DIRECT AND INDIRECT EFFECTS OF NATIVE AND INVASIVE PLANTS ON MOSQUITO ECOLOGY

BY

ALLISON MAE GARDNER

DISSERTATION

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Entomology in the Graduate College of the University of Illinois at Urbana-Champaign, 2016

Urbana, Illinois

Doctoral Committee:

Assistant Professor Brian F. Allan, Chair, Co-Director of Research Dr. Ephantus J. Muturi, Co-Director of Research Professor May R. Berenbaum Professor Lawrence M. Hanks Assistant Professor Allison K. Hansen Abstract

Container-breeding mosquitoes (Diptera: Culicidae), including the vectors of human and wildlife pathogens, interact with terrestrial plants throughout their life cycles. Inputs of leaf detritus into the aquatic habitat provide an energy base for developing larvae, and plants mediate the distribution of adult mosquitoes by influencing microclimate conditions, supplying sugar-feeding sources, and altering communities of wildlife blood-meal hosts. This dissertation examines direct and indirect effects of understory shrubs, including species both native and invasive to North

America, on the ecology of Culex pipiens, an important vector of West Nile virus in the northeastern and midwestern United States. Laboratory and field bioassays demonstrated that leaf detritus from different plant species in the aquatic environment alter two key components of mosquito production (i.e., oviposition site selection and adult emergence) via the abundance and composition of bacterial flora that form on different leaf species as they decompose. In particular, an invasive plant (Lonicera maackii, Amur honeysuckle) yielded high oviposition and adult emergence rates, while in contrast, a native plant (Rubus allegheniensis, common blackberry) was identified to function as an ecological trap for Cx. pipiens, attracting gravid females to oviposit and yet deleterious to larvae yielding low emergence rates. Subsequent laboratory bioassays in which first instar larvae were exposed to mixtures of leaves from different plant species revealed that while leaf resource diversity generally yields an increase in

Cx. pipiens adult emergence rates, addition of high-quality resources is not sufficient to offset the deleterious effect of R. allegheniensis leaves. I then explored two integrated vector management applications of these findings. First, a field experiment demonstrated the feasibility of exploiting a naturally-occurring ecological trap (R. allegheniensis leaves) and an artificial ecological trap

ii (L. maackii leaves mixed with Bacillus thuringiensis var. israelensis larvicide) for attract-and- kill mosquito control in storm water catch basins, in which gravid females are lured to oviposit in a low-quality environment. This result provides experimental proof of concept for a novel integrated vector management tool that may enhance the effectiveness and sustainability of existing mosquito abatement strategies with minimal non-target effects and reduced potential to select for insecticide resistance. A second field experiment showed that removal of L. maackii decreases abundance of adult Culex spp. mosquitoes in forest fragments within a residential neighborhood. The mechanisms underlying this reduction in mosquito abundance most likely include effects of L. maackii removal on microclimate conditions and the availability of avian blood-meal hosts. Collectively, these studies reveal multiple ecological pathways by which terrestrial plants interact with, and alter the abundance, distribution, and life history characteristics of mosquitoes, and suggest landscape modification strategies that may be used to manage an important disease vector species in residential ecosystems.

iii Acknowledgments

I owe my great experience as a graduate student at the University of Illinois to numerous individuals with whom I have had the pleasure of working in the classroom and the lab. First and foremost, I would like to thank my advisors, Dr. Brian Allan and Dr. Ephantus Muturi, as well as

Dr. Don Bullock and Dr. Marilyn O’Hara Ruiz, for their instruction, guidance, and unfailing support of my academic goals. Further I would like to thank my thesis committee members, Dr.

May Berenbaum, Dr. Larry Hanks, and Dr. Allison Hansen. Through direct feedback on my work and by example, my professors have shown me – and enabled me to discover for myself – intellectual paths that I could not have imagined six years ago. There is little for which I could be more grateful both professionally and personally, and I intend to spend my career paying forward this exceptional mentorship and investment in my education.

My research was accomplished through the dedication and energy of dozens of people, and acknowledgments for specific projects are included at the conclusion of each chapter, but I would like to begin by thanking a few current and former members of my advisors’ labs for their ideas and technical assistance on all aspects of my dissertation: Dr. Andrew Mackay, Allison

Parker, Dr. Chang-Hyun Kim, Millon Blackshear, Dr. Page Fredericks, and Dr. Jeff Bara. Thank you to the independent research students who conducted lab and field work with me: Brit’nee

Haskins, Lauren Frisbie, Matt Edinger, Jackie Duple, Leah Overmier, Noor Malik, Reanna

Kayser, Jake Gold, Chase Robinson, Michelle Manious, Audrey Killarney, Kristen Chen, Katie

Lincoln, Ali Braet, Billy LeGare, and Grace Cahill.

Finally, I would like to acknowledge my family and friends who shared my experience at the University of Illinois. I would like to thank my parents and my sister, Wendy, Tom, and

iv Kristy Gardner, and my friends in the department, Allison Parker, Tyler Hedlund, Tanya Josek,

Erin Allmann Updyke, Erin Welsh, Natalie Pawlikowski, Elijah Juma, and Catherine Wangen, for their thoughtful comments on my writing and presentations and their encouragement and good humor as they plowed through hours and hours of conscripted field work service and witnessed their yards, homes, and freezers being transformed into field lab extensions. I would like to acknowledge my earliest research mentors, Zenaida Bongaarts and the late Dr. Robert

Pavlica, for their high expectations and commitment to teaching that have been a source of motivation and inspiration for twelve years and counting. I would like to thank Brandon

Lieberthal for his emotional and technical support, and for many memories – that I’m sure will become fond eventually – of late night Rural King runs and low-budget field equipment construction projects. If there is a limit to your patience and kindness, I haven’t discovered it yet!

v Table of Contents

Introduction ...... 1

CHAPTER 1: Asymmetric effects of native and exotic invasive shrubs on ecology of the West Nile virus vector Culex pipiens (Diptera: Culicidae) ...... 10

CHAPTER 2: Leaf detritus mixtures alter microbial resource composition in aquatic habitats and performance of container-breeding mosquito larvae (Diptera: Culicidae) ...... 38

CHAPTER 3: Exploitation of ecological traps for the control of a mosquito vector for West Nile virus ...... 80

CHAPTER 4: Removal of invasive Amur honeysuckle (Lonicera maackii) decreases abundance of mosquitoes (Diptera: Culicidae) ...... 129

vi Introduction

Invasive species, defined here as non-native species whose introduction and subsequent spread are likely to cause economic or environmental harm, rank second only to habitat destruction as a threat to native biodiversity (Wilcove et al. 1998, Pimentel et al. 2005).

Mechanisms by which invasive species may reduce biodiversity include altering habitat structure, acting as predators and competitors of native species, and hybridizing with native species (Mooney and Cleland 2001, Lee 2002, Levine et al. 2003). There also is evidence that invasive species may cause increased risk of exposure to infectious diseases in wildlife and human populations via the introduction of novel pathogens and parasites into susceptible populations (Daszak et al. 2000) and by altering the ecology of established infectious agents that are embedded in biological communities through complex networks of energy flow (Pejchar and

Mooney 2009). The distribution and prevalence of arthropod-borne diseases depend on the interactions between an arthropod vector, the pathogen, and a human or wildlife reservoir host, and these interactions may be altered by invasive species via a diversity of mechanisms.

In particular, invasive plants are among the most important threats to native ecosystems and recent research has demonstrated their potential to trigger a cascade of ecological interactions that alter vector-borne disease transmission (Mack and Smith 2011, Mackay et al.

2016). This is a problem of global health significance because nearly one-third of emerging infectious diseases that threaten ecosystem health are vector-borne (Jones et al. 2008) and terrestrial plant invasions are increasingly widespread phenomena aided by human activities

(Reichard and White 2001). For example, in systems of diseases vectored by ticks (Acari:

Ixodidae), invasive bush honeysuckle (Lonicera maackii) appears to enhance the risk of exposure to human tick-borne ehrlichiosis (Ehrlichia chaffeensis and E. ewingii) by increasing

1 the abundance and encounter frequency of lone star ticks (Amblyomma americanum) and their white-tailed deer (Odocoileus virginianus) hosts (Allan et al. 2010). Similarly, invasive Japanese barberry (Berberis thunbergii) directly improves habitat quality and recruitment of the black- legged tick (Ixodes scapularis) via increased leaf litter depth and soil moisture (Elias et al. 2006).

Mosquitoes (Diptera: Culicidae), other important arthropod vectors of wildlife and human pathogens, interact with plants throughout their life cycles and thus invasive plants also may influence risk of exposure to mosquito-borne illnesses, although the potential mechanisms by which plants alter vector and host ecology and the consequences for ecosystem health remain poorly understood.

During the juvenile life cycle stages, habitat quality for container-breeding mosquitoes is driven largely by inputs of allochthonous plant-based detritus from the terrestrial environment, an important energy source. As leaves decay in the aquatic environment through leaching of soluble compounds into the water (Kuiters and Sarink 1986) and decomposition by detritivorous macroinvertebrates (McArthur and Barnes 1988) and microorganisms (Fish and Carpenter 1982), mosquito larvae filter bacteria, fungi, and protozoa suspended in the water column and graze on leaf surfaces and the sides of artificial containers (Wallace and Merritt 1980, Merritt et al. 1992).

Mosquito adult emergence rates are related to the quantity and age of leaf detritus, whereby large masses of green leaves support higher emergence compared to small masses and senescent leaves (Walker et al. 1997). Emergence rates also vary due to species of leaf detritus in the aquatic habitat (Yanoviak 1999, Kominoski et al. 2007), and multiple biological mechanisms could underlie this observation. Habitat quality may be altered by plant secondary compounds that leach into the aquatic environment during decomposition, including toxic phytochemicals that act directly on developing mosquitoes (Kuiters and Sarink 1986) and antimicrobials that reduce the food available to larvae (Rios et al. 1988). Alternatively, the importance of bacteria

2 and fungi to the mosquito diet (Fish and Carpenter 1982) and the variability in the microbial communities that form on different leaf species (Findlay and Arsuffi 1989) are well-known, and habitat quality in the aquatic environment may be related to composition of the microorganisms associated with different leaves. The leaves of many invasive plants decompose rapidly and are associated with a greater abundance of microbial flora compared to native plants and thus may provide particularly high-quality resources for mosquito larvae (Poulette and Arthur 2012).

There also is some evidence that leaf detritus in the aquatic environment mediates choice of oviposition site by container-breeding mosquitoes. The ecological pathways governing habitat selection are less well-established experimentally than the mechanisms underlying habitat quality. Again, plant secondary compounds released during leaf decay differ among leaf species

(Horner et al. 1988, Cornelissen et al. 1999), and some of these volatiles (e.g., phenolics) may include chemicals that attract adult mosquitoes (Kramer and Mulla 1979, Bentley and Day 1988) as well as oviposition stimulants (Ponnusamy et al. 2008). Other studies of oviposition sites selection have linked semiochemicals produced by the bacterial flora as byproducts of metabolism to habitat selection by gravid female Aedes aegypti and Ae. albopictus (Ponnusamy et al. 2008, Ponnusamy et al. 2010). Optimal oviposition theory suggests that organisms should interpret environmental cues to choose the highest-quality habitat for the development of progeny (Gripenberg et al. 2010), but this hypothesis also has not been evaluated experimentally in container-breeding mosquitoes as it relates to leaf detritus.

Adult mosquitoes use plants as resting sites and their flowers as sources of sugar meals.

The high vegetation biomass characteristic of plant invasions may provide a rich supply of sugar meals and retain heat and humidity in the understory, both contributing to mosquito longevity

(Foster 1995, Delatte 2009). This potentially has consequences for transmission rates of mosquito-borne pathogens, since biological transmission is only possible when the vector

3 survives long enough for an ingested pathogen to complete the extrinsic incubation period.

Moreover, the ecology of some mosquitoes, such as Culex pipiens and Cx. restuans, may be influenced by the effects of invasive plants on the community of avian hosts of these mosquitoes.

For example, an important reservoir host for West Nile virus, the American robin (Turdus migratorius), is a frugivore and may select habitats infested with fruit-bearing invasive plants

(Ingold and Craycraft 1983), causing mosquitoes to aggregate in these areas (Diuk-Wasser et al.

2010) with potential downstream consequences for disease transmission.

This dissertation examines direct and indirect effects of understory shrubs, including both native and invasive species, on the ecology of Cx. pipiens, an important vector of West Nile virus in the northeastern and midwestern United States (Turell et al. 2001, Andreadis et al.

2001). With over 43,000 human cases including over 1,800 deaths documented since its introduction into the New York City metropolitan area in 1999 (Lanciotti et al. 1999) and subsequent expansion throughout the contiguous states (Hayes and Gubler 2006), West Nile virus is the most prevalent mosquito-borne pathogen in North America. Because there are no human vaccines available for West Nile virus and the majority of other mosquito-borne viruses, vector control is the most effective and in many cases the only viable option for the prevention and management of these infectious illnesses. I first examined the role of leaf detritus from terrestrial plants in mediating the aquatic ecology of Cx. pipiens, particularly two important components of mosquito production: oviposition site selection and adult emergence rates. After identifying leaves from different plant species (hereafter ‘leaf species’) that vary in their attractiveness to mosquitoes as well as their quality as nutritional resources, I next assessed variation in the microbial community associated with different leaf detritus species as a potential mechanism underlying variation in oviposition and emergence rates. Finally, I explored two vector management applications of these findings. First, I investigated the feasibility of

4 exploiting ecological traps (i.e., habitats that are attractive to mosquitoes for oviposition yet deleterious to developing larvae) for attract-and-kill mosquito control in storm water environments, in which an attractant is combined with a toxin to lure mosquitoes to lay eggs in low-quality habitats. Second, I conducted a Before-After/Control-Impact experiment to test the hypothesis that removal of a widespread invasive plant (Lonicera maackii) decreases the abundance of mosquitoes belonging to multiple genera in forest fragments embedded within a residential neighborhood. I also investigated mechanisms underlying a reduction in mosquito abundance, including the effects of honeysuckle removal on microclimate conditions and the avian host community for Cx. pipiens and Cx. restuans. Collectively, these studies reveal multiple ecological pathways by which terrestrial plants interact with and alter the abundance, distribution, and life histories of mosquitoes and suggest landscape modification strategies that may be used to control an important disease vector species in residential ecosystems.

References Cited

Allan BF, Dutra HP, Goessling LS, Barnett K, Chase JM, Marquis RJ, Pang G, Storch GA,

Thach RE, Orrock JL. 2010. Invasive honeysuckle eradication reduces tick-borne disease

risk by altering host dynamics. Proceedings of the National Academy of Sciences

USA, 107, 18523-18527.

Andreadis TG, Anderson JF, Vossbrinck CR. 2001. Mosquito surveillance for West Nile virus in

Connecticut, 2000: isolation from Culex pipiens, Cx. restuans, Cx. salinarius, and

Culiseta melanura. Emerging Infectious Diseases, 7, 670.

Bentley MD, Day JF. 1989. Chemical ecology and behavioral aspects of mosquito oviposition.

Annual Review of Entomology, 34, 401-421.

5

Cornelissen JH, Perez‐Harguindeguy N, Díaz S, Grime JP, Marzano B, Cabido M, Vendramini

F, Cerabolini B. 1999. Leaf structure and defence control litter decomposition rate across

species and life forms in regional floras on two continents. New Phytologist, 143, 191-

200.

Daszak P, Cunningham AA, Hyatt AD. 2000. Emerging infectious diseases of wildlife – threats

to biodiversity and human health. BioScience, 287, 443-449.

Delatte H, Gimonneau G, Triboire A, Fontenille D. 2009. Influence of temperature on immature

development, survival, longevity, fecundity, and gonotrophic cycles of Aedes albopictus,

vector of chikungunya and dengue in the Indian Ocean. Journal of Medical

Entomology, 46, 33-41.

Diuk-Wasser MA, Molaei G, Simpson JE, Folsom-O’Keefe CM, Armstrong PM, Andreadis TG.

2010. Avian communal roosts as amplification foci for West Nile virus in urban areas in

northeastern United States. American Journal of Tropical Medicine and Hygiene, 82,

337-343.

Elias SP, Lubelczyk CB, Rand PW, Lacombe EH, Holman MS, Smith RP. 2006. Deer browse

resistant exotic-invasive understory: an indicator of elevated human risk of exposure to

Ixodes scapularis (Acari: Ixodidae) in southern coastal Maine woodlands. Journal of

Medical Entomology, 43, 1142-1152.

Findlay SE, Arsuffi TL. 1989. Microbial growth and detritus transformations during

decomposition of leaf litter in a stream. Freshwater Biology, 21, 261-269.

Fish D, Carpenter SR. 1982. Leaf litter and larval mosquito dynamics in tree-hole ecosystems.

Ecology, 283-288.

Foster WA. 1995. Mosquito sugar feeding and reproductive energetics. Annual Review of

Entomology, 40, 443-474.

6

Gripenberg S, Mayhew PJ, Parnell M, Roslin T. 2010. A meta‐analysis of preference-

performance relationships in phytophagous insects. Ecology Letters, 13, 383-393.

Hayes EB, Gubler DJ. 2006. West Nile virus: Epidemiology and clinical features of an emerging

epidemic in the United States. Annual Review of Medicine, 57, 181-194.

Horner JD, Gosz JR, Cates RG. 1988. The role of carbon-based plant secondary metabolites in

decomposition in terrestrial ecosystems. American Naturalist, 869-883.

Ingold JL, Craycraft MJ. 1983. Avian frugivory on honeysuckle (Lonicera) in soutwestern Ohio

in fall. Ohio Journal of Science, 83, 256-258.

Jones KE, Patel NG, Levy MA, Storeygard A, Balk D, Gittleman JL, Daszak P. 2008. Global

trends in emerging infectious diseases. Nature, 451, 990-993.

Kominoski JS, Pringle CM, Ball BA, Bradford MA, Coleman DC, Hall DB, Hunter MD. 2007.

Nonadditive effects of leaf litter species diversity on breakdown dynamics in a detritus-

based stream. Ecology, 88, 1167-1176.

Kramer WL, Mulla MS. 1979. Oviposition attractants and repellents of mosquitoes: oviposition

responses of Culex mosquitoes to organic infusions. Environmental Entomology, 8, 1111-

1117.

Kuiters AT, Sarink HM. 1986. Leaching of phenolic compounds from leaf and needle litter of

several deciduous and coniferous trees. Soil Biology and Biochemistry, 18, 475-480.

Lanciotti RS, Roehrig JT, Deubel V, Smith J, Parker M, Steele K, Crise B, Volpe KE, Crabtree

MB, Scherret JH, Hall RA. 1999. Origin of the West Nile virus responsible for an

outbreak of encephalitis in the northeastern United States. Science, 286, 2333-2337.

Lee CE. 2002. Evolutionary genetics of invasive species. Trends in Ecology and Evolution, 17,

386-391.

7

Levine JM, Vila M, Antonio CM, Dukes JS, Grigulis K, Lavorel S. 2003. Mechanisms

underlying the impacts of exotic plant invasions. Proceedings of the Royal Society of

London B: Biological Sciences, 270, 775-781.

Mack R, Smith M. 2011. Invasive plants as catalysts for the spread of human parasites.

NeoBiota, 9, 13-29.

Mackay AJ, Muturi EJ, Ward MP, Allan BF. 2016. Cascade of ecological consequences for West

Nile virus transmission when aquatic macrophytes invade storm water habitats.

Ecological Applications, 26, 219-232.

McArthur JV, Barnes JR. 1988. Community dynamics of leaf litter breakdown in a Utah alpine

stream. Journal of the North American Benthological Society, 37-43.

Merritt RW, Dadd RH, Walker ED. 1992. Feeding behavior, natural food, and nutritional

relationships of larval mosquitoes. Annual Review of Entomology, 37, 349-374.

Mooney HA, Cleland EE. 2001. The evolutionary impact of invasive species. Proceedings of the

National Academy of Sciences USA, 98, 5446-5451.

Pejchar L, Mooney HA. 2009. Invasive species, ecosystem services and human well-being.

Trends in Ecology and Evolution, 24, 497-504.

Pimentel D, Zuniga R, Morrison D. 2005. Update on the environmental and economic costs

associated with alien-invasive species in the United States. Ecological Economics, 52,

273-288.

Ponnusamy L, Xu N, Nojima S, Wesson DM, Schal C, Apperson CS. 2008. Identification of

bacteria and bacteria-associated chemical cues that mediate oviposition site preferences

by Aedes aegypti. Proceedings of the National Academy of Sciences USA, 105, 9262-

9267.

8

Ponnusamy L, Wesson DM, Arellano C, Schal C, Apperson CS. 2010. Species composition of

bacterial communities influences attraction of mosquitoes to experimental plant

infusions. Microbial Ecology 59, 158-173.

Poulette MM, Arthur MA. 2012. The impact of the invasive shrub Lonicera maackii on the

decomposition dynamics of a native plant community. Ecological Applications, 22, 412-

424.

Reichard SH, White P. 2001. Horticulture as a pathway of invasive plant introductions in the

United States. BioScience, 51, 103-113.

Rios JL, Recio MC, Villar A. 1988. Screening methods for natural products with antimicrobial

activity: a review of the literature. Journal of Ethnopharmacology, 23, 127-149.

Turell MJ, Dohm DJ, Sardelis MR, O’Guinn ML, Andreadis TG, Blow JA. 2005. An update on

the potential of North American mosquitoes (Diptera: Culicidae) to transmit West Nile

virus. Journal of Medical Entomology, 42, 57-62.

Walker ED, Merritt RW, Kaufman MG, Ayres MP, Riedel MH. 1997. Effects of variation in

quality of leaf detritus on growth of the eastern tree-hole mosquito, Aedes triseriatus

(Diptera: Culicidae). Canadian Journal of Zoology, 75, 706-718.

Wallace JB, Merritt RW. 1980. Filter-feeding ecology of aquatic insects. Annual Review of

Entomology, 25, 103-132.

Wilcove DS, Rothstein D, Dubow J, Phillips A, Losos E. 1998. Quantifying threats to imperiled

species in the United States. BioScience, 607-615.

Yanoviak SP. 1999. Effects of leaf litter species on macroinvertebrate community properties and

mosquito yield in Neotropical tree hole microcosms. Oecologia, 120, 147-155.

9

CHAPTER 1

Asymmetric effects of native and exotic invasive shrubs on ecology of the West Nile virus

vector Culex pipiens (Diptera: Culicidae)1

Abstract

Introduction. Exotic invasive plants alter the structure and function of native ecosystems and may influence the distribution and abundance of arthropod disease vectors by modifying habitat quality. This study investigated how invasive plants alter the ecology of Culex pipiens, an important vector of West Nile virus (WNV) in northeastern and midwestern United States.

Methods. Field and laboratory experiments were conducted to test the hypothesis that three native leaf species (Rubus allegheniensis, blackberry; Sambucus canadensis, elderberry; and

Amelanchier laevis, serviceberry), and three exotic invasive leaf species (Lonicera maackii,

Amur honeysuckle; Elaeagnus umbellata, autumn olive; and Rosa multiflora, multiflora rose) alter Cx. pipiens oviposition site selection, emergence rates, development time, and adult body size. The relative abundance of seven bacterial phyla in infusions of the six leaf species also was determined using quantitative real-time polymerase chain reaction to test the hypothesis that variation in emergence, development, and oviposition site selection is correlated to differences in the diversity and abundance of bacteria associated with different leaf species, important determinants of nutrient quality and availability for mosquito larvae.

Results. Leaf detritus from invasive honeysuckle and autumn olive yielded significantly higher adult emergence rates compared to detritus from the remaining leaf species and honeysuckle

1 This chapter appeared in its entirety in Parasites and Vectors and is referred to later in this thesis as Gardner et al. 2015. Gardner AM, Allan BF, Frisbie LA, and Muturi EJ. 2015. Asymmetric effects of native and exotic invasive shrubs on ecology of the West Nile virus vector Culex pipiens (Diptera: Culicidae). Parasites and Vectors 8: e329. This article is reprinted with the permission of the publisher and is available using doi: 10.1186/s13071-015-0941-z. 10 alleviated the negative effects of intraspecific competition on adult emergence. Conversely, leaves of native blackberry acted as an ecological trap, generating high oviposition but low emergence rates. Variation in bacterial flora associated with different leaf species may explain this asymmetrical production of mosquitoes: emergence rates and oviposition rates were positively correlated to bacterial abundance and diversity, respectively.

Conclusions. We conclude that the displacement of native understory plant species by certain invasive shrubs may increase production of Cx. pipiens with potential negative repercussions for human and wildlife health. These findings may be relevant to mosquito control and invasive plant management practices in the geographic range of Cx. pipiens. Further, our discovery of a previously unknown ecological trap for an important vector of WNV has the potential to lead to novel alternatives to conventional insecticides in mosquito control by exploiting the apparent

“attract-kill” properties of this native plant species.

Introduction

Exotic shrubs, aided mostly by human activities, have become dominant features of urban and suburban landscapes in North America (Bertin 2002, DeCandido et al. 2004). Deliberate introduction of non-native turf grass lawns and ornamental plants remains a common practice in residential neighborhoods despite the well-documented potential for these species to become invasive in their introduced ranges (Mack and Lonsdale 2001). On a broader scale, the disturbed structure of urban communities has given rise to an altered natural selection process that often functions to the competitive detriment of native species (Byers 2002). As a consequence, propagation of exotic invasive plants (hereafter “invasive plants”) may reduce the ecosystem services typically provided by native plant communities, including nutrient cycling, prevention of stream erosion, air filtration, and preservation of wildlife diversity (Kareiva et al. 2007,

11

Masters and Sheley 2001, Bolund and Hunhammar 1999, Smith et al. 2006, Burghardt et al.

2009).

There is also recent evidence that invasive plants can cause ecological cascades that alter human risk of exposure to diseases vectored by arthropods. For example, invasive plants may enhance the risk of exposure to tick-borne pathogens by increasing the abundance and encounter frequency of ticks and their vertebrate hosts (Williams et al. 2009, Williams et al. 2010, Allan et al. 2010). In mosquito-borne disease systems, changes in habitat complexity resulting from plant invasions may alter oviposition site selection by some vector species (Conley et al. 2011), and inputs of fruits and leaves of some terrestrial invasive plant species into aquatic larval habitats may provide a high-quality nutritional base for mosquito larvae (Reiskind and Zarrabi 2011). Yet despite the strong potential for vectors to interact with invasive plants, we have limited understanding of the diversity of potential mechanisms by which these plants alter vector ecology and the consequences for human health.

Container-dwelling mosquitoes provide an ideal model system for understanding the effects of native and invasive plants on mosquito ecology and the risk of human exposure to vector-borne pathogens. Culex pipiens pipiens Linnaeus (Diptera: Culicidae), an important vector for West Nile virus in urban landscapes throughout the northeastern and midwestern

United States (Turell et al. 2005), oviposits in a variety of natural and artificial containers such as small ponds, discarded tires, and storm water catch basins (Geery and Holub 1989, Yee 2008).

These habitats are mainly fueled by plant-based detritus from the surrounding terrestrial vegetation. This often includes leaves of non-native understory plant species (Gardner, pers. obs.), which commonly are introduced to residential environments for home landscaping and often become invasive in small urban forest fragments (Reichard and White 2001). Detritus type and quantity determine the composition and abundance of microbial communities that form in

12 container habitats as microbes break down terrestrial leaf litter (Fish and Carpenter 1982, Walker and Merritt 1988, Murrell and Juliano 2008, Ponnusamy et al. 2010, Muturi et al. 2013). In turn, these bacteria and fungi provide a direct food source for mosquito larvae (Merritt et al. 1992,

Kitching 2001) and influence oviposition behavior of gravid female mosquitoes through emission of oviposition attractants and stimulants (Bentley and Day 1988, Ponnusamy et al.

2008). Thus, detritus from terrestrial plants and its associated microbes play a critical role in determining vector distribution, relative abundance, and life history traits that are important for vector-borne pathogen transmission including adult body size, longevity, biting rates, and vector competence (Leonard and Juliano 1995, Walker et al. 1997, Bevins 2008).

It has been proposed that detritus from invasive plants benefits mosquito vectors and enhances vector-borne disease transmission potential because these leaves decompose faster and facilitate more rapid microbial growth compared to native plants (Blair and Stowasser 2009,

Poulette and Arthur 2012, Watling et al. 2011). In this study, we test the hypotheses that A) leaf detritus of three native and three invasive shrubs asymmetrically affects the emergence rates, development, and oviposition site selection by Cx. pipiens, and B) variation in emergence, development, and oviposition site selection is correlated with differences in the diversity and abundance of bacterial flora among leaf detritus types.

Methods

Selection of plant species. Six focal plant species were selected among shrubs common within the geographic range of Cx. pipiens (Darsie and Ward 2004). The invasive shrubs were Lonicera maackii (Dipsacales: Caprifoliaceae; Amur honeysuckle), Elaeagnus umbellata (Rosales:

Elaeagnaceae; autumn olive), and Rosa multiflora (Rosales: Rosaceae; multiflora rose). Found in both natural and domestic environments – including forest fragments, landscaped parks, and

13 residential neighborhoods – these three exotics are highly invasive throughout much of the

United States, and in some areas of the northeast and Midwest dominate over 80 percent of land cover (Luken 1988, Meiners et al. 2002). The native shrubs were Rubus allegheniensis (Rosales:

Rosaceae; blackberry), Sambucus canadensis (Dipsacales: Adoxaceae; elderberry), and

Amelanchier laevis (Rosales: Rosaceae; serviceberry), all three of which are common understory species in mixed deciduous forests and residential landscapes.

Attractiveness of leaf species for Cx. pipiens oviposition. Six oviposition traps (Reiter 1986) each containing 4 L of tap water and 80 g of fresh whole leaves of one of the six shrub species were placed 1 m apart from each other in partial shade at five sites located within a 5 km radius in residential Champaign-Urbana, Illinois. The 30 oviposition traps were monitored for egg rafts daily and the number of egg rafts collected in each substrate from June 24 to August 5, 2013 (39 days) was recorded. This design, which was informed by a smaller pilot study performed in

2012, tested for temporal variation in the attractiveness of the leaf species to gravid females based on the number of egg rafts collected, which may vary as the microbial and chemical composition of the infusion changes with decomposition of the leaves.

Data analyses were conducted in SAS 9.3 (SAS Institute Inc., Cary, NC). A general linear mixed model (GLMM) with repeated measures was used to compare the abundance of egg rafts collected by leaf substrate, day, and their interaction throughout the 39 day period (Eqn. 1), where 훽푖 represents the random effect of site (block), 퐿푗 represents the fixed effect of leaf litter species, 퐷푘 represents the fixed effect of day, 퐿퐷푗푘 represents their interaction.

푦푖푗푘 = 휇 + 훽푖 + 퐿푗 + 휀1(푖푗) + 퐷푘 + 퐿퐷푗푘 + 휀2(푖푗푘) (Eqn. 1)

Day was the repeated variable, and an autoregressive-1 covariance structure with restricted maximum likelihood (REML) was used for estimation of covariance parameters. Orthogonal 14 contrasts were used to test the linear and quadratic components of the quantitative day variable.

