THE ROLE OF BIOTIC INTERACTIONS IN BIOLOGICAL INVASIONS By Elizabeth H. Schultheis A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Plant Biology Ð Doctor of Philosophy Ecology, Evolutionary Biology and Behavior Ð Dual Major 2015 2 ABSTRACT THE ROLE OF BIOTIC INTERACTIONS IN BIOLOGICAL INVASIONS By Elizabeth H. Schultheis Invasive species are one of the major drivers of biodiversity loss, and it is estimated that invasive species cost billions in damage per year, globally. Given the economic costs and the potential ecological consequences of invasive species, it is importan t to understand how introduced species become integrated into natural communities and the consequences of invasion over longer time scales. To better predict and prevent future invasions , we must identify the mechanisms driving a small proportion of introd uced species to become invasive. Biotic interactions , such as herbivores and competitors , are among the major drivers of plant community structure and population dynamics. Release from antagonistic biotic interactions during the process of introduction may drive the explosive population growth rates of invasive species when they are transported to new ranges. However, subsequent acquisition of novel biotic interactions in the introduced range could explain why so many of the plants introduced around the wo rld fail to become invasive . The Enemy R elease Hypothesis (ERH) is one of the leading hypotheses explaining the success of invasive species and stat es that species once controlled by antagonistic biotic interactions in their native range will be able to reach high abundances once released from this control. My dissertation research takes an integrative approach to rigorously test the oft -cited Enemy Release Hypothesis. Using field experiments including over 50 plant species, and a meta-analysis of the published literature to test ERH across a wider range of environments and species , I address four main questions: 3 1. Do invasive, noninvasive exotic, and native species experience different amounts of damage from enemies in the introduced range ? 2. Does enemy release result in increased performance for invasive spe cies compared to native and non invasive exotic species? 3. Is enemy release lost with increased residence time and geographic spread in the introduced range ? 4. Does tolerance to enemy damage or competitive ability drive invasiveness? I found no evidence suggesting enemy release is a general mechanism contributing to invasiveness. Invasive species received the most damage from enemies and were equally affected by the presence of antagonistic biotic interactions, compared to native and noninvasive exotic species. Invasive species were no more tolerant to enemy damage than were native or noninvasive exotic species. For both invasive and noninvasive introduced plants, dam age and the performance effects of that damage, increased with longer residence times and larger areas of spread in the introduced range . Our results show that invasive and exotic species fail to escape enemies, particularly over longer temporal and larger spatial scales. Key differences between introduced species that become invasive and those that do not may be the formation of successful mutualisms in the introduced range, and release from competition compared to native and noninvasive exotic species in the introduced range. 4 Copyright by ELIZABETH H. SCHULTHEIS 2015 v To my fianc” , Jim. I canÕt wait to spend our lives together on the shores of Lake Michigan. vi ACKNOWLEDGMENTS This work was funded by NSF DDIG -1210436, The Hanes Trust, Michigan Botanical Foundation, G.H. Lauff Research Awards, and Kathryn Porter Graduate Fellowships from the W.K. Kellogg Biological Station (KBS). I would like to thank members of the KBS community and undergraduate, high school , and teacher volunteers and researchers who helped me construct experimental plots an d plant over 10,000 seedlings. Special thanks to J.P. Springer, D. Williams, T. Van Doornik, D.J. MacGuigan, G. Stewart, C. Portales -Reyes , D. Cronkright, M. Herman, M. Tillotson, K. Davis, S. Mooney, J. Sharp, T. Suwa, K. Keller , S. Magnoli , R. Prunier , C. Kremer , J. Li, C. Monroe, M. McKenzie, and J. Jubenville . Thank you also to M. Hammond for logistical support and help collecting seeds, S. Bassett for site preparation, and T. Bassett for help choosing experimental species. Thank you to all members of the J.A. Lau laborator y and University of Denver Organismal Biologists Group who provided useful feedback on earlier drafts of my dissertation . I would especially like to thank my committee, J.A. Lau, G.G. Mittelbach, R.K. Kobe, and D. Schemske , whose advice greatly improved my research over the years. vii TABLE OF CONTENTS LIST OF TABLESÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...x LIST OF FIGURESÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..xiii CHAPTER 1: INTRODUCTIONÉ ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ1 Background ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..1 Organization of the Dissertation ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..2 Chapter 2ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..2 Chapter 3ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..3 Chapter 4ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..3 Chapter 5ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..4 CHAPTER 2 : NO RELEASE FOR THE WICKED: ENEMY RELEA SE IS DYNAMIC AND NOT ASSOCIATED WITH IN VASIVENESSÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.. 5 IntroductionÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..5 Dynamic Invasions ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...8 Methods ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..10 Study Species ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.10 Experimental Design ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..12 Residence Time and Spread Data ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..14 Phylogenetic Reconstruction ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.15 Statistical Analysis ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.16 Results ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ18 Discussion ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..23 Conclusion ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.28 CHAPTER 3 : PERFORMANCE CONSEQUENCES OF ENEMY RELEASE DEPEND ON INVASION AGE BUT DO NOT EXPLAIN INVASIVENESSÉÉÉÉÉÉÉÉÉÉÉÉ.. 30 Introduction ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ30 Community Complexity and Enemy Release ÉÉÉÉÉÉÉÉÉÉÉÉÉÉ31 Tolerance ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ32 Dynamic Enemy Release ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...33 Methods ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..35 Experimental Design ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..35 Statistical Analysis ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.38 Damage ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..38 Performance ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...39 Tolerance ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ40 TimeÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...41 Results ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ42 Damage ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..42 Performance ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...42 Survival ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..42 viii Vegetative Biomass ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...43 Flower Number and Reproductive Biomass ÉÉÉÉÉÉÉÉÉÉÉ..46 Tolerance ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ46 TimeÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...50 Discussion ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..50 Community Complexity and Enemy Release ÉÉÉÉÉÉÉÉÉÉÉÉÉÉ51 Tolerance ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ54 Dynamic Invasions ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.55 Treatment Effects on Background Community ÉÉÉÉÉÉÉÉÉÉÉÉÉ.56 Conclusion ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.57 CHAPTER 4 : COMPETITIVE ABILITY, NOT TOLERANCE, MAY EXPLAIN SUCCESS OF INTRODUCED PLANTS OVE R NATIVESÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.5 8 Introduction ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ58 Enemy Release and Tolera nce ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...58 Competitive Ability of Invasive Species ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...60 Methods ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..61 Study Species ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.61 Experimental Design ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..62 Statistical Analysis ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.64 Tolerance and Competitive Response ÉÉÉÉÉÉÉÉÉÉÉÉÉÉ64 Competitive Effects on Elymus canadensis ÉÉÉÉÉÉÉÉÉÉÉ...65 Results ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ66 Tolerance and Competitive Response ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ66 Competitive Effects on Elymus canadensis ÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...67 Discussion ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..69 Competitive Ability of Invasive Species: Response and Effects ÉÉÉÉÉÉ...70 Enemy Release and Tolerance ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...71 Conclusion ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.72 CHAPTER 5: MUTUALISM GAIN AND COMPETITIVE ABILITY, NOT ENEMY RELE ASE, MAY EXPLAIN SUCCESS OF INVASIVE SPE CIES: A META -ANALYSISÉ. 74 Introduction ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ73 Methods ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..78 Literature Search and Data Collection ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...78 Statistical Analysis ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.80 Vote Count ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.83 Results ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ84 Cross Continental Studies Ð Plants ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ84 Native/Introd uced Comparison Studies Ð Plants ÉÉÉÉÉÉÉÉÉÉÉÉ...86 Native/Introduced Comparison Studies Ð Animals ÉÉÉÉÉÉÉÉÉÉÉÉ87 Vote Count ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.88 Discussion ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..90 Species Specific Case Studies and Context Dependency of Biotic Release ÉÉ..92 Interactive and Synergistic Effects of Multiple Biotic Interactions ÉÉÉÉÉ...95 Conclusion ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.96 ix APPENDICESÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...97 Appendix A: Supplemental Tables and Figures for Chapter 2ÉÉÉ ÉÉÉÉÉÉÉ..98 Appendix B: Statistical Methods and Results for Plant Family AnalysisÉÉÉÉÉ...104 Appendix C: List of Experimental S pecies in the 2012 -2014 experimentÉÉÉÉÉ..109 Appendix D: Analysis of B ackground Community ChangesÉÉÉÉÉÉÉÉÉÉ..110 Appendix E: Flow er Number AnalysisÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ112 Appendix F: Hierarchical M eta-analysis ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.116 REFERENCESÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ128 x LIST OF TABLES Table 1 Results from mixed model analysis of variance (ANOVA) testing the fixed effects of plant status (native, noninvasive exotic, invasive), family (Asteraceae, Fabaceae, Poaceae), and mammalian browser removal (fencing treatment, control) on plant performance. Biomass and floral biomass were log transform ed to fit normality assumptions. Binomial (survival) and count (flower number) data were analyzed using general linearized mixed models; all other data were analyzed using linear mixed models. Chi square statistics based on log -likelihood ratio tests are p resented for random factors and general linearized models. Statistically significant (P ! 0.05) effects are in bold ÉÉÉÉÉÉÉÉÉÉÉ43 Table 2 Results from tolerance mixed model analysis of variance (ANOVA) testing the fixed effects of plant status (native, non invasive exotic, invasive) and enemy damage (insect herbivory and mammalian browsing) on plant performance (vegetative biomass, reproductive biomass, and flower number). Vegetative and reproductive biomass and damage data were log transformed to fit normal ity assumptions. Statistically significant (P ! 0.05) effects are in bold ÉÉÉÉÉ..48 Table 3 Results from mixed model analysis of variance (ANOVA) testing the fixed effects of plant status (noninvasive exotic, invasive), insecticide treatment (sprayed, cont rol), fencing treatment (fenced, control), and time on plant performance. Vegetative and reproductive biomass data were log transformed to fit normality assumptions. Chi square statistics based on log -likelihood ratio tests are presented for random factors . Statistically significant (P ! 0.05) effects are in bold ÉÉÉ51 Table 4 List of the 18 experimental species, and one competitor species, used in the experiment, along with their family and status designation. The competitor species, Elymus canadensis , is indicated with an * ÉÉÉÉÉÉÉÉÉÉÉ...62 Table 5 Results from mixed model analysis of variance (ANOVA). Results show the effects of plant status (native, noninvasive exotic, or invasive), family (Asteraceae, Fabaceae, Poaceae), clipping treatment (clipped, con trol), and competition treatment (competitor present, no competition) on experimental plant biomass and height. Statistically significant (P ! 0.05) effects are in bold ÉÉÉÉÉÉÉÉ65 Table 6 Results from mixed model analysis of variance (ANOVA). Results show the effect of competitor status, family, and whether the competitor was clipped for Elymus canadensis biomass and height. Statistically significant (P ! 0.05) effects are in bold. Non -significant interaction terms we re dropped from the final model ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..66 Table 7 Total heterogeneity (Q T) and between -group heterogeneity (Q B) of effect sizes in studies comparing the effects of biotic interactions on native, noninvasive exotic, invasive species performance. A significant Q B indicates that status explained a xi significant portion of the overall variation in effect sizes, meaning that mean effect size of biotic interaction removal differs between invasive, noninvasive, and native species. Significant p -values (p ! 0.05) for QB are shown in bold. Missing cells represent categories with insufficient repl ication for analysis (n < 5)ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ84 Table 8 Summary table of vote count results. For our vote count, significance is determined by statistics reported in the origina l papers. Significant effect sizes are indicated with + and -, where non -signi ficant effects indicated by n.s ÉÉÉÉ.89 Table A1 List of the 61 species planted into the 2011 and 2012 -2013 common gardens. Species are color coded by plant status: native (white ), exotic (gray), and invasive (black). In the columns for year, presence of a particular species is indicated with an ÔXÕ. If the cell is grayed out, it indicates that survival was low and the species was not included in the analysis for that year. GenBan k accession numbers of genes used for phylogeny construction are listed. When a species was not located in GenBank, a close relative was used and noted with (*).ÉÉÉÉÉÉÉÉ...99 Table A2 Results from phylogenetic generalized least squares (PGLS) analysis of variance (ANOVA) testing the effects of plant status (native, noninvasive exotic, or invasive) and phylogeny on insect herbivory and mammal browsing. Statistically significant (P < 0.05) effects are in bold .ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.100 Table A3 Results from analysis of covariance (ANCOVA) testing the effects of plant status (native, noninvasive exotic, or invasive) and geographic spread (at three spatial scales) or time on insect herbivory and mammal browsing. Statistically significant (P < 0.05) effect s are in boldÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ101 Table B1 Results from analysis of variance (ANOVA) testing the effects of plant status (invasive, noninvasive exotic, or native) , family, and their interaction on insect herbivory and mammal browsing. Statistically sig nificant (P < 0.05) effects are in boldÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..106 Table C1 List of the 50 species planted into the 2012 -2014 experimental plots . Species are color coded by plant status: native (white), noninvasive exotic (gray), and invasive (black). In the columns for year, presence of a particular species is indicated with an ÔXÕ. For invasive species lists: WTP = Wild Type Plants, MNFI = Michigan Natural Features Inventory, MSL = Michigan Seed Law, PMW = Invasive Plants of the Upper Midwest (Czarapata 2005)ÉÉÉÉÉÉÉÉÉ109 Table E1 Results from mixed model analysis of variance (ANOVA) showing the effects of status, family, clipping, and competition on experimental plant flower number. Statistically significant (P ! 0.05) effects are in bold. All non -significant interaction terms were dropped from the final modelÉÉÉÉÉÉÉÉÉÉ..114 Table F1 To ease the process of comparing results from our vote count, traditional meta - xii analysis, and hierarchical meta -analysis we summarized the results of all three analyses here. Summary table of e ffect sizes from studies comparing the effects of biotic interactions on status performance. Symbols represent positive (+), negative (-), and neutral (n.s.) effects of biotic interaction removal. For our traditional (a) and hierarchical meta -analyses (b), effect sizes are calculated as HedgesÕ d, and significance is determined as whether 95% CIs cross zero. For our vote count, significance is determined by statistics reported in the original papers . Significant effect sizes are indicated with + and -, where non -significant effects indicated by n.sÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ. 121 xiii LIST OF FIGURES Figure 1 Three years of (a) insect herbivore and (b) mammal browser damage data on native (white bars), noninvasive exotic (gray bars), and invasive (black bars) plants. Bars indicate mean ± SE. Means with the same letter are not significantly different (P ! 0.05) based on post -hoc contrasts ÉÉÉÉÉÉÉÉÉÉÉÉÉ19 Figure 2 Boxplots of the median (black line), first and third quartiles (box), maximum, and minimum for (a) insect herbivory and (b) mammalian browsing for each species. Native species (N) are listed in white, noninvasive exotic species (E) in gray, and invasive species (I) in black. Species with one year of data, or the same amount of damage in all years, have boxplots only showing the median without quartiles. Species are organi zed in descending order by meanÉÉÉÉÉÉÉÉÉÉÉ. 20 Figure 3 Insect herbivore (a) and mammalian browser (b) damage on noninvasive exotic (gray points) and invasive (black points) plants with increasing residence time in MI. Analysis was performed on species averages across the three study years; each point represents one species. The gray r egression line indicates insect herbivory increases with residence time for non -invasive exotic specie s, but not invasive species ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..22 Figure 4 Insect herbivore (a -c) and mammal browser (d -f) damage on noninvasive exotic (gray poi nts) and invasive (black points) plants with increasing spread. Spread measures for counties within MI (a & d), spread within MI, WI, IL, IN, OH (b & e), and spread within the U .S. (c & f). Analysis was performed on species averages across the three study years; each point represents one species. Regression lines show significant relationships (P ! 0.05). A dashed black and gray regression line indicates insect herbivory increases with spread, but no difference between non -invasive exotic and invasive speci es. Black and gray lines indicate patterns only significant for invasive and non -invasi ve exotic species, respectively ÉÉÉÉÉÉÉ................................................................................23 Figure 5 Three years of survival data for nativ e, noninvasive exotic, and invasive plants in control (white bars) and fenced (gray bars) plots. Bars indicate mean ± SE. Status and family means with the same letter are not significantly different (P ! 0.05) based on post -hoc contrasts. ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..44 Figure 6 Biomass data for native, noninvasive exotic, and invasive plants in control (white bars) and fenced (gray bars) plots, divided by plant family. Bars indicate mean ± SE. Within family, means with the same letter are not significantly differ ent (P ! 0.05) based on post -hoc contrasts. ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.45 Figure 7 Flower number (a -c) and reproductive biomass (d -f) data for native, noninvasive exotic, and invasive plants in control (white bars) and fenced (gray bars) plots, divided by plant fa mily. Bars indicate mean ± SE. Within family, means with the xiv same letter are not significantly different (P ! 0.05) based on post -hoc contrasts ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.47 Figure 8 The effects of plant status and family on tolerance to insect herbivory (a-c) and mammalian browsing (d -h) in the field. Negative slopes indicate undercompensation and that the biomass and flower number of damaged plants is less than that of undamaged plants (undercompensation). Positive slopes indicate increased mass and flo wer number due to damage (overcompensation), and values of zero indicate compensation and no net change in growth rate. Solid regression lines show slopes that are significantly different from zero (P ! 0.05), and dashed lines represent slopes that are not significantly different from zero. ÉÉÉÉÉ.49 Figure 9 Natural log flower number (a) and natural log reproductive biomass (b) data for noninvasive exotic and invasive plants in control (light gray bars) and fenced (dark gray bars) plots, divided by species residence time. ÉÉÉÉÉÉÉÉÉ52 Figure 10 Biomass (a -b) and height (c -d) of native, noninvasive exotic, and invasive plants in clipped and unclipped treatments (a, c) or grown in the presence and absence of competition (b, d). Bars indicate mean ± SEÉÉÉÉÉÉÉÉÉÉÉÉÉÉ.67 Figure 11 Elymus canadensis biomass (a) and height (b) when grown with native, noninvasive exotic, or invasive species competitors. Bars indicate mean ± SE. Means with different letters are significantly different (P ! 0.05) based on pos t-hoc contrasts ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...68 Figure 12 Elymus canadensis biomass (a) and height (b) when grown with different competing species. Bars are labeled by competing species status (native = white, noninvasive exotic = light gray, invasive = da rk gray) and are ordered by descending mean values. The black bar indicates E. canadensis performance when grown alo ne. Bars indicate mean ± SE ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..69 Figure 13 Mean +/ - 95% CI effect size of each type of biotic interaction on native, noninva sive exotic, and invasive plants for cross continental studies. One asterisk (*) indicates effect sizes that differ significantly from zero, and two stars ( **) indicate significant effects of status. The number of studies in each category is indicated in p arentheses. Positive and negative values indicate that removal of the interaction increases or dec reases performance respectively ÉÉÉÉÉÉÉÉ.85 Figure 14 Mean +/ - 95% CI effect size of each type of biotic interaction on native, noninvasive exotic, and invasi ve plants for native/introduced comparison studies. One star ( *) indicates effect sizes that differ significantly from zero. The number of studies in each category is indicated in parentheses. Positive and negative values indicate that removal of the inter action increases or dec reases performance respectivelyÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 86 xv Figure 15 Mean +/ - 95% CI effect size of each type of biotic interaction on native, noninvasive exotic, and invasive animals for native/introduced comparison studies. One star ( *) indicates effect sizes that differ significantly from zero. The number of studies in each category is indicated in parentheses. Positive and negative values indicate that removal of the interaction increases or dec reases performance respectively ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...87 Figure A1 The best -scoring ML tree from a rapid bootstrap analysis in RAxML from the analysis of the concatenated sequences of matK, ITS, and rbcL. ML bootstrap frequencies are the numbers associated with nodes, and branch lengths are proportional to the number of nucleotide changesÉÉÉÉÉÉÉÉÉÉÉ...102 Figure A2 Images showing (a) the experimental common garden in 2012, (b) E.H. Schultheis in the field measuring insect herbivory and mammalian browsing on experimental seedlings, and (c) an experimental Lupinus perennis seedling.ÉÉÉÉÉÉ...103 Figure B1 Three years o f (a) insect herbivore and (b) mammal browser damage data on Asteraceae (hatched bars), Fabaceae (empty bars), and Poaceae (striped bars) plants. All analysis was performed within year. Bars indicate mean ± SE. Means with the same letter are not statistica lly different (P ! 0.05) based on post -hoc contrasts ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...107 Figure B2 Data from 2012 showing the family by status interaction for insect herbivore damage data. Species statuses shown with different color bars: native (white bars), noninvasive exotic (gr ay bars), and invasive (black bars). Bars indicate mean ± SE. Means within family with the same lett er are not statistically different (P ! 0.05) based on post -hoc contrasts ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ108 Figure E1 Flower number data for native, noninvasive exotic, and invasive plants that flowered during the course of the experiment. Graph a displays data by sta tus, while graph b displays data by species. Different colored bars represent the clipping and competition treatments. Bars indicate mean ± SE. Means with different letters are significantly different (P ! 0.05) based on post -hoc contrastsÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ ÉÉÉÉÉÉÉÉÉÉ...115 Figure F1 Effects of biotic interactions on native (light gray) , exoti c (medium gray), and invasive (black) species for cross continental studies, and native/introduced species comparison studies. Points show means bracketed by 95% confid ence intervals. Asterisk represents significant effect when 95% confidence intervals did not cross zero. The number of studies in each category is given in parenthesisÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ 122 Figure F2 Effects biotic interactions on native, exotic, and invasive animals for (a) native/introduced comparison studies, and (b) cross continental studies. Results are split up by biotic interaction manipulated. Points show means bracketed by 95% confidence intervals. Black asterisk s represent effects are sig nificant and xvi 95% confidence intervals do not cross zero. The number of studies in each category is given in parenthesis ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...123 Figure F3 Effects biotic interactions on native, exotic, and invasive plants for (a) native/introduced comparis on studies, and (b) cross continental studies. Results are split up by biotic interaction manipulated. Points show means bracketed by 95% confidence intervals. Black asterisk s represent effects are significant and 95% confidence intervals do not cross zero. The number of studies in each category is given in parenthesis ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ...124 Figure F4 Effects biotic interactions on native, exotic, and invasive animals for (a) native/introduced comparison studies, and (b) cross continental studies. Results are split up by biotic interaction manipulated and performance response variable measured (fecundity, growth, population growth, and survival) . Points show means bracketed by 95% confidence intervals. Black asterisk s represent effects are significant and 95% confidence intervals do not cross zero. The number of studies in each category is given in parenthesis ÉÉÉÉÉÉÉÉÉÉÉÉ125 Figure F5 Effects biotic interactions on native, exot ic, and invasive plants for native/introduced comparison studies. Results are split up by biotic interaction manipulated and performance response variable measured (fecundity, growth, population growth, and survival) . Points show means bracketed by 95% con fidence intervals. Black asterisk s represent effects are significant and 95% confidence intervals do not cross zero. The number of studies in each category is given in parenthesis ÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉÉ..126 Figure F6 Effects biotic interactions on native , exotic, and invasive plants for cross continental studies. Results are split up by biotic interaction manipulated and performance response variable measured (fecundity, growth, population growth, and survival) . Points show means bracketed by 95% confiden ce intervals. Black asterisk s represent effects are significant and 95% confidence intervals do not cross zero. The number of studies in each category is given in parenthesisÉÉ 127 1 CHAPTER 1 : INTRODUCTION Background Biotic interactions , such as herbivores and competitors , are major drivers of plant community structure and population dynamics (Crawley 1989, Lubchenco 1978, Louda 1982, Klironomos 2002, Fitzsimons and Miller 2010). When species are introduced into new ranges, they leave behind many biotic interactions from their native range, however they are immediately met with a new suite of species in the invaded range. These species may eventually become new competitors, herbivores, pathogens and mutualists for the invader and may be just as influential on performance as lost interactions. Elton (1958) was the first to describe the two potential roles played by biotic interactions for introduced species: release from antagonistic interactions from the native range may play a role in the explosive population growth rates of invasive species, and acquisition of antagonists in the introduced range may explain why so many of the species introduced around the world fail to establish and spread. The enemy release hypothesis (ERH) posits that invasiveness is driven by escape from enemies that constrained performance in the native range of an introduced species , and predicts that invasive species are less damaged and controlled by antagonistic biotic interactions than native species or noninvasive exotics. Therefore, escape may only be a temporary feature experienced d uring early phases of invasion, and the magnitude of enemy release could change over time. Enemy release may be ephemeral, and p lants and animals with longer residence times and larger areas of spread in their introduced range may be more strongly controlled by biotic interactions (Hawkes 2007, Mitchell et al. 2010). 