The Ecological Impacts of Non-Native Annual and Native Perennial Floral Insectaries on Beneficial Insect Activity Density and Arthropod-Mediated Ecosystem Services Within Ohio Pumpkin (Cucurbita pepo) Agroecosystems

Thesis

Presented in Partial Fulfillment of the Requirements for the Degree of Master of Science in the Graduate School of The Ohio State University

By

Benjamin W. Phillips, B.S.

Graduate Program in Department of Entomology

The Ohio State University

2013

Thesis Committee:

Mary Gardiner, Advisor – The Ohio State University, Wooster, OH

Karen Goodell – The Ohio State University, Newark, OH

Robin Taylor –Texas A&M University, Temple, TX

Celeste Welty – The Ohio State University, Columbus, OH

Copyright by

Benjamin Walter Phillips

2013

Abstract

Pumpkins (Cucurbita pepo) rely on insect-mediated pollination, and host a distinct community of pests, natural enemies and pollinators. My goal was to determine if biocontrol and pollination services in pumpkins were affected by local habitat management and landscape composition in Ohio. I measured biocontrol through predation and parasitism rates of sentinel egg cards of squash bug (Anasa tristis) and the spotted cucumber beetle (Diabrotica undecimpunctata howardii), and collected adults of striped cucumber beetle (Acalymma vittatum) to determine parasitism activity in 2011-

2012. I used pitfall traps to determine the activity density of ground-dwelling predators per field per sample period, and video cameras to determine the taxa responsible for egg mortality. I measured visitation frequency and duration of Apis mellifera, Bombus spp., and Peponapis pruinosa in male and female flowers of pumpkins in 2011-2012, and pollen deposition across the pollination window (0600-1200 hr) in 2012. I tested the

Intermediate Landscape-Complexity Hypothesis in one year by determining the combined effects of surrounding landscape composition and local habitat management on the relative visit frequency of pollinators, activity density of predators, and rates of predation and parasitism services by ranking general linear mixed models. I found that only D. undecimpunctata experienced a significant amount of egg predation, which was positively correlated to the percentage of field crops within a 1500 m radius of pumpkin fields. The parasitism of A. vittatum and the visitation frequency of A. mellifera was

ii diluted in the presence of fruit and vegetable habitats within a 1500 m radius, and P. pruinosa visit frequency was diluted within a 500 m radius. Parasitism of A. vittatum was positively associated with urban habitats within a 500 m radius, and the visit frequency of

P. pruinosa was positively associated with urban habitats within a 1500 m radii.

Predation of A. tristis and D. undecimpunctata eggs and parasitism of A. vittatum adults were not significantly affected by the addition of annual non-native floral insectaries of sweet alyssum (Lobularia maritima) or a perennial native insectary planted adjacent to the crop. Formicidae were the largest contributor to egg predation, and also responded positively to urban habitats. Activity density of Carabidae and Orthoptera captured in pitfalls located in alyssum insectaries increased with higher percentages of mowed turfgrass habitats.

In 2011 A. mellifera was more abundant in flowers than other bees, and in 2012

Bombus spp. was the most abundant. A. mellifera spent more time in flowers, and had a higher visit frequency in female flowers. In both years, Bombus spp. had a significantly higher visit frequency after 0700 hr, and both Bombus spp. and P. pruinosa spent less time in flowers after 0800 hr. Pollen loads on female flowers indicated that the majority of pollen deposited across the 6 hr window was transferred between 0600-0800 hr, which is when all three bee species foraged with equal frequency and similar visit duration, though Bombus spp. was the largest contributor. Alyssum and perennial floral insectaries did not have an effect on the foraging activity of bees. However, visits to pumpkins by A. mellifera showed that pumpkin fields close to an increased percentage of forest habitats supported higher visit frequencies to pumpkins with alyssum floral insectaries.

iii

Dedication

To Albert Reading (1944 - 2013), who abstained advanced treatment of mesothelioma, and instead donated much of the compensation money he was receiving to his nieces and nephews so we could pay off our college debts.

iv

Acknowledgments

Alphabetical order by contribution, which unfortunately does not emphasize the substantial overlap between these categories…

Lab and Professional Supporters:

Caitlin Burkman, Brenda Franks, Mary Gardiner, Karen Goodell, Matt Grieshop, Jim Jasinski, Lori Jones, Andrea Kautz, Scott Prazjner, Ian McIlvaine, Chelsea Smith, Rebecca Smyth, Robin Taylor, and Celeste Welty

Farmers and Collaborators on the Project:

Jim Badger, Brad Bergefurd, Jon Branstrator, Steve Cory, Steve Finney, Lee French, Thom Harker, Mark Hoverstock, Lloyd King, Carmella Massaro, The Pollinator Partnership, Ben Richardson, Norm Staufer, Randy Tegtmeier, Cameron Way, Fred Weaver, Brandon Weber, and Chris Vodraska

Statistical Help:

Larry Phelan, Diego Rincon, and Alain Zuur

Family & Friends:

Teresa Campton, Sarah Esralew, Marla Greanya, Tony Gregorc, J. Randall Hicks, Jason Hudson, Kyle Hutson, Liz Kolbe, Mitch Lettow, Larry Long, Kayla Perry, Bob/Mary/Jackie Phillips, Rob Stuckert, Laura Willis

v

Vita

20 December 1985……………………………...Born a gentleman

2004…………………………………………… East Kentwood High School

2009…………………………………………… B.S. in Fisheries and Wildlife, Michigan State University

2010-Present……………………………………Graduate Research Assistant, Department of Entomology, The Ohio State University

Fields of Study

Major Field: Entomology

vi

Table of Contents ABSTRACT!...... !II! DEDICATION!...... !IV! ACKNOWLEDGMENTS!...... !V! LIST OF TABLES!...... !VIII! LIST OF FIGURES!...... !IX! CHAPTER 1. INTRODUCTION!...... !1! BACKGROUND ON PUMPKIN (CUCURBITA PEPO) AGROECOSYSTEMS!...... !1! HABITAT MANAGEMENT TO CONSERVE WILDLIFE, INCLUDING BENEFICIAL INSECTS!...... !6! LANDSCAPE EFFECTS ON INSECTS IN AGROECOSYSTEMS!...... !14! THIS STUDY!...... !17! OBJECTIVES!...... !18! CHAPTER 2. PREDATION AND PARASITISM OF CUCURBITA PEPO PESTS IN OHIO AGROECOSYSTEMS IN THE PRESENCE OF AN NON-NATIVE ANNUAL FLORAL INSECTARY (LOBULARIA MARITIMA) AND A NATIVE PERENNIAL FLORAL INSECTARY MIX!...... !22! ABSTRACT!...... !22! INTRODUCTION!...... !23! METHODS!...... !31! RESULTS!...... !41! DISCUSSION!...... !49! TABLES!...... !59! FIGURES!...... !65! CHAPTER 3. POLLINATION OF CUCURBITA PEPO IN OHIO AGROECOSYSTEMS BY NATIVE AND MANAGED BEES IN THE PRESENCE OF A NON-NATIVE ANNUAL FLORAL INSECTARY (LOBULARIA MARITIMA)!...... !72! ABSTRACT!...... !72! INTRODUCTION!...... !73! METHODS!...... !79! RESULTS!...... !86! DISCUSSION!...... !93! TABLES!...... !104! FIGURES!...... !107! REFERENCES!...... !112!

vii

List of Tables

Table 2.1. The location and 2012 treatment assignments for pumpkin farms sampled in 2011 and 2012 for the predation and parasitism studies...... 59 Table 2.2. Seed mix of native perennial insectaries consisting of 23 forbs and 2 grasses...... 60 Table 2.3. A list of 26 models that were used on the predation and parasitism data...... 61 Table 2.4. AICc table showing the top candidate predation models for each year...... 62 Table 2.5. AICc table showing the top candidate activity density models for each year...... 63 Table 2.6. AICc table showing the top candidate parasitism models for each year...... 64

Table 3.1. The location, 2012 treatment assignments and video sample dates for pumpkin farms sampled in 2011 and 2012...... 104 Table 3.2. A list of 26 models that were used on the pollinator visit frequency data...... 105 Table 3.3. AICc table showing the top candidate visit frequency models for each year...... 106

viii

List of Figures

Figure 2.1. 2011 and 2012 site maps...... 65 Figure 2.2. Percentage of eggs missing after 48 hrs from caged and open egg cards from experiments in 2011 and 2012 for A. tristis, and D. undecimpunctata across experiment and site treatment...... 66 Figure 2.3. Proportions of predators observed interacting with sentinel eggs of A. tristis and D. undecimpunctata eggs in 2011 and 2012...... 67 Figure 2.4. Clock plot histograms illustrating activity patterns of predators on A. tristis and D. undecimpunctata eggs across a 24 hr foraging period...... 68 Figure 2.6. Proportions of predators caught in all pitfalls deployed in 2011, and 2012...... 69 Figure 2.7. Percent parasitism of A. vittatum adults collected from pumpkin fields adjacent to site treatments in 2012...... 70 Figure 2.8. Interaction plots showing the activity densities of predators caught in pitfalls placed in floral insectaries in the presence of different landscape-scale habitat types within 500 m...... 71

Figure 3.1. Map of sites used in 2011 and 2012 for the pollination study...... 107 Figure 3.2. The average number of visits, and average visit duration of bees in male and female flowers in 2011, and 2012...... 108 Figure 3.3. The average number of visits, and average visit duration of bees across the pollination window between 0600 hr and 1200 hr in 2011, and 2012...... 109 Figure 3.4. The average number of visits, and average visit duration of bees between grassy control and alyssum insectaries in 2012...... 110 Figure 3.5. The average number of pollen grains deposited on pumpkin stigmas between the hours of 0600- 0800, 0800-1000, 1000-1200, and 0600-1200, and between site treatments...... 110 Figure 3.6. Interaction plots showing the visit frequency of A. mellifera in pumpkin flowers planted adjacent to grassy control insectaries and alyssum insectaries with an increasing percentage of forested areas within 500 m, 1000 m, and 1500 m from my sites...... 111

ix

Chapter 1. Introduction

BACKGROUND ON PUMPKIN (CUCURBITA PEPO) AGROECOSYSTEMS

In the north central United States, pumpkin (Cucurbita pepo) is damaged primarily by the spotted cucumber beetle (Diabrotica undecimpunctata howardi), striped cucumber beetle (Acalymma vittatum), squash bug (Anasa tristis), and horned squash bug

(A. armigera). Both A. vittatum and A. tristis/armigera overwinter near crop habitat, and in addition to damage caused by herbivory, these pests also vector several damaging pathogens (Ellers-Kirk and Fleischer, 2006). Through defecating, cucumber beetles vector the pathogen, Erwinia tracheiphila, which causes bacterial wilt, and the squash bugs vector the bacterium, Serratia marcescens, which causes cucurbit yellow vine disease. In order to manage these pests, growers may use insecticide-treated seeds and a spray schedule that consists of a rotation of pesticides to prevent pest populations from becoming resistant, totaling approximately $10.63 billion annually in The United States

(NASS Quick Stats 2.0).

Spray regimes are one factor that can interfere with pollination of pumpkin.

Monoecious cucurbits are highly dependent on pollinators, like the honey bee (Apis mellifera), that transfer pollen grains from the stamens of male flowers to the stigmas of female flowers. In The United States, pollinators account for $40 billion per year in fruit, fiber, vegetable and legume crops (Pimentel et al., 1997), with an estimated $1.6-$14.8

1 billion of that attributed to A. mellifera alone (Southwick and Southwick, 1992; Morse and Calderone, 2000; Losey and Vaughan, 2006). In 2006, A. mellifera suffered major population crashes, from Colony Collapse Disorder, that caused the prices of hive rentals to increase more than two-fold from $54 to $136 per colony in 2004 and 2006, respectively (Sumner and Boriss, 2006; Stokstad, 2007; vanEngelsdorp et al., 2008). To help reduce the economic costs for growers and the environmental costs of associated management practices, researchers have been exploring strategies to promote naturally- occurring biocontrol and pollination services (Landis et al., 2000; Kremen and Ostfeld,

2005; Klein et al., 2007; Chaplin-Kramer et al., 2011; Tscharntke et al., 2012; Garibaldi et al., 2013; Kennedy et al., 2013).

The three most damaging herbivores are vulnerable to mortality from natural enemies at all life stages. Cucumber beetle larvae are susceptible to infection by entomopathogenic nematodes (Ellers-Kirk et al., 2000). Eggs of cucumber beetle are eaten by generalist predators such as hunting and sedentary spiders (Snyder and Wise,

2000; Williams and Wise, 2003), carabid and soldier beetles (Platt et al., 1999), and bats

(Whitaker Jr, 1995), and adult cucumber beetles are attacked by parasitoids such as braconid Centistes wasps, and tachinid Celatoria flies (Smyth and Hoffman, 2010;

Toepfer et al., 2008). Squash bug eggs are susceptible to parasitism by encyrtid wasps

(Ooencyrtus anasae, and O. sp. (Nechols et al., 1989), and scelionid wasps (Gryon pennsylvanicum, Eumicrosoma benefica (Metcalf and Metcalf, 1993; Driesche and

Bellows, 1996), and Hadronotus ajax (Schell, 1943)). Late instar nymphs and adults of squash bugs are parasitized by the tachinid fly Trichopoda pennipes (Beard, 1940a,

1940b, 1942). In laboratory studies, egg, nymph, and adult squash bugs were consumed

2 by carabids (Harpalus pennsylvanicus, Evarthrus sodalist, and Scarites spp. (Snyder and

Wise, 1999)); by geocorids (Geocoris punctipes), nabids (Pagasa fusca (Rondon et al.,

2003; Decker and Yeargan, 2008)), and lycosids (Hogna spp., and Pardosa spp. (Snyder and Wise, 1999)). Under field conditions, only G. punctipes, P. fusca, and Coleomegilla maculata (Coccinellidae) were observed eating squash bug eggs and nymphs (Rondon et al., 2003; Decker and Yeargan, 2008).

Previous studies have found that the dipteran and hymenopteran parasitoids

(Beard, 1940a; Shahjahan, 1968; Platt et al., 1999), and generalist predators (Gardiner et al., 2009; Woltz et al., 2012) of herbivorous pests are influenced by local and landscape habitat management (Batáry et al., 2011; Concepción et al., 2012). Additionally, studies on high value vegetable crops have found that natural enemies are attracted to floral insectaries in or near the crops can move into the crop (Platt et al., 1999; Gillespie et al.,

2011), and can provide biocontrol services to the insect pests within (Baggen and Gurr,

1998; Lee and Heimpel, 2005; Pease and Zalom, 2010).

Cucurbits in Ohio are typically visited by three types of pollinators, spread across a gradient of domestication and commercial availability: fully domesticated and feral

Apis mellifera, wild and semi-domesticated Bombus spp. (Stanghellini et al., 1998;

Stubbs and Drummond, 2001; Thomson and Goodell, 2001; Fuchs and Müller, 2004;

Artz and Nault, 2011), and undomesticated cucurbit-specialists, Peponapis pruinosa

(Hurd, Jr et al., 1971; Hurd et al., 1974). High seed numbers, successful maturation, and fruit weight are highly correlated with the number of visits of these bees (Brewer, 1974;

Jaycox et al., 1975; Stanghellini et al., 1998; Garibaldi et al., 2013), and the subsequent amount of pollen transferred (Canto-Aguilar and Parra-Tabla, 2000; Winfree et al.,

3 2007a; Goodell, 2008; Graças Vidal et al., 2010; Artz and Nault, 2011) to female flowers.

Because of this tight relationship, research on pollinators of cucurbits has often focused the abundance of these pollinators found inside flowers, and the duration of their visitation (Tepedino, 1981; Cane et al., 2000; Kremen et al., 2002; Fuchs and Müller,

2004; Shuler et al., 2005; Winfree et al., 2008; Julier and Roulston, 2009; Nicodemo et al., 2009; Barber et al., 2011; Artz et al., 2011; Garibaldi et al., 2013).

Pollination contributions of A. mellifera in cucurbits, compared to other bee species, has been variable in past studies. Artz & Nault (2011) found that, in New York,

A. mellifera was more likely to visit female flowers of C. pepo and to spend more time in them, though supplemental Bombus impatients hives improved fruit set because they carried three times as many pollen grains per visit as A. mellifera. Artz et al. (2011) also found that supplemental A. mellifera hives increased their abundance in C. pepo flowers, but decreased the abundance of P. pruinosa in pumpkin flowers. Shuler et al. (2005) found P. pruinosa was the most abundant pollinator at sites in Virginia, West Virginia, and Maryland, and that the density of A. mellifera observed in flowers was actually unrelated to the presence of commercial hives, possibly due to competitive blooms.

Goodell (2008) in Ohio also found P. pruinosa to have the highest visit frequency, compared to other bees, between 0630 and 0900 hr, compared to other species. Cane et al. (2011) found that male P. pruinosa in Mississippi were more commonly observed in flowers than females, and that just 7 visits by males maximized Cucurbita pepo fruit set and growth. Canto-Aguilar and Parra-Tabla (2000) in Mexico found that Peponapis limitaris deposited four times as much pollen per visit to Cucurbita moschata plants, and visited flowers significantly more frequently than A. mellifera. Tepedino (1981) found

4 that in Utah, P. pruinosa began flying 15-30 minutes before sunrise, and before the first

A. mellifera was found in the field, and preferred male flowers over female flowers.

Walters and Taylor (2006) found that Illinois pumpkin fields lacking A. mellifera colonies produced similar numbers of C. pepo fruits, but that maximum fruit weights could not be reached with naturally occurring wild pollinators. Fuchs and Müller (2004) found that Bombus terrestris visited up to five times as many flowers in one minute than

A. mellifera, and also foraged in adverse and rainy weather when A. mellifera would not fly. However, they stated that there are no natural populations of B. terrestris in Austria, and maintaining 4-5 hive boxes per hectare would not be sustainable without conservation of natural habitat.

In other Cucurbitaceae vine crops, Stanghellini et al. (1998) found that cucumber

(Cucumis sativus), and watermelon (Citrullus lanatus) in North Carolina experienced lower abortion rates, and higher seed sets when pollinated by 1, 6, and 18 Bombus spp. visits compared to the same number of A. mellifera visits. Winfree et al. (2007a, 2008) found that in New Jersey and Pennsylvania, A. mellifera contributed significantly fewer visits to C. lanatus than all wild pollinators combined, that flowers received full pollination even with few A. mellifera visits, and that increased pollen deposition was significantly correlated with higher populations of wild native bee visitation rates.

Further, wild bees were the dominant population of pollinators that visited flowers at

75% of the study sites. Similarly, Kremen et al. (2002) found that in multiple crops, including C. lanatus, wild pollinators provided enough visitations to deposit the required amount of pollen for a marketable crop. However, their effect was greatly depressed in

5 intensified agricultural areas, which reduced alternative forage and nesting habitats that support a higher diversity and abundance of native bees.

HABITAT MANAGEMENT TO CONSERVE WILDLIFE, INCLUDING

BENEFICIAL INSECTS

Habitat management incorporates many ecological principles by purposefully designing agricultural landscapes to promote arthropod biodiversity and to maximize the services provided by a stable ecosystem for the production of food (Dufour 2000).

Arthropods provide important pollination services as well as pest management services in the forms of parasitism and predation. Intensive agricultural production can have multiple negative impacts on these beneficial arthropod communities and the biocontrol and pollination ecosystem services they support through varying degrees of habitat disturbance, such as insecticides and soil cultivation (Isaacs et al. 2009). Many conventional techniques have been developed to treat the most common problems of soil conservation and pest management through chemical inputs, yet fail to retain soil fertility and productivity over time. Fertilizer inputs often come from off-farm sources, and leave the farm through leaching and non-point source runoff, sometimes resulting in cascading ecological effects in a watershed. To lessen these impacts, the incorporation of semi- natural habitat can enhance agricultural landscapes for beneficial arthropods.

The study of habitat management in agricultural entomology examines methods for altering agroecosystems to optimize the abundance of and services provided by a diversity of beneficial insects by providing needed resources including pollen and nectar as food, alternative prey, vegetative structure for shelter, mating sites and overwintering habitat (Landis et al. 2000). Depending on a grower’s goals, several different habitat 6 management strategies can be incorporated into agroecosystems. Herein, I discuss the diversity of conservation and non-crop habitat management methods currently in use and under development to improve agricultural sustainability.

Soil conservation through cover cropping, crop rotation, conservation tillage, and no-till production

One way to improve the habitat of ground-dwelling arthropods is through conservation tillage, and cover cropping techniques, which minimize soil disturbances and leave vegetative structure throughout the year. Tilling soil is a disturbance that occurs before, during and after a crop is planted. The goals of tilling are to remove unwanted vegetation, aerate the soil, and incorporate new organic material in the form of dead and decomposing organic matter. This frequently discourages ground-dwelling arthropods from dispersing, feeding, and proliferating (Shuler et al., 2005; Goodell, 2008; Julier and

Roulston, 2009), and leads to topsoil deficiencies and increased fertilizer inputs (Lal,

1993). Alternatively, no-tilling and strip-tilling aim to maintain organic material and nitrogen in the topmost soil horizon over time. This benefits below-ground and ground- level arthropod communities by increasing vegetative vertical structure at the soil surface for refugia, and by providing raw organic matter which becomes the basis of an arthropod-centric trophic cascade (Bugg and Waddington, 1994). However, herbicidal spray schedules tend to increase in no-till farms to compensate for the lack of mechanical weed control (Stougaard et al., 1984).

Crop rotation can also benefit soil nutrients and soil food webs. The best example of this in the United States is the corn-soybean rotation that utilizes the nitrogen fixing ability of the soybean planted in the stubble of nitrogen-depleting corn. Crop rotation and 7 staggered planting and harvest can also benefit arthropods by allowing them to migrate from one field to another without completely removing its habitat (Summers 1976). For example, many farmers harvest alfalfa fields in succession, to allow beneficial predators of pest Helicoverpa moths to migrate to unharvested fields on the farm (Hossain et al.

2002). This technique does not sacrifice the service provided by these natural enemies by depriving them of habitat.

Cover cropping refers to the planting of multiple crops in the same field across seasons, and can also promote a diverse and abundant assemblage of ground-dwelling arthropods, including herbivores, predators and detritivores. In vegetable crops, late fall cover cropping with rye holds nitrogen over the winter and slowly releases it after it is killed in the early summer when vegetable crops are seeded. This dead rye also serves as allelopathic mulch which blocks light and reduces weed competition (Barnes and

Putnam, 1983; Putnam et al., 1983; Sullivan, 2002). Cover crops also provide microhabitats (Bugg and Waddington, 1994) and floral resources (English-Loeb et al.,

2003) for beneficial arthropods.

Non-crop habitat introductions

Several other types of habitat management techniques introduce non-crop habitat.

These habitats may be established near agricultural sites, such as artificial wetlands or perennial buffer strips to filter leachates from the soil. These can positively impact insects, by providing a range of microhabitats, and possible food sources. Other types of habitat management introduce non-crop habitat directly within or adjacent to crop fields.

These include companion plantings, which are defined as within-crop plantings that provide services to the crop through nutrient supplementation, allelopathy, light blocking, 8 moisture retention and/or insect habitat. Floral strips are one form of companion planting that can be sowed in or near the crop and provide insect habitat and encourage pollination, predation, and parasitism services within nearby cropland.

Water retention and prevention of water contamination

Wetlands left to grow in field margins along waterways and collection ponds provide additional habitat that is not disturbed by typical farming practices like tilling or direct spraying. These provide habitat for both terrestrial and aquatic species, and could provide over-wintering sites for arthropods. However, the primary focus of this management technique is to hold water and sequester leachates from contaminating ground and stream water (Bennett et al. 2005; Moore et al. 2009)

Conservation lands

Conservation Reserve Programs (CRP) to increase biodiversity and conserve soil nutrients are often promoted by State governments and non-profits to favor game species.

Depending on the mix of plants, these leave undisturbed habitat and food resources for a variety of arthropods as well (Olson & Wackers 2007). Some of these beneficial arthropods, such as ground beetles, actually serve as a food source for avian game species

(Marshall et al. 2003). Used in a rotation, these fallow lands can improve the condition of carabid beetles, and harbor other beneficial ground predators for an extended period of time (Halaj et al. 2000; Oberg et al. 2008; Ostman et al. 2001). Conservation headlands, where tractor equipment makes turns, are often weedy and produce poor crops, but have provided pollen sources for adult syrphids whose larvae are predators of aphids (Cowgill et al., 1993; MacLeod, 1999).

Companion plantings

9 Different from the aforementioned artificial wetlands and CRP fields introduced for runoff mitigation and soil conservation, plants can be introduced along with a conventional crop during the growing season in strips adjacent or within the main crop.

Companion planting aims to benefit the crop by providing services such as nutrient fixing, allelopathy against weeds, shade, mulching, or as a trap crop for pests. Companion planting practices have been used in food production for hundreds of years. For example,

Native Americans and European colonists employed the “Three Sisters” companion plants of corn, beans and squash. The corn uses the nitrogen that the legumes fix in the soil, the legume uses the corn stalks as a lattice on which to climb, and the cucurbits use their big leaves to act as a shade mulch to prevent weed competition (Kuepper & Dodson

2001). More recently, companion planting methods have been refined to provide additional pest management benefits. For example, trap crops are planted in field margins to attract pests away from the main economic crop (Jasinski & Welty 2002). Also, artificial kairomone simulations of squash flowers have been used to attract and trap cucumber beetle away from cucurbit crops in a similar fashion (Welty 2008).

Alternatively, strips of purposefully-chosen plant species can be selected to attract and protect beneficial arthropods like pollinators (Holzschuh et al. 2009), predators

(MacLeod, 1999; Smith et al., 2008) and parasitoids (Platt et al., 1999; Wäckers, 2004;

Lavandero et al., 2005) to a focal crop field. For example, “beetle banks” made from tussock-forming grasses planted in conservation headlands in England have had positive effects on populations of generalist beetles, spiders and bird species of special concern

(Thomas et al., 1991; Marshall et al., 2003; MacLeod et al., 2004). Further, Cline et al.

(2008) found a significant reduction in trapped cucumber beetles when intercropped with

10 a combination of three plants (radish, nasturtium and tansy) thought to repel the beetle and three plants insects (buckwheat, cowpea and sweetclover) known for attracting beneficial.

Floral additions to agroecosystems

The use of floral resources to provide habitat for beneficial insects has been shown to increase the diversity of beneficial insects in agricultural systems including vegetable and field crops, tree fruit, and turfgrass (Braman et al. 2002; Bigger & Chaney

1998; Gurr et al. 2003; Lee & Heimpel 2005; Patt et al. 1997). Both the use of annuals and perennials have been illustrated to be a successful management strategy to recruit and support beneficial insects within agricultural systems (Fiedler et al. 2007; Tuell et al.

2008i; Pontin et al. 2005h). A review by Fiedler et al. (2008) found that the majority of studies focus on the use of exotic annual plants such as buckwheat (Fagopyrum esculentum), alyssum (Lobularia maratima), phacelia (Phacelia tanacetifolia), and coriander (Coriandrum sativa). The use of annual species has several benefits: the seed is readily available and cost effective, strips are relatively easy to plant and maintain as insectaries, plants provide prolific floral displays, and these insectaries can be moved to different locations on-farm each year. Monocropped insectaries of exotic annuals have been used to facilitate biocontrol (Thomas et al., 1991; White et al., 1995; Platt et al.,

1999; Ambrosino, 2006; Haenke et al., 2009; Pfiffner et al., 2009). However, annuals must be planted each year, and can become invasive.

To maximize the environmental benefits supplied by habitat management using floral insectaries, Fielder and Landis (2007b) investigated the use of native plants to attract beneficial insects. In a study comparing 43 native Michigan perennial plants with

11 the five most commonly recommended exotic annuals, multiple native species were found to be as or more attractive to pollinators and natural enemies than the typically recommended exotic annuals (Fiedler and Landis, 2007b; Tuell et al., 2008). They hypothesized that native wildflowers are species adapted to local climatic and soil conditions, and the local native pollinators are evolutionarily familiar with these plants

(Fiedler and Landis, 2007a; Isaacs et al., 2009). In addition to supporting natural enemy and pollinator populations during the growing season, native perennial plants supply additional environmental benefits by establishing overwintering habitat essential for beneficial insects, enhancing wildlife habitat throughout the year, stabilizing soils, reducing agricultural runoff, and enhancing native biodiversity through the re- incorporation of rare or formerly common native regional plants into agricultural landscapes.

