Replace This with the Actual Title Using All Caps

DIFFERENTIAL IMPACTS OF POLLEN QUALITY AND MICROBIAL COMMUNITIES ON GENERALIST AND SPECIALIST BEES VISITING A SHARED FOOD RESOURCE

A Dissertation Presented to the Faculty of the Graduate School of Cornell University In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

by Kristen Brochu May 2018

DIFFERENTIAL IMPACTS OF POLLEN QUALITY AND MICROBIAL COMMUNITIES ON GENERALIST AND SPECIALIST BEES VISITING A SHARED FOOD RESOURCE

Kristen Brochu, Ph. D. Cornell University 2018

Abstract: There is mounting evidence for declines in both managed and wild pollinator species, necessitating support for wild bee populations in order to maintain pollination services. Solitary, ground-nesting bees comprise the vast majority of bees in the world; however, these important pollinators are relatively understudied and we know little about how they are impacted by resource availability, pollen quality, and microbial associates. My dissertation research uses plants of the genus Cucurbita as a case study to assess differential health concerns associated with generalist (Bombus impatiens and Apis mellifera) and specialist (Peponapis pruinosa) pollinators on a shared resource. I begin by examining how cucurbit plant chemistry varies across plant tissue types and varieties and how this variation impacts the foraging choices of both generalist and specialist bees in an agricultural system. Next, I assess the physiological costs associated with digesting cucurbit pollen for these generalist and specialist bees. Finally, I explore the microbial community associated with a specialist ground-nesting bee’s brood cells, evaluate how this community changes over space and time, and assess the most likely routes of microbial colonization into brood cells.

BIOGRAPHICAL SKETCH

Kristen was born on March 11, 1986 to Joanne Labre and Kenneth Brochu. Growing up in Montreal, she has always been fascinated by the natural world. From an early age she loved learning and was relentlessly curious about the world around her. Encouraged by her parents, she completed a BSc in Wildlife Biology at McGill University, where her love of biology was reinforced by excellent field courses and research opportunities. It was at the Macdonald Campus that she discovered her passion for insects and science. Several wonderful mentors fostered her enthusiasm for research, which led her to pursue an MSc in Evolutionary Biology at the University of Toronto. Here she developed international collaborations and was captivated by the diversity she observed during her tropical fieldwork. While her foray into fish biology taught her many valuable skills, her interest in insects won out and she began her PhD in Entomology at Cornell University. Throughout her education, her passion for sharing science with others has only grown with many wonderful opportunities to be involved in science communication and outreach activities. When not travelling or spending quality time with her collecting net, Kristen can usually be found tending to her insect pets, who are the stars of her outreach program.

In memory of Terry Wheeler, my first mentor

ACKNOWLEDGMENTS

It has been an incredible privilege to work with so many wonderful people during the course of my PhD. First, I’d like to thank my advisor, Bryan Danforth, for his support throughout the ups and downs of my various projects. Bryan was always ready to share his amazing wealth of knowledge about bees and I can only aspire to one day know as much. His great enthusiasm for natural history is infectious and I have many fond memories of collecting bees with Bryan. I would also like to thank my special committee, Andre Kessler and Anurag Agrawal, for their invaluable expertise which truly enriched my graduate experience. Andre was critical in helping me to develop my ideas and brainstorming with him was always enlightening, motivating, and best of all, fun. Anurag’s astute insights improved the quality of my work and his talent for seeing the ‘big picture’ perspective helped me to hone my research interests. The Entomology Department as a whole has been a wonderful academic home for the last six years. Thank you to Scott McArt for generously giving me access to his lab space and equipment, and to Nelson Milano for helping me to set up my microcolonies. Collaborating with the McArt lab has been a wonderful experience, both in and out of the darkroom. Thank you to Kyle Wickings for offering helpful advice about soil sampling and lending equipment to help me accomplish it. I loved spending time up in Geneva and learning more about all that’s underfoot. Thank you to Jessica Petersen and Brian Nault for sharing data and ideas about cucurbit pollinators. Thank you to Michael Mazourek for guidance on growing squash and pumpkin and for his wonderful work on other aspects of cucurbit biology. Thank you to Rob Raguso for many wonderful conversations about pollinators and to Anne Hajek for her fantastic suggestions on working with fungi. Thank you to Laura Harrington

for her advice on professional development and for generally being a great role model. Thank you to Jason Dombroskie for always helping me to ID odd bugs, and for bringing a bit of Canadian culture to Cornell. Thank you to Linda Rayor for introducing me to entomological outreach at Cornell, for helping me to hone my presentation skills, and above all, for all the bugs. I had so much fun taking and TAing the Naturalist Outreach class and continuing to do outreach afterwards! It will continue to be a big part of my commitment to sharing my love of science. Thank you to Cheryl Gombas, Lisa Westcott, Lisa Marsh, Stephanie Westmiller, Amy Arsenault, Rita Stucky, Laurie McCall, and Alicia Caswell for patiently answering all my weird questions, consistently solving strange administrative problems, and just generally making my life so much easier. I don’t know what I would have done without you! Further afield at Cornell, I would like to thank the staff at the Cornell Greenhouse facilities for all their help growing my plants and for their patience with the cucurbit jungle that I cultivated. I would also like to thank the Statistical Consulting Unit for confirming that although my analyses were always more complicated than they should be (because live animals just do weird things), they were statistically correct. Thank you to Susi Varvayanis, Colleen McLinn, Kimberly Williams, and Derina Samuel for making me a better communicator and teacher. I feel so fortunate that I was able to learn from all of you and I look forward to sharing the skills I gained at all my future academic institutions. I want to thank past and present members of the Danforth lab for their support, encouragement, and camaraderie during my time here. Thanks to Shannon Hedtke, EJ Blitzer, Jason Gibbs, Margarita Lopez-Uribe, Mia Park, and Laura Russo for helping me to focus my early research interests and providing wonderful insights into my experiment designs. I would particularly like to thank Elizabeth Murray, Heather Grab, Maria van Dyke, Mary Centrella, Silas Bossert and Kass Urban-Mead for

vii

helpful comments on my data analysis, results, and manuscripts. It has been a joy to work with all of you and I’m a much better scientist for your feedback. I want to thank everyone who helped me with fieldwork, especially Maria van Dyke without whom my days in the field would have been much less fun and much less productive. You are absolutely wonderful and I will miss our long talks in the dirt. Thank you to all the squash and pumpkin growers in the Finger Lakes region who let me look for bees and dig giant holes in their fields. All the sites I visited provided a beautiful backdrop for so many lovely sunrises. Thank you also to NSERC, Cornell University, the Entomology Department, the Griswold Endowment, the Rawlins Endowment, and the Hatch Grant for funding my projects and sending me to conferences. Thank you to all my friends who have made my time here so enjoyable. Thank you to Laura, Heather, Elizabeth, Silas, Leticia, Maria, Mary, Kalyn, John, Kass, Joanna, Zoe, Anyi, Nelson, Ashley, Tim, and Maria for many fantastic times in Ithaca. I have so many great memories with you all and I look forward to making many more. Thank you to Roger & Carmela for all the movies. Thank you to Val for always coming to visit me. Thank you to Suji, Jackie, Miao, Kin, and Alice for making my visits home so much fun. Thank you to my family for always supporting me and for listening to me talk endlessly about bugs. I would not be here without your encouragement. Your astute questions have always made my presentations so much better. Please keep sending me photos of bugs to ID. A little less blurry if you can. Thank you to my wonderful partner Paul. For everything.

viii

TABLE OF CONTENTS

BIOGRAPHICAL SKETCH…………………………………………………………….. iv

ACKNOWLEDGEMENTS………………………………………………………………. vi

LIST OF FIGURES……………………………………………………………………... x

LIST OF TABLES……………………………………………………………………… xi

CHAPTER 1. POLLEN DEFENSES NEGATIVELY IMPACT FORAGING AND FITNESS IN A

GENERALIST BEE………………………………………………………………………. 1

CHAPTER 2. CHEMICAL ECOLOGY OF POLLINATOR DIETARY CHOICES……………... 36

CHAPTER 3. COMMUNITY ECOLOGY OF MICROBES IN THE NESTS OF A SOLITARY,

OLIGOLECTIC, GROUND-NESTING BEE………………………………………………… 59

LIST OF FIGURES

CHAPTER 1

FIGURE 1. Cucurbita pollen use in field and lab experiments……………………….. 13

FIGURE 2. Fitness effects on adult B. impatiens fed various diet treatments………... 15

FIGURE S1. Mass loss due to evaporation over time by treatment…………………... 34

FIGURE S2. Effect of average weight on sucrose consumption……………………… 35

FIGURE S3. Effect of treatment on pollen efficiency………………………………… 35

CHAPTER 2

FIGURE 1. Comparison of chemistry across tissue types…………………………….. 43

FIGURE 2. Non-metric multidimensional scaling plot (NMDS) to compare compound bouquet composition in leaves, nectar, petals, and pollen (all cultivars pooled)……. 44

FIGURE 3. Non-metric multidimensional scaling plot (NMDS) to compare compound bouquet composition in cultivars within leaves, petals, nectar, and pollen…………. 45

FIGURE 4. Bee visitation to Cucurbita pepo cultivars……………………………….. 47

CHAPTER 3

FIGURE 1. Pareto chart showing the overall abundance of core bacterial and fungal communities of pollen provisions…………………………………………………… 68

FIGURE 2. Comparison of alpha diversity metrics across sample types……………... 69

FIGURE 3. Non-metric multidimensional scaling (NMDS) plot comparing microbial communities of pollen provisions…………………………………………………… 71

FIGURE 4. Non-metric multidimensional scaling (NMDS) plot comparing microbial communities across sample type…………………………………………………….. 72

FIGURE 5. Overlap between microbial communities in different sample types……... 73

LIST OF TABLES

CHAPTER 1

TABLE 1. Summary of diet treatments provided to Bombus impatiens.………………. 8

TABLE S1. Varieties of Cucurbita pepo used in the cucurbit pollen diet mix.………. 33

TABLE S2. Detailed results for all statistical tests performed……………………….. 33

TABLE S3. Summary of analysis for pollen mass lost to evaporation……………….. 34

CHAPTER 2

TABLE 1. Varieties of Cucurbita pepo used in bee choice and chemistry assessment. 42

CHAPTER 3

TABLE 1. Adonis and anosim results………………………………………………… 70

TABLE S1. Locality information for nest aggregations……………………………… 91

TABLE S2. Dunn’s rest results for alpha diversity metrics…………………………... 91

Pollen defenses negatively impact foraging and fitness in a generalist

bee

Kristen Brochu *a, Maria van Dyke a, Nelson Milano a, Jessica Petersen b,

Scott McArt a, Brian Nault c, Andre Kessler d, and Bryan Danforth a

a Department of Entomology, Cornell University, Ithaca, USA

b Department of Natural Resources, Madelia, MN, USA

c Department of Entomology, New York State Agricultural Experiment Station, Cornell

University, Geneva, USA

d Department of Ecology and Evolutionary Biology, Cornell University, Ithaca, USA

* Corresponding author: [email protected]

Summary

1. Plants may benefit from reducing the fauna of generalist floral visitors if, by doing so,

they increased the fidelity and effectiveness of the remaining pollinators.

2. We observed that Peponapis pruinosa, a specialist bee on cucurbit plants, collected pure

loads of cucurbit pollen in contrast with generalist honey bees and bumble bees which

collected very small amounts of cucurbit pollen, even though all groups of bees visited

cucurbit flowers.

3. Cucurbit flowers have no morphological adaptations to limit pollen collection by bees,

thus we assessed the potential effects of physical and chemical pollen defenses as a likely

method for cucurbit plants to limit pollen loss to generalist pollinators.

4. Bumble bee (Bombus impatiens) microcolonies fed non-cucurbit pollen showed increased

pollen consumption over time, while microcolonies fed natural cucurbit pollen and

crushed cucurbit pollen (to remove physical defenses) showed markedly decreased pollen

consumption over time.

5. Furthermore, bees from microcolonies fed crushed cucurbit pollen had higher proportions

of gut melanization and mortality overall.

6. Microcolonies fed both cucurbit pollen treatments never reared offspring to adulthood,

while all other treatments were more likely to do so, and more larvae were ejected from

microcolonies fed natural cucurbit pollen than in any other treatment.

7. Together, these results suggest that Bombus impatiens workers avoid collecting cucurbit

pollen due to the physiological costs of pollen defenses.

Key-words: Bombus impatiens, Cucurbita, mechanical defenses, microcolonies, nutrition, secondary plant metabolites, squash bee

Introduction

Many studies attribute the vast diversity of bees to their mutualistic pollenivorous lifestyle, yet bee-plant interactions are much more complex than simple mutualisms. The conflict

in bee-plant interactions arises because bees (Apoidea) feed solely on vast quantities of pollen, but vary markedly in effectiveness as pollinators (L. D. Harder & Barclay, 1994; Hargreaves,

Harder, & Johnson, 2009). Thus, bees are more accurately described as highly specialized, extremely efficient, herbivores. Consequently, plants must balance their need to be pollinated against the loss of pollen from foraging bees.

Plants have evolved to limit detrimental pollen loss in various ways. Many plants use volatile cues as exclusive channels of communication to attract specific pollinators, frequently specialists (Cane & Sipes, 2006). Specialist bees that visit a restricted set of plant genera or species are generally assumed to be the most efficient pollinators (Burger, Ayasse, Dötterl,

Kreissl, & Galizia, 2013; Dobson, 1987). Plants have also evolved various morphological adaptations that limit pollen collection, such as hidden anthers (Westerkamp, 1997; Westerkamp

& Claßen-Bockhoff, 2007), or poricidal anthers that require specialized behavior (i.e. buzz- pollination) to release pollen (Buchmann, 1983; L. D. Harder & Barclay, 1994). Finally, plants can also reduce pollen consumption directly through pollen defenses, such as chemical defenses

(Arnold, Idrovo, Arias, Belmain, & Stevenson, 2014; Detzel & Wink, 1993; Eckhardt, Haider,

Dorn, & Müller, 2014; Praz, Müller, & Dorn, 2008; Sedivy, Piskorski, Müller, & Dorn, 2012), mechanical defenses (Lunau, Piorek, Krohn, & Pacini, 2015), and even lack of essential nutrients

(Dobson, 1988; Johri & Vasil, 1961; T. H. Roulston & Cane, 2000; T’ai H. Roulston &

Buchmann, 2000; T’ai H. Roulston, Cane, & Buchmann, 2000; Stanley & Linskens, 1974).

Plants may actually benefit from reducing the fauna of possible floral visitors if, by doing so, they increase the fidelity and effectiveness of the remaining pollinators (L. D. Harder & Barclay,

1994; Lawrence D. Harder & Thomson, 1989).

If plants are chemically or mechanically defending their pollen, we might expect

specialist bee species to have physiological adaptations that allow them to feed exclusively on

their preferred host pollen, but this same pollen source could be indigestible or nutritionally

inadequate for a range of generalist species (Janz & Nylin, 2008; Müller & Kuhlmann, 2008;

Sedivy, Dorn, Widmer, & Müller, 2013; Sedivy, Praz, Müller, Widmer, & Dorn, 2008; Sipes &

Tepedino, 2005). Physiological restrictions could be caused by nutritional requirements not

served by single-plant diets, a lack of mechanisms to deal with pollen defenses that interfere with

digestion, reproduction, or growth processes, or even direct toxic effects (Archer, Pirk, Wright,

& Nicolson, 2014; Arnold et al., 2014; Detzel & Wink, 1993; Dobson & Peng, 1997; Eckhardt et al., 2014; Lunau et al., 2015; Manson & Thomson, 2009; Palmer-Young et al., 2017; Peng, Nasr,

& Marston, 1986; Peng, Nasr, Marston, & Fang, 1985; Praz et al., 2008; Sedivy, Müller, & Dorn,

2011; Suárez-Cervera, Marquez, Bosch, & Seoane-Camba, 1994; Vaudo et al., 2016). These strategies to limit pollen consumption are not mutually exclusive (Haider, Dorn, & Müller,

2014), but little is known about the relative costs of investing in each strategy. Studies on pollen defenses generally consider these defenses in isolation, confounding mechanical or chemical defenses with effects of poor nutrition (Eckhardt et al., 2014; Praz et al., 2008), demonstrating deterrence without actual consumption costs (Kevan & Ebert, 2005; London-Shafir, Shafir, &

Eisikowitch, 2003; Lunau et al., 2015) or, more rarely, toxic effects that do not translate to

changes in foraging behavior (Arnold et al., 2014). Considering how important direct pollen

defenses could be in determining patterns of bee foraging and health, there is surprisingly little

work examining how the nutritional quality of pollen and its direct defenses intersect to impact

bee health and fitness.

Squash and pumpkin (genus Cucurbita) are obligate outcrossing plants with a diverse

pollinator fauna of both host-plant generalists (Bombus, Apis, Melissodes, Lasioglossum,

Agapostemon, and Halictus) and narrow host-plant specialists (Peponapis and Xenoglossa) across their geographic range (Artz, Hsu, & Nault, 2011; Artz & Nault, 2011; Cane & Sipes,

2006; Paul D. Hurd, Linsley, & Whitaker, 1971; Paul David Hurd, Linsley, & Michelbacher,

1974; Julier & Roulston, 2009; Kevan, Mohr, Offer, & Kemp, 1989; Mathewson, 1968; Petersen,

Reiners, & Nault, 2013; Shuler, Roulston, & Farris, 2005). Cucurbit flowers have no morphological adaptations to limit pollen collection, thus mechanical or chemical pollen defenses are likely mechanisms for cucurbit plants to limit pollen loss to generalist, low-fidelity visitors. Cucurbit pollen is large, spiky, has a sticky pollenkitt and contains many compounds, which could all act as defenses (Lundgren, 2009). Cucurbit specialist bees, like Peponapis pruinosa, thrive on a solely cucurbit pollen diet, despite its potential pollen defenses; however, it is unknown if generalist pollinators, such as the common eastern bumble bee (Bombus impatiens), collect large quantities of cucurbit pollen, or are negatively impacted by feeding on a diet of cucurbit pollen. Accordingly, we evaluated the effects of pollen defenses on the foraging behaviour and physiology of specialist and generalist pollinators, which largely overlap in their geographical and seasonal range, in this model system.

Our first objective evaluated the frequency of cucurbit pollen collected by generalists

(honey bees, Apis mellifera; and bumble bees, Bombus spp.), often used for commercial pollination of cucurbits, in the field. We predicted that generalist pollinators would collect few cucurbit pollen grains compared with those from other plant families. Our second objective examined fitness costs associated with a generalist pollinator feeding on cucurbit pollen. We predicted that a generalist would suffer increased fitness costs by consuming cucurbit pollen.

Cucurbit pollen may exhibit three levels of defenses: chemical defenses (secondary plant metabolites), mechanical defenses (large size or spines) and poor nutrition (a lack of essential

nutrients). Our study was designed to distinguish among the effects of specific pollen defenses as

well as the impact of multiple defenses. We predicted that a generalist would suffer reduced

fitness when consuming a diet of multiple pollen defenses. We predicted that bees would incur

these costs as a combination of increased mortality, reduced or inefficient resource utilization,

reduced reproduction, and increased stress responses.

Methods

Assessing Pollen Collection

In the Finger Lakes Region of New York in 2011 and 2012, we supplemented cucurbit

fields (0.5-10 ha) with commercially produced B. impatiens colonies (Koppert Biological

Systems, Inc.) or with locally rented A. mellifera hives. Sampling was conducted from 16 July-

27 August in 2012 in ten fields (n = 4 honey bee supplemented, n = 6 bumble bee supplemented), and from 15 July-21 August in 2013 in ten fields (n = 5 honey bee, n = 5 bumble

bee). Pollen from the corbiculae was collected from ten bees returning to the colonies during

each of three rounds of sampling (total of n = 30 individuals per field). Bees were sampled

between 0600-1200 h, during the time when pumpkin flowers were open. Bees were captured by

aerial net, and pollen from their corbicula was removed, placed into a centrifuge tube and placed

on ice for transport to the lab.

Pollen samples collected from corbiculae were placed in centrifuge tubes and 150µL of

95% ethanol was added. A drop of this mixture was applied to a microscope slide and then

covered with fuschia-dyed glycerin jelly, melted on a hotplate, and sealed with clear nail polish.

For each slide a random subsample of 100 pollen grains was counted under a compound light

microscope and identified to the lowest possible taxonomic classification using a reference

library of pollen created from local concurrently blooming flora.

In the Finger Lakes Region of New York in August 2014 female P. pruinosa (n = 37) were collected while foraging on cucurbit plants from seven sites, six of which were adjacent to

Peponapis nesting aggregations). Bees were washed in ethyl acetate to remove pollen, which was then slide-mounted in Calberla’s fluid (recipe in Supplemental). We performed six randomly selected field of view transects across the slide at 20X magnification, counting all grains except those that were broken or incomplete. Pollen was then identified to the lowest taxonomic rank feasible. For all bee species, pollen data were pooled across all collection sites and collection times to determine the overall percentage of cucurbit pollen collected. Our primary interest was to discern relative percentages of cucurbit pollen collected by each bee species, not to compare these amounts among species.

Microcolony Experimental Protocol

From March-May 2017, we used a microcolony protocol to test whether or not B. impatiens is negatively impacted by consuming cucurbit pollen. This technique has been successfully used in previous studies to assess the effects of diet on bee fitness (Genissel,

Aupinel, Bressac, Tasei, & Chevrier, 2002; Larrere & Couillaud, 1993; Richardson et al., 2015;

Tasei & Aupinel, 2008a, 2008b). Each experiment was replicated using three source colonies of

Bombus impatiens purchased from Biobest Canada Ltd. Leamington, ON, Canada. Each source colony supplied three microcolony replicates for each of five treatments. Microcolonies consisted of five workers taken from a source colony at least 24 hours after adult eclosion and were kept in a growth chamber at 27°C and 60%-80% humidity. We attempted to standardize bee size as much as possible within a microcolony, but limitations due to the number and size of bees eclosing from source colonies on any given day meant that there was variation in the size of individual bees and the average weight of microcolonies. A standard diet of pesticide-free, honey

bee collected pollen (CC Pollen Company High Desert Fresh Raw Bee Pollen Granules) mixed with 30% sucrose solution was provided initially while bees acclimated to the microcolony and established a dominance hierarchy (Genissel et al., 2002; Larrere & Couillaud, 1993). After this period, one bee typically develops into a pseudoqueen capable of laying eggs.