Tukey’s mean separation test was used for comparisons of leaf detritus species treatment means.

Number of egg rafts was log(x+0.1)-transformed to meet the assumptions of normally, identically, and independently distributed residuals.

Effect of leaf detritus species on Cx. pipiens emergence and development. Culex pipiens larvae were obtained by collecting egg rafts from five sites using standard grass infusion-baited oviposition traps. Egg rafts were individually hatched in petri dishes containing deionized water and first instar larvae of Cx. pipiens were distinguished from those of Culex restuans Theobald based on the presence or absence of a clear area anterior to the sclerotized egg-breaker which is present in Cx. restuans and absent in Cx. pipiens (Dodge 1966). Culex pipiens larvae collected from different sites were pooled prior to their random allocation to experimental treatments.

To test for the effect of leaf substrate on intraspecific competition, 18 treatments were established with five replicates per treatment (90 containers). Each treatment included one of three densities of first instar larvae of Cx. pipiens (10, 20, or 40 per container) and 360 mL infusion of one of the six leaf detritus species in 400 mL tri-pour beakers. Infusions were prepared by fermenting 80 g of fresh leaves of each plant species in 4 L of tap water for 7 days, a standard infusion age used in comparable studies. Fresh leaves constitute a substantial portion of leaf litter inputs in container habitats and are superior food source for mosquito larvae compared to senescent leaves (Walker et al. 1997). The containers were monitored daily and pupae were removed from containers and housed individually in cotton-sealed plastic vials with deionized water. The experiment was conducted under ambient conditions of 25ºC, 70% relative humidity, and a 16:8 (L:D) photoperiod. Adults were sorted by sex and date of eclosion and their wing lengths determined. Time to eclosion and adult body size are important determinants of the

15 potential for disease transmission (i.e., “vectorial capacity”) in mosquitoes. Rapid time to eclosion facilitates high emergence rates of mosquitoes even from ephemeral habitats, while adult body size is directly related to longevity and potential to transmit viruses; only a vector that survives the duration of the viral extrinsic incubation period can infect a human or wildlife host

(Stiles 1982).

Separate Multivariate Analysis of Variance (MANOVA) tests were used to test for the fixed effects of leaf detritus species, competition, and their interaction on male and female development time to eclosion and adult wing length. Standardized canonical coefficients (SCCs) were used to describe the relative contributions of development time and wing length to significant treatment effects as well as the relationship between the two variables. A GLMM with a factorial treatment structure was used to test the fixed effects of leaf species, competition, and their interaction on Cx. pipiens emergence rates (male and female combined) with Tukey’s separation of means. Development time and wing length were log(x+0.1)-transformed and emergence data were arcsine-square root(x)-transformed to meet the assumptions of the analysis.

Relationship between leaf detritus species and microbial composition and abundance.

Before adding larvae to treatments as described above, 2 mL aliquots of 7-day old leaf infusions were taken from each container and stored at -80°C until further processing. Genomic DNA was extracted from the samples using the UltraClean® Soil DNA Isolation Kit (Mo Bio Laboratories

Inc., Carlsbad CA, Cat. No. 12800-50) according to the manufacturer’s instructions. Real-time polymerase chain reaction was conducted to detect total bacterial abundance and abundance of seven bacterial phyla (α-proteobacteria, β-proteobacteria, γ-proteobacteria, firmicutes, acidobacteria, actinobacteria, and bacteroidetes) in aliquots of leaf infusions, according to the methods described in Muturi et al. 2013.

16

The Shannon diversity index (H) was used to calculate bacterial diversity at the phylum level associated with each leaf detritus species. Further analysis was carried out in three steps.

First, GLMs were used to compare total bacterial abundance and bacterial diversity among leaf species, with Tukey’s separation of treatment means. Second, multiple linear regression models were used to assess the relationship between bacterial abundance, bacterial diversity, larval density, and their interaction and mosquito emergence rates, and the relationship between bacterial abundance, bacterial diversity, and their interaction and oviposition rates. Finally, multiple linear regression models were used to identify specific bacterial phyla that are related to mosquito emergence and oviposition rates with phylum-level abundance of seven bacterial phyla as independent effects. Emergence rates were arcsine-square root(x)-transformed and oviposition rates and bacterial abundance and diversity measures were log(x+0.1)-transformed to meet the assumptions of all tests.

Results

Attractiveness of leaf species for Cx. pipiens oviposition. The number of egg rafts laid in oviposition traps containing the leaves of different native and invasive shrubs varied within and among leaf detritus species over the collection period, with significant effects of leaf species (F =

7.25; df = 5, 20; P = 0.0005) and day (F = 23.70; df = 39, 912; P < 0.0001) but not their interaction (Figure 1.1). Throughout the experiment, the greatest number of egg rafts was collected in blackberry and elderberry leaf infusion; the lowest number of egg rafts per day were collected from water containing serviceberry, autumn olive, and honeysuckle leaves; and an intermediate number of egg rafts were collected from water containing multiflora rose. We found a strong quadratic effect of time (F = 586.89; df = 1, 912; P < 0.0001) and a significant though

17 less important linear effect (F = 19.79; df = 1, 912; P < 0.0001) of time. The polynomial function relating oviposition response (y) to time (x) is given by

2 푦푖 = −0.029푥 + 1.117푥 − 0.308 (Eqn. 2)

Thus, all leaf detritus species initially increased in attractiveness and then declined in attractiveness after the critical point at 19 days.

Effect of leaf detritus species on Cx. pipiens emergence and development. Mosquito emergence rates varied across leaf detritus types and larval densities, with a significant interaction between leaf species and density (F = 7.33; df = 10, 72; P < 0.0001; Figure 1.2). The lowest emergence rates were observed in blackberry and multiflora rose infusion; no adults emerged in the latter leaf detritus species at any larval density. The highest emergence rates were observed in honeysuckle and autumn olive infusions, although autumn olive-reared mosquitoes experienced a significant decline in emergence at the highest density while honeysuckle infusion mitigated the deleterious effects of intraspecific competition even at high larval densities.

Among all other leaf species except for elderberry, higher larval densities yielded significantly lower emergence rates than lower larval densities.

Development time to adult eclosion and adult wing length also were influenced by leaf detritus type and larval density, with significant interactions between leaf species and density for females (F = 5.05; df = 16, 644; P < 0.0001) and males (F = 3.03; df = 16, 614; P < 0.0001;

Figure 1.3). SCCs show that wing length and development time both contributed strongly to the multivariate effect for females, while wing length explained most variance for males (Table 1.1).

The longest times to eclosion for females were observed in blackberry infusions, while the shortest times to eclosion were observed in honeysuckle infusions. Similarly, the shortest wing lengths for both females and males were observed in blackberry-reared mosquitoes, while the

18 longest wing lengths were observed in honeysuckle-reared mosquitoes. Multiflora rose infusions were excluded from this analysis because no mosquitoes survived in these treatments. Time to eclosion generally increased and adult wing length decreased with higher larval densities, although as in the case for emergence, honeysuckle infusion mitigated the effects of intraspecific competition on development time in particular.

Relationship between leaf detritus species and microbial composition and abundance. Total abundance (F = 7.24; df = 5, 24; P = 0.0003) and diversity of bacterial flora (F = 12.30; df = 5,

24; P < 0.0001) varied significantly with leaf detritus species (Figure 1.4). A high abundance of bacteria was present in honeysuckle, autumn olive, and elderberry infusion, while a low abundance was present in multiflora rose, blackberry, and serviceberry infusions. High phylum- level diversity of bacteria was present in blackberry and elderberry infusions, while low diversity was present in honeysuckle, multiflora rose, autumn olive, and serviceberry infusions.

Total bacterial abundance and its interaction with bacterial diversity were positively correlated to mosquito emergence rates, with no significant interaction between these effects and larval density (Table 1.2). Interestingly, the direction of the abundance and diversity interaction was negative, indicating that when bacterial diversity is increased, the positive effect of bacterial abundance on emergence is decreased and vice versa. In particular, abundance of β- proteobacteria was negatively correlated with rate of mosquito emergence while bacteroidetes and acidobacteria were positively correlated with emergence (Table 1.3). Again, these effects did not interact with larval density.

Bacterial diversity but not total bacterial abundance or their interaction was positively correlated to mosquito oviposition rate (Table 1.2). No particular bacterial phyla were identified that were significantly correlated to oviposition rate (Table 1.3).

19

Discussion

Using a combination of laboratory and field experiments, we identified two invasive shrubs that may promote growth and emergence of an important mosquito vector relative to native shrub species by improving the nutritional quality of the larval environment via leaf detritus inputs. Culex pipiens emergence rates were significantly higher in leaf infusions of honeysuckle and autumn olive compared to the other shrub species. We also noted that the deleterious effects of intraspecific larval competition were mitigated in honeysuckle treatments.

The reduction of this important constraint on mosquito population carrying capacity (Larson

2005) indicates that this invasive leaf species may support higher mosquito densities in heavily invaded areas. Similarly, mosquitoes reared in honeysuckle and autumn olive infusion had the shortest development time to eclosion and the longest adult wing lengths in both females and males compared to the other leaf detritus species. These life history characteristics are positively associated with efficiency of mosquito-borne pathogen transmission (i.e., “vectorial capacity”).

Therefore, the widespread distribution of these two invasive shrubs in North America raises concerns regarding their impact on the transmission of several mosquito-borne viruses such as

West Nile virus and St. Louis encephalitis.

Although not the most attractive leaf detritus species observed in the experiment, honeysuckle and autumn olive were by far the most favorable leaf detritus species for development and emergence of laboratory-reared mosquitoes. These results complement a growing body of field studies that suggest landscaping with exotic – and potentially invasive – plants has the potential to influence local larval and adult mosquito abundance and distribution

(Conley et al. 2011, Shewart et al. 2014). For instance, the presence of ornamental lawn shrubs, including many invasive species, is positively associated with abundance of Cx. pipiens and Cx. restuans larvae in roadside storm water catch basins (Gardner et al. 2013). However, it is

20 noteworthy that our experiments also identified a third invasive plant, multiflora rose, that is lethal to Cx. pipiens, suggesting that the effect of displacement of native plants by invasive species has neither uniformly positive nor negative impacts on the abundance and distribution of mosquitoes.

Our comparisons of native and invasive leaf detritus species on larval development and oviposition facilitated the discovery of a naturally-occurring ecological trap for Cx. pipiens.

Ecological traps occur due to a mismatch between the attractiveness of a habitat and its quality for reproduction (Schlaepfer et al. 2002). The greatest number of egg rafts was collected in water containing leaves of blackberry, a native plant species found throughout the geographic range of

Cx. pipiens. However, in laboratory assays, exceptionally low mosquito emergence rates were observed in blackberry infusions, with fewer than 20 percent of larvae surviving to eclosion even at the lowest larval density. Blackberry-reared mosquitoes also exhibited significantly longer development times to eclosion and the shortest adult wing lengths compared to those exposed to other leaf detritus species. Infusion of multiflora rose, an exotic shrub of limited importance in the Midwest but highly invasive in the northeastern United States (Stiles 1982, Larson 2005), similarly yielded lower emergence with no mosquitoes developing to eclosion across all density treatments. However, gravid females were better able to discriminate against this leaf detritus species and consequently water containing multiflora rose leaves collected fewer egg rafts than water containing blackberry leaves. Future research will determine whether exploitation of this ecological trap may yield a novel “attract-kill” approach to control mosquito larvae in closed aquatic environments, such as rain barrels, buckets, and storm water catch basins.

A potential mechanism to explain both asymmetrical oviposition rates and emergence and development rates of mosquitoes with respect to leaf detritus species is differences in the associated microbial flora. Honeysuckle, autumn olive, and elderberry infusions contained

21 greater abundances of bacterial flora than multiflora rose, serviceberry, and blackberry infusion, an observation that likely reflects more rapid leaf decomposition rates among the former species

(Blair and Stowasser 2009, Poulette and Arthur 2012), yielding larger amounts of bacteria. For the most part, mosquito emergence reflected this pattern; the notable exception was serviceberry infusion, in which a moderate proportion of mosquitoes developed to eclosion despite low bacterial abundance. This result suggests total bacterial abundance may be a good indicator of habitat quality under most but not all conditions (Murrell and Juliano 2008). Microbial abundance also varied significantly between different phyla across leaf detritus species, and was statistically correlated to mosquito emergence. In particular, β-proteobacteria was negatively correlated with emergence rates while bacteroidetes and acidobacteria were positively correlated with emergence rates. This result is consistent with studies of other mosquitoes; for example,

Aedes triseriatus Say and Culex tarsalis Coquillett appear to feed preferentially on bacteroidetes

(Duguma et al. 2013, Kaufman et al. 2008).

Further, irrespective of bacterial abundance, phylum-level bacterial diversity was positively correlated with oviposition rates. This result supports previous research which indicates that mosquito oviposition site discovery and selection may be mediated by the odors produced by a higher diversity of bacteria (Ponnusamy et al. 2010, Ponnusamy et al. 2008).

Thus, we propose that optimal habitats for mosquito production may contain both higher bacterial abundance and higher bacterial diversity, which jointly facilitate discovery of the habitat and the emergence and development of larvae within. Meanwhile, high-diversity and low-abundance habitats such as blackberry-infused water may constitute ecological traps, while low-diversity and high-abundance habitats such as honeysuckle-infused water appear to attract fewer gravid females but promote higher rates of mosquito emergence. It is possible that observed differences in oviposition and emergence rates were not driven by bacterial diversity

22 per se, but rather variation in bacterial community composition between leaf species. It also is noteworthy that oviposition and emergence rates associated with different leaf species may be influenced by phytochemicals unrelated to microbial flora; for example, tannins are abundant in blackberry leaves (Foo and Porter 1981, Gudej et al. 2004) and are known to be toxic to mosquitoes in other systems (Mercer 1993, Rey et al. 1999). However, these hypotheses were not tested thoroughly in our current experiment.

Our understanding of the mechanistic role potentially played by variation in microbial growth is limited by our focus on a higher taxonomic level of bacteria (i.e., phylum) and at the exclusion of fungi and other microbes that may be consumed by mosquito larvae. Additional studies are required to identify the particular microbial taxa that may be involved in determining mosquito emergence and oviposition site selection. Further, our molecular approach is subject to limitations such as potential underestimation of bacterial abundance due to DNA extraction bias and PCR bias (Kirk et al. 2004, Martin-Laurent et al. 2001). However, important conclusions can still be made; our findings corroborate previous research which has demonstrated that resource type and availability may influence emergence and oviposition. For example, variation in microbial communities among leaf detritus treatments have been shown to be related to Culex mollis Dyar life history traits (Yanoviak 1999), and binary choice laboratory assays revealed that

Aedes aegypti Linnaeus oviposition site selection may be related to extracts released by the microbes present in infusions of different leaf species (Ponnusamy et al. 2008).

In summary, we observed elevated emergence rates and more rapid development among

Cx. pipiens mosquitoes reared in infusions of honeysuckle and autumn olive leaves, two exotic, invasive shrubs that occur throughout much of the northeastern and midwestern United States. In contrast, we discovered mosquito emergence was significantly reduced among mosquitoes reared in infusions of native blackberry and exotic multiflora rose leaves compared to those exposed to

23 other leaf detritus species. Our results have applications in two areas. First, our findings that some exotic, invasive shrubs are favorable for mosquito production may be relevant to mosquito control and invasive plant management practices in the geographic range of Cx. pipiens. Second, our discovery of a previously unknown ecological trap for an important vector of West Nile virus has the potential to lead to novel alternatives to conventional insecticides in mosquito control, exploiting the apparent “attract-kill” properties of this native plant species.

Acknowledgments

We thank Nina Krasavin, Millon Blackshear, Chang-Hyun Kim, Matthew Edinger, Audrey

Killarney, and William LeGare for their technical assistance, Andrew Mackay and Richard

Lampman for their advice on the experimental design, and the Muturi and Allan lab groups for their comments on the manuscript. We also thank homeowners in the City of Urbana and the

City of Champaign, Illinois, for land use permission. This study was supported by the Used Tire

Fund and Emergency Public Health Act from the State of Illinois to EJM. Summer salary for

AMG was supported by the University of Illinois at Urbana-Champaign (UIUC) School of

Integrative Biology, and research assistant support for AMG for fall 2013 was provided by a

UIUC Center for a Sustainable Environment Sustainability Fellowship to BFA.

References Cited

Allan BF, Dutra HP, Goessling LS, Barnett K, Chase JM, Marquis RJ, Pang G, Storch GA,

Thach RE, Orrock JL. 2010. Invasive honeysuckle eradication reduces tick-borne disease

risk by altering host dynamics. Proceedings of the National Academy of Sciences USA,

107, 18523-18527.

24

Bentley MD, Day JF. 1988. Chemical ecology and behavioral aspects of mosquito oviposition.

Annual Review of Entomology, 34, 401-421.

Bertin RI. 2002. Losses of native plant species from Worcester, Massachusetts. Rhodora, 104,

325-349.

Bevins SN. 2008. Invasive mosquitoes, larval competition, and indirect effects on the vector

competence of native mosquito species (Diptera: Culicidae). Biological Invasions, 10,

1109-1117.

Blair BC, Stowasser A. 2009. Impact of Lonicera maackii on decomposition rates of native leaf

litter in a southwestern Ohio woodland. Ohio Journal of Science, 109, 43-47.

Bolund P, Hunhammar S. 1999. Ecosystem services in urban areas. Ecological Economics, 29,

293-301.

Burghardt KT, Tallamy DW, Shriver WG. 2009. Impact of native plants on bird and butterfly

biodiversity in suburban landscapes. Conservation Biology, 23, 219-224.

Byers JE. 2002. Impact of non-indigenous species on natives enhanced by anthropogenic

alteration of selection regimes. Oikos, 97, 449-458.

Conley AK, Watling JI, Orrock JL. 2011. Invasive plant alters ability to predict disease vector

distribution. Ecological Applications, 21, 329-334.

Darsie RF, Ward RA. 2004. Identification and geographical distribution of the mosquitoes of

North America, north of Mexico. Gainesville: University Press of Florida.

DeCandido R, Muir AA, Gargiullo MB. 2004. First approximation of the historial and extant

vascular flora of New York City: Implications for native plant species conservation.

Journal of the Torrey Botanical Society, 131, 243-251.

Dodge HR. 1966. Studies on mosquito larvae III. The first instar larvae of North American

Culicidae and of world Anophelinae. Canadian Entomologist, 98, 337-393.

25

Duguma D, Rugman-Jones P, Kaufman MG, Hall MW, Neufeld JD, Stouthamer R, Walton WE.

2013. Bacterial communities associated with Culex mosquito larvae and two emergent

aquatic plants of bioremediation importance. PLoS One, 8, 72522.

Fish D, Carpenter SR. 1982. Leaf litter and larval mosquito dynamics in tree-hole ecosystems.

Ecology, 63, 283-288.

Foo LY, Porter LJ. 1981. The structure of tannins of some edible fruits. Journal of Science,

Food, and Agriculture, 32, 711-716.

Gardner AM, Anderson TK, Hamer GL, Johnson DE, Varela KE, Walker ED, Ruiz MO. 2013.

Terrestrial vegetation and aquatic chemistry influence larval mosquito abundance in catch

basins, Chicago, USA. Parasites and Vectors, 6, 9.

Geery PR, Holub RE. 1989. Seasonal abundance and control of Culex spp in catch basins in

Illinois. Journal of the American Mosquito Control Association, 5, 537-540.

Gudej J, Tomczyk M. 2004. Determination of flavonoids, tannins, and ellagic acid in leaves from

Rubus L. species. Archives of Farm Research, 27, 1114-1119.

Kareiva P, Watts S, McDonald R, Boucher T. 2007. Domesticated nature: Shaping landscapes

and ecosystems for human welfare. Science, 316, 1866-1869.

Kaufman MG, Chen S, Walker ED. 2008. Leaf-associated bacterial and fungal taxa shifts in

response to larvae of the tree hole mosquito, Ochlerotatus triseriatus. Microbial Ecology,

55, 673-684.

Kirk JL, Beaudette LA, Hart M, Moutoglis P, Kilronomos JN, Lee H, Trevors JT. 2004. Methods

of studying soil microbial diversity. Journal of Microbial Methods, 2, 169-188.

Kitching RL. 2001. Food webs in phytotelmata: “bottom-up” and “top-down” explanations for

community structure. Annual Review of Entomology, 46, 729-760.

26

Larson BM. 2005. The war of the roses: demilitarizing invasion biology. Frontiers in Ecology

and the Environment, 3, 495-500.

Leonard PM, Juliano SA. 1995. Effect of leaf litter and density on fitness and population

performance of the hole mosquito Aedes triseriatus. Ecological Entomology, 20, 125-

136.

Luken JO. 1988. Population structure and biomass allocation of the naturalized shrub Lonicera

maackii (Rupr.) Maxim. in forest and open habitats. American Midland Naturalist, 119,

258-267.

Mack RN, Lonsdale WM. 2001. Humans as global plant dispersers: Getting more than we

bargained for. Bioscience, 51, 95-102.

Martin-Laurent F, Philippot L, Hallet S, Chaussod R, Germon JC, Soulas G, Catroux G. 2001.

DNA extraction from soils: old bias for new microbial diversity analysis methods.

Applied Environmental Microbiology, 67, 2354-2359.

Masters RA, Sheley RL. 2001. Principles and practices for managing rangeland invasive plants.

Journal of Range Management, 54, 502-517.

Meiners SJ, Pickett ST, Cadenasso ML. 2002. Exotic plant invasions over 40 years of old field

successions: community patterns and associations. Ecography, 25, 215-223.

Mercer DR. 1993. Effect of tannic acid concentration on development of the western treehole

mosquito, Aedes sierrensis (Diptera: Culicidae). Journal of Chemical Ecology, 19, 1119-

1127.

Merritt RW, Dadd RH, Walker ED. 1992. Feeding behavior, natural food, and nutritional

relationships of larval mosquitos. Annual Review of Entomology, 37, 349-376.

27

Murrell EG, Juliano SA. 2008. Detritus type alters the outcome of interspecific competition

between Aedes aegypti and Aedes albopictus (Diptera: Culicidae). Journal of Medical

Entomology, 45, 375-383.

Muturi EJ, Allan BF, Ricci J. 2012. Influence of leaf detritus type on production and longevity of

container-breeding mosquitoes. Environmental Entomology, 41, 1062-1068.

Muturi EJ, Orindi BO, Kim CH. 2013. Effect of leaf type and pesticide exposure on abundance

of bacterial taxa in mosquito larval habitats. PLoS One, 8, 71812.

Ponnusamy L, Wesson DM, Arellano C, Schal C, Apperson CS. 2010. Species composition of

bacterial communities influences attraction of mosquitoes to experimental plant

infusions. Microbial Ecology, 59, 158-173.

Ponnusamy L, Xu N, Nojima S, Wesson DM, Schal C, Apperson CS. 2008. Identificiation of

bacteria and bacteria-associated chemical cues that mediate oviposition site preferences

by Aedes aegypti. Proceedings of the National Academy of Sciences USA, 105, 9262-

9267.

Poulette MM, Arthur MA. 2012. The impact of the invasive shrub Lonicera maackii on the

decomposition dynamics of a native plant community. Ecological Applications, 22, 412-

424.

Reichard SH, White P. 2001. Horticulture as a pathway of invasive plant introductions in the

United States. Bioscience, 51, 103-113.

Reiskind MH, Zarrabi AA. 2011. The importance of an invasive tree fruit as a resource for

mosquito larvae. Journal of Vector Ecology, 36, 197-203.

Reiskind MH, Zarrabi AA, Lounibos LP. 2012. Effects of combination of leaf resources on

competition in container mosquito larvae. Bulletin of Entomological Research, 7, 1-11.

28

Reiter P. 1986. A standardized procedure for the quantitative surveillance of certain Culex

mosquitoes by egg raft collection. Journal of the American Mosquito Control

Association, 2, 219-221.

Rey D, Cuany A, Pauton MP, Meyran JC. 1999. Differential sensitivity of mosquito taxa to

vegetable tannins. Journal of Chemical Ecology, 25, 537-548.

Schlaepfer MA, Runge MC, Sherman PW. 2002. Ecological and evolutionary traps. Trends in

Ecology and Evolution, 17, 474-480.

Shewart L, McEwan RW, Benbow ME. 2014. Evidence for facilitation of Culex pipiens

(Diptera: Culicidae) life history traits by the nonnative invasive shrub Amur honeysuckle

(Lonicera maackii). Environmental Entomology, 43, 1584-1593.

Smith RM, Thompson K, Hodgson JG, Warren PH, Gaston KJ. 2006. Urban domestic gardens

(IX): Composition and richness of the vascular plant flora, and implications for native

biodiversity. Biological Conservation, 129, 312-322.

Stiles EW. 1982. Expansions of mockingbird and multiflora rose in the northeastern United

States and Canada. American Birds, 36, 358-364.

Turell MJ, Dohm DJ, Sardelis MR, O’Guinn ML, Andreadis TG, Blow JA. 2005. An update on

the potential of North American mosquitoes (Diptera: Culicidae) to transmit West Nile

virus. Journal of Medical Entomology, 42, 57-62.

Walker ED, Merritt RW. 1988. The significance of leaf detritus to mosquito (Diptera: Culicidae)

productivity from treeholes. Environmental Entomology, 17, 199-206.

Walker ED, Kaufman MG, Ayres MP, Riedel MH, Merritt RW. 1997. Effects of variation in

quality of leaf detritus on growth of the eastern tree-hole mosquito, Aedes triseriatus

(Diptera: Culicidae). Canadian Review of Zoology, 75, 706-718.

29

Watling JL, Hickman CR, Lee E, Wang K, Orrock JL. 2011. Extracts of the invasive shrub

Lonicera maackii increase mortality and alter behavior of amphibian larvae. Oecologia,

165, 153-159.

Williams SC, Ward JS, Worthley TE, Stafford KC. 2009. Managing Japanese barberry

(Ranunculales: Berberidaceae) infestations reduces blacklegged tick (Acari: Ixodidae)

abundance and infection prevalence with Borrelia burgdorferi (Spirochaetales:

Spirochaetaceae). Environmental Entomology, 38, 977-984.

Williams SC, Ward JS. 2010. Effects of Japanese barberry (Ranunculales: Berberidaceae)

removal and resulting microclimatic changes on Ixodes scapularis (Acari: Ixodidae)

abundances in Connecticut, USA. Environmental Entomology, 39, 1911-1921.

Yanoviak SP. 1999. Effects of leaf litter species on macroinvertebrate community properties and

mosquito yield in Neotropical tree hole microcosms. Oecologia, 120, 147-155.

Yee DA. 2008. Tires as habitats for mosquitoes: a review of studies within the eastern United

States. Journal of Medical Entomology, 45, 581-593.

30

Figures and Tables

Figure 1.1. Mean (±1 standard error) for Culex pipiens egg rafts collected in oviposition traps per day from June 24 to August 5, 2013 (6 weeks) by leaf detritus treatment. Letters indicate significant pairwise differences at α = 0.05.

31

Figure 1.2. Mean (±1 standard error) for Culex pipiens male and female emergence rates across intraspecific competition by leaf detritus treatments. Letters indicate significant pairwise differences at α = 0.05.

32

Figure 1.3. Mean (±1 standard error) for Culex pipiens female and male time to eclosion and wing length across intraspecific competition by leaf detritus treatments. The following treatments were excluded because no females survived to eclosion: all multiflora rose treatments; blackberry: 20 larvae and 40 larvae; autumn olive: 40 larvae.

33

Figure 1.4. Mean (±1 standard error) for cumulative abundance and Shannon’s diversity index

(H) of bacteria by leaf detritus treatment. Upper case letters indicate significant pairwise differences at α = 0.05 for bacterial diversity; lower case letters indicate significant pairwise differences for bacterial abundance.

34

Table 1.1. Multivariate Analysis of Variance (MANOVA) for the effect of leaf detritus species and intraspecific competition on Culex pipiens female and male time to eclosion (days) and wing length (mm). Standardized canonical coefficients (SCCs) describe the relative contribution of each life history trait to the multivariate effect. Asterisks indicate statistically significant effects at α = 0.05.

Pillai’s SCCs Sex Variable df P trace Eclosion time Wing length Leaf 8 22.07 <0.001* 1.41 -0.33 Female Competition 4 18.81 <0.001* 1.01 -0.94 Leaf*Competition 16 5.05 <0.001* -0.92 1.02 Leaf 8 26.52 <0.001* 1.25 -0.27 Male Competition 4 22.79 <0.001* -0.44 1.21 Leaf*Competition 16 3.03 <0.001* 0.38 1.46

35

Table 1.2. Multiple linear regressions for the effect of total bacterial abundance, bacterial diversity, larval density (10, 20, and 40 larvae), and their interaction on Culex pipiens female and male emergence rate and oviposition rates. Asterisks indicate statistically significant effects at α

= 0.05.

Emergence rate Oviposition rate Variable T P T P Intercept -1.22 0.2282 -1.37 0.1877 Abundance 4.78 <0.0001* -0.75 0.4644 Diversity 0.59 0.5562 2.20 0.0411* Density -2.76 0.0073* - - Abundance*Diversity -2.71 0.0090* -0.77 0.4534 Abundance*Density -0.73 0.4706 - - Diversity*Density 0.03 0.9794 - - Abundance*Diversity*Density -0.02 0.9836 - -

36

Table 1.3. Multiple linear regressions for the effect of phylum-level bacterial composition, larval density (10, 20, and 40 larvae), and their interaction on Culex pipiens female and male emergence and oviposition rates. Asterisks indicate statistically significant effects at α = 0.05.

