<<

Bumblebee diversity and floral resource preferences: Working toward the development of comprehensive ecological conservation strategies for North American bumblebees. A Thesis

Submitted to the Faculty of WORCESTER POLYTECHNIC INSTITUTE in Partial Fulfillment of the Requirement for the Degree of Master of Science in Biology and Biotechnology

February 2018

Ellen C. Pierce

Approved by:

Dr. Robert J. Gegear Dr, Elizabeth F. Ryder Dr. Marja Bakermans Assistant Professor Associate Professor Assistant Teaching Professor Dept. Biology & Biotechnology Dept. Undergraduate Studies Dept. Biology & Biotechnology Associate Director, Biology & Computational Biology

Worcester Polytechnic Institute Worcester Polytechnic Institute Worcester Polytechnic Institute

Table of Contents Table of Contents ...... i Acknowledgements ...... iii Abstract ...... iv Chapter 1: Bumblebee ecology and conservation in North America: a review...... 1 Pesticide exposure: ...... 3 Pathogens and Parasites: ...... 4 Climate change: ...... 5 loss: ...... 5 Exotic :...... 5 Conclusion/Thesis goals: ...... 6 Chapter 2: General Methods for Chapters 3‐5 ...... 7 Field sites ...... 7 Bumblebee Surveys ...... 7 2015 season ...... 8 2016 and 2017 seasons ...... 9 Flower surveys ...... 9 Chapter 3: Bumblebee decline in Massachusetts ...... 11 Methods ...... 12 Results...... 15 Discussion ...... 20 Chapter 4: Landscape‐scale analysis of resource partitioning in bumblebee pollination networks ...... 23 Introduction ...... 23 Methods ...... 24 Results...... 26 Discussion ...... 38 Chapter 5: Influence of Exotic Species in Bumblebee Community Dynamics: A Manipulative Study ...... 43 Introduction ...... 43 Methods I ...... 45 Results I ...... 46 Effect of loosestrife removal on bumblebee‐native plant interactions ...... 48

i

Methods II ...... 49 Results II ...... 50 Discussion ...... 57 Chapter 6: Conclusions ...... 60 References ...... 62 Appendix A: List and visuals of Bumblebee (Bombus) Species in Massachusetts ...... 73 Appendix B: List of bumblebee visited flower species in each site (2015‐ 2017) ...... 77 Appendix C: Screenshots of ArcGIS maps of field sites with transects drawn ...... 82 Appendix D: Shannon’s diversity (H’) for all field sites/seasons ...... 84 Appendix E: B. vagans historical data high (>1000’) vs. low elevation (<1000’) ...... 85 Appendix F: List of wildflowers visited by bumblebee species from 2015‐2017. *=indicates nectar robbing observations...... 85 Appendix G: Conservation ranking definitions for bumblebee species...... 89 Appendix H: Long vs. medium vs. short tongued bees at early, mid and late season ...... 90 Appendix I: Long vs. medium vs. short tongued workers (within tongue groups) at high and low elevation ...... 95 Appendix J: Long vs. medium vs. short tongued males (within tongue groups) at high and low elevation ...... 99

ii

Acknowledgements

I’d like to thank my advisor, Dr. Robert Gegear, for his continual encouragement, troubleshooting, and guidance throughout the entire process of my thesis. I’d also like to thank the other members of my committee, Dr. Liz Ryder and Dr. Marja Bakermans, for providing help and insight on both the field protocol and on the thesis writing. I’d like to thank the entire Biology and Biotechnology department, particularly my colleagues in the graduate program, who helped me through all the frustrating moments and found my preference for working outside interesting rather than odd. I’d like to thank my family and friends for their unwavering support and encouragement for the past two and a half years. Finally, I’d like to thank WPI and the Dean of Graduate Studies for their funding and support, without which I would have been unable to pursue my degree.

iii

Abstract Bumblebees and other wild pollinators are in a state of decline worldwide. While the cause of these declines remains unknown, several contributing factors have been proposed, including pesticide exposure, disease, climate change, and habitat loss. My thesis explores ecological processes driving the structure, dynamics, and diversity of wild bumblebee communities in Massachusetts. My ultimate goal is to identify key features of human‐introduced ecological stressors that cause some species to decline and others to thrive in the same habitat. The first research chapter examines the status of 11 bumblebee species that were historically present in Massachusetts by comparing historical and current data on the relative abundance of each species. Results showed that relative abundance has dramatically shifted for the overwhelming majority of our native bumblebee species, with some historically abundant species now rare or absent and other historically common species now more abundant and widely distributed in the state. The second research chapter explored floral resource partitioning among bumblebee species. Extensive surveys of bumblebee‐plant species interactions were conducted at several field sites in Massachusetts from June‐October. My data revealed that bumblebees vary considerably in their preference for nectar resources but have strikingly similar preferences for pollen resources. Although some of variation in nectar plant preference among bumblebee species can be attributed to differences species in proboscis length, species with similar tongue lengths also showed divergent floral preferences suggesting resource partitioning based on an as yet undetermined behavioral trait. The final research chapter focused on the potential role of exotic floral resources in the dramatic decline of many bumblebee species eastern North America over recent years. Field survey data had also shown that bumblebee species differ in their preference for exotic over native flowers. To gain insight into the potential effects of such variation in exotic flower preference on the structure and dynamics of bumblebee‐native plant pollination systems, I then conducted a landscape‐scale removal experiment involving purple loosestrife (Lythrum salicaria), a highly invasive exotic plant. My results showed that purple loosestrife ‘steals’ the pollination services of certain bumblebee species by offering a more highly preferred nectar reward. I further show that loosestrife is highly preferred by some bumblebee species but avoided by others, suggesting that loosestrife and other exotic plant species may alter the competitive dynamics of bumblebee communities. Collectively, the results of my thesis have important implications for the conservation and restoration of bumblebee habitat in North America. My work also identifies areas in Massachusetts that are of particular conservation concern.

iv

Chapter 1: Bumblebee ecology and conservation in North America: a review.

Bumblebees (Bombus spp.) are one of the most widely recognized insects in the world due to their relatively large size, ‘fuzzy’ black and yellow appearance, and distinctive buzzing sound as they visit flowers for nectar and pollen, their only food source (Goulson and Darvill 2004, Goulson et al. 2005). There are over 250 bumblebee species found in temperate regions worldwide, (Plowright and Laverty 1984, Williams and Osborne 2009), and approximately 50 species native to North America (Plowright and Laverty 1984, Laverty and Harder 2012). In Massachusetts, historical records indicate that there were 11 species present (Plath 1927) that vary in abundance and distribution in an elevation‐dependent manner (Table 1.1).

Like all social insects native to temperate regions, bumblebees have an annual cycle that begins in the spring and concludes in mid‐summer to late fall, depending on the species (Figure 1.1). It is important to note that the number of individuals at each stage is critical for maintaining healthy bumblebee populations from year to year. The cycle is initiated when mated queens emerge from their hibernation sites in the spring and search for an appropriate site to build a nest. Once the queen has found a suitable site, she lays the first brood of workers (sterile females) and cares for them by collecting nectar and pollen. Once the first brood of workers has emerged, they take over care of foraging while the queen remains in the colony for the remainder of her life. Over the rest of the spring and the summer when floral resources are abundant, the colony grows and the number of workers increase to collect as many resources as possible. From mid‐summer to fall, the queen stops producing workers and starts producing males and daughter queens. Once these have emerged as adults, they leave the colony and search for mates. Once the daughter queens are mated, they will search out a place to hibernate over the winter. The cycle begins again when the daughter queens emerge from hibernation in the spring. While the general cycle is similar for all bumblebee species, there are variations between species on when the cycle starts and finishes. For example, not all species emerge from hibernation at the same time. As seen below in Table 1.1, some species emerge early in the spring (e.g. B. bimaculatus) while others emerge late (e.g. B. pensylvanicus). Bumblebee species can also vary in other characteristics such as nesting preferences and tongue length (Table 1.1).

