratus could not be mass- reared. For testing in the lab, adult beetles were placed singly in 11-cm petri dishes with a moist filter paper and a leaf from a single test-plant species per petri dish. Plants were collected as young plants in the field from the Royal Botanical Gardens in Hamilton, ON and grown in a greenhouse in 4L pots with potting soil (Sungro ® Sunshine Mix #1). Each leaf used for testing was selected from the middle stratum of the plant to maintain consistency in leaf size and scanned prior to placement using ImageJ software v.1.47 (Bethesda, Maryland) to determine the surface area (mm 2). Petri dishes were placed on the test bench so as to ensure no dish with the same treatment was adjacent to another and to account for possible variation in the lab environment. All dishes were kept at ambient lab conditions (20-22°C) with a photoperiod of

16:8 L:D. To reduce error due to surface area reduction from water loss, leaves were removed after two days and feeding galleries were traced with initial scan overlaid to record the amount of leaf surface area (mm 2) removed. All leaves were reviewed visually to confirm presence or absence of feeding. All adult feeding tests described below followed this protocol.

Age and sex – As age and sex have been shown to influence adult beetle feeding (Jaenike 1990), we compared the adult feeding results from beetles that had recently emerged from the soil

(within 24 hours) and had not yet initiated feeding with older beetles collected at least one week later in the field that were actively feeding. To obtain recently-emerged C. auratus beetles, sites were surveyed daily from the beginning of June (before the beetles have emerged) through to mid-June 2012. Sites were searched thoroughly in order to collect all newly-emerged beetles

36 present on Apocynum spp. in each patch at each date. No evidence of adult C. auratus feeding was observed on Apocynum plants during these collections. All 60 recently-emerged beetles were collected from four surveyed sites (Dundas, Copetown, Guelph, and Toronto (Table 2.1)) between 16 and 23 June 2012. Adults were distributed evenly between petri dishes containing leaves of Ap. androsaemifolium and V. rossicum (30 beetles/plant species). These same sites were visited at least one week later to collect 120 older beetles, the last of which were collected on 8 July 2012. Older beetles were also distributed evenly on Ap. androsaemifolium and V. rossicum . Older beetles are identified as those that had already initiated feeding on Apocynum spp. whereas in contrast, younger, newly emerged beetles observed in the greenhouse and the field remain relatively stationary and do not initiate feeding, mating or oviposition immediately

(i.e., naïve or with very little exposure to Apocynum spp.) (personal observations, R. deJonge).

To determine whether sex has an effect on feeding presence/absence with V. rossicum and amount on native hosts, Apocynum spp., we compared the adult feeding test results of 33 females and 47 males (80 beetles total) collected from Ap. cannabinum in Richland, WA in

2013 (Table 2.1). Forty beetles (15 female and 25 male) were fed leaves of V. rossicum , the remaining 40 (18 female and 22 males) were fed Ap. cannabinum as a control. All beetles were measured from the front of the head to the tip of the elytron to the nearest 0.01mm using electronic calipers. Beetles were sexed following the procedure by dissection. For all other tests, beetles were sexed using external morphology.

Lab Exposure - To determine whether lab exposure to V. rossicum increased adult feeding, the presence/absence and extent of feeding were examined by comparing beetles exposed to the vine with those that had not been exposed over a 2-day period. In June 2013, 70 beetles were collected from Ap. androsaemifolium plants at the Copetown site and tested using the standard

37 adult feeding procedure detailed above with leaves from the following leaf species: Apocynum androsaemifolium, Ap. cannabinum, Asclepias eriocarpa Benth. , Asclepias fascicularis Dcne. ,

Asclepias speciosa Torr. , Solidago canadensis L. and V. rossicum (10 beetles per plant species).

In the event we did not observe feeding on V. rossicum , we added Asclepias spp. to determine whether this result was due to host fidelity to Apocynum spp., or whether V. rossicum leaves in particular are unpalatable to C. auratus adults. In the event we observed a high amount of feeding on V. rossicum, we added S. canadensis (an unrelated common plant often found in the field with Apocynum spp. where C. auratus is present) to determine whether this result was due to a lack of phylogenetic specificity in exploratory feeding demonstrated by C. auratus adults.

Immediately after the initial procedure was completed, the same beetles were subjected to the identical procedure, with each beetle receiving the same leaf species as they had two days before. The results of the two procedures were compared to each other to determine whether mean feeding had changed for beetles following a 2-day exposure to the same leaf species.

Field Exposure - To determine whether exposure to V. rossicum (as measured by proximity to

V. rossicum in the field) affected feeding, beetles were collected from Apocynum spp. at sites at varying distances from the introduced vine in both Ontario, Canada and Washington State,

USA. Specifically, we collected beetles from sites intermixed with V. rossicum (i.e. the

‘exposed group’) (from Toronto, ON (17) and Dundas, ON (49)), and those without V. rossicum nearby ( i.e. the ‘unexposed group’): (Copetown, ON (13), Guelph, ON (22)), which are both

5km distant from V. rossicum , and (Mabton, WA (25) and Richland, WA (40)), which are both over 500km distant from V. rossicum (Table 2.1). Distance to closest V. rossicum was determined using an invasive plant database (EDDMapS 2016), and personal observation

(R.deJonge). Leaves of native hosts Ap. cannabinum were tested concurrently as a control

38 (Toronto (19), Dundas (59), Copetown (12), Guelph (16) Mabton (27) and Richland (42)). In all cases, beetles were returned to the lab for testing using the standard adult feeding protocol with cut leaves in petri dishes.

Statistical analysis - G-tests were used to compare the presence/absence of feeding between; a) recently-emerged and older adult beetles, b) males and females, and c) different collection sites/distances from V. rossicum in the field. The G-test was used rather than the more common

χ2 test as it is based on a multinomial distribution and is robust to smaller sample sizes (Gotelli

2004). McNemar’s test was used to compare feeding presence/absence on all leaf species pooled by genera before and after 2-day exposure to plant leaves.

ANOVAs were used with Tukey’s HSD (Tukey 1953, Kramer 1956) when comparing the amount of feeding by; a) recently-emerged and older adult beetles and b) between the collection sites. To compare feeding amounts by male/female beetles and between feeding on Ap. cannabinum and V. rossicum, t-tests were conducted. All statistical analyses in this study were carried out using R software, version 3.2.2 (R Core Team 2015) except for the G-tests, which were calculated using the G-test calculator (McDonald 2014).

Adult survival, oviposition, and egg eclosion

No-choice survival test - During the first week of July 2011, shortly after initial C. auratus emergence had been observed, 120 adult beetles were collected from Apocynum spp. at four sites in Ontario: Copetown, Guelph, Dundas, and King City (Table 2.1) and subjected to a no- choice test. Mated pairs were allocated evenly between single potted plants of either; Ap. androsaemifolium (n=10), Ap. cannabinum (10), Vincetoxicum nigrum (L.) Moench (a second introduced, Vincetoxicum species but less common in Ontario) (20) or V. rossicum (20). Pots

39 were tightly netted and placed on a greenhouse bench with no plants of the same species directly adjacent to each other. Beetles were monitored daily to determine survival. The procedure was stopped one week after the last beetle on a Vincetoxicum plant died; at which time only those beetles on Apocynum spp. remained alive.

Choice oviposition test - To determine whether beetle oviposition was highly host-specific (only on Apocynum spp. ) or whether females would also use plants from other genera within

Apocynaceae for oviposition, and if so, the ranking of their oviposition preference, a choice test using tightly-netted pairs of potted plants in a common garden was set up. In July 2012, 60 mated beetle pairs were collected from Ap. androsaemifolium plants at the Copetown site (Table

2.1) and placed on paired plants in a common garden within a single tightly-netted pot containing either: a) Ap. androsaemifolium and V. rossicum ; b) Ap. androsaemifolium and

Asclepias incarnata L. (Apocynaceae); or c) Ap. androsaemifolium and Ap. androsaemifolium

(20 replicates of each plant-pair). In the event that no females oviposited on V. rossicum plants, we added As. incarnata , a North American native Apocynaceae that is not typically used as a host by C. auratus , in order to determine whether C. auratus females are highly specific to

Apocynum spp., or whether V. rossicum itself is not preferred as an oviposition site. As female preference for oviposition sites was being measured, not absolute fecundity, we chose not to count individual eggs within each egg mass due to lack of high variation observed. All egg masses laid on plants and non-plant substrates during the 14-day procedure were collected and counted. Egg masses were removed from leaves and stems then stored separately in micro- centrifuge tubes in the lab, where they were monitored daily for six weeks to record hatching.

Statistical Analysis – No-choice adult survival on Apocynum spp . and Vincetoxicum spp. was compared using an ANOVA with a post-hoc Tukey’s test. In the choice oviposition test, a type

40 II ANOVA ( Anova function; (car package Fox and Weisberg 2016)), which uses Wald χ 2 tests to generate P-values, was used to determine whether the number of egg masses laid on each plant species was significantly different from each other. A type II ANOVA was used as it allows the testing of each of the two main effects: (1) plant species; and (2) the random effect of total number of egg masses laid by the female in each pair, after testing for the other main effect. The total number of egg masses laid by the female in each pair was included in the model because the beetles would often lay egg masses on the plant and on non-plant substrates like the screen and pot and this parameter captures that response. Total egg masses laid by each pair was set as a random effect with the lmer function (lme4 package) (Bates et al. 2016). As multiple

ANOVAs were calculated in order to identify differences between all three species, a

Bonferonni-corrected alpha was used.

Larval feeding and development

Early larval feeding and development on excised roots - Adult beetles were placed in 4-L clear plastic containers (12cm x 12cm x 26cm) containing host plant foliage ( Apocynum spp.). Egg masses laid on the foliage were removed from stems and leaves every three days and placed separately in microcentrifuge tubes where they were monitored daily for hatching. Following hatching, 120 1 st -instar larvae from Richland, Washington, 190 from Mabton, Washington, and

180 from Dundas, Ontario, were placed in 10 petri dishes in the lab in groups of 12, 19, and 18, respectively, each containing cut root segments (8-10mm in length) of either Ap. cannabinum or

V. rossicum (5 petri dishes of each root species). To maximize data collection, we used all available larvae for each site instead of an equal number because pre-testing suggested variances between sites were relatively uniform. Each week the larvae were given freshly-cut roots, and dead larvae and shed head capsules were collected and preserved in 75% ethanol. To determine

41 the feeding instar or instar at death, head capsules were measured using a digital microscope

(Dino-Lite AM413TA) and image processing software (ImageJ). Head capsules were oriented with the mouth parts at the bottom and the distance at the widest point between sides was measured for head capsule width (Delbac et al. 2010). ANOVA was used to compare larval survival and head capsule widths between collection sites and root species.

Late larval development on potted plants – In July of 2012, 15 potted plants each of V. rossicum and Ap. androsaemifolium (as a control) were placed on a greenhouse bench and received 20 1 st - instar larvae from the Copetown site (Table 2.1). Larvae were produced using the adult rearing and egg mass collection described above. Larvae were placed at the base of each plant stem and the plants were grown in 4-L pots with potting soil (Sungro ® Sunshine Mix #1). A screen (mesh size = 0.5mm) at the base of each pot and tightly secured netting over the above-ground plant matter prevented escape of the larvae. Pots were held in the greenhouse under ambient light conditions (temperature ranging 16-34°C with an average of 24°C cooled by cold-water air vents). The pots were dissected after 85 days (allowing for sufficient time for beetles to develop into late-instar larvae/pupae). All live larvae in the soil were counted and head capsules measured as described above.

Results

Adult feeding

Age and sex - Feeding by C. auratus adults on V. rossicum leaves was characterized only by nibbling/exploratory feeding (<5 bites per leaf). When comparing the presence or absence of

42 feeding, older beetles fed significantly more often than recently-emerged (24 h) beetles on V. rossicum (G 1=7.992, P<0.050), thus they were used for all subsequent adult feeding tests. There was no significant difference between age ( F1,82 =0.264; P=0.608), or collection site

(F3,80 =1.319; P=0.274) for the amount of beetle feeding on Ap. androsaemifolium leaves.

Females were larger ( t111.63 =2.719; P<0.010) and fed in greater amounts on Ap. cannabinum

(t22.65 =4.123; P<0.001) than males. There was no difference between sexes for either presence/absence of feeding on V. rossicum (G 1=0.404, P=0.525) nor amount of feeding on the non-native plant ( t14.03 =0.964; P=0.353).

Lab Exposure - There was no significant increase or decrease in the presence of feeding by C. auratus adults on V. rossicum following short-term (2-day) exposure to cut leaves of the same species, with 2% feeding initially and 1% following exposure (n=10) (χ 2=0, d.f.=1, P=1). In addition, no increase or decrease in the presence of feeding was observed on native Apocynum spp. hosts with 86% feeding initially and 76% following exposure (n=30) (χ 2=1.333, d.f.=1,

P=0.248), nor Asclepias spp. (native plants in the same family, Apocynaceae, yet not used as hosts) with 80% feeding initially and 83% following exposure (n=30) (χ 2=0, d.f.=1, P=1) or the common native weed S. canadensis (Asteraceae) (χ 2=0, d.f.=1, P=1), on which only one beetle fed prior to exposure and none following exposure to S. canadensis (n=10).

When comparing the presence or absence of feeding on V. rossicum by beetles at sites where this non-native plant was intermixed with their host plants, Apocynum spp. (Dundas, ON and

Toronto, ON) to those found at sites where V. rossicum was not known to be present within at least 5km (Copetown, ON, Guelph, ON) and 500km (Mabton, WA and Richland, WA), we first determined there was no significant difference in the presence or absence of feeding by beetles

43 either 5km or 500km distant from V. rossicum (G1=0.111, P=0.739). The beetles from intermixed sites (‘exposed’) were much less likely to feed on V. rossicum than those from sites where beetles were unexposed to V. rossicum (from sites 5km to 500km away) ( G1=7.950,

P<0.010).). There was no difference between presence of feeding on native hosts Ap. cannabinum either 5 km or 500 km distant from V. rossicum as well ( G1=1.090, P=0.296), nor between sites exposed or unexposed ( G1=0.28, P=0.866). Overall, beetles from all sites fed significantly more often ( G1=153.351, P<0.001) and in higher amounts ( t108.08 =13.185,

P<0.001) on Ap. cannabinum than on V. rossicum.

Adult survival, oviposition, and egg eclosion

No-choice survival test - Adult survival on potted plants in the no-choice test was not significantly affected by collection site ( F3,56 =0.976, P=0.411) or date in July on which they were collected ( F5,54 =0.793, P=0.559). Adult beetles survived significantly longer on Apocynum spp. (13.60 ± 0.837 days), than on Vincetoxicum spp., (5.45 ± 0.288) (mean ± SE) ( F3,56 =42.010,

P<0.001), with adults on Apocynum spp. living over twice as long as those on the Vincetoxicum spp. (max. 10 days on V. rossicum ) (Fig. 2.1). As the experiment was stopped while C. auratus beetles remained alive on Apocynum spp., this survival data is truncated, and does not reflect the maximum life-length of these beetles.

Choice oviposition test - Female beetles laid significantly fewer egg masses on V. rossicum than they did on Ap. androsaemifolium (χ 2= 28.257, d.f.=1, P<0.001) or Asclepias incarnata

(χ 2=15.118, d.f.=1, P<0.001) (Fig. 2.2). There was no significant difference between the number of larvae hatching per mass laid on the three plant species ( F2,6.96 =2.145, P=0.118).

44 Larval feeding and development

Early larval feeding and development on excised roots - Larvae from all three sites (Richland,

WA, Mabton, WA, and Dundas, ON) fed on both Ap. cannabinum and V. rossicum but survived significantly longer on Ap. cannabinum (F1,957 =23.595; P<0.001), with no larvae living on V. rossicum beyond week seven of the procedure (Fig. 2.3). Of larvae given V. rossicum roots

(n=245), only 20.0% were able to undergo one molt, with no larvae molting a second time. No head capsules were found for larvae feeding on V. rossicum following the 5th week of the procedure. Larvae feeding on Ap. cannabinum molted a maximum of two times during this same time period, with 19.6% (n=245) undergoing the first molt and 10.41% of these going on to shed a second head capsule. The longest larval survival on Ap. cannabinum was 31 weeks. During the first four weeks of the procedure (when larvae were present on both root species) there was no significant difference between the shed head capsule widths of larvae fed V. rossicum or Ap. cannabinum (F1,66 =1.794; P=0.185). Throughout the procedure there was no difference in shed head capsule widths of larvae collected from the different sites and fed V. rossicum in the laboratory ( F2,31 =3.026; P=0.063). However, there was between-site variation in the width of head capsules from larvae fed Ap. cannabinum. Larvae from the Dundas, ON site were smaller than both Richland, WA ( P<0.050) and Mabton, WA ( P<0.050) sites (Tukey’s HSD), when comparing head capsule width of dead larvae measured throughout the duration of the procedure.

Late larval development on potted plants - No C. auratus larvae were recovered from any of the pots containing V. rossicum in this no-choice test. Larvae given Ap. androsaemifolium (n=300) had a survival rate of 9.7%, with an average of 8.3 ± 1.4% (mean ± SE) per pot. Larvae or pupae were found in eight of the 15 pots with Ap. androsaemifolium 85 days after larvae placement.

45 Discussion

Chrysochus auratus initially accepts the non-native Vincetoxicum rossicum vine for adult and larval feeding and oviposition in the lab, but is unable to feed beyond nibbling or complete larval development on this novel host suggesting that it is not a viable host and may act as an oviposition sink, very similar to that observed for other specialists on plants in the Apocynaceae such as monarch butterflies (DiTommaso and Losey 2003, Mattila and Otis 2003, Casagrande and Dacey 2007). It is likely that C. auratus responds to flavonoid glycosides in V. rossicum that are common within the Apocynaceae (Haribal and Renwick 1998) . The low incidence of feeding by later instar larvae and adults may be due to compounds found in V. rossicum that deter feeding, as was observed for other insect species (Mogg et al. 2008). The implication of the novel chemistry of V. rossicum preventing herbivory has been made in earlier studies

(Cappuccino and Arnason 2006). North American herbivores may need to adapt to the unique chemistry of V. rossicum in order to sustain feeding and complete development. Overall, V. rossicum seems to provide feeding cues for adult beetles (unlike the native common weed,

Solidago canadensis ), initial feeding cues for motile 1 st instar larvae, and ovipositional cues for female beetles, however its leaves do not permit feeding beyond the initial exploratory stage, and its roots are unable to support complete larval development, likely due to its chemical compounds.

Beetle age, but not sex, predicted feeding by adult beetles on V. rossicum, with younger (<24 h) beetles feeding less frequently than those at least a week older. This is in-line with predictions made in optimality models for host specialization, which predict that older beetles are more likely to accept poor quality hosts when compared with younger adults of the same species

46 (Jaenike 1990). Female C. auratus fed on native Apocynum hosts in greater amounts than the smaller males. However, there was no difference between the sexes and their feeding presence/absence and amounts on non-native V. rossicum . If feeding had been greater by the larger females on V. rossicum as compared to males ( i.e. the amount of feeding would be directly tied to food requirements of the individual beetles), this would suggest that C. auratus accepts V. rossicum as a host, albeit less preferred. The lack of difference in feeding on V. rossicum between sexes instead suggests that C. auratus does not accept the non-native vine as a food source.

Chrysochus auratus beetles found on their host plants intermixed with V. rossicum in the field appeared to avoid feeding on this introduced vine when compared to those collected from sites with no known V. rossicum . This held true for beetles collected greater than 5km (Ontario) and

500km (Washington State) from V. rossicum . The hypothesis that gene introgression due to hybridization with a western sibling species, Chrysochus cobaltinus LeConte (which sustains feeding on V. rossicum (deJonge et al. unpublished)) caused an increase in feeding, could be explored if only beetles from Washington State (500km), and not Ontario (5km), had fed on V. rossicum in significantly higher amounts. As we saw no difference in feeding levels by beetles independent of continental location, we instead propose two possible explanations for the observation of close-proximity beetles (from sites intermixed with V. rossicum ) feeding less on the novel host. First, either individual beetles from sites closest to V. rossicum had learned to avoid unacceptable hosts, as has been observed in locusts on an unpalatable forb, Senecio vulgaris (L.) (Blaney et al. 1985) or second, beetle populations in these locations may have adapted over decades of exposure to avoid V. rossicum. Early adaptation to avoid an unpalatable host within decades had been seen in a leaf mining fly, Amauromyza flavifrons

47 (Meigen) on sugar beets (Uesugi 2008), and more recently with the European corn borer

(Ostrina nublilalis (Hubner)) on maize (Orsucci et al. 2016). The lack of increase/decrease in presence of feeding by adult C. auratus on V. rossicum following two days of exposure suggests that the reduction in feeding on V. rossicum by adults intermixed with the vine in the field may be an adaptation to avoid V. rossicum ; however, longer term lab exposure studies are necessary as this shorter length of time may not be sufficient to test for changes in adult feeding.

Our study is one of the first to examine the potential relationship between a native herbivorous insect and an invasive non-native plant before actual feeding and use have been observed in the field (see also Dalosto et al. 2015; Pfammatter et al. 2015). Investigating such novel associations before they are altered through longer-term interaction is important as it describes a baseline from which to compare any future relationship between these species. By studying the herbivore-host association before it is observed in the field, we can better determine whether any possible association between these two species is due to adaptation following some specified lag period ( i.e. the current time period), and if so, the time taken, or in contrast, whether the herbivore and host plant required no adaptation to form an association or

‘ecological fit’ (Agosta 2006, Harvey et al. 2010). Use of V. rossicum by C. auratus from the initial point of contact would support the latter mechanism.

The investigation of novel associations before they occur in the field may also help to identify potential native biological control agents to help limit the spread of invasive species. If we can confirm the potential for a native herbivore to assist in controlling an invasive plant, then its augmentation and relocation would be supported, much like that undertaken with the native weevil, Euhrychiopsis lecontei (Dietz) to control Eurasian watermilfoil (Sheldon and Creed

48 2003). Lastly, the investigation of such novel associations before they are actually observed in the field can help identify species of concern that may be affected directly when an invasive arrives and becomes an oviposition sink or ecological trap. Dalosto et al. (2015) used just such a strategy to protect native species at risk by helping make recommendations for the management and control of an introduced crayfish in South America. By studying this herbivore-host relationships before it actually occurs in the field, we have been able to add valuable information on how novel associations form, as well as improve our understanding as to the potential threat and risk of invasion for at least one native beetle.

