Title of My Thesis

Home , Chaitophorus, Halictus

UC San Diego UC San Diego Electronic Theses and Dissertations

Title The broader ecological effects of species invasion on protection mutualisms /

Permalink https://escholarship.org/uc/item/82p479kq

Author LeVan, Katherine Elizabeth

Publication Date 2013

Peer reviewed|Thesis/dissertation

eScholarship.org Powered by the California Digital Library University of California UNIVERSITY OF CALIFORNIA, SAN DIEGO

The broader ecological effects of species invasion on protection mutualisms

A dissertation submitted in partial satisfaction of the

requirements for the degree Doctor of Philosophy

Biology

Katherine Elizabeth LeVan

Committee in charge:

Professor David A. Holway, Chair Professor Elsa Cleland Professor Micky D. Eubanks Professor Phil Hastings Professor James Nieh

2013

Katherine Elizabeth LeVan, 2013

This dissertation of Katherine Elizabeth LeVan is approved, and it is acceptable in quality and form for publication on microfilm and electronically:

______

Chair

University of California, San Diego

2013

iii DEDICATION

To my parents, family, and friends that provided me emotional support without which this dissertation would have been exceedingly less likely to have been completed. In some sense, this is as much your accomplishment as mine. Thanks guys.

iv TABLE OF CONTENTS

Signature Page…..…………………………………………………………………………...... iii

Dedication……………………………………………………………………………………...... iv

Table of Contents…………….………………………………………………………………..... v

List of Figures…………………………………………………………………...... vii

List of Tables………………………………………………………………………………….... viii

Acknowledgements…………………………………………………………………………...... ix

Vita……………………………………………...……………………………………………..... xi

Abstract of the Dissertation……..……………………………………………………….……. xii

Introduction of the Dissertation.…………………………………………………………...... 1

References...... …………………………………………………..…....…...... 6

Floral visitation by the Argentine ant reduces pollinator visitation and seed set in the coast barrel cactus, Ferocactus viridescens ………………………………….……………. 12

Abstract…………………………………………………………………..…....………. 12

Introduction..……………………………………………………………..…....………. 12

Methods...... ……………………………………………………………..…....………. 13

Results...... ……………………………………………………………..…....………. 15

Discussion...……………………………………………………………..…....………. 18

Acknowledgements……………………………………………………..…....………. 19

References...... ……………………………………………………..…....………. 19

Acknowledgements……………………………………………………..…....………. 20

Argentine ant invasion reduces mutualist guild diversity and food web complexity…… 21

Abstract…………………………………………………………………..…....………. 21

Introduction..……………………………………………………………..…....………. 22

v Methods...... ……………………………………………………………..…....………. 26

Results...... ……………………………………………………………..…....………. 29

Discussion...……………………………………………………………..…....………. 32

Acknowledgements……………………………………………………..…....………. 35

References...... ……………………………………………………..…....………. 36

Ant-aphid mutualism increases floral visitation by ants and reduces plant fitness via decreased pollinator visitation …………………………………………………………….… 47

Abstract…………………………………………………………………..…....………. 47

Introduction..……………………………………………………………..…....………. 48

Methods...... ……………………………………………………………..…....………. 51

Results...... ……………………………………………………………..…....………. 55

Discussion...……………………………………………………………..…....………. 58

Acknowledgements……………………………………………………..…....………. 61

References...... ……………………………………………………..…....………. 62

An experimental test of how mutualisms between the Argentine ant and cotton aphid structure arthropod food webs in cotton...... …………………………...... 73

Abstract…………………………………………………………………..…....………. 73

Introduction..……………………………………………………………..…....………. 74

Methods...... ……………………………………………………………..…....………. 75

Results...... ……………………………………………………………..…....………. 78

Discussion...……………………………………………………………..…....………. 81

Acknowledgements……………………………………………………..…....………. 84

References...... ……………………………………………………..…....………. 84

vi LIST OF FIGURES

Figure 1.1. Argentine ant vs. native ant abundance in flowers………………………….. 16

Figure 1.2. A test of ant aversion toward petal compounds………………………..…..... 16

Figure 1.3. Number and duration of cactus bee visits………………………….……….... 17

Figure 1.4. Cactus reproduction versus ant species identity………….……………...…. 17

Figure 2.1. A map of 6 paired riparian sites…………………………………………..….... 43

Figure 2.2. Aphid richness versus ant invasion status………………………………….... 43

Figure 2.3. Interaction networks for invaded and uninvaded sites...……………….….... 44

Figure 2.4. Ant:aphid ratio decreases with aggregation size……………………….….... 45

Figure 2.5. Measures of food web complexity change with ant richness………….….... 46

Figure 3.1. Estimated cotton aphid abundance in three years………………………….. 69

Figure 3.2. Effect of aphids on ant abundance in leaves and flowers………………….. 70

Figure 3.3. Effect of ant floral visitation on pollinator and bee visit duration………….... 71

Figure 3.4. Effect of aphid abundance on seed production……………………………... 72

Figure 4.1. Mean time-averaged aphid and ant abundance on cotton...... ……………. 93

Figure 4.2. Effect of aphid density treatment on morphospecies composition...... 94

Figure 4.3. Effect of aphid density treatment on linkage within food webs…………….. 95

Figure 4.4. Effect of ant presence on generalist and specialist predator behavior toward aphid prey...... ……………. 96

Figure 4.5. Effect of aphid-density treatment on Lepidotera abundance and feeding… 97

vii LIST OF TABLES

Table 1.1: Reproductive traits of coast barrel cacti (Ferocactus viridescens).………..... 14

Table 1.2: The effects of ant species identity and year on plant reproduction...... 17

Table 2.1: Species distributions of ants, aphids, and wasps……………………….…….. 42

Table 3.1: Details of the experimental regime that differed between years……………. 67

Table 3.2: Results of GLM analyzing the effect of aphid abundance on plant fitness… 68

Table 4.1: Identities and abundance of species in the study…………………………..…. 88

viii ACKNOWLEDGEMENTS

The support that I received from mentors, colleagues, and family were integral to the success of this dissertation. Without their assistance and guidance, this thesis would not have been possible. Of course, I must acknowledge Professor David Holway for his support as the chair of my committee. Through multiple drafts, his guidance has proved to be invaluable. However, an advisor is only one of the many people that are a support network. My committee has also provided expertise and guidance on experimental design and statistical analysis, without which, this dissertation would have been the poorer. Thus, I gratefully acknowledge the assistance received from Elsa Cleland,

James Nieh, Phil Hastings, and Micky Eubanks. Discussions with this group of committee members have enriched my scientific development and this project.

Beyond the official assistance that a committee provides, I also benefited from conversations and advice from members within the lab and grad community. For long talks about science, crazy news stories, and international differences between Canada and the US, I have to thank Erin Wilson, Elinor Lichtenberg, James Hung, Chris Kopp,

Akana Noto, Ellen Esch and Celia Symons. The success of my research was improved by access to field sites, for which Isabelle Kay and Pablo Bryant were extremely helpful.

Finally, I must thank my family and close friends who endured a lot of late night conversations and supported me emotionally through the roller coaster that is a PhD thesis. Their encouragement has been priceless to me and has significantly (p < 0.05) contributed to me completing the dissertation. For that I will always be grateful. In particular, I want to say that Meghan F. Hogan and Peter C. St. John are probably the best two people in the world and I hope that they both know that their support was of great value to me in this difficult process.

ix Chapter 1, in full, is a reprint of an article as it appears in Oecologia. LeVan, KE,

KJ Hung, KR McCann, JT Ludka and DA Holway. 2013. “Floral visitation by the

Argentine ant reduces pollinator visitation and seed set in the coast barrel cactus,

Ferocactus viridescens.” The dissertation author was the primary investigator and author of this manuscript.

Chapter 2, in part, has been submitted for publication as it may appear in

Molecular Ecology. LeVan, KE and DA Holway. “Argentine ant invasion reduces mutualist guild diversity and food web complexity.” The dissertation author was the primary investigator and author of this paper.

Chapter 3, in part, has been submitted for publication as it may appear in

Ecology. LeVan, KE and DA Holway. “Ant-aphid mutualism increases floral visitation by ants and reduces plant fitness via decreased pollinator visitation.” The dissertation author was the primary investigator and author of this paper.

Chapter 4, in part is currently being prepared for submission for publication.

LeVan, KE and DA Holway. “The mutualism between the Argentine ant and cotton aphid structures food webs in cotton.” The dissertation author was the primary investigator and author of this material.

x VITA

2008 Bachelor of Science, Tufts University

2013 Doctor of Philosophy, University of California, San Diego

AWARDS

2012 - 2013 “Achievement Rewards for College Scientists” Foundation Fellow

2010 - 2011 Mildred Mathias Natural Reserve System Grantee

2009 - 2012 National Science Foundation Graduate Research Fellow

2008 - 2009 Jeanne Marie Messier Awardee

PUBLICATIONS

LeVan, KE, K-L James Hung, KR McCann, JT Ludka and DA Holway. 2013. Floral visitation by the Argentine ant reduces pollinator visitation and seed set in the coast barrel cactus, Ferocactus viridescens. Oecologia.

Wilson, EE, CS Sidhu, KE LeVan and DA Holway. 2010. Pollen foraging behaviour of solitary Hawaiian bees revealed through molecular pollen analysis. Molecular Ecology.

19(21): 4823-4829.

LeVan, KE, TY Fedina and SM Lewis. 2009. Testing multiple hypotheses for the maintenance of homosexual copulatory behavior in flour beetles. Journal of Evolutionary

Biology. 22: 60-70.

South, A, K LeVan, L Leombruni, CM Orians, SM Lewis. 2008. Examining the role of cuticular hydrocarbons in firefly species recognition. Ethology. 114 (9): 916-924.

LeVan, KE and N Wilson-Rich. 2006. Behavioral Ecology: An honest and deceitful review. Science. 314 (5801):927-928

xi ABSTRACT OF THE DISSERTATION

The broader ecological effects of species invasion on protection mutualisms

Katherine Elizabeth LeVan

Doctor of Philosophy in Biology

University of California, San Diego, 2013

Professor David A. Holway, Chair

Species introductions dramatically alter the diversity and abundance of native species via competitive and predatory interactions that negatively influence native biota.

Although invasive species sometimes form mutualisms within their introduced range, the role of mutualistic interactions on species abundance and food web composition is less known. In this study I examined the invasive Argentine ant (Linepithema humile), a species which readily interacts with carbohydrate-producers within its introduced range.

This study investigates these (presumably) reciprocally beneficial food-for-protection mutualisms to determine the broader ecological effects of the mutualism on species external to mutualism. Using field experiments and molecular techniques, I quantify the effects of the Argentine ant on food web structure, diversity, and plant reproduction. This study identifies the broader ecological effects of invasion, and provides insight into

xii management strategies that may minimize the ecological impacts of invasive social insects on native biota.

xiii

INTRODUCTION OF THE DISSERTATION

Mutualisms are the most ubiquitous interactions in ecology, occurring in diverse species from microbes (Wernegreen 2002) to mammals (Willson 1993). All higher terrestrial plants, many vertebrates, and a variety of arthropods are involved in mutualisms (Boucher et al. 1982). Defined as an interaction between species that is beneficial to both, mutualisms are generally categorized as nutritional (e.g., endosymbionts, nitrogen fixation by microbes, plant-pollinator mutualisms), protective

(e.g., from predators, parasites, disease, toxins) or transportive (e.g., seed dispersal)

(Boucher et al. 1982). Mutualisms can be obligate (coevolved interactions between species dependent upon one another for survival or reproduction) or facultative (loose associations where one or both partners can survive or reproduce without the other partner). Generalized facultative interactions are most prevalent (Way 1963; Bentley

1977; Hölldobler & Wilson 1990; Stadler & Dixon 2005) and a single species can be involved in one or more types of mutualism (Boucher et al. 1982).

Mutualisms are conventionally viewed as interactions between two partners but not necessarily between equal partners. In symmetric mutualisms, both partners gain similar benefits, while asymmetric interactions result in one partner receiving the greater net benefit (Bronstein 2001). Mechanistic analyses of mutualisms can provide information regarding the benefits that each partner derives, the strength of the association (Bronstein 1994), or the degree of symmetry (Bronstein 2001). Because studies of mutualism occur in isolation, they are unable to quantify how these binary relationships influence the food webs in which they are embedded (Polis & Strong 1996;

Stanton 2003; Styrsky & Eubanks 2007). This binary analysis is incomplete because pairwise mutualisms frequently interact with many species in natural systems (Polis &

Strong 1996; Kaplan & Eubanks 2005; Schumacher & Platner 2009).

1 2

When mutualisms are placed in a food web context, species outside of the mutualism can respond in unexpected ways to the presence and strength of mutualisms

(Stachowicz 2001). Recent reviews suggest that mutualisms drive species composition

(Bruno et al. 2003), though few studies explicitly test this hypothesis (Wimp & Whitham

2001; Styrsky & Eubanks 2007). From what information is available, mutualistic interactions change the structure of trophic levels above (Kaplan & Eubanks 2002) and below (O’Dowd et al. 2003). In facultative associations, participants are somewhat interchangeable with the potential to impact a food web more than an obligate pair.

Flexibility in pairings has important implications for changes in trophic interactions. In these loose associations, each participant is one of many potential partners and a particular pairwise interaction may only exist because a more desirable partner is unavailable (Cushman & Whitham 1991; Fischer et al. 2001; Blüthgen &

Fiedler 2004). In this situation, the identity of the mutualist and strength of the mutualism determines the impact of the relationship on its food web. As a result, abundances of available mutualists may become important to the pressure that a mutualism exerts on community structure.

Ant mutualisms

Many ant species form facultative, protective mutualisms (Hölldobler & Wilson

1990), receiving rewards from their partners in exchange for protection from enemies. In mutualistic interactions, ants may receive carbohydrates (Völkl et al. 1999), amino acids

(Gonzalez-Teuber & Heil 2009), or nesting sites (Gaume et al. 2005). Because of these incentives, a variety of ant species will diligently tend plant extrafloral nectaries (Bentley

1977), many Hemiptera (Way 1963; Stadler & Dixon 2005) and some lycaenids (Pierce et al. 2002). However, the services that the ants provide are variable. For example, not all ant species provide high quality protection (Ness et al. 2006) and ants are completely

ineffective at warding off leaf miners and gall-forming herbivores (Ito & Higashi 1991). In the case of honeydew-producing insects, ants may provide some level of protection

(Wimp & Whitham 2001; Grover et al. 2008), but are less effective against specialist predators (Wimp & Whitham 2001) or parasitoids (Herbert & Horn 2008).

Generally speaking, plants that form mutualisms with ants derive a variety of benefits. Plants receive protection from some herbivores (Rudgers 2004) because of increased ant presence. These ant-plant mutualisms can result in increased growth

(Sipura 2002; Mooney 2007), fitness (Frederickson et al. 2005; Gaume et al. 2005), and dispersal (Christian 2001; Carney et al. 2003) of dependent taxa. In return, plants provide nutrients (Bentley 1977) or nesting sites (Gaume et al. 2005). In spite of the cost of producing rewards (Keeler 1985; Mondor et al. 2006), many plants will induce EFN secretions in response to herbivore damage to attract ants (Mondor & Addicott 2003;

Pulice & Packer 2008).

Mutualisms between ants and carbohydrate-producers are extremely labile.

Variation in tending ability translates into changes in population growth (Nygard et al.

2008) and amount of honeydew harvested by ants (Paris & Espadaler 2009). Gaume et al. (1998) indirectly demonstrated that a single species of ant may effectively tend one hemipteran, but ineffectually tend another type. Differences in tending may arise from ant preferences for more nutritious honeydew (Fischer & Shingleton 2001; Blüthgen &

Fiedler 2004), certain morphological features (Mondor et al. 2002) or behavioral characteristics (Dixon 1998). Currently, it is unclear why one honeydew producer forms a mutualism when another does not. Studies comparing the maintenance of mutualisms within a clade of Chaitophorus aphids determined that these mutualisms have been gained or lost several times in their evolutionary history (Shingleton & Stern 2003), but have not offered explanations for this phenomenon.

Effects of mutualisms between honeydew-producing insects and ants on host plants

Quantifying net effects of mutualisms is complicated by the fact that mutualisms are not just a positive-feedback loop; instead, they are a combination of positive and negative effects that vary in strength. Direct mutualisms between ants and plants are typically beneficial (Bentley 1977), but indirect mutualisms are more variable.

Honeydew-producing insects provide carbohydrate resources to ants (Buckley 1982) and increase plant protection through elevated ant presence (Grover et al. 2008), thus increasing the suppression of certain types of herbivores (Fowler & Macgarvin 1985).

Although hemipterans can be beneficial to the host plant (Messina 1981), they are themselves a sap-sucking herbivore and sometimes vector disease (Kaloshian & Walling

2005). Moreover, ant-hemipteran mutualisms are known to negatively impact plant growth (Buckley 1982) and fitness (Banks & Macaulay 1967) in herbaceous plants.

Lowered seed set may reflect increased seed abortion due to a depletion of phloem nutrients (Dixon 1998) or decreased pollination as a result of diminished floral nectar palatability.