Figure 1.1: Visual representation of the bumblebee annual cycle. Diagram created by Rachel Blakely.

1

Table 1.1: Bumblebee species historically found in Massachusetts and their characteristics. For visuals of each bumblebee species, see Appendix A.

Species Historical flower Nesting Time of queen Tongue length preferences preferences emergence (Laverty (Laverty and (Plath 1927) (Laverty and and Harder 2012) Harder 2012) Harder 2012) B. affinis Rhododendron, Underground Early Short mountain laurel, jewelweed B. bimaculatus Bush Underground Early Medium/Long honeysuckle, red clover, cow vetch B. borealis Bush Underground Late Long honeysuckle, honey locust, cow vetch B. fervidus Red clover, cow Surface Intermediate/ Late Long vetch, toad flax B. griseocollis Milkweed, red Surface Intermediate Medium clover, cow vetch B. impatiens Goldenrod, cow Underground Intermediate Medium vetch, purple loosestrife B. pensylvanicus Bush Surface Late Long honeysuckle, red clover, cow vetch B. perplexus Raspberry, Surface Early Medium basswood, bush honeysuckle B. ternarius Rhododendron, Underground Early Short basswood, clover B. terricola Rhododendron, Underground Early Short mountain laurel, bush honeysuckle B. vagans Bush Surface/ Early Medium/Long honeysuckle, red Underground clover, cow vetch

Over the past decade, bumblebees have declined in abundance, diversity, and geographic range at an alarming rate worldwide (Cox and Elmqvist 2000). In eastern North America alone, several species have been extirpated from areas they were historically abundant, with one (B. affinis) recently listed an endangered species by the US Fish and Wildlife service (Colla and Packer 2008, Cameron et al. 2011). Intriguingly, while some bumblebee species face extinction, others are more abundant and widely distributed than historical records indicate. Bumblebee species native to the mid‐west region of the United States are also experiencing population declines and local extinctions (Grixti et al. 2009). These

2 changes have raised significant global concern due to the critical role that bumblebees play as pollinators in crop and wildflowers. In agricultural context bumblebees, and their managed cousin the honeybee, are responsible for one out of every three bites of food that we take and contribute billions of dollars per year to agro‐businesses. Bumblebees are noted for their usefulness in pollinating some native crops (like tomatoes and blueberries) that the non‐native honeybees cannot pollinate as efficiently (Javorek et al. 2002, Velthuis and Doorn 2006). Consequently, the overwhelming focus of government bodies, conservation groups, and the general public has been on maintaining adequate numbers of bumblebees and honeybees in agricultural areas.

However, bumblebees and the thousands of other species from other wild pollinator groups (solitary bees, butterflies, flies, beetles) play an equally (and some would argue more) important “keystone” role in natural ecosystems, meaning that the pollination services that they provide to the vast majority of our wildflowers is essential for producing food, nesting sites, and shelter needed to sustain populations of hundreds of other animal species. Keystone species are defined as species in a community that when lost would cause loss of many others in the community, in essence the foundation that holds up the community (Mills et al. 1993). As their loss would cause a ripple effect of loss through the community, keystone species such as pollinators are viewed as integral to the ecosystems they inhabit. Collectively, the pollinators (such as bumblebees) and the plants they visit in a community form what is known as a “pollination network”, or a web of interactions between the pollinators and the plants they pollinate. By studying a pollination network, scientists can use the entire community (rather than a few species) to examine aspects such as resiliency to extinction events and specialization (Memmott 2004, Tur et al. 2014, Tur et al. 2015). A diverse group of pollinators can benefit the reproductive success of the receiving flowers (Klein et al. 2003), and could help buffer the effects of pollinator extinction (Memmott 2004). As one of our most important native pollinators, maintaining bumblebee species diversity means healthy ecosystems and biodiversity at the regional scale. It is therefore utterly necessary to focus more attention on their plight, identify what major stressors are driving their decline, and mitigate them.

At present, the causes and ecological consequences of bumblebee decline are unknown. Below, I outline potential contributing factors, which include pesticide exposure, novel pathogens and parasites, climate change, habitat change/loss, and exotic species (Goulson 2015).

Pesticide exposure: Although wild bees are likely exposed to a wide variety of xenobiotics throughout their lifecycle in urban and agricultural areas, a new class of pesticides called neonicotinoids are thought to pose a significant threat to wild populations. Neonicotinoids are neurotoxins that selectively kill insects, leading to worldwide popularity as a pest control agent (Tomizawa and Casida 2005, Goulson 2013). Neonicotinoids are also systemic and are easily taken up through the roots and translocated to all parts of the plant (Krupke et al. 2012, Wood and Goulson 2017). However, neonicotinoids also can be transferred to areas away from the application site where they can contaminate areas with wildflowers for extended periods of time (has been shown to persist in soil for years), and thus pose a significant hazard to non‐target insect pollinator species including bumblebees. Due to honeybees’ large role in agriculture, honeybee impacts are the best studied. Neonicotinoids have a high acute toxicity on honeybees, ranging from 5‐500ng/bee LD50 (Pisa et al. 2015). While there is not as much information or studies focusing on neonicotinoid effects on bumblebees, it has been established that the pesticide group has a high toxicity on bumblebees, and may have an even worse effect when combined with other pesticide or herbicide residues like propiconazole and permethrin (Sanchez‐Bayo and Goka 2014, Riaño Jiménez and Cure 2016). It has also been found that bumblebees are exposed to neonicotinoids in the

3 pollen, both in food crops and in nearby wildflowers (Botías et al. 2015, David et al. 2016). Studies vary on how much exposure bees receive from both sources, so there is no clear rate of exposure for either honeybees or bumblebees. In addition, impacts and toxicity in wild bumblebees in the field and long term population studies have not fully been accomplished, and are needed to gain a clearer picture on how badly neonicotinoids are negatively impacting bumblebee species (Pisa et al. 2015).

Even if the neonicotinoids do not kill bumblebees outright, there is increasing evidence that “sublethal” doses of pesticide can impair bees, thus indirectly causing increased mortality. Behavioral tests of bumblebee workers exposed to sublethal doses of various neonicotinoids showed altered foraging preferences compared to control groups, negative impacts on foraging, and also led to fewer reproductives produced by the colony (Mommaerts et al. 2009, Stanley and Raine 2016, Arce et al. 2017). Feltham et al.’s (2014) experiments with imidacloprid at “field realistic” sublethal doses showed a significant decrease in workers bringing back pollen, which is consistent with the negative impacts on foraging, and a potential cause for the lower numbers of reproductives in colonies exposed to neonicotinoids (Feltham et al. 2014). Sublethal chronic doses (1‐10ppb) of clothianidin caused 50% mortality in B. impatiens workers within a week and B. impatiens males within a few days. This led to a claim that males were more strongly affected than workers, which would then have implications spanning generations (Mobley 2016). Because bumblebees have an annual cycle, reduced males and daughter queens due to pesticides leads to a cascading effect of fewer colonies and even fewer reproductives in future years. However, there is not yet a consensus over the impacts of sublethal exposure to neonicotinoids. Several studies have claimed instead that there was no significant impact of sublethal doses on bumblebees, but only on honeybees (Piiroinen et al. 2016, Piiroinen and Goulson 2016).