Emergence rate Oviposition rate Variable T P T P Intercept 3.80 0.0003* 7.18 <0.0001* α-proteobacteria -1.69 0.0958 -0.36 0.7258 Bacteroidetes 3.86 0.0003* -2.10 0.0580 Firmicutes -1.07 0.2867 0.60 0.5572 β-proteobacteria -2.75 0.0077* 0.93 0.3726 Acidobacteria 1.98 0.0437* 0.21 0.8338 γ-proteobacteria 0.49 0.6263 -0.39 0.7037 Actinobacteria -0.22 0.8273 1.43 0.1788 Density -3.03 0.0035* - - Density*α-proteobacteria 0.08 0.9365 - - Density*Bacteroidetes 2.39 0.0196* - - Density*Firmicutes 1.06 0.2944 - - Density*β-proteobacteria -1.44 0.1560 - - Density*Acidobacteria 0.97 0.3370 - - Density*γ-proteobacteria 0.20 0.8437 - - Density*Actinobacteria -0.67 0.5068 - -

37

CHAPTER 2

Leaf detritus mixtures alter microbial resource composition in aquatic habitats and

performance of container-breeding mosquito larvae (Diptera: Culicidae)

Abstract

The enhancement of animal fitness by resource diversity is a well-documented phenomenon in many ecosystems. In aquatic communities fueled by plant-based detritus, a major source of resource diversity is mixing of leaf detritus species from the terrestrial environment. This study investigated the direct and indirect effects of leaf diversity in mediating the development and adult emergence rates of Culex pipiens (Diptera: Culicidae), a container-breeding mosquito vector for West Nile virus in the northeastern and midwestern United States. Laboratory mosquito development assays and next-generation sequencing of DNA from bacterial and fungal taxa were used to test three hypotheses regarding the effect of leaf detritus inputs from native and invasive shrubs (Amur honeysuckle, Lonicera maackii; autumn olive, Elaeagnus umbellata; blackberry, Rubus allegheniensis; and elderberry, Sambucus canadensis) in the aquatic habitat on Cx. pipiens emergence rates, development time to emergence, and adult body size. First, we found support for the hypothesis that leaf resource diversity in the aquatic larval environment increases emergence rates, increases adult body size, and reduces development time to emergence. Second, we found support for the hypothesis that variation in emergence rates across single and mixed leaf species treatments is related to the microbial community associated with the different leaves and their combinations, in contrast to plant secondary compounds leaching from decaying leaf detritus. Third, we found support for the hypothesis that abundance and composition of bacterial taxa are predictive of mosquito emergence rates and vary predictably

38 among aquatic habitats containing different leaf detritus species. However, we did not observe the same effects among fungal taxa, nor did we find support for the hypotheses that alpha diversity of bacteria or fungi is predictive of emergence rates or varies across detritus species when considered independently of the composition of the microbial community. We conclude that resource diversity generally enhances production of this filter-feeding aquatic macroinvertebrate and the main mechanism underlying variation in mosquito emergence rates due to leaf detritus treatment is the indirect effect of leaf detritus species on the bacterial community.

Introduction

Numerous freshwater macroinvertebrates depend upon leaf detritus from terrestrial plants as a food and energy source (Anderson and Sedell 1979, Wallace et al. 1982, Webster and

Benfield 1986, Graça 2001). Some detritivorous fauna directly feed on coarse organic particulates in the aquatic environment, and the activity of these “shredders” often accelerates the initial decomposition of leaf detritus (McArthur and Barnes 1988, Cuffney et al. 1990). These invertebrates, including amphipods, plecopterans, trichopterans, and some dipterans (Arsuffi and

Suberkropp 1989), generally are selective feeders and preferentially consume leaves that promote growth via high nutrient content (e.g., nitrogen and carbon), low toughness, and low chemical defenses (i.e., plant secondary components) (Arsuffi and Suberkropp 1984, Irons et al.

1988, Bastian et al. 2007). Meanwhile, other aquatic fauna are supported by leaf detritus indirectly. As feeding by shredders and physical abrasion break down leaves, further degradation occurs by the leaching of soluble compounds into the water (Kuiters and Sarink 1986) and decomposition by microbial flora, particularly bacteria and fungi (Fish and Carpenter 1982).

39

Higher-order macroinvertebrate consumers in turn filter microorganisms suspended in the water column or graze on leaf surfaces (Cummins and Klug 1979, Wallace and Merritt 1980) or in some cases the sides of artificial containers (Merritt et al. 1992). While a few filter feeder species have evolved physiological mechanisms to gain some control over their dietary intake (Lehman

1976, Caine 1977), these invertebrates usually are non-selective feeders that indiscriminately consume microbial flora and fine organic particulates present in the aquatic habitat (Jørgensen

1955) and thus their food resource diversity is driven almost exclusively by the amount and type of leaves and associated bacteria and fungi available in the aquatic habitat (Esler et al. 2000).

Container-breeding mosquito larvae are among this latter group of non-selective feeders supported indirectly by detritus from terrestrial plants (Kitching 2001), and differences in quality of leaf detritus for mosquito development are well-established across mosquito species (David et al. 2000, David et al. 2002, Reiskind et al. 2010, Muturi et al. 2012) and aquatic mesocosms

(e.g., tree holes, drainage ditches, storm water catch basins) (Walker et al. 1991, Gardner et al.

2013). Mosquito adult emergence rates are related to the quantity and age of leaf detritus, where large masses of green leaves support higher emergence rates compared to small masses and senescent leaves (Walker et al. 1997). Emergence rates also vary due to species identity of leaf detritus in the aquatic environment. For example, our research showed that certain invasive plant species (e.g., Amur honeysuckle, Lonicera maackii) support higher emergence rates in Culex pipiens (Diptera: Culicidae) compared to native plants (Gardner et al. 2015). This finding contributes to a growing body of research demonstrating variation in the quality of leaf detritus species for development in mosquitoes and other arthropods with aquatic juvenile stages

(Yanoviak 1999, Kominoski et al. 2007), a result with implications for understanding the drivers of the spatial-temporal abundance and distribution of filter-feeding invertebrates.

40

Many aquatic habitats for mosquitoes and other freshwater macroinvertebrates contain leaves from multiple plant species (e.g., Swan and Palmer 2006, Lecerf et al. 2007), and the consequences of resource diversity for habitat quality for these organisms is incompletely understood. Alhough filter feeders generally lack physiological mechanisms to select for a diverse diet and instead extract essential nutrients post-ingestion (Price 1988), the benefits of resource diversity for animal fitness have been established in a variety of terrestrial and aquatic arthropod communities (Moore et al. 2004, Hättenschwiler et al. 2005). Much of this research has focused on the role of food diversity in alleviating interspecific competition between ecologically similar species via resource partitioning (Reiskind et al. 2012), but the positive effect of resource diversity extends to simple systems with a single consumer as well. For example, the co-occurrence of multiple leaf species in both aquatic and terrestrial environments has been linked to more rapid leaf decay rates (Salamanca et al. 1998, Kominoski et al. 2007) and increased growth of microorganisms (Gartner and Cardon 2004), potentially providing additional food to invertebrate detritivores. In mosquito communities, there is some evidence that aquatic mesocosms containing mixtures of two leaf detritus species yield faster development and higher adult emergence rates than expected compared to single leaf species (Reiskind et al.

2012), but results among a limited number of such studies are conflicting (Muturi et al. 2015).

Furthermore, the majority of this work focuses on Aedes species which have different feeding ecologies from Culex species mosquitoes (Yee et al. 2004).

The biological mechanisms underlying leaf detritus species dependent variation in habitat quality for mosquitoes in particular also are underexplored, and there are multiple possible interacting pathways by which leaf detritus may influence adult emergence rates. First, habitat quality may be altered by the plant secondary components that leach into the aquatic

41 environment during decomposition, including compounds that serve as defensive mechanisms against herbivores (Kuiters and Sarink 1986). These toxic phytochemicals may directly affect nervous system function in developing mosquitoes, and numerous larvicidal botanicals have been identified and extracted from plants (Shaalan et al. 2005). Plant secondary components also may act as antimicrobials, reducing the abundance of bacteria and fungi available to developing larvae as food resources (Rios et al. 1988, Silva et al. 2010). Second, habitat quality in response to leaf detritus species present in the aquatic environment may be related to the microbial flora associated with the leaf species. Both the importance of bacteria and fungi to the mosquito diet

(Fish and Carpenter 1982, Walker and Merritt 1988) and the variability in microorganisms that grow on different leaf species (Findlay and Arsuffi 1989, Baldy et al. 1995) are well- documented. Mosquito emergence may be mediated by microbial abundance if all bacteria and fungi are assumed equal in nutritional value to larvae (Kaufman et al. 1999). Alternatively, emergence may be driven by microbial diversity if different microorganisms produce different nutrients and a variety is required for larvae to extract essential dietary components, or by microbial community composition if some microbes produce toxic semiochemicals as byproducts of metabolism that may be deleterious to mosquitoes (Singer 1973, Lacey and

Undeen 1986).

In this study, we tested three hypotheses regarding the effect of resource diversity on adult emergence rates and development of Culex pipiens, an important mosquito vector for West

Nile virus in the northeastern and midwestern United States (Turell et al. 2001, Andreadis et al.

2001). First, we tested the hypothesis that leaf resource diversity in the aquatic larval environment increases adult emergence rates, increases adult body size, and reduces development time to emergence. Second, we tested the hypothesis that variation in emergence

42 rates across single and mixed leaf species is related to the microbial community associated with the different leaves and their combinations, as opposed to plant secondary compounds leaching from the different leaf species. Third, we tested the hypotheses that alpha diversity and abundance and composition of bacterial and fungal operational taxonomic units (OTUs) are predictive of mosquito emergence rates and vary predictably among aquatic habitats containing different leaf detritus species.

Methods

Mosquitoes. Culex pipiens larvae were obtained by collecting egg rafts from five residential sites in Champaign-Urbana, Illinois, using grass infusion-baited oviposition traps (Reiter 1986).

Egg rafts were individually hatched in 24-well plates and Cx. pipiens were distinguished from

Culex restuans based on the presence or absence of a clear area anterior to the sclerotized egg- breaker that is present in Cx. restuans and absent in Cx. pipiens (Dodge 1966). Larvae from different sites were pooled prior to their allocation to experimental treatments.

Leaf detritus species and source. The four focal plant species considered in this study were two exotic and invasive and two native shrubs common within the geographic range of Cx. pipiens

(Darsie and Ward 2004), all of which we used in previous experiments examining the effects of single leaf species on Cx. pipiens emergence rates, development, and oviposition site selection and represent a range of habitat quality (Gardner et al. 2015). The exotic shrubs were Lonicera maackii (Amur honeysuckle, ‘H’) and Elaeagnus umbellata (autumn olive, ‘A’), two highly invasive plants found in both natural and built environments (e.g., forest fragments, landscaped parks, and residential neighborhoods) throughout much of the United States (Deering and Vaknat

43

1999, Orr et al. 2005). The native shrubs were Rubus allegheniensis (blackberry, ‘B’), a bramble that occurs in thickets throughout the eastern and central United States, and Sambucus canadensis (elderberry, ‘E’), a deciduous shrub occurring east of the Rocky Mountains. In our previous studies, honeysuckle leaves supported rapid development and high emergence success of Cx. pipiens, while autumn olive and elderberry leaves yielded moderate adult emergence rates and blackberry leaves were deleterious to mosquito larvae.

Green leaves were collected from each plant species at three sites in Urbana, IL,

(University of Illinois Pollinatarium, Illini Forestry Plantation, and Busey Woods) and pooled prior to their allocation to experimental treatments. We used fresh leaves rather than senescent leaves because while there is no interaction in response between the effect of leaf age and the effect of leaf species on mosquito emergence rates, the magnitude of the main effect of leaf species is larger for green leaves than senescent leaves (Walker et al. 1997). Moreover, in a survey of naturally occurring aquatic mosquito habitats (i.e., storm water catch basins) conducted in Decatur, Illinois, in 2013, we observed high masses of green leaves and low masses of senescent leaves due to the more rapid decomposition of senescent leaves (personal observation).

Experimental design. A completely randomized design was established composed of 15 treatments with five replicates per treatment (for a total of 75 containers). Four treatments consisted of single leaf species (H, A, B, E) as a baseline for comparison of mixture treatments.

Six treatments consisted of mixtures of two leaf species (HA, HB, HE, AB, AE, BE); four treatments consisted of mixtures of three leaf species (HAB, HAE, HBE, ABE); and one treatment was a mixture of all four leaf species (HABE). Infusions were prepared by fermenting a total of 10 g green leaves in closed plastic 2 L buckets filled to the brim with tap water for 7

44 days, a standard infusion age used in comparable studies (e.g., Muturi et al. 2012, Gardner et al.

2015). Thus, for the mixtures of two leaf species, 5 g of each species were infused together; for mixtures of three species, 3.3 g of each species were infused together; and for the mixture of four species, 2.5 g of each species were infused together. After the 7 day period, the leaves and large particulates were strained out of the buckets by passing the infusion through cheesecloth. Using this “whole infusion” with leaf particulates removed, two mosquito development assays were established.

Experiment 1: whole infusion. In the first assay, 320 mL whole infusion were removed from each of the buckets in 360 mL tri-pour beakers, and 20 first instar Cx. pipiens were reared in each beaker. The containers were monitored daily and pupae were transferred individually to cotton- sealed plastic vials with deionized water until all mosquitoes had either died or emerged as adults. The experiment was conducted in a walk-in incubator under ambient conditions of 25ºC,

70% relative humidity, and a 16:8 (L:D) photoperiod. Adults were sorted by sex and date of emergence and their wing lengths determined. Development rate and adult body size are important determinants of vectorial capacity in mosquitoes (Ameneshewa and Service 1996, Alto et al. 2008). Rapid development rate facilitates high emergence success of mosquitoes from ephemeral habitats following rainfall, while adult body size is linearly related to longevity, an important component of vectorial capacity; only a vector that survives the duration of the viral extrinsic incubation period can infect an avian or mammalian host (Bellan 2010).

Experiment 2: filtered infusion. In the second assay, the same experimental conditions and setup were maintained with first instar larvae from the same collection as the first assay. However,

45 instead of supplying the larvae with whole infusion, 320 mL whole infusion were passed through

0.22 μm pore Stericup filters (Fisher Scientific, Hampton NH) powered by a belt drive rotary vacuum pump (Precision Scientific Instruments, Buffalo NY) to remove microorganisms while retaining the plant secondary components associated with the different leaf detritus mixtures.

Every two days over a 20-day period, 0.1 mg ground TetraMin® tropical fish food flakes were added to each container as a uniform diet across all leaf detritus treatments. The fish food contains a mixture of crude protein, crude fat, crude fiber, and vitamins and minerals including phosphorus, Vitamin A, Vitamin B12, ascorbic acid, Vitamin D3, Vitamin E, niacin, inositol, choline, biotin, and omega-3 fatty acids and is used as a standard diet for colony-reared mosquitoes.

Bacterial and fungal composition. Before adding larvae (day 0) to treatments as described above, 15 mL aliquots of 7-day-old whole infusion were taken from each container and stored at

-80°C until further processing. Genomic DNA was extracted from the samples using the

UltraClean® Soil DNA Isolation Kit (Mo Bio Laboratories Inc., Carlsbad CA, Cat. No. 12800-

50) according to the manufacturer’s instructions with the following modifications. Prior to the first step of the protocol, 15 mL infusion samples were concentrated by centrifugation at 5000 rpm for 20 minutes. Next, 14.5 mL supernatant were removed and the pellet was re-suspended in the remaining supernatant by vortexing; these 500 μL aliquots of the sample were added to the bead solution tubes in the first step of the extraction protocol. The fifteenth step of the protocol

(i.e., addition of Solution S4 and centrifugation) was repeated three times to maximize purity of the samples.

The 75 DNA extracts were sequenced using Illumina MiSeq Bulk v3 analysis at the W.

46

M. Keck Center for Comparative and Functional Genomics at the University of Illinois at

Urbana-Champaign. We targeted three hypervariable regions of the 16S rRNA gene for bacterial surveys and the ITS1-ITS4 for fungal surveys using the following primer sets: V1-V3 forward

5’–GAGTTTGATCNTGGCTCAG–3’, reverse 5’–GTNTTACNGCGGCKGCTG–3’; V3-V5 forward 5’–CCTACGGGAGGCAGCAG–3’, reverse 5’–CCGTCAATTCMTTTRAGT–3’; V4 forward 5’–GTGCCAGCMGCCGCGGTAA–3’, reverse 5’–GGACTACHVGGGTWTCTAAT–

3’; and ITS1-ITS4 forward 5’–TTCGTAGGTGAACCTGCGG–3’, reverse 5’–

TCCTCCGCTTATTGATATGC–3’. For Illumina library preparation by Fluidigm protocols,

DNA from each water sample was amplified using the four primer sets added with Fluidigm

Illumina linkers and unique barcodes. The final Fluidigm libraries were pooled for sequencing at the Keck Center’s DNA Services laboratory. The V4 region yielded 8.5 million reads, more than two times the number of reads for the V3-V5 region (3.9 million reads) and more than three times the number of reads for the V1-V3 region (2.6 million reads); therefore the V4 region was used for subsequent analyses. The ITS1-ITS4 region yielded 2.0 million reads.

The IM-TORNADO v2.0.3.2 pipeline (Jeraldo et al. 2014) was used on the University of

Illinois at Urbana-Champaign Biocluster to demultiplex and quality filter both bacterial and fungal forward and reverse reads. IM-TORNADO trims low quality bases using Trimmomatic

(Bolger et al. 2014) before running de novo OTU picking. For this analysis, we used

R1_TRIM=150 as the cut-off length for the R1 read and R2_TRIM=150 as the cut off length for the R2 read. For the taxonomy assignment, the RDP10 database (latest version accessed and downloaded April, 2015) (Cole 2009) was used for bacterial reads and the UNITE ITS database

(latest version accessed and downloaded April, 2015) (Kõljalg et al. 2005) was used for fungal reads. Sampling effort was standardized per sample using rarefaction to normalize read depth to

47

38,773 reads per sample for bacterial reads, and 3,746 reads per sample for fungal reads. The lower number was chosen for fungal reads to retain sufficient sample sizes because read depth was much lower for the ITS amplicons compared to the 16S amplicons per sample. In addition, reads were filtered using the criterion that an OTU must constitute a minimum total observation count of 0.001 (i.e., 0.1 percent) for retention.

Statistical analysis. All statistical analyses were conducted in SAS 9.3 using PROC GLM (SAS

Institute Inc., Cary NC), with the exception of bacterial and fungal community analyses which were conducted using PC-ORD 6 (MjM Software Design, McCune and Mefford 2011) and distance-based regression models which were conducted using the ecodist v1.2.9 package in R

(Goslee and Urban 2013).

Effects of resource diversity. To test the hypothesis that resource diversity in the aquatic habitat increases adult mosquito emergence rates, a general linear model first was fit to the proportion of mosquitoes emerged (males and females combined) given by

푌푖푗푘 = 휇 + 훼푖 + 훽(푖)푗 + 휀(푖푗)푘 (Eqn. 2.1) where 훼푖 represents the main effect of the number of leaf species contained in the treatment (i.e., one, two, three, or four leaf species) and 훽(푖)푗 represents the main effect of the leaf mixture nested within the number of leaf species (e.g., HAB, HAE, HBE, and ABE nested within three leaf species). Tukey’s mean separation test was conducted to detect significant pairwise differences among numbers of leaf species and among leaf mixtures within number of leaf species. If the effect of resource diversity on emergence were additive, we would expect to see no significant effect of number of leaf species, whereas a significant F test would indicate a

48 synergistic or antagonistic effect of leaf species diversity. To further clarify the magnitude and direction of non-additive effects of resource diversity on emergence, linear contrasts were performed using ESTIMATE statements in SAS and a Bonferroni adjustment for multiple comparisons. In brief, we compared the mean emergence rate for a leaf mixture treatment to the combined mean response for its constituent individual leaves. For example, to test the effect of mixing honeysuckle and elderberry leaves for additivity, we conducted a T test for the null hypothesis given by

ℓ = 휇퐻퐸 − (휇퐻 + 휇퐸)/2 = 0 (Eqn. 2.2)

A non-significant contrast indicated an additive effect of mixing leaf species, while a significant positive T statistic indicated a synergistic effect and a significant negative T statistic indicated an antagonistic effect.

To test the hypothesis that resource diversity increases adult body size and reduces the development time to adult emergence, Multivariate Analysis of Variance (MANOVA) tests were fit with average body size (wing length) and average development time per container as the responses and number of leaf species and leaf mixture nested within number of species as predictors. Standardized canonical coefficients (SCCs) were used to describe the relative contributions of development time and wing length to significant main effects. Separate analyses were conducted for male and female mosquitoes, and the need to separate the sexes for examination of these life history traits is the reason emergence rate data were not included in the same multivariate model.

Effects of uniform diet. To test the hypothesis that variation in emergence rates across single and mixed leaf species is related to the microbial community associated with the different leaves and

49 their combinations, we repeated the same statistical analyses described above (Eqn. 1 and Eqn.

2) for the second assay, in which microbes were removed from the leaf infusions and replaced with a uniform diet while retaining the secondary components associated with the different leaf detritus mixtures. If plant secondary compounds were the main driver of emergence rates, we would expect to see a significant effect of leaf mixture within leaf species. If the microbial community were related to emergence rates, we would expect to see a reduction in magnitude or no significance of the main effect of leaf mixture.

Effects of bacteria and fungi on emergence. To test the hypothesis that diversity of bacterial

OTUs is predictive of mosquito emergence rates, the Shannon index was estimated for each replicate container and a simple least squares linear regression model was fit to the proportion of mosquitoes emerged (males and females combined) with this diversity measure as the single predictor. This model ignored the identity of the leaf species associated with each replicate and simply related emergence rates to bacterial diversity. Because fungal and bacterial sequencing were carried out using different amplicons and thus cannot be combined for downstream analyses, the same analysis was repeated separately to test the hypothesis that diversity of fungal

OTUs is linearly related to emergence rates.

To test the hypothesis that abundance and composition of bacterial OTUs are predictive of mosquito emergence rates, non-metric multidimensional scaling (NMDS) first was conducted with abundances of unique bacterial OTUs as the raw explanatory variables (Kruskal 1964) and a

Bray-Curtis dissimilarity index which takes into account both the presence/absence and the relative abundance of species. Ordination is a commonly used dimension reduction procedure in community ecology designed to circumvent the problem that many statistical techniques used for

50 hypothesis testing (e.g., regression, MANOVA) cannot handle high-dimensional data matrices containing many zeroes (Legendre and Legendre 1998) and become difficult to interpret in the presence of multicollinearity (Graham 2003), both typical of species abundance data. NMDS in particular is advantageous for community analyses compared to other ordination techniques (e.g., principal components analysis, canonical correspondence analysis) because it does not assume linear relationships among variables, its use of ranked distances makes the method appropriate to small and non-normal data sets, and it allows the use of any distance matrix of dissimilarities for viewing underlying data structures (McCune et al. 2002, Ramette 2007). The number of NMDS axes was selected based on a stress test, using the criterion that the addition of an axis must reduce stress by at least five. Next, a distance-based regression model was fit with the proportion of mosquitoes emerged (males and females combiend) as the response and the predictors contained in the NMDS distance matrix. This non-parametric approach provides permutational tests of significance for regression coefficients (Legendre et al. 1994, Lichstein 2007), and the test was conducted with 1000 iterations. The same analyses were repeated separately to test the hypothesis that abundance and composition of fungal OTUs are predictive of mosquito emergence rates.

Bacterial and fungal community analyses. If characteristics of the microbial community were found predictive of emergence rates, our next objective was to describe the variation in bacterial and fungal flora associated with the different leaf detritus treatments. To test the hypothesis that diversity of bacterial OTUs varies among aquatic habitats containing different leaf detritus species, the linear model including the number of leaf species and leaf mixture nested within number of leaf species (Eqn. 2.1) was fit with the Shannon index for each container as the

51 response. Tukey’s mean separation test was conducted to detect significant pairwise differences among numbers of leaf species and among leaf mixtures within number of leaf species. The same analysis was repeated separately to compare the diversity of fungal OTUs across leaf detritus treatments.

To test the hypothesis that abundance and composition of bacterial OTUs vary predictably among aquatic habitats containing different leaf detritus species, multi-response permutation procedure (MRPP) tests were used to test for differences among groups established a priori based on leaf detritus treatments (Miekle 1976). MRPP is a non-parametric analog to

MANOVA, in which the average within-group distance in multidimensional space is calculated for the specified groups, then all experimental replicates are randomly assigned to new groups and the average within-group distances calculated again over a series of iterations. The statistical significance of the observed average within-group distances is determined by comparing the observed test statistic to the distribution of within-group distances generated by the randomization test. Separate MRPP tests with 1000 iterations were conducted comparing bacterial community composition and structure among treatments containing different numbers of leaf species and among leaf mixtures within a given number of species, with a Bonferroni adjustment for multiple comparisons. These analyses were repeated separately to compare the abundance and composition of fungal OTUs across leaf detritus treatments.

Finally, to identify key bacterial OTUs that are highly characteristic of (i.e., both faithful to and exclusive to) particular leaf detritus treatments, indicator species analysis (ISA) was carried out comparing the single leaf species treatments (Dufrêne and Legendre 1997). The goal of this analysis was to identify taxa that may be responsible for strong effects on emergence rates such as the deleterious effect of blackberry leaves on mosquito larvae (Gardner et al. 2015). A

52 perfect indicator score of 100 indicates that a taxon occurs in all samples for a given treatment and occurs in no samples for any other treatments; here, we required a cutoff indicator value of

80 for a bacterial OTU to be considered an indicator taxon for a leaf species. The same analysis was repeated separately to identify indicator species among fungal OTUs across leaf detritus species.

Results

Non-additive effects of resource diversity. Emergence rates varied within and among number of leaf species contained in treatments, with significant main effects of number of leaf species (F

= 34.08; df = 3, 58; P < 0.01) and mixture nested within number of species (F = 42.16; df = 11,

58; P < 0.01). On average, the treatment containing all four leaf species yielded the highest emergence rate, followed by treatments containing three leaf species, with no significant difference between treatments containing one versus two leaf species (Figure 2.1a). While we found variation in emergence among the single leaf species and two species mixtures, with honeysuckle leaves supporting the highest emergence rates (0.89 ± 0.04) and blackberry leaves yielding no survival (0 ± 0.04), there were no significant pairwise differences between leaf mixtures containing three species regardless of the identity of the three species (Figure 2.1b-e).

Further, linear contrasts demonstrated a synergistic emergence response to mixtures of three and four leaf species (with the exception of the additive HAB mixture), but more variability in responses to mixtures of two leaf species (Table 2.1). Among the two leaf species mixtures, three treatments (HB, AE, and BE) had antagonistic effects on mosquito emergence, while all other treatments had additive effects.

Development time to emergence and adult wing length also varied within and among

53 number of leaf species contained in treatments, with significant main effects of number of leaf species and mixture nested within number of species for males and females (Table 2.2). On average the treatments containing three and four leaf species yielded more rapid development and larger adult sizes compared to the treatments containing one and two leaf species in males and females (Figure 2.2). Among the single leaf species, the most rapid development time was observed in honeysuckle for males (7.73 ± 0.24 days) and females (8.93 ± 0.33 days) and the slowest development time was observed in elderberry for males (9.97 ± 0.24 days) and females

(10.85 ± 0.33 days) (Figure 2.2b). Similarly, the largest wing lengths were observed in honeysuckle for males (3.85 ± 0.04 mm) and females (4.37 ± 0.04 mm), while the shortest wing lengths were observed in autumn olive for males (3.70 ± 0.04 mm) and elderberry for females

(4.02 ± 0.04 mm). Here, linear contrasts were not conducted to test for additive effects of leaf mixtures due to lack of survivors among larvae reared in the blackberry leaf infusion.

Role of microbial flora in habitat quality. Following filtering of whole infusion to remove the bacteria and fungi and addition of a uniform larval diet across all treatments, emergence rates did not vary within or among number of leaf species contained in treatments, with non-significant main effects of number of leaf species (F = 0.02; df = 3, 60; P = 0.99) and mixture nested within number of species (F = 1.38; df = 11, 60; P = 0.21). Similarly, there were no significant main effects of number of leaf species and mixture on development time to emergence or wing length for males or females.

Bacterial and fungal diversity and emergence. Bacterial and fungal OTUs were profiled across 15 leaf detritus mixtures. After rarefaction, quality filtering, and criteria filtering of raw

54 reads, 227 OTUs were retained for bacterial (16S V4) amplicons and 142 OTUs were retained for fungal (ITS1-ITS4) amplicons. Bacterial taxa were dominated by the phyla Bacteroidetes,

Firmicutes, and Proteobacteria while fungal taxa were dominated by the phylum Ascomycota.

Bacterial diversity (computed using the Shannon index) alone was not a significant predictor of adult mosquito emergence rates (F = 1.33; df = 1, 68; P = 0.19) and did not vary within or among number of leaf species contained in detritus treatments, with non-significant main effects of number of leaf species (F = 0.61; df = 3, 55; P = 0.61) and mixture nested within number of leaf species (F = 1.11; df = 11, 55; P = 0.37). Similarly, fungal diversity was not a significant predictor of adult mosquito emergence rates (F = 0.11; df = 1, 60; P = 0.74) and did not vary within or among number of leaf species contained in detritus treatments, with non- significant main effects of number of leaf species (F = 2.85; df = 3, 47; P = 0.06) and mixture nested within number of leaf species (F = 1.38; df = 11, 47; P = 0.21).

Bacterial and fungal composition and emergence. Ordination using NMDS reduced 227 unique bacterial OTUs to three axes. NMDS axis 1 explained 35 percent of variation in the distance matrix, axis 2 explained 20 percent, and axis 3 explained 15 percent. Permutational distance-based regression showed the bacterial distance matrix was predictive of emergence rates

(F = 42.75; P < 0.01). Similarly, ordination using NMDS reduced 142 unique fungal OTUs to three orthogonal axes, where NMDS axis 1 explained 22 percent of variation in the distance matrix, axis 2 explained 27 percent, and axis 3 explained 31 percent. In contrast to the bacteria, however, scores extracted from the fungal NMDS axes for each replicate were not predictive of emergence rates (F = 10.21; P = 0.10). Thus, no additional analyses were carried out related to the fungal community.

55

Bacterial abundance and composition varied within and among number of leaf species contained in treatments, with significant main effects of number of leaf species (T = -11.17; P <

0.001) and mixture nested within number of species (T = -19.46; P < 0.001). Significant pairwise differences between treatments based on MRPP analysis generally mirrored those observed in comparisons of mosquito emergence rates (Table 2.3). No significant differences were found between the centroids for the treatments containing all four leaf species compared to three leaf species, and similarly no significant differences were found between the treatments containing two leaf species compared to single leaf species, although the three/four-leaf species and one/two-leaf species groups were different from each other (Figure 2.3a). While we found variation in bacterial composition among the single leaf species and two species mixtures, there were no significant pairwise differences in bacterial composition between leaf mixtures containing three species or four species regardless of the identity of the species (with the exception of the ABE mixture which was significantly different from all other three species mixtures; Figure 2.3b-d).

ISA identified bacterial OTUs that are highly characteristic of (i.e., both consistent with and exclusive to) particular leaf detritus treatments, and taxa with indicator scores greater than

80 for each single leaf species are tabulated in Table 2.4 (two taxa for honeysuckle, two taxa for autumn olive, 12 taxa for elderberry, and six taxa for blackberry). In particular, the highest scoring indicator taxa for each treatment were as follows: honeysuckle – Anaerovorax sp.

(Firmicutes, Clostridia), Comamonas sp. (Proteobacteria, Betaproteobacteria); autumn olive –

Dyadobacter (Bacteroidetes, Cytophagia), Acidovorax (Proteobacteria, Betaproteobacteria); blackberry – Paludibacter (Bacteroidetes, Bacteroidia), Flavonifractor sp. (Firmicutes,

Clostridia), Acidovorax sp. (Proteobacteria, Betaproteobacteria); elderberry – Sporobacter sp.

56

(Firmicutes, Clostridia), Erwinia sp. (Proteobacteria, Gammaproteobacteria), and Clostridium sp.

(Firmicutes, Clostridia).