Nectar (30% sucrose solution in 1/2oz cups) and pollen (mixed with 30% sucrose solution according to treatment, detailed below, in 0.3g portions) were provided ad libitum.

Treatments were produced in bulk before the start of the experiment and then stored at -20°C until use. Cucurbit pollen was collected from plants grown in a greenhouse at Cornell University

(in July-August 2016 and January-February 2017), in order to acquire pollen from non-herbivore damaged, pesticide-free plants. A large number of plants were required to obtain sufficient pollen, so several varieties of Cucurbita pepo pollen were mixed for the cucurbit pollen diet (see

Table S1 for proportions and varieties used). We used five treatments (Table 1) to differentiate between the effects of chemical and mechanical defenses as well as nutrition: 1) Diet Control, 2)

Solvent Control, 3) Added Chemistry, 4) Crushed Cucurbit, and 5) Natural Cucurbit.

Table1. Summary of diet treatments provided to Bombus impatiens. Each microcolony consisted of 5 worker bees. Treatments were replicated three times for each source colony for a total of 9 replicates across colonies.

Treatment Treatment Contents Possible Possible Possible Name Chemical Poor Diet Mechanical Defenses Defenses Diet Standard pollen diet + 30% sucrose Control Solvent Standard pollen diet + 5%DMSO in 30% sucrose Control Added Standard pollen diet + extracted cucurbit chemistry X Chemistry dissolved in 5% DMSO in 30% sucrose Crushed Homogenized cucurbit pollen in 30% sucrose X X Cucurbit Natural Unmanipulated cucurbit pollen in 30% sucrose X X X Cucurbit

The diet control treatment was a multi-floral diet with no pollen defenses that is sufficient for B. impatiens development and the solvent control contained the solvent on the standard pollen diet to control for potential negative effects of the solvent when assessing the chemistry treatment. Control and Solvent treatment pollen consisted of the standard pollen diet mixed with

30% sucrose solution and 5% DMSO in 30% sucrose solution, respectively. The chemistry treatment contained chemicals extracted from cucurbit pollen into a solvent on the standard pollen diet, which eliminated potential mechanical defenses and should have been nutritionally sufficient, but may have contained chemical defenses. The cucurbit chemical extract was obtained by bead homogenization (using a FASTPREP for 45s at 6.5m/s, twice) of the cucurbit pollen diet into methanol. The methanol was then evaporated using a vacuum centrifuge and the chemical extract was resuspended in a 5% DMSO solution in 30% sucrose solution. This solution was then mixed with a standard pollen diet in a 1:1 cucurbit pollen to standard pollen mass ratio. The crushed treatment consisted of crushed cucurbit pollen to eliminate mechanical defenses, but still retained chemicals that could act as chemical defenses, and additionally, may not have been nutritionally sufficient for B. impatiens development. Cucurbit pollen was homogenized (using a FASTPREP for 30s at 4.0m/s, twice) in water (to remove mechanical defenses), with the water evaporated using a nitrogen evaporator (to retain chemical defenses), then mixed with 30% sucrose solution to obtain the Crushed cucurbit treatment. The natural treatment consisted of pure cucurbit pollen, which retained physical features that could act as mechanical defenses, chemicals that could act as chemical defenses, and may not have been nutritionally sufficient for B. impatiens development. Natural treatment pollen consisted of unmanipulated cucurbit pollen diet mixed with 30% sucrose solution. Treatment pollen was added two to six days after the microcolony was formed.

Nectar refills were monitored to assess how much sucrose the bees were consuming.

Pollen was weighed daily to record bee consumption. Pollen for each treatment was also

maintained without exposure to bees and weighed daily in order to control for weight loss due to

evaporation. The following measures of fitness were recorded daily: 1) number of dead workers,

2) number of larval cells produced, and 3) number of ejected larvae. Microcolonies were

terminated once the first adult offspring eclosed or at 50 days from inception, whichever came

first. Following termination, all adult bees including newly eclosed offspring were weighed and

then euthanized by freezing. These bees were later dissected to observe gut morphology changes

as a result of the treatments. Melanization and hindgut expansion were assessed qualitatively

(presence/absence) and quantitatively, comparing the size of the affected area using the image

analysis software, Fiji (Schindelin et al., 2012; Schindelin, Rueden, Hiner, & Eliceiri, 2015).

Microcolonies were allowed to remain intact for 24 hours following termination to allow

for the emergence of any additional adult offspring, and following this incubation period they

were assessed for total reproductive output. All remaining larval offspring were counted and

weighed, then euthanized by freezing.

Statistical Analyses

All analyses were conducted in R version 3.3.3 (https://www.r-project.org/) using the following packages: AICcmodavg, car, coxme, emmeans, ggplot2, lattice, lme4, MuMIN, multcomp, plyr, reshape2, scales, survival and the HighStat Library (Bartoń, 2017; Bates,

Mächler, Bolker, & Walker, 2015; Fox et al., 2017; Hothorn, Bretz, & Westfall, 2008; Lenth,

Love, & Herve, 2018; Mazerolle, 2017; Sarkar, 2017; Therneau, 2015; Therneau & Lumley,

2017; Wickham, 2007, 2011, 2017; Wickham & Chang, 2016; Zuur, Ieno, & Elphick, 2010).

Individual functions are cited in the text.

Results

Pollen Collection

All P. pruinosa individuals carried pure cucurbit pollen loads, while both generalist

species (B. impatiens and A. mellifera) collected little cucurbit pollen (Figure 1a). A pollen load

was considered pure when it consisted of 90% or more of one type of pollen (Sipes & Tepedino

2005). With a minimum of 93% cucurbit grains, and an average of 97%, with the remainder

made up of other Cucurbitaceae pollen, P. pruinosa well exceeded the minimum threshold to be

considered cucurbit specialists. In contrast, cucurbit pollen made up a small percentage of the

total pollen collected by the generalist A. mellifera and B. impatiens, only 2.0% and 0.4%

respectively. Pollen types represented by less than 3% of the sample are generally considered to

be ‘accidental contact’ and are not normally recorded as host-plant pollen (Edens-Meier, Joseph,

Arduser, Westhus, & Bernhardt, 2011; Russo & Danforth, 2017). Only 2.0% and 0.3% of A. mellifera and B. impatiens bees sampled, respectively, had cucurbit pollen in quantities greater than 3%, suggesting that few bees are actively collecting cucurbit pollen.

Effect of mechanical vs. chemical defenses on colony fitness

All analyses used the ‘lmer’ function (Bates et al., 2015) to fit linear mixed effects models, unless otherwise indicated. We confirmed the absence of multicollinearity using the function ‘corvif’ (Zuur et al., 2010). For each analysis we compared all possible models

(including treatment and average worker weight as fixed effects and replicate within source colony as random effects) and selected the model with the lowest AICc score using the ‘dredge’ and ‘AICc’ functions (Bartoń, 2017). Best models are described in Table S2.

Resource Utilization

Pollen consumption was standardized to consumption per bee for each day of the

experiment and compared over time using a repeated measures mixed effects regression with a fully crossed fixed effects design. We used the same method to assess the mass lost to evaporation and used these results to correct our measures of pollen consumption. Day of experiment was the repeated measure and was added to these models as a fixed effect.

We found that the Crushed and Natural cucurbit pollen lost more mass due to evaporation than the other treatments, but only on day one (for more details of this analysis see Table S3,

Figure S1). Mass loss due to evaporation was not significant from days two through seven. We thus corrected daily pollen consumption for microcolonies fed the Crushed and Natural cucurbit treatments by the mean value lost to evaporation only on the first day of replacement. We found a significant effect of diet treatment (F(4,1274) = 230.305, p < 0.001), day (F(1,1) = 26.776, p =

0.049), average weight (F(1,78) = 6.701, p = 0.012), diet treatment by day (F(4,1330) = 60.095, p <

0.001), diet treatment by average weight (F(4,1394) = 42.691, p < 0.001), and diet treatment by day by average weight (F(4,290) = 16.594, p = 0.001) on pollen consumption by B. impatiens over time

(Figure 1b).

Post-hoc analyses indicated that B. impatiens pollen consumption increased over time for all diet treatments except the Crushed and Natural treatments, and varied across diet treatments.

Microcolonies fed the Control, Solvent, and Chemistry treatments increased their pollen consumption over time for all average weights, except for lighter microcolonies fed the Solvent treatment and heavier microcolonies fed the Control treatment. Microcolonies fed the Crushed and Natural treatments tended to decrease pollen consumption over time. As the experiment progressed, heavier microcolonies fed the Solvent and Chemistry treatment increased their pollen consumption more than lighter microcolonies. The reverse trend was true for microcolonies fed the Control treatment, whereby the lighter microcolonies increased their consumption more over

time than heavier microcolonies. This is likely due to two heavy microcolonies fed the Control treatment that never produced adult offspring, while small microcolonies increased pollen use to provide for their offspring. Microcolonies with higher average weights consumed more nectar per bee per day (F(1,14) = 16.001, p = 0.001, Figure S2).

Figure 1. Cucurbita pollen use in field and lab experiments. a) Percentage of Cucurbita pollen grains observed in typical pollen loads collected by three bee species located in cucurbit fields. These field studies were conducted in New York in 2011 and 2012 (A. mellifera and B. impatiens) and 2014 (P. pruinosa). Error bars represent standard errors. b) Pollen consumption by B. impatiens over time. Each horizontal line represents the consumption values for a single microcolony over time. The width of the line indicates the average value of pollen consumption for each bee in the microcolony on that date. The colour of each line indicates the average weight of the microcolony (which does not change over time).

Asterisks indicate when a microcolony was terminated early due to the production of an adult offspring.

Mortality

We used the ‘coxme’ function (Therneau, 2015) to fit a Cox Proportional Hazards mixed effects model, to assess the probability of individual bee mortality over 27 days. This time period is before the first microcolony was terminated due to the production of an adult offspring and avoids biasing our analysis with differential termination dates. Treatment and weight did not have significant effects on mortality over time, but there was a significant treatment by weight

2 interaction (Χ (4)=13.329, p = 0.01). Compared to bees fed the Control treatment, there was a

trend (p < 0.1) for increased mortality in bees fed the Solvent treatment and in larger bees fed the

Crushed treatment. There was a trend (p < 0.1) for decreased mortality risk in bees fed the

Crushed treatment, and in larger bees fed on both the Solvent and Natural treatments (Figure

3A). Using the ‘emmeans’ function (Lenth et al., 2018; Searle, Speed, & Milliken, 1980), we

then assessed if the predicted mortality hazard was greater than zero after 27 days for each

treatment across seven weight classes (minimum, 5th percentile, 25th percentile, median, 75th

percentile, 95th percentile, maximum), with a Holm-Bonferroni (Holm, 1979) correction for

multiple comparisons (Figure 3B). Smaller bees (below the 25th percentile) fed the Solvent and

Natural treatment had a higher mortality risk (p < 0.05), and a trend (p < 0.1) for higher mortality

risk at the 25th percentile weight. Bees fed the Control treatment had a trend (p < 0.1) for higher mortality risk for weights at and below the median. Bees fed the Crushed treatment had a trend

(p < 0.1) for higher mortality risk for weights at and above the 75th percentile.

We also found variation in the overall proportion of individual bee mortality among microcolonies in different diet treatments (F(4,32) = 20.327, p < 0.001). A post-hoc Tukey test

showed that microcolonies fed the Crushed treatment exhibited higher overall mortality than

microcolonies fed all other treatments (p < 0.001), while microcolonies fed the Natural treatment

exhibited significantly and marginally higher overall mortality than microcolonies fed the

Solvent (p < 0.05) and Chemistry (p < 0.1) treatments, respectively, but did not differ from

microcolonies fed the Control diet (Figure 3C).

Figure 2. Fitness effects on adult B. impatiens fed various diet treatments. a) Log Mortality Risk based on

Cox Proportional Hazards Mixed-Effects model coefficients with standard error bars. Error bars that do

not cross the red dotted line indicate effects significant at p < 0.1. b) Overall proportion of mortality in

each microcolony across treatments with standard error. c) Percentage of microcolonies producing adult

offspring across treatments. d) Average number of larvae ejected per day across treatments.

Reproduction

While all microcolonies produced eggs, we found that treatment significantly affected the

probability of a microcolony rearing their offspring to adulthood (Fisher’s Exact test, p < 0.001).

Post-hoc Fisher exact pairwise comparisons showed that microcolonies fed Control, Solvent, and

Chemistry treatments were more likely to produce adult offspring than microcolonies fed

Crushed and Natural cucurbit treatments, which never produced adults (Figure 4A). Because

microcolonies fed Crushed and Natural cucurbit treatments never produced adult offspring, we restricted all following analyses of reproduction to microcolonies that produced adult offspring.

We found no significant effects of treatment or average weight on the number of days to the first eclosed offspring, the number of eclosed offspring per bee per day, or the average eclosed offspring weight.

Stress Responses

When larvae are sick or adult bees are stressed, they will eject larvae from their brood cells and discard them in the trash (where they defecate) or smear the larvae along the wall of the microcolony chamber (Genissel et al., 2002; Tasei & Aupinel, 2008a). We found a significant effect of diet treatment (F(4,32) = 6.413, p = 0.001, Figure 4B) on the number of ejected larvae per day, with post-hoc Tukey analyses showing that microcolonies fed the Natural cucurbit treatment ejected more larvae than microcolonies in all other treatments (p < 0.005). Another stress response evaluated was pollen diet efficiency, which is the number of eclosed offspring divided by the total pollen consumed per microcolony. Therefore, lower diet efficiency indicates that bees need to consume more pollen to produce each eclosed offspring, and could be due to an inability to adequately digest pollen or assimilate nutrients from their diet. We found a significant effect of treatment (F(2,14) = 5.310, p = 0.019, Figure S3) on pollen diet efficiency, with microcolonies fed the Chemistry treatment having a lower pollen efficiency than bees in the

Control treatment (p < 0.005); however, they did not differ from the Solvent treatment, suggesting a possible subtle effect of the chemistry + solvent combination that merits further investigation. This analysis was restricted to microcolonies that produced adult offspring.

Discussion

Our study is the first to explicitly assess the interacting impacts of physical, chemical,

and nutritional pollen defenses on the foraging behavior and fitness of a generalist pollen forager. We addressed this question via a pollen foraging study in the field and a mechanistic lab experiment. We observed that Peponapis pruinosa, a specialist bee on cucurbit plants, collected pure loads of cucurbit pollen in contrast to generalist honey bees and bumble bees which collected very small amounts of cucurbit pollen, although all bees visited cucurbit flowers.

Microcolonies fed non-cucurbit pollen increased pollen consumption over time, while microcolonies fed natural cucurbit pollen and crushed cucurbit decreased pollen consumption over time. Microcolonies fed crushed cucurbit pollen has higher mortality overall, with large bees in this treatment particularly at risk. Microcolonies fed both cucurbit pollen treatments never reared offspring to adulthood, while all other treatments always produced adult offspring, except for two microcolonies fed the control diet. Additionally, more larvae were ejected from microcolonies fed natural cucurbit pollen than in any other treatment. Together, these results suggest that Bombus impatiens workers avoid collecting cucurbit pollen due to the physiological costs associated with the consumption of pollen with multiple levels of defenses.

In the field, B. impatiens and A. mellifera foraged for pollen away from where their colonies were located in cucurbit fields. Only a very small percentage of cucurbit pollen grains were found within their corbiculae. Yet these generalist bees spent time foraging in cucurbit flowers, presumably for nectar (Artz & Nault, 2011; Petersen & Nault, 2014; Petersen et al.,

2013). This foraging behaviour is important because it suggests that bees recognize cucurbit flowers as a nectar source, but avoid collecting the pollen.

There are several mechanisms that could explain why generalist bees avoid collecting certain pollens (Freeland & Janzen, 1974; Hägele & Rowell-Rahier, 1999; Pulliam, 1975; Singer,

Bernays, & Carrière, 2002; Tadmor-Melamed et al., 2004; Westoby, 1978). Pre-ingestive effects

(e.g. mechanical defenses, cues correlated with poor nutrition, chemical defenses, or chemical

tastes) deter feeding (changing consumption behaviour), while post-ingestive effects, which can be pre-digestive (e.g. bees have different capacities for pollen digestion or toxins reduce nutrient digestibility) or post-digestive (e.g. a diet has insufficient nutrients or bees cannot metabolize diet toxins), can reduce growth and reproduction through malnutrition or direct toxic effects.

Pre-ingestive effects were evident in our study with reduced pollen consumption in microcolonies fed both Crushed and Natural cucurbit pollen treatments, compared with microcolonies fed all other diet treatments, and a tendency to decrease pollen consumption over time. Reduced feeding in the cucurbit pollen treatments could be a response to a poor diet, or an attempt to minimize consumption of plant defenses in the diet (Bernays, Bright, Gonzalez, &

Angel, 1994; Hägele & Rowell-Rahier, 1999; Pennings, Nadeau, & Paul, 1993). Several studies have suggested that animals fed single-plant diets will do worse than on mixed diets, but if a lack of nutrients is to blame they may increase their feeding to compensate, while if toxins are to blame the animal should reduce consumption (Bernays et al., 1994; Pennings et al., 1993); however, this pattern does not always prove true if the poor diet is lacking in some essential nutrient or contains the wrong ratio of nutrients (Hägele & Rowell-Rahier, 1999; Vaudo et al.,

2016). Our pattern of reduced consumption suggests that cucurbit pollen is lacking in an essential component of the B. impatiens diet and/or contains pollen defenses.

Post-ingestive effects in B. impatiens were also observed in our study in the form of reduced reproduction and increased mortality. Bees in the Crushed and Natural cucurbit treatments were unable to rear any offspring to adulthood and microcolonies fed Crushed cucurbit pollen had higher proportions of mortality. If B. impatiens simply cannot digest cucurbit pollen this effect would be pre-digestive (Hägele & Rowell-Rahier, 1999), but this outcome

would also have been minimized in the Crushed treatment, where broken exines should have

made nutrients more readily available. Since we observed severe negative effects in both

Crushed and Natural cucurbit pollen treatments, we hypothesized that these effects are primarily

post-digestive, where cucurbit pollen is either missing essential dietary components and/or has physical or chemical properties that interfere with physiological processes.

Our study was designed to provide evidence for the effects of both malnutrition and

pollen defenses, but due to the unexpected strength of the responses in both Crushed and Natural

cucurbit treatments, particularly the increased mortality in the Crushed cucurbit treatment, it is

difficult to parse out the mechanisms of these effects. Such high mortality in the Crushed

cucurbit treatment was unexpected as we predicted that the Natural cucurbit pollen would have

the most severe effects on both adults and offspring. It is possible that in crushing the cucurbit

pollen, we changed the pollen by: 1) increasing evaporation and drying of the pollen thus

reducing its nutritional value, 2) increasing the level of chemical defenses by releasing additional

chemicals contained within the exine that would otherwise have been inaccessible, or 3)

increasing the level of mechanical damage by creating smaller shards of exines that had a more

severe effect than the intact exine. The amount of pollen consumed by larval bumble bees

determines their size as adults; therefore, smaller adult bees may already exhibit sub-optimal

health condition (Ribeiro, 1994; Sutcliffe & Plowright, 1988). In contrast, smaller bees were

more resilient to nectar shortages, likely due to an increased proportion of lipid tissues compared

to larger bees (M. J. Couvillon & Dornhaus, 2010; Margaret J. Couvillon, Jandt, Bonds, Helm, &

Dornhaus, 2011). Larger bees also tended to be more dominant, and sometimes restricted access to food for smaller bees (Ayasse, Marlovits, Tengö, Taghizadeh, & Francke, 1995; Cnaani,

Wong, & Thomson, 2007). This reduced access to food could cause smaller bees to become

stressed and even more susceptible to the effects of pollen defenses (Manson & Thomson, 2009).

Since larger bees fed crushed cucurbit pollen tended to have a higher risk of mortality, our

hypothesis that the crushed treatment had lower nutritional value was supported (M. J. Couvillon

& Dornhaus, 2010; Margaret J. Couvillon et al., 2011). In contrast, smaller bees fed the solvent

control and natural cucurbit pollen had higher mortality risks, suggesting that they could have

been suffering from increased toxic effects as a result of their small size (Ayasse et al., 1995;

Manson & Thomson, 2009). Interestingly, this effect did not extend to bees in the Chemistry

treatment.

Our results on pollen consumption over time also provide some evidence for post-

ingestive effects. Microcolonies of larger average weights fed the Solvent and Chemistry

treatments consumed more pollen over time than smaller microcolonies in the same treatment

(with microcolonies fed the Chemistry treatment consuming more pollen than microcolonies fed

the Solvent treatment at all weights), but did not produce more or larger offspring, suggesting

compensatory feeding to overcome a pre-digestive constraint whereby some component of the

diet reduced digestibility (Cruz-Rivera & Hay, 2003; Hägele & Rowell-Rahier, 1999; Singer et al., 2002; Tadmor-Melamed et al., 2004). We would expect this effect to be more pronounced in larger microcolonies which would need to consume proportionally more pollen than smaller microcolonies.

While our results suggest no effects of chemical defenses alone, we believe they merit further investigation. Our chemical extraction method can only succeed in isolating a subset of the potential chemicals found in cucurbit pollen, thus both cucurbit pollen treatments would have a more complex suite of chemicals which could have contributed to the stronger effects in these treatments. Our results suggested that while there may be subtle effects of chemical defenses on

pollen efficiency, there are unlikely to be strong mortality effects due to toxins in pollen alone.