Trophic effects

The effects of ant-hemipteran mutualisms are not limited to direct and indirect participants. Mutualisms between ants and honeydew-producing insects also affect higher trophic levels, reducing abundances of generalist predators (Wimp & Whitham

2001) and pollinators (Lach 2007). Some ant species can harass pollinators (Ness

2006), which may be facilitated with ant-hemipteran mutualisms (Lach 2007). In contrast, specialist predators of tended hemipterans are most common where ant-hemipteran mutualisms occur (Way 1963; Wimp & Whitham 2001). Moreover, rate of parasitism by parasitoids specializing on aphids are reduced by ants in one study (Fischer et al. 2002), but not another (Savage & Peterson 2007). As a result, it is unclear if protection from

parasitoids depends on the ant species defending the hemipteran, the aggregation size that the ants must patrol, or the density of parasitoids. Thus, the effect of mutualisms can permeate multiple trophic levels.

Linepithema humile

The Argentine ant (Linepithema humile) is a widespread, abundant, and disruptive invasive species (Holway et al. 2002). Linepithema humile foragers are small in size and recruit readily to available resources, preferring liquid food over seeds or arthropods (Newell & Barber 1913; Markin 1970). Although L. humile is native to central

South America, the Argentine ant has spread throughout Mediterranean-type climates across the globe (Roura-Pascal et al. 2004) and thrives in riparian habitats in both its native and introduced ranges (Menke & Holway 2006). Argentine ants are polydomous and polygynous, where queens and workers form large, diffuse colonies (Suarez et al.

1999). Unicoloniality, high interspecific aggression coupled with low intraspecific conflict are the most notable behavioral traits exhibited by L. humile in its introduced range. The widespread success of L. humile may be explained in part by its ability to tend aphids.

Because L. humile is aggressive and has a high foraging tempo (Holway 1998a; Holway

& Suarez 1999), Argentine ants may be well suited to access and defend honeydew resources in the field. As a result of carbohydrate supplementation, they may more greatly alter species composition in arthropod food webs and derive increased success at the individual and population levels.

In their introduced range, L. humile displaces native ants (Holway 1998b, 2005;

Tillberg et al. 2007) and may decrease overall arthropod diversity (Human & Gordon

1997). Several studies have indirectly examined how native (Fowler & Macgarvin 1985;

Ito & Higashi 1991) or invasive (Altfeld & Stiling 2009) ant predators influence ant biodiversity. Despite its extensive distribution and ecological dominance, the specific

effects of the Argentine ant on arthropod food webs remain incompletely studied.

Because of limitations in experimental design, it is unclear whether invasive ant species generally suppress diversity or only target specific groups within a food web.

The broader ecological effects of species invasion on protection mutualisms

Because ant mutualisms can influence host plant reproduction (Banks &

Macaulay 1967), herbivore load (Kaplan & Eubanks 2005), predator and parasite abundance (Wimp & Whitham 2001) and pollinator visitation (Lach 2007), my thesis investigates the species composition effects of mutualisms with an introduced ant.

Chapter 1 of the thesis focuses on ant-plant mutualisms in coastal San Diego. Chapter 2 examines the effect of an invasive ant on the composition of the hemipteran mutualist guild. Chapter 3 determines the effects of ant-aphid mutualisms on pollinator visitation and Chapter 4 evaluates the effect of ant-aphid mutualisms on food web structure.

REFERENCES

Altfeld L, Stiling P (2009) Effects of aphid-tending Argentine ants, nitrogen enrichment and early-season herbivory on insects hosted by a coastal shrub. Biological Invasions, 11, 183–191.

Banks C, Macaulay E (1967) Effects of Aphis fabae Scop. and of its attendant ants and insect predators on yields of field beans (Vicia faba L.). Annals of Applied Biology, 60, 445–453.

Bentley BL (1977) Extrafloral nectaries and protection by pugnacious bodyguards. Annual Review of Ecology and Systematics, 8, 407–427.

Blüthgen N, Fiedler K (2004a) Preferences for sugars and amino acids and their conditionality in a diverse nectar-feeding ant community. Journal of Animal Ecology, 73, 155–166.

Blüthgen N, Fiedler K (2004b) Competiting for composition: lessons from nectar-feeding ant communities. Ecology, 85, 1479–1485.

Boucher DH, James S, Keeler KH (1982) The ecology of mutualism. Annual Review of Ecology and Systematics, 13, 315–347.

Bronstein JL (1994) Conditional outcomes in mutualistic interactions. Trends in Ecology & Evolution, 9, 214–217.

Bronstein JL (2001) The exploitation of mutualisms. Ecology Letters, 4, 277–287.

Bruno JF, Stachowicz JJ, Bertness MD (2003) Inclusion of facilitation into ecological theory. Trends in Ecology & Evolution, 18, 119–125.

Buckley R (1982) Interactions between ants and membracid bugs decreases growth and seed set of host plant bearing extrafloral nectaries. Oecologia, 58, 132–136.

Carney SE, Brooke Byerley M, Holway DA (2003) Invasive Argentine ants (Linepithema humile) do not replace native ants as seed dispersers of Dendromecon rigida (Papaveraceae) in California, USA. Oecologia, 135, 576–582.

Christian CE (2001) Consequences of a biological invasion reveal the importance of mutualism for plant communities. Nature, 413, 635–639.

Cushman JH, Whitham TG (1991) Competition mediating the outcome of a mutualism: protective services of ants as a limiting resource for membracids. Ecological Entomology, 31, 568–574.

Dixon AFG (1998) Aphid ecology: an optimization approach. Chapman and Hall, London.

Fischer MK, Hoffmann KH, Völkl W (2001) Competition for mutualists in an ant- homopteran interaction mediated by hierarchies of ant attendance. Oikos, 92, 531– 541.

Fischer MK, Shingleton AW (2001) Host plant and ants influence the honeydew sugar composition of aphids. Functional Ecology, 15, 544–550.

Fischer MK, Völkl W, Schopf R, Hoffmann KH (2002) Age-specific patterns in honeydew production and honeydew composition in the aphid Metopeurum fuscoviride: Implications for ant-attendance. Journal of Insect Physiology, 48, 319–326.

Fowler SV, Macgarvin M (1985) The impact of hairy wood ants, Formica lugubris, on the guild structure of herbivorous insects on birch, Betula pubescens. Journal of Animal Ecology, 54, 847–855.

Frederickson ME, Greene MJ, Gordon DM (2005) “Devil”s gardens’ bedevilled by ants. Nature, 437, 495–496.

Gaume L, McKey D, Terrin S (1998) Ant-plant-homopteran mutualism: how the third partner affects the interaction between a plant-specialist ant and its myrmecophyte host. Proceedings of the Royal Society B-Biological Sciences, 265, 569–575.

Gaume L, Zacharias M, Grosbois V, Borges RM (2005) The fitness consequences of bearing domatia and having the right ant partner:experiments with protective and non-protective ants in a semi-mymecophyte. Oecologia, 145, 76–86.

Gonzalez-Teuber M, Heil M (2009) The role of extrafloral nectar amino acids for the preferences of facultative and obligate ant mutualists. Journal of Chemical Ecology, 35, 459–468.

Grover CD, Dayton KC, Menke SB, Holway DA (2008) Effects of aphids on foliar foraging by Argentine ants and the resulting effects on other arthropods. Ecological Entomology, 33, 101–106.

Herbert JJ, Horn DJ (2008) Effect of ant attendance by Monomorium minimum (Buckley) (Hymenoptera: Formicidae) on predation and parasitism of the soybean aphid Aphis glycines Matsumura (Hemiptera: Aphididae). Environmental Entomology, 37, 1258–1263.

Hölldobler H, Wilson EO (1990) The Ants. Harvard University Press, Cambridge, Massachusetts.

Holway DA (1998a) Effect of Argentine ant invasions on ground-dwelling arthropods in northern California riparian woodlands. Oecologia, 116, 252–258.

Holway DA (1998b) Factors governing rate of invasion: a natural experiment using Argentine ants. Oecologia, 115, 206–212.

Holway DA (2005) Edge effects of an invasive species across a natural ecological boundary. Biological Conservation, 121, 561–567.

Holway DA, Lach L, Suarez A V, Tsutsui ND, Case TJ (2002) The causes and consequences of ant invasions. Annual Review of Ecology and Systematics, 33, 181–233.

Holway DA, Suarez A V (1999) Animal behavior: An essential component of invasion biology. Trends in Ecology & Evolution, 14, 328–330.

Human KG, Gordon DM (1997) Effects of Argentine ants on invertebrate biodiversity in northern California. Conservation Biology, 11, 1242–1248.

Ito F, Higashi S (1991) An indirect mutualism between oaks and wood ants via aphids. Journal of Animal Ecology, 60, 463–470.

Kaloshian I, Walling LL (2005) Hemipterans as plant pathogens. Annual Review of Phytopathology, 43, 491–521.

Kaplan I, Eubanks MD (2002) Disruption of cotton aphid (Homoptera: Aphididae) - Natural enemy dynamics by red imported fire ants (Hymenoptera: Formicidae). Environmental Entomology, 31, 1175–1183.

Kaplan I, Eubanks MD (2005) Aphids alter the community-wide impact of fire ants. Ecology, 86, 1640–1649.

Keeler KH (1985) Extrafloral nectaries on plants in communities without ants: Hawaii. Oikos, 44, 407–414.

Lach L (2007) A mutualism with a native membracid facilitates pollinator displacement by Argentine ants. Ecology, 88, 1994–2004.

Markin GP (1970) The seasonal life cycle of the Argentine ant, Iridomyrmex humilis (Hymenoptera: Formicidae), in Southern California. Annals of the Entomological Society of America, 63, 1238–1242.

Menke SB, Holway DA (2006) Abiotic factors control invasion by Argentine ants at the community scale. Journal of Animal Ecology, 75, 368–376.

Messina FJ (1981) Plant-protection as a consequence of an ant-membracid mutualism - interactions on goldenrod (Solidago sp). Ecology, 62, 1433–1440.

Mondor EB, Addicott JF (2003) Conspicuous extra-floral nectaries are inducible in Vicia faba. Ecology Letters, 6, 495–497.

Mondor EB, Roitberg BD, Stadler B (2002) Cornicle length in Macrosiphini aphids: A comparison of ecological traits. Ecological Entomology, 27, 758–762.

Mondor EB, Tremblay MN, Messing R (2006) Extrafloral nectary phenotypic plasticity is damage- and resource-dependent in Vicia faba. Biology Letters, 2, 583–585.

Mooney KA (2007) Tritrophic effects of birds and ants on a canopy food web, tree growth, and phytochemistry. Ecology, 88, 2005–2014.

Ness JH (2006) A mutualism’s indirect costs: the most aggressive plant bodyguards also deter pollinators. Oikos, 113, 506–514.

Ness JH, Morris WF, Bronstein JL (2006) Integrating quality and quantity of mutualistic service to contrast ant species protecting Ferocactus wislizeni. Ecology, 87, 912– 921.

Newell W, Barber TC (1913) The Argentine ant. USDA Bureau Entomological Bulletin, 122.

Nygard JP, Sanders NJ, Connor EF (2008) The effects of the invasive Argentine ant (Linepithema humile) and the native ant Prenolepis imparis on the structure of insect herbivore communities on willow trees (Salix lasiolepis). Ecological Entomology, 33, 789–795.

O’Dowd DJ, Green PT, Lake PS (2003) Invasional “meltdown” on an oceanic island. Ecology Letters, 6, 812–817.

Paris CI, Espadaler X (2009) Honeydew collection by the invasive garden ant Lasius neglectus versus the native ant L. grandis. Arthropod-Plant Interactions, 1–11.

Pierce NE, Braby MF, Heath A et al. (2002) The ecology and evolution of ant association in the Lycaenidae (Lepidoptera). Annual Review of Entomology, 47, 733–771.

Polis GA, Strong DR (1996) Food web complexity and community dynamics. The American Naturalist, 147, 813–846.

Pulice CE, Packer A (2008) Simulated herbivory induces extrafloral nectary production in Prunus avium. Functional Ecology, 22, 801–807.

Roura-Pascal N, Suarez AV, Gomez C et al. (2004) Geographical potential of Argentine ants (Linepithema humile Mayr) in the face of global climate change. Proceedings of the Royal Society B-Biological Sciences, 271, 2527–2534.

Rudgers JA (2004) Enemies of herbivores can shape plant traits: selection in a facultative ant-plant mutualism. Ecology, 85, 192–205.

Savage AM, Peterson MA (2007) Mutualism in a community context: the positive feedback between an ant-aphid mutualism and a gall-making midge. Oecologia, 151, 280–291.

Schumacher E, Platner C (2009) Nutrient dynamics in a tritrophic system of ants, aphids and beans. Journal of Applied Ecology, 133, 33–46.

Shingleton AW, Stern DL (2003) Molecular phylogenetic evidence for multiple gains or losses of ant mutualism within the aphid genus Chaitophorus. Molecular Phylogenetics and Evolution, 26, 26–35.

Sipura M (2002) Contrasting effects of ants on the herbivory and growth of two willow species. Ecology, 83, 2680–2690.

Stachowicz JJ (2001) Mutualism, facilitation, and the structure of ecological communities. BioScience, 51, 235–246.

Stadler B, Dixon AFG (2005) Ecology and evolution of aphid-ant interactions. Annual Review of Ecology, Evolution, and Systematics, 36, 345–372.

Stanton ML (2003) Interacting guilds: moving beyond the pairwise perspective on mutualisms. The American Naturalist, 162, S10–23.

Styrsky JD, Eubanks MD (2007) Ecological consequences of interactions between ants and honeydew-producing insects. Proceedings of the Royal Society B-Biological Sciences, 274, 151–64.

Suarez AV, Tsutsui ND, Holway DA, Case TJ (1999) Behavioral and genetic differentiation between native and introduced populations of the Argentine ant. Biological Invasions, 1, 43–53.

Tillberg CV, Holway DA, Lebrun EG, Suarez AV (2007) Trophic ecology of invasive Argentine ants in their native and introduced ranges. Proceedings of the National Academy of Sciences (USA), 104, 20856–20861.

Völkl W, Woodring J, Fischer MK, Lorenz MW, Hoffmann KH (1999) Ant-aphid mutualisms: The impact of honeydew production and honeydew sugar composition on ant preferences. Oecologia, 118, 483–491.

Way MJ (1963) Mutualism between ants and honeydew-producing Homoptera. Annual Review of Entomology, 8, 307–344.

Wernegreen JJ (2002) Genome evolution in bacterial endosymbionts of insects. Nature Reviews Genetics, 3, 850–861.

Willson MF (1993) Mammals as seed-dispersal mutualists in North America. Oikos, 67, 159–176.

Wimp GM, Whitham TG (2001) Biodiversity consequences of predation and host plant hybridization on an aphid-ant mutualism. Ecology, 82, 440–452.

12 13

ACKNOWLEDGEMENTS

Chapter 1, in full, is a reprint of an article as it appears in Oecologia. LeVan, KE,

KJ Hung, KR McCann, JT Ludka and DA Holway. 2013. “Floral visitation by the

Argentine ant reduces pollinator visitation and seed set in the coast barrel cactus,

Ferocactus viridescens.” The dissertation author was the primary investigator and author of this manuscript.

Argentine ant invasions homogenize networks of interacting aphid-tending ants,

ant-tended aphids and parasitoid wasps

ABSTRACT

Mutualisms contribute in fundamental ways to the origin, maintenance, and organization of biological diversity. Introduced species are well known to infiltrate mutualisms, and their participation in these interactions can come at the expense of other mutualists. The degree to which the usurpation of mutualisms by introduced species affects the diversity of native mutualists remains incompletely understood. As the transport and establishment of non-native species throughout the world increases, an appreciation of the effect on mutualisms is of clear importance. Here we use DNA barcoding to clarify how Argentine ant invasions affect the diversity of a multi-species, ant-aphid mutualism and in turn alter the structure of networks involving ants and aphids and those involving aphids and parasitoids. Invaded sites supported only one ant-tended aphid species, which was exclusively tended by the Argentine ant. In contrast, the interaction network between ant-tended aphids and aphid-tending ants at uninvaded sites included 16 different ant-aphid pairings involving four species of aphids and eight species of ants. We detected only one species of parasitoid wasp in our aphid samples, but the greater number of potential aphid host species at uninvaded sites led to pronounced differences in the structure of the willow-aphid-parasitoid food web. Food webs at uninvaded sites, for example, had more links per site and lower per site connectance compared to food webs at invaded sites. These results illustrate the potential for introduced species to reduce the diversity present in multi-species mutualisms and to homogenize the structure of associated interaction networks.

21 22

INTRODUCTION

Mutualisms contribute in fundamental ways to the origin, maintenance and organization of biological diversity (Bertness & Callaway 1994; Bruno et al. 2003;

Bascompte & Jordano 2007). Environmental changes brought about by human activity can alter the currencies that underlie mutualistic interactions and shift selective forces away from those that favor mutualism (Kiers et al. 2010). Species introductions represent an accelerating and dynamic form of environmental change that can disrupt mutualisms between native species (Traveset & Richardson 2006; Kiers et al. 2010).

However, the degree to which the cooption of mutualisms by introduced species affects the diversity of species that participate in mutualisms remains incompletely understood

(Traveset & Richardson 2006). Here we use DNA barcoding to identify aphids and their parasitoids to examine (1) how ant invasions affect the diversity of an assemblage of ant-tended aphids following displacement of native ants, and (2) how changes in aphid diversity in turn affect the structure of the ant-aphid interaction network and the structure of the host plant-aphid-parasitoid food web.

Interactions between ants and honeydew-producing insects are ancient, geographically widespread, and common in both temperate and tropical ecosystems

(Way 1963; Hölldobler & Wilson 1990; Grimaldi & Engel 2005; Stadler & Dixon 2005;

Oliver et al. 2008). In exchange for honeydew, which is partially digested plant sap rich in carbohydrates and (variably) amino acids (Blüthgen & Fiedler 2004; Blüthgen et al.