Floral insectaries are a potential strategy to mitigate pollinator decline. For example, Kells et al. (2001) found that wildflower species within weedy field margins experienced significantly more visitation by Bombus and A. mellifera than plants within cropped field margins, and that different bees showed preferences for different flowers.

Pontin et al. (2005), Pywell et al. (2006), and Tuell et al. (2008) investigated the relative attractiveness of plant mixes to wild and managed bees, and confirmed that different wild bees prefer different flowers. These researchers concluded that highly-attractive seed mixes could be made to maintain diverse communities of wild native and managed foraging bees.

Annual and perennial plants have been evaluated for their potential to attract natural enemies in the field with brassica crops (Bigger and Chaney, 1998; Lee and

12 Heimpel, 2005; Pontin et al., 2005; Lee et al., 2006; Pascual-Villalobos et al., 2006;

Schellhorn et al., 2008; Gillespie et al., 2011), cereals (Cowgill et al., 1993; MacLeod,

1999; Sutherland et al., 2001; Collins et al., 2002, 2003; Haenke et al., 2009), field crops

(Nentwig et al., 1998; Cottrell and Yeargan, 1999; Koji et al., 2007; Atakan, 2010; Woltz et al., 2012), cotton (Olson and Wäckers, 2007), ornamentals (Rebek et al., 2005), fruit crops (Berndt et al., 2006; Burgio et al., 2006; Walton and Isaacs, 2011; Gontijo et al.,

2013), and non-brassica vegetables (Baggen and Gurr, 1998; Platt et al., 1999; Colley and

Luna, 2000). A common response variable in these studies is the relative attractiveness of the local habitat additions to predators, parasitoids and pests as measured by abundance, richness, or diversity (Cowgill et al., 1993; Bigger and Chaney, 1998; Platt et al., 1999;

Thomas and Marshall, 1999; Sutherland et al., 2001; Collins et al., 2003; MacLeod et al.,

2004; Wäckers, 2004; Pontin et al., 2005; Rebek et al., 2005; Burgio et al., 2006;

Pascual-Villalobos et al., 2006; Cole et al., 2007; Fiedler and Landis, 2007a; Koji et al.,

2007; Haenke et al., 2009; Atakan, 2010; Gillespie et al., 2011; Walton and Isaacs, 2011).

However, the function of these natural enemies is far less frequently measured (Kremen and Ostfeld, 2005; Chaplin-Kramer et al., 2011; Tscharntke et al., 2012). In the cases where biocontrol services have been directly examined, varied relationships were found, with some illustrating a positive impact (Hickman and Wratten, 1996; Baggen and Gurr,

1998; Collins et al., 2002; Lee and Heimpel, 2005; Pease and Zalom, 2010) and others a neutral impact (Berndt et al., 2006; Lee et al., 2006; Olson and Wäckers, 2007; Woltz et al., 2012) of floral insectary addition. However, less research has investigated whether floral insectaries established to improve predation and parasitism services in a crop can also attract bees and their pollination service to crops a well.

13 LANDSCAPE EFFECTS ON INSECTS IN AGROECOSYSTEMS

Farm habitats are nested within a larger landscape, consisting of a variety of habitats that can support an even larger variety of organisms. Some habitats are more favorable toward beneficial insects, and these could have implications for farmers anticipating their services. Klein et al. (2007) reviewed studies on 115 world crops and found that 30 of them are highly dependent on pollinators, and agricultural intensification reduces the composition of native bee communities and their pollination services.

Garibaldi et al. (2013) reviewed pollen deposition and fruit set data from studies of 41 world crops and found that fruit set increased two-fold with visitation from wild bees compared to that pollinated by A. mellifera alone, and concluded that A. mellifera provides only partial pollination in these crops. Kennedy et al. (2013) reviewed studies on local management, and landscape composition and configuration effects on 39 crop systems. They found that bee abundance and richness were higher if more natural habitats surrounded fields, and the beneficial effect of natural habitat types decreased when farm fields were more locally diversified. Winfree & Kremen (2009) researched potential mechanisms used by bee communities to regulate their populations amidst a changing landscape, and found that bee communities do not experience density compensation when populations of specific taxa fluctuate, which highlighted the importance of species richness.

Many solitary native bees provision brood nests in excavated tunnels in soil and wood, and also in hollow tubular grass stalks (Winfree et al. 2007) found in early successional grasslands and forests. Steffan-Dewenter et al. (2002) found that the species richness and abundance of wild solitary bees was significantly correlated with an increase

14 in the percentage of semi-natural habitat, such as grasslands and forests, within a 750 m radius of sites, but social Bombus and A. mellifera were unaffected by habitat changes at that scale. Additionally, A. mellifera abundance actually decreased in the presence of high percentages of semi-natural habitat within a 3000 m radius, and Bombus was only weakly correlated with landscape features within all radii measured. Westphal et al.

(2003) also found that Bombus did not respond to semi-natural habitat, and instead were positively related to the availability of mass flowering crops. They also found evidence that Bombus responded at larger spatial scales between 1750 m and 3000 m radii of sites.

Hines and Hendrix (2005) found Bombus abundance responded to the percentage of grasslands within 700 m of sites, and their diversity responded to the percentage of grasslands within 500 m. Pywell et al. (2006) and Öckinger and Smith (2007) also found that Bombus responded positively to the proportion of grasslands. Julier and Roulston

(2009) found that Bombus and A. mellifera had a nonsignificant positive effect in response to the percentage of forests, and Bombus had a nonsignificant negative effect in response to flowering crops (orchards) at 500 and 300 m. P. pruinosa did not respond to the landscape predictors, but responded to local habitat management of the pumpkin crops on which they specialize.

Proximity to resources also has implications for biocontrol services.

Concentration of floral resources has been found to affect the abundance of parasitic wasps. The nectar provides a food source to the wasps, and protein found in pollen can increase their longevity and fertility (Shahjahan, 1968). If their nutrition is provisioned for, then they spend more time reproducing through parasitism of herbivores (Baggen and

Gurr, 1998; Berndt et al., 2006; English-Loeb et al., 2003; Lavandero et al., 2005;

15 Wäckers, 2004). During times of asynchrony, where the main economic pest may not be present in the adjacent crop, alternative prey species in the floral insectaries could keep the wasps and syrphids active in the area around the crop (Fiedler & Landis 2007; Landis et al. 2000). This knowledge impacts growers deciding how to promote species abundance, diversity and their associated ecosystem services within a variety of landscapes.

A meta-analysis of 46 landscape-scale studies on natural enemies by Chaplin-

Kramer et al. (2011) found a more natural landscape, dominated by non-crop habitats was positively associated with abundance and diversity of natural enemies. Also, generalist natural enemies were more likely to be affected by landscape complexity at a larger scale

(1500-2000 m) than specialists (500-1000 m). However, fewer studies examining the influence of landscape on natural enemies have focused on biocontrol services, than on diversity or abundance. Gardiner et al. (2009), Roschewitz et al. (2005), Thies and

Tscharntke (1999), and Thies et al. (2003, 2005, 2008) have led such studies. Gardiner et al. (2009) found that predation of soybean aphids was greater, and that the abundance of the coccinellid predators was positively related to the abundance of forest and grassland habitats surrounding soybean fields. Roschewitz et al. (2005) investigated the effects of the percent of arable land on parasitism of cereal aphids and found parasitism and aphid populations increased with a reduction in arable land, resulting in no net benefit. Thies and Tscharntke (1999), and Thies et al. (2003, 2005, 2008) found that parasitism of cereal aphid in winter wheat fields, and rape pollen beetle in rape crops, were associated with the amount of crop habitat within a 2 km radius of sites.

16 THIS STUDY

Knowing how habitat additions affect crops in different landscapes can help farmers know what cost-effective and environmentally beneficial practices they can employ on their farm while maintaining crop productivity. However, the use floral insectaries have not been fully incorporated into commercial vegetable production systems in The United States. To be adopted by large and small scale growers, the costs and benefits of incorporating habitat management must be fully evaluated and information on the practical aspects of implementing this type of management must be readily available. My research examines the effects of landscape composition, and incorporation of floral insectaries in pumpkin agroecosystems on activity density of ground-dwelling predatory arthropods, predation of A. trisitis, D. undecimpunctata eggs, parasitism of A. vittatum adults, as well as the pollination from A. mellifera, Bombus spp., and P. pruinosa.

The hypotheses tested were three-fold: 1) when floral insectaries are planted adjacent to a crop, populations of beneficial insects enter the crop from the established plantings (Platt et al., 1999; Pontin et al., 2005), 2) natural landscapes act as a source for beneficial insects into the floral insectaries (Isaacs et al., 2009; Pywell et al., 2006;

Rodriguez-Saona et al., 2012; Woltz et al., 2012), and 3) an intermediate landscape- complexity maximizes the effects of a floral insectary addition (Isaacs et al., 2009; Batáry et al., 2011; Concepción et al., 2012; Tscharntke et al., 2012). The Intermediate

Landscape-Complexity Hypothesis (Tscharntke et al., 2012) states that the addition of a local habitat management scheme, such as floral insectaries, to a crop field within an already complex and supportive landscape (> 20% non-crop areas) would not make a

17 difference to populations of beneficial arthropods and their ecosystems services provided to the crop. Additionally, placing floral insectaries next to a crop field within an already simple and unsupportive landscape (< 1% non-crop areas) would not make a difference either, because the landscape as a whole would not support high enough populations to enter and utilize the floral insectary. In an intermediately-complex landscape is where local habitat management strategies would have the most observable effect.

OBJECTIVES

Given the importance of pollination, predation, and parasitism of pumpkins, and the hypotheses above, I examined the following objectives, organized by chapter:

Chapter 2 Objectives

2.1) Determine if local addition of floral insectaries in landscapes that vary in

composition influenced predation of cucumber beetle or squash bug eggs.

2.2) Measure activity density of potential generalist ground predators and assess the

guild of actual predators of A. tristis and D. undecimpunctata eggs with video

surveillance.

2.3) Examine whether localized habitat management and/or landscape composition

influenced the frequency of A. vittatum or A. tristis adult parasitism.

Chapter 2 Predictions

2.1) The addition of native perennial insectaries would support the highest level of

egg predation and adult parasitism of cucurbit pests in pumpkin by providing a

more diverse habitat structure, and a more continuous bloom period (Landis et

al., 2000; Fiedler et al., 2008; Isaacs et al., 2009).

18 2.2) The addition of annual alyssum insectaries would support a higher level of

predation and parasitism than pumpkins grown on farms with grassy control

insectaries by providing a rich, though more temporary, nectar source for

natural enemies than the perennials (Fiedler and Landis, 2007a).

2.3) Predator activity density would be higher in the floral insectary plots than in

the pumpkin plots (Bigger and Chaney, 1998; Platt et al., 1999; Olson and

Wäckers, 2007; Woltz et al., 2012).

2.4) Higher percentages of semi-natural, non-crop habitat would support higher

activity density of predators and a higher level of predation and parasitism

services (Thies et al., 2003; Gardiner et al., 2009; Woltz et al., 2012).

2.5) In concordance with the Intermediate Landscape-Complexity Hypothesis

(Tscharntke et al., 2012), I predicted that in the presence of intermediate

habitat-complexity, predator abundance, predation, and parasitism occurring at

sites with local floral insectary additions (in 2012) would be significantly

greater than at sites with a control insectary.

Chapter 3 Objectives

3.1) Measure the visitation frequency and duration of A. mellifera, Bombus spp.,

and P. pruinosa to male and female pumpkin flowers across the pollination

window (0600-1200 hr).

3.2) Measure pollen deposition across the pollination window (0600-1200 hr).

3.3) Determine whether the surrounding landscape composition and/or local habitat

management influences the relative visit frequency of pollinators to pumpkin

flowers. 19 Chapter 3 Predictions

3.1) A. mellifera visit frequency and duration would be higher in female flowers,

but that flower sex would not affect visit frequency and duration of other bee

species (Tepedino, 1981; Artz and Nault, 2011).

3.2) P. pruinosa and Bombus spp. would be exhibit greater visit frequency and have

longer visit durations earlier in the morning, resulting in more pollen grains

deposited earlier in the morning, and A. mellifera would exhibit greater visit

frequency and have longer visit durations later in the pollination window (Hurd

et al., 1974; Canto-Aguilar and Parra-Tabla, 2000; Goodell, 2008; Nicodemo et

al., 2009; Graças Vidal et al., 2010; Artz et al., 2011; Artz and Nault, 2011).

3.3) Local habitat management in the form of sweet alyssum floral insectaries

would increase the visit frequency of A. mellifera and Bombus spp., by acting

as alternative nectar sources before and after pumpkin bloom (Pontin et al.,

2005; Pywell et al., 2006; Tuell et al., 2008), but decrease visit frequency of P.

pruinosa by decreasing pumpkin habitat on which they specialize.

3.4) Higher percentages of semi-natural and less-disturbed habitat surrounding my

sites would result in higher visit frequency of A. mellifera and Bombus spp.

visiting pumpkin flowers (Kremen et al., 2002; Steffan-Dewenter et al., 2002;

Shuler et al., 2005; Julier and Roulston, 2009), but decrease visit frequency of

P. pruinosa by competing with the area pumpkin crop on which they

specialize.

3.5) Finally, in concordance with the Intermediate Landscape-Complexity

Hypothesis (Tscharntke et al., 2012), I predicted that in the presence of

20 intermediate habitat-complexity, visit frequencies of A. mellifera and Bombus spp. to pumpkin flowers at sites with local floral insectary additions (in 2012) would be significantly greater than at sites with a grassy control insectary. !

21

Chapter 2. Predation and parasitism of Cucurbita pepo pests in Ohio agroecosystems in the presence of an non-native annual floral insectary

(Lobularia maritima) and a native perennial floral insectary mix

ABSTRACT

Cucurbit crops, such as pumpkins (Cucurbita pepo) are attacked by a pest complex that threatens production via direct feeding and disease transmission. My goal was to determine if predation and parasitism services of these pests in pumpkins were affected by local habitat management, in the form of floral insectaries, and landscape composition. At each site I examined the impact of egg predation on squash bug (Anasa tristis) and spotted cucumber beetle (Diabrotica undecimpunctata howardii), and parasitism of the adult striped cucumber beetle (Acalymma vittatum) in 2011-2012. I used video surveillance and pitfall traps to determine the taxa responsible for egg predation and their activity densities. I also determined the effects of surrounding landscape composition and local habitat management on the relative activity density of predators and rates of predation and parasitism by ranking general linear mixed models. I found that only D. undecimpunctata experienced a significant amount of egg predation, which was positively correlated to the percentage of field crops within 1500 m of pumpkin fields (in one year). Parasitism of A. vittatum was diluted in the presence of fruit and vegetable habitats within 1500 m, and was positively associated with urban landscapes.

22 Predation of A. tristis and D. undecimpunctata eggs and parasitism of A. vittatum adults were not affected significantly by the addition of an annual non-native floral insectary or a native perennial insectary planted adjacent to the crop. Formicidae were the largest contributor to egg predation of both A. tristis and D. undecimpunctata, and responded to positively to urban landscapes. Activity density of Carabidae and Orthoptera captured in pitfalls located in alyssum insectaries increased in the presence of higher percentages of mowed turfgrass environments.

INTRODUCTION

Agricultural landscapes are subject to frequent disturbances including planting, harvesting, chemical and nutrient inputs, and tillage. Habitat management using floral insectaries seeks to mediate some of the negative impacts of these practices by providing alternative food and shelter resources for beneficial arthropods (Landis et al., 2000;

Zehnder et al., 2007). When these practices are incorporated into a farmscape, larger scale landscape composition and heterogeneity can influence the species pool of organisms supplied to the floral insectary, and the arthropod mediated ecosystem services they are able to support (Isaacs et al., 2009; Batáry et al., 2011; Concepción et al., 2012;

Rodriguez-Saona et al., 2012). Further, Tscharntke et al. (2012) introduced the

Intermediate Landscape Complexity Hypothesis, which states that in highly heterogeneous landscapes (> 20% non-crop habitats), stable populations of beneficial organisms already exist which limited the effect of local habitat management; and extremely simplified landscapes (< 1 % non-crop habitats) do not have enough supporting habitats for a substantial species pool to take advantage of local habitat amendments. As such, local habitat management is theoretically most successful with 23 intermediate landscape-complexity. To date, the majority of studies investigating how landscape complexity mediates the impact of local habitat management on biocontrol services, such as predation and parasitism, have focused on measuring natural enemy diversity and abundance, but few have approached the question by measuring biocontrol services directly (Kremen and Ostfeld, 2005; Chaplin-Kramer et al., 2011; Tscharntke et al., 2012). This study aimed to test the Intermediate Landscape-Complexity Hypothesis by measuring activity density of predators, and the rate of predation and parasitism supplied to pumpkin fields in the presence of floral insectaries nested within landscapes that ranged in composition from simple to complex.

Floral insectaries (Colley and Luna, 2000; Hogg et al., 2011) are commonly referred to as floral strips (Gillespie et al., 2011), flower strips (Bianchi and Wäckers,

2008; Haenke et al., 2009; Walton and Isaacs, 2011), beetle banks (MacLeod et al.,

2004), grass banks (Thomas and Marshall, 1999), conservation headlands (Cole et al.,

2007), refuges (Schellhorn et al., 2008), and floral resources (Lee et al., 2006). The flowering plants within the insectaries provide carbohydrate and protein sources in the form of pollen and nectar that have been shown to increase fecundity and longevity of natural enemies (Shahjahan, 1968; Baggen and Gurr, 1998; Johanowicz and Mitchell,

2000; Lee et al., 2006). Annual plants are commonly used as floral insectaries because they are inexpensive to establish, moveable from year-to-year, fast-growing, and some make effective cover crops or are economically important themselves (Bugg and

Waddington, 1994; Prasifka et al., 1999; Dufour, 2000; Landis et al., 2000). Some of the commonly studied annual plants include candytuft (Iberis umbellate), lacey phacelia

(Phacelia tanacetifolia), buckwheat (Fagopyrum esculentum), sweet alyssum (Lobularia

24 maritima), coriander (Coriandrum sativum), and chrysanthemum (Chrysanthemum coronarium) (MacLeod, 1999; Wäckers, 2004; Rebek et al., 2005; Lavandero et al.,

2006; Fiedler and Landis, 2007b). However, some of the desired annuals do not thrive in an agricultural setting without careful management, others can re-seed prolifically and become a nuisance, and many are exotic plants with a limited bloom period (Landis et al.,

2000; Fiedler and Landis, 2007a).

Native perennials have also been explored as potential floral insectaries. Fiedler and Landis (2007b) and Tuell et al. (2008) investigated and rated 43 native Eastern US perennial plants based on their bloom size, length of flowering period and attractiveness to natural enemies and pollinators. They found that mixes of native perennials could be designed to bloom continuously throughout the growing season, and increase the biodiversity of a landscape. However, investing in a proper planting takes a larger initial financial investment than typically expected for an annual floral insectary, and a two to three-year establishment period until all of the plants have accumulated a large enough root mass to flower. Investment in annual or perennial floral insectaries could be of particular interest to growers of specialty crops that are grown on smaller scales where their addition could support a community of beneficial insects to provide ecosystem services.

Many laboratory and field studies have measured the attractiveness of varied plant species and mixes (Wäckers, 2004; Lavandero et al., 2005, 2006; Fiedler and Landis,

2007a; Amaral et al., 2012). These plants have also been evaluated in the field with brassica crops (Bigger and Chaney, 1998; Lee and Heimpel, 2005; Pontin et al., 2005;

Lee et al., 2006; Pascual-Villalobos et al., 2006; Schellhorn et al., 2008; Gillespie et al.,

25 2011), cereals (Cowgill et al., 1993; MacLeod, 1999; Sutherland et al., 2001; Collins et al., 2002, 2003; Haenke et al., 2009), field crops (Nentwig et al., 1998; Cottrell and

Yeargan, 1999; Koji et al., 2007; Atakan, 2010; Woltz et al., 2012), cotton (Olson and

Wäckers, 2007), ornamentals (Rebek et al., 2005), fruit crops (Berndt et al., 2006; Burgio et al., 2006; Walton and Isaacs, 2011; Gontijo et al., 2013), and non-brassica vegetables

(Baggen and Gurr, 1998; Platt et al., 1999; Colley and Luna, 2000). A common response variable in these studies is the relative attractiveness of the local habitat additions to predators, parasitoids and pests as measured by abundance, richness, or diversity

(Cowgill et al., 1993; Bigger and Chaney, 1998; Platt et al., 1999; Thomas and Marshall,

1999; Sutherland et al., 2001; Collins et al., 2003; MacLeod et al., 2004; Wäckers, 2004;

Pontin et al., 2005; Rebek et al., 2005; Burgio et al., 2006; Pascual-Villalobos et al.,

2006; Cole et al., 2007; Fiedler and Landis, 2007a; Koji et al., 2007; Haenke et al., 2009;

Atakan, 2010; Gillespie et al., 2011; Walton and Isaacs, 2011). However, the function of these natural enemies is far less frequently measured (Kremen and Ostfeld, 2005;

Chaplin-Kramer et al., 2011; Tscharntke et al., 2012). In the cases where rates of biocontrol services have been examined, varied relationships were found, with some illustrating a positive impact (Hickman and Wratten, 1996; Baggen and Gurr, 1998;

Collins et al., 2002; Lee and Heimpel, 2005; Pease and Zalom, 2010) and others a neutral impact (Berndt et al., 2006; Lee et al., 2006; Olson and Wäckers, 2007; Woltz et al.,

2012) of floral insectary addition.

Natural enemies that respond to local habitat plantings can also respond to the landscape at large, and these landscape-scale differences could be a key player in how local habitat management affects insect dispersal and behavior in an agricultural setting.

26 A meta-analysis of 46 landscape-scale studies on natural enemies by Chaplin-Kramer et al. (2011) found a more natural landscape, dominated by non-crop habitats was positively associated with abundance and diversity of natural enemies. Also, generalist natural enemies were more likely to be affected by landscape complexity within larger radii of sites (1500-2000 m) than specialists (500-1000 m). However, presence of natural enemies may not necessarily predicate predation or parasitism service.

Similar to local habitat mangement, fewer studies examining the influence of landscape on natural enemies have focused on biocontrol services, than on diversity or abundance. Examples include Gardiner et al. (2009), Roschewitz et al. (2005), Thies and

Tscharntke (1999), and Thies et al. (2003, 2005, 2008). Gardiner et al. (2009) found that predation of soybean aphids was greater, and that the abundance of the coccinellid predators was positively related to the abundance of forest and grassland habitats surrounding soybean fields. Roschewitz et al. (2005) investigated the effects of the percent of arable land on parasitism of cereal aphids and found parasitism and aphid populations increased with a reduction in arable land, resulting in no net benefit. Thies and Tscharntke (1999), and Thies et al. (2003, 2005, 2008) tested the hypothesis that organisms of higher trophic levels respond to large-scale landscape changes more than lower trophic levels and found that parasitism of cereal aphid in winter wheat fields, and rape pollen beetle in rape crops, were associated with the amount of crop habitat at radii within 2 km.

Researchers are just beginning to focus on the important interaction between landscapes and local habitat management on biocontrol services. One example comes from Woltz et al. (2012), who examined the effects of adding annual buckwheat floral

27 insectaries adjacent to soybean on predation of soybean aphid. They found that the abundance of coccinellids in the crop was positively associated with semi-natural habitat, but found no evidence that the amount of semi-natural habitat affected abundance of coccinellids in the buckwheat insectaries. Additionally, predation of soybean aphids in the crop was not significantly higher in the presence of buckwheat insectaries. Studies of biocontrol services are beginning to elucidate how local habitat management and landscape complexity interact, but more evidence is needed to predict how these factors will influence the outcomes of local habitat management.

Vegetable cropping systems are ideal for habitat management due to smaller field size, high crop value, high pest pressure, and growing demand for organic produce.

Studies on high value vegetable crops have found that natural enemies attracted to floral insectaries in or near the crops can move into the crop (Platt et al., 1999; Gillespie et al.,

2011), and can provide biocontrol services to the insect pests within (Baggen and Gurr,

1998; Lee and Heimpel, 2005; Pease and Zalom, 2010).

In the north central United States, pumpkin (Cucurbita pepo) is damaged primarily by the spotted cucumber beetle (Diabrotica undecimpunctata howardi), striped cucumber beetle (Acalymma vittatum), squash bugs (Anasa tristis, A. armigera). Both A. vittatum and A. tristis/armigera overwinter near crop habitat, and in addition to damage caused by herbivory, these pests also vector several damaging pathogens (Ellers-Kirk and

Fleischer, 2006). Through defecating in wounded tissue, cucumber beetles vector the pathogen, Erwinia tracheiphila, which causes bacterial wilt, and the feeding squash bugs vector the bacterium, Serratia marcescens, which causes cucurbit yellow vine disease.

These herbivores are vulnerable to mortality at all life stages from natural enemies.

28 Cucumber beetle larvae are susceptible to infection by entomopathogenic nematodes

(Ellers-Kirk et al., 2000). Generalist predatory arthropods are known to attack eggs of A. vittatum and D. undecimpunctata (Whitaker Jr, 1995; Platt et al., 1999; Snyder and Wise,

2000; Williams and Wise, 2003), and parasitoid wasps and flies (Toepfer et al., 2008;

Smyth and Hoffman, 2010) are known to attack their adult life stage. Additionally, in laboratory studies eggs, nymphs, and adults of A. tristis were consumed by carabids

(Snyder and Wise, 1999), geocorids and nabids (Rondon et al., 2003; Decker and

Yeargan, 2008), and lycosids (Snyder and Wise, 1999). Under field conditions, geocorids, nabids and coccinellids have been observed eating squash bug eggs and nymphs (Rondon et al., 2003; Decker and Yeargan, 2008). A. tristis/armigera is also parasitized by Trichopoda pennipes as a late-instar nymph and adult (Beard, 1940a,

1940b, 1942) and as eggs (Schell, 1943; Nechols et al., 1989; Metcalf and Metcalf, 1993;

Driesche and Bellows, 1996).

This study tested three hypotheses in relation to biocontrol services to pumpkin pests in the presence of local habitat management and landscape complexity: 1) when floral insectaries are planted adjacent to a crop, populations of natural enemies enter the crop from the established plantings (Platt et al., 1999; Pontin et al., 2005), 2) natural landscapes act as a source for natural enemies to move into the floral insectaries (Isaacs et al., 2009; Pywell et al., 2006; Rodriguez-Saona et al., 2012; Woltz et al., 2012), and 3) an intermediate landscape-complexity maximizes the effects of a floral insectary addition on natural enemy communities, and predation and parasitism services (Isaacs et al., 2009;

Batáry et al., 2011; Concepción et al., 2012; Tscharntke et al., 2012). To test these hypotheses, my objectives were to: 1) determine if local addition of floral insectaries in

29 landscapes that vary in composition influenced cucumber beetle or squash bug egg predation services, 2) measure activity density of potential generalist ground predators and assess the guild of actual predators with video surveillance of A. tristis and D. undecimpunctata eggs, and 3) examine whether localized habitat management and/or landscape composition influenced the frequency of A. vittatum and A. tristis/armigera adult parasitism. In 2011, I measured how the landscape composition and diversity influenced predation and parasitism services in the pumpkins. In 2012, I evaluated how the addition of non-native sweet alyssum (L. maritima) and native perennial floral insectary management, as well as landscape composition influences the activity density and foraging activity of predators in pumpkin. I predicted that 1) the addition of native perennial floral insectaries would support the highest level of egg predation and adult parasitism of cucurbit pests in pumpkin by providing a more diverse habitat structure, and a more continuous bloom period (Landis et al., 2000; Fiedler et al., 2008; Isaacs et al.,

2009), 2) the addition of annual alyssum insectaries would support a higher level of predation and parasitism than pumpkins grown on farms with grassy control insectaries by providing a rich, though more temporary, nectar source for natural enemies than the perennials (Fiedler and Landis, 2007a), 3) predator activity density would be higher in the floral insectary plots than in the adjacent pumpkin plots (Bigger and Chaney, 1998;

Platt et al., 1999; Olson and Wäckers, 2007; Woltz et al., 2012), 4) higher percentages of semi-natural, non-crop habitat would support higher activity density of predators and a higher level of predation and parasitism services (Thies et al., 2003; Gardiner et al., 2009;

Woltz et al., 2012), and 5) in the presence of intermediate habitat-complexity (Tscharntke et al., 2012), predator activity density, predation, and parasitism occurring at sites with

30 local floral insectary additions (in 2012) would be significantly greater than at sites with a grassy control insectary.