Additionally, since most studies on toxins in pollen fail to account for the interacting effects of nutrition (Eckhardt et al., 2014; Gosselin et al., 2013; Haider et al., 2014; Praz et al., 2008;

Sedivy et al., 2011, 2012, 2008), the strongest effects may occur when toxins interfere with digestion in an already poor nutritional diet, as in our cucurbit treatments (Cruz-Rivera & Hay,

2003; Simpson & Raubenheimer, 2001; Wahl & Ulm, 1983).

Interestingly, adults and larvae responded differently to our treatments, suggesting life- stage specific physiological adaptations. Other studies have also found that digestion is affected by the age of the animal, indicating that larval and adult digestion may differ (T. H. Roulston &

Cane, 2000; Unsicker, Oswald, Köhler, & Weisser, 2008). Adult bees are capable of reducing their own diet consumption, but they may continue to provide larval bees with a set amount of food, which could have increased larval exposure to pollen defenses. This could explain the increased rates of larval ejection in our Natural cucurbit pollen treatment. Some of these larvae were blackened in appearance, suggesting that they were already sick when ejected, but some of the larvae appeared to be healthy, suggesting that their removal was more likely due to stress in the worker bees (Genissel et al., 2002; Tasei & Aupinel, 2008a). This observation supports the hypothesis that deleterious effects of the diet on the adult bees themselves could have contributed to reduced reproduction in our cucurbit treatments. Malnutrition or increased physiological costs of metabolizing toxins could reduce investment in producing or caring for offspring (Cruz-

Rivera & Hay, 2003; Tasei & Aupinel, 2008a). Microcolonies fed the Crushed cucurbit pollen treatment exhibited high adult mortality, particularly for large bees, and lower investment in offspring production. Microcolonies fed the Natural cucurbit pollen treatment exhibited high mortality risk only for small bees and high levels of larval ejection. These patterns suggest that

our manipulation of the crushed cucurbit pollen could have exacerbated effects due to

malnutrition, which combined with pollen defenses, were particularly difficult to overcome for

the adult bees. Effects of feeding on Natural cucurbit pollen, while likely not the optimal diet for

bumble bees, appear to be consistent with patterns of post-digestive stress due to pollen defenses.

Overall our study provides evidence that pollen defenses impact both larval and adult bees, through pre-ingestive and post-ingestive effects. Bees were deterred from feeding on

cucurbit pollen both in nature and in our lab experiment, suggesting some cue indicates the

suitability of the pollen diet for consumption. When feeding on the cucurbit diet, we found that

microcolonies suffered severe fitness effects of both increased mortality and reduced

reproduction. Deterrence mechanisms in this system could thus serve as honest signals of

defense allowing bees to avoid physiological damage caused by ingesting defended pollen.

Particularly if pre-ingestive defenses are less costly for plants when compared to post-ingestive

defenses, this may be an efficient mechanism to reduce pollen loss while minimizing costly

defenses. We also found that secondary plant chemicals in cucurbit pollen and DMSO may act as

chemical defenses by reducing the digestibility of nutrients in pollen, but more study is needed to

verify this finding. Ultimately, it would appear that cucurbit pollen is not an optimal diet for B.

impatiens and contains pollen defenses that cause B. impatiens to incur severe physiological

costs. Different combinations of these pollen defenses could allow plants to selectively attract

and deter particular suites of pollinators that have physiological adaptations to different defenses.

Future research should be directed at how widespread pollen defenses are, and how they may

shape the evolution of pollinator floral preferences. Understanding how generalist bees respond

to pollen defenses can provide new insights into digestive adaptations to the pollen diet as well

as elucidate the context for trade-offs between diet generalization and specialization.

References

Archer, C. R., Pirk, C. W. W., Wright, G. A., & Nicolson, S. W. (2014). Nutrition affects survival in

African honeybees exposed to interacting stressors. Functional Ecology, 28(4), 913–923.

doi:10.1111/1365-2435.12226

Arnold, S. E. J., Idrovo, M. E. P., Arias, L. J. L., Belmain, S. R., & Stevenson, P. C. (2014). Herbivore

defence compounds occur in pollen and reduce bumblebee colony fitness. Journal of Chemical

Ecology, 40(8), 878–881. doi:10.1007/s10886-014-0467-4

Artz, D. R., Hsu, C. L., & 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(5),

1144–1158. doi:10.1603/EN10218

Artz, D. R., & Nault, B. A. (2011). Performance of Apis mellifera, Bombus impatiens, and Peponapis

pruinosa (Hymenoptera: Apidae) as pollinators of pumpkin. Journal of Economic Entomology,

104(4), 1153–1161. doi:10.1603/EC10431

Ayasse, M., Marlovits, T., Tengö, J., Taghizadeh, T., & Francke, W. (1995). Are there pheromonal

dominance signals in the bumblebee Bombus hypnorum L (Hymenoptera, Apidae)? Apidologie,

26(3), 163–180. doi:10.1051/apido:19950301

Bartoń, K. (2017). MuMIn: Multi-model inference (Version 1.40.0). Retrieved from https://cran.r-

project.org/web/packages/MuMIn/index.html

Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects models using lme4.

Journal of Statistical Software, 67(1). doi:10.18637/jss.v067.i01

Bernays, E. A., Bright, K. L., Gonzalez, N., & Angel, J. (1994). Dietary mixing in a generalist herbivore:

Tests of two hypotheses. Ecology, 75(7), 1997–2006. doi:10.2307/1941604

Buchmann, S. L. (1983). Buzz pollination in angiosperms. In C. E. Jones & R. J. Little (Eds.), Handbook

of experimental pollination biology (pp. 73–113). New York, NY: Van Nostrand Reinhold

Company. Retrieved from https://www.cabdirect.org/cabdirect/abstract/19840215543

Burger, H., Ayasse, M., Dötterl, S., Kreissl, S., & Galizia, C. G. (2013). Perception of floral volatiles

involved in host-plant finding behaviour: Comparison of a bee specialist and generalist. Journal

of Comparative Physiology A, 199(9), 751–761. doi:10.1007/s00359-013-0835-5

Cane, J. H., & Sipes, S. D. (2006). Characterizing floral specialization by bees: Analytical methods and a

revised lexicon for oligolecty. In Plant-Pollinator Interactions: From Specialization to

Generalization (p. 99:122). University of Chicago Press.

Cnaani, J., Wong, A., & Thomson, J. D. (2007). Effect of group size on ovarian development in

bumblebee workers (Hymenoptera: Apidae: Bombus). Entomologia Generalis, 305–314.

doi:10.1127/entom.gen/29/2007/305

Couvillon, M. J., & Dornhaus, A. (2010). Small worker bumble bees (Bombus impatiens) are hardier

against starvation than their larger sisters. Insectes Sociaux, 57(2), 193–197. doi:10.1007/s00040-

010-0064-7

Couvillon, Margaret J., Jandt, J. M., Bonds, J., Helm, B. R., & Dornhaus, A. (2011). Percent lipid is

associated with body size but not task in the bumble bee Bombus impatiens. Journal of

Comparative Physiology A, 197(11), 1097. doi:10.1007/s00359-011-0670-5

Cruz-Rivera, E., & Hay, M. E. (2003). Prey nutritional quality interacts with chemical defenses to affect

consumer feeding and fitness. Ecological Monographs, 73(3), 483–506. doi:10.1890/0012-

9615(2003)073[0483:PNQIWC]2.0.CO;2

Detzel, A., & Wink, M. (1993). Attraction, deterrence or intoxication of bees (Apis mellifera) by plant

allelochemicals. Chemoecology, 4(1), 8–18. doi:10.1007/BF01245891

Dobson, H. E. ., & Peng, Y.-S. (1997). Digestion of pollen components by larvae of the flower-specialist

bee Chelostoma florisomne (Hymenoptera: Megachilidae). Journal of Insect Physiology, 43(1),

89–100. doi:10.1016/S0022-1910(96)00024-8

Dobson, H. E. M. (1987). Role of flower and pollen aromas in host-plant recognition by solitary bees.

Oecologia, 72(4), 618–623. doi:10.1007/BF00378991

Dobson, H. E. M. (1988). Survey of pollen and pollenkitt lipids -- Chemical cues to flower visitors?

American Journal of Botany, 75(2), 170. doi:10.2307/2443884

Eckhardt, M., Haider, M., Dorn, S., & Müller, A. (2014). Pollen mixing in pollen generalist solitary bees:

a possible strategy to complement or mitigate unfavourable pollen properties? Journal of Animal

Ecology, 83(3), 588–597. doi:10.1111/1365-2656.12168

Edens-Meier, R., Joseph, M., Arduser, M., Westhus, E., & Bernhardt, P. (2011). The pollination biology

of an annual endemic herb, Physaria filiformis (Brassicaceae), in the Missouri Ozarks following

controlled burns. The Journal of the Torrey Botanical Society, 138(3), 287–297.

doi:10.3159/TORREY-D-11-00024.1

Fox, J., Weisberg, S., Adler, D., Bates, D., Baud-Bovy, G., Ellison, S., … Zeileis, A. (2017). car:

Companion to applied regression (Version 2.1-5). Retrieved from https://cran.r-

project.org/web/packages/car/index.html

Freeland, W. J., & Janzen, D. H. (1974). Strategies in herbivory by mammals: The role of plant secondary

compounds. The American Naturalist, 108(961), 269–289.

Genissel, A., Aupinel, P., Bressac, C., Tasei, J.-N., & Chevrier, C. (2002). Influence of pollen origin on

performance of Bombus terrestris micro-colonies. Entomologia Experimentalis et Applicata,

104(2–3), 329–336. doi:10.1046/j.1570-7458.2002.01019.x

Gosselin, M., Michez, D., Vanderplanck, M., Roelants, D., Glauser, G., & Rasmont, P. (2013). Does

Aconitum septentrionale chemically protect floral rewards to the advantage of specialist

bumblebees?: Impact of toxic rewards on visitors. Ecological Entomology, 38(4), 400–407.

doi:10.1111/een.12032

Hägele, B. F., & Rowell-Rahier, M. (1999). Dietary mixing in three generalist herbivores: Nutrient

complementation or toxin dilution? Oecologia, 119(4), 521–533.

Haider, M., Dorn, S., & Müller, A. (2014). Better safe than sorry? A Fabaceae species exhibits

unfavourable pollen properties for developing bee larvae despite its hidden anthers. Arthropod-

Plant Interactions, 8(3), 221–231. doi:10.1007/s11829-014-9299-y

Harder, L. D., & Barclay, R. M. R. (1994). The functional significance of poricidal anthers and buzz

pollination: Controlled pollen removal from Dodecatheon. Functional Ecology, 8(4), 509–517.

doi:10.2307/2390076

Harder, Lawrence D., & Thomson, J. D. (1989). Evolutionary options for maximizing pollen dispersal of

animal-pollinated plants. The American Naturalist, 133(3), 323–344.

Hargreaves, A. L., Harder, L. D., & Johnson, S. D. (2009). Consumptive emasculation: The ecological

and evolutionary consequences of pollen theft. Biological Reviews, 84(2), 259–276.

doi:10.1111/j.1469-185X.2008.00074.x

Holm, S. (1979). A simple sequentially rejective multiple test procedure. Scandinavian Journal of

Statistics, 6(2), 65–70.

Hothorn, T., Bretz, F., & Westfall, P. (2008). Simultaneous inference in general parametric models.

Biometrical Journal, 50(3), 346–363. doi:10.1002/bimj.200810425

Hurd, Paul D., Linsley, E. G., & Whitaker, T. W. (1971). Squash and gourd bees (Peponapis,

Xenoglossa) and the origin of the cultivated Cucurbita. Evolution, 25(1), 218–234.

doi:10.2307/2406514

Hurd, Paul David, Linsley, E. G., & Michelbacher, A. D. (1974). Ecology of the squash and gourd bee,

Peponapis pruinosa, on cultivated cucurbits in California (Hymenoptera: Apoidea). Smithsonian

Contributions to Zoology, (168), 1–17. doi:10.5479/si.00810282.168

Janz, N., & Nylin, S. (2008). The oscillation hypothesis of host-plant range and speciation. In

Specialization, Speciation, and Radiation: the evolutionary biology of herbivorous insects (pp.

203–215). University of California Press.

Johri, B. M., & Vasil, I. K. (1961). Physiology of pollen. Botanical Review, 27(3), 325–381.

Julier, H. E., & Roulston, T. H. (2009). Wild Bee Abundance and Pollination Service in Cultivated

Pumpkins: Farm Management, Nesting Behavior and Landscape Effects. Journal of Economic

Entomology, 102(2), 563–573. doi:10.1603/029.102.0214

Kevan, P. G., & Ebert, T. (2005). Can almond nector & pollen poison honey bees? American Bee

Journal. Retrieved from http://agris.fao.org/agris-search/search.do?recordID=US201301000539

Kevan, P. G., Mohr, N. A., Offer, M. D., & Kemp, J. R. (1989). The squash and gourd bee, Peponapis

pruinosa (Hymenoptera). Proceedings of the Entomological Society of Ontario, 119, 9–15.

Larrere, M., & Couillaud, F. (1993). Role of juvenile hormone biosynthesis in dominance status and

reproduction of the bumblebee, Bombus terrestris. Behavioral Ecology and Sociobiology, 33(5).

doi:10.1007/BF00172932

Lenth, R., Love, J., & Herve, M. (2018). emmeans: Estimated marginal means, aka least-squares means

(Version 1.1). Retrieved from https://cran.r-project.org/web/packages/emmeans/index.html

London-Shafir, I., Shafir, S., & Eisikowitch, D. (2003). Amygdalin in almond nectar and pollen – Facts

and possible roles. Plant Systematics and Evolution, 238(1–4), 87–95. doi:10.1007/s00606-003-

0272-y

Lunau, K., Piorek, V., Krohn, O., & Pacini, E. (2015). Just spines — Mechanical defense of malvaceous

pollen against collection by corbiculate bees. Apidologie, 46(2), 144–149. doi:10.1007/s13592-

014-0310-5

Lundgren, J. G. (2009). Pollen nutrition and defense. In Relationships of Natural Enemies and Non-Prey

Foods (Vol. 7, pp. 127–139). Springer, Dordrecht. doi:10.1007/978-1-4020-9235-0_8

Manson, J. S., & Thomson, J. D. (2009). Post-ingestive effects of nectar alkaloids depend on dominance

status of bumblebees. Ecological Entomology, 34(4), 421–426. doi:10.1111/j.1365-

2311.2009.01100.x

Mathewson, J. A. (1968). Nest construction and life history of the eastern cucurbit bee, Peponapis

pruinosa (Hymenoptera: Apoidea). Journal of the Kansas Entomological Society, 41(2), 255–

261.

Mazerolle, M. J. (2017). AICcmodavg: Model selection and multimodel inference based on (Q)AIC(c)

(Version 2.1-1). Retrieved from https://cran.r-project.org/web/packages/AICcmodavg/index.html

Müller, A., & Kuhlmann, M. (2008). Pollen hosts of western palaearctic bees of the genus Colletes

(Hymenoptera: Colletidae): The Asteraceae paradox. Biological Journal of the Linnean Society,

95(4), 719–733. doi:10.1111/j.1095-8312.2008.01113.x

Palmer-Young, E. C., Hogeboom, A., Kaye, A. J., Donnelly, D., Andicoechea, J., Connon, S. J., … Adler,

L. S. (2017). Context-dependent medicinal effects of anabasine and infection-dependent toxicity

in bumble bees. PLOS ONE, 12(8), e0183729. doi:10.1371/journal.pone.0183729

Peng, Y.-S., Nasr, M. E., & Marston, J. M. (1986). Release of alfalfa, Medicago sativa, pollen cytoplasm

in the gut of the honey bee, Apis mellifera (Hymenoptera: Apidae). Annals of the Entomological

Society of America, 79(5), 804–807. doi:10.1093/aesa/79.5.804

Peng, Y.-S., Nasr, M. E., Marston, J. M., & Fang, Y. (1985). The digestion of dandelion pollen by adult

worker honeybees. Physiological Entomology, 10(1), 75–82. doi:10.1111/j.1365-

3032.1985.tb00021.x

Pennings, S. C., Nadeau, M. T., & Paul, V. J. (1993). Selectivity and growth of the generalist herbivore

Dolabella auricularia feeding upon complementary resources. Ecology, 74(3), 879–890.

doi:10.2307/1940813

Petersen, J. D., & Nault, B. A. (2014). Landscape diversity moderates the effects of bee visitation

frequency to flowers on crop production. Journal of Applied Ecology, 51(5), 1347–1356.

doi:10.1111/1365-2664.12287

Petersen, J. D., Reiners, S., & Nault, B. A. (2013). Pollination services provided by bees in pumpkin

fields supplemented with either Apis mellifera or Bombus impatiens or not supplemented. PLoS

ONE, 8(7), e69819. doi:10.1371/journal.pone.0069819

Praz, C. J., Müller, A., & Dorn, S. (2008). Specialized bees fail to develop on non-host pollen: do plants

chemically protect their pollen? Ecology, 89(3), 795–804. doi:10.1890/07-0751.1

Pulliam, H. R. (1975). Diet optimization with nutrient constraints. The American Naturalist, 109(970),

765–768.

Ribeiro, M. F. (1994). Growth in bumble bee larvae: Relation between development time, mass, and

amount of pollen ingested. Canadian Journal of Zoology, 72(11), 1978–1985. doi:10.1139/z94-

270

Richardson, L. L., Adler, L. S., Leonard, A. S., Andicoechea, J., Regan, K. H., Anthony, W. E., … Irwin,

R. E. (2015). Secondary metabolites in floral nectar reduce parasite infections in bumblebees.

Proceedings of the Royal Society B: Biological Sciences, 282(1803), 20142471–20142471.

doi:10.1098/rspb.2014.2471

Roulston, T. H., & Cane, J. H. (2000). Pollen nutritional content and digestibility for animals. Plant

Systematics and Evolution, 222(1–4), 187–209. doi:10.1007/BF00984102

Roulston, T’ai H., & Buchmann, S. L. (2000). A phylogenetic reconsideration of the pollen starch–

pollination correlation. Evolutionary Ecology Research, 2(5), 627–643.

Roulston, T’ai H., Cane, J. H., & Buchmann, S. L. (2000). What governs protein content of pollen:

Pollinator preferences, pollen-pistil interactions, or phylogeny? Ecological Monographs, 70(4),

617–643. doi:10.2307/2657188

Russo, L., & Danforth, B. (2017). Pollen preferences among the bee species visiting apple (Malus pumila)

in New York. Apidologie, 48(6), 806–820. doi:10.1007/s13592-017-0525-3

Sarkar, D. (2017). lattice: Trellis graphics for R (Version 0.20-35). Retrieved from https://cran.r-

project.org/web/packages/lattice/index.html

Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T., … Cardona, A.

(2012). Fiji: An open-source platform for biological-image analysis. Nature Methods, 9(7), 676.

doi:10.1038/nmeth.2019

Schindelin, J., Rueden, C. T., Hiner, M. C., & Eliceiri, K. W. (2015). The ImageJ ecosystem: An open

platform for biomedical image analysis. Molecular Reproduction and Development, 82(7–8),

518–529. doi:10.1002/mrd.22489

Searle, S. R., Speed, F. M., & Milliken, G. A. (1980). Population marginal means in the linear model: An

alternative to least squares means. The American Statistician, 34(4), 216–221.

doi:10.2307/2684063

Sedivy, C., Dorn, S., Widmer, A., & Müller, A. (2013). Host range evolution in a selected group of

osmiine bees (Hymenoptera: Megachilidae): The Boraginaceae-Fabaceae paradox: Osmiine bee

host range evolution. Biological Journal of the Linnean Society, 108(1), 35–54.

doi:10.1111/j.1095-8312.2012.02013.x

Sedivy, C., Müller, A., & Dorn, S. (2011). Closely related pollen generalist bees differ in their ability to

develop on the same pollen diet: Evidence for physiological adaptations to digest pollen.

Functional Ecology, 25(3), 718–725. doi:10.1111/j.1365-2435.2010.01828.x

Sedivy, C., Piskorski, R., Müller, A., & Dorn, S. (2012). Too low to kill: Concentration of the secondary

metabolite ranunculin in buttercup pollen does not affect bee larval survival. Journal of Chemical

Ecology, 38(8), 996–1002. doi:10.1007/s10886-012-0153-3

Sedivy, C., Praz, C. J., Müller, A., Widmer, A., & Dorn, S. (2008). Patterns of host-plant choice in bees

of the genus Chelostoma: The constraint hypothesis of host-range evolution in bees. Evolution,

62(10), 2487–2507. doi:10.1111/j.1558-5646.2008.00465.x

Shuler, R. E., Roulston, T. H., & Farris, G. E. (2005). Farming practices influence wild pollinator

populations on squash and pumpkin. Journal of Economic Entomology, 98(3), 790–795.