2 Previous studies testing the ERH have focused primarily on enemy damage levels , and few have taken the next step to see if damage drives the increased performance of invasive species compared to native and noninvasive exotic species . Further, identifying the mechanisms that differ between introduced species that become invasive and those that do not is crucial to understand invasiveness. In addition , the effects of enemy release may be dynamic over the course of invasion as enemies acc umulate in the introduced range . My dissertation addresses these shortcomings in the E RH literature by focusing on four main questions: (1) Do invasive, noninvasive exotic, and native species experience different amounts of damage from enemies in the introduced range? (2) Does enemy release result in increased performance for invasive speci es compared to native and noninvasive exotic species? (3) Is enemy release lost with increased residence time and geographic spread in the introduced range? (4) Does tolerance to enemy damage or competitive ability drive invasiveness? Organization of the Dissertation Chapter 2 In collaboration with Jennifer A. Lau and Andrea E. Berardi, I used three years of data from 61 plant species planted into common gardens to determine whether invasive, noninvasive exotic, and native species experience different ial damage from insect herbivores and mammalian browsers, and whether enemy release is lost with increased residence time and geographic spread in the introduced range. We found no evidence suggesting enemy release is a general mechanism contributing to invasi veness in this region. Invasive species received the most insect herbivory, compared to native and noninvasive exotic species, and damage increased with longer residence times and larger range sizes at three spatial scales. Our results show that invasive a nd 3 noninvasive exotic species fail to escape enemies, particularly over longer temporal and larger spatial scales. Chapter 3 To test whether enemy release explains invasive success, and how the benefits of enemy escape may change with increasing residence time, I conducted a three -year field experiment, manipulating the presence of insect herbivores, mammalian browsers and fungal disease on 50 native, nonin vasive exotic, and invasive species. Counter to predictions, I found that native, noninvasive exotic, and invasive species experienced similar fitness benefits from enemy removal. On average, invasive species were no more tolerant to enemy damage than nati ve or noninvasive exotic species, though some invasive species showed evidence for increased performance when damaged. Additionally, I found that enemy release was lost over time and strongest for recently introduced species. Though ERH was not a mechanism shared by invaders in this system, it may operate for some introduced species and early in the invasion process. Chapter 4 In collaboration with Daniel MacGuigan , I performed a greenhouse experiment to investigate whether increased tolerance to herbivore damage or competitive ability could contribute to invasive plant success. We investigated whether invasive plants are more competitive than native and noninvasive exotics, and whether they are less affected by competition or exert stronger competitive effects on native species. We found the effects of competition and herbivory to be additive; the presence of our competition treatment did not affect the outcome of o ur herbivory clipping treatment, and vice versa. We also found that 4 introduced species were equally affected by herbivory and competition, compared to native species, yet they exerted stronger competitive effects on the native grass, Elymus canadensis . The refore, competitive effects on native competitors , potentially through allelopathy and novel weapons, may contribute to the success of introduced species. Chapter 5 In collaboration with Jennifer A. Lau and Ines Ib⁄Œez , I performed a meta -analysis on stud ies that experimentally tested effects of biotic interactions (competition, disease /parasitism , herbivory, mutualism, plant -soil feedbacks, predation) on native, invasive, and noninvasive exotic plant and animal performance . We included cross -continental s tudies that manipulated biotic interactions in both the native and introduced range of a species, and studies comparing introduced species to co -occurring natives in the introduced range. O ur meta -analysis provides strong evidence that antagonistic biotic interactions do not generally differ between native, invasive, and noninvasive exotic species. Removal of antagonistic biotic interactions increased performance equally for native, invasive, and noninvasive exotic species, indicating that biotic resistance is occurring for introduced species and that release is not a general mechanism explaining invasive speciesÕ success . However, while competition reduced performance for native and noninvasive exotic species, invasive species were generally not affected. I n general, native and invasive species were negatively affected by the experimental removal of mutualist partners, while noninvasive exotic species were not. Therefore, key differences between introduced species that become invasive and those that do not m ay be the formation of successful mutualisms in the introduced range, and release from competition compared to native and noninvasive exotic species in the introduced rang e. 5 CHAPTER 2 : NO RELEASE FOR THE WICKED: ENEMY RELEASE IS DYNAMIC AND NOT ASSOCIATED WITH INVASIVENESS Introduction Most introduced species do not establish, and even fewer become invasive ( Williamson and Fitter 1996 ). Although invasive species have fascinated scientists for decades ( Thellung 1912, Darwin 1895), the causal mechanisms of invasiveness are still undetermined. Some of the earliest writers on invasiveness predicted that loss of enemies in the introduced range might drive the success of invasive species over natives (Thellung 1912, citations within Kowarik and Py " ek 2012). Toda y, the Enemy Release Hypothesis (ERH) is the predominant , and most extensively tested, mechanism addressing the success of invasive species , and posits that invasive species gain a competitive advantage in their introduced range by escaping enemies that constrained their growth in their native range (Elton 1958, Callaway and Aschehoug 2000 , Keane and Crawley 2002) . An extension of the ERH is that invasive species are expected to receive reduced damage from enemies compared to co -occurring native species i n their introduced range. Enemy release may result in increased population densities, and could explain how invasive species overcome usual controls on population growth such as density dependence and life history tradeoffs (Blair and Wolfe 2004, Martin et al. 2010). Species introduced into new ranges sometimes experience reduced enemy diversity and damage compared to their native ranges (Mitchell and Power 2003, Torchin et al. 2003, Liu and Stiling 2006), and this reduced damage may translate into increase d performance (Maron and Vila 2001). For example, parallel experiments in the native and introduced ranges of Cynoglossum officinale found that reduced insect herbivory in the introduced range led to increased performance and population growth rates for th is species (Williams et al. 2010, see 6 also DeWalt et al. 2004). Similarly, a review of 473 species found that plants were attacked by 24% fewer virus and 84% fewer pathogen species in their introduced range, compared to their native range, and those specie s with a lower diversity of pathogens were more invasive, supporting ERH (Mitchell and Power 2003). However, release from enemies found in the native range does not mean complete release from all enemy pressures. While enemy release may play a role in the explosive population growth rate of invasive species , the acquisition of new enemies in the introduced range could explain why so many introduced species fail to become invasive (Carpenter and Cappuccino 2005, Hawkes 2007). Elton (1958) was the first to describe the dual role s played by enemies during biological invasions: an introduced species leaves behind many of its enemies, but is immediately met with a novel set of potential enemies in its introduced range. These new interactions could be just as important as those initially lost, limiting the establishment and geographic spread of an introduced species, preventing it from becoming invasive (Elton 1958, Maron and Vila 2001 ). Introduced species released from enemy pressures are likely to experience increased population growth and competitive ability and as a result become invasive, while introduced species that do not experience release should not ( Keane and Crawley 2002). In this paper, we will follow the conventions used in previous studies (Cappu ccino and Carpenter 2005, Liu et al. 2007, Parker and Gilbert 2007, Jogesh et al. 2008) and consider the following patterns of enemy damage evidence for ERH: (1) if enemy release explains the success of invasive species in their introduced range, we expect invaders to receive reduced damage compared to the native species with which they now compete (invasives < natives) and (2) if enemy release explains the differential success between introduced species that become invasive and introduced species that fail to become invasive (i.e., noninvasive exotics), we expect 7 invasives to receive less damage than noninvasive exotics (invasives < noninvasive exotics). These same patterns would hold for ERH studies looking at other responses to enemies, such as enemy effe cts on plant performance and survival. Many previous studies on ERH do not differentiate between introduced invasive and noninvasive exotic species, allowing them to test only the first prediction. Given the dual roles enemies play in invasions, homogeniz ing invasive and noninvasive exotic species into one group could miss important information on the drivers of invasiveness, and thus provide only a conservative estimate for whether invasive species experience enemy release. Studies that do not differentia te between these two types of introduced species have found that introduced plants receive less (Agrawal et al. 2005 ), no difference (Agrawal and Kotanen 2003, Hawkes 2007, Chun et al. 2010), or more (Ashton and Lerdau 2008, Stricker and Stiling 2014) enem y damage compared to natives . These same patterns are found in studies looking at enemy abundance or the performance consequences of enemy damage (reviewed in Colautti et al. 2004) . For example, seed pathogens and predators have similar effects on the fecu ndity of native and introduced species ( Blaney and Kotanen 2001a, Blaney and Kotanen 2001 b, Blaney and Kotanen 200 2). A recent systematic review of the ERH literature found that there is as much evidence for ERH as there is against it (Heger and Jeschke 20 14), and studies that find support for ERH tend to include just one pair of native and introduced congeners, while large multi -species experiments tend to find no difference in enemy effects between native and introduced species (Colautti et al. 2004). A m eta-analysis by Chun and collaborators (2010) found that introduced plants in general receive similar amounts of damage as native species and their performance was reduced to a greater degree than was nativesÕ. This lack of evidence for ERH may be due to combining invasive and noninvasive exotic species in analyses. 8 Studies that partition introduced species into invasive and noninvasive exotics are more rare (Liu and Stiling 2006) and find mixed support for ERH as well (ex. Cappuccino and Carpenter 2005, Carpenter and Cappuccino 2005 , Parker et al. 2006, Liu et al. 2007, Parker and Gilbert 2007, Jogesh et al. 2008 , Dawson et al. 2014). In a study of native, noninvasive exotic, and invasive Eugenia species, native species did in fact receiv e higher damage levels than invasives, supporting ERH Prediction 1 (Liu et al. 2007; see also Dietz et al. 2004 and Liu and Stiling 2006). The same study found no support for ERH Prediction 2 Ð invasive and noninvasive exotic Eugenia species received simil ar amounts of damage. Other studies support ERH Prediction 2, finding that invasive species received less enemy damage than noninvasive exotic species, or that introduced species that are more invasive tend to receive less herbivore damage or disease ( Mitchell and Power 2003; Cappuccino and Carpenter 2005; Carpenter and Cappuccino 2005 ). These studies reveal that the relationship between enemy pressures and invasiveness is complex and variable across species, study systems, and time (e.g., Agrawal and Kotan en 2003 , Agrawal et al. 2005). Dynamic Invasions An extension of the ERH is that the effects of enemy release may be dynamic over the course of invasion as enemies accu mulate in the introduced range. As an introduced species spends more time in its introduced range, expanding into new habitats and occupying a greater area, its likelihood of encountering an enemy that can attack it increases, potentially leading to increased damage with increased reside nce time and geographic spread. While enemy release may facilitate colonization and establishment during the early stages of an invasion, these benefits could be lost over time as introduced species acquire enemies (Elton 1958, Mitchell et 9 al. 2006, Mitche ll et al. 2010) . Species can accumulate enemies in their introduced range in three ways: (1) as invaders expand their range, they increase their probability of encountering an enemy in the introduced range that can attack them, (2) new introductions may b ring enemies from the speciesÕ native range from which they had previously escaped, and (3) evolutionary changes or plasticity in native enemies or the introduced species may result in enemies being able to exploit an introduced species as a novel resource (Go §ner et al. 2009). Therefore, the magnitude of enemy release is predicted to decrease over time and with range expansion into new habitats (Hawkes 2007, Mitchell et al. 2010). Studies on both crops and undomesticated species find that introduced speci es accumulate enemies over time and with increasing geographic spread in the introduced range (Strong et al. 1977, Hawkes 2007). In a study of 124 plant species introduced to North America from Europe, Mitchell and collaborators (2010) found that pathogen richness increased with speciesÕ time since introduction and geographic range. Other work has found no relation between time since introduction and damage from enemies (Carpenter and Cappuccino 2005). Despite the widespread ecological and evolutionary pro cesses that vary over the course of an invasion, only 40% of recent invasion literature mentions the residence times of species (Strayer et al. 2006), and even fewer factor this into their experimental design. If enemy release is dynamic, it could explain some of the contradictory findings in previous studies comparing enemy attack on noninvasive exotic, invasive, and native species (Colautti et al. 2004) . Understanding whether invasive species acquire enemies over time and with range expansion will help to predict the long -term effects of biological invasions (Mitchell et al. 2006 , Strayer et 10 al. 2006). These distinctions underscore the need for multi -species experiments to test the generality and persistence of enemy release. Here, we address this need by testing the dynamic nature of enemy release, while differentiating between damage received by invasive and noninvasive introduced species. We conducted multi -year field experiments using 61 plant species from multiple families, three provenanc es (status = native, noninvasive exotic, invasive), and with a variety of introduction dates and areas of geographic spread. The objectives of our study are to test the major predictions of the ERH (listed above) , and determine how damage from insect herbi vores and mammalian browsers change s over the course of invasion. We address two questions : (1) Do invasive, noninvasive exotic, and native species experience different amounts of damage from enemies? (2) Is enemy release lost with increased residence time and geographic spread in the introduced range ? We predict that if ERH contributes to invasiveness, invasive plants should receive less damage from insect herbivores and mammalian browsers, compared to native and noninvasive exotic plants. Further, i f intr oduced species lose the benefits of enemy release over time and with increased geographic spread in the introduced range , we predict that noninvasive exotic and invasive plants with earlier introduction dates and larger regional distributions will experience increased insect herbivory and browsing damage . Methods Study Species We planted 61 plant species into an old field community in Michigan, near the W.K. Kellogg Biological Station (Latitude: 42¡24' N, Longitude: 85¡23' W) (Table A1). Species were categorized as native , noninvasive exotic, or invasive (n = 25, 25, and 11 species respectively). 11 We define d native species as those naturally occurring in Michigan, prior to widespread European settlement. Invasive and noninvasive exotic species were both introduced to Michigan from outside the U.S., either accidentally or intentionally by humans, according to herbarium and historical records (Reznicek et al. 2011). While noninvasive exotic species assimilate d into the native community with little effect, invasive plants aggr essively colonized natural areas, threatening biodiversity and human interests . Invasiveness for this study was determined by inclus ion on one or more of the following local invasive species lists, as of June 2014 (Table A1): (1) Michigan Natural Features Inventory (Borland 2009) , (2) Listed by Czarapata (2005) as Òmajor invader of natural areasÓ and not categorized as needing disturba nce to establish, (3) Wild Type Plants (http://www.wildtypeplants.com/invasive.html ), and (4) the Michigan Seed Law (Act 329 of 1965) (http://www.legislature.mi.gov/(S(4zhwk1by1svk1hgs1ooxf03f))/d ocuments/mcl/pdf/mcl -act-329-of-1965.pdf). Inclusion on these lists means the species have been categorized as invasive within the Midwestern United States based on reports from land managers, inclusion on government invasive species lists, or published do cumentation of their effects on native plant and animal communities. Final decisions on status were made in consult with local land managers. We acknowledge that the classification of 'invasive' is not an absolute; it can depend on many biotic and abiotic factors [i.e., depends on context] . For invasive species found on only one list, we conducted a second analysis of our data listing them as noninvasive exotic; this analysis did not alter the main findings of the paper so we only report the analysis with them listed as invasive. We chose species based on the following criteria: First, to test for the generalizability of the ERH, we used a species mix that represented a wide range of phylogenetic diversity, 12 residence times (number of years in Michigan), an d geographic spread (number of counties occupied) (Ahern et al. 2010). Second, we included only herbaceous species to control for life form. Third, we used species already reported in herbarium records for Kalamazoo County (Reznicek et al. 2011) and common ly found in old field or grassland habitats to ensure that experimental plants grew in conditions similar to where they typically occur and also to make certain that we did not introduce species into parts of Michigan where they were not previously found. Finally , we preferentially chose species for which we could obtain seeds from nearby populations, either from personal field collections or orders from local growers, although some species were obtained from a broader geographic region (Table A1). Experi mental Design We planted two common garden field experiments: the first running from June through November 2011 and the second from May 2012 through September 2014. For Experiment 1 in 2011, we germinated seedlings of 30 species from six plant families (13 native, 11 noninvasive exotic, and 6 invasive) (Table A1) in greenhouses at the W.K. Kellogg Biological Station. We then trans planted two to three replicate seedlings of each species into randomly assigned locations within a 10 x 10 planting grid located within each of nine field plots (N = 540 seedlings) . These nine field plots represented the control plots of a large manipulative field experiment. Field plots were 2m x 2m in size, with 2m separating each plot. Species were planted within a grid of 100 cells within each plot, and were separated from the nearest experimental seedling by 20 cm. From October 11 to November 3, we measured damage from insect herbivores as the proportion of leaf area removed on 10 leaves per plant , selected as every third leaf starting at the top of the plant, and damage from mammalian browsers as the 13 proportion of aboveground vegetation removed by browsing damage , calculated as the proportion of stems with browsing damage for all plant families except the Poaceae. For the Poac eae, browsing was calculated as the proportion of tillers with browsing damage . If the individual was fully browsed down to the soil surface, we recorded this as 100% browsed. Author Schultheis collected all damage data to ensure estimates of aboveground v egetation removal were consistent. In 2012 we established Experiment 2, which included 50 species from three plant families (20 native, 2 0 noninvasive exotic, and 10 invasive) ( Table A1). We transplanted two replicate seedlings of every species into rando mly assigned locations within each of five field plots (N = 500 seedlings). These five field plots represent the control plots of a large common garden experiment manipulating insect herbivores , mammalian browsers , and disease to study their fitness effect s on native, noninvasive exotic, and invasive plants. Plots were 2m x 2m in size, with 2m separating each of the 40 plots. Within each plot, species were located within a grid of 100 cells and were separated from the nearest experimental seedling by 20 cm. From September 10 to October 4 in 2012 and August 26 to September 12 in 2013, we estimated insect herbivore damage and mammalian browsing damage as in Experiment 1. Annuals were harvested at the end of the 2012 growing season and are not present in the 20 13 census. In Experiment 2 we focused on the plant families Asteraceae, Fabaceae, and Poaceae, which represent three of the four plant families with the most invasive species in Michigan (Ahern et al. 2010). Additionally, these families vary widely in chem ical and structural traits, which could play a large role in herbivore defense strategies (Agrawal 2007 and citations within). Both experiments were planted in the same old field in Hickory Corners, MI . Old field habitats are common in the area and are for med when abandoned agricultural areas convert back 14 to unmanaged land. These communities consist of a wide diversity of both native and introduced plant and animal species . Based on field observations and trapping experiments, the dominant mammalian browser s in this community were Peromyscus maniculatus bairdii , Tamias striatus , Spermophilus tridecemlineatus , Sylvilagus floridanus, and Odocoileus virginianus (P. Howell, unpublished data ). These mammals are native to the area (Baker 1983) , with O. virginianus existing at a moderate to high density of ~30 individuals per square mile ( MDNR 2010). Detailed sampling of the insect community was performed in nearby prairie habitats (Robertson et al. 2011 ), however we were unable to find records that identified insec t herbivores down to species, precluding our ability to assign them native or introduced status. It is likely that the community consists of a mix of native and introduced insect herbivores, which differ in their effects on introduced plants (Parker et al. 2006). Residence Time and Spread Data To study the dynamic nature of enemy release, we determined the residence time and geographic spread for our invasive and noninvasive exotic species. We defined s pread as the number of counties occupied by a species, according to presence in herbarium records. We defined r esidence time as the number of years a species has occurred in Michigan, and calculated it as the year from the first herbarium specimen or histo rical record of introduction, subtracted from 2014. The method of using herbarium records to define spread and residence time of species is well established (citations within Ahern et al. 2010), but not without bias, including differential accessibility of field sites and variable sampling efforts over time. Therefore, residence time may actually indicate dates when an introduced species became apparent and occurred at high enough densities to be sampled, especially for species not intentionally 15 introduced and for which we have no historical record of introduction . We collected Michigan spread data from a published dataset, constructed from herbarium and historical records compiled in the Michigan Flora ( Reznicek et al. 2011 ) and updated with recent herbariu m records from the University of Michigan Herbarium (Ann Arbor, MI) (Ahern et al. 2010). Spread data at a regional and broader geographic scale was collected from the USDA PLANTS Database ( http://plants.usda.gov/ ), which records county level occurrence data for plant species (accessed January 2015). The USDA assigns county level occurrence by the presence of herbarium records and the scientific literature, similar to our dataset for Michigan. From these datasets we re corded (a) the number of counties invaded in Michigan (local scale), (b) the number of counties invaded in the five nearest states surrounding our study site (Michigan, Wisconsin, Illinois, Indiana, and Ohio ) (regional scale), and (c ) the number of countie s invaded in the U.S. (broad geographic scale). Phylogenetic R econstruction To control for phylogenetic non -independence in our study, we accounted for phylogenetic relatedness in all ANOVA analyses. Nucleotide sequences for matK, rbcL, and ITS were retrieved from NCBI Genbank for each species (accessed February 2015) (Table A1). If a species had no accession for a gene, a sequence from a closely related taxon was chosen if available . Gene sequences we re aligned using the MUSCLE algorithm in Geneious v . 6.1.8 (Kearse et al. 2012). The ends of sequences were trimmed from each gene, and the three genes were concat enated using phyutility (Smith and Dunn 2008). We determined t he optimal model of molecular e volution for the alignment using the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Performance Based Selection (DT) using jMo delTest2 v . 2.1.7 16 (Darriba et al. 2012). All three methods selected the General Time Reversible mod el, with rate heterogeneity including invariable sites and the rate of evolution at other sites as a gamma distribution (GTR + I + # ), as the optimal model. Maximum likelihood (ML) analysis with 100 bootstrap replicates was implemented with the high perfor mance computing version of RAxML v. 8.1.17 (Stamatakis 2014). We included a partition file for ML analysis to account for gene regions in the concatenated alignment. To assess phylogenetic structure for insect herbivory and mammalian browsing across the st udy species , we calculated BlombergÕs K (Blomberg et al. 2003) separately for each year of the study , following methods found in Swenson (2014) and using the phytools package in R (v. 0.4-21, Revell, 2012) . BlombergÕs K is a measure of whether a trait shows more or less phylogenetic divergence than expected by a null model of Brownian Motion. Statistical Analysis We performed all analyses in R ( v. 3.0.2, R Core Team 2015 ). Due to shared ancestry, traits in related species cannot always be viewed as being independent. We therefore incorporated comparative methods with linear models to determine whether invasive, noninvasive exotic, or native species differ in her bivore damage. We performed phylogenetic generalized least squares (PGLS) with Brownian Motion and Ornstein -Uhlenbeck (OU) models of trait evolution (Garland et al. 1993, Martins and Hansen 1997), with subsequent AIC model selection. PGLS was implemented b y incorporating the constructed phylogeny into the covariance stru cture using the ape package (v . 3.1-4, Paradis 2012), after which the linear models were fit using the gls function in the nlme package in R (v . 3.1-119, Pinheiro et al. 2015). Proportion of leaf area removed and proportion of stems (or tillers) browsed were included as 17 separate response variables, and plant status (native, noninvasive exotic, or invasive) was included as a fixed predictor variable. Because species is our unit of replication for questions on status, we averaged individuals within a species within a year. Analyses on herbivor y and browsing were conducted on within -year species averages; separate analyses were run for each year of data because species composition varied. To dete rmine whether there is a relationship between the damage a species received from insect herbivores and that received from mammalian browsers , we performed a regression using the lm function in R . Species were excluded from some analyses due to high mortality in the field (Table A1, grayed out boxes), likely due to limited rainfall and water availability at the time of plantin g, competition from the background community, and enemy damage . Post -hoc tests were used to evaluate differences between treatment combinations when the main effect of status was significant (P ! 0.05), and were implemented with a Holm multiple comparisons correction using the phylANOVA function in the phytools package in R. An additional analysis, including plant Family in our models in place of phylogenetic structure, can be found in Appendix B. To determine whether enemy damage changes with increased residence time or spread, we performed non -linear ANCOVAs using the glm func tion in R (v. 3.1.1, R Core Team 2015) . We included a logit link transformation in the generalized linear model to accommodate the nonlinear associations with county spread and time variables ( Bolker 2008 ). We did not incorporate phylogeny into these model s. Current phylogenetic methods that incorporate nonlinear relationships (such as independent contrasts) can reduce statistical power, and ignoring nonlinearity can affect biological inferences (Quader et al. 2004). Additionally , alternative techniques suc h as PGLS assume linear relationships and could not be used for our data. Only invasive and noninvasive exotic species were included in spread and residence time 18 analyses. In our dataset, an introduced speciesÕ range size is a function of its residence tim e in the introduced range, meaning that generally when given more time, an introduced species will continue to expand its range (Ahern et al. 2010). Because of the high degree of correlation between time and spread ( r = 0.70, P < 0.001), and between our different measures of spread (MI and 5 state spread: r = 0.86, P < 0.001 ; MI and U .S. spread: r = 0.72, P < 0.001), these predictor variables could not be tested simultaneously in one ANCOVA model (Underwood 1997, Miller and Chapma n 2001). Therefore, when discussing how residence time and spread relate to herbivory and browsing damage, we cannot differentiate between their effects, though we can still explore the relationships between these variables and enemy damage (Miller and Cha pman 2001). Here, we explore the effects of these variables separately, testing their individual influences in different models (Underwood 1997), and then discuss their effects making clear that either residence time or spread could be driving the observed patterns. To test whether enemy damage increases with residence time, proportion of leaf area removed and proportion of stems (or tillers) browsed were included as response variables; plant status, residence time, and their interaction were included as fi xed factors. We tested the same model for each of our spread measures, substituting spread for time as a predictor variable. Model fit and hypothesis testing were conducted using likelihood ratio tests, and significance was assessed from the $2 distributio n. Post -hoc contrasts were used to evaluate whether the slopes for invasive or noninvasive exotic species were significantly different from zero when a status by time, or status by spread, interaction was significant (P ! 0.05). Results In our system, we found no evidence that invasive species receive reduced enemy 19 damage, compared to native and noninvasive exotic species. Enemy damage from insect herbivores and mammalian browsers tended to be higher on invasive species, compared to nat ive and noninvasive exotic species (Fig. 1, Table A2). In 2011, invasive species received significantly more damage from insect herbivores than natives, and noninvasive exotics received intermediate amounts of damage (Fig. 1a; 2011: t 2,27 = 2.20, P = 0.04) . Though not significant, this trend remained consistent through 2012 and 2013 (Fig. 1a; 2012: t 2,43 = 0.24, P = 0.81; 2013: t2,31 = 1.20, P = 0.24). Notably, out of the top 10 species with the most insect herbivore damage, Figure 1: Three years of (a) insect herbivore and (b) mammal browser damage data on native (white bars), noninvasive exotic (gray bars), and invasive (black bars) plants. Bars indicate mean ± SE. Means with the same letter are not significantly different (P ! 0.05) based on post-hoc contrasts. 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 2011 2012 2013 Proportion Branches with Browsing Damage native exotic invasive a a a a a a ab a b 0 0.05 0.1 0.15 0.2 0.25 2011 2012 2013 Proportion Leaf Area Removed by Herbivory native exotic invasive a ab b a a a a a a (b) Proportion Leaf Area Removed by Herbivory Proportion Vegetation Browsed MELAL COSSU CENCY COSBI MELOF GAIPU SOLGR CORTRCENST SOLCA CORLAEUPPE LOTCOSYMPICORPA BROHO LESCU DESCA ERIAN POANE HELAU CORTITRIPR LESCA SOLRI CACAR MEDLU DAUCA POTAR POTRE SORNU ACHMI AGRRE AMCAN BROIN BROKA CICINCONCACORVA HELFL HESMA LACSA LEUVU PANVI PHLPR POACO POAPR POATR POTAG SCHSC SPOHE TAROF TEPVI TRIHYTRIRE0.00.20.40.60.81.0All Years Together Proportion Branches with Browsing Damage Native Exotic Invasive (a) 20 MELAL COSSU CENCY COSBI MELOF GAIPU SOLGR CORTRCENST SOLCA CORLAEUPPE LOTCOSYMPICORPA BROHO LESCU DESCA ERIAN POANE HELAU CORTITRIPR LESCA SOLRI CACAR MEDLU DAUCA POTAR POTRE SORNU ACHMI AGRRE AMCAN BROIN BROKA CICINCONCACORVA HELFL HESMA LACSA LEUVU PANVI PHLPR POACO POAPR POATR POTAG SCHSC SPOHE TAROF TEPVI TRIHYTRIRE0.00.20.40.60.81.0All Years Together Proportion Branches with Browsing Damage MELAL TRIRECICINCORVA CENST MELOF POTAR EUPPE TRIPR TRIHYAMCAN MEDLU HELAU LOTCOHESMA DESCA POTRE DAUCA POAPR SOLCA CONCACORTRLESCU SYMPICORLALESCA HELFL BROIN AGRRE BROHO COSBI GAIPU POANE PHLPR POACO BROKA ACHMI LEUVU CACAR CORTIERIAN PANVI SOLRI POATR SOLGR SORNU POTAG CORPA COSSU TEPVI SCHSC CENCY LACSA SPOHE TAROF 0.00.10.20.3All Years Together Proportion Leaf Area Removed By Herbivory Proportion Leaf Area Removed by Herbivory Proportion Vegetation Browsed MELAL COSSU CENCY COSBI MELOF GAIPU SOLGR CORTRCENST SOLCA CORLAEUPPE LOTCOSYMPICORPA BROHO LESCU DESCA ERIAN POANE HELAU CORTITRIPR LESCA SOLRI CACAR MEDLU DAUCA POTAR POTRE SORNU ACHMI AGRRE AMCAN BROIN BROKA CICINCONCACORVA HELFL HESMA LACSA LEUVU PANVI PHLPR POACO POAPR POATR POTAG SCHSC SPOHE TAROF TEPVI TRIHYTRIRE0.00.20.40.60.81.0All Years Together Proportion Branches with Browsing Damage Native Exotic Invasive (a) (b) (I) (E) (E) (I) (I) (I) (E) (N) (I) (I) (N) (I) (E) (E) (E) (I) (E) (N) (N) (I) (N) (E) (E) (I) (I) (N) (E) (E) (I) (N) (N) (N) (I) (N) (N) (N) (E) (I) (I) (E) (E) (E) (N) (E) (I) (N) (N) (E) CARCA CARCA (E) (E) (N) (N) (N) (E) (N) (N) (N) (N) (E) (N) (N) (E) (E) (N) (E) (N) (N) (I) (N) (N) (E) (I) (N) (N) (N) (E) (E) (E) (N) (N) (E) (E) (E) (E) (E) (N) (N) (I) (N) (I) (N) (E) (N) (I) (E) (I) (E) (E) (E) (N) (I) (I) (E) (N) (N) (N) (E) (N) (E) (E) Figure 2: Boxplots of the median (black line), first and third quartiles (box), maximum, and minimum for (a) insect herbivory and (b) mammalian browsing for each species. Native species (N) are listed in white, noninvasive exotic species (E) in gray, and invasive species (I) in black. Species with one year of data, or the same amount of damage in all years, have boxplots only showing the median without quartiles. Species are organized in descending order by mean. 21 six were invasive and three are noninvasive exotic. Only one of the top 10 species with the most insect damage was native (Fig. 2). Browsing was variable across years, but again we find no evidence consistent with the ERH (Fig. 1b, Table A2). In all years, native, noninvasive exotic, and invasive species did not differ significantly in browsing, but our post hoc contrasts revealed differences between noninvasive exotic and invasive species in 2013. In 2011 and 2013, invasives and natives generally received the most browsing dam age (Fig. 1b; 2011: t 2,27 = -0.16, P = 0.56; 2013: t2,31 = 0.28, P = 0.78). In 2012, invasive and noninvasive exotic species tended to receive more browsing damage than natives (2012: t2,43 = 0.96, P = 0.34). Of the 10 species with the most browsing damage , three were invasive, four were noninvasive exotic, and three were native. About half of species received little or no damage from mammalian browsers (Fig. 2). There was no relationship between the damage a species received from insect herbivores and that received from mammalian browsers (R 2 = 0.01, F1,31 = 0.31, P = 0.58). The phylogeny created for PGLS analyses was well resolved with high bootstrap support at the nodes and expected grouping by genus and family (F ig. A1). All of our values for BlombergÕs K were less than 1 (Table A2), indicating that leaf damage and stem browsing in close relatives were more divergent than expected across the phylogeny (Blomberg et al. 2003) and provides additional evidence that the lack of control for phylogeny in our ANCOVA analyses described below likely does not bias results. Enemy damage was dynamic, depending on residence time and areas of spread in the introduced range (Fig. 3 & 4, Table A3 ), and these patterns were consistent across multiple geographic s cales. With increased residence time, noninvasive exotic species experienced increased insect herbivore damage (Fig. 3a: $2 = 2.57, P = 0.02), but this pattern was not 22 observed for invasive species (Fig. 3a ; status x time interaction: $2= 1.76, P < 0.001). Introduced species with longer residence times tended to experience less mammalian browsing, although this trend was marginally non -significant (Fig. 3b; $2 = 9.82, P = 0.06). With increasing area occupied in Michigan (spread), noninvasive exotic s pecies experienced greater insect herbivory (Fig. 4a; $2 = 2.04, P < 0.001), and both invasive and noninvasive exotics experienced greater insect herbivory with increasing area occupied at larger spatial scales. (Fig. 4b -c; five states: $2 = 1.74, P < 0. 