METHODS Study sites

The relative activity density of ground-dwelling arthropod predators, and rates of predation and parasitism of pumpkin pests were measured on pumpkin farms throughout

Ohio in 2011-2012. Ohio is the 2nd largest pumpkin-producing state in The United States, and two regions within Ohio were selected that represent major production areas (USDA-

NASS, 2013). In 2011, 12 sites were included; six in Wayne, Stark, and Medina counties in northern Ohio, and six in Jackson, Pike, Highland, and Warren counties in southern

Ohio. In 2012, one site was dropped (site 3) and three sites were added (sites 7, 8, and 15) in Ashland county in northern Ohio, and Ross and Clinton county in southern Ohio

(Figure 2.1). The distance between the two closest sites was 4.25 km.

In 2011 and 2012, a plot consisting of four rows of jack-o-lantern pumpkins (var.

“Gladiator”) were established between 10 June and 8 July within each site. No insecticides were applied to pumpkin plots throughout the study. Each plot was divided into 4 equally-spaced sub-plots along the length of the study site where all data were collected. Sites were chosen based on grower interest in participating, and by visual- assessment of the complexity of habitats in their surrounding landscape.

Habitat management

In 2012, sites in the northern and southern regions were assigned to one of three randomized treatments: 1) control: pumpkin plots planted adjacent to a 6 x 60 m grass

31 area, mowed approximately once per month; 2) alyssum: pumpkin plots planted between two 60 m rows of the non-native annual, L. maritima; and 3) perennial: pumpkin plots planted adjacent to a 6 x 60 m swath of native perennials (Table 2.1). Each individual farm site had one pumpkin plot and one floral insectary treatment except for sites 2, 9, and 10. In these cases, both an annual plot and a perennial plot were established on two separated fields. The distances between paired plots ranged from 51 m at site 10, to 570 m at site 9.

Establishing native perennial floral insectaries

In October 2010, six sites were selected to establish a 6 x 60 m perennial floral insectary treatment of 23 native forbs and 2 grasses (Table 2.2). Each grower cleared the area with field cultivators and herbicide, and rolled the soil flat. I mixed the perennial seeds with sawdust at a ratio of 1:2 and spread 1.3 kgs of that mixture at each site to overwinter (Landis et al., 2000; Fiedler and Landis, 2007a; Fiedler et al., 2007). The perennial floral insectary plots were mowed by the growers once per month to enhance root mass growth during the 2011 growing season. Each perennial insectary plot was divided into 4 equally-spaced sub-plots along the length of the adjacent pumpkin plot, where all data were collected

Establishing non-native annual floral insectaries

In 2012, I planted two rows of L. maritima as the annual floral insectary treatment adjacent to pumpkins at six sites in northern and southern Ohio. For this treatment, one row of L. maritima was established on either side of the pumpkin plot. The sweet alyssum was started from seed in 72-cell plug trays in a greenhouse in early May and

32 fertilized twice per week for two weeks. The plants were hardened off outside for an additional two weeks before being transplanted with a pottiputki planter (Stand ‘n Plant,

Saltsburg, PA) into the alyssum insectary plots between 7-14 June 2012. Plants were watered and Preen Garden Weed Preventer (Lebanon Seaboard Corp., Lebanon, PA) was applied. The transplants were watered via drip irrigation and hand containers (~190 L) twice per week in the field through July. Each alyssum insectary was divided into 4 equally-spaced sub-plots along the length of the adjacent pumpkin plot, where all data were collected

Measuring predation of pumpkin pest eggs

In 2011-2012, I deployed sentinel eggs of two pumpkin pests to assess predation services. Eggs of D. undecimpunctata were purchased from French Agricultural

Research, Inc. (Lamberton, MN). Eggs were maintained in a refrigerator at 4 °C for no longer than 8 days prior to use in the field. A. tristis eggs were reared in a greenhouse from a colony of field-captured adults. They were fed on young pumpkin plants, and fresh 1-2 day old eggs were collected directly from leaves one day prior to the experiment.

The eggs were attached to 2 x 2 cm construction paper cards by being set in a thin layer of non-toxic silicon fish tank sealant (Hoffmann et al., 1991; Tillman, 2010). They were deployed in paired treatments: an exclusion cage treatment, where egg cards were covered with a clip cage to prevent predation but allow measurement of egg loss due to environmental conditions; and an open treatment, where cards were left uncovered and open to natural enemy attack (Jones et al., 2001; Tillman, 2010; Gardiner et al., 2013). In

2011 D. undecimpunctata eggs were glued to brown cards in sets of 30 and secured to the 33 ground with pins at the base of plants. Though they naturally occur more often on the undersides, 16 A. tristis eggs were glued to green cards and secured to the topsides of pumpkin leaves to aid observation by video cameras. They were deployed twice in southern Ohio (8 July, 29 July) and twice in northern Ohio (14 July, 3 Aug). In 2012, I used 18 D. undecimpunctata eggs and 10 A. tristis eggs, and they were deployed twice in southern Ohio (4 July, 17 July) and twice in northern Ohio (26 June, 11 July). In both years, one set of exclusion cage/open treatment cards were placed each pumpkin plot for

48 hours, and recounted in the laboratory after retrieval. A. tristis eggs were placed in a growth chamber held at 28 °C, 80% RH, and 12:12 light:dark to assess parasitism, though none was observed (Nechols et al., 1989).

Surveillance of egg predation

A modified 4-channel security camera system (Q-see, model no. QSC26404,

Anaheim, CA) was used to monitor predator activity on two D. undecimpunctata egg cards and two A. tristis egg cards (Grieshop et al., 2012). Cameras recorded predator activity at two sets of Open treatment D. undecimpunctata and A. tristis egg cards within two of the four sub-plots within a field for 24 hours. In 2011, this was conducted at two sites in southern Ohio (29 July) and two sites in northern Ohio (3 Aug). In 2012, data was collected once at each of 9 sites in southern Ohio (3 July or 17 July) and once at each 8 sites in northern Ohio (11 July). The video footage was transferred to portable hard- drives and stored until viewed. When an organism was observed interacting with the sentinel eggs, the time of arrival and departure was recorded. All predators were identified to the lowest taxonomic level possible with video recorded at 16 frames per second, and a playback pixel resolution of 352 x 240 (aspect ratio ~1.222:1). 34 Ground level arthropod collection

Pitfall traps were deployed in 2011 and 2012 to measure numbers of ground- dwelling predatory arthropods in response to landscape and local habitat management.

Since pitfalls are passive intercept traps, their catch more accurately reflects an estimate of activity density of mobile arthropods per field per sample period, each with their own likelihood of being captured, rather than abundance (Pavuk et al., 1997; Thomas et al.,

1998; Gardiner et al., 2010). Pitfalls were constructed from two 946 mL cups (Solo Cup

Company, Lake Forest, IL) with an 11 cm inside diameter. The outer cup was modified with drainage holes, and the inside cup held ~473 mL of a 10% dish soap solution (Laub et al., 2009). I used a golf course cup cutter to establish one hole in each plot in the pumpkins and floral insectary treatments. Pitfall placement in pumpkin sub-plots (2011-

2012), grassy control sub-plots (2011-2012), and perennial sub-plots (2012) was determined with a 6-sided die. Two additive rolls determined the number of paces to take along the length of the sub-plot, and one roll determined the number of paces to take into the perennial/control, or pumpkin sub-plot. The pitfalls were placed in the nearest pumpkin row to the last pace. In alyssum insectaries, an even or odd roll determined which alyssum row to use, and two additive rolls determined the number of paces within a sub-plot to place the pitfall. Empty pitfall cups were left in place for two weeks before taking data. In 2011, I conducted two collection periods of three days each in southern

Ohio (8 July, 11 July) and northern Ohio (14 July, 3 Aug). In 2012, I deployed pitfalls for seven days (to increase catch size), twice in southern Ohio (26 June, 17 July) and twice in northern Ohio (2 July, 25 July). I sorted pitfalls into important arthropod predator groups

(Carabidae, Cantharidae, Geocoridae, Nabidae, Formicidae, Orthoptera, Pulmonatta,

35 Araneae, and Opiliones) based on both previous evidence of key cucurbit pest natural enemies (Platt et al., 1999; Snyder and Wise, 2000; Rondon et al., 2003) and the results of previous video analysis (Grieshop et al., 2012; Smith, 2012). Dates for these trap collections coincided with two predation service experiments.

Measuring parasitism of adults

I also measured the activity of parasitoids on adult A. vittatum, and the combined capture of adult A. tristis and A. armigera collected by hand and aspirator. Within each site, I searched for adults by checking every plant in the plot, or until 45 minutes had passed; whichever came first. Equal sample sizes across sites were difficult to attain, and

I dropped collections from the analysis if they had fewer than 10 insects to generate a percentage of parasitism. For D. undecimpunctata, counts ranged from 0-51 in 2011 and

1-66 in 2012, while A. tristis/armigera counts varied from 0-9 in 2011 and 1-27 in 2012.

In 2011, collections were made twice in southern Ohio (8 July, 29 July) and once in northern Ohio (3 Aug, 17 Aug one site). In 2012 two collections were made in southern

Ohio (11-12 July, 25-26 July) and northern Ohio (12-13 July, 28-31 July). After the insects had been collected, they were all transferred to 3 x 8 cm clear plastic tubes with mesh windows cut into the tube and/or end caps, along with a slice of cucumber. I labeled and kept the insects in a growth chamber held at 28 °C, 80% RH, and 12:12 light:dark for

8 days. They were then frozen in a -80 °C freezer and dissected for developing parasitoids (Smyth, 2011). The response variable was the percentage of adults of each species parasitized at each site, pooled across sample dates.

36 Collecting landscape data

I obtained aerial image mosaics of each county that contained a research site from the year 2010 (OGRIP, 2010) and uploaded them into ArcMap (version 9.3; ESRI, 2011) and QGIS (version 1.8.0; Quantum GIS Development Team, 2012) to digitize all land cover elements. I determined the area of each distinct landscape feature within a 1500,

1000, and 500 m buffer around the geographic center of each plot and ground verified them with a classification system including 22 habitat types. Excluding open water, 21 fine-grain cover types were combined into 7 coarse-grain habitat categories totaling

100% of the land area, and the percentages of each habitat types were aggregated as predictor variables within each landscape buffer for analysis (%FIELD = annual field crops; %GRASSLAND = perennial grassland, fallow, and pastureland; %FORAGE = perennial alfalfa/oats; %FRUITVEG = mixed fruit and vegetable cropland; %FOREST = contiguous woodlands and hedgerows; %URBAN = impervious surfaces and buildings;

%TURF = mowed turfgrass).

Statistical analyses Assessment of predation and parasitism services Predation

The extent of predation on eggs of A. tristis and D. undecimpunctata in both years was modeled using a generalized linear mixed model (glmer function) with a Laplace maximum likelihood approximation that allowed for specification of a binomial distribution and logistic link function in the lme4 package of R (version 3.0.0; R

Development Core Team, 2013). I followed the modeling protocol described by Zuur et al. (2010) by exploring the data for outliers in the response variable, and the continuous 37 covariates using Cleveland dotplots. I also assessed collinearity between continuous variables with variance inflation factors (VIF) with the corvif function (Highland

Statistics Ltd. Newburgh, UK), and used the mixed model parameter estimates to determine whether a particular parameter, or combination of parameters, had a positive, negative, or neutral effect on the response variables.

The binomial response variable, Predij, was defined as the probability of egg removal from egg cards of A. tristis and D. undecimpunctata eggs, respectively in the jth plot at farm i at 48 hours. Predij was predicted by the single or combined fixed covariates of cage treatment (categorical: open or caged), site treatment (categorical: control, alyssum, or perennial), habitat variables (2012 only, continuous: percentages of field crops, grasslands, forage crops, fruit and vegetable crops, forests, impervious urban surfaces, and mowed turfgrass), and experimental period (categorical: experiment 1 and

2). Random covariates ai and xi set up a correlation structure for each farm and experimental period, and random covariate ɛij was introduced as an observation level random effect to function as a latent variable for handling overdispersion (Zuur et al.,

2009; Zuur and Saveliev, 2012).

Parasitism

Collections had to meet the criteria of having 10 or more specimens to be included in analysis. In 2011, no collections met this minimum for A. tristis, and for D. undecimpunctata nine collections in experiment 1, and one collection in experiment 2 were dropped. In 2012, 12 collections were omitted for A. tristis in both experiments, and for D. undecimpunctata four collections were dropped from experiment 1, and one collection from experiment 2. These omissions limited the number of parameters that 38 could be used simultaneously in models. As a result, I dropped A. tristis from analysis, and pooled data on A. vittatum from both experiments within years.

Parasitism was modeled using similar methods to the predation analysis, where the binomial response variable, Paraij, was defines as the probability of parasitism of adult A. vittatum in the jth observation at farm i. Paraij was predicted by the single fixed covariates of site treatment (categorical: control, alyssum, or perennial), and habitat type

(continuous: percentages of field crops, grasslands, forage crops, fruit and vegetable crops, forests, impervious urban surfaces, and mowed turfgrass). In 2012, random covariate ai was used to account for correlation between farm sites because some farms contained both the alyssum and perennial treatment.

I conducted multiple comparisons procedures with binomial models contrasting cage treatment (predation data only), experimental period (predation data only), site treatment (predation and parasitism data in 2012 only), and their interactions using the general linear hypothesis test (glht) function in the multcomp package to separate means. When contrasting the two experimental periods as a categorical fixed factor, I removed the random covariate xi. The multiple comparisons procedure adjusted z-scores and P-values with a single-step method. Significance was determined at the α = 0.05 level.

Assessment of ground level predator activity densities

I modeled the activity densities of the most common taxa using a generalized linear mixed model. I chose the glmmadmb function (glmmADMB package in R) with a

Laplace maximum likelihood approximation that allowed for specification of a negative binomial distribution for count data and logistic link function. I used the random factor ai, 39 to account for variability between farms. I conducted hypothesis tests by contrasting the differences in mean activity densities within each predator taxa between fixed covariates of pitfall placement (categorical: pitfalls placed in the pumpkin crop, or pitfalls placed in the floral insectary treatments), experimental periods, site treatment (in 2012 only), and their interactions using the general linear hypothesis test (glht) function in the multcomp package (α = 0.05).

Examining landscape and local habitat management impacts on predation and parasitism

To find the optimal set of continuous and fixed covariates to predict predator activity density, predation service, and parasitism service at each landscape scale (500,

1000, and 1500 m), I used an information-theory approach (Bolker et al., 2009; McElduff et al., 2010; Zuur et al., 2010) to make the 13 a priori models in 2011 and 13 models in

2012 (Table 2.3), including a null model representing random chance. In 2012, where local habitat management was evaluated, models were constructed to test the

Intermediate Landscape-Complexity hypothesis with activity density and predation data

(Tscharntke et al., 2012). I used the mixed model regression estimates to determine the inequalities between the local site treatments.

Following the information-theoretic approach, the 13 models for each species and each landscape radii (Anderson et al., 2001; Bolker et al., 2009) were compared using a corrected Akaike’s Information Criterion (AICc: for small samples sizes). Each model was given an AIC score, and the model with the highest score and weight was considered the most parsimonious explanation of the data. The AICc weight is a probability proportion between 0 and 1 that describes the likelihood that a model will be ranked in 40 the same position over infinite iterations. Models were considered competing candidate models if the AICc score differences were ≤ 2. Additionally, if the null model was ranked as a top or competing model, that was interpreted to mean that none of the 13 a priori models were any better at explaining the data than random chance. Significant interactions were plotted and interpreted using ggplot2 in R.

RESULTS Predation by experiment

In 2011, I did not detect significant predation of A. tristis eggs, indicated by no significant difference between open and exclusion cage egg cards. In the first experiment of 2011 (8 July, 14 July) 16.64% (± 4.49 SEM) of A. tristis eggs were removed from open egg cards, and 11.46% (± 3.92 SEM) were removed in the second experiment (29

July and 3 Aug). The egg predation was not significantly different between experiments

(Figure 2.2 A). For D. undecimpunctata in 2011, a significant interaction was observed between experiment and cage treatment (z = -3.96, P < 0.001), indicating that the level of predation varied across experiments. In both experiments, I detected significant egg removal (z = -7.53, P < 0.002 in experiment 1, and z = -9.16, P <0.001 in experiment 2)

(Figure 2.2 B). Yet, the 82.45% (± 4.33 SEM) of eggs removed from open egg cards during the second experiment (29 July, and 3 Aug) was significantly greater than the

67.67% (± 6.00 SEM) removed during the first experiment (8 July and 14 July) (z = 3.63,

P = 0.002).

In 2012, predators again removed a significant proportion of D. undecimpunctata eggs (z = -9.36, P < 0.001 in experiment 1, and z = -9.79, P < 0.001 in experiment 2),

41 but not A. tristis eggs (Figure 2.2 C and D). Additionally, egg predation during the first experiment (4 July, 26 June) and second experiment (17 July, 11 July) was not significantly different for either pest species.

Influence of floral insectaries on predation

In 2012, I found that local habitat treatment (grassy control, alyssum, and perennial insectaries) did not significantly impact predation of either egg species, in either experiment (Figure 2.2 C and D). Numerically, greater egg predation occurred in perennial treatments for both species, but there were no significant differences between the numbers of eggs removed from cards across the three types of site treatments. I observed 63.66% (± 8.01 SEM), 75.69% (± 6.50 SEM), and 89.82% (± 4.12 SEM) of eggs removed from open D. undecimpunctata egg cards in experiment 1 in the control, alyssum, and perennial insectaries, respectively; and 78.77% (± 6.21 SEM), 68.51% (±

5.92 SEM), and 71.03% (± 8.76 SEM) in experiment 2. For A. tristis I observed 6.13% (±

5.00 SEM), 7.92% (± 5.70 SEM), and 24.00% (± 12.22 SEM) of eggs removed from open A. tristis egg cards in the control, alyssum, and perennial insectary treatments, respectively; and 11.43% (± 5.14 SEM), 7.00% (± 3.17 SEM), and 12.86% (± 5.07 SEM) in experiment 2.

Video surveillance of egg predation

In 2011 the feeding guild attacking A. tristis and D. undecimpunctata eggs consisted of ants (Formicidae, n = 73), harvestmen (Opiliones, n = 39), and crickets

(Gryllidae, n = 27), while A. tristis eggs were only visited by ants (n = 7) and mice

(Cricetidae, n = 4). In 2012 a similar guild of predators fed upon A. tristis and D.

42 undecimpunctata eggs, with the addition of springtails (Entomobryidae) (Figure 2.3 A-

D). I observed no parasitism of A. tristis eggs in the day or night of either year. Foraging

Formicidae was more active on D. undecimpunctata eggs between 1800-2300 hr,

Opiliones between 2000-0400 hr, and Gryllidae between 0200-0600 hr (2011) and 1000-

2400 hr (2012). Entomobryidae was only observed in 2012, and was most active between

2200-0100 hr (Figure 2.4 A-C).

Predator activity densities

In 2011, pitfall traps collected Araneae (n = 1537), Formicidae (n = 721),

Carabidae (n = 309), Opiliones (n = 41), and Orthoptera (n = 7) (Figure 2.5 A). In 2012

Formicidae (n = 4251), Araneae (n = 3461), Carabidae (n = 1534), Orthoptera (n = 302),

Opiliones (n = 207) were collected (Figure 2.5 B). I dropped Orthoptera from analysis in

2011 for having fewer than 30 observations. Opiliones activity densities were greater in experiment 2 in 2011 (z = 3.88, P < 0.001) and 2012 (z = 2.11, P = 0.035). In 2011,

Formicidae were fewer in experiment 2 (z = -2.36, P = 0.018). There were no other differences in pitfall catch between experiments.

Influence of floral insectaries on predator activity densities

Floral insectary treatments had no effect on the activity densities of ground- dwelling predators in the adjacent pumpkins plots.

Adult parasitism and the influence of floral insectaries

In 2011, I found 20.45% (± 4.10 SEM) parasitism of 312 collected A. vittatum. In

2012, I found 19.61% (± 2.75 SEM) of 909 A. vittatum adults were parasitized. In 2012, I found that local habitat treatments did not impact parasitism of naturally occurring A. 43 vittatum adults, though numerically greater parasitism was detected in the alyssum and perennial floral insectary treatments. In 2012, 11.65% (± 3.61 SEM), 22.96% (± 4.92

SEM), and 24.81% (± 4.86 SEM) of collected beetles were parasitized in pumpkins adjacent to grassy control areas, annual alyssum insectaries, and perennial insectaries

(Figure 2.6). However, there were no statistically significant differences.

Landscape model comparisons

Many predation, parasitism, and activity density model sets ranked The Intercept

Only model M1 with < 2 AICc difference. This indicated that no a priori model explained more variation from a random distribution, and conclusions drawn from these model sets could be spurious. I omitted the models sets that ranked the Intercept Only model M1.

2011 models – A. tristis egg predation

In 2011, I found that A. tristis egg predation was negatively associated with, and best predicted by the proportion of forested habitat within 500 m of my sites (z = -3.18, P

= 0.002 in M6, and z = -2.93, P = 0.003 in M13). Within 1000 m the percentage of urban areas in model M7 was the best-fit model with no significant predictor, with competing model M6 showing a negative relationship with forested areas (z = -3.53, P < 0.001).

Within 1500 m, urban areas in model M7 were again the most predictive habitat type on

A. tristis, as well as the model M11 containing both urban and mowed turfgrass habitats, though no relationship with these habitats were significant. Summing the AICc weights of models containing important predictors illustrated that forested areas became a less likely predictor of A. tristis egg predation at larger radii (0.875 at 500 m, 0.191 at 1000 m, and

44 0.001 at 1500 m), but urban areas became more likely at larger radii (0.012 at 500 m,

0.608 at 1000 m, and 0.709 at 1500 m) (Table 2.4).

2011 models – D. undecimpunctata egg predation

In 2011, D. undecimpunctata egg predation was positively associated with, and best predicted by field crop areas at 1000 (z = 3.55, P < 0.001) and 1500 m (z = 3.01, P =

0.003) in model M2, and in model M12 within 1500 m (z = 4.02, P < 0.001). Egg removal was also positively associated with fruit and vegetable areas (z = -2.32, P =

0.020) within 1500 m in model M12 (Table 2.4).

2012 models and site treatment predictors – A. tristis and D. undecimpunctata egg predation

In 2012, Intercept Only models M1 were consistently ranked with high Akaike weights (0.143 at 1000 m, and 0.19 at 1500 m for A. tristis, and 0.57 at 500 m, 0.722 at

1000 m, and 0.638 at 1500 m for D. undecimpunctata), suggesting that habitat and site treatment parameters did not describe the egg predation well. However, A. tristis predation was positively associated with the percentage of grasslands (z = 2.07, P =

0.036) within 500 m of sites in model M26 (Table 2.4).

2011 models – Pitfall predator activity densities in pumpkins

In pitfalls placed in pumpkin plots in 2011, the activity density of Araneae was best predicted by forage crops within 500 m in model M4 (z = 3.01, P = 0.003) and M9

(z = 3.02, P = 0.003). The activity density of Opiliones was positively associated with and best predicted by field (z = 2.29, P = 0.002), and forage crops (z = 2.35, P = 0.019) in model M12, and fruit and vegetable crops in model M12 (z = 5.50, P < 0.000), and

45 M5 (z = 3.40, P < 0.001) within 1000 m. Formicidae was positively associated with impervious urban structures within 500 m (z = 1.96, P = 0.050) in model M11, and negatively associated with mowed turfgrass in model M11 (z = -3.37, P = 0.001), and

M8 (z = -2.51, P = 0.012) (Table 2.5).

In pitfalls placed within the grass border of pumpkin fields in 2011, Carabidae were best described by and positively associated with forage crops within 1000 m in model M4 (z = 2.47, P = 0.013), and M9 (z = 2.64, P = 0.008). Opiliones was also best predicted by the tall perennial grassland model M9, but no predictors were significant

(Table 2.5).

2012 models – Predator activity densities in pumpkins

In pitfalls placed in pumpkins in 2012, Formicidae was again positively associated with urban areas within 1000 m (z = 2.46, P = 0.014) and 1500 m (z = 2.55, P

= 0.011), and negatively associated with mowed turfgrass areas within 1000 m (z = -

5.05, P < 0.001) and 1500 m (z = -3.72, P < 0.001) in model M24. Orthoptera was positively associated with and best predicted by fruit and vegetable crops at 500 m (z =

2.63, P < 0.001), 1000 m (z = 3.68, P < 0.001), and 1500 m (z = 2.70, P = 0.007), and perennial insectaries at 500 m (z = 3.45, P < 0.001), 1000 m (z = 4.13, P < 0.001), and

1500 m (z = 3.73, P < 0.001) in model M25. Orthoptera was also negatively associated with field crops within 500 m (z = -2.69, P = 0.007) and positively associated with forage crops within 1500 m (z = 2.04, P = 0.041) in model M25. Opiliones was best described by the all agriculture model M25 and tall perennial grassland model M22 at 500 m.

However, no predictors were significant for Opiliones at that scale. For Carabidae, there was a significant interaction between the percentage of fruit and vegetable crop areas 46 within 500 m and the floral insectary treatments (z = -3.25, P = 0.001). Carabidae in pumpkins planted adjacent to grassy control insectaries increased significantly more than those in sites planted adjacent to both types of floral insectaries with an increase in fruit and vegetable crop habitats (Table 2.5).

In pitfalls placed within floral insectaries adjacent to pumpkins in 2012, Opiliones was best predicted by, and positively associated with mowed turfgrass within 500 m (z =

2.47, P = 0.013), and 1000 m (z = 3.66, P < 0.001) in model M24. Opiliones also showed a positive association with alyssum (z = 2.08, P = 0.037) and perennial insectaries (z =

2.56, P = 0.010), and a negative relationship with urban areas (z = -3.07, P = 0.002) within 1000 m in model M24. Formicidae was negatively associated with mowed turfgrass within 1000 m in models M21 (z = -2.05, P < 0.001), M23 (z = -3.83, P <

0.001), and M24 (z = -3.85, P < 0.001). Orthoptera was positively associated fruit and vegetable crops within 1000 m (z = 4.38, P < 0.001) and 1500 m (z = 3.44, P = 0.001), perennial insectaries within 1000 m (z = 4.52, P = 0.010) and 1500 m (z = 4.32, P <

0.001), and forage crops within 1500 m (z = 2.11, P = 0.035) in model M25. Araneae was best predicted by, and positively associated with forage crops within 1000 m (z =

2.51, P = 0.012) and 1500 m (z = 2.88, P = 0.004), and perennial insectaries within 1000 m (z = 2.27, P = 0.023) and 1500 m (z = 2.47, P = 0.014) in model M22. Araneae was also negatively associated with grasslands at 1000 m (z = -3.72, P < 0.001) and 1500 m

(z = -3.81, P < 0.001) in model M22 (Table 2.5).

An interaction was found for the activity density of Araneae between the floral insectaries and the proportion grassland habitats within 500 m (z = 2.68, P = 0.007). In the presence of higher percentages of grassland habitats, the activity densities of Araneae

47 were not strongly affected within both alyssum and perennial insectaries, but Araneae found in grassy control insectaries decreased (Figure 2.7 C). A significant interaction was found between activity density of Carabidae found in the floral insectaries and the amount of mowed turfgrass within 500 m of pumpkin sites (z = 4.16, P < 0.001).