Simpson, S. J., & Raubenheimer, D. (2001). The geometric analysis of nutrient-allelochemical

interactions: A case study using locusts. Ecology, 82(2), 422–439. doi:10.2307/2679870

Singer, M. S., Bernays, E. A., & Carrière, Y. (2002). The interplay between nutrient balancing and toxin

dilution in foraging by a generalist insect herbivore. Animal Behaviour, 64(4), 629–643.

doi:10.1006/anbe.2002.3082

Sipes, S. D., & Tepedino, V. J. (2005). Pollen-host specificity and evolutionary patterns of host switching

in a clade of specialist bees (Apoidea: Diadasia). Biological Journal of the Linnean Society, 86,

487–505. doi:10.1111/j.1095-8312.2005.00544.x

Stanley, R. G., & Linskens, H. F. (1974). Pollen: Biology, Biochemistry, Management. Berlin,

Heidelberg: Springer Berlin Heidelberg. doi:10.1007/978-3-642-65905-8

Suárez-Cervera, M., Marquez, J., Bosch, J., & Seoane-Camba, J. (1994). An ultrastructural study of

pollen grains consumed by larvae of Osmia bees (Hymenoptera, Megachilidae). Grana, 33(4–5),

191–204. doi:10.1080/00173139409429000

Sutcliffe, G. H., & Plowright, R. C. (1988). The effects of food supply on adult size in the bumble bee

Bombus terricola Kirby (Hymenoptera: Apidae). The Canadian Entomologist, 120(12), 1051–

1058. doi:10.4039/Ent1201051-12

Tadmor-Melamed, H., Markman, S., Arieli, A., Distl, M., Wink, M., & Izhaki, I. (2004). Limited ability

of Palestine Sunbirds Nectarinia osea to cope with pyridine alkaloids in nectar of Tree Tobacco

Nicotiana glauca. Functional Ecology, 18(6), 844–850. doi:10.1111/j.0269-8463.2004.00929.x

Tasei, J.-N., & Aupinel, P. (2008a). Nutritive value of 15 single pollens and pollen mixes tested on larvae

produced by bumblebee workers (Bombus terrestris, Hymenoptera: Apidae). Apidologie, 39(4),

397–409. doi:10.1051/apido:2008017

Tasei, J.-N., & Aupinel, P. (2008b). Validation of a method using queenless Bombus terrestris micro-

colonies for testing the nutritive value of commercial pollen mixes by comparison with

queenright colonies. Journal of Economic Entomology, 101(6), 1737–1742. doi:10.1603/0022-

0493-101.6.1737

Therneau, T. M. (2015). coxme: Mixed effects cox models (Version 2.2-5). Retrieved from https://cran.r-

project.org/web/packages/coxme/index.html

Therneau, T. M., & Lumley, T. (2017). survival: Survival analysis (Version 2.41-3). Retrieved from

https://cran.r-project.org/web/packages/survival/index.html

Unsicker, S. B., Oswald, A., Köhler, G., & Weisser, W. W. (2008). Complementarity effects through

dietary mixing enhance the performance of a generalist insect herbivore. Oecologia, 156(2), 313–

324. doi:10.1007/s00442-008-0973-6

Vaudo, A. D., Stabler, D., Patch, H. M., Tooker, J. F., Grozinger, C. M., & Wright, G. A. (2016). Bumble

bees regulate their intake of essential protein and lipid pollen macronutrients. The Journal of

Experimental Biology, 219(24), 3962–3970. doi:10.1242/jeb.140772

Wahl, O., & Ulm, K. (1983). Influence of pollen feeding and physiological condition on pesticide

sensitivity of the honey bee Apis mellifera carnica. Oecologia, 59(1), 106–128.

doi:10.1007/BF00388082

Westerkamp, C. (1997). Keel blossoms: Bee flowers with adaptations against bees. Flora, 192(2), 125–

132. doi:10.1016/S0367-2530(17)30767-3

Westerkamp, C., & Claßen-Bockhoff, R. (2007). Bilabiate flowers: The ultimate response to bees? Annals

of Botany, 100(2), 361–374. doi:10.1093/aob/mcm123

Westoby, M. (1978). What are the biological bases of varied diets? The American Naturalist, 112(985),

627–631.

Wickham, H. (2007). Reshaping data with the reshape package. Journal of Statistical Software, 21(12).

doi:10.18637/jss.v021.i12

Wickham, H. (2011). The split-apply-combine strategy for data analysis. Journal of Statistical Software,

40(1). doi:10.18637/jss.v040.i01

Wickham, H. (2017). scales: Scale functions for visualization (Version 0.5.0). Retrieved from

https://cran.r-project.org/web/packages/scales/index.html

Wickham, H., & Chang, W. (2016). ggplot2: Create elegant data visualisations using the grammar of

graphics (Version 2.2.1). Retrieved from https://cran.r-

project.org/web/packages/ggplot2/index.html

Zuur, A. F., Ieno, E. N., & Elphick, C. S. (2010). A protocol for data exploration to avoid common

statistical problems: Data exploration. Methods in Ecology and Evolution, 1(1), 3–14.

doi:10.1111/j.2041-210X.2009.00001.x

Supplementary Materials

Calberla’s Fluid Recipe

Mix 5 ml glycerol, 10 ml 95% ethanol, 15 ml distilled water, and 3 to 4 drops saturated aqueous

basic fuchsin. If the glycerin does not mix well, heat the solution in a water bath until it does.

Table S1. Varieties of Cucurbita pepo used in the cucurbit pollen diet mix.

Variety Fruit Percentage Black Beauty Zucchini 11.68 Baby Pam Pie Pumpkin 7.05 Cougar Summer Squash 3.08 Flying Saucer Patty pan Squash 6.77 Golden Arrow Summer Squash 6.26 Howden Carving Pumpkin 15.89 New England Pie Pie Pumpkin 16.33 Success PM Summer Squash 13.61 Tom Fox Carving Pumpkin 11.60 Zephyr Summer Squash 7.73

Table S2. Detailed results for all statistical tests performed. Fixed effects listed represent the optimal

model as indicated by AICc scores. Effects in bold indicates significance.

Response Variable Fixed Effects Statistic

1) Proportion adult mortality Treatment F(4,32) = 20.327, p < 0.001

2) Pollen consumption per bee over Treatment F(4,1274) = 230.305, p < 0.001 time (with repeated measures) Day F(1,1) = 26.776, p = 0.049

AverageWeight F(1,78) = 6.701, p = 0.012

Treatment:Day F(4,1330) = 60.095, p < 0.001

Treatment:AverageWeight F(4,1394) = 42.691, p < 0.001

Day:AverageWeight F(1,31) = 0.003, p = 0.961

Treatment:Day:AverageWeight F(4,290) = 16.594, p < 0.001

3) Average sucrose consumption per Average Weight F(1,14) = 16.001, p = 0.001 bee per day

4) Days to first eclosed offspring Average Weight F(1,7) = 0.251, p = 0.633

5) Number of eclosed offspring per Average Weight F(1,6) = 1.780, p = 0.229 bee per day

6) Average eclosed offspring weight Average Weight F(1,13) = 0.010, p = 0.922

7) Number of ejected larva per day Treatment F(4,32) = 6.413, p = 0.001

8) Pollen efficiency Treatment F(2,14) = 5.310, p = 0.019

Table S3. Summary of analysis for pollen mass lost to evaporation.

Response Variable Fixed Effects Statistic

Pollen Mass Lost to Evaporation Time Interval F(6,1) = 0.916, p = 0.649

Treatment F(4,571) = 22.091, p < 0.001

Treatment : Time Interval F(24,569) = 17.042, p < 0.001

Figure S1. Mass loss due to evaporation over time by treatment.

Figure S2. Effect of average weight on sucrose consumption.

Figure S3. Effect of treatment on pollen efficiency.

Chemical Ecology of Pollinator Dietary Choices

Kristen Brochu *a, Maria van Dykea, Andre Kessler b, and Bryan Danforth a

a Department of Entomology, Cornell University, Ithaca, USA

b Department of Ecology and Evolutionary Biology, Cornell University, Ithaca, USA

* Corresponding author: [email protected]

Abstract

Chemical signals have mediated interactions between plants and insects for millions of years, yet few studies have assessed variation in the chemical signals of plant reward tissues and how they impact pollinator foraging behavior. We examined how variation in floral chemistry impacts bee visitation choices between Cucurbita pepo cultivars. We found that the Cucurbit specialist bee,

Peponapis pruinosa, exhibited significant differences in preference for Cucurbita pepo cultivars, but the generalist visitors, Bombus spp. and Apis mellifera did not. We also found that chemistry in leaves, petals, nectar, and pollen were significantly different, and chemistry of the individual tissues differed by cultivar, clustering by subspecies. Varieties preferred by squash bees were more distinct in floral chemistry than other varieties. Certain floral and nectar chemicals were found to predict bee choices, largely due to the preference for the Zephyr cultivar. Our results show that floral chemistry varies by subspecies and cultivar in C. pepo, impacting pollinator preferences in specialist but not generalist bees. Variation in floral chemistry represents an important, yet under-studied, determinant of plant-pollinator interactions.

Introduction

Chemically-mediated interactions between plants and insects are ubiquitous in ecosystems. Plants use chemical communication to attract pollinators (Burger et al. 2013), defend their tissues (Carter and Thornburg 2004, Lev-Yadun et al. 2009, Schiestl 2010,

Nocentini et al. 2015), and recruit predators or parasitoids of their herbivores (Turlings and

Wäckers 2004). This diversity of roles for chemical signals has contributed to high variability in the presence and quantity of secondary metabolites both within and across plant species (Detzel and Wink 1993, Weng et al. 2012, Wetzel et al. 2016, Wetzel and Thaler 2016). The importance of this variation has traditionally been overlooked and is now starting to be recognized as an important contributor to herbivore performance and community function (Albert et al. 2011,

Wetzel et al. 2016, Wetzel and Thaler 2016, Wright et al. 2016). In particular, few studies have assessed variation in the chemical signals of plant reward tissues and how they impact pollinator foraging behaviour (Adler et al. 2012, Richardson et al. 2016).

Secondary metabolites in floral rewards may simply be present as a consequence of the defense of other tissues (Adler 2000); however, some plants are known to actively regulate defense chemicals in floral reward tissue, suggesting an important role for pollen chemistry in shaping foraging preferences (Detzel and Wink 1993). In addition, a growing body of literature has suggested that secondary metabolites in nectar play a variety of functions, from attracting specialist pollinators and discouraging nectar robbers (Burger et al. 2013), to defending against pathogens for both the plant and its pollinators (Thornburg et al. 2003, Richardson et al. 2015) or encouraging better pollination by altering pollinator behavior (Adler 2000). While many plants use secondary defense chemicals to deter generalist herbivores, these same defense chemicals can be attractive to specialist herbivores by providing benefits through pathogen resistance or

predator deterrence (Gosselin et al. 2013, Muller and Kuhlmann 2008, Praz et al. 2008, Sedivy et al. 2008). In the context of plant-pollinator interactions, these defense chemicals can function as a specialized channel of communication that targets specialist pollinators and deters generalist pollinators that may be less efficient (Canto-Aguilar and Parra-Tabla 2000, Tepedino 1981).

Bees are highly specialized herbivores, collecting pollen in large quantities to feed their developing offspring. For bees, the pollen and nectar mass collected by the adult female is the only nourishment that the larva will receive to complete its development, rendering the foraging choices made by female bees extremely important. Obtaining quality pollen resources is therefore critical to larval survival. Bees use multiple cues to assess floral resources, but visual and chemical cues are the most well-studied (Burger et al. 2010, 2013). Chemical cues are important for attracting bees at a distance or when visual cues are less distinct (Raguso 2008).

These long-distance cues may be especially important in agricultural landscapes where crops tend to be rotated, and thus these resource-rich patches move from year to year.

Agricultural crops usually have altered plant chemistry relative to wild progenitors, especially in reproductive tissues, which is often characteristic to particular cultivars (Rosenthal and Dirzo 1997, Turcotte et al. 2017, Whitehead et al. 2017). This range of floral chemistry makes crop plants an ideal system for investigating how floral chemistry impacts foraging choices. Squash and pumpkin (genus Cucurbita) are completely dependent on biotic pollination, with a diverse fauna of bee pollinators across their geographic range, including both specialist and generalist species (Mathewson 1968, Hurd et al. 1971, 1974, Kevan et al. 1989, Shuler et al.

2005, Cane and Sipes 2006, Julier and Roulston 2009, Artz and Nault 2011, Artz et al. 2011,

Petersen et al. 2013). There is also a wide diversity in cultivars of Cucurbita pepo (e.g. pumpkin, squash, and gourds) with potentially large variation in their floral rewards. With low

morphological variability in flowers among cultivars, chemical cues are likely to be important for bees to distinguish resource quality. The squash bee, Peponapis pruinosa, is a specialist on the genus Cucurbita (Hurd et al. 1971). Squash bees are extremely efficient pollinators of cucurbits, often completing pollination early in the morning, before other bees even begin foraging (Tepedino 1981, Canto-Aguilar and Parra-Tabla 2000). This pollination efficiency may provide incentives for the plant to target squash bees over generalist pollinators such as honey bees (Apis mellifera) and bumblebees (Bombus spp.). Since squash bees are active before sunrise, it is likely that they locate Cucurbita plants through chemical rather than visual cues

(Hurd et al. 1971, 1974).

In this study we examined how variation in floral chemistry impacts bee visitation choices between Cucurbita pepo cultivars. Hypothesizing that floral chemical cues mediate pollinator choice and community composition, we predict that: 1) Reward tissue (pollen and nectar) will have reduced levels of secondary metabolites relative to other plant tissues, and 2) specialist bees (squash bees) will respond to chemical variation differently than generalists, preferring cultivars with higher levels of secondary metabolites. We predicted that reward tissue chemistry will be distinct from other tissue types and distinct among cultivars. We hypothesize that plants are able to regulate the concentration of secondary defense chemicals in their reward tissue, allowing them to selectively attract specialist pollinators. We also predicted that specialist bees (i.e., squash bees) would demonstrate strong foraging preferences for certain cultivars of C. pepo compared to generalist bees (i.e., honeybees and bumblebees). Generalists may be deterred by these same secondary metabolites or may not even differentiate among them. To test this we assessed plant chemistry profiles of several cultivars of C. pepo, compared the chemistry of the various floral tissues (petals, pollen, and nectar) to leaves, and then assessed bee foraging

choices on these cultivars. Finally we correlated bee preferences with the floral chemistry of the

C. pepo cultivars.

Methods

Plant Chemistry Assessment

In 2014, fifteen commonly available cultivars of C. pepo representing several market classes from two subspecies were selected for chemical analyses (Table 1). These cultivars are known to vary in their attractiveness to the striped cucumber beetle (Acalymma vittatum), a common herbivore of squash and pumpkin (Brzozowski et al. 2016) and thus represent good candidate cultivars to test for foraging preferences in bees. These cultivars were grown in a greenhouse at Cornell University from January to March under standardized organic conditions and natural light. Four tissues were examined (leaf, flower, nectar, and pollen) with samples collected from one male and one female flower on three plants for each of fifteen different C. pepo cultivars. Samples were immediately frozen in liquid nitrogen and stored at -80°C until needed for analysis. Samples were then extracted into methanol by bead homogenization (using a

FASTPREP for 45s at 6.5m/s, twice) and analyzed by HPLC using a Phenomenex® C-18 column, installed in a HP 1100 HPLC-instrument (Agilent Technologies, Santa Clara, CA,

USA), with a UV/VIS diode array detector (DAD) measuring at 229 nm. We used an acetonitrile

(CH3CN, Solvent A) and 0.25% phosphoric acid (H3PO4, Solvent B) solvent system. The 15 µl injection was eluted at 1ml/min with the following gradients: 1- 4 minutes, 90% Solvent B; 4-45 minutes, 80% Solvent B; 45-50 minutes, 50% Solvent B; and 50-55 minutes, 20% Solvent B.

This procedure targets phenolic compounds.

Chromatographs from the HPLC analysis were aligned using a custom Python program

(Python Software Foundation. Python Language Reference, version 2.7. Available at

http://www.python.org) with additional manual screening, and adjusted based on the relative concentration of each sample. We compared the total number of peaks and the total sum of relative peak area across tissue type and cultivar using Kruskal-Wallis tests (due to unequal variances) with post-hoc Dunn’s tests using the R package “FSA” (Dunn 1964, Ogle 2017).

Chemical bouquet composition, or chemotype, of these tissues was compared using non-metric multidimensional scaling (NMDS) of Bray-Curtis dissimilarity matrices performed with the R package “vegan” (Oksanen et al. 2018) taking into account both presence/absence of peaks and peak area. Plant tissue chemistry was compared across type, cultivar, subspecies, and market class using ‘adonis’ and ‘anosim’ analyses.

Bee Choice Test

Six squash cultivars were chosen for choice tests that represented several market classes within two subspecies of C. pepo (subsp. pepo and subsp. texana) with the addition of four cultivars of pumpkin in 2016 to represent an additional market class (Table 1). Plants for the choice tests were grown in a greenhouse at Cornell University under standardized organic conditions from June through September of 2015 and in June of 2016. In 2015, choice tests were conducted in August and September at three sites with potted greenhouse plants of similar age that were moved to the field each day of the experiment. Plants were used only once. Each day, plants were chosen so that there were a similar number of flowers in bloom between cultivars.

We initially used fluorescent powder on the flowers to indicate which bees were repeat visitors, but all bees successfully avoided contacting the powder. In 2016, we used a different protocol, with choice tests conducted in two locations (separated by a distance of 200 m) at one farm with sentinel plants (i.e., plants located at the periphery of a large squash field) that were seeded in the greenhouse in mid June and transplanted into the field in early July in a randomized array. Field

sites were selected from farms that grow cucurbits in the Finger Lakes region of upstate New

York where known large nesting aggregations of P. pruinosa could be found, in addition to large

populations of generalist bees, such as honey bees (A. mellifera) and bumble bees (Bombus spp.).

Two individuals monitored bee choices between 06:00 – 10:00 hrs and recorded the following

information: 1) bee species, 2) plant visited, 3) behavior of the bee, i.e., nectar feeding or pollen collection, and 4) in the case of P. pruinosa, the gender of the individual bee. Cultivars were labeled with numbers in the field, and later transcribed so observers were unaware of cultivar identity during the observations. We assessed bee choices on seven days over the course of two weeks from August 18 to September 3 in 2015 and on August 9 and 16 in 2016. Since we used different methods in 2015 and 2016, we analyzed each year separately.

Table 1. Varieties of Cucurbita pepo used in bee choice and chemistry assessment. Cultivar Code Subspecies Market Class Black Beauty BB pepo Zucchini Baby Pam BP pepo Pie Pumpkin Costata Ct pepo Cocozelle Cougar Cg texana Straightneck Dunja D pepo Zucchini Zucchini Elite EZ pepo Zucchini Flying Saucer FS texana Scallop Golden Arrow GA pepo Zucchini Golden Zucchini GZ pepo Zucchini Greentint Scallop GS texana Scallop Howden H pepo Carving Pumpkin Magda M pepo Vegetable Marrow New England Pie NP pepo Pie Pumpkin Partenon P pepo Zucchini PMR Costata PC pepo Cocozelle Success PM SP texana Straightneck Superpik S texana Straightneck Tom Fox TF pepo Carving Pumpkin Zephyr Z texana Straightneck

We used the R package “nnet” (Venables and Ripley 2002) to conduct a multinomial

logistic regression with Date nested within Site as explanatory variables to verify that choices did

not vary by Date or Site. The number of visits to each cultivar were then pooled and compared

with either a Chi-Square test or an exact multinomial test of goodness-of-fit using the

“XNomial” package in R (Engels 2015) for small sample sizes, with the null hypothesis being that bees visit each of the plant cultivars with equal frequency.

Plant Chemistry Influences on Bee Choice

We performed Mantel tests with the R package “vegan” (Oksanen et al. 2018) to compare

how differences in the chemistry of each cultivar varied with bee preferences. We also used

correlation analyses to test whether the 1) total number of chemicals, 2) sum of chemical

abundance, or 3) any individual chemical was correlated with bee preferences.

Results

Figure 1. Comparison of chemistry across tissue types. Boxplots show distribution of a) Total number of

peaks and b) Sum of relative peak area for samples of each tissue type. Dot indicates mean and dashed

line indicates standard error.

Plant Chemistry

We found that tissue types varied in the total number of peaks (H = 51.486, p < 0.001)

present and the sum of relative peak area (abundance of chemicals) (H = 51.216, p < 0.001)

(Figure 1). Post-hoc Dunn’s tests with a Holm correction for multiple tests (Holm 1979) showed

that all tissues have different total numbers of peaks, except for Petals and Pollen (petals

marginally higher, p < 0.1) and all tissues have different total amounts of chemicals, except for

Nectar and Pollen (pollen marginally higher, p < 0.1). Cultivars did not vary in either of these

traits (H = 1.685, p = 1, and H = 2.965, p = 0.999; respectively). We found that plant tissue types

varied in their chemical bouquet composition, or chemotype (adonis: R2 = 0.369, p = 0.001;

anosim: R = 0.708, p = 0.001; Figure 2), with each tissue type exhibiting distinct chemistry in

post-host pairwise adonis tests with a Holm correction for multiple tests (Holm 1979). Reward

tissue (i.e. pollen and nectar) exhibited greater variation in chemotype than leaves or petals.

Figure 2. Non-metric multidimensional scaling (NMDS) plot to compare compound bouquet composition in leaves, nectar, petals, and pollen (all cultivars pooled). Ellipses represent 95% confidence intervals. 44

We also found significant differences in chemotype among cultivars in flowers (adonis:

R2 = 0.572, p = 0.001; anosim: R = 0.551, p = 0.001), leaves (adonis: R2 = 0.444, p = 0.001;

anosim: R = 0.394, p = 0.001), nectar (adonis: R2 = 0.391, p = 0.002; anosim: R = 0.359, p =

0.001), and pollen (adonis: R2 = 0.783, p = 0.001; anosim: R = 0.588, p = 0.001; Figure 3). No

post-hoc pairwise adonis tests between cultivars were significant with a Holm correction for any tissue; however, with an FDR correction, Flying Saucer and Zephyr floral tissues were different from all but three other cultivars, including each other.

Figure 3. Non-metric multidimensional scaling (NMDS) plot to compare compound bouquet composition in cultivars within a) leaves, b) petals, c) nectar, and d) pollen. Ellipses represent 95% confidence intervals for C. pepo subsp. pepo (black) and C. pepo subsp. texana (brown).