2006), ants provide sanitation services and protect honeydew producers from their natural enemies, such as parasitoid wasps (Way 1963; Hölldobler & Wilson 1990;

Stadler & Dixon 2005). Ant species opportunistically tend multiple honeydew-producing insect taxa (Fischer et al. 2001; Smith et al. 2008; Chong et al. 2010; Kaminski et al.

2010; Yoo et al. 2013), which in turn may be tended by more than one ant species

(Addicott 1979; Bristow 1984; Sipura 2002; Blüthgen et al. 2006). These interactions may thus commonly represent multi-species mutualisms consisting of two interacting mutualist guilds (sensu Stanton 2003): multiple ant species tending multiple honeydew- producing insect species. Introduced species nicely illustrate the open nature of these partnerships in that mutualisms often develop between ants and honeydew-producing insects that share no evolutionary history (Way 1963; Bach 1991; Helms & Vinson 2002;

Lach 2003; O’Dowd et al. 2003; Ness & Bronstein 2004; Green et al. 2011).

Pair-wise interactions between introduced ants and honeydew-producing insects often result in increased numbers of honeydew-producing insects relative to those present in ant exclusions (Lach 2003; Ness & Bronstein 2004; Helms & Vinson 2008).

Whether introduced ants differ from native ants with respect to their ability to tend honeydew-producing insects and whether these differences are sufficient to affect the abundance or diversity of honeydew-producing insects remain largely unexplored questions (Ness & Bronstein 2004). Introduced ants, however, can become abundant and recruit large numbers of workers to carbohydrate resources (Holway et al. 2002;

Lach 2003). To the extent that introduced ants differ from native ants with respect to these attributes, higher levels of tending (e.g., higher ant:aphid ratios) provided by introduced ants could translate into superior protection from enemies (Lach 2003).

Enhanced protection could potentially foster diversity if introduced ants do not discriminate with respect to which honeydew-producing insect species they tend.

Alternatively, introduced ants might selectively interact with the most prolific honeydew- producers and ignore or prey upon remaining taxa (Fischer et al. 2001; Powell &

Silverman 2010). If so, then honeydew-producing insect diversity might decrease relative to that present in areas occupied by multiple native ant species where territoriality and species-specific preferences on the part of native ants might maintain greater diversity.

To the extent that introduced ants differ from native ants with respect to how they tend honeydew-producing insects, disruptions to the network of interactions between ants and the insects that they tend may result in downstream effects on the consumers of honeydew-producing insects. If the enemies of honeydew-producing insects are host specific, for example, changes in the diversity of honeydew-producing insects that result from ant invasions could directly alter the composition of the enemy assemblage, perhaps via reduced ability to partition resources among enemies (Finke & Snyder

2008). In addition, mutualisms between ants and honeydew-producing insects often exert strong effects on the diversity and composition of arthropod assemblages present on host plants (Wimp & Whitham 2001; Styrsky & Eubanks 2007). Recent work illustrates that even the composition of the honeydew-producing insect assemblage itself can affect the behavior and local abundance of enemies as a result of changes in the foraging behavior of tending ants (Kaminski et al. 2010; Yoo et al. 2013). Especially in cases where the consistency, duration or dispersion of honeydew production increases with the diversity of honeydew-producing insects present, one might expect that arthropods sensitive to ants would be less abundant in the presence of multiple species of honeydew producers compared to situations involving just one species.

Species identification is a key impediment to clarifying how ant invasions affect assemblages of non-ant arthropods (Krushelnycky & Gillespie 2008; Lach et al. 2010).

This challenge may be particularly difficult in cases involving honeydew-producing insects and their enemies. Honeydew-producing hemipterans, and aphids in particular, exhibit complex life cycles (Stadler & Dixon 2005) and can vary in appearance as a function of temperature and nutrition (Nevo & Coll 2001; Blackman et al. 2011;

Richardson et al. 2011). DNA barcoding has eliminated many of the difficulties associated with species identifications of these insects (Traugott & Symondson 2008;

Traugott et al. 2008; Santos et al. 2011) and can also be used to differentiate difficult to identify enemy taxa (Andrew et al. 2013), such as hymenopteran parasitoids of aphids.

Identification of parasitoids formerly required rearing of larvae from hosts (Henneman &

Memmott 2001), but the development of molecular tools for parasitoid identification and detection inside hosts (Weathersbee et al. 2004; Santos et al. 2011; Derocles et al.

2012) has advanced an understanding of these interactions. DNA barcoding, for example, can reveal the presence of aphid parasitoids in the early stages of development and yield more accurate assessment of parasitism levels compared to those resulting from simple visual inspection (Hrcek et al. 2011).

In this study we investigate how invasion by the Argentine ant (Linepithema humile) affects the species composition of the ant-tended aphid assemblage on arroyo willow (Salix lasiolepis) in Southern California, and how changes in aphid diversity in turn affect species that interact with aphids. The Argentine ant tends a wide variety of honeydew-producing insects (Way 1963; Lach 2003; Ness & Bronstein 2004) and displaces aphid-tending native ants from riparian habitats (Ward 1987; Holway 1998a,

2005). Using DNA barcoding to identify aphids and their parasitoids, we address the following questions. (1) Does the diversity of the aphid mutualist assemblage increase or decrease as a function of ant invasion? (2) Do native ants differ from the Argentine ant with respect to their levels of aphid tending and or the aphid species that they tend? (3)

Do levels of aphid parasitism increase or decrease as a function of ant invasion? (4) Do invaded and uninvaded sites differ from one another with respect to the structure of the ant-aphid interaction network or the structure of the host plant-aphid-parasitoid food web? Answers to these questions will clarify how the assimilation of introduced species into multi-species mutualisms affects the diversity of native mutualists and the structure of associated interaction networks and food webs.

METHODS

Study system

The Argentine ant now occurs throughout many portions of California, where its ability to displace native ants from a variety of natural ecosystems is widely documented

(Tremper 1977; Ward 1987; Human & Gordon 1996; Holway 1998ab, 2005; Suarez et al. 1998; Mitrovich et al. 2010). The Argentine ant requires adequate levels of soil moisture to invade seasonally dry environments (Menke & Holway 2006; Menke et al.

2007) and attains exceptionally high abundance in moist riparian woodlands that border perennial streams and rivers (Holway 1998ab, 2005), ecosystems that resemble the preferred habitat of L. humile in its native Argentina (LeBrun et al. 2007). We selected six paired sites in riparian woodlands in San Diego and Riverside counties; each pair of sites was located in a different watershed and included an invaded portion of a riparian corridor and an uninvaded portion (Figure 2.1; Holway 2005). The Argentine ant is actively spreading at each invaded site (DAH, unpubl. data); native ants persist along uninvaded stretches of riparian corridor because L. humile has not yet reached these sites. Sites within each pair were separated by 3.9 (± 2.2 SE) km and were matched as closely as possible with respect to elevation, soil type, dominant perennial vegetation and total cover. The landscape-level pattern of invasion depicted in Figure 2.1 provides an opportunity to compare paired invaded and uninvaded sites with respect to naturally occurring populations of ants, aphids and parasitoids across a large spatial scale (see

Holway 1998a; Krushelnycky & Gillespie 2008, 2010). This experimental design differs from a manipulation in that the main treatment (i.e., invader presence) is a pre-existing condition, but it nonetheless provides insights into invasion processes that would be impossible to study using small-scale manipulations (Krushelnycky & Gillespie 2010).

Sampling

Sampling was conducted three times per site in early (June), mid (1-14 July) and late (21 July-5 Aug) summer 2012. At every site, 4 or 5 (4.75 trees per visit ± 0.45 SD) arroyo willow trees, each separated by at least 10 m, were searched for 1 to 2 hours per visit for all ant-tended aphid aggregations present on above-ground vegetation. For each aphid morphospecies at each site, we collected three to five individuals from at least one aggregation. At sites where we found more than one aggregation of a particular aphid morphospecies, we sampled from at least three different aggregations. In total we collected aphids from 187 aggregations (660 aphids): 54 aggregations (190 aphids) at sites invaded by the Argentine ant and 133 aggregations (470 aphids) at sites still occupied by native ants. We did not sample aphid aggregations that were not being actively tended by ants. While sampling aphid aggregations, we also estimated the level of tending exhibited by each ant species at aggregations of Chaitophorus nigrae; this aphid species was present at every site and was tended by almost every ant species.

We estimated tending levels by calculating the ratio between the number of tending ants at a particular C. nigrae aggregation and the number of aphids in that aggregation.

Molecular analysis of species identity

We used polymerase chain reaction (PCR) to amplify the cytochrome oxidase I

(COI) barcode region, from which we used Sanger sequencing to determine aphid species identity. Because PCR amplification of a parasitized aphid also allows for the detection of parasitoids (Traugott et al. 2008; Lee & Lee 2012) within 2 to 5 days after parasitoid oviposition (Traugott & Symondson 2008), this method can be used to indicate parasitoid presence and identity and to estimate the level of parasitism at early stages of wasp development. We extracted whole aphids using a DNeasy Blood &

Tissue kit (Qiagen). In this study, 95% of extracted aphid samples successfully

amplified. PCRs for aphids were performed in a 15-µL volume on a BioRAD iCycler gradient thermal cycler with primers (HCO2198 & LCO1490) to amplify the COI locus

(see Folmer et al. 1994). We used 3 µL of template DNA, 1.5 µL 10X PCR buffer

(Teknova), 1.5 µL of 15 mM MgCl2, 1.5 µL of 10X BSA, 0.3 µL of 10 mM dNTPs, 0.6 µL of each 10 µM primer and 0.5 U of Taq DNA Polymerase (Apex). For PCR, we used a modification of the protocol of Folmer et al. (1994) with the following reaction conditions: an initial denaturation step of 94○ C for 3 min, followed by 35 cycles of denaturation at

94○ C for 30 s, annealing at 45○ C for 1 min and extension at 72○ C for 90 s. A final extension step was performed at 72○ C for 10 min. PCR product was sequenced by

Retrogen (San Diego, CA), and COI sequences were aligned to reference sequences of known aphids and parasitoids. Contig assembly was performed in CodonCode using reference sequences from GenBank. Additionally, vouchers of each aphid morphospecies from each site were identified by the Systematic Entomology Laboratory

(Agricultural Research Service, US Department of Agriculture) to confirm that species identified based on DNA barcoding matched the species identity of vouchers. In all cases aphid identifications based on morphology matched those based on DNA barcoding.

Statistical analysis

After molecular analysis allowed us to determine species identities (of aphids and parasitoids) and levels of parasitism, we compared sites with and without the Argentine ant with respect to ant-tended aphid richness, degree of parasitism and level of ant tending. We used paired t-tests to compare invaded and uninvaded sites with respect to aphid species richness (rarefied) and the proportion of aggregations in which we detected at least one parasitoid. No seasonal trends in aphid parasitism were apparent, and thus samples taken throughout the season were pooled within sites prior to analysis.

We used analysis of covariance (ANCOVA) to compare ant:aphid ratios across aphid- tending ant species as a function of aphid aggregation size (covariate). Statistical analyses were conducted in R (R Development Core Team).

We also compared sites with and without the Argentine ant with respect to (1) the interaction network between ants and the aphid species they tended, and (2) the food web that included willows, aphids and parasitoids. Interaction networks (Bascompte &

Jordano 2007) for ants and aphids were constructed based on the observed occurrences of particular two-way interactions at a site. In the analyses of willow-aphid- wasp food webs, we considered how invasion status (invaded, uninvaded) and tending ant richness influenced the number of links, connectance (= observed links / all possible links), and linkage density (= observed links / number of species) among willows, aphids and parasitoid wasps. Connectance is a measure of generalized consumption among food web members that tends to decline with increasing network size (Williams &

Martinez 2000). The upper bound of linkage density is based on web size, but for food webs containing the same number of species, linkage density increases as consumption becomes generalized. We used t-tests and simple linear regressions to test the effects of invasion status and ant species richness on different metrics of food web complexity.

RESULTS

Interactions between ants and aphids

Sites without the Argentine ant supported a greater number of aphid-tending ant species and a greater number ant-tended aphid species compared to sites with the

Argentine ant. The Argentine ant was the only aphid-tending ant species detected at invaded sites, whereas eight native ant species were observed tending aphids at uninvaded sites (Table 2.1) with an average of 2.72 ± 0.7 (mean ± SE) ant species per site (paired t-test: t5 = 6.03, p = 0.0018). The richness of ant-tended aphid species at

uninvaded sites (2.13 ± 0.77 species) also exceeded that at invaded sites (1.00 ± 0.00 species) (Figure 2.2a; paired t-test: t5 = 3.60, p = 0.016). Chaitophorus nigrae was the most common aphid species observed; it was present at all sites, represented 85%

(159/187) of all aphid aggregations sampled, and was the only aphid species found at sites with the Argentine ant (Table 2.1). At sites without the Argentine ant, ant-tended aphids detected included Pterocomma bicolor (9.6% (18/187) of all aggregations), Aphis farinosa (4.8% (9/187) of all aggregations), and Tuberolachnus salignus (< 1% (1/187) of all aggregations) (Table 2.1). Across all sites, ant-tended aphid richness increased with the richness of aphid-tending ant species (Figure 2.2b; simple linear regression: F1,10 =

5.36, p = 0.043).

Reductions in ant and aphid richness at sites with the Argentine ant led to only one possible type of ant-aphid interaction (Figure 2.3a): C. nigrae tended by L. humile.

At uninvaded sites, in contrast, the interaction network between ant-tended aphids and aphid-tending ants included 16 unique ant-aphid pairings (Figure 2.3b). The number of observed pair-wise interactions (4.2 ± 0.40 per site) was only slightly less than the number of possible interactions (6.2 ± 1.1 per site) given the ant and aphid species detected at each site (paired t-test: t5 = 2.58, p = 0.049). The pattern of interactions between ants and aphids exhibited two commonly observed properties of ecological networks: asymmetry and nestedness. Evidence for asymmetry comes from the tendency for the least connected species in the network to interact with the most connected species. Evidence for nestedness comes from the tendency for the most connected species in the network to interact with sets of species that include all (or at least most) of those species that the next most connected species interacts with.

Per capita levels of tending exhibited by the Argentine ant did not differ from those of all native ant species except for Formica moki. Ant:aphid ratios declined with

aphid aggregation size for all ant species (Figure 2.4; ANCOVA: F1,150 = 16.1, p <

0.0001), and there was also a significant effect of tending ant species (Figure 2.4;

ANCOVA: F5,150 = 4.23, p = 0.001) when all ant species were included in the analysis.

However, the effect of tending ant species appeared driven by F. moki. When F. moki is excluded from the analysis, the effect of tending ant species is no longer significant

(ANCOVA: F4,120 = 2.02, p = 0.10). Compared to the other aphid-tending ant species in this study, F. moki workers are larger and exhibited lower levels of tending (i.e., ant:aphid ratios).

Interactions between aphids and parasitoids

We only detected one species of parasitoid wasp (Adialytus salicaphis

(Hymenoptera: Braconidae)) but found that this species parasitized three of the four most common aphid species in our study (Figure 2.3). More than three-quarters (39/51) of the individual parasitoids detected (across all aphid species) were in the early stages of wasp development before it was visually apparent that aphids were parasitized. For C. nigrae, the only aphid species present at all sites, sites with and without the Argentine ant did not differ with respect to the percentage of aggregations in which we detected at least one A. salicaphis (invaded: 16.7% ± 8.5%), uninvaded: 28.7% ± 1.0%; paired t-test: t5 = 0.418, p = 0.694). Although sample sizes are low for some of the aphid species, the percentage of aggregations in which we detected at least one A. salicaphis did not differ

2 among aphid host species (G-test; X 2 = 2.75, p = 0.25). At uninvaded sites, Adialytus salicaphis parasitized 11.1% ± 10% of P. bicolor aggregations, 44.4% ± 33% of A. farinosa aggregations, and 28.7% ± 1.0% of C. nigrae aggregations. Pooling across all aphid species, the proportion of aggregations in which we detected at least one A. salicaphis was independent of per site aphid richness (simple linear regression; F1,10 =

0.35, p = 0.57).

We detected only one species of parasitoid, but the greater number of potential aphid host species at uninvaded sites (Figure 2.3) led to marked differences the willow- aphid-parasitoid food web between invaded and uninvaded sites. Food webs at uninvaded sites, for example, had more links per site (Figure 2.5a; paired t-test: t5 =

2.67, p = 0.04) and lower per site connectance (paired t-test: t5 = 3.43, p = 0.019) compared to food webs at invaded sites. Aphid-tending ant species were excluded from the calculation of food web metrics, but both the number of links (Figure 2.5b) and linkage density (Figure 2.5c) increased with aphid-tending ant richness (simple linear regression: number of links F1,10 = 5.82, p = 0.036; linkage density F1,10 = 5.031, p =

0.049).

DISCUSSION

At sites invaded by the Argentine ant we found that the diversity of ant-tended aphids was lower than that at uninvaded sites and that the structure of the network between aphid-tending ants and ant-tending aphids was homogenized as a result. We detected only one species of parasitoid wasp in our aphid samples, and the proportion of aphid aggregations (C. nigrae only) in which we detected a parasitoid did not differ between invaded and uninvaded sites. The greater number of potential aphid host species at uninvaded sites, however, led to marked differences the structure of the willow-aphid-parasitoid food web. Food webs at uninvaded sites, for example, had more links per site and lower per site connectance compared to food webs at invaded sites.

Our results illustrate the potential for introduced species to reduce the diversity present in multi-species mutualisms and to homogenize the structure of associated interaction networks.