Activity density of Carabidae within alyssum insectaries increased with the percentage of mowed turfgrass in the area, while activity densities within both the grassy control insectaries and perennial plots decreased (Figure 2.7 A). Similarly, the activity density of

Orthoptera within alyssum insectaries exhibited an interaction with the proportion of mowed turfgrass within 500 m of the pumpkin sites (z = 3.95, P < 0.001). Orthoptera within alyssum insectaries increased with the percentage of mowed turfgrass in the area, while activity densities within both the grassy control insectaries and perennial plots decreased (Figure 2.7 B).

2011 and 2012 models – A. vittatum adult parasitism

In 2011, parasitism of adult A. vittatum was positively associated with urban habitats within 500 m (z = 2.30, P = 0.022) in model M7 and forage crop habitats within

1500 m (z = -2.48, P = 0.013) in model M4 (Table 2.6). In 2012, parasitism was again positively associated with urban habitats within 500 m (z = 3.25, P = 0.001) in model

M20, but also negatively associated with fruit and vegetable crops in model M18 within

500 m (z = -3.75, P < 0.001), 1000 m (z = -4.32, P < 0.001), and 1500 m (z = -3.73, P <

0.001) (Table 2.6).

48 DISCUSSION

This study examined the effects of local habitat management and landscape composition on the activity density of ground-dwelling insect predators, the levels of predation on the eggs of A. tristis and D. undecimpunctata, and the parasitism of adult A. vittatum. Other research has measured the abundance of natural enemies as a proxy for biocontrol service, yet fewer studies measured the actual services in the field (Kremen and Ostfeld, 2005; Chaplin-Kramer et al., 2011; Tscharntke et al., 2012). Those that measured the influence of habitat management on services have found positive and neutral impacts of floral insectaries. My study also found a neutral effect of non-native annual sweet alyssum and native perennial floral insectaries planted adjacent to pumpkins. Studies that investigated the effect of landscape composition on biocontrol services have also found positive and neutral effects of non-crop habitats. I found inconsistent effects of surrounding habitats on predation of A. tristis and D. undecimpunctata eggs. Parasitism of A. vittatum was negatively related to the percentage of fruit and vegetable crops and positively related to urban landscapes. Turfgrass, grassland, and fruit and vegetable habitat within 500 m interacted with local habitat management to affect the activity densities of Carabidae, Araneae and Orthoptera.

However, video footage of egg cards revealed that the dominant predators of A.tristis and

D. undecimpunctata eggs were Formicidae, which made up between 45-87% of the observations. Formicidae was also consistently associated with urban and mowed turfgrass habitats.

49 Egg predation services in pumpkin plots

I observed significant predation of D. undecimpunctata eggs in both years. In

2011, between 67.67% and 82.45% of eggs were removed from open egg cards in the early and late sample periods, and this difference was significant. In 2012 there was no significant difference between sample periods. There was no significant predation of A. tristis eggs in either year.

In my study, all predators that appeared on video were also captured in pitfalls, though not with the same relative frequencies. The predators observed attacking eggs were Formicidae, Opiliones, Gryllidae and Entomobryidae (2012 only), with a majority of the pitfall catches made up of Araneae, Formicidae, and Carabidae. Entomobryidae are largely detritivores, but have been found to feed on other Collembolans and micro- organisms (Rusek, 1998). However, they were not known to be important predators of D. undecimpunctata eggs until after the 2012 field season, and therefore were not counted in pitfalls. Formicidae has been found to be an important predator of eucalyptus borer

(Phoracantha semipunctata) eggs in Portugal (Way et al., 1992), and eastern tent caterpillar (Malacosoma amaericanum) in Michigan (Tilman, 1978).

The use of video cameras demonstrated the importance of research on predation services, rather than simply abundance and diversity of natural enemies. The predators observed interacting with eggs were not species typically targeted for predation studies.

Grieshop et al. (2012) also utilized video surveillance to observe predation on cranberry fruitworm pupa (Acrobasis vaccinii) and Japanese beetle (Popillia japonica) eggs in blueberry, and restrained Galleria mellonella in corn and grasslands. Smith (2012) observed predation on coccinellid eggs in alfalfa crops and grasslands, and Pfannenstiel

50 and Yeargen (2002) investigated predation of Helicoverpa zea eggs in soy bean. In all species and in all environments, they found similar nighttime responses by Formicidae,

Gryllidae and Opiliones. Grieshop et al. (2012) also compared the observed predators and paired pitfall catches in blueberries, and demonstrated that though a consistently high proportion of carabid beetles were captured in pitfall traps, they showed up much less frequently as foraging predators on video. It is possible that the natural enemies caught in pitfalls were predating on organisms other than the bait used for these experiments, and the measure of their abundance and diversity is still a good measure of potential predation services. However, their measure can overstate the actual functions of natural enemies when specific questions are being asked about target pests system (Kremen and Ostfeld,

2005; Chaplin-Kramer et al., 2011; Grieshop et al., 2012; Tscharntke et al., 2012).

Influence of floral insectaries on egg predation

I predicted that the addition of native perennial floral insectaries would support the highest level of predator activity density and predation of D. undecimpunctata and A. tristis eggs in pumpkin by providing a more diverse habitat structure and a more continuous bloom period, and that the alyssum insectaries would only support more predators and predation services than the grassy control insectaries. I did not find evidence that floral insectaries had an effect on predation services. Predation of A. tristis eggs was 12 % higher and D. undecimpunctata was 10% higher in pumpkin sites adjacent to the perennial insectaries compared to the control, and nearly equal to alyssum treatments. There also appeared to be more Araneae, Formicidae, and Orthoptera found in pitfall traps located in pumpkin plots adjacent to perennial insectaries, though fewer

Carabidae. In the presence of alyssum insectaries, activity density of Formicidae caught 51 in pitfall traps placed in the pumpkin plots were the only taxa that increased, and

Araneae, Carabidae, and Orthoptera decreased. I also predicted that the activity density of ground-dwelling predators caught in pitfalls would be higher in the floral insectary plots than in the pumpkin plots, but did not find evidence to support this.

Other researchers have found an increase in predator numbers in a floral insectary addition. For example, Woltz et al. (2012) found significantly more coccinellids in insectaries of buckwheat planted adjacent to soybean, Platt et al. (1999) found more predators in a buckwheat border than in adjacent cucumber and squash fields, and

Macleod et al. (2004) found an increase in ground-dwelling predators overwintering in perennial overgrown grass banks in field margins. Some researchers have found evidence of dispersal into the crop. For example, Walton and Isaacs (2011) found that natural enemies increased in a blueberry field adjacent to a conservation planting of wildflowers, and Thomas et al. (1991) found positive effects overgrown grassy field margins on the, diversity and dispersal capabilities of carabids into a crop. However, similar to my results, those that measured biocontrol services found no effect: Woltz et al. (2012) found that sentinel aphids were not significantly predated upon within the crop, and Platt et al.

(1999) found that biocontrol of cucumber beetles was not enough to improve yield with the addition of a floral insectary. Additionally, Olson and Wäckers (2007) investigated predation and services on corn earworm (Helicoverpa zea) eggs in cotton crops planted near infrequently-disturbed perennial vegetative buffers. They found no effect of the type of field margin on the mortality of H. zea eggs by predation, and predation occurred consistently at every distance measured within the crop, regardless of field margin

52 In my study, transplanting sweet alyssum into plots required labor-intensive greenhouse logistics, careful transplanting, fertilizer application, frequent irrigation, and plants were scorched by sun, and bloomed weakly in the shade. They were fickle, and an experiment direct seeding into fields resulted in extremely patchy distributions of plants that did not experience a substantial bloom until October, well after the cucurbit growing season had ended. Additionally, though the transplants established well and bloomed, after 1 month of growth, pumpkin leaves were large enough to block the sun from alyssum rows at sites with narrow row spacing. These factors could explain the neutral response by predators and parasitoids on pumpkin pests with alyssum floral insectary additions. Despite the low quality of the alyssum, Carabidae, and Orthoptera appeared to rely on alyssum insectaries at sites with high percentages of mowed turfgrass. Perhaps with careful site selection and management with more space to grow, alyssum could flourish and attract these predators more effectively.

In 2012, the perennial plots maintained a large and constant bloom all spring, summer, and fall from four plants from the custom seed mixture (yarrow (Achillea millefolium), bergamot (Mondarda fistulosa), three-lobed coneflower (Rudbeckia triloba) yellow coneflower (Rudbeckia, pinata)), and three wild flowering weeds (black medic

(Medicago lupulina), fleabane (Erigeron philadelphicus), cinquefoils (Potentilla spp.), and clovers (Trifolium spp.)). My data showed a trend in numerically higher levels of predation, parasitism and activity densities of Formicidae, Araneae, and Orthoptera associated with perennial insectaries, and I expect that over the course of at least 3 years, more of the perennials planted in 2010 will have developed a large enough root mass to bloom as well. Blaauw (2013) recorded how perennials that were started from seed

53 increased their percent coverage area each year across 4 years of development, and had an increased floral density across 2 sampled years. This increase in nectar resources could enhance resident parasitoid populations in step (Landis et al., 2000), and ants and spiders have both been documented feeding from flowers and extra-floral nectaries (Tilman,

1978; Jackson et al., 2001; Chen et al., 2010). In addition, their above-ground vegetative structure provides vital nesting, foraging, and overwintering habitat. Macleod et al.

(2004) found that abundance of Araneae, Staphylinidae and Carabidae increased in perennial grass banks, that their abundance positively correlated with dispersal into agricultural fields, and that carabid diversity increased significantly after 4 years of development.

Influence of landscape composition on activity density of predators and egg predation

I predicted that higher percentages of semi-natural, non-crop habitat would support higher activity density of predators and a higher level of predation services. I found that there were inconsistent impacts of habitat predictors on the predation of A. tristis and D. undecimpunctata eggs across years, but some patterns emerged in the activity densities of predators. In both years, Formicidae was positively correlated with the proportion of urban habitats, and negatively correlated with mowed turfgrass within

500 m of my sites. This could reflect the encroachment of residential areas into farmland.

More urban areas may have concentrated the numbers of ants I found foraging in my fields, but mowed turfgrass may have been a competing habitat drawing ants away from pumpkin sites. Importantly, these types of non-crop habitats are not often differentiated

54 from other more natural types of non-crop habitats, like forests or grasslands (Bianchi et al., 2006). These results suggest that they should be.

I also predicted that in the presence of intermediate habitat-complexity, predator activity density and predation occurring at sites with local floral insectary additions (in

2012) would be significantly greater than at sites with a grassy control insectary. I found little evidence supporting the Intermediate Landscape-Complexity Hypothesis for

Carabidae, Orthoptera and Araneae caught in pitfall traps placed in the floral insectaries.

With this hypothesis I expected to see the biggest differences between floral insectary treatments with between 1-20% non-crop habitats surrounding my sites, with smaller difference or no difference between treatments with < 1 % non-crop habitat, or > 20% non-crop habitat. This would manifest itself as a significant interaction term that could be interpreted with an interaction plot. I found that Carabidae and Orthoptera found in pitfalls within the alyssum insectaries increased with more mowed turfgrass habitat, while those taxa in the grassy control insectaries and perennial insectaries decreased

(Figure 2.7 A and B). Within 500 m of my sites, no more than 20% of the landscape was comprised of mowed turfgrass, within the intermediate landscape-complexity range. The smallest differences in activity density of these predators between the floral insectaries and grassy control insectaries occurred on the outer edges of this range, suggesting that the Intermediate Landscape-Complexity Hypothesis may not hold for this type of anthropic non-crop habitat. Further, in landscapes with high percentages of mowed turfgrass, it appears as though the non-native annual, sweet alyssum, is an attractive resource. Araneae caught in pitfalls in the perennial insectaries were not strongly affected by an increase in grasslands within 500 m, but those located in alyssum and grassy

55 control insectaries decreased (Figure 2.7 C). Only one of my sites had greater than 20% grassland habitat within 500 m, but there was still a trend showing greater difference occurring with more of this non-crop habitat. In sites with more grassland, Araneae may have found it easy to migrate to and utilize the perennial insectaries for its complex vegetation structure more than the alyssum or grassy control insectaries.

Influence of floral insectaries on parasitism

I predicted that the addition of native perennial floral insectaries would support the highest level of parasitism of A. vattatum eggs in pumpkin by providing a more diverse habitat structure and a more continuous bloom period, and that the alyssum insectaries would only support more parasitism services than the grassy control insectaries. I found that parasitism of A. vittatum in pumpkin sites adjacent to the perennial insectaries (24.81%) more than double that of the grassy control insectaries

(11.65%), and greater than that of the alyssum (22.96%) insectaries.

Other research has suggested that floral insectaries may attract more parasitoids, but they simply stay inside the floral insectary. Lee et al. (2006) found that that there was a positive relationship between the presence of flowers and the amount of sugars found in the guts of the parasitoid, Diadegma insulare, in annual buckwheat margins bordering cabbage, but there was no relationship between parasitism of diamondback moth

(Plutella xylostella) and floral resources in study plots within the crop. This suggests, that if a floral insectary is successful, it may increase parasitoid abundance, but not necessarily translate to higher parasitism rates outside of the floral insectary.

Additionally, Olson and Wäckers (2007) also investigated parasitism services on corn earworm (Helicoverpa zea) eggs in cotton crops planted near infrequently-disturbed 56 perennial vegetative buffers, but found no effect of the type of field margin on the mortality of H. zea eggs by parasitism. In contrast to Lee et al. (2006), they hypothesized that unstable populations of natural enemies could have been caused by a lack of carbohydrates available in the vegetative buffer, which prevented recruitment of natural enemy populations large enough to appear more frequently in the crop.

Influence of landscape composition on parasitism

I predicted that higher percentages of semi-natural, non-crop habitat would support higher activity density of predators and a higher level of parasitism services. In both years, parasitism of adult A. vittatum significantly increased with the presence of a higher proportion of impervious urban surfaces within 500 m of my sites, which along with the association between Formicidae and urban environments could reflect encroaching residential developments in agricultural areas in The United States.

Additionally, models of A. vittatum parasitism in 2012 showed a significant negative association with the proportion of fruit and vegetable crop habitats between 500-1500 m.

This is interesting because the parasitoids that were found emerging from A. vittatum adults were specialized Celatoria spp. (Tachinidae) that would be found more often where they could find their hosts. The areas with more mixed fruit and vegetable crops were dominated by many cucurbit varieties of melons, squashes, and pumpkins where this specialist fly genera would be expected to proliferate on their cucumber beetle hosts

(Elzinga et al., 2007). However, the increase in fruit and vegetable habitat could have caused a dilution effect in these parasitoid populations. I also predicted that in the presence of intermediate habitat-complexity, parasitism occurring at sites with local floral insectary additions (in 2012) would be significantly greater than at sites with a grassy 57 control insectary. However, I did not find evidence of this occurring, with no significant interactions specifying a larger difference in parasitism at sites with floral insectaries with non-crop habitats between 1-20%.

This study opens doors for future research. For example, non-crop habitat was further classified into anthropic habitats in this study, such as urban areas consisting of impervious concrete, metal and wooden structures, and mowed turfgrass areas. More significant correlations were found between predator activity density, predation and parasitism services, and mowed turfgrass and urban habitats than between grassland and forest habitats. It is important to differentiate from semi-natural non-crop habitat, like forests and grasslands, because urbanized areas are a fast-growing habitat type in agricultural lands as cities expand to support more than 50% of the world’s population.

Further, a field study investigating biocontrol that consists of a mixture of candidate insect bait species in different life-stages could elucidate these uncertainties. For example, bait prey harboring certain characteristics like a hard and soft-bodied larvae, hard and soft-shelled eggs, mobile and stationary larvae and adults, and presence/absence of pharmacophagy. Alternatively, since some target pests are hard to control for in mark- recapture, or field arena studies, other techniques that can discern predation on target insects are gut content analysis (Szendrei et al., 2010), and fecal analysis (Whitaker Jr,

1995). Further, more research into the dispersal and host-colonizing mechanisms of the

Celatoria spp. parasitoids across multiple years should be performed to elucidate dilution effects in the presence of more cucurbit crops.

58 TABLES

2011 Site/ 2012 Site Latitude Longitude Treatment Latitude Longitude

1 40°/54'/37.94" 82°/6'/35.06" a 40°/54'/9.69" 82°/6'/44.11" 2 40°/55'/6.92" 82°/2'/57.66" a 40°/54'/58.95" 82°/2'/48.14" 2 > > p 40°/55'/5.27" 82°/2'/38.15" 3 40°/56'/25.06" 82°/6'/58.21" > > > 4 41°/5'/2.65" 81°/57'/1.51" c 41°/5'/3.28" 81°/57'/8.13" 5 40°/42'/37.87" 81°/58'/16.31" c 40°/42'/23.5" 81°/57'/56.45" 6 40°/55'/17.93" 81°/18'/33.26" c 40°/55'/17.29" 81°/18'/31.78" 7 > > p 40°/44'/12.27" 82°/11'/48.86" 8 > > a 40°/58'/13.68" 81°/44'/25.37" 9 39°/26'/5.63" 83°/59'/26.59" a 39°/26'/4.39" 83°/59'/1.35" 9 > > p 39°/26'/4.01" 83°/59'/25.23" 10 39°/2'/50.88" 82°/59'/37.4" a 39°/2'/49.35" 82°/59'/38.15" 10 > > p 39°/2'/50.88" 82°/59'/37.4" 11 39°/13'/13.41" 83°/25'/36.81" p 39°/13'/13.41" 83°/25'/36.81" 12 39°/10'/58.65" 83°/21'/3.09" c 39°/10'/55" 83°/21'/11.37" 13 38°/59'/29.9" 82°/46'/4.54" c 38°/59'/37.44" 82°/45'/51.76" 14 39°/8'/16.65" 82°/58'/58.47" c 39°/8'/11.46" 82°/58'/59.39" 15 > > a 39°/24'/41.94" 83°/9'/27.33"

Table!2.1.!The!location!and!2012!treatment!assignments!for!pumpkin!farms!sampled!in!2011!and!2012.!All! farms!had!one!pumpkin!field!except!numbers!2,!9,!and!10,!which!consisted!of!two!pumpkin!fields!each!with!both! floral!insectaries!in!2012.!Site treatments were two 60 m rows of an annual non-native plant, sweet alyssum, flanking 4 rows of pumpkins (a); a native perennial mixture planted in a 6 x 60 m area adjacent to the pumpkins (p); or a 6 x 60 m grassy control insectary adjacent to the pumpkins (c). At!farms!with!two!pumpkin!fields,!the!fields!were! separated!by!304,!51,!and!570!m,!respectively.!

59 Percent' Bloom'Period Common'Name Species by'weight May June July Aug Sept Oct

Spiderwort Tradescancia*ohiensis 1.75 Canada'anemone Anemone*canadensis* 2.50 Cow'parsnip Heracleum*maximum 2.50 Common'milkweed Asclepias*syriaca 2.50 Three'lobed'coneflower Rudbeckia*triloba 1.75 Yarrow Achillea*millifolium 4.25 Sand'tickseed Coreopsis*lanceolata 4.25 Culver's'root Veronicastrum*virginicum 1.00 Swamp'milkweed Asclepias*incarnata 3.50 Bergamot Monarda*fistulosa 1.75 Canada'wild'rye* Elymus*canadensis* 15.00 Cup'plant Silphium*perfoliatum 3.50 Rosinweed Silphium*integrifolium 1.75 Blue'lobelia Lobelia*siphilitica 1.75 Yellow'coneflower Ratibida*pinnata 3.50 Spotted'JoeNPye'weed Eupatorium*maculatum 1.75 Horsemint Monarda*punctata 1.75 Boneset Eupatorium*perfoliatum 3.50 False'boneset Kuhnia*eupatorioides 1.75 Little'bluestem* Schizachyrium*scoparius* 25.00 Stiff'goldenrod Solidago*rigida 4.25 Showy'goldenrod Solidago*speciosa 2.00 Riddell's'goldenrod Solidago*riddellii 3.50 Tall'coreopsis Coreopsis*tripteris 1.75 New'England'aster Aster*novaeEangliae 3.50

Table!2.2. Native perennial floral insectaries consisting of 23 forbs and 2 grasses* were established in 6 x 60 m plots in 2012. The impact of these habitats on predation and parasitism in pumpkin was assessed in 2012. The seed mix was designed following Fielder et al. (2007) to support the production of floral resources throughout the growing season.

60 Year Model Description 20123Model3Codea,b Predictionsc

2011 M1 Random3distribution Intercept Random3distribution3(0) FrequentlyAdisturbed3333333333333333333333333333 M2 %FIELD %FIELD3(A) annual3field3crops Uncultivated3grasslands3and33333333333333333333 M3 %GRASSLAND %GRASSLAND3(+) openAcanopy3river3corridors InfrequentlyAdisturbed3333333333333333333 M4 %FORAGE %FORAGE3(0) perennial3crops M5 Fruit3and3vegetable3crops %FRUITVEG %FRUITVEG3(+) Uncultivated3forest,3fencerows,3and3 M6 %FOREST %FOREST3(+) closedAcanopy3river3corridors M7 ManAmade3impervious3surfaces %URBAN %URBAN3(A)

M8 Mowed3turf %TURF %TURF3(A) Perennial3crops3and3333333333333333333 M9 %GRASSLAND3+3%FORAGE %GRASSLAND3(+)3and3%FORAGE3(0) uncultivated3areas %GRASSLAND3+3%FORAGE3+3 %GRASSLAND3(+),3%FORAGE3(0),3and3 M10 All3grassy3areas %TURF %TURF3(A) M11 All3urbanized3areas %URBAN3+3%TURF %URBAN3(A)3and3%TURF3(A) %FRUITVEG3+3%FIELD3+3 %FRUITVEG3(+),3%FIELD3(A),3and3 M12 All3forms3of3agriculture %FORAGE %FORAGE3(0) M13 All3forms3of3uncultivated3land %FOREST3+3%GRASSLAND %FOREST3(+)3and3%GRASSLAND3(+)

2012 M1 Random3distribution Intercept Random3distribution3(0)

FrequentlyAdisturbed3333333333333333333333333333 %FIELD3(A),3333333333333333333333333333333333333333333 M15 %FIELD3*3SITETRT annual3field3crops SITETRT3(c3<3a3<3p) Uncultivated3grasslands3and33333333333333333333 %GRASSLAND3(+),333333333333333333333333333333 M16 %GRASSLAND3*3SITETRT openAcanopy3river3corridors SITETRT3(c3=3a3=3p) InfrequentlyAdisturbed3333333333333333333 %FORAGE3(0),3333333333333333333333333333333333333 M17 %FORAGE3*3SITETRT perennial3crops SITETRT3(c3<3a3<3p) %FRUITVEG3(+),3333333333333333333333333333333333 M18 Fruit3and3vegetable3crops %FRUITVEG3*3SITETRT SITETRT3(c3=3a3=3p) Uncultivated3forest,3fencerows,3and3 %FOREST3(+),33333333333333333333333333333333333333 M19 %FOREST3*3SITETRT closedAcanopy3river3corridors SITETRT3(c3=3a3=3p) %URBAN3(A),3333333333333333333333333333333333333333 M20 ManAmade3impervious3surfaces %URBAN3*3SITETRT SITETRT3(c3<3a3<3p) %TURF3(A),3333333333333333333333333333333333333333333 M21 Mowed3turf %TURF3*3SITETRT SITETRT3(c3<3a3<3p) Perennial3crops3and3333333333333333333 %GRASSLAND3+3%FORAGE3+3333%GRASSLAND3(+)3and3%FORAGE3(0),3333 M22 uncultivated3areas SITETRT SITETRT3(c3=3a3=3p) %GRASSLAND3+3%TURF3+333333 %GRASSLAND3(+),3and3%TURF3(A),3333333 M23 All3grassy3areas SITETRTb SITETRT3(c3<3a3<3p) %URBAN3+3%TURF3+33333333333333333333%URBAN3(A)3and3%TURF3(A),3333333333333333333 M24 All3urbanized3areas SITETRT SITETRT3(c3<3a3<3p) %FRUITVEG3+3%FIELD3+333333333 %FRUITVEG3(+),3%FIELD3(A),3and3 M25 All3forms3of3agriculture %FORAGE3+3SITETRT %FORAGE3(0),3SITETRT3(c3<3a3<3p) %FOREST3+3%GRASSLAND3+3333333%FOREST3(+)3and3%GRASSLAND3(+),33333 M26 All3forms3of3uncultivated3land SITETRT SITETRT3(c3=3a3=3p)

Table!2.3. A total of 26 models were used on the predation and parasitism data. Models M1-M13 were used for 2011 for predation and parasitism data. M1-M13 were used again for parasitism data in 2012. Models M1,M15-M26 were used for predation data in 2012. a An astrisk in R denotes that each factor is included individually in the model, and also interacted. b %Forage was removed from the 2012 All Grass model at 1500 m because of a high VIF score. c Inequality statements indicate more, less, or equal effect of floral insectary treatment on the response variable, where c = control, a = alyssum, and p = perennial.! 61 Year Response*Species Scale*(m) Model LL K AICc Δi Wi Significant*factor*estimates*(p>value) 2011 A.#tristis 500 M6 '94.640 5 199.290 0.000 0.612 %FOREST2'0.1112(0.002) M13 >94.350 6 200.709 1.686 0.263 %FOREST*>0.083*(0.003)

1000 M7 '96.300 5 202.599 0.000 0.448 M6 >97.160 5 204.320 1.721 0.190 %FOREST*>0.115*(0.000)

1500 M7 '96.550 5 203.100 0.000 0.483 M11 >96.180 6 204.351 1.518 0.226

D.#undecimpunctata 1000 M2 '129.600 5 269.242 0.000 0.630 %FIELD20.0932(0.000)

1500 M12 '128.300 7 270.552 0.000 0.402 %FIELD20.1252(0.000),2%FRUITVEG2'0.5542(0.020) M2 >130.800 5 271.524 0.402 0.329 %FIELD*0.101*(0.003)

2012 A.#tristis 500 M26 '80.43 8 176.867 0 0.553 %GRASSLAND20.1892(0.036) M17 >80.25 9 178.498 1.962 0.207

Table!2.4. AICc table showing the top candidate egg predation models for each year, landscape scale and species (bold), followed by models that were ranked with differences ≤ 2. LL = log-likelihood, K = number of parameters, Δi = AICc difference, Wi = AICc weight. *%FORAGE was dropped from this a priori model in 2012 because of a high VIF score. Factors within models that were significant are listed, along with their model estimates and p-values.