49 Tables

Table 2.1. Field collection sites for Chrysochus auratus adults in Ontario, Canada and eastern Washington State, USA. Nearest

Host Vincetoxicum

Site Location (DD) Year(s) collected plant spp.

Washington State, USA Badger Road, Richland 46.191944, -119.355556 2013,2015 2 >500km

Hybrid Site, Mabton 46.245556, -120.110278 2013,2015 2 >500km

Ontario, Canada Farm Field, Guelph 43.527778, -80.322778 2011, 2012,2015 2 (1) <5km

Copetown Field, Flamborough 43.224051, -80.055077 2011,2012 1 (2) <5km

Koffler Scientific Reserve 44.035556, -79.540833 2011 1 NA b

Cootes Drive, Dundas 43.266308, -79.941197 2011 2 (1) intermixed

High Park, Toronto 43.648866, -79.462608 2012 1 (2) intermixed a1, Apocynum androsaemifolium ; 2, Apocynum cannabinum . Numbers in brackets refer to host plants present at the same site but in reduced numbers (less than 20 stems) that had evidence of feeding by C. auratus . b Beetles from this site were not included in the field exposure test due to the small size of this colony.

50 51 Figures

16 a a 14

12

10

8 b b 6

Mean Mean no. days alive on plant 4

2

10 10 20 20 0 A. androsaemifolium A. cannabinum V. nigrum V. rossicum Plant species

Figure 2.1. Mean (±SE) number of days adult Chrysochus auratus lived when caged on

potted plants of either Apocynum androsaemifolium, Ap. cannabinum, Vincetoxicum nigrum

or V. rossicum in greenhouse experiments, 2011. Bars with the same letter are not

significantly different from each other. Numbers within bars reflect sample size (number of

potted plants). Test was stopped one week after the last beetle on Vincetoxicum spp. died, at

which time only beetles on Apocynum spp. remained alive.

52 18 a 16

14

12 b 10

8

6

Mean Mean no. egg masses/ plant c 4

2 16 18 17 0 Apocynum androsaemifolium Asclepias incarnata Vincetoxicum rossicum

Figure 2.2. Mean (±SE) number of Chrysochus auratus egg masses laid by beetle pairs on each plant species in a common garden experiment, 2012. Bars with the same letter are not significantly different from each other. Numbers within bars reflect sample size (number of potted plants) for each species. Although 20 tested pairs were used for each plant treatment, only beetle pairs later confirmed through dissection to be male-female were used for statistical tests.

53 Vincetoxicum rossicum Apocynum cannabinum

Larvae Larvae alive (%) g10$alive[g10$species == "APC"] == g10$alive[g10$species g10$alive[g10$species == "DSV"] == g10$alive[g10$species 0 20 40 60 80 100 0 20 40 60 80 100

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

Week g10$week[g10$species == "DSV"]"APC"]

Figure 2.3. Percentage of Chrysochus auratus larvae found alive each week on excised root segments of either Apocynum cannabinum (native host) or Vincetoxicum rossicum (non-native host) in lab experiments, 2015 (means ±95% confidence intervals) (n=245).

54

Chapter 3: Western North American Chrysochus Beetles: Weak Candidates for Conservation Biological Control of the Invasive Vine, Vincetoxicum rossicum (Gentianales: Apocynaceae)

Abstract

The introduced pale swallowwort ( Vincetoxicum rossicum ) (Apocynaceae) is a European vine that is invasive in eastern North America. I investigate the ability of the native leaf beetle,

Chrysochus cobaltinus, as well as a North American hybrid between C. cobaltinus and C. auratus , to feed and complete development on the vine to determine whether the beetles could act as native biocontrol agents. Results are compared with these beetles’ ability to feed on native

North American plants in the same family. Lab tests demonstrate larval C. cobaltinus are able to undergo one moult on V. rossicum roots and some adult C. cobaltinus can feed on V. rossicum foliage in amounts similar as on their native hosts; however, this beetle species is unable to reduce above-ground plant height or weight. The Chrysochus hybrid is unable to feed on V. rossicum (similar to C. auratus parents), but can feed on North American Asclepias spp.,

(similar to C. cobaltinus parents). Gene introgression into parental species populations within the hybrid zone does not appear to confer the ability for C. auratus adults to feed on V. rossicum ; however, it does appear to confer the ability for increased feeding by C. auratus and decreased feeding by C. cobaltinus on native Asclepias spp. Neither C. cobaltinus nor

Chrysochus hybrids are likely to be effective native biological control agents to slow the spread of V. rossicum across North American forest ecosystems unless adaptation to V. rossicum occurs.

55

Introduction

When a non-native plant invades a native insect’s habitat it is often unclear whether this novel interaction will result in a positive or negative outcome. The majority of interactions between invasive and native species are negative, and this is particularly true for herbivorous insects

(Schirmel et al. 2016). In contrast, there is mounting evidence that suggests many native species can form novel associations with these invaders and take advantage of a novel source of abundant food (Schlaepfer et al. 2005, Agosta 2006, Carroll 2007, Carlsson et al. 2009). Native insects that form novel associations with invasive plants are more likely to be ecotophagous herbivores (Lawton and Schroder 1977) closely related to herbivores of the introduced species in the country of origin (Futuyma and Mitter 1996). In addition, they are more likely to demonstrate broader host ranges (Bertheau et al. 2010), have higher genetic diversity (Frankham

2005), and/or feed on native plants carrying similar traits or sharing genetics with the introduced plant (Futuyma and Mitter 1996, Jobin et al. 1996, Agrawal and Kotanen 2003, Dalin and

Bjorkman 2006, Pearse et al. 2013). Depending upon the nature and extent of the interaction, adaptation to an introduced plant by native insects has the potential to reduce the invader’s spread and provide some level of biological control.

Hybridization of insects can lead to novel associations with non-native plants. For example, the hybridization of tephritid fruit flies Rhagoletis pomonella (Walsh) (Diptera: Tephritidae) and

Rhagoletis zephryia Snow, created by parental interbreeding on a novel secondary host, invasive honeysuckle, has led to preference of the hybrid insects for this host (Schwarz et al. 2005). Gene introgression, which occurs when a hybrid backcrosses with either of its parents resulting in

56 flow of genetic material between all three populations (Rhymer and Simberloff 1996), can also cause changes in host use. Hybridization and subsequent gene introgression of two swallowtail species, Papilio glaucus L. (Lepidoptera: Papilionidae), and Papilio canadensis (Rothschild and

Jordan) resulted in increased feeding ability/detoxification of Liriodendron tulipifera L. by P. canadensis , a host previously unused by the butterfly (Scriber 2002). Thus, I expect that if a novel association forms between introduced plants and native herbivore species, the existence of hybrids and parental populations with introgressed genes could increase the potential for novel host use. This is the situation for the C. auratus x C. cobaltinus hybrid system I report on in this paper. Hybridization and genetic introgression may enhance the ability for North American

Chrysochus hybrids to feed on novel hosts including invasive Vincetoxicum .

Vincetoxicum rossicum (Kleopow) Barbar (Apocynaceae; syn. Cynanchum rossicum (Kleopow)

Borhidi), commonly known as pale swallow-wort, is an invasive exotic plant from Europe that considered a major threat to native biodiversity in North America, specifically in the Northeast

USA and the provinces of Ontario and Quebec (Sheeley and Raynal 1996). Vincetoxicum rossicum ’s current western extent in North America is disturbed areas in Kansas and Missouri

(Sheeley and Raynal 1996) . An early mention of its collection in cultivated ground in Victoria,

British Columbia in the late 1800s has not persisted (Pringle 1973, Scoggan 1979). Recent studies indicate that there is little to no climatic barrier for the vine’s geographic expansion in

North America (Sanderson and Antunes 2013, R. Dickinson, personal communication). It appears then that unless effective management options are found to prevent its westward spread, this introduced plant will continue to establish in forested landscapes across North America.

57 Vincetoxicum rossicum is difficult to control mechanically and chemically; therefore, biological control is suggested as a long-term option for management at the scale of the invasion (Lawlor and Raynal 2002, Averill et al. 2008, Douglass et al. 2011). Initial research into several classical biological control agents showed a leaf-feeding beetle from southeastern Europe, Chrysochus asclepiadeus (Pallas) (Coleoptera: Chrysomelidae), to be an important herbivore of the vine

(Weed et al. 2011a, Weed et al. 2011b). The screening of the beetle as a potential biocontrol agent was suspended due to its ability to feed, oviposit and develop on native North American milkweeds ( Asclepias spp.) (Gassmann et al. 2011, Gassmann et al. 2012, Sforza 2011).

Concurrent studies identified the European leaf-feeding moth, Hypena opulenta (Christoph)

(Lepidoptera: Erebidae) , as a potential agent for V. rossicum (Weed and Casgrande 2010).

Following host-specificity testing of Hypena opulenta, the insect was approved for release in

Canada in 2013 and is now established at least one location in southern Ontario (R. S.

Bourchier, personal communication). Lab studies have shown the potential for H. opulenta to successfully reduce above-ground biomass of the vine and seed set, with as few as two larvae per V. rossicum plant (Weed and Casagrande 2010, Milbrath and Biazzo 2016). To date H. opulenta is still at low densities in the field and is not causing visible biomass reductions (R. S.

Bourchier, personal communication). Based on lab studies, the expected biomass reductions will only rarely kill the plants (Weed and Casagrande 2010, Milbrath et al. 2016) and thus additional biocontrol agents with the potential to damage roots of V. rossicum, such as Chrysochus spp. are desirable.

To date, insects native to eastern North America have not shown potential to limit the spread of

V. rossicum in its current invaded range (Sheeley and Raynal 1996, Lawlor 2000, Ernst and

Cappuccino 2005, Milbrath 2010, Milbrath and Biazzo 2012). Those that have the highest

58 probability of doing so are likely to have biological characteristics capable of forming novel associations with V. rossicum . Two insects that feed on North American plants closely related to

V. rossicum have been studied in-depth; first is a North American relative of C. asclepiadeus, the eastern leaf beetle, Chrysochus auratus Fabricius (Coleoptera: Chrysomelidae), which feeds exclusively on Apocynum cannabinum L. and Apocynum androsaemifolium L. (Apocynaceae).

In feeding and development experiments this insect did not consume V. rossicum beyond exploratory feeding, and did not develop past the second instar when placed as larvae on the roots of V. rossicum (deJonge et. al 2017). Similarly, monarch butterflies ( Danaus plexippus

(L.) (Lepidoptera: Nymphalidae) that specialize on Asclepias spp. (within the same subfamily as

V. rossicum (Asclepiadeae)) lay eggs on V. rossicum, but their larvae are unable to develop on the novel host ( i.e. V. rossicum acts as an oviposition sink for the butterflies) (DiTommaso and

Losey 2003, Mattila and Otis 2003, Casagrande and Dacey 2007).

A second North American relative of C. asclepiadeus , Chrysochus cobaltinus LeConte (Col.:

Chrysomelidae), is found in western North America and may have greater potential to form a novel association with V. rossicum than C. auratus . It has a higher genetic diversity and expresses a broader host range, feeding on plants in both the Asclepias and Apocynum genera throughout the western USA (Dobler and Farrell 1999). In addition, C. cobaltinus and C. auratus are known to hybridize in at least one location in western North America, along the

Yakima River Valley in Washington State, USA (Peterson et al. 2001). As mentioned previously, high genetic diversity and hybridization are associated with an increased ability to host shift (Frankham 2005, Schwarz et al. 2005), making the western beetles likely candidates to feed on a novel host. To date, host plant use by hybrid beetles in this area is unknown, except that they have been found feeding on Apocynum cannabinum L. (Apocynaceae) in the field

59 (Peterson et al. 2001). The hybrid beetles demonstrate a similar life cycle to both parental species, however they exhibit much lower fertility and F 2 hybrids are rare (Peterson et al. 2005).

Due to their specialization on plants closely-related to V. rossicum, their close-phylogenetic relationship with one of the plants most damaging herbivores, and increased potential for novel host due to higher genetic variability, I hypothesized both C. cobaltinus and the Chrysochus hybrid may have the potential to form a novel association with V. rossicum and assist in its biological control.

Here, I test whether western C. cobaltinus beetles and/or their hybrids could feed on the introduced vine as an initial step in forming the necessary novel association for biocontrol.

Chrysochus cobaltinus from populations along the Pacific coast and in the Sierra Nevada range in California along with sites in eastern Washington State were tested to determine whether variability between these populations affects feeding. Results of adult feeding tests for C. cobaltinus and C. auratus collected from sites within and outside the hybrid zone in Washington were compared in order to determine whether previously observed gene introgression (Peterson et al. 2001, 2005a) affected host use within the hybrid zone. For example, heterospecific mating may cause genetic material from C. cobaltinus (a species with wider host specificity) to introgress into C. auratus within the hybrid zone and cause it to accept novel hosts more often than in previous studies on C. auratus (deJonge et al. 2017). The prediction is that there would be a greater likelihood of long-term adaptation and ultimate suppression of V. rossicum if C. cobaltinus and its hybrids were able to feed on this novel plant prior to its introduction into this region. The long-term goal of this work is to identify native natural enemies that could increase mortality on invading V. rossicum and become effective biological control agents to limit its spread.

60

Methods

Adult feeding: No-choice experiments

Chrysochus cobaltinus- Adult C. cobaltinus beetles were collected from six sites in the western

United States in 2013: three sites in California along a north-south gradient and three sites in central Washington State, one of which was within the Chrysochus hybrid zone (Table 1).

Throughout their geographic range, beetles were collected from native host plants in the family

Apocynaceae, including: pubescent, thick-leaved Asclepias eriocarpa Benth. and Asclepias speciosa Torr .; glabrous, thin-leaved Apocynum cannabinum ; and thin, narrow-leaved southern milkweed Asclepias fascicularis Decne. (Table 1). Adult beetles were placed in 4-L (12cm x

12cm x 26cm) clear plastic containers with mesh lids and provided foliage collected from plants at the site and held in ambient lab conditions (16:8 L:D, 20-25°C) prior to being brought to the lab for testing.

To assess if C. cobaltinus would feed on V. rossicum or other Apocynaceae hosts native to

North America yet novel to the population, the beetle’s potential native host plants ( Ap. cannabinum, Asclepias eriocarpa , As. fascicularis and As. speciosa ) were tested concurrently with leaves of V. rossicum in no-choice feeding tests. In the lab, individual beetles were placed into 11-cm petri dishes with a moist filter paper and one leaf of the test species. Beetles were distributed evenly between leaves of test species: V. rossicum , Apocynum cannabinum , and the

Asclepias species present at the site from which they were collected (as a control) (Table 1). For the Pine Mountain Club, CA site where more than one Asclepias spp. was present, both As. eriocarpa and As. fascicularis were tested as controls. Additionally, at two sites where C.

61 cobaltinus was collected (Yosemite, CA and Ellensburg, WA), the leaves of Solidago canadensis L. (Asteraceae), a plant commonly found in association with Chrysochus spp., were collected and tested in order to determine a base level of exploratory feeding by adult beetles on non-host plants. Each leaf was scanned prior to the beetle placement and petri dishes were set up in ambient lab conditions (20-25 °C) with a photoperiod of 16:8 L:D; dishes were set up so that no leaves of the same species were adjacent to each other to account for possible variation in the lab environment. To reduce error due to surface area reduction from water loss, leaves were removed after two days and feeding galleries were traced with initial scan overlaid to record the amount of leaf surface area (mm 2) removed. When comparisons were made between leaf species, mm 2 was converted to grams to account for the difference in leaf densities. The mm 2/g ratio for each species was then calculated at each site by weighing 10 dried leaves for which the mm 2 value had been determined by scanning. All beetles were measured for length (front of the head to the tip of the elytron) and head capsule width (the distance between the widest point of the capsule) to the nearest 0.01mm using electronic calipers.

Chrysochus hybrids- Hybrid Chrysochus adults, along with adults of both parental species ( C. auratus and C. cobaltinus), were collected from a single location in Mabton, WA(Table 1), considered the center of the Chrysochus hybrid zone (Peterson et al. 2001). A total of 120 beetles (45 C. auratus , 45 C. cobaltinus, and 30 hybrids) were collected during 2013. Species were separated on-site based on their colour (Fig. 3.1); C. cobaltinus is iridescent blue, C. auratus iridescent green/gold (and sometimes reddish (personal observation)), and hybrids are typically dull purple with brown ventral regions (Peterson et al. 2001). The identification of each individual beetle was confirmed at the end of the experiment by measuring the l:w ratio of the 8 th flagellomere of each using a digital microscope, as reported by Peterson (2001). This

62 method of combining colour and flagellomere ratios has been shown to be over 97% accurate for taxonomic identification (Peterson et al. 2005a). As above, adult beetles were stored in clear,

4-L meshed plastic containers with foliage collected from the site and maintained in ambient lab conditions until testing on leaves of Ap. cannabinum, As. speciosa, and V. rossicum using the no-choice adult feeding procedure previously described .

The potential effect of gene introgression on feeding of adult beetles was investigated by comparing no-choice adult feeding test results between C. cobaltinus and C. auratus collected from sites within and outside the Washington Chrysochus hybrid zone. In 2013, the adult no- choice feeding procedure was conducted with 60 C. cobaltinus beetles collected from outside of the hybrid zone on (Ellensburg, WA) and 45 C. cobaltinus beetles collected from the centre of the hybrid zone (Mabton, WA). Beetles were distributed evenly between petri dishes containing

Ap. cannabinum, V. rossicum, and As. speciosa leaves. Apocynum cannabinum, the field host at both sites , was included in the test as a control, and As. speciosa was used to test levels of feeding on the Asclepias genus, as this species is found in close proximity to both Ellensburg and Mabton sites. In 2015, the adult no-choice feeding procedure was conducted with 55 C. auratus beetles collected from outside of the hybrid zone (Cootes and Guelph, ON, Canada) and

36 C. auratus beetles collected from the center of the hybrid zone. Beetles were evenly distributed between petri dishes containing leaves of either Ap. cannabinum , As. eriocarpa, or V. rossicum . To assure that all beetles in this experiment were equally naïve to the Asclepias spp. tested, Asclepias eriocarpa , (a milkweed found in southern California and Mexico) was used in this test to represent Asclepias spp. instead of the common northern Asclepias (As. speciosa in

WA and Asclepias syriaca L. in Ontario).

63 Statistics - G-tests were used to compare the presence of feeding between the different geographic sites and host plants for all adult no-choice feeding tests conducted in the lab. The

G-test test was used rather than the more common χ2 test as it is based on a multinomial distribution and is robust to smaller sample sizes (Gotelli 2004). ANOVA was used to compare the amount of leaf material removed by C. cobaltinus adults from Ap. cannabinum, and

Asclepias spp., at each site and between regions (grouped as California, Washington hybrid zone, Washington outside of hybrid zone) except for the comparison between native hosts at both Yosemite and Hastings sites, in which the more appropriate t-test was used. The factors analyzed were collection site, beetle head width, body length, lab temperature, and date of test.

To account for multiple host plants at the Pine Mountain, CA site, the comparison of Asclepias spp. and Ap. cannabinum leaves at the Pine Mountain, CA site required a post-hoc Tukey’s

HSD test (Tukey 1953, Kramer 1956). Feeding on Ap. cannabinum, Asclepias spp. and V. rossicum by adult beetles of both C. auratus and C. cobaltinus within and outside of the hybrid zone were compared using t-tests, except for C. auratus feeding on V. rossicum, in which a permutation ANOVA was used due to the low amount of feeding by this beetle. Permutation

ANOVA using the vegan package in R (Oksanen et al. 2016) also compared the amount of leaf material removed for both V. rossicum and S. canadensis among and between sites. The rda function with “terms” selected enabled variables to be tested sequentially and a Bonferroni correction adjusted the value for alpha to account for multiple comparisons.

Adult development: Chrysochus cobaltinus longevity, oviposition and damage

To determine whether C. cobaltinus adults were able to survive, oviposit on, and cause damage to live V. rossicum plants, a no-choice test with adults on potted plants within a greenhouse was conducted. Four beetles (two male, two female) collected from Ellensburg, WA (outside the

64 hybrid zone) were placed on each of nine Apocynum cannabinum and nine V. rossicum plants on

1 August 2013. Pots were tightly netted and placed on a greenhouse bench with no plants of the same species directly adjacent to each other. The beetles were monitored daily; any dead beetles were replaced with beetles of the same sex from the same collection site/date. Egg masses were counted daily. The test was ended after 12 days and the height and number of leaves on each plant were measured before and after the test. At the end of the test, egg masses were collected, placed singly in microcentrifuge tubes to monitor larval hatching and the plants were dried and weighed. Adults were frozen and later dissected to confirm sex and count the number of eggs retained by females.

Statistics

To compare final dry weight and height using gain scores (post-height - pre-height) of untreated control pots with pots treated with four adult C. cobaltinus beetles t-tests were used. An

ANOVA was used to compare the number of beetles that died on Ap. cannabinum and V. rossicum during the test. Both the mean numbers of egg masses laid and retained by female C. cobaltinus placed on V. rossicum or Ap. cannabinum plants were compared using t-tests.

C. cobaltinus larval feeding and development

To determine whether early-instar larvae can feed on the roots of V. rossicum, a no-choice lab experiment with C. cobaltinus larvae was conducted in petri dishes using root segments of V. rossicum , Apocynum cannabinum and Asclepias syriaca . Adult C. cobaltinus were kept in clear plastic, vented, 4-L breeding chambers with host foliage; egg masses from these adults were collected daily and stored singly in microcentrifuge tubes until hatched. On 11 August 2015, 27 recently-hatched 1 st -instar larvae were collected from egg masses laid on Ap. cannabinum by

65 beetles from Ellensburg, WA. Available larvae were placed in groups of nine per petri dish containing cut root segments (8-10mm in length) of either V. rossicum, Apocynum cannabinum

(as a native host plant), and Asclepias syriaca . In the event that larval feeding was not observed on V. rossicum , the native non-host Apocynaceae, As. syriaca, was added to the experiment.

The use of this additional plant in the same family (Apocynaceae) allowed me to determine whether avoidance of V. rossicum was because of fidelity to its native Apocynum spp. hosts, or whether the roots of the introduced species in particular are unpalatable to the larvae. Each week, larvae were given freshly-cut roots and were counted to determine their longevity and survival. Dead larvae and head capsules were preserved in 75% ethanol for later measurement using a digital microscope (Dino-Lite AM413TA) and image processing software (ImageJ software (v.1.47)).