Pathogens and Parasites: Bumblebees are host to a wide variety of naturally occurring pathogenic organisms, including bacteria, fungi, wasps, nematodes, and viruses (Gomez‐Moracho et al. 2017). Due to globalization, bumblebees are also host to several new diseases from honeybees. For example, the Varroa mite, a honeybee brood parasite (Daszak et al. 2000) was transferred to wild bumblebees in North America. While the bumblebees aren’t strongly affected by the mite itself (Goulson and Hughes 2015), they are negatively impacted by diseases for which the Varroa mite is a vector (e.g. deformed wing virus) (Rosenkranz et al. 2010, Nazzi et al. 2012, Goulson and Hughes 2015, Gomez‐Moracho et al. 2017). Another example of cross‐transmission from honeybees is the fungus Nosema ceranae, previously classified as a protozoan. This followed the same path as the Varroa mite, transferring from Asian honeybees to Western honeybees, and then from Western honeybees to native bumblebee species (Graystock et al. 2013).

One of the most prevalent diseases for bumblebees appears to be fungal pathogens in the genus Nosema, such as N. ceranae and N. bombi. N. bombi has been seen in multiple declining species in North America (Cameron et al. 2016). Studies have found that while there did not appear to be evidence for N. bombi being introduced from Europe, the strains seen in historical specimens correspond to those seen in commercially raised bumblebee outbreaks (known as spillover or horizontal transfer) (Daszak et al. 2000, Goulson and Hughes 2015, Cameron et al. 2016, Gomez‐Moracho et al. 2017). This spillover from commercial colonies is also the case with the bumblebee parasite Crithidia bombi (Meeus et al. 2011, Schmid‐Hempel et al. 2014). N. bombi and other diseases can cause the bumblebees to have impaired foraging abilities, leading to decreased fitness and decreased reproductive output (Gomez‐Moracho et al. 2017). Decreased reproductive output from pathogens or parasites, like that of pesticides, negatively

4 affects multiple generations of bumblebees given their annual colony cycle. Due to the amount of evidence showing both presence of spillover and negative impacts of the spillover on the wild bumblebee populations, a solid argument can be made that disease is a factor in global bumblebee decline.

Climate change: In a 2015 report, researchers found that compared to bumblebees’ historical ranges, the southern boundary was moving north, but not expanding in return into farther northern regions (Kerr et al. 2015). In other words, their current ranges are shrinking, rather than shifting northward. This phenomenon was consistent between Europe and the Americas (Martins et al. 2015). The Kerr et al. study, along with another, also claimed that the bumblebees are not just moving northward from the southern end of their ranges, they are also generally moving to higher elevations (Ploquin et al. 2013, Kerr et al. 2015). If a species has a narrow climactic range, then that would put them at a greater risk of decline than one that is capable of a greater temperature or elevation range (Williams et al. 2009). Extreme weather events such as droughts can negatively impact the flower species, and in turn negatively impact the bumblebees that visit them. A recent study in 2016 found evidence that in addition to drought shrinking floral display, it also altered the flower volatiles by altering their relative abundance (e.g. doubling α‐pinene relative abundance in flowers impacted by drought), which then impacted the rate of pollinator visitation (Burkle and Runyon 2016). This loss of range and elevation changes can then tie in with habitat loss, and the accompanying ill effects (Elias et al. 2017, Papanikolaou et al. 2017).

Habitat loss: Agricultural intensification has transformed natural habitat of bumblebees, grasslands, into farmland. In Britain, it is estimated that 97% of grasslands have been lost, which sharply limits the suitable habitat for bumblebees to thrive in (Goulson 2015). In North America, Grixti et al. (2009) found that the time period where bumblebee species richness declined coincided with large scale agricultural intensification. In addition, a monoculture of crop leads to a “glut” of resources during only a short portion of the season, with low available resources during the rest of the season (Dramstad and Fry 1995, Carvell et al. 2007, Goulson and Nicholls 2016). Increased urbanization is also contributing to the loss of bumblebee habitat (Osborne et al. 2008).

In the U.K. and other countries in Europe, there have been several studies showing that longer tongued bumblebees are more likely to be in decline than their shorter tongued counterparts. The argument is that this is due to longer tongued bees having a more specialized diet, less flexibility to adapt to cultivated plants, and therefore more sensitivity to habitat change/loss (Goulson and Darvill 2004, Goulson et al. 2005, Williams 2005, Goulson et al. 2008b, Miller‐Struttmann et al. 2015). Diet breadth in pollen as well as nectar could also create an impact. Kämper et al. found after an examination of B. terrestris pollen collection in different landscapes that the species preferred pollen from woody plants (in particular maple trees) regardless of the landscape (Kämper et al. 2016). Given the importance of pollen for colony growth and reproductive, loss of those preferred pollen plants could create a definite negative impact on bumblebee species, especially taking into account their annual life cycle.

Exotic species: Although critical for the pollination of crop plants in agricultural areas, non‐native honeybee and bumblebee species have the potential to negatively impact native bumblebees and other pollinator groups through competitive interactions. While there is no direct evidence tying the presence of non‐native pollinators to a decline in native pollinator abundance or diversity (Goulson 2003, Forup and Memmott 2005, Stout and Morales 2009), there has been some indirect evidence in recent years to

5 suggest a potential impact. Thomson in 2004 introduced honeybee and B. occidentalis (a native bumblebee) colonies to field sites, and examined worker travel and reproductive success in B. occidentalis colonies that were either near or far away from the honeybee hives (Thomson 2004). The study showed that bumblebee reproductive success declined the closer the colonies were to the honeybee hives, and workers from colonies near the honeybee hives showed fewer colony returns with pollen, the primary nutrition for larval growth and development (Thomson 2004). In a later publication, Thomson showed an inverse relationship between honeybee and native bumblebee abundance in California. Due to drought, there were also fewer floral resources, which correlated with lower bumblebee abundances (Thomson 2016). In 2009, a UK study showed that bumblebee workers in survey areas where honeybees were present were significantly smaller than those in areas without honeybees, implying increased competition for resources (Goulson and Sparrow 2009).

Other species of bumblebees can even become invasive when introduced, and hurt the native populations. In Chile, introduced populations of European bumblebees B. ruderatus and especially B. terrestris have spread rapidly over the country, and appear to have displaced the native populations of B. dahlbomii (Schmid‐Hempel et al. 2014). Conclusion/Thesis goals: In conclusion, bumblebee species are declining at an alarming rate in some areas of North America for unknown reasons, posing a significant threat to ecosystem health and the biodiversity that it supports. Steps need to be taken to preserve the populations of stable bumblebee species, and help bolster declining species; however, we presently lack sufficient ecological data on bumblebees at the species level to do so. My thesis work aims to fill this knowledge gap by 1) determining the status of bumblebee species in Massachusetts (my study location); 2) identifying floral resources preferences of different bumblebee species and advance our understanding of the mechanisms underlying them; 3) testing the effects of exotic floral resources on the structure and dynamics of bumblebee pollination networks.