62 Response* Year Pitfall*Placement Scale*(m) Model LL K AIC Δ W Significant*factor*estimates*(p?value) Species c i i 2011 Pumpkins 500 Araneae M4 3291.12 5 592.246 0 0.509 %FORAGE=0.075=(0.003) M9 ?291.1 6 594.202 2.222 0.168 %FORAGE*0.076*(0.003)

Formicidae M11 3186.74 6 385.478 0 0.398 %URBAN=0.115=(0.050),=%TURF=30.308=(0.001) M8 ?188.72 5 387.442 1.698 0.171 %TURF*?0.181*(0.012)

%FRUITVEG=0.271=(0.000),=%FIELD=0.041=(0.002),=== 1000 Opiliones M12 339.22 7 92.441 0 0.639 %FORAGE=0.155=(0.019) M5 ?42.61 5 95.216 2.182 0.215 %FRUITVEG*0.216*(0.000)

1500 Opiliones M12 340.26 7 94.525 0 0.393 %FRUITVEG=0.488=(0.000),=%FIELD=0.043=(0.026) M5 ?42.97 5 95.944 0.827 0.26 %FRUITVEG*0.484*(0.002)

Insectary=Strip 1000 Carabidae M4 3157.49 5 324.974 0 0.362 %FORAGE=0.152=(0.013) M9 ?156.86 6 325.71 1.002 0.219 %FORAGE*0.151*(0.008)

Opiliones M9 35.14 6 22.281 0 1

%FRUITVEG=0.104=(0.000),=PERENNIAL=1.117=(0.008),= 2012 Pumpkins 500 Carabidae M18 3314.51 9 645.75 0 0.6 FRUITVEG=:=PERENNIAL=30.108=(0.001) M25 ?313.86 9 647.028 1.278 0.317 %FRUITVEG=0.049=(0.001)

Opiliones M25 3110.46 9 238.914 0 0.529 M22 ?112.3 8 240.6 1.233 0.285

%FRUITVEG=0.069=(0.000),=%FIELD=30.025=(0.007),= Orthoptera M25 3121.48 9 260.954 0 0.617 PERENNIAL=1.827=(0.000) M18 ?122.3 9 262.598 1.644 0.271 %FRUITVEG*0.073*(0.009),*PERENNIAL*2.102*(0.001)

1000 Formicidae M24 3439.92 8 895.844 0 0.863 %URBAN=0.048=(0.014),=%TURF=30.322=(0.000)

Orthoptera M25 3123.27 9 264.548 0 0.634 %FRUITVEG=0.200=(0.000),=PERENNIAL=2.043=(0.000)

%FRUITVEG=0.295=(0.007),=%FORAGE=0.203=(0.041),= 1500 Orthoptera M25 3124.72 9 267.432 0 0.524 PERENNIAL=1.932=(0.000)

Formicidae M24 3443.05 8 902.106 0 0.706 %URBAN=0.082=(0.011),=%=TURF=30.222=(0.000)

%GRASSLAND=30.143=(0.001),======Insectary=Strip 500 Araneae M16 3405.15 9 828.296 0 0.398 %GRASSLAND=:=PERENNIAL=0.144=(0.007) M26 ?407.12 8 830.236 1.487 0.189 PERENNIAL*1.568*(0.004)

%TURF=30.170=(0.019),=ALYSSUM=32.730=(0.000),======Carabidae M21 3325.91 9 669.82 0 0.787 %TURF=:=ALYSSUM=0.366=(0.000)

Opiliones M24 3107.6 8 231.22 0 0.816 %TURF=0.269=(0.013)

ALYSSUM=32.531=(0.000),=PERENNIAL=1.473=(0.002),===== Orthoptera M21 3175.7 9 369.402 0 0.955 %TURF=:=ALYSSUM=0.309=(0.000)

%FORAGE=0.218=(0.012),=%GRASSLAND=30.152= 1000 Araneae M22 3404.54 8 825.086 0 0.667 (0.000),=PERENNIAL=1.055=(0.023)

Formicidae M23* 3444.34 8 904.686 0 0.394 %TURF=30.283=(0.000) Formicidae M24 ?444.54 8 905.082 0.396 0.323 %TURF*?0.271*(0.000) Formicidae M21 ?444 9 905.992 1.759 0.164 %TURF*?0.195*(0.040)

%URBAN=30.446=(0.002),=%TURF=0.630=(0.000),======Opiliones M24 3104.01 8 224.024 0 0.993 ALYSSUM=2.500=(0.037),=PERENNIAL=3.708=(0.010)

Orthoptera M25 3179.15 9 376.3 0 0.473 %FRUITVEG=0.209=(0.000),=PERENNIAL=2.050=(0.000)

%FORAGE=0.286=(0.004),=%GRASSLAND=30.182= 1500 Araneae M22 3403.61 8 823.216 0 0.759 (0.000),=PERENNIAL=1.077=(0.014)

Orthoptera M24 3180.95 8 377.9 0 0.343 %URBAN=30.177=(0.036),=PERENNIAL=1.66=(0.015) %FRUITVEG*0.326*(0.001),*%FORAGE*0.218*(0.035),* M25 ?179.87 9 377.742 0.295 0.296 PERENNIAL*2.087*(0.000)

Table!2.5.!AICc table showing the top candidate pitfall activity density models for each year, landscape scale and species (bold), followed by models that were ranked with differences ≤ 2. LL = log-likelihood, K = number of parameters, Δi = AICc difference, Wi = AICc weight. *%FORAGE was dropped from this a priori model in 2012 because of a high VIF score. Factors within models that were significant are listed, along with their model estimates and p-values. 63 Year Response*Species Scale*(m) Model LL K AICc Δi Wi Signifcant*factor*estimates*(p>values) 2011 A.#vittatum 500 M7 '31.64 2 67.281 0 0.491 %URBAN40.0394(0.022)

1500 M4 '30.69 2 65.369 0 0.567 %FORAGE4'0.1564(0.013)

2012 A.#vittatum 500 M18 '33.88 3 73.759 0 0.739 %FRUITVEG4'0.0664(0.000) 500 M20 >35.06 3 76.129 2.37 0.226 %URBAN*0.040*(0.001)

1000 M18 '32.63 3 71.267 0 0.97 %FRUITVEG4'0.1974(0.000)

1500 M18 '33.81 3 73.623 0 0.934 %FRUITVEG4'0.3334(0.000)

Table!2.6. AICc table showing the top candidate parasitism models for each year and landscape scale (bold), followed by models that were ranked with differences ≤ 2. LL = log-likelihood, K = number of parameters, Δi = AICc difference, Wi = AICc weight. Factors within models that were significant!

64 FIGURES

Figure 2.1. 2011 sites were located in pumpkin growing regions in northern and southern Ohio (A). In 2012 site 3 was dropped, and sites 7, 8 and 15 were added (B). Sweet alyssum and perennial floral insectary site treatments were added in 2012 and both landscape-level and local habitat effects were measured. See Table 2.1 for site coordinates.

65 A. tristis D. undecimpunctata

2011 A. 2011 B. 100 Open 100 a Caged b 75 75

50 50 a c 25 ab 25 ab b d

Percent predation Percent (+/- SEM) 0 0 1 2 1 2 Experiment 1 Experiment 2 Experiment 1 Experiment 2

2012 C. 2012 a D. 100 100 a a a a a

75 75 66

50 a 50 b b b b a a a b 25 a a a a 25 b a a a a

Percent predation Percent (+/- SEM) 0 0 Control Alyssum Perennial Control Alyssum Perennial Control Alyssum Perennial Control Alyssum Perennial Experiment 1 Experiment 2 Experiment 1 Experiment 2

Figure 2.2. Percentage of eggs missing after 48 hrs from caged and open egg cards from experiments in 2011 for A) A. tristis, and B) D. undecimpunctatal. Also, percentage of eggs missing after 48 hrs from caged and open cards across experiment and site treatment in 2012 for C) A. tristis, and D) D. undecimpunctata. Letters indicate differences between cage, experimental period, and site treatments. In 2011, experiment 1 occurred on 8 July in southern Ohio and 14 July in northern Ohio, and experiment 2 occurred on 29 July in southern Ohio and 3 Aug in northern Ohio. In 2012 experiment 1 occurred 4 July in southern Ohio and 26 June in northern Ohio, and experiment 2 on 17 July in southern Ohio and 11 July in northern Ohio. Site treatments were two 60 m rows of an annual non-native plant, sweet alyssum, flanking 4 rows of pumpkins; a native perennial mixture planted in a 6 x 60 m area adjacent to the pumpkins; or a 6 x 60 m grassy control insectary adjacent to the pumpkins.

66 2011* 2012* A.## B.## C.## D.## A.#tris(s* D.#undecimpunctata* A.#tris(s* D.#undecimpunctata*

Formicidae* Opiliones* Araneae* Crice7dae* Entomobryidae* Gryllidae* Carabidae*

Figure 2.3. Proportions of predators observed interacting with sentinel eggs of A) A. tristis in 2011, B) D. undecimpunctata eggs in 2011, C) A. tristis in 2012, and D) D. undecimpunctata in 2012. Numbers indicate the total number of that taxa observed.

67 D.#undecimpunctata# A.#tris1s# 2011% 2012% 2012% A.% B.% C.% 24hr% 24hr% 24hr% %

18hr% 06hr% Formicidae

12hr% %

18hr% 06hr% Opiliones %

18hr% 06hr% Gryllidae

12hr% %

18hr% 06hr% Entomobryidae

12hr%

Figure'2.4.!Clock plot histograms illustrate activity patterns of predators across a 24 hr foraging period. Each colored wedge represents the proportion of a predator occurring during 1 hour periods across a 24 hour day. A taller wedge indicates higher predator activity. The most abundant predators that attacked eggs more than 10 times were Formicidae, Opiliones, Gryllidae, and Entomobryidae. They were observed feeding on D. undecimpunctata eggs in 2011 (A), and 2012 (B), and on A. tristis eggs in 2012 (C). A. tristis eggs in 2011 were only attacked by ants (n=7), and mice (n=4), and were not graphed this way.

68 Formicidae* Gryllidae* Carabidae* Opiliones* Araneae*

Figure 2.5. Proportions of predators caught in all pitfalls deployed in 2011 (A) and 2012 (B). Numbers indicate the total number collected. Entomobryidae was present in traps, but not counted because they were unknown predators until after reviewing the video data from the 2012 field season.

69

Figure 2.6. Percent parasitism of A. vittatum adults collected from pumpkin fields adjacent to site treatments in 2012. Though parasitism doubled with the addition of floral insectaries, the data was highly variable. Site treatments were two 60 m rows of an annual non-native plant, sweet alyssum, flanking 4 rows of pumpkins; a native perennial mixture planted in a 6 x 60 m area adjacent to the pumpkins; or a 6 x 60 m grassy control insectary adjacent to the pumpkins.

70

# A.# Carabidae Ac8vity#density#of#

# %TURF#within#500m# B.## Orthoptera Ac8vity#density#of#

%TURF#within#500m# # C.## Araneae Ac8vity#density#of#

%GRASSLAND#within#500m#

Figure'2.7. Interaction plots showing the activity densities of predators caught in pitfalls placed in floral insectaries in the presence of different landscape-scale habitat types within 500 m. A) Carabidae in the presence of mowed turfgrass habitats. B) Orthoptera in the presence of mowed turfgrass habitats. C) Araneae in the presence of grassland habitats. Site treatments were two 60 m rows of an annual non-native plant, sweet alyssum, flanking 4 rows of pumpkins (triangles); a native perennial mixture planted in a 6 x 60 m area adjacent to the pumpkins (squares); or a 6 x 60 m grassy control insectary adjacent to the pumpkins (circles).! 71

Chapter 3. Pollination of Cucurbita pepo in Ohio agroecosystems by

native and managed bees in the presence of a non-native annual floral

insectary (Lobularia maritima)

ABSTRACT

Pumpkin (Cucurbita pepo) production relies on insect-mediated pollination, provided by managed and unmanaged pollinators, and the goal of this study was to determine if local habitat management and landscape composition affected the pollination services provided to pumpkins. I measured visitation frequency and duration of Apis mellifera, Bombus spp., and Peponapis pruinosa in male and female flowers of pumpkins in 2011-2012, and pollen deposition across the pollination window (0600-1200 hr) in

2012. I also tested the Intermediate Landscape-Complexity Hypothesis by determining the effects of surrounding landscape composition and the addition of sweet alyssum

(Lobularia maritima) floral insectaries on the relative abundance of pollinators. In 2011

A. mellifera was more abundant in flowers than other bees, and in 2012 Bombus spp. was most abundant. A. mellifera spent more time in flowers, and preferred female flowers. In one year, Bombus spp. spent more time in female flowers, and P. pruinosa visited male flowers more frequently than male flowers. In both years, Bombus spp. had a significantly higher visit frequency after 0700 hr, and both Bombus spp. and P. pruinosa spent less time in flowers after 0800 hr. Pollen loads on female flowers indicated the

72 majority of pollen deposited across the 6 hr window was transferred between 0600-0800 hr, which is when all three bee species foraged with equal frequency and similar visit duration. Local additions of sweet alyssum floral insectaries did not have an effect on the foraging activity of bees. However, visits to pumpkins by A. mellifera showed that increased percentage of forest habitats supported higher visit frequencies to pumpkins with alyssum insectaries.

INTRODUCTION

Pollinators are important for agriculture worldwide, including pumpkins and other cucurbits (Michelbacher et al., 1964; Klein et al., 2007). However, the domesticated honey bee (Apis mellifera) is often solely depended on for crop pollination in the United

States. In The United States, pollinators account for $40 billion per year in fruit, fiber, vegetable and legume crops (Pimentel et al., 1997), with an estimated $1.6-$14.8 billion of that attributed to A. mellifera alone (Southwick and Southwick, 1992; Morse and

Calderone, 2000; Losey and Vaughan, 2006). Recently, A. mellifera has suffered major population crashes attributed to Colony Collapse Disorder (Stokstad, 2007), that caused the prices of hive rentals to more than double from $54 to $136 per colony in 2004 and

2006, respectively (Sumner and Boriss, 2006; vanEngelsdorp et al., 2008). In light of this, the potential of native wild bees as crop pollinators has become a greater research focus. Reviews have found that 45 world crops are highly reliant on pollinators (Klein et al., 2007), fruit set of many of these crops is higher when pollinated by wild native pollinators (Garibaldi et al., 2013), and wild bee abundance and richness is greater if more high-quality habitats surround agricultural fields (Kennedy et al., 2013). However, as agricultural intensification reduces the amount of favorable habitat available, wild bee 73 communities have not been found to compensate for population losses at the landscape scale (Winfree and Kremen, 2009), highlighting the importance of maintaining high bee richness for successful pollination of the world’s crops.

Habitat management using annual floral insectaries seeks to mitigate some of the negative impacts of agricultural intensification by providing alternative food and shelter resources for beneficial arthropods, such as predators, parasitoids, and pollinators (Landis et al., 2000; Zehnder et al., 2007). When habitat management practices are incorporated into a farmscape, larger scale landscape composition and heterogeneity can influence the pool of beneficial species supplied to the floral insectary, and the arthropod mediated ecosystem services they are able to support (Isaacs et al., 2009; Batáry et al., 2011;

Concepción et al., 2012; Rodriguez-Saona et al., 2012). Further, Tscharntke et al. (2012) introduced the Intermediate Landscape Complexity Hypothesis, which states that in highly heterogeneous landscapes (> 20 % non-crop habitats), stable populations of beneficial organisms already exist which limited the effect of local habitat management; and extremely simplified landscapes (< 1 % non-crop habitats) do not have enough supporting habitats for a substantial species pool to take advantage of local habitat amendments. As such, local habitat management is theoretically most useful to enhance arthropod-mediated ecosystem services within intermediately-complex landscapes.

Monoecious cucurbits are highly dependent on pollinators that transfer pollen grains from the stamens of male flowers to the stigmas of female flowers (Wien, 1997).

Cucurbits also provide a unique study system because they are typically pollinated by social and solitary bees that vary in their domestication and commercial availability.

Fully domesticated A. mellifera are utilized through rental hives and feral populations.

74 Wild Bombus spp. are a very common visitor of cucurbit flowers, and a niche commercial industry is developing around domestication of a few species for use in multiple crops

(Stanghellini et al., 1998; Stubbs and Drummond, 2001; Thomson and Goodell, 2001;

Fuchs and Müller, 2004; Artz and Nault, 2011). Finally, Peponapis pruinosa is a solitary soil-nesting bee native to the Midwest that specializes on cucurbits, and have not been domesticated (Hurd, Jr et al., 1971; Hurd et al., 1974).

Large seed set, successful maturation, and fruit weight are highly correlated with the number of pollinator visits to cucurbit flowers (Brewer, 1974; Jaycox et al., 1975;

Stanghellini et al., 1998; Garibaldi et al., 2013), and the subsequent amount of pollen transferred (Canto-Aguilar and Parra-Tabla, 2000; Winfree et al., 2007a; Goodell, 2008;

Graças Vidal et al., 2010; Artz and Nault, 2011) to female flowers. Because of this close relationship, research on pollinators of cucurbits has often focused on the abundance of these pollinators found inside flowers, and the duration of their visitation (Tepedino,

1981; Cane et al., 2000; Fuchs and Müller, 2004; Shuler et al., 2005; Julier and Roulston,

2009; Nicodemo et al., 2009; Barber et al., 2011; Artz et al., 2011). Some researchers found A. mellifera to be integral to achieving marketable fruit production (Fuchs and

Müller, 2004; Walters and Taylor, 2006; Nicodemo et al., 2009; Graças Vidal et al.,

2010; Artz et al., 2011), while others found better performance from Bombus spp.

(Stanghellini et al., 1998; Artz and Nault, 2011), P. pruinosa (Michelbacher et al., 1964;

Hurd et al., 1974; Canto-Aguilar and Parra-Tabla, 2000; Shuler et al., 2005; Goodell,

2008; Julier and Roulston, 2009; Delaplane and Mayer, 2010; Cane et al., 2011), or a mixture of native bees (Hoehn et al., 2008).

75 Kremen et al. (2002) found that in multiple crops, wild pollinators provided enough visits to deposit the required amount of pollen for a marketable crop. However, their effect was greatly depressed in intensified agricultural areas, which reduced alternative forage and nesting habitats that support a higher diversity and abundance of native bees. Other research has also shown that wild bees require semi-natural habitat at the landscape scale, but responses to specific habitat types vary (Kremen and Ostfeld,

2005; Steffan-Dewenter et al., 2005; Klein et al., 2007; Kennedy et al., 2013; Scheper et al., 2013). Other researchers have begun focusing on the effects of localized management of floral resources, and have found that these wild stands of annual and perennial flowers can potentially be managed selectively to target particular guilds of pollinators (Kells et al., 2001; Pywell et al., 2006).

Floral insectaries (Colley and Luna, 2000; Hogg et al., 2011), are commonly referred to as floral strips (Gillespie et al., 2011), flower strips (Bianchi and Wäckers,

2008; Haenke et al., 2009), beetle banks (MacLeod et al., 2004), grass banks (Thomas and Marshall, 1999), conservation headlands (Cole et al., 2007), refuges (Schellhorn et al., 2008), and floral resources (Lee et al., 2006). The flowering plants within floral insectaries have been shown to provide carbohydrate and protein sources in the form of pollen and nectar that have been shown to increase fecundity and longevity of natural enemies and pollinators (Shahjahan, 1968; Baggen and Gurr, 1998; Johanowicz and

Mitchell, 2000; Pontin et al., 2005; Lee et al., 2006; Pywell et al., 2006; Tuell et al.,

2008).

Annual plants are commonly used as floral insectaries because they are inexpensive to establish, moveable from year-to-year, fast-growing, and some make

76 effective cover crops or are economically important themselves (Bugg and Waddington,

1994; Prasifka et al., 1999; Dufour, 2000; Landis et al., 2000). Some of the commonly studied annual plants include candytuft (Iberis umbellate), lacey phacelia (Phacelia tanacetifolia), buckwheat (Fagopyrum esculentum), sweet alyssum (Lobularia maritima), coriander (Coriandrum sativum), and chrysanthemum (Chrysanthemum coronarium)

(MacLeod, 1999; Wäckers, 2004; Rebek et al., 2005; Lavandero et al., 2006; Fiedler and

Landis, 2007b). However some annuals do not thrive in an agricultural setting without careful management, others can re-seed prolifically and become a nuisance, and some may have a limited bloom period (Landis et al., 2000; Fiedler and Landis, 2007a).

Research on annual floral insectaries in agriculture has been largely investigated for their effects on natural enemies, and biocontrol services occurring within the crop

(Platt et al., 1999; Lee and Heimpel, 2005; Woltz et al., 2012; Gontijo et al., 2013), but fewer have investigated whether an annual floral insectary established for enhancing natural enemies has similar effects on pollinators and pollination services. Pontin et al.

(2005) performed one such study by planting five treatments of single species

(buckwheat, Phacelia, or crop control only) and multiple species strips (buckwheat +

Phacelia, and buckwheat + Phacelia + commercial seed mix) in the margins of broccoli and lucerne crops. They found that Bombus and A. mellifera were significantly more abundant in floral strips containing Phacelia flowers, while syrphid flies showed no preference for floral treatment. However, within mixed flower strips A. mellifera,

Bombus, and syprhid flies each focused on particular flower species, despite the presence of many, including sweet alyssum (Maritima lobularia), Bishop's flower (Ammi majus), coriander (Coriandrum sativum), dill (Anethum graveolens), and fennel (Foeniculum

77 vulgare). When lucerne crops where blooming adjacent to the floral strips, the abundance, total visit duration, visitation rates declined in the floral strips for some pollinators. The authors concluded that attraction of bees to the floral strips might improve longevity and fecundity of bee populations, which would in turn improve pollination services within the crop.

Therefore, this study aimed to test three hypotheses: 1) when annual floral insectaries are planted adjacent to a crop for use enhancing biocontrol (Platt et al., 1999), the communities of bees provided pollination services are affected (Pontin et al., 2005),

2) natural landscapes act as a source for bees to move into the floral insectaries (Isaacs et al., 2009; Pywell et al., 2006; Rodriguez-Saona et al., 2012; Woltz et al., 2012), and 3) an intermediate landscape-complexity maximizes the effects of a floral insectary addition on bee communities and pollination services (Isaacs et al., 2009; Batáry et al., 2011;

Concepción et al., 2012; Tscharntke et al., 2012). To address these hypotheses, I examined 3 research objectives. 1) Measure the visitation frequency and duration of A. mellifera, Bombus spp., and P. pruinosa to male and female pumpkin flowers across the pollination window (0600-1200 hr), 2) Measure pollen deposition across the pollination window (0600-1200 hr), and 3) Determine whether the surrounding landscape composition or local sweet alyssum insectaries affected relative abundance of pollinators visiting pumpkin flowers. My predictions were that 1) A. mellifera visit frequency and duration would be higher in female flowers, but that flower sex would not affect visit frequency and duration other bee species (Tepedino, 1981; Artz and Nault, 2011), 2) P. pruinosa and Bombus spp. would exhibit greater visit frequency and have longer visit durations earlier in the morning, resulting in more pollen grains deposited earlier in the

78 morning, and A. mellifera would exhibit greater visit frequency and have longer visit durations later in the pollination window (Hurd et al., 1974; Canto-Aguilar and Parra-

Tabla, 2000; Goodell, 2008; Nicodemo et al., 2009; Graças Vidal et al., 2010; Artz et al.,

2011; Artz and Nault, 2011), 3) local habitat management in the form of sweet alyssum floral insectaries would increase the visit frequency of A. mellifera and Bombus spp., by acting as alternative nectar sources before and after pumpkin bloom (Pontin et al., 2005;

Pywell et al., 2006; Tuell et al., 2008), 4) higher percentages of semi-natural and less- disturbed habitat surrounding my sites would result in higher visit frequency of A. mellifera and Bombus spp. visiting pumpkin flowers (Kremen et al., 2002; Steffan-

Dewenter et al., 2002; Shuler et al., 2005; Julier and Roulston, 2009), but decrease visit frequency of P. pruinosa as a result of reduced cucurbit habitat, and 5) in the presence of intermediate habitat-complexity (Tscharntke et al., 2012), visit frequencies of A. mellifera and Bombus spp. to pumpkin flowers at sites with local floral insectary additions (in

2012) would be significantly greater than at sites with a grassy control insectary.

METHODS Study sites

The visit frequency and visit duration of pollinating arthropods were measured on pumpkin farms throughout Ohio in 2011-2012. Ohio is the 2nd largest pumpkin-producing state, and two regions were selected that represent major production areas (USDA-NASS,

2013). In 2011, 12 sites were included; six in Wayne, Stark, and Medina counties in northern Ohio, and six in Jackson, Pike, Highland, and Warren counties in southern Ohio.

In 2012, one site was dropped (site 3) and two sites were added (sites 8 and 15) in Wayne county in northern Ohio, and Ross county in southern Ohio (Figure 3.1). In 2011 and 79 2012 a plot consisting of four rows of jack-o-lantern pumpkins (var. “Gladiator”) was established between 10 June and 8 July within each site. No insecticides were applied to these sites. Each plot was divided into 4 sub-plots where all data were collected.

Habitat management

A habitat management strategy using a floral insectary was examined in 2012.

Three sites in the northern and southern regions were randomly assigned to one of two treatments: 1) control: pumpkin plots planted adjacent to a 6 x 60 m grassy area, mowed approximately once per month; and 2) alyssum: pumpkin plots planted between two 60 m rows of the exotic annual, L. maritima (Table 3.1). Each individual farm site was established with one pumpkin plot and one floral insectary treatment.

Establishing non-native annual floral insectaries

In 2012, I planted two rows of L. maritima as the annual floral insectary treatment adjacent to pumpkins at six sites in northern and southern Ohio. For this treatment, one row of L. maritima was established on either side of the pumpkin plot. The sweet alyssum was started from seed in 72-cell plug trays in a greenhouse in early May and fertilized twice per week for two weeks. The plants were hardened off outside for an additional two weeks before being transplanted with a pottiputki planter (Stand ‘n Plant,

Saltsburg, PA) into the alyssum treatment plots between 7-14 June 2012. Plants were watered and Preen Garden Weed Preventer (Lebanon Seaboard Corp., Lebanon, PA) was applied. The transplants were watered via drip irrigation and hand containers (~190 L) twice per week in the field through July.

80 Measuring pollinator activity with video cameras

A modified 4-channel security camera system (Q-see, model no. QSC26404,

Anaheim, CA) was used to monitor pollinator activity across the entire 6 hr pollination window on two female pumpkin flowers and two male pumpkin flowers within each plot

(Grieshop et al., 2012). Cameras recorded pollinator activity between 0600 hr and 1200 hr. In both years, this was conducted once at each site during peak bloom in late-July through August (Table 3.1). However, sites 5, 10, and 13 in 2011 were not sampled because a wet spring prevented an expedient planting, and the peak bloom period occurred in September, which was much too late to accurately represent the pollinator community that focuses on the pumpkin flower resource pulse. Additionally, site 3 and

11 in 2012 were omitted because of heavy weed pressures that drastically altered pumpkin bloom availability. The video footage was transferred to portable hard-drives and stored until viewed on a computer at 8-16 times normal speed. When an organism was observed crossing the plane made by the open corolla, the time of arrival and departure was recorded as a measure of the amount of time spent inside the flower. All pollinators were identified to the lowest taxonomic level possible with video recorded at

16 frames per second, and a playback pixel resolution of 352 x 240 (aspect ratio

~1.222:1).

Measuring pollen deposition throughout the pollination window

In 2012, I conducted pollen counts as a direct measure of pollination service. One day prior to the collection of video surveillance data, I identified 24 mature female flower buds per plot. All buds that were at least 5 cm in length and turning deep yellow were

81 fitted with a mesh paint strainer bag (Reaves & Co. Durham, NC) as a pollinator excluder, and marked with a step-in poly post (Gempler’s, Madison, WI).

I examined the amount of pollen deposited during three segments of the pollination window (T1 = 0600-0800, T2 = 0800-1000, and T3 = 1000-1200 hr), as well as across the entire period (T4 = 0600-1200 hr). Bags were left on flowers until the beginning of the treatment time upon which they were removed and pollinators were allowed to access flowers. Six flowers were assigned to each treatment. Since pumpkin flowers only open once for one morning, if six flowers could not be found for each treatment on the morning of the video experiment I returned within seven days of the first attempt, and in comparable weather conditions to attempt to collect additional replicates.

However, some replicates were still missed. For T1, sites 1 and 8 missed one replicate, and sites 2 and 5 missed all replicates. For T2, site 2 missed four replicates, site 4 missed one replicate, site 5 missed all replicates, and site 6 missed three replicates. For T3, site 6 missed two replicates, sites 2 and 5 missed all replicates, and site 8 missed one replicate.

For T4, sites 5 and 9 missed one replicate.

Pollen filtering process

I used an Aeropress espresso maker and the stock filters (Aerobie, Inc., Palo Alto,

CA) to sieve pollen grains from collected stigmas. The filters were modified with a 1 x 1 cm grid pattern prior to the experiment. Once stigmas were collected, I placed them in a

120 mL urine specimen cup with ~44 mL of a dish soap and water solution (4 drops of dish soap per 2 L of water) and shook them vigorously for 20 seconds. I decanted the solution into a separate cup and washed the stigma a second time. The stigma was then removed from the solution and washed with 70% ethanol from a 500 mL wash bottle. 82 The pollen solution was then poured into the Aeropress, and expunged. The inside of the

Aeropress was washed with ethanol so that any pollen that was sticking to the sides was collected on the filter. After each stigma was processed the filters were allowed to dry, packaged individually in labeled petri dishes, and frozen until they were counted under a microscope. I counted pollen grains from 6 full grid squares, and 6 partial grid squares to make an extrapolated estimation of the total pollen load on each filter.