001; U.S.: $2 = 2.16, P = 0.001). Noninvasive exotic species experienced reduced mammalian browsing with increasing spread at all three scales, and at larger scales invasive species experienced increased mammalian browsing with increasing spread (Fig. 4 d -f, status x county interaction at all three scales; Michigan: $2 = 7.83, P = 0.009; five states: $2 = 5.43, P < 801001201401601800.00.10.20.30.40.50.6Number of Years in MI Proportion Branches Browsed Proportion Vegetation Browsed 801001201401601800.000.050.100.150.200.25Number of Years in MI Proportion Leaf Herbivory Exotic Invasive (a) (b) Proportion Leaf Area Removed by Herbivory R2 = 0.44 R2 = 0.16 Number of Years in MI Figure 3: Insect herbivore (a) and mammalian browser (b) damage on noninvasive exotic (gray points) and invasive (black points) plants with increasing residence time in MI. Analysis was performed on species averages across the three study years; each point represents one species. The gray regression line indicates insect herbivory increases with residence time for non-invasive exotic species, but not invasive species. 23 0.001; U.S.: $2 = 5.86, P < 0.001). Discussion We found no evidence consistent with ERH contributing to invasiveness in this system. We detected few significant differences in damage between native, invasive, and noninvasive 01002003004000.000.050.100.150.200.25Five State Counties Proportion Leaf Herbivory 01002003004000.00.10.20.30.40.50.6Number of Counties in 5 States Proportion Branches Browsed 0204060800.00.10.20.30.40.50.6Number of MI Counties Proportion Branches Browsed 0204060800.000.050.100.150.200.25Number of MI Counties Proportion Leaf Herbivory Exotic Invasive (a) (b) (c) Number of Counties Proportion Leaf Area Removed by Herbivory Proportion Vegetation Browsed R2 = 0.49 R2 = 0.45 R2 = 0.29 R2 = 0.41 R2 = 0.51 (d) (e) R2 = 0.47 (f) MI MI, WI, IL, IN, OH United States 05001000150020000.00.10.20.30.40.50.6Number of US Counties Proportion Branches Browsed 05001000150020000.000.050.100.150.200.25Number of US Counties Proportion Leaf Herbivory Figure 4: Insect herbivore (a-c) and mammal browser (d-f) damage on noninvasive exotic (gray points) and invasive (black points) plants with increasing spread. Spread measures for counties within MI (a & d), spread within MI, WI, IL, IN, OH (b & e), and spread within the U.S. (c & f). Analysis was performed on species averages across the three study years; each point represents one species. Regression lines show significant relationships (P ! 0.05). A dashed black and gray regression line indicates insect herbivory increases with spread, but no difference between non-invasive exotic and invasive species. Black and gray lines indicate patterns only significant for invasive and non-invasive exotic species, respectively. 24 exotic species, although invasives tended to receive more damage from insect herbivores than did native or noninvasive exotic plant species across all study years ( Fig. 1a). Browsing damage did not differ based on status, however native and invasive plants tended to get more browsing damage than did noninvasive exotics, not supporting either ERH prediction ( Fig. 1b). Our results are consistent with other experiments using the common garden approach. For example, in a study of 12 temperate vine species, native and invasive vines experienced more foliar damage from insect and mammal herbivores than noninvasive exotics, not supporting ERH predictions (Ashton and Lerdau 2 008). Similarly, invasive Eugenia uniflora sustained more insect herbivore damage than congeneric native and noninvasive exotic species in a common garden experiment, also not supporting ERH Predictions 1 and 2 (Stricker and Stiling 2014). In a study using 18 clover species, introduced and native species experienced similar amounts of disease, and the most invasive introduced species experienced the most disease (Parker and Gilbert 2007). Further, we found that invasions are dynamic and enemy release from insect herbi vores is lost over time for noninvasive exotic species, and with increasing spread at all scales for both invasive and noninvasive introduced species (Fig. 3a & 4a -c). Our herbivory results on the dynamic nature of invasions are consistent with other studies that have found that enemy release is lost with increased residence time ( Siemann et al. 2006, Hawkes 2007, Diez et al. 2010; but see Carpenter and Cappuccino. 2005 ) and spread (Mitchell and Power 2003, Diez et al. 2010 ) in the introduced range . While there was no relationship between residence time and browsing (Fig. 3b), we found that the most widespread noninvasive exotic species actually received the least amount of browsing, contrary to our predictions (Fig. 4d -f). Because we are not able to determine the direction of causality, browsers may in fact be driving the pattern in spread, acting as a filter and determining which species can spread furthest in the landscape. In contrast, 25 invasive species with the largest ranges experienced higher amounts of browsing damage (Fig. 4e-f), indicating that these two types of introduced plants might interact differently with mammalian browsers , although this result should be interpreted cautiously given that two species in a single genus ( Melilotus officinalis and M. albus ) drive the observed patterns for invasives . Due to the tight correlation between time and spread, we are unable to determine which variable is driving the patterns observed. Additionally, because we are unable to manipulate these two variables, other unmeasured correlated variables could be acting in our system. Previous analyses on our Michigan dataset of residence time and spread have found that the average introduced specie s will be present in 50% of counties after 160 years, with only the most invasive species spreading more quickly (Ahern et al. 2010). Given sufficient time, 10 -20% of introduced plants will be listed as invasive, indicating that invasiveness may be a funct ion of residence time in the introduced range (Ahern et al. 2010). In our study, we find consistent patterns across all three years of data collection, despite slight variations in experimental species composition (Table A1). Enemy pressures can vary great ly across years and growing seasons ( Agrawal and Kotanen 2003, Agrawal et al. 2005, Parker and Gilbert 2007). For example, in the first year of a common garden experiment, Agrawal and Kotanen (2003) found that introduced plants experienced more herbivory t han did natives, similar to the results of our own study. They collected a second year of data on the same common garden and found that introduced plants now received less herbivory, supporting the ERH ( Agrawal et al. 2005). They hypothesized that variable herbivore communities could drive these yearly differences, as well as ontogenetic changes in study plants and a potentially delayed response of the herbivore community to the establishment of their experiment. Thus, in their system, time periods where na tive species receive high amounts of enemy attack, but introduced 26 species receive little damage, may provide an opportunity window for introduced plants to dominate the system. In our experiment, no opportunity windows were apparent; the consistency of our results across three years provides strong support against ERH and suggests that enemy release windows, where invasive species experience reduced damage for a particular growing season, may be relatively infrequent in this system. Several mechanisms could explain higher enemy damage to invasives than noninvasive exotics or natives . First, fast growing species tend to allocate less to defense, resulting in higher amounts of herbivore damage than slow growing species (Cebrian and Duarte 1994, Endara and Cole y 2011), and invasive species may have faster growth rates than noninvasive exotic and native plant species (van Kleunen et al. 2010). Second, a lack of a shared coevolutionary history between introduced species and enemies in their introduced ranges could lead to higher amounts of damage because introduced species may lack defenses against these enemies, unlike native plants with coevolved defenses (increased susceptibility hypothesis; Hokkanen and Pimentel 1989; Colautti et al. 2004 , Verhoeven et al. 2009 ). Consistent with this hypothesis, herbivore feeding trials on aquatic (Parker and Hay 2005 , Morrison and Hay 2011) and terrestrial (Parker and Hay 2005) plants have shown that native herbivores preferentially consume introduced plants over nat ive s, and t he defensive chemistry of invasive plants serves as no more of a deterrent to herbivores than does the defensive chemistry of natives (Lind and Parker 2010). Third , invasive species may have higher local population densities than native or noninvasive exot ic species (e .g. Herrera et al. 2011), which could potentially increase the abundance of enemies feeding on these species or make invasive plants more apparent to insect herbivores and mammalian browsers (Feeny 1976). This final hypothesis is unlikely in o ur system. In our common gardens, all species were planted at equal densities, and although some experimental 27 species also naturally occurred at our experimental site, we found that, h erbivore ( r = 0.19, P = 0.42) and browsing ( r = 0.003, P = 0.99) damage were not correlated with species abundance at our site (percent cover estimated at 1% intervals for each species based on visual observation of 100 20x20 cm cells nested within 2x2 m experimental plots ). The question remains, if invasive plants tend to receive the most enemy damage , how is it that they are still invasive? There now exist over two dozen hypotheses attempting to explain invasiveness (Catford et al. 2009), and it is clear that no single hypothesis can explain the diversity of invasion strateg ies employed by todayÕs invaders (Gurevitch et al. 2011 , Lau and Schultheis 2015 ). Invasiveness could be driven, not by enemy release, but instead by performance and defense strategy traits of invasive species. In this study, we have identified differences in enemy damage between native, invasive, and noninvasive exotic species, and in future studies we will determine whether this damage translates into different effects on performance. Invasive species may lie on one end of a tradeoff between an individual Õs ability to resist damage and ability to maintain performance when damaged (i.e., tolerance) (Strauss and Agrawal 1999). Though invasive species in our system received the most insect herbivore damage, if they are also more tolerant, then their performan ce (growth, survival, fecundity) may be less affected by this damage compared to noninvasive exotics and natives . Contradictory to this hypothesis, a meta -analysis found introduced species to be less tolerant to damage (Chun et al. 2010). In a different me ta-analysis, Parker and colleagues (2006) classified introduced species along a spectrum of invasiveness and determined that herbivores had similar effects on the performance of both noninvasive exotic and invasive species. Alternatively , release from ene mies not tested in our system, such as disease or belowground enemies, could contribute to invasion success . Mitchell and Power (2003) 28 demonstrated that plant species that experienced release from fungal and viral pathogens were more widely invasive than t hose that did not. Similarly, invasive plants experienced lesser effects from belowground enemies than rare, native plants (Klironomos 2002). In contrast, a meta -analysis by Levine and collaborators (2004) found herbivores (as well as competition and diver sity in the native community) provided resistance to invasion, while fungal pathogens did not (Levine et al. 2004), indicating that some enemies may contribute more to ERH than others (Levine et al. 2004). Though we find that ERH, mediated through abovegr ound herbivores, was not a common pattern across the species tested in our study location , some invasive and noninvasive exotic species did receive low amounts of enemy damage ( Fig. 2), indicating that the success of these species may be driven by enemy release. During at least one study year, the invasive Poaceae species Bromus inermis (2012), Poa compressa (2013), and Poa pratensis (2013) experienced no damage from either insect her bivores or mammal ian browsers. These species could be candidates for further study to assess whether enemy release contributes to increased fitness over native competitors. Conclusion The ERH remains among the most popular hypotheses explaining the succ esses and failures in introduced species, despite mixed support. A review on the ERH found that 36% of studies support it, while 43% do not (Heger and Jeschke 2014, see also Colautti et al. 2004). Meta-analyses and reviews on enemy richness and damage for introduced and native species find results both for (Liu and Stiling 2006, Hawkes 2007) and against the ERH (Chun et al. 2010). Our study helps to identify some of the sources of variation in previous Enemy Release 29 Hypothesis studies, namely distinguishing between invasive and noninvasive exotic species and considering the dynamic interplay between an introduced species and their enemies over decadal timescales. Our findings indicate that invasive species generally receive more damage from enemies, compared to native and noninvasive exotic species, not supporting key predictions arising from the ERH. Therefore, we conclude that enemy release is not a general mechanism associated with invasiveness in our system, although enemy release could apply to specific cases of invasion and early on in the invasion process. 30 CHAPTER 3 : PERFORMANCE CONSEQUENCES OF ENEMY RELEASE DEPEND ON INVASION AGE BUT DO NOT EXPLAIN INVASIVENESS Introduction The enemy release hypothesis (ERH) is among the most commonly invoked hypotheses to explain invasiveness, and states that the success of invasive species in their introduced range is driven by the loss of enemies, such as herbivores and disease, that constrain their performance in their native range ( Elton 1958, Callaway and Aschehoug 2000 , Maron and Vila 2001, Keane and Crawley 2002). According to the ERH, invaders should experience reduced enemy damage and as a result, fewer negative fitness effects of enemies in their introduced range compared to their native range. Additi onally, the ERH predicts that invasive species will receive reduced damage, and less negative effects on performance from enemies compared to co -occurring native and noninvasive exotic species in their introduced range . To date, most ERH studies have focu sed on measuring damage, and many studies have compared enemy damage on introduced and co -occurring native species in the introduced range (Agrawal and Kotanen 2003, Colautti et al. 2004, Agrawa l et al. 2005, Chun et al. 2010). Few have explored whether damage differs for invasive and noninvasive exotic species ( Mitchell and Power 2003, Parker et al. 2006, Liu et al. 2007, Parker and Gilbert 2007, Ashton and Lerdau 2008, Schultheis et al. 2015), which could reveal important differences between successful invaders and introduced species that do not become invasive. These o bservational studies of enemy damage have provided limited support for ERH, with only a few studies finding that invasive species receive less damage from enemies than natives or noninvasi ve exotics (Maron and Vila 2001, Colautti et al . 2004, Liu and Stiling 2006), and a number of studies and meta -analyses finding that invaders receive equal or even more damage (Parker et al. 2006, Ashton 31 and Lerdau 2008, Zou et al. 2008, Chun et al. 2010, Morrison and Hay 2011, Dawson et al. 2014, Schultheis et al. 2015) . Three hypotheses may explain how invaders are able to receive high amounts of damage while still performing better than native and noninvasive exotic species. First, many ERH studies mani pulate only a single enemy or examine a particular type of enemy damage, and release from multiple enemies may be necessary for invasive success. Second, invasive species may be more tolerant to enemy damage, maintaining performance and fitness despite rec eiving high levels of damage. Third, the fitness effects of enemies in the introduced range may increase over time as introduced species form novel biotic interactions with members of their new community, and ERH may only contribute to success early in the invasion process. Community Complexity and Enemy R elease In native systems, plant performance is typically influenced by a multitude of biotic factors (Harper 1977, Louda 1982, Crawley 1989), and the effect of a particular biotic interaction often depend s on a speciesÕ community context ( Wootton 1994, Agrawal et al. 2007 ). However, ERH studies manipulating only one biotic interaction are unable to identify interactive and synergistic effects between antagonists . Perhaps, as in native systems, the r emoval of multiple antagonists will have additive or synergistic effects, and enemy release will only promote invasiveness when more than one antagonist is escaped. For example, only when competitors and herbivores were present was the performance of invaders Triadica sebifera and Cirsium vulgare reduced (Huang et al. 2012 , Suwa and Louda 2012) ; escape from either competitors or herbivores alone would not have been sufficient for release, but escape from both antagonists could drive invasiveness . 32 Tolerance Differences in damage between species may not directly translate into differences in performance. Tolerance, or a plantÕs ability to regrow and reproduce after damage, differs greatly between species (Marquis 1992, Strauss and Agrawal 1999). Therefore, t ests of the ERH must go beyond observations of enemy damage and quantify performance effects of damage by manipulating the presence of enemies in the field ( Maron and Vila 2001, Keane and Crawley 2002). Given the number of studies suggesting that invasive spec ies receive as much or even more damage than native and noninvasive exotics, tolerance to, rather than release from , enemy damage may explain invasiveness. Invasive species tend to be larger, grow more quickly, and produce denser stands compared to competi ng species in their introduced range and conspecific populations in their native ranges (Crawley 1987, Hinz and Schwarzlaender 2004 , van Kleunen et al. 2010, Dawson et al. 2014; but see Thebaud and Simberloff 2001, Daehler 2003). Larger, more locally abundant species often lo se more leaf tissue to herbivory and disease (Feeny 1976, Rhoades and Cates 1976, Paker et al. 2015), potentially due to increased detection by herbivores and increased density promoting disease spread (Anderson and May 1979, Burdon and Chilvers 1982). Additionally, invasive plants frequently have lower construction costs for tissues, and cheap tissues may be less defended against damage (Daehler 2003). Therefore, invaders may lie on one end of a tradeoff spectrum between defending against damage and tolerating damage once it occurs ( Strauss and Agrawal 1999). 33 Dynamic Enemy R elease The lack of generalizability observed in ERH studies may be due to the fact that the traits and causal processes d riving invader success early during invasion differ from those acting later (Dietz and Edwards 2006, Theoharides and Dukes 2007 ). For example, factors determining establishment success might differ from those controlling spread in the landscape (Richardson et al. 2000, Kolar and Lodge 2001, Heger and Trepl 2003, Levine et al. 2004, Kempel et al. 2013). In a study of 48 introduced plant species, propagule pressure contributed to speciesÕ establishment, but after three years of growth, traits related to inter actions with insect herbivores and plant competitors became more important determinants of success (Kempel et al. 2013). Additionally, the probability of encountering new antagonists could increase with longer residence times; for example, native enemies o f closely related species may evolve to use invaders as a host or food source, or enemies from the invaderÕs native range could be introduced (Go§ner et al. 2009, Mitchell et al. 2010, Flory and Clay 2013). Initial loss of enemies can be overshadowed by ga in of new enemies, and the diversity of herbivores on introduced plants can be equal to that of natives (Maron and Vila 2001). ERH therefore may only be important and operate during the initial phases of invasion, and may be lost over time and with increas ing spread in the introduced range (Elton 1958, Mitchell et al. 2006, Mitchell et al. 2010) . Dynamic ERH predicts that introduced species with longer residence times in their new range will be more negatively affected by biotic interactions than species wi th more recent dates of introduction (Hawkes 2007, Mitchell et al. 2010). There is now mounting evidence for the ephemeral nature of ERH (Hawkes 2007, Hayes and Barry 2008 , Flory and Clay 2013 ). For example, in a study of 36 invasive and noninvasive exotic species, Schultheis and collaborators (2015) found that noninvasive exotic species 34 experienced increased insect herbivory with increasing residence times, while invasive species did not. In a study of 124 plant species, pathogen richness was six times higher on introduced plants that had been in their introduced range 400 years, compared to more recently introduced plants (Mitchell et al. 2010). However, these studies do not determine whether e nemy accumulation translates to effects on performance (Flory and Clay 2013). Several studies of belowground enemies and plant -soil feedbacks (PSF) have demonstrated increasingly negative fitness effects of the soil community on introduced plants with time . The invasive plant, Heracleum mantegazzianum , accumulated belowground enemies and experienced reduced survival, biomass, and competitive ability when grown in soil collected from sites invaded longer ago compared to newly invaded sites ( Dost⁄l et al. 201 3). In New Zealand, species with longer residence times experienced more negative PSF (Diez et al. 2010). However, t o the best of my knowledge, no study has demonstrated whether the accumulation of aboveground enemies translates into performance difference s over time for introduced species. Here, I test the ERH by manipulating the presence of insect herbivores, mammalian browsers, and fungal disease on 20 native, 10 invasive, and 20 noninvasive exotic plant species in Michigan. Previously in this system, I found no evidence for reduced damage from insect herbivores or mammalian browsers on invasive species compared to native and noninvasive exotics (Schultheis et al. 2015), leading me to hypothesize that release from multiple antagonists or tolerance may pla y an important role in invasiveness. Additionally, enemy damage was dynamic; insect herbivory increased for noninvasive exotic species with longer residence times, but did not increase for invasives (Schultheis et al. 2015). Therefore, the performance effe cts of enemies may also increase for introduced species with longer residence times in the introduced range. 35 If ERH determines invasion success, I expect that invaders will experience fewer fitness benefits from the removal of enemies compared to native a nd noninvasive exotic species that are still controlled by enemies, either because invaders experience reduced damage or because invaders are highly tolerant. In addition, I predict that the removal of enemies will reduce the reproductive and growth advant ages commonly observed for invasive species, relative to native species. Further, I predict that ERH will be ephemeral and lost for invasive and noninvasive exotic species with longer residence times in Michigan. Methods Experimental D esign To determine how antagonistic biotic interactions affect the performance of native, noninvasive exotic, and invasive plants, I conducted an enemy exclusion experiment in an old field community in southwest Michigan, near the W.K. Kellogg Biological Station (Latitude: 42¡24' N, Longitude: 85¡23' W). I constructed 40 2x2m field plots, with a 2m buffer between each plot, in which I experimentally manipulated the presence and absence of mammalian browsers, insect herbivores, and fungal disease in a 2x2x2 factor ial design (n=5 plots per treatment). To exclude mammalian browsers I constructed four -foot deer fencing around the perimeter of treated plots and buried 0.64cm grid hardware cloth to a depth of 0.2 meters with a bent 3cm lip facing outwards to re -direct d igging mammals away from the plot interior (Munger and Brown 1981) . Hardware cloth extended 0.6m above the soil surface and was secured to deer fencing. To bury hardware cloth, I trenched the perimeter of each plot and backfilled once fencing was in place; to control for effects of trenching, all plots without fencing were also trenched and backfilled. To exclude insect he rbivores and fungal pathogens I sprayed 36 experimental plots with insecticide or fungicide (Merit 75 WP at 0.031g/L and Heritage at 0.062g/L, respectively). Treatment applications used approximately 8L of liquid, and c ontrol plots were sprayed with an equal volume of water. I applied treatments biweekly throughout the growing season. I germinated experimental seedlings in greenhouses at the W .K. Kellogg Biological Station, beginning in April 2012. From 14 -22 May, 2012 I planted seedlings into the existing background community present in the field. Within experimental field plots, species locations were randomized in a 10x10 grid with 20cm betw een each seedling. Each plot contained two seedlings of 50 experimental species (N=4,000 seedlings), representing three plant families (Asteraceae, Fabaceae, Poaceae) and three provenances (status = native [n=20], noninvasive exotic [n=20], invasive exotic [n=10]) (Table C1). Species used in this experiment naturally occur in Kalamazoo County and grow in similar conditions to those present at the field site . When possible, I collect ed seeds from nearby populations or used local seed sources (Table C1). Nat ive status was assigned if a species occurred in Michigan prior to widespread European settlement. Invasive and noninvasive exotic species were introduced from outside the U.S. , according to herbarium and historical records (Reznicek et al. 2011), but noni nvasive e xotic plants assimilate into the native community with little effect, while invasive plants invade natural areas and threaten biodiversity . Invasiveness of introduced species was determined for this study by inclusion on local invasive species lists and in consultation with local land managers (Schultheis et al. 2015). In addition, I focused on three plant families (Asteraceae, Fabaceae, Poaceae) that vary widely in chemical, structural, and growth traits, po tentially playing a large role in enemy tolerance and resistance strategies (Agrawal 2007), and represent three of the four plant families with the most invasive species in 37 Michigan (Ahern et al. 2010). At the end of each growing season I estimated enemy damage on the 4,000 experimental seedlings. I calculated d amage from insect herbivores as the proportion of leaf area missing on 10 leaves per plant , selected as every third leaf starting at the top of the plant. I recorded leaves that were totally removed as 100% herbivory when I could observe the petiole was intact. Because it remains constant over the age of a leaf, percent leaf area missing serves as a reliable estimate of herbivory (Lowman and Heatwole 1992). To estimate disease incidence, I recorded the proportion of necrotic and chlorotic leaf tissue. I measured damage from mammalian browsers as the proportion of stems or tillers with evidence of browsing damage. If browsing resulted in complete removal of a focal plant, I recorded the individual as 1 00% browsed. In addition, I measured performance metrics for each species, including survival, flower (or inflorescence) number, reproductive biomass, and vegetative biomass. An individual was recorded as alive in a particular year if aboveground biomass was present at any time during the growing season. Reproduction for the Asteraceae was estimated as the number of composite flower heads, and for Poaceae as the number of spikelets . Reproduction for al l species was calculated as the sum of buds, flowers, and seed heads produced by an individual throughout the growing season. To estimate reproductive I collected all buds, flowers, and seed heads at the end of each growing season. To assess vegetative bio mass, I harvested annuals at the end of the 2012, biennials in 2013, and perennials in 2014. I dried reproductive and vegetative biomass at 70¡C for 72 hours before weighing. To assess whether enemy release is dynamic and lost over time, I assigned invasiv e and noninvasive exotic species a residence time in Michigan based on a published dataset constructed from herbarium and historical records (Ahern et al. 2010). I define r esidence time as 38 the number of years a species has occurred in Michigan, and calcula ted it for each species as the year a species was first present in records, subtracted from 2014 (Schultheis et al. 2015). Using herbarium records to define residence time in a speciesÕ introduced range is a well -established method, particularly when other resources, such as historical introduction documentation, are not available (Ahern et al. 2010). However, this method is potentially biased due to differential accessibility of field sites, inconsistent sampling efforts over time, and introduced species becoming more apparent over time as they increase in number in their introduced range. Therefore, herbarium records may underestimate residence time, especially for introduced species with a long lag phase before they became abundant. Statistical Analysis Damage: I performed all analyses in R (v. 3.0.2, R Core Team 2015). To determine whether enemy removal treatments were effective at reducing enemy damage, I tested the effects of enemy exclusion on enemy damage with mixed model ANOVA using the lmer function in the lme4 package in R (v. 1.1 -7, Bates et al. 2015) . To test the efficacy of the fencing treatment, mammalian browsing was included as the response variable and fencing treatment as the fixed predictor variable. To test the efficacy of the insecticide treatment, the model included insect herbivory as the response variable and insecticide treatment as the fixed predictor variable. I was unable to run a model testing the efficacy of the fungicide treatment because no visible infection was detected on experimental plants throughout the experiment. Experimental plot is the unit of replication for questions on treatment effectiveness , so I included plot nested within treatment as a random factor in t hese models. P -values for mixed models were obtained using the lmerTest package in R (v. 2.0 -20, Kuznetsova et al. 2015). 39 Performance: To determine whether enemies significantly altered the performance of experimental species, I tested the effects of enemy exclusion on plant performance with mixed model ANOVA. Status (native, noninvasive exotic, invasive), plant family (Asteraceae, Fabaceae, Poaceae), plant type (annual, biennial, perennial), fencing treatment (fenced, control), insecticide treatment (insec ticide spray, control), fungicide treatment (fungicide spray, control), were included as fixed predictor variables. Species nested within status, plot nested within enemy treatment, and the species x enemy treatment interaction were included as random vari ables. Because of the potential for many interactions between predictor variables in the experiment, I lacked sufficient power to include all interaction terms in the models. I addressed this power issue using two methods. (1) First, I ran a series of mode ls where I included all possible two and three way interactions between enemy removal treatments, which allowed me to identify any interactive or synergistic treatment effects. However, due to power issues, these models could not contain all interactions b etween treatments and status, or all necessary random terms to eliminate pseudoreplication from models. (2 ) Because these initial models revealed that only fencing significantly affected performance and enemy removal treatments did not interact, I conducted a second set of analyses including the fencing treatment only. These models allowed me to include all interactions between plant family, status, type, and fencing treatment, and include all necessary random terms. Both analyses yielded consistent results, so here I present only the second set of models. I conducted separate analyses for each performance metric [survival, vegetative biomass (g), flower number, reproductive biomass (g)]. I calculated an individual plantÕs flower number and reproduct ive biomass by summing data across all years of the experiment. I only collected vegetative biomass once for each species (2012 for annuals, 2013 biennials, 2014 perennials) and 40 all years were analyzed together. To test for the effects of treatments on sur vival, and to explore how survival differed across years, I analyzed survival for each year of the experiment separately. For survival models using 2014 data, I did not include a plant perenniality (type) term because only perennials remained in the experi ment. For lmer models, I determined significance of fixed and random effects using the lmerTest package in R, and for random effects , I used chi -squared tests using the rand function . Flower number (Poisson distribution) and survival (binomial distributio n) were analyzed using general linear mixed models with the glmer function in the lme4 package in R. For these glmer models, I determined significance for fixed and random effects using chi -squared tests based on log -likelihood ratio tests. Reproductive an d vegetative biomass data were log transformed to improve normality and analyzed using linear mixed models and the lmer function in the lme4 package in R. When I found a significant main or interactive effect of status, plant family, plant type, or fencing treatment (P ! 