6

Chapter 2: General Methods for Chapters 3‐5 In order to achieve the goals stated in the last paragraph of Chapter 1, I conducted intensive field surveys of bumblebee‐flower interactions at several locations in Massachusetts. The data collected from these field surveys forms the Results section of each of my research chapters. While the data analysis is different for each chapter, the Methods sections have a considerable amount of overlap. To minimize redundancy, I describe the general field survey method here and place any additional methodological detail to the Methods section of each chapter. Field sites A total of 4 field sites in central and northwestern Massachusetts were surveyed over the course of 2015‐2017 (Figure 2.1). Breakneck Hill Conservation Land in Southborough (46 acres, elevation 93m) and Wachusett Meadow Wildlife Sanctuary in Princeton (16.7 acres, elevation 358m) were surveyed in 2016 and 2017, while Crowningshield Conservation Area in Heath (8.9 acres, elevation 512m) and Bullitt Reservation in Ashfield (8.7 acres, elevation 379m) were surveyed in 2017. All sites had formerly been farms, but had since been converted to conservation land. The field sites were defined as “high” elevation (elevation over 305m [1,000ft]) and “low” elevation (elevation below 305m). Wachusett, Heath, and Ashfield were high elevation, while Breakneck was low elevation. They were separated as such due to differential bumblebee species composition. Bombus ternarius, B. borealis, and B. terricola were only found at sites classified as high elevation.

Figure 2.1: Location of field sites in Massachusetts. Green=Worcester, location of lab. Yellow= 2015‐ 2017 field site. Blue=2016 and 2017 field site. Orange=2017 field site. Bumblebee Surveys The surveying season started when both bumblebees and flowers were present at the field sites, around the beginning of June, and ended at either first frost or when the landowners mowed the fields (Table 2.1). The field seasons were separated into three sections: early season (beginning of June to mid‐July), mid‐season (mid‐July to end of August), and late season (beginning of September to end of field season). The field sites were surveyed on non‐rainy days with dry ground conditions, and a minimum starting temperature of 60°F. Surveys typically started between 9:30 and 10:30am, and ran until the entire site was surveyed. If the peak temperature was going to be excessively hot (>90°F) during the day, the field survey would be started earlier than 9:30 (8:30‐9am), as long as the starting temperature was above the

7

60°F threshold. The surveys at each field site were run on as close to a weekly basis as possible, depending on the weather.

Table 2.1: Start and end dates for each field season at each site. The reason for ending the survey is also listed as it changed depending on the year and field site.

Field site (year) Start date (first survey) End date (last survey) Reason for ending Breakneck (2015) 6/7/15 10/14/15 First frost Breakneck (2016) 6/1/16 10/7/16 First frost Breakneck (2017) 6/9/17 10/4/17 Mowing Wachusett (2016) 6/2/16 9/16/16 Mowing Wachusett (2017) 6/8/17 9/4/17 Mowing Heath (2017) 6/12/17 9/11/17 Construction/Mowing Ashfield (2017) 7/5/17 10/5/17 First frost

When a bumblebee was observed, the species, caste, the flower it was feeding from (if applicable) and whether the bee was collecting nectar or pollen were recorded using a handheld recorder. Bumblebees for the most part were non‐invasively identified “on the wing” throughout surveying, rather than capturing or terminating the bee. Previous training was required in species identification to ensure accuracy. In rare cases when one was not easily identifiable, the bee was either videoed or captured through sweep netting, brought back to the lab for a second opinion and released back at the field site. 2015 season The first field season in 2015 was run at Breakneck Hill Conservation Land, where a preset path was followed around the property (Figure 2.2). The survey route started at the parking lot and the trail was followed counterclockwise while observations were recorded.

Figure 2.2: Map of Breakneck Hill Conservation Land. Diagram created by Robert Gegear. Dotted red line shows survey path used in 2015. Yellow circles indicate areas of interest to the creator (e.g. sole location of a flower species on the land).

8

2016 and 2017 seasons During the 2016 and 2017 field seasons, all sites were surveyed in transects 20m apart, with searching for and recording observations of bumblebees 2m on either side of the transect (4m wide belt transects) (Carvell et al. 2007, Colla and Packer 2008). Breakneck 2017 transect distance was changed to 40m apart to enable the entire site to be covered in one day. ArcGIS was used to draw transects and determine the appropriate number (Figure 2.3, see Appendix C for other sites’ transect lines). Using a random number generator, the survey started on a random transect every survey to account for any potential activity differences due to time of day.

Figure 2.3: Screenshot of Wachusett Meadow Wildlife Sanctuary transects drawn in ArcGIS. Transect lines were drawn 20m apart. See Appendix C for other field site transect maps. Flower surveys In addition to the bumblebee surveys, flower surveys focusing on wildflower diversity and abundance were also carried out in 2017 for the same period of time as the bumblebee surveys. Within a 2m width x 5m length rectangle along the transect line (10m2 total area), all plants, flower clusters if applicable on each plant and number of flowers on each plant and flower cluster were counted and noted on a field sheet. In order to conduct surveys in different places on the transects, the flower survey was either done at the beginning, middle or end of the transect, with a different location for the next surveyed transect. For example, after doing a survey at the beginning of transect #2, the next flower survey was carried out in the middle of transect #4. The entire transect was not surveyed in order to not detract from the bumblebee survey. Every other transect was surveyed in order to gain a clear picture of the survey area while continuing to not take away from the bumblebee survey. If a floral species occurred in only one location (a patch), the number of flowers in the patch were estimated by counting the flowers in a fraction of the patch and then extrapolating to estimate total flower numbers in the patch. Any patches too big to estimate in a timely manner were photographed, and estimates were completed

9 using the pictures. The data were summarized for each week of surveying and overall season in the total number of flowers for each species divided by the number of transects surveyed, giving an average number of flowers per transect sample (10m2).

Even numbered transects were surveyed on even numbered survey weeks and vice versa, with patches being surveyed every week. The exception to this was Breakneck Hill Conservation Land in 2017; when the transects for Breakneck were halved for the 2017 season to the entire site to be covered in one day, all transects were surveyed on each survey date that season to continue to cover half the survey area.

10

Chapter 3: Bumblebee decline in Massachusetts Bumblebees are one of the most important pollinators of native species in temperate regions around the world. As keystone species, bumblebees also maintain biodiversity as the pollination services that they provide to wild plants has cascading positive effects on the survival of wildlife at other trophic levels (Mills et al. 1993). In an agricultural context, bumblebees are essential for the pollination of many crop plants (Javorek et al. 2002, Velthuis and Doorn 2006), although their contribution as crop pollinators is relative small compared to their cousin the honeybees. Determining the status of bumblebee species throughout their native range therefore has important socio‐economic and ecological implications.

Recent studies have shown that bumblebees are in a state of unprecedented decline in several European countries (Rasmont and Mersch 1988, Sárospataki et al. 2005, Biesmeijer et al. 2006, Kosior et al. 2007) and parts of North America (Colla and Packer 2008, Grixti et al. 2009, Cameron et al. 2011). Although some of the North American studies have assessed population status based on small scale observations over a few years (Kearns et al. (2017), most have been conducted on a large scale (state to country level) and spanned decades (Colla and Packer 2008, Grixti et al. 2009, Cameron et al. 2011). In southern Ontario, for instance, Colla and Packer demonstrated changes in the status of multiple bumblebee species by comparing current species abundances in an area with those obtained through museum specimens collected in the same area two decades earlier (Colla and Packer 2008). Using a similar technique, Cameron et al. (2011) demonstrated status changes in the same bumblebee species throughout the United States. Grixti et al. (2009) found similar changes in bumblebee species and abundance in Illinois, discovering that two species, B. borealis and B. terricola, had been extirpated at the state level. Collectively, these findings facilitated the listing of B. affinis as the first and only endangered bumblebee species in the U.S. (Szymanski et al. 2016, USFWS 2017). However, more information on the abundance and diversity of bumblebee species in different areas of their native range, particularly at the state level, is desperately needed to determine whether other species should also be listed as endangered. Such comparisons with historical data also yield novel insights into factors that may be contributing to bumblebee decline, such as habitat loss/modification, pesticide use, and exotic species introductions (Thomson 2004, Williams et al. 2009, Ploquin et al. 2013, Kerr et al. 2015, Goulson and Nicholls 2016, Gomez‐Moracho et al. 2017).