Aphid assemblages at invaded sites were depauperate compared to those at uninvaded sites (Figure 2.2), but the mechanism responsible for this difference remains

unclear. Similarities in the levels of parasitism and ant:aphid tending ratios suggest that, at least with respect to C. nigrae, the Argentine ant provides comparable protective services (at least from parasitoids) compared to those offered by most native ants. The

Argentine ant could be reducing aphid diversity through direct culling of less-preferred aphid species or indirectly by ignoring certain aphid species and leaving them vulnerable to their enemies. Both mechanisms have precedent in other systems (Fischer et al.

2001; Billick et al. 2007; Powell & Silverman 2010). In the absence of tending by ants, aphids typically have smaller aggregation sizes (Fischer et al. 2001), which can decrease the rate that aggregations grow (Yoo & Holway 2011) and increase the risk of local extinction from predation or parasitism (Way 1963; Sakata & Katayama 2001;

Stadler & Dixon 2005; Pareja et al. 2008).

The higher aphid richness observed at sites without the Argentine ant could also be a byproduct of territorial interactions among native ant species. Due to the spatially patchy nature of where aphid aggregations occur (Way 1963), an individual ant colony might only have access to a less-preferred aphid species within its foraging range.

Assuming such colonies will tend these less-preferred aphid species, a mosaic of different aphid species could develop across the host plant canopy as a result of ant territoriality. This mechanism is consistent with the observation that particular aphid aggregations were tended by a single ant species over time and that most native ant species tended a variety of aphid species (Figure 2.3). In invaded areas, however, the

Argentine ant was the only aphid-tending ant species present, and intraspecific territoriality is absent (Thomas et al. 2006). As a consequence, less-preferred aphid species might disappear through a lack of effective protection or through direct consumption by the ants themselves (Fischer et al. 2001; Powell & Silverman 2010).

Three-quarters of the parasitoids detected in this study were identified from aphids that were in the early stages of wasp development; this finding is consistent with that of Hrcek et al. (2011) who argued that visual identification of parasitized aphids can underestimate levels of parasitism revealed through barcoding. Nonetheless, the percentage of aphid aggregations in which we detected at least one parasitoid (17.6% of all aphid aggregations sampled) appears comparable to levels of aphid parasitism reported in other studies (Horn 1989; Barzman & Daane 2001; Daane et al. 2007; Pareja et al. 2008). All of the aphid species as well as the single wasp species detected in this study are believed to be native to Southern California.

Despite finding only one species of parasitoid wasp in our aphid samples, the greater number of potential aphid host species at uninvaded sites led to marked differences the structure of the willow-aphid-parasitoid food web between invaded and uninvaded sites (Figure 2.3). The number of links and linkage density increased with increasing ant richness even though per site ant richness was not included in the calculation of food web statistics. Given that the lifespan of an ant colony typically exceeds that of an aphid aggregation, it seems plausible that local ant richness drives aphid richness and not vice versa, perhaps through the spatial partitioning of the willow canopy that results from territorial interactions among native ant colonies.

Consensus has yet to emerge regarding the extent to which ant invasions affect assemblages of non-ant arthropods (Lach et al. 2010). Definitive information regarding the effects of ant introductions on non-ant arthropods primarily comes from the minority of studies that identify focal taxa to a level appropriate to the questions considered

(Bolger et al. 2008; Krushelnycky & Gillespie 2008). In this study DNA barcoding opened a window onto the cryptic diversity of an assemblage of interacting ant-tended aphids and the wasps that parasitize them. Our results provide a dramatic, and somewhat

unexpected, illustration of the extent to which biological invasions can homogenize diversity within a multi-species mutualism and restructure patterns of interactions among associated species (Figure 2.3). These results suggest that additional research could be directed at better understanding how ant invasions alter diversity within multi-species mutualisms involving ants and honeydew-producing insects. Although numerous studies now document that introduced ants benefit from their interactions with honeydew- producing insects (Helms & Vinson 2002; O’Dowd et al. 2003; Wilder et al. 2011), surprisingly little information has been published documenting how the assemblages of honeydew-producing insects that interact with introduced ants differ from those that interact with native ants (Ness & Bronstein 2004). If ant invasions cause these assemblages to diverge with respect to diversity or species composition, then associated species seem likely to be affected as well. Studies that address how introduced plants and pollinators disrupt interactions between native plants and pollinators (Morales &

Traveset 2009; Traveset et al. 2013; Palladini & Maron 2013), illustrate the clear potential for introduced species to have far reaching species-level and assemblage-level effects on native taxa.

ACKNOWLEDGEMENTS

A. Ashbacher, S. Durfey, N. Gillcrist, R. Graham, and B. Summerhays provided assistance in the field and at the bench. E. Cleland, M. Eubanks, H. Henter, J. Nieh, E.

Wilson, and H. Yoo offered helpful comments on earlier drafts of this manuscript. D.

Miller, G. Miller and M. Buffington at the Systematic Entomology Laboratory (Agricultural

Research Service, US Department of Agriculture) confirmed aphid and wasp identifications. This work was supported by an NSF Graduate Research Fellowship

(KEL), an ARCs Foundation Fellowship (KEL), a Mildred Mathias UC Natural Reserve

System Grant (KEL), and NSF DEB 07-16966 (DAH).

REFERENCES

Addicott JF (1979) A multispecies aphid–ant association: density dependence and species-specific effects. Canadian Journal of Zoology, 57, 558–569.

Andrew R, Bernatchez L, Bonin A et al. (2013) A road map for molecular ecology. Molecular ecology, 22, 2605–2626.

Bach CE (1991) Direct and indirect interactions between ants (Pheidole megacephala), scales (Coccus viridis) and plants (Pluchea indica). Oecologia, 87, 233–239.

Barzman MS, Daane KM (2001) Host-handling behaviours in parasitoids of the black scale: a case for ant-mediated evolution. Journal of Animal Ecology, 70, 237–247.

Bascompte J, Jordano P (2007) Plant-animal mutualistic networks: the architecture of biodiversity. Annual Review of Ecology, Evolution, and Systematics, 38, 567–593.

Bertness MD, Callaway R (1994) Positive interactions in communities. Trends in ecology & evolution, 9, 191–3.

Billick I, Hammer S, Reithel JS, Abbot P (2007) Ant – aphid interactions: are ants friends, enemies, or both? Annals of the Entomological Society of America, 100, 887–892.

Blackman RL, Sorin M, Miyazaki M (2011) Sexual morphs and colour variants of Aphis (formerly Toxoptera) odinae (Hemiptera, Aphididae) in Japan. Zootaxa, 3110, 53– 60.

Blüthgen N, Fiedler K (2004) Preferences for sugars and amino acids and their conditionality in a diverse nectar-feeding ant community. Journal of Animal Ecology, 73, 155–166.

Blüthgen N, Mezger D, Linsenmair KE (2006) Ant-hemipteran trophobioses in a Bornean rainforest – diversity, specificity and monopolisation. Insectes Sociaux, 53, 194– 203.

Bolger DT, Beard KH, Suarez AV, Case TJ (2008) Increased abundance of native and non-native spiders with habitat fragmentation. Diversity and Distributions, 14, 655– 665.

Bristow CM (1984) Differential benefits from ant attendance to two species of Homoptera on New York ironweed. Journal of Animal Ecology, 53, 715–726.

Bruno JF, Stachowicz JJ, Bertness MD (2003) Inclusion of facilitation into ecological theory. Trends in Ecology & Evolution, 18, 119–125.

Chong CS, D’Alberto CF, Thomson LJ, Hoffmann AA (2010) Influence of native ants on arthropod communities in a vineyard. Agricultural and Forest Entomology, 12, 223– 232.

Daane KM, Sime KR, Fallon J, Cooper ML (2007) Impacts of Argentine ants on mealybugs and their natural enemies in California’s coastal vineyards. Ecological Entomology, 32, 583–596.

Derocles SAP, Plantegenest M, Simon J-C, Taberlet P, Le Ralec A (2012) A universal method for the detection and identification of Aphidiinae parasitoids within their aphid hosts. Molecular ecology resources, 12, 634–45.

Finke DL, Snyder WE (2008) Niche partitioning increases resource exploitation by diverse communities. Science, 321, 1488–90.

Fischer MK, Hoffmann KH, Völkl W (2001) Competition for mutualists in an ant- homopteran interaction mediated by hierarchies of ant attendance. Oikos, 92, 531– 541.

Folmer O, Black M, Hoeh W, Lutz R, Vrijenhoek R (1994) DNA primers for amplification of mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates. Molecular Marine Biology and Biotechnology, 3, 294–299.

Green PT, O’Dowd DJ, Abbott KL et al. (2011) Invasional meltdown: invader-invader mutualism facilitates a secondary invasion. Ecology, 92, 1758–68.

Grimaldi D, Engel MS (2005) Evolution of the insects. Cambridge University Press, Cambridge.

Helms KR, Vinson BS (2002) Widespread association of the invasive ant Solenopsis invicta with an invasive mealybug. Ecology, 83, 2425–2438.

Helms KR, Vinson SB (2008) Plant resources and colony growth in an invasive ant: the importance of honeydew-producing Hemiptera in carbohydrate transfer across trophic levels. Environmental Entomology, 37, 487–493.

Henneman ML, Memmott J (2001) Infiltration of a Hawaiian community by introduced biological control agents. Science, 293, 1314–1316.

Hölldobler H, Wilson EO (1990) The Ants. Harvard University Press, Cambridge, Massachusetts.

Holway DA (1998a) Effect of Argentine ant invasions on ground-dwelling arthropods in northern California riparian woodlands. Oecologia, 116, 252–258.

Holway DA (1998b) Factors governing rate of invasion: a natural experiment using Argentine ants. Oecologia, 115, 206–212.

Holway DA (2005) Edge effects of an invasive species across a natural ecological boundary. Biological Conservation, 121, 561–567.

Holway DA, Lach L, Suarez AV, Tsutsui ND, Case TJ (2002) The causes and consequences of ant invasions. Annual Review of Ecology and Systematics, 33, 181–233.

Horn DJ (1989) Secondary parasitism and population dynamics of aphid parasitoids (Hymenoptera : Aphidiidae). Journal of the Kansas Entomological Society, 62, 203– 210.

Hrcek J, Miller SE, Quicke DLJ, Smith MA (2011) Molecular detection of trophic links in a complex insect host-parasitoid food web. Molecular Ecology Resources, 11, 786– 794.

Human KG, Gordon DM (1996) Exploitation and interference competition between the invasive Argentine ant, Linepithema humile, and native ant species. Oecologia, 105, 405–412.

Kaminski LA, Freitas AVL, Oliveira PS (2010) Interaction between mutualisms: ant- tended butterflies exploit enemy-free space provided by ant-treehopper associations. American Naturalist, 176, 322–334.

Kiers ET, Palmer TM, Ives AR et al. (2010) Mutualisms in a changing world: an evolutionary perspective. Ecology Letters, 13, 1459–1474.

Krushelnycky PD, Gillespie RG (2008) Compositional and functional stability of arthropod communities in the face of ant invasions. Ecological Applications, 18, 1547–62.

Krushelnycky PD, Gillespie RG (2010) Sampling across space and time to validate natural experiments: an example with ant invasions in Hawaii. Biological Invasions, 12, 643–655.

Lach L (2003) Invasive ants: unwanted partners in ant-plant interactions? Annals of the Missouri Botanical Garden, 90, 91–108.

Lach L, Tillberg CV, Suarez AV (2010) Contrasting effects of an invasive ant on a native and an invasive plant. Biological Invasions, 12, 3123–3133.

LeBrun EG, Tillberg CV, Suarez AV et al. (2007) An experimental study of competition between fire ants and Argentine ants in their native range. Ecology, 88, 63–75.

Lee W, Lee S (2012) Unexpected problem in aphid DNA barcoding by universal primers. Entomological Science, 15, 121–126.

Menke SB, Fisher RN, Jetz W, Holway DA (2007) Biotic and abiotic controls of Argentine ant invasion success at local and landscape scales. Ecology, 88, 3164–3173.

Menke SB, Holway DA (2006) Abiotic factors control invasion by Argentine ants at the community scale. Journal of Animal Ecology, 75, 368–376.

Mitrovich MJ, Matsuda T, Pease KH, Fisher RN (2010) Ants as a measure of effectiveness of habitat conservation planning in southern California. Conservation Biology, 24, 1239–1248.

Morales CL, Traveset A (2009) A meta-analysis of impacts of alien vs. native plants on pollinator visitation and reproductive success of co-flowering native plants. Ecology letters, 12, 716–28.

Ness JH, Bronstein JL (2004) The effects of invasive ants on prospective ant mutualists. Biological Invasions, 6, 445–461.

Nevo E, Coll M (2001) Effect of nitrogen fertilization on Aphis gossypii (Homoptera : Aphididae): variation in size , color, and reproduction. Journal of Economic Entomology, 94, 27–32.

O’Dowd DJ, Green PT, Lake PS (2003) Invasional “meltdown” on an oceanic island. Ecology Letters, 6, 812–817.

Oliver TH, Leather SR, Cook JM (2008) Macroevolutionary patterns in the origin of mutualisms involving ants. Journal of Evolutionary Biology, 21, 1597–1608.

Palladini JD, Maron JL (2013) Indirect competition for pollinators is weak compared to direct resource competition: pollination and performance in the face of an invader. Oecologia, 172, 1061–9.

Pareja M, Brown VK, Powell W (2008) Aggregation of parasitism risk in an aphid- parasitoid system: Effects of plant patch size and aphid density. Basic and Applied Ecology, 9, 701–708.

Powell BE, Silverman J (2010) Impact of Linepithema humile and Tapinoma sessile (Hymenoptera: Formicidae) on three natural enemies of Aphis gossypii (Hemiptera: Aphididae). Biological Control, 54, 285–291.

Richardson ML, Lagos DM, Mitchell RF, Hartman GL, Voegtlin DJ (2011) Life history and morphological plasticity of the soybean aphid, Aphis glycines. Entomologia Experimentalis et Applicata, 140, 139–145.

Sakata H, Katayama N (2001) Ant defence system: A mechanism organizing individual responses into efficient collective behavior. Ecological Research, 16, 395–403.

Santos AMC, Besnard G, Quicke DLJ (2011) Applying DNA barcoding for the study of geographical variation in host-parasitoid interactions. Molecular Ecology Resources, 11, 46–59.

Sipura M (2002) Contrasting effects of ants on the herbivory and growth of two willow species. Ecology, 83, 2680–2690.

Smith RA, Mooney KA, Agrawal AA (2008) Coexistence of three specialist aphids on common milkweed, Asclepias syriaca. Ecology, 89, 2187–96.

Stadler B, Dixon AFG (2005) Ecology and evolution of aphid-ant interactions. Annual Review of Ecology, Evolution, and Systematics, 36, 345–372.

Stanton ML (2003) Interacting guilds: moving beyond the pairwise perspective on mutualisms. The American naturalist, 162, S10–23.

Styrsky JD, Eubanks MD (2007) Ecological consequences of interactions between ants and honeydew-producing insects. Proceedings of the Royal Society B-Biological Sciences, 274, 151–64.

Suarez A V., Bolger DT, Case TJ (1998) Effects of fragmentation and invasion on native ant communities in coastal southern California. Ecology, 79, 2041–2056.

Thomas ML, Payne-Makrisâ CM, Suarez AV, Tsutsui ND, Holway DA (2006) When supercolonies collide: Territorial aggression in an invasive and unicolonial social insect. Molecular Ecology, 15, 4303–4315.

Traugott M, Bell JR, Broad GR et al. (2008) Endoparasitism in cereal aphids: molecular analysis of a whole parasitoid community. Molecular Ecology, 17, 3928–3938.

Traugott M, Symondson WOC (2008) Molecular analysis of predation on parasitized hosts. Bulletin of Entomological Research, 98, 223–31.

Traveset A, Heleno R, Chamorro S et al. (2013) Invaders of pollination networks in the Galápagos Islands : emergence of novel communities. Proceedings of the Royal Society B-Biological Sciences, 280, 20123040.

Traveset A, Richardson DM (2006) Biological invasions as disruptors of plant reproductive mutualisms. Trends in Ecology & Evolution, 21, 208–216.

Tremper BS (1977) Distribution of the Argentine ant, Iridomyrmex humilis Mayr, in relation to certain native ants of California: ecological, physiological and behavioural aspects. Ph.D. Thesis: University of California, Berkeley.

Ward PS (1987) Distribution of the introduced Argentine ant (Iridomyrmex humilis) in natural habitats of the lower Sacramento valley and its effects on the indigenous ant fauna. Hilgardia, 55, 1–16.

Way MJ (1963) Mutualism between ants and honeydew-producing Homoptera. Annual Review of Entomology, 8, 307–344.

Weathersbee AA, Shufran KA, Panchal TD et al. (2004) Detection and differentiation of parasitoids (Hymenoptera : Aphidiidae and Aphelinidae) of the brown citrus aphid (Homoptera : Aphididae): Species-specific polymerase chain reaction amplification of 18S rDNA. Annals of the Entomological Society of America, 97, 286–292.

Wilder SM, Holway DA, Suarez A V, LeBrun EG, Eubanks MD (2011) Intercontinental differences in resource use reveal the importance of mutualisms in fire ant invasions. Proceedings of the National Academy of Sciences (USA), 108, 20639– 20644.

Williams RJ, Martinez ND (2000) Simple rules yield complex food webs. Nature, 404, 180–183.

Wimp GM, Whitham TG (2001) Biodiversity consequences of predation and host plant hybridization on an aphid-ant mutualism. Ecology, 82, 440–452.