Discussion

In this study, we tested three hypotheses regarding the effect of resource diversity on adult emergence rates and development of Cx. pipiens mosquitoes. First, we found support for the hypothesis that leaf resource diversity in the aquatic larval environment increases emergence rates, increases adult body size (wing length), and shortens the development time to emergence.

Second, we found support for the hypothesis that variation in emergence rates across single and mixed leaf species treatments is related to the microbial community associated with the different leaves and their combinations, in contrast to plant secondary compounds leaching from decaying leaf detritus. Third, we found support for the hypothesis that abundance and composition of bacterial taxa are predictive of mosquito emergence rates and vary predictably among aquatic habitats containing different leaf detritus species. However, we did not observe the same effects among fungal taxa, nor did we find support for the hypotheses that diversity per se of bacteria and fungi are predictive of emergence rates or that diversity varies across detritus species when considered independently of the composition of the microbial community.

Our study contributed to a well-established body of literature suggesting that leaf detritus species vary in quality for mosquito development and adult emergence (Yanoviak 1999, Yee and

Juliano 2006, Reiskind et al. 2010, Muturi et al. 2012), and confirmed our previously reported results that Amur honeysuckle leaves are a high-quality resource for Cx. pipiens larvae while blackberry leaves are deleterious to this species (Gardner et al. 2015). While numerous studies have addressed the effects of single leaf species on mosquito performance, the aquatic

57 environments inhabited by larvae in naturally-occurring container habitats (e.g., ditches, used tires, tree holes, storm water catch basins) typically contain mixtures of a variety of leaf detritus species and comparatively few experiments mostly with Aedes mosquitoes have examined the impact of leaf detritus mixtures on mosquito performance. Further, these limited studies often have focused on mixtures of only two leaf species, potentially yielding biased conclusions regarding the impact of leaf resource diversity if one of the leaf species selected was highly nutritious. For example, Reiskind et al. (2012) found a strong synergistic effect of mixing leaf detritus species on Aedes aegypti and Ae. albopictus emergence rates, while Muturi et al. (2015) found no support for the hypothesis that resource diversity results in non-additive effects on production of adult Ochlerotatus triseriatus, Oc. japonicus, and Cx. restuans.

In our current study, we deliberately selected leaf species representing a range of qualities for Cx. pipiens development and a larger number of leaf combination treatments than has been considered in prior work (four single leaf species, six two-leaf species mixtures, four three-leaf mixtures, and one four-leaf mixture). We observed the general trend that increasing leaf diversity in the aquatic environment yields higher adult emergence rates, larger adult body sizes, and more rapid development time to emergence compared to single leaf species, consistent with the result of Reiskind et al. (2012). We also used linear contrasts to evaluate the nature of the effect of number of leaf species on performance of larvae. We found mixtures of three leaf species yielded statistically equal emergence rates and consistently produced a synergistic effect compared to the mean emergence rate for the constituent single leaf species, regardless of the identity of the particular leaves. However, in mixtures of two leaf species we observed more variability in the emergence response. While the majority of two leaf mixtures produced an additive effect, the mixtures containing blackberry leaves often yielded an antagonistic effect on emergence,

58 indicating that addition of a higher-quality resource did not offset the deleterious effect of the blackberry leaves.

The design of our experiment was motivated in part by an interest in identifying whether the primary mechanism underlying leaf species dependent variation in mosquito adult emergence rates, and in particular the deleterious effect of some leaves such as blackberry, is related to the nutritional quality of the microbial communities associated with the different leaf species versus plant secondary compounds that leach from the leaves as they decay and may be toxic to larvae

(Kuiters and Sarink 1986). In one iteration of the development assay, infusions were filtered to remove the bacteria and fungi associated with different leaf detritus species while retaining the chemical background of those treatments, and the microbial flora were replaced with a uniform diet as a food supply. In these assays, we found no variation in emergence rates within or among number of leaf species contained in the treatments, which we interpret as an indication that the emergence response to different leaf detritus mixtures is not driven primarily by plant secondary compounds and may be related to the microbial community; if plant secondary compounds were the main driver of emergence, we would have expected to continue to see a significant effect of leaf mixture under these experimental conditions. This observation was consistent with the findings of other studies that suggest the importance of microbial flora to the fitness of filter- feeding invertebrates (Fish and Carpenter 1982, Walker and Merritt 1988).

We conducted a correlational analysis to assess characteristics of the bacterial and fungal communities that may be predictive of adult mosquito emergence rates. Initially we hypothesized that diversity of bacteria and fungi are predictive of emergence rates, when considered independently of the composition of microbial taxa, based on the reasoning that different microorganisms produce different nutrients and filter-feeding aquatic macroinvertebrates require

59 resource variety to extract essential dietary components. We did not find support for this hypothesis in either bacteria or fungi. We also did not find support for the hypothesis that fungal composition varies among leaf detritus treatments and is predictive of Cx. pipiens emergence rates, but we did find support for the hypothesis that bacterial composition is predictive of emergence rates. The relative importance of bacteria compared to fungi most likely reflects a substantially greater abundance of bacteria compared to fungi across all detritus treatments.

Multivariate community analysis metrics first were used to compare the bacterial flora associated with the four individual leaf species and then to compare the composition of bacterial taxa among the different leaf mixtures in an effort to clarify the ecological mechanisms underlying variation in emergence rates. Pairwise differences in observed emergence rates generally reflected pairwise differences in bacterial composition; again we found that bacterial composition did not vary substantially among mixtures of three or four leaf species, although significant pairwise differences were found among the four single leaf species and some mixtures of two leaf species. Although a few other studies have characterized the microbial flora associated with aquatic mosquito habitats (Ponnusamy et al. 2008, Duguma et al. 2013, Muturi et al. 2013), none have attempted to relate bacterial community composition and structure to emergence outcomes.

The deleterious blackberry leaf treatment was of particular interest to us for potential downstream mosquito control applications in aquatic habitats (e.g., storm water catch basins), and key taxa differentiating this treatment from the other single leaf species included taxa belonging to the phyla Proteobacteria and Firmicutes (particularly the class Clostridia). This result was consistent with our previous experiment that showed relative abundance of the class

Betaproteobacteria is negatively correlated to emergence rates of Cx. pipiens (Gardner et al.

60

2015). However, we also found that members of the same groupings were indicator taxa for honeysuckle and autumn olive treatments, in which mosquito performance was substantially better. This result indicates that the phylum or class taxonomic level may not be the appropriate taxonomic grouping at which to attempt to draw correlations with mosquito emergence, and a higher degree of taxonomic detail is needed to identify bacteria that may be exploited for vector control strategies. Additional studies are required to identify the particular bacterial taxa that may be involved in determining Cx. pipiens emergence, and this will be a challenging avenue to pursue experimentally because it is unknown whether these bacteria are culturable.

We wish to note that there are multiple possible explanations for the patterns we observed in mosquito emergence rates in response to leaf detritus mixture. First, we did not allow the mass of leaves to vary between container replicates, only the proportion of leaves belonging to each species, so it is impossible to rule out the interpretation that the 10 g of blackberry leaves in the single species treatment and the 5 g of blackberry leaves in the two leaf species treatments were sufficient to inhibit larval growth whereas the 3.3 g of leaves in the three species treatments and the 2.5 g of leaves in the four leaf species treatment were not. Second, it may be that we consistently observed the synergistic effect of mixing three or four leaf species as an artifact of the length of time for which infusion was prepared prior to allocation of larvae to experimental replicates. We argue that the infusion age considered in this study is biologically relevant, because experiments conducted in the field have demonstrated that aquatic mesocosms containing leaf detritus generally become attractive to Cx. pipiens for oviposition after approximately seven days of leaf decay (Gardner et al. 2015). However, it has been shown that combinations of leaves decompose more rapidly than single leaf species (Salamanca et al. 1998,

Kominoski et al. 2007), and perhaps given a longer fermentation time, the single and two leaf

61 species mixtures would have yielded equal emergence rates compared to the three and four leaf species mixtures.

Further, although our results provide experimental support for the concept that the microbial community does have a function, albeit one that is incompletely understood, in mediating leaf species dependent variation in development and emergence of Cx. pipiens, our findings are subject to a number of experimental limitations and thus should be interpreted with caution. The fish food diet supplied to the developing larvae in lieu of the microbial flora is a highly nutritious resource for mosquitoes, containing proteins and vitamin supplements that are not necessarily characteristic of aquatic communities fueled by plant detritus (Yee and Juliano

2006). There is evidence that a high-quality diet may enable some aquatic macroinvertebrates to persist in a toxic environment (David et al. 2000), and Cx. pipiens in particular is well-known to tolerate poor quality aquatic habitats; in fact this mosquito is an indicator species commonly used for water quality assessment to identify polluted environments (Gaufin and Tarzwell 1952).

However, at least one study of container-breeding mosquitoes used path analysis to demonstrate that the potentially toxic effects of leaf detritus itself may be offset by the indirect effect of bacterial production by decomposing leaves in Ae. albopictus and Oc. triseriatus (Yee et al.

2007), so we argue that the addition of fish food in our experiment may not represent a strong departure from food dynamics in naturally-occurring aquatic environments. Based on our current experimental design, we also cannot evaluate the possibilities that the leaves leach antimicrobial compounds, reducing the overall abundance of bacteria and fungi available to larvae as food resources (Rios et al. 1988, Silva et al. 2010), that microbes themselves produce semiochemicals that are toxic to mosquitoes as byproducts of metabolism (Singer 1973, Lacey and Undeen

1986), or that variation in ion concentration and ratios (e.g., carbon and nitrogen) across leaf

62 detritus treatments is responsible for some leaf species supporting a greater abundance of microbial flora than others (Walker et al. 1997).

The current study has provided additional evidence that the adult emergence success of container-breeding mosquitoes depends largely upon the quality of the aquatic environment, and a high diversity of leaf detritus resources may yield an altered microbial community that better supports the development and eventual survival of Cx. pipiens larvae. Given that filter-feeding macroinvertebrates generally have little choice in the resources they consume so are beholden to the quality of their environment, a logical corollary question is whether leaf detritus and microbial flora have a similar role in mediating mosquito oviposition site selection. Optimal oviposition theory suggests that organisms should interpret cues to choose the highest-quality habitat for the development of progeny (Gripenberg et al. 2010). However, our previous research which focused strictly on leaf detritus species and did not address the bacterial community with a high level of taxonomic resolution suggested that in fact mosquitoes are not always capable of recognizing the optimal habitat for oviposition (Gardner et al. 2015), and other studies have linked semiochemicals produced by the bacterial flora as byproducts of metabolism to habitat selection by gravid female Ae. aegypti and Ae. albopictus (Ponnusamy et al. 2008, Ponnusamy et al. 2010). Future experimental studies should address whether Cx. pipiens are more likely to choose habitats containing multiple leaf species for oviposition, reflecting the high quality of these habitats, and investigate potential mechanisms for any deviations from this expectation including characteristics of the microbial community associated with aquatic habitats in order to more fully capture the complex ecology of mosquito production in variable environments.

63

Acknowledgments

We thank Jackie Duple, Noor Malik, Chang-Hyun Kim, Millon Blackshear, Madhu Siddappaji,

Brandon Lieberthal, and Stephen House for their technical assistance and Allison Hansen for her advice on sample size. We also thank the DNA Services laboratory at the W. M. Keck Center for

Comparative and Functional Genomics at the University of Illinois at Urbana-Champaign for sequencing services and Ravikiran Donthu, Christopher Wright, and the High Performance

Biological Computing (HPCBio) group for preparation of the bacterial and fungal OTU tables and instruction in use of the IM-TORNADO pipeline. We gratefully acknowledge Elizabeth

Bach, Angela Kent, Anthony Yannarell, and Alexander Alekseyenko, Paul Joseph McMurdie, and students and staff of the University of Washington Summer Institute in Statistics and

Modeling of Infectious Diseases for advice on downstream analysis of the metagenomic data.

We thank members of the Allan lab for their thoughtful comments on the manuscript draft. This study was supported by a U.S. Environmental Protection Agency (EPA) STAR Graduate

Fellowship (award no. F13F11066) to AMG, a University of Illinois Institute for Sustainability,

Energy, and Environment grant to BFA, the Used Tire Fund and Emergency Public Health Act from the State of Illinois to EJM.

References Cited

Anderson NH, Sedell JR. 1979. Detritus processing by macroinvertebrates in stream ecosystems.

Annual Review of Entomology, 24, 351-377.

Andreadis TG, Anderson JF, Vossbrinck CR. 2001. Mosquito surveillance for West Nile virus in

Connecticut, 2000: Isolation from Culex pipiens, Cx. restuans, Cx. salinarius, and

Culiseta melanura. Emerging Infectious Diseases, 7, 670.

64

Arsuffi TL, Suberkropp K. 1984. Leaf processing capabilities of aquatic hyphomycetes:

interspecific differences and influence on shredder feeding preferences. Oikos, 144-154.

Arsuffi TL, Suberkropp K. 1989. Selective feeding by shredders on leaf-colonizing stream fungi:

comparison of macroinvertebrate taxa. Oecologia, 79, 30-37.

Baldy V, Gessner MO, Chauvet E. 1995. Bacteria, fungi and the breakdown of leaf litter in a

large river. Oikos, 93-102.

Bastian M, Boyero L, Jackes BR, Pearson RG. 2007. Leaf litter diversity and shredder

preferences in an Australian tropical rain-forest stream. Journal of Tropical Ecology, 23,

219-229.

Bolger AM, Lohse M, Usadel B. 2014. Trimmomatic: a flexible trimmer for Illumina sequence

data. Bioinformatics, btu170.

Caine EA. 1977. Feeding mechanisms and possible resource partitioning of the Caprellidae

(Crustacea: Amphipoda) from Puget Sound, USA. Marine Biology, 42, 331-336.

Cole JR, Wang Q, Cardenas E, Fish J, Chai B, Farris RJ, Kulam-Syed-Mohideen AS, McGarrell

DM, Marsh T, Garrity GM, Tiedje JM. 2009. The Ribosomal Database Project: improved

alignments and new tools for rRNA analysis. Nucleic Acids Research, 37, 141-5.

Cuffney TF, Wallace JB, Lugthart GJ. 1990. Experimental evidence quantifying the role of

benthic invertebrates in organic matter dynamics of headwater streams. Freshwater

Biology, 23, 281-299.

Cummins KW, Klug MJ. 1979. Feeding ecology of stream invertebrates. Annual Review of

Ecology and Systematics, 147-172.

David JP, Rey D, Pautou MP, Meyran JC. 2000. Differential toxicity of leaf litter to dipteran

larvae of mosquito developmental sites. Journal of Invertebrate Pathology, 75, 9-18.

65

David JP, Rey D, Cuany A, Bride JM, Meyran JC. 2002. Larvicidal properties of decomposed

leaf litter in the subalpine mosquito breeding sites. Environmental Toxicology and

Chemistry, 21, 62-66.

Dufrêne M, Legendre P. 1997. Species assemblages and indicator species: the need for a flexible

asymmetrical approach. Ecological Monographs, 67, 345-366.

Duguma D, Rugman-Jones P, Kaufman MG, Hall MW, Neufeld JD, Stouthamer R, Walton WE.

2013. Bacterial communities associated with Culex mosquito larvae and two emergent

aquatic plants of bioremediation importance. PLoS One, 8, e72522.

Elser JJ, Fagan WF, Denno RF, Dobberfuhl DR, Folarin A, Huberty A, Interlandi S. 2000.

Nutritional constraints in terrestrial and freshwater food webs. Nature, 408, 578-580.

Findlay SE, Arsuffi TL. 1989. Microbial growth and detritus transformations during

decomposition of leaf litter in a stream. Freshwater Biology, 21, 261-269.

Fish D, Carpenter SR. 1982. Leaf litter and larval mosquito dynamics in tree-hole ecosystems.

Ecology, 283-288.

Gardner AM, Anderson TK, Hamer GL, Johnson DE, Varela KE, Walker ED, Ruiz MO. 2013.

Terrestrial vegetation and aquatic chemistry influence larval mosquito abundance in catch

basins, Chicago, USA. Parasites and Vectors, 6, 9.

Gardner AM, Allan BF, Frisbie LA, Muturi EJ. 2015. Asymmetric effects of native and exotic

invasive shrubs on ecology of the West Nile Virus vector Culex pipiens (Diptera:

Culicidae). Parasites and Vectors, 16, 329.

Gartner TB, Cardon ZG. 2004. Decomposition dynamics in mixed‐species leaf litter. Oikos, 104,

230-246.

66

Gaufin AR, Tarzwell CM. 1952. Aquatic invertebrates as indicators of stream pollution. Public

Health Reports, 67, 57.

Goslee SC, Urban DL. 2007. The ecodist package for dissimilarity-based analysis of ecological

data. Journal of Statistical Software, 22, 1-19.

Graça MA. 2001. The role of invertebrates on leaf litter decomposition in streams: a review.

International Review of Hydrobiology, 86, 383-393.

Graham MH. 2003. Confronting multicollinearity in ecological multiple regression. Ecology, 84,

2809-2815.

Gripenberg S, Mayhew PJ, Parnell M, Roslin T. 2010. A meta‐analysis of preference-

performance relationships in phytophagous insects.Ecology Letters, 13, 383-393.

Hättenschwiler S, Tiunov AV, Scheu S. 2005. Biodiversity and litter decomposition in terrestrial

ecosystems. Annual Review of Ecology, Evolution, and Systematics, 191-218.

Irons III JG, Oswood MW, Bryant JP. 1988. Consumption of leaf detritus by a stream shredder:

influence of tree species and nutrient status. Hydrobiologia, 160, 53-61.

Jeraldo P, Kalari K, Chen X, Bhavsar J, Mangalam A, White B, Nelson H, Kocher JP, Chia N.

2014. IM-TORNADO: A tool for comparison of 16S reads from paired-end

libraries. PLoS One, 9, e114804.

Jørgensen CB. 1955. Quantitative aspects of filter feeding in invertebrates. Biological Reviews,

30, 391-453.

Kaufman MG, Walker ED, Smith TW, Merritt RW, Klug MJ. 1999. Effects of larval mosquitoes

(Aedes triseriatus) and stemflow on microbial community dynamics in container habitats.

Applied and Environmental Microbiology, 65, 2661-2673.

67

Kitching RL. 2001. Food webs in phytotelmata: “bottom-up” and “top-down” explanations for

community structure. Annual Review of Entomology, 46, 729-760.

Kõljalg U, Larsson KH, Abarenkov K, Nilsson RH, Alexander IJ, Eberhardt U, Erland S,

Høiland K, Kjøller R, Larsson E, Pennanen T. 2005. UNITE: a database providing web‐

based methods for the molecular identification of ectomycorrhizal fungi. New

Phytologist, 166, 1063-1068.

Kominoski JS, Pringle CM, Ball BA, Bradford MA, Coleman DC, Hall DB, Hunter MD. 2007.

Nonadditive effects of leaf litter species diversity on breakdown dynamics in a detritus-

based stream. Ecology, 88, 1167-1176.

Kruskal JB. 1964. Nonmetric multidimensional scaling: a numerical method. Psychometrika, 27,

115-29.

Kuiters AT, Sarink HM. 1986. Leaching of phenolic compounds from leaf and needle litter of

several deciduous and coniferous trees. Soil Biology and Biochemistry, 18, 475-480.

Lacey LA, Undeen AH. 1986. Microbial control of black flies and mosquitoes. Annual Review of

Entomology, 31, 265-296.

Lecerf A, Risnoveanu G, Popescu C, Gessner MO, Chauvet E. 2007. Decomposition of diverse

litter mixtures in streams. Ecology, 88, 219-227.

Legendre P, Lapointe F, Casgrain P. 1994. Modeling brain evolution from behavior: A

permutational regression approach. Evolution, 48, 1487-1499.

Legendre P, Legendre L. 1998. Numerical Ecology, Volume 24.

Lehman JT. 1976. The filter‐feeder as an optimal forager, and the predicted shapes of feeding

curves. Limnology and Oceanography, 21, 501-516.

68

Lichstein J. 2007. Multiple regression on distance matrices: A multivariate spatial analysis tool.

Plant Ecology, 188, 117-131.

McArthur JV, Barnes JR. 1988. Community dynamics of leaf litter breakdown in a Utah alpine

stream. Journal of the North American Benthological Society, 37-43.

McCune B, Mefford MJ. 2011. PC-ORD. Multivariate Analysis of Ecological Data. Version 6.

MjM Software, Gleneden Beach, Oregon.

McCune B, Grace JB, Urban DL. 2002. Analysis of Ecological Communities (Vol. 28). Gleneden

Beach, OR: MjM software design.

Merritt RW, Dadd RH, Walker ED. 1992. Feeding behavior, natural food, and nutritional

relationships of larval mosquitoes. Annual Review of Entomology, 37, 349-374.

Mielke Jr. PW, Berry KJ, Johnson ES. 1976. Multi-response permutation procedures for a priori

classifications. Communications in Statistics Theory and Methods, 5, 1409-1424.

Moore JC, Berlow EL, Coleman DC, Ruiter PC, Dong Q, Hastings A, Collins Johnson N. 2004.

Detritus, trophic dynamics and biodiversity. Ecology Letters, 7, 584-600.

Muturi EJ, Allan BF, Ricci J. 2012. Influence of leaf detritus type on production and longevity of

container-breeding mosquitoes. Environmental Entomology, 41, 1062-1068.

Muturi EJ, Orindi BO, Kim CH. 2013. Effect of leaf type and pesticide exposure on abundance

of bacterial taxa in mosquito larval habitats. PLoS One, 8, 8.

Muturi EJ, Gardner AM, Bara JJ. 2015. Impact of an alien invasive shrub on ecology of native

and alien invasive mosquito species (Diptera: Culicidae). Environmental Entomology, 44,

1308-1315.

Ponnusamy L, Xu N, Stav G, Wesson DM, Schal C, Apperson CS. 2008. Diversity of bacterial

communities in container habitats of mosquitoes. Microbial Ecology, 56, 593-603.

69

Ponnusamy L, Xu N, Nojima S, Wesson DM, Schal C, Apperson CS. 2008. Identification of

bacteria and bacteria-associated chemical cues that mediate oviposition site preferences

by Aedes aegypti. Proceedings of the National Academy of Sciences, 105, 9262-9267.

Ponnusamy L, Wesson DM, Arellano C, Schal C, Apperson CS. 2010. Species composition of

bacterial communities influences attraction of mosquitoes to experimental plant

infusions. Microbial Ecology 59, 158-173.

Price HJ. 1988. Feeding mechanisms in marine and freshwater zooplankton. Bulletin of Marine

Science, 43, 327-343.

Ramette A. 2007. Multivariate analyses in microbial ecology. FEMS Microbiology Ecology, 62,

142-160.

Reiskind MH, Zarrabi AA, Lounibos LP. 2010. Invasive leaf resources alleviate density

dependence in the invasive mosquito, Aedes albopictus. Biological Invasions, 12, 2319-

2328.

Reiskind MH, Zarrabi AA, Lounibos LP. 2012. Effects of combination of leaf resources on

competition in container mosquito larvae. Bulletin of Entomological Research, 102, 424-

434.

Rios JL, Recio MC, Villar A. 1988. Screening methods for natural products with antimicrobial

activity: a review of the literature. Journal of Ethnopharmacology, 23, 127-149.

Salamanca EF, Kaneko N, Katagiri S. 1998. Effects of leaf litter mixtures on the decomposition

of Quercus serrata and Pinus densiflora using field and laboratory microcosm methods.

Ecological Engineering, 10, 53-73.

70

Shaalan EAS, Canyon D, Younes MWF, Abdel-Wahab H, Mansour AH. 2005. A review of

botanical phytochemicals with mosquitocidal potential. Environment International, 31,

1149-1166.

Silva NCC, Fernandes Jr. A. 2010. Biological properties of medicinal plants: a review of their

antimicrobial activity. Journal of Venomous Animals and Toxins Including Tropical

Diseases, 16, 402-413.

Singer S. 1973. Insecticidal activity of recent bacterial isolates and their toxins against mosquito

larvae. Nature, 110-111.

Swan CM, Palmer MA. 2006. Composition of speciose leaf litter alters stream detritivore

growth, feeding activity and leaf breakdown. Oecologia, 147, 469-478.

Tu YK, Kellett M, Clerehugh V, Gilthorpe MS. 2005. Problems of correlations between

explanatory variables in multiple regression analyses in the dental literature. British

Dental Journal, 199, 457-461.

Turell MJ, O’Guinn ML, Dohm DJ, Jones JW. 2001. Vector competence of North American

mosquitoes (Diptera: Culicidae) for West Nile virus. Journal of Medical Entomology, 38,

130-134.

Walker ED, Merritt RW. 1988. The significance of leaf detritus to mosquito (Diptera: Culicidae)

productivity from treeholes. Environmental Entomology, 17, 199-206.

Walker ED, Merritt RW, Kaufman MG, Ayres MP, Riedel MH. 1997. Effects of variation in

quality of leaf detritus on growth of the eastern tree-hole mosquito, Aedes triseriatus

(Diptera: Culicidae). Canadian Journal of Zoology, 75, 706-718.

71

Walker ED, Lawson DL, Merritt RW, Morgan WT, Klug MJ. 1991. Nutrient dynamics, bacterial

populations, and mosquito productivity in tree hole ecosystems and microcosms.

Ecology, 1529-1546.

Wallace JB, Webster JR, Cuffney TF. 1982. Stream detritus dynamics: regulation by invertebrate

consumers. Oecologia, 53, 197-200.

Wallace JB, Merritt RW. 1980. Filter-feeding ecology of aquatic insects. Annual Review of

Entomology, 25, 103-132.

Webster JR, Benfield EF. 1986. Vascular plant breakdown in freshwater ecosystems. Annual

Review of Ecology and Systematics, 567-594.

Yanoviak SP. 1999. Effects of leaf litter species on macroinvertebrate community properties and

mosquito yield in Neotropical tree hole microcosms.Oecologia, 120, 147-155.

Yee DA, Kaufman MG, Juliano SA. 2007. The significance of ratios of detritus types and micro‐

organism productivity to competitive interactions between aquatic insect detritivores.

Journal of Animal Ecology, 76, 1105-1115.

Yee DA, Kesavaraju B, Juliano SA. 2004. Larval feeding behavior of three co-occurring species

of container mosquitoes. Journal of Vector Ecology, 29, 315.

Yee DA, Juliano SA. 2006. Consequences of detritus type in an aquatic microsystem: effects on

water quality, micro‐organisms and performance of the dominant consumer. Freshwater

Biology, 51, 448-459.

72

Figures and Tables

Figure 2.1. Proportion of 20 Culex pipiens larvae that emerged as adults from whole leaf infusion across a) different numbers of leaf detritus species, b) single leaf species treatments, c) two leaf species treatments, d) three leaf species treatments, and e) four leaf species. Grouping letters represent significant within-number of species pairwise differences at 훼 = 0.05.

73

Figure 2.2. Mean wing length (mm) and development time (days) for Culex pipiens males and females reared in whole leaf infusion across a) different numbers of leaf detritus species, b) single leaf species treatments, c) two leaf species treatments, and d) three leaf species treatments.

74

Figure 2.3. Non-metric multidimensional scaling (NMDS) plot of 70 replicate containers on 227 bacterial taxa with a Bray-Curtis dissimilarity index across a) different numbers of leaf detritus species, b) single leaf species treatments, c) two leaf species treatments, and d) three leaf species treatments.

75

Table 2.1. Linear contrasts testing for additive effects of mixing four leaf detritus species on

Culex pipiens adult emergence rates by comparing mean emergence rates for mixture treatments to the combined mean response for their constituent individual leaf species. Red indicates a synergistic effect, blue indicates an antagonistic effect, and white indicates an additive effect.

Leaf mixture Estimate SE T P Effect HA 0.035 0.053 0.67 0.508 Additive HB -0.170 0.057 -2.99 0.004 Antagonistic HE 0.075 0.053 1.43 0.159 Additive AB -0.073 0.057 -1.28 0.207 Additive AE -0.200 0.053 -3.80 < 0.001 Antagonistic BE -0.100 0.053 -1.90 0.062 Antagonistic HAB 0.077 0.050 1.55 0.127 Additive HAE 0.127 0.050 2.56 0.013 Synergistic HBE 0.113 0.050 2.29 0.026 Synergistic ABE 0.203 0.050 4.10 < 0.001 Synergistic HABE 0.343 0.048 7.14 < 0.001 Synergistic

76

Table 2.2. Multivariate Analysis of Variance (MANOVA) for the effect of number of leaf species and mixture nested within number of leaf species on Culex pipiens female and male development time to emergence (days) and wing length (mm).

Pillai’s SCCs Sex Variable df P trace Development time Wing length Female Number of leaves 6 13.41 < 0.001 1.465 -1.986 Mixture(Number) 20 5.99 < 0.001 1.364 -2.084 Male Number of leaves 6 17.72 < 0.001 3.042 -0.982 Mixture(Number) 20 8.56 < 0.001 2.751 1.909

77

Table 2.3. Multiple response permutation procedure (MRPP) for the effect of number of leaf species and mixture nested within number of leaf species on 227 bacterial taxa. Grouping letters represent significant pairwise differences at 훼 = 0.05.

Test 1: Number of leaf species Treatment group n Average distance Significance letters 1 leaf species 19 0.764 A 2 leaf species 29 0.719 A 3 leaf species 18 0.584 B 4 leaf species 4 0.749 B Test 2: Single leaf species mixtures Treatment group n Average distance Significance letters H 5 0.701 A A 5 0.699 B E 5 0.695 C B 4 0.557 D Test 3: Two leaf species mixtures Treatment group n Average distance Significance letters HA 5 0.622 A HE 5 0.641 A HB 4 0.534 B AB 5 0.735 C AE 5 0.540 C BE 5 0.349 D Test 4: Three leaf species mixtures Treatment group n Average distance Significance letters HAB 5 0.504 A HAE 4 0.506 A HBE 5 0.411 A ABE 4 0.561 B

78

Table 2.4. Bacterial OTU indicator taxa (IV cutoff of 80) for each of four single leaf species.