To determine whether C. cobaltinus larvae could complete development on V. rossicum roots, a no-choice test was conducted using potted plants. On 3 August 2015, 20 pots of Apocynum cannabinum and V. rossicum each received 1 st -instar larvae hatched from eggs laid by adults collected from Ellensburg, WA as above. Two larval densities (11 and 16 larvae) were tested to assess the effect on plant biomass with 10 replicates per plant species. Larvae were placed on the soil at the base of each plant stem. An additional 20 pots of each plant species were used as controls to monitor the growth of plants without larvae in order to determine whether plant biomass of either species was reduced by larval feeding. Plants were grown in 3.5L-pots with potting soil (Sungro ® Sunshine Mix #1) in a greenhouse and screened both top and bottom with netting to prevent beetle escape. Plants were placed on a greenhouse bench in a wooden garden box filled with sawdust up to the pot rims to provide insulation. The pots were dissected after 85 days (allowing for sufficient time for beetles to develop into late-instar larvae/pupae). All live

66 larvae in the soil were counted and head capsules measured. Roots were cleaned, dried at 50 °C for 10 days, and then the dry weight recorded.

Statistics

Due to the limited number of early-instar larvae of C. cobaltinus available, no statistical analyses could be conducted for no-choice early-instar larvae feeding test. Mean dry root weights of potted V. rossicum and Ap. cannabinum plants with C. cobaltinus larvae and controls in the larval development test were compared using an ANOVA with a post-hoc Tukey’s HSD.

Throughout this study, all analyses were performed using R software version 3.2.4 (Very Secure

Dishes) (R Development Core Team 2016), except G-tests, which were carried out using the G- test calculator (McDonald 2014).

Results

No-choice feeding experiments with adults

Chrysochus cobaltinus - Adult C. cobaltinus beetles from all sites fed on V. rossicum and there was no difference in the presence/absence of feeding on V. rossicum across the sites

(G12 =17.242, P=0.141). There was also no difference in the presence/absence of feeding on native Apocynaceae host plant leaves when comparing adults between collection sites

(G12 =14.425, P=0.274). Chrysochus cobaltinus adults from all sites fed more often on their field-collected native host plants than they did on on V. rossicum (G1=7.461, P<0.010).

67 Site was the only significant factor affecting the amount of V. rossicum leaf material removed by adult C. cobaltinus (F5, 126 = 3.984, P<0.010). Native host species at beetle collection sites

(F2, 126 = 0.180, P=0.803), head width ( F1, 126 = 0.253, P=0.621), beetle length ( F1, 126 = 0.646,

P=0.435), lab temperature ( F1, 126 = 0.968, P=0.310), and date of test ( F1, 126 = 0.025, P=0.875) did not affect the amount of leaf material removed, after accounting for the main effect of site.

Similar to results with V. rossicum , the site where the beetles were collected affected the amount of leaf material C. cobaltinus adults removed from the native host Ap. cannabinum (F5, 130 =

3.738, P<0.001). Beetle size (body length) of adult beetles also affected the amount of feeding on Ap. cannabinum (F1, 130 = 11.964, P<0.001), although the two factors, length and site, did not covary (F5, 130 = 0.868, P=0.504).

When beetles are grouped by geographic location (California sites, Washington hybrid zone, and Washington outside of hybrid zone), beetles from within the hybrid zone fed in greater amounts and expressed greater variability in feeding on V. rossicum than C. cobaltinus beetles collected from California sites (Fig 3.2). There was no difference between V. rossicum feeding by adults from California and those collected from Washington outside of the hybrid zone (Fig

2). Feeding on Ap. cannabinum was not different between the Washington beetles from inside and outside the hybrid zone, however beetles from within the hybrid zone fed on Ap. cannabinum significantly less than beetles collected in California (Fig. 3.2).

When comparing host use of a native plant from which the insect was collected with other potential native hosts, C. cobaltinus adults fed equally or in greater amounts on native hosts from which they were not collected. At the Yosemite, CA site beetles collected from As.

68 speciosa (t30.524 = 6.202 P<0.001) fed more on Ap. cannabinum than on As. speciosa. When comparing the amount of feeding on the three native hosts tested ( As. fascicularis , As. eriocarpa and Ap. cannabinum ) at the Pine Mountain Club, CA site, beetles collected from As. eriocarpa

(F2,12 = 1.198 P=0.334) fed equally on all three hosts, and a post-hoc Tukey’s test revealed beetles collected from A. fascicularis prefer A. fascicularis over the novel Ap. cannabinum

(P<0.050), but not over As. eriocarpa (P<0.149). This was contrary to what was found at the site on the Hastings Reserve, CA, where beetles similarly unexposed to Ap. cannabinum in the field fed on this native host plant in greater amounts than on the As. eriocarpa from which the beetles were collected ( t41.987 = 2.014 P=0.050).

When feeding by adult C. cobaltinus on V. rossicum was compared with that on the non-familial plant Solidago canadensis , beetles fed more often ( G23.5 =19.021, P<0.001) and in greater amounts on V. rossicum than on S. canadensis at all sites ( F45 =4.726, P<0.050) (Fig. 3.3).

Chrysochus hybrid - The ability to distinguish between the different adult Chrysochus beetle species and their hybrids using colour was confirmed by the flagella l:w ratios as initially presented in Peterson et al. 2001. All 45 beetles identified by morphospecies as C. auratus had flagella l:w ratios larger than 1.546 (1.733 ± 0.021mm (mean ± SE)), while all 45 C. cobaltinus beetles had flagella l:w ratios smaller than 1.472 (1.249 ± 0.026mm), and all 30 hybrid beetles had flagella l:w ratios intermediate between both parents (1.487 ± 0.030mm).

The presence of feeding by the Chrysochus hybrids was not significantly different from either parent on the novel non-native host V. rossicum (G2=2.963, P=0.227) or the native host from which the beetles had been collected, Ap. cannabinum (G2=4.639, P=0.098). Hybrid beetles

69 initiated feeding on As. speciosa significantly more often than C. auratus parents ( G1=13.104,

P<0.001) but initiated feeding as often on As. speciosa as C. cobaltinus parents ( G1=0.089,

P=0.766). When comparing the amount of leaf material fed on between the hybrids and their parents by plant species, the amount of V. rossicum removed by hybrids appeared similar to C. auratus , however, only feeding on Ap. cannabinum differed, with C. auratus feeding in greater amounts than both C. cobaltinus and the hybrid (Fig. 3.4). Asclepias speciosa feeding by hybrids was more similar to C. cobaltinus feeding on this plant (Fig. 3.4).

Chrysochus cobaltinus beetles from outside the hybrid zone (Ellensburg, WA) and from the center of the hybrid zone (Mabton, WA) showed no difference in feeding amounts on either the native host plant Apocynum cannabinum (t26.48 =1.125, P=0.270) or introduced vine V. rossicum

(t17.313 =1.053, P=0.307). Beetles outside of the hybrid zone did however demonstrate a higher amount of feeding on As. speciosa compared to those from the center of the zone ( t26.82 =4.727,

P<0.001) (Fig. 3.5).

In the no-choice lab feeding tests with C. auratus collected from within and outside the hybrid zone, adult C. auratus inside the zone (Mabton, WA) exhibited lower feeding on Ap. cannabinum than those outside of the zone (Dundas and Guelph, ON) ( t23.581 =2.874, P=0.008).

Adults of C. auratus inside the hybrid zone also fed more often on Asclepias eriocarpa than those outside of the zone ( G1=3.632, P=0.057). There was no difference in presence of feeding

(G1=0.080, P=0.777) or the amount of feeding ( F1,44 =0.2711, P=0.583) on V. rossicum between adults within or outside of the zone.

70 Chrysochus cobaltinus longevity, oviposition and damage

In the no-choice greenhouse experiment on live potted plants, the weight of V. rossicum plants treated with four adult C. cobaltinus was the same as that of the untreated (no beetle) control plants ( t15.305 =0.3743, P=0.713). The mean height of treated V. rossicum plants was 17% less

(23.33 ±4.72 mm; mean ± SE) than plants without beetles (27.94±5.02 mm), however this was not statistically significant ( t15.941 =-0.669, P< 0.513). More beetles died on V. rossicum than on

Ap. cannabinum over the 12-day test ( F1,16 =23.579, P<0.001). Female C. cobaltinus laid egg masses on V. rossicum (2.55 ± 0.63; mean ± SE), but in far lower numbers than on Ap. cannabinum (31.44 ± 6.44) (t8.06 =4.078, P< 0.010). Females on V. rossicum had a mean of 8.66

± 1.04 eggs (mean ± SE) remaining in their ovaries at the end of the test while those on Ap. cannabinum had 6.55 ± 1.06 eggs but this was not significantly different ( t33.992 =-1.417,

P=0.165).

C. cobaltinus larval experiments

Early-instar larvae of C. cobaltinus fed and completed at least one moult on all three species of plant roots ( V. rossicum , Ap. cannabinum , As. syriaca ) during the lab test in petri dishes. The last surviving larva on both V. rossicum and As. syriaca died 11 weeks into the test, while the final larva on Ap. cannabinum survived just one additional week. Only larvae given As. syriaca completed a second moult during the test.

During the no-choice greenhouse trials with potted plants, no C. cobaltinus larvae were found on the roots or within any of the pots containing V. rossicum , while developing larvae or pupae were found in four of the 20 pots containing Ap. cannabinum . There was no significant difference between the weight of control roots (no larvae) with that of V. rossicum plants given

71 either 11 or 16 larvae ( F2,37 =0.997, P=0.378). Roots from Ap. cannabinum plants treated with 16 larvae weighed significantly less than those from the control group ( F2,37 =4.589, P<0.050) (Fig.

3.6).

Discussion

Results here indicate that the western leaf beetle, C. cobaltinus, is able to recognize invasive V. rossicum as a potential host plant and can feed on this novel host for up to two weeks, although both feeding amounts and longevity of the adults are less than on native host plants. Even though the amount of feeding on V. rossicum was not comparable to that on the native host plants at most sites, the beetles did feed on V. rossicum in much higher amounts than exploratory feeding on novel host plants outside of Apocynaceae (Fig. 3.3). The mean feeding on V. rossicum from all C. cobaltinus sites in this study (14.835 ± 3.662 mm 2) was significantly higher than what was observed for their monophagous sister-species C. auratus as described in

Chapter Two (0.644 ± 0.218 mm 2) (deJonge et al. 2017). The difference between these two species is biologically significant with C. auratus never feeding more than a few bites as compared to C. cobaltinus ’ ability to exhibit visible feeding damage on V. rossicum leaves and stems. In some cases, individual C. cobaltinus adults fed on V. rossicum in amounts above the mean feeding observed on their native hosts. For example, one beetle collected from the central hybrid zone site fed over 300 mm 2 on a V. rossicum leaf while another fed over 160 mm 2, and multiple beetles at this and other sites fed on V. rossicum over 100 mm 2 during the 2 days they were given cut leaves in no-choice tests; feeding on North American hosts Apocynum cannabinum and Asclepias speciosa was 187.75 ± 10.93mm 2 (mean ± SE) and 74.77 ±

8.40mm 2, respectively. Thus, there were some individuals who fed on V. rossicum well within

72 the normal range as that on native hosts. Interestingly, the highest feeding levels on V. rossicum were observed in beetles collected from the central hybrid zone. As hybridization can add genetic diversity leading to an increase in novel host use (Seehausen 2004, Abbott et al. 2013), it is possible that Chrysochus hybridization in this region may be the mechanism behind the high range in feeding amounts expressed by these adults. The variable feeding demonstrated by

C. cobaltinus may enable this native beetle to use V. rossicum as an intermittent host for adult feeding if it eventually spreads into the region.

Chrysochus cobaltinus adults demonstrated a propensity to feed upon Apocynaceae hosts they had not been reared on. This was observed most clearly at the Yosemite, CA and Hasting, CA sites, where Ap. cannabinum was a preferred food of beetles collected from both As. speciosa and A. eriocarpa . In the case of Yosemite, beetles fed over twice as much on Ap. cannabinum

(0.0095 ± 0.001g) compared to As. speciosa (0.003 ± 0.001g) (mean ± SE). Contrary to what was recorded in Dobler and Farrell (1999), which reported C. cobaltinus as ‘never flying’ at the

Yosemite site, I observed beetles here flying from As. speciosa plants to Ap. cannabinum plants

(<200m) on multiple occassions. It is likely that for this population, As. speciosa is the preferred ovipositional host (or a better larval host if oviposition amounts are equal), and Ap. cannabinum is the preferred host for adult feeding. In addition, this ability to use multiple host plants in the field and to feed extensively on novel Apocynaceae hosts demonstrates that individual C. cobaltinus populations are not locally-specialized to a single host. The previously identified high genetic variability in C. cobaltinus (Dobler and Farrell 1999) may be one mechanism behind this increased ability to use novel hosts. The fact that that novel hosts were not preferred

(or in some cases were more preferred) suggests that C. cobaltinus has the ability to introduce a novel host within the Apocynaceae into their diet.

73

Genetic introgression from hybridization between C. auratus and C. cobaltinus likely did not affect feeding or damage on V. rossicum, even though it could have had an effect on Aslepias spp. feeding. The ability of C. cobaltinus to feed on V. rossicum did not appear to be a dominant trait, as Chrysochus hybrids exhibited only exploratory feeding on it similar to their C. auratus parents. Gene introgression from hybridization (as identified by Peterson et. al 2001, 2005) may be one mechanism behind the observed increase in feeding on Asclepias spp. and decrease in feeding on Ap. cannabinum by C. auratus adults within the hybrid zone (Fig. 3.4), as well as the decrease in feeding on Asclepias spp. by C. cobaltinus within the hybrid zone, when compared to other members of their species outside of the hybrid zone (Fig. 3.5). In all cases of assumed gene introgression, adults within the hybrid zone increased/decreased their feeding to be more similar to their sister species. However, there was a lack of evidence of such introgression based on feeding by C. auratus on V. rossicum within the zone. Therefore, even though hybridization may increase genetic diversity leading to increased, although variable, feeding levels on V. rossicum by C. cobaltinus adults, it appears unlikely that Chrysochus hybrids or C. auratus within or outside of the hybrid zone will adapt to V. rossicum and assist in reducing its spread.

There is still opportunity for C. cobaltinus to adapt to V. rossicum once it is introduced to western North America. Although tests allowing adult C. cobaltinus beetles to choose between

V. rossicum and their native host plants were not included in the current study, when their host plant As. speciosa was placed side-by-side in a 4-L, clear plastic, meshed container with V. rossicum , beetles from the Hastings Reserve completely defoliated leaves of both species demonstrating that they will feed on V. rossicum even when their native host is available (R. deJonge, personal observation). With future exposure, C. cobaltinus is likely to feed on V.

74 rossicum in the field, and thus will have greater opportunity to adapt to this novel host.

Although uncommon, adaptation to a non-native invasive host plant, which initially is unable to support insect development (possibly due to plant toxicity), has been observed in other systems; i.e. Pieris oleracea Harris (Lep.: Pieridae) has adapted to invasive garlic mustard, Alliaria petiolata (Bieb.) Cavara and Grande (Brassicaceae) (Morton et al. 2015) and Papilio zelicaon

Lucas (Lep.: Papilionidae) has adapted to Ammi visnaga (Apiaceae) (L.) Lam. (Graves and

Shapiro 2003, Strauss et al. 2006). As mentioned above, some individual C. cobaltinus beetles fed on V. rossicum in amounts similar to feeding on this beetle’s native hosts. If beetles that express increased feeding on V. rossicum pass on their genes, the potential for adaptation by C. cobaltinus to the introduced vine would increase.

In summary, although C. cobaltinus may have the ability to adapt to V. rossicum at some point in the future following continued exposure, it is unlikely that Chrysochus hybrids will do so.

This suggests that the potential for a host shift to a novel host is high within this system, however without this adaptation, native Chrysochus beetles will unlikely be effective native biological control agents for V. rossicum at the outset of its westward expansion.

75

Tables

Table 3.1. Collection locations of North American Chrysochus spp. (eastern C. auratus, western C. cobaltinus and their hybrids found in Washington State), their native field host plants, and Asclepias spp. found within this region. To determine host plant acceptance by beetles in this genus, adult beetles collected from these sites in the province of Ontario (ON), Canada, and the USA states of California (CA) and Washington (WA) were tested for feeding on cut leaves of invasive V. rossicum , their native field host plants, and nearby Asclepias spp. (all in Apocynaceae). Beetle collection locations are ordered first by beetle species and then geographically from north to south. Asclepias spp. Relation to present near WA hybrid beetle species Collection site Decimal Degrees (DD) Field host plant this location zone C. cobaltinus Ellensburg, WA 46.945833, -120.517778 Apocynum cannabinum As. speciosa outside Granger, WA 46.320556, -120.225833 Apocynum cannabinum As. speciosa central Mabton, WA 46.245556, -120.110278 Apocynum cannabinum As. speciosa central Yosemite National Park, CA 37.739354, -119.595166 Apocynum cannabinum, Asclepias speciosa (field host) outside UC Davis Hastings Research Centre, CA 36.362724, -121.565709 Asclepias speciosa (field host) outside Pine Mountain Club, CA 34.853610, -119.149212 Asclepias eriocarpa, Asclepias fascicularis (field host) outside Chrysochus hybrid Mabton, WA 46.245556, -120.110278 Apocynum cannabinum As. speciosa central C. auratus Mabton, WA 46.245556, -120.110278 Apocynum cannabinum As. syriaca central Guelph, ON 43.527778, -80.322778 Apocynum cannabinum As. syriaca outside Dundas, ON 43.266308, -79.941197 Apocynum cannabinum As. syriaca outside

76 Figures

Figure 3.1. North American adult Chrysochus beetles can be distinguished by colour

(left to right): a) Chrysochus auratus : both dorsal and ventral sides exhibit iridescent greenish gold coloration, often with red tones (ovipositing female); b) Chrysochus hybrid

(dorsal): lacking iridescence, typically purple, though some exhibit colouration intermediate of parents (Peterson et al. 2001); c) Chrysochus hybrid (ventral): abdominal sterna are typically dull brown; and d) Chrysochus cobaltinus : both dorsal and ventral sides exhibit iridescent metallic blue colouration .

77 Apocynum cannabinum 250 b a, b

200

a 150

100 ) 2 (mm 50

83/87 25/28 35/37 0

consumed California Hybrid Zone, WA Outside Hybrid zone, WA 1 2 3 Vincetoxicum rossicum area

45 a leaf

40

Mean 35 30 25 20 a, b 15 10 b 5 66/86 18/25 32/37 0 California1 Hybrid Zone,2 WA Outside Hybrid3 zone, WA

Figure 3.2. Mean (±SE) leaf area of Apocynum cannabinum and Vincetoxicum rossicum removed by Chrysochus cobaltinus beetles collected from sites in California and

Washington State in 2013. Beetles are grouped into California (4 sites), inside the

Hybrid zone in WA (2 sites) and Washington outside of the Hybrid Zone (1 site).

Numbers below the bars are the number of beetles that fed versus the number of beetles tested for each site. Sites with the same letter are not significantly different from each other according to a post-hoc Tukey’s HSD test (α = 0.05). Bars are ordered south to north.

78

Figure 3.3. Mean (±SE) leaf surface area removed by adult C. cobaltinus beetles fed V. rossicum and Solidago canadensis at Yosemite National Park Site, CA during 2013.

Solidago canadensis , a common weed found at this site (and on which C. cobaltinus has been observed), was included to estimate the base level of adult exploratory feeding on a non-host plant by C. cobaltinus beetles. Numbers at the bottom of the bars are the number of beetles that fed versus the number of beetles tested for each site. Bars with the same letter are not significantly different from each other, according to a PERMANOVA

(P<0.050).

79

Figure 3.4. Mean (± SE) feeding by adult C. auratus , C. cobaltinus and their hybrid on cut leaves of test species. All beetles were collected in the center of the hybrid zone

(Mabton, WA) in 2013. Numbers below the bars are the number of beetles that fed versus the number of beetles tested for each site. Bars with the same letter are not significantly different from each other, according to ANOVA ( Ap. cannabinum and As. speciosa ) and

PERMANOVA ( V. rossicum ) (Bonferonni corrected α = 0.016).

80

Figure 3.5. Mean (±SE) feeding on Asclepias speciosa by adult C. cobaltinus collected from outside of C. auratus x C. cobaltinus hybrid zone (Ellensburg, WA) and in the central hybrid zone (Mabton, WA). Numbers at the bottom of the bars are the number of beetles tested for each site (all tested beetles fed). Bars with the same letter are not significantly different from each other, according to a t-test (P<0.001).

81

Figure 3.6. Mean (±SE) dry weight (g) of Apocynum cannabinum potted plant roots following treatment of either 0 (control), 11, or 16 C. cobaltinus larvae per plant.

Numbers within the bars are the number of plant roots tested. Bars with the same letter are not significantly different from each other according to ANOVA (Bonferonni corrected α = 0.025).

82 Chapter 4: Host Specificity Testing of North American Chryoshus to Enhance the Ecological Host Range Prediction for European Chrysochus asclepiadeus (Col: Chrysomelidae)

Abstract

It is widely understood than an insect’s fundamental host range is broader than its ecological host range. In weed biological control it can be difficult to predict which plant species in the fundamental host range of a biological control candidate will be realized in its ecological host range. In this study on congeneric leaf-feeding beetles, I draw conclusions about the predictive ability of the fundamental host range of Chrysochus asclepiadeus (Coleoptera: Chrysomelidae) , a classical biological control candidate for the invasive vine Vincetoxicum rossicum, by studying how the North American Chrysochus congeners’ fundamental host ranges translate into their confirmed ecological host ranges.

Testing for C. asclepiadeus as a biocontrol agent for V. rossicum was suspended in 2011 because of its ability to oviposit, develop and feed on native North American plants, including members of the Asclepias genus (Apocynaceae). My results demonstrate that

C. auratus , a North American congener highly-specific to the Apocynum genus

(Apocynaceae) in the field, is able to complete development on the roots of plants in the

Asclepias genus in the lab, and adult females oviposit on their leaves. The use of the

Asclepias genus by C. auratus in the lab, but not in the field, shows that: (1) this plant genus is able to create a ‘false positive’ for one member of the Chrysochus genus.