Yoo HJS, Holway DA (2011) Context-dependence in an ant-aphid mutualism: direct effects of tending intensity on aphid performance. Ecological Entomology, 36, 450– 458.

Yoo HJS, Kizner MC, Holway DA (2013) Ecological effects of multi-species, ant – hemipteran mutualisms in citrus. Ecological Entomology, 38, 505–514.

Figure 2.1. Map of California with inset showing the locations of riparian corridor sites in

San Diego and Riverside Cos. Each watershed included a pair of sites: one site with the

Argentine ant (closed circle) and one site without (open circle) the Argentine ant.

Figure 2.2. (a) Mean (± SE) species richness (rarefied) of ant-tended aphids at paired invaded or uninvaded sites (n = 6) (* p < 0.02). (b) Ant-tended aphid richness versus aphid-tending ant richness at sites with (closed circles) or without (open circles) the

Argentine ant. Identical data points are jittered to show overlapping points.

Figure 2.3. Interaction networks for sites with (a) and without (b) the Argentine ant.

Lines show interactions between a parasitoid wasp (Adialytus salicaphis) and its aphid hosts and interactions between ant-tended aphids and the ants that tend them. The number of lines connecting each pair of species depicts the number of sites at which that particular interaction was observed. The width of each species bar in (a) and (b) indicates the number of sites (out of six) where that species was present.

1 C. hespera T. sessile L.humile

0.8 L. occidentale F. mccooki F. moki 0.6

0.4

Ratio of ants to aphids to ants of Ratio 0.2

0 1 10 100 1000

Aphid aggregation size

Figure 2.4. Levels of ant tending (ant:aphid ratios) versus the size of Chaitophorus nigrae aggregations for the Argentine ant and for the five most common species of native ants.

Figure 2.5. (a) Mean (± SE) number of links in willow-aphid-parasitoid food webs at sites with or without the Argentine ant (* p < 0.05). (b) The number of food web links versus species richness of aphid-tending ants. (c) Linkage density versus species richness of aphid-tending ants. In (b) and (c) identical data points are jittered to show overlapping points; closed circles indicate sites with the Argentine ant, and open circles indicate sites without the Argentine ant.

Ant-aphid mutualism increases floral visitation by ants and reduces plant

reproduction via decreased pollinator visitation

ABSTRACT

For plant species that depend on animal-mediated pollination, reproduction hinges on adequate access to pollinators. Even in the presence of intact pollinator assemblages, negative interactions among floral visitors can compromise pollination services. Ants, for example, often visit flowers and discourage visitation by other insects, but usually do not perform pollination themselves. Effects on plant reproduction that result from this type of interaction may be compounded by factors that increase floral visitation by ants. Mutualisms between ants and honeydew-producing insects alter the activity and local abundance of ants on plants, but the extent to which these interactions increase floral visitation by ants and in turn disrupt pollination services remains incompletely understood. We manipulated the abundance of cotton aphids (Aphis gossypii) on cotton (Gossypium hirsutum) over three years to test how the Argentine ant

(Linepithema humile), which collects aphid honeydew and floral nectar, affects pollinator visitation and seed production. Increasing aphid abundance increased ant abundance on cotton leaves, and floral visitation by ants was positively related to ant abundance on leaves. The duration of visits by putative pollinators (mostly bees) declined with both increasing aphid abundance and increasing ant floral visitation. Two measures of seed production (seed set and seed mass) declined with increasing aphid abundance and decreasing bee visitation. Aphid herbivory alone did not affect plant reproduction; seed production was independent of aphid abundance on plants that were manually cross- pollinated under greenhouse conditions. Our results illustrate that mutualisms between ants and honeydew-producing insects can enhance levels of floral visitation by ants and

47 48

in turn disrupt pollination services enough to have measurable effects on plant reproduction.

INTRODUCTION

More than 85% of flowering plant species rely on animal pollination for successful reproduction (Delaplane & Mayer 2000; Potts et al. 2010; Ollerton et al. 2011). These services are becoming increasingly jeopardized as a consequence of declining pollinator populations (Winfree et al. 2009; Potts et al. 2010), land-use change and habitat fragmentation (Aguilar et al. 2006), and the establishment of non-native plants (Morales

& Traveset 2009). The interdependence of plants and their pollinators is highlighted by studies that show parallel declines between plants and the pollinators on which they rely

(Biesmeijer et al. 2006; Pauw & Hawkins 2011). Pollination services may also be compromised when floral visitors negatively interact with one another.

Pollen limitation resulting from negative interactions among floral visitors can affect seed production, which will negatively affect plant fitness when plant populations are seed limited (Irwin et al. 2001). Negative effects can include predation on pollinators by wasps (Dukas 2005; Wilson & Holway 2010), spiders (Heiling et al. 2003; Suttle

2003; Dukas & Morse 2013) and other predators (Romero et al. 2011). Non-consumptive effects such as pollinator harassment (Ness 2006; Lach 2007) or predator avoidance by pollinators (Dukas 2001; Abbott & Dukas 2009) can also affect plant reproduction

(Gonçalves et al. 2008). In addition to direct interactions like predation, indirect interactions among floral visitors (e.g., exploitative competition) can discourage visitation by pollinators and negatively impact plant reproduction (Roubik 1982; Irwin et al. 2001).

Negative interactions among floral visitors may be especially common for interactions that involve ants, which commonly visit flowers in a wide variety of environments but rarely provide effective pollination services (Hölldobler & Wilson 1990).

Floral visitation by ants can result in damage to floral structures (Palmer et al. 2010), reduced pollen viability (Beattie et al. 1984; Cushman & Beattie 1991; Galen 1999), pollinator harassment (Ness 2006), learned avoidance by pollinators in response to olfactory cues produced by ants (Ballantyne & Willmer 2012), and exploitation or alteration of floral resources (Irwin et al. 2001; Lach 2008a; de Vega & Herrera 2013).

Harassment by ants may deter ineffective floral visitors in favor of effective pollinators

(Gonzálvez et al. 2013) or potentially enhance the male component of plant fitness by increasing pollen transfer (Traveset & Richardson 2006), but perhaps more often ants interfere with legitimate pollinators (e.g., Tsuji et al. 2004; Ness 2006). Floral visitation by ants is known to reduce visitation by potential pollinators (see Tsuji et al. 2004; Ness

2006; Lach 2007, 2008b), but surprisingly few studies investigate in detail how floral visitation by ants affects plant reproduction (Ashman & King 2005; Galen & Geib 2008).

Effects of floral visitation by ants may be amplified when ants interact with honeydew-producing insects, which typically alter the behavior, activity and local abundance of ants on plants. Honeydew is a highly attractive resource to many ant species (Way 1963; Völkl et al. 1999; Engel et al. 2001), and ants that tend honeydew- producing insects will commonly consume other insects, often to the indirect benefit of the host plant (Styrsky & Eubanks 2007). Although honeydew-producing insects could potentially distract ants from flowers in a manner similar to that proposed for extrafloral nectar (Wagner & Kay 2002), enhanced floral visitation by ants may be more common

(Lach 2007). Despite that interaction between ants and honeydew-producing insects are widespread and conspicuous, few experimental studies have investigated how interactions between ants and honeydew-producing insects affect floral visitation by ants, influence how pollinators are affected by such visitation, and in turn how changes in pollinator behavior affect pollen limitation and seed set.

In this study we manipulated an ant-aphid mutualism in upland cotton

(Gossypium hirsutum) to determine whether or not this interaction affected floral visitation by ants, pollinator behavior, and plant reproductive performance. We focus on the Argentine ant (Linepithema humile), which commonly forages on floral nectar (Lach

2007, 2008ab), and the cotton aphid (Aphis gossypii), which is a honeydew-producing hemipteran that readily forms food-for-protection mutualisms with ants (Kaplan &

Eubanks 2005; Powell & Silverman 2010; Styrsky & Eubanks 2010). Previous work has demonstrated that interactions between the cotton aphid and the red imported fire ant

(Solenopsis invicta) increased cotton reproduction because fire ants consumed non- aphid herbivores that were more damaging to cotton compared to aphids (Styrsky &

Eubanks 2010). Solenopsis invicta, however, rarely forages in cotton flowers (MD

Eubanks, pers. comm.). Switching the identity of the mutualist partner (i.e., from the red imported fire ant to the Argentine ant), provides a unique opportunity to examine how ant-aphid interactions affect plant reproduction via floral visitation by ants. Negative interactions between the Argentine ant and potential pollinators may affect cotton reproduction because pollinator visitation increases seed set in cotton, and ineffective pollination can result in abnormal fruit development and reduced yield (Delaplane &

Mayer 2000). Moreover, bee pollination (Waller et al. 1985), especially by honey bees

(Rhodes 2002) and bumblebees (Berger et al. 1988; Van Deynze et al. 2005), can enhance reproduction in cotton up to 38 percent (Delaplane & Mayer 2000).

In a three-year field study, we manipulated the abundance of cotton aphids on cotton plants to test how the mutualism between Argentine ants and cotton aphids (1) affects floral visitation by ants, (2) the duration of visits by pollinators, and (3) seed production. To our knowledge, this is the first study to manipulate mutualisms between ants and honeydew-producing insects to test how these interactions affect pollinator

activity and plant reproduction. Given the ubiquity of interactions between ants and honeydew-producing insects (Way 1963; Hölldobler & Wilson 1990; Davidson et al.

2003) and the ability of these interactions to affect the composition and structure of food webs on plants (Wimp & Whitham 2001; Styrsky & Eubanks 2007), the implications of our study broadly extend to a wide variety of systems. More generally, these experiments provide a novel test of the capacity for mutualisms to structure species interactions external to the mutualism itself (Bruno et al. 2003).

METHODS

Study system

Over three field seasons (2011-2013), we manipulated cotton aphid abundance on cotton plants in the presence of the Argentine ant at the UC San Diego Biology Field

Station (32° 53' 12" N, 117° 13' 48" W). The specific focus of the experimental work differed across years (see descriptions of separate experiments below), but all plants were treated as follows. Individual plants were fertilized at the start of each field season

(5:1:2 NPK ratio; see Tewolde et al. 2009) and then provided with 10 liters of water weekly. To test the ecological role of ant-aphid mutualisms on plant reproduction, we performed aphid-removal experiments in each year of the study. Manipulations of this kind are considered to be the appropriate type of experiment to examine the broader ecological consequences of mutualisms between ants and honeydew-producing insects

(Wimp & Whitham 2001; Styrsky & Eubanks 2007). In each year we randomly assigned seedlings to one of two aphid treatments: high-density and low-density aphid groups.

Seedlings of each treatment type were interspersed evenly throughout experimental plots in each year, and aphid treatments were maintained throughout each season.

Plants in the high-aphid density treatment were inoculated with cotton aphids at the start of the season; a single inoculation was usually sufficient to maintain high aphid

abundances on plants throughout the season. Aphids naturally colonized plants in the low-density aphid treatment, and plants were checked weekly (or more often as needed) for aphids, which were manually removed.

In all years, plants in the low-density aphid treatment had densities of aphids that were, on average, less than 18 percent of those in the high-density group. Time- averaged counts of aphid abundance differed across treatment groups (Figure 3.1; two- way ANOVA: aphid group, F1,174 = 4.08, p = 0.045; survey year, F1,174 = 2.37, p = 0.12; interaction, F1,174 = 2.47, p = 0.12). Although aphid removal reduced aphid numbers in the low-aphid density group, the number of aphid individuals present on plants in both treatment groups varied over the course of each season. This variation was in part driven by exponential growth of aphid populations and inter-plant dispersal, which is typical of this species (Lombaert et al. 2006). Recognizing that grouping aphid abundance data across replicates into categories can result in misleading outcomes

(Helms & Hunter 2005), we took a more statistically powerful approach that employed regression (Cottingham et al. 2005). For most analyses, we thus treated aphid abundance as a continuous independent variable but indicate on figures the experimental group (low-density or high-density) to which each replicate belonged.

Unless indicated otherwise, individual plants (or the insects on them) represent the experimental unit in all analyses. Analyses were typically generalized linear models

(GLM) that we used to determine (1) the effect of aphids on ant foraging behavior, (2) the effect of ant-aphid mutualisms on pollinator activity, and (3) the effect of ant-aphid mutualisms on cotton reproduction. We used a Poisson link function for these GLMs because the data showed a strong positive skew and is appropriate distribution for count data (Quinn & Keough 2002).

Effects of aphids on ant foraging behavior

In each field season, we determined how the mutualism between cotton aphids and the Argentine ant affected ant foraging behavior, especially with respect to whether ants visited flowers. First, we estimated the number of aphids and Argentine ant workers per leaf and per flower. Table 3.1 provides details about the specifics of experiments conducted in different years. For each year separately, we used generalized linear models to compare time-averaged means of (1) ant abundance versus aphid abundance on leaves, and (2) ant abundance on flowers versus ant abundance on leaves. In the first analysis, we treat ant abundance as a dependent variable because it was unmanipulated (unlike aphid abundance) but recognize that the presence of ants can influence the abundance of aphids. In the second analysis, we consider ant abundance in flowers to be a dependent variable because the number of ants present on all of the leaves within a plant always greatly exceeded the number of ants in flowers. Sample sizes were 40, 24, and 31 plants in the high-density treatment and 40, 26, and 39 plants in the low-density treatment, in 2011, 2012, and 2013, respectively.

Effects of ant-aphid mutualisms on pollinator activity

To quantify the effect of the mutualism between cotton aphids and the Argentine ant on pollinator activity (via floral visitation by the Argentine ant), we conducted 294 pollinator surveys (spanning 2,092 total minutes of observation) from 2011 to 2013 (see

Table 3.1 for experimental details). For each plant during each survey, we recorded the number of Argentine ant workers present in each flower and the number and identity of floral visitors over 2 min (2011) or 10 min (2012-2013) intervals. During these pollinator surveys, we found that some floral visitors stayed in flowers for periods exceeding 10 min. To gather additional data on visits that exceeded our survey window, we also watched plants for extended periods of 2 to 3 hours (only in year 2013) and recorded the

entire duration of each pollinator visit, the identity of the visitor, and the number of ants in the flower during the visit.

Using the pollinator data described above, we tested the extent to which floral visitation by the Argentine ant was related to the duration of visits by other insects. First, we used GLM to compare the duration of all observed visits (by non-ants) at flowers (all years, N = 146 individual visits) as a function of ant abundance in flowers, aphid density treatment, and survey year. Sample sizes for this analysis were 47, 67, and 32 plants in the high-aphid density treatment and 59, 73, and 79 plants in the low-aphid density treatment, in 2011, 2012, and 2013, respectively. We also separately compared visitation by honey bees on plants in the high-aphid and low-aphid density groups. In this latter analysis, we also used GLM to examine the effect of aphid density treatment and ant abundance in flowers on the duration of honey bee visits; we included survey year as an additional fixed effect. Sample sizes for this GLM were 15 visits by honey bees in the high-aphid density group and 23 visits in the low-aphid density group. Finally, we used contingency table analysis to compare the types of floral visitors (bees vs. flies) that commonly occurred on cotton plants in the two aphid density treatments.

Effects of ant-aphid mutualisms on plant reproduction

To determine the roles of the ant-aphid mutualism and pollinator visitation on host plant reproduction, we evaluated how seed production (number of seeds, mass of individual seeds) changes with (1) aphid density and (2) pollinator visit duration. At the end of each year’s field season, we harvested and dried bolls (cotton fruit), counted the number of seeds, and estimated the mass of individual seeds produced by each plant; see Table 3.1 for sample sizes. Additionally, we determined plant reproductive allocation

(i.e., the number of flowers produced by plants) as a function of aphid density treatments. In 2013, we quantified the effects of pollinator activity on plant reproduction

by individually marking flowers and then testing how observed pollinator visitation

(cumulative duration of all pollinator visits) affected seed production. Because cotton flowers only remain open for one day, reproduction of each individual boll provides an estimate of the cross-pollen received by that flower. We used GLMs to test for the role of aphid density and pollinator activity on seed production. In a separate analysis we tested the effects of aphid density on seed production for plants that received supplemental cross pollination under greenhouse conditions. In this experiment, all pollination was performed manually (i.e., insect pollinators were absent). Cotton is self-compatible but exhibits increased seed set when cross-pollinated by insects (Delaplane & Mayer 2000).

We used a t- test to compare seed production as a function of aphid density (low-density vs. high-density). Eight plants were equally divided across each aphid treatment group.

RESULTS

Effects of aphids on ant foraging behavior

Argentine ant abundance increased with increasing cotton aphid abundance on cotton leaves, and the numbers of ants present in flowers were positively related to the number of ants on leaves. In each year of the study, per leaf ant abundance increased with per leaf aphid abundance (Figure 3.2a-c; GLM 2011: Wald z = 3.8, df = 79, p =

0.0001; GLM 2012: Wald z = 16.5, df = 49, p < 0.00001; GLM 2013: Wald z = 6.74, df =

50, p < 0.00001). Removal of the data point in the upper right hand corner of Fig. 1b did not qualitatively affect the results of this analysis. Increased ant abundance on leaves was also positively related to the number of ants visiting cotton flowers (Figure 3.2d-f;

GLM 2011: Wald z = 3.04, df = 34, p = 0.001; GLM 2012: Wald z = 3.3, df = 49, p =

0.001; GLM 2013: Wald z = 7.76, df = 69, p < 0.00001). Up to 3% of ants present on plants were found foraging within flowers (GLM 2011: b = 0.031 ± .033 (95% CI; GLM

2012: b = 0.003 ± .001 (95% CI); GLM 2013: b = 0.034 ± .019 (95% CI)). Ants collected

floral nectar while in flowers; ants with empty gasters descended into nectar wells and left flowers with full gasters.