Collecting landscape data

I obtained aerial image mosaics of each county that contained a research site from the year 2010 (OGRIP, 2010) and uploaded them into ArcMap (version 9.3; ESRI, 2011) and QGIS (version 1.8.0; Quantum GIS Development Team, 2012) to digitize all land cover elements. I determined the area of each distinct landscape feature within a 1500,

1000, and 500 m buffer around the geographic center of each plot and ground verified them with a classification system including 22 habitat types. The 22 fine-grain cover types were combined into 7 coarse-grain habitat categories, and the percentages of each habitat type were aggregated as predictor variables within each landscape buffer for analysis (%FIELD = annual field crops; %GRASSLAND = perennial grassland, fallow, and pastureland; %FORAGE = perennial alfalfa/oats; %FRUITVEG = mixed fruit and vegetable cropland; %FOREST = contiguous woodlands and hedgerows; %URBAN = impervious surfaces and buildings; %TURF = mowed turfgrass). Landscapes ranged in non-crop habitat from 2-91%. This allowed me to test the effects of landscape composition on the study system.

83 Statistical analyses Assessment of pollination services Pollinator visitation frequency and visit duration

The amount of flower visitation frequency and duration of A. mellifera, Bombus spp., and P. pruinosa in both years was modeled using a generalized linear mixed model.

I used the glmmadmb function (glmmADMB package in R (version 3.0.0; R Development

Core Team, 2013)) with a Laplace maximum likelihood approximation that allowed for specification of a logistic link function, a negative binomial error distribution for visit frequency data, and a gamma distribution or visit duration data (Bolker, 2007). I followed the modeling protocol described by Zuur et al. (2010) by exploring the data for outliers in the response variable, and the continuous covariates using Cleveland dotplots. I also assessed collinearity between continuous variables with variance inflation factors (VIF) with the corvif function (Highland Statistics Ltd. Newburgh, UK), and used the mixed model parameter estimates to determine whether a particular parameter, or combination of parameters, had a positive, negative, or neutral effect on the response variables.

I used multiple comparisons procedures to contrast the fixed covariates of flower sex (categorical: male or female), hour of the pollination window (categorical: 6, 7, 8, 9,

10, or 11), bee (categorical: A. mellifera, Bombus spp., or P. pruinosa), and habitat management treatment (categorical: control, or alyssum (2012 only)) to predict visit frequency and visit duration between and within species, using the general linear hypothesis test (glht) function in the R multcomp package to separate means.

Random covariate, ai, set up a correlation structure for each farm, and significance was

84 determined at the α = 0.05 level. The multiple comparisons procedure adjusted z-scores and P-values with a single-step method.

Pollen deposition

I modeled the average number pollen grains collected from stigmas of female flowers from 4 different times throughout the pollination window. Pollen counts were predicted by the contrasted categorical fixed covariates of flower sex, time of pollinator exposure (categorical: T1 = stigmas exposed to pollinators between 0600-0800, T2 =

0800-1000, T3 = 1000-1200, and T4 = 0600-1200 hr) and habitat management treatment separately, using the glmmadmb mixed model function with a negative binomial distribution, and the general linear hypothesis test (glht) function from the multcomp package in R to separate means. Random covariate, ai, set up a correlation structure for each farm, and significance was determined at the α = 0.05 level.

Landscape models

To find the optimal set of continuous and fixed covariates to predict pollination service at each landscape scale (500, 1000, and 1500 m), I used an information-theory approach (Bolker et al., 2009; McElduff et al., 2010; Zuur et al., 2010) to construct 13 a priori models in 2011 and 13 models in 2012 (Table 3.2). The following land cover categories served as predictors in these models: percentages of field crops, grasslands, forage crops, fruit and vegetable crops, forests, impervious urban surfaces, and mowed turfgrass. The models were compared using a corrected Akaike’s Information Criterion

(AICc: for small samples sizes). Each model was given an AIC score, and the model with the highest score and weight was considered the most parsimonious explanation of the

85 data. The AICc weight is a probability proportion between 0 and 1 that describes the likelihood that a model will be ranked in the same position over infinite iterations.

Models were considered competing candidate models if the AICc score differences were

≤ 2. Additionally, if the null model was ranked as a top or competing model, that was interpreted to mean that none of the 13 a priori models were any better at explaining the data than random chance. Interactions between local and landscape predictors were interpreted using ggplot2 in R.

RESULTS

In 2011 a total of 1294 A. mellifera, 573 Bombus spp., and 867 P. pruinosa were observed in flowers. In 2012, observations of A. mellifera and P. pruinosa decreased to

542 and 728, respectively, while observations Bombus spp. increased to 5069. A. mellifera was significantly more abundant in flowers in 2011 compared to Bombus spp (z

= -6.23, P < 0.001) and P. pruinosa (z = -5.85, P < 0.001). In 2012, Bombus spp. were the most abundance bees in flowers compared to A. mellifera (z = 7.42, P < 0.001) and P. pruinosa (z = -6.31, P < 0.001).

Pollinator activity in male and female flowers

In 2011, A. mellifera visitation frequency was significantly higher in female flowers compared to male flowers (z = -3.26, P = 0.001), while the visit frequency of

Bombus spp. and P. pruinosa did not vary among male and female flowers. A. mellifera also spent significantly more minutes in female flowers than in male flowers (z = -8.10, P

< 0.001). Between other bee species, A. mellifera had a higher visitation frequency than

Bombus spp. in male flowers (z = -3.09, P = 0.005) and longer visit duration on female 86 flowers than both Bombus spp. (z = -5.28, P < 0.001) and P. pruinosa (z = -4.56, P <

0.001) (Figure 3.2 A and C).

In 2012, A. mellifera visitation was again significantly higher in female flowers (z

= -5.42, P < 0.001), and individuals spent more time within them compared to male flowers (z = -4.28, P < 0.001). Bombus spp. visitation frequency did not vary by flower sex, but they spent significantly more time in female flowers (z = -3.24, P = 0.001). P. pruinosa visitation was significantly higher in male flowers (z = 2.48, P = 0.013), and individuals spent equal time in male and female flowers. Among species, Bombus spp. visitation frequency was more than double that of A. mellifera (z = 8.65, P < 0.001 in male and z = 4.48, P < 0.001 in female flowers), and P. pruinosa (z = -4.14, P < 0.001 in male and z = -5.66, P < 0.001 in female flowers). A. mellifera showed higher visit frequency in female flowers compared to P. pruinosa in female flowers (z = -3.16, P =

0.004), but exhibited a significantly lower frequency than P. pruinosa in male flowers (z

= 2.59, P =0.024). A. mellifera showed higher visit duration in female flowers compared to Bombus spp. (z = 3.11, P = 0.005) and P. pruinosa (z = -4.44, P < 0.001) in female flowers (Figure 3.2 B an D).

Pollinator activity throughout the pollination window

In 2011, A. mellifera visitation frequency peaked at 0900 hr, which was significantly greater than visitation frequency at 0600 hr (z = 3.07, P = 0.024), and individual A. mellifera spent the same number of minutes inside flowers throughout the pollination window. Bombus spp. visitation frequency was significantly greater after

0600 hr (z > 3.44, P < 0.007 when 0600 was compared to all other hours), and individuals spent significantly fewer minutes inside flowers after 0800 hr (z < -3.42, P < 87 0.008 when 0600 hr was compared to hours after 0700). P. pruinosa visitation frequency did not vary by time of day, but individuals spent significantly more time in the flowers at 0600 hr than at times between 0800-1000 hr (z < -3.84, P < 0.001 when 0600 was compared to 0800-1000). Among species, A. mellifera had a higher visitation frequency than P. peponapis (z = -2.44, P = 0.038) at 0900 hr, and a significantly higher visit frequency than Bombus spp. (z = -3.15, P = 0.005) at 1100 hr. All bee taxa spent equal time in flowers early in the morning between 0600-0800 hr. A. mellifera spent significantly more time in flowers than Bombus spp. at 0800 hr (z = -2.59, P = 0.024),

0900 hr (z = -4.84, P < 0.001), and 1000 hr (z = -3.05, P = 0.006). A. mellifera spent significantly more time in flowers than P. pruinosa at 0800 hr (z = -3.13, P = 0.005),

0900 (z = -4.97, P < 0.001), and 1000 hr (z = -3.73, P < 0.001). At 1100 hr, P. pruinosa increased visit durations, but A. mellifera was still spending significantly more time in flowers than Bombus spp. (z = -5.26, P < 0.001) (Figure 3.3 A and C).

In 2012, A. mellifera visit frequency and duration did not vary across the pollination window. Bombus spp. visit frequency was significantly greater after 0700 hr

(z > 4.72, P < 0.001 when 0600 was compared to all other hours), though it peaked at

0800 hr. Individuals spent equal time in the flowers throughout the morning. P. pruinosa visitation frequency did not vary by time of day, but individuals spent significantly more minutes inside flowers before 0800 hr (z = -3.53, P < 0.01 when 0600 hr was compared to 0800 hr), and after 1100 hr (z = 3.21, P < 0.01 when 1100 hr was compared to 0800 hr, and z = 3.31, P = 0.012 when 1100 hr was compared to 1000 hr). Among species,

Bombus spp. was a much more frequent visitor, with a significantly higher visitation frequency than A. mellifera at 0700 hr (z = 2.76, P = 0.015), 0800 hr (z = 4.57, P <

88 0.001), 0900 hr (z = 4.10, P < 0.001), 1000 hr (z = 4.74, P < 0.001), and 1100 hr (z =

4.94, P < 0.001). Bombus spp. exhibited a higher visit frequency than P. pruinosa at

0800 hr (z = -3.18, P =0.004), 0900 hr (z = -2.98, P = 0.007), 1000 hr (z = -3.89, P <

0.001), and 1100 hr (z = -3.39, P = 0.002). However, A. mellifera spent more time inside flowers, with a significantly longer duration than P. pruinosa at 0700 hr (z = -4.24, P <

0.001), 0800 hr (z = -5.23, P < 0.001), 0900 hr (z = -2.50, P = 031), and 1000 hr (z = -

3.35, P = 0.003). A. mellifera also spent more time in flowers than Bombus spp. at 0800

(z = -2.55, P =0.027), and 0900 hr (z = -2.59, P = 0.024) (Figure 3.3 B and D).

Influence of floral insectaries on pollinators

In 2012, the floral insectaries had no effect on the visit frequency or duration when each taxa was compared between treatments. For visit frequency, Bombus spp. were more frequent visitors than A. mellifera (z = 7.72, P < 0.001) and P. pruinosa (z = -

4.49, P < 0.001) to pumpkins adjacent to the grassy control insectaries (Figure 3.4 A). At pumpkins adjacent to the alyssum insectaries, A. mellifera had a higher visit frequency than P. pruinosa (z = -3.32, P =0.003), but Bombus spp. showed a higher frequency than both A. mellifera (z = 3.59, P < 0.001) and P. pruinosa (z = 5.24, P < 0.001). For duration of visits, A. mellifera spent more time in flowers than Bombus spp. (z = -2.69, P

= 0.018), and P. pruinosa (z = -5.14, P < 0.001) in pumpkin plots planted adjacent to the grassy control insectary, and the alyssum insectary (z = -2.99, P = 0.007 compared to

Bombus spp. and z = -2.85, P = 0.012 compared to P. pruinosa) (Figure 3.4 B).

89 Pollen deposition throughout the pollination window

In 2012, the most pollen grains were found in flowers between 0600-0800 hr, which was numerically greater than pollen found between 0800-1000 hr, and significantly greater that pollen transferred between 1000-1200 hr (z = -3.98, P < 0.001).

Pollen transferred after a full morning (0600-1200 hr) was not significantly different from the 0600-0800 hr pollination period (z = 0.74, P = 0.871), suggesting that most of the pollination occurred before 0800 hr (Figure 3.5 A).

Influence of floral insectaries on pollen deposition

In 2012, there were numerically more pollen grains found in female flowers in pumpkin plots adjacent to the grassy control insectary (22407 ± 3673 SEM), than the alyssum insectary (17855 ± 2979 SEM). However, this was not statistically significant (z

= -0.87, P = 0.384) (Figure 3.5 B).

Landscape model comparisons 2011 models –Visitation frequency of bees

In 2011, I found that A. mellifera was positively associated with, and best predicted by the percentage of forest areas at the smallest scale of 500 m (z = 2.16, P =

0.031). A. mellifera also had a negative association with mowed turfgrass areas in model

M8 (z = -2.08, P =0.038) and model M11 (z = -2.25, P = 0.024). Bombus spp. was positively associated with, and best predicted by percentage of urban impervious surfaces in model M7 (z = 3.32, P < 0.001) and M11 (z = 3.20, P = 0.001) within the larger scale of 1500 m. P. pruinosa was negatively associated with, and best predicted by the percentage of fruit and vegetable crop areas in model M12 (z = -5.63, P < 0.001), and

90 model M5 (z = -4.08, P < 0.001). A. mellifera model sets at 1000 and 1500 m, Bombus spp. model sets at 500 and 1000 m, and P. pruinosa model sets at 1000 and 1500 m ranked the Intercept Only model M1, which indicated that landscapes were not important predictors within those radii (Table 3.3).

2012 models – Visitation frequency of bees

In 2012, Bombus spp. model sets at all landscape scales ranked the Intercept Only model M1, which indicated that landscapes were not important predictors for this species within those radii. However, A. mellifera within 500 m of my sites was negatively associated with and best predicted by the percentage of fruit and vegetable crop areas in model M18 (z = -2.79, P = 0.005), and model M25 (z = -2.71, P = 0.007), and by the interaction of %FOREST and the ALYSSUM insectary variables in model M19 (z =

2.43, P = 0.015). Within 1000 m, A. mellifera was best predicted by the interaction between the percentage of forested areas and the alyssum insectaries (z = -4.22, P =

0.019) in model M19, but also by fruit and vegetable crop areas in model M25 (z = -2.24,

P = 0.025). Within a 1500 m radius of my sites, A. mellifera was negatively associated with and best predicted by the percentage of fruit and vegetable crops (z = -2.12, P =

0.034), field crops (z = 3.12, P = 0.002), forage crops (z = -2.91, P = 0.004) in model

M25, and again with the interaction between the percentage of forested areas and the alyssum insectaries (z = 3.06, P = 0.002) in model M19. Interaction plots (Figure 3.6) indicated that A. mellifera visit frequency was higher in the pumpkins planted adjacent to the grassy control insectary with low percentage of forest area, and decreased to a lower level than A. mellifera observed in pumpkins planted adjacent to the alyssum insectaries with an increase in forest area. At the 500 m scale, A. mellifera frequency in pumpkins 91 adjacent to alyssum increased with an increase in forest area, while A. mellifera frequency in pumpkins adjacent to the grassy control insectaries decreased. But, at the

1000 and 1500 m scales, A. mellifera frequency in pumpkins adjacent to alyssum insectaries slightly decreased (Table 3.3).

In 2012, P. pruinosa within 500 m was negatively associated with and best predicted by the percentage of fruit and vegetable crops (z = -3.35, P = 0.001) in model

M18, and positively associated with the area of urban impervious surfaces (z = 3.04, P =

0.002) in model M20. Model M18 also showed that pumpkins planted adjacent to the alyssum insectaries had significantly fewer visits from P. pruinosa than those planted next to the grassy control (z = -3.13, P = 0.002) in the presence of the percentage of fruit and vegetable crops, though the interaction term was not significant. Within larger radii of my sites, the percentage of urban impervious surfaces and mowed turfgrass became more important, as evidenced by the higher AICc weight given to model M24 within

1000 (0.98) 1500 m (0.93) of my sites. Within 1000 m, P. pruinosa was positively associated with urban areas (z = 5.45, P < 0.001) and negatively associated with mowed turfgrass (z = -3.74, P < 0.001) in model M24. Within 1500 m, this was also observed

(%URBAN z = 5.19, P < 0.001 and %TURF z = -3.20, P = 0.001). Additionally, model

M24 showed that pumpkins planted adjacent to the alyssum insectaries had significantly fewer visits from P. pruinosa than those planted next to the grassy control within 1000 m

(z = -2.97, P = 0.003) and 1500 m (z = -2.86, P = 0.004) in the presence of urban impervious surfaces and mowed turfgrass (Table 3.3).

92 DISCUSSION

This study examined the foraging preferences of A. mellifera, Bombus spp. and P. pruinosa towards male and female pumpkin flowers and across a 6 hr pollination window, and the effects of local habitat management and landscape composition on the frequency of flower visits. Unlike Shuler et al. (2005) and Goodell (2008), whose sample dates and observation times differed from mine, I found that P. pruinosa was the least abundant pollinator observed in my pumpkin fields in 2011 and 2012. I also found that A. mellifera spent more time in female flowers than male flowers, and more time in female flowers compared to Bombus spp. and P. pruinosa in both years. However, between years

2011 and 2012 the visit frequency relationships varied. In particular, Bombus spp. visit frequencies increased to more than double the visit frequencies of other bees in 2012, and this elevated population spent more time in female flowers.

In 2011, bees frequented flowers consistently throughout the pollination window, though A. mellifera visited more frequently overall. In 2011, visit duration of Bombus spp. and P. pruinosa decreased to a point significantly less than the amount of time A. mellifera between 0800-1000 hr, but P. pruinosa increased its visit duration at 1100 hr. In

2012, a similar pattern emerged but pollinator visit durations converged later in the morning, and A. mellifera only spent significantly greater time than Bombus spp. between

0800-0900 hr, and from P. pruinosa between 0700-1000 hr. Pollen loads on female flowers indicated the majority of pollen deposited across the 6 hr window was transferred between 0600-0800 hr, which is when all three bee species foraged with equal frequency and similar visit duration. Local habitat management alone did not have an effect on the foraging activity of bees. However, in the presence of certain landscapes, visitation

93 frequency was shown to vary between pumpkins planted adjacent to the alyssum and grassy control insectaries. Visits to pumpkins by A. mellifera showed that increased percentage of forest habitats supported higher visit frequencies to pumpkins with an alyssum insectary.

Bee activity in male and female flowers

I found my first prediction was that A. mellifera abundance and visit duration would be higher in female flowers, but that flower sex would not have an effect on other bee species. This was supported for A. mellifera, in both years, and for Bombus and P. pruinosa only in 2011. A. mellifera did have a higher visit frequency and visit duration in female flowers in both years, and in 2011 Bombus and P. pruinosa did not have a strong preference for either flower sex. However, in 2012 P. pruinosa were more abundant in male flowers, and Bombus spent significantly more time in female flowers. Tepedino et al. (1981) also found that A. mellifera spends significantly more time inside flowers than

P. pruinosa, and that P. pruinosa visited more male flowers than female flowers in Utah.

Artz & Nault (2011) found that, in New York, A. mellifera was more likely to visit female flowers of C. pepo and to spend more time in them, and in a separate study, Artz et al. (2011) found that supplemental A. mellifera hives increased their abundance in C. pepo flowers, but decreased the abundance of P. pruinosa. Female flowers produce significantly more nectar than male flowers (Tepedino, 1981; Ashworth and Galetto,

2001; Artz et al., 2011). Part of the reason A. mellifera and Bombus spp. may prefer female flowers is because it is their prime directive to forage nectar to improve over- wintering capabilities and rapidly expand their brood in the summer months (Heinrich,

2004; Seeley, 2009). 94 Bee activity across the pollination window

My second prediction was that P. pruinosa and Bombus spp. would exhibit greater visit frequency and have longer visit durations earlier in the morning, resulting in more pollen grains deposited earlier in the morning, and A. mellifera would exhibit greater visit frequency and have longer visit durations later in the morning. This was partially supported. All taxa experienced an increase in visit frequency inside flowers that peaked between 0800-1000 hr. A. mellifera did not vary in their visit frequency or duration after 0600 hr. But, evidence suggested that Bombus spp. were actually less active earlier in the morning. In both years Bombus spp. visit frequency between 0600-

0700 hr was significantly less than their visit frequency at all other times throughout the pollination window. However, in 2011 Bombus spp. did have longer visit durations between 0600-0700 hr, than between 0800-1100 hr. P. pruinosa also spent more time in flowers between 0600-0700 hr in 2011, and a similar pattern was observed in 2012.

However, in both years, their visitation duration spiked at 1100 hr. The increase in visit duration exhibited by P. pruinosa near the end of the pollination window could be attributed to males looking for flowers to sleep in (Michelbacher et al., 1964; Hurd et al.,

1974). Hurd et al. (1974) found P. pruinosa to be active 22-55 minutes before sunrise, and Tepedino (1981) found that most pollination occurred by P. pruinosa before A. mellifera became prominent in the crop after 0800 hr. However my data show that there was no difference in visit frequency or visit duration between species at 0600 hr. Further, though the time that Bombus spp. and P. pruinosa spent inside flowers in 2012 shadowed the pollen deposition data, showing long visit durations and high pollen deposition

95 between 0600-0800 hr in 2012, their combined activity with A. mellifera at early hours necessitates differentiating their individual contributions.

Pollination contributions of A. mellifera in cucurbits, has been shown to be variable in past studies in The United States, Indonesia, Mexico, Austria, and Brazil.

Research on pollen deposition has shown that A. mellifera deposit 50-253, Bombus spp. deposit 75-252, and Peponapis spp. deposit 60-481 pollen grains per visit (Canto-Aguilar and Parra-Tabla, 2000; Winfree et al., 2007a; Goodell, 2008; Graças Vidal et al., 2010;

Phillips et al., 2010; Artz and Nault, 2011). Averaged across studies, A. mellifera,

Bombus spp., and P. pruinosa can potentially deposit 125.5, 167, and 217 grains of cucurbit pollen. Applied to the cumulative number per flower per hour of A. mellifera

(2.91), Bombus spp. (25.87), and P. pruinosa (3.91) in my 2012 study revealed that A. mellifera, Bombus spp., and P. pruinosa hypothetically transferred approximately 366,

4320, and 849 respective pollen grains to female flowers, totaling a potential 5535 pollen grains transferred before 0800 hr. This estimate exceeds the actual average of 4188 (±

294.49 SEM) pollen grains deposited in female flowers across sites between 0600-0800 hr, but this quick estimate should not replace more advanced measures of pollen deposition by species across the pollination window without taking into account pollen depletion in male flowers.

However, with the individual contributions of these pollinators approximated, the required pollen load to maximize fruit production (~2250 grains (Goodell, 2008)) in my pumpkins in 2012 was surpassed by 0800 hr, and the largest contributor was Bombus spp.

This is supported by Artz & Nault et al. (2011), who found that supplemental Bombus impatients hives improved fruit set of C. pepo by carrying three times as many pollen

96 grains per visit as A. mellifera, and Canto-Aguilar and Parra-Tabla (2000), who found that Peponapis limitaris deposited more pollen per visit to Cucurbita moschata plants in

Mexico, and visited flowers significantly more than A. mellifera. Finally, though Cane et al. (2011) did not quantify pollen deposition, they found that just seven visits by male P. pruinosa maximized C. pepo fruit set and growth. These studies and mine suggest that native wild bees can provide a majority of pollen deposition to C. pepo.

Further research on other Cucurbitaceae vine crops has also found that native wild bees could successfully pollinate watermelon (Citrullus lanatus) in Pennsylvania

(Winfree et al., 2007a), California (Kremen et al., 2002), and both C. lanatus and cucumber (Cucumis sativus) in North Carolina (Stanghellini et al., 1998). Winfree et al.

(2007a) modeled pollen deposition C. lanatus in Pennsylvania and found that pollen deposition was unrelated to A. mellifera visitation rate. Based on the simulation, they also found that native bees alone could deposit enough pollen for proper fruit set at 91% of the study farms compared to 73% if pollinated solely by A. mellifera. Stanghellini et al.

(1998) found that cucumber (Cucumis sativus), and watermelon (Citrullus lanatus) in

North Carolina experienced lower abortion rates, and higher seed sets when pollinated by

1, 6, and 18 Bombus spp. visits compared to the same number of A. mellifera visits.

The influence of floral insectaries on bee activity

My third prediction was that sweet alyssum floral insectaries would increase the visit frequency of A. mellifera and Bombus spp., by acting as alternative nectar sources before and after pumpkin bloom. My results did not support this, and showed that there were no significant differences between the site treatments. Additionally, the pollen deposition in female flowers adjacent to alyssum insectaries was numerically lower than 97 those adjacent to the grassy control insectary. This could be evidence of a competitive effect of the alyssum blooms that drew pollinators away from the pumpkin flowers, which has been shown in A. mellifera (Shuler et al., 2005; Delaplane and Mayer, 2010).

Additionally, Kells et al. (2001) found that wildflower species within naturally regenerated field margins experienced significantly more visitation by Bombus and A. mellifera than plants within cropped field margins. P. pruinosa was not predicted to be affected by the alyssum insectary, though they have been documented visiting morning glory (Williams et al., 2009) and blackberry flowers (Hurd et al., 1974) prior to pumpkin blooming. Though P. pruinosa could possibly utilize alyssum, the addition of alyssum rows in place of pumpkins would actually reduce the bloom area of their preferred pumpkin and squash flowers. Additionally, other local factors, such soil clay content, and tillage and irrigation practices could have had a stronger effect on the cucurbit specialist

P. pruinosa than the addition of a floral insectary (Shuler et al., 2005; Goodell, 2008;

Julier and Roulston, 2009), because those factors are known to influence their nesting behaviors.

Despite the lack of a response to the floral insectaries in my study, other researchers have found that insectaries of annual and perennial flowers can potentially be managed selectively to target generalist pollinators, and that the pollinators would enter a blooming adjacent crop. For example, Pywell et al. (2006) found that Bombus spp. richness was significantly higher in nectar and pollen rich wildflower plant mixes planted adjacent to cereal crops, and Pontin et al. (2005) observed that bees visiting floral insectaries adjacent to lucerne reduced their foraging behavior in the floral insectaries to take advantage of the blooming crop.

98 The influence of landscape composition on the visit frequency of bees

My fourth prediction was that higher percentages of semi-natural and less- disturbed habitat surrounding my sites would result in higher visit frequency of A. mellifera and Bombus spp. visiting pumpkin flowers, but decrease visit frequency of P. pruinosa by competing with the area pumpkin crop on which they specialize. My results variably supported this. A. mellifera was positively associated with the percentage of high quality forested areas within 500 m in 2011, which can support feral hives and multiple floral resources, but this trend did not appear in 2012. A. mellifera responded negatively to the percentage of mowed turfgrass within 500 m in 2011, and P. pruinosa did as well within large landscape scales. This could reflect a higher quality or quantity of floral resources, like clover or dandelion, and nesting habitat in those areas, which pulled bees away from pumpkins. Contrary to my prediction, Bombus spp. (in 2011), and P. pruinosa

(in 2012) responded positively to urban areas, which may have been a caused by the increased residential development of agricultural lands (Steffan-Dewenter et al., 2005). P. pruinosa and A. mellifera also responded negatively to the percentage of fruit and vegetable crops. The %FRUITVEG classification was made up of the individual percentages of leafy greens, tomatoes, peppers, grapes, berries, and stone fruit, in addition to cucurbits within the chosen radius of my site. However, cucurbits were the dominant contributor of agricultural landscapes to this category. The negative response to the percentage of fruit and vegetable crops could be indication of a dilution effect for P. pruinosa and a source of bloom competition for A. mellifera (Shuler et al., 2005;

Delaplane and Mayer, 2010). Julier and Roulston (2009) have been the only researchers to investigate landscape effects on P. pruinosa, and found that they did not respond to

99 their selected landscape predictors, and were instead more affected by localized management of the pumpkin crops on which they specialize.

Bombus spp. did not respond to any habitat predictors, within any radius in 2012, thought their population size dwarfed other bee taxa. The Intercept Only models (M1) acted as a null model predicted by a random distribution. All AICc model selections for

Bombus spp. ranked the Intercept Only model with an AICc difference ≤ 2, which made interpretation of landscape effects spurious. This could reflect previous findings that indicate Bombus spp. respond to landscape changes within landscape radii larger than

1000 m, up to radii of 1750 and 3000 m (Steffan-Dewenter et al., 2002; Westphal et al.,

2003). Still, others have found Bombus spp. respond to habitats within 1000 m, and are in fact positively correlated with the proportion of grasslands in the surrounding landscapes

(Hines and Hendrix, 2005; Pywell et al., 2006; Öckinger and Smith, 2007). Julier and

Roulston (2009) found that Bombus and A. mellifera did not respond to grasslands, but had a nonsignificant positive response to the percentage of forests, and Bombus had a nonsignificant negative response to flowering crops (orchards) within the small landscape radii (300 and 500 m). Interestingly, Westphal et al. (2003) found that Bombus spp. experienced a significant positive relationship with the proportion of mass flowering crops (oilseed rape) within a very large landscape scale (1750 and 3000 m). Further, their large population size in 2012, compared to 2011 could have been the result of a particularly wet spring in 2011 followed be a very dry spring in 2012. Bombus spp. queens survive the winter to establish nests in animal burrows, tufts of grass, and many abandoned structures. Perhaps the wet spring in 2011 made these microhabitats more

100 difficult to find or utilize, or the dry 2012 spring improved them in some way to enhance nest-finding and brood building.