We found that cultivars clustered by subspecies in flowers (adonis: R2 = 0.29, p = 0.001;

anosim: R = 0.646, p = 0.001), leaves (adonis: R2 = 0.06461, p = 0.001; anosim: R = 0.134, p =

0.002), nectar (adonis: R2 = 0.081, p = 0.001; anosim: R = 0.164, p = 0.003), and pollen (adonis:

R2 = 0.081, p = 0.001; anosim: R = 0.164, p = 0.003). Leaves also clustered by market class

(adonis: R2 = 0.150, p = 0.009; anosim: R = 0.158, p = 0.011) with FDR-corrected post-hoc tests indicating Scallop and Straightneck cultivars to be different from all other market classes.

Bee Choice

We observed too few pollen collection events to analyze, so our results focus solely on observations of nectar foraging. In 2015, honey bees and bumblebees were observed in very low frequencies (n = 16 and n = 62 respectively). Observations were therefore pooled across Site and

Date for exact multinomial goodness-of-fit tests. Honey bees and bumblebees did not exhibit preferences for any cultivar (p = 0.295 and p = 0.458 respectively, Figure 4a), although power analyses (α = 0.05 and β = 0.2) indicated that we only had sufficient power to detect large effects

(w = 0.895 and w = 0.455, respectively) for these species (Cohen 1988). Multinomial logistic regression showed that P. pruinosa choices in 2015 were not significantly different within Site and Date (Χ2 = 60.898, df = 60, p = 0.443) therefore the observations were pooled. Peponapis pruinosa were observed in large numbers (ntotal = 283, nfemale = 36, nmale = 247), with males visiting more frequently than females. Overall P. pruinosa exhibited significant preferences for cultivars (Χ2 = 38.986, df = 5, p < 0.001), with both males and females exhibiting significant differences in the cultivars that they visited (Χ2 = 33.543, df = 5, p < 0.001 and p = 0.0462 respectively). Post-hoc tests with the FDR correction for multiple tests (Benjamini and Hochberg

1995), indicates that overall, P. pruinosa visited Flying Saucer and Zephyr cultivars more than expected with fewer bees than expected visiting the Cougar cultivar. Peponapis pruinosa males most strongly prefer Flying Saucer, while females most strongly prefer Success PM (Figure 4b).

Figure 4. Bee visitation to Cucurbita pepo cultivars. Dashed lines represent expected visitations based on total number of visits. a) Generalist bee visits in 2015. b) Specialist bee visits in 2015. c) Specialist bee visits in 2016. Data in 2016 was adjusted to counts per flower based on total number of visits.

In 2016, due to the use of sentinel plants grown in the field, we were limited to the use of

plants that were flowering on any given day. Due to the non-significant effects of date and site in

2015, we pooled our dates and sites for 2016 and divided the visitation frequency by the total

number of flowers for each cultivar to account for an effect of floral display size. We again

observed P. pruinosa in large numbers (ntotal = 313, nfemale = 8, nmale = 305), but fewer females.

P. pruinosa exhibited significant preferences for cultivars (Χ2 = 305.07, df = 9, p < 0.001), with

males and females exhibiting significant differences in their preferences for different cultivars

(Χ2 = 294.24, df = 9, p < 0.001and p = 0.009 respectively). Post-hoc tests indicated that P. pruinosa visits the Flying Saucer and Zephyr cultivars more than expected, with males most strongly preferring Flying Saucer and Zephyr and females preferring Zephyr (Figure 4c).

Chemical Influence on Bee Choice

Mantel tests with FDR-corrected p-values for multiple tests were not significant for any individual chemicals, but without the correction, one floral chemical (both its mean and se) was found to predict preference of both males and females (Retention Time = 6.357 minutes). With

Holm-corrected p-values for multiple tests, two nectar chemicals (Retention Time = 46.896 minutes, rm = 0.980; Retention Time = 48.187 minutes, rf = 0.950; p < 0.001) were correlated

with both male and female preference while three floral chemicals (Retention Time = 7.140

minutes, r1 = 0.983; Retention Time = 8.818 minutes, r2 = 0.979; Retention Time = 17.127

minutes, r3 = 0.995; p < 0.001) were correlated with female preference. These five chemicals

were found in high abundance in the Zephyr cultivar. The total number of floral peaks was also

correlated with male preference (r = -0.848, p = 0.033).

Discussion

We found significant differences in preferences for Cucurbita pepo cultivars in the

Cucurbit specialist bee, Peponapis pruinosa, but not in the generalist visitors, Bombus spp. or

Apis mellifera. We also found that chemistry in leaves, petals, nectar, and pollen were significantly different. Leaves had more peaks and a higher abundance of chemicals than any other tissue, while nectar had the least numbers and abundance. The chemistry of the individual tissues also differed by cultivar, clustering by subspecies although not strongly in leaves or nectar. Varieties preferred by squash bees were more distinct in floral chemistry than other varieties. Certain floral and nectar chemicals were found to predict bee choices, largely due to the preference for the Zephyr cultivar.

Our results support the prediction that specialist bees respond differently to chemical variability than generalist bees. Squash bees exhibited preferences for Zephyr and Flying Saucer cultivars, with both male and female squash bees exhibiting similar preferences across the two years of the study. Male bees are unusually important in cucurbits as they actually contribute to pollination through their foraging and mate searching behaviour (Cane et al. 2011). Squash bee biology is intimately tied to the host plant; males typically sleep within flowers overnight, and mating occurs within flowers, so there may be selective pressure for males to have floral preferences that align with those of females. Male bees of other species have been found to sleep in flowers preferred by females of their species, which may function as a strategy to find mates

(Pinheiro et al. 2017). Over the course of a season it is common to observe several males mobbing a single female as she lands on a flower to forage (pers. obs). The generalist bees in our study did not exhibit any preferences for cultivars suggesting that perhaps generalist bees do not distinguish between chemicals that delineate fine-scale differences between varieties, as has been demonstrated in other systems (Burger et al. 2013, Brandt et al. 2017). Studies have shown that different species respond in different ways to plant defenses (Afik et al. 2014, Okamura et al.

2016), although in some cases specialists and generalists may respond similarly to particular chemical profiles (Poelman et al. 2009).

We also found that squash bees exhibited preferences for C. pepo subsp. texana cultivars over the C. pepo subsp. pepo cultivars. These two subspecies explained significant variation in the chemistry of all tissues we examined, although this effect was largest in floral tissues. These two subspecies are thought to be the result of two independent domestication events (Decker

1988, Sanjur et al. 2002) with distinct genetic differences existing between them and their market classes, including separation by species for important floral traits, like corolla length (Paris et al.

2003, Gong et al. 2012, Theis et al. 2014). Additionally, subspecies identity has been found to explain striped cucumber beetle preference, with beetles preferring C. pepo subsp. pepo over C. pepo subsp. texana (Brzozowski et al. 2016), in direct contrast to squash bee preferences observed in this study. Theis et al. (2014) found that squash bees and striped cucumber beetles preferred flowers of the same varieties, and suggested they are using similar floral cues to make foraging decisions (Theis et al. 2014) although only one wild variety of C. pepo subsp. texana was assessed in that study. Squash bees and striped cucumber beetles are known to have similar preferences for certain floral volatiles, but contrasting preferences for others (Andrews et al.

2007), suggesting that plants can mediate potential conflicts between pollinator attraction and herbivore defense through variations in floral reward chemistry (Kessler and Halitschke 2009).

Overall pollen and nectar tissues in our study had the most variability in chemical bouquet composition, but chemotype variation in subspecies and cultivar was strongest in petals and pollen. Floral tissues have been found to vary in their chemical composition and abundance in other studies (Milet-Pinheiro et al. 2013), with petals of wild radish flowers exhibiting higher levels of glucosinolates than undamaged leaves (Strauss et al. 2004). In fact, squash bees

preferred cultivars that were most different from other cultivars. Unique or rare floral chemicals have been shown to be important for host plant recognition in specialist species (Milet-Pinheiro et al. 2013, Carvalho et al. 2014, Brandt et al. 2017), suggesting that cultivars with a distinct chemical profile may be more conspicuous for specialist bees within a noisy chemical environment. Interestingly, these same varieties tended to have fewer compounds in the bouquet, suggesting reduced floral chemical diversity, which was preferred by squash bees. Other studies have suggested that pollinator preferences for low levels of defense may actually maintain variation in plant defensive chemistry by exerting opposing selective pressure to high levels of defense against herbivores (Strauss et al. 2004).

Floral and nectar chemistry was found to be correlated with squash bee preferences, which raises the question: are bees foraging for a particular nectar chemical profile? Several studies have shown that chemical profiles or chemotypes predict insect species abundances or community composition (Kleine and Müller 2011, Bálint et al. 2016). For example, specialist herbivore abundance in wild cabbage (Brassica oleracea) was correlated to total glucosinolate concentration (Bustos‐Segura et al. 2016) and differences in secondary metabolites predicted caterpillar and parasitoid diversity in a wild species (Piper kelleyi) related to black pepper

(Glassmire et al. 2016). It is not then surprising that specialist bees may be responding to variation in floral chemistry. Floral chemicals may represent honest signals of reward quality, which often predict plant visitation by bees (Knauer and Schiestl 2015). Secondary metabolites in floral rewards have been hypothesized to have many possible beneficial roles including microbial defense and pollinator attraction (Adler 2000). Combinations of phytochemicals were found to synergistically reduce a bumblebee pathogen, although the effect varied by pathogen strain and phytochemical environment (Palmer-Young et al. 2017) and secondary metabolites in

fruits were found to defend against fungal pathogens (Whitehead and Bowers 2014). In

cucurbits, certain volatiles are correlated with nectar production (Theis et al. 2014) directly

linking chemical cues with reward quality. Floral chemistry could directly impact choices

through taste. Bumble bees are more likely to collect sweet pollen than bitter pollen, suggesting

that taste helps bees to determine aspects of pollen chemistry that impact foraging choices (Muth

et al. 2016).

Our findings indicate that floral chemistry varies by subspecies and cultivar in petals and

pollen of C. pepo, impacting pollinator preferences in specialist but not generalist bees. These

results support our initial predictions that plants of C. pepo can regulate the chemistry in reward

tissues, selectively attracting specialist pollinators. Future work should focus on why certain

chemicals are attractive to bees. Identifying the mechanisms by which bees evaluate floral

resources is critical to understanding how bees forage in a potentially noisy chemical

environment. Variation in floral chemistry represents an important, yet under-studied,

determinant of plant-pollinator interactions.

References

Adler, L. S. 2000. The ecological significance of toxic nectar. Oikos 91:409–420.

Adler, L. S., M. G. Seifert, M. Wink, and G. E. Morse. 2012. Reliance on pollinators predicts defensive

chemistry across tobacco species. Ecology Letters 15:1140–1148.

Afik, O., K. S. Delaplane, S. Shafir, H. Moo-Valle, and J. J. G. Quezada-Euán. 2014. Nectar minerals as

regulators of flower visitation in stingless bees and nectar hoarding wasps. Journal of Chemical

Ecology 40:476–483.

Albert, C. H., F. Grassein, F. M. Schurr, G. Vieilledent, and C. Violle. 2011. When and how should

intraspecific variability be considered in trait-based plant ecology? Perspectives in Plant Ecology,

Evolution and Systematics 13:217–225.

Andrews, E. S., N. Theis, and L. S. Adler. 2007. Pollinator and herbivore attraction to Cucurbita floral

volatiles. Journal of Chemical Ecology 33:1682–1691.

Artz, D. R., C. L. Hsu, and B. A. Nault. 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.

Artz, D. R., and B. A. Nault. 2011. Performance of Apis mellifera, Bombus impatiens, and Peponapis

pruinosa (Hymenoptera: Apidae) as pollinators of pumpkin. Journal of Economic Entomology

104:1153–1161.

Bálint, J., S. E. Zytynska, R. V. Salamon, M. Mehrparvar, W. W. Weisser, O. J. Schmitz, K. Benedek,

and A. Balog. 2016. Intraspecific differences in plant chemotype determine the structure of

arthropod food webs. Oecologia 180:797–807.

Benjamini, Y., and Y. Hochberg. 1995. Controlling the false discovery rate: A practical and powerful

approach to multiple testing. Journal of the Royal Statistical Society. Series B (Methodological)

57:289–300.

Brandt, K., S. Dötterl, W. Francke, M. Ayasse, and P. Milet-Pinheiro. 2017. Flower visitors of

Campanula: Are oligoleges more sensitive to host-specific floral scents than polyleges? Journal

of Chemical Ecology 43:4–12.

Brzozowski, L., B. M. Leckie, J. Gardner, M. P. Hoffmann, and M. Mazourek. 2016. Curcurbita pepo

subspecies delineates striped cucumber beetle (Acalymma vittatum) preference. Horticulture

Research 3:16028.

Burger, H., M. Ayasse, S. Dötterl, S. Kreissl, and C. G. Galizia. 2013. Perception of floral volatiles

involved in host-plant finding behaviour: Comparison of a bee specialist and generalist. Journal

of Comparative Physiology A 199:751–761.

Burger, H., S. Dötterl, and M. Ayasse. 2010. Host‐plant finding and recognition by visual and olfactory

floral cues in an oligolectic bee. Functional Ecology 24:1234–1240.

Bustos‐Segura, C., E. H. Poelman, M. Reichelt, J. Gershenzon, and R. Gols. 2016. Intraspecific chemical

diversity among neighbouring plants correlates positively with plant size and herbivore load but

negatively with herbivore damage. Ecology Letters 20:87–97.

Cane, J. H., B. J. Sampson, and S. A. Miller. 2011. Pollination value of male bees: The specialist bee

Peponapis pruinosa (Apidae) at summer squash (Cucurbita pepo). Environmental Entomology

40:614–620.

Cane, J. H., and S. D. Sipes. 2006. Characterizing floral specialization by bees: Analytical methods and a

revised lexicon for oligolecty. Page 99:122 Plant-Pollinator Interactions: From Specialization to

Generalization. University of Chicago Press.

Carter, C., and R. W. Thornburg. 2004. Is the nectar redox cycle a floral defense against microbial attack?

Trends in Plant Science 9:320–324.

Carvalho, A. T., S. Dötterl, and C. Schlindwein. 2014. An aromatic volatile attracts oligolectic bee

pollinators in an interdependent bee-plant relationship. Journal of Chemical Ecology 40:1126–

1134.

Cohen, J. 1988. Statistical Power Analysis for the Behavioral Sciences. Academic Press.

Decker, D. S. 1988. Origin(s), evolution, and systematics of Cucurbita pepo(Cucurbitaceae). Economic

Botany 42:4–15.

Detzel, A., and M. Wink. 1993. Attraction, deterrence or intoxication of bees (Apis mellifera) by plant

allelochemicals. Chemoecology 4:8–18.

Dunn, O. J. 1964. Multiple comparisons using rank sums. Technometrics 6:241–252.

Engels, B. 2015. XNomial: Exact goodness-of-fit test for multinomial data with fixed probabilities.

Glassmire, A. E., C. S. Jeffrey, M. L. Forister, T. L. Parchman, C. C. Nice, J. P. Jahner, J. S. Wilson, T.

R. Walla, L. A. Richards, A. M. Smilanich, M. D. Leonard, C. R. Morrison, W. Simbaña, L. A.

Salagaje, C. D. Dodson, J. S. Miller, E. J. Tepe, S. Villamarin‐Cortez, and L. A. Dyer. 2016.

Intraspecific phytochemical variation shapes community and population structure for specialist

caterpillars. New Phytologist 212:208–219.

Gong, L., H. S. Paris, M. H. Nee, G. Stift, M. Pachner, J. Vollmann, and T. Lelley. 2012. Genetic

relationships and evolution in Cucurbita pepo (pumpkin, squash, gourd) as revealed by simple

sequence repeat polymorphisms. TAG. Theoretical and applied genetics. Theoretische und

angewandte Genetik 124:875–891.

Holm, S. 1979. A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics

6:65–70.

Hurd, P. D., E. G. Linsley, and A. D. Michelbacher. 1974. Ecology of the squash and gourd bee,

Peponapis pruinosa, on cultivated cucurbits in California (Hymenoptera: Apoidea). Smithsonian

Contributions to Zoology:1–17.

Hurd, P. D., E. G. Linsley, and T. W. Whitaker. 1971. Squash and gourd bees (Peponapis, Xenoglossa)

and the origin of the cultivated Cucurbita. Evolution 25:218–234.

Julier, H. E., and T. H. Roulston. 2009. Wild Bee Abundance and Pollination Service in Cultivated

Pumpkins: Farm Management, Nesting Behavior and Landscape Effects. Journal of Economic

Entomology 102:563–573.

Kessler, A., and R. Halitschke. 2009. Testing the potential for conflicting selection on floral chemical

traits by pollinators and herbivores: Predictions and case study. Functional Ecology 23:901–912.

Kevan, P. G., N. A. Mohr, M. D. Offer, and J. R. Kemp. 1989. The squash and gourd bee, Peponapis

pruinosa (Hymenoptera). Proceedings of the Entomological Society of Ontario 119:9–15.

Kleine, S., and C. Müller. 2011. Intraspecific plant chemical diversity and its relation to herbivory.

Oecologia 166:175–186.

Knauer, A. C., and F. P. Schiestl. 2015. Bees use honest floral signals as indicators of reward when

visiting flowers. Ecology Letters 18:135–143.

Lev-Yadun, S., G. Ne’eman, and U. Shanas. 2009. A sheep in wolf’s clothing: Do carrion and dung

odours of flowers not only attract pollinators but also deter herbivores? BioEssays 31:84–88.

Mathewson, J. A. 1968. Nest construction and life history of the eastern cucurbit bee, Peponapis pruinosa

(Hymenoptera: Apoidea). Journal of the Kansas Entomological Society 41:255–261.

Milet-Pinheiro, P., M. Ayasse, H. E. M. Dobson, C. Schlindwein, W. Francke, and S. Dötterl. 2013. The

chemical basis of host-plant recognition in a specialized bee pollinator. Journal of Chemical

Ecology 39:1347–1360.

Muth, F., J. S. Francis, and A. S. Leonard. 2016. Bees use the taste of pollen to determine which flowers

to visit. Biology Letters 12.

Nocentini, D., M. Guarnieri, and C. Soligo. 2015. Nectar defense and hydrogen peroxide in floral nectar

of Cucurbita pepo. Acta Agrobotanica 68:187–193.

Ogle, D. 2017. FSA: Simple fisheries stock assessment methods.

Okamura, Y., Y. Sawada, M. Y. Hirai, and M. Murakami. 2016. Effects of different secondary metabolite

profiles in plant defense syndromes on specialist and generalist herbivores. Entomological

Science 19:97–103.

Oksanen, J., F. G. Blanchet, M. Friendly, R. Kindt, P. Legendre, D. McGlinn, P. R. Minchin, R. B.

O’Hara, G. L. Simpson, P. Solymos, M. H. H. Stevens, E. Szoecs, and H. Wagner. 2018. vegan:

Community ecology package.

Palmer-Young, E. C., A. Hogeboom, A. J. Kaye, D. Donnelly, J. Andicoechea, S. J. Connon, I. Weston,

K. Skyrm, R. E. Irwin, and L. S. Adler. 2017. Context-dependent medicinal effects of anabasine

and infection-dependent toxicity in bumble bees. PLOS ONE 12:e0183729.

Paris, H. S., N. Yonash, V. Portnoy, N. Mozes-Daube, G. Tzuri, and N. Katzir. 2003. Assessment of

genetic relationships in Cucurbita pepo (Cucurbitaceae) using DNA markers. TAG. Theoretical

and applied genetics. Theoretische und angewandte Genetik 106:971–978.

Petersen, J. D., S. Reiners, and B. A. Nault. 2013. Pollination services provided by bees in pumpkin fields

supplemented with either Apis mellifera or Bombus impatiens or not supplemented. PLoS ONE

8:e69819.

Pinheiro, M., I. Alves-dos-Santos, and M. Sazima. 2017. Flowers as sleeping places for male bees:

Somehow the males know which flowers their females prefer. Arthropod-Plant Interactions

11:329–337.

Poelman, E. H., N. M. Van Dam, J. J. A. Van Loon, L. E. M. Vet, and M. Dicke. 2009. Chemical

diversity in Brassica oleracea affects biodiversity of insect herbivores. Ecology 90:1863–1877.

Raguso, R. A. 2008. Start making scents: The challenge of integrating chemistry into pollination ecology.

Entomologia Experimentalis et Applicata 128:196–207.

Richardson, L. L., L. S. Adler, A. S. Leonard, J. Andicoechea, K. H. Regan, W. E. Anthony, J. S.

Manson, and R. E. Irwin. 2015. Secondary metabolites in floral nectar reduce parasite infections

in bumblebees. Proceedings of the Royal Society B: Biological Sciences 282:20142471–

20142471.

Richardson, L. L., M. D. Bowers, and R. E. Irwin. 2016. Nectar chemistry mediates the behavior of

parasitized bees: consequences for plant fitness. Ecology 97:325–337.

Rosenthal, J. P., and R. Dirzo. 1997. Effects of life history, domestication and agronomic selection on

plant defence against insects: Evidence from maizes and wild relatives. Evolutionary Ecology

11:337–355.

Sanjur, O. I., D. R. Piperno, T. C. Andres, and L. Wessel-Beaver. 2002. Phylogenetic relationships among

domesticated and wild species of Cucurbita (Cucurbitaceae) inferred from a mitochondrial gene:

Implications for crop plant evolution and areas of origin. Proceedings of the National Academy of

Sciences of the United States of America 99:535–540.

Schiestl, F. P. 2010. The evolution of floral scent and insect chemical communication. Ecology Letters

13:643–656.

Shuler, R. E., T. H. Roulston, and G. E. Farris. 2005. Farming practices influence wild pollinator

populations on squash and pumpkin. Journal of Economic Entomology 98:790–795.

Strauss, S. Y., R. E. Irwin, and V. M. Lambrix. 2004. Optimal defence theory and flower petal colour

predict variation in the secondary chemistry of wild radish. Journal of Ecology 92:132–141.