Therefore, the ability to feed or develop on Asclepias by other Chrysochus spp., particularly C. asclepiadeus , may be worth further investigation to determine whether

83 Asclepias acts as a false positive for other beetles in this genus; and (2) adaptation to a novel host may not occur in all situations, even when conditions appear ideal. In addition, both field surveys and greenhouse expirements in this study demonstrate that ovipositional preference by Chrysochu s spp. is not a good indicator of ecological host use. By drawing conclusions about the predictive ability of C. asclepiadeus’ demonstrated fundamental host range through the study of its North American congeners we can enhance the risk assessment for non-target attack for a classical biocontrol candidate.

Introduction

Classical weed biological control programs aim to use natural enemies for effective suppression of target invasive plants while preventing unacceptable damage to non-target species. The centrifugal phylogenetic method first tests the potential of a candidate agent to attack non-target plants that are most closely-related (phylogenetically) to the target weed in order to efficiently identify an agent’s fundamental host range (Wapshere 1974,

Briese 2005) and assure a candidate agent is sufficiently host-specific. The basis of this method is the fundamental concept that closely-related plants are similar biochemically and morphologically (Wapshere 1974). As closely-related insects often demonstrate similar ecological niches (Freckleton et al. 2002, Wiens and Graham 2005) and feed on similar species (Futuyma and Mitter 1996, Ødegaard et al. 2005), there are benefits to studying phylogenetically similar insects as well when testing candidate agents.

84 Insects native to the proposed region of introduction that are closely-related to a candidate agent could be subjected to the same host-specificity tests as the candidate agent in order to better interpret the candidate agent’s fundamental host range results.

If insects that are closely related to a candidate agent are found to feed on plants in the lab that they would otherwise not use in the field, these plants can be identified as ‘highly palatable’ in the lab and their use by the candidate agent should be further investigated.

The use of species closely-related to the candidate agent that are native or naturalized to the area of introduction have been used as surrogates for efficacy testing (Puliafico et al.

2008); however, I am not aware of any previous studies using native insects related to agents for the purpose of interpreting fundamental host range results. The study of insects closely related to candidate agents may assist in clearer identification of the host specificity of agents, thereby improving the use and application of biological control.

Vincetoxicum rossicum (Kleopow) Barbar. (Apocynaceae; syn. Cynanchum rossicum

(Kleopow) Borhidi), commonly known as pale swallowwort or dog strangling vine

(DSV), is invasive in open fields, alvars, tree plantations and forest understories throughout eastern North America (Sheeley and Raynal 1996, Lawlor 2000, DiTommaso et al. 2005, Milbrath 2010). Biological control is seen by some as the best opportunity for controlling this vine, as it is time-consuming and expensive to control physically and chemically (Lawlor and Raynal 2002, Averill et al. 2008, Douglass et al. 2011).

Chrysochus asclepiadeus (Pallas) (Coleoptera: Chrysomelidae; syn. Atymius asclepiadeus Gistel, syn. Eumolpus asclepiadeus Illiger) is a leaf-feeding beetle found

85 throughout southern Europe (Jolivet and Verma 2008, Schmitt 2011) feeding and developing exclusively on Vincetoxicum spp (Weed 2010). Female beetles oviposit on the base of the plants stems, adults feed on the leaves and larvae feed, develop and overwinter on plant roots (Weed 2010). The beetle was selected as a classical biocontrol candidate for V. rossicum and was determined to be its most effective herbivore in lab and common garden tests conducted in Europe (Weed et al. 2011a, Weed et al. 2011b).

However, the beetle also demonstrated the ability to feed and develop on several native

North American plants in lab tests, including Asclepias spp. (milkweeds) (Gassmann et al. 2010, Gassmann et al. 2011, Sforza 2011). Therefore, even though it was the most destructive feeder on roots of V. rossicum , additional screening of C. asclepiadeus ’ as a biocontrol agent of DSV was suspended out of concern for non-target impacts to native

North American species.

Here, I study the host ranges of two members of Chrysochus asclepiadeus ’ genus that are native to North America in order to improve predictions about the potential ecological host range of European C. asclepiadeus if introduced into North America for biocontrol of DSV. Chrysochus auratus Fabricius (Coleoptera: Chrysomelidae) is found in eastern

North America and feeds exclusively on Apocynum spp. (Apocynaceae) (Arnett 1968,

Doussourd and Eisner 1987, Williams 1991, Dobler and Farrell 1999, deJonge et al.

2017). Chrysochus cobaltinus LeConte in western North America specializes on plants in both Apocynum and Asclepias genera (Apocynaceae: Asclepiadeae) (Dobler and Farrell

1999, deJonge et al. 2017). The two North American species hybridize in a 25-km section of the Yakima River Valley in eastern Washington State (Peterson et al. 2001, Peterson et

86 al. 2005a). The North American and European Chrysochus beetles are appropriate to compare as they are known specialists on plants within the Apocynaceae, demonstrate similar life-cycles (Weiss and West 1921, Arnett 1968, Dobler and Farrell 1999, Peterson et al. 2001, Jolivet and Verma 2008), and share a similar form of self-defense using chemical secretions (Dobler et al. 1998). By studying how the fundamental host ranges of

North American Chrysochus spp. translate into their confirmed ecological host ranges, conclusions about the predictive ability of C. asclepiadeus’ demonstrated fundamental host range can be made.

Here, I surveyed populations of North American Chrysochus spp. to confirm their ecological host ranges. Host range testing was carried out in the lab, greenhouse, and common garden with North American Chrysochus beetles at all life stages to determine their fundamental host ranges. The ecological and fundamental host ranges of the beetles were compared in order to identify plant species that may be highly palatable to this genus in lab tests (and therefore part of the fundamental host range but not the ecological). By subjecting native congeners to similar host range testing as C. asclepiadeus , a better understanding of how the palatability of specific plant species used in lab tests may affect host range testing results will be gained.

87 Methods

Ecological host range determination

Qualitative surveys were conducted to determine the ecological host ranges of North

American Chrysochus spp. at sites in Ontario, British Columbia, Washington, and

California (Table 4.1). The site boundaries were determined by the presence of host plants; namely Asclepias spp. and/or Apocynum spp. Locations where Chrysochus spp. populations were observed but not surveyed or collected are listed in Appendix A. At survey sites, adult beetles were collected due to their low mobility, iridescent colouration, and aposematic behaviour. Egg masses were also collected from the underside of leaves, along stems (particularly at the base for C. cobaltinus (Dickinson 1995)), and on nearby vegetation (personal observation, Weiss and West 1921). Following adult beetle collection, each site was surveyed to record vegetative cover. At sites over 100 m 2, a transect was set lengthwise with five transects evenly spaced perpendicular to the main transect. Four 1-m2 quadrats were evenly spaced along each perpendicular transect to determine % coverage (up to 100%) of the available host plants at the site ( Asclepias spp. or Apocynum spp.). Sites less than 100 m 2 were divided into 1-m2 quadrats and surveyed.

All Asclepias spp. plants at C. auratus sites, as well as all plants within each quadrat were thoroughly inspected for signs of Chrysochus spp. feeding. Following site surveys, all vegetation peripheral (~50 m) to the site was searched to note any beetles or egg masses found on plant species away from higher density beetle populations.

88 Fundamental host range determination

In order to determine the fundamental host range of North American Chrysochus beetles, choice and no-choice experiments were conducted with adults and larvae in the lab and greenhouse. Host range tests were conducted using a plant list based on the demonstrated fundamental host range of C. asclepiadeus . A greater number of tests were conducted with eastern C. auratus beetles compared to western C. cobaltinus in part to facilitate the comparison with C. asclepiadeus , as both C. auratus and C. asclepiadeus only use plants in one genus within Apocynanceae as hosts.

Adult beetle feeding, survival and oviposition - No-choice tests were conducted with adult C. auratus , C. cobaltinus, and Chrysochus hybrids using cut leaves in the lab. All adults were collected in the field and placed singly in 11-cm petri dishes with moist filter paper and a single leaf of one of the test species (Table 4.2). Complete methods for this experiment and the specific comparison of feeding on V. rossicum and Apocynum spp. appear in deJonge et al. 2017a. Plant leaves of Apocynum spp. , As. speciosa , Solidago canadensis L., and Vincetoxicum spp. were sourced from the field. The remaining leaves were taken from plants grown in the greenhouse: As. eriocarpa , As. fascicularis, and As. tuberosa were grown from seed, and As. incarnata and As. syriaca were purchased as small plants from an Ontario grower specializing in wild-collected native seeds (Native

Plants in Claremont).

A no-choice test to determine C. auratus adult beetle feeding, oviposition, and survival on potted plants was conducted in July 2011. Adult C. auratus beetles were collected

89 from sites in Ontario: Copetown, Dundas, Koffler Scientific Reserve, and Guelph

(n=260) (Table 4.1). Mated beetle pairs were allocated evenly between tightly-netted test plants of the following species: Apocynum androsaemifolium (10), Ap. cannabinum (10),

As. incarnata (20), As. speciosa (10), As. syriaca (20), As. tuberosa (20), Vincetoxicum nigrum (L.) Moench (a second introduced Vincetoxicum species, but less common in

Ontario) (20), and V. rossicum (20). Apocynum cannabinum and Vincetoxicum spp. were grown in the greenhouse from roots collected in the field near Toronto, Ontario.

Apocynum androsaemifolium and Asclepias spp . were grown from plants purchased in

Ontario. Beetles were monitored daily to determine feeding, survival, and to count egg masses. The experiment was stopped one week after the last beetle on a non-host plant died and only those on Apocynum spp. remained alive. The specific comparison of survival between beetles placed on Apocynum spp. and Vincetoxicum spp. is outlined in deJonge et al. 2017a.

Larval development and choice - A no-choice larval development test with C. auratus was initiated in July 2012 using potted plants grown in potting soil (Sungro ® Sunshine

Mix #1) and placed on a greenhouse bench. Larvae were collected from egg masses laid by adult beetles collected from the Copetown, ON site (Table 4.1) and held in 4-L clear plastic containers (12 cm x 12 cm x 26 cm) with foliage from host sites. Egg masses laid on the plant material were removed every three days and placed singly in microcentrifuge tubes, which were then stored in ambient lab conditions and monitored daily for emergence. Following emergence, 20 1 st -instar larvae were placed at the base of each plant stem of 15 pots each of As. incarnata , Ap. androsaemifolium, and V. rossicum . A

90 similar no-choice larval test in potted plants was conducted in the greenhouse August

2015. Twenty larvae from egg masses laid by adults collected from Dundas, ON (Table

4.1) and hatched in the manner described above, were placed at the base of 10 Ap. cannabinum plants and one As. eriocarpa (220 larvae total). In addition, five pots of As. syriaca were treated with 16 larvae each from the Dundas site and two pots of Ap. cannabinum were treated with 20 larvae from the Guelph, ON site (Table 4.1). Beetles from the Guelph site were tested concurrently with larvae from the Dundas site to identify whether there were any site-dependent differences in larval growth/development between Ontario sites, as previous work demonstrated site-dependent size differences between C. auratus larvae from Ontario and Washington State sources (deJonge et al.

2017). Apocynum cannabinum and V. rossicum plants were both grown from rootstock collected in Toronto, ON. Apocynum androsaemifolium , As. incarnata, and As. syriaca were grown in a greenhouse from plants purchased in Ontario. In both the 2012 and 2015 experiments, a screen at the base of the pot and tightly secured netting overtop prevented larvae escape (mesh size = 0.5 mm). The pots were dissected after 85 days (allowing for sufficient time for beetles to develop into late-instar larvae/pupae), and all larvae found were counted and head capsule widths were measured using a digital microscope (Dino-

Lite AM413TA). Head capsules were oriented with the mouth parts at the bottom and the distance at the widest point between sides was measured for head capsule width

(Delbac et al. 2010). The specific comparison of the presence/absence of larvae given Ap. androsaemifolium and V. rossicum spp. in 2012 is described in deJonge et al. 2017a.

91 On 24 July 2015, a choice larval test was initiated in petri dishes using C. auratus larvae recently hatched from egg masses laid by adults collected from the Dundas, ON site and stored in microcentrifuge tubes as described above. First-instar larvae were placed in the center of 12 sterilized 11-cm petri dishes in groups of ten. Excised root segments of V. rossicum, Ap. cannabinum, As. syriaca, and S. canadensis were placed at the edges of the petri dishes in the four cardinal directions. S. canadensis (Asteraceae) was tested alongside potential Apocynaceae hosts to ensure that host choice made by C. auratus larvae was not random. Root segments were excised from plants that had been collected as rootstock in Ontario during early spring and grown in 4-L pots with potting soil in a common garden. Each petri dish had roots placed in a different order to avoid any interference from environmental factors. Larvae were monitored after 30 min, 60 min, and 120 min to determine location within the petri dish. Larvae were considered to have

‘chosen’ a root species if they were touching the root segment, curled underneath it, feeding on it or actively crawling over it.

Statistical analysis - ANOVA was used to compare feeding amounts by adults in the no- choice experiments in petri dishes and early larval survival between root species tested in the no-choice experiment in petri dishes. ANOVAs with Tukey’s HSD (Tukey 1953,

Kramer 1956) compared mean number of larvae that had chosen each root species in each dish for each time interval in the multiple choice test with C. auratus larvae and also compared adult survival between plant species in the no-choice test pooled by genera.

Head capsule width of C. auratus larvae reared on Apocynum spp. vs. Asclepias spp .

92 were compared using t-tests . All analyses were performed using R software version 3.2.4

(Very Secure Dishes) (R Development Core Team 2016).

Results

Ecological host range determination

Field surveys demonstrated a clear preference by C. auratus adults for Apocynum species.

Adult beetles oviposited on plants from a wide breadth of plant families (Table 4.1), but no feeding damage was observed on plants outside the Apocynum genus. No evidence of

C. auratus feeding was observed on Asclepias spp. during field surveys, even though plants of this genus were present at three of the seven C. auratus sites (Guelph, ON,

Toronto, ON and Kamloops, BC). The majority of C. auratus egg masses found in the field (86%) were laid on Apocynum spp ., ( n = 877) although females oviposited on surrounding substrates and plant species outside of the Apocynum genus (Table 4.1). All egg masses not laid on Apocynum plants were within quadrats that also contained C. auratus’ host plant ( Apocynum spp.).

Chrysochus cobaltinus adults at sites in California, Washington, and British Columbia were observed feeding on As. speciosa, As. eriocarpa, As. fascicularis, Ap. cannabinum, and Ap. androsaemifolium . Similar to its eastern congener, C. cobaltinus oviposited primarily on host plants (98.76%, n = 404), in addition to nearby vegetation (Table 4.1).

Chrysochus hybrid adults were only observed feeding on Ap. cannabinum in the field. No ovipositional survey was conducted at the hybrid site, as egg masses of the parental

93 species and hybrid appear indistinguishable in the field; therefore, the ovipositional ecological host range for the Chryschu s hybrid could not be determined.

Fundamental host range determination

Adult feeding, survival, and oviposition - In no-choice tests with cut leaves in petri dishes, Chrysochus auratus adults initiated feeding on seven separate plant species (Table

4.2). However, the adults only fed extensively on Apocynum spp. (known ecological hosts) (Table 4.2). Chrysochus cobaltinus adults fed on nine Apocynaceae species, four of which were not part of the beetle’s current geographical range (V. rossicum, As. syriaca, As. tuberosa, and As. incarnata ) (Table 4.2). Chrysochus hybrids fed on both

Apocynum spp . and Asclepias spp. (Table 4.2).

In no-choice tests conducted in the greenhouse with potted plants, there was a significant difference between the survival of C. auratus beetles between plant species ( F7, 251 =

51.082, P<0.001), with those on both Apocynum species surviving longer than the other species. There was no difference in length of survival between beetles on Asclepias and

Vincetoxicum genera (P=0.100). Limited feeding was observed on the leaves of two of the ten As. speciosa plants tested. A beetle that had fed upon an As. speciosa plant lived the longest of all beetles placed on non-Apocynum plants (13 days). One egg mass was laid on one As. speciosa plant and one As. syriaca plant as well as on screens within the cages of one As. speciosa , two V. nigrum, and one As. tuberosa plant. By comparison, females laid 50.40 ± 9.28 egg masses on the ten Apocynum cannabinum plants and

94 46.70 ± 7.59 on the ten Apocynum androsaemifolium plants during this same experiment (mean ± SE).

Larval development and choice - In the 2012 no-choice larval development test on potted plants, no C. auratus larvae were found on any of the roots or within any of the pots containing V. rossicum . Seven of the 15 pots containing As. incarnata and eight of the 15 pots containing Ap. androsaemifolium pots held developing larvae or pupae (Table 4.3).

In 2015, larvae developed on both As. eriocarpa and As. syriaca in addition to Ap. cannabinum (Table 4.3). In both 2012 and 2015, larvae reared on Asclepias spp. had head capsule widths smaller than those reared on Apocynum spp. (Fig. 4.1). There was no difference in the size of larvae depending on collection site (Guelph, ON vs. Dundas,

ON) ( t11.586 =0.872, P=0.401).

In the multiple-choice expirment with early instar C. auratus larvae given excised root segments in petri dishes, larvae never chose the roots of S. canadensis . Thirty minutes after they were placed in the center of the dish, larvae were evenly distributed between the three Apocynaceae roots: Ap. cannabinum, As. syriaca, and V. rossicum (F2,33 =0.910,

P=0.427). After 60 minutes, the number of larvae selecting V. rossicum was fewer than that on Ap. cannabinum (P=0.030) although it was not significantly different from As. syriaca (P=0.194). After 120 minutes, larvae equally preferred Ap. cannabinum and As. syriaca (P=0.952), over V. rossicum (F2,33 =4.848, P<0.050).

95 Discussion

Surveys from the field confirmed previous studies about the ecological host range of C. auratus and C. cobaltinus . There is errant mention of Asclepias spp. use in some of the earliest mention of C. auratus (Weiss and West 1921), although recent species descriptions indentify C. auratus as specific to Apocynum spp. (Arnett 1968, Doussourd and Eisner 1987, Williams 1992, Dobler and Farrell 1999). No feeding was observed on

Asclepias spp. in my surveys of C. auratus populations, even though three of the seven sites had plants of this genus intermixed with beetle host plants. Chrysochus cobaltinus feeds on both Apocynum spp. and Asclepias spp. in the field, as previously described

(Hatch 1953, Arnett 1968, Sady 1994, Dickinson 1995, Dobler and Farrell 1999), although, to the best of my knowledge, the ecological host use of As. fascicularis by this species that was observed here has not been reported previously in academic literature

(Tables 4.1 and 4.4). As Asclepias spp. was not observed at the site where hybrid

Chrysochus were surveyed, it is impossible to determine whether this plant genus is used as an ecological host in addition to Apocynum cannabinum .

The broad range ovipositional host plants in both fundamental and ecological host ranges by North American Chrysochus spp. (Tables 4.1 and 4.4) indicates this variable is not a good indicator of non-target risk for this genus. It is likely that the ovipositional behaviour of females in this species is similar to that of Pieris brassicae L. (Lepidoptera:

Pieridae), which oviposits within an acceptable micro-habitat, and whose larvae ‘update’ their mother’s choice by moving to the most palatable hosts within reach (Soler et al.

2012). In other words, adult female Chrysochus spp. may oviposit on vegetation within a

96 span of distance that their larvae can reasonably travel to find Apocynum spp. or

Asclepias spp. roots. This hypothesis is supported by the observation of C. auratus larvae exhibiting host-choice in the multiple-choice test here with early-instar larvae. The observation of ovipositional choice in the field and lab, along with demonstrated host- choice by larvae, indicates that by itself, ovipositional host range is not a good indicator of overall host acceptance by Chrysochus spp.

Potential for false positives

The results of the fundamental host range tests for C. auratus suggest that Asclepias spp. is broadly palatable in the lab to both adults and larvae (Tables 4.2, 4.3, 4.4). However, this palatability in the lab has created a ‘false-positive’ for C. auratus in that the insect feeds on a plant that it would not attack in the field (Marohasy 1998). Here, the lab and greenhouse tests demonstrated that Chrysochus auratus larvae could develop on

Asclepias spp., albeit more slowly than on Apocynum spp. (Fig 4.1 & 4.2), and that early- instar larvae show no preference between Asclepias spp. and Apocynum spp, roots.

Larval development on a non-target plant is not usually of concern in species where oviposition is highly specific, however here, C. auratus females oviposited on vegetation outside of the Apocynum genus. In addition, in Chapter two I show that C. auratus females will lay more egg masses on Asclepias spp. compared to other non-Apocynum members of the Apocynaceae in common garden tests (deJonge et. al 2017a), which suggests females may receive oviposition cues from Asclepias species. Thus, the host specificity tests here would predict that Asclepias spp. is used by C. auratus in its ecological range. In fact, the inclusion of Asclepias plants in the fundamental host range

97 of C. auratus as demonstrated in my study could halt its candidacy if the insect were being considered for classical biocontrol. The predicted use of this same plant genus

(Asclepias ) in the North American ecological host range of the European C. asclepiadeus should therefore be investigated further.

A common argument against classical biological control is that insect agents may adapt to non-target hosts not initially part of the predicted ecological range and cause damage

(Simberloff and Stiling 1996). Results here demonstrate how minimal the risk of this can be in some systems. Conditions appear ideal for C. auratus to use Asclepias spp. in their ecological host range, and yet they do not. In addition to the test results here showing that

C. auratus uses Asclepias spp. as a fundamental host, other compounding factors might include: a) Asclepias spp. commonly occurs within or near C. auratus sites; b) the hybridization of C. auratus with Asclepias -feeding western C. cobaltinus could cause introgression of the Asclepias -feeding trait (deJonge et al. 2017); c) C. auratus shares the same amino acid sequence associated with the ability to digest specific plant toxins as its

Asclepias -feeding western sister species, C. cobaltinus (Labeyrie and Dobler 2004); and d) it is likely that the genus Asclepias is an ancestral host plant for C. auratus (Dobler and

Farrell 1999), all which may enable C. auratus to add this genus to its host range more easily than a novel plant (Gassmann et al. 2006). Surprisingly, despite millenia of opportunity for interaction, C. auratus still does not include Asclepias spp. in its ecological host range, and this suggests that the risk of adaptation to a novel host by a highly specialized insect can be minimal in some cases.