Leaf species OTU IV Phylum Class Genus Autumn 275 95.5 Bacteroidetes Cytophagia Dyadobacter olive 58 89.4 Proteobacteria Betaproteobacteria Acidovorax 2726 93.3 Bacteroidetes Bacteroidia Paludibacter 1676 86.5 Firmicutes Clostridia Flavonifractor 2 86.4 Proteobacteria Betaproteobacteria Acidovorax Blackberry 387 81.5 Proteobacteria Betaproteobacteria Comamonas 2228 80.1 Proteobacteria Alphaproteobacteria Pleomorphomonas 39 80 Bacteroidetes Cytophagia Siphonobacter 1707 100 Firmicutes Clostridia Sporobacter 38 99.7 Proteobacteria Gammaproteobacteria Erwinia 124 99.2 Firmicutes Clostridia Clostridium 41 98.7 Proteobacteria Gammaproteobacteria Cronobacter 14 97.1 Firmicutes Negativicutes Zymophilus 6 97 Proteobacteria Gammaproteobacteria Acinetobacter Elderberry 137 94.8 Firmicutes Clostridia Clostridium 1791 89 Actinobacteria Actinobacteria Mycobacterium 3131 86.6 Proteobacteria Betaproteobacteria Comamonas 79 85.3 Proteobacteria Gammaproteobacteria Nevskia 129 80.5 Proteobacteria Alphaproteobacteria Saccharibacter 1388 80.2 Firmicutes Bacilli Paenibacillus 3026 86.3 Firmicutes Clostridia Anaerovorax Honeysuckle 121 82.1 Proteobacteria Betaproteobacteria Comamonas

79

CHAPTER 3

Exploitation of ecological traps for the control of a mosquito vector for West Nile virus

Abstract

For container-breeding mosquitoes, both the attractiveness of aquatic mesocosms for oviposition and the quality of these habitats for developing larvae are determined in part by inputs of leaf detritus from the terrestrial environment. But contrary to the expectations of optimal oviposition theory, this does not always result in selection for high-quality habitats. When organisms select a low-quality habitat, the mismatch between the attractiveness of a habitat and its quality for reproduction is known as an ecological trap. In urban ecosystems, Culex pipiens (Diptera:

Culicidae), an important vector for West Nile virus in the northeastern and midwestern United

States, typically lays eggs in roadside storm water catch basins and these permanent and semi- permanent water sources are longstanding targets for mosquito abatement via application of insecticides. In this study, we investigated the potential to improve the efficacy of existing control strategies for Cx. pipiens and provide botanical alternatives to conventional larvicides by exploiting naturally-occurring and artificial ecological traps for attract-and-kill mosquito control.

First, we conducted a field trial that found support for the hypotheses that inputs of two ecological traps – attractive yet deleterious common blackberry (Rubus allegheniensis) leaves and attractive Amur honeysuckle (Lonicera maackii) leaves combined with a Bacillus thuringiensis var. israelensis (Bti) briquette – in catch basins in a residential neighborhood increase mosquito oviposition rates and reduce adult emergence rates compared to a positive control. Second, we used field bioassays and binary choice cage bioassays to find support for the hypothesis that the attractiveness of these leaf species is related to plant secondary compounds

80 leaching from the leaf detritus. Third, we used next-generation sequencing and multivariate community analysis approaches to find support for the hypothesis that composition of bacterial flora varies predictably among habitats containing different leaf detritus species, although these differences are not predictive of mosquito oviposition rates. The results of this study provide experimental proof of concept for further exploration of a cheap and effective integrated vector management tool that may have minimal non-target effects and reduced potential to select for insecticide resistance.

Introduction

Arthropod-vectored diseases now constitute nearly 30 percent of emerging infectious diseases worldwide, and recent decades have seen numerous vector-borne pathogens spread at unprecedented rates throughout human and wildlife populations (Jones et al. 2008). In particular, more than half the global human population is at risk of exposure to pathogens transmitted via the bite of an infected mosquito, including dengue fever, chikungunya, Zika, and yellow fever, with an estimated 725,000 deaths due to mosquito-borne disease annually. In the United States, the most prevalent mosquito-borne pathogen is West Nile virus (WNV), with over 43,000 human cases including over 1,800 deaths documented (CDC 2015) since the virus was introduced to the

New York City metropolitan area in 1999 (Lanciotti et al. 1999) and subsequently expanded throughout the continental U.S. (Hayes and Gubler 2006). Because there are no human vaccines available for WNV and the majority of other mosquito-borne viruses, vector control is the most effective and, in many cases, only viable option for the prevention and management of these infectious illnesses.

Since the development of the insecticide dichlorodiphenyltrichloroethane (DDT) during

81

World War II, mosquito abatement strategies generally have relied upon application of synthetic organic insecticides. Most of the important vector mosquitoes in North America are container- breeding species, including Culex pipiens (Diptera: Culicidae), Cx. tarsalis, and Cx. quinquefasciatus (WNV; Goddard et al. 2002, Turell et al. 2005), Culiseta melanura (eastern equine encephalitis; Scott et al. 1984), and Ochlerotatus triseriatus, Oc. japonicus, and Aedes albopictus (La Crosse encephalitis; Thompson and Beaty 1977, Sardelis et al. 2002), and Ae. aegypti (dengue fever; O’Meara et al. 1995). Due to the obligate confinement of mosquitoes to closed aquatic habitats during the early life cycle, abatement efforts are typically more successful when they target the larval stage. Roadside storm water catch basins provide widespread permanent and semi-permanent water sources that support production of Culex species mosquitoes in particular (e.g., Munstermann and Craig 1976, Geery and Holub 1989), and these habitats are common targets for abatement efforts in urban environments (e.g., Stockwell et al.

2006, Anderson et al. 2011).

The first generation of larvicides developed and widely used for storm water application after DDT was banned in the U.S. in 1972 was methoprene, a compound that mimics the action of insect growth regulating hormones (Hazelrigg et al. 1980). During the 1980s a second generation of larvicides, the microbial insecticide Bacillus thuringiensis var. israelensis (Bti)

(Mulligan et al. 1980), was certified by the U.S. Environmental Protection Agency. There are currently 26 licensed granular and sustained-release briquette Bti formulations on the market for larval mosquito abatement. While these larvicides have been established as the “state of the art” in mosquito control for the past 30 years, rapid evolution of insecticide resistance in vector species (Goldman et al. 1986, Hemingway and Ranson 2000), effects on non-target organisms

(Milam et al. 2000), and concerns regarding the environmental and public health safety of

82 insecticide application (Green et al. 1990, Paris et al. 2011) contribute to a growing need for sustainable alternatives to conventional pesticides for mosquito abatement. Furthermore, the abundance of mosquitoes in catch basins and the efficacy of larvicides are influenced by environmental variables, including rainfall, water chemistry, and vegetation structure (Rey et al.

2006, Gardner et al. 2013, Harbison et al. 2015), suggesting that ecologically-based vector management strategies could complement insecticide use for mosquito abatement in storm water infrastructure.

A more sustainable, ecological alternative to exclusive reliance upon insecticides that has gained traction in the management of agricultural and forest pests is the strategy of “attract-and- kill.” This approach often uses insect attractants (e.g., pheromones) to lure target species to ingest toxins, thus eliminating the need for widespread and often non-specific insecticide application (de Groot and Nott 2001, Vargas et al. 2003). For example, a blend of the pheromone codlemone with permethrin has been used to control the codling moth Cydia pomonella

(Charmillot et al. 2000). Attract-and-kill has been explored for control of adult mosquitoes using attractive toxic sugar baits, which lure sugar-feeding adults to consume boric acid, an oral insecticide (Müller et al. 2010). However, this approach has not been investigated for control of larval mosquitoes. An important first step in evaluating the potential for a novel attract-and-kill larval control approach is beginning to identify the naturally-occurring ecological drivers of two components of mosquito production: habitat quality and habitat attractiveness.

Habitat quality for Culex spp. mosquitoes is driven largely by inputs of allochthonous plant-based detritus from the terrestrial environment, an important energy source for the aquatic larval stage. As leaves decay in the aquatic environment through leaching of soluble compounds into the water (Kuiters and Sarink 1986) and decomposition by detritivorous macroinvertebrates

83

(McArthur and Barnes 1988) and microorganisms (Fish and Carpenter 1982), developing mosquito larvae filter bacteria, fungi, and protozoa suspended in the water column and graze on leaf surfaces and the sides of artificial containers (Wallace and Merritt 1980, Merritt et al. 1992).

Mosquito adult emergence rates are related to the quantity and age of leaf detritus, and large masses of green leaves support higher emergence compared to small masses and senescent leaves (Walker et al. 1997). Emergence rates also vary due to species of leaf detritus in the aquatic habitat (Yanoviak 1999, Kominoski et al. 2007), and multiple biological mechanisms could underlie this observation. Habitat quality may be altered by plant secondary compounds that leach into the aquatic environment during decomposition, including toxic phytochemicals that act directly on developing mosquitoes (Kuiters and Sarink 1986) and antimicrobials that reduce the food available to larvae (Rios et al. 1988). Alternatively, the importance of bacteria and fungi to the mosquito diet (Fish and Carpenter 1982) and the variability in the microbial communities that form on different leaf species (Findlay and Arsuffi 1989) are well-documented, and our recent work suggests that habitat quality in the aquatic environment may be related to composition of the bacterial flora associated with different leaves (see Chapter 2).

There also is some evidence that leaf detritus in the aquatic environment mediates choice of oviposition site in female container-breeding mosquitoes; however, the ecological pathways governing habitat selection are less well-established experimentally than the mechanisms underlying habitat quality. Again, plant secondary compounds released during leaf decay differ among leaf species (Horner et al. 1988, Cornelissen et al. 1999), and some of these volatiles

(e.g., phenolics) may include attractants enabling mosquitoes to discover an oviposition site

(Kramer and Mulla 1979, Bentley and Day 1988) as well as contact oviposition stimulants

(Ponnusamy et al. 2008). Other studies of oviposition site selection have linked semiochemicals

84 produced by the bacterial flora as byproducts of metabolism to habitat selection by gravid female

Aedes aegypti and Aedes albopictus (Ponnusamy et al. 2008, Ponnusamy et al. 2010). However, the characteristics of the microbial community that are key to mediating oviposition site selection in mosquitoes (e.g., bacterial and/or fungal abundance, alpha diversity, composition) remain underexplored.

Optimal oviposition theory suggests that organisms should interpret environmental cues to choose the highest-quality habitat for the development of progeny (Gripenberg et al. 2010).

However, our previous experiments testing the hypotheses that Cx. pipiens adult emergence and oviposition rates vary among aquatic environments containing different leaf detritus species revealed that this is not always the case (Gardner et al. 2015). Although we found that female mosquitoes recognized and preferentially laid eggs in some high-quality habitats, such as those containing leaves of invasive Amur honeysuckle (Lonicera maackii), we also identified a naturally-occurring ecological trap; Cx. pipiens were strongly attracted to oviposit in a poor- quality habitat containing leaves of common blackberry (Rubus allegheniensis). Ecological traps occur due to a mismatch between the attractiveness of a habitat and its quality for reproduction, wherein organisms select for a low-quality habitat. The discovery of leaf species that mediate habitat attractiveness for this important enzootic vector for WNV in the northeastern and midwestern United States (Turell et al. 2001, Andreadis et al. 2001) suggests the potential for a novel attract-and-kill approach for mosquito control in closed aquatic environments including storm water catch basins.

In this study, we investigated the potential for exploitation of ecological traps for control of Cx. pipiens in storm water environments. First, we conducted a field trial to test the hypothesis that inputs of two ecological traps (i.e., attractive yet deleterious common blackberry leaves and

85 attractive Amur honeysuckle leaves combined with a Bti larvicidal briquette) in catch basins in a residential neighborhood increase mosquito oviposition rates and reduce adult emergence rates compared to a positive control (i.e., no alteration of the existing detritus in the catch basin).

Second, we used field bioassays and “binary choice” cage bioassays to test the hypothesis that the attractiveness of these ecological traps to Cx. pipiens is related to the bacterial community associated with the leaf detritus species rather than plant secondary compounds leaching from the leaf detritus. Third, we used next-generation sequencing and multivariate community analysis approaches to test the hypothesis that abundance and composition of bacterial flora vary predictably among habitats containing different leaf detritus species and are predictive of mosquito oviposition rates. The results of this study provide experimental proof of concept for a cheap and effective integrated vector management tool with minimal non-target effects and reduced potential to select for insecticide resistance. Our findings may also yield insights into potential mechanisms by which ecological traps occur, with applications to conservation biology.

Methods

Oviposition and emergence in catch basins. Our field trial testing the efficacy of exploitation of ecological traps for control of Cx. pipiens was conducted over a two-year period in Paxton,

Illinois (40°27’N, 88°05’W). Paxton has an area of 5.95 km2 and a human population of 4,345 as of 2014. The first human case of WNV in Ford County was detected in 2002 (CDC MMWR

2002), and we selected the study area because no mosquito abatement efforts are in place in this city (e.g., adulticide spraying, larvicide application in catch basins) whereas other more populated cities in central Illinois (e.g., Champaign, Bloomington, Decatur) have established mosquito control programs.

86

In 2013, four experimental treatments (described below; negative control, positive control, blackberry, and honeysuckle) were applied to 20 catch basins in Paxton (5 catch basins per treatment) from 16 August to 13 September. In 2014, the same four treatments plus two additional treatments (Bti and Bti+honeysuckle) were applied to 60 catch basins (10 catch basins per treatment) from 18 July to 15 August. All catch basins had open grates and were located on the edges of residential streets, and the mean depth from street surface to the bottom of the catch basin was 55 ± 20.38 cm. Treatments were spatially aggregated within the study area to prevent carryover effects of one treatment into adjacent connected catch basins (Figure 3.1). We used three measures to protect against the confounding of treatment effects by this non-random spatial distribution of treatments. First, during each year of the experiment we sampled larvae using methods described below prior to the allocation of treatments to establish that there was no spatial dependence associated with the baseline abundance of mosquitoes across treatment areas.

Only catch basins where a minimum of 15 larvae were found in this initial sampling were included in the study. Second, areas of the city that received each treatment were selected randomly in 2013, and re-randomized in 2014. Third, during the second year of the experiment, the negative control catch basins were redistributed throughout the study area.

The six experimental treatments are summarized as follows: 1) Negative control: all organic detritus dredged from the catch basins on a weekly basis throughout the duration of the study using a 20.3 X 20.3 cm aquarium net attached to the end of PVC pipe 3 m in length; 2)

Positive control: no modification of debris in catch basins; 3) Blackberry (naturally-occurring ecological trap): all organic detritus dredged from the catch basins during the first week of the experiment and 100 g fresh blackberry leaves added, submerged underwater in Woolite© mesh delicates bags which were tied to the grates using bailing twine to prevent the leaf bags from

87 being flushed out of the catch basins; 4) Honeysuckle (attractant only): same as blackberry treatment, except with the addition of 100 g fresh honeysuckle leaves instead of blackberry leaves; 5) Bti (toxin only): FourStar® 45-day sustained release Bti briquettes (Central Life

Sciences, Phoenix AZ) were added to catch basins, with no additional modification of debris in catch basins; 6) Bti+honeysuckle (artificial ecological trap): same as honeysuckle treatment, with the addition of a FourStar® Bti briquette as described for the Bti treatment. For the treatments containing leaf detritus, green leaves were collected in the research woods at the University of

Illinois Pollinatarium in Urbana, Illinois, and weighed and bagged 24 hours prior to their allocation to catch basins. We used fresh leaves rather than senescent leaves because while there is no interaction in response between the effect of leaf age and the effect of leaf species on mosquito emergence rates, the magnitude of the main effect of leaf species is larger for green leaves compared to senescent leaves (Walker et al. 1997).

To estimate relative adult mosquito emergence rates from the catch basins and prevent mosquitoes from escaping into the surrounding environment, floating emergence traps modeled after those described in Hamer et al. (2011) were established in the catch basins after allocation of experimental treatments. These traps float on the surface of the water, covering approximately two-thirds of the water surface, and capture adult mosquitoes as they fly upward out of the catch basins. Adults were collected once per week for four weeks, frozen, and identified to species using taxonomic keys in Andreadis et al. (2005). Due to the timing of the experiment, the only species observed was Cx. pipiens.

Abundance of larvae was used as a proxy for an estimate of relative mosquito oviposition rates due to the logistical difficulties of collecting egg rafts inside deep catch basins and the effect that removal and handling of egg rafts would have on the abundance of adult mosquitoes

88 emerging from these habitats. Larvae were collected using a 12.7 X 12.7 cm aquarium net attached to the end of PVC pipe 3 m in length. The pole was passed over the water surface in two figure eights and the net was then inverted into a container and flushed with water, and larvae were stored on ice for further processing. Larvae of all instars were counted in aggregate and identified to species using taxonomic keys. Larvae were collected once per week for five weeks, once prior to allocation of experimental treatments and four times after the treatments were applied. Again, the only species observed was Cx. pipiens.

All data analyses were conducted in SAS 9.3 (SAS Institute Inc., Cary NC) using PROC

MIXED. To confirm that there were no significant differences in abundance of larvae across catch basin treatment areas prior to the allocation of experimental treatments, a general linear mixed model (GLMM) was fit to the number of larvae per catch basin during the baseline week given by

푌푖푗푘 = 휇 + 푌푖 + 푇(푖)푗 + 휀푖푗푘 (Eqn. 3.1) where 푌푖 represents the random main effect of year and 푇(푖)푗 represents the fixed main effect of treatment nested within year.

To test the hypothesis that inputs of attractive leaf detritus species in catch basins increase mosquito oviposition rates, a GLMM with repeated measures was fit to the number of larvae per catch basin given by

푌푖푗푘 = 휇 + 푌푖 + 푇(푖)푗 + 휀1(푖푗푘) + 푊푙 + 푇푊(푖)푗푙 + 휀2(푖푗푘푙) (Eqn. 3.2) where 푌푖 represents the random main effect of year, 푇(푖)푗 represents the fixed main effect of treatment nested within year, 푊푙 represents the fixed main effect of week, and 푇푊(푖)푗푙 represents the interaction between the main effects of week and treatment. Catch basin ID was the subject, week was the repeated variable, and an AR(1) covariance structure with restricted maximum

89 likelihood (REML) was used for estimation of covariance parameters. Linear contrasts using the

ESTIMATE statement in PROC MIXED were used to compare each of the three attractive treatments (blackberry, honeysuckle, and Bti+honeysuckle) to the positive control, the

Bti+honeysuckle treatment to the Bti-only treatment, and the negative control to the positive control. A Bonferroni correction was applied to adjust for multiple comparisons. Significant interactions were interpreted using SLICE statements.

To test the hypothesis that inputs deleterious leaf detritus species or Bti larvicide) in catch basins decrease adult mosquito emergence rates and high-quality leaf detritus species increase adult emergence rates, the same linear model was fit to the number of adults per catch basin

(Eqn. 3.2). Here, linear functions were used to compare each of the three deleterious treatments

(blackberry, Bti+honeysuckle, and Bti) to the positive and negative controls, the high-quality leaf detritus treatment (honeysuckle) to the positive control, and the negative control to the positive control. Again, a Bonferroni correction was used to adjust for multiple comparisons and significant interactions were interpreted using SLICE statements.

Oviposition field bioassay. To test whether Cx. pipiens oviposition rates respond similarly to the treatments compared to abundance of larvae, a separate oviposition field assay was conducted using artificial containers in which direct measurement of mosquito oviposition (i.e., egg raft counts) was more tractable. In 2015, six oviposition traps (Reiter 1986) were placed 1 m apart from each other in partial shade at five locations within a 2 km radius in residential Urbana,

Illinois. The 30 oviposition traps were monitored for egg rafts weekly and the number of egg rafts collected in each treatment was recorded over four weeks from 24 July to 21 August.

During the first week of the study period, a subset of the egg rafts collected at each location were

90 hatched in petri dishes and identified as Cx. pipiens or Culex restuans based on the presence or absence of a clear area anterior to the sclerotized egg-breaker that is present in Cx. restuans and absent in Cx. pipiens. The only species observed was Cx. pipiens.

Each oviposition trap contained 7 L of tap water and one of six treatments designed to imitate the treatments applied to catch basins in Paxton: 1) Negative control: tap water only; 2)

Positive control: mixture of 80 g senescent leaves from native trees (described below); 3)

Blackberry (naturally-occurring ecological trap): 80 g fresh blackberry leaves; 4) Honeysuckle

(attractant only): 80 g fresh honeysuckle leaves; 5) Bti (toxin only): same as positive control, with the addition of a FourStar® 45-day sustained release Bti briquette; 6) Bti+honeysuckle

(artificial ecological trap): same as honeysuckle treatment, with the addition of a FourStar® Bti briquette as described for the Bti treatment. All leaves were collected from the same sources as for the catch basin experiment. For the positive control and the Bti treatment, the leaf species were a mixture of sugar maple (Acer saccharum) and red oak (Quercus rubra), which were the two most common species identified in a survey of catch basins conducted in Decatur, Illinois in

2013 (personal observation).

To test the hypothesis that inputs of attractive leaf detritus species increase mosquito oviposition rates, a GLMM with repeated measures was fit to the number of egg rafts per week per oviposition trap given by

푦푖푗푘 = 휇 + 퐿푖 + 푇푗 + 휀1(푖푗) + 푊푘 + 푇푊푗푘 + 휀2(푖푗푘) (Eqn. 3.3) where 퐿푖 represents the random block effect of location, 푇푗 represents the fixed main effect of treatment, 푊푘 represents the fixed main effect of week, and 푇푊푗푘 represents the interaction between the main effects of week and treatment. Oviposition trap ID was the subject, week was the repeated variable, and an AR(1) covariance structure with restricted maximum likelihood

91

(REML) was used for estimation of covariance parameters. As in the catch basin experiment, linear contrasts were used to compare each of the three attractive treatments (blackberry, honeysuckle, and Bti+honeysuckle) to the positive control, the Bti+honeysuckle treatment to the

Bti-only treatment, and the negative control to the positive control. A Bonferroni correction was applied to adjust for multiple comparisons. Significant interactions were interpreted using SLICE statements.

Microbial communities and oviposition. On the same days as egg raft counts in the oviposition field bioassay, 15 mL aliquots of infusion were taken from each oviposition trap (for a total of 30 samples per week and 120 samples total) and stored at -80°C until further processing. Genomic

DNA was extracted from the samples using the UltraClean® Soil DNA Isolation Kit (Mo Bio

Laboratories Inc., Carlsbad CA, Cat. No. 12800-50) according to the manufacturer’s instructions with the following modifications. Prior to the first step of the protocol, 15 mL water samples were concentrated by centrifugation at 5000 rpm for 20 minutes. Next, 14.5 mL supernatant were removed and the pellet was re-suspended in the remaining supernatant by vortex; these 500 μL aliquots of the sample were added to the bead solution tubes in the first step of the extraction protocol. The fifteenth step of the protocol (i.e., addition of Solution S4 and centrifugation) was repeated three times to maximize purity of the samples.

The 120 DNA extracts were sequenced using Illumina MiSeq Bulk v3 analysis at the W.

M. Keck Center for Comparative and Functional Genomics at the University of Illinois at

Urbana-Champaign. Seventeen samples failed to amplify and were excluded from further consideration. Initially, we targeted three hypervariable regions of the 16S rRNA gene for bacteria and the ITS1-ITS4 interspacer regions for fungi using the following primer sets: V1-V3

92 forward 5’–GAGTTTGATCNTGGCTCAG–3’, reverse 5’–GTNTTACNGCGGCKGCTG–3’;

V3-V5 forward 5’–CCTACGGGAGGCAGCAG–3’, reverse 5’–CCGTCAATTCMTTTRAGT–

3’; V4 forward 5’–GTGYCAGCMGCCGCGGTAA–3’, reverse 5’–

GGACTACNVGGGTWTCTAAT–3’; ITS1 5’–TCCGTAGGTGAACCTGCGG–3’; and ITS4

5’–TCCTCCGCTTATTGATATGC–3’. For Illumina library preparation by Fluidigm protocols,

DNA from each water sample was amplified using the four primer sets added with Fluidigm

Illumina linkers and unique barcodes. The final Fluidigm libraries were pooled for sequencing at the Keck Center’s DNA Services laboratory. The V4 region yielded 6.1 million reads, more than two times the number of reads for the V3-V5 region (2.2 million reads) and more than four times the number of reads for the V1-V3 region (1.5 million reads), and therefore the V4 region was used for subsequent analyses. The ITS region yielded 846,000 reads and was excluded from subsequent analyses, which focused on bacteria rather than fungi.

As read1 and read2 overlap for the V4 region, read1 and read2 data were merged for each sample using PEAR with default parameters. The IM-TORNADO v2.0.3.2 pipeline (Jeraldo et al. 2014) was used on the University of Illinois Biocluster to demultiplex and quality filter bacterial reads. IM-TORNADO requires both read1 and read2 data as input. Data merged by

PEAR were used as read1 data, and read1 data for all samples were reverse complemented using

Perl and input as read2 data. IM-TORNADO trims low quality bases using Trimmomatic (Bolger et al. 2014) before running de novo operational taxonomic unit (OTU) picking. For this analysis, a minimum length of 150 (MINLEN:150) was required for a read. The RDP10 database was used for the bacterial taxonomy assignment (latest version accessed and downloaded in January,

2016) (Cole 2009). Sampling effort was standardized per sample using rarefaction to normalize read depth to 6,229 reads per sample. In addition, reads were filtered using the criterion that

93

0.001 (i.e., 0.1 percent) is the minimum total observation count of an OTU for retention. The reason we used a conservative filtering criterion rather than, for example, requiring a minimum of 10 reads for retention, was that our study questions were concerned with OTUs present in sufficient abundance to have a role in mediating mosquito oviposition site selection. However, we conducted the same set of analyses described below using a less stringent filtering criterion

(i.e., a minimum of 10 reads is required for retention of an OTU) for comparison and arrived at the same qualitative conclusions.

To test the hypothesis that abundance and composition of bacterial OTUs vary predictably among aquatic habitats containing different treatments and over time, multi-response permutation procedure (MRPP) tests were used to test for differences among groups established a priori based on treatments and time points (Miekle 1976). MRPP is a non-parametric permutation test, in which the average within-group distance in multidimensional space is calculated for the specified groups, then all experimental replicates are randomly assigned to new groups and the average within-group distances calculated again over a series of iterations. The statistical significance of the observed average within-group distances is determined by comparing the observed test statistic to the distribution of within-group distances generated by the randomization test. A disadvantage of MRPP is that it cannot accommodate factorial designs and the permutation test cannot model autocorrelated error structures, and therefore complex experimental designs must be sliced to answer different research questions (McCune et al. 2002).

Therefore, separate tests with 1000 iterations were conducted comparing bacterial flora among treatments within a given week and among weeks within a given treatment, with Bonferroni adjustments applied for multiple comparisons.

94

To test the hypothesis that abundance and composition of bacterial OTUs are predictive of Cx. pipiens oviposition rates, non-metric multidimensional scaling (NMDS) first was conducted with abundances of unique bacterial OTUs as the raw explanatory variables (Kruskal

1964) and a Bray-Curtis dissimilarity index, which takes into account both the presence/absence and the relative abundance of species. The number of NMDS axes was selected based on a stress test, using the criterion that the addition of an axis must reduce stress by at least five. Next, a distance-based regression model was fit with the number of egg rafts laid per week as the response, predictors expressed as the NMDS distance matrix, and a block effect of location, using the R package ecodist v1.2.9 (Goslee and Urban 2007). This non-parametric approach provides permutational tests of significance for regression coefficients (Legendre et al. 1994,

Lichstein 2007), and the test was conducted with 1000 iterations.

Oviposition cage bioassay. To further assess whether the attractiveness of blackberry and honeysuckle leaves to Cx. pipiens is related to the microbial community associated with the leaf species versus plant secondary compounds leaching from the leaf detritus, a series of “binary choice” cage behavioral bioassays was conducted. Infusions of blackberry and honeysuckle leaves were prepared by fermenting 15 g fresh leaves collected from the same sources as for the previous experiments in closed plastic 2 L buckets filled to the brim with tap water for 7 days, a standard infusion age used in comparable studies (e.g., Gardner et al. 2015, see Chapter 2).

These “whole” infusions were used to prepare three treatments used for subsequent bioassays: 1) whole infusion; 2) leachate-only; and 3) microbes-only. The leachate-only treatment was prepared by passing 100 mL whole infusion through 0.22 μm pore Stericup filters (Fisher

Scientific, Hampton NH) powered by a vacuum pump (Precision Scientific Instruments, Buffalo

95

NY) to remove microorganisms while retaining the plant secondary components associated with the different leaf detritus mixtures. The microbes-only treatment was prepared by centrifuging

100 mL whole infusion in 25 mL aliquots at 5000 rpm for 15 minutes, removing the supernatant, and re-suspending the pellet containing the microbes in 200 mL deionized water.

Gravid female Cx. pipiens were collected at two residences in Urbana, Illinois, using grass infusion-baited gravid traps set for a 16-hour period (1600h to 0800h). The mosquitoes were knocked down in a -20°C freezer for two minutes, then sorted by sex and identified to species. Female Cx. pipiens were placed in one of ten 1 ft3 cages (24.88 ± 2.36 mosquitoes per cage). Two 125 mL polypropylene cups each containing 100 mL of test and control treatments were placed in diagonal corners of each cage. The five treatment pairs repeated for both honeysuckle and blackberry leaf infusion were: 1) whole infusion versus deionized water; 2) microbes-only versus deionized water; 3) leachate-only versus deionized water; 4) microbes-only versus whole infusion; and 5) leachate-only versus whole infusion. Treatment cups were fitted with black construction paper sleeves to mask the visual cues associated with dark-colored treatments (Bentley and Day 1988). Mosquitoes were provided a vial of honey water with cotton wicking for sugar feeding and deionized water with cotton wicking for hydration throughout the duration of the experiment. The cages were stored in a dark room for a minimum of three days, and egg rafts were counted in each cup every 12 hours.

To compare mosquito oviposition rates within each pair of treatments, separate chi- square tests of goodness of fit were conducted for each treatment pair. The expected values for each treatment within a pair were the total number of egg rafts laid in the cage divided by two. A

Bonferroni correction was applied to adjust for multiple comparisons.

96

To verify that the filtering for the leachate-only treatment successfully removed microbes and that the centrifugation for the microbes-only treatment successfully retained microbes, four

15 mL aliquots of sample were taken from each treatment (i.e., whole infusion, microbes-only, and leachate-only for blackberry and honeysuckle infusions, for a total of 24 samples) over the duration of the cage bioassay. Genomic DNA extraction and sequencing were carried out for these samples as described above, and again 16S V4 reads were filtered using the criterion that

0.001 (i.e., 0.1 percent) is the minimum total observation count of an OTU for retention. Because the goal of this sequencing effort was to validate our experimental protocol, rarefaction was not conducted for these samples.

Results

Oviposition and emergence in catch basins. Abundance of Cx. pipiens larvae in catch basins did not vary among treatment areas prior to allocation of experimental treatments in either 2013 or 2014 (F = 0.60; df = 9, 70; P = 0.21). After treatments were applied, we found significant main effects of treatment (F = 9.78; df = 9, 60; P < 0.01) but not week (F = 1.31; df = 3, 188; P =

0.27) and a significant treatment by week interaction effect (F = 2.64; df = 27, 188; P < 0.01) on abundance of larvae. The significant interaction reflected differences in the magnitude but not the direction of main effects over time, with the largest effects of treatment observed during the second and third weeks of the study period (Figure 3.2). Linear contrasts showed that abundance of larvae was greater than the positive control for the blackberry treatment, lower for the honeysuckle+Bti treatment, and not significantly different for the honeysuckle treatment. We found no significant difference in abundance of larvae for the honeysuckle+Bti treatment

97 compared to the Bti-only treatment. The positive control had a significantly greater abundance of larvae than the negative control (Table 3.1).