Effects of ant-aphid mutualisms on pollinator activity

The majority of non-ant floral visitors observed in cotton flowers were bees, and the duration of bee visits declined with an increasing number of Argentine ants in cotton flowers. During pollinator surveys, we observed eleven insect taxa visiting cotton flowers; 72.6% of these visitors were bees. Honey bees were the most common floral visitor and made up 26% (38/146) of all floral visits and 35.8% (38/106) of all visits by bees. Other bee visitors included Agapostemon texanus (18.4% of all floral visits),

Melissodes tessellata (13.9% of all floral visits), Lasioglossum incompletum (12.5% of all floral visits), and Halictus tripartitus (5.1% of all floral visits). We also observed two fly morphospecies (18.4% of all floral visits; one of these morphospecies was a hoverfly

(Diptera: Syrphidae), whereas the other dipteran remained unidentified), 10 visits by

Diabrotica balteata (Coleoptera: Chrysomelidae), 4 visits by thrips and one visit by

Strymon melinus (Lepidoptera: Lycaenidae).

The duration of visits by non-ant floral visitors declined with the number of ants present in a flower (Figure 3.3a; GLM: Wald z = 7.29, df = 377, p < 0.000001). Moreover, this analysis revealed that floral visitors spent more time in flowers on plants in the low- density group compared to plants in the high-density aphid group (GLM: Wald z = 16.4, df = 377, p < 0.000001). The overall trend summarized in Figure 3.3a was also evident when we restricted the analysis to just honey bee visitation (Figure 3.3b). Honey bees reduced the duration of their floral visits by 56.3% in flowers occupied by ants compared to flowers without ants. As the number of ants in flowers increased, the duration of honeybee visits declined (Figure 3.3b; GLM: Wald z = 5.45, df = 35, p < 0.00001).

Honey bee visits were also longer in the low-density group compared to the high-density

aphid group (GLM: Wald z = 3.69, df = 35, p = 0.0002). Bees visited plants in the low- density aphid treatment almost twice as often as they visited plants in the high-density treatment (70 vs. 36 visits), whereas plants in the high-density treatment had three times as many visits by flies (mostly syrphids) compared to plants in the low-density group (19

2 vs. 6 visits; X 1 = 14.7, p = 0.00013).

Effects of ant-aphid mutualisms on plant reproduction

The reproductive performance of cotton declined with increasing numbers of aphids on individual plants and increased with the duration of floral visits by pollinators.

In each year of the study, seed set was negatively related to aphid abundance (Figure

3.4, Table 3.2). In Figure 3.4b, removal of the data point in the lower right hand corner of the plot did not qualitatively affect the outcome of the analysis. In two of three years, the mass of individual seeds also declined with increasing aphid abundance (Table 3.2). The lack of a significant effect between individual seed mass and aphid abundance in 2012 might have resulted from low levels of pollinator activity in 2012 compared to 2011 and

2013. Increasing visitation by pollinators enhanced production of cotton seeds. Seed mass increased with the cumulative duration of pollinator visits to individual flowers

(GLM: Wald z = 11.8, df = 72, p < 0.00001). When this analysis was restricted to the duration of visits only by honey bees, we likewise found that longer visits corresponded to greater individual seed mass (GLM: Wald z = 5.62, df = 72, p < 0.00001). Although seed set decreased with aphid density, the number of flowers produced per plant was independent of aphid density (2011: t48 = 1.82, p = 0.08; 2012: t173 = 0.90, p = 0.37).

Plants in the high-density treatment produced less seed (Figure 3.4), but this reduction appears to result from compromised pollination services rather than from the costs of herbivory. Hand cross-pollinated plants set similar numbers of seeds in the low- density treatment compared to plants in the high-density group (29.30 vs. 29.25 seeds; t8

= 0.034, p = 0.97). Moreover, the mass of seeds produced by plants in the low-density group was similar to mass of seeds set by plants in the high-density group when flowers were hand cross-pollinated (49.7 mg vs. 57 mg; t8 = 1.53, p = 0.16).

DISCUSSION

The mutualism between Argentine ants and cotton aphids increased levels of floral visitation by ants and in turn, disrupted pollination services enough to have measurable effects on plant reproduction. In each year of the study, increasing aphid abundance increased ant abundance on cotton leaves, and floral visitation by ants was positively related to per leaf ant abundance. The duration of visits by non-ant floral visitors (mostly bees) declined with both increasing aphid abundance and increasing ant floral visitation. Greater numbers of ants in flowers reduced the duration of visits by all non-ant floral visitors and by honey bees alone. Two measures of seed production (seed set and seed mass) declined with increasing aphid abundance and decreasing bee visitation. Under controlled greenhouse conditions, seed production for plants receiving supplemented non-self pollen was independent of aphid abundance.

Cotton aphid abundance strongly increased the abundance of ant on leaves, consistent with previous work showing that Argentine ants are strongly attracted to honeydew (Way 1963; Holway et al. 2002; Ness & Bronstein 2004). Heightened ant traffic on cotton leaves also resulted in a fraction of ants (up to 3% of those present on plants) foraging within flowers for nectar. Some ant species tend hemipterans without foraging in flowers and thus do not disrupt floral visitation (Styrsky & Eubanks 2010), perhaps because of repellent chemicals present in pollen, nectar or petals (Ness 2006;

Nicklen & Wagner 2006; Willmer et al. 2009; Junker et al. 2011). However, these chemicals do not affect some ant species or are not present in high enough concentrations to entirely deter ant floral foraging (Chamberlain & Holland 2008). The

modification of ant floral foraging by hemipterans has broad implications because food- for-protection mutualisms between ants and hemipterans are ubiquitous and ecologically important interactions (Way 1963; Hölldobler & Wilson 1990; Davidson et al. 2003;

Styrsky & Eubanks 2007).

The duration of visits by bees in cotton flowers declined with increasing aphid abundance and increasing floral visitation by ants. Lach (2008b) found that honey bees shortened the duration of their floral visits in the presence of the Argentine ant and left flowers following direct encounters with the Argentine ant. We observed the former phenomenon (Figure 3.3), but further research is required to determine the mechanism responsible for reduced duration of bee visits. The abundance of aphids on cotton appeared to produce an interaction visitation by syrphids versus bees. Bees and syrphids preferentially visited low-aphid density and high-aphid density plants, respectively. Both honey bees and bumble bees can detect and avoid flowers previously visited by syrphid flies (Reader et al. 2005). Syrphid larvae are well-known aphid predators, and we commonly observed syrphids copulating and ovipositing on or near flowers of plants in high-aphid density treatments. Syrphids can be effective pollinators

(Branquart & Hemptinne 2000), but in our system adult syrphids typically did not contact the sexual organs of cotton flowers during visits.

Ant-hemipteran mutualisms may commonly enhance ant floral visitation and reduce pollinator visitation (e.g., Lach 2007), but this is the first study we are aware of to demonstrate that these interactions reduce plant reproduction. Floral visitation by ants may only affect plant reproduction under conditions of pollen limitation, barring direct damage to floral tissue. In this study, hand cross-pollinated plants set similar quantities of seeds in high-density and low-density aphid treatment groups. This result indicates that pollen limitation brought about by the interaction between aphids and ants, not by

aphid herbivory itself, is the driving force behind reduced seed production. An alternative explanation for the decreased duration of pollinator visits under high-aphid densities could entail reduced reproductive allocation to flowers. Such an effect could cause pollinators to reduce the length of their visits to cotton flowers. Two lines of evidence suggest that this alternative hypothesis fails to provide a strong explanation for our results. First, under greenhouse conditions, aphid herbivory did not affect seed production; this finding suggests that aphids do not reduce plant allocation to reproduction at least under controlled circumstances. Second, flower production per plant was independent of aphid density under field conditions in two different years. This latter result further suggests that herbivory by aphids does not affect reproductive allocation in cotton.

Although a growing number of studies show that ant-hemipteran mutualisms enhance plant growth and reproduction (reviewed in Styrsky & Eubanks 2007), our research illustrates that these positive indirect effects depend on the foraging behavior of the focal ant species. Positive effects can manifest themselves when ants consume herbivores on plants (Messina 1981; Skinner & Whittaker 1981; Floate & Whitham 1994;

Kaplan & Eubanks 2005); however, if honeydew-producing insects encourage ants to forage in flowers, then the potential exists for plant reproduction to suffer as a result of negative interactions between ants and other floral visitors. The outcome of ant- hemipteran mutualisms as outlined in this paper may depend on an ant colony’s nutritional needs, which can influence preferences for carbohydrate-rich resources (e.g., honeydew) versus protein-rich resources (Ness et al. 2009; Petry et al. 2012). Plant floral defense compounds may modify the response of ants to aphid honeydew and the net effect of ant-hemipteran mutualisms. Certain ant lineages may be prone to respond to hemipterans by visiting flowers as opposed to consuming herbivores; some

introduced ants visit flowers even in the absence of hemipterans (Lach 2005). If floral visitation by ants discourages visitation by putative pollinators, especially bees (Lach

2007, 2008a), then the continued spread of invasive ants will likely only cause further pollinator disruption (Lach 2003).

ACKNOWLEDGEMENTS

We gratefully acknowledge A. Ashbacher, C. K. Dao, J. Wong, R. Graham, S.

Durfey, A. Schochet, A. Aguilar, V. Hernandez, K. Bergeson, B. White, K. Kwok, A.

Valdez, T.G. Ho, T. Smith, J. Scioli and A. Shum for their assistance in the field. We thank K. J. Hung for his assistance providing identification of bees to species. This work was supported by a NSF graduate research fellowship (to KEL) as well as grants from the USDA (NRI-CGP 2006-35302-17255) and NSF (DEB 07-16966) to D. A. Holway.

REFERENCES

Abbott KR, Dukas R (2009) Honeybees consider flower danger in their waggle dance. Animal Behaviour, 78, 633–635.

Aguilar R, Ashworth L, Galetto L, Aizen MA (2006) Plant reproductive susceptibility to habitat fragmentation: review and synthesis through a meta-analysis. Ecology letters, 9, 968–80.

Ashman T-L, King EA (2005) Flower-visiting ants mutualists or antagonists? A study in a gynodioecious wild strawberry. American Journal of Botany, 92, 891–895.

Ballantyne G, Willmer P (2012) Floral visitors and ant scent marks: noticed but not used? Ecological Entomology, 37, 402–409.

Beattie AJ, Turnbull C, Knox RB, Williams EG (1984) Ant inhibition of pollen function: a possible reason why ant pollination is rare. American Journal of Botany, 71, 421– 426.

Berger LA, Vaissiere BE, Moffett JO, Merritt SJ (1988) Bombus spp. (Hymenoptera: Apidae) as pollinators of male-sterile upland cotton on the Texas high plains. Environmental Entomology, 17, 789–794.

Biesmeijer JC, Roberts SPM, Reemer M et al. (2006) Parallel declines in pollinators and insect-pollinated plants in Britain and the Netherlands. Science, 313, 351–354.

Branquart E, Hemptinne J (2000) Selectivity in the exploitation of floral resources by hoverflies (Diptera: Syrphinae). Ecography, 23, 732–742.

Bruno JF, Stachowicz JJ, Bertness MD (2003) Inclusion of facilitation into ecological theory. Trends in Ecology & Evolution, 18, 119–125.

Chamberlain SA, Holland JN (2008) Density-mediated, context-dependent consumer- resource interactions between ants and extrafloral nectar plants. Ecology, 89, 1364–1374.

Cottingham KL, Lennon JT, Brown BL (2005) Knowing when to draw the line: designing more informative ecological experiments. Frontiers in Ecology and the Environment, 3, 145–152.

Cushman JH, Beattie AJ (1991) Mutualisms: assessing the benefits to hosts and visitors. Trends in Ecology & Evolution, 6, 193–5.

Davidson DW, Cook SC, Snelling RR, Chua TH (2003) Explaining the abundance of ants in lowland tropical rainforest canopies. Science, 300, 969–972.

Delaplane KS, Mayer DF (2000) Crop pollination by bees. University Press, Cambridge, London.

Van Deynze AE, Sundstrom FJ, Bradford KJ (2005) Pollen-mediated gene flow in California cotton depends on pollinator activity. Crop Science, 45, 1565–1570.

Dukas R (2001) Effects of perceived danger on flower choice by bees. Ecology Letters, 4, 327–333.

Dukas R (2005) Bumble bee predators reduce pollinator density and plant fitness. Ecology, 86, 1401–1406.

Dukas R, Morse DH (2013) Crab spiders affect flower visitation by bees. Oikos, 101, 157–163.

Engel V, Fischer MK, Wackers FL, Völkl W (2001) Interactions between extrafloral nectaries, aphids and ants: Are there competition effects between plant and homopteran sugar sources? Oecologia, 129, 577–584.

Floate KD, Whitham TG (1994) Aphid-ant interaction reduces Chrysomelid herbivory in a cottonwood hybrid zone. Oecologia, 97, 215–221.

Galen C (1999) Flowers and enemies: Predation by nectar-thieving ants in relation to variation in floral form of an alpine wildflower, Polemonium viscosum. Oikos, 85, 426–434.

Galen C, Geib JC (2008) Density-dependent effects of ants on selection for bumble bee pollination in Polemonium viscosum. Ecology, 88, 1202–1209.

Gonçalves JM, Omena PM, Souza JC, Romero GQ (2008) Trait-mediated effects on flowers: artifical spiders decieve pollinators and decrease plant fitness. Ecology, 89, 2407–2413.

Gonzálvez FG, Santamaría L, Corlett RT, Rodríguez-Gironés MA (2013) Flowers attract weaver ants that deter less effective pollinators (K Whitney, Ed,). Journal of Ecology, 101, 78–85.

Heiling AM, Herberstein ME, Chittka L (2003) Pollinator attraction - Crab-spiders manipulate flower signals. Nature, 421, 334.

Helms SE, Hunter MD (2005) Variation in plant quality and the population dynamics of herbivores: there is nothing average about aphids. Oecologia, 145, 197–204.

Hölldobler H, Wilson EO (1990) The Ants. Harvard University Press, Cambridge, Massachusetts.

Holway DA, Lach L, Suarez A V, Tsutsui ND, Case TJ (2002) The causes and consequences of ant invasions. Annual Review of Ecology and Systematics, 33, 181–233.

Irwin RE, Brody AK, Waser NM (2001) The impact of floral larceny on individuals, populations, and communities. Oecologia, 129, 161–168.

Junker RR, Daehler CC, Doetterl S, Keller A, Blüthgen N (2011) Hawaiian ant-flower networks: nectar-thieving ants prefer undefended native over introduced plants with floral defenses. Ecological Monographs, 81, 295–311.

Kaplan I, Eubanks MD (2005) Aphids alter the community-wide impact of fire ants. Ecology, 86, 1640–1649.

Lach L (2003) Invasive ants: unwanted partners in ant-plant interactions? Annals of the Missouri Botanical Garden, 90, 91–108.

Lach L (2005) Interference and exploitation competition of three nectar-thieving invasive ant species. Insectes Sociaux, 52, 257–262.

Lach L (2007) A mutualism with a native membracid facilitates pollinator displacement by Argentine ants. Ecology, 88, 1994–2004.

Lach L (2008a) Floral visitation patterns of two invasive ant species and their effects on other hymenopteran visitors. Ecological Entomology, 33, 155–160.

Lach L (2008b) Argentine ants displace floral arthropods in a biodiversity hotspot. Diversity and Distributions, 14, 281–290.

Lombaert E, Boll R, Lapchin L (2006) Dispersal strategies of phytophagous insects at a local scale: adaptive potential of aphids in an agricultural environment. Bmc Evolutionary Biology, 6, 75.

Messina FJ (1981) Plant-protection as a consequence of an ant-membracid mutualism - interactions on goldenrod (Solidago sp). Ecology, 62, 1433–1440.

Morales CL, Traveset A (2009) A meta-analysis of impacts of alien vs. native plants on pollinator visitation and reproductive success of co-flowering native plants. Ecology letters, 12, 716–28.

Ness JH (2006) A mutualism’s indirect costs: the most aggressive plant bodyguards also deter pollinators. Oikos, 113, 506–514.

Ness JH, Bronstein JL (2004) The effects of invasive ants on prospective ant mutualists. Biological Invasions, 6, 445–461.

Ness JH, Morris WF, Bronstein JL (2009) For ant-protected plants, the best defense is a hungry offense. Ecology, 90, 2823–31.

Nicklen EF, Wagner D (2006) Conflict resolution in an ant-plant interaction: Acacia constricta traits reduce ant costs to reproduction. Oecologia, 148, 81–87.

Ollerton J, Winfree R, Tarrant S (2011) How many flowering plants are pollinated by animals? Oikos, 120, 321–326.

Palmer TM, Doak DF, Stanton ML et al. (2010) Synergy of multiple partners, including freeloaders, increases host fitness in a multispecies mutualism. Proceedings of the National Academy of Sciences (USA), 107, 17234–17239.

Pauw A, Hawkins J a (2011) Reconstruction of historical pollination rates reveals linked declines of pollinators and plants. Oikos, 120, 344–349.

Petry WK, Perry KI, Mooney KA (2012) Influence of macronutrient imbalance on native ant foraging and interspecific interactions in the field. Ecological Entomology, 37, 175–183.

Potts SG, Biesmeijer JC, Kremen C et al. (2010) Global pollinator declines: trends, impacts and drivers. Trends in Ecology & Evolution, 25, 345–353.