98 There are many lab effects that may have increased the palatability of Asclepias spp. in the host range tests with Chrysochus beetles. Asclepias plants used in these experiments were regularly watered and protected from weather stress and natural enemies making them more palatable to insects and liable for attack than if they had developed naturally in the field (Karban et al. 1997, Bernays and Graham 1998). In addition, plants used in host range testing of C. asclepiadeus in Europe were fertilized (personal observation), which further increases their palatability (Van Hezewijk et al. 2008). In larval tests conducted with both North American and European Chrysochus , otherwise mobile larvae were given the artificial choice of feed or starve, which can cause them to use a plant they would move away from in the field (Marohasy 1996; van Klinken 2000). Asclepias species in particular may be susceptible to increased palatability when grown in the greenhouse or lab. Specifically, A. syriaca has been shown to reduce any induced defenses when not grown in direct sunlight (Agrawal et al. 2012) and to have lower cardenolide content when not exposed to droughty conditions (Agrawal et al. 2014). The use of As. incarnata by larvae of C. auratus and C. asclepiadeus may be due to the physical environment of the soil used in the greenhouse. This plant is typically found in wet soils (Kirk and Belt 2011), and the water itself may act as a physical barrier against soil-dwelling enemies in the field, causing them to be more vulnerable than in drier soils of the lab or greenhouse.

Recommendations

Work here clearly demonstrates that Asclepias spp. can create false-positives in lab tests for at least one member of the Chrysochus genus. The question arises then, is the use of

99 Asclepias spp. in the lab by C. asclepiadeus also a false-positive, similar to C. auratus , or instead does it accurately reflect the potential for use of Asclepias spp. in the field, similar to C. cobaltinus ? To answer this question, field trials using C. asclepiadeus could be conducted with Asclepias spp. in the beetles’ home region with at least one representative of the genus. These experiments may be relatively simple to conduct, as

As. syriaca has naturalized in some areas of Europe, and is considered invasive from

France to western (Pauková et al. 2014; DAISIE Consortium 2016). Chrysochus asclepiadeus could be tested in both adult and larval stages on As. syriaca under natural conditions alongside Vincetoxicum spp. in order to determine whether fundamental host range results are capable of being realized in the field. If C. asclepiadeus does not use As. syriaca, then its candidacy as a biocontrol agent may be worth reconsideration and further field testing. If however, C. asclepiadeus does use As. syriaca in the field, its candidacy as a classical biological agent for V. rossicum in North America should be permanently terminated, while simultaneously, its candidacy as a native biological agent for As. syriaca in Europe could be initiated. This is not to suggest that C. asclepiadeus should be permitted for release in North America at this time, but instead, that additional field trials with Asclepias spp. be conducted within Europe to determine whether use by

C. asclepiadeus of this plant genus is a false positive and further testing is warranted for this otherwise promising classical biocontrol agent.

Conclusion

By studying how the fundamental host range can translate into the ecological host range in a native insect related to a biocontrol candidate, I have shown how we can better

100 interpret the candidate’s host range tests, specifically by identifying insect traits common in the genus (eg. catholic ovipositional behaviour) and plant species that are likely to create false-positives for an insect in the lab. There may be multiple other candidate agents in which the ecological host range is difficult to assess or adaptation to a fundamental host is of concern, in which a study as carried out here, may assist in improving the prediction of the ecological host range. Thus, the study of native insects closely related to candidate agents may bring additional benefit in the risk assessment process and provide more effective biological control.

101 Tables

102 Table 4.1. Survey locations for North American Chrysochus species conducted in 2012 and 2013. Each site was surveyed once. The site boundaries were determined by known ecological host plant p resence (Asclepias spp. and/or Apocynum spp.). Ecological hosts of C. auratus are limited to Apocynum spp., whereas hosts for C. cobaltinus are both Asclepias spp. and Apocynum spp. Sites were searched thoroughly to count adults and egg masses. Each site was surveyed to record vegetative cover. At sites over 100 m 2, a transect was set lengthwise with five transects evenly spaced perpendicular to the main transect. Four 1-m2 quadrats were evenly spaced along each perpendicular transect to determine % coverage (up to 100%) of the available host plants at the site ( Asclepias spp. or Apocynum spp.). Sites less than 100 m 2 were divided into 1-m2 quadrats and surveyed. A Total no. egg Patch du Host plant (vegetative masses Locations of egg masses not laid on Species Site Location size lts cover % ± SD) (% on ecological ecological host plants (m2) (n host plants) ) C. auratus Kamloops, BC 50.694167, -120.380000 400 Ap. cannabinum (NA a) 31 250 (100.00) -

Richland, WA 46.191944, -119.355556 400 Ap. cannabinum (19.38 ± 25.93) 35 308 (69.48) dead wood, fireweed: Chamerion sp. (Onagraceae) , St John’s wort: Hypericum perforatum L. (Clusiaceae) , and sowthistle: Sonchus arvensis L. (Asteraceae) Guelph, ON 43.527778, -80.322778 350 Ap. cannabinum (44.33 ± 20.69) 44 270 (91.85) common milkweed: Asclepias syriaca, soybean: Glycine max (L.) Merr. (Fabaceae) , grass: Poaceae sp., raspberry: Rubus idaeus L. (Rosaceae) Copetown, ON 43.224051, -80.055077 500 Ap. androsaemifolium (27.47 ± 20.50) 65 13 (92.30) Canada goldenrod: Solidago canadensis (Asteraceae) Dundas, ON 43.266308, -79.941197 75 Ap. cannabinum (57.00 ± 25.55) 11 21 (85.71) thistle: Cirsium vulgare (Savi) Ten. 2 (Asteraceae) , grapevine: Vitis riparia Michx. (Vitaceae) Toronto, ON 43.648866, -79.462608 250 Ap. androsaemifolium (22.00 ± 22.10) 79 13 (76.92) dead wood Mabton, WA 46.245556, -120.110278 400 Ap. cannabinum (66.77 ± 20.00) 90 NA b NA Chrysochus Mabton, WA 46.245556, -120.110278 400 Ap. cannabinum (66.77 ± 20.00) 11 NA b NA hybrid C. cobaltinus Mabton, WA 46.245556, -120.110278 400 Ap. cannabinum (66.77 ± 20.00) 67 NA b NA Ellensburg, WA 46.945833, -120.517778 150 Ap. cannabinum (37.33 ± 22.51) 53 20 (95.00) multiflora rose: Rosa multiflora Thunb. (Rosaceae) Kelowna, BC 49.946783, -119.401973 500 Ap. androsaemifolium (NA a) 5 2 (100.00) - Yosemite National Park, CA 37.739354, -119.595166 25 Asclepias speciosa (5.66 ± 6.54) 28 49 (93.88) horsetail: Equisetum sp. (Equisetaceae), 3 knapweed: Centaurea sp. (Asteraceae), sagebrush: Artemisia sp. Hastings Research Centre, CA 36.362724, -121.565709 205 Asclepias speciosa (13.00 ± 9.08) 59 55 (100.00) - Pine Mountain Club, CA 34.853610, -119.149212 12 Asclepias eriocarpa (NA a) 22 28 (100.00) - Pine Mountain Club, CA 34.853610, -119.149212 56 Asclepias fascicularis (5.40 ± 3.44) 45 17 (94.12) lamb’s quarters: Chenopodium album (Amaranthaceae) Riley Wilderness Park, CA 33.571428, -117.593021 23 Asclepias eriocarpa (18.38 ± 28.10) 11 163 (100) - 3 Riley Wilderness Park, CA 33.576043, -117.596677 56 Asclepias fascicularis (3.00 ± 2.71) 26 0 - a Vegetative surveys were not conducted at this site. b No. of egg masses not determined as masses from both Chrysochus species and hybrid are indecipherable from each other in the field.

103

Table 4.2

Replicates (no. tested) and mean (±SE) amount of leaf surface removed from feeding (mm 2) and % of adults that fed on leaf species by both North American Chrysochus spp . and their hybrid in no-choice tests conducted with cut leaves in petri dishes in the lab 2012-2015. Beetles were collected from sites in Ontario and British Columbia in Canada as well as California and Washington in the USA (Table 4.1) Dashes (-) indicate no testing. C. auratus x C. cobaltinus hybrid C. auratus C. cobaltinus Adults that fed Mean feeding Adults that fed Mean feeding Adults that fed Mean feeding Leaf species % (no. tested) (±SE) mm 2 % (no. tested) (±SE) mm 2 % (no. tested) (±SE) mm 2 Apocynum androsaemifolium 88.0 (25) 244.38 (32.29) 100 (5) 159.52 (14.81) 100 (3) 145.86 (66.17) Ap. cannabinum 85.7 (91) 237.76 (23.71) 94.4 (179) 228.30 (13.37) 100 (10) 117.94 (21.04) Asclepias eriocarpa 33.3 (57) 22.46 (13.48) 73.4 (139) 137.64 (13.05) 100 (3) 50.28 (43.10) As. fascicularis 0 (10) 0.00 84.1 (44) 93.24 (13.28) - - As. incarnata 18.3 (60) 3.14 (1.55) 100 (9) 183.46 (51.40) 100 (4) 91.95 (31.62) As. speciosa 24.6 (65) 18.10 (12.00) 92.8 (83) 80.61 (8.71) 90.0 (10) 46.23 (20.22) As. syriaca 66.7 (6) 11.57 (6.95) 88.9 (9) 67.83 (39.45) 66.7 (3) 120.61 (19.62) As. tuberosa 0 (6) 0.00 88.9 (9) 53.52 (29.22) 100 (2) 0.12 (0.08) Vincetoxicum rossicum 26.6 (154) 1.09 (0.40) 66.2 (207) 12.96 (3.12) 40.0 (10) 1.75 (1.31) Solidago canadensis b 1.4 (70) 0.53 (NA a) 20.6 (68) 0.29 (0.07) - - an=1 bSpecies tested include S. canadensis (Asteraceae) concurrent with potential Apocynaceae hosts to assure that host choice was not random.

104 Table 4. 3

Results of no-choice larval development tests conducted with Chrysochus asclepiadeus in Switzerland and with C. auratu s in Canada. Recently hatched larvae were placed at the base of screened potted plants. Pots were dissected 85 days after initial larval placement to determine larval survival on plant roots. Dashes (-) indicate no testing. C. asclepiadeus a C. auratus b Mean ± SD Mean ± SD No. of pots survival (%) No. of pots survival (%)

Ap. cannabinum 4 2.5 ± NA 12 12.08± 15.6

Apocynum androsaemifolium - - 15 22.00 ± 25.8

Asclepias eriocarpa - - 1 10.00 ± NA c

As. fascicularis 6 16.7 ± 20.4 - -

As. incarnata 8 9.3 ± 13.0 15 15.33± 4.3

As. speciosa 5 27 ± 24.1 - -

As. syriaca 6 0 5 10. 00 ± 3.0

As. tuberosa 3 10 ± 12.7 - -

Vincetoxicum nigrum 14 22.3 ± 21.2 - -

V. rossicum 11 27.7 ± 25.6 15 0 a Each pot given 20 larvae. Tests occurred 2007 -2009 at the University of Rhode Island and were dissected after 4 months (Gassman et al. 2009, and 2011) b Each pot given 20 larvae except As. syriaca which only received 16. Tests occurred in 2012 and 2015 at the University of Toronto in Canada in a greenhouse and dissected after 85 days cn=1

105 Table 4.4

Summary of fundamental host ranges of North American Chrysochus species and European Chrysochus asclepiadeus compiled from both no- choice and choice tests on leaves and plants and available survey data (deJonge et al. 2017a & 2017b, Gassman et al. 2009 & 2010, and Sforza 2009, 2011). Dashes (-) refer to lack of use by beetle at this life stage. Blank spaces refer to no testing done. Scores in bold are plant species only in beetles' fundamental host range. Scores in grey are plant species in beetles' ecological host range. C. asclepiadeus C. auratus C. cobaltinus Adult a Larvae b Ovpstn c Adult Larvae Ovpstn Adult Larvae Ovpstn Apocynum androsaemifolium +++ +++ +++ +++ +++ +++ Ap. cannabinum - +++ - +++ +++ +++ +++ +++ +++ Asclepias eriocarpa ++ ++ +++ +++ +++ A. fascicularis +++ +++ - + +++ +++ +++ A. incarnata ++ +++ + + +++ ++ +++ A. speciosa ++ +++ + ++ + +++ +++ +++ A. syriaca ++ +++ + ++ ++ + +++ ++ A. tuberosa ++ +++ + - + +++ V. nigrum +++ +++ +++ - - + Vincetoxicum rossicum +++ +++ +++ + + + ++ + ++ Solidago canadensis - - + + - - a Adult feeding: + nibbling, ++ moderate feeding, +++ extensive defoliation. b Larval development: + one molt, ++ development beyond 2nd instar, +++ complete development. c Oviposition: + very few masses observed on plant in lab tests, ++ moderate oviposition +++ extensive oviposition observed in lab and field.

106 Figures

Fig. 4.1. Mean (±95% CI) head capsule width of Chrysochus auratus larvae on potted plants of either Apocynum androsaemifolium (n = 15) or Asclepias incarnata (n = 15) in

2012, or on Apocynum cannabinum (n = 12) or Asclepias syriaca (n = 5) in 2015 in the greenhouse. Numbers below bars reflect no. of larvae found on roots of each plant species when pots were dissected 85 days after initial placement. In both 2012

(t90.358 =2.252, P<0.050) and 2015 ( t15.847 =3.9301, P<0.010), head capsules were larger on

Apocynum spp. (ecological hosts) than on Asclepias spp. (not ecological hosts of C. auratus ).

107 Appendix A

Table 1A

Chrysochus beetle populations observed but not collected or surveyed due to the small number of beetles present or because a larger population was available nearby (<250m distant). Not all sites were visited each year of this study, and therefore only the date of last observance is listed. No. Species Location Coordinates (DD) Host plant(s) adults Date C. auratus Kamloops, BC Canada 50.699050, -120.4308 Apocynum cannabinum 12 July 2013 Pinery Provincial Park, ON Canada 43.250083, -81.849611 Apocynum androsaemifolium 12 Sept. 2017 Ancaster, ON Canada 43.243766, -79.934648 Apocynum androsaemifolium 18 Aug. 2016 Ancaster, ON Canada 43.241594, -79.938002 Apocynum cannabinum 18 Aug. 2016 Dundas, ON Canada 43.286704, -79.902659 Apocynum androsaemifolium 5 July 2015 Dundas, ON Canada 43.253128, -80.033630 Apocynum cannabinum 5 July 2015 Toronto, ON Canada 43.730680, -79.618458 Apocynum cannabinum 6 July 2015 Toronto, ON Canada 43.726823, -79.616993 Apocynum cannabinum 3 Aug. 2013 King City, ON Canada 44.037226, -79.529002 Apocynum cannabinum 10 July 2013 King City, ON Canada 44.033868, -79.526583 Apocynum cannabinum 8 July 2013 King City, ON Canada 44.034183, -79.527797 Apocynum androsaemifolium 4 July 2013 C. cobaltinus Caspers Wilderness Park, CA USA 33.547789, -117.556610 Asclepias eriocarpa 12 June 2013 Yosemite National Park, CA USA 37.744489, -119.593428 Asclepias speciosa 40 June 2013 Yosemite National Park, CA USA 37.739706, -119.570918 Apocynum cannabinum 25 June 2013 UC Hastings Reserve, CA USA 36.363068, -121.564186 Asclepias speciosa 20 June 2013 Granger, WA USA 46.323056, -120.231389 Asclepias speciosa 35 July 2015 Apocynum cannabinum and Granger, WA USA 46.323611, -120.233889 Asclepias speciosa 20 July 2015 Mabton, WA USA 46.191111, -119.911389 Asclepias speciosa 25 July 2015

108 Chapter 5: Discussion

The goal of this thesis is to assist in reducing the negative impacts of invasive plants, particularly the European vine V. rossicum by providing new insight into novel associations that may help to reduce its spread and negative effects on native species. The objectives to reach this goal were: (1) to predict the novel associations that will occur between V. rossicum and North American Chrysochus spp. in order to determine whether these native insects will be harmed by the vine or assist in reducing its spread; (2) to improve interpretation of the host range testing results of the European C. asclepiadeus, a previous biological control candidate for V. rossicum ; and (3) to use this study system to make management suggestions and improve predictions of novel associations both for ecological and biological control purposes. This thesis is the first known example of a comparison of the ecological and fundamental host range for native insects in order to provide information on the predicted ecological range of a European congener. In this way, the study provides more information for biological control practitioners to accurately predict the ecological host range of insect candidates in other systems. The thesis is also one of the few studies to predict the effects of an invasive weed on native insects prior to the development of a novel association in the field. In the first chapter, a case for the importance of novel associations to the establishment success, invasive potential, and possible future control of introduced species is made. I outline the diversity of directions novel associations can take and describe our current knowledge on predicting novel associations. Importantly, this first chapter frames methods for host

109 specificity testing in biological control within the ecological concept of novel associations and provides a clear outline of how such methods can better inform risk assessment of non-target plants by classical biological control candidates. This final chapter connects the knowledge on novel associations described in the first chapter with the key findings of this thesis and suggests approaches to encourage the formation of novel associations that may reduce the harm caused by invasive plants, such as V. rossicum .

Predictions for Novel Associations

Chrysochus auratus

The results of Chapters 2, 3, and 4 demonstrate that North American Chrysochus spp. will form novel associations with the invasive European vine V. rossicum . As outlined in the first chapter of this thesis, the occurrence of a novel association can have negative, neutral and positive impacts on the species involved. The evidence provided in Chapter 2 suggests that a negative novel association may be forming between C. auratus and V. rossicum , possibly leading the latter to act as an oviposition sink (deJonge et al. 2017).

Female C. auratus beetles will oviposit on V. rossicum leaves and larvae encountering V. rossicum roots will initiate feeding even though they are unable to develop on V. rossicum . The results of host specificity testing of C. auratus with V. rossicum provide a clear baseline from which to compare future novel associations between these species in the field, as the two species are likely to intermix with increasing frequency as V. rossicum continues to spread. It remains to be seen whether the prediction made in

110 Chapter 2 suggesting V. rossicum will act as an ovipositional sink for C. auratus will occur. Given that the current baseline of their relationship is well understood, similar host specificity tests and field surveys could be undertaken in the future at intervals of 10-20 years (enough time to allow adaptation by these univoltine beetles) to verify the prediction. Such monitoring would allow us to: (1) determine whether V. rossicum acts as an ovipositional sink once it invades C. auratus populations at higher densities, and (2) observe the progress of the association between these two species to identify whether there is any difference in the adaptation of beetles found geographically closer to or further away from the vine following their initial contact studied here.

Vincetoxicum rossicum may also cause direct harm to C. auratus adults. Chrysochus auratus adults traveling long distances may visit V. rossicum patches instead of continuing on to patches of more appropriate host plants, Apocynum spp. This could occur as: (1) V. rossicum gives feeding and oviposition cues attractive to C. auratus

(deJonge et al. 2017), and (2) C. auratus adults may be capable of travelling very long distances between acceptable host plant patches (over 3km). When conducting common garden experiments for this study, there were over 30 instances of C. auratus beetles not from the tested population attaching themselves to the outside of screened potted plants, likely attracted by the flowering Apocynum spp. and/or the fertile adults held within. The distance these migrant beetles travelled is unknown; however, the closest Apocynum spp. patch is just over 3km to the SW of the common garden site (personal knowledge). The ability to travel long distances is in contrast to previous studies on C. auratus that suggested limited dispersal (Williams 1992, Decker 2000). However, the genetic

111 homogeneity of C. auratus (Dobler and Farrell 1999) would suggest abundant gene flow between populations requiring long distance dispersal. If C. auratus is attracted to V. rossicum instead of continuing on to Apocynum spp. patches, the adults may be harmed by a loss of mating and/or feeding opportunities. Conversely, if some C. auratus individuals are able to feed on V. rossicum (beyond the intial feeding demonstrated in this study (deJonge et al. 2017), V. rossicum patches between C. auratus populations may increase connectivity between beetle populations and improve survival of these long- distance travellers.

The finding that V. rossicum may act as an ovipositional sink for C. auratus is not without uncertainty. If Apocynum spp. roots are intermixed with V. rossicum in the field, larvae that hatch on the invasive species’ leaves and fall to the ground may eventually choose to feed on Apocynum roots if they are present, thereby preventing their mother’s ovipositional error from harming the eggs/larvae (Chapter 4). Additionally, it is possible that C. auratus populations may adapt to avoid ovipositing on V. rossicum , making the negative association of V. rossicum as an ovipositional sink only temporary. This adaptive avoidance may already be occurring, as following exposure to V. rossicum in the field, feeding initiation by adult C. auratus decreases as compared to adults collected from populations were V. rossicum is not found (>5km away) (deJonge et al. 2017). Lab experiments testing feeding on V. rossicum leaves by C. auratus following different lengths of exposure to the vine could help determine whether the decrease in feeding observed in beetle populations intermixed with V. rossicum is adaptive avoidance or a learned behaviour. Alternatively, if alleles associated with V. rossicum -avoidance can be

112 identified, the prevalence of these alleles within populations could be measured over time to see if they increase within populations. If this decrease in feeding at sites intermixed with V. rossicum is due to adaptation, then the adaptation to avoid V. rossicum could continue throughout the overlapping geographic range of both species with all other factors remaining stable. Additionally, it is possible that C. auratus populations may adapt to avoid not only feeding, but also ovipositing on V. rossicum , making the negative association of V. rossicum as an ovipositional sink only temporary. If C. auratus populations adapt to avoid feeding initiation on V. rossicum , it is likely that the amount of time they spend on the vine will also decrease. As I show that Chrysochus can be opportunistic in its choice for oviposition sites (Chapter 4), the reduction in time spent on

V. rossicum initiating feeding may also lead to a reduction in the number of ovipositional errors on the vine by C. auratus females, thereby reducing the vine’s negative impact on the beetles as an oviposition sink.