Theis, N., N. A. Barber, S. D. Gillespie, R. V. Hazzard, and L. S. Adler. 2014. Attracting mutualists and

antagonists: Plant trait variation explains the distribution of specialist floral herbivores and

pollinators on crops and wild gourds. American Journal of Botany 101:1314–1322.

Thornburg, R. W., C. Carter, A. Powell, R. Mittler, L. Rizhsky, and H. T. Horner. 2003. A major function

of the tobacco floral nectary is defense against microbial attack. Plant Systematics and Evolution

238:211–218.

Turcotte, M. M., H. Araki, D. S. Karp, K. Poveda, and S. R. Whitehead. 2017. The eco-evolutionary

impacts of domestication and agricultural practices on wild species. Philosophical Transactions of

the Royal Society of London. Series B, Biological Sciences 372.

Turlings, T. C. J., and F. Wäckers. 2004. Recruitment of predators and parasitoids by herbivore-injured

plants. Pages 21–75 in R. T. Cardé and J. G. Millar, editors. Advances in Insect Chemical

Ecology. Cambridge University Press.

Venables, W. N., and B. D. Ripley. 2002. Modern Applied Statistics with S, 4th ed. Springer, New York.

Weng, J.-K., R. N. Philippe, and J. P. Noel. 2012. The rise of chemodiversity in plants. Science (New

York, N.Y.) 336:1667–1670.

Wetzel, W. C., H. M. Kharouba, M. Robinson, M. Holyoak, and R. Karban. 2016. Variability in plant

nutrients reduces insect herbivore performance. Nature 539:425–427.

Wetzel, W. C., and J. S. Thaler. 2016. Does plant trait diversity reduce the ability of herbivores to defend

against predators? The plant variability-gut acclimation hypothesis. Current Opinion in Insect

Science 14:25–31.

Whitehead, S. R., and M. D. Bowers. 2014. Chemical ecology of fruit defence: Synergistic and

antagonistic interactions among amides from Piper. Functional Ecology 28:1094–1106.

Whitehead, S. R., M. M. Turcotte, and K. Poveda. 2017. Domestication impacts on plant-herbivore

interactions: A meta-analysis. Philosophical Transactions of the Royal Society of London. Series

B, Biological Sciences 372.

Wright, J. P., G. M. Ames, and R. M. Mitchell. 2016. The more things change, the more they stay the

same? When is trait variability important for stability of ecosystem function in a changing

environment. Philosophical Transactions of the Royal Society B: Biological Sciences 371.

Community ecology of microbes in the nests of a solitary, oligolectic,

ground-nesting bee

Kristen Brochu *a and Bryan Danforth a

a Department of Entomology, Cornell University, Ithaca, USA

* Corresponding author: [email protected]

Abstract

The nutrient rich provision mass of pollen and nectar that provides the base of all larval bee nutrition also acts as a substrate for an incredible diversity of microbes, i.e. bacteria and fungi, which play largely unknown roles. Many of bee microbial studies have focused on the effects of pathogens as a major source of colony mortality for managed bees in agricultural settings, but the microbial communities of wild unmanaged pollinators are relatively understudied, with the majority of studies focusing on generalist stem-nesting bees. Solitary, specialist, ground-nesting bees may present a very different situation for microbial community assembly due to consistent microbial associates and homogeneous environmental conditions. We examined the microbial community present in the nests of an oligolectic ground-nesting bee (the squash bee, Peponapis pruinosa) in order to investigate the diversity of the microbial community in healthy nests, how this microbial community varies across space and time, and the likely sources of acquisition for microbes associated with these nests. We found a diverse community with a total of 137 bacterial

OTUs and 226 fungal OTUs in our pollen provisions with greater than 1% relative abundance.

Microbial communities were extremely stable at the regional scale across years, with site-to-site

variation within years accounting for about 25% of the heterogeneity across samples. Pollen provision communities were also found to be distinct from communities of other sample types.

Finally, the core bacterial community in pollen provisions likely originated from pollen and adult bee digestive tract microbial communities, while the core fungal community in pollen provisions is likely influenced by soil and pollen microbial communities. Our results suggest that the microbial community associated with Peponapis pruinosa is diverse, stable, and unique to the brood cell environment. These reliable and persistent associations may pave the way for the evolution of mutualistic interactions and co-evolution.

Introduction

Microbes are often found associated with insects in a continuum of interactions that vary in their complexity and impact (Degli Esposti and Martinez Romero 2017). Bacterial and fungal diseases of insects are common (Huang 2003, Forsgren 2010), but microbes also play a variety of beneficial roles. Microbes are involved in highly obligate symbioses with insects, both nutritional (Mueller et al. 2005, Paludo et al. 2018) and defensive (Flórez et al. 2015). Bees may have an especially close relationship with microbes as they spend a large proportion of their life nestled amongst a nutrient-laden provision mass of pollen and nectar that is a rich substrate for microbial growth. This provision mass provides the base of all larval bee nutrition and is also host to an incredible diversity of microbes, including bacteria and fungi, which play largely unknown roles. Early culture-based studies provided the foundational work that showed a diverse community of bacteria (Lactobacillus), yeasts (Saccharomyces, Pichia, Hansenula,

Candida, Kloeckera, and Rhodotorula), and other fungi (Fusarium and Rhizopus) associated with both wild and domesticated bees (Batra et al. 1973, Goerzen 1991, Inglis et al. 1993, Rosa et al. 1999, 2003, Pimentel et al. 2005). Advancements in molecular techniques in recent years 60

have greatly increased our knowledge of microbial communities associated with provision

masses, with most of the focus centered on managed bees such as honey bees and bumblebees

(Sabree et al. 2012, Koch et al. 2012, Graystock et al. 2017). Many of these studies have focused

on the effects of pathogens, such as Nosema or European foulbrood, as a major source of colony

mortality for managed bees (e.g. honey bees and bumblebees) in agricultural settings (Forsgren

2010, Evans and Schwarz 2011, Graystock et al. 2015, Erban et al. 2017, Li et al. 2017).

However, there is an increasing body of evidence that some bees can benefit from mutualistic

microbes that inhabit their pollen/nectar provisions (Roberts 1971, Gilliam et al. 1990, Rosa et

al. 1999, McFrederick et al. 2012, 2013, 2014a, Menezes et al. 2015).

The microbial communities of wild unmanaged pollinators are relatively understudied,

with the majority of studies focusing on generalist stem-nesting bees (Gilliam et al. 1990,

Goerzen 1991, Inglis et al. 1993, Rosa et al. 2003, Keller et al. 2013, 2013, McFrederick et al.

2014a, 2014b, Menezes et al. 2015, McFrederick and Rehan 2016, Graystock et al. 2017). In

particular, there is a notable lack of metagenomic studies of solitary ground-nesting bees

(Roberts 1971, Batra et al. 1973, Gilliam et al. 1984, Rosa et al. 1999, Pimentel et al. 2005),

which actually comprise the majority of currently described bee species (Cane and Neff 2011)

and specialist bees, which collect pollen from a restricted set of plant genera or species (Cane

and Sipes 2006). Less attention has been paid to solitary bees, but studies have revealed a lack of

consistent microbiomes (Martinson et al. 2011, Koch et al. 2013, Graystock et al. 2017),

suggesting that these species acquire microbes stochastically from the environment, most likely

through pollen and nectar feeding (Inglis et al. 1993, Brysch-Herzberg 2004, McFrederick et al.

2012, 2013, 2014a, 2014b, 2017, Golonka and Vilgalys 2013, McFrederick and Rehan 2016).

Studies on bacteria and yeasts in floral nectar suggest that nectar microbial communities are

highly variable depending on the environmental conditions (Herrera et al. 2009, Pozo et al. 2011,

Jacquemyn et al. 2013a, 2013b, Golonka and Vilgalys 2013, Álvarez-Pérez et al. 2013, Schaeffer et al. 2015, Canto et al. 2017) and are likely to act as sites of reciprocal microbial exchange with floral visitors (Durrer and Schmid-Hempel 1994, Brysch-Herzberg 2004, Junker et al. 2011,

Belisle et al. 2012, Canto and Herrera 2012, Martinson et al. 2012, Shade et al. 2013b,

Aizenberg-Gershtein et al. 2013, Samuni-Blank et al. 2014, Pozo et al. 2014, Graystock et al.

2015, Junker and Keller 2015, Kwon et al. 2018). Habitat filtering can occur in nectar, resulting in relatively species-poor microbial communities that typically possess special adaptations to survive in these unique conditions (i.e., high sucrose levels resulting in osmotic stress, low nitrogen content, and possible growth inhibitors or toxins; de Vega et al. 2009, Herrera et al.

2010, Álvarez-Pérez et al. 2012). For ground-nesting bees, soil may also be a likely source of microbial acquisition and has been found to have a highly variable microbial community, but this pattern is more likely due to the predominant importance of abiotic factors (Andrew et al. 2012).

In particular, characteristics such as carbon availability can predict the abundance and

composition of bacteria found in soils (Fierer et al. 2007, Trivedi et al. 2017).

Solitary, specialist, ground-nesting bees may present a very different situation for

microbial community assembly. Pollen specialist bees which visit fewer flower species than

generalists may be exposed to reduced microbial diversity, unless their host plant has an

unexpectedly diverse microbiome (McFrederick and Rehan 2016, McFrederick et al. 2017). In

addition, the conditions in a brood cell under the ground are likely to be much more

environmentally homogeneous than for brood cells that occur above-ground in stems or in

colonies (DeBano et al. 1979, Cane and Neff 2011, Menezes et al. 2015), potentially weakening

priority effects of community assembly (i.e., the dominance of early colonizing species),

allowing particular microbes to consistently outcompete others in this environment (Tucker and

Fukami 2014). This combination of regular microbial associates with stable environmental

conditions could create the ideal circumstances for a microbial community to consistently

assemble in predictable ways (Nemergut et al. 2013, Adair et al. 2018). Specialist bees also

allow for a direct comparison of microbial communities between locally available pollen sources

and the provision masses since the identity of pollen in the provision can be determined.

Insights into the composition of a healthy microbial community of ground-nesting bee

provision masses and how they are influenced over time at local and regional scales can help to

gain a better understanding of microbial community assembly. We examined the microbial

community present in the nests of an oligolectic ground-nesting bee (the squash bee, Peponapis

pruinosa) in order to investigate the following questions: 1) how diverse is the bacterial and

fungal community, hereafter referred to collectively as the microbial community, in healthy

squash bee nests, 2) does this microbial community vary across space and time, 3) what are the

likely source communities for microbes associated with squash bee nests (e.g. pollen, nectar, soil, and adult bee digestive systems), and 4) how does the nest microbial community compare to environmental microbial communities? If microbial communities in solitary, specialist, ground- nesting bees are similar to those of stem-nesting generalist bees, we hypothesize that microbial communities will vary strongly across space and time. We expect strong local environmental influences, with greater overlap in the microbiota of provision masses and locally available pollen and nectar, rather than from the soil, due to the former being the primary constituents of provision masses. If, however; microbial communities in solitary, specialist, ground-nesting bees do not behave similarly to those of other bees, we might predict a more stable community due to consistent microbial associates in the local environmental community. To assess this, we

compared microbial communities from pollen provisions, pollen, nectar, soil, and adult bee

digestive systems. We also test the hypothesis that pollen provisions are less diverse than

environmental samples with less even communities, due to particular environmental conditions.

Methods

Study System

Bees in the genus Peponapis (Apidae, Eucerini) are abundant and effective specialist pollinators of cucurbits, squash and pumpkin, which are highly dependent on bee pollinators to produce good quality fruit, rendering pollination services especially important in this high-value crop (Tepedino 1981, Canto-Aguilar and Parra-Tabla 2000, Meléndez-Ramirez et al. 2002,

Shuler et al. 2005, Julier and Roulston 2009). The only species in this genus found in the United

States, Peponapis pruinosa, is widespread throughout North America and nests in aggregations in the ground on the edges of cucurbit fields. This economically important solitary bee is abundant in agricultural fields and its close association with cucurbits allows us to assess the local floral microbiome, making it an ideal model organism for investigating the environmental influences of the microbial community associated with ground-nesting bees.

Sample Acquisition

Field nest surveys were performed in cucurbit fields at 25 farms across the Finger Lakes

Region of upstate New York over the course of three summers (2014, 2015, and 2016). Surveys were performed between 5 – 10am to coincide with peak squash bee activity in order to visually assess the presence of squash bee nest aggregations. Sixteen nesting aggregations were found over the three years (six in 2014, and five in both 2015 and 2016). Locality information for each nest site can be found in Table S1. Nests were then excavated and brood cell contents separated, with pollen provisions and egg/larva comprising separate samples. We collected 5-10 pollen

provisions from each of the 16 nesting aggregations for a total of 136 samples. In order to

compare pollen provisions to the local environmental microbial community, we also collected

soil samples at the same depth as nest brood cells at each site as well as pollen and nectar

samples from a random sampling of Cucurbita flowers in fields adjacent to nest sites. Adult bees

were collected at each site both on flowers and in nests in order to assess the microbial

community of adult digestive systems.

All field samples were collected using ethanol sterilized tools and stored in sterile screw-

cap vials which were immediately preserved in liquid nitrogen. Samples were stored at -80°C

until DNA extraction. Adult bees were defrosted, washed in 95% ethanol, 6% bleach, and a

second wash of 95% ethanol before dissection in sterile Ringer’s solution. Their digestive system

was then extracted using sterile tools and preserved at -80°C until DNA extraction.

From each nest site in each year, we chose five brood cells containing larvae and five brood cells containing eggs for analysis from across our sampling dates to ensure brood cells were from different nests, as well as three samples each of soil, pollen, nectar, and adult bee digestive systems.

DNA Extraction and Illumina Sequencing

We used a phenol-chloroform protocol to extract DNA from all our samples which can be found in detail on the Danforth Lab website (Danforth et al. 2011). We mechanically ground samples under liquid nitrogen for 120 seconds before the chemical lysis step. DNA quality was assessed using a NanoDrop 2000 and only pure DNA samples were sent for sequencing.

Sequencing was conducted at MR DNA (www.mrdnalab.com, Shallowater, TX, USA) for the rRNA gene V4 variable region (hereafter 16S, primers 27F/519) to assess bacterial

communities and for the internal transcribed spacer 1 gene region (herafter ITS1, primers

ITS1F/ITS4R) to assess fungal communities. Each primer set, with the barcode located on the

forward primer, was used with the HotStarTaq Plus Master Mix Kit (Qiagen, USA) to complete

the following PCR steps: 94°C for 3 minutes, followed by 28 cycles of 94°C for 30 seconds,

53°C for 40 seconds and 72°C for 1 minute, with a final elongation step at 72°C for 5 minutes.

These PCR products were then run on 2% agarose gel to verify amplification success and

relative band intensity. Samples were then pooled (proportionally according to their molecular

weight and DNA concentration) and purified using calibrated Ampure XP beads with the resulting product used to construct the Illumina DNA library. Samples were sequenced according to the manufacturer’s guidelines on a MiSeq instrument.

Read Processing and OTU Assignment

Sequence data were processed with QIIME software. Sequences were joined, demultiplexed and filtered for quality (removing sequences <150bp or with ambiguous base calls). Chimaeras were detected and removed and Operational Taxonomic Units (OTUs) generated by clustering at 97% sequence similarity. Reference sequences were picked from each cluster and taxonomy was assigned by three methods. We compared reference OTUs to the

Greengenes database for 16S and the UNITE repository for ITS1 genes, as well as BLASTn searches on NCBI’s Nucleotide Collection database (www.ncbi.nlm.nih.gov) and RDP Classifier

(http://rdp.cme.msu.edu). We removed mitochondria, chloroplast, and Viridiplantae sequences.

Community Ecology Analysis

We defined the core microbial community as bacterial and fungal OTUs that were

present in greater than 50% of samples and had a greater than 1% relative abundance within the

sample type (Graystock et al. 2017). In order to test the hypothesis that provision mass

communities are different from environmental communities we compared alpha diversity of the

bacterial and fungal community across our sample types by performing Kruskal-Wallis tests (due to unequal variances) on OTU richness and PIE (the probability of an inter-specific encounter, a common sample-size-independent measure of evenness, Hurlbert 1971) diversity metrics with post-hoc Dunn’s tests of multiple comparisons using the R package “FSA” (Dunn 1964, Ogle

2017). We also compared bacterial and fungal alpha diversity for these sample types using

Student’s t-tests. In order to test the hypotheses that provision mass communities are distinct from environmental communities and that they vary across space and time, we compared beta diversity using non-metric multidimensional scaling (NMDS) of Bray-Curtis dissimilarity matrices performed with the R package “vegan” (Oksanen et al. 2018). Functions ‘adonis’ and

‘anosim’ were used to assess differences in community composition across sample types as well as across years, sites (strata=years), months (strata = years and sites), and developmental stage

(strata = years, sites, and months), in provision masses. Dispersion of groups was assessed using the ‘betadisper’ function in the R package “vegan” (Oksanen et al. 2018).

Results

We found a total of 162 bacterial OTUs and 243 fungal OTUs in our pollen provisions

that had greater than 1% relative abundance (Figure 1). The bacterial community was dominated

by Lactobacillus at 79% relative abundance, while the fungal community was more evenly

distributed with the four most common fungi representing approximately 50% of the relative

abundance (Fusarium oxysporum at 18%, Cladosporium cladosporioides at 14%, Alternaria

alternata at 11%, and Starmerella sp. at 6%). We found four bacteria and 14 fungi that

comprised the core microbiota of the pollen provisions (present in greater than 50% of samples

and had a greater than 1% relative abundance within the samples, Figure 1).

Figure 1. Pareto chart showing the overall abundance of core bacterial (a) and fungal (b) communities of

pollen provisions. Lines represent cumulative percentages.

We found that OTU richness and PIE (the probability of an inter-specific encounter, a

common sample-size-independent measure of evenness, Hurlbert 1971) differed by sample type

(Figure 2) for both bacterial (H=114.07, p < 0.001, and H=137.71, p < 0.001) and fungal

communities (H=78.622, p < 0.001, and H=45.71, p < 0.001). Results of post-hoc Dunn’s tests with the Holm correction (Holm 1979) showed that both pollen and pollen provisions had greater bacterial OTU richness than either adult digestive systems or nectar, while soil had higher bacterial OTU richness than all other sample types (Table S2). Soil had higher fungal OTU richness than all other samples, but was not significantly different from adult digestive systems, while nectar and pollen provisions had lower fungal OTU richness than all other sample types.

Soil communities had higher bacterial and fungal PIE than the other sample types, while bacterial PIE of pollen provisions was lower than all other types.

Figure 2. Comparison of alpha diversity metrics across sample types. DS = Adult bee digestive system,

PP = pollen provision mass. Boxplots show distribution of a) Bacterial OTU Richness, b) Fungal OTU

Richness, c) Bacterial PIE, and d) Fungal PIE across sample types. Dot indicates mean and dashed line indicates standard error.

Bacterial OTU richness was higher than fungal OTU richness in pollen provisions

(t=15.517, p<2.2x10-16), pollen (t=4.501, p=6.124x10-5), and soil (t=9.446, p=2.607x10-10), but

lower than fungal OTU richness in adult digestive systems (t=4.364, p=6.634x10-5). Nectar OTU

richness was not different in bacterial and fungal communities. Fungal PIE was higher than

bacterial PIE in pollen provisions (t=16.534, p<2.2x10-16) and adult digestive systems (t=4.231,

p=1.339x 10-4), but lower than bacterial PIE in soil (t=4.043, p=0.001), pollen (t=2.432,

p=0.018), and nectar (t=2.336, p=0.035).

Table 1. Adonis and anosim results Community Comparison Test Test-Statistic p-value Bacteria PP by Year adonis R2=0.116 p = 0.001 Bacteria PP by Year anosim R=0.098 p = 0.001 Fungi PP by Year adonis R2=0.073 p = 0.001 Fungi PP by Year anosim R=0.196 p = 0.001 Bacteria PP by Site adonis R2=0.227 p = 0.036 Bacteria PP by Site anosim R=0.081 p = 0.001 Fungi PP by Site adonis R2=0.232 p = 0.001 Fungi PP by Site anosim R=0.284 p = 0.001 Bacteria PP by Month adonis R2=0.026 p = 0.43 Bacteria PP by Month anosim R=-0.042 p = 0.855 Fungi PP by Month adonis R2=0.034 p = 0.156 Fungi PP by Month anosim R=0.160 p = 0.001 Bacteria PP by Developmental Stage adonis R2=0.025 p = 0.049 Bacteria PP by Developmental Stage anosim R=0.076 p = 0.003 Fungi PP by Developmental Stage adonis R2=0.064 p = 0.001 Fungi PP by Developmental Stage anosim R=0.186 p = 0.001 Bacteria Samples By Type adonis R2=0.408 p = 0.001 Bacteria Samples By Type anosim R=0.871 p = 0.001 Fungi Samples By Type adonis R2=0.149 p = 0.001 Fungi Samples By Type anosim R=0.375 p = 0.001

Bacterial and fungal communities in pollen provisions are different by year (Figure 3a &

3b, Table 2), with the bacterial community in 2016 being different from 2014 or 2015 (p =

0.003) and fungal communities different in all years (p = 0.003). Bacterial and fungal communities in pollen provisions are also different by site (Figure 3c & 3d, Table 2), although no sites were found to be different from each other with pairwise post-hoc anosim tests with a

Holm correction for multiple comparisons (Holm 1979), which tends to be very conservative with a large number of pairwise tests. Communities were also different by month, and developmental stage. Although we found many significant differences in community composition, the magnitude of these effects was small and they explained only a small component of the overall community variation (low R2 and R values) suggesting that these grouping factors are not, in fact, very important in differentiating the community, signifying that

these communities are relatively stable across sites, years, months, and even developmental

stages.

Figure 3. Non-metric multidimensional scaling (NMDS) plot comparing microbial communities of pollen

provisions. a) Bacterial samples across years. b) Fungal samples across years. c) Bacterial samples across

sites. d) Fungal samples across sites. Ellipses represent 95% confidence intervals. Colours represent years

or sites.