Powell BE, Silverman J (2010) Population growth of Aphis gossypii and Myzus persicae (Hemiptera: Aphididae) in the presence of Linepithema humile and Tapinoma sessile (Hymenoptera: Formicidae). Environmental Entomology, 39, 1492–1499.

Quinn GP, Keough MJ (2002) Experimental design and data analysis for biologists. Cambridge University Press.

Reader T, MacLeod I, Elliott PT, Robinson OJ, Manica A (2005) Inter-order interactions between flower-visiting insects: foraging bees avoid flowers previously visited by hoverflies. Journal of Insect Behavior, 18, 51–57.

Rhodes J (2002) Cotton pollination by honey bees. Australian Journal of Experimental Agriculture, 42, 513–518.

Romero GQ, Antiqueira PAP, Koricheva J (2011) A meta-analysis of predation risk effects on pollinator behaviour. Plos One, 6, e20689.

Roubik DW (1982) The ecological impact of nectar-robbing bees and pollinating hummingbirds on a tropical shrub. Ecology, 63, 354–360.

Skinner GJ, Whittaker JB (1981) An experimental investigation of inter-relationships between the wood-ant (Formica rufa) and some tree-canopy herbivores. Journal of Animal EcologyS, 50, 313–326.

Styrsky JD, Eubanks MD (2007) Ecological consequences of interactions between ants and honeydew-producing insects. Proceedings of the Royal Society B-Biological Sciences, 274, 151–64.

Styrsky JD, Eubanks MD (2010) A facultative mutualism between aphids and an invasive ant increases plant reproduction. Ecological Entomology, 35, 190–199.

Suttle KB (2003) Pollinators as mediators of top-down effects on plants. Ecology letters, 6, 688–694.

Tewolde H, Shankle MW, Adeli A, Sistani KR, Rowe DE (2009) Macronutrient concentration in plant parts of cotton fertilized with broiler litter in a marginal upland soil. Soil and Tillage Research, 105, 1–11.

Traveset A, Richardson DM (2006) Biological invasions as disruptors of plant reproductive mutualisms. Trends in Ecology & Evolution, 21, 208–216.

Tsuji K, Hasyim A, Nakamura K (2004) Asian weaver ants, Oecophylla smaragdina, and their repelling of pollinators. Ecological Research, 19, 669–673.

De Vega C, Herrera CM (2013) Microorganisms transported by ants induce changes in floral nectar composition of an ant-pollinated plant. American Journal of Botany, 100, 792–800.

Völkl W, Woodring J, Fischer MK, Lorenz MW, Hoffmann KH (1999) Ant-aphid mutualisms: The impact of honeydew production and honeydew sugar composition on ant preferences. Oecologia, 118, 483–491.

Wagner DL, Kay AD (2002) Do extrafloral nectaries distract ants from visiting flowers? An experimental test of an overlooked hypothesis. Evolutionary Ecology Research, 4, 293–305.

Waller GD, Vaissiere BE, Moffett JO, Martin JH (1985) Comparison of carpenter bees (Xylocopa varipuncta Patton) (Hymenoptera: Anthophoridae) and honey bees (Apis mellifera L.) (Hymenoptera: Apidae) as pollinators of male-sterile cotton in cages. Journal of Economic Entomology, 78, 558–561.

Way MJ (1963) Mutualism between ants and honeydew-producing Homoptera. Annual Review of Entomology, 8, 307–344.

Willmer P, Nuttman C V, Raine NE et al. (2009) Floral volatiles controlling ant behaviour. Functional Ecology, 23, 888–900.

Wilson EE, Holway DA (2010) Multiple mechanisms underlie displacement of solitary Hawaiian Hymenoptera by an invasive social wasp. Ecology, 91, 3294–3302.

Wimp GM, Whitham TG (2001) Biodiversity consequences of predation and host plant hybridization on an aphid-ant mutualism. Ecology, 82, 440–452.

Winfree R, Aguilar R, Vasquez DP, Lebuhn G, Aizen MA (2009) A meta-analysis of bees’ responses to anthropogenic disturbance. Ecology, 90, 2068–2076.

Table 3.2. Results of generalized linear models analyzing the effect of aphid abundance on plant fitness. Parameter estimates (b), standard error (SE), the test statistic (Wald z) and significance (p) values are reported.

Response variable Year b SE z p

Number of Seeds

2011 -0.0010 0.0001 7.30 < 0.001

2012 -0.0004 0.0001 6.09 < 0.001

2013 -0.0001 0.0000 4.66 < 0.001

Mass per seed (mg)

2011 -0.0004 0.0001 3.11 0.001

2012 0.0000 0.0003 0.13 0.89

2013 -0.0002 0.0000 10.00 < 0.001

1 1000 400 4000 - a: 2011 b: 2012 c: 2013 750 300 3000

500 200 2000

250 100 1000 Number ofaphids plant 0 0 0 High Low High Low High Low

Aphid Density Treatment

Figure 3.1. Estimated cotton aphid abundance on cotton in high-aphid density and low- aphid density experimental treatment groups.

5 10

a: 2011 -

d: 2011 1 - 4 8

3 6

2 4

1 2

Numberofants leaf Numberofants flower 0 0 0 25 50 75 100 0 0.3 0.6 0.9 1.2 1.5

10 5

1 -

b: 2012 e: 2012 1 - 8 4

6 3

4 2

2 1

Numberofants leaf Numberofants flower 0 0 0 100 200 300 400 0 2 4 6 8 10 5 10

c: 2013 f: 2013

1 - 4 8

3 6

2 4

1 2

Numberofants leaf Numberofants flower 0 0 0 100 200 300 400 0 1 2 3 4 5 -1 -1 Number of aphids leaf Number of ants leaf

Figure 3.2. The per leaf abundance of Argentine ants and cotton aphids on cotton in (a)

2011, (b) 2012, and (c) 2013; the abundance of Argentine ants on cotton flowers versus the number of ants on leaves in (d) 2011, (e) 2012, and (f) 2013. Data points in each graph (a-f) indicate plants in the high-density aphid treatment (closed circles) and those in the low-density aphid treatment (open circles).

250

200

150

100

50 Duration of visit floral by (s) visitor

0 0 5 10 15 20 125

100

25 Duration of visit honey by bees(s)

0 0 1 2 3 4 5 6 Number of ants in flowers

Figure 3.3. Floral visit duration by (a) all non-ant floral visitors and (b) honey bees as a function of the number of ants foraging in flowers throughout a visit. Data points in each graph indicate individual floral visits to flowers on plants in the high-density aphid treatment (closed circles) and in the low-density aphid treatment (open circles). Data from all years are combined.

Figure 3.4. The number of seeds produced by cotton plants in (a) 2011, (b) 2012 and (c)

2013 as a function of cotton aphid abundance on plants. Data points in each graph indicate plant replicates in the high-density treatment (closed circles) and replicates in the low-density treatment (open circles).

An experimental test of how mutualisms between the Argentine ant and cotton

aphid structure arthropod food webs in cotton

ABSTRACT

Although mutualisms have the potential to influence biotic interactions external to the mutualism itself, few experimental studies address how reciprocally positive interactions between species structure the food webs in which these interacting players are embedded. Food-for-protection mutualisms between ants and honeydew-producing insects are common and widespread interactions that alter the activity and local abundance of ants on plants. Changes in ant foraging behavior that result from the presence of honeydew-producers often reduce the diversity and abundance of plant- feeding insects and their enemies. Here, we test how the food-for-protection mutualism between the cotton aphid and the Argentine ant alters the abundance, diversity and composition of the arthropod assemblages on cotton. We manipulated aphid density

(low and high) on cotton plants growing under field conditions. Compared to plants in the low-density aphid treatment, plants in the high-density aphid treatment had higher numbers of ants on cotton and strongly differed in food web composition. Plants in the high-density aphid treatment supported higher abundances of aphid-specialist predators and non-aphid phloem-feeders. In the low-density aphid treatment, generalist predators had more links between themselves and prey items, whereas generalist predators consumed fewer prey species in the high-density aphid treatment. These differences perhaps reflect a diminished need for alternate prey given the abundance of cotton aphids as a food source. Behavioral assays indicate that the effectiveness of aphid protection by ants differed between different types of aphid enemies (aphid specialist versus generalist predator) and may explain observed numerical increases by specialist predators and dietary shifts by generalist predators. Leaf-chewers were not strongly

affected by aphids, and leaf damage was independent of aphid-density treatment; however, overall lepidopteran abundance was low. These results illustrate the potential importance of mutualisms in shaping the structure of arthropod food webs.

INTRODUCTION

Reciprocally positive species interactions are an important and common type of biotic interaction (Boucher et al. 1982; Bronstein 1994; Bertness & Callaway 1994;

Bruno et al. 2003) with the potential to influence diversity (Palmer et al. 2003; Lach

2007; Cavieres & Badano 2009), food web structure (Gross 2008; Crowley & Cox 2011) and ecosystem services (Ness 2006; Baiser et al. 2011). Positive interactions also help to explain patterns of invasibility (Bulleri et al. 2008) and the success of some species invasions (Wilder et al. 2011).

Of the diversity of mutualistic interactions that occur, food-for-protection mutualisms seem likely to affect other trophic levels. Mutualists that offer protective services often interact broadly with other taxa that are present. Ants, for example, often protect honeydew-producing insects from their enemies in exchange for sugar-rich honeydew (Way 1963; Stadler & Dixon 2005) but also consume or scare off leaf- chewing herbivores (Messina 1981; Ito & Higashi 1991; Wimp & Whitham 2001; Sipura

2002; Styrsky & Eubanks 2010) with indirect benefits to host plant growth and reproduction (Styrsky & Eubanks 2007). The broader ecological effects of mutualisms between ants and honeydew-producers can also extend to the diversity or abundance of pollinating insects (Lach 2007), generalist predators (Wimp & Whitham 2001; Kaplan &

Eubanks 2002, 2005), and aphid specialists (Wimp & Whitham 2001; Sanders & van

Veen 2010).

In this study we build on a body of research that has investigated the broader ecological consequences of mutualisms between aphid-tending ants and cotton aphids

on cotton (Kaplan & Eubanks 2002, 2005; Styrsky & Eubanks 2010). Interactions between cotton aphids and red imported fire ants, for example, lead to reduced generalist predator abundance (Kaplan & Eubanks 2002) and modifications in food web composition (Kaplan & Eubanks 2005). Leaf-chewing herbivores are also negatively affected by the fire ant-cotton aphid mutualism (Styrsky & Eubanks 2010).

Here, we further examine how food-for-protection mutualisms between ants and cotton aphids may affect the composition and structure of arthropod food webs on cotton. Using a similar experimental design to Styrsky and Eubanks (2010), we examined the effects of ant-aphid mutualisms in which the Argentine ant protected cotton aphids on cotton. Using this framework we asked the following questions. (1) Do ant-aphid mutualisms affect the diversity, abundance or composition of arthropods external to the mutualism itself? (2) Do ant-aphid mutualisms affect food web structure and do different feeding guilds respond differently to the mutualism?

METHODS

In this study we manipulated the abundance of cotton aphids (Aphis gossypii) and Argentine ants (Linepithema humile) on cotton at the UC San Diego Biology Field

Station (32° 53' 12" N, 117° 13' 48" W) during the 2011 growing season. We planted 80 cotton seedlings in May, fertilized each plant at the start of the experiment (5:1:2 NPK ratio; see Tewolde et al. 2009), and provided them with 10 liters of water weekly. We then randomly assigned seedlings to one of two aphid treatments: a low-density group and a high-density group. Seedlings of each treatment type were interspersed evenly throughout experimental plots, and aphid treatments were maintained through

September. Plants in the high-density aphid treatment were inoculated with cotton aphids at the start of the season; a single, initial inoculation was usually sufficient to keep aphids present on plants all season long. Aphids naturally colonized plants in the

low-density aphid treatment, and plants were checked weekly (or more often as needed) for aphids, which were manually removed. Aphid removals successfully reduced, but did not entirely eliminate aphids; aphid population growth and dispersal contributed to the presence of small numbers of aphids on plants in the low-density treatment.

Nonetheless, time-averaged aphid densities were 86% lower, on average, in the low- density aphid group compared to the high-density aphid group (Figure 4.1a; ANOVA

F1,78 = 5.24, p = 0.025). Ant abundances on the low-density aphid group were 41% of those in the high-density aphid treatment (Figure 4.1b; ANOVA F1,78 = 7.96, p = 0.006).

Arthropod surveys

To determine the abundance, richness and morphospecies composition of arthropod assemblages on cotton we surveyed plants in each aphid-density group for arthropods on a weekly basis from May to September. Voucher specimens of each observed morphospecies were collected and identified (see Supplementary Table 1 for details). Based on observed feeding patterns and literature records for each species (or morphospecies), we categorized each taxon as a sap-sucking herbivore, a leaf-chewing herbivore, a generalist arthropod predator, or an aphid specialist.

We compared differences in the abundance, richness and morphospecies composition of arthropod assemblages present on plants across the two aphid-density treatments. In all analyses, aphid-density treatment is the categorical independent variable, and the abundance of cotton aphids and Argentine ants themselves are not included as part of the response variable. We used one-way ANOVA to compare arthropod richness between plants in the two aphid-density treatment groups. We also calculated time-averaged abundance values for each arthropod taxon on each plant, summed these values for each plant, and then used one-way ANOVA to compare summed abundances between plants in the two aphid-density treatment groups. We

used non-metric multidimensional scaling (NMDS) and permutational multivariate analysis of variance (PERMANOVA) to test for differences in species composition between plants in the two aphid-density treatment groups (Anderson 2001; McArdle &

Anderson 2001). We first performed NMDS using Bray-Curtis dissimilarities calculated from time-averaged abundance data for each plant. Next, we performed a PERMANOVA

(with 1,000 permutations) on the whole arthropod assemblage and then separate

PERMANOVAs (each with 1,000 permutations) for each of the four feeding guilds.

NMDS and PERMANOVA were run in R using the vegan package (Oksanen et al.

2012).

To further assess how food webs differed between aphid-density treatments, we examined food web structure within arthropod assemblages. We compared the number of links, connectance (actual number of links per number of possible links), and linkage density (actual number of links per number of species) between plants in each treatment group. On each plant, a link was considered to occur between any consumer morphospecies and any resource morphospecies if (1) both morphospecies occurred together on the plant, and (2) the consumer was observed to feed on the resource morphospecies or literature records confirmed that the consumer could consume that type of prey. We used one-way ANOVAs to test if plants in each aphid-density treatment group differed with respect to the number of links, connectance, and linkage density. In a second series of analyses, we compared estimated diet breadth of generalist predators and specialist predators in each aphid treatment. Diet breadth for each group was estimated by counting the number of links between predators and their prey.

To determine the role of ant activity on different types of predatory arthropods, we measured behavioral responses exhibited by generalist and specialist predators in response to aphid tending by the Argentine ant. We used field-caught Hippodamia

convergens (Coleoptera: Coccinellidae) larvae and syrphid (Diptera: Syrphidae) larvae as the generalist and specialist predators, respectively. Paired trials were conducted in which we placed a predator on a leaf with only cotton aphids or on a leaf with both

Argentine ants and cotton aphids (47.4 ± 7.3 SE aphids per aggregation). We recorded how long individual predators stayed on single leaves and also how long they stayed within 2 cm of aphid prey before moving off of leaves. If a predator stayed on a leaf for more than 5 min, then we ended the trial. Linear regressions were used to relate ant abundance to predator behavior (i.e., how long they spent on leaves).

To test the hypothesis that food-for-protection mutualisms between ants and aphids would negatively affect the abundance of leaf-chewing herbivores and the leaf damage caused by their feeding, we collected additional data on the abundanceof leaf chewers and the leaf area they consumed. We measured changes in leaf area on plants in each aphid-density treatment by marking and photographing one leaf per plant against a white background with a reference coin. We resampled leaves the following week, and the difference in area was calculated using ImageJ. This procedure was repeated every two weeks throughout the peak of arthropod activity in the field (n = 7 leaves sampled per plant).

RESULTS

Overall, ant-aphid mutualisms had little effect on the richness or abundance of arthropods on cotton. Morphospecies richness did not differ between aphid-density treatments (high density: 23.6 ± 0.91 (mean ± 1 SE); low density: 25.7 ± 1.07; one-way

ANOVA: F1,78 = 2.29, p = 0.14). Although plants in the high-density aphid treatment did numerically host more arthropods compared to plants in the low-density aphid treatment

(one-way ANOVA: F1,78 = 5.39, p = 0.02), this difference reflects the higher abundance of

Aphis fabae on plants in the high-density group (high density: 996.4 ± 323; low density:

154.6 ± 28; one-way ANOVA: F1,78 = 6.73, p = 0.01). When A. fabae is excluded from the abundance comparison, arthropod abundance did not differ across aphid-density treatments (high density: 361.1 ± 38 individual arthropods; low density: 395.3 ± 45 individual arthropods; one-way ANOVA: F1,78 = 0.34, p = 0.56).

Although the ant-aphid mutualism did not change the richness or abundance of the arthropods on cotton (the abundance of A. fabae aside), the composition of the arthropod assemblages did differ between aphid-density treatments. Aphid density, for example, modified assemblage composition when all arthropods (except for ants and cotton aphids) are included in the analysis (Figure 4.2a; PERMANOVA, r2 = 0.038, stress = 0.07, F1,78 = 3.12, p = 0.006). Further analyses revealed that species composition of both sap-sucking herbivores (PERMANOVA; r2 = 0.03, stress = 0.009,

2 F1,78 = 3.15, p = 0.014) and aphid specialist predators (PERMANOVA; r = 0.04, stress =

0.03, F1,78 = 3.28, p = 0.003) differed between treatment groups. This may reflect the fact that aphid specialists (particularly parasitoids) were numerically more common on plants that had aphids present (Figure 4.2b; one-way ANOVA: F1,78 =7.555, p = 0.007).