0.05), I conducted post hoc Tukey tests using the multcomp package in R (v. 1.4 -0, Hothorn et al. 2015). I removed species perenniality from final models as it did not contribute to predictive ability and was never significant. Non -signific ant interaction terms were removed from final models to increase power for testing main effects (Crawley 2007). Tolerance: To determine if invasive species are more tolerant of enemy damage than native or noninvasive exotic species, I tested the effects of enemy damage on plant performance with mixed model ANCOVA . To test whether invasive species are more tolerant than natives or noninvasive exotics , I included performance (vegetative biomass, reproductive biomass, flower number) as response variables. Plan t status (native, noninvasive exotic, invasive), plant family (Asteraceae, Fabaceae, Poaceae), damage (proportion of leaves with insect herbivory or 41 proportion of stems/tillers browsed), and all interactions were included as fixed factors. Because toleranc e can be measured as the slope of the regression of plant performance on damage (McNaughton 1983, Strauss and Agrawal 1999), a significant interaction between plant status and damage would indicate that native, noninvasive exotic, and invasive species diff er in tolerance. Species nested within status and damage x species interactions were included as random factors. If the damage x species random term was significant, then I concluded that species differed in their tolerance to damage. I conducted separate analyses for tolerance to insect herbivores and mammalian browsers. Only data from individuals for which I had performance and damage data could be used for tolerance analysis; species that died before I could estimate damage were therefore not included, possibly leading me to overestimate tolerance if these individuals died due to high levels of enemy damage. Enemy damage and biomass data were natural log transformed to improve normality. When the main effect of damage or the damage by status interaction was significant (P ! 0.05), I used post -hoc contrasts to evaluate whether the slopes for native, invasive, or noninvasive exotic species were significantly different from zero . Time: Species in the experiment spanned a range of residence times from 72 to 176 years. Based on preliminary data exploration, I grouped species into two categories for analysis Ð those that had been in Michigan for less than 120 years, or those that had been i n Michigan % 120 years. I was unable to treat time as a continuous variable in models due to insufficient replication for the invasive status (n = 10). To determine whether mammalian browser or insect herbivore removal treatments affected plant performance differently based on speciesÕ residence time in Michigan, I tested the effects of fencing, insecticide, and residence time on plant performance with mixed model ANOVAs. Response variables included vegetative biomass (g), reproductive 42 biomass (g), and flowe r number. Status (native, noninvasive exotic, invasive), insecticide (sprayed, control), fencing (fenced, control), residence time ( < 120 years, % 120 years) , and all interactions were included as fixed predictor variables. Species nested within status, plo t nested within insecticide treatment, and plot nested within fencing treatment were included as random variables. Biomass response variables were natural log transformed to improve normality. Because the insecticide treatment did not influence speciesÕ pe rformance, all non -significant (p > 0.05) interaction terms with insecticide were dropped from final models. Results Damage Fencing and insecticide treatments effectively reduced damage from mammalian browsers and insect herbivores in all years of study. Mammalian browsing was reduced by 97.4%, 99.0%, and 97.0% in 2012, 2013, and 2014 respectively (mean ± SE branches browsed. 2012: control 11.7 ± 0.9%, fenced 0.3 ± 0.1% [F 1,27 = 65.3, p < 0.001; 2013: control 19.9 ±1.6%, fenced 0.2 ± 0.1%: F 1,32 = 102.7, p < 0.001; 2014: control 29.9 ± 3.0%, fenced 0.9 ± 0.6% [F 1,34 = 65.3, p < 0.001). Insect herbivory was reduced by 42.9%, 74.6%, and 56.5% in 2012, 2013, and 2014 respectively (mean ± SE % leaf area removed. 2012: control 5.6 ± 0.4%, insecticide 3.2 ± 0.3% [F1,2214 = 27.0, p < 0.001], 2013: control 6.7 ± 0.5%, insecticide 1.7 ± 0.2% [F1,44 = 61.6, p < 0.001]; 2 014: control 6.9 ± 0.9%, insecticide 3.0 ± 0.5% [F1,44 = 9.2, p = 0.004]). Performance Survival : In 2012 plants survived significantly better in fenced plot s for all plant statuses (Fig. 5 a; F 1,11 = 40.38, P = 0.01). The effects of the mammalian browser removal on survival did 43 not differ based on status, family, or plant type (Table 1). In 2013, rem oval of mammalian browsers did not significantly increase survival, and survival did not depend on plant status or family (Table 1). In 2014, removal of mammalian browsers did not significantly increase survival (fencing: F 1,11 = 0.11, P = 0.12), and survival was higher in Asteraceae and Fabaceae compared to Poaceae (family: F 2, 11 = 11.04, P < 0.001). Vegetative Biomass : Fencing treatments differentially affected native, noninvasive exotic and invasive species, but these effects depended on plant fam ily (status x fencing x family; F 4,36 = 3.36, P = 0.02). Fencing treatments did not affect Asteracea e biomass for any status (Fig. 6 a), but significantly increased invasive Fabaceae biomass and decreased native Poaceae biomass. Additionally, averaged acros s all treatments, biomass for invasive Fabaceae was significantly Model Terms dfF!2PdfF!2PdfF!2Pstatus 2, 445.480.0082, 3713.56<0.001 4, 142.9814.80.005family 2, 434.180.022, 400.240.794, 1413.0232.5<0.001 fencing 1, 119 0.760.391, 1640.330.573, 141.1512.10.007status x fencing 2, 911.480.232, 293.580.042, 161.222.20.33status x family 4, 423.390.024, 383.870.014, 200.301.20.88family x fencing 2, 415.600.0072, 830.420.662, 208.8512.30.002status x family x fencing 4, 363.360.021, 2290.400.532, 230.842.40.496Random Effects (species)status 22.8<0.001 13.20.00117.2<0.001 (plot)fencing 23.3<0.001 7.80.0223.8<0.001 species x fencing 1.50.202.00.1614.3<0.001 Model Terms dfF!2PdfF!2PdfF!2Pstatus 2, 11 2.064.20.122, 11 2.670.10.742, 11 2.824.40.11 family 2, 11 0.651.30.522, 11 0.511.00.612, 11 11.04 18.3<0.001 fencing 1, 11 40.386.30.011, 11 0.11 4.30.121, 11 2.061.80.18Random Effects (species)status 73.1<0.001 11.0 0.00427.4<0.001 (plot)fencing 53.1<0.001 12.80.0025.00.08species x fencing 7.80.0054.20.040.10.77Vegetative Biomass (g) Reproductive Biomass (g) Flower Number Survival 2012 Survival 2013 Survival 2014 Table 1: Results from mixed model analysis of variance (ANOVA) testing the fixed effects of plant status (native, noninvasive exotic, invasive), family (Asteraceae, Fabaceae, Poaceae), and mammalian browser removal (fencing treatment, control) on plant performance. Biomass and floral biomass were log transformed to fit normality assumptions. Binomial (survival) and count (flower number) data were analyzed using general linearized mixed models; all other data were analyzed using linear mixed models. Chi square statistics based on log-likelihood ratio tests are presented for random factors and general linearized models. Statistically significant (P ! 0.05) effects are in bold. 44 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 native exotic invasive Survival 2013 control fenced 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 native exotic invasive Survival 2012 control fenced 0 0.1 0.2 0.3 0.4 0.5 0.6 Asteraceae Fabaceae Poaceae Survival 2014 (a) (b) (c) Survival 2014 Survival 2013 Survival 2012 Asteraceae Fabaceae Poaceae Native Exotic Invasive Native Exotic Invasive a a b a b a b a b 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 native exotic invasive Survival 2014 control fenced a a a a a a Figure 5: Three years of survival data for native, noninvasive exotic, and invasive plants in control (white bars) and fenced (gray bars) plots. Bars indicate mean ± SE. Status and family means with the same letter are not significantly different (P ! 0.05) based on post-hoc contrasts. 45 0 2 4 6 8 10 12 14 16 18 native exotic invasive Biomass (g) Poaceae Status control fenced Native 0 10 20 30 40 50 60 70 native exotic invasive Biomass (g) Fabaceae Status control fenced 0 1 2 3 4 5 6 native exotic invasive Biomass (g) Asteraceae Status control fenced Native Exotic Invasive Native Exotic Invasive Exotic Invasive 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 native exotic invasive Survival 2014 control fenced Asteraceae Fabaceae Poaceae Biomass (g) Biomass (g) Biomass (g) a a ab b b b cd b c d d a b a c c bc bc (a) (b) (c) Figure 6: Biomass data for native, noninvasive exotic, and invasive plants in control (white bars) and fenced (gray bars) plots, divided by plant family. Bars indicate mean ± SE. Within family, means with the same letter are not significantly different (P ! 0.05) based on post- hoc contrasts. 46 higher than that of native s or noninvasive exotics (Fig. 6 b). Vegetative biomass was unaffected by insecticide and fungicide treatments (Table 1). Flower Number and Reproductive Biomass : Flower number increased when mammalian browsers were removed, and was highest for the Fabaceae in fenced plots (family x fencing interaction: F 2,20 = 8.85, P = 0.002). Plants in the Fabaceae produced by far the most flowers, while species in the Poaceae produced the least (family: F 4,14 = 13.02, P < 0.001). Invasive species produced more flowers than did noninvasive exotic and native plants (status: F 4,14 = 2.98, P = 0.005), however, though not statistically significant (status x family x fencing interaction: F2,23 = 0.84, P = 0.50), this pattern is drive n by the Fabaceae family (Fig. 7 b). Reproductive biomass differed between native, noninvasive exotic, and in vasive species (status: F 2,37 = 13.56, P < 0.001), but this relationship depended on plant family and browser removal (Table 1; status x fencing interaction: F 2,29 = 3.58, P = 0.04; status x family interaction: F4,38 = 3.87, P = 0.01). Invasive plants prod uced significantly more floral biomass when fenced, while floral biomass for native and noninvasive exoti c species was unaffected (Fig. 7 d-f). In the Asteraceae and Poaceae, native species had the highest floral biomass; in the Fabaceae invasive species ha d the highest f loral biomass (Fig. 7 d-f). Tolerance Species differed in their ability to tolerate insect herbivory and regrow vegetative biomass and produce flowers after damage. Tolerance to insect herbivory depended on speciesÕ status (status x herbivory interaction; vegetative biomass: F 2,375 = 3.95, P = 0.02; flower number: F2,11 = 5.46, P = 0.005) (Table 2). Native species were able to, on average, compensate for insect herbivore damage in their regrowth of vegetative biomass and flowers, whil e noninvasive exotics 47 undercompensated and produced less biomass when damaged (Fig. 8a). Invasive species undercompensated for herbivory and produced fewer flowers when damaged, but fully compensated in term s of vegetative regrowth (Fig. 8 c). Species iden tity contributed to tolerance above and beyond the variation explained by status for vegetative biomass regrowth and flower 0 2 4 6 8 10 12 native exotic invasive Floral Biomass (g) Poaceae Status control fenced (f) 0 2 4 6 8 10 12 native exotic invasive Flower Number Poaceae Status control fenced 0 100 200 300 400 500 600 native exotic invasive Flower Number Fabaceae Status control fenced 0 2 4 6 8 10 12 14 16 18 20 native exotic invasive Flower Number Asteraceae Status control fenced 0 10 20 30 40 50 60 70 native exotic invasive Floral Biomass (g) Fabaceae Status control fenced 0 1 2 3 4 5 6 7 8 native exotic invasive Floral Biomass (g) Asteraceae Status control fenced Native Native Exotic Invasive Native Exotic Invasive Exotic Invasive 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 native exotic invasive Survival 2014 control fenced Asteraceae Fabaceae Poaceae Flower Number Flower Number Flower Number a a ab b b b a b b b a b a a a a a Native Native Exotic Invasive Native Exotic Invasive Exotic Invasive Asteraceae Fabaceae Poaceae Reproductive Biomass (g) Reproductive Biomass (g) Reproductive Biomass (g) a a ab b b c a c d a b a c bc ab (a) (b) (c) (d) (e) Figure 7: Flower number (a-c) and reproductive biomass (d-f) data for native, noninvasive exotic, and invasive plants in control (white bars) and fenced (gray bars) plots, divided by plant family. Bars indicate mean ± SE. Within family, means with the same letter are not significantly different (P ! 0.05) based on post-hoc contrasts. 48 production after damage (vegetative biomass: 2 = 5.0, P = 0.03 ; flower number: 2 = 85413, P < 0.001). The ability to maintain reproductive biomass when damaged did not vary based on species identity ( 2 = 0.0, P = 0.80) or plant status (Fig. 8 a; F 2,154 = 1.19, P = 0.31). Tolerance to mammalian browsing, or the ability to maintain vegetative and reproductive biomass when browsed, depended on status but these effects varied across families (damage x status x family interaction; vegetative biomass: F 3,1105 = 6.91, P < 0.001; reproductive biomass: F2,405 = 3.89, P = 0.02) (Table 2). In the Asteraceae, native and invasive species overc ompensated and produced more vegetative and/or reproductive biomass (invasives only) when browsed, while exotic species compensated for browsing and maintaine d similar biomass levels (Fig. 8 d). . In the Fabaceae, invasive species undercompensated and produ ced less vegetative and Table 2: Results from tolerance mixed model analysis of variance (ANOVA) testing the fixed effects of plant status (native, noninvasive exotic, invasive) and enemy damage (insect herbivory and mammalian browsing) on plant performance (vegetative biomass, reproductive biomass, and flower number). Vegetative and reproductive biomass and damage data were log transformed to fit normality assumptions. Statistically significant (P ! 0.05) effects are in bold. Source dfF!2PdfF!2PdfF!2Pstatus 2, 425.140.012, 400.120.892, 11 5.642.90.09family 2, 404.920.012, 335.490.0092, 11 8.7015.2<0.001 herbivory 1, 4286.670.011, 960.100.751, 93.783.40.06status x herbivory 2, 3753.950.022, 1541.190.312, 11 5.4610.60.005status x family 4, 423.610.014, 341.300.294, 150.451.70.78family x herbivory 2, 3681.840.162, 1540.050.952, 170.11 0.20.89status x family x herbivory 4, 3452.170.073, 1940.580.634, 210.381.50.83Random Effects (species)status 187.1<0.001 70.0<0.001 61.6<0.001 species x herbivory 5.00.030.00.8085413<0.001 status 2, 804.11 0.012, 590.11 0.902, 97.223.60.16family 2, 115 0.980.382, 1260.090.922, 911.87 20.0<0.001 browsing 1, 1106 2.460.121, 4300.020.901, 98.336.50.01status x browsing 2, 1105 2.630.072, 4052.800.062, 11 0.931.90.39status x family 4, 900.990.404, 471.770.154, 150.592.20.71family x browsing 2, 1105 6.330.0022, 4220.430.652, 171.442.90.24status x family x browsing 3, 1105 6.91<0.001 2, 4053.890.022, 190.380.80.67Random Effects (species)status 67.0<0.001 17.3<0.001 5.50.06species x browsing 0.01.000.00.9010111 <0.001 (b) Mammal Browser Mixed Model ANCOVA Reproductive Biomass (g) Flower Number (a) Insect Herbivory Mixed Model ANCOVA Vegetative Biomass (g) 49 -3 -2.5 -2 -1.5 -1 -0.5 0 -0.6 -0.1 0.4 Log (reproductive biomass) Log (mamalian browsing) nativea exotica invasivea 0 2 4 6 8 10 12 14 16 18 -0.6 -0.1 0.4 Flower number Log (mamalian browsing) native exotic invasive -4 -3 -2 -1 0 1 2 3 -0.6 -0.1 0.4 Log (vegetative biomass) Log (mamalian browsing) nativef exoticf invasivef -1 -0.5 0 0.5 1 1.5 2 2.5 3 -0.6 -0.1 0.4 Log (vegetative biomass) Log (mamalian browsing) nativea exotica invasivea 0 2 4 6 8 10 12 14 16 18 0 1 2 3 4 Flower number Log (insect herbivory) native exotic invasive -2 -1.8 -1.6 -1.4 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0 0 1 2 3 Log (reproductive biomass) Log (insect herbivory) native exotic invasive -2.5 -2 -1.5 -1 -0.5 0 0.5 0 1 2 3 4 Log (vegetative biomass) Log (insect herbivory) native exotic invasive Tolerance to Insect Herbivory Tolerance to Mammalian Browsing Ln (Insect Herbivory) Flower Number R2 = 0.29 Ln (Vegetative Biomass [g]) Ln (Vegetative Biomass [g]) (a) (b) (d) R2 = 0.33 Flower Number R2 = 0.43 Ln (Reproductive Biomass [g]) (c) Ln (Vegetative Biomass [g]) -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 -0.6 -0.1 0.4 Log (reproductive biomass) Log (mamalian browsing) nativef exoticf invasivef (f) Ln (Reproductive Biomass [g]) Ln (Reproductive Biomass [g]) Ln (Mammalian Browsing) Ln (Mammalian Browsing) (e) (h) R2 = 0.38 R2 = 0.45 R2 = 0.38 R2 = 0.33 R2 = 0.33 (g) Fabaceae Asteraceae All All Figure 8: The effects of plant status and family on tolerance to insect herbivory (a-c) and mammalian browsing (d-h) in the field. Negative slopes indicate undercompensation and that the biomass and flower number of damaged plants is less than that of undamaged plants (undercompensation). Positive slopes indicate increased mass and flower number due to damage (overcompensation), and values of zero indicate compensation and no net change in growth rate. Solid regression lines show slopes that are significantly different from zero (P ! 0.05), and dashed lines represent slopes that are not significantly different from zero. 50 reproductive biomass when browsed. Native and exotic species were able to compensate for mammalian browsing and produce similar levels of vegetative and reproduct ive biomass when browsed (Fig. 8 g). In the Poaceae, most individuals that survived until biomass harvest did not receive any browsing damage and there was insufficient variation to calculate slopes using regression. For flower number, species differed in their ability to tolerate mammalian browsing (2 = 10111, P < 0.001), but this relationship did not depend on status (F 2,11 = 0.93, P = 0.39) (Table 2). Native, invasive, and noninvasive exotic species all produced fe wer flowers when browsed (Fig. 8 f). Time Invasive species with longer residence times benefited more from mammalian browser removal than invasives more recently introduced (Fig. 9 , Table 3). Invasives with residence times in Michigan equal to or longer than 120 years grew significantly larger and tended to produce more flowers when fenced (status x fencing x time; vegetative biomass: F 1,510 = 16.88, P < 0.001; flower number: F 3,15 = 8.47, P = 0.04). Species with residence times less than 120 years did not benefit from mammalian browser removal. Fe ncing effects on reproductive biomass did not differ based on status or residence time (status x fencing x time: F 1,284 = 0.01, P = 0.91). The insecticide treatment did not affect species performance for any status or residence time (Table 3). Discussion I found no evidence that enemy release is a general mechanism contributing to invasiveness in this system. Invasive species were affected by enemy removal treatments, 51 similarly to native and noninvasive exotic species, and their performance was affected t o the same degree by insect herbivores, mammalian browsers, and fungal disease. These findings are consistent with the few other studies that have experimentally manipulated enemy presence, which also found invasive species to be equally affected by biotic interactions (Parker et al. 2006, Heard and Sax 2012, Stricker and Stiling. 2012, Dawson et al. 2014). A meta -analysis of studies that manipulated biotic interactions on native, invasive, and noninvasive exotic species found that, in general, species bene fitted from the removal of enemies, however this effect did not differ between native, noninvasive exotic and invasive species (Chapter 5). Community Complexity and Enemy R elease Removal of mammalian browsers was the only treatment to significantly affect performance of experimental plants, and I found no significant interactions between insect herbivore, mammalian browser, or fungal disease removal treatments. These results suggest that escape from multiple enemies also is unlikely to explain invasiveness in this system. These Model Terms dfF!2PdfF!2PdfF!2Pstatus 1, 226.260.021, 281.840.191, 91.951.940.16time 1, 220.130.731, 281.770.191, 90.791.1350.29insecticide 1, 370.000.961, 2831.810.181, 92.492.3060.13fencing 1, 412.630.11 1, 28410.350.0011, 942.336.5130.01status x fencing 1, 5042.520.11 1, 2832.460.121, 1252.7421.79<0.001 status x time 1, 221.570.221, 280.130.721, 120.310.20.70time x fencing 1, 5009.900.0021, 2830.930.341, 12441.40390.6<0.001 status x time x fencing 1, 51016.88<0.001 1, 2840.010.913, 158.478.50.04Random Effects (species)status 275.6<0.001 141.5<0.001 23300<0.001 (plot)insecticide 01.0001.0001.00(plot)fencing 9.90.0070.11.0001.00Flower Number Vegetative Biomass (g) Reproductive Biomass (g) Table 3: Results from mixed model analysis of variance (ANOVA) testing the fixed effects of plant status (noninvasive exotic, invasive), insecticide treatment (sprayed, control), fencing treatment (fenced, control), and time on plant performance. Vegetative and reproductive biomass data were log transformed to fit normality assumptions. Chi square statistics based on log-likelihood ratio tests are presented for random factors. Statistically significant (P ! 0.05) effects are in bold. 52 findings are consistent with other studies that have manipulated multiple biotic interactions on native and introduced species. Prior to this study, 23 studies manipulated more than one biotic interaction on native and introduced sp ecies, and of those, nine found interactive effects of their treatments (Chapter 5). Only one previous study factorially manipulated three biotic interactions (competition, soil microbial community, and arbuscular mycorrhizal fungi) on invasive Centaurea s toebe and native Ammophila breviligulata and found no interactive effects between 0 1 2 3 4 5 6 exotic short exotic long invasive short invasive long Ln Flower Number control fenced 0 1 2 3 4 5 6 exotic short exotic long invasive short invasive long Ln Flower Number control fenced -2 -1 0 1 2 3 4 exotic short exotic long invasive short invasive long Ln Biomass (g) control fenced (b) (a) < 120 years ! 120 years < 120 years ! 120 years exotic invasive Ln Biomass (g) Ln Flower Number !"!!"!"Figure 9: Natural log flower number (a) and natural log reproductive biomass (b) data for noninvasive exotic and invasive plants in control (light gray bars) and fenced (dark gray bars) plots, divided by species residence time. 53 treatments (Emery and Rudgers 2012). This study is the first to factorially manipulate three biotic interactions for a diverse set of native, noninvasive exotic, and invasive species, but also finds that escape from multiple enemies does not explain invasiveness. Despite the fact that enemies had similar effects across plant status, invaders on average had higher biomass than native and noninvasive exotic species. This pattern was primarily driven by invasive Fabaceae, which were the largest and highest performing species in the experiment. Invasiveness in the Fabaceae may occur despite strong effects from enemies and instead, may be related to performance traits. Consistent wi th this hypothesis, a meta -analysis comparing the traits of invasive and exotic species found that i nvasiveness in plants is positively associated with performance -related traits, such as growth rate, size, and flower and seed number (van Kleunen et al. 20 10). Interestingly, invasives tended to have higher biomass, reproductive biomass, and flower numbers, but only for those individuals that were less damaged by herbivores and browsers. Across years and growing seasons, enemy abundance and damage can fluctuate greatly ( Agrawal and Kotanen 2003, Agrawal et al. 2005, Parker and Gilbert 2007), and i nvasive species may be better able to exploit opportunity windows during periods with low enemy damage, while performing similarly to native and noninvasive ex otics when enemy damage is high. During the first year of their study, Agrawal and Kotanen (2003) found that introduced plants were more damage by herbivores than were natives, similar to our own system (Schultheis et al. 2015). However, during the second year of their study, they found that introduced plants were less damaged, supporting the ERH ( Agrawal et al. 2005). Therefore, invaders in their system may be able to dominate during these periods of low enemy damage. However, in this system, invasive spec ies consistently received more damage from insect herbivores and mammalian browsers 54 across years (Schultheis et al. 2015), though enemy fluctuations may occur over longer timescales, providing opportunity windows in some years. Tolerance Because invasive species in this system tended to experience higher insect herbivore and mammalian browsing damage than did native and noninvasive exotic species, I hypothesized that tolerance could play a role in invasive success (Schultheis et al. 2015). However, in this study I found that invasive species were generally no more tolerant to insect herbivory or mammalian browsing than were native and noninvasive exotic species. Invasive and native species compensated equally for insect herbivore damage, although reduced to lerance of noninvasive exotics may explain their lack of invasion success. In general, the findings presented here are similar to those of a recent review of studies that measured both damage and performance on native and invasive plants, which found no ev idence that invasive plants were more tolerant than natives (Chun et al. 2010) . Additionally, Ashton and Lerdau (2008) found no difference in tolerance among native, noninvasive exotic, and invasive species under field levels of insect herbivory and mammal ian browser damage. Interestingly, invasive Asteraceae overcompensated for mammalian browsing and were more tolerant than native Asteraceae. I included 13 noninvasive exotic Asteraceae species in the experiment (Table C1), however there was only one invasi ve Asteraceae, Centaurea stoebe . Though these results for invasive species cannot be extrapolated beyond this one species, the data indicate that tolerance could play an important role in the success of this invader and for noninvasive exotics in general. In contrast, invasive Fabaceae undercompensated for browsing while native species fully compensated; therefore, increased tolerance cannot explain 55 invasiveness in this family. However, tolerance measures that do not take into account the full lifespan of a plant should be interpreted with caution, as they do not represent lifetime performance (Stowe et al. 2000). For annuals and biennials in this experiment, tolerance measures represent lifetime values, however perennials may have survived if not harvested. Compensation could be due to reallocation of belowground biomass to aboveground tissues, which could result in lower lifetime performance even though we observe no decrease in performance in one growing season. For example, while certain Ipomopsis aggrega ta genotypes exhibit overcompensation over their lifetime after mammalian browsing (Paige and Whitham 1987, Paige 1992), this species represents the extreme end of a continuum of known plant responses to antagonistic interactions (Maschinski and Whitham 19 89). Dynamic I nvasions While enemy release was not operating generally for invaders, I found evidence that release is dynamic and experienced only by invasives with shorter residence times in their introduced range. Both invasive and noninvasive exotic species with residence tim es equal to or longer than 120 years tended to produce more flowers and vegetative biomass when fenced, while species with residence times shorter than 120 years on average did not. These results indicate that both exotic and invasive species that have bee n in their introduced range for less than 120 years experienced minimal effects from mammalian browsers, supporting the ERH. Enemy release could be lost over time in this system due to accumulation of both native and introduced enemies on introduced plant s, which switch from species found in the resident community. Both disease and herbivore pressure can be driven local population abundances and phylogenetic relatedness to the resident community ( Agrawal and Kotanen 2003, Parker and 56 Gilbert 2007, Dost⁄l et al. 2013, Parker et al. 2015). Compared to species that grow in less dense populations, invasive species growing in dense populations are expected to accumulate pathogens at a faster rate ( Bever 1994, Mitchell et al. 2010). Pathogens and herbivores are mo re likely able to infect close relatives of their hosts (Gilbert and Webb 2007, Go§ner et al. 2009, Pearse and Hipp 2009, Hill and Kotanen 2010). Additionally, plants with longer residence times have greater geographic spread (Ahern et al. 2010), raising t heir encounter rates with novel enemies as they enter new habitats and come into contact with more enemy species ( Go§ner et al. 2009, Flory and Clay 2013 ). Treatment Effects on Background Community Given that this experiment was conducted in an existing old field community, enemy removal treatments may have had unintended effects on the resident community, contributing to the observed patterns. For example, fungicide increased the amount of thatch pr esent in experimental plots by 8.5% ( F1,28 = 4.8, p = 0.04), potentially by reducing the number of fungal decomposers in the community (Appendix D) . Accumulation of thatch from grass leaf tissue can alter abiotic conditions in a habitat, for example reduci ng light availability, inhibit ing nitrogen fixation and CO 2 uptake in the soil, and decreasing soil temperatures (Knapp and Seastedt 1986) . Similarly, fencing increased the standing stock of the background community by 18.6% ( F1,28 = 6.38, p = 0.02), poten tially increasing the competitive environment for experimental seedlings (Appendix D) . Removal of mammalian browsers also would have removed the disturbance effect from trampling by ungulates, which potentially favors introduced species ( Vavra et al. 2007) . At this field site, Odocoileus virginianus (white tailed deer) occurred at a moderate to high density of ~30 individuals per square mile ( MDNR 2010). Therefore, removal of deer and other 57 mammals may have harmed species that are adapted to ungulate feeding and trampling, while benefitting those that are more palatable to these browsers ( Augustine and McNaughton 1998, Vavra et al. 2007). Native grasses, which have many adaptations decreasing vulnerability to mammalian browsing, such as the presence of silica in tissues and rhizomatous growth form , had reduced biomass in fenced plots where ungulates were excluded, potentially because of increased competiti on in these treatments. These non -target effects on background community may have negated any direct effects of enemy removal on plant performance, and could have altered the strength or direction of treatments on experimental plants. Conclusion Due to accidental and purposeful transport of species into new regions, today introduced species are present in most communities across the globe (Lonsdale 1999). Introduced plants make up 34% of the flora in Michigan, and an average of 55.5 new species establis h in Michigan each decade (Ahern et al. 2010) . Loss of antagonistic biotic interactions during the introduction process is hypothesized to drive the population growth and success of invasive species. In this study, I manipulated three classes of enemies to study their effects on multiple native, invasive, and noninvasive exotic speciesÕ performance. I did not find evidence that ERH was a general mechanism explaining the success of invaders in this system. However, tolerance and competitive traits may explai n the invasive success of some taxa (e.g., Asteraceae and Fabaceae respectively). Though ERH was not generally supported , I found evidence that ERH is dynamic and potentially lost with increasing time in the introduced range . Thus, while ERH may not be a universal mechanism behind the success of all invaders, it may still be important for some species at certain points in the invasion process (Heger and Jeschke 2014). 