In contrast, we found that sample type substrate (i.e. soil, pollen, nectar, adult digestive system, and pollen provision) explained a much greater amount of variation in our bacterial

communities. Both the bacterial and fungal communities differed across sample types, with pairwise post-hoc anosim tests showing that each sample type is distinct from every other in both bacterial and fungal communities (p = 0.01, Figure 4, Table 2). This pattern is stronger in the bacterial community as evidenced by the higher R2 and R values compared to the fungal community.

Figure 4. Non-metric multidimensional scaling (NMDS) plot comparing microbial communities across sample type. DS = Adult bee digestive system, PP = pollen provision mass. a) Bacteria across sample type. b) Fungi across sample type. Samples are highly clustered by type. Ellipses represent 95% confidence intervals.

Pollen provisions share 84 bacterial OTUs and 99 fungal OTUs with soil, 82 bacterial

OTUs and 51 fungal OTUs with pollen, 14 bacterial OTUs and 42 fungal OTUs with adult bee digestive systems, and 40 bacterial OTUs and 19 fungal OTUs with nectar (Figure 5a & 5b). In particular, the core bacteria of pollen provisions were all found in pollen, adult bee digestive systems, and nectar, while only Lactobacillus spp. was found in soil (Figure 5c). The core fungi of pollen provisions were almost all found in soil (13/14), with a lower proportion present in

adult bee digestive systems (9/14), nectar (7/14), and pollen (6/14), suggesting that soil is an

important source of fungi in pollen provisions (Figure 5c).

Figure 5. Overlap between microbial communities in different sample types. a) Venn Diagram showing overlap between bacteria among types. b) Venn Diagram showing overlap between fungi among types.

Numbers within ellipses refer to the number of shared taxa. c) Heatmap comparing occurrence of core microbiota in samples across sample types. PP = pollen provisions, BeeDS = adult bee digestive system.

Discussion

Our study is the first to assess the composition and assemblage of the microbial community associated with an economically important, solitary ground-nesting bee species. We found a diverse community with a total of 137 bacterial OTUs and 226 fungal OTUs in our pollen provisions with greater than 1% relative abundance. Microbial communities were stable at the regional scale across years, with site-to-site variation within years accounting for about 25% of the heterogeneity across samples. Pollen provision communities were also found to be distinct from communities of other sample types. Finally, the core bacterial community in pollen provisions likely originated from pollen and adult bee digestive system microbial communities, while the core fungal community in pollen provisions is likely influenced by soil and pollen microbial communities.

Although our study does not allow us to infer active functional roles for the microbes we detected, those that are consistently present are likely to be active with a potential functional role

(Aleklett et al. 2014). Many studies agree upon the core bacterial species of the honey bee digestive system microbiome, which represent, on average, 95% of the worker digestive system bacteria (Sabree et al. 2012). In Peponapis, the core pollen provision bacterial community represents 84% of the community and the core fungal community represents 75%. The three most abundant core bacterial OTUs in pollen provisions are also found in high abundance in adult bee digestive systems, with Lactobacillus spp. and Saccharibacter floricola also found in moderate abundance in pollen. The dominant bacteria in our pollen provisions, Lactobacillus, has also been found to be common in the digestive systems and pollen masses of honey bees and bumblebees, along with Bifidobacterium, where they are thought to contribute to the conversion of nectar into honey through fermentation as well as protecting the provisions against pathogenic

bacteria (Olofsson and Vásquez 2008, 2009, Vásquez and Olofsson 2009, Vásquez et al. 2012).

Species of Lactobacillus have also been found in the digestive systems of several species of wild

bees, and in the pollen provision masses of two species, where it most likely serves to defend

pollen provisions from other bacteria through fermentation or production of anti-microbial compounds (McFrederick et al. 2012, 2013, 2014b). The presence of these core bacteria in adult bee digestive systems suggests that they may be inoculating pollen provisions when regurgitating nectar for storage (Powell et al. 2014). Female burying beetles inoculate larval food sources with digestive system microbes in order to transmit symbionts to their offspring, where they outcompete any new larval or pupal digestive system colonizer microbes (Wang and Rozen

2017).

The core fungal community was more diverse but also contained a large number of plant pathogens (Cladosporium cladosporioides, Fusarium oxysporum, Alternaria alternata,

Podosphaera fusca, Epicoccum nigrum, Fusarium nectria haematococca) (Crowhurst et al.

1997, Pérez‐García et al. 2009). Pollen has been found to be an excellent vector of plant viruses,

but there are few reports of pollen or pollinators vectoring fungal pathogens (Huang 2003, Card

et al. 2007, Marques et al. 2013). In a European cultivar of pumpkin, flowers were found to host

important plant pathogens, both bacteria and fungi, and these pathogens were suggested to be

transported by pollen (Fürnkranz et al. 2012). Based on the presence of these fungal plant pathogens in our pollen and pollen provision samples, greater attention should be paid to

pollinators as potential vectors of fungal diseases. Additionally, fungal plant pathogens in pollen

provisions could act as competitors for bees if they also feed on the pollen provision. In contrast,

some fungi may play a defensive role for bees if they produce anti-microbial compounds, such as

Cladosporium cladosporioides, Candida sake, Epicoccum nigrum, Clonostachys bionectria

ochroleuca, and Trichoderma hamatum (Larena et al. 2005, El-Hassan et al. 2013, Samaga et al.

2014). Aside from the potential anti-microbial effects of various glandular secretions that waterproof the brood-cell (Cane et al. 1983), ground-nesting bees may be in particular need of symbionts that can reduce the activity of harmful microbes. In some cases fermentation of the pollen provision can lead to secondary fungal infection (Aspergillus, Penicillium, Ascosphaera,

Eurolium, Carpenteles, and Mucor) and ultimately to larval mortality (Batra et al. 1973, Goerzen

1991, Inglis et al. 1993, Rosa et al. 1999, 2003, Pimentel et al. 2005). Yeasts, including those in

the Starmerella clade, have been suggested to play a beneficial role for bees, often by providing

indirect (Rosa et al. 1999, 2003, Pimentel et al. 2005) or direct (Roberts 1971, Menezes et al.

2015) nutritional benefits, although one study found that yeasts were found in the lowest

densities in plants that are pollinated by Hymenoptera (de Vega et al. 2009). Some of the core

pollen provision fungi are common soil saprotrophs (Gibellulopsis (Verticillium) nigrescens and

Mortierella), while Reniforma strues is a species of yeast that we know very little about, but appears to be commonly found in adult bee digestive systems as well as pollen provisions

(Warcup 1951). Several core pollen provision fungi are abundant in pollen (Cladosporium

cladosporioides, Alternaria alternata, Podosphaera fusca, Epicoccum nigrum), but some are

more abundant in soils (Fusarium oxysporum, Fusarium nectria haematococca, Trichoderma

hamatum, Gibellulopsis (Verticillium) nigrescens, and Mortierella). Most pollinator studies tend

to focus on floral microbes as the major source of microbial acquisition, but for many plants, the

initial source of microbes is the soil (Zarraonaindia et al. 2015, de Souza et al. 2016, Wagner et

al. 2016, Sánchez-Cañizares et al. 2017), and based on the abundance of soil fungi in our core

pollen provision fungal community, soil is likely an important microbial reservoir for fungi

associated with ground-nesting bees.

The most likely source of microbial acquisition for pollen provisions is from pollen and

nectar, as its major constituents. Studies on bacteria and yeasts in floral nectar are common

(Herrera et al. 2009, Pozo et al. 2011, Jacquemyn et al. 2013a, 2013b, Golonka and Vilgalys

2013, Álvarez-Pérez et al. 2013, Schaeffer et al. 2015, Canto et al. 2017) and suggest that nectar

microbial communities are highly variable depending on the environmental conditions,

particularly biotic interactions with floral visitors (Brysch-Herzberg 2004, Canto and Herrera

2012, Martinson et al. 2012, Aizenberg-Gershtein et al. 2013, Samuni-Blank et al. 2014, Pozo et al. 2014). Soil has also been found to have a highly variable environmental community, but this pattern is more likely due to the predominant importance of abiotic factors (Andrew et al. 2012).

Soil pH and organic matter content have been found to influence the composition of soil microbial communities (Fierer and Jackson 2006, Lauber et al. 2009, Rousk et al. 2010, Griffiths et al. 2011). In our study, soil was the most species rich sample type, for both bacteria and fungi.

Bacterial communities were also generally more species rich than fungal communities in soil, pollen, and pollen provisions, but less species rich in adult bee digestive systems. Soil was also found to have the most even communities (both bacteria and fungi), while fungal communities were generally much more even than bacterial communities for all our other sample types.

Bacterial communities in our pollen provisions were highly uneven, dominated by a single bacterial OTU, Lactobacillus spp.

The microbial community associated with pollen provisions is distinct from other sample types suggesting that, although they acquire microbes from environmental sources, the specific conditions of the brood cell foster a microbial community that is unique. We also found that variation in the microbial community associated with pollen provisions was not well explained by year, season, or developmental stage. Site explained the most variation in our communities,

consistent with the expectation that microbes are acquired from local microbial communities

acting as reservoirs. This result disagrees with studies that found greater variation over time than

across spatial scales (Belisle et al. 2014) and variation across developmental stages (Ramalho et

al. 2017). Persistence and stability can be important criteria for assessing the core microbial

community (Shade et al. 2012, 2013a, Shade and Handelsman 2012). The high level of stability

in our pollen provisions could be due to several factors. Lower complexity of agricultural

landscapes may reduce the number of microbes that are associated with domesticated crop

plants, resulting in a relatively homogenous microbial community (Levine et al. 2011, Gámez-

Virués et al. 2015, Trivedi et al. 2016). In urban landscapes, birds were found to have altered microbial digestive system community composition, specifically lower species richness and changes in metabolic functional roles, compared with birds in landscapes with less urban area

(Teyssier et al. 2018). Diet can also be an important determinant of associated microbial communities (Russell et al. 2009, Anderson et al. 2012, Téfit et al. 2017). Bacterial communities in digestive systems of lady beetles were more diverse when fed a greater prey diversity and in certain landscapes where prey items were assumed to be more diverse (Tiede et al. 2017). Some

plants are associated with a disproportionally high number of microbes (McFrederick and Rehan

2016), and since specialist bees forage on a small subset of plants they may encounter a reduced

microbial fauna, although many of these bees will visit additional plant species for nectar. In

addition, soil likely acts as insulation for brood cells under the ground, moderating conditions

(like temperature and humidity) to create a more homogeneous environmental than that

experienced by brood cells occurring above-ground in stems or in colonies (DeBano et al. 1979,

Cane and Neff 2011, Menezes et al. 2015). Priority effects (i.e. the dominance of early colonizing species) in floral nectar communities can last across multiple floral generations of the

same plant (Toju et al. 2018), but homogeneous environments may weaken the importance of

these effects, allowing some microbes to consistently outcompete others in favorable

environments, regardless of the order of arrival (Tucker and Fukami 2014). This combination of regular microbial associates with stable environmental conditions could create the ideal circumstances for a microbial community to consistently assemble in predictable ways. Given the importance of local environmental variation in determining the potential colonizing microbial community, it is noteworthy that squash bee pollen provisions maintain such a stereotypical community.

Our results suggest that the microbial community associated with Peponapis pruinosa is diverse, stable, and unique to the brood cell environment. These reliable and persistent associations may pave the way for the evolution of mutualistic interactions and co-evolution.

Future work should continue to focus on evaluating the core microbiota in pollen provision of

solitary bees as this is an understudied extension of the larval microbiome and an over-looked

source of potential nutritional and defensive microbial associations.

References

Adair, K. L., M. Wilson, A. Bost, and A. E. Douglas. 2018. Microbial community assembly in wild

populations of the fruit fly Drosophila melanogaster. The ISME Journal 12:959–972.

Aizenberg-Gershtein, Y., I. Izhaki, and M. Halpern. 2013. Do honeybees shape the bacterial community

composition in floral nectar? PLOS ONE 8:e67556.

Aleklett, K., M. Hart, and A. Shade. 2014. The microbial ecology of flowers: An emerging frontier in

phyllosphere research. Botany 92:253–266.

Álvarez-Pérez, S., C. M. Herrera, and C. de Vega. 2012. Zooming-in on floral nectar: A first exploration

of nectar-associated bacteria in wild plant communities. FEMS Microbiology Ecology 80:591–

602.

Álvarez-Pérez, S., B. Lievens, H. Jacquemyn, and C. M. Herrera. 2013. Acinetobacter nectaris sp. nov.

and Acinetobacter boissieri sp. nov., isolated from floral nectar of wild Mediterranean insect-

pollinated plants. International Journal of Systematic and Evolutionary Microbiology 63:1532–

1539.

Anderson, K. E., J. A. Russell, C. S. Moreau, S. Kautz, K. E. Sullam, Y. Hu, U. Basinger, B. M. Mott, N.

Buck, and D. E. Wheeler. 2012. Highly similar microbial communities are shared among related

and trophically similar ant species. Molecular Ecology 21:2282–2296.

Andrew, D. R., R. R. Fitak, A. Munguia-Vega, A. Racolta, V. G. Martinson, and K. Dontsova. 2012.

Abiotic factors shape microbial diversity in Sonoran Desert soils. Applied and Environmental

Microbiology 78:7527–7537.

Batra, L. R., S. W. T. Batra, and G. E. Bohart. 1973. The mycoflora of domesticated and wild bees

(Apoidea). Mycopathologia et Mycologia Applicata 49:13–44.

Belisle, M., C. D. Mendenhall, F. Oviedo Brenes, and T. Fukami. 2014. Temporal variation in fungal

communities associated with tropical hummingbirds and nectarivorous bats. Fungal Ecology

12:44–51.

Belisle, M., K. G. Peay, and T. Fukami. 2012. Flowers as islands: Spatial distribution of nectar-inhabiting

microfungi among plants of Mimulus aurantiacus, a hummingbird-pollinated shrub. Microbial

Ecology 63:711–718.

Brysch-Herzberg, M. 2004. Ecology of yeasts in plant-bumblebee mutualism in Central Europe. FEMS

microbiology ecology 50:87–100.

Cane, J. H., S. Gerdin, and G. Wife. 1983. Mandibular gland secretions of solitary bees (Hymenoptera:

Apoidea): Potential for nest cell disinfection. Journal of the Kansas Entomological Society

56:199–204.

Cane, J. H., and J. L. Neff. 2011. Predicted fates of ground-nesting bees in soil heated by wildfire:

Thermal tolerances of life stages and a survey of nesting depths. Biological Conservation

144:2631–2636.

Cane, J. H., and S. D. Sipes. 2006. Characterizing floral specialization by bees: Analytical methods and a

revised lexicon for oligolecty. Page 99:122 Plant-Pollinator Interactions: From Specialization to

Generalization. University of Chicago Press.

Canto, A., and C. M. Herrera. 2012. Micro-organisms behind the pollination scenes: Microbial imprint on

floral nectar sugar variation in a tropical plant community. Annals of Botany 110:1173–1183.

Canto, A., C. M. Herrera, and R. Rodriguez. 2017. Nectar-living yeasts of a tropical host plant

community: Diversity and effects on community-wide floral nectar traits. PeerJ 5:e3517.

Canto-Aguilar, M. A., and V. Parra-Tabla. 2000. Importance of conserving alternative pollinators:

Assessing the pollination efficiency of the squash bee, Peponapis limitaris in Cucurbita moschata

(Cucurbitaceae). Journal of Insect Conservation 4:201–208.

Card, S. D., M. N. Pearson, and G. R. G. Clover. 2007. Plant pathogens transmitted by pollen.

Australasian Plant Pathology 36:455–461.

Crowhurst, R. N., S. J. Binnie, J. K. Bowen, B. T. Hawthorne, K. M. Plummer, J. Rees-George, E. H. A.

Rikkerink, and M. D. Templeton. 1997. Effect of disruption of a cutinase gene ( cutA ) on

virulence and tissue specificity of Fusarium solani f. sp. cucurbitae race 2 toward Cucurbita

maxima and C. moschata. Molecular Plant-Microbe Interactions 10:355–368.

Danforth, B. N., J. Fang, S. D. Sipes, S. G. Brady, E. A. B. Almeida, J. R. Litman, and S. Cardinal. 2011.

Phylogeny and molecular systematics of bees (Hymenoptera: Apoidea). Cornell University,

Ithaca, NY.

DeBano, L. F., R. M. Rice, and C. C. Eugene. 1979. Soil heating in chaparral fires: Effects on soil

properties, plant nutrients, erosion, and runoff. Res. Paper PSW-RP-145. Berkeley, CA: U.S.

Department of Agriculture, Forest Service, Pacific Southwest Forest and Range Experiment

Station. 21 p 145.

Degli Esposti, M., and E. Martinez Romero. 2017. The functional microbiome of arthropods. PloS One

12:e0176573.

Dunn, O. J. 1964. Multiple comparisons using rank sums. Technometrics 6:241–252.

Durrer, S., and P. Schmid-Hempel. 1994. Shared use of flowers leads to horizontal pathogen

transmission. Proc. R. Soc. Lond. B 258:299–302.

El-Hassan, S. A., S. R. Gowen, and B. Pembroke. 2013. Use of Trichoderma hamatum for biocontrol of

lentil vascular wilt disease: Efficacy, mechanisms of interaction and future prospects. Journal of

Plant Protection Research 53:12–26.

Erban, T., O. Ledvinka, M. Kamler, B. Hortova, M. Nesvorna, J. Tyl, D. Titera, M. Markovic, and J.

Hubert. 2017. Bacterial community associated with worker honeybees (Apis mellifera) affected

by European foulbrood. PeerJ 5:e3816.

Evans, J. D., and R. S. Schwarz. 2011. Bees brought to their knees: Microbes affecting honey bee health.

Trends in Microbiology 19:614–620.

Fierer, N., M. A. Bradford, and R. B. Jackson. 2007. Toward an ecological classification of soil bacteria.

Ecology 88:1354–1364.

Fierer, N., and R. B. Jackson. 2006. The diversity and biogeography of soil bacterial communities.

Proceedings of the National Academy of Sciences 103:626–631.

Flórez, L. V., P. H. W. Biedermann, T. Engl, and M. Kaltenpoth. 2015. Defensive symbioses of animals

with prokaryotic and eukaryotic microorganisms. Natural Product Reports 32:904–936.

Forsgren, E. 2010. European foulbrood in honey bees. Journal of Invertebrate Pathology 103:S5–S9.

Fürnkranz, M., B. Lukesch, H. Müller, H. Huss, M. Grube, and G. Berg. 2012. Microbial diversity inside

pumpkins: Microhabitat-specific communities display a high antagonistic potential against

phytopathogens. Microbial Ecology 63:418–428.

Gámez-Virués, S., D. J. Perović, M. M. Gossner, C. Börschig, N. Blüthgen, H. de Jong, N. K. Simons,

A.-M. Klein, J. Krauss, G. Maier, C. Scherber, J. Steckel, C. Rothenwöhrer, I. Steffan-Dewenter,

C. N. Weiner, W. Weisser, M. Werner, T. Tscharntke, and C. Westphal. 2015. Landscape

simplification filters species traits and drives biotic homogenization. Nature Communications

6:8568.

Gilliam, M., S. L. Buchmann, and B. J. Lorenz. 1984. Microbial flora of the larval provisions of the

solitary bees, Centris pallida and Anthophora sp. Apidologie 15:1–10.

Gilliam, M., S. L. Buchmann, B. J. Lorenz, and R. J. Schmalzel. 1990. Bacteria belonging to the genus

Bacillus associated with three species of solitary bees. Apidologie 21:99–105.

Goerzen, D. W. 1991. Microflora associated with the alfalfa leafcutting bee, Megachile rotundata (Fab)

(Hymenoptera: Megachilidae) in Saskatchewan, Canada. Apidologie 22:553–561.

Golonka, A. M., and R. Vilgalys. 2013. Nectar inhabiting yeasts in Virginian populations of Silene

latifolia (Caryophyllaceae) and coflowering species. The American Midland Naturalist 169:235–

258.

Graystock, P., D. Goulson, and W. O. H. Hughes. 2015. Parasites in bloom: Flowers aid dispersal and

transmission of pollinator parasites within and between bee species. Proc. R. Soc. B

282:20151371.

Graystock, P., S. M. Rehan, and Q. S. McFrederick. 2017. Hunting for healthy microbiomes: Determining

the core microbiomes of Ceratina, Megalopta, and Apis bees and how they associate with

microbes in bee collected pollen. Conservation Genetics 18:701–711.

Griffiths, R. I., B. C. Thomson, P. James, T. Bell, M. Bailey, and A. S. Whiteley. 2011. The bacterial

biogeography of British soils. Environmental Microbiology 13:1642–1654.

Herrera, C. M., A. Canto, M. I. Pozo, and P. Bazaga. 2010. Inhospitable sweetness: Nectar filtering of

pollinator-borne inocula leads to impoverished, phylogenetically clustered yeast communities.

Proceedings of the Royal Society of London B: Biological Sciences 277:747–754.

Herrera, C. M., C. de Vega, A. Canto, and M. I. Pozo. 2009. Yeasts in floral nectar: A quantitative survey.

Annals of Botany 103:1415–1423.

Holm, S. 1979. A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics

6:65–70.

Huang, H. c. 2003. Verticillium wilt of alfalfa: Epidemiology and control strategies. Canadian Journal of

Plant Pathology 25:328–338.

Hurlbert, S. H. 1971. The nonconcept of species diversity: A critique and alternative parameters. Ecology

52:577–586.

Inglis, G. D., L. Sigler, and M. S. Goettel. 1993. Aerobic microorganisms associated with alfalfa

leafcutter bees (Megachile rotundata). Microbial Ecology 26:125–143.