In addition to the general lack of data on the status of bumblebee species in North America, we also have a gap in our knowledge of how human‐induced stressors have affected population dynamics or ‘phenology’ of different species over the past couple of decades. The bumblebee population cycle resembles a classic Gaussian distribution curve, beginning with a small number of mated queens emerging from hibernation in the spring followed large number of workers foraging for floral resources mid‐cycle, and then a small number of males and queens mating at the end of the cycle (see Figure 3.1). If species are less able to ‘shift’ their phenology in response to anthropogenic stressors, then they may be more susceptible to decline. For example, global warming may cause queens to emerge from hibernation much earlier than have historically resulting in corresponding shift of peak abundance and male/queen production to earlier in the season. While there is some evidence that climate change has shifted flower phenology and consequently bee abundance (Ranta et al. 1981, Pyke et al. 2016, Ogilvie et al. 2017), few studies have directly tested for shifts in bee phenology. Bartomeus et al. (2011) found that phenological advances in queen emergence for two bumblebee species over the past several decades, but did not examine the rest of the phenological cycle and focused only on stable bumblebee

11 species (B. impatiens and B. bimaculatus) (Bartomeus et al. 2011). Clearly, more direct tests for changes in phenology among bumblebee species would yield greater insight into potential factors driving differential declines in wild populations.

Figure 3.1: Visual representation of the bumblebee annual cycle. Diagram created by Rachel Blakely.

In this study, we first determine the status of bumblebee species in Massachusetts, using comparable historical data (Yale 2017) to compare with species relative abundances from current field surveys. In order to make the most detailed analysis possible, the data were also separated by high (>305m) and low (<305m) elevation. We then compare the phenology curves of the historical data of common bumblebee species in Massachusetts to those of the current field sites to determine if there has been a significant ‘phenological shift’ over time. With currently little data on changes in bumblebee phenology (Bartomeus et al. 2011), any information gained from this study would be useful in determining if population declines are related to major changes in phenology. Methods Historical data Historical data for Massachusetts counties was taken from the collections of the Yale Peabody Museum of Natural History in New Haven, CT. The records of collections were publicly available online (http://collections.peabody.yale.edu/search/Search/Advanced?collection=Entomology), and included information such as collection date, state, county, species, and recorder of the specimen. These historical data were collected by Russell B. Miller, Sophie Coe, and Michael Coe from all over the state, between 1969‐1986 (Yale 2017). They collectively surveyed mainly central and eastern Massachusetts counties, ranging from Franklin County to the north west to Barnstable County on Cape Cod (Figure 3.2). They also surveyed along both low and high elevations (11‐518m).

12

Figure 3.2: Counties surveyed by Russell B. Miller, Sophie Coe, and Michael Coe for the historical data between 1969 and 1986. Red dots signify surveyed counties at low elevation (<305m), blue dot signifies surveyed county at high elevation (>305m).

Data analysis The historical data did not differentiate between the different bumblebee castes (worker, male, queen) so the current field survey data were pooled together in the same manner for more accurate comparison.

For relative abundance comparisons between historical and current data, a Z test for equal proportions was used (Equation 3.1) (Colla and Packer 2008, Cameron et al. 2011). A Chi‐square test was not used due to a large number of expected values lower than 5, which would make the test invalid.

Equation 3.1: Z test for equal proportions formula 1 2 1 1 1 1 2 In Equation 3.1, p1 stands for proportion of historical abundance, p2 for proportion of current abundance, n1=total number of historical observations, n2= total number of current observations, and p stands for the sum of historical and current observations of a species divided by the sum of total historical and current observations.

There was a lack of historical data for Worcester County, the location of both the Wachusett and Breakneck survey sites. Therefore, the historical relative abundance for Franklin County was used to compare to the current field data because of its similar “high” elevation (elevation over 305m [1,000ft]).

13

The same was done with the current low elevation data, with the Barnstable, Hampshire, Hampden, Suffolk, Norfolk, and Middlesex Counties pooled to provide a large enough historical sample size for “low” elevation (elevation below 305m). The current and historical data were separated by low and high elevation due to different bumblebee species composition. B. ternarius and B. borealis were only found at sites or counties that would be considered “high” elevation. In addition, B. pensylvanicus was only found at sites or counties that would be considered “low” elevation. To have current data in another site in Massachusetts to compare to the Breakneck data at low elevation, 2006 survey data from Gillespie (2010) and survey data from Lerman and Milam (2016) were used.

Analysis of bumblebee species phenology curves comparing historical and current data was performed through non‐linear regression (curve fitting analysis), with Gaussian distribution as the curve while amplitude, mean, and SD were used as comparative parameters to determine whether to reject or not reject the null hypothesis. The null hypothesis used for all groups of phenology data was that one curve would fit all data sets within the group.

To ensure that variation in elevation would not impact historical data in the phenology curves, the data from “high” (over 305m) and “low” (below 305m) elevation were compared to the total pooled data for an example bumblebee species (B. vagans) using non‐linear regression (curve fitting analysis) with a Gaussian bell curve as the base curve, and found that the null hypothesis was rejected, meaning the “high” and “low” data could not be pooled and had to be shown separately (Appendix E). The null hypothesis for this, like all subsequent non‐linear regression curves, was that one curve could reasonably fit all data sets. A minimum of n>10 sample size was instituted for the curves in order to show only the phenologies that had a coherent curve. Because of this, only 5 species were analyzed as they had large enough sample sizes to compare data at both high and low elevation. Each species’ phenology curve was analyzed for differences between current field sites and historical data.

Shannon’s diversity index (H’) was used to measure bumblebee species diversity for comparison between historical and current data at both high and low elevation, due to it being a standard biodiversity test in literature (Morris et al. 2014) (Equation 3.2). The higher the Shannon’s diversity index, the more diverse the surveyed bumblebee community. Shannon’s diversity was chosen over Simpson’s due to all species examined having equal importance (as opposed to dominant species being more important in Simpson’s diversity) (Morris et al. 2014).

Equation 3.2: Shannon’s diversity index formula

H=Shannon’s diversity index

S=number of species pi=proportion of total observations that are species i

14

Results Several bumblebee species in decline in Massachusetts After comparing the proportion of observations from historical data and current field data, it was evident that there are several bumblebee species that are in decline in Massachusetts. At all current field sites, B. vagans has significantly declined in relative abundance compared to the historical records, especially at high elevation (0.40 of observations to 0.09) (Figure 3.3). While the low elevation decline for B. vagans was less severe (0.23 of observations to 0.004), it was still significant (Figure 3.4). B. fervidus also declined at both elevations, going from 0.03 of observations to 0.0002 at low elevation and 0.02 of observations to 0.002 at high elevation (Figures 3.3 and 3.4). B. terricola has declined significantly in high elevation sites from 0.36 of observations to 0.03, and from 0.08 to none seen at low elevation (Figures 3.3 and 3.4). B. affinis and B. pensylvanicus were not seen at all in the recent field surveys, though they were present in the historical data (Figures 3.3 and 3.4). At low elevation, B. bimaculatus (0.21 of observations to 0.02) and B. perplexus (0.05 of observations to 0.006) relative abundances were significantly lower than the historical data (Figure 3.4).