58 CHAPTER 4 : COMPETITIVE ABILITY, NOT TOLERANCE, MAY EXPLAIN SUCCESS OF INTROD UCED PLANTS OVER NATIVES Introduction Invasive species are one of the greatest threats to biodiversity (Vil‹ et al. 2011, Powell et al. 2011, Powell et al. 2013) , on par with habitat destruction and climate change ( Sala et al. 2000, Tylianakis et al. 200 8). While invasive species are rarely competitively dominant or major components in their native systems, in novel communities they often have larger populations, grow more densely, have higher fitness, and are able to outcompete natives (Hinz et al. 2004, Vil‹ et al. 2011, but see Firn et al. 2011). B iologi sts have struggled to identify the underlying mechanisms driving invasiveness and the effects of invaders on native communities. One feature that is shared by all introduced species is that, during the process of introduction, they disassociate from many biotic interactions from their native range while simultaneously forming new biotic interactions in their introduced range (Hallet t 2006, Mitchell et al. 2006). Novel biotic interactions could influence the performance of introduced species and promote invasiveness. For example, release from antagonists has been hypothesized to play a role in the prolific success of some of the most invasive species (Thellung 1912, Elton 1958 ). Additionally, evolutionary naŁve native species may be more susceptible to novel competitive mechanisms, leading to decreased performance of native populations and loss of community diversity (Callaway and Asch ehoug 2000). Enemy Release and T olerance Enemies, such as insect herbivores, mammalian browsers, competitors, disease, and predators, may all regulate the population dynamics and performance of native and introduced 59 plants ( Harper 1977, Louda 1982, Crawl ey 1989 , Levine et al. 2004, Chapter 5). Therefore, loss of key enemies from the native rage may explain the increased performance experienced by invasive species in their introduced ranges (Enemy Release Hypothesis [ERH]; Elton 1958, Callaway and Aschehou g 2000, Maron and Vila 2001 , Keane and Crawley 2002). However, comparisons between native and introduced range populations and between native and invasive species in introduced communities find that invaders are not consistently less damaged by enemies (Ch un et al. 2010, Dost⁄l et al. 2013 ), and often times are more damaged ( Colautti et al. 2004, Torchin and Mitchell 2004, Agrawal et al. 2005, Carpenter and Cappuccino 2005, Morrison and Hay 2011, Dawson et al. 2014, Schultheis et al. 2015). Therefore, the ability to maintain performance when damaged may play an important role in invasiveness. Invasive species may not be those released from enemy damage, but instead those that are better able to tolerate high levels of damage in their introduced ranges ( Maro n and Vila 2001, Keane and Crawley 2002) . Plants defend against enemies in two ways Ð resistance and tolerance (Marquis 1992, Stowe et al. 2000). Resistance traits reduc e the amount of enemy damage sustained, while tolerance traits allow the plant to main tain performance once damaged ( Strauss and Agrawal 1999, Stowe et al. 2000) . Plant architecture and resource allocation patterns both contribute to tolerance; for example, individuals that store more resources belowground may be more tolerant to abovegroun d damage ( Hochwender et al. 2000 ). Additionally, plants with a greater number of meristems can be more tolerant to herbivory; for instance, mammalian browsing can release from dormancy secondary meristems when the primary meristem is damaged in grass speci es (Olson and Richards 1988). Tolerance can also vary based on traits related to performance under different abiotic conditions, for example tolerance to antagonistic soil microbes was correlated 60 with ability to maintain performance under low -light condit ions in 21 tropical tree species (McCarthy -Neumann and Kobe 2008). Competitive Ability of Invasive Species The mechanisms responsible for invasive species establishment and effects on the native community are rarely identified (Levine et al. 2003). Howeve r, a review of the studies that identified mechanisms found that most invaders had strong negative effects on native community members through competition for resources like light and water and through allelopathy (Levine et al. 2003). Thus, successful inv aders may be those plants that are competitively superior in their new communities ( Crawley 1987, Hinz and Schwarzlaender 2004 , van Kleunen et al. 2010, Dawson et al. 2014) , utilizing resources more efficiently and growing larger and more densely in their introduced range (Hinz et al. 2004). Additionally, invaders may compete through mechanisms novel to the community, such as allelochemicals not previously present, leading to competitive dominance of naŁve native neighbors (Callaway and Aschehoug 2000). Both herbivory and competition contribute to biotic resistance of the native community to invasion (Levine et al. 2004), yet f ew studies have explored the effects of competition and herbivory on invaders simultaneously (Heard and Sax 2013 ), while many have studied them independently (Chun et al. 2010, Levine et al. 2004 , Chapter 5 ). Release from enemies may increase competitive ability, by making more resources available for competitive traits, or over longer timescales as invasive species evolve to reallocate resources from defensive to competitive traits ( Blossey and Nıtzold 1995 ). The simultaneous manipulation of both herbivory and competition not only tests the two major hypothesis addressing invasive species success and 61 effect on nativ e species, but could reveal non -additive or synergistic effects that can not be observed when both are studied in isolation. Here, we test whether invasive species are more tolerant to herbivory or are more competitive compared to native and noninvasive ex otic species. Using a manipulative greenhouse experiment we ask the following questions: (1) Do invasive plant species have higher tolerance to simulated herbivory compared to native and noninvasive exotic plants? (2) Do invasive plants demonstrate a great er competitive ability (competitive effects and response) than native and noninvasive exotic plant species? (3) Are the effects of competition and herbivory synergistic, reducing performance to a greater degree when both are present? If tolerance contribut es to invasiveness, we predict that invaders will experience minimal effects from simulated herbivory, while native and noninvasive exotic species will be more negatively affected. If competitive ability contributes to invasiveness, we predict that invader s will both experience minimal effects from the presence of a competitor, while simultaneously reducing native speciesÕ performance to a greater degree than native and noninvasive exotic species. Finally, if competition and simulated herbivory have synergi stic effects on performance, we expect plants grown in the presence of clipping and competition to have reduced performance below that predicted by the additive effects of both treatments. Methods Study Species In our study, we included 19 old field plant species representing three of the four plant families (n = 6 Asteraceae, 6 Fabaceae, and 7 Poaceae) that have contributed most to invasive plant species in Michi gan (Ahern et al. 2010) (Table 4 ). We categorized species as native , 62 noninvasive exotic, or invasive (n = 7, 5, and 7 species respectively) based on presence on local invasive species lists and herbarium records, and in consultation with local land managers (Schultheis et al. 2015). Invasive and noninvasive exotic species are both introduced to Mi chigan from outside the U.S. by human actions, either accidentally or intentionally (Reznicek et al. 2011). Noninvasive exotic plants assimilate into the native community with little effect, while invasive plants aggr essively colonize natural areas and thr eaten biodiversity and human interests . Experimental Design To test tolerance to herbivory an d competition, we initiated a greenhouse experiment at the W.K. Kellogg Biological Station, factorially manipulating simulated herbivory (clipping FamilySpecies Name Abbrev. Status Asteraceae Centaurea cyanus CENCY exotic Asteraceae Coreopsis tinctoria CORTI exotic Asteraceae Sonchus oleraceus SONOL exotic Asteraceae Centaurea stoebe CENST invasive Asteraceae Coreopsis lanceolata CORLAnative Asteraceae Erigeron annuus ERIAN native Fabaceae Coronilla varia CORVA invasive Fabaceae Lespedeza cuneata LESCU invasive Fabaceae Lotus corniculatus LOTCOinvasive Fabaceae Melilotus officinalis MELOF invasive Fabaceae Desmodium canadense DESCA native Fabaceae Lespedeza capitata LESCA native Poaceae Bromus hordeaceus BROHO exotic Poaceae Poa trivialis POATR exotic Poaceae Bromus inermis BROIN invasive Poaceae Poa compressa POACO invasive Poaceae Bromus kalmii BROKA native Poaceae Elymus canadensis *ELYCA native Poaceae Sorghastrum nutans SORNU native Table 4: List of the 18 experimental species, and one competitor species, used in the experiment, along with their family and status designation. The competitor species, Elymus canadensis , is indicated with an *. 63 treatment, con trol) and competition (competitor present, absent ) (n = 5 replicates per species per treatment) (N = 370 pots) . In addition, we included ten replicates of the competitor species, Elymus canadensis , grown alone , half of which were subjected to the c lipping treatment. We germinated seeds and then directly transplanted experimental seedlings into 656ml pots (D40 Deepots, Stuewe & Sons, LLC.) containing a mixture of potting soil (Sunshine Mix #5; SunGro Horticulture Canada Ltd., Alberta, Canada), peat moss (Pro -Moss Hort, Premier Tech Ltd, Pennsylvania USA), sand (Tubesand Quikrete International, Inc, Georgia, USA) and perlite (Horticultural Perlite, Midwest Perlite, Wisconsin, USA) in a 3:3:3:1 ratio on 20 June 2013 . We watered plants as needed during the cours e of the experiment. Three weeks after planting, we added 50 mL of water to each pot containing dissolved fertilizer (Miracle -Gro Al l Purpose Plant Food, NPK 24 -8-16) at a concentration of 1.2 g/L. The location of each species and treatment was randomized at the pot level. Pots were spaced a minimum of 12cm apart to prevent shading and light competition between seedlings not growing within the same pot. To manipulate competition, we grew half of our experimental seedlings in pots alone, while the other half grew with one individual of a competitor species, Elymus canadensis (Canada wild rye) . Elymus canadensis is a grass native to Michigan, and was chosen as our competitor species because it overlaps in geographic range and habitat preference with all of our experimental species. On 12 August 2013 we administered a simulated herbivory treatment to half of our e xperimental seedlings. We measured the height of each seedling and clipped 50% of each individualÕs height, which was similar to herbivory from insect herbivores and mammalian browsers observed on these species in the field (Schultheis et al. 2015). On 2 October 2013 we harvested the experiment and measured plant performance metrics on both the experimental species and the competitor E. canadensis , including height (cm) from 64 the soil surface to apical meristem, aboveground biomass (g), and flower number. F lower number analysis and results can be found in Appendix E, but are not presented in the main text because most species produced no flowers during the course of the experiment, and because flower number data could be misleading due to differences in phen ology between experimental species . Harvested biomass was dried at 65 ¡C for three days and weighed. Statistical Analysis Tolerance and Competitive Response : We performed all analyses in R ( v. 3.2.0, R Core Team 2015 ). To determine whether our treatments influenced plant performance, we tested the effects of simulated herbivory and competition on plant biomass and height with mixed model ANOVA using the lmer function in the lme4 package in R (v. 1.1-7, Bates et al. 2015 ). Our models included plant biomass (g) or plant height (cm) as response variables and clipping treatment (clipped, unclipped), competition treatment (competitor present, absent), status (native, noninvasive exotic, invasive), family (Asteraceae, Poaceae, Fabaceae), and all possible interactions as fixed predictor variables. A significant negative effect of our competition treatment indicates a negative competitive response in our experimental species. A significant negative effect of our clipping treatmen t indicates that performance is reduced when clipped, indicating a negative tolerance value (undercompensation). Full compensation occurs when an individualÕs performance is the same in the presence and absence of clipping, and overcompensation results whe n clipping increases individual performance (Strauss and Agrawal 1999). Non -significant (p > 0.05) interaction terms were dropped from final models to increase our power to detect significant main effects. The number of species in each status is the unit of replication for questions on whether 65 treatment effects differed between native, noninvasive exotic and invasive species, so we included species nested within status as a random factor in our models. To determine whether species responded differently to our treatments, we included species x clipping, species x competition, and species x competition x clipping interactions as random terms in our models. Because we were interested in proportional responses to our treatments, and to improve normality, heigh t and biomass data were natural log transformed for analysis. P -values for fixed effects were obtained using the lmerTest package in R, and for random terms we used chi -squared tests and the rand function (v. 2.0-20, Kuznetsova et al. 2015). Competitive E ffects on Elymus canadensis : To determine whether invasive, noninvasive exotic, and native species differ in competitive effects, we tested the effects of competitor identity on E. canadensis performance with mixed model ANOVA. We measured competitive effe ct as the degree to which our experimental species reduced E. canadensis performance. Our model included E. canadensis biomass (g) or height (cm) as the response variable and competitor status (native, noninvasive exotic, invasive), competitor family (Asteraceae, Fabaceae, Poaceae), Source dfF!2PdfF!2Pstatus 2, 182.510.11 2, 181.230.32family 2, 180.040.962, 181.960.17clipped 1, 5420.35< 0.001 1. 1811.58 0.003competition 1, 5427.00< 0.001 1, 366.090.02clipped x competition 1, 540.310.581, 360.450.51Random Effects (species)status 33.8< 0.001 27.3< 0.001 species x clipped 0.01.000.50.50species x competition 0.01.000.01.00species x competition x clipped 2.20.100.50.50Biomass (g) Height (cm) Native, Exotic, and Invasive Species Table 5: Results from mixed model analysis of variance (ANOVA). Results show the effects of plant status (native, noninvasive exotic, or invasive), family (Asteraceae, Fabaceae, Poaceae), clipping treatment (clipped, control), and competition treatment (competitor present, no competition) on experimental plant biomass and height. Statistically significant (P ! 0.05) effects are in bold. 66 and whether the competitor species received the clipping treatment (clipped, unclipped), a nd all possible interactions as fixed predictor variables. No E. canadensis individuals flowered during the course of the experiment, so we were unable to determine competitive effects on fitness. Non -significant (p > 0.05) interaction terms were dropped f rom final models. We included competitor species nested within competitor status and the species x clipping treatment interaction as random factors in our models. All E. canadensis performance data was natural log transformed for analysis. Results Toleran ce and Competitive Response Invasive, noninvasive exotic, and native species responded similarly to treatments, indicating that invasive species are no more tolerant to simulated herbivory and respond similarly to competition (Table 5 , Fig. 10 ). Clipping and competition both reduced plant height and biomass (Fig. 10) , but there was no interaction between the clipping and competition treatments, meaning that effects were additive (clipped x competition; biomass: F 1,54 = 0.31, p = 0.58; height: F 1,36 = 0.45, p = 0.51). Surprisingly, clipping and competition reduced plant Source dfF!2PdfF!2PElymus canadensis Competitor competitor status 2, 224.670.022, 1903.280.04competitor family 2, 223.430.052, 1902.740.07competitor clipped 1, 1730.210.651, 1900.010.92Random Effect (comp. species)comp. status 2.00.200.01.00comp. species x comp. clipped 0.01.000.01.00Biomass (g) Height (cm) Table 6: Results from mixed model analysis of variance (ANOVA). Results show the effect of competitor status, family, and whether the competitor was clipped for Elymus canadensis biomass and height. Statistically significant (P ! 0.05) effects are in bold. Non-significant interaction terms were dropped from the final model. 67 biomass (species x clipped : $2 = 0.0, P = 1.0; species x competition $2 = 0.0, P = 1.00) and height (species x clipped : $2 = 0.5, P = 0. 50; species x competition $2 = 0.0, P = 1.00) similarly for all study species, indicating that species did not differ in tolerance or competitive response. Competitive E ffects on Elymus canadensis Competition marginally reduced Elymus canadensis biomass from 0.54g ± 0.07 to 0.40g ± 0.02 (mean ± SE ) (F 1,188 = 2.87, p = 0.09), and did not affect height (F 1,188 = 1.47, p = 0.23). Invasive and noninvasive exotic species had the greatest competitive effects on E. canadensis , (significant effect of s tatus on biomass: F 2,22 = 4.67, p = 0.02; height: F 2,190 = 3.28, p = 0.04; Fig. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Native Exotic Invasive Biomass (g) Status Clipped Control 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Native Exotic Invasive Biomass (g) Status In Competition Alone 0 5 10 15 20 25 30 35 Native Exotic Invasive Height (cm) Status Clipped Control 0 5 10 15 20 25 30 35 Native Exotic Invasive Height (cm) Status In Competition Alone 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Native Exotic Invasive Biomass (g) Status Clipped Control 0 5 10 15 20 25 30 35 Native Exotic Invasive Height (cm) Status In Competition Alone (a) (b) (c) (d) Figure 10: Biomass (a-b) and height (c-d) of native, noninvasive exotic, and invasive plants in clipped and unclipped treatments (a, c) or grown in the presence and absence of competition (b, d). Bars indicate mean ± SE. 68 11). Notably, of the six species with the strongest competitive effect on E. canadensis , three were invasive and three were noninvasive exotic species (Fig. 12). Native, invas ive, noninvasive exotic species exhibited similar competitive effects on E. canadensis when they were clipped or unclipped (competitor clipped x competitor status: p > 0.05), indicating that competitive ability was not affected by simulated herbivory. When grown with species in the Fabaceae, E. canadensis also tended to produce more biomass and was taller than when grown with species in the Asteraceae and Poaceae (Table 6 ). Elymus canadensis biomass, but not height, depended on 0 0.1 0.2 0.3 0.4 0.5 0.6 Native Exotic Invasive Elymus Biomass (g) Status 30 31 32 33 34 35 36 37 38 Native Exotic Invasive Elymus Height (cm) Status (a) (b) a b b ab a b Figure 11: Elymus canadensis biomass (a) and height (b) when grown with native, noninvasive exotic, or invasive species competitors. Bars indicate mean ± SE. Means with different letters are significantly different (P ! 0.05) based on post-hoc contrasts. 69 competitor species identity (biomass: F 18,171 = 2.17, p = 0.006; height: F 18,171 = 1.50, p = 0.11; Fig. 12). Discussion We found no evidence that invasive species were more tolerant to simulated herbivory or experienced less of a response to competition compared to native and noninvasive exotic species (Fig. 1 0). However, we found invasive and noninvasive exotic species exhibited the strongest 20 25 30 35 40 45 CONTROL CORVA SORNU LESCU POATR LOTCO LESCA CORTI CORLA ERIAN MELOF BROKA DESCA CENCY BROIN SONOL BROHO POACO CENST Elymus Height (cm) Species 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 CONTROL LESCA CORVA CORLA POATR LOTCO ERIAN SORNU CORTI LESCU BROKA MELOF DESCA POACO CENCY CENST BROIN BROHO SONOL Elymus Biomass (g) Species (a) (b) Figure 12: Elymus canadensis biomass (a) and height (b) when grown with different competing species. Bars are labeled by competing species status (native = white, noninvasive exotic = light gray, invasive = dark gray) and are ordered by descending mean values. The black bar indicates E. canadensis performance when grown alone. Bars indicate mean ± SE. 70 competitive effects on a native grass, E. canadensis. When grown with introduced competitors, E. canadensis produced less biomass and wa s shorter than when grown with natives, indicating that introduced species in this system may negatively effect native populations more so than other native competitors. Competitive Ability of Invasive Species : Response and Effects Invasive species often reduce native diversity and alter community structure through competitive effects on native species ( Vil‹ et al. 2011, Levine et al. 2003 ), and our study is consistent with a review by Levine and colleagues (2003) that found when native and introduced species competed, introduced species often had stronger competitive effects on natives than natives on introduced species. These competitive effects could be driven by three mechanisms. First, because of lack of a shared evolutionary history between an introduced species and the new community (Verhoeve n et al. 2009) , anta gonistic traits of the invader, such as allelopathy, often prove effective against native community members that lack previous experience with such tactics (Novel Weapons Hypothesis ; Callaway and Ridenour 2004). Second, invasive species may be those that are simply more competitive for limiting resources, for example effectively driving down light and resources levels, excluding other species. Third, escape from enemies may increase introduced species growth and competitive effects on surrounding species ( Keane and Crawley 2002, Klironomos 2002, Blair and Wolfe 2004). Allelopathy , or chemically mediated plant interactions (Rice 1974, Meiners and Kong 2012), contributes to many successful plant invasions and may explain some of the strongest competitive effects observed in this experiment (Callaway and Aschehoug 2000, Callaway and Ridenour 2004). The success of the invasive plant, Centaurea stoebe , is partially attributed to the 71 novel weapon ( Ð)-catechin, which it excretes from its roots serves as an allelochemical (Bias et al. 2003, but see Duke et al. 2009) . Soils supporting popul ations of C. stoebe in the invasive range contain levels of ( Ð)-catechin that are twice that found in the native range (Bias et al. 2003). This allelochemical negatively affects performance of natives in its invasive range (Ridenour and Callaway 2001), and to a lesser degree, natives in its native range (Bias et al. 2003). Interestingly, C. stoebe was still able to outcompete native species even when allelochemicals were inactivated using carbon, indicating that allelopathy only partially explains its compe titive dominance in its invasive range (Ridenour and Callaway 2001). Invasive Bromus inermus has allelopathic effects on native and introduced species (Stowe 1979); native grass growth was halved in patches containing B. inermu s, presumably due to the comb ined effects of competition and allelopathy (Dillemuth et al. 2009). In agricultural systems, fields planted with Melilotus officinalis had up to 97% lower unplanted weed densities due to direct competitive effects and release of allelochemicals by decompo sing tissues (Blackshaw et al. 2001). These three invaders exhibited some of the strongest competitive effects observed in our experiment, suggesting that allelopathy may play an important role in invasiveness in our system. Enemy Release and T olerance Interactions with native community members provide biotic resistance to introduced species, significantly reducing their performance (Levine et al. 2004), and often effect introduced species to an equal degree as natives (Chapter 5). Consistent with these findings, our simulated herbivory and competition treatments significantly reduced performance of invasive and noninvasive exotic species to a similar degree as natives. These results are also consistent with previous experiments in our system that determi ned that, in old field communities, invasive 72 species experienced similar performance effects of enemy damage and were no more tolerant of damage, compared to native and noninvasive exotic species (Schultheis et al. 2015). Similar to our own results, Ashton and Lerdau (2008) found that invasive temperate vine species were more damaged and were no more tolerant to browsing in the field, compared to native and noninvasive exotics. However, their simulated greenhouse manipulations revealed that invasive species were in fact more tolerant under controlled damage levels (Ashton and Lerdau 2008). Our clipping treatment was very similar theirs, where clipping stems removed 50% of all leaves, and we observed similar effects of our treatments on mean plant performance . The lack of higher tolerance exhibited by invasive species in our system could be due to the fact that invasive vines (Ashton and Lerdau 2008) and herbaceous species (this experiment) invade by different mechanisms. Although our simulated herbivory and competition treatments reduced performance on average, most species in our experiment were able to maintain performance when experiencing simulated herbivory and competition, however some species were negatively affected by our treatments. Our tolerance me asures represent just one growing season and should be interpreted with caution, as they do not represent species lifetime performance (Stowe et al. 2000). Compensation could be due to reallocation of belowground biomass to aboveground tissues, which may r esult in lower lifetime performance even though we observed no decrease in performance during the course of one growing season. Conclusion Due to unprecedented rates of transport of species across the globe, invasions are today common features shared by most ecosystems ( Lonsdale 1999 ). Invaders threaten biodiversity, 73 often outcompeting and displacing native species. Here, we find evidence that competitive effects of introduced species on a native species likely contribute to their negative effects on nat ive populations, but compared to native species, invasive and noninvasive exotic species were similarly affected by simulated herbivory and competition. 74 CHAPTER 5 : MUTUALISM GAIN AND COMPETITIVE ABILITY, NOT ENEMY RELEASE, MAY EXPLAIN SUCCESS OF INVASIVE SPECIES: A META -ANALYSIS Introduction Identifying the mechanisms underlying the success of invasive species is one of the most challenging and pressing goals in the field of invasion biology. However, after almost 60 years of intensive study, no prevailing mechanism has yet been identified (Elton 1958 ). Dozens of hypotheses attempt to explain the increased population growth, size, and competitive ability of invasive species ( van Kleunen et al. 2010), and many of th ese cite altered biotic interactions in the introduced range as potentially contributing to invasiveness (e.g. Enemy Release Hypothesis [Keane and Crawley 2002]) . Here, we performed a meta -analysis of the literature to test whether altered biotic interacti ons generally contribute to invasive speciesÕ success, and investigated whether invasive species are released from antagonistic biotic interactions or benefit more from acquired mutualists compared to populations in their native range or co -occurring nativ e and noninvasive exotic competitors in their introduced range. Biotic i nteractions are major drivers of plant and animal community structure and population dynamics ( Harper 1977, Crawley 1989, Louda 1982, Klironomos 2002, Morris et al. 2007). Thus, any al teration to these biotic interactions could have major effects on the populations of the species involved, be they native or introduced. For example, seed predators reduce plant population growth rates ( Louda 1982 ), and herbivores limit species abundance a nd distributions to subsets of available habitat (Lau et al. 2008). In animals, predators greatly reduce the density of prey species (Krebs et al. 1995), and pathogens alter host population dynamics (Anderson and May 1981). Alternatively, mutualists increa se individual performance, influencing local abundance and extending speciesÕ range sizes (Anacker et al. 2014 ). 75 When a species is introduced into a new community , it leave s behind native biotic interactions and en counters new mutualists, predators, herbivores, competitors, and diseases . Lack of a shared coevolutionary history between introduced species and new community members can lead to two potential outcomes (Elton 1958) : (1) Biotic release Ð an introduced species may interact weakly with its inv aded community, experiencing less damage from enemies, reduced suppression from competitors, and less benefit from mutualist partners. For example, native herbivores and pathogens may not recognize an introduced plant as a resource, which could result in r educed damage and increased fitness for the invader (Keane and Crawley 2002, Hallett 2006 ). Alternatively, (2) Biotic resistance Ð an introduced species may be equally, or even more , affected than native species by novel enemies and competitors , as it will have few evolved strategies to defend against them. For example, an introduced plant may lack defenses against unfamiliar herbivores, thus experiencing high attack rates and reduced fitness . In this case, intense novel antagonistic interactions may limit an introduced speciesÕ performance, preventing invasion (Levine et al. 2004 ). Two complementary approaches have been used to compare effects of biotic interactions on native and introduced species (Liu and Stiling 2006): (1) Cross continental comparisons, which compare biotic interaction effects on populations of a single species in its native and introduced range, and (2) Native/introduced comparisons, which compar e the effects of biotic interactions on co-occurring native and introduced species in a given location. Cross continental studies determine whether invasive species are doing something different in their new range compared to their native range (Hierro et al. 2005), while native/introduced comparison studies determine whether invasive s pecies are doing something different compared to competing native or noninvasive exotics species in their new range. 76 To u nderstand the role of biotic interactions in the invasion process, we must determine if and when positive and antagonistic interaction s affect the fitness of introduced and native species, and whether differences in the magnitude of fitness effect can explain increased performance for invasive species, relative to native and noninvasive exotic species (Maron and Vila 2001). Ideally, e xperiments would manipulate biotic interactions to study their effects on individual performance or population growth rates (Maron and Vila 2001, Keane and Crawley 2002, Liu and Stiling 2006 ), although correlational approaches can also provide evidence on the fitness effects of biotic interactions . Many manipulative studies testing the effects of biotic interactions on native and introduced species have been conducted, yet no quantitative synthesis has been performed to determine whether biotic release is a prevailing mechanism shared by most invaders. Previous meta -analyses have considered specific types of biotic interactions that may differ between native and invasive taxa (e.g., herbivory: Parker et al. 2006, Chun et al. 2010; plant -soil feedback: Kulmatisk i et al. 2008 , Suding et al. 2013). These studies all focused on one or two biotic interactions, or only considered introduced species without making comparisons to co-occurring native competitors or distinguishing between invasive and noninvasive exotics. For example, competitors, herbivores, and diversity of the resident community contribute to biotic resistance, reducing introduced species establishment and performance (Levine et al. 2004, Kimbro et al. 2013), but whether these effects are equivalent to those on native species was not considered. Another meta -analysis of nine studies found introduced species to be more affected by herbivores and disease than co -occurring native species, however invasive and noninvasive exotics were not considered separate ly (Chun et al. 2010). A meta -analysis on performance effects of herbivores by Parker and colleagues (2006) found no relationship between 77 invasiveness of introduced plants and herbivore effects, and compared these effects to those on native plants. Interes tingly, native plants were most controlled by introduced herbivores, and introduced plants were most controlled by native herbivores (Parker et al. 2006). Meta -analysis of plant -soil feedback (PSF) studies manipulating entire soil communities found that na tive species were most negatively affected by PSF, invasive species least, and noninvasive exotic species to an intermediate degree ( Kulmatiski et al. 2008) . Similarly, introduced species were more likely to develop soil communities that facilitate their own growth (positive PSF), while native species cultivated soil communities that were detrimental to growth (negative PSF) (Suding et al. 2013). Even though many reviews and meta -analyses exist, no analysis has yet considered the multitude of biotic intera ctions experienced by introduced species in nature, nor compared the fitness effects of different types of mutualistic and antagonistic interactions on native, noninvasive exotic, and invasive species. Here, we investigated whether altered biotic interacti ons during the process of introduction drive biological invasions. Our meta-analysis included studies that manipulated biotic agents to determine their effects on native, invasive , and noninvasive exotic speciesÕ performance (i.e., individual growth, fecun dity, survival, and population growth). We included studies conducted in both the native and introduced range of a species (cross -continental), or conducted in the introduced range on native and introduced species (native/introduced comparison). Because in troduced species commonly leave behind many strongly interacting species when they colonize new habitats, we predicted that introduced species (both invasive and noninvasive species) would be less strongly affected by biotic interactions, compared to co -occurring native species or populations in their historic range . Furthermore, if biotic release explains invasion success, we predicted that invasive species would be less affected by biotic 78 interactions than co -occurring noninvasive exotic species. Support for our latter prediction would provide evidence for biotic release as a general mechanism explaining invasion success. Methods We used traditional meta -analysis, a hierarchical framework ( Appendix F), and vote counting approaches to test if and when alt ered biotic interactions facilitate biological invasions by comparing the fitness effects of biotic interactions among native, invasive, or noninvasive exotic species, type of organism (animal or plant), type of biotic interaction (competition, disease, herbivory, mutualism, predation, or plant -soil feedback) and performance metrics (hierarchical analysis only; individual growth, fecundity, survival, and population growth). Meta -analysis provided a tool to combine data from many individual studies and draw more general conclusions about whether the performance effects of biotic interactions differed for invasive species , compared to native and noninvasive exotic species across a wide range of taxa and types of biotic interactions. The multilevel framework al lowed for testing how the magnitude of effect depended on the type of performance metric measured. Literature Search and Data Collection We searched ISI Web of Science for studies, published between 2000 and 2014, which manipulated biotic interactions on native and introduced species. Our searches included the topic search terms ([invasi*] OR [exotic *] OR [introduced] ) AND ( [enemy release ] OR [enemy escape ] OR [biotic resistance ]) AND ( terms describing biotic interaction , see below ). We searched for studies covering the following types of biotic interactions: competition ([compet*]), disease ([disease] OR [fung*]), parasitism ([parasit*]), herbivory (folivory and browsing) 79 ([herbivore] OR [herbivory] OR [brows*]), mutualism ([mutualis*]), plant -soil f eedbacks (PSF) ([soil] OR [microb* community]), and predation (seed predation in plants) ([predate*]). We also searched for relevant reviews on the topic and used their bibliographies to crosscheck our own lists. These searches returned over 3,000 studies, from which we identified appropriate studies. We included only: ( 1) Studies that experimentally manipulated the presence or intensity of biotic interactions under natural or realistic field conditions, or microcosm conditions if the manipulation could not be conducted in the field (mostly soil manipulations). For PSF, we included studies that grew plants in both live and sterilized soil, or in live soil conditioned by conspecific plants (home) and heterospecific plants (away). ( 2) Studies that performed manipulations on (a) both an introduced (noninvasive exotic or invasive) species and a co -occurring native species (native/introduced comparison studies) , or (b) populations of an introduced species in its native and introduced range (cross -continental studi es). If the genotype of the focal species was identified, we only included studies that collected data on local genotypes, such as native genotypes for manipulations conducted in the native range or introduced genotypes for manipulations conducted in the i ntroduced range. ( 3) Studies that measured at least one performance metric (individual growth, fecundity, survival, or population growth). From each study, we recorded species name, species status (native or introduced), and type of introduced species acco rding to author classification in the text (invasive or noninvasive exotic). Though there can be some subjectivity about whether introduced species are described as noninvasive or invasive, we relied on the classification provided by the authors in the pub lication. Native species were generally defined as those occurring at a site without the aid of human introduction. Introduced species were those that occurred outside their native range, and were 80 classified as noninvasive exotic if they naturalized into t he introduced community with little effect, or invasive if the authors listed the species as spreading rapidly or outcompeting native species. We also recorded whether the study was cross continental or a native/introduced species comparison, the type of o rganism (plant or animal), the performance metrics measured, and mean performance value and associated standard deviation and samples sizes within each treatment. To extract data from figures, we used the software program xyExtract Graph Digitizer version 5.1 (developed by W.P. da Silva ). If a paper reported standard errors, we transformed them into standard deviations using reported sample sizes. We recorded all performance metrics reported for each species in the study. If repeated measures of the same s tudy were reported, we only included data from the final time point. We attempted to contact authors to fill in missing data if papers were missing key summary statistics. In total, we extracted a complete set of summary statistics (mean, standard deviatio n [SD], and sample size [n]) from figures, tables, and text of 98 studies that met our criteria for inclusion, resulting in a total of 1,030 effect sizes . Statistical Analysis We calculated the effect size ( d) as HedgesÕ d (Rosenberg et al. 1999). HedgesÕ d performed well for our data as many studies had small sample sizes (n < 10), unequal sampling variances between experimental and control treatments, experimental and control groups with different signs (+ or -), and zeroes (Rosenberg et al. 1999 ). We conducted similar analyses using the log response ratio (Hedges et al. 1999), however this analysis yielded similar results so here we present only results based on HedgesÕ d. HedgesÕ d was calculated as: !!!!!!!!!"