My last prediction was that in the presence of intermediate habitat-complexity, visit frequencies of A. mellifera and Bombus spp. to pumpkin flowers at sites with local floral insectary additions (in 2012) would be significantly greater than at sites with a grassy control insectary. My results refuted this prediction. With the Intermediate

Landscape-Complexity hypothesis I expected to see the biggest differences between floral insectary treatments with between 1-20% non-crop habitats surrounding my sites, with smaller difference or no difference between treatments with < 1% non-crop habitat, or > 20% non-crop habitat. This would manifest itself as a significant interaction term that could be interpreted with an interaction plot. I found that A. mellifera visitation frequency did not respond strongly to alyssum insectaries at farms in the presence of high quality forested habitat within large landscape radii (1000 and 1500 m) compared to bees at farms with grassy control insectaries (Figure 3.6 B and C). Additionally, visitation frequency increased with forested habitat within small landscape radii (500 m) at farms with alyssum insectaries, and decreased at farms with grassy control insectaries (Figure

3.6 A). The percentage of forest around my sites was the largest non-crop habitat category, and made up between 1-50% of the landscapes. Between 1-20% forest habitats within the 500 and 1000 m radii, I observed the smallest difference in visit frequency of

A. mellifera between floral insectaries and grassy control insectaries, with larger difference at the extremes. Within 1500 m, the smallest difference was observed where forest habitats made up > 20% of the landscape, after which the visit frequencies between the two treatments continued to increase. More forested areas may have attracted A.

101 mellifera away from the crop, though Steffan-Dewenter et al. (2002) found that A. mellifera was negatively associated with semi-natural habitat. However, the response they observed occurred at larger landscape scales. Naturally, fewer croplands occur in landscapes with increased forest habitat, and feral colonies are in decline. The observed data could suggest that crop pollination requirements could be the main driver for the presence of A. mellifera in the form of commercial rental hives.

This project showed that all three bee taxa provide an important role in pollinating pumpkin flowers prior to 0800 hr. However, management of wild bees with the non- native annual, sweet alyssum used to enhance biocontrol, was not an effective way to enhance pumpkin pollination from managed and wild bees. Additionally, bees did not respond to intensified annual field crop agriculture within landscapes surround my sites, but instead to forested areas, flowering fruit and vegetable crop areas and urbanized environments. It will be important to continue to differentiate between non-crop habitats in future studies. Particularly between natural habitats, such as grasslands and forests, and intensified anthro-habitats, such as mowed turfgrass and artificially built impervious surfaces. The natural and semi-natural habitats utilized by wild bees are not only encroached upon by intensified agriculture, like annual field cropping. In The United

States, many natural areas are turned into residential sectors, with their own implications for pollinator survival. Perhaps selective re-incorporation of native perennials into the landscape could have a more positive, yet long-term effect on pumpkin pollinators by increasing the diversity of habitats and bloom resources before and after pumpkin anthesis (Tuell et al., 2008), with the stacked benefit of enhancing overwintering

102 structure for natural enemies (Fiedler et al., 2008), and restoring natural areas lost to human expansion.

103 TABLES 2011 2012 Site0 Site Latitude Longitude Video0date Treatment Latitude Longitude Video0date

1 40°054'037.94" 82°06'035.06" 16?Aug a 40°054'09.69" 82°06'044.11" 17?Aug 2 40°055'06.92" 82°02'057.66" 26?Aug a 40°054'058.95" 82°02'048.14" 13?Aug 3 40°056'025.06" 82°06'058.21" 31?Aug ? ? ? ? 4 41°05'02.65" 81°057'01.51" 25?Aug c 41°05'03.28" 81°057'08.13" 16?Aug 5 40°042'037.87" 81°058'016.31" ? c 40°042'023.5" 81°057'056.45" 16?Aug 6 40°055'017.93" 81°018'033.26" 14?Aug c 40°055'017.29" 81°018'031.78" 15?Aug 8 ? ? ? a 40°058'013.68" 81°044'025.37" 15?Aug 9 39°026'05.63" 83°059'026.59" 10?Aug a 39°026'04.39" 83°059'01.35" 8?Aug 10 39°02'050.88" 82°059'037.4" ? a 39°02'049.35" 82°059'038.15" 1?Aug 11 39°013'013.41" 83°025'036.81" 9?Aug ? ? ? ? 12 39°010'058.65" 83°021'03.09" 11?Aug c 39°010'055" 83°021'011.37" 3?Aug 13 38°059'029.9" 82°046'04.54" ? c 38°059'037.44" 82°045'051.76" 11?Aug 14 39°08'016.65" 82°058'058.47" 19?Aug c 39°08'011.46" 82°058'059.39" 2?Aug 15 ? ? ? a 39°024'041.94" 83°09'027.33" 31?Jul

Table 3.1. The location, 2012 treatment assignments and video sample dates for pumpkin farms sampled in 2011 and 2012. All farms had one pumpkin field. Site treatments were two 60 m rows of an annual non-native plant, sweet alyssum, flanking 4 rows of pumpkins (a); or a 6 x 60 m grassy control insectary adjacent to the pumpkins (c).

104 Year Model Description 20123Model3Codea,b Predictionsc

2011 M1 Random3distribution Intercept Random3distribution3(0) FrequentlyAdisturbed3333333333333333333333333333 M2 %FIELD %FIELD3(A) annual3field3crops Uncultivated3grasslands3and33333333333333333333 M3 %GRASSLAND %GRASSLAND3(+) openAcanopy3river3corridors InfrequentlyAdisturbed3333333333333333333 M4 %FORAGE %FORAGE3(0) perennial3crops M5 Fruit3and3vegetable3crops %FRUITVEG %FRUITVEG3(+) Uncultivated3forest,3fencerows,3and3 M6 %FOREST %FOREST3(+) closedAcanopy3river3corridors M7 ManAmade3impervious3surfaces %URBAN %URBAN3(A)

M8 Mowed3turf %TURF %TURF3(A) Perennial3crops3and3333333333333333333 M9 %GRASSLAND3+3%FORAGE %GRASSLAND3(+)3and3%FORAGE3(0) uncultivated3areas %GRASSLAND3+3%FORAGE3+3 %GRASSLAND3(+),3%FORAGE3(0),3and3 M10 All3grassy3areas %TURF %TURF3(A) M11 All3urbanized3areas %URBAN3+3%TURF %URBAN3(A)3and3%TURF3(A) %FRUITVEG3+3%FIELD3+3 %FRUITVEG3(+),3%FIELD3(A),3and3 M12 All3forms3of3agriculture %FORAGE %FORAGE3(0) M13 All3forms3of3uncultivated3land %FOREST3+3%GRASSLAND %FOREST3(+)33and3%GRASSLAND3(+)

2012 M1 Random3distribution Intercept Random3distribution3(0) FrequentlyAdisturbed3333333333333333333333333333 %FIELD3(A),3333333333333333333333333333333333333333333 M15 %FIELD3*3SITETRT annual3field3crops SITETRT3(c3<3a) Uncultivated3grasslands3and33333333333333333333 %GRASSLAND3(+),333333333333333333333333333333 M16 %GRASSLAND3*3SITETRT openAcanopy3river3corridors SITETRT3(c3=3a) InfrequentlyAdisturbed3333333333333333333 %FORAGE3(0),3333333333333333333333333333333333333 M17 %FORAGE3*3SITETRT perennial3crops SITETRT3(c3<3a) %FRUITVEG3(+),3333333333333333333333333333333333 M18 Fruit3and3vegetable3crops %FRUITVEG3*3SITETRT SITETRT3(c3=3a) Uncultivated3forest,3fencerows,3and3 %FOREST3(+),33333333333333333333333333333333333333 M19 %FOREST3*3SITETRT closedAcanopy3river3corridors SITETRT3(c3=3a) %URBAN3(A),3333333333333333333333333333333333333333 M20 ManAmade3impervious3surfaces %URBAN3*3SITETRT SITETRT3(c3<3a) %TURF3(A),3333333333333333333333333333333333333333333 M21 Mowed3turf %TURF3*3SITETRT SITETRT3(c3<3a) Perennial3crops3and3333333333333333333 %GRASSLAND3+3%FORAGE3+3333%GRASSLAND3(+)3and3%FORAGE3(0),3333 M22 uncultivated3areas SITETRT SITETRT3(c3=3a) %GRASSLAND3+3%TURF3+333333 %GRASSLAND3(+),3and3%TURF3(A),3333333 M23 All3grassy3areas SITETRTb SITETRT3(c3<3a) %URBAN3+3%TURF3+33333333333333333333%URBAN3(A)3and3%TURF3(A),3333333333333333333 M24 All3urbanized3areas SITETRT SITETRT3(c3<3a) %FRUITVEG3+3%FIELD3+333333333 %FRUITVEG3(+),3%FIELD3(A),3and3 M25 All3forms3of3agriculture %FORAGE3+3SITETRT %FORAGE3(0),3SITETRT3(c3<3a) %FOREST3+3%GRASSLAND3+3333333%FOREST3(+)3and3%GRASSLAND3(+),33333 M26 All3forms3of3uncultivated3land SITETRT SITETRT3(c3=3a)

Table 3.2. A total of 26 models were used on the pollinator visit frequency data. Models M1-M13 were used in 2011. Models M1-M26 were used in 2012. a An astrisk in R denotes that each factor is included individually in the model, and also interacted. b %Forage was removed from the 2012 All Grass model at 1500 m because of a high VIF score. c Inequality statements indicate more, less, or equal effect of floral insectary treatment on the response variable, where c = control, and a = alyssum.

105 Year Species Scale+(m) Model LL K AICc Δi Wi Signifcant+factor+estimates+(p=values) 2011 A.#mellifera 500 M6 '267.66 4 543.32 0 0.224 %FOREST30.0213(0.031) M13 =266.68 5 543.354 0.243 0.199 M8 =268.12 4 544.23 0.91 0.142 %TURF+=0.067+(0.038) M3 =268.57 4 545.134 1.814 0.091 M11 =267.58 5 545.16 2.049 0.081 %TURF+=0.073+(0.024)

Bombus#spp. 1500 M7 '190.86 4 389.726 0 0.506 %URBAN30.0863(0.000) M11 =190.58 5 391.158 1.641 0.223 %URBAN+0.101+(0.001)

P.#pruinosa 500 M12 '283.71 6 579.41 0 0.487 %FRUITVEG3'0.0793(0.000) M5 =286.18 4 580.364 0.479 0.383 %FRUITVEG+=0.074+(0.000)

2012 A.#mellifera 500 M18 '201.55 6 415.10 0.00 0.37 %FRUITVEG3'0.0593(0.005) M25 =200.51 7 415.03 0.25 0.33 %FRUITVEG3'0.0483(0.007) %FOREST+=0.035+(0.001),++++++++++++++++++++++++++++++++++ M19 =202.61 6 417.23 2.12 0.13 %FOREST+:+ALYSSUM+0.060+(0.015)

%FOREST+=0.065+(0.000),++++++++++++++++++++++++++++++++++ 1000 M19 '200.24 6 412.49 0.00 0.59 %FOREST+:+ALYSSUM+0.057+(0.019) M25 =200.17 7 414.33 2.17 0.20 %FRUITVEG+=0.097+(0.025)

%FRUITVEG3'0.1363(0.034),3%FIELD30.0213(0.002),3 1500 M25 '200.54 7 415.09 0.00 0.46 %FORAGE3'0.2933(0.004) %FOREST+=0.692+(0.019),+ALYSSUM+=1.54+(0.014),+ M19 =202.14 6 416.27 0.86 0.30 %FOREST+:+ALYSSUM+0.068+(0.002)

P.#pruinosa 500 M18 '269.56 6 551.11 0.00 0.46 %FRUITVEG3'0.0753(0.001),3ALYSSUM3'1.173(0.002) M20 =270.31 6 552.62 1.51 0.22 %URBAN+0.050+(0.002)

%URBAN30.0473(0.000),3%TURF3'0.1293(0.000),3 1000 M24 '265.02 6 542.03 0.00 0.98 ALYSSUM3'0.5583(0.003)

%URBAN30.0623(0.000),3%TURF3'0.0833(0.001),3 1500 M24 '266.46 6 544.92 0.00 0.93 ALYSSUM3'0.5473(0.004)

Table 3.3. AICc table showing the top candidate visit frequency models for each year, landscape scale and species (bold), followed by models that were ranked with differences ≤ 2. LL = log-likelihood, K = number of parameters, Δi = AICc difference, Wi = AICc weight. *%FORAGE was dropped from this a priori model in 2012 because of a high VIF score. Factors within models that were significant are listed, along with their model estimates and p-values.

106 FIGURES

Figure 3.1. 2011 sites were located in pumpkin growing regions in northern and southern Ohio (A). In 2012 site 3 and 11 were dropped, and sites 7, 8 and 15 were added (B). Sweet alyssum insectary site treatments were added in 2012 and both landscape-level and local habitat effects were measured. See Table 3.1 for site coordinates.

107 2011 2012 80 80 A. mellifera A. B. Bombus spp. P. pruinosa 60 60 A a A a

40 40 A a B A A 20 A A a A A 20 b B b b b b ab B c c Average number of bees +/- SEM +/- bees of number Average

0 0 Female Male Female Male 6 6 C. D.

5 5

108 A 4 4 a

3 3 A A a A 2 2 b B b A A B A A A b b A b b b 1 b b b 1

Average visit duration of bees +/- SEM +/- bees of duration visit Average 0 0 Female Male Female Male

Flower sex Flower sex

Figure 3.2. The average number of visits, and average visit duration of bees in male and female flowers in 2011 (A and C), and 2012 (B and D), as observed by video cameras. The Greek letters above the bars indicate significant differences within species across flower sex, and Arabic letters denote differences between species within flower sexes.

108 2011 2012 80 80 A. mellifera A B. A. A a A Bombus spp. a a P. pruinosa 60 60 A a

40 A 40 A AB a AB a AB a AB A a A a A A a A A A A A A 20 A a ab A A 20 B ab A b A A a b a A A b A B a A b b A a a ab A a b b A B a b a b a a b b

Average number of bees +/- SEM +/- bees of number Average a 0 0 6 7 8 9 10 11 6 7 8 9 10 11

6 6 C. D. A 5 5 a A

109 a 4 4 A

a A A 3 A A 3 a a A a A A A A a A A A A A A a a a A AB a a a 2 AB a 2 a a A AB a ab A A a a AB b a B AB b B B B ab B b ab B B B B b 1 b b b b 1 b b b b b

Average visit duration of bees +/- SEM +/- bees of duration visit Average 0 0 6 7 8 9 10 11 6 7 8 9 10 11

Hour of day Hour of day

Figure 3.3. The average number of visits, and average visit duration of bees across the pollination window between 0600 hr and 1200 hr in 2011 (A and C), and 2012 (B and D), as observed by video cameras. The Greek letters above the bars indicate significant differences within species across hour, and Arabic letters denote differences between species within hours.

109 Flower visitation frequency of bees Flower visitation duration of bees

80 6 Average visit duration (min) +/- SEM A. A. mellifera B. A Bombus spp. 5 a P. pruinosa 60 4 A A a 40 A 3 a a A A A b 2 A A b b A 20 A b

Average number +/- SEM +/- number Average b A b b 1 c

0 0 Control Alyssum Control Alyssum

Site treatment Site treatment

Figure 3.4. The average number of visits (A), and average visit duration (B) of bees between grassy control and alyssum insectaries in 2012 The letters above the bars indicate significant differences within species. Site treatments were two 60 m rows of an annual non-native plant, sweet alyssum, flanking 4 rows of pumpkins, or a 6 x 60 m grassy control insectary adjacent to the pumpkins. The Greek letters above the bars indicate significant differences within species across site treatment, and Arabic letters denote differences between species within site treatments.

Figure 3.5. The average number of pollen grains deposited on pumpkin stigmas between the hours of 0600-0800, 0800- 1000, 1000-1200, and 0600-1200 (A), and between site treatments (B). Site treatments were two 60 m rows of an annual non-native plant, sweet alyssum, flanking 4 rows of pumpkins, or a 6 x 60 m grassy control insectary adjacent to the pumpkins.

110 A.# # B.# mellifera A.# Visit#frequency#of#

C.#

%FOREST#within#1500m#

Figure 3.6. Interaction plots showing the visit frequency of A. mellifera in pumpkin flowers planted adjacent to grassy control insectaries and alyssum insectaries with an increasing percentage of forested areas within 500 m (A), 1000 m (B), and 1500 m (C) from my sites. Site treatments were two 60 m rows of an annual non-native plant, sweet alyssum, flanking 4 rows of pumpkins (triangles); or a 6 x 60 m grassy control insectary adjacent to the pumpkins (circles).

111

References

Amaral, D.S., Venzon, M., Duarte, M.V., Sousa, F.F., Pallini, A., and Harwood, J.D. (2012). Non-crop vegetation associated with chili pepper agroecosystems promote the abundance and survival of aphid predators. Biological Control 64, 338–346.

Ambrosino, M.D. (2006). Enhancing the predatory potential of hoverflies on aphids in Oregon broccoli fields with floral resources. Oregon State University.

Anderson, D.R., Link, W.A., Johnson, D.H., and Burnham, K.P. (2001). Suggestions for presenting the results of data analyses. The Journal of Wildlife Management 373–378.

Artz, D.R., and Nault, B.A. (2011). Performance of Apis mellifera, Bombus impatiens, and Peponapis pruinosa (Hymenoptera: Apidae) as pollinators of pumpkin. Journal of Economic Entomology 104, 1153–1161.

Artz, D.R., Hsu, C.L., and Nault, B.A. (2011). Influence of honey bee, Apis mellifera, hives and field size on foraging activity of native bee species in pumpkin fields. Environmental Entomology 40, 1144–1158.

Ashworth, L., and Galetto, L. (2001). Pollinators and reproducting success of the wild cucurbit Cucurbita maxima spp. andreana (Cucurbitaceae). Plant Biol 3, 398–404.

Atakan, E. (2010). Influence of weedy field margins on abundance patterns of the predatory bugs Orius spp. and their prey, the western flower thrips (Frankliniella occidentalis), on faba bean. Phytoparasitica 38, 313–325.

Baggen, L.R., and Gurr, G.M. (1998). The influence of food on Copidosoma koehleri (Hymenoptera: Encyrtidae), and the use of flowering plants as a habitat management tool to enhance biological control of potato moth, Phthorimaea operculella (Lepidoptera: Gelechiidae). Biological Control 11, 9–17.

Barber, N.A., Adler, L.S., and Bernardo, H.L. (2011). Effects of above and below-ground herbivory on growth, pollination, and reproduction in cucumber. Oecologia 165, 377– 386.

Barnes, J.P., and Putnam, A.R. (1983). Rye residues contribute weed suppression in no- tillage cropping systems. Journal of Chemical Ecology 9, 1045–1057.

112 Batáry, P., Fischer, J., Báldi, A., Crist, T.O., and Tscharntke, T. (2011). Does habitat heterogeneity increase farmland biodiversity? Frontiers in Ecology and the Environment 9, 152–153.

Beard, R.L. (1940a). The biology of Anasa tristis DeGeer with particular reference to the tachinid parasite, Trichopoda pennipes Fabr. (Connecticut Agricultural Experiment Station).

Beard, R.L. (1940b). Parasitic vastration of Anasa tristis DeG. by Trichopoda pennipes Fab., and its effect on reproduction. Journal of Economic Entomology 33, 269–272.

Beard, R.L. (1942). On the formation of the tracheal funnel in Anasa tristis DeG. induced by the parasite Trichopoda pennipes Fabr. Annals of the Entomological Society of America 35, 68–72.

Bennett, E.R., Moore, M.T., Cooper, C.M., Smith Jr, S., Shields Jr, F.D., Drouillard, K.G., and Schulz, R. (2005). Vegetated agricultural drainage ditches for the mitigation of pyrethroid-associated runoff. Environmental Toxicology and Chemistry 24, 2121–2127.

Berndt, L.A., Wratten, S.D., and Scarratt, S.L. (2006). The influence of floral resource subsidies on parasitism rates of leafrollers (Lepidoptera: Tortricidae) in New Zealand vineyards. Biological Control 37, 50–55.

Bianchi, F.J., and Wäckers, F.L. (2008). Effects of flower attractiveness and nectar availability in field margins on biological control by parasitoids. Biological Control 46, 400–408.

Bianchi, F.J.J.A., Booij, C.J.H., and Tscharntke, T. (2006). Sustainable pest regulation in agricultural landscapes: a review on landscape composition, biodiversity and natural pest control. Proceedings of the Royal Society B: Biological Sciences 273, 1715–1727.

Bigger, D., and Chaney, W. (1998). Effects of Iberis umbellata (Brassicaceae) on insect pests of cabbage and on potential biological control agents. Environmental Entomology 27, 161–167.

Blaauw, B.R. (2013). Composition, size, and effectiveness of native plant conservation plantings for supporting beneficial insects (Michigan State University).

Bolker, B.M. (2007). Probability and statistical distributions for ecological modeling. In Ecological Models and Data in R, (Princeton, New Jersey: Princeton University Press),.

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

113 Braman, S.K., Pendley, A.F., and Corley, W. (2002). Influence of commercially available wildflower mixes on beneficial arthropod abundance and predation in turfgrass. Environmental Entomology 31, 564–572.

Brewer, J.W. (1974). Pollination requirements for watermelon seed production. Journal of Apicultural Research 13, 207–212.

Bugg, R.L., and Waddington, C. (1994). Using cover crops to manage arthropod pests of orchards: A review. Agriculture, Ecosystems & Environment 50, 11–28.

Burgio, G., Ferrari, R., Boriani, L., Pozzati, M., and van Lenteren, J. (2006). The role of ecological infrastructures on Coccinellidae (Coleoptera) and other predators in weedy field margins within northern Italy agroecosystems. Bulletin of Insectology 59, 59–67.

Cane, J.H., Minckley, R.L., and Kervin, L.J. (2000). Sampling bees (Hymenoptera: Apiformes) for pollinator community studies: pitfalls of pan-trapping. J Kansas Entomol Soc 73, 225–231.

Cane, J.H., Sampson, B.J., and Miller, S.A. (2011). Pollination value of male bees: The specialist bee Peponapis pruinosa (Apidae) at summer squash (Cucurbita pepo). Environmental Entomology 40, 614–620.

Canto-Aguilar, M.A., and Parra-Tabla, V. (2000). Importance of conserving alternative pollinators: assessing the pollination efficiency of the squash bee, Peponapis limitaris in Cucurbita moschata (Cucurbitaceae). J Insect Conserv 4, 203–210.

Chaplin-Kramer, R., O’Rourke, M.E., Blitzer, E.J., and Kremen, C. (2011). A meta- analysis of crop pest and natural enemy response to landscape complexity. Ecology Letters 14, 922–932.

Chen, X., Chen, Y., Wu, L., Peng, Y., Chen, J., and Liu, F. (2010). A survey of nectar feeding by spiders in three different habitats. Bulletin of Insectology 63, 203–208.

Cline, G.R., Sedlacek, J.D., Hillman, S.L., Parker, S.K., and Silvernail, A.F. (2008). Organic management of cucumber beetles in watermelon and muskmelon production. Hort Tech 18, 436–444.

Cole, L.J., McCracken, D.I., Baker, L., and Parish, D. (2007). Grassland conservation headlands: their impact on invertebrate assemblages in intensively managed grassland. Agr Ecosyst Environ 122, 252–258.

Colley, M.R., and Luna, J.M. (2000). Relative attractiveness of potential beneficial insectary plants to aphidophagous hoverflies (Diptera: Syrphidae). Environmental Entomology 29, 1054–1059.

114 Collins, K., Boatman, N., Wilcox, A., Holland, J., and Chaney, K. (2002). Influence of beetle banks on cereal aphid predation in winter wheat. Agriculture, Ecosystems and Environment 93, 337–350.

Collins, K., Boatman, N., Wilcox, A., and Holland, J. (2003). A 5-year comparison of overwintering polyphagous predator densities within a beetle bank and two conventional hedgebanks. Annals of Applied Biology 143, 63–71.

Concepción, E.D., Díaz, M., Kleijn, D., Báldi, A., Batáry, P., Clough, Y., Gabriel, D., Herzog, F., Holzschuh, A., Knop, E., et al. (2012). Interactive effects of landscape context constrain the effectiveness of local agri-environmental management. Journal of Applied Ecology 49, 695–705.

Cottrell, T.E., and Yeargan, K.V. (1999). Factors influencing dispersal of larval Coleomegilla maculata from the weed Acalypha ostryaefolia to sweet corn. Entomologia Experimentalis et Applicata 90, 313–322.

Cowgill, S., Sotherton, N., and Wratten, S. (1993). The selective use of floral resources by the hoverfly Episyrphus balteatus (Diptera: Syrphidae) on farmland. Annals of Applied Biology 122, 223–231.

Decker, K.B., and Yeargan, K.V. (2008). Seasonal phenology and natural enemies of the squash bug (Hemiptera: Coreidae) in Kentucky. Environ Entomol 37, 670–678.

Delaplane, K., and Mayer, D. (2010). Squash, pumpkin, and gourd. In Crop Pollination by Bees, pp. 257–262.

Driesche, R.G.V., and Bellows, T.S.J. (1996). Biology of Arthropod Parasitoids and Predators. In Biological Control, (Springer US), pp. 309–336.

Dufour, R. (2000). Farmscaping to enhance biological control (ATTRA).

Ellers-Kirk, C., and Fleischer, S.J. (2006). Development and life table of Acalymma vittatum (Coleoptera: Chrysomelidae), a vector of Erwinia tracheiphila in cucurbits. Environ Entomol 35, 875–880.

Ellers-Kirk, C.D., Fleischer, S.J., Snyder, R.H., and Lynch, J.P. (2000). Potential of entomopathogenic nematodes for biological control of Acalymma vittatum (Coleoptera: Chrysomelidae) in cucumbers grown in conventional and organic soil management systems. Journal of Economic Entomology 93, 605–612.

Elzinga, J.A., van Nouhuys, S., van Leeuwen, D.-J., and Biere, A. (2007). Distribution and colonisation ability of three parasitoids and their herbivorous host in a fragmented landscape. Basic and Applied Ecology 8, 75–88.

English-Loeb, G., Rhainds, M., Martinson, T., and Ugine, T. (2003). Influence of flowering cover crops on Anagrus parasitoids (Hymenoptera: Mymaridae) and 115 Erythroneura leafhoppers (Homoptera: Cicadellidae) in New York vineyards. Agricultural and Forest Entomology 5, 173–181.

ESRI (2011). ArcMap (Redlands, California: ESRI (Environmental Systems Resource Institute)).

Fiedler, A.K., and Landis, D.A. (2007a). Attractiveness of Michigan native plants to arthropod natural enemies and herbivores. Environmental Entomology 36, 751–765.

Fiedler, A.K., and Landis, D.A. (2007b). Plant characteristics associated with natural enemy abundance at Michigan native plants. Environmental Entomology 36, 878–886.

Fiedler, A., Tuell, J., Isaacs, R., and Landis, D. (2007). Attracting beneficial insects with native flowering plants (Michigan State University Extension).

Fiedler, A.K., Landis, D.A., and Wratten, S.D. (2008). Maximizing ecosystem services from conservation biological control: the role of habitat management. Biological Control 45, 254–271.

Fuchs, R., and Müller, M. (2004). Pollination problems in Styrian oil pumpkin plants: can bumblebees be an alternative to honeybees? Phyton 44, 155–165.