* * * *

*

* * * * *

Figure 3.3: Bar graph of relative abundance of species for high elevation. Current “high” elevation (n=3196) and historical “high elevation” (n=988). Current field sites had all data pooled for an overall proportion. * indicates significant difference at P ≤ 0.05.

15

*

* *

*

* * * * *

Figure 3.4: Bar graph of relative abundance of species for low elevation. Top: current “low” elevation (n=10319), Gillespie 2010 (n=271), Lerman and Milam 2016 (n=234), and historical “low elevation” (n=100) for B. impatiens, Bottom: current “low” elevation (n=782), Gillespie 2010 (n=60), Lerman and

16

Milam 2016 (n=26), and historical “low elevation” (n=310) for other Bombus species. Current field sites had all data pooled for an overall proportion. * indicates significant difference at P ≤ 0.05.

Bumblebee species diversity also decreased when comparing historical to current data at low elevation. The overall Shannon’s diversity for historical data at low elevation was 1.88, while the current Shannon’s diversity was 0.33 (Table 3.1). Conversely, the bumblebee species diversity increased at high elevation, when comparing current and historical data, from 1.37 to 1.57 (Table 3.1).

Table 3.1: Shannon’s diversity of historical vs. current data at high and low elevation

Overall Shannon’s diversity Historical (high) 1.37 Current (high) 1.57 Historical (low) 1.88 Current (low) 0.33

Some bumblebee species have increased relative abundance and range in Massachusetts There were some bumblebee species that have significantly increased in abundance in current times compared to the historical surveys. The clearest example of this is B. impatiens which has jumped from a small portion of the bumblebees seen (0.01 of observations at high elevation, 0.26 at low) to being one of the most common species seen (0.35 of observations at high elevation, 0.93 at low) (Figures 3.3 and 3.4). In high elevation sites, B. griseocollis and B. bimaculatus have jumped from negligible abundance (0.002 of observations and 0 respectively) to larger abundances seen (0.35 of observations and 0.05 respectively) (Figure 3.3). In addition, B. borealis was observed in current field surveys, when the historical data had indicated that the species was not present in the state of Massachusetts.

Some of these bumblebee species that have increased in relative abundance showed signs of range expansion. The clearest example of this is B. borealis. B. borealis was only seen in the state of Massachusetts two times in the entirety of the 1969‐1986 historical recordings (Yale 2017), but was seen over 50 times in the current field data at high elevation, for a relative abundance of 0.02 (Figure 3.3). Less drastic examples are B. griseocollis and B. bimaculatus expanding from low elevation sites to high elevation sites. Both species were very uncommon at high elevations historically (0.002 and 0 respectively), but were seen in significantly higher numbers currently (0.35 and 0.05 respectively).

Phenologies of several bumblebee species have altered compared to historical data When historical data was compared against current data for phenology curves, it demonstrated that for many species, observations of bumblebees were historically seen at least one month after the last observation in the current field studies. Examples of this are B. griseocollis at current low elevation, B. vagans at current high elevation, and B. perplexus for both current data sets. In contrast, the B. griseocollis data for current low elevation showed extremely similar curves for both historical and current field surveys (Figure 3.5).

17

ns

ns

18

**** P<0.0001

**** P<0.0001

19

ns

Figure 3.5: Curves of phenologies for common species in current field surveys (low and high elevation) compared to historical species phenologies (low and high elevation) in Massachusetts. X‐axis substitutes numbers for corresponding months (January=1, February=2, etc.). Solid lines indicate the phenology curves and dashed lines indicate the best‐fit Gaussian curves. Only samples with n>10 with a clear curve were included for clarity. ns = not significant. Discussion The overall conclusion of the study is that all bumblebee species previously seen in the state have significantly changed in relative abundance in general compared to the historical data. In addition, all seen species either increased or decreased significantly when focusing on just high or low elevation, with the exception of B. griseocollis at low elevation (Table 3.2). Some, like B. fervidus, have significantly declined at both high and low elevation, while some species like B. impatiens have significantly increased (Table 3.2). Some species seen only at high elevation like B. terricola and B. ternarius declined, while B. borealis increased. In addition, B. affinis and B. pensylvanicus were not seen at all in current field sites, while they were observed historically. Because of this, it is possible that these two species may be extirpated from the state (Table 3.2). These results are consistent with other North American decline studies (Colla and Packer 2008, Cameron et al. 2011).

Table 3.2: Summary of bumblebee species found in Massachusetts and their status present day when comparing current to historical relative abundances (Figures 2 and 3). + = increasing, ‐ = decreasing, 0 = stable or no change. Gillespie 2010 and Lerman and Milam 2016 data not included in this table. For definitions of conservation ranking given by Massachusetts Division of Fisheries and Wildlife, see Appendix G.

High elevation (>305m) Low elevation (<305m) Bumblebee Historical Current Historical Current Conserva‐ species relative relative Status relative relative Status tion abundance abundance abundance abundance Ranking B. impatiens 0.01 0.35 + 0.26 0.93 + S5

20

B. griseocollis 0.002 0.34 + 0.04 0.04 0 S5 B. 0 + 0.21 0.02 ‐ S5 bimaculatus 0.05 B. vagans 0.40 0.09 ‐ 0.23 0.004 ‐ S3 B. perplexus 0.03 0.06 + 0.05 0.01 ‐ S5 B. fervidus 0.02 0.002 ‐ 0.03 0.0002 ‐ S2 B. ternarius 0.15 0.08 ‐ 0 0 N/A S4 B. borealis 0.002 0.02 + 0 0 N/A S3 B. terricola 0.36 0.005 ‐ 0 0 N/A S1 B. affinis 0.03 0 ‐ 0.16 0 ‐ SH B. 0 N/A 0.01 0 ‐ SH pensylvanicus 0

Some species (such as B. borealis) appear to be expanding their range either into or throughout Massachusetts. Several possible theories could explain the difference that could be seen compared to historical data. Firstly, some bumblebee species (e.g. B. vagans, B. fervidus, and B. terricola) have declined significantly at high elevation. It is possible that because of their lower abundance, floral resources became more available for species that were not declining which then expanded into the area. Land use change could also be a factor. As noted in Chapter 1, the different bumblebee species in Massachusetts show variation in several aspects of their lives, including nesting preferences. It is possible that a change from farmland to forest (or vice versa) could cause a change in available nesting habitat and create more suitable habitat for queens of a particular species to raise colonies in. While a couple studies in the U.K. have looking at nesting habitat preferences, there have been no equivalents examining North American species (R Kells and Goulson 2003, Lye et al. 2009).

Interestingly, there were some differences in relative abundance between the current “low” elevation sites (Breakneck and data from two Massachusetts studies). While B. griseocollis remained the same across field sites and historically, species such as B. vagans and B. perplexus showed higher proportions in Gillespie 2010 data than both Breakneck and Lerman and Milam 2016 (which were comparable). One possibility for this is the type of field site surveyed. Gillespie 2010 surveyed “old‐field meadows”, Breakneck was a fairly disturbed site having been sprayed with herbicides a few years before surveying, and Lerman and Milam 2016 surveyed suburban lawns (Gillespie 2010, Lerman and Milam 2016). The amount of human disturbance in the area (lower=Gillespie, higher=Breakneck/Lerman and Milam) could have had an impact in the bumblebee species diversity and relative abundances at the sites (Ahrné et al. 2009, Grixti et al. 2009).