!""#$% ! 81 where X C is t he mean performance in the presence of a biotic interaction (control treatment) and XE is the mean performance without the biotic interaction (removal, experimental treatment). The effect size d represents the performance difference for a focal species in the presence and absence of a biotic interaction. A large positive or negative value for d represent s a strong effect from a biotic interaction . For example, if competition strongly reduces performance, removing competitors would result in a positive d. Al ternatively, if the presence of mutualists improves performance, removing mutualists would result in a negative d. J weighted each study by its sample size: !!!!!!!!!!!!!!!!!!! The pooled standard deviation was calculated as: !"!""#$% !!!!!!"!!!!!!!!!!"!!!!!!!!!! where n is the studyÕs sample size for each treatment , and SD is the standard deviation of the control ( C) or experimental ( E) treatment. This analysis results in large studies with small SD receiving the highest weigh ts. Using these effect sizes we calculated a cumulative effect size (!!!, as: !!!!!!!!!!!!!!!!!! where n is the number of studies and di is the effect size for the ith study. The ds from each study were weighted by the reciprocal of their samp ling variance, wi = 1/vnp i. We calculated nonparametric sampling variances ( vnp) as: !!"!!!!!!!!!!!!!!!! 82 where !!! and !!! are the sample sizes from the experimental and control group of the ith study . Nonparametric sampling variances may be less constrained by the need for large sample sizes, compared to typical variances ( Rosenberg et al. 1999 ). To test whether the cumulative effect size (!!!!for each status differs significantly from zero, we used 95% bias -corrected bootstrap confidence intervals (CI) obtained using 9999 iterations (Dixon 1993) . When 95% CIs did not overlap zero, on average the removal of the biotic interaction significantly affected performance. To test for differences among native, invasive, and noninvasive exotic species , we performed a categorical random effects meta -analysis (Raudenbush 1994). The Q statistic assesses the homogeneity of effect sizes and determines whether all studies share a common effect size; the null hy pothesis is that all effect sizes are equal and there is no difference between invasive, noninvasive, and native species (our status moderator variable) ( Rosenberg et al. 1999). When the between -group heterogeneity (Q B) was significant, status explained a significant portion of the overall variation in effect sizes, meaning that the mean effect size of biotic interaction removal differ ed between invasive, noninvasive, and native species . In cases where QB was significant, we tested for pairwise differences between native, noninvasive exotic, or invasive species by comparing 95% CIs. We further explored our data to test for outliers and publication bias, or the tendency of authors to publish certain types of results over others (Begg 1994) . We tested the re lationship between the standardized effect size and sample size using funnel plots and Spearman rank correlation s ( Rosenberg et al. 1999 ). The graphical output showed decreasing variation around the cumulative effect size with increasing sample size and th at the effect sizes were independent of the study sample sizes; our statistical tests revealed that these relationships were non -significant for each biotic interaction manipulated , consistent with a lack of publication bias 83 (Rosenberg et al. 1999). In add ition, because we were concerned about several studies with extreme values of Hedges Õ d (# effects sizes < -3 or > +3), we conducted parallel analyses excluding those extreme effects sizes; results were qualitatively similar to our original analyses , so h ere we present the results from data analys es o f our full range of effect sizes. Initial data exploration revealed differences between (1) plant vs. animal studies, (2) native/introduced comparison studies vs. cross -continental studies, and (3) type of bio tic interaction manipulated ( Appendix F). We therefore conducted separate tests of status on each of these data subdivisions, or study groups. Due to insufficient replication (n < 5), we could not run analysis for cross continental studies on animals, on disease and mutualism for animals in native/exotic comparison studies, and on seed predation for plants in cross continental studies. Because animals do not experience herbivory or PSF, these data are also absent from our analysis. A nalys es were performed using MetaWin version 2.0 (Rosenberg et al. 1999). Vote Count To further corroborate our results, we conducted two additional analyses, including a hierarchical framework (Appendix F) and vote count. Our hierarchical analysis allowed us to test whether b iotic interaction removal had different effects depending on the performance metric measured in the study (i.e., fecundity, growth, population growth and survival), and our vote count allowed us to better explore the variation between our studies. From ea ch study included in the meta -analysis we recorded the (1) the directionality of responses to the removal of biotic interactions for native, noninvasive exotic, and invasive species, and (2) the proportion of our data that supports biotic release. We deter mined that a study supported biotic release when an introduced invasive or noninvasive species was less 84 affected by the removal of a biotic interaction than was a native. We were only able to collect this data from studies that reported statistics on whet her biotic interaction removal/addition treatments affected the performance of the focal species, and whether these performance effects differed by status (native, noninvasive exotic, invasive). Studies that included more focal species, or that manipulated more than one biotic interaction contribute more to our vote count data than do smaller studies, due to the fact that they reported a greater number of effect sizes. Therefore, we also calculated the proportion of studies that support biotic release, weig hting each study the same. Results Cross Continental S tudies Ð Plants Statistically significant and large effect sizes for invasive and noninvasive exotic species indicate that competition, herbivory, and PSF all reduce performance of introduced species, however the magnitude of these effects did not differ fr om those on native taxa (Table 7 , Fig. 1 3). Removing competition and herbivory increased performance of native, noninvasive exotic, and Table 7: Total heterogeneity (Q T) and between-group heterogeneity (Q B) of effect sizes in studies comparing the effects of biotic interactions on native, noninvasive exotic, invasive species performance. A significant Q B indicates that status explained a significant portion of the overall variation in effect sizes, meaning that mean effect size of biotic interaction removal differs between invasive, noninvasive, and native species. Significant p-values (p ! 0.05) for QB are shown in bold. Missing cells represent categories with insufficient replication for analysis (n < 5). QTQBDFpQTQBDFpQTQBDFpCompetition 33.16.4130.23596.812.71330.281434.6152.5350.66Disease 123.40.3230.7597.31.7910.38Herbivory 426.00.25400.895802.716.92190.11 Mutualism 783.5363.0470.001782.37.0810.64Predation 120.40.2860.7642.16.1370.07PSF1439.16.8720.852994.315.1117 0.71Effect of Status (Moderator Variable) Native/Introduced Comparison Studies Plants Animals Cross Continental Studies Plants 85 invasive species, while removing mutualisms decreased performan ce of native and invasive species. We found that native species were most negatively affected by mutualism removal, that mutualism removal also reduced invasive species performance, but that exotic species were not generally affected by mutualisms (signifi cant effect of status, p = 0.001). PSFs were generally negative for native, noninvasive exotic, and invasive species, meaning that species did best in sterilized soil and in soil conditioned by other species. Experimental reduction or removal of disease di d not, in general, affect performance. -5 -4 -3 -2 -1 0 1 2 3 4 competition disease herbivory mutualism predation PSF Response to removal of interaction (Hedges d) Plants - Cross Continental Comparisons Native Exotic Invasive * * * * ** * * * Figure 1: Effects biotic interactions on native, exotic, and invasive plants for cross continental studies, divided by biotic interaction manipulated. Points show means bracketed by 95% confidence intervals. Stars represent effects are significant and 95% confidence intervals do not cross zero. The number of studies in each category are given in parenthesis. !(7) (2) (5) (12) (12) (19) (20) (15) (4) (29) (31) (5) (35) * * -5 -4 -3 -2 -1 0 1 2 3 4 competition disease herbivory mutualism predation PSF Response to removal of interaction (Hedges d) Plants - Cross Continental Comparisons Native Exotic Invasive * * * * ** * * * Figure 1: Effects biotic interactions on native, exotic, and invasive plants for cross continental studies, divided by biotic interaction manipulated. Points show means bracketed by 95% confidence intervals. Stars represent effects are significant and 95% confidence intervals do not cross zero. The number of studies in each category are given in parenthesis. !(7) (2) (5) (12) (12) (19) (20) (15) (4) (29) (31) (5) (35) * * Figure 13: Mean +/- 95% CI effect size of each type of biotic interaction on native, noninvasive exotic, and invasive plants for cross continental studies. One asterisk ( *) indicates effect sizes that differ significantly from zero, and two stars ( **) indicate significant effects of status. The number of studies in each category is indicated in parentheses. Positive and negative values indicate that removal of the interaction increases or decreases performance respectively. 86 Native/Introduced Comparison S tudies Ð Plants The removal of competitors significantly increased the performance of native and noninvasive exotic species, but did not generally affect invas ive species performance ( Table 7, Fig. 14 ). The opposite pattern was present for disease Ð removal or reductions of pathogens significantly increased the performance of invasive species, while native and exotic species were not significantly affected (Fig. 14 ). Native, noninvasive exotic, and invasive speciesÕ performances were significantly increased when herbivores were removed. The removal of seed predators benefitted both native and noninvasive exotic species. PSF had no signific ant effect on any status (Fig. 14 ), but large CIs indicate substantial variation among studies and soil microbes -1.5 -1 -0.5 0 0.5 1 1.5 competition disease herbivory mutualism predation PSF Response to removal of interaction (Hedges d) Plants - Native/Introduced Comparisons Native Exotic Invasive * * * * * * * Figure 2: Effects biotic interactions on native, exotic, and invasive plants for native/introduced comparison studies, divided by biotic interaction manipulated. Points show means bracketed by 95% confidence intervals. Stars represent effects are significant and 95% confidence intervals do not cross zero. The number of studies in each category are given in parenthesis. !(81) (43) (17) (51) (33) (8) (110) (52) (58) (44) (4) (34) (43) (44) (73) (10) (31) * -1.5 -1 -0.5 0 0.5 1 1.5 competition disease herbivory mutualism predation PSF Response to removal of interaction (Hedges d) Plants - Native/Introduced Comparisons Native Exotic Invasive * * * * * * * Figure 2: Effects biotic interactions on native, exotic, and invasive plants for native/introduced comparison studies, divided by biotic interaction manipulated. Points show means bracketed by 95% confidence intervals. Stars represent effects are significant and 95% confidence intervals do not cross zero. The number of studies in each category are given in parenthesis. !(81) (43) (17) (51) (33) (8) (110) (52) (58) (44) (4) (34) (43) (44) (73) (10) (31) * -1.5 -1 -0.5 0 0.5 1 1.5 competition disease herbivory mutualism predation PSF Response to removal of interaction (Hedges d) Plants - Native/Introduced Comparisons Native Exotic Invasive * * * * * * * Figure 2: Effects biotic interactions on native, exotic, and invasive plants for native/introduced comparison studies, divided by biotic interaction manipulated. Points show means bracketed by 95% confidence intervals. Stars represent effects are significant and 95% confidence intervals do not cross zero. The number of studies in each category are given in parenthesis. !(81) (43) (17) (51) (33) (8) (110) (52) (58) (44) (4) (34) (43) (44) (73) (10) (31) * Figure 14: Mean +/- 95% CI effect size of each type of biotic interaction on native, noninvasive exotic, and invasive plants for native/introduced comparison studies. One star ( *) indicates effect sizes that differ significantly from zero. The number of studies in each category is indicated in parentheses. Positive and negative values indicate that removal of the interaction increases or decreases performance respectively. 87 may act as both antagonists and mutualists for plants. Native, noninvasive exotic, and invasive speciesÕ performances were not generally lowered by mutualist removal treatments. Native/Intr oduced Comparison S tudies Ð Animals Competition removal treatments did not significantly influence native, noninvasive exotic, or inva sive animal performance (Table 7 ). Invasive speciesÕ performance significantly increased -2 -1 0 1 2 3 4 5 6 competition predation Response to removal of interaction (Hedges d) Animals - Native/Introduced Comparisons Native Exotic Invasive (18) (11) (7) (23) (8) (7) Figure 15: Mean +/- 95% CI effect size of each type of biotic interaction on native, noninvasive exotic, and invasive animals for native/ introduced comparison studies. One star ( *) indicates effect sizes that differ significantly from zero. The number of studies in each category is indicated in parentheses. Positive and negative values indicate that removal of the interaction increases or decreases performance respectively. * 88 whe n predators were removed (Fig . 15), however there was no effect of removal of native or noninvasive exotic animals. Vote Count From the 98 studies included in our analysis, we collected 687 responses to treatments manipulating the presence of biotic interactions (Table 8 ). Of these, 42.4% (291studies) found that focal species responded positively to the removal of antagonistic biotic interactions, and 43.6% (300 studies) showed no response. Some studies (8.4% , 58 studies ) found that removing an antagonistic biotic interaction actually decreased performance, potentially due to some unmeasured factor. For example, if the removal of herbivores has a greater benefit to plant A than plant B, plant B may show a negative response to herbivore removal when in fact they are responding to increa sed competition from plant A. The remaining negative responses to biotic interaction removal were due to the removal of mutualists (5.5% , 38 studies ). When we look at the level of study (published paper), 28 found evidence in support of biotic release, whi le 58 found that biotic release did not drive the success of introduced plants in their study. We collected 146 data points from these 86 studies, recording whether an invasive or noninvasive introduced species was released, compared to native competitors in its introduced range or compared to conspecific populations in its native range. These measures give more weight to studies that manipulated more than one biotic interaction, or studied more than one introduced species, as these yielded separate data po ints for each. We found that 26.7% (39 studies) of these data points supported biotic release, while 73.3% (107 studies) did not. The finding that the majority of studies do not find evidence for biotic release supports the conclusions from our meta -analys is. 89 Table 8: Summary table of vote count results. For our vote count, significance is determined by statistics reported in the original papers. Significant effect sizes are indicated with + and -, where non-significant effects indicated by n.s. Plants Effect of Biotic Interaction Removal All (+) n.s. (-)(+) n.s. (-)(+) n.s. (-)43.2% Y, 56.8% NCompetition 61232100% NDisease 5121250% Y, 50% NHerbivory 123171360.0% Y, 40.0% NMutualism 12315250% Y, 50% NPredation PSF236351612842.1% Y, 57.9% NPlants Effect of Biotic Interaction Removal All (+) n.s. (-)(+) n.s. (-)(+) n.s. (-)20.1% Y, 79.9% NCompetition 333041417384523.5% Y, 76.5% NDisease 41311215100% NHerbivory 2330291632115117.6% Y, 82.4% NMutualism 12191012179933.3% Y, 66.7% NPredation 445466.7% Y, 33.3% NPSF2920124426139100% NAnimals All (+) n.s. (-)(+) n.s. (-)(+) n.s. (-)30.8% Y, 69.2% NCompetition 5505233.3% Y, 66.7% NDisease 552100% NMutualism Predation 4181175283.3% Y, 16.7% NCross Continental Comparisons Native/Introduced Comparisons Biotic Release? Biotic Release? Native Exotic Invasive Native Exotic Invasive 90 Discussion In our study , we find that biotic release is not a prevailing mechanism explaining invasive species success; antagonistic interaction effects did not differ significantly for invasive species, compared to populations in their native range, or native and noninvasive exotic competitors in their introduced ranges. These results were consistent across our traditional meta -analysis approach, hierarchical meta -analysis (Appendix F), and vote count. We found that exotic and invasive plants are negatively affected by competition, predation, PSF, disease (invasive plants only), and herbivory (invasive plants only). However, the removal of competitors in native/introduced comparison studies significantly improved performance of native and noninvasive exotic plants, while not improving performance of invas ives. Additionally, we found that native and invasive plants relied substantially on mutualist partners, while noninvasive exotic plants did not. Therefore, we detected evidence that biotic interactions can limit introduced species performance and that inv asiveness in plants may be driven by competitive release and the formation of successful mutualisms in the introduced range. Our results are consistent with a previous meta -analysis, which found that introduced plant establishment and performance was reduc ed by competition and herbivory (Levine et al. 2004), and with previous studies and meta -analyses which found introduced species are not generally released from herbivores compared to natives (Parker et al. 2006, Chun et al. 2010, Schultheis et al. 2015). Chun and collaborators (2010) found that introduced and native species received equal damage from herbivores and disease. The same meta -analysis identified nine studies that manipulated the presence herbivores and disease, and consistent with our study, th ey found no difference in the performance response of native and introduced species to herbivores and diseases (Chun et al. 2010). Only one prior meta -analysis separated out the effects of 91 herbivores on invasive and noninvasive exotic species (Parker et al . 2006), and they too found that invasive and noninvasive exotics responded similarly to herbivore removal . In our study, native and introduced species responded similarly to the soil community (Fig. 1 3 and 14 ), consistent with a study on an entire pl ant communityÕs response to PSF (Anacker et al. 2014 ), but inconsistent with Klironomos (2002) , which found native plants experience strong negative PSF s, while the most abundant invasive plants experience positive PSFs. While positive PSFs often correlate wit h greater field abundances, studies included in our analysis often did not report natural field densities, so we were unable to determine whether invasive species in these studies were in fact most abundant in the community. Our results demonstrate invasi ve species are significantly harmed by disease and parasites, while native and exotic species were generally not affected by disease and par asite removal treatments (Fig. 14 ), supporting findings by Parker and Gilbert (2007), which found that invasive Trifolium and Medicago species had the highest levels of disease prevalence and greatest performance increases in response to disease removal. However, most enemy removal studies included in our analysis manipulated entire fungal communities with fungicide tre atments. The lack of performance effects from disease in cross -continental studies could be driven by the fact that these treatments indiscriminately removed antagonistic and mutualistic fungi species, resulting in no net effect of removal on performance. While invasive species were just as strongly affected by most antagonistic biotic interactions as native and noninvasive exotic species, we found evidence that competition may contribute to invasion success. While native and noninvasive exotic plants sign ificantly benefitted from the removal of competition in native/introduced comparison studies, invasive species did not (Fig. 14 ). However, large CIs indicate that the effects of competition vary widely 92 for invasive species. Additionally, in cross -continent al comparisons, invasive and noninvasive exotic populations tended to be less affected by competition than were natives (Fig. 1 3). Therefore, release from competition may play a role in invasiveness. Another key difference between introduced species that become invasive and those that do not may be the formation of successful mutualisms in the introduced range. In cross -continental studies, both native and invasive populations of introduced plants were negatively affected by the experimental removal of mut ualist partners. This pattern was consistent with native/introduced comparison studies. In both study types, noninvasive exotic species were not affected by the removal of mutualists, suggesting that these species receive minimal benefit from mutualist par tners in the introduced range and could even be limited by lack of mutualisms . Studies manipulating mycorrhizal mutualists dominated the literature, but we observed similar patterns across all mutualism types. Other mutualisms studied included rhizobia (Parker et al. 2007, Rodr™guez -Echeverr™a et al. 2012 , Horn et al. 2014 ), ants ( Lach et al. 2010, Prior et al. 2015 unpublished data ), earthworms (Wurst et al. 2011), frugivores (Zuel et al. 2012), and endophytes ( Aschehoug et al. 2012) . All mutualisms tested in our study were facultative, as missing obligate mutualisms would have pre vented the establishment of introduced species, making further experimentation impossible. Species Specific Case Studies and Context Dependency of Biotic Release Although biotic interactions are not a general explanation for biological invasions, there is substantial variation around our cumulative effect sizes ( !!!; thus , altered biotic interactions may contribute to some invasions . Within the studies included in our analysis, several found evidence that enemy release contributed to the success of invade rs in their system. For example, 93 herbivory by white -tailed deer ( Odocoileus virginianus ) increased the relative abundance of invasive Alliaria petiolata and Microstegium vimineum , while reducing native plant species abundance (Knight et al. 2009). In a coa stal wetland, invasive Lythrum salicaria abundance was not significantly affected by the presence of herbivores, while many native species Õ abundances decreased (Barry et al. 2004). In a study of 30 native and introduced species, Agrawal and colleagues (2005) found that introduced species experienced half the negative performance effects from soil microbes compared to native species. Similarly, Klironomos (2002) found that invasive sp ecies in his system, which included A. petiolata and L. salicaria , had positive PSF, while rare, native species experienced negative PSF. Species interactions are frequently context -dependent, varying in strength or even direction depending on environment al conditions (Chamberlain et al. 2014) ; thus, a particular invasive species may experience release under certain environmental conditions but not others, potentially accounting for some of the variation found between studies. For example, temperature can affect predator -prey interactions (Fey and Herren 2014) , and release could occur in some climates but not others because of temperature -driven mismatch between interacting species. Additionally, plant interactions with microbes can vary from mutualism to p arasitism depending on soil nutrient cond itions (Hoeksema et al. 2010 ) or competitive environment experienced by a plant (Casper and Castelli 2007). The Resource -Enemy Release Hypothesis predicts that invaders growing in high nutrient conditions will benef it more from release from herbivory if higher levels of resource lead to faster growth and more poorly defended tissues in invaders (Blumenthal 2006). In this scenario, only studies manipulating herbivory in high nutrient environments would reveal evidence for enemy release. 94 Enemy and mutualist acquisition in the introduced range, over time and with increasing spread , also may explain some of the variation among studies. Introduced species leave behind biotic interactions from their native range, yet concu rrently encounter a new suite of species in the introduced range. While release from negative biotic interactions may facilitate colonization and establishment during the early stages of an invasion, these benefits may be lost over time as introduced speci es accumulate enemies and competitors in their introduced range (Elton 1958; Mitchell et al. 2006, Mitchell et al. 2010). For example, s tudies on introduced plants find that release from herbivory and disease is lost over a period of a few hundred years ( Mitchell et al. 2010). Processes affecting aboveground enemy acquisition might function similarly belowground as well; plants with longer residence times in New Zealand had more negative interactions with soil organisms than newly introduced plants (Diez e t al. 2010). Unfortunately, very few studies on introduced species report information on introduction dates (Strayer et al. 2006), and for many species this data is unknown. Future studies on introduced species that elucidate the changing effects of biotic interactions over time will help determine the long -term effects of biological invasions (Mitchell et al. 2006). Finally, b iotic release is predicted to occur when an introduced species leaves behind coevolved antagonists and enter a community where co-evolved relationships are lacking (Hallett 2006) . However, closely related native species often occur in the introduced community. DarwinÕs Naturalization Hypothesis predicts that species closely related to the invaded community are less likely to establish because they tend to have more similar traits, and as a result, are more likely to compete for resources (Darwin 1859). This hypothesis can be extended to traits that mediate interactions with antagonists and mutualists as well Ð if defense or mutualism tr aits are phylogenetically conserved between close relatives, introduced species may 95 be more likely to acquire biotic interactions from close relatives present in the community (Gilbert and Webb 2007). In support of this hypothesis, several recent studies h ave shown that phylogenetically dissimilar introduced species are more likely to establish and become invasive in novel communities, compared to introduced species with close relatives present (Strauss et al. 2006, Jiang et al. 2010, Schaefer et al. 2011; but see Duncan and Williams 2002). Interactive and S ynergistic Effects of Multiple Biotic Interactions Many of our CIs include d zero , resulting in the counter -intuitive interpretation that removal of a biotic interaction did not always significantly affe ct performance . Natural systems are complex, and survival and other performance metrics are typically simultaneously influenced by a multitude of abiotic and biotic factors. The vast majority of studies included in our analysis manipulated a single type of biotic interaction, but perhaps release from one interaction is not enough to cause significant shifts in fitness (potentially explaining why many of our CIs included zero) or drive invasiveness (potentially explaining the lack of significant differences between native, noninvasive and invasive species for most biotic interactions). Out of the 98 studies included in our analysis, 23 manipulated more than one biotic interaction, and nine of those found a significant interaction between treatments. Only two studies found that biotic interactions acted synergistically to suppress invaders, together reducing invader performance more than when acting alone. For example, competition from a native thistle and herbivory interact to resist invasion by the introduced thistle, Cirsium vulgare (Suwa and Louda 2012). In this case, release from multiple interactions might be necessary to drive invasiveness (Huang et al. 2012, Suwa and Louda 2012). 96 Conclusion We found that predation, herbivory, disease, and PSFs generally decrease the fitness of native, noninvasive exotic, and invasive species similarly, providing little evidence that enemy release is a general mechanism facilitating invasions. However, invasiv e species were less affected by competition than native or noninvasive exotic species, and the removal of mutualists decreased performance for native and invasive plant species, but not noninvasive exotics, indicating that escape from competition and the f ormation of mutualisms in the introduced range may be important in promoting the success of introduced species. The earliest writings on biotic release recognized that interactions gained in the introduced range might be just as limiting to performance as those lost (Elton 1958), and that release would not operate for all introduced species. Dozens of hypotheses attemp t to explain invasiveness, and it is becoming clear that no one hypothesis will serve as a Òmagic bulletÓ explaining the diversity of strate gies employed by invasive plants and animals (Gurevitch et al. 2011). Just as a variety of biotic and abiotic factors control the performance of native populations, introduced species likely succeed and fail due to a variety of mechanisms ( Gurevitch et al. 2011). To determine if and when biotic release operates, future studies must focus on the features that may drive context dependency in release, and the factors influencing how introduced species acquire novel antagonistic biotic interactions in their int roduced range. 