2 Assemblages of generalist predators (PERMANOVA: r = 0.01, stress = 0.07, F1,78 =

0.96, p = 0.48) and leaf-chewing herbivores (PERMANOVA: r2 = 0.01, stress = 0.17,

F1,69 = 0.35, p = 0.88), however, did not differ between aphid-density treatments.

Changes in species composition brought about by ant-aphid mutualisms are reflected by changes in food web structure. Food webs on plants in the high-density aphid treatment had fewer links (Figure 4.3a; one-way ANOVA: F1,78 = 4.098, p = 0.046) and reduced linkage density (Figure 4.3b; one-way ANOVA: F1,78 = 4.089, p = 0.046) compared to those on plants in the low-density aphid treatment. Connectance did not differ between aphid-density groups (one-way ANOVA: F1,78 = 0.04, p = 0.84). When we further examined the number of links present within subsets of the complete food web,

we found that generalist predators interacted with fewer taxa in the high-density aphid treatment compared to the low-density aphid treatment (Figure 4.3c; one-way ANOVA:

F1,78 = 4.49, p = 0.037). In contrast, aphid specialists did not differ in diet breadth (i.e., interacted with similar numbers of taxa) between aphid-density treatments (Figure 4.3d; one-way ANOVA: F1,78 = 0.19, p = 0.66).

Behavioral assays demonstrated that ant-aphid mutualisms had contrasting effects on predator behavior in the presence of aphid prey. Generalist predators spent less time on leaves (Figure 4.4a; paired t25 = 4.43, p = 0.0002) and less time in proximity to aphid prey (Figure 4.4a; paired t25 = 4.04, p = 0.0004) in the presence of ants and aphids compared to when only aphids were present. Generalist predators also decreased the amount of time spent on individual leaves (linear regression: F1,50 = 23.8, p < 0.0001) and near aphid prey (linear regression: F1,50 = 12.1, p = 0.001) with increasing numbers of ants present. In contrast, specialist predators did not modify time spent on individual leaves (Figure 4.4b; paired t8 = 0.61, p = 0.56) or near aphid prey

(Figure 4.4b; paired t8 = 0.41, p = 0.69) based on ant presence. Moreover, the number of ants present on cotton plants did not affect the amount of time syrphid larvae spent on leaves (linear regression: F1,16 = 1.05, p = 0.32) or near prey (linear regression: F1,16 =

0.40, p = 0.54).

Ant-aphid mutualisms did not affect the abundance of leaf-chewing herbivores or alter the amount of damage they caused to cotton leaves. Lepidopteran larvae did not differ in abundance between aphid-density treatments (Figure 4.5a) and average leaf damage (as measured by weekly changes in leaf area) did not differ between aphid- density treatments (Figure 4.5b; one-way ANOVA: F1,78 = 0.06, p = 0.81).

DISCUSSION

Food-for-protection mutualism between the Argentine ant and cotton aphid modified the composition and structure of arthropod food webs on cotton. Specialist predators numerically responded to increased aphid abundance, whereas generalist predators did not differ in their abundance between aphid-density treatments. Behavioral trials indicated that increases in specialist abundance might result from ineffective deterrence by ants. Aphid-tending ants consistently drove off a generalist predator, a result that perhaps explains changes in diet breadth observed among generalist predators. We found no evidence that the Argentine ant deterred leaf chewers from consuming leaves, but ant-aphid mutualisms were associated with shifts in the composition of the phloem-feeder assemblage.

Ant-aphid mutualism did not change the abundance or composition of generalist predator assemblages. While generalist predators can be more abundant and diverse in the presence of ant-hemipteran mutualisms (Marín & Perfecto 2013), in the present system generalist predators did not respond numerically to ant-aphid mutualisms.

Instead, we found that abundant aphid populations appeared to subsidize predators and to reduce the diversity of insects consumed by predators. Behavioral trials confirmed that generalist predators likely reduce their consumption of aphids in situations where per-capita aphid defense is high. When aphids are abundant (e.g., in the high-density aphid treatment), the number of available tending ants is proportionally lower (i.e., per aphid), and generalist predators may increase their aphid consumption. If ant- hemipteran mutualisms increase generalist predator diversity (Marín & Perfecto 2013) and diverse predators consume more herbivores (Cardinale et al. 2003; Snyder et al.

2006), then ant-hemipteran mutualisms can indirectly reduce herbivory without requiring ants to directly consume herbivores. However, if aphids subsidize predator diets without

increasing generalist diversity (as was our finding), then the benefits provided by ant- aphid mutualisms may be limited. The ability of ant-hemipteran mutualisms to increase predator richness may rely, in part, on aggression by ants during tending. Perhaps other ants that offer superior protection limit the ability of any one predator to monopolize increased food resources (i.e., hemipterans), which increases generalist predator diversity.

Aphid specialist predators responded numerically to ant-aphid mutualism and were more commonly observed when their prey were abundant, a phenomenon documented in other studies (Völkl 1994; Wimp & Whitham 2001; Stadler 2002).

Behavioral data provide a mechanism for this result; specialist predators interacted with aphid prey similarly regardless of ant presence or abundance. When ants are competent hemipteran defenders, specialist consumers are more abundant where ants are absent

(Herbert & Horn 2008). If ants are less proficient at aphid defense, however, then aphid specialists will undergo numerical responses to abundant prey resources. Other studies indicate that consumption of aphids by both generalists and specialists can additively

(Snyder & Ives 2003; Schmidt et al. 2003) or synergistically (Cardinale et al. 2003) suppress aphid populations in the field. Alternatively, generalist predators can disrupt specialist parasitoids from consuming untended aphids (Snyder & Ives 2001). However, our results suggest that ant-aphid mutualisms may limit the extent to which generalists interfere with specialists.

Other studies routinely find that when ants engage in food-for-protection mutualisms, there are often negative effects on the abundance of herbivores aside from tended honeydew producers (Messina 1981; Ito & Higashi 1991; Kaplan & Eubanks

2005; Styrsky & Eubanks 2010) and that these reductions in abundance in turn reduce damage caused by leaf chewers (Karhu 1998; Rudgers et al. 2003). When ants do not

directly interact with herbivores (e.g., leaf miners and plant gallers), the mutualism either has no effect on herbivores (Grinath et al. 2012) or indirectly benefits the herbivore

(Savage & Peterson 2007; Yoo et al. 2013). The primary effectiveness of ants as an herbivore deterrent appears strongly predicated on their predatory abilities (possession of a functional stinger) and adequate access to herbivores (i.e., exposed leaf-chewers vs. protected leaf-miners). In this study, lepidopterans and other leaf-chewers were distributed equally across both treatment groups, indicating that Argentine ants did not consume or remove chewing-herbivores. Moreover, the Argentine ant did not affect leaf consumption by chewing herbivores. Although the overall abundance of lepidopterans and other chewers appeared relatively low, this study at least suggests that the

Argentine ant does not effectively remove leaf-chewing herbivores.

Phloem-feeders were more abundant in treatments with high-densities of aphids.

This result indicates that non-aphid phloem feeders (e.g., thrips, mealybugs, whiteflies) were likely not strongly competing with A. gossypii for plant resources, because phloem feeders can decline in the presence of competition from untended aphids (Ando et al.

2011) or ant-tended honeydew-producing insects (Smith et al. 2008; Grinath et al. 2012).

Previous work has demonstrated that other phloem-feeding insects can benefit from co- occurring with ant-tended species (Martinez-Ferrer et al. 2003; Yoo et al. 2013) in a byproduct mutualism, even though non-aphid phloem-feeders are not ant-tended. In part, ant identity seems may play some role in moderating the degree of indirect benefit received by non-tended species. However, ant activity is also affected by ant nutritional needs (Ness et al. 2009), which may also modify whether phloem-feeding insects receive any indirect benefits.

ACKNOWLEDGEMENTS

We gratefully acknowledge A. Ashbacher, C.K. Dao, J. Wong, R. Graham, S.

Durfey, A. Schochet, A. Aguilar, V. Hernandez, K. Bergeson, B. White, K. Kwok, A.

Valdez and T.G. Ho for their assistance in the field. This work was supported by a NSF graduate research fellowship (KEL) as well as grants from the USDA (NRI-CGP 2006-

35302-17255 to DAH) and NSF (DEB 07-16966 to DAH).

REFERENCES

Anderson MJ (2001) A new method for non-parametric multivariate analysis of variance. Austral Ecology, 26, 32–46.

Ando Y, Utsumi S, Ohgushi T (2011) Community-wide impact of an exotic aphid on introduced tall goldenrod. Ecological Entomology, 36, 643–653.

Baiser B, Ardeshiri RS, Ellison AM (2011) Species richness and trophic diversity increase decomposition in a co-evolved food web. Plos One, 6.

Bertness MD, Callaway R (1994) Positive interactions in communities. Trends in ecology & evolution, 9, 191–3.

Boucher DH, James S, Keeler KH (1982) The ecology of mutualism. Annual Review of Ecology and Systematics, 13, 315–347.

Bronstein JL (1994) Conditional outcomes in mutualistic interactions. Trends in Ecology & Evolution, 9, 214–217.

Bruno JF, Stachowicz JJ, Bertness MD (2003) Inclusion of facilitation into ecological theory. Trends in Ecology & Evolution, 18, 119–125.

Bulleri F, Bruno JF, Benedetti-Cecchi L (2008) Beyond competition: incorporating positive interactions between species to predict ecosystem invasibility. PLoS biology, 6, e162.

Cardinale BJ, Harvey CT, Gross K, Ives AR (2003) Biodiversity and biocontrol: emergent impacts of a multi-enemy assemblage on pest suppression and crop yield in an agroecosystem. Ecology Letters, 6, 857–865.

Cavieres L a., Badano EI (2009) Do facilitative interactions increase species richness at the entire community level? Journal of Ecology, 97, 1181–1191.

Crowley PH, Cox JJ (2011) Intraguild mutualism. Trends in Ecology & Evolution, 26, 627–633.

Grinath JB, Inouye BD, Underwood N, Billick I (2012) The indirect consequences of a mutualism: comparing positive and negative components of the net interaction between honeydew-tending ants and host plants. The Journal of animal ecology, 81, 494–502.

Gross K (2008) Positive interactions among competitors can produce species-rich communities. Ecology letters, 11, 929–36.

Ito F, Higashi S (1991) An indirect mutualism between oaks and wood ants via aphids. Journal of Animal Ecology, 60, 463–470.

Kaplan I, Eubanks MD (2002) Disruption of cotton aphid (Homoptera: Aphididae) - Natural enemy dynamics by red imported fire ants (Hymenoptera: Formicidae). Environmental Entomology, 31, 1175–1183.

Kaplan I, Eubanks MD (2005) Aphids alter the community-wide impact of fire ants. Ecology, 86, 1640–1649.

Karhu KJ (1998) Effects of ant exclusion during outbreaks of a defoliator and a sap- sucker on birch. Ecological Entomology, 23, 185–194.

Lach L (2007) A mutualism with a native membracid facilitates pollinator displacement by Argentine ants. Ecology, 88, 1994–2004.

Marín L, Perfecto I (2013) Spider diversity in coffee agroecosystems: the influence of agricultural intensification and aggressive ants. Environmental Entomology, 42, 204–213.

Martinez-Ferrer MT, Grafton-Cardwell EE, Shorey HH (2003) Disruption of parasitism of the California red scale (Homoptera: Diaspididae) by three ant species (Hymenoptera: Formicidae). Biological Control, 26, 279–286.

McArdle BH, Anderson MJ (2001) Fitting multivariate models to community data: A comment on distance-based redundancy analysis. Ecology, 82, 290–297.

Messina FJ (1981) Plant-protection as a consequence of an ant-membracid mutualism - interactions on goldenrod (Solidago sp). Ecology, 62, 1433–1440.

Ness JH (2006) A mutualism’s indirect costs: the most aggressive plant bodyguards also deter pollinators. Oikos, 113, 506–514.

Ness JH, Morris WF, Bronstein JL (2009) For ant-protected plants, the best defense is a hungry offense. Ecology, 90, 2823–31.

Oksanen J, Blanchet FG, Kindt R et al. (2012) Ordination methods, diversity analysis and other functions for community and vegetation ecologists.

Palmer TM, Stanton ML, Young TP (2003) Competition and coexistence: exploring mechanisms that restrict and maintain diversity within mutualist guilds. The American naturalist, 162, S63–79.

Rudgers JA, Hodgen JG, White JW (2003) Behavioral mechanisms underlie an ant-plant mutualism. Oecologia, 135, 51–59.

Sanders D, van Veen FJF (2010) The impact of an ant-aphid mutualism on the functional composition of the secondary parasitoid community. Ecological Entomology, 35, 704–710.

Savage AM, Peterson MA (2007) Mutualism in a community context: the positive feedback between an ant-aphid mutualism and a gall-making midge. Oecologia, 151, 280–291.

Schmidt MH, Lauer A, Purtauf T et al. (2003) Relative importance of predators and parasitoids for cereal aphid control. Proceedings of the Royal Society B-Biological Sciences, 270, 1905–9.

Sipura M (2002) Contrasting effects of ants on the herbivory and growth of two willow species. Ecology, 83, 2680–2690.

Smith RA, Mooney KA, Agrawal AA (2008) Coexistence of three specialist aphids on common milkweed, Asclepias syriaca. Ecology, 89, 2187–96.

Snyder WE, Ives AR (2001) Generalist Predators Disrupt Biological Control by a Specialist Parasitoid. Ecology, 82, 705.

Snyder WE, Ives AR (2003) Interactions between specialist and generalist natural enemies: Parasitoids, predators, and pea aphid biocontrol. Ecology, 84, 91–107.

Snyder WE, Snyder GB, Finke DL, Straub CS (2006) Predator biodiversity strengthens herbivore suppression. Ecology letters, 9, 789–96.

Stadler B (2002) Determinants of the size of aphid-parasitoid assemblages. Journal of Applied Entomology, 126, 258–264.

Stadler B, Dixon AFG (2005) Ecology and evolution of aphid-ant interactions. Annual Review of Ecology, Evolution, and Systematics, 36, 345–372.

Styrsky JD, Eubanks MD (2007) Ecological consequences of interactions between ants and honeydew-producing insects. Proceedings of the Royal Society B-Biological Sciences, 274, 151–64.

Styrsky JD, Eubanks MD (2010) A facultative mutualism between aphids and an invasive ant increases plant reproduction. Ecological Entomology, 35, 190–199.

Völkl W (1994) The effect of ant-attendance on the foraging behaviour of the aphid parasitoid Lysiphlebus cardui. Oikos, 70, 149–155.

Way MJ (1963) Mutualism between ants and honeydew-producing Homoptera. Annual Review of Entomology, 8, 307–344.

Wimp GM, Whitham TG (2001) Biodiversity consequences of predation and host plant hybridization on an aphid-ant mutualism. Ecology, 82, 440–452.

Yoo HJS, Kizner MC, Holway DA (2013) Ecological effects of multi-species, ant – hemipteran mutualisms in citrus. Ecological Entomology, 38, 505–514.

800

a *

600

400

200 Mean aphid abundance aphid Mean

0 20 b

5 Mean ant abundance ant Mean

0 High-density Low-density

Figure 4.1. Mean (+ SE) time-averaged (a) aphid and (b) ant abundance on cotton plants in low-density and high-density aphid treatment groups. (*p < 0.05).

1 a

0.5

0 NMDS2

-0.5

-1 -0.5 0 0.5 NMDS1 0.3

b ** 0.25

0.2

0.15

0.1

0.05 Abundance perplant Abundance

0 Leaf Chewing Generalist Specialist Herbivores Predators Predators

Feeding Guild Figure 4.2. (a) A non-metric multidimensional scaling ordination indicating mean (± SD) species composition of insects in high-density (black circle) and low-density (white circle) aphid groups. Stress is 0.071. (b) The mean (± SE) time-averaged abundances of leaf chewing herbivores, generalist predators and aphid specialist predators in low-density and high-density aphid treatments. (** p < 0.01). Note: the Argentine ant and cotton aphid are excluded from these analyses.

150 5 * a * b

120 4

90 3

60 2

Linkage Linkage density Numberoflinks 30 1

0 0

120 30

c d * 100 25 ns

80 20

60 15

40 10

20 Number ofspecialist links Numberof generalist links

0 0 High-density Low-density High-density Low-density Figure 4.3. (a) Mean (± SE) number of links and (b) linkage density in arthropod food webs on plants in low-density and high-density aphid treatment groups. Subsets of the total food web were created that only include (c) links between generalist predators and their prey, or (d) links between aphid specialists and their prey.

a Ants present Ants removed

300

250 *

200 * 150

100

50 Duration of coccinelid visit (s) Duration 0 300 b

ns ns 250

200

150

100

50 Duration of syrphid visit (s) of syrphid Duration

0 On the leaf Near aphids

Figure 4.4. Mean (± SE) time spent on leaves near aphids by (a) generalist and (b) specialist aphid predators in the presence (black bars) and absence (white bars) of the

Argentine ant.

3 a

1 Lepedoptera abundance Lepedoptera

0.2 b ns 0.15

0.1

0.05 Average change in leaf area leaf in change Average 0 High-aphid Low-aphid Aphid density treatment

Figure 4.5. (a) Mean (+ SE) lepidopteran abundance and (b) damage by leaf-chewing insects in high-density (black bars) and low-density (white bars) aphid groups.