Although unlikely, there remains the possibility that C. auratus will eventually adapt to

V. rossicum and use it as a food source, as both adults and larvae initiate feeding on the vine and females use it as an ovipositional site (deJonge et al. 2017). The mass of V. rossicum roots is extensive (DiTommaso et al. 2005, Mckague and Cappuccino 2005). At sites where Apocynum spp. and V. rossicum plants are intermixed, it is conceivable that

C. auratus larvae, upon hatching from eggs laid on leaves overhanging V. rossicum roots, encounter V. rossicum roots first, due to the high density of their root systems. The larvae may then initiate feeding on V. rossicum as initially they do not show a preference between V. rossicum and Apocynum spp. roots (Chapter 4). Novel host use typically

113 requires an increase in preference by females to oviposit on the new host (Jaenike 1990)

(i.e. mothers’ choose the host plant for their larval offspring). However, in cases where larvae ‘update’ the choice of their mothers, and can move off of hosts on which they were laid as eggs, their ability to develop on the novel host can be key to the host selection process (Soler et al. 2012) and an important first step in novel host adaptation. Although an increase in use of a previously unpalatable host has been observed in other systems

(Graves and Shapiro 2003, Morton et al. 2015, Strauss et al. 2006), it is important to put this in the context of novel associations typically harming native insects overall (Schirmel et al. 2016). The research presented in my thesis demonstrates that the likelihood of C. auratus adapting to V. rossicum is extremely low, especially when considering C. auratus’ inability to use another novel, yet viable, native host, Asclepias spp. Larvae of

C. auratus are able to complete development on Asclepias spp., adults initiate feeding on this plant, and females will oviposit on it in higher numbers than on V. rossicum

(Chapters 2 & 4). However, following millenia of interaction, C. auratus still does not use Asclepias spp. in its ecological host range (Chapter 4). In short, C. auratus has not demonstrated the ability to include and otherwise ‘ideal’ host plant ( Asclepias spp.) as a food source. Therefore, the likelihood of C. auratus adapting to V. rossicum and forming a positive novel association allowing the beetle to feed and develop on the vine to assist in reducing its spread is highly unlikely.

Chrysochus cobaltinus

This thesis predicts a novel association to form between C. cobaltinus and V. rossicum , resulting in a neutral or positive outcome for these native beetles. The mean feeding on V.

114 rossicum at all C. cobaltinus sites was substantially higher than what was observed for C. auratus at V. rossicum sites (Chapter 3) . Adult feeding and longevity tests conducted with

C. cobaltinus collected from Ellensburg, WA demonstrate the beetles can survive up to two weeks on the vine (Chapter 3). It is possible that with a larger sample size, the reduction in height of potted V. rossicum I observed following feeding by C. cobaltinus would be statistically significant. Ideally, the longevity and feeding impact test would be replicated using C. cobaltinus collected from the Mabton, WA site where the highest feeding on V. rossicum was observed (Chapter 3). Following invasion of V. rossicum into western North America, the variable feeding I observe on V. rossicum may enable C. cobaltinus to use the invasive vine as an intermittent host.

The higher amount of feeding on V. rossicum shown by C. cobaltinus compared to C. auratus fits the prediction that insects with a broader host range and higher genetic variability are more likely to attack novel hosts. In fact, the highest feeding on V. rossicum by C. cobaltinus was observed at Mabton site, which is located in the centre of the Washington-Chrysochus -hybrid zone, and therefore presumably had the highest genetic variation of all the study sites (Chapter 3). Contrary to previous findings (Dobler and Farrell 1999), Chrysochus cobaltinus throughout California and Washington did not demonstrate specialization on local hosts and, as observed at multiple sites, will feed to a greater extent on a novel native hosts compared to the native host from which they were collected (Chapter 3). This provides further evidence that the lack of specialization on local hosts and the broad host range of these beetles may allow them to attack novel hosts.

115

The results of my host specificity testing provide a clear baseline at a point in time when there has been no interaction between any C. cobaltinus populations and V. rossicum . Any current use of the vine by C. cobaltinus does not account for subsequent adaptation that could occur between the vine and the beetle following interaction in the field. It remains to be seen whether C. cobaltinus will adapt to V. rossicum following its introduction into western North America. Overall, the potential for adaptation to a novel host is high within this system, however without adaptation, native Chrysochus beetles will unlikely be effective native biological control agents for V. rossicum at the outset of expansion. Once V. rossicum spreads to western North America, host specificity testing with C. cobaltinus should be repeated to determine whether C. cobaltinus has adapted to include the vine in its ecological host range.

Chrysochus hybrid

The increased ability to feed on V. rossicum demonstrated by C. cobaltinus does not appear to have been inherited by Chrysochus hybrids. Instead, the hybrids feed in very low amounts on the vine, similar to the eastern C. auratus (Chapter 3). The differences in

V. rossicum feeding between the Chrysochus hybrid and C. auratus, as compared to C. cobaltinus in Table 4 (Chapter 3), are not significant although the trend is very clear. I would expect significant differences if the replicate numbers were higher, but unfortunately this experiment was limited by the difficulty in collecting Chrysochus hybrids in the field (45 total). The number of Chrysochus hybrids and C. cobaltinus found throughout eastern Washington was much lower than anticipated based on the

116 numbers reported in Peterson et al. (2001, 2005a, 2005b). There are many factors that may cause this steep decrease in beetle population, including over-collection, disease/virus, normal population fluctuations, the reduction in habitat for agricultural purposes, or, in this region, a multi-year drought. Low numbers of replicates were also an issue when comparing feeding levels of the Chrysochus hybrids to parental species on

Asclepias spp. hosts. My tests demonstrated a clear trend: Chrysochus hybrids fed more on Asclepias spp. than their C. auratus parents (Chapter 3). The hybrids appear to have inherited the ability to feed on Asclepias spp. from their C. cobaltinus parents even though feeding on V. rossicum was not inherited. This suggests that separate unlinked loci are responsible for each trait (Sheck and Gould 1996). Further support for this theory is given by the possible introgression of feeding behaviours in C. auratus within the hybrid zone. I observed an increased use of Asclepias spp. but not V. rossicum by C. auratus within the hybrid zone compared to those outside of the hybrid zone (Chapter 3).

If the traits that control feeding on V. rossicum and Asclepias spp. were linked or pleiotropic, I could expect to see the ability to feed on V. rossicum and Asclepias both inherited in the hybrid beetles and introgressed equally into C. auratus within the hybrid zone. Therefore, even though in other systems the increase in host-breadth or inclusion of all hosts accepted by both parents can occur (Mathenge et al. 2010 and references therein), this does not appear to be the case for Chrysochus hybrids. It seems that

Chrysochus hybrids are unlikely to form a novel association with invasive V. rossicum to the degree that their feeding will result in control of the vine.

117 Chrysochus asclepiadeus

Based on the fundamental host ranges for the two North American Chrysochus spp . seen here compared to their ecological host ranges , the prediction that European C. asclepiadeus will form a novel association with North American Asclepias spp. as initially predicted by Gassmann et al. (2011, 2012) and Sforza (2011) (Chapter 4) can be re-examined. The risk to Asclepias spp. should be taken seriously (as C. asclepiadeus is able to use this genus in lab tests); however, the use of this genus by C. auratus in the lab, but not in the field clearly demonstrates that these plants are liable to create false- positives in the lab. Although Asclepias spp. are often found near C. auratus populations, are used in the field by sister species, C. cobaltinus, and support C. auratus feeding and development in the lab, this genus is not used as a host by C. auratus in the field

(Chapters 2, 3 & 4). This demonstrates unequivocally that plants included in an insect’s fundamental host range are not necessarily at imminent risk for use in their ecological host range even when conditions appear ideal and given ample time for adaptation.

Further field testing is needed to help determine whether the use of Asclepias spp . by C. asclepiadeus is a false positive, or whether this plant genus would not be at risk of attack in the field.

Recommendations for Biological Control

As clearly shown here with Chrysochus and Apocynaceae host plants, host range testing of closely-related native insects can enhance the interpretation of host specificity testing for candidate biocontrol agents and better predict their future ecological ranges. The

118 approach described in my thesis can be applied to other systems if native relatives appropriate for comparison are available for study (e.g. share similar life-history traits, feeding behaviour, etc.). For example, five European members of the Ceutorhynchus genus have been investigated as potential classical biological control agents for garlic mustard Alliaria petiolata (M. Bieb.) Cavara and Grande (Brassicales: Brassicaceae)

(Van Driesche et al. 2002). Larvae of one of the candidate agents, Ceutorhynchus scrobicollis Nerensheimer and Wagner (Coleoptera: Curculionidae), were able to complete development on one North American Brassicaceae, spreading yellowcress

(Rorippa sinuata [Nutt.] Hitchc.), although with minimal damage (Gerber et al. 2009).

Induced defenses are common in Brassicaceae (Hopkins et al. 2009), which could lead to variation in the palatability of test plants for this study. There a number of native North

American Ceutorhynchus spp. specializing on plants in the Brassicaceae that could be investigated as either native biological control agents and/or to assist in better interpretation of the European Ceutorhynchus spp. fundamental host range. If native

Ceutorhynchus spp. that do not use R. sinuata in the field are able to complete development on this plant in the lab (similar to what we have observed with C. auratus larvae developing on native Asclepias spp. in the lab, but not in the field), the use of this native Brassicaceae by C. scrobicollis would be worth further investigation/field testing for a more accurate prediction of its true ecological range. This is just one example of another system where the testing of native congeners may assist biological control. There may be other native insects worth investigating in order to improve the interpretation of classical host specificity test results for potential native biological control candidates.

119 Initial attempts at creating predictive models for novel associations have identified that information on known plant-insect associations is needed to create accurate models, however no clear methodology on how to gather this information is suggested (Pearse et al. 2013). Biological control host specificity testing provides a largest dataset of information to predict novel associations. Fundamental host range data gathered to assess potential non-target impacts, coupled with data on the realized ecological host range of the agent, can be applied to predictive models for novel associations. Although follow-up studies to determine the ecological host range of agents following their release are required, these could be more consistent and thorough (Hinz et al. 2014). By adding data gleaned from fundamental and ecological host ranges demonstrated by the classical agents’ native relatives, the dataset is enhanced. When coupled with information on the insect and plants’ biological traits and phylogenies, data initially gathered from host specificity tests for biological control purposes could provide information integral to the formation of the large datasets necessary to create accurate predictive models for novel associations.

Management Applications of Novel Associations

Prescriptive evolution

The lack of observation of a novel association between an invasive species and potential native ‘enemies’ within the area of introduction does not mean that such an association could not form, especially if given assistance. Methods to increase the number of insects with specific genotypes that have already adapted to invasive plants, while reducing the genetic diversity of the invasive plant to prevent co-adaptation, could occur

120 concurrently. Termed ‘prescriptive evolution’, the aim is to use evolutionary processes in order to help manage the impacts of invasive species. (Smith et al. 2014). We can use these principles to reduce potential harm caused to C. auratus by V. rossicum . For example, C. auratus from sites intermixed with V. rossicum that demonstrate an avoidance of the vine (deJonge et al. 2017) could be spread to other sites within the same region that have not been exposed to V. rossicum yet may intermix with the vine in the near future. If this trait is inherited, the offspring of beetles at these sites may be less likely to waste time initiating feeding on the vine and reduce their likelihood of ovipositional errors on its leaves. Similar tactics are suggested by Schlaepfer et al.

(2002) and have already been undertaken for native seedlings at sites exposed to invasives and demonstrating some ability to adapt to invasive competitors (Deck et al.

2013, Lankau 2012). Unfortunately, this strategy may not be available to reduce harm to the North American monarch butterfly ( Danaus plexippus) caused by V. rossicum acting as an oviposition sink. Even though sub-populations are somewhat isolated in the summer months (Brower and Boyce 1991) the multi-generational migration and subsequent grouping up at overwintering sites (Brower 1995, Flockhart et al. 2013), make the selection of distinct populations that have adapted to avoid ovipositing on V. rossicum nearly impossible to identify. Without specific populations expressing the avoidance of

V. rossicum , the initiation of a breeding program for accelerating adaptation of this trait would be very difficult. As it is clear that a higher genetic diversity (often due to admixture) in an introduced plant can contribute to its invasiveness (Novak & Mack

2005), a plant’s genetic pool could be reduced in order to limit its ability to adapt in its novel environment. Some studies suggest using periodic disturbance of mowing, burning,

121 or seed removal in order to limit gene flow between patches of invasive plants (Leger and

Espeland 2010). This purposeful genetic bottlenecking may be an appropriate strategy for some invasives, however it should be noted that a lack of genetic diversity has assisted in the invasiveness of others (Dlugosch and Parker 2008). Although it is a compelling strategy, the purposeful genetic bottlenecking of V. rossicum is not something to recommend. Vincetoxicum rossicum appears to lack genetic diversity (Douglass 2008), making gene flow between sites an unlikely factor leading to its invasive success. Even if a high genetic diversity was identified as a causal factor in its invasion, mowing will not consistently reduce seed set in V. rossicum unless done four times a season (Milbrath et al. 2016). Therefore, I suggest that if prescriptive evolution is to be applied in this system, the first priority should be maintenance of North American Chrysochus beetle populations, followed by the transportation of adapted genotypes to sites likely to be invaded, instead of using time and resources to attempt genetic bottlenecking of a plant whose invasability does not seem to be hindered by its lack of genetic diversity.

Accelerated adaptation

Biological control practitioners could consider accelerating adaptation as a potential solution to assist in the management of the invasive vine. Accelerated adaptation is the breeding of insects within a species for purpose of creating strains suited for biological control of target invasive plants. Similar processes, termed ‘biological improvement’, are used in agricultural systems, typically with a focus on selecting for pesticide resistance or better efficacy of biological control agents under different climatic conditions (Atwa

2014). For example, naturally-occurring pesticide-resistant strains of a mite, Metaseiulus

122 occidentalis (Nesbitt), have been selected for in the lab and reared to assist in biological control of orchard and vineyard pests, in conjunction with chemical treatments in

California (Atwa 2014). Others have genetically-engineered strains of fungal pathogens for better efficacy in controlling insect pests, including the insertion of scorpion venom toxic to the insects (St. Leger and Wang 2010). No known studies have yet focused on selecting strains of insects likely to exhibit a host-switch to an invasive plant. Colonies of

C. cobaltinus could be reared in the greenhouse and individuals/strains that express greater V. rossicum feeding and development success could be selected for in order to accelerate such a host switch. The V. rossicum -preferring strain could be released as a conservation biological control agent in western regions where and when the vine has become invasive. Under this scenario, efficacy tests and assurances that the introduction of V. rossicum strains of C. cobaltinus would not cause undo harm to Asclepias spp., and

Apocynum spp. in the area of introduction would be required. The use of accelerated adaptation to mimic slower natural host-switching processes on a faster time-scale may be worth investigating to contain invasive species before they reach widespread establishment in new geographic areas.

If a V. rossicum -damaging strain of C. cobaltinus is identified, its release into central and eastern regions of North America where V. rossicum is currently prevalent, could be investigated. There are currently no restrictions on the transport of native insects from one area of Canada to the other (except in the form of firewood quarantines), making such a move lawful. If strains of V .rossicum -damaging C. cobaltinus were bred in the lab, the only law currently in Canada somewhat pertainting to this situation is the New

123 Substances Notification for Products Regulated under the Food and Drug Act

(Environment Canada 2009), which may or may not prevent such a release from occuring. However, even without studying the possible non-target impacts of such a move, the use of C. cobaltinus in such a scenario is unwise and unethical. Concerns include: 1) As C. cobaltinus lacks regional host specialization (Chapter 3), V. rossicum - damaging strains may be just as likely to feed on other hosts in the Apocynaceae that they encounter, including native Asclepias spp. and Apocynum spp. If so, the high numbers of

C. cobaltinus required for an innundative release at V. rossicum sites could eventually disperse to native Apocynaceae and cause little long-term damage to V. rossicum in the field. 2) Hybridization with C. auratus found throughout C. cobaltinus ’ area of proposed introduction could reduce the presence of the V. rossicum -feeding trait, as feeding ability on V. rossicum does not appear to be a dominant inherited trait in this genus (Chapter 3).

3) Currently, hybridization between C. auratus and C. cobaltinus is rare due to steep pre and post zygotic barriers including a lack of overlap between their geographic ranges

(Peterson et al. 2005). However, if C. cobaltinus are introduced throughout the current area of V. rossicum ’s invasion in sufficient numbers to cause damage to the vine, the opportunity for hybridization with C. auratus will increase. The impacts of increased hybridization could include an escalation of genetic introgression of the traits responsible for the feeding ability on Asclepias spp. into C. auratus , as suggested in Chapter 3. This could increase the opportunity for some C. auratus populations to include Asclepias spp. in their ecological host range. Therefore, any consideration of transporting C. cobaltinus eastward should not be undertaken without adequate risk assessment.

124 Transfer of endosymbionts

Even with extensive genetic and behavioural knowledge of a species, predictions of host use can be incorrect. For example, in lab tests, C. asclepiadeus is able to feed and develop upon plants in the Asclepias genus (Chapter 4). However, previous research on the biology of this C. asclepiadeus predicted the use of Asclepias spp. or other cardenoloid-laden plants to be very rare if not impossible (Labeyrie and Dobler 2004).

The amino acid sequences in the Chrysochu s genome that have been correlated with the ability to detoxify cardenoloids found in Asclepias spp. and Apocynum spp. are absent in

C. asclepiadeus (Labeyrie and Dobler 2004), leading a follow-up study to conclude that the European Chrysochus is unable to digest cardenoloids (Dobler et al. 2012). It is suggested that the possession of a gut impermeable to cardenoloids may be the reason for cardenoloid-use by insects like C. asclepiadeus that lack the usual amino acid substitution needed to permit feeding on otherwise toxic plants (Agrawal et al. 2012). However, the direct reason for feeding and development on cardenoloid-laden plants by C. asclepiadeus has not yet been determined (Dobler et al. 2015). In addition, C. auratus and C. cobaltinus show identical amino acid sequences at the site responsible for cardenoloid insensivity (Labeyrie and Dobler 2004), which would imply an identical ability to feed on the same toxic hosts if these sequences alone predicted host use. This thesis clearly demonstrates that the feeding and development abilities by these two North

American Chrysochus , although similar, are far from identical (Chapter 4). Clearly, a key factor outside of the beetle’s amino acid substitutions lies behind the different host-use abilities in Chrysochus . I suggest gut microbes, or intestinal endosymbionts that permit

125 ingestion and digestion of highly toxic plants could be the missing factor in explaining host use observed in this system.

Gut microbes in insects can play a leading role in host specificity and preference

(Sudakaran et al. 2015). Research on the role of the microbiome and how it impacts human health and digestion is extensive (Cho and Blaser 2012, Kinross et al. 2008); even leading to the coining of the term ‘holobiont’, which describes humans and other macrobial life as an not just one organism, but an assemblage of many working in symbiosis (Margulis 1991). Some of the best-known research on this topic within the context of host use and insects has been conducted with Wolbachia (Wolbachia pipientis) , which is primarily known as a widespread reproductive parasite of insects. The

Wolbachia virus also plays a key role in the digestion of some insects by metabolizing and synthesizing key vitamins and minerals, without which, the insect hosts would experience a lower level of fitness (Brownlie et al. 2009, Moriyama et al. 2015, Nikoh et al. 2014). Outside of Wolbachia, other studies demonstrate the transmission of specific endosymbionts increased performance on hosts that had typically been of poor quality for a variety of insects. These examples include: (1) increased survivorship on soybean by shieldbugs (Hosokawa et al. 2007); (2) improved survival on locusts by aphids (Wagner et al. 2015); (3) improvement in overall egg hatch in stinkbugs (Hosokawa et al. 2007). In addition, bacteria and yeast gut symbionts isolated from bark beetles have been implicated in the hydrolyzation of carbohydrates and organic compounds not typically easily digested by insects (Briones-Roblero et al. 2017). The acquisition of a complex of symbionts appears to have facilitated the diversification of host use by bugs within the

126 family Pyrrhocoridae, without which sister families are unable to feed upon these chemically-defended seeds (Sudakaran et al. 2015). Some have hypothesized that adaptation accelerating host differentiation could be initiated by symbiosis with particular endosymbionts (Vavre and Kremer 2014). However, even with the abundance of knowledge on endosymbionts impacting host breadth, the research on microbes in insects is lacking in the otherwise-rigorous process of host-specificity testing. I suggest testing the impact of microbes on host use within the Chrysochus and Apocynaceae system.

In order to test whether endosymbionts may be a factor determining host use in this genus, an experiment using early C. auratus or C. cobaltinus larvae and the fecal matter of female C. asclepiadeus would be informative. As mentioned in the introduction, following hatch and before the North American larvae drop to the ground to feed on roots, early-instar larvae chew through a fecal encasement created by their mothers. In other systems, this maternal fecal matter has been shown to be an important vector for key endosymbionts required for digestion of chemically-defended plants (Hammer and

Bowers 2015). The early larvae of C. auratus or C. cobaltinus could be removed from fecal encasements prior to hatch and maintained inside a microcentrifuge tube with the fecal matter of C. asclepiadeus mothers. Following observed ingestion, larvae would then be placed on roots of V. rossicum, and their survival and development would be compared with larvae not fed fecal matter of C. asclepiadeus mothers. This test could be similarly done with C. auratus larvae and fecal encasements from C. cobaltinus mothers to determine whether microbes ingested from fecal egg encasements could increase survivorship and development time on Ascelpias spp. roots. As well, antibiotics could be

127 given to C. cobaltinus and/or C. asclepiadeus adults to determine if it limits their ability to feed on Asclepias or Vincetoxicum spp. Ideally, gut symbionts would be isolated to determine, which bacteria are associated with host feeding ability. However, these simple lab experiments would be an important first step in investigating the role that endosymbionts may play in Chrysochus ’ digestion of the well-defended Apocynaceae family.

Conclusion

For each of the Chrysochus species covered in this thesis, I have recommended further testing. This is not to suggest that more testing should have been done with each species in preparation for this thesis, but instead is recommended in order to avoid predicting novel associations based on a single snapshot in time. As C. auratus is only now being exposed to higher densities of V. rossicum in the field, and both C. cobaltinus and the

Chrysochus hybrid have no experience with the invasive vine, the predictions herein are based primarily on the ability of the insects to demonstrate ecological fitting. Host specificity testing with C. auratus and V. rossicum should be done 10-20 years from now to observe whether the novel association has changed following a longer opportunity for adaption. Similarly, when seeking out native biological control agents, it is recommended that surveys should be done every 10-20 years to allow sufficient time for adaptation in order to identify any novel associations that have formed in the intervening years.

However, it is important to recognize that most novel associations with invasive species result in negative outcomes for native insects, and even though an insect may have the ability to use a novel host, this doesn’t guarantee its use in the field. Therefore, only

128 insects most likely to form a novel association should be investigated. Their adaptation to the novel host resulting in biotic resistance is predicted to be rare. By studying these novel associations through time, management decisions may be better informed and the host specificity tests for classical agents may be interpreted better. Both of which will inform management and research in order to assist in reducing the impact of invasive species such as V. rossicum.