97 APPENDICES 98 Appendix A: Supplemental Tables and Figures for Chapter 2 1 2 This appendix c ontains s upplemental tables and figures for Chapter 2 . Table A 1 lists all 3 experimental species and detailed information on their status, seed origin, years planted into the 4 experiment, and GenBank accession numbers. Table A2 gives r esults from phylogenetic 5 generalized least squares (PGLS) analysis of variance (ANOVA) testin g the effects of plant 6 status (native, noninvasive exotic, or invasive) and phylogeny on insect herbivory and mammal 7 browsing. Table A3 gives r esults from analysis of covariance (ANCOVA) testing the effects of 8 plant status and geographic spread (at three spatial scales) or time on herbivory and browsing. 9 Fig. A1 gives the bes t scoring maximum likelihood ( ML) phylogenetic tree, and Fig. A2 shows 10 images of the experimental common garden. 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 99 Table A1 : List of the 61 species planted into the 2011 and 2012 -2013 common gardens. Species 34 are color coded by plant status: native (white), exotic (gray), and invasive (black). In the columns 35 for year, presence of a particular species is indicated with an ÔXÕ. If the cell is grayed out, it 36 indicates that survival was low and the species was not included in the analysis for that year. 37 GenBank accession numbers of genes used for phylogeny construction are listed. When a species 38 was not located in GenBank, a close relative was used and note d with (*). 39 40 41 FamilySpecies Name Abbrev. Status Lists Seed Source 2011 20122013matK ITSrbcLAsteraceae Achillea millefolium ACHMI native Michigan XXXEU385315.1 AY603185.1 JX848399.1 Poaceae Agropyron repens AGRRE invasive MSLMichigan XXFJ395421.1 GQ365145.1KJ841296.1 Fabaceae Amorpha canescens AMCAN native Michigan X*AY426773.1 Fabaceae Amorpha fruticosa AMFRU exotic noneMissouri XKC584927.1 GQ281030.1KC584888.1 Poaceae Bromus hordeaceus BROHO exotic noneCalifornia XXAM889695.1 KM077298.1 GQ248557.1Poaceae Bromus inermis BROIN invasive WTPMichigan XXAF164398.1 KF713194.1 KJ841141.1 Poaceae Bromus kalmii BROKA native Michigan XXXAY367916.1 Apiaceae Carum carvi CARCA exotic nonePennsylvania XU58553.1JQ792209.1 KF602102.1 Asteraceae Centaurea cyanus CENCY exotic nonePennsylvania XJN894130.1 KC603919.1 AB530955.1 Asteraceae Centaurea stoebe CENST invasive MNFI, WTP Michigan XXKC969492.1 JF914072.1 KJ746252.1 Asteraceae Cichorium intybus CICINexotic nonePennsylvania XXXAJ633131.1 HM921413.1 HQ590035.1Asteraceae Conyza canadensis CONCAnative Michigan XHM850627.1 AY875695.1 HQ590045.1Asteraceae Coreopsis lanceolata CORLAnative Michigan XXXAY551495.1 KM347947.1 HM849915.1 Asteraceae Coreopsis palmata CORPA native Pennsylvania XXAY551480.1 AY553673.1, AY553674.1 Asteraceae Coreopsis tinctoria CORTI exotic nonePennsylvania XXHM989735.1 KM347935.1 GU724222.1Asteraceae Coreopsis tripteris CORTR native Michigan XXXAY551499.1 KM347917.1 Fabaceae Coronilla varia CORVA invasive WTPPennsylvania XXHM049547.1 AF218537.1 U74222.1Asteraceae Cosmos bipinnatus COSBI exotic nonePennsylvania XHM989783.1 KM347948.1 GQ436474.1Asteraceae Cosmos sulphureus COSSU exotic noneOhioXEU049362.1 KM347949.1 Apiaceae Daucus carota DAUCA exotic noneKansas XHQ593265.1KJ415356.1 KJ841260.1 Fabaceae Desmodium canadense DESCA native Michigan XXHQ593266.1KM098891.1 KJ841264.1 Asteraceae Erigeron annuus ERIAN native Michigan XXHM989796.1 GU724302.1KJ841309.1 Asteraceae Eupatorium perfoliatum EUPPE native Michigan XXXEU749317.1 DQ415741.1KJ841315.1 Asteraceae Gaillardia pulchella GAIPU exotic nonePennsylvania XHM989787.1 KF607074.1 HQ590105.1Asteraceae Helianthus petiolaris HELAU exotic noneMichigan XXX*AY009458.1 JX121556.1 Asteraceae Helenium flexuosum HELFL exotic noneOhioXXX*AY215804.1 KF607070.1 *AY215123.1 Brassicaceae Hesperis matronalis HESMA invasive MNFI, WTP New York XHQ593319.1DQ357547.1HQ590129.1Asteraceae Lactuca saligna LACSA exotic noneMichigan XX*AJ633239.1 HQ161960.1*JN893847.1 Asteraceae Lactuca serriola LACSE exotic noneMichigan XXHQ593336.1HQ172902.1HQ590149.1Fabaceae Lespedeza capitata LESCA native Michigan XXXGU572331GU572172.1Fabaceae Lespedeza cuneata LESCU invasive nonePennsylvania XXXEU717416.1 GU572175.1EU717275.1 Asteraceae Leucanthemum vulgare LEUVU exotic nonePennsylvania XXHQ593344.1EF091600.1 KJ841377.1 Fabaceae Lotus corniculatus LOTCOinvasive PMW, WTP Pennsylvania XXXHM049505.1 JN861076.1 KJ841388.1 Fabaceae Lupinus perennis LUPPE native Michigan XXZ72162.1, Z72163.1KF613009.1 Fabaceae Medicago lupulina MEDLU exotic noneNebraska XXHE966952.1 JQ858257.1 KJ841412.1 Fabaceae Melilotus albus MELAL invasive MNFI, PMW, WTP Wisconsin XXXHE967441.1 DQ006009.1DQ006095.1Fabaceae Melilotus officinalis MELOF invasive MNFI, PMW, WTP Pennsylvania XXXHE970723.1 KJ999362.1 KJ841414.1 Poaceae Panicum virgatum PANVI native Michigan XXEU434294.1 DQ005062.1EF125135.1 Poaceae Phleum pratense PHLPR exotic noneMichigan XXHQ593382.1HQ600524.1KJ841460.1 Poaceae Poa compressa POACO invasive WTPPennsylvania XXXKJ599232.1 KJ598896.1 KJ599121.1 Poaceae Poa nemoralis POANE native CanadaXXXJN894815.1 GQ324529.1KJ841479.1 Poaceae Poa pratensis POAPR invasive WTPPennsylvania XXKJ599261.1 KJ598925.1 KJ599150.1 Poaceae Poa trivialis POATR exotic nonePennsylvania XXXFJ395369.1 GQ324555.1JN893080.1 Rosaceae Potentilla arguta POTAG native Michigan XHQ593397.1U90787.1HQ590221.1Rosaceae Potentilla anserina POTAN native California XKJ840972.1 KF954772.1 KJ841496.1 Rosaceae Potentilla argentea POTAR exotic noneCanadaXKJ840973.1 AB894151.1 KJ841497.1 Rosaceae Potentilla recta POTRE exotic noneOregon XHQ593398.1AB894160.1 HQ590222.1Rosaceae Rosa setigera ROSSE native Michigan XAB048601.1 AB048596.1 Poaceae Schizachyrium scoparium SCHSC native Michigan XXFR832830.1DQ005072.1HE577863.1 Asteraceae Solidago canadensis SOLCA native Michigan XXEU749415.1 HQ142591.1EU677023.1 Asteraceae Solidago graminifolia SOLGR native Michigan XXKM212072.1 HQ142624.1HQ590098.1Asteraceae Solidago rigida SOLRI native Michigan XXXHQ142603.1JX848426.1 Asteraceae Sonchus oleraceus SONOL exotic noneMichigan XJN894897.1 AY458002.1 KF196024.1 Poaceae Sorghastrum nutans SORNU native Michigan XXEF137473.1 DQ005080.1EF125121.1 Poaceae Sporobolus heterolepis SPOHE native Michigan XXAF164429.1 *GU359228.1KJ740997.1 Asteraceae Symphyotrichum pilosum SYMPInative Michigan XXEU749444.1 JQ360419.1 EU677053.1 Asteraceae Taraxacum officinale TAROF exotic noneMichigan XXFJ395377.1 HQ161934.1FJ395571.1 Fabaceae Tephrosia virginiana TEPVI native Michigan X*AF467499.1 *KF511648.1 Fabaceae Trifolium hybridum TRIHYexotic nonePennsylvania XXAF522125.1 AF053159.1 KJ841632.1 Fabaceae Trifolium pratense TRIPR exotic noneMichigan XXEU749448.1 AF053171.1 KJ841633.1 Fabaceae Trifolium repens TRIREexotic noneMichigan XXKJ841029.1 AF053172.1 KJ841634.1 100 Table A2: Results from phylogenetic generalized least squares (PGLS) analysis of variance 42 (ANOVA) testing the effects of plant status (native, noninvasive exotic, or invasive) and 43 phylogeny on insect herbivory and mammal browsing. Statistically significant (P < 0.05) effects 44 are in bold. 45 46 47 48 49 50 51 Source dftPdftPdftPStatus 2,272.200.042,430.240.812,311.200.24Blomberg's K = 0.13 Blomberg's K = 0.12 Blomberg's K = 0.03 Status 2,27-0.61 0.562,430.960.342,310.280.78Blomberg's K = 0.03 Blomberg's K = 0.12 Blomberg's K = 0.11 Note: Significant effects of status (P ! 0.05) are in bold. (a) Insect Herbivory PGLS (b) Mammal Browser PGLS 2011 20122013 101 Table A3: Results from analysis of covariance (ANCOVA) testing the effects of plant status 52 (native, noninvasive exotic, or invasive) and geographic spread (at three spatial scales) or time 53 on insect herbivory and mammal browsing. Statistically significant (P < 0.05) effects are in bold. 54 55 Source df!2Pdf!2Pdf!2PSource df!2PStatus 1, 301.990.241, 301.640.161, 302.080.29Status 1, 302.400.08County1, 312.04< 0.001 1, 311.74< 0.001 1, 312.160.001Residence Time 1, 312.570.02Status x County 1, 290.170.041, 291.580.281, 292.040.42Status x Time 1, 291.76< 0.001 Status 1, 309.520.221, 3010.140.341, 3010.520.40Status 1, 309.720.59County1, 319.880.041, 3110.340.101, 3110.670.24Residence Time 1, 319.820.06Status x County 1, 297.830.0091, 295.43< 0.001 1, 295.86< 0.001 Status x Time 1, 299.180.20Note: Significant differences (P ! 0.05) are in bold. (a) Insect Herbivory ANCOVA (b) Mammal Browser ANCOVA Five State Spread MI Spread Time US Spread 102 56 0.2Lespedeza_capitataAmorpha_fruticosaTaraxacum_officinaleDesmodium_canadensePoa_trivialisPotentilla_anserinaElymus_repensCoreopsis_lanceolataCichorium_intybusPoa_compressaSchizachyrium_scopariumCoreopsis_tinctoriaPhleum_pratenseLactuca_salignaPotentilla_argenteaCoreopsis_palmataLactuca_serriolaMelilotus_officinalisErigeron_annuusAmorpha_canescensPotentilla_argutaGaillardia_pulchellaSymphyotrichum_pilosumSolidago_canadensisPoa_pratensisCentaurea_stoebeCoronilla_variaSporobolus_heterolepisMelilotus_albusLeucanthemum_vulgareHelianthus_petiolarisTrifolium_repensLotus_corniculatusBromus_kalmiiLespedeza_cuneataTrifolium_hybridumHelenium_flexuosumEuthamia_graminifoliaConyza_canadensisLupinus_perennisMedicago_lupulinaCosmos_sulphureusCentaurea_cyanusDaucus_carotaCosmos_bipinnatusSorghastrum_nutansTephrosia_virginianaRosa_setigeraSonchus_oleraceusCoreopsis_tripterisPoa_nemoralisSolidago_rigidaBromus_inermisBromus_hordeaceusPotentilla_rectaAchillea_millefoliumCarum_carviPanicum_virgatumTrifolium_pratenseHesperis_matronalisEupatorium_perfoliatum1009910093100100100100100100981005881100869210010010010010099100100100735557999994100100849785986710086798610051100407998988210080100100531004326Poaceae Figure A1: The best-scoring ML tree from a rapid bootstrap analysis in RAxML from the analysis of the concatenated sequences of matK , ITS, and rbcL . ML bootstrap frequencies are the numbers associated with nodes, and branch lengths are proportional to the number of nucleotide changes. Brassicaceae Fabaceae Rosaceae Apiaceae Asteraceae 103 57 Figure A2: Images showing (a) the experimental common garden in 2012, (b) E.H. Schultheis in 58 the field measuring insect herbivory and mammalian browsing on experimental seedlings, and 59 (c) an experimental Lupinus perennis seedling . 60 61 62 63 64 65 66 67 68 (a) (b) (c) 104 Appendix B: Statistical Methods and R esults for Plant Family Analysis 69 70 Statistical Analysis 71 To determine whether invasive, noninvasive exotic, or native species differ in herbivore 72 damage, and whether plant family influenced damage, we performed ANOVA using the aov 73 function in R. Proportion leaf area removed and proportion of stems browsed were i ncluded as 74 response variables, and plant status (invasive, noninvasive exotic, or native), family, and the 75 status x family interaction were included as fixed predictor variables. Analyses were conducted 76 on within -year species averages; separate analyses we re run for each year of data because species 77 composition varied. In 2011 it was not possible to test for the interaction between status and 78 family due to lack of replication of status within family. All non -significant interaction terms 79 were dropped from t he 2012 and 2013 models to increase power for testing main effects. Tukey -80 adjusted post -hoc contrasts were used to evaluate differences between treatment combinations 81 when main effects or interaction terms were significant (P ! 0.05). Response variables we re not 82 transformed because species mean data met ANOVA normality assumptions . 83 84 Results 85 Plant families received different amounts of herbivory and browsing damage (Table B1). 86 Fabaceae tended to receive the most insect herbivore damage and Poaceae the least, with 87 Asteraceae receiving intermediate amounts (Fig. B1a). In 2012, exotics and invasives in the 88 Fabaceae tended to receive more insect herbivore damage than natives, and in the Asteraceae and 89 Poaceae natives and invasives tended to receive more insect he rbivore damage than exotics (Fig. 90 B2). Plant families differed in susceptibility to browsing; in 2013, Fabaceae and Asteraceae 91 105 received more browsing damage than Poaceae, and though not statistically significant, similar 92 patterns were obser ved in 2011 and 2012 (Fig. B1b). 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 106 Table B1: Results from analysis of variance (ANOVA) testing the effects of plant status 114 (invasive, noninvasive exotic, or native) , family, and their interaction on insect herbivory and 115 mammal browsing. Statistically significant (P < 0.05) effects are in bold. 116 117 118 2011 2012 2013 Source df F P df F P df F P (a) Insect Herbivory ANOVA Status 2, 21 5.35 0.01 2, 36 4.23 0.02 2, 28 0.51 0.35 Family 5, 21 0.98 0.45 2, 36 23.6 < 0.001 2, 28 4.17 0.03 Status x Family 4, 36 6.55 < 0.001 (b) Mammal Browser ANOVA Status 2, 21 2.07 0.15 2, 36 2.52 0.09 2, 28 3.09 0.06 Family 5, 21 0.60 0.70 2, 36 1.48 0.24 2, 28 6.92 0.004 Status x Family 4, 36 1 107 Figure B1: Three years of (a) insect herbivore and (b) mammal browser damage data on Asteraceae (hatched bars), Fabaceae (empty bars), and Poaceae (striped bars) plants. All analysis was performed within year. Bars indicate mean ± SE. Means with the same letter are n ot statistically different (P ! 0.05) based on post -hoc contrasts . 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 2011 2012 2013 Proportion Branches with Browsing Damage Asteraceae Fabaceae Poaceae a a a a a a b b a 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 2011 2012 2013 Proportion Leaf Area Removed by Herbivory Asteraceae Fabaceae Poaceae a a a a a b a ab b Proportion Leaf Area Removed by Herbivory Proportion Branches with Browsing Damage 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 2011 2012 2013 Proportion Branches with Browsing Damage Asteraceae Fabaceae Poaceae a a a b ab a b b a (a) (b) 108 Figure B2: Data from 2012 showing the family by status interaction for insect herbivore damage data. Species statuses shown with different color bars: native (white bars), noninvasive exotic (gray bars), and invasive (black bars). Bars indicate mean ± SE. Means within family with the same letter are not statistically different (P ! 0.05) based on post -hoc contrasts . 0 0.05 0.1 0.15 0.2 0.25 0.3 Asteraceae Fabaceae Poaceae Percent Leaf Area Removed in 2012 by Herbivory native exotic invasive a a b a a a a a a 2012 Proportion Leaf Area Removed by Herbivory MELAL COSSU CENCY COSBI MELOF GAIPU SOLGR CORTRCENST SOLCA CORLAEUPPE LOTCOSYMPICORPA BROHO LESCU DESCA ERIAN POANE HELAU CORTITRIPR LESCA SOLRI CACAR MEDLU DAUCA POTAR POTRE SORNU ACHMI AGRRE AMCAN BROIN BROKA CICINCONCACORVA HELFL HESMA LACSA LEUVU PANVI PHLPR POACO POAPR POATR POTAG SCHSC SPOHE TAROF TEPVI TRIHYTRIRE0.00.20.40.60.81.0All Years Together Proportion Branches with Browsing Damage Native Exotic Invasive 109 Appendix C: List of Experimental S pecies in the 2012-2014 experiment Table C1 : List of the 50 species planted into the 2012 -2014 experimental plots . Species are color coded by plant status: native (white), noninvasive exotic (gray), and invasive (black). In the columns for year, presence of a particular species is indicated with an ÔXÕ. For invasive species lists: WTP = Wild Type Plants, MNFI = Michigan Natural Features Inventory, MSL = Michigan Seed Law, PMW = Invasive Plants of the Upper Midwest (Czarapata 2005). FamilySpecies Name Abbrev. Status Lists Seed Source Perenniality 201220132014Asteraceae Achillea millefolium ACHMI native Michigan perennial XXXAsteraceae Conyza canadensis CONCAnative Michigan annualXAsteraceae Coreopsis lanceolata CORLAnative Michigan perennial XXXAsteraceae Coreopsis palmata CORPA native Pennsylvania perennial XXXAsteraceae Coreopsis tripteris CORTR native Michigan perennial XXXAsteraceae Erigeron annuus ERIAN native Michigan biennialXXAsteraceae Eupatorium perfoliatum EUPPE native Michigan perennial XXXAsteraceae Solidago canadensis SOLCA native Michigan perennial XXXAsteraceae Solidago graminifolia SOLGR native Michigan perennial XXXAsteraceae Solidago rigida SOLRI native Michigan perennial XXXAsteraceae Symphyotrichum pilosum SYMPInative Michigan perennial XXXAsteraceae Centaurea cyanus CENCY exotic nonePennsylvania annualXAsteraceae Cichorium intybus CICINexotic nonePennsylvania perennial XXXAsteraceae Coreopsis tinctoria CORTI exotic nonePennsylvania annualXAsteraceae Cosmos bipinnatus COSBI exotic nonePennsylvania annualXAsteraceae Cosmos sulphureus COSSU exotic noneOhioannualXAsteraceae Gaillardia pulchella GAIPU exotic nonePennsylvania annualXAsteraceae Helenium flexuosum HELFL exotic noneOhioperennial XXXAsteraceae Helianthus petiolaris HELPE exotic noneMichigan annualXAsteraceae Lactuca saligna LACSA exotic noneMichigan biennialXXAsteraceae Lactuca serriola LACSE exotic noneMichigan biennialXXAsteraceae Leucanthemum vulgare LEUVU exotic nonePennsylvania perennial XXXAsteraceae Sonchus oleraceus SONOL exotic noneMichigan annualXAsteraceae Taraxacum officinale TAROF exotic noneMichigan perennial XXXAsteraceae Centaurea stoebe CENST invasive MNFI, WTP Michigan biennialXXFabaceae Desmodium canadense DESCA native Michigan perennial XXXFabaceae Lespedeza capitata LESCA native Michigan perennial XXXFabaceae Lupinus perennis LUPPE native Michigan perennial XXXFabaceae Medicago lupulina MEDLU exotic noneNebraska annualXFabaceae Trifolium hybridum TRIHYexotic nonePennsylvania perennial XXXFabaceae Trifolium pratense TRIPR exotic noneMichigan perennial XXXFabaceae Trifolium repens TRIREexotic noneMichigan perennial XXXFabaceae Coronilla varia CORVA invasive WTPPennsylvania perennial XXXFabaceae Lespedeza cuneata LESCU invasive nonePennsylvania perennial XXXFabaceae Lotus corniculatus LOTCOinvasive PMW, WTP Pennsylvania perennial XXXFabaceae Melilotus albus MELAL invasive MNFI, PMW, WTP Wisconsin biennialXXFabaceae Melilotus officinalis MELOF invasive MNFI, PMW, WTP Pennsylvania biennialXXPoaceae Bromus kalmii BROKA native Michigan perennial XXXPoaceae Panicum virgatum PANVI native Michigan perennial XXXPoaceae Poa nemoralis POANE native Canadaperennial XXXPoaceae Schizachyrium scoparium SCHSC native Michigan perennial XXXPoaceae Sorghastrum nutans SORNU native Michigan perennial XXXPoaceae Sporobolus heterolepis SPOHE native Michigan perennial XXXPoaceae Bromus hordeaceus BROHO exotic noneCalifornia annualXPoaceae Phleum pratense PHLPR exotic noneMichigan perennial XXXPoaceae Poa trivialis POATR exotic nonePennsylvania perennial XXXPoaceae Agropyron repens AGRRE invasive MSLMichigan perennial XXXPoaceae Bromus inermis BROIN invasive WTPMichigan perennial XXXPoaceae Poa compressa POACO invasive WTPPennsylvania perennial XXXPoaceae Poa pratensis POAPR invasive WTPPennsylvania perennial XXX 110 Appendix D: Analysis of Background Community Changes Methods In September 2013 I estimated light competition and productivity of the background community in each experimental plot because it appeared that competition was becoming more intense in some of my enemy removal treatments. As an estimation of productivity, I harvested aboveground biomass and dead thatch in a "x#m sub -plot within my 40 2x2m experimental plots. Biomass was dried at 70¡C for 72 hours and weighed. To assess light availability, I used a ceptometer (Decagon LP -80 AccuPAR) to measure photosynthetically active radiation (PAR) above the plant canopy and 10cm above the soil surface. I collected three measurements above and three below, averag ed them, and took the difference between these two averages. Larger values represent potentially increased light competition in experimental field plots. I tested the effects of enemy exclusion on background community, amount of thatch, and light competiti on with mixed model ANOVA using the aov function in R. PAR, thatch biomass, and aboveground biomass were included as response variables, and my three treatments (fencing, insecticide, and fungicide) and all interactions were included as fixed predictor var iables. When a significant interaction between treatments was found, post hoc Tukey tests were used to determine which treatment combinations differed from one another (P ! 0.05). Data was untransformed as it satisfied normality assumptions. Results I fou nd evidence that enemy removal treatments affected standing stock of the background community, thatch biomass, and light availability (PAR) . The fencing treatment 111 increased the productivity of the background community ( F1,28 = 6.38, p = 0.02), raising biomass 18.6% from 692.0 ± 47.6 g/m 2 to 820.5 ± 69.1 g/m 2 [mean ± SE]. There was also a marginally significant interaction between fencing and insecticide treatments ( F1,28 = 3.7, p = 0.06); biomass was lowest at 621.8 ± 45.5 g/m 2 in control plots, and removal of either insects or browsers was enough to raise biomass up to levels found in plots where both types of herbivores were excluded (806.5 ± 92.7 g/m 2). The fungicide treatment significantly increased the amount of thatch pres ent in experimental plots ( F1,28 = 4.8, p = 0.04). Thatch increased 8.5% from 86.0 ± 6.2 g/m 2 in untreated plots, to 93.3 ± 7.6 g/m 2 in treated plots. There was also a significant interaction between fungicide and insecticide treatments ( F1,28 = 4.8, p = 0.04); thatch was highest in plots that received the fungicide, but not insecticide, treatment (103.5 ± 12.7 g/m 2). Fencing and fungicide treatments significantly increased light availability in experimental plots (fencing: F1,32 = 18.9, p < 0.001; fungicide: F1,32 = 4.3, p = 0.04). PAR increased from 1,234.4 ± 20.9 µmol/m 2s to 1,385.0 ± 7.1 µmol/m 2s in fenced plots, and from 1,288.3 ± 28.9 µmol/m 2s to 1,331.1 ± 14.1 µmol/m 2s in plots that received the fungicide treatment. 112 Appendix E: Flower Number A nalysis Methods At the end of the experiment we measured plant performance metrics, including height (cm) from the soil surface to apical meristem, aboveground biomass (g), and flower number. To determine whether our treatments influenced plant performance, we tested the effects of simulated herbivory and competition on plant biomass and height with mixed model ANOVA using the lmer function, and flower number with the glmer function, in the lme4 package in R (v. 1.1-7, Bates et a l. 2015). To test treatment effects, our model included plant biomass (g), plant height (cm), or flower number as the response variable and clipping (clipped, unclipped), competition (competitor present, absent), status (native, noninvasive exotic, invasiv e), family (Asteraceae, Poaceae, Fabaceae), and all possible interactions as fixed predictor variables. Flower number data was analyzed using the Poisson distribution, and because only a small number of individuals flowered during the course of the experi ment, we analyzed only data for those individuals and species that flowered. To test significance fixed and random effects for flower number, we used chi -squared tests. Results No native species flowered during the experiment (Fig. E1a), and only noninvasive exotic Centaurea cyanus , Sonchus oleraceus , and Bromus hordeaceus , and invasive Lotus corniculatus , Melilotus officinalis , and Poa compressa flowered; only one individual of M. officinalis and P. compressa produced any flowers. Flower number depended on the interaction between status, clipping, and the competition treatment (Table E1); invasive species in unclipped competition 113 pots produced significantly more flowers than did exotic species where either competition or clipping treatments were applied (Fig. E1a). This pattern was driven by invasive L. corniculatus , which produced significantly more flowers when grown in competition and without clipping compared to the control (F ig. E1b). 114 Table E1: Results from m ixed model analysis of variance (ANOVA) showing the effects of status, family, clipping, and competition on experimental plant flower number. Statistically significant (P ! 0.05) effects are in bold. All non-significant interaction terms were dropped from the final model. Source dfF!2Pstatus 1, 100.026.60.01family 2, 104.346.20.04clipped 1, 1012.846.50.01competition 1, 100.980.20.63status x clipped 1, 120.570.30.58status x competition 1, 1213.458.50.004clipped x competiton 1, 131.791.70.19status x clipped x competition 1, 1419.1519.7< 0.001 Random Effects (species)status 1.10.59species x clipped 0.01.00species x competition 8.00.005Flower Number Native, Exotic, and Invasive Species 115 Figure E1: Flower number data for native, noninvasive exotic, and invasive plants that flowered during the course of the experiment. Graph a displays data by status, while graph b displays data by species. Different colored bars represent the clipping and competition treatments. Bars indicate mean ± SE. Means with different letters are significantly different (P ! 0.05) based on post -hoc contrasts. 0 5 10 15 20 25 30 35 CENCY (e) SONOL (e) BROHO (e) LOTCO (i) MELOF (i) POACO (i) Flower Number Species Control Unclipped, Competition Clipped, No Competition Clipped, Competition 0 5 10 15 20 25 30 35 Native Exotic Invasive Flower Number Status Control Unclipped, Competition Clipped, No Competition Clipped, Competition 0 5 10 15 20 25 30 35 Native Exotic Invasive Flower Number Status Control Unclipped, Competition Clipped, No Competition Clipped, Competition (a) (b) ab b b b ab ab ab a bc c c bc bc bc bc bc ab bc abc bc bc a abc abc c c 116 Appendix F: Hierarchical M eta-analysis Methods To further explore how biotic interactions influence perfor mance and also to better account for the low number of observations in some categories, we analyzed the data following a multilevel, or hierarchical, framework where the different categories of the data were nested within each other (Clark 2007, Ibanez et al. 2014). By using a multilevel/hierarchical framework we thoroughly document if altered biotic interactions are driving the invasion process by assessing difference among species status (native, invasive, or noninvasive exotic), type of organism (animal or plant), type of biotic interaction (competition, disease, herbivory, mutualism, predation, or plant -soil feedback), and among fitness metrics (fecundity, growth, population growth, and survival). We used the same effect size (Hedges d) estimated for the main analysis, and their associated SD. In this case, instead of using non -parametric variance, we included study random effects to account for any bias that could have been associated with any particular study and used variance estimates calculated as: !!!!!!!!!!!!!!!!!!!!!!!!! where NE is the sample size of the experimental group, N C the sample size of the control group, and d is the measure of Hedges d (Rosenberg et al. 1999). Unlike the traditional meta -analysis approach, our hierarchical approach allowed us analyze data with smaller sample sizes, such as studies that measured disease effects on animals, and to further explore differences among fitness metrics. The significance of the effect size values (different from zero or not) were first estimated for each status (native, invasive, or noninvasive exotic); within each status, we then separated data by organism type (animal or 117 plant), followed by type of biotic interaction ma nipulated (competition, disease/parasite, herbivory, mutualism, predation or plant -soil feedbacks [PSF]), and finishing with fitness metric as the lowest level of our hierarchical analysis (fecundity, growth, population growth, or survival). Given the mult ilevel structure of the data, and the large number of parameters involved, we used a Bayesian framework to estimate parameter values from non -informative distributions. We independently analyzed cross continental studies from native/introduced comparison s tudies. Observation i, ESobs i, of status(i), organism type(i), biotic interaction(i) and fitness metric(i) was modeled as: !"!"#!!!!"#$%& !!"!"#"$! !!!!!"#$%&'( !!!!!"#$%&'#!(" !!!!!"#$%&& !!!!!"#!!!!!!"#!!!!! where !"!"#"$! !!!!!"#$%&'( !!!!!"#$%&'#!(" !!!!!"#$%&& !!! is the mean ES for the combination of fitness metric, biotic interaction, type of organism and species status that observation i belongs to. SRE repr esents study random effects, !"#!!!"#$%& !!!!!"# !! and !!"# !!"#$%&' !!!!"!. The variability around ESobs was the estimated variability in the original study, !!"#!. Mean parameters were then estimated from hyperparameter values th at follow the multilevel structure of the data. -Species status, organism type, biotic interaction and fitness metrics ES: !"!"#"$! !!!"#$%&'( !!!"#$%&'#!(" !!!"#$%&& !! !"#$%& !!"!"#"$! !!!"#$%&'( !!!"#$ !"#$%&' !!!!"#"$! !!!"#$%&'( !!!"#$%&'#!(" !!! -Species status, type of organism and biotic interaction ES: !"!"#"$! !!!"#$%&'( !!!"#$%&'#!(" !!!"#$%& !!"!"#"$! !!!"#$%&'( !!!!"#"$! !!!"#$%&'( !!!"#$%&'#!(" !!! -Species status and type of organism ES: !"!"#"$! !!!"#$%&'( !!!"#$%& !!"!"#"$! !!!!"#"$! !!!"#$%&'( !!! -Species status, overall, ES: 118 !"!"#"$! !!!"# !"#!!!!""""! All variances were estimated from non -informative prior distributions, !!!!"#$%&' !!!!"!. Analyses were performed in OPENBUGS (Thomas et al. 2006) and ran for 100000 iterations, after the 25000 initial burn -in period parameter values were estimated by thinning every 100 th iteration. Results Species Status and Type of Organism L evels Overall predicted effect sizes for each species status (native, exotic and invasive) in the native/introduced comparisons show no effect of the removal of the biotic interaction (Fig. F1). When we divided studies by type of organism (animals or plants), effect sizes were mainly positive but none was significantly different from zero. For the cross continental comparisons, noninvasive exotics had effect si zes significantly different from zero, driven by the animal studies (Fig. F1). Biotic Interactions L evel Among native/introduced comparisons studying animals only, three biotic interactions were reported: competition, disease and predation. Removal of com petition only significantly affected the performance of native animals (Fig. F2a). In cross continental studies of animals, only two interactions were reported, competition for native species and disease for exotic species and only ES associated with the l atest was significantly different from zero (Fig. F2b). Among plant studies for native/introduced comparisons, none of the effect sizes at this level were statistically different from zero (Fig. F3a). Effect sizes for cross continental 119 comparisons were si gnificantly positive for disease, herbivory and PSF removal in native species, PSF removal in exotic species, and disease, herbivory and PSF removal for invasive species (Fig. F3b). Fitness Metric L evel Animal studies in the native/introduced comparisons showed positive effect sizes, significantly different from zero for population growth of native species and invasive species when released from competition, for survival of invasive species when released from competition, and for population growth of invasive species when release from predation (Fig. F4a). Across cross continental studies, only survival of noninvasive exotic species when released from disease was statistically significant (Fig. F4b). Among plant native/introduced comparison studies, release from competition had a positive effect on the fecundity, growth, and survival of native species and the population growth and survival of noninvasive exotic species. Competition had a negative effect on the growth of invasive species (Fig. F5). Release from disease only benefited growth of invasive species, while release from herbivory benefited growth of native and invasive species and survival of noninvasive exotic species (Fig. F5). Released from preda tion also had a positive effect on survival of native and noninvasive exotic species (Fig. F5). From the analyses of the cross continental comparisons, native and invasive species experienced significant release from disease and herbivory and experienced i ncreased fecundity, individual growth, and survival (Fig. F6). Disease also significantly reduced population growth of native and invasive species, while herbivory only reduced population growth of native species. PSF significantly reduced individual growt h of native and invasive species. The removal of 120 mutualists had a negative effect on growth of native species (Fig. F6). Noninvasive exotic species experienced improved fecundity when herbivores were removed, and increased growth when growing in sterilized soil or soil trained by heterospecifics (Fig. F6). 121 Table F1: To ease the process of comparing results from our vote count, traditional meta -analysis, and hierarchical meta -analysis we summarized the results of all three analyses here. Summary table of e ffect sizes from studies comparing the effects of biotic interactions on status performance. Symbols represent positive (+), negative ( -), and neutral (n.s.) effects of biotic interaction removal. For our traditional (a) and hierarchical meta -analyses (b), effect sizes are calculated as HedgesÕ d, and significance is determined as whether 95% CIs cross zero. For our vote count, significance is determined by statistics reported in the original papers. Sig nificant effect sizes are indicated with + and -, where non -significant effects indicated by n.s. Native Exotic Invasive Native Exotic Invasive Plants Effect of Biotic Interaction Removal Plants Effect of Biotic Interaction Removal Plants Effect of Biotic Interaction Removal All n.s. n.s. n.s. All n.s. n.s. n.s. All (+) n.s. (-)(+) n.s. (-)(+) n.s. (-)43.2% Y, 56.8% NCompetition +++Competition n.s. n.s. n.s. Competition 61232100% NDisease n.s. n.s. Disease ++Disease 5121250% Y, 50% NHerbivory ++Herbivory +n.s. +Herbivory 123171360.0% Y, 40.0% NMutualism -n.s. -Mutualism n.s. n.s. n.s. Mutualism 12315250% Y, 50% NPredation Predation Predation PSF+++PSF+++PSF236351612842.1% Y, 57.9% NNative Exotic Invasive Native Exotic Invasive Plants Effect of Biotic Interaction Removal Plants Effect of Biotic Interaction Removal Plants Effect of Biotic Interaction Removal All n.s. +n.s. All n.s. n.s. n.s. All (+) n.s. (-)(+) n.s. (-)(+) n.s. (-)20.1% Y, 79.9% NCompetition ++n.s. Competition +n.s. n.s. Competition 333041417384523.5% Y, 76.5% NDisease n.s. n.s. +Disease n.s. n.s. +Disease 41311215100% NHerbivory +++Herbivory +n.s. n.s. Herbivory 2330291632115117.6% Y, 82.4% NMutualism n.s. n.s. n.s. Mutualism n.s. n.s. n.s. Mutualism 12191012179933.3% Y, 66.7% NPredation ++Predation n.s. n.s. Predation 445466.7% Y, 33.3% NPSFn.s. n.s. n.s. PSFn.s. n.s. n.s. PSF2920124426139100% NAnimals Animals Animals All n.s. n.s. n.s. All n.s. n.s. n.s. All (+) n.s. (-)(+) n.s. (-)(+) n.s. (-)30.8% Y, 69.2% NCompetition n.s. n.s. n.s. Competition n.s. +n.s. Competition 5505233.3% Y, 66.7% NDisease Disease n.s. n.s. Disease 552100% NMutualism Mutualism Mutualism Predation n.s. n.s. +Predation n.s. n.s. n.s. Predation 4181175283.3% Y, 16.7% NNative/Introduced Comparisons (b) Hierarchical Meta-analysis (a) Traditional Meta-analysis Biotic Release? Biotic Release? (c) Vote Count Cross Continental Comparisons Native/Introduced Comparisons Cross Continental Comparisons Native/Introduced Comparisons Native Exotic Invasive Native Exotic Invasive Cross Continental Comparisons 122 Figure F1: Effects of biotic interactions on native (light gray) , exoti c (medium gray), and invasive (black) species for cross continental studies, and native/introduced species comparison studies. Points show means bracketed by 95% confidence intervals. Asterisk represents significant effect when 95% confidence intervals did not cross zero. The number of studies in each category is given in parenthesis. 