We also found significant main effects of treatment (F = 24.92; df = 9, 70; P < 0.01) and week (F = 10.32; df = 3, 204; P < 0.01), and a significant treatment by week interaction effect (F

= 4.16; df = 27, 204; P < 0.01), on abundance of adult mosquitoes emerging from catch basins.

Again, the significant interaction reflected differences in magnitude rather than direction of effects, and treatment effects were largest during the second and third weeks of the study period

(Figure 3.2). Linear contrasts showed that abundance of adults were lower than the positive control and not significantly different from the negative control for the blackberry, honeysuckle+Bti, and Bti-only treatments. Furthermore, the abundance of adults was higher for the honeysuckle treatment compared to the positive control. Again, the positive control had a significantly greater abundance of adults than the negative control (Table 3.2).

Oviposition field bioassay. We found significant main effects of treatment (F = 5.37; df = 5, 20;

P < 0.01) and week (F = 4.99; df = 3, 72; P < 0.01) but not their interaction (F = 1.08; df = 15,

72; P = 0.39) on the number of Cx. pipiens egg rafts laid in oviposition traps (Figure 3.3). Linear contrasts showed that the number of egg rafts was greater than the positive control for the blackberry, honeysuckle, and honeysuckle+Bti treatments. The number of egg rafts also was greater for the honeysuckle+Bti treatment compared to the Bti-only treatment. We found no significant difference between the number of egg rafts collected in the positive control versus the negative control (Table 3.3). In general, the largest numbers of egg rafts were collected during the second and third weeks of the study period.

98

Bacterial communities and oviposition. Bacterial OTUs were profiled across six treatments at four time points. After rarefaction, quality filtering, and criteria filtering of raw reads, 169 OTUs were retained for the 16S V4 amplicon. Ordination using NMDS reduced these taxa to three axes. NMDS axis 1 explained 31 percent of variation in the distance matrix, axis 2 explained 25 percent, and axis 3 explained 12 percent. Bacterial flora varied predictably among treatments (T

= -20.20; P < 0.01) and over time (T = -3.38; P = 0.02). At each time point, we found significant pairwise differences in the distances between the negative control and all other treatments and blackberry leaves compared to all other treatments except the positive control (Table 3.4). We found no significant pairwise differences in the distances among the honeysuckle, Bti-only, honeysuckle+Bti, and positive control treatments. In the blackberry, honeysuckle, honeysuckle+

Bti, and positive control treatments, we found significant pairwise differences in the distances between week 1, weeks 2 and 3, and week 4, with fewer apparent changes in bacterial flora over time observed in the Bti and negative control treatments (Table 3.5). These pairwise differences did not reflect the observed differences in oviposition rates. Further, permutational distance- based regression showed the bacterial distance matrix was not predictive of oviposition rates (F =

7.36; P = 0.41).

Oviposition cage bioassay. For the blackberry leaf infusion treatments, we found that a larger proportion of egg rafts were laid in the whole infusion and leachate-only treatments compared to the deionized water treatment, but there was no significant difference between the microbes-only treatment and the deionized water treatment (Figure 3.4). We found no significant difference in the number of egg rafts laid in the whole infusion versus the leachate, but a significantly greater number of egg rafts were laid in the whole infusion compared to the microbes-only (Table 3.6).

99

Similarly, for the honeysuckle leaf infusion treatments, we found that a larger proportion of egg rafts were laid in the whole infusion and leachate-only treatments compared to the deionized water treatment, but there was no significant difference between the microbes-only treatment and the deionized water treatment. We found no significant difference in the number of egg rafts laid in the whole infusion versus the leachate-only, but again a significantly greater number of egg rafts were laid in the whole infusion versus the microbes-only treatment.

We compared the abundance of reads for the 16S V4 amplicon across treatments to verify that the filtering for the leachate-only treatment successfully removed microbes and that the centrifugation for the microbes-only treatment successfully retained microbes. After quality filtering and criteria filtering of raw reads, 116 OTUs were retained: 47,651 ± 551 reads for the blackberry whole infusion treatment; 43,372 ± 1,395 for the blackberry microbes-only treatment

(91% of whole infusion reads); 5,192 ± 1,731 for the blackberry leachate-only treatment (11% of whole infusion reads); 40,691 ± 876 for the honeysuckle whole infusion treatment; 39,924 ± 960 for the honeysuckle microbes-only treatment (98% of whole infusion reads); and 12,218 ± 2,493 for the honeysuckle leachate-only treatment (30% of whole infusion reads).

Discussion

Due to the rapid evolution of insecticide resistance in mosquitoes, effects on non-target organisms, and concerns regarding the environmental and public health safety of insecticide application, there is a growing need for sustainable alternatives to conventional pesticide use for mosquito abatement. The approach of attract-and-kill, which has been used successfully for the control of agricultural and forest pests but remains underexplored for arthropod vector species, has the potential to yield an integrated vector management tool that may reduce the need for

100 insecticide use and enhance the efficacy of existing mosquito control strategies. In this study, we investigated the potential for exploitation of ecological traps for the larval-stage control of an important vector for West Nile virus in roadside storm water catch basins, longstanding targets for mosquito abatement efforts in urban environments. We used inputs of leaf detritus species that are known attractants for oviposition in Cx. pipiens combined with conventional insecticides and leaf species that are deleterious to developing larvae to lure gravid females to poor quality habitats. We then conducted field and laboratory bioassays to examine the poorly-understood ecological mechanisms underlying variation in the attractiveness of leaf detritus species. This is an area in critical need of further research to realize the potential of attract-and-kill as a cheap and effective supplement or alternative to conventional mosquito control approaches.

We found support for the hypothesis that inputs of a natural and an artificial ecological trap (i.e., blackberry leaves and Amur honeysuckle leaves combined with a Bti larvicidal briquette, respectively) in catch basins in a residential neighborhood reduce adult emergence rates compared to a positive control, with emergence rates on par with a negative control and the current “state of the art” insecticide. The variability in quality of leaf detritus species as food resources for developing mosquito larvae is well-established by laboratory experiments

(Reiskind et al. 2009, Muturi et al. 2012) and studies of treehole systems (Yanoviak 1999, David et al. 2000) across multiple mosquito species. Underlying mechanisms are thought to include variation in the abundance and composition of bacterial flora that form on different leaf species as they decay (see Chapter 2) as well as phytochemicals that leach into the aquatic environment during leaf decomposition. These phytochemicals may include compounds that serve as defensive mechanisms against herbivores and may alter nervous system function in developing mosquitoes, and antimicrobials that reduce the abundance of bacteria and fungi available to

101 developing larvae as food resources. Previous research also has demonstrated correlations between the abundance of Cx. pipiens and Cx. restuans larvae in catch basins and the species and structure of vegetation in the surrounding environment (Gardner et al. 2013). However, our current study is the first to experimentally demonstrate that inputs of leaf detritus influence adult mosquito emergence rates even in the dynamic aquatic mesocosms of storm water systems, where mosquito production may be altered by a variety of physical and biological factors such as sunlight exposure, water chemistry, inputs of fertilizers via runoff from lawns, rainfall, ambient temperature, and predation by other aquatic macroinvertebrates (Rey et al. 2006, Gardner et al.

2012). This result indicates that naturally-occurring botanicals may provide viable alternatives or supplements to conventional insecticides for mosquito control in the storm water environment.

Using abundance of Cx. pipiens larvae as a proxy for oviposition rates, we further found support for the hypothesis that inputs of blackberry leaves increase the attractiveness of catch basins, while inputs of honeysuckle leaves mixed with Bti resulted in a significant decrease in the abundance of larvae collected compared to a positive control. However, our complementary assay conducted using oviposition traps in residential neighborhoods suggested that abundance of larvae is not always a good measure of oviposition rates, most likely because exposure to Bti causes larvae to die in the first instar whereas the deleterious effects of blackberry leaves are slower-acting. In this secondary experiment in which we counted egg rafts directly, we found that like the blackberry treatment, the honeysuckle+Bti treatment yielded a significant increase in oviposition rates compared to the positive control and the Bti-only treatment. While previous research has demonstrated a role of leaf detritus from terrestrial plants in mediating mosquito oviposition site selection (Rajkumar and Jebanesan 2005, Reiskind et al. 2009, Gardner et al.

2015), this is the first study to experimentally demonstrate that this observation holds even in

102 highly dynamic storm water environments, which provide a diversity of visual and chemical cues that are atypical of more controlled laboratory conditions, such as water color, pollutants, and presence of conspecifics (Mokany and Shine 2003, Beehler et al. 1993). Thus, our results suggest that larval Cx. pipiens control efforts may be enhanced by use of oviposition attractants to lure gravid females to lay eggs in deleterious habitats.

Our oviposition trap bioassay also began to explore potential mechanisms for variation in the Cx. pipiens oviposition response to different leaf detritus species, in particular characteristics of the bacterial communities associated with different treatments that may be predictive of female mosquito oviposition rates. Using multivariate community metrics, we found support for the hypothesis that abundance and composition of bacterial flora varies among treatments, a result consistent with our previous work (see Chapter 2), and additionally found support for the hypothesis that bacterial communities vary over time within treatment as the leaves decompose.

However, pairwise differences among bacteria associated with different treatments did not reflect differences in oviposition rates observed in the oviposition trap assay; distance-based regression analysis also failed to find support for the hypothesis that abundance and composition of bacteria are predictive of oviposition rates. Our prior research revealed that mosquito emergence rates can be predicted by bacterial composition via the same regression approaches (see Chapter 2), suggesting that different factors govern emergence and oviposition rates in Cx. pipiens.

Binary choice oviposition cage bioassays conducted in the laboratory provided further support against the hypothesis that the attractiveness of blackberry and honeysuckle leaves is due primarily to the bacterial communities associated with these leaf detritus species. Even after we controlled for the effect of infusion color by masking visual cues, mosquitoes laid eggs in equal proportions in containers of whole infusion versus the leachate-only treatment for both

103 blackberry and honeysuckle leaves, whereas whole infusions consistently were selected over the microbes-only treatments and in fact the microbes-only treatments were not selected over the deionized water negative control. Collectively, these experiments suggest that it is not strictly the microbial community that determines oviposition site selection by Cx. pipiens, but another component of the leaf infusion (e.g., plant secondary compounds that leach into the water during leaf decomposition). These results contrast with the findings of Ponnusamy et al., who demonstrated using similar binary choice laboratory experiments that bamboo and white oak leaf infusions collect large numbers of Ae. aegypti eggs due to bacteria-associated carboxylic acids and methyl esters that stimulate oviposition (Ponnusamy et al. 2008), and showed that the composition of bacterial flora in aquatic habitats mediates production of oviposition-stimulating kairomones (Ponnusamy et al. 2010). However, other studies of Anopheles stephensi and An. subpictus oviposition ecology have demonstrated that plant secondary compounds obtained from leaf tissue using acetone, ethyl acetate, and methanol extraction alter oviposition site selection, indicating that leaf material itself as well as microbial flora can provide the source of chemicals that drive choice of oviposition habitat (Rajkumar and Jebanesan 2005, Rajkumar and Jebanesan

2009, Elango et al. 2009).

Although our experiments provide multiple lines of support for a role of leachate from blackberry and honeysuckle leaves in altering Cx. pipiens oviposition habitat choice, we wish to note some limitations of the technical and statistical approaches in our studies of bacterial flora and oviposition that may be relevant to the interpretation of our results. First and foremost, all of our analyses were conducted at the bacterial community level, using multivariate community analysis techniques (i.e., NMDS and MRPP) to test for among-treatment differences in bacterial composition and distance-based regression to test average distances between containers as a

104 predictor of oviposition rates. In general, community-level analysis is the approach most commonly used with Illumina metagenomic data because it minimizes the influence of the biases in this sequencing technique (Gomez-Alvarez et al. 2009, Aird et al. 2011, Steven et al. 2012,

Degnan and Ochman 2012), and we chose to use Illumina sequencing over other methods to capture low abundance and unculturable bacteria. However, use of this correlational approach does not allow for experimental tests of the contribution of individual bacterial taxa to mediating oviposition site selection across a range of bacterial abundance in a way that has been achieved using culture-dependent methods (Ponnusamy et al. 2010). Second, although the centrifugation method used to prepare the microbes-only treatment in the oviposition cage bioassays appeared to retain bacteria, the filtering method used to prepare the leachate-only treatment was less effective at removing bacteria and may have removed plant compounds larger than the filter pore size as well. We chose to use the filtering method rather than autoclaving or heat shock to minimize the effect on the composition of volatiles contained in the infusion, but the technique was imperfect and the inadvertent retention of certain bacterial flora may have contributed to the attractiveness of the leachate-only treatment in the cage bioassay.

Although it may remain unclear whether oviposition attractants are produced as byproducts of microbial metabolism or leach directly from the leaves, it is evident that there is a chemical basis to oviposition site selection in Cx. pipiens (Leal et al. 2008) that requires further study to realize potential applications of ecological traps for control of container-breeding mosquito disease vectors. A question that is not assessed within the scope of our current experimental design is whether the increased oviposition rate in blackberry and honeysuckle leaf infusions is due mainly to oviposition attractants that enable discovery of the aquatic habitat or contact oviposition stimulants that induce egg-laying behavior. This distinction is important

105 because attractants may be more valuable than oviposition stimulants for an attract-and-kill mosquito control approach because mosquitoes typically do not visit many potential oviposition habitats prior to selecting a site (Bentley and Day 1989). Future studies could use sticky-screen cage bioassays with the same paired treatments and compare these results to our existing data to assess whether the blackberry and honeysuckle leaf infusions contain oviposition attractants

(Isoe et al. 1995). Moreover, there is a need to identify compounds that are attractive to Cx. pipiens to apply the attract-and-kill concept to vector control. A variety of phenolic compounds are abundant in blackberry (Wang and Lin 2000) and honeysuckle leaves (Cipollini et al. 2008) and have been shown attractive to mosquitoes in systems with different mosquito species and leaf species (Beehler et al. 1994, Du and Millar 1999).

In summary, our current study provided experimental proof of concept for a novel, cheap, and effective integrated vector management tool for mosquito control in storm water habitats.

Due to a widespread lack of vaccines, vector control is the most effective and often the only viable option for the prevention or management of mosquito-borne viruses, and attract-and-kill mosquito control has the potential to improve upon existing proven strategies in several areas.

First, while the deleterious effect of blackberry leaves likely is subject to the same evolutionary pressures in Cx. pipiens as Bti and other insecticides, we believe there is less potential for selection for resistance to the attractiveness of blackberry and honeysuckle leaves given that the components of infusions of these leaves (i.e., bacterial flora, plant compounds, and volatiles) are ubiquitous in the environment without the presence of toxins. Second, there is experimental evidence to suggest larvicidal botanicals are less likely to have effects on non-target organisms compared to conventional insecticides (Sivagnaname and Kalyanasundaram 2004, Shaalan et al.

2005), and thus use of blackberry leaves may provide a more sustainable management solution

106 for Cx. pipiens in storm water environments. Our findings also yield insights into why ecological traps occur. Studies of ecological traps are limited because such traps are difficult to identify, and particularly to distinguish from “sinks” in which a low-quality habitat is avoided. In the current study and our previous research (Gardner et al. 2015, see Chapter 2), we have documented the existence of an ecological trap for Cx. pipiens. We have provided experimental evidence that the combination of the bacterial communities that form as blackberry leaves decay and the leaching of secondary plant components during the microbial decomposition process work in tandem to create habitat that is attractive to adult female mosquitoes for oviposition yet deleterious to developing larvae.

Acknowledgments

We thank Allison Parker, Matt Edinger, Chase Robinson, Lauren Frisbie, Noor Malik, Jackie

Duple, Jake Gold, and Kristen Chen for their assistance with catch basin sampling in 2013 and

2014. We thank Brit’nee Haskins, Reanna Kayser, and Leah Overmier for their assistance with the oviposition field and cage bioassays in 2015, and Chang-Hyun Kim, Millon Blackshear,

Dohyup Kim, Katie Lincoln, and Leah Overmier for their assistance with DNA extractions. We thank the following individuals for their assistance with construction of mosquito emergence traps: Page Fredericks, Steven Huckett, Allison Parker, Erin Allmann Updyke, Jackie Duple,

Noor Malik, Chase Robinson, Jake Gold, Kristen Chen, Grace Cahill, Tara Stewart, and Brandon

Lieberthal. We also thank the DNA Services laboratory at the W. M. Keck Center for

Comparative and Functional Genomics at the University of Illinois at Urbana-Champaign for sequencing services and Ravikiran Donthu, Christopher Wright, and the High Performance

Biological Computing (HPCBio) group for instruction in use of the IM-TORNADO pipeline and

107 preparation of the bacterial OTU table. We thank the City of Paxton, Illinois, and homeowners in

Urbana, Illinois, for land use permission. This study was supported by a U.S. Environmental

Protection Agency (EPA) STAR Graduate Fellowship (award no. F13F11066) to AMG, a

University of Illinois Francis M. and Harlie M. Clark Summer Fellowship to AMG, an Illinois

Natural History Survey Philip W. Smith Memorial Award for Ecology, Evolution, and

Conservation Biology to AMG, a University of Illinois Institute for Sustainability, Energy, and

Environment seed grant to BFA, the Used Tire Fund and Emergency Public Health Act from the

State of Illinois to EJM, and an NIH NRSA Short-Term Reseach Training Grant (award no.

OD011145) to Lois Hoyer and the University of Illinois College of Veterinary Medicine for summer support for BH.

References Cited

Aird D, Ross MG, Chen WS, Danielsson M, Fennell T, Russ C, Jaffe DB, Nusbaum C, Gnirke

A. 2011. Analyzing and minimizing PCR amplification bias in Illumina sequencing

libraries. Genome Biology, 12, 18.

Anderson JF, Ferrandino FJ, Dingman DW, Main AJ, Andreadis TG, Becnel JJ. 2011. Control of

mosquitoes in catch basins in Connecticut with Bacillus thuringiensis israelensis,

Bacillus sphaericus, and spinosad. Journal of the American Mosquito Control

Association, 27, 45-55.

Andreadis TG, Thomas MC, Shepard JJ. 2005. Identification guide to the mosquitoes of

Connecticut (No. 966). Connecticut Agricultural Experiment Station.

108

Beehler JW, Millar JG, Mulla MS. 1993. Synergism between chemical attractants and visual

cues influencing oviposition of the mosquito, Culex quinquefasciatus (Diptera:

Culicidae). Journal of Chemical Ecology, 19, 635-644.

Beehler JW, Millar JG, Mulla MS. 1994. Field evaluation of synthetic compounds mediating

oviposition in Culex mosquitoes (Diptera: Culicidae). Journal of Chemical Ecology, 20.

Bentley MD, Day JF. 1989. Chemical ecology and behavioral aspects of mosquito

oviposition. Annual Review of Entomology, 34, 401-421.

Bolger AM, Lohse M, Usadel B. 2014. Trimmomatic: a flexible trimmer for Illumina sequence

data. Bioinformatics, 170.

Charmillot PJ, Hofer D, Pasquier D. 2000. Attract and kill: a new method for control of the

codling moth Cydia pomonella. Entomologia Experimentalis et Applicata, 94, 211-216.

Cipollini D, Stevenson R, Enright S, Eyles A, Bonelio P. 2008. Phenolic metabolites in leaves of

the invasive shrub, Lonicera maackii, and their potential phytotoxic and anti-herbivore

effects. Journal of Chemical Ecology, 34.

Cole JR, Wang Q, Cardenas E, Fish J, Chai B, Farris RJ, Kulam-Syed-Mohideen AS, McGarrell

DM, Marsh T, Garrity GM, Tiedje JM. 2009. The Ribosomal Database Project: improved

alignments and new tools for rRNA analysis. Nucleic Acids Research, 37, 141-5.

Cornelissen JH, Perez‐Harguindeguy N, Díaz S, Grime JP, Marzano B, Cabido M, Vendramini

F, Cerabolini B. 1999. Leaf structure and defence control litter decomposition rate across

species and life forms in regional floras on two continents. New Phytologist, 143, 191-

200.

David JP, Rey D, Pautou MP, Meyran JC. 2000. Differential toxicity of leaf litter to dipteran

larvae of mosquito developmental sites. Journal of Invertebrate Pathology, 75, 9-18.

109

Degnan PH, Ochman H. 2012. Illumina-based analysis of microbial community diversity. The

ISME Journal, 6, 183-194.

De Groot P, Nott R. 2001. Evaluation of traps of six different designs to capture pine sawyer

beetles (Coleoptera: Cerambycidae). Agricultural and Forest Entomology, 3, 107-111.

Du Y, Millar JG. 1999. Electroantennogram and oviposition bioassay responses of Culex

quinquefasciatus and Culex tarsalis (Diptera: Culicidae) to chemicals in odors from

Bermuda grass infusions. Journal of Medical Entomology, 36.

Elango G, Bagavan A, Kamaraj C, Zahir AA, Rahuman AA. 2009. Oviposition-deterrent,

ovicidal, and repellent activities of indigenous plant extracts against Anopheles subpictus

Grassi (Diptera: Culicidae). Parasitology Research, 105, 1567-1576.

Findlay SE, Arsuffi TL. 1989. Microbial growth and detritus transformations during

decomposition of leaf litter in a stream. Freshwater Biology, 21, 261-269.

Fish D, Carpenter SR. 1982. Leaf litter and larval mosquito dynamics in tree-hole ecosystems.

Ecology, 283-288.

Gardner AM, Hamer GL, Hines AM, Newman CM, Walker ED, Ruiz MO. 2012. Weather

variability affects abundance of larval Culex (Diptera: Culicidae) in storm water catch

basins in suburban Chicago. Journal of Medical Entomology, 49, 270-276.

Gardner AM, Anderson TK, Hamer GL, Johnson DE, Varela KE, Walker ED, Ruiz MO. 2013.

Terrestrial vegetation and aquatic chemistry influence larval mosquito abundance in catch

basins, Chicago, USA. Parasites and Vectors, 6, 9.

Gardner AM, Allan BF, Frisbie LA, Muturi EJ. 2015. Asymmetric effects of native and exotic

invasive shrubs on ecology of the West Nile Virus vector Culex pipiens (Diptera:

Culicidae). Parasites and Vectors, 16, 329.

110

Geery PR, Holub RE. 1989. Seasonal abundance and control of Culex spp. in catch basins in

Illinois. Journal of the American Mosquito Control Association, 5, 537-540.

Goddard LB, Roth AE, Reisen WK, Scott TW. 2002. Vector competence of California

mosquitoes for West Nile virus. Emerging Infectious Diseases, 8, 1385-1391.

Goldman IF, Arnold J, Carlton BC. 1986. Selection for resistance to Bacillus thuringiensis

subspecies israelensis in field and laboratory populations of the mosquito Aedes aegypti.

Journal of Invertebrate Pathology, 47, 317-324.

Gomez-Alvarez V, Teal TK, Schmidt TM. 2009. Systematic artifacts in metagenomes from

complex microbial communities. The ISME Journal, 3, 1314-1317.

Goslee SC, Urban DL. 2007. The ecodist package for dissimilarity-based analysis of ecological

data. Journal of Statistical Software, 22, 1-19.

Green M, Heumann M, Sokolow R, Foster LR, Bryant R, Skeels M. 1990. Public health

implications of the microbial pesticide Bacillus thuringiensis: an epidemiological study,

Oregon, 1985-86. American Journal of Public Health, 80, 848-852.

Gripenberg, S., Mayhew, P.J., Parnell, M. and Roslin, T., 2010. A meta‐analysis of preference–

performance relationships in phytophagous insects.Ecology Letters, 13(3), pp.383-393.

Hamer GL, Kelly PH, Focks DA, Goldberg TL, Walker ED. 2011. Evaluation of a novel

emergence trap to study Culex mosquitoes in urban catch basins. Journal of the American

Mosquito Control Association, 27, 142-147.

Harbison JE, Layden JE, Xamplas C, Zazra D, Henry M, Ruiz MO. 2015. Observed loss and

ineffectiveness of mosquito larvicides applied to catch basins in the northern suburbs of

Chicago IL, 2014. Environmental Health Insights, 9, 1.

111

Hayes EB, Gubler DJ. 2006. West Nile virus: Epidemiology and clinical features of an emerging

epidemic in the United States. Annual Review of Medicine, 57, 181-194.

Hazelrigg JE, Pelsue FW, Grant CD. 1980. A technique for controlling mosquito breeding in

underground storm drains using methoprene: Altosid (California SLN-780183).

Proceedings and Papers of the 48th Annual Conference of the California Mosquito and

Vector Control Association. CMVCA Press.

Hemingway J, Ranson H. 2000. Insecticide resistance in insect vectors of human disease. Annual

Review of Entomology, 45, 371-391.

Horner JD, Gosz JR, Cates RG. 1988. The role of carbon-based plant secondary metabolites in

decomposition in terrestrial ecosystems. American Naturalist, 869-883.

Isoe J, Millar JG, Beehler JW. 1995. Bioassays for Culex (Diptera: Culicidae) mosquito

oviposition attractants and stimulants. Journal of Medical Entomology, 32, 475-483.

Jeraldo P, Kalari K, Chen X, Bhavsar J, Mangalam A, White B, Nelson H, Kocher JP, Chia N.

2014. IM-TORNADO: A tool for comparison of 16S reads from paired-end

libraries. PLoS One, 9, 114804.

Jones KE, Patel NG, Levy MA, Storeygard A, Balk D, Gittleman JL, Daszak P. 2008. Global

trends in emerging infectious diseases. Nature, 451, 990-993.

Kominoski JS, Pringle CM, Ball BA, Bradford MA, Coleman DC, Hall DB, Hunter MD. 2007.

Nonadditive effects of leaf litter species diversity on breakdown dynamics in a detritus-

based stream. Ecology, 88, 1167-1176.

Kramer WL, Mulla MS. 1979. Oviposition attractants and repellents of mosquitoes: oviposition

responses of Culex mosquitoes to organic infusions. Environmental Entomology, 8, 1111-

1117.

112

Kruskal JB. 1964. Nonmetric multidimensional scaling: a numerical method. Psychometrika, 27,

115-29.

Kuiters AT, Sarink HM. 1986. Leaching of phenolic compounds from leaf and needle litter of

several deciduous and coniferous trees. Soil Biology and Biochemistry, 18, 475-480.

Lanciotti RS, Roehrig JT, Deubel V, Smith J, Parker M, Steele K, Crise B, Volpe KE, Crabtree

MB, Scherret JH, Hall RA. 1999. Origin of the West Nile virus responsible for an

outbreak of encephalitis in the northeastern United States. Science, 286, 2333-2337.

Leal WS, Barbosa RM, Xu W, Ishida Y, Syed Z, Latte N, Chen AM, Morgan TI, Cornel AJ,

Furtado A. 2008. Reverse and conventional chemical ecology approaches for the

development of oviposition attractants for Culex mosquitoes. PloS One, 3, 3045.

Legendre P, Lapointe F, Casgrain P. 1994. Modeling brain evolution from behavior: A

permutational regression approach. Evolution, 48, 1487-1499.

Lichstein J. 2007. Multiple regression on distance matrices: A multivariate spatial analysis tool.

Plant Ecology, 188, 117-131.

McArthur JV, Barnes JR. 1988. Community dynamics of leaf litter breakdown in a Utah alpine

stream. Journal of the North American Benthological Society, 37-43.

McCune B, Grace JB, Urban DL. 2002. Analysis of Ecological Communities (Vol. 28). Gleneden

Beach, OR: MjM Software Design.

Merritt RW, Dadd RH, Walker ED. 1992. Feeding behavior, natural food, and nutritional

relationships of larval mosquitoes. Annual Review of Entomology, 37, 349-374.

Mielke Jr. PW, Berry KJ, Johnson ES. 1976. Multi-response permutation procedures for a priori

classifications. Communications in Statistics Theory and Methods, 5, 1409-1424.

113

Milam CD, Farris JL, Wilhide JD. 2000. Evaluating mosquito control pesticides for effect on

target and nontarget organisms. Archives of Environmental Contamination and

Toxicology, 39, 324-328.

Mokany A, Shine R. 2003. Oviposition site selection by mosquitoes is affected by cues from

conspecific larvae and anuran tadpoles. Austral Ecology, 28, 33-37.

Monath TP. 2001. Prospects for development of a vaccine against the West Nile virus. Annals of

the New York Academy of Sciences, 951, 1-12.

Müller GC, Beier JC, Traore SF, Toure MB, Traore MM, Bah S, Doumbia S, Schlein Y. 2010.

Successful field trial of attractive toxic sugar bait (ATSB) plant-spraying methods against

malaria vectors in the Anopheles gambiae complex in Mali, West Africa. Malaria

Journal, 9, 210.

Mulligan FS, Schaefer CH, Wilder WH. 1980. Efficacy and persistence of Bacillus sphaericus

and B. thuringiensis against mosquitoes under laboratory and field conditions. Journal of

Economic Entomology, 73, 684-688.

Munstermann LE, Craig Jr. GB. 1976. Culex mosquito populations in the catch basins of

northern St. Joseph County, Indiana. Proceedings of the Indiana Academy of Science, 86,

246-252.

Muturi EJ, Allan BF, Ricci J. 2012. Influence of leaf detritus type on production and longevity of

container-breeding mosquitoes. Environmental Entomology, 41, 1062-1068.

O’Meara GF, Evans Jr. LF, Gettman AD, Cuda JP. 1995. Spread of Aedes albopictus and decline

of Ae. aegypti (Diptera: Culicidae) in Florida. Journal of Medical Entomology, 32, 554-

62.

114

Paris M, Tetreau G, Laurent F, Lelu M, Despres L, David JP. 2011. Persistence of Bacillus

thuringiensis israelensis (Bti) in the environment induces resistance to multiple Bti toxins

in mosquitoes. Pest Management Science, 67, 122-128.

Ponnusamy L, Xu N, Nojima S, Wesson DM, Schal C, Apperson CS. 2008. Identification of

bacteria and bacteria-associated chemical cues that mediate oviposition site preferences

by Aedes aegypti. Proceedings of the National Academy of Sciences, 105, 9262-9267.

Ponnusamy L, Wesson DM, Arellano C, Schal C, Apperson CS. 2010. Species composition of

bacterial communities influences attraction of mosquitoes to experimental plant

infusions. Microbial Ecology, 59, 158-173.

Rajkumar S, Jebanesan A. 2005. Oviposition deterrent and skin repellent activities of Solanum

trilobatum leaf extract against the malarial vector Anopheles stephensi. Journal of Insect

Science, 5, 15.

Rajkumar S, Jebanesan A. 2009. Larvicidal and oviposition activity of Cassia obtusifolia Linn

(Family: Leguminosae) leaf extract against malarial vector, Anopheles stephensi Liston

(Diptera: Culicidae). Parasitology Research, 104, 337-340.

Reiskind MH, Greene KL, Lounibos LP. 2009. Leaf species identity and combination affect

performance and oviposition choice of two container mosquito species. Ecological

Entomology, 34, 447-456.