Jacquemyn, H., M. Lenaerts, R. Brys, K. Willems, O. Honnay, and B. Lievens. 2013a. Among-population

variation in microbial community structure in the floral nectar of the bee-pollinated forest herb

Pulmonaria officinalis L. PloS One 8:e56917.

Jacquemyn, H., M. Lenaerts, D. Tyteca, and B. Lievens. 2013b. Microbial diversity in the floral nectar of

seven Epipactis (Orchidaceae) species. MicrobiologyOpen 2:644–658.

Julier, H. E., and T. H. Roulston. 2009. Wild bee abundance and pollination service in cultivated

pumpkins: Farm management, nesting behavior and landscape effects. Journal of Economic

Entomology 102:563–573.

Junker, R. R., and A. Keller. 2015. Microhabitat heterogeneity across leaves and flower organs promotes

bacterial diversity. FEMS microbiology ecology 91:fiv097.

Junker, R. R., C. Loewel, R. Gross, S. Dötterl, A. Keller, and N. Blüthgen. 2011. Composition of

epiphytic bacterial communities differs on petals and leaves. Plant Biology 13:918–924.

Keller, A., G. Grimmer, and I. Steffan-Dewenter. 2013. Diverse microbiota identified in whole intact nest

chambers of the red mason bee Osmia bicornis (Linnaeus 1758). PLOS ONE 8:e78296.

Koch, H., D. P. Abrol, J. Li, and P. Schmid-Hempel. 2013. Diversity and evolutionary patterns of

bacterial gut associates of corbiculate bees. Molecular Ecology 22:2028–2044.

Koch, H., G. Cisarovsky, and P. Schmid-Hempel. 2012. Ecological effects on gut bacterial communities

in wild bumblebee colonies. The Journal of Animal Ecology 81:1202–1210.

Kwon, Y., J.-T. Lee, H. S. Kim, C. Jeon, and Y.-S. Kwak. 2018. Comparative tomato flower and

pollinator hive microbial communities. Journal of Plant Diseases and Protection 125:115–119.

Larena, I., R. Torres, A. De Cal, M. Liñán, P. Melgarejo, P. Domenichini, A. Bellini, J. F. Mandrin, J.

Lichou, X. O. de Eribe, and J. Usall. 2005. Biological control of postharvest brown rot (Monilinia

spp.) of peaches by field applications of Epicoccum nigrum. Biological Control 32:305–310.

Lauber, C. L., M. Hamady, R. Knight, and N. Fierer. 2009. Pyrosequencing-based assessment of soil pH

as a predictor of soil bacterial community structure at the continental scale. Applied and

Environmental Microbiology 75:5111–5120.

Levine, U. Y., T. K. Teal, G. P. Robertson, and T. M. Schmidt. 2011. Agriculture’s impact on microbial

diversity and associated fluxes of carbon dioxide and methane. The ISME journal 5:1683–1691.

Li, J. H., J. D. Evans, W. F. Li, Y. Z. Zhao, G. DeGrandi-Hoffman, S. K. Huang, Z. G. Li, M. Hamilton,

and Y. P. Chen. 2017. New evidence showing that the destruction of gut bacteria by antibiotic

treatment could increase the honey bee’s vulnerability to Nosema infection. PLOS ONE

12:e0187505.

Marques, J. P. R., L. Amorim, M. B. Spósito, D. Marin, and B. Appezzato-da-Glória. 2013. Infection of

citrus pollen grains by Colletotrichum acutatum. European Journal of Plant Pathology 136:35–40.

Martinson, E. O., E. A. Herre, C. A. Machado, and A. E. Arnold. 2012. Culture-free survey reveals

diverse and distinctive fungal communities associated with developing figs (Ficus spp.) in

Panama. Microbial Ecology 64:1073–1084.

Martinson, V. G., B. N. Danforth, R. L. Minckley, O. Rueppell, S. Tingek, and N. A. Moran. 2011. A

simple and distinctive microbiota associated with honey bees and bumble bees. Molecular

Ecology 20:619–628.

McFrederick, Q. S., J. J. Cannone, R. R. Gutell, K. Kellner, R. M. Plowes, and U. G. Mueller. 2013.

Specificity between Lactobacilli and Hymenopteran hosts is the exception rather than the rule.

Applied and Environmental Microbiology 79:1803–1812.

McFrederick, Q. S., U. G. Mueller, and R. R. James. 2014a. Interactions between fungi and bacteria

influence microbial community structure in the Megachile rotundata larval gut. Proceedings of

the Royal Society B: Biological Sciences 281:20132653–20132653.

McFrederick, Q. S., and S. M. Rehan. 2016. Characterization of pollen and bacterial community

composition in brood provisions of a small carpenter bee. Molecular Ecology 25:2302–2311.

McFrederick, Q. S., J. M. Thomas, J. L. Neff, H. Q. Vuong, K. A. Russell, A. R. Hale, and U. G. Mueller.

2017. Flowers and wild megachilid bees share microbes. Microbial Ecology 73:188–200.

McFrederick, Q. S., W. T. Wcislo, M. C. Hout, and U. G. Mueller. 2014b. Host species and

developmental stage, but not host social structure, affects bacterial community structure in

socially polymorphic bees. FEMS Microbiology Ecology 88:398–406.

McFrederick, Q. S., W. T. Wcislo, D. R. Taylor, H. D. Ishak, S. E. Dowd, and U. G. Mueller. 2012.

Environment or kin: Whence do bees obtain acidophilic bacteria? Molecular Ecology 21:1754–

1768.

Meléndez-Ramirez, V., S. Magaña-Rueda, V. Parra-Tabla, R. Ayala, and J. Navarro. 2002. Diversity of

native bee visitors of cucurbit crops (Cucurbitaceae) in Yucatán, México. Journal of Insect

Conservation 6:135–147.

Menezes, C., A. Vollet-Neto, A. J. Marsaioli, D. Zampieri, I. C. Fontoura, A. D. Luchessi, and V. L.

Imperatriz-Fonseca. 2015. A Brazilian social bee must cultivate fungus to survive. Current

Biology 25:2851–2855.

Mueller, U. G., N. M. Gerardo, D. K. Aanen, D. L. Six, and T. R. Schultz. 2005. The evolution of

agriculture in insects. Annual Review of Ecology, Evolution, and Systematics 36:563–595.

Nemergut, D. R., S. K. Schmidt, T. Fukami, S. P. O’Neill, T. M. Bilinski, L. F. Stanish, J. E. Knelman, J.

L. Darcy, R. C. Lynch, P. Wickey, and S. Ferrenberg. 2013. Patterns and processes of microbial

community assembly. Microbiology and Molecular Biology Reviews : MMBR 77:342–356.

Ogle, D. 2017. FSA: Simple fisheries stock assessment methods.

Oksanen, J., F. G. Blanchet, M. Friendly, R. Kindt, P. Legendre, D. McGlinn, P. R. Minchin, R. B.

O’Hara, G. L. Simpson, P. Solymos, M. H. H. Stevens, E. Szoecs, and H. Wagner. 2018. vegan:

Community ecology package.

Olofsson, T. C., and A. Vásquez. 2008. Detection and identification of a novel lactic acid bacterial flora

within the honey stomach of the honeybee Apis mellifera. Current Microbiology 57:356–363.

Olofsson, T. C., and A. Vásquez. 2009. Phylogenetic comparison of bacteria isolated from the honey

stomachs of honey bees Apis mellifera and bumble bees Bombus spp. Journal of Apicultural

Research 48:233–237.

Paludo, C. R., C. Menezes, E. A. Silva-Junior, A. Vollet-Neto, A. Andrade-Dominguez, G. Pishchany, L.

Khadempour, F. S. Nascimento, C. R. Currie, R. Kolter, J. Clardy, and M. T. Pupo. 2018.

Stingless bee larvae require fungal steroid to pupate. Scientific Reports 8:1122.

Pérez‐García, A., D. Romero, D. Fernández‐Ortuño, F. López‐Ruiz, A. D. Vicente, and J. A. Torés. 2009.

The powdery mildew fungus Podosphaera fusca (synonym Podosphaera xanthii), a constant

threat to cucurbits. Molecular Plant Pathology 10:153–160.

Pimentel, M. R. C., Y. Antonini, R. P. Martins, M.-A. Lachance, and C. A. Rosa. 2005. Candida

riodocensis and Candida cellae, two new yeast species from the Starmerella clade associated

with solitary bees in the Atlantic rain forest of Brazil. FEMS Yeast Research 5:875–879.

Powell, J. E., V. G. Martinson, K. Urban-Mead, and N. A. Moran. 2014. Routes of acquisition of the gut

microbiota of the honey bee Apis mellifera. Applied and Environmental Microbiology 80:7378–

7387.

Pozo, M. I., C. M. Herrera, and C. Alonso. 2014. Spatial and temporal distribution patterns of nectar-

inhabiting yeasts: How different floral microenvironments arise in winter-blooming Helleborus

foetidus. Fungal Ecology 11:173–180.

Pozo, M. I., C. M. Herrera, and P. Bazaga. 2011. Species richness of yeast communities in floral nectar of

southern Spanish plants. Microbial Ecology 61:82–91.

Ramalho, M. O., O. C. Bueno, and C. S. Moreau. 2017. Species-specific signatures of the microbiome

from Camponotus and Colobopsis ants across developmental stages. PLOS ONE 12:e0187461.

Roberts, R. B. 1971. Biology of the crepuscular bee Ptiloglossa guinnae n sp. with notes on associated

bees, mites, and yeasts. Journal of the Kansas Entomological Society 44:283–294.

Rosa, C. A., M.-A. Lachance, J. O. C. Silva, A. C. P. Teixeira, M. M. Marini, Y. Antonini, and R. P.

Martins. 2003. Yeast communities associated with stingless bees. FEMS Yeast Research 4:271–

275.

Rosa, C. A., E. M. Viana, R. P. Martins, Y. Antonini, and M.-A. Lachance. 1999. Candida batistae, a

new yeast species associated with solitary digger nesting bees in Brazil. Mycologia 91:428–433.

Rousk, J., E. Bååth, P. C. Brookes, C. L. Lauber, C. Lozupone, J. G. Caporaso, R. Knight, and N. Fierer.

2010. Soil bacterial and fungal communities across a pH gradient in an arable soil. The ISME

journal 4:1340–1351.

Russell, J. A., C. S. Moreau, B. Goldman-Huertas, M. Fujiwara, D. J. Lohman, and N. E. Pierce. 2009.

Bacterial gut symbionts are tightly linked with the evolution of herbivory in ants. Proceedings of

the National Academy of Sciences of the United States of America 106:21236–21241.

Sabree, Z. L., A. K. Hansen, and N. A. Moran. 2012. Independent studies using deep sequencing resolve

the same set of core bacterial species dominating gut communities of honey bees. PLOS ONE

7:e41250.

Samaga, P. V., V. R. Rai, and K. M. L. Rai. 2014. Bionectria ochroleuca NOTL33—an endophytic

fungus from Nothapodytes foetida producing antimicrobial and free radical scavenging

metabolites. Annals of Microbiology 64:275–285.

Samuni-Blank, M., I. Izhaki, S. Laviad, A. Bar-Massada, Y. Gerchman, and M. Halpern. 2014. The role

of abiotic environmental conditions and herbivory in shaping bacterial community composition in

floral nectar. PloS One 9:e99107.

Sánchez-Cañizares, C., B. Jorrín, P. S. Poole, and A. Tkacz. 2017. Understanding the holobiont: The

interdependence of plants and their microbiome. Current Opinion in Microbiology 38:188–196.

Schaeffer, R. N., R. L. Vannette, and R. E. Irwin. 2015. Nectar yeasts in Delphinium nuttallianum

(Ranunculaceae) and their effects on nectar quality. Fungal Ecology 18:100–106.

Shade, A., J. Gregory Caporaso, J. Handelsman, R. Knight, and N. Fierer. 2013a. A meta-analysis of

changes in bacterial and archaeal communities with time. The ISME Journal 7:1493–1506.

Shade, A., and J. Handelsman. 2012. Beyond the Venn diagram: The hunt for a core microbiome.

Environmental Microbiology 14:4–12.

Shade, A., P. S. McManus, and J. Handelsman. 2013b. Unexpected diversity during community

succession in the apple flower microbiome. mBio 4:e00602-12.

Shade, A., H. Peter, S. D. Allison, D. Baho, M. Berga, H. Buergmann, D. H. Huber, S. Langenheder, J. T.

Lennon, J. B. Martiny, K. L. Matulich, T. M. Schmidt, and J. Handelsman. 2012. Fundamentals

of microbial community resistance and resilience. Frontiers in Microbiology 3.

Shuler, R. E., T. H. Roulston, and G. E. Farris. 2005. Farming practices influence wild pollinator

populations on squash and pumpkin. Journal of Economic Entomology 98:790–795. de Souza, R. S. C., V. K. Okura, J. S. L. Armanhi, B. Jorrín, N. Lozano, M. J. da Silva, M. González-

Guerrero, L. M. de Araújo, N. C. Verza, H. C. Bagheri, J. Imperial, and P. Arruda. 2016.

Unlocking the bacterial and fungal communities assemblages of sugarcane microbiome.

Scientific Reports 6:28774.

Téfit, M. A., B. Gillet, P. Joncour, S. Hughes, and F. Leulier. 2017. Stable association of a Drosophila-

derived microbiota with its animal partner and the nutritional environment throughout a fly

population’s life cycle. Journal of Insect Physiology.

Tepedino, V. J. 1981. The pollination efficiency of the squash bee (Peponapis pruinosa) and the honey

bee (Apis mellifera) on summer squash (Cucurbita pepo). Journal of the Kansas Entomological

Society 54:359–377.

Teyssier, A., L. O. Rouffaer, N. Saleh Hudin, D. Strubbe, E. Matthysen, L. Lens, and J. White. 2018.

Inside the guts of the city: Urban-induced alterations of the gut microbiota in a wild passerine.

Science of The Total Environment 612:1276–1286.

Tiede, J., C. Scherber, J. Mutschler, K. D. McMahon, and C. Gratton. 2017. Gut microbiomes of mobile

predators vary with landscape context and species identity. Ecology and Evolution 7:8545–8557.

Toju, H., R. L. Vannette, M.-P. L. Gauthier, M. K. Dhami, and T. Fukami. 2018. Priority effects can

persist across floral generations in nectar microbial metacommunities. Oikos 127:345–352.

Trivedi, P., M. Delgado-Baquerizo, I. C. Anderson, and B. K. Singh. 2016. Response of soil properties

and microbial communities to agriculture: Implications for primary productivity and soil health

indicators. Frontiers in Plant Science 7.

Trivedi, P., M. Delgado-Baquerizo, T. C. Jeffries, C. Trivedi, I. C. Anderson, K. Lai, M. McNee, K.

Flower, B. Pal Singh, D. Minkey, and B. K. Singh. 2017. Soil aggregation and associated

microbial communities modify the impact of agricultural management on carbon content.

Environmental Microbiology 19:3070–3086.

Tucker, C. M., and T. Fukami. 2014. Environmental variability counteracts priority effects to facilitate

species coexistence: Evidence from nectar microbes. Proceedings of the Royal Society of London

B: Biological Sciences 281:20132637.

Vásquez, A., E. Forsgren, I. Fries, R. J. Paxton, E. Flaberg, L. Szekely, and T. C. Olofsson. 2012.

Symbionts as major modulators of insect health: Lactic acid bacteria and honeybees. PLoS ONE

Vásquez, A., and T. C. Olofsson. 2009. The lactic acid bacteria involved in the production of bee pollen

and bee bread. Journal of Apicultural Research 48:189–195. de Vega, C., C. M. Herrera, and S. D. Johnson. 2009. Yeasts in floral nectar of some South African

plants: Quantification and associations with pollinator type and sugar concentration. South

African Journal of Botany 75:798–806.

Wagner, M. R., D. S. Lundberg, T. G. Del Rio, S. G. Tringe, J. L. Dangl, and T. Mitchell-Olds. 2016.

Host genotype and age shape the leaf and root microbiomes of a wild perennial plant. Nature

Communications 7:12151.

Wang, Y., and D. E. Rozen. 2017. Gut microbiota colonization and transmission in the burying beetle

Nicrophorus vespilloides throughout development. Applied and Environmental Microbiology

83:e03250-16.

Warcup, J. H. 1951. The ecology of soil fungi. Transactions of the British Mycological Society 34:376–

399.

Zarraonaindia, I., S. M. Owens, P. Weisenhorn, K. West, J. Hampton-Marcell, S. Lax, N. A. Bokulich, D.

A. Mills, G. Martin, S. Taghavi, D. van der Lelie, and J. A. Gilbert. 2015. The soil microbiome

influences grapevine-associated microbiota. mBio 6:e02527-14.

Supplementary Materials

Table S1. Locality information for nest aggregations.

Year Site Code Lat Long 14 GR N 42.65 – N 42.66 W 76.47 – W 76.48 14 IK N 42.19 – N 42.20 W 76.31 – W 76.32 14 LY N 42.69 – N 42.70 W 77.02 – W 77.03 14 PD N 42.88 – N 42.89 W 77.07 – W 77.08 14 SH N 42.11 – N 42.12 W 76.92 – W 76.93 14 TP N 42.90 – N 42.91 W 76.34 – W 76.35 15 EM N 42.64 – N 42.65 W 76.82 – W 76.83 15 LY2 N 42.71 – N 42.72 W 77.02 – W 77.03 15 PD2 N 42.91 – N 42.92 W 77.10 – W 77.11 15 SH N 42.11 – N 42.12 W 76.92 – W 76.93 15 TM N 42.75 – N 42.76 W 77.11 – W 77.12 16 IK3 N 42.19 – N 42.20 W 76.31 – W 76.32 16 PD3 N 42.86 – N 42.87 W 77.14 – W 77.15 16 PG N 42.82 – N 42.83 W 77.12 – W 77.13 16 RV N 43.18 – N 42.19 W 76.41 – W 76.42 16 TP3 N 42.89 – N 42.90 W 76.34 – W 76.35

Table S2. Dunn’s test results for alpha diversity metrics. Significant results in bold.

Community Metric Comparison z p unadjusted p adjusted Bacteria OTU Richness Guts-Nectar 0.06761669 9.460908e-01 9.460908e-01 Bacteria OTU Richness Guts-Pollen -5.01084219 5.419234e-07 3.251541e-06 Bacteria OTU Richness Nectar-Pollen -4.23741994 2.261030e-05 6.783090e-05 Bacteria OTU Richness Guts-PP -6.83835109 8.010987e-12 6.408790e-11 Bacteria OTU Richness Nectar-PP -5.37569917 7.628607e-08 5.340025e-07 Bacteria OTU Richness Pollen-PP -0.77755435 4.368318e-01 8.736636e-01 Bacteria OTU Richness Guts-Soil -8.23802561 1.750757e-16 1.750757e-15 Bacteria OTU Richness Nectar-Soil -7.43589809 1.038599e-13 9.347390e-13 Bacteria OTU Richness Pollen-Soil -4.66519683 3.083218e-06 1.233287e-5 Bacteria OTU Richness PP-Soil -4.70665247 2.518177e-06 1.259088e-5 Fungi OTU Richness Guts-Nectar 3.7621597 1.684525e-04 1.010715e-03 Fungi OTU Richness Guts-Pollen 1.5359427 1.245524e-01 2.491049e-01 Fungi OTU Richness Nectar-Pollen -2.7381140 6.179266e-03 2.471706e-02 Fungi OTU Richness Guts-PP 5.5521343 2.822026e-08 2.539824e-07 Fungi OTU Richness Nectar-PP -0.5512677 5.814502e-01 5.814502e-01

Fungi OTU Richness Pollen-PP 4.0389538 5.369013e-05 3.758309e-04 Fungi OTU Richness Guts-Soil -1.7413693 8.161887e-02 2.448566e-01 Fungi OTU Richness Nectar-Soil -4.6167480 3.898001e-06 3.118401e-05 Fungi OTU Richness Pollen-Soil -2.9371279 3.312675e-03 1.656337e-02 Fungi OTU Richness PP-Soil -5.7216577 1.054897e-08 1.054897e-07 Bacteria PIE Guts-Nectar 0.5920256 5.538334e-01 1.000000e+00 Bacteria PIE Guts-Pollen 0.1271522 8.988200e-01 8.988200e-01 Bacteria PIE Nectar-Pollen -0.5093265 6.105234e-01 1.000000e+00 Bacteria PIE Guts-PP 4.6989807 2.614632e-06 2.091706e-05 Bacteria PIE Nectar-PP 2.9381701 3.301559e-03 1.320623e-02 Bacteria PIE Pollen-PP 4.9923951 5.963511e-07 5.367160e-06 Bacteria PIE Guts-Soil -3.0848326 2.036668e-03 1.018334e-02 Bacteria PIE Nectar-Soil -3.2556699 1.131252e-03 6.787511e-03 Bacteria PIE Pollen-Soil -3.2850594 1.019609e-03 7.137261e-03 Bacteria PIE PP-Soil -6.8415987 7.831424e-12 7.831424e-11 Fungi PIE Guts-Nectar 2.1863127 2.879273e-02 1.727564e-01 Fungi PIE Guts-Pollen 1.0090924 3.129303e-01 6.258607e-01 Fungi PIE Nectar-Pollen -1.5046256 1.324204e-01 5.296814e-01 Fungi PIE Guts-PP 1.7670135 7.722597e-02 3.861299e-01 Fungi PIE Nectar-PP -1.3026784 1.926846e-01 5.780537e-01 Fungi PIE Pollen-PP 0.5880966 5.564675e-01 5.564675e-01 Fungi PIE Guts-Soil -3.1288376 1.754993e-03 1.228495e-02 Fungi PIE Nectar-Soil -4.4587228 8.244946e-06 7.420452e-05 Fungi PIE Pollen-Soil -3.9763601 6.997813e-05 5.598250e-04 Fungi PIE PP-Soil -4.7555329 1.979236e-06 1.979236e-05