123 Figure F2: Effects biotic interactions on native, exotic, and invasive animals for (a) native/introduced comparison studies, and (b) cross continental studies. Results are split up by biotic interaction manipulated. Points show means bracketed by 95% confidence intervals. Black asterisk s represent effects are significant and 95% confidence intervals do not cross zero. The number of studies in each category is given in parenthesis. 124 Figure F3: Effects biotic interactions on native, exotic, and invasive plants for (a) native/introduced comparison studies, and (b) cross continental studies. Results are split up by biotic interaction manipulated. Points show means bracketed by 95% confidence intervals. Black asterisk s represent effects are significant and 95% confidence intervals do not cross zero. The number of studies in each category is given in parenthesis. * 125 Figure F4: Effects biotic interactions on native, exotic, and invasive animals for (a) native/introduced comparison studies, and (b) cross continental studies. Results are split up by biotic interaction manipulated and performance response variable measured (fecundit y, growth, population growth, and survival) . Points show means bracketed by 95% confidence intervals. Black asterisk s represent effects are significant and 95% confidence intervals do not cross zero. The number of studies in each category is given in paren thesis. 126 Figure F5: Effects biotic interactions on native, exotic, and invasive plants for native/introduced comparison studies. Results are split up by biotic interaction manipulated and performance response variable measured (fecundity, growth, popul ation growth, and survival) . Points show means bracketed by 95% confidence intervals. Black asterisk s represent effects are significant and 95% confidence intervals do not cross zero. The number of studies in each category is given in parenthesis. 127 Figure F6: Effects biotic interactions on native, exotic, and invasive plants for cross continental studies. Results are split up by biotic interaction manipulated and performance response variable measured (fecundity, growth, population growth, and surviv al). Points show means bracketed by 95% confidence intervals. Black asterisk s represent effects are significant and 95% confidence intervals do not cross zero. The number of studies in each category is given in parenthesis. 128 REFERENCES 129 REFERENCES Agrawal, A.A. and P.M. Kotanen . 2003. Herbivores and the success of exotic plants: a phylogenetically controlled experiment. Ecology Letters 6:712-715. Agrawal, A.A. , P.M. Kotanen, C.E. Mitchell, A.G. Power, W. Godsoe, and J. Klironomos. 2005. Enemy release? An experiment with congeneric plant pairs and diverse above - and belowground enemies. Ecology 86:2979-2989. Agrawal, A.A. 2007. Macroevolution of plant defense strategies. Trends in Ecology and Evolution 22:103-109. Agrawal, A.A., D.D. Ackerly, F. Adler, A.E. Arnold, C. Caceres, D.F. Doak, E. Post, P.J. Hudson, J. Maron, K.A. Mooney, M. Power, D.W. Schemske, J. Stachowicz, S.Y. Strauss, M.G. Turner, and E. Werner. 2007. Filling key gaps in population and community ecology. Frontiers in Ecology and the Environment 5:145 -152. Ahern, R.G. , D.A. Landis, A.A. Reznicek, and D.W. Schemske . 2010. Spread of exotic plants in the landscape: the role of time, growth habit, and his tory of invasiveness. Biological Invasions 12:3157-3169. Anacker, B.L., J.N. Klironomos, H. Maherali, K.O. Reinhart, and S.Y. Strauss. 2014. Phylogenetic conservatism in plant -soil feedback and its implications for plant abund ance. Ecology Letters 17:1613 -1621. Anderson, R.M. and R.M. May. 1979. Population biology of infectious diseases. Part1. Nature 280:361 -367. Anderson, R.M. and R.M. May. 1981. The population dynamics of microparasites and their invertebrate hosts. Phi losophical Transactions B 291:451 -524. Ashton, I.W. and M.T. Lerdau . 2008. Tolerance to herbivory, and not resistance, may explain differential success of invasive, naturalized, and native North American temperate vines. Diversity and Distributions 14:169-178. Augustine, D.J. and S.J. McNaughton. 1998. Ungulate effects on the functional species composition of plant communities: herbivore selectivity and plant tolerance. Journal of Wildlife Management 62:1165 -1183. Baker R. 1983. Michigan Mammals . Michigan State University Press, East Lansing, MI . Barry, M.J., R. Bowers, and F.A. De Szalay. 2004. Effects of hydrology, herbivory and sediment disturbance on plant recruitment in a Lake Erie coastal wetland. The American Midland Naturalist 151:217 -232. 130 Bates, D., M. Maechler, B. Bolker, S. Walker, R.H.B. Christensen, H. Singmann, and B. Dai. 2015. lme4: Linear mixed -effects models using Eigen and S4 . R package version 1.1 -7, Available at: http://lme4.r -forge.r -project.org/. Bais, H.P., R. Vepachedu, S. Gilroy, R. M. Callaw ay, and J. M. Vivanco. 2003. Allelopathy and exotic plant invasion: from molecules and genes to species interactions. Science 301:1377 -1380. Begg, C.B. 1994. Publication bias. In H. Cooper and L. V. Hedges (eds.). The Handbook of research synthesis. (pp. 399 -409). Russell Sage Foundation, New York. Bever, J.D. 1994. Feedback between plants and their soil communities in an old field community. Ecology 75:1965 -1977. Blackshaw, R.E., J.R. Moyer, R.C. Doram, and A.L. Boswell. 2001. Yellow sweetclover, green manure, and its residues effectively suppress weeds during fallow. Weed Science 49:406 -413. Blair, A.C. and L.M. Wolfe . 2004. The evolution of an invasive plant: A n expe rimental study with Silene latifolia . Ecology 85:3035-3042. Blaney, C.S. and P.M. Kotanen. 2001a . Effects of fungal pathogens on seeds of native and exotic plants: a test using congeneric pairs. J ournal of Applied Ecology 38:1104-1113. Blaney, C.S. and P.M. Kotanen. 2001b . Post -dispersal losses to seed predators: an experimental comparison of native and exotic old field plants. Canadian Journal of Botany 79: 284-292. Blaney, C.S. and P.M. Kotanen. 2002 . Persistence in the seed bank: the effects of fungi and invertebrates on seeds of native and exotic plants. E coscience 9 :509-517. Blomberg, S.P., T. Garland Jr., and A.R. Ives. 2003. Testing for phylogenetic signal in comparative data: behavioral tr aits are more labile. Evolution 57:717 -745. Blossey, B. and R. N ıtzold. 1995. Evolution of Increased Competitive Ability in Invasive Nonindigenous Plants: A Hypothesis. Journal of Ecology 83:887 -889. Blumenthal, D. M. 2006. Interactions between resourc e availability and enemy release in plant invasion. Ecology Letters 9:887 -895. Bolker, B.M. 2008. Ecological Data and Models in R. Princeton University Press Borland, K. , S. Campbell, R. Schillo, and P. Higman . 2009. A field identification guide to invasive plants in Michigan's natural communities, Michigan Natural Features Inventory. Lansing, MI. Available at: http://mnfi.anr.msu.edu/invasive species/InvasivePlantsFieldGuide.pdf 131 Burdon, J.J. and G.A. Chilvers. 1982. Host density as a factor in plant disease ecology. Annual Review of Phytopathology 20:143 -166. Callaway, R.M. and E.T. Aschehoug. 2000. Invasive plants versus their new and old neighbors: A mechanism for exotic invasion. Science 2 90:521-523. Callaway R .M. and W.M. Ridenour. 2004. Novel weapons: invasive success and the evolution of increased competitive ability. Frontiers in Ecology and the Environment 2:436 -443. Cappuccino, N. and D. Carpenter . 2005. Invasive exotic plants suffer less herbivory than non - invasive exotic plants. Bi ology Letters 1:435 -438. Carpenter, D. and N. Cappuccino . 2005. Herbivory, time since introduction and the invasiveness of exotic plants. Journal of Ecology 93:315-321. Casper, B.B. and J.P. Castelli. 2007. Evaluating plant Ðsoil feedback together with competition in a serpentine grassland. Ecology Letters 10:394 -400. Catford, J.A., R. Jansson, and C. Nilsson. 2009. Reducing redundancy in invasion ecology by integratin g hypotheses into a single theoretical framework . Diversity and Distributions 15:22-40. Cebrian, J. and C.M. Duarte. 1994. The dependence of herbivory on growth rate in natural plant communities. Functional Ecology 8 : 518 -525. Chamberlain, S.A., J.L. Bronstein, and J.A. Rudgers. 2014. How context dependent are species interactions? Ecology Letters 17:881 -890. Chun, Y. J., M. Van Kleunen, and W. Dawson. 2010. The role of enemy release, tolerance and resistance in plant invasio ns: linking damage to performance. Ecology Letters 13:937 -946. Clark, J. S. 2007. Models for ecological data . Princeton University Press, Princeton, NJ. Cleveland, C.C., A.R. Townsend, D.S. Schimel, H. Fisher, R.W. Howarth, L.O. Hedin, S.S. Perakis, E.F. Latty, J.C.V. Fischer, A. Elseroad, and M.F. Wasson. 1999. Global patterns of terrestrial biological nitrogen (N 2) fixation in natural ecosystems. Global Biogeochemical Cycles 13:623 -645. Colautti, R.I., A. Ricciardi, I.A. Grigorovich and H.J . MacIsaac . 2004. Is invasion success explained by the enemy release hypothesis? Ecology Letters 7: 721-733. Crawley, M.J. 1987. What makes a community invasible? Colonization, Succession, and Stability (eds A.J. Gr ay, M.J. Crawley & P.J. Edwards), pp. 429 Ð 453. Blackwell, Oxford. 132 Crawley, M.J. 1989. Insect herbivores and plant population dynamics. Annual Review of Entomology 34:531 -534. Crawley, M.J. 2007 . The R Book. John Wiley & Sons , Ltd. West Sussex, England. Czarapata, E.J. 2005. Invasive Plants of the Upper Midwest. Univer sity of Wisconsin Press, Madison, WI. Daehler, C.C. 2003. Performance comparisons of co -occurring native and alien invasive plants: implications for conservation and restoration. Annual Review of Ecology, Evolution, and Systemat ics 34:183 -211. Darriba, D., G.L. Taboada, R. Doallo, and D. Posada 2012. jModelTest 2: more models, new heuristics and parallel computing. Nature Methods 9:772. Darwin, C. 1859. On the origin of species by means of natural selection. London . Dawson, W., A. Bottini, M. Fischer, M. van Kleunen, and E. Knop. 2014. Little evidence for release from herbivores as a driver of plant invasiveness from a multi -specie s herbivore -removal experiment . Oikos 123:1509 -1518. DeWalt S .J., J.S. Denslow and K. Ickes. 2004. Natural -enemy release facilitates habitat expansion of the invasive tropical shrub Clidemia hirta . Ecology 85:471 -483. Dietz H ., L.R. Wirt h, H. Buschmann. 2004. Variation in herbivore damage to invasive and native woody plant species in ope n forest vegetation on Mahe, Seychelles. Biol ogical Invasions 6:511 Ð521 Dietz, H. and P. Edwards. 2006. Recognition that causal processes change during plant invasion helps explain conflicts in evidence. Ecology 87:1359 -1367. Diez, J. M., I. D ickie, G. Edwards, P.E. Hulme, J. J. Sullivan, and R. P. Duncan. 2010. Negative soil feedbacks accumulate over time for non -native plant species. Ecology Letters 13:803 -809. Dillemuth, F.P., E.A. Rietschier, and J. T. Cronin. 2008. Patch dynamics of a native grass in relation to the spread of invasive smooth brome ( Bromus inermis ). Biological Invasions 11:1381 -1391. Dixon, P.M. 1993. The bootstrap and the jackknife: Describing the precision of ecological indices. In S. M. Scheiner and J. Gurevitch (eds.). Design and analysis of ecological experiments. (pp. 290 -318). Chapman and H all, New York. Dost⁄l, P., E. Allan, W. Dawson, M. van Kleunen, I. Bartish, and M. Fischer. 2013. Enemy damage of exotic plant species i s similar to that of natives and increases with productivity. Journal of Ecology 101:388 -399. 133 Duke, S.O., F.E. Dayan, J. Bajsa, K.M. Meepagala, R.A. Hufbauer, and A.C. Blair. 2009 . The case against ( -)-catechin involvement in allelopathy of Centaurea sto ebe (spotted knapweed) Plant Signaling and Behavior 4:422 -424. Duncan, R.P. and P.A. Williams. 2002. Ecology: DarwinÕs naturalization hypothesis challenged. Nature, 417:608 -609. Elton, C.S. 1958. The ecology of invasions of animals and plants . Methuen , London. Emery, S. M. and J.A. Rudgers. 2012. Impact of Competition and Mycorrhizal Fungi on Growth of Centaurea stoebe , an Invasive Plant of Sand Dunes. The American Midland Naturalist 167:213 -222. Endara, M.J. and P.D. Coley. 2011. The resource availabili ty hypothesis revisited: a meta - analysis. Functional Ecology 25:389 -398. Feeny, P. P. 1976. Plant apparency and chemical defense. Pages 1 Ð40 in J. W. Wallace and R. L. Mansell, editors. Biochemical interaction between plants and insects. Plenum, New York, New York, USA. Fey, S.B. and C.M. Herren. 2014. Temperature -mediated biotic interactions influence enemy release of nonnative species in warming environments. Ecology 95:2246 -2256. Firn, J., J.L. Moore, A.S. MacDougall, E.T. B orer, E. W. S eabloom, J. HilleRisLambers, W. S. Harpole, E.E. Cleland, C.S. Brown, J. M. Knops, et al. 2011. Abundance of introduced species at home predicts abundance away in herbaceous communities. Ecology Letters 14:274 -281. Flory, S.L. and K. Clay. 2013. Pathogen accumulation and long -term dynamics of plant invasions. Journal of Ecology 101:607 -613. Garland, T., A.W. Dickerman, C. M. Janis, and J.A. Jones. 1993. Phylogenetic analysis of covariance by computer simulation. Systematic Biology 42:265 -292. Gilbert, G.S. and C.O. Webb. 2007. Phylogenetic signal in plant pathogen -host range. Proceedings of the National Academy of Sciences 104:4979 -4983. Go§ner , M.M. , A. Chao, R.I. Bailey, and A. Prinzing . 2009. Native fauna on exotic trees: phylogenetic conservatism and geographic contingency in two lineages of phytophages on two lineages of trees. The American Naturalist 173: 599-614. Gurevitch, J., G.A. Fox, G.M. Wardle, Inderjit, !and D. Taub . 2011. Emergent insights from the synthesis of conceptual frameworks for biological invasions. Ecology Letters 14:407-418. Hallett, S.G. 2006. Dislocation from coevolved relationships: a unifying theory for plant invasion 134 and naturalization? Weed Science 54:282 -290. Harper, J.L. 1977. P opulation Biology of Plants. Academic Press, London. Hawkes, C.V. 2007. Are invaders moving targets? The generality and persistence of advantages in size, reproduction, and enemy release in invasive plants species with time since introduction. The Americ an Naturalist 170: 832 -843. Hayes, K.R. and S.C. Barry. 2008. Are there any consistent predictors of invasion success? Biological Invasions 10:483 -506. Heard, M.J. and D.F. Sax. 2013. Coexistence between native and exotic species is facilitated by asymm etries in competitive ability and susceptibility to herbivores. Ecology Letters 16:206 -213. Hedges, L.V., J. Gurevitch , and P.S. Curtis. 1999. The meta -analysis of respons e ratios in experimental ecology. Ecology 80:1150 -1156. Heger, T. and L. Trepl. 2003. Predicting biological invasions. Biological Invasions 5:313 -321. Heger, T. and J.M. Jeschke. 2014. The enemy release hypothesi s as a hierarchy of hypotheses. Oikos 123:741-750. Herrera, A. M., R.I. Carruthers, and N.J. Mills. 2011. Introduced populations of Genista monspessulana (French broom) are more dense and produce a greater seed rain in California, USA, than native populations in the Mediterranean Basin of Europe. Biological Invasions 13: 369-380. Hierro, J.L., J.L. Maron, and R. M. Callaway. 2005. A biogeographical approach to plant invasions: the importance of studying exotics in their introduced and native range. Journal of Ecology 93:5 -15. Hill, S.B. and P.M. Kotanen. 2010. Phylogenetically structured damage to Asteraceae: susceptibility of native and exotic species to foliar herbivores. Biological Invasions 12:3333 -3342. Hinz, H.L. and M. Schwarzlaender. 2004. Comparing Invasive Plants from Their Native and Exotic Range: What Can We Learn for Biological Control? Weed Technology 18:1533 -1541. Hochwender C., R. Marquis, and K. Stowe. 2000. The potential for and constraints on the evolution of compensatory ability in Asclepias syriaca . Oecologia 122:361 -370. Hoeksema, J.D., V.B. Chaudhary, C.A. Gehring, N.C. Johnson, J. Karst, R.T. Koide, A. Pringle, C. Zabinski, J.D. Bever, J.C. Moore, et al. 2010. A meta -analysis of context -dependency in plant response to inoculation with mycorrhizal fungi. Ecology Letters 13:394 -407. 135 Hokkanen, H.M.T. and D. Pimentel. 1989 . New associations in biological control Ð theory and practice. The Canadian Entomologist 121: 829 -840. Hothorn, T., F. Bretz, P. Westfall, R.M. Heiberger, A. Schuetzenmeister, and S. Scheibe. 2015. multcomp: Simultaneous Inference in General Parametric Models. R package version 1.4-0, Available at: http://multcomp.R -forge.R -project.org. Huang, W., J. Carrillo, J. Ding, and E. Siemann. 2012. Interactive effects of herbivory and competition intensity determine invasive plant performance. Oecologia 170:373 -382. Ib⁄Œez, I., D.W. Katz, D. Peltier, S.M. Wolf, and B.T. Connor Barrie. 2014. Assessing the integrated e ffects of landscape fragmentation on plants and plant communities: The challenge of a multiprocess -multiresponse dynamics . Journal of Ecology 102:882-895. Jiang, L., J. Tan, and Z. Pu. 2010. An Experimental Test of DarwinÕs Naturalization Hypothesis. The American Naturalist 175:415 -423. Jogesh, T., D. Car penter, and N. Cappuccino. 2008. Herbivory on i nvasive exotic plants and their noninvasive relatives. Biological Invasions 10:797 -804. Keane, R.M. and M.J. Crawley . 2002. Exotic plant invasions and the enemy release hypothesis. Trends in Ecology and Evolution 17:164 -170. Kearse, M., R. Moir, A. Wilson, S. Stones -Havas, M. Cheung, S. Sturrock, S. Buxton, A. Cooper, S. Markowitz, C. Duran, et al. 2012. Geneious Basic: an integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics 28:1647 -1649. Kempel, A., T. Chrobock, M. Fischer, R.P. Rohr, and M. van Kleunen. 2013. Determina nts of plant establishment success in a multispecies introduction experiment with native and alien species. Proceedings of the National Academy of Sciences of the United States of America 110:12727 -12732. Kimbro, D.L., B.S. Cheng, and E.D. Grosholz. 2013 . Biotic resistance in marine environments. Ecology Letters 16:821 -833 Klironomos, J.N. 2002. Feedback with soil biota contributes to plant rarity and invasiveness in communities. Nature 417:67 -70. Knapp A. and T. Seastedt. 1986. Detritus accumulation limits productivity of tallgrass prairie. Bioscience 36:662 -668. Knight, T.M., J.L. Dunn, L.A. Smith, J. Davis, and S. Kalisz. 2009. Deer facilitate invasive plant success in a Pennsylvania forest understory. Natural Areas Journal 29:110 -116. 136 Kolar C.S. and D.M. Lodge. 2001. Progress in invasion biology: Predicting invaders. Trends in Ecology and Evolution 164:199 -204. Kowarik, I. and P. Py " ek . 2012. The first steps towards unifying concepts in invasion ecology were made one hundred yea rs ago: revisiting the work of the Swiss botanist Albert Thellung. Diversity and Distributions 18 :1243-1252. Krebs, C.J., S. Boutin, R. Boonstra, A.R.E Sinclair, J.N.M Smith, M.R.T. Dale, K. Martin, and R. Turkington. 1995. Impact of Food and Predation o n the Snowshoe Hare Cycle. Science 269:1112 -1115. Kulmatiski, A., K.H. Beard, J.R. Stevens, and S.M. Cobbold. 2008. Plant -soil feedbacks: a meta - analytical review. Ecology Letters 11:980 -992. Kuznetsova, A., P. Bruun Brockhoff, and R.H.B. Christensen. 2015. lmerTest: Tests in Linear Mixed Effects Models R package version 2.0 -20 , http://cran.r -project.org/package=lmerTest. Lau, J.A., A.C. McCall, K.F. Davies, J.K. McKay, and J.W. Wright. 2008. Herbivores and edaphic factors constrain the realized niche of a native plant. Ecology 89:754 -762. Lau, J.A., and E.H. Schultheis. 2015. When two invasion hypotheses are better than one. New Phytologist 205:958 -960. Levine, J.M. 2000 . Species diversity and biological invasio ns: relating local process to community pattern. Science 288:852 -854. Levine, J.M., M. Vila, C.M.D. Antonio, J.S. Dukes, K. Grigulis, and S. Lavorel. 2003. Mechanisms underlying the impacts of exotic plant invasions. Proceedings of the Royal Society B : Biological Sciences 270:775 -781. Levine, J.M., P.B. Adler, and S.G. Yelenik. 2004. A meta -analysis of biotic resistance to exotic plant invasions. Ecology Letters 7:975 -989. Lind, E.M. and J. D. Parker. 2010. Novel weapons testing: are invasive plants more chemically defended than native plants? PloS one 5:e10429. Liu, H. and P. Stiling. 2006. Testing the enemy release hypothesis: a review and meta -analysis. Biological Invasions 8:1535 -1545. Liu, H., P. Stiling, and R. Pemberton. 2007. Does enemy release matter for invasive plants? evidence from a comparison of insect herbivore damage among invasive, noninvasive and native congeners. Biological Invasions 9:773 -781. Lonsdale, W.M. 1999. Global patterns of plant invasions and the concept of invisibility. Ecology 80:1522 -1536. 137 Louda, S. M. 1982. Distribution Ecology: Variation in plant recruitment over a gradient in relation to insect seed predation. Ecological Mon ographs 51:25 -41. Lowman, M.D. and H. Heatwole. 1992. Spatial and temporal variability in defoliation of Australian eucalypts. Ecology 73:129 -142. Maron, J.L. and M. Vila. 2001. When do herbivores affect p lant invasion? Evidence for the natural enemies and biotic resistance hypotheses. Oikos 9 5:361-373. Marquis R.J. and C.J. Whelan. 1994. Insectivorous birds increase growth of white oak through consumption of leaf -chewing insects. Ecology 75:2007 -2014. Martin, P.H., C.D. Canham, and R .K. Kobe. 2010. Divergence from the growth -survival trade - off and extreme high growth rates drive patterns of exotic tree invasions in closed -canopy forests. Ecology 98:778-789. Martins, E.P. and T.F. Hansen. 1997. Phylogenies and the comparative method : a general approach to incorporating phylogenetic in - formation into the analysis of interspecific data. The American Naturalist 149:646 -667. Maschinski J, and T.G. Whitham. 1989. The continuum of plant responses to herbivory: the influence of plant as sociation, nutrient availability, and timing. The American Naturalist 134:1 -19. McCarthy -Neumann, S. and R. K. Kobe. 2008. Tolerance of soil pathogens co -varies with shade tolerance across species of tropical tree seedlings. Ecology 89:1883 -1892. McNaughton, S.J. 1983. Compensatory plant growth as a response to herbivory. Oikos 40:329 - 336. Meiners S .J. and C. -H. Kong. 2012. Introduction to the special issue on al lelopathy. Plant Ecology 213: 1857-1859. Michigan Department of Natural Resources and Environment [MDNR]. 2010. Michigan Deer Management Plan. Lansing, Michigan. Available at: http://www.michigan.gov/dnr/0,4570,7 -153-10363_41153---,00.html Miller, G.A., a nd J.P. Chapman. 2001. Misunderstanding Analysis of Covariance. Journal of Abnormal Psychology 110:40 -48. Mitchell, C. E., and A. G. Power. 2003. Release of invasive plants from fungal and viral pathogens. Nature 421:625 -627. Mitchell, C.E., A.A. Agr awal, J.D. Bever, G.S. Gilbert, R.A. Hufbauer, J.N. Klironomos, J.L. Maron, W.F. Morris, I.M. Parker, A.G. Power, E.W. Seabloom, M.E. Torchin, and D.P. 138 Vazquez. 2006. Biotic interactions and plant invasions. Ecology Letters 9:726 -740. Mitchell, C.E., D. Blumenthal, V. Jaro " ™k, E.E. Puckett, and P. Py " ek. 2010. Controls on pathogen species richness in plantsÕ introduced and native ranges: roles of residence time, range size and host traits. Ecology Letters 13:1525 -1535. Morris, W.F., R.A. Hufbauer, A.A. Agrawal, J.D. Bever, and V.A. Borowicz. 2007. Direct and interactive effects of enemies and mutualists on plant performance: a meta -analysis. Ecology 88:1021 -1029. Morrison, W.E. and M.E. Hay. 2011. Herbivore preference for native vs. exotic plants: generalist herbivores from multiple continents prefer exotic plants that are evolutionarily naŁve. PloS one 6:e17227. Munger, J.C. and J.H. Brown. 1981. Competition in desert rodents: an experiment using semipermeable excl osures. Science 211:510 -512. Olson B.E. and J.H. Richards. 1988. Tussock regrowth after grazing: intercalary meristem and axillary bud activity of tillers of Agropyron desertorum . Oikos 51:374 Ð82 Paige K.N. 1992. Overcompensation in response to mammalian herbivory from mutualistic to antagonistic interactions. Ecology 73:2076 -2085. Paige K.N., and T.G. Whitham TG. 1987. Overcompensation in response to mammalian herbivory: the advantage of bei ng eaten. The American Naturalist 129:407 -416. Paradis, E. 2012. Analysis of Phylogenetics and Evolution with R (Second Edition). New York: Springer. Parker, I .M. and G.S. Gilbert. 2007. When There Is No Escape: The Effects of Natural Enemies on Nati ve, In vasive, and Noninvasive Plants. Ecology 88:1210-1224. Parker, I.M., M. Saunders, M. Bontrager, A.P. Weitz, R. Hendricks, R. Magarey, K. Suiter, and G.S. Gilbert. 2015. Phylogenetic structure and host abundance drive disease pressure in communities. Nature 520:542 -544. Parker, J.D., and M.E. Hay. 2005. Biotic resistance to plant invasions? Native herbivores prefer non-native plants. Ecology Letters 8:959 -967. Parker, J.D., D.E. Burkepile, and M.E. Hay. 2006. Opposing Effects of Native and Exotic Herbivores on Plant Invasions. Science 311:1459 -1461. Pearse, I.S. and A.L. Hipp. 2009. Phylogenetic and trait similarity to a native species predict herbivory on non -native oaks. Proceedings of the National Academy of Sciences 106:18097 -18102. 139 Pinheir o, J., D. Bates, S. DebRoy, D. Sarkar and R Core Team. 2015. nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.1 -119, Available at: http://CRAN.R -project.org/package=nlme. Powell, K.I., J.M. Chase, and T.M. Knight. 2011. A synthesis of plant invasion effects on biodiversity across spatial scales. American Journal of Botany 98:539 -548. Powell, K.I., J.M. Chase, and T.M. Knight. 2013. Invasive plants have scale -dependent ef fects on diversity by altering species -area relationships. Science 339:316 -318. Quader, S., K. Isvaran , R.E. Hale , B.G. Miner, and N.E. Seavy . 2004. Nonlinear relationships and phylogenetically independent contrasts. Journal of Evolutionary Biology 17:709 -715. R Core Team. 2015 . R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria . Available at: http://www.R -project.org/ Raudenbush, S.W. 1994. Random effects models. In H. Cooper and L. V . Hedges ( eds.). The Handbook of research synthesis. (pp. 301 -321). Russell Sage Foundation, New York. Revell, L.J. (2012) phytools: An R package for phylogenetic comparative biology (and other things). Methods in Ecology and Evolution 3:217 -223. Rez nicek, A.A., E.G. Voss, and B.S. Walters. 2011. Michigan Flora Online. University of Michigan. Available at: http://michiganflora.net/home.aspx Rhoades, D.F. and R.G. Cates. 1976. Towards a general theory of plant antiherbivore chemistry. Recent Advances in Phytochemistry. 10:168 -213. Richardson D.M., P. Pysek, M. RejmaneK, M.G. Barbour, F.D. Panetta, and C.J. West. 2000. Naturalization and invasion of alien plants: Concepts and definitions. Diversity and Distributions 6:93 -107. Rice E .L. 1974. Allelopathy. Academic Press, New York Ridenour, W.M. and R. M. Callaway. 2001. The relative importance of allelopathy in interference: the effects of an invasive weed on a native bunchgrass. Oecologia 126:444 -450. Robertson, B.A., C. Porter, D.A . Landis, and D. W. Schemske. 2011. Agroenergy Crops Influence the Diversity, Biomass, and Guild Structure of Terrestrial Arthropod Communities. BioEnergy Research 5:179 -188. Rodriguez, L.F. 2006. Can invasive species facilitate native species? Evidence o f how, when, and why these impacts occur. Biological Invasions 8:927 -939. Rosenberg, M.S., D.C. Adams , and J. Gurevitch. 1 999. MetaWin: statistical software for meta - analysis . Version 2.0. Sinauer Associates, Sunderland. 140 Sala, O.E., F.S. Chapin, J.J. Armesto, E. Berlow, J. Bloomfield, R. Dirzo, E. Huber -Sanwald , L.F. Huenneke , R.B. Jackson , A. Kinzig , et al. 2000. Global biodiversity scenarios for the year 2100. Science 287:1770 -1774. Schultheis, E.H., A.E. Berardi, and J.A. Lau. 2015. No release for the wicked: enemy releas e is dynamic and not associated with invasiveness. Ecology in press (Chapter 2) 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. Proce edings of the National Academy of Sciences 100:13384 -13389. Shea, K. and P. Chesson. 2002. Community ecology theory as a framework for biological invasions. Trends in Ecology & Evolution 17:170 -176. Siemann, E., W.E. Rogers, and S.J. Dewalt. 2006. Rap id adaptation of insect herbivores to an invasive plant. Proceedings of the Royal Society B: Biological Sciences 273:2763 -2769. Smith, S.A. and C.W. Dunn. 2008. Phyutility: a phyloinformatics tool for trees, alignments, and molecular data. Bioinformatics 24:715 -716. Stamatakis, A. 2014. RAxML version 8: a tool for phylogenetic analysis and post -analysis of large phylogenies. Bioinformatics 30: 1312 Ð3. Stowe, L.G. 1979. Allelopathy and its Influence on the Distribution of Plants in an Illinois Old - Field. Journal of Ecology 67:1065 -1085. Stowe, K.A., R.J. Marquis, C.G. Hochwender, and E.L. Simms. 2000. The evolutionary ecology of tolerance to consumer damage. Annual Review of Ecology, Evolution and Systematics 31:565 Ð595. Strauss, S.Y. and A.A. Agrawal. 1999. The ecology and evolution of plant tolerance to herbivory. Trends in Ecology and Evolution 14:179 -185. Strauss, S.Y., C.O. Webb, and N. Salamin. 2006. Exotic taxa less related to native species are more invasive. Proceedin gs of the National Academy of Sciences 103:5841 -5845. Strayer, D.L., V. T. Eviner , J.M. Jeschke, and M.L. Pace. 2006. Understanding the long -term effects of species invasions. Trends in Ecology and Evolution 21:64 5-651. Stricker, K .B. and P. Stiling . 2014. Release from herbivory does not confer invasion success for Eugenia uniflora in F lorida. Oecologia 174:817 -826. Strong, D.R., E.D. McCoy, and J.R. Rey . 1977. Time and the number of herbivore species: the pests of sugarcane . Ecology 58:167-175. 141 Stowe, K.A., R.J. Marquis, C.G. Hochwender, and E.L. Simms. 2000. The evolutionary ecology of tolerance to consumer damage. Annual Review of Ecology, Evolution and Systematics 31:565 Ð595. Stricker, K.B. and P. Stiling. 2012. Herbivory by an introduced A sian weevil negatively affects population growth of an invasive Brazilian shrub in Florida. Ecology 93:1902 -1911. Strauss, S.Y. and A.A. Agrawal. 1999. The ecology and evolution of plant tolerance to herbivory. Trends in Ecology & Evolution 14:179 -185. Strayer, D.L., V.T. Eviner, J.M. Jeschke, and M.L. Pace. 2006. Understanding the long -term effects of species invasions. Trends in Ecology and Evolution 21:645 -651. Suding, K.N., W. Stanley Harpole, T. Fukami, A. Kulmatiski, A.S. MacDougall, C. Stein , W.H. van der Putten, and M. Hutchings. 2013. Consequences of plant -soil feedbacks in invasion. Journal of Ecology 101:298 -308. Suwa, T. and S.M. Louda. 2012. Combined effects of plant competition and insect herbivory hinder invasiveness of an introduced thistle. Oecologia 169:467 -476. Swenson, N.G. 2014. Functional and Phylogenetic Ecology in R. Springer, New York. Thebaud, C.and D. Simberloff. 2001. Are plants really larger in their introduced ranges? The American Naturalist 157:231 -236. Thellung, A. 1912. La flore adventice de Montpellier. M”moires de la Soci”t” nationale des sciences naturelles de Cherbourg 38:57 -728. Theoharides, K.A. and J.S. Dukes. 2007. Plant invasion across space and time: factors affecting nonindigenous speci es success during four stages of invasion. New Phytologist 176:256 - 273. Thomas, A., R. O'Hara, U. Ligges, and S. Sturts . 2006. Making BUGS Open. R News 6:12-17. Torchin M .E., K.D. Lafferty, A.P. Dobson , V.J. McKenzie , and A.M. Kuris . 2003. Introduced species and their mis sing parasites. Nature 421: 628 -630. Torchin, M.E. and C.E. Mitchell. 2004. Parasites, Pathogens, and Invasions by Plants and Animals. Frontiers in Ecology and the Environment 2:183 -190. Tylianakis, J.M., R.K. Didham, J. Bascompte , and D.A. Wardle. 2008. Global change and species interactions in terrestrial ecosystems. Ecology Letters 11,:1 -13. Underwood, A.J. 1997. Experiments in Ecology. Cambridge , UK. van Kleunen, M., E. Weber, and M. Fischer . 2010. A meta -analysis of trait differences between 142 invasive and noninvasive plant species. Ecology Letters 13: 235-245. Vavra, M., C.G. Parks, and M.J. Wisdom. 2007. Biodiversity, exotic plant species, and herbivory: The good, the bad, and the ungulate. Fore st Ecology and Management 246:66 -72. Verhoeven, K.J. F., A. Biere, J.A. Harvey, and W.H. van der Putten . 2009. Plant invaders and their novel natural enemies: who is naŁve ? Ecology Letters 12:107-117. Vil‹, M., J.L. Espinar, M. Hejda, P.E. Hulme, V. Jarosik, J.L. Maron, J. Pergl, U. Schaffner, Y. Sun, and P. Pysek. 2011. Ecological impacts of invasive alien plants: a meta -analysis of their effects on species, communities and ecosystems. Ecology Letter s 14:702 -708. Williams, J.L., H. Auge, and J. L. Maron. 2010. Testing hypotheses for exotic plant success: parallel experiments in the native and introduced ranges. Ecology 91:1355 -1366. Williamson, M. and A. Fitter. 1996. The varying success of invader s. Ecology 77:1661 -1666. Wootton, J.T. 1994. The nature and consequences of indirect effects in ecological communities. Annual Review of Ecology and Systematics 25:443 Ð466. Zou, J., W.E. Rogers, and E. Siemann. 2008. Increased competitive ability and herbivory tolerance in the invasive plant Sapium sebiferum . Biological Invasions 10:291 -302.