Rey JR, O’Meara GF, O’Connell SM, Cutwa-Francis MM. 2006. Factors affecting mosquito

production from storm water drains and catch basins in two Florida cities. Journal of

Vector Ecology, 31, 334-343.

Rios JL, Recio MC, Villar A. 1988. Screening methods for natural products with antimicrobial

activity: a review of the literature. Journal of Ethnopharmacology, 23, 127-149.

115

Sardelis MR, Turell MJ, Andre RG. 2002. Laboratory transmission of La Crosse virus by

Ochlerotatus j. japonicus (Diptera: Culicidae). Journal of Medical Entomology, 39, 635-

639.

Scott TW, Hildreth SW, Beaty BJ. 1984. The distribution and development of eastern equine

encephalitis virus in its enzootic mosquito vector, Culiseta melanura. The American

Journal of Tropical Medicine and Hygiene, 33, 300-310.

Shaalan EAS, Canyon D, Younes MWF, Abdel-Wahab H, Mansour AH. 2005. A review of

botanical phytochemicals with mosquitocidal potential. Environment International, 31,

1149-1166.

Sivagnaname N, Kalyanasundaram M. 2004. Laboratory evaluation of methanolic extract of

Atlantia monophylla (Family: Rutaceae) against immature stages of mosquitoes and non-

target organisms. Memórias do Instituto Oswaldo Cruz, 99, 115-118.

Steven B, Gallegos‐Graves LV, Starkenburg SR, Chain PS, Kuske CR. 2012. Targeted and

shotgun metagenomic approaches provide different descriptions of dryland soil microbial

communities in a manipulated field study. Environmental Microbiology Reports, 4, 248-

256.

Stockwell PJ, Wessell N, Reed DR, Kronenwetter-Koepel TA, Reed KD, Turchi TR, Meece JK.

2006. A field evaluation of four larval mosquito control methods in urban catch basins.

Journal of the American Mosquito Control Association, 22, 666-671.

Thompson WH, Beaty BJ. 1977. Venereal transmission of La Crosse (California encephalitis)

arbovirus in Aedes triseriatus mosquitoes. Science, 196, 530-531.

116

Turell MJ, Dohm DJ, Sardelis MR, O’Guinn ML, Andreadis TG, Blow JA. 2005. An update on

the potential of North American mosquitoes (Diptera: Culicidae) to transmit West Nile

virus. Journal of Medical Entomology, 42, 57-62.

Vargas RI, Miller NW, Stark JD. 2003. Field trials of spinosad as a replacement for naled,

DDVP, and malathion in methyl eugenol and cue-lure bucket traps to attract and kill male

oriental fruit flies and melon flies (Diptera: Tephritidae) in Hawaii. Journal of Economic

Entomology, 96, 1780-1785.

Walker ED, Merritt RW, Kaufman MG, Ayres MP, Riedel MH. 1997. Effects of variation in

quality of leaf detritus on growth of the eastern tree-hole mosquito, Aedes triseriatus

(Diptera: Culicidae). Canadian Journal of Zoology, 75, 706-718.

Wallace JB, Merritt RW. 1980. Filter-feeding ecology of aquatic insects. Annual Review of

Entomology, 25, 103-132.

Wang SY, Lin HS. 2000. Antioxidant activity in fruits and leaves of blackberry, raspberry, and

strawberry varies with cultivar and developmental stage. Journal of Agriculture and Food

Chemistry, 48.

Webster DP, Farrar J, Rowland-Jones S. 2009. Progress towards a dengue vaccine. The Lancet

Infectious Diseases, 9, 678-687.

Yanoviak SP. 1999. Effects of leaf litter species on macroinvertebrate community properties and

mosquito yield in Neotropical tree hole microcosms. Oecologia, 120, 147-155.

117

Figures and Tables

Figure 3.1. Maps of catch basin treatments in Paxton, Illinois, in 2013 (left) and 2014 (right).

The treatments included a positive control, a negative control, Bti larvicidal briquettes (toxin only), honeysuckle leaves (attractants only), and blackberry leaves and honeysuckle leaves mixed with Bti (natural and artificial ecological traps, respectively).

118

Figure 3.2. Mean (± SE) number of Culex pipiens larvae (a proxy for oviposition rates; left), and adults (right) collected across treatments in catch basins in Paxton, Illinois over four-week study periods. The six treatments included a positive control, a negative control, Bti larvicidal briquettes (toxin only), honeysuckle leaves (attractants only), and blackberry leaves and honeysuckle leaves combined with Bti (ecological traps). Five replicates per treatment were measured in 2013 (top) and ten replicates per treatment were measured in 2014 (bottom).

119

Figure 3.3. Mean (± SE) number of Culex pipiens egg rafts collected across treatments from oviposition traps in Urbana, Illinois, over a four-week study period. The six treatments included a positive control, a negative control, Bti larvicidal briquettes (toxin only), honeysuckle leaves

(attractants only), and blackberry leaves and honeysuckle leaves combined with Bti (ecological traps).

120

Figure 3.4. Proportion of Culex pipiens egg rafts laid in binary choice behavioral cage bioassays for blackberry leaf infusion (top; purple) and honeysuckle leaf infusion (bottom; red). Pictured left is the configuration of treatment pairs in insect rearing cages (top), and absence (center) and presence (bottom) of egg rafts in experimental treatments.

121

Figure 3.5. Non-metric multidimensional scaling (NMDS) plot of 99 infusion samples from 30 oviposition traps in Urbana, Illinois over a four-week study period. The six treatments included a positive control, a negative control, Bti larvicidal briquettes (toxin only), honeysuckle leaves

(attractants only), and blackberry leaves and honeysuckle leaves combined with Bti (ecological traps).

122

Table 3.1. Linear contrasts comparing abundance of Culex pipiens larvae across catch basin treatments to test the hypothesis that inputs of attractive leaf detritus species increase mosquito oviposition rates in these habitats.

Catch basin treatment Estimate SE T P Blackberry vs. positive control 19.96 5.93 3.37 < 0.01 Honeysuckle vs. positive control 5.37 5.94 0.90 0.37 Honeysuckle+Bti vs. positive control -9.93 3.36 -2.95 < 0.01 Honeysuckle+Bti vs. Bti 0.27 3.33 0.08 0.94 Negative control vs. positive control -23.50 5.97 -3.94 < 0.01

123

Table 3.2. Linear contrasts comparing abundance of Culex pipiens adults across catch basin treatments to test the hypothesis that inputs of toxins (i.e., larvicides and deleterious leaf detritus species) reduce mosquito emergence rates in these habitats while inputs of high-quality leaf detritus resources increase emergence reates.

Catch basin treatment Estimate SE T P Blackberry vs. positive control -24.56 4.58 -5.36 < 0.01 Honeysuckle+Bti vs. positive control -11.08 2.63 -4.21 < 0.01 Bti vs. positive control -10.69 2.65 -4.03 < 0.01 Blackberry vs. negative control -2.76 4.58 -0.60 0.55 Honeysuckle+Bti vs. negative control -1.70 2.63 -0.65 0.52 Bti vs. negative control -1.32 2.65 -0.50 0.62 Honeysuckle vs. positive control 26.85 4.55 5.90 < 0.01 Negative control vs. positive control -21.80 4.55 -4.79 < 0.01

124

Table 3.3. Linear contrasts comparing number of Culex pipiens egg rafts across treatments in oviposition traps to test the hypothesis that inputs of attractive leaf detritus species increase mosquito oviposition rates in these habitats.

Treatment Estimate SE T P Blackberry vs. positive control 51.00 22.93 2.22 0.03 Honeysuckle vs. positive control 59.80 22.93 2.61 0.02 Honeysuckle+Bti vs. positive control 64.25 22.93 2.80 0.01 Honeysuckle+Bti vs. Bti 53.05 22.93 2.31 0.03 Negative control vs. positive control -28.20 22.93 -1.23 0.23

125

Table 3.4. Multiple response permutation procedure (MRPP) for the effect of catch basin treatment within week on 169 bacterial taxa. Grouping letters represent significant pairwise differences at 훼 = 0.05.

Test 1: Within week 1 Treatment group n Average distance Significance letters Blackberry 5 0.8193 A Positive control 5 0.8485 AB Honeysuckle 3 0.8905 B Honeysuckle+Bti 4 0.8278 B Bti 4 0.7898 B Negative control 3 0.7126 C Test 2: Within week 2 Treatment group n Average distance Significance letters Blackberry 5 0.7890 A Positive control 5 0.7633 AB Honeysuckle 3 0.8124 B Honeysuckle+Bti 4 0.8668 B Bti 5 0.7946 B Negative control 5 0.8820 C Test 3: Within week 3 Treatment group n Average distance Significance letters Blackberry 2 0.6881 A Positive control 5 0.8542 AB Honeysuckle 5 0.8978 B Honeysuckle+Bti 5 0.7865 B Bti 2 0.8287 B Negative control 5 0.7365 C Test 4: Within week 4 Treatment group n Average distance Significance letters Blackberry 5 0.8625 A Positive control 2 0.9578 AB Honeysuckle 5 0.8155 B Honeysuckle+Bti 5 0.9164 B Bti 4 0.9239 B Negative control 3 0.7591 C

126

Table 3.5. Multiple response permutation procedure (MRPP) for the effect of week within catch basin treatment on 169 bacterial taxa. Grouping letters represent significant pairwise differences at 훼 = 0.05.

Test 1: Blackberry Treatment group n Average distance Significance letters Week 1 5 0.8193 A Week 2 5 0.7890 B Week 3 2 0.6881 BC Week 4 5 0.8625 C Test 2: Honeysuckle Treatment group n Average distance Significance letters Week 1 3 0.8905 A Week 2 3 0.8124 B Week 3 5 0.8978 BC Week 4 5 0.8155 C Test 3: Honeysuckle + Bti Treatment group n Average distance Significance letters Week 1 4 0.8278 A Week 2 4 0.8668 B Week 3 5 0.7865 C Week 4 5 0.9164 C Test 4: Bti Treatment group n Average distance Significance letters Week 1 4 0.7898 A Week 2 5 0.7946 B Week 3 2 0.8287 B Week 4 4 0.9239 B Test 5: Positive control Treatment group n Average distance Significance letters Week 1 5 0.8485 A Week 2 5 0.7633 B Week 3 5 0.8542 B Week 4 2 0.9578 C Test 6: Negative control Treatment group n Average distance Significance letters Week 1 3 0.7126 A Week 2 5 0.8820 A Week 3 5 0.7365 A Week 4 3 0.7591 B

127

Table 3.6. Chi-square tests of significance for proportion of Culex pipiens egg rafts laid in binary choice behavioral cage bioassays for blackberry leaf infusion and honeysuckle leaf infusion.

Blackberry leaf infusion Treatment 1 Treatment 2 n χ2 P DI water Whole infusion 22 14.73 < 0.01 DI water Microbes-only 29 0.03 0.85 DI water Leachate-only 30 16.13 < 0.01 Microbes-only Whole infusion 21 10.71 < 0.01 Leachate-only Whole infusion 29 0.86 0.35 Honeysuckle leaf infusion Treatment 1 Treatment 2 n χ2 P DI water Whole infusion 24 6.00 0.01 DI water Microbes-only 22 0.18 0.67 DI water Leachate-only 22 10.00 < 0.01 Microbes-only Whole infusion 30 4.80 0.03 Leachate-only Whole infusion 36 1.00 0.32

128

CHAPTER 4

Removal of invasive Amur honeysuckle (Lonicera maackii) decreases abundance of

mosquitoes (Diptera: Culicidae)

Abstract

Invasive species rank second only to habitat destruction as a threat to native biodiversity, and one consequence of biological invasions is increased risk of exposure to infectious diseases in human and animal populations. The distribution and prevalence of mosquito-borne diseases in particular depend on the complex interactions between the vector, the pathogen, and the human or wildlife reservoir host. These interactions are highly susceptible to disturbance by invasive species, including terrestrial plants. This study examined the effect of removal of an invasive bush honeysuckle (Amur honeysuckle, Lonicera maackii) on the abundance of mosquitoes, including two regional vectors of West Nile virus (Culex pipiens and Cx. restuans), in a forest fragment embedded within a residential neighborhood. Our two-year field experiment used a Before-

After/Control-Impact design to find support for the hypothesis that removal of Amur honeysuckle reduces the abundance of both Culex spp. and non-Culex spp. “nuisance” mosquitoes. We assessed several ecological mechanisms that potentially explain observed differences in the mosquito community due to Amur honeysuckle removal, and found support for the hypotheses that 1) the abundance and composition of avian hosts is altered by honeysuckle removal and 2) areas invaded with honeysuckle support local microclimates (i.e., high temperature and humidity) that are favorable to mosquitoes. Collectively, our experiments demonstrate the role of a highly invasive understory shrub in determining the abundance and distribution of mosquitoes and suggest potential mechanisms underlying this pattern. Our results

129 also motivate questions regarding the effect of invasive plants on arthropod disease vector ecology, such as the generalizability of our results to other disease vector-invasive plant systems and the spatial scale at which removal of invasive plants may reduce the abundance of mosquitoes.

Introduction

Invasive species, defined here as non-native species whose introduction and subsequent spread are likely to cause economic or environmental harm, rank second only to habitat destruction as a threat to native biodiversity (Wilcove et al. 1998, Pimentel et al. 2005).

Mechanisms by which invasive species may reduce biodiversity include altering habitat structure, acting as predators and competitors of native species, and hybridizing with native species (Mooney and Cleland 2001, Lee 2002, Levine et al. 2003). There also is evidence that invasive species may cause increased risk of exposure to infectious diseases in wildlife and human populations via the introduction of novel pathogens and parasites into susceptible populations (Daszak et al. 2000) and by altering the ecology of established infectious agents that are embedded in biological communities through complex networks of energy flow (Pejchar and

Mooney 2009). The distribution and prevalence of arthropod-borne diseases depend on the interactions between an arthropod vector, the pathogen, and a human or wildlife reservoir host, and these interactions may be facilitated by invasive species.

In particular, invasive plants are among the most important threats to native ecosystems and recent research has demonstrated their potential to trigger a cascade of ecological interactions that alter vector-borne disease transmission (Mack and Smith 2011, Mackay et al.

2016). This is a problem of global health significance because nearly one-third of emerging

130 infectious diseases that threaten ecosystem health are vector-borne (Jones et al. 2008) and terrestrial plant invasions are increasingly widespread phenomena aided by human activities

(Reichard and White 2001). Examples from diseases vectored by ticks (Acari: Ixodidae) include that invasive bush honeysuckle (Lonicera maackii, Dipsacales: Caprifoliaceae; hereafter

‘honeysuckle’) appears to enhance the risk of exposure to human tick-borne ehrlichiosis

(Ehrlichia chaffeensis and E. ewingii) by increasing the abundance and encounter frequency of lone star ticks (Amblyomma americanum) and their white-tailed deer (Odocoileus virginianus) hosts (Allan et al. 2010). Similarly, invasive Japanese barberry (Berberis thunbergii) directly improves habitat quality and recruitment of the black-legged tick (Ixodes scapularis) via increased leaf litter depth and soil moisture (Elias et al. 2006).

Mosquitoes (Diptera: Culicidae) are another group of important arthropod vectors of wildlife and human pathogens, and may interact with plants throughout their life cycles. Thus invasive plants also may influence risk of exposure to mosquito-borne illnesses, although the potential mechanisms by which plants alter vector and host ecology and the consequences for ecosystem health remain poorly understood. During the juvenile stages of the mosquito life cycle, inputs of fruits and leaf detritus into the aquatic habitat provide an energy base for developing larvae; leaf detritus type and quantity determine the composition and abundance of bacterial and fungal flora that form in container habitats as the microbes break down the leaf litter (Fish and Carpenter 1982, and see Chapter 2). In turn, these bacteria and fungi provide a direct food source for mosquito larvae (Walker et al. 1991, Kitching 2001). Several studies have suggested that leaf detritus from invasive plants, including honeysuckle and autumn olive

(Elaeagnus umbellata), may provide high-quality nutritional resources for larvae of two West

Nile virus vectors (Culex pipiens and Cx. restuans) and two La Crosse encephalitis vectors

131

(Ochlerotatus japonicus and Oc. triseriatus) compared to native plants (Shewhart et al. 2014,

Gardner et al. 2015, Muturi et al. 2015).

As adults, mosquitoes use plants as resting sites as well as sources of sugar meals. The high vegetation biomass characteristic of plant invasions may provide a rich supply of sugar meals and retain heat and humidity in the understory, both contributing to mosquito longevity

(Foster 1995, Delatte 2009). This potentially has consequences for transmission rates of mosquito-borne pathogens, since transmission is only possible when the vector survives long enough for an ingested pathogen to complete the extrinsic incubation period. Moreover, the ecology of some mosquitoes such as Culex spp. may be influenced by the effects of invasive plants on the community of their passerine bird hosts. For example, an important reservoir host for West Nile virus, the American robin (Turdus migratorius), is a frugivore and may select habitats infested with fruit-bearing invasive plants (Ingold and Craycraft 1983), causing mosquitoes to aggregate in these areas (Diuk-Wasser et al. 2010) with potential downstream consequences for disease transmission.

In central Illinois and much of the midwestern and northeastern United States, one of the dominant invasive plants is Amur honeysuckle, an invasive woody shrub that arrived in North

America from Korea during the mid-1800s and subsequently spread throughout 27 states (Luken and Thieret 1996). Due to its long growing season, allelopathic effects, and dense structure which blocks sunlight and inhibits the growth of competitors, honeysuckle has outcompeted native understory plants throughout the region (Miller and Gorchov 2004, Dorning and Cipollini

2006). While the effects of honeysuckle on insect and vertebrate communities are incompletely understood, there is evidence that this plant can alter the abundance and composition of bird and rodent populations including the reservoir hosts of several vector-borne diseases (McCusker et

132 al. 2010, Mattos and Orrock 2010). Honeysuckle leaf detritus is an attractive and highly nutritious food resource for mosquitoes, alleviating the effects of intraspecific competition on adult emergence rates of Cx. pipiens (Gardner et al. 2015), and mixing honeysuckle leaves with detritus of native leaf species accelerates the decomposition rate of the native leaves, further enhancing the quality of aquatic habitats for developing larvae (Arthur et al. 2012, Poulette and

Arthur 2012).

In this study, we tested the hypotheses that removal of honeysuckle in a forest fragment embedded within a residential neighborhood 1) reduces the abundance of Culex pipiens and Cx. restuans, two important regional vectors for West Nile virus (Turell et al. 2001, Turell et al.

2005); and 2) reduces the abundance of non-Culex spp. mosquitoes (i.e., belonging to the genera

Aedes, Ochlerotatus, Anopheles, Culiseta, Coquillettidia, Orthopodomyia, Uranotaenia, and

Psorophora at our study locations) including several common “nuisance” species that feed preferentially and aggressively on human hosts. Additionally, we assessed several ecological mechanisms that potentially explain observed differences in the mosquito community due to honeysuckle removal. Specifically, we tested the hypotheses that 1) the abundance and species diversity of bird populations is altered by honeysuckle removal, and 2) microclimate conditions relevant to mosquito life history characteristics, particularly vapor pressure deficit (i.e., the difference between the amount of moisture in the air and how much moisture the air can hold when saturated), vary with honeysuckle density.

Methods

Experimental design. Our honeysuckle removal experiment was conducted using a Before-

After/Control-Impact design (Smith 2002) at two sites in Mahomet, Illinois (40°11’N, 88°24’W),

133 over a two-year period. We selected the study area largely because land use permits could be obtained at two highly comparable sites. The first site (randomly selected to be the ‘Control’ site) was 6.4 acres and the second site (randomly selected to be the ‘Treatment’ site) was 4.1 acres.

Both study sites consisted of oak-hickory forest embedded in residential neighborhoods within

2.5 km of each other (Figure 4.1a). Prior to the beginning of the experiment, the density of honeysuckle at each location was estimated by counting the number of stems greater than 2 cm in diameter in five 5 X 5 m plots randomly selected and spaced throughout the study sites. In the

Control site, the estimated honeysuckle density was 0.62 ± 0.09 stems/m2, and in the Treatment site the estimated honeysuckle density was 1.77 ± 0.20 stems/m2.

In 2014, response variables described below were measured with honeysuckle intact at both sites to provide a baseline for the main effect of site. During the winter of 2015, the honeysuckle at the Treatment site was removed (Figure 4.2). In 2015, response variables were measured again with honeysuckle removed at the Treatment site and honeysuckle intact at the

Control site to provide a baseline for the main effect of year. The honeysuckle removal effort was conducted by volunteers over seven days for a total of 620 personnel hours of labor, using chainsaws and spraying the stumps with 50% Drexel Imitator Plus® glyphosate herbicide.

Detritus was mechanically chipped to eliminate any effect on the biological community potentially caused by the remaining structure of dead branches. During the summer of 2015, honeysuckle suppression was maintained by spraying the foliage of resprouts with 5% glyphosate once per week.

Mosquito trapping. Adult mosquito trapping was conducted eleven times per year at each site from CDC MMWR weeks 27 to 37 (29 June to 14 September 2014 and 28 June to 13 September

134

2015). Five pairs of CDC CO2-baited miniature light traps and grass infusion-baited gravid traps powered by 6V lantern batteries were operated in each of the two study sites. The two trap types were placed at locations approximately 10 m apart and the distance to the next pair of traps was at least 100 m, with light traps at 1.5 m above the ground (Figure 4.1b-c). Light traps are designed to capture a representative cross-section of host-seeking mosquitoes, including males and females of mixed parity, while gravid traps are designed to collect female mosquitoes seeking stagnant water sources to oviposit (McCardle et al. 2004). Mosquito collections were conducted one night per week, and the average trap night lasted from 1600h to 1100h. If the trap malfunctioned any given night, the data were discarded and denoted as ‘NA’ for statistical analyses. Adults were identified morphologically to species using dichotomous keys (Darsie and

Ward 2005), and Cx. pipiens and Cx. restuans were pooled for statistical analyses due to difficulty in accurately distinguishing their morphology.

To test the hypothesis that removal of honeysuckle reduces the abundance of adult Cx. pipiens and Cx. restuans, separate general linear mixed models (GLMMs) with repeated measures were fit to the number of mosquitoes per trap for light trap and gravid trap data using

PROC MIXED in SAS 9.3 (SAS Institute Inc., Cary NC). The linear model was given by

푌푖푗푘푙 = 휇 + 푆푖 + 푌푗 + 푆푌푖푗 + 휀1(푖푗푘) + 푊(푗)푙 + 푆푊(푗)푖푙 + 휀2(푖푗푘푙) (Eqn. 4.1) where 푆푖 represents the fixed main effect of site, 푌푗 represents the fixed main effect of year, 푆푌푖푗 represents the interaction between the main effects of site and year, 푊(푗)푙 represents the fixed main effect of week nested within year, and 푆푊(푗)푖푙 represents the interaction between the main effects of site and week. Trap ID was the subject, week was the repeated variable, and an 퐴푅(1) covariance structure with restricted maximum likelihood (REML) was used for estimation of covariance parameters. A significant site by year interaction was expected to indicate an effect of

135 honeysuckle removal on the response variable. SLICE statements were used to interpret significant interaction effects. To test the hypothesis that removal of honeysuckle reduces the abundance of adult non-Culex spp. mosquitoes, the same linear models (Eqn. 4.1) were carried out using the abundance of non-Culex spp. mosquitoes as the response.

Host community. To examine the effect of honeysuckle removal on the abundance of avian hosts for Culex spp. mosquitoes, bird surveys were performed at the Control and Treatment sites three times in both 2014 and 2015 using limited radius (100 m) ten minute auditory and visual survey point counts following the methods of Bibby et al. (2000). One point was established at a central location in each site. All surveys were conducted between 0.5 hours before sunrise and

0.5 hours after sunrise on days with no precipitation or extreme wind. The dates of the surveys, corresponding with the avian breeding season and dates of mosquito trapping, were 31 July, 20

August, and 16 September in 2014 and 10 July, 20 August, and 12 September in 2015. The species and estimated distance to each observed bird were recorded.

To test the hypothesis that removal of honeysuckle reduces the abundance of birds, a two-way Analysis of Variance (ANOVA) model was fit to the bird abundance data with the main effects of site and year and the interaction between the main effects of site and year. Again, a significant site by year interaction would indicate an effect of honeysuckle removal on the abundance of birds. SLICE statements were used to interpret the interaction.

Microclimate conditions. To assess the effect of honeysuckle density on vapor pressure deficit, iButton temperature/humidity loggers (Maxim Integrated, San Jose CA, catalog no. DS1923) were deployed from 29 August to 8 September 2015, with simultaneous temperature (°C) and

136 relative humidity (%) measurements continuously recorded at 10 minute intervals. The vapor pressure deficit (Pa) was calculated for each time point from these temperature and humidity values as follows

100−푅퐻 푉푃퐷 = ( ) (610.7 × 107.5푇/(237.3+푇)) (Eqn. 2) 100 where 푅퐻 represents relative humidity (%) and 푇 represents temperature (°C) (Eqn. 2). Four data loggers were established at the Treatment site and eight data loggers were established at the

Control site. To capture the effect of fine-scale variation in honeysuckle density on microclimate conditions, half of the loggers at the Control site were placed in locations that were heavily invaded with honeysuckle while half were placed in locations where honeysuckle was comparatively sparse. Data loggers were suspended at 1.5 m above the ground.

To test the hypothesis that vapor pressure deficit varies with honeysuckle density, a multivariate time series model was fit to the time series for the Treatment site and high honeysuckle density and low honeysuckle density areas of the Control site using robust linear regression and a novel application of the generalized method of wavelet moments to model the error processes (described in Guerrier and Gardner in prep.). The final model was given by

2 3 푌푖푗푘 = 휇푖 + 휇푖푗 + 푆푗 + 훽1푡 + 훽2푡 + 훽3푡 + 휀푖푗푘 (Eqn. 3) where 휇푖 represents the grand mean, 휇푖푗 represents the mean for each sensor, 푆푗 represents the

2 3 fixed main effect of site (i.e., Treatment, low density, high density), and 훽1푡, 훽2푡 , and 훽3푡 represent the linear, quadratic, and cubic components of a fixed temperature covariate, with the error term defined by

(푗) (푖푗) 휀푖푗푘 = 푍푡 + 푊푡 + 푌푡 (Eqn. 4)

(푗) where 푍푡 represents a global 퐴푅(2) process estimated for all sites, 푊푡 represents a local

137

(푗) 2 (1) (2) 2 2 퐴푅1(휙 , 휎(푗)) process estimated for each site separately with 휙 ≠ 휙 and 휎1 ≠ 휎2 , and

(푖푗) 푌푡 represents a white noise process associated with individual sensor error. A permutation test with 1000 Monte Carlo iterations was used to test the significance of the site effect.

Results

Mosquito composition. We collected a total of 31,837 adult mosquitoes over the two-year study period (Table 4.1). At the Control site, we collected 7,872 mosquitoes in 2014 and 8,659 mosquitoes in 2015, and at the Treatment site, we collected 13,816 mosquitoes in 2014 and

1,490 mosquitoes in 2015. Approximately 30 percent of the mosquitoes collected at each site belonged to the genus Culex while 70 percent were non-Culex spp. Among the non-Culex spp. mosquitoes, over 97 percent belonged to seven species: Aedes triseriatus, Ae. trivittatus, Ae. vexans, Ochlerotatus japonicus, Psorophora ferox, Anopheles punctipennis, and An. quadrimaculatus. Other infrequently observed species were An. walkeri, Coquillettidia perturbans, Culiseta inornata, Cs. morsitans, Orthopodomyia signifera, Uranotaenia sapphirina, and Ps. ciliata.

Mosquito abundance. Abundance of adult Cx. pipiens and Cx. restuans varied in response to honeysuckle removal (Figure 4.3). We found a significant interaction between the main effects of site and year on the abundance of Culex spp. mosquitoes collected in both light traps (F =

147.76; df = 1, 16; P < 0.01) and gravid traps (F = 111.11; df = 1, 16; P < 0.01) (Table 4.2).

Contrasts demonstrated that while abundance was higher at the Treatment site compared to the

Control site in 2014, after honeysuckle removal the abundance was lower at the Treatment site compared to the Control site in 2015 (Table 4.3). There was no significant difference in Culex

138 spp. abundance at the Control site between 2014 and 2015, whereas the Treatment site had lower abundance in 2015 after honeysuckle removal. We also found a significant interaction between the main effects of site and week nested within year on the abundance of Culex spp. mosquitoes collected in both light traps (F = 4.12; df = 20, 150; P < 0.01) and gravid traps (F = 2.77; df = 20,

154; P < 0.01). Here, contrasts demonstrated that this interaction reflected equal mosquito abundance (i.e., no mosquitoes) across sites during the final week of the experiment during each year.

Similarly, abundance of non-Culex spp. mosquitoes varied in response to honeysuckle removal (Figure 4.4). Again we found a significant interaction between the main effects of site and year on the abundance of non-Culex spp. mosquitoes collected in both light traps (F =

123.35; df = 1, 16; P < 0.01) and gravid traps (F = 5.17; df = 1, 16; P = 0.04) (Table 4.2).

Contrasts showed that while abundance was higher at the Treatment site compared to the Control site in 2014, after honeysuckle removal the abundance was lower at the Treatment site compared to the Control site in 2015 (Table 4.3). There also was no significant difference in non-Culex spp. abundance at the Control site between 2014 and 2015, while the Treatment site had lower abundance in 2015 after honeysuckle removal. Finally, we found a significant interaction between the main effects of site and week nested within year on the abundance of non-Culex spp. mosquitoes collected in light traps (F = 8.05; df = 20, 150; P < 0.01) but not gravid traps (F =

0.94; df = 20, 154; P = 0.53). The interpretation of the significant interaction for the light traps was the same as for the Culex spp. mosquitoes.

Host abundance. Abundance of birds varied in response to honeysuckle removal (Table 4.44).

We found a significant interaction between the main effects of site and year on the abundance of

139 birds (F = 30.86; df = 1, 8; P < 0.01; Figure 4.6). Contrasts showed that while bird abundance was equal at the Treatment and Control sites in 2014 (F = 0.05; df = 1, 8; P = 0.83), after honeysuckle removal the abundance was lower at the Treatment site compared to the Control site in 2015 (F = 58.33; df = 1, 8; P < 0.01). There also was no significant difference in bird abundance at the Control site between 2014 and 2015 (F = 3.05; df = 1, 8; P = 0.12), while the

Treatment site had lower abundance in 2015 after honeysuckle removal (F = 37.33; df = 1, 8; P <

0.01).

Microclimate conditions. Mean vapor pressure varied between the Treatment site and locations within the Control site with high honeysuckle density (Figure 5). We found vapor pressure deficit was significantly higher at the Treatment site following honeysuckle removal compared to the Control site (P = 0.01), indicating honeysuckle creates a more stable microclimate for mosquito survival. However, we found no significant difference in the vapor pressure deficit observed in the high-density versus low-density areas of the Control site, indicating a lack of an effect of fine-scale variation in honeysuckle density on microclimate. Overall, the majority of error variance at the Control site was described by the global 퐴푅(2) process (~72%) compared to the local 퐴푅(1) process (~27%) and the white noise process associated with individual sensors (~1%), while the error variance at the Treatment site was more evenly described by the global 퐴푅(2) process (~50%) and the local 퐴푅(1) process (~49%) and the white noise process associated with individual sensors (~1%).