Some species phenology curves, when compared to the historical data, are significantly different. For example, B. vagans (a bumblebee species shown to be in decline currently) shows a peak abundance in August/September according to the historical data, but the current high and low elevation field sites show a current peak abundance around July. In contrast, the historical and current phenology data for B. griseocollis and B. perplexus (both shown to not have significant declines currently) show that no significant change in phenology has happened. It is possible that in addition to declining overall, some bumblebee species such as B. vagans are also having shortened phenologies, meaning that they are not active for as long as they used to be. Interestingly, the current high elevation phenology curves showed a significant difference when compared to current low elevation phenology curves for B. bimaculatus

21 and B. vagans. For both species, the phenology curve was slightly later at high elevation than at low elevation. However, more data would be needed to confirm this and make the study more robust.

There were some limitations to this study. First, the historical data, while useful, was also limited. It would have been preferable to have larger sample sizes for all counties other than Franklin County. This would have negated the need to combine multiple counties for the Breakneck field site, some of which may have had different bee diversity due to geographical differences between western and eastern Massachusetts despite their elevational similarity. Another study limitation is that the current field surveys did not cover all regions of Massachusetts. The west, south west, east, north east, and south east parts of Massachusetts did not have any sites, leaving only the central/south and north/north east regions of Massachusetts covered. As such, the entire state was not surveyed and examined for evidence of species declines. This would have helped create a more complete picture of current bumblebee species abundance across the state, and serve as “baseline” data for future studies.

With the information from this study, recommendations can be made as to which bumblebee species are in need of conservation assistance in the state. The highest priority species, should they be seen, are B. affinis and B. pensylvanicus. Their conservation ranking is SH (possibly extirpated) due to no records seen currently when they were seen historically, so if any individuals are seen anywhere in the state immediate action would need to be taken to keep them from becoming extirpated (Table 3.2). B. terricola is considered critically imperiled (S1), due to its severe decline from historical data, and its very few (<20 total) current observations limited to a single high elevation site (Table 3.2). Of the bumblebee species that have been confirmed to still be present in the state, B. terricola should have the strongest conservation action taken. B. fervidus with a ranking of S2 should also have strong conservation action, due to its severe decline at both high and low elevation and very few individuals seen (<20 total). Finally, B. vagans and B. borealis should have some conservation action with a ranking of S3, but they are not in as much peril as the previously mentioned species. B. borealis has expanded its range into the state (rather than declining), but it appears to be restricted to a limited area (a single high elevation site), and is not observed to be an extremely abundant species (<100 total). B. vagans has declined at both high and low elevations, but there are still many individuals seen (>100 total) so it appears to be in a better position than B. terricola and B. fervidus but should still be watched in the event that their decline becomes worse.

For future studies, more field studies should be conducted, in all Massachusetts counties if possible. In order to get the most accurate data, the field surveys should be completed on a regular basis (every other year, for example). These could be used to compare with both historical data and the current field surveys completed for this study. Through this study, knowledge has been gained as to which bumblebee species have declined compared to historical data, but nothing is known on whether these species are still continuing to decline. Thus, the goal of these future surveys would be to gain a more complete picture not only of which species are declining across the state, but also the rate at which they are declining, to help with conservation efforts. In the same vein, these surveys could be used by conservation authorities to determine if any (currently declining) bumblebee species (like B. fervidus) have started to bounce back and no longer need special protection.

22

Chapter 4: Landscape‐scale analysis of resource partitioning in bumblebee pollination networks Introduction Exploitive resource competition has been a significant driving force in the immense diversification of flowering plants and their animal pollinators (Fründ et al. 2010, Rodríguez‐Gironés and Santamaría 2010, Temeles et al. 2016). As flowering plants cannot search for an appropriate mate due to their immobility, they require an appropriate abiotic (wind, water) or biotic (animal) pollen vector in order to reproduce. In order to entice animals to do so, plants provide animals with floral rewards in the form of nectar (carbohydrate source) and pollen (protein source). Plant reproductive success is therefore dictated by the foraging decisions of pollinators. However, the reproductive and foraging interests of the two groups diverge, creating a source of conflict. From the plant perspective, the ideal pollinator is one that specializes on flowers of the same species, but pollinators are focused on obtaining the most high‐ quality resources as possible regardless of what is in the best interest of the plant. Since multiple pollinator species are competing for the same array of floral resources (Zimmerman and Pleasants 1982, Rodríguez‐Gironés and Llandres 2008), there is strong selection on pollinators to evolve traits that enable them to more effectively exploit a subset of resources thereby providing them with a competitive advantage. For example, many bumblebee species have evolved longer tongues which give them a competitive advantage over shorter‐tongued species on long, tubular flowers. In addition to morphological traits, it is also known that some bumblebee species have behavioral adaptations such as “nectar robbing”, or biting a hole at the base of a long tubular flower to steal the nectar, in order to compete effectively against the long tongued bees. This niche partitioning by the different pollinators leads to a responding change in the plants, as the pollinators apply selective pressure onto their chosen flowers for characteristics such as larger corollas thereby benefiting both plant and pollinator species. These relatively specialized plant‐pollinator species interactions within the community form what is known as a “pollination network”. Investigating the degree of specialization in pollination networks can therefore novel insight into ecological factors influencing the abundance of different species, such as competition and resource availability (Memmott 2004, Tur et al. 2014, Tur et al. 2015).

In this study, we examine mechanisms of floral resource partitioning in bumblebee communities. Although long considered floral generalists (Laverty and Plowright 1988, Garbuzov and Ratnieks 2014), there is some evidence that bumblebees vary in their floral resource preferences and thus exhibit resource partitioning at the species level. For example, almost a century ago Otto Plath noted in his PhD dissertation on the ecology of bumblebees in New England that Bombus griseocollis had a ‘particular fondness for the flowers of milkweed’ (Plath 1927). Since that time, other studies have found that some long tongued bumblebee species fit into “compartments” with butterflies in community food webs, in terms of their preference for flowers with long corollas (Dicks et al. 2002). A few groups have focused on the pollen reward of the flower rather than the more commonly studied nectar reward, and found how the pollen content from flowers can influence bumblebee species visitation (Russell et al. 2016, Kriesell et al. 2017, Nicholls and Hempel de Ibarra 2017). However, the most direct evidence of species‐level resource partitioning in bumblebees by far has been gained from preference studies on morphological traits such as tongue length and body size. For example, studies including Heinrich and Inouye in the 1970’s have found that longer tongued bees prefer flowers with longer corollas, while shorter tongued bees prefer shorter, less tubular flowers (Heinrich 1976, Inouye 1978, Pyke 1982, Harder 1983, Harder 1985). Barrow in 1984 also found a positive correlation between tongue length and

23 corolla length within individuals of the same species (Barrow 1984, Peat et al. 2005). This resource partitioning based on morphological differences viewpoint has also been supported by more recent studies (Johnson 1986, Graham and Jones 1996, Plowright and Plowright 1997, Arbulo et al. 2011, Sikora and Kelm 2012, Balfour et al. 2013, Pellissier et al. 2013). Part of the preference may be due to foraging efficiency, as studies have found that longer tongued bees are more efficient on longer corolla flowers than short tongued bees, and vice versa (Inouye 1980, Graham and Jones 1996, Plowright and Plowright 1997, Raine and Chittka 2005). Ultimately, the main result to come out of these myriad and varied studies on bumblebee flower preferences is evidence of resource partitioning, exhibited through different morphologies.