129 References

Abbott, R., D. Albach, S. Ansell, J. W. Arntzen, S. J. E. Baird, N. Bierne, J. Boughman, A. Brelsford, C. A. Buerkle, R. Buggs, R. K. Butlin, U. Dieckmann, F. Eroukhmanoff, A. Grill, S. H. Cahan, J. S. Hermansen, G. Hewitt, A. G. Hudson, C. Jiggins, J. Jones, B. Keller, T. Marczewski, J. Mallet, P. Martinez- Rodriguez, M. Möst, S. Mullen, R. Nichols, A. W. Nolte, C. Parisod, K. Pfennig, A. M. Rice, M. G. Ritchie, B. Seifert, C. M. Smadja, R. Stelkens, J. M. Szymura, R. Väinölä, J. B. W. Wolf, and D. Zinner . 2013 . Hybridization and speciation. J. Evol. Biol. 26: 229–246.

Agosta, S. J. 2006 . On ecological fitting, plant-insect associations, herbivore host shifts, and host plant selection. Oikos. 114: 556–565.

Agosta, S. J., and J. A. Klemens . 2008 . Ecological fitting by phenotypically flexible genotypes: implications for species associations, community assembly and evolution. Ecol. Lett. 1123–1134.

Agrawal, A. A., E. E. Kearney, A. P. Hastings, and T. E. Ramsey . 2012 . Attenuation of the jasmonate burst, plant defensive traits, and resistance to specialist monarch caterpillars on shaded common milkweed ( Asclepias syriaca ). J. Chem. Ecol. 38: 893–901.

Agrawal, A. A., and P. M. Kotanen . 2003 . Herbivores and the success of exotic plants: a phylogenetically controlled experiment. Ecol. Lett. 6: 712–715.

Agrawal, A. A., E. T. Patrick, and A. P. Hastings . 2014 . Tests of the coupled expression of latex and cardenolide plant defense in common milkweed ( Asclepias syriaca ). Ecosphere. 5: 126.

Allison, J. D., and J. D. Hare . 2009 . Learned and naive natural enemy responses and the interpretation of volatile organic compounds as cues or signals. New Phytol. 184: 768–782.

Alpert, P., E. Bone, and C. Holzapfel . 2000 . Invasiveness, invasibility and the role of environmental stress in the spread of non-native plants. Perspect. Plant Ecol. Evol. Syst. 3: 52–66.

Araujo, S. B. L., M. P. Braga, D. R. Brooks, S. J. Agosta, E. P. Hoberg, F. W. Von Hartenthal, and W. A. Boeger . 2015 . Undestanding host-switching by ecological fitting. PLoS One. 10: 1–17.

Arnett, R. 1968 . The beetles of the United States (A manual for identification). The American Entomological Institute, Ann Arbor, MI.

130 Averill, K. M., A. DiTommaso, C. L. Mohler, and L. R. Milbrath . 2010 . Establishment of the invasive perennial Vincetoxicum rossicum across a disturbance gradient in New York State, USA. Plant Ecol. 211: 65–77.

Averill, K. M., A. DiTommaso, and S. H. Morris . 2008 . Response of Pale Swallow- wort ( Vincetoxicum rossicum ) to Triclopyr Application and Clipping. Invasive Plant Sci. Manag. 1: 196–206.

Baker, H. G. 1974 . The evolution of weeds. Annu. Rev. Ecol. Syst. 5: 1–24.

Bates, D., M. Maechler, B. Bolker, and S. Walker . 2016 . lme4: Linear mixed-effects models using Eigen and S4. R Packag. version 1.1-12.

Battin, J. 2004 . Bad habitats: ecological traps and the conservation of populations. Soc. Conserv. Biol. 18: 1482–1491.

Bernays, E. A., and M. Graham . 1998 . On the evolution of host specificity in phytophagous arthropods. Ecology. 69: 886–892.

Bertheau, C., E. G. Brockerhoff, G. Roux-Morabito, F. Lieutier, and H. Jactel . 2010 . Novel insect-tree associations resulting from accidental and intentional biological “invasions”: a meta-analysis of effects on insect fitness. Ecol. Lett. 13: 506–15.

Blaney, W. M., C. Winstanley, and M. S. J. Simmonds. 1985 . Food selection by locusts: An analysis of rejection behaviour. Entomol. Exp. Appl. 38: 35–40.

Blossey, B., and R. Notzold . 2016 . Evolution of increased competitive ability in invasive nonindigenous plants: A hypothesis. J. of Ecol. 4. 83: 887–889.

Bossdorf, O., H. Auge, L. Lafuma, W. E. Rogers, E. Siemann, and D. Prati . 2005 . Phenotypic and genetic differentiation between native and introduced plant populations. Oecologia. 144: 1–11.

Briese, D. T. 1989 . Host-specificity and virus-vector potential of Aphis chloris [Hemiptera: Aphididae], a biological control agent for St John’s wort in . Entomophaga. 34: 247–264.

Briese, D. T. 2005 . Translating host-specificity test results into the real world: The need to harmonize the yin and yang of current testing procedures. Biol. Control. 35: 208– 214.

Brower, L. P. 1995 . Understanding and misunderstanding the migration of the monarch butterfly (Nymphalidae) in North America: 1857-1995. J. Lepid. Soc. 49: 304–385.

Cadotte, M. W., R. Dinnage, and D. Tilman . 2012 . Phylogenetic diversity promotes ecosystem stability. Ecology. 8: S223–S233.

131

Cappuccino, N., and J. T. Arnason . 2006 . Novel chemistry of invasive exotic plants. Biol. Lett. 2: 189–193.

Carlsson, N. O. L., H. Bustamante, D. L. Strayer, and M. L. Pace . 2011 . Biotic resistance on the increase: native predators structure invasive zebra mussel populations. Freshw. Biol. 56: 1630–1637.

Carlsson, N. O. L., O. Sarnelle, and D. L. Strayer . 2009 . Native predators and exotic prey –an acquired taste? Front. Ecol. Environ. 7: 525–532.

Carroll, S. P. 2007 . Brave New World : the epistatic foundations of natives adapting to invaders. Genetica. 193–204.

Carroll, S. P., and J. E. Loye . 2012 . Soapberry Bug (Hemiptera: Rhopalidae: Serinethinae) native and introduced host plants: biogeographic background of anthropogenic evolution. Ann. Entomol. Soc. Am. 105: 671–684.

Casagrande, R. A., and J. E. Dacey . 2007 . Monarch butterfly oviposition on swallow- worts ( Vincetoxicum spp.). Environ. Entomol. 36: 631–6.

Chupp, A. D., and L. L. Battaglia . 2014 . Potential for host shifting in Papilio palamedes following invasion of laurel wilt disease. Biol. Invasions. 16: 2639–2651.

Colautti, R. I., A. Ricciardi, I. a. Grigorovich, and H. J. MacIsaac . 2004 . Is invasion success explained by the enemy release hypothesis? Ecol. Lett. 7: 721–733.

Creed, R. P. J., and S. P. Sheldon . 1993 . The effect of feeding by a North American weevil, Euhrychiopsis lecontei , on Eurasian watermilfoil ( Myriophyllum spicatum ). Aquat. Bot. 45: 245–256.

Creed, R. P. J., and S. P. Sheldon . 1995 . Weevils and watermilfoil: Did a North American herbivore cause the decline of an exotic plant? Ecol. Appl. 5: 1113–1121.

Cristofaro, M., A. De Biase, and L. Smith . 2013 . Field release of a prospective biological control agent of weeds, Ceratapion basicorne , to evaluate potential risk to a nontarget crop. Biol. Control. 64: 305–314.

Crooks, J. 2011 . Lag times, pp. 404–410. In Simberloff, D., Rejmánek, M. (eds.), Encycl. Biol. Invasions. University of California Press.

Cullen, R., K. D. Warner, M. Jonsson, and S. D. Wratten . 2008 . Economics and adoption of conservation biological control. Biol. Control. 45: 272–280.

Daehler, C. C. 1998 . The taxonomic distribution of invasive angiosperm plants: ecological insights and comparison to agricultural weeds. Biol. Conserv. 84: 167–

132 180.

DAISIE Consortium . 2016 . Asclepias syriaca fact sheet. Springer (http://www.europe- aliens.org/speciesFactsheet.do?speciesId=17716#).

Dalin, P., and C. Bjorkman . 2006 . Chapter four: Native insects colonizing introduced tree species — patterns and potential risks, pp. 63–77. In Paine, T.D. (ed.), Invasive For. Insects, Introd. For. Trees Altered Ecosyst. Springer. Dordrecht, The Netherlands.

Dalosto, M. M., A. V. Palaoro, C. Souty-Grosset, S. L. D. S. Bueno, T. G. Loureiro, M. P. Almerão, P. B. de Araujo, and S. Santos . 2015 . One step ahead of the enemy: Investigating aggressive interactions between invasive and native crayfish before the contact in nature. Biol. Invasions. 17: 3503–3515.

Davis, S., D. Cipollini . 2014 . Do mothers always know best? Oviposition mistakes and resulting failure of Pieris virginiensis on Alliaria petiolata , a novel, toxic host. Biol. Invasions. 16: 1941–1950.

Decker, M. D. 2000 . Gene Flow and Genetic Structure in a Spatially-Structured Beetle Population. (Doctoral thesis). University of Minnesota. Minneapolis, MN, USA. deJonge, R. B., R. S. Bourchier, and S. M. Smith . 2017 . Initial response by a native beetle, Chrysochus auratus (Coleoptera: Chrysomelidae), to a novel introduced host- plant, Vincetoxicum rossicum (Gentianales: Apocynaceae). Environ. Entomol. 46: 617–625.

Delbac, L., P. Lecharpentier, and D. Thiery . 2010 . Larval instars determination for the European grapevine moth (Lepidoptera: Tortricidae) based on the frequency distribution of head-capsule widths. Crop Prot. 29: 623–630.

Dethier, V. 1982 . Mechanism of host-plant recognition. Entomol. exp. appl. 31: 49–56.

Diaz R, V. Manrique, K. Hibbard, A. Fox, A. Roda, D. Gandolfo, F. McKay, J. Medal, S. Hight, W. Overholt. 2014. Successful biological control of tropical soda apple (Solanales: Solanaceae) in Florida: a review of key program components. Fla Entomol 1: 179–190

Dickinson, J. L. 1995 . Trade-offs between postcopulatory riding and mate location in the blue milkweed beetle. Behav. Ecol. 6: 280–286.

Dietze, M. C. 2017 . Prediction in ecology: A first-principles framework: A. Ecol. Appl. 27: 2048–2060.

DiTommaso, A., F. M. Lawlor, and S. J. Darbyshire . 2005 . The biology of invasive

133 alien plants in Canada. 2. Cynanchum rossicum (Kleopow) Borhidi [= Vincetoxicum rossicum (Kleopow) Barbar.] and Cynanchum louiseae (L.) Kartesz & Gandhi [= Vincetoxicum nigrum (L.) Moench]. Can. J. Plant Sci. 85: 243–263.

DiTommaso, A., and J. E. Losey . 2003 . Oviposition preference and larval performance of monarch butterflies ( Danaus plexippus ) on two invasive swallow-wort species. Entomol. Exp. Appl. 108: 205–209.

Dobler, S., S. Dalla, V. Wagschal, and A. Agrawal . 2012 . Community-wide convergent evolution in insect adaptation to toxic cardenolides by substitutions in the Na,K- ATPase. Proc. Natl. Acad. Sci. U. S. A. 109: 13040–5.

Dobler, S., D. Daloze, and J. M. Pasteels . 1998 . Sequestration of plant compounds in a leaf beetle’s defensive secretion: cardenolides in Chrysochus. Chemoecology. 8: 111–118.

Dobler, S., and B. D. Farrell . 1999 . Host use evolution in Chrysochus milkweed beetles: evidence from behaviour, population genetics and phylogeny. Mol. Ecol. 8: 1297– 307.

Dobler, S., G. Petschenka, V. Wagschal, and L. Flacht . 2015 . Convergent adaptive evolution - how insects master the challenge of cardiac glycoside-containing host plants. Entomol. Exp. Appl. 157: 30–39.

Dostál, P., E. Allan, W. Dawson, M. van Kleunen, I. Bartish, and M. Fischer . 2013 . Enemy damage of exotic plant species is similar to that of natives and increases with productivity. J. Ecol. 101: 388–399.

Douglass, C. H., L. A. Weston, and A. Ditommaso . 2009 . Black and pale swallow-wort (Vincetoxicum nigrum and V. rossicum ): The biology and ecology of two perennial, exotic and invasive vines, pp. 261–277. In Inderjit (ed.), Manag. Invasive Weeds. Springer Science and Business Media.

Douglass, C. H., L. A. Weston, and D. Wolfe . 2011 . Phytotoxicity and potential allelopathy in pale ( Cynanchum rossicum ) and black swallowwort ( C. nigrum ). Invasive Plant Sci. Manag. 4: 133–141.

Doussourd D. E., and Eisner T. 1987 . Vein-cutting behavior: insect counterploy to the latex defense of plants. Science (80-. ). 237: 898–901.

Elton, C. A. 1958 . The ecology of invasions by animals and plants. Methuen, London.

Environment Canada. 2009. New substances notification regulations (organisms), Legislation Revision and Consolidation . Retrieved from the Justice Laws website: http://laws-lois.justice.gc.ca

134 Ernst, C. M., and N. Cappuccino . 2005 . The effect of an invasive alien vine, Vincetoxicum rossicum (Asclepiadaceae), on populations in Ontario old fields. Biol. Invasions. 7: 417–425.

Ewel, J. J. 1986 . Invasibility: Lessons from South Florida, pp. 214–30. In Mooney, H.A., Drake, J.A. (eds.), Ecol. Biol. Invasions North Am. Hawaii. Springer-Verlag, New York.

Felker-Quinn, E., J. A. Schweitzer, and J. K. Bailey . 2013 . Meta-analysis reveals evolution in invasive plant species but little support for Evolution of Increased Competitive Ability (EICA). Ecol. Evol. 3: 739–751.

Flockhart, D. T. T., L. I. Wassenaar, T. G. Martin, K. a Hobson, M. B. Wunder, and D. R. Norris . 2013 . Tracking multi-generational colonization of the breeding grounds by monarch butterflies in eastern North America. Proc. Biol. Sci. 280: 20131087.

Fox, J., and S. Weisberg . 2016 . car: Companion to applied regression. R Packag. version 2.1-2.

Frankham, R. 2005 . Genetics and extinction. Biol. Conserv. 126: 131–140.

Freckleton, R. P., P. H. Harvey, and M. D. Pagel . 2002 . Phylogenetic analysis and comparative data: a test and review of evidence. Am. Nat. 160: 712–726.

Fritts, T. H., and G. H. Rodda . 1998 . The role of introduced species in the degradation of island ecosystems:a case history of Guam. Annu. Rev. Ecol. Syst. 29: 113–140.

Futuyma, D. J., and C. Mitter . 1996 . Insect-plant interactions: the evolution of component communities. Philisophica Trans. Biol. Sci. 351: 1361–1366.

Gandhi, K. J. K., and D. A. Herms . 2010 . Direct and indirect effects of alien insect herbivores on ecological processes and interactions in forests of eastern North America. Biol. Invasions. 12: 389–405.

Gandolfo, D., F. McKay, J. Medal, J. Cuda. 2007 Open-field host specificity test of Gratiana boliviana (Coleoptera: Chrysomelidae), a biological control agent of tropical soda apple (Solanaceae) in the United States. Fla Entomol 90:223–228

Gassmann, A., M. Bennett, M. Penic, and A. M. Leroux . 2012 . Biological control of swallow-worts, Vincetoxicum rossicum and V. nigrum - Annual Report 2011 of CABI Europe. Delemont, CH.

Gassmann, A., A. Delaplace, D. Nguyen, and A. S. Weed . 2010 . Evaluating the potential for biological control of swallow-worts, Vincetoxicum nigrum and V.

135 rossicum. Joint Annual Report 2009 of CABI Europe-Switzerland and University of Rhode Island. Delemont, CH.

Gassmann, A., A. M. Leroux, M. Bennett, and M. Penic . 2011 . Weed Biological Control: Quarterly Report 30 June 2011 of CABI Europe. Delemont, CH.

Gassmann, A., A. M. Leroux, and A. S. Weed . 2009 . Evaluating the potential for biological control of swallow-worts, V incetoxicum nigrum and V. rossicum. Annual Report 2008 of CABI Europe. Delemont, CH.

Gassmann, A., A. Levy, T. Tran, and D. J. Futuyma . 2006 . Adaptations of an insect to a novel host plant: A phylogenetic approach. Funct. Ecol. 20: 478–485.

Gleditsch, J. M., and T. A. Carlo . 2011 . Fruit quantity of invasive shrubs predicts the abundance of common native avian frugivores in central Pennsylvania. Divers. Distrib. 17: 244–253.

Gleditsch, J. M., and T. A. Carlo . 2014 . Living with aliens: Effects of invasive shrub honeysuckles on avian nesting. PLoS One. 9:1-8.

Goryacheva, I., A. Blekhman, B. Andrianov, and I. Zakharov . 2017 . Heritable bacterial endosymbionts in native and invasive populations of Harmonia axyridis. Biol. Invasions. 19: 493–502.

Gratton, C., and R. F. Denno . 2005 . Restoration of arthropod assemblages in a Spartina salt marsh following removal of the invasive plant Phragmites australis . Restor. Ecol. 13: 358–372.

Graves, S. D., and A. M. Shapiro . 2003 . Exotics as host plants of the California butterfly fauna. Biol. Conserv. 110: 413–433.

Groenteman, R., S. V. Fowler, and J. J. Sullivan . 2011 . St. John’s wort beetles would not have been introduced to New Zealand now: A retrospective host range test of New Zealand’s most successful weed biocontrol agents. Biol. Control. 57: 50–58.

Hajek, A. E. 2004 . Natural Enemies: An Introduction to Biological Control. Cambridge University Press, Cambridge, UK.

Hajek, A. E., B. P. Hurley, M. Kenis, J. R. Garnas, S. J. Bush, M. J. Wingfield, J. C. van Lenteren, and M. J. W. Cock . 2016 . Exotic biological control agents: A solution or contribution to arthropod invasions? Biol. Invasions. 18: 953–969.

Haribal, M., and J. A. A. Renwick . 1998 . Identification and distribution of oviposition stimulants for monarch butterflies in hosts and nonhosts. J. Chem. Ecol. 24: 891– 905.

136 Harvey, J. A., A. Biere, T. Fortuna, L. E. M. Vet, T. Engelkes, E. Morriën, R. Gols, K. Verhoeven, H. Vogel, M. Macel, H. M. Heidel-Fischer, K. Schramm, and W. H. Putten . 2010 . Ecological fits, mis-fits and lotteries involving insect herbivores on the invasive plant, Bunias orientalis . Biol. Invasions. 12: 3045–3059.

Hatch, M. 1953 . The beetles of the Pacific Northwest. Washington University Press. Seattle, USA. pp. 157-239.

Herlihy, M. V., R. G. Van Driesche, and D. L. Wagner . 2014 . Persistence in Massachusetts of the veined white butterfly due to use of the invasive form of cuckoo flower. Biol. Invasions. 16: 2713–2724.

Higuti, J., and K. Martens . 2016 . Invasive South American floating plants are a successful substrate for native Central African pleuston. Biol. Invasions. 18: 1191– 1201.

Hinz, H. L., M. Schwarzländer, A. Gassmann, and R. S. Bourchier . 2014 . Successes we may not have had: A retrospective analysis of selected weed biological control agents in the United States. Invasive Plant Sci. Manag. 7: 565–579.

Hoffmann, J. H., F. A. C. Impson, and C. R. Volchansky . 2002 . Biological control of cactus weeds: implications of hybridization between control agent biotypes. J. Appl. Ecol. 39: 900–908.

Howarth, F. G. 1983 . Classical biocontrol: Panacea or pandora’s box. Proceedings, Hawaiian Entomolgical Soc. 24: 239–244.

Hufbauer, R. A., and G. K. Roderick . 2005 . Microevolution in biological control: Mechanisms, patterns, and processes. Biol. Control. 35: 227–239.

Jaenike, J. 1990 . Host specialization in phytophagous insects. Annu. Rev. Ecol. Syst. 21: 243–73.

Janzen, D. H. 1985 . On Ecological Fitting. Oikos. 45: 308–310.

Jeschke, J., L. Gómez Aparicio, S. Haider, T. Heger, C. Lortie, P. Pyšek, and D. Strayer . 2012 . Support for major hypotheses in invasion biology is uneven and declining. NeoBiota. 14: 1–20.

Jobin, A., U. Schaffner, and W. Nentwig . 1996 . The structure of the phytophagous insect fauna on the introduced weed Solidago altissima in Switzerland. Entomol. Exp. Appl. 79: 33–42.

Jolivet, P., and K. Verma . 2008 . – a widely distributed and much diversified subfamily of leaf beetles (Coleoptera, Chrysomelidae). Terr. Arthropod

137 Rev. 1: 3–37.

Karban, R., A. A. Agrawal, and M. Mangel . 1997 . The benefits of induced defenses against herbivores. Ecology. 78: 1351-1355.

Keane, R. M., and M. J. Crawley . 2002 . Exotic plant invasions and the enemy release hypothesis. Trends Ecol. Evol. 17: 164–170.

Keeler, M. S., and F. S. Chew . 2008 . Escaping an evolutionary trap: preference and performance of a native insect on an exotic invasive host. Oecologia. 156: 559–68.

Keller, R., J. Geist, J. Jeschke, and I. Kuhn . 2011 . Invasive species in Europe: Ecology, status, and policy. Environ. Sci. Eur. 23: 23.

Kirk, S., and S. Belt . 2011 . Swamp milkweed plant fact sheet. USDA-NRCS. Beltsville, MD. Available from: https://plants.usda.gov/factsheet/pdf/fs_asin.pdf

Klinken, R. D. van 2000 . Host specificity testing: Why do we do it and how we can do it better, pp. 54–68. In Van Driesche, R.L., Heard, T., McClay, A., Reardon, R. (eds.), Host specif. test. exot. arthropod biol. control agents biol. basis improv. safety. X Int. Symp. Biol. Control ofWeeds, Bozeman, Mont. USA July 4-14, 1999. USDA- Forest Service, West Virginia, USA.