Discussion

Terrestrial plant invasions are emerging as increasingly widespread phenomena aided by

140 human activities, including global climate change (Reichard and White 2001, Hellmann et al.

2008). Previous studies suggest that the ecological disturbances associated with these plants can alter the transmission dynamics of arthropod-borne pathogens (Pejchar and Mooney 2009). In this study, we tested several hypotheses regarding the effect of removal of Amur honeysuckle, an invasive understory shrub that occurs throughout the midwestern United States (Luken and

Thieret 1996), on mosquito abundance and composition in a forest fragment within a residential neighborhood. We found support for the hypotheses that honeysuckle removal greatly reduces the abundance of two regional mosquito vectors for West Nile virus, Culex pipiens and Cx. restuans, as well as other species of medical importance (e.g., Ochlerotatus japonicus and Aedes triseriatus) and nuisance species that preferentially and aggressively feed on human hosts (e.g.

Aedes vexans, Oc. trivittatus). These results join a growing body of literature that suggests that the ecosystem services provided by native plants – and the benefits of management of invasive species – may extend beyond oft-considered factors such as nutrient cycling, prevention of stream erosion, air filtration, and preservation of wildlife diversity (Loomis et al. 2000, Fiedler et al. 2008, Chazdon 2008) and include direct ramifications for human and wildlife risk of exposure to vector-borne pathogens.

We also examined several ecological mechanisms that potentially explain observed differences in the mosquito community due to honeysuckle removal. We found support for the hypothesis that the abundance and composition of bird populations, the most important blood- meal hosts for Cx. pipiens and Cx. restuans (Molaei et al. 2006, Hamer et al. 2009), are altered by honeysuckle removal. These results are consistent with studies conducted in other arthropod- borne disease systems that suggest invasive plants may increase the abundance and encounter frequency between vectors and their hosts (Williams et al. 2009, Allan et al. 2010). Second, we

141 found support for the hypothesis that microclimate conditions relevant to mosquito life history traits, particularly vapor pressure deficit, varies with honeysuckle density. This result is important because vapor pressure deficit is related to mosquito longevity (Delatte 2009) with potential consequences for risk of exposure to West Nile virus because only a vector that becomes infected and survives the extrinsic incubation period is capable of transmitting a pathogen. This result is also consistent with previous observations in tick-borne disease systems, where changes in microclimate conditions after management of the invasive Japanese barberry was associated with a reduction in the abundance of Ixodes scapularis ticks (Williams and Ward

2010).

Our current study represents a set of exploratory, hypothesis-generating experiments that provide initial evidence for the role of a widespread invasive plant in mediating the abundance and distribution of mosquito disease vectors. Future research in this system is required to assess the relative importance of the mechanisms we have identified to explain the response of the mosquito community to honeysuckle removal, and additional mechanisms remain to be addressed experimentally (e.g., the effect of honeysuckle on mosquito sugar-feeding rates and availability of resting habitat for gravid females). In addition, the findings of our two-year field experiment suggest other questions critical to understanding the impact of invasive species on the complex ecology of mosquito-borne disease transmission. Here we discuss these avenues that require further research.

First, an important question that extends beyond the scope of the current study is whether our observations were primarily due to the effect of removal of honeysuckle in particular or the effect of removal of over 80 percent of the understory. Due to its long growing season, allelopathic effects, and dense structure that blocks sunlight to inhibit the growth of competitors,

142 honeysuckle is a noxious invader that, once introduced, often dominates local plant assemblages

(Miller and Gorchov 2004, Dorning and Cipollini 2006). So pernicious is honeysuckle as an invader that in Champaign County we were unable to identify a forest fragment in a residential environment that was not dominated by honeysuckle as a basis to compare the effects of native plants versus honeysuckle on mosquito communities. The duration and budget of our study precluded the feasibility of a multi-year experiment investigating the effects of restoration of native plants on mosquito abundance at the Treatment site. However, we hypothesize that many of our proposed mechanisms by which honeysuckle removal alters mosquito abundance are in fact particular to honeysuckle because invasions of this plant are characterized by a denser understory compared to native shrub species (Hutchinson and Vankat 1997, White et al. 2009).

Prior research has provided mixed evidence regarding the effect of honeysuckle on avian community composition, nesting behavior, nest predation, and food resource availability (Ingold and Craycraft 1983, Schmidt and Whelan 1999), although a recent study conducted in Vermilion,

Piatt, and Champaign Counties, Illinois, showed that honeysuckle is associated with an increased density of understory bird species and frugivores, including several common hosts for Cx. pipiens and Cx. restuans and reservoirs for West Nile virus (McCusker et al. 2010). Honeysuckle also has been linked to local increases in temperature and humidity in the understory compared to ecosystems dominated by native plants (Hartman and McCarthy 2004, Loomis and Cameron

2014).

Second, a critical question motivated by the current study is to what extent our results can be generalized to other invasive terrestrial plants and mosquito-borne disease systems. We first argue that these findings are important even if not generalizable because honeysuckle is such a significant invasive plant in the United States, and the effects of honeysuckle on the environment

143 have been experimentally studied in contexts as diverse as leaf decay in streams, microbial flora in soil, and foraging of large mammals (Ehrenfeld et al. 2001, Arthur et al. 2012, Castellano and

Gorchov 2013). Several other studies have documented facilitation of ticks and tick-borne pathogens by invasive plants (Elias et al. 2006, Williams et al. 2009, Williams and Ward 2010).

However, the hypothesized biological mechanisms underlying these patterns have varied and even often conflicted. For example, Allan et al. (2010) found no effect of abiotic factors related to invasive plant removal on the abundance and recruitment of A. americanum, while Civitello et al. (2008) found ground-level temperature and humidity positively associated with the presence of Japanese stiltgrass (Microstegium vimineum) and the survival of A. americanum and

Dermacentor variabilis ticks. A study of oviposition behavior of Oc. triseriatus found that habitats containing leaf detritus of honeysuckle are not attractive to these mosquitoes (Conley et al. 2011), while our own research found that habitats containing honeysuckle leaf detritus are attractive to Cx. pipiens and Cx. restuans (Gardner et al. 2015, Muturi et al. 2015). Our own research also has shown that not all mosquito species respond similarly to invasive plants, and while honeysuckle leaf detritus provides a high-quality resource base for Cx. pipiens, it is not a superior source of food for Ae. triseriatus and Oc. japonicus compared to native plants (Muturi et al. 2015). And finally, our prior research has demonstrated that the effect of invasive plants on mosquito production is not uniform; while honeysuckle provided a high-quality resource to Cx. pipiens, leaf detritus from multiflora rose (Rosa multiflora) in fact was deleterious to larvae

(Gardner et al. 2015). Collectively, this limited number of studies suggests that generalizations regarding the effect of invasive plants on vector ecology should be made with caution, and an improved mechanistic understanding is required to inform broader conclusions regarding the net impact of invasive plants on human and wildlife risk of exposure to vector-borne disease.

144

Third, another important area for future research is analysis of the spatial scale at which removal of invasive honeysuckle reduces the abundance of mosquitoes. The current study ffocused exclusively on the abundance and composition of mosquitoes within two sites. While our experiment provides quantitative and fairly conclusive evidence that honeysuckle increases the abundance of Cx. pipiens, Cx. restuans, and other mosquito species within-patch, the effect of honeysuckle removal at the landscape level remains unexplored. An increased abundance of mosquitoes within honeysuckle-dominated forest fragments may yield a net decrease in mosquito abundance in the surrounding residential neighborhoods and other natural areas where this invasive plant is not present. Alternatively, the presence of honeysuckle could lead to a

“spillover” effect resulting in increased abundance of mosquitoes in adjacent sites due to the proximity to honeysuckle within mosquito dispersal ranges. Monitoring across the landscape, including sites within honeysuckle-invaded patches as well as sites along a distance gradient from invaded patches, is required to more fully understand the spatial scale of the effect of honeysuckle on mosquito abundance. Furthermore, because honeysuckle has a longer

“photosynthetic” season than the majority of native plants (McEwan et al. 2009) and continues to provide favorable microclimate conditions, sugar-feeding resources (i.e., berries), and habitat for hosts, its presence potentially could lengthen the active time period for Cx. pipiens and other mosquitoes of medical and veterinary importance.

Finally, a question that was not addressed by the current study is the implications of our results for disease transmission. Vector presence certainly is a prerequisite for mosquito-borne disease transmission, and the importance of entomological risk is well-documented in other systems (Scott and Morrison 2003). However, there is not necessarily a linear correspondence between mosquito abundance and transmission rates. For instance, previous studies have

145 documented a spatial mismatch between density of West Nile virus vectors and the locations of human cases (Ruiz et al. 2004), indicating the importance of environmental and host-related factors in determining risk of exposure to mosquito-borne pathogens. However, at least one study of tick-borne diseases documented a decline in tick abundance and a concomitant decrease in tick infection rates with Ehrlichia chaffeensis and E. ewingii after removal of an invasive plant

(Allan et al. 2010). The reduced abundance of avian reservoir hosts for West Nile virus observed at the Treatment location after honeysuckle removal suggests the possibility of reduced West

Nile virus prevalence in mosquitoes, although testing is required to make a firmer assessment of the implications of invasive plant removal for transmission. In summary, our experiments demonstrate the role of an invasive understory shrub in determining the abundance and distribution of mosquitoes, including species of medical importance, suggest several potential mechanisms underlying this pattern, and motivate a number of questions regarding the effect of invasive plants on arthropod disease vector ecology that require additional sustained research efforts.

Acknowledgments

We thank Jackie Duple, Noor Malik, Chase Robinson, Jake Gold, Kristen Chen, Leah Overmier,

Brit’nee Haskins, and Reanna Kayser for their assistance with mosquito trapping in 2014 and

2015. We thank Leah Overmier for her additional assistance with mosquito identification and processing. We gratefully acknowledge Dr. Stéphane Guerrier and James Balamuta (University of Illinois Department of Statistics) for their analysis of the vapor pressure deficit time series data and Dr. Michael Ward (University of Illinois Department of Natural Resources and

Environmental Sciences) for discussing the analysis of the bird abundance data. We thank the

146 following individuals and organizations for their assistance with the honeysuckle removal during the winter of 2015: Jay Hayek, Tom Schmeelk, Natalie Pawlikowski, Dr. Andrew Mackay, Dan

Swanson, Tyler Hedlund, Jocelyn Sullivan, Eric South, Dr. Brandon Lieberthal, the University of

Illinois Extension Master Naturalists and Master Gardeners of Champaign County, the

University of Illinois Entomology Graduate Students Association, the University of Illinois Crop

Sciences Graduate Student Organization, and students in the University of Illinois Conservation

Biology course. We thank Mahomet-Seymour High School and the Village of Mahomet, Illinois for land use permission. This study was supported by a U.S. Environmental Protection Agency

(EPA) STAR Graduate Fellowship (award no. F13F11066) to AMG, a University of Illinois

Francis M. and Harlie M. Clark Summer Fellowship to AMG, an Illinois Natural History Survey

Philip W. Smith Memorial Award for Ecology, Evolution, and Conservation Biology to AMG, a

University of Illinois Institute for Sustainability, Energy, and Environment grant to BFA, and

Illinois Waste Tire and Emergency Public Health Funds to EJM.

References Cited

Anderson NH, Sedell JR. 1979. Detritus processing by macroinvertebrates in stream ecosystems.

Annual Review of Entomology, 24, 351-377.

Allan BF, Dutra HP, Goessling LS, Barnett K, Chase JM, Marquis RJ, Pang G, Storch GA,

Thach RE, Orrock JL. 2010. Invasive honeysuckle eradication reduces tick-borne disease

risk by altering host dynamics. Proceedings of the National Academy of Sciences

USA, 107, 18523-18527.

147

Arthur MA, Bray SR, Kuchle CR, McEwan RW. 2012. The influence of the invasive shrub,

Lonicera maackii, on leaf decomposition and microbial community dynamics. Plant

Ecology, 213, 1571-1582.

Bibby CJ. 2000. Bird census techniques. Elsevier, Philadelphia PA.

Castellano SM, Gorchov DL. 2013. White-tailed deer (Odocoileus virginianus) disperse seeds of

the invasive shrub, Amur honeysuckle (Lonicera maackii). Natural Areas Journal, 33,

78-80.

Chazdon RL. 2008. Beyond deforestation: restoring forests and ecosystem services on degraded

lands. Science, 320, 1458-1460.

Civitello DJ, Flory SL, Clay K. 2008. Exotic grass invasion reduces survival of Amblyomma

americanum and Dermacentor variabilis ticks (Acari: Ixodidae). Journal of Medical

Entomology, 45, 867-872.

Conley AK, Watling JI, Orrock JL. 2011. Invasive plant alters ability to predict disease vector

distribution. Ecological Applications, 21, 329-334.

Darsie RF, Ward RA. 1981. Identification and geographical distribution of the mosquitoes of

North America, north of Mexico. Walter Reed Army Institute of Research, Washington

D.C.

Daszak P, Cunningham AA, Hyatt AD. 2000. Emerging infectious diseases of wildlife – threats

to biodiversity and human health. BioScience, 287, 443-449.

Delatte H, Gimonneau G, Triboire A, Fontenille D. 2009. Influence of temperature on immature

development, survival, longevity, fecundity, and gonotrophic cycles of Aedes albopictus,

vector of chikungunya and dengue in the Indian Ocean. Journal of Medical

Entomology, 46, 33-41.

148

Diuk-Wasser MA, Molaei G, Simpson JE, Folsom-O’Keefe CM, Armstrong PM, Andreadis TG.

2010. Avian communal roosts as amplification foci for West Nile virus in urban areas in

northeastern United States. American Journal of Tropical Medicine and Hygiene, 82,

337-343.

Dorning M, Cipollini D. 2006. Leaf and root extracts of the invasive shrub, Lonicera maackii,

inhibit seed germination of three herbs with no autotoxic effects. Plant Ecology, 184,

287-296.

Ehrenfeld JG, Kourtev P, Huang W. 2001. Changes in soil functions following invasions of

exotic understory plants in deciduous forests. Ecological Applications, 11, 1287-1300.

Elias SP, Lubelczyk CB, Rand PW, Lacombe EH, Holman MS, Smith RP. 2006. Deer browse

resistant exotic-invasive understory: an indicator of elevated human risk of exposure to

Ixodes scapularis (Acari: Ixodidae) in southern coastal Maine woodlands. Journal of

Medical Entomology, 43, 1142-1152.

Fiedler AK, Landis DA, Wratten SD. 2008. Maximizing ecosystem services from conservation

biological control: the role of habitat management. Biological Control, 45, 254-271.

Fish D, Carpenter SR. 1982. Leaf litter and larval mosquito dynamics in tree-hole ecosystems.

Ecology, 283-288.

Foster WA. 1995. Mosquito sugar feeding and reproductive energetics. Annual Review of

Entomology, 40, 443-474.

Gardner AM, Allan BF, Frisbie LA, Muturi EJ. 2015. Asymmetric effects of native and exotic

invasive shrubs on ecology of the West Nile Virus vector Culex pipiens (Diptera:

Culicidae). Parasites and Vectors, 16, 329.

149

Guerrier S, Gardner AM, Balamuta J, Molinari R. The generalized method of wavelet moments

for multivariate latent composite stochastic processes. In preparation.

Hamer GL, Kitron UD, Goldberg TL, Brawn JD, Loss SR, Ruiz MO, Hayes DB, Walker ED.

2009. Host selection by Culex pipiens mosquitoes and West Nile virus amplification.

American Journal of Tropical Medicine and Hygiene, 80, 268-278.

Hartman KM, McCarthy BC. 2004. Restoration of a forest understory after the removal of an

invasive shrub, Amur honeysuckle (Lonicera maackii). Restoration Ecology, 12, 154-

165.

Hellmann JJ, Byers JE, Bierwagen BG, Dukes JS. 2008. Five potential consequences of climate

change for invasive species. Conservation Biology, 22, 534-543.

Hutchinson TF, Vankat JL. 1997. Invasibility and effects of Amur honeysuckle in southwestern

Ohio forests. Conservation Biology, 11, 1117-1124.

Ingold JL, Craycraft MJ. 1983. Avian frugivory on honeysuckle (Lonicera) in soutwestern Ohio

in fall. Ohio Journal of Science, 83, 256-258.

Jones KE, Patel NG, Levy MA, Storeygard A, Balk D, Gittleman JL, Daszak P. 2008. Global

trends in emerging infectious diseases. Nature, 451, 990-993.

Kitching RL. 2001. Food webs in phytotelmata: “bottom-up” and “top-down” explanations for

community structure. Annual Review of Entomology, 46, 729-760.

Lee CE. 2002. Evolutionary genetics of invasive species. Trends in Ecology and Evolution, 17,

386-391.

Levine JM, Vila M, Antonio CM, Dukes JS, Grigulis K, Lavorel S. 2003. Mechanisms

underlying the impacts of exotic plant invasions. Proceedings of the Royal Society of

London B: Biological Sciences, 270, 775-781.

150

Loomis J, Kent P, Strange L, Fausch K, Covich A. 2000. Measuring the total economic value of

restoring ecosystem services in an impaired river basin: results from a contingent

valuation survey. Ecological Economics, 33, 103-117.

Loomis JD, Cameron GN. 2014. Impact of the invasive shrub Amur honeysuckle (Lonicera

maackii) on shrub-layer insects in a deciduous forest in the eastern United

States. Biological Invasions, 16, 89-100.

Luken JO, Thieret JW. 1996. Amur honeysuckle, its fall from grace. BioScience, 18-24.

Mack R, Smith M. 2011. Invasive plants as catalysts for the spread of human parasites.

NeoBiota, 9, 13-29.

Mackay AJ, Muturi EJ, Ward MP, Allan BF. 2016. Cascade of ecological consequences for West

Nile virus transmission when aquatic macrophytes invade storm water habitats.

Ecological Applications, 26, 219-232.

Mattos KJ, Orrock JL. 2010. Behavioral consequences of plant invasion: an invasive plant alters

rodent antipredator behavior. Behavioral Ecology, 21, 556-561.

McCardle PW, Webb RE, Norden BB, Aldrich JR. 2004. Evaluation of five trapping systems for

the surveillance of gravid mosquitoes in Prince Georges County, Maryland. Journal of

the American Mosquito Control Association, 20, 254-260.

McCusker CE, Ward MP, Brawn JD. 2010. Seasonal responses of avian communities to invasive

bush honeysuckles (Lonicera spp.). Biological Invasions, 12, 2459-2470.

McEwan RW, Birchfield MK, Schoergendorfer A, Arthur MA. 2009. Leaf phenology and freeze

tolerance of the invasive shrub Amur honeysuckle and potential native competitors. The

Journal of the Torrey Botanical Society, 136, 212-220.

151

Miller KE, Gorchov DL. 2004. The invasive shrub, Lonicera maackii, reduces growth and

fecundity of perennial forest herbs. Oecologia, 139, 359-375.

Molaei G, Andreadis TG, Armstrong PM, Anderson JF, Vossbrinck CR. 2006. Host feeding

patterns of Culex mosquitoes and West Nile virus transmission, northeastern United

States. Emerging Infectious Diseases, 12, 468.

Mooney HA, Cleland EE. 2001. The evolutionary impact of invasive species. Proceedings of the

National Academy of Sciences, 98, 5446-5451.

Muturi EJ, Gardner AM, Bara JJ. 2015. Impact of an alien invasive shrub on ecology of native

and alien invasive mosquito species (Diptera: Culicidae). Environmental Entomology, 44,

1308-1315.

Pejchar L, Mooney HA. 2009. Invasive species, ecosystem services and human well-being.

Trends in Ecology and Evolution, 24, 497-504.

Pimentel D, Zuniga R, Morrison D. 2005. Update on the environmental and economic costs

associated with alien-invasive species in the United States. Ecological Economics, 52,

273-288.

Poulette MM, Arthur MA. 2012. The impact of the invasive shrub Lonicera maackii on the

decomposition dynamics of a native plant community. Ecological Applications, 22, 412-

424.

Reichard SH, White P. 2001. Horticulture as a pathway of invasive plant introductions in the

United States. BioScience, 51, 103-113.

Ruiz MO, Tedesco C, McTighe TJ, Austin C, Kitron U. 2004. Environmental and social

determinants of human risk during a West Nile virus outbreak in the greater Chicago

area, 2002. International Journal of Health Geographics, 3, 8.

152

Schmidt KA, Whelan CJ. 1999. Effects of exotic Lonicera and Rhamnus on songbird nest

predation. Conservation Biology, 13, 1502-1506.

Scott TW, Morrison AC. 2003. Aedes aegypti density and the risk of dengue virus. Ecological

Aspects for Application of Genetically Modified Mosquitoes, 2, 187.

Shewhart L, McEwan RW, Benbow ME. 2014. Evidence for facilitation of Culex pipiens

(Diptera: Culicidae) life history traits by the nonnative invasive shrub Amur honeysuckle

(Lonicera maackii). Environmental Entomology, 43, 1584-1593.

Smith EP. 2002. BACI design. Encyclopedia of Environmetrics.

Tsuda Y, Komagata O, Kasai S, Hayashi T, Nihei N, Saito K, Mizutani M, Kunida M, Yoshida

M, Kobayashi M. 2008. A mark-release-recapture study on dispersal and flight distance

of Culex pipiens pallens in an urban area of Japan. Journal of the American Mosquito

Control Association, 24, 339-343.

Turell MJ, O’Guinn ML, Dohm DJ, Jones JW. 2001. Vector competence of North American

mosquitoes (Diptera: Culicidae) for West Nile virus. Journal of Medical Entomology, 38,

130-134.

Turell MJ, Dohm DJ, Sardelis MR, O’Guinn ML, Andreadis TG, Blow JA. 2005. An update on

the potential of North American mosquitoes (Diptera: Culicidae) to transmit West Nile

virus. Journal of Medical Entomology, 42, 57-62.

Walker ED, Lawson DL, Merritt RW, Morgan WT, Klug MJ. 1991. Nutrient dynamics, bacterial

populations, and mosquito productivity in tree hole ecosystems and microcosms. Ecology

1529-1546.

153

White MA, Beurs D, Kirsten M, Didan K, Inouye DW, Richardson AD, Jensen OP, O’Keefe J,

Zhang G, Nemani RR, Leeuwen V. 2009. Intercomparison, interpretation, and assessment

of spring phenology in North America estimated from remote sensing for 1982-

2006. Global Change Biology, 15, 2335-2359.

Wilcove DS, Rothstein D, Dubow J, Phillips A, Losos E. 1998. Quantifying threats to imperiled

species in the United States. BioScience, 607-615.

Williams SC, Ward JS, Worthley TE, Stafford KC. 2009. Managing Japanese barberry

(Ranunculales: Berberidaceae) infestations reduces blacklegged tick (Acari: Ixodidae)

abundance and infection prevalence with Borrelia burgdorferi (Spirochaetales:

Spirochaetaceae). Environmental Entomology, 38, 977-984.

Williams SC, Ward JS. 2010. Effects of Japanese barberry (Ranunculales: Berberidaceae)

removal and resulting microclimatic changes on Ixodes scapularis (Acari: Ixodidae)

abundances in Connecticut, USA. Environmental Entomology, 39, 1911-1921.

154

Figures and Tables

Figure 4.1. Maps of spatial configuration of study locations in Mahomet IL (left) and CDC light traps (yellow) and infusion-baited gravid traps (red) at the Control (center) and Treatment (right) sites in 2014 and 2015.

155

Figure 4.2. Treatment site pre- (left) and post- (right) Amur honeysuckle eradication.

156

Figure 4.3. Mean number of Culex spp. mosquitoes (± SE) collected per trap night from the week of 29-Jun 2014 to 6-Sep 2015 at Control and Treatment sites using CDC light traps (top) and infusion-baited gravid traps (bottom). The vertical black line indicates the time point at which honeysuckle was removed at the Treatment site (winter of 2015).

157

Figure 4.4. Mean number of non-Culex spp. mosquitoes (± SE) collected per trap night from the week of 29-Jun 2014 to 6-Sep 2015 at Control and Treatment sites using CDC light traps (top) and infusion-baited gravid traps (bottom). The vertical black line indicates the time point at which honeysuckle was removed at the Treatment site (winter of 2015).

158

Figure 4.5. Vapor pressure deficit (Pa) measured at the Treatment site (orange lines) and at high

(dark blue lines) and low (light blue lines) densities of honeysuckle within the Control site from

29-Aug to 8-Sep 2015 (following honeysuckle eradication from the Treatment site).

159

Figure 4.6. Counts of birds observed in survey point counts at the Control and Treatment sites in

Mahomet, IL, at time points in July, August, and September 2014 and 2015.

160

Table 4.1. Composition and total abundance of mosquitoes collected in CDC light traps and infusion-baited gravid traps in Mahomet IL, 29-Jun to 14-Sep 2014 and 28-Jun to 13-Sep 2015.

2014 (pre-eradication) 2015 (post-eradication) Species Control Treatment Control Treatment CDC Gravid CDC Gravid CDC Gravid CDC Gravid Culex spp. 1117 1097 1758 1493 1236 1590 214 627 Aedes triseriatus 165 58 1028 58 203 56 58 26 Ae. trivittatus 443 3 544 0 486 0 45 1 Ae. vexans 4363 3 7306 3 4708 2 670 4 Ochlerotatus japonicus 97 43 281 63 85 54 13 29 Anopheles punctipennis 203 16 554 11 186 11 38 10 An. quadrimaculatus 72 5 419 2 88 6 25 0 An. walkeri 10 2 6 0 4 0 1 0 Psorophora ferox 156 2 267 0 130 0 6 0 Ps. ciliata 4 0 15 0 3 0 0 0 Culiseta inornata 5 2 0 0 2 0 0 0 Cs. morsitans 2 0 0 0 4 0 0 0 Uranotaenia sapphirina 0 0 8 0 0 0 0 0 Orthopodomyia signifera 3 0 0 0 4 0 0 0 Coquillettidia perturbans 1 0 0 0 2 0 0 0 TOTAL 6641 1231 12186 1630 5905 1719 990 500

161

Table 4.2. Type 3 tests of fixed effects of site (푆푖), year (푌푗), week within year (푊(푗)푙), and their interactions on abundance of Culex spp. and non-Culex spp. mosquitoes collected in CDC light traps and infusion-baited gravid traps in Mahomet, IL.

Culex spp., CDC light trap Culex spp., gravid trap Effect DF F P Effect DF F P 푆푖 1, 16 7.54 0.02 푆푖 1, 16 25.31 < 0.01 푌푗 1, 16 128.79 < 0.01 푌푗 1, 16 16.51 < 0.01 푆푌푖푗 1, 16 147.76 < 0.01 푆푌푖푗 1, 16 111.11 < 0.01 푊(푗)푙 20, 150 14.02 < 0.01 푊(푗)푙 20, 154 13.15 < 0.01 푆푊(푗)푖푙 20, 150 4.12 < 0.01 푆푊(푗)푖푙 20, 154 2.77 < 0.01 Non-Culex spp., CDC light trap Non-Culex spp., gravid trap Effect DF F P Effect DF F P 푆푖 1, 16 0.31 0.59 푆푖 1, 16 2.60 0.13 푌푗 1, 16 88.57 < 0.01 푌푗 1, 16 4.85 0.04 푆푌푖푗 1, 16 123.35 < 0.01 푆푌푖푗 1, 16 5.17 0.04 푊(푗)푙 20, 150 28.85 < 0.01 푊(푗)푙 20, 154 3.66 < 0.01 푆푊(푗)푖푙 20, 150 8.05 < 0.01 푆푊(푗)푖푙 20, 154 0.94 0.53

162

Table 4.3. Contrasts/SLICEs of the site by year interaction effect for Culex spp. and non-Culex spp. mosquitoes collected in Mahomet, IL, in CDC light traps and infusion-baited gravid traps.

Model SLICE DF F P Control 1, 16 0.32 0.58 Culex spp., Treatment 1, 16 279.18 < 0.01 CDC light trap 2014 (pre-eradication) 1, 16 45.12 < 0.01 2015 (post-eradication) 1, 16 109.00 < 0.01 Control 1, 16 20.83 < 0.01 Culex spp., Treatment 1, 16 107.42 < 0.01 Gravid trap 2014 (pre-eradication) 1, 16 15.12 < 0.01 2015 (post-eradication) 1, 16 121.69 < 0.01 Control 1, 16 1.43 0.25 Non-Culex spp., Treatment 1, 16 211.49 < 0.01 CDC light trap 2014 (pre-eradication) 1, 16 56.15 < 0.01 2015 (post-eradication) 1, 16 67.42 < 0.01 Control 1, 16 0.01 0.96 Non-Culex spp., Treatment 1, 16 10.09 < 0.01 Gravid trap 2014 (pre-eradication) 1, 16 0.22 0.65 2015 (post-eradication) 1, 16 7.58 0.02

163

Table 4.4. Numbers and species of birds observed in survey point counts in Mahomet, IL, at time points in July, August, and September 2014 and 2015 (presented using standardized bird species “alpha” codes).

Year 1, pre-eradication (2014) Year 2, post-eradication (2015) July August September July August September AMCR (3) AMCR (3) AMCR (2) AMCR (3) AMCR (2) AMCR (3) BLJA (2) BCCH (1) AMGO (1) AMRO (1) BLJA (2) AMRO (3) BCCH (2) BLJA (1) AMRO (1) BCCH (4) COGR (2) BCCH (1) DOWO (5) COGR (2) BLJA (2) BLJA (1) DOWO (2) BLJA (1) Control HOSP (2) DOWO (3) COGR (2) DOWO (4) HOSP (5) COGR (2) WBNU (2) HOSP (3) DOWO (5) MODO (1) INBU (1) DOWO (6) NOCA (1) HOSP (2) NOCA (2) TUTI (4) HOSP (2) TUTI (1) NOCA (2) WBNU (5) NOCA (1) WBNU (1) WBNU (1) July August September July August September AMCR (2) AMCR (3) AMCR (4) AMCR (3) AMCR (1) AMCR (3) AMRO (1) AMGO (1) AMGO (1) BCCH (1) COGR (2) AMRO (3) BLJA (3) AMRO (2) AMRO (2) DOWO (3) DOWO (2) DOWO (1) Treatment DOWO (5) BLJA (1) BLJA (1) HOSP (2) WBNU (2) HOSP (3) COGR (1) COGR (3) DOWO (4) DOWO (6) HOSP (3) HOSP (3) WBNU (2)

Code Species AMCR American crow AMGO American goldfinch AMRO American robin BCCH Black-capped chickadee BLJA Blue jay COGR Common grackle DOWO Downy woodpecker HOSP House sparrow INBU Indigo bunting MODO Mourning dove NOCA Northern cardinal TUTI Tufted titmouse WBNU White-breasted nuthatch

164