Gardiner, M., Landis, D., Gratton, C., DiFonzo, C., O’Neal, M., Chacon, J., Wayo, M., Schmidt, N., Mueller, E., and Heimpel, G. (2009). Landscape diversity enhances biological control of an introduced crop pest in north-central USA. Ecol Appl 19, 143– 154.

Gardiner, M., Landis, D., Gratton, C., Schmidt, N., O’Neal, M., Mueller, E., Chacon, J., and Heimpel, G. (2010). Landscape composition influences the activity density of Carabidae and Arachnida in soybean fields. Biol Control 55, 11–19.

Gardiner, M.M., Prajzner, S.P., Burkman, C.E., Albro, S., and Grewal, P.S. (2013). Vacant land conversion to community gardens: influences on generalist arthropod predators and biocontrol services in urban greenspaces. Urban Ecosystems 1–22.

Garibaldi, L.A., Steffan-Dewenter, I., Winfree, R., Aizen, M.A., Bommarco, R., Cunningham, S.A., Kremen, C., Carvalheiro, L.G., Harder, L.D., Afik, O., et al. (2013). Wild pollinators enhance fruit set of crops regardless of honey bee abundance. Science 339, 1608–1611.

Gillespie, M., Wratten, S., Sedcole, R., and Colfer, R. (2011). Manipulating floral resources dispersion for hoverflies (Diptera: Syrphidae) in a California lettuce agro- ecosystem. Biological Control 59, 215–220.

Gontijo, L.M., Beers, E.H., and Snyder, W.E. (2013). Flowers promote aphid suppression in apple orchards. Biological Control 66, 8–15.

116 Goodell, K. (2008). Contributions of native, unmanaged bees to pollination of pumpkin and squash in Ohio correlates to their densities.

Graças Vidal, M. das, De Jong, D., Wien, H.C., and Morse, R.A. (2010). Pollination and fruit set in pumpkin (Cucurbita pepo) by honey bees. Revista Brasil Bot 33, 107–113.

Grieshop, M., Werling, B., Buehrer, K., Perrone, J., Isaacs, R., and Landis, D. (2012). Big brother is watching: studying insect predation in the age of digital surveillance. Amer Entomol 58, 172–182.

Gurr, G.M., Wratten, S.D., and Luna, J.M. (2003). Multi-function agricultural biodiversity: pest management and other benefits. Basic and Applied Ecology 4, 107– 116.

Haenke, S., Scheid, B., Schaefer, M., Tscharntke, T., and Thies, C. (2009). Increasing syrphid fly diversity and density in sown flower strips within simple vs complex landcapes. J Appl Ecol 46, 1106–1114.

Halaj, J., Cady, A., and Uetz, G. (2000). Modular habitat refugia enhance generalist predators and lower plant damage in soybeans. Environmental Entomology 29, 383–393.

Heinrich, B. (2004). Bumblebee economics (Harvard University Press).

Hickman, J.M., and Wratten, S.D. (1996). Use of Phelia tanacetifolia strips to enhance biological control of aphids by hoverfly larvae in cereal fields. Journal of Economic Entomology 89, 832–840.

Hines, H.M., and Hendrix, S.D. (2005). Bumble bee (Hymenoptera: Apidae) diversity and abundance in tallgrass prairie patches: effects of local and landscape floral resources. Environmental Entomology 34, 1477–1484.

Hoehn, P., Tscharntke, T., Tylianakis, J.M., and Steffan-Dewenter, I. (2008). Functional group diversity of bee pollinators increases crop yield. Proc. R. Soc. B 275, 2283–2291.

Hoffmann, M., Davidson, N., Wilson, L., Ehler, L., Jones, W., and Zalom, F. (1991). Imported wasp helps control southern green stink bug. California Agriculture 45, 20–22.

Hogg, B.N., Bugg, R.L., and Daane, K.M. (2011). Attractiveness of common insectary and harvestable floral resources to beneficial insects. Biological Control 56, 76–84.

Holzschuh, A., Steffan-Dewenter, I., and Tscharntke, T. (2009). How do landscape composition and configuration, organic farming and fallow strips affect the diversity of bees, wasps and their parasitoids? J Anim Ecol 79, 491–500.

Hossain, Z., Gurr, G.M., Wratten, S.D., and Raman, A. (2002). Habitat manipulation in lucerne Medicago sativa!: arthropod population dynamics in harvested and “refuge” crop strips. Journal of Applied Ecology 39, 445–454. 117 Hurd, P.D., Linsley, E.G., and Michelbacher, A.E. (1974). Ecology of the squash and gourd bee, Peponapis pruinosa, on cultivated cucurbits in California (Hymenoptera: Apoidea) (Smithsonian Institution Press).

Hurd, Jr, P.D., Linsley, E.G., and Whitaker, T.W. (1971). Squash and gourd bees (Peponapis, Xenoglossa) and the origin of the cultivated cucurbita. Evolution 25, 218– 234.

Isaacs, R., Tuell, J., Fiedler, A., Gardiner, M., and Landis, D. (2009). Maximizing arthropod-mediated ecosystem services in agricultural landscapes: the role of native plants. Frontiers in Ecology and the Environment 7, 196–203.

Jackson, R.R., Pollard, S.D., Nelson, X.J., Edwards, G.B., and Barrion, A.T. (2001). Jumping spiders (Araneae: Salticidae) that feed on nectar. Journal of Zoology 255, 25– 29.

Jasinski, J., and Welty, C. (2002). Perimeter trap cropping in pumpkin production (Ohio State University Extension).

Jaycox, E.R., Guynn, G., Rhodes, A.M., and Vandemark, J.S. (1975). Observation on pumpkin pollination in Illinois. Am. Bee J. 115.

Johanowicz, D., and Mitchell, E. (2000). Effects of sweet alyssum flowers on the longevity of the parasitoid wasps Cotesia marginiventris (Hymenoptera: Braconidae) and Diadegma insulare (Hymenoptera: Ichneumonidae). Florida Entomologist 83, 41–47.

Jones, V.P., Westcott, D.M., Finson, N.N., and Nishimoto, R.K. (2001). Relationship between community structure and southern green stink bug (Heteroptera: Pentatomidae) damage in macadamia nuts. Environmental Entomology 30, 1028–1035.

Julier, H.E., and Roulston, T.H. (2009). Wild bee abundance and pollination service in cultivated pumpkins: farm management, nesting behavior and landscape effects. J Econ Entomol 102, 563–573.

Kells, A.R., Holland, J.M., and Goulson, D. (2001). The value of uncropped field margins for foraging bumblebees. J Insect Conserv 5, 283–291.

Kennedy, C.M., Lonsdorf, E., Neel, M.C., Williams, N.M., Ricketts, T.H., Winfree, R., Bommarco, R., Brittain, C., Burley, A.L., Cariveau, D., et al. (2013). A global quantitative synthesis of local and landscape effects on wild bee pollinators in agroecosystems. Ecology Letters 16, 584–599.

Klein, A.M., Vaissiere, B.E., Cane, J.H., Steffan-Dewenter, I., Cunningham, S.A., Kremen, C., and Tscharntke, T. (2007). Importance of pollinators in changing landscapes for world crops. Proc R Soc Lond B 274, 303–313.

118 Koji, S., Khan, Z.R., and Midega, C. a. O. (2007). Field boundaries of Panicum maximum as a reservoir for predators and a sink for Chilo partellus. Journal of Applied Entomology 131, 186–196.

Kremen, C., and Ostfeld, R.S. (2005). A call to ecologists: measuring, analyzing, and managing ecosystem services. Front. Ecol. Environ. 3, 540–548.

Kremen, C., Williams, N.W., and Thorp, R.W. (2002). Crop pollination from native bees at risk from agricultural intensification. Pest Management Science 99, 16812–16816.

Kuepper, G., and Dodson, M. (2001). Companion planting: basic concepts and resources (ATTRA).

Lal, R. (1993). Tillage effects on soil degradation, soil resilience, soil quality, and sustainability. Soil and Tillage Research 27, 1–8.

Landis, D.A., Wratten, S.D., and Gurr, G.M. (2000). Habitat management to conserve natural enemies of arthropod pests in agriculture. Annu Rev Entomol 45, 175–201.

Laub, C., Youngman, R.R., Love, K., and Mize, T. (2009). Using pitfall traps to monitor insect activity (Virginia Cooperative Extension).

Lavandero, B., Wratten, S., Shishehbor, P., and Worner, S. (2005). Enhancing the effectiveness of the parasitoid Diadegma semiclausum (Helen): movement after use of nectar in the field. Biol Control 34, 152–158.

Lavandero, B., Wratten, S.D., Didham, R.K., and Gurr, G. (2006). Increasing floral diversity for selective enhancement of biological control agents: a double-edged sward? Basic Appl Ecol 7, 236–243.

Lee, J.C., and Heimpel, G.E. (2005). Impact of flowering buckwheat on lepidopteran cabbage pests and their parasitoids at two spatial scales. Biological Control 34, 290–301.

Lee, J.C., Andow, D.A., and Heimpel, G.E. (2006). Influence of floral resources on sugar feeding and nutrient dynamics of a parasitoid in the field. Ecological Entomology 31, 470–480.

Losey, J., and Vaughan, M. (2006). The economic value of ecological services provided by insects. BioScience 56, 311–323.

MacLeod, A. (1999). Attraction and retention of Episyrphus balteatus DeGeer (Diptera: Syrphidae) at an arable field margin with rich and poor floral resources. Agriculture, Ecosystems and Environment 73, 237–244.

MacLeod, A., Wratten, S., Sotherton, N., and Thomas, M. (2004). “Beetle banks” as refuges for beneficial arthropods in farmland: long-term changes in predator communities and habitat. Agr Forest Entomol 6, 147–154. 119 Marshall, E., Brown, V., Boatman, N., Lutman, P., Squire, G., and Ward, L. (2003). The role of weeds in supporting biological diversity within crop fields. Weed Research 43, 77–89.

McElduff, F., Cortina-Borja, M., Chan, S.K., and Wade, A. (2010). When t-tests or Wilcoxon-Mann-Whitney tests won’t do. Advan in Physiol Edu 34, 128–133.

Metcalf, R., and Metcalf, R. (1993). Destructive and Useful Insects: Their Habits and Control (NY: McGraw Hill).

Michelbacher, A., Smith, R.F., and Hurd, P. (1964). Bees are essential...pollination of squashes, gourds and pumpkins. California Agriculture 18, 2–4.

Moore, M.T., Kröger, R., Cooper, C.M., and Smith, S. (2009). Ability of four emergent macrophytes to remediate permethrin in mesocosm experiments. Archives of Environmental Contamination and Toxicology 57, 282–288.

Morse, R.A., and Calderone, N.W. (2000). The value of honey bees as pollinators of US crops in 2000. Bee Culture 128, 1–15.

NASS (2012). Farm production Expenditures 2011 Summary (National Agricultural Statistics Service).

Nechols, J., Tracy, J., and Vogt, E. (1989). Comparative ecological studies of indigenous egg parasitoids (Hymenoptera: Scelionidae; Encyrtidae) of the squash bug, Anasa tristis (Hemiptera: Coreidae). Journal of the Kansas Entomological Society 62, 177–188.

Nentwig, W., Frank, T., and Lethmayer, C. (1998). Sown weed strips: artificial ecological compensation areas as an important tool in conservation biological control. Conservation Biological Control. Academic Press, San Diego 133–153.

Nicodemo, D., Nogueira Couto, R.H., Malheiros, E.B., and De Jong, D. (2009). Honey bee as an effective pollination agent of pumpkin. Sci Agr 66, 476–480.

Oberg, S., Mayr, S., and Dauber, J. (2008). Landscape effects on recolonisation patterns of spiders in arable fields. Agr Ecosyst Environ 123, 211–218.

Öckinger, E., and Smith, H.G. (2007). Semi-natural grasslands as population sources for pollinating insects in agricultural landscapes. J Appl Ecol 44, 50–59.

OGRIP (2010). Ohio Statewide Imagery Program (OGRIP (Ohio Geographically Referenced Information Program)).

Olson, D.M., and Wäckers, F.L. (2007). Management of field margins to maximize multiple ecological services. Journal of Applied Ecology 44, 13–21.

120 Ostman, O., Ekbom, B., Bengtsson, J., and Weibull, A.C. (2001). Landscape complexity and farming practice influence the condition of polyphagous carabid beetles. Ecological Applications 11, 480–488.

Pascual-Villalobos, M., Lacasa, A., Gonzalez, A., Varo, P., and Garcia, M. (2006). Effect of flowering plant strips on aphid and syrphid populations in lettuce. European Journal of Agronomy 24, 182–185.

Patt, J.M., Hamilton, G.C., and Lashomb, J.H. (1997). Foraging success of parasitoid wasps on flowers: interplay of insect morphology, floral architecture and searching behavior. Entomologia Experimentalis et Applicata 83, 21–30.

Pavuk, D.M., Purrington, F.F., Williams, C.E., and Stinner, B.R. (1997). Ground beetle (Coleoptera: Carabidae) activity density and community composition in vegetationally diverse corn agroecosystems. American Midland Naturalist 14–28.

Pease, C.G., and Zalom, F.G. (2010). Influence of non-crop plants on stink bug (Hemiptera: Pentatomidae) and natural enemy abundance in tomatoes. Journal of Applied Entomology 134, 626–636.

Pfannenstiel, R.S., and Yeargan, K.V. (2002). Identification and diel activity patterns of predators attacking Helicoverpa zea (Lepidoptera: Noctuidae) eggs in soybean and sweet corn. Environmental Entomology 31, 232–241.

Pfiffner, L., Luka, H., Schlatter, C., Juen, A., and Traugott, M. (2009). Impact of wildflower strips on biological control of cabbage lepidopterans. Agriculture, Ecosystems & Environment 129, 310–314.

Phillips, B., Prajzner, S., and Gardiner, M. (2010). Quantifying the pollination service supplied to pumpkins (Cucurbita pepo) by multiple bee species in Ohio (Entomological Society of America Annual Meeting).

Pimentel, D., Wilson, C., McCullum, C., Huang, R., Dwen, P., Flack, J., Tran, Q., Saltman, T., and Cliff, B. (1997). Economic and environmental benefits of biodiversity. BioScience 47, 747–757.

Platt, J., Caldwell, J., and Kok, L. (1999). Effect of buckwheat as a flowering border on populations of cucumber beetles and their natural enemies in cucumber and squash. Crop Protect 18, 305–313.

Pontin, D., Wade, M., Kehril, P., and Wratten, S. (2005). Attractiveness of single and multiple species flower patches to beneficial insects in agroecosystems. Ann Appl Biol 148, 39–47.

Prasifka, J.R., Krauter, P.C., Heinz, K.M., Sansone, C.G., and Minzenmayer, R.R. (1999). Predator conservation in cotton: using grain sorghum as a source for insect predators. Biological Control 16, 223–229. 121 Putnam, A., DeFrank, J., and Barnes, J. (1983). Exploitation of allelopathy for weed control in annual and perennial cropping systems. Journal of Chemical Ecology 9, 1001– 1010.

Pywell, R., Warman, E., Hulmes, L., Nuttall, P., Sparks, T., Critchley, C., and Sherwood, A. (2006). Effectiveness of new agri-environment schemes in providing foraging resources for bumblebees in intensively farmed landscapes. Biol Conserv 129, 192–206.

Quantum GIS Development Team (2012). Quantum GIS Geographic Information System (Open Source Geospatial Foundation Project).

R Development Core Team (2013). R: A language and environment for statistical computing (Vienna, Austria: R Foundation for Statistical Computing).

Rebek, E.J., Sadof, C.S., and Hanks, L.M. (2005). Manipulating the abundance of natural enemies in ornamental landscapes with floral resource plants. Biol Control 33, 203–216.

Rodriguez-Saona, C., Blaauw, B.R., and Isaacs, R. (2012). Manipulation of natural enemies in agroecosystems: habitat and semiochemicals for sustainable insect pest control. Integrated Pest Management and Pest Control, Current and Future Tactics. Soloneski S, Larramendy ML (Eds.), InTech Publishing, ISBN 978–953.

Rondon, S.I., Cantliffe, D.J., and Price, J.F. (2003). Anasa tristis (Heteroptera: Coreidae) development, survival and egg distribution on beit alpha cucumber and as prey for Coleomegilla maculata (Coleoptera: Coccinellidae) and Geocoris punctipes (Heteroptera: Lygaeidae). Florida Entomologist 86, 488–490.

Roschewitz, I., Hücker, M., Tscharntke, T., and Thies, C. (2005). The influence of landscape context and farming practices on parasitism of cereal aphids. Agriculture, Ecosystems & Environment 108, 218–227.

Rusek, J. (1998). Biodiversity of Collembola and their functional role in the ecosystem. Biodiversity and Conservation 7, 1207–1219.

Schell, S., C. (1943). The biology of Hadronotus ajax (Hymenoptera - Scelionidae), a parasite in the eggs of squash-bug (Anasa trists DeGreer). Annals of the Entomological Society of America 36, 625–635.

Schellhorn, N.A., Bellati, J., Paull, C.A., and Maratos, L. (2008). Parasitoid and moth movement from refuge to crop. Basic and Applied Ecology 9, 691–700.

Scheper, J., Holzschuh, A., Kuussaari, M., Potts, S.G., Rundlöf, M., Smith, H.G., and Kleijn, D. (2013). Environmental factors driving the effectiveness of European agri- environmental measures in mitigating pollinator loss – a meta-analysis. Ecology Letters 16, 912–920.

122 Seeley, T.D. (2009). The wisdom of the hive: the social physiology of honey bee colonies (Harvard University Press).

Shahjahan, M. (1968). Effect of diet on the longevity and fecundity of the adults of the tachinid parasite Trichopoda pennipes pilipes. Journal of Economic Entomology 61, 1102–1103.

Shuler, R.E., Roulston, T.H., and Farris, G.E. (2005). Farming practices influence wild pollinator populations on squash and pumpkin. Journal of Economic Entomology 98, 790–795.

Smith, C. (2012). Testing an interference competition hypothesis to explain the decline of the convergent lady beetle, Hippodamia convergens (Coleoptera: Coccinellidae), in Ohio (The Ohio State University).

Smith, H.A., Chaney, W.E., and Bensen, T.A. (2008). Role of Syrphid larvae and other predators in suppressing aphid infestations in organic lettuce on California’s central coast. Journal of Economic Entomology 101, 1526–1532.

Smyth, R. (2011). Personal correspondence discussing how to dissect cucumber beetles for developing parasitoids.

Smyth, R.R., and Hoffman, M.P. (2010). Seasonal incidence of two co-occuring adult parasitoids of Acalymma vittatum in New York State: Centistes (Syrrhizus) diabroticae and Celatoria setosa. BioControl 55, 219–228.

Snyder, W.E., and Wise, D.H. (1999). Predator interference and the establishment of generalist predator populations for biocontrol. Biol Control 15, 283–292.

Snyder, W.E., and Wise, D.H. (2000). Antipredator behavior of spotted cucumber beetles (Coleoptera: Chrysomelidae) in response to predators that pose varying risks. Environmental Entomology 29, 35–42.

Southwick, E., and Southwick, L. (1992). Estimating the economic value of honey bees (Hymenoptera: Apidae) as agricultural pollinators in the United States. Journal of Economic Entomology 85, 621–633.

Stanghellini, M.S., Ambrose, J.T., and Schultheis, J.R. (1998). Using commercial bumble bee colonies as backup pollinators for honey bees to produce cucumbers and watermelons. HortTechnology 8, 590–594.

Steffan-Dewenter, I., Munzenberg, U., Burger, C., Thies, C., and Tscharntke, T. (2002). Scale-dependent effects of landscape context on three pollinator guilds. Ecology 83, 1421–1432.

Steffan-Dewenter, I., Potts, S.G., and Packer, L. (2005). Pollinator diversity and crop pollination services are at risk. TRENDS in Ecology and Evolution 20, 651–652. 123 Stokstad, E. (2007). The case of the empty hives. Science 316, 970–972.

Stougaard, R.N., Kapusta, G., and Roskamp, G. (1984). Early preplant herbicide applications for no-till soybean (Glycine max) weed control. Weed Science 32, 293–298.

Stubbs, C.S., and Drummond, F.A. (2001). Bombus impatiens (Hymenoptera: Apidae): an alternative to Apis mellifera (Hymenoptera: Apidae) for lowbush blueberry pollination. Journal of Economic Entomology 94, 609–616.

Sullivan, P. (2002). Rye as a cover crop (ATTRA).

Summers, C. (1976). Population fluctuations of selected arthropods in alfalfa: influence of two harvesting practices. Environmental Entomology 5, 103–110.

Sumner, D., and Boriss, H. (2006). Bee-conomics and the leap in pollination fees. Giannini Foundation of Agricultural Economics 9, 9–11.

Sutherland, J.P., Sullivan, M.S., and Poppy, G.M. (2001). Distribution and abundance of aphidophagous hoverflies (Diptera: Syrphidae) in wildflower patches and field margin habitats. Agricultural and Forest Entomology 3, 57–64.

Szendrei, Z., Greenstone, M.H., Payton, M.E., and Weber, D.C. (2010). Molecular gut- content analysis of a predator assemblage reveals the effect of habitat manipulation on biological control in the field. Basic and Applied Ecology 11, 153–161.

Tepedino, V.J. (1981). The pollination efficiency of the squash bee (Peponapis pruinosa) and the honey bees (Apis mellifera) on summer squash (Cucurbita pepo). J Kansas Entomol Soc 54, 359–377.

Thies, C., and Tscharntke, T. (1999). Landscape structure and biological control in agroecosystems. Science 285, 893–895.

Thies, C., Steffan-Dewenter, I., and Tscharntke, T. (2003). Effects of landscape context on herbivory and parasitism at different spatial scales. Oikos 101, 18–25.

Thies, C., Roschewitz, I., and Tscharntke, T. (2005). The landscape context of cereal aphid–parasitoid interactions. Proc. R. Soc. B 272, 203–210.

Thies, C., Steffan-Dewenter, I., and Tscharntke, T. (2008). Interannual landscape changes influence plant–herbivore–parasitoid interactions. Agriculture, Ecosystems & Environment 125, 266–268.

Thomas, C., and Marshall, E. (1999). Arthropod abundance and diversity in differently vegetated margins of arable fields. Agriculture, Ecosystems and Environment 72, 131– 144.

124 Thomas, C.F.G., Parkinson, L., and Marshall, E.J.P. (1998). Isolating the components of activity-density for the carabid beetle Pterostichus melanarius in farmland. Oecologia 116, 103–112.

Thomas, M.B., Wratten, S.D., and Sotherton, N.W. (1991). Creation of “island” habitats in farmland to manipulate populations of beneficial arthropods: predator densities and emigration. Journal of Applied Ecology 28, 906–917.

Thomson, J., and Goodell, K. (2001). Pollen removal and deposition by honeybee and bumblebee visitors to apple and almond flowers. J Appl Ecol 38, 1032–1044.

Tillman, P.G. (2010). Parasitism and predation of stink bug (Heteroptera: Pentatomidae) eggs in Georgia corn fields. Environmental Entomology 39, 1184–1194.

Tilman, D. (1978). Cherries, Ants and Tent Caterpillars: Timing of Nectar Production in Relation to Susceptibility of Caterpillars to Ant Predation. Ecology 59, 686–692.

Toepfer, S., Walsh, G.C., Eben, A., Alverez-Zagoya, R., Haye, T., Zhang, F., and Kuhlmann, U. (2008). A critical evaluation of host ranges of parasitoids of the subtribe Diabroticina (Coleoptera: Chrysomelidae: Galeucinae: Luperini) using field and laboratory host records. Bicontrol Sci Tech 18, 483–504.

Tscharntke, T., Tylianakis, J.M., Rand, T.A., Didham, R.K., Fahrig, L., Batáry, P., Bengtsson, J., Clough, Y., Crist, T.O., Dormann, C.F., et al. (2012). Landscape moderation of biodiversity patterns and processes - eight hypotheses. Biological Reviews 87, 661–685.

Tuell, J.K., Fiedler, A.K., Landis, D., and Isaacs, R. (2008). Visitation by wild and managed bees (Hymenoptera: Apoidea) to eastern U.S. native plants for use in conservation programs. Environ Entomol 37, 707–718. vanEngelsdorp, D., Cox-Foster, D., and Frazier, M. (2008). Research Continues As Honey Bee Losses Rise In U.S., Fall In Pa.

Wäckers, F. (2004). Assessing the suitability of flowering herbs as parasitoid food sources: flower attractiveness and nectar accessibility. Biol Control 29, 307–314.

Walters, S.A., and Taylor, B.H. (2006). Effects of honey bee pollination on pumpkin fruit and seed yield. Hortscience 41, 370–373.

Walton, N., and Isaacs, R. (2011). Influence of native flowering plant strips on natural enemies and herbivores in adjacent blueberry fields. Environ Entomol 40, 697–705.

Way, M.J., Cammell, M.E., and Paiva, M.R. (1992). Studies on egg predation by ants (Hymenoptera: Formicidae) especially on the eucalyptus borer Phoracantha semipunctata (Coleoptera: Cerambycidae) in Portugal. Bulletin of Entomological Research 82, 425–432. 125 Welty, C. (2008). Cucumber beetle management by mass trapping (Ohio State University Extension).

Westphal, C., Steffan-Dewenter, I., and Tscharntke, T. (2003). Mass flowering crops enhance pollinator densities at a landscape scale. Ecology Lett 6, 961–965.

Whitaker Jr, J.O. (1995). Food of the big brown bat Eptesicus fuscus from maternity colonies in Indiana and Illinois. American Midland Naturalist 346–360.

White, A.J., Wratten, S.D., Berry, N.A., and Weigman, U. (1995). Habitat manipulation to enhance biological control of brassica pests by hover flies (Diptera: Syrphidae). Journal of Economic Entomology 88, 1171–1176.

Wien, H. (1997). The cucurbits: cucumber, melon, squash and pumpkin. In The Physiology of Vegetable Crops, (New York (USA): CAB International, Oxford), pp. 345–386.

Williams, J.L., and Wise, D.H. (2003). Avoidance of wolf spiders (Araneae: Lycosidae) by striped cucumber beetles (Coleoptera: Chrysomelidae): laboratory and field studies. Environmental Entomology 32, 633–640.

Williams, R., Fickle, D., Michel, A.P., and Goodell, K. (2009). Pumpkin pollinator: biology and behaviour of the squash bee (Peponapis pruinosa (Say)) (The Ohio State University Extension).

Winfree, R., and Kremen, C. (2009). Are ecosystem services stabilized by differences among species? A test using crop pollination. Proc R Soc Lond B 276, 229–237.

Winfree, R., Williams, N.M., Dushoff, J., and Kremen, C. (2007a). Native bees provide insurance against ongoing honey bee losses. Ecology Letters 10, 1105–1113.

Winfree, R., Griswold, T., and Kremen, C. (2007b). Effect of human disturbance on bee communities in a forested ecosystem. Conserv Biol 21, 213–223.

Winfree, R., Williams, N.M., Gaines, H., Ascher, J.S., and Kremen, C. (2008). Wild bee pollinators provide the majority of crop visitation across land-use gradients in New Jersey and Pennsylvania, USA. Journal of Applied Ecology 45, 793–802.

Woltz, J.M., Isaacs, R., and Landis, D.A. (2012). Landscape structure and habitat management differentially influence insect natural enemies in an agricultural landscape. Agriculture, Ecosystems & Environment 152, 40–49.

Zehnder, G., Gurr, G.M., Kühne, S., Wade, M.R., Wratten, S.D., and Wyss, E. (2007). Arthropod pest management in organic crops. Annual Review of Entomology 52, 57–80.

Zuur, A., and Saveliev, I. (2012). Zero Inflated Models and Generalized Linear Mixed Models with R (Highland Statistics Ltd.). 126 Zuur, A., Ieno, E., Walker, N., Saveliev, A., and Smith, G. (2009). Mixed Effects Models and Extensions in Ecology with R (Springer).

Zuur, A.F., Ieno, E.N., and Elphick, C.S. (2010). A protocol for data exploration to avoid common statistical problems. Methods in Ecology and Evolution 1, 3–14.

(2013). Vegetables 2012 Summary (USDA, National Agricultural Statistics Service).

127