Kraft, N. J. B., P. B. Adler, O. Godoy, E. C. James, S. Fuller, and J. M. Levine . 2015 . Community assembly, coexistence and the environmental filtering metaphor. Funct. Ecol. 29: 592–599.

Kramer, C. Y. 1956 . Extension of multiple range tests to group means with unequal number of replications. Biometrics. 12: 307–310.

Labeyrie, E., and S. Dobler . 2004 . Molecular adaptation of Chrysochus leaf beetles to toxic compounds in their food plants. Mol. Biol. Evol. 21: 218–21.

Lankau, R. A., W. E. Rogers, and E. Siemann . 2004 . Constraints on the utilisation of the invasive Chinese tallow tree Sapium sebiferum by generalist native herbivores in coastal prairies. Ecol. Entomol. 29: 66–75.

Lawlor, F. M. 2000 . Herbicidal treatment of the invasive plant Cynanchum rossicum : and experimental post control restoration of infested sites. M.S. thesis. Syracuse, NY: State University of New York.

Lawlor, F. M., and D. J. Raynal . 2002 . Response of swallow-wort to herbicides. Weed Sci. 50: 179–185.

Lawton, J. H., and D. Schroder . 1977 . Effects of plant type, size of geographical range and taxonomic isolation on number of insect species associated with British plants.

138 Nature. 265: 137–140.

Louda, S. M., R. W. Pemberton, M. T. Johnson, and P. A. Follett . 2003 . Nontarget effects: The Achilles’ heel of biological control? Retrospective analyses to reduce risk associated with biocontrol introductions. Annu. Rev. Entomol. 48: 365–96.

Marohasy, J. 1996 . Host shifts in biological weed control: real problems, semantic difficulties or poor science? Int. J. Pest Manag. 42: 71–75.

Marohasy, J. 1998 . The design and interpretation of host-specificity\ntests for weed biological control with particular reference to insect behaviour. Biocontrol News Inf. 19: 13N–20N.

Maron, J. L., and M. Vila . 2001 . When do herbivores affect plant invasion? Evidence for the natural enemies and biotic resistance hypotheses. Oikos. 95: 361–373.

Mathenge, C. W., P. Holford, J. H. Hoffmann, H. G. Zimmermann, R. N. Spooner- Hart, and G. A. C. Beattie . 2010 . Hybridization between Dactylopius tomentosus (Hemiptera: Dactylopiidae) biotypes and its effects on host specificity. Bull. Entomol. Res. 100: 331–8.

Mason, P. G., R. A. De Clerck-Floate, B. Gallant, D. R. Gillespie, K. Floate, R. Bourchier, H. Douglas, C. Vincent, and G. Boivin . 2017 . Guide for the importation and release of arthropod biological control agents. Available from: www.agr.gc.ca

Mattila, H. R., and G. W. Otis . 2003 . A comparison of the host preference of monarch butterflies ( Danaus plexippus ) for milkweed ( Asclepias syriaca ) over dog-strangler vine ( Vincetoxicum rossicum ). Netherlands Entomol. Soc. 193–199.

McClay, A. S., and J. K. Balciunas . 2005 . The role of pre-release efficacy assessment in selecting classical biological control agents for weeds - Applying the Anna Karenina principle. Biol. Control. 35: 197–207.

McDonald, J. H. 2014 . Handbook of biological statistics. 3 rd ed. Sparky house publishing. Baltinmore, Maryland. Available from: http://www.biostathandbook.com/

Mcevoy, P. B., K. M. Higgs, E. M. Coombs, E. Karaçetin, and L. Ann Starcevich . 2012 . Evolving while invading: Rapid adaptive evolution in juvenile development time for a biological control organism colonizing a high-elevation environment. Evol. Appl. 5: 524–536.

Medal J, W. Overholt, P. Stansly, A. Roda, L, Osborne, K. Hibbard, R. Gaskalla, E. Burns, J. Chong, B. Sellers, S. Hight, J. Cuda, M. Vitorino, E. Bredow, J. Pedrosa-Macedo, C. Wikler. 2008. pp. 389–393 In Julien, M.H., Sforza, R., Bon,

139 M.C., Evans, H.C., Hatcher, P.E., Hinz, H.L., Rector, B.G. (eds.), Proc. XII Int. Symp. Biol. Control Weeds. CAB International, International Wallingford, UK.

Medal J., D. Sudbrink, D. Gandolfo, D. Ohashi, J. Cuda. 2002. Gratiana boliviana , a potential biocontrol agent of Solanum viarum : quarantine host-specificity testing in Florida and field surveys in South America. BioControl 47:445–461.

Messing, R. H., and M. G. Wright . 2006 . Biological control of invasive species: solution or pollution? Front. Ecol. Environ. 4: 132–140.

Milbrath, L. R. 2010 . Phytophagous arthropods of invasive swallow-wort vines (Vincetoxicum spp.) in New York. Environ. Entomol. 39: 68–78.

Milbrath, L. R., and J. Biazzo . 2012 . Development and reproduction of the foxglove aphid (Hemiptera: Aphididae) on invasive swallow-worts ( Vincetoxicum spp.). Environ. Entomol. 41: 665–668.

Milbrath, L. R., and J. Biazzo . 2016 . Impact of the defoliating moth Hypena opulenta on invasive swallow-worts ( Vincetoxicum species) under different light environments. Biol. Control. 97: 1–12.

Milbrath, L. R., A. DiTommaso, J. Biazzo, and S. H. Morris . 2016. Tolerance of swallowworts ( Vincetoxicum spp.) to multiple years of artificial defoliation and clipping. Invasive Plant Sci. Manag. 9: 1–11.

Mogg, C., P. Petit, N. Cappuccino, T. Durst, C. McKague, M. Foster, J. E. Yack, J. T. Arnason, and M. L. Smith . 2008 . Tests of the antibiotic properties of the invasive vine Vincetoxicum rossicum against bacteria, fungi and insects. Biochem. Syst. Ecol. 36: 383–391.

Morton, T. A. L., A. Thorn, J. M. Reed, R. G. Van Driesche, R. A. Casagrande, and F. S. Chew . 2015 . Modeling the decline and potential recovery of a native butterfly following serial invasions by exotic species. Biol. Invasions. 17: 1683–1695.

Norbury, G., A. Byrom, R. Pech, J. Smith, D. Clarke, D. Anderson, and G. Forrester . 2013 . Invasive mammals and habitat modification interact to generate unforeseen outcomes for indigenous fauna. Ecol. Appl. 23: 1707–21.

Novak, S. J., and R. N. Mack . 2005 . Genetic bottlenecks in alien plant species: influences of mating systems and introduction dynamics. In Sax, D.F., Stachowicz, J.J., Gaines, S.D. (eds.), Species Invasions Insights into Ecol. Evol. Biogeogr. NRC, Sinauer, Sunderland, MA, USA. pp. 201–228.

Ødegaard, F., O. H. Diserud, and K. Østbye . 2005 . The importance of plant relatedness for host utilization among phytophagous insects. Ecol. Lett. 8: 612–617.

140 Oksanen, J., F. G. Blanchet, M. Friendly, R. Kindt, P. Legendre, D. McGlinn, P. R. Minchin, R. B. O’Hara, G. L. Simpson, P. Solymos, M. H. Stevens, E. Szoecs, and H. Wagner . 2016 . Vegan: Community ecology package.

Orsucci, M., P. Audiot, A. Pommier, C. Raynaud, B. Ramora, A. Zanetto, D. Bourguet, and R. Streiff . 2016 . Host specialization involving attraction, avoidance and performance, in two phytophagous moth species. J. Evol. Biol. 29: 114–125.

Papaj, D. R., and R. J. Prokopy . 1989 . Ecological and evolutionary aspects of learning in phytophagous insects. Annu. Rev. Entomol. 34: 315–350.

Pauková, Ž., M. Knápeková, and M. Hauptvogl . 2014 . Mapping of alien species of Asclepias syriaca and Fallopia japonica populations in the agricultural landscape. J. Cent. Eur. Agric. 15: 12–22.

Pearse, I. S., and F. Altermatt . 2013 . Predicting novel trophic interactions in a non- native world. Ecol. Lett. 16: 1088–1094.

Pearse, I. S., D. J. Harris, R. Karban, and A. Sih . 2013 . Predicting novel herbivore- plant interactions. Oikos. 122: 1554–1564.

Peterson, A. T., J. Soberon, R. B. Pearson, R. P. Anderson, E. Martinez-Meyer, M. Nakamura, and M. B. Araujo . 2011 . Ecological niches and geographic distributions. Princeton University Press, Princeton and Oxford.

Peterson, M. A., S. Dobler, J. Holland, L. Tantalo, and S. Locke . 2001 . Behavioral, molecular, and morphological evidence for a hybrid zone between Chrysochus auratus and C . cobaltinus (Coleoptera: Chrysomelidae). Entomol. Soc. Am. 94: 1– 9.

Peterson, M. A., B. Honchack, S. Locke, T. Beeman, J. Mendoza, J. Green, K. Buckingham, M. A. White, and K. J. Monsen . 2005 . Relative abundance and the species-specific reinforcement of male mating preference in the Chrysochus (Coleoptera: Chrysomelidae) hybrid zone. Evolution. 59: 2639–2655.

Peterson, M. A., K. J. Monsen, H. Pedersen, T. McFarland, and J. Bearden . 2005 . Direct and indirect analysis of the fitness of Chrysochus (Coleoptera: Chrysomelidae) hybrids. Biol. J. Linn. Soc. 84: 273–286.

Pfammatter, J. A., A. Krause, and K. F. Raffa . 2015 . Evaluating predators and competitors in Wisconsin red pine forests for attraction to mountain pine beetle pheromones for anticipatory biological control. Environ. Entomol. 44: 1161–1171.

Pimentel, D., R. Zuniga, and D. Morrison . 2005 . Update on the environmental and economic costs associated with alien-invasive species in the United States. Ecol. Econ. 52: 273–288.

141

Pringle, J. S. 1973 . The spread of Vincetoxicum species (Asclepiadaceae) in Ontario. Can. Field-Naturalist. 87: 27–33.

Puliafico, K. P., M. Schwarzländer, B. L. Harmon, and H. L. Hinz . 2008 . Effect of generalist insect herbivores on introduced Lepidium draba (Brassicaceae): Implications for the enemy release hypothesis. J. Appl. Entomol. 132: 519–529.

Puliafico, K. P., M. Schwarzländer, H. L. Hinz, and B. L. Harmon . 2008 . The use of surrogate herbivores for the pre-release efficacy screening of biological control agents of Lepidium draba , pp. 429–434. In Julien, M.H., Sforza, R., Bon, M.C., Evans, H.C., Hatcher, P.E., Hinz, H.L., Rector, B.G. (eds.), Proc. XII Int. Symp. Biol. Control Weeds. CAB International, International Wallingford, UK.

R Core Team . 2015 . R: A language and environment for statistical computing. R foundation for statistical computing, Vienna, . (http://www.r-project.org/.).

R Core Team . 2016 . R: A language and environment for statistical computing. R foundation for statistical computing, Vienna, Austria. (http://www.r-project.org/.).

Rhymer, J. M., and D. Simberloff . 1996 . Extinction by hybridization and introgression. Annu. Rev. Ecol. Syst. 27: 83–109.

Richardson, D. M., N. Allsopp, C. M. D’Antonio, S. J. Milton, and M. Rejmánek . 2000 . Plant invasions -- the role of mutualisms. Biol. Rev. Camb. Philos. Soc. 75: 65–93.

Richardson, D. M., and P. Pyšek . 2006 . Plant invasions: Merging the concepts of species invasiveness and community invasibility. Prog. Phys. Geogr. 30: 409–431.

Richardson, D. M., P. P. Y. S. Pysek, M. Rejmánek, M. G. Barbour, F. D. Panetta, and C. J. West . 2000 . Naturalization and invasion of alien plants concepts and definitions. Divers. Distrib. 6: 93–107.

Rodewald, A. D. 2012 . Spreading messages about invasives. Divers. Distrib. 18: 97–99.

Rodewald, A. D., D. P. Shustack, and L. E. Hitchcock . 2010 . Exotic shrubs as ephemeral ecological traps for nesting birds. Biol. Invasions. 12: 33–39.

Sady, M. B. 1994 . Survey of the blue milkweed beetle, Chrysochus colbaltinus LeConte (Coleoptera : Chrysomelidae ), in Western Nevada. Coleopt. Bull. 48: 299.

Sanderson, L. A., and P. M. Antunes . 2013 . The exotic invasive plant Vincetoxicum rossicum is a strong competitor even outside its current realized climatic temperature range. NeoBiota. 16: 1–15.

142 Santana, A. F. K., and F. S. Zucoloto . 2011 . Influence of previous experience on the preference, food utilization and performance of Ascia monuste orseis wild larvae (Godart) ( Lepidoptera Pieridae ) for three different hosts. Neotrop. Entomol. 40: 631–638.

Schaffner, U. 2001 . Host range testing of insects for biological weed control: How can it be better interpreted? Bioscience. 51: 951–959.

Schirmel, J., M. Bundschuh, M. H. Entling, I. Kowarik, and S. Buchholz . 2016 . Impacts of invasive plants on resident animals across ecosystems, taxa, and feeding types: A global assessment. Glob. Chang. Biol. 22: 594–603.

Schlaepfer, M. A., M. C. Runge, and P. W. Sherman . 2002 . Ecological and evolutionary traps. Trends Ecol. Evol. 17: 474–480.

Schlaepfer, M. A., P. W. Sherman, B. Blossey, and M. C. Runge . 2005 . Introduced species as evolutionary traps. Ecol. Lett. 8: 241–246.

Schmitt, M. 2011 . Types of geographical distribution of leaf beetles (Chrysomelidae) in Central Europe. Zookeys. 157: 131–158.

Schwarz, D., B. M. Matta, N. L. Shakir-Botteri, and B. A. Mcpheron . 2005 . Host shift to an invasive plant triggers rapid animal hybrid speciation. Nature. 436: 546– 549.

Scoggan, H. J. 1979 . The flora of Canada. Part 4 Dicotyledoneae (Loasaceae to Compositae). In Nat. Mus. Nat. Sci. Publ. Bot., 7. pp. 117–1711.

Seabloom, E. W., W. S. Harpole, O. J. Reichman, and D. Tilman . 2003 . Invasion, competitive dominance, and resource use by exotic and native California grassland species. Proc. Natl. Acad. Sci. U. S. A. 100: 13384–9.

Seehausen, O. 2004 . Hybridization and adaptive radiation. Trends Ecol. Evol. 19: 198– 207.

Sforza, R. 2011 . Testing agents for the biological control of swallow worts (Vincetoxicum spp.) in North America Phase II USDA-ARS project 2009-2010.

Sheeley, S. E., and D. J. Raynal . 1996 . The distribution and status of species of Vincetoxicum in Eastern North America. Bull. Torrey Bot. Club. 123: 148–156.

Sheldon, S. P., and R. P. J. Creed . 2003 . The effect of a native biological control agent for Eurasian watermilfoil on six North American watermilfoils. Aquat. Bot. 76: 259–265.

Sheppard, A. W., R. D. van Klinken, and T. A. Heard. 2005 . Scientific advances in

143 the analysis of direct risks of weed biological control agents to nontarget plants. Biol. Control. 35: 215–226.

Siemann, E., W. E. Rogers, and S. J. Dewalt . 2006 . Rapid adaptation of insect herbivores to an invasive plant. Proc. Biol. Sci. 273: 2763–9.

Simberloff, D. 2011 . How common are invasion-induced ecosystem impacts? Biol. Invasions. 13: 1255–1268.

Simberloff, D., and B. Von Holle . 1999 . Positive interactions of nonindigenous species: invasional meltdown? Biol. Invasions. 1: 21–32.

Simberloff, D., and P. Stiling . 1996 . How risky is biological control? Ecology. 77: 1965–1974.

Slippers, B., J. Stenlid, and M. J. Wingfield . 2005 . Emerging pathogens: Fungal host jumps following anthropogenic introduction. Trends Ecol. Evol. 20: 420–421.

Smith, L., M. Cristofaro, C. Tronci, and R. Hayat . 1999 . Refining methods to improve pre-release risk assessment of prospective agents: the case of Ceratapion basicorne . XII Int. Symp. Biol. Control Weeds. 321–327.

Smith, L. L., A. Ditommaso, J. Lehmann, and S. Greipsson . 2006 . Growth and reproductive potential of the invasive exotic vine Vincetoxicum rossicum in northern New York State. Can. J. Bot. 84: 1771–1780.

Solarz, S. L., and R. M. Newman . 2001 . Variation in hostplant preference and performance by the milfoil weevil, Euhrychiopsis lecontei Dietz, exposed to native and exotic watermilfoils. Oecologia. 126: 66–75.

Soler, R., A. Pineda, Y. Li, C. Ponzio, J. J. A van Loon, B. T. Weldegergis, and M. Dicke . 2012 . Neonates know better than their mothers when selecting a host plant. Oikos. 121: 1923–1934.

Strauss, S., J. A Lau, and S. P. Carroll . 2006 . Evolutionary responses of natives to introduced species: what do introductions tell us about natural communities? Ecol. Lett. 9: 357–74.

Sudakaran, S., F. Retz, Y. Kikuchi, C. Kost, and M. Kaltenpoth . 2015 . Evolutionary transition in symbiotic syndromes enabled diversification of phytophagous insects on an imbalanced diet. ISME J. 9: 1–18.

Sunny, A., S. Diwakar, and G. P. Sharma . 2015 . Native insects and invasive plants encounters. Arthropod. Plant. Interact. 9: 323–331.

Sutherland, S. 2004 . What makes a weed a weed: Life history traits of native and exotic

144 plants in the USA. Oecologia. 141: 24–39.

Tamayo, M., and C. E. Grue . 2004 . Developmental performance of the milfoil weevil (Coleoptera: Curculionidae) on watermilfoils in Washington State. Environ. Entomol. 33: 872–880.

Tewksbury, L., R. A. Casagrande, and A. Gassmann . 2002 . 16 Swallow-worts, pp. 209–216. In Van Drieshce, R. et. al. (ed.), Biol. Control Invasive Plants East. United States. USDA Forest Service Publication.

Tukey, J. 1953 . The Problem of Multiple Comparisons. unpublished manuscript reprinted in: The collected works of John W. Tukey, Volume 8, 1994 H. I. Braun (Ed.), Chapman and Hall, New York.

Uesugi, A. 2008 . Evolution of host avoidance in a leafmining fly, Amauromyza flavifrons . Entomol. Exp. Appl. 128: 398–402.

Van Driesche, R. G. 2011 . The role of biological control in wildlands. BioControl. 57: 131–137.

Van Driesche, R. G., T. Murray, and R. Reardon . 2004 . Assessing host ranges for parasitoids and predators used for classical biological control: a guide to best practice. Morgantown, WV: Forest Health Technology Enterprise Team, USDA- Forest Service.

Van Hezewijk, B. H., R. A. De Clerck-Floate, and J. R. Moyer . 2008 . Effect of nitrogen on the preference and performance of a biological control agent for an invasive plant. Biol. Control. 46: 332–340.

Vitousek, P. M., J. D. Aber, R. W. Howarth, G. E. Likens, P. A. Matson, D. W. Schindler, W. H. Schlesinger, and D. G. Tilman . 1997 . Human alteration of the global nitrogen cycle: Sources and consequences. Ecol. Appl. 7: 737–750.

Wapshere, A. J. 1974 . A strategy for evaluating the safety of organisms for biological weed control. Ann. App. Biol. 77: 20–211.

Weed, A. S. 2010 . Biology and ecology of European natural enemies of swallow-worts (Vincetoxicum ) and the potential for biological control (Doctoral thesis). University of Rhode Island. Kingston, RI, USA.

Weed, A. S., and R. A. Casagrande . 2011 . Evaluation of host range and larval feeding impact of Chrysolina aurichalcea asclepiadis (Villa): considerations for biological control of Vincetoxicum in North America. Environ. Entomol. 40: 1427–1436.

Weed, A. S., A. Gassmann, and R. A. Casagrande . 2011 . Effects of leaf and root herbivory by potential insect biological control agents on the performance of

145 invasive Vincetoxicum spp. Biol. Control. 56: 50–58.

Weed, A. S., A. Gassmann, A. M. Leroux, and R. A. Casagrande . 2011 . Performance of potential European biological control agents of Vincetoxicum spp. with notes on their distribution. J. Appl. Entomol. 135: 700–713.

Weiss, H. B., and E. West . 1921 . Notes on the insects of the spreading dogbane, Apocynum androsaemifolium L. with a description of a new dogbane midge, by Dr. E. P. Felt. Can. Entomol. 53: 147.

Weston, L. A., J. N. Barney, and A. DiTommaso . 2005 . A review of the biology and ecology of three invasive perennials in New York State: Japanese knotweed (Polygonum cuspidatum ), mugwort ( Artemisia vulgaris ) and pale swallow-wort (Vincetoxicum rossicum ). Plant Soil. 277: 53–69.

White, D. W., and E. W. Stiles . 1992 . Bird dispersal of fruits of species introduced into eastern North-America. Can. J. Bot. Can. Bot. 70: 1689–1696.

Whitney, K. D., and C. A. Gabler . 2008 . Rapid evolution in introduced species, “invasive traits” and recipient communities: Challenges for predicting invasive potential. Divers. Distrib. 14: 569–580.

Wiens, J. J., and C. H. Graham . 2005 . Niche conservatism: Integrating evolution, ecology, and conservation biology. Annu. Rev. Ecol. Evol. Syst. 36: 519–539.

Williams, C. 1991 . Scientific note: Host plant latex and the feeding behaviour of Chrysochus auratus (Coleoptera: Chrysomelidae). Coleopt. Bull. 45: 195–196.

Williams, C. E. 1992 . Movement of the cogbane beetle, Chrysochus auratus (Coleoptera: Chrysomelidae), in a patchy environment. Banisteria. 1: 8–10.

Wingfield, M., B. Slippers, B. Hurley, T. Coutinho, B. Wingfield, and J. Roux . 2008 . Eucalypt pests and diseases: growing threats to plantation productivity. South. For. a J. For. Sci. 70: 139–144.

Winston, R. L., M. Schwarzländer, H. L. Hinz, M. D. Day, M. J. W. Cock, M. H. Julien, and M. Lewis . 2014 . Biological control of weeds: A world catalogue of agents and their target weeds. USDA For. Serv. For. Heal. Technol. Enterp. 5: 1– 848.

146