Intriguingly, there has also been some suggestions that traits other than morphology can drive resource partitioning in bumblebees (Heinrich 1976). The best example of non‐morphological floral resource partitioning in bumblebees is “nectar robbing” behavior, which is the propensity of some short‐tongued bumblebee species to bite holes in long tubular flowers to access the nectar. (Stout et al. 2000, Barlow et al. 2017). Goulson et al. (2008), when examining diet breadth in European bumblebees, surmises that the differences between physically similar species could be accounted for by high levels of taxonomic diversity (physically similar species belonging in different genera and thus showing differences in diet due to evolutionary separation). Another proposed explanation is temporal differences and ecological divergence (Scriven et al. 2016).

To date, there have not been any definitive studies demonstrating that behavioral traits other than nectar robbing behavior mediating floral resource partitioning in bumblebees. In addition, many recent field studies looking at flower preferences or efficiency have focused solely on a short tongued and a long tongued species in their research area, rather than examining the bumblebee community in its entirety. In order to provide more comprehensive insight into mechanisms of resource partitioning in bumblebees, we conducted intensive landscape‐scale field surveys of bumblebee‐flower interactions around the state of Massachusetts over a three year period. Here, we define landscape scale as an area that is greater than 8 acres and covers multiple habitat types (e.g. old field and wetland). The goal of these surveys was to determine if there were any differences in floral nectar and pollen preferences among morphologically similar and dissimilar bumblebee species in different regions (high and low elevation and at different times of the year (early, mid and late season). In addition to examining flower preferences between species, we also tested for variation in floral resource preference within species (males and workers). Methods Data analysis The bumblebee preference data was processed using the bipartite package of R. The data used in the bipartite plots were separated by low and high elevation due to different bumblebee species composition. B. ternarius and B. borealis were only found at sites or counties that would be considered “high” elevation, and their presence or absence could impact how floral resources are partitioned. For the bipartite plots used for visualization of partitioning between the different morphological groups (Table 4.1), the data were separated into early, middle, and late season (see Chapter 2 for demarcations of early, middle, and late season) to allow good visualization of all the data. For the bipartite plots used for visualization of partitioning within morphological groups, the data during the time of greatest diversity were used. The time of greatest diversity was the time where all the bumblebee species were

24 present at the site, which provided the most accurate picture of any niche partitioning. The function “plotweb” was used to create bipartite webs for all relevant data. “Specieslevel” functions were used to analyze the data (F. Dormann et al. 2009). While many indices were given for an output, only those most relevant to the results are mentioned below.

Table 4.1: Bumblebee species in each morphological group

Short tongued species Medium tongued species Long tongued species B. ternarius B. griseocollis B. bimaculatus B. terricola B. impatiens B. borealis B. perplexus B. fervidus B. vagans

To determine which bumblebee species are more specialists (if any), the specieslevel function “d’” was used as a specialization index by considering interaction frequencies, accounting for both the diversity of partners and their availability, where 1=complete specialization and 0=complete generalization (Blüthgen et al. 2006, F. Dormann et al. 2009, Dormann 2017, Magrach et al. 2018) (Equation 4.1). d’ can also have larger values for species that are found on otherwise rarely visited flower species (Blüthgen 2010). “Nested rank”, which also quantifies the generalization of the species (the higher the value the more specialist the species), was also used as a measure of specialization (Alarcón 2008, F. Dormann et al. 2009, Dormann 2017, Magrach et al. 2018). Nested rank is determined by examining the number of interactions a species has; the greater number of interactions, the lower the nested rank and the more generalized a species is considered. A one‐way ANOVA was run for each of these specialization indices for both workers and males, to compare between species.

Equation 4.1: Formula for d’ specialization index

∗ln di=specialization index of species i c=number of animal species p’ij=distribution of interactions with each partner qj=overall partner availability

For determining species‐specific preferences, Chi‐squared tests were used. The time of greatest species diversity was used for the analysis to show the most accurate picture of niche partitioning and competition happening at the field site. Only workers were analyzed due to males and queens having small sample sizes. The estimated total flower abundances were used to formulate the expected values for each field site. Only floral resources that bumblebees had been observed to visit were considered in the flower abundances. The total flower abundance assumed that all resources were available to all bumblebee species. The null hypothesis for the species‐specific preference data was that there were no species‐specific preferences, and the bumblebee visits are proportional to according to what flowers are proportionally most abundant at the site at the time of the greatest diversity.

25

Results Bumblebee species show resource partitioning based on morphological traits (tongue length) The first results to be established is whether the morphologically dissimilar bumblebee species showed different preferences, as has been established through multiple prior studies. The example bipartite plot, shown below, shows that while some flower species are only visited by one tongue group (e.g. comfrey by long tongued bees, 8/14 species total in plot below), many (6/14) flower species were visited by multiple tongue groups. In the case of common milkweed, all three tongue groups were seen to be visiting it.

26

27

Figure 4.1: Bipartite plots of short tongued vs. medium tongued vs. long tongued bumblebee species during early season at high elevation. Middle and late season at high elevation and early, middle, and late season at low elevation can be found in Appendix H. Only worker data included for clarity. See Table 4.1 for list of bumblebee species in each tongue group

As there were visual overlaps between the different tongue groups for multiple flower species indicating that not all flower species were only visited by one bumblebee tongue group (Figure 4.1), all visited flowers were categorized by flower shape type to determine if there were any specific flower shape types (rather than flower species) that each bumblebee tongue group selected (see Figure 4.2). The flowers, their color and flower shape, and the types of bumblebees that were observed visiting them can be seen summarized in Appendix F.

Figure 4.2: Diagram of flower shape type categories. From left to right: no tube, open tube, closed tube, narrow tube, and spiked tube.

Using the flower shapes seen in Figure 4.2, long tongued bees generally visited closed tube (4/33 visited species), spiked tube (2/33), or narrow tube flowers (7/33) (that have longer corollas). Short tongued bees utilized no tube (13/25 visited species), open tube (5/25), and some narrow tube flower species (5/25) (no tube typically have no corollas), with nectar robbing short tongued bees robbing from flower shape types used by long tongued bees. All flower types were more or less equally used by medium tongued bees, utilizing everything (5/5 closed tube flower species, 7/8 narrow tube, 2/2 spiked tube, 29/30 no tube, 24/26 open tube).

Bumblebee workers show resource partitioning among morphologically similar species As it had been determined above that the field studies were robust and showed consistency with the many studies that demonstrated niche partitioning among morphologically different bumblebee species, the flower preferences of morphologically similar bumblebee species were examined. The summary of top worker flower preferences can be seen below in Table 4.2. Some species, like B. griseocollis, are consistent with their top flower preference through field sites and years. Others, like B. perplexus, appear to have different top flower preferences depending on the site.

Table 4.2: Top preferences for each bumblebee species per field site per year (*no purple loosestrife in Breakneck 2017). Worker data only included. Percentage of visits shown below flower species. Yellow color indicates short tongue, green for medium tongue, and blue for long tongue.

Bombus Breakneck Breakneck Breakneck Wachusett Wachusett Ashfield Heath 2015 2016 2017* 2016 2017 2017 2017

Grass leaved Canada Canada Early ternarius N/A N/A N/A goldenrod goldenrod goldenrod goldenrod (Euthamia (Solidago (38%) (Solidago

28

graminifolia) canadensis) juncea) (32%) (37%) (39%) Common milkweed terricola N/A N/A N/A N/A N/A N/A (Asclepias syriaca) (60%) Common Common Common Common Common Common Common griseocollis milkweed milkweed milkweed milkweed milkweed milkweed milkweed (69%) (50%) (51%) (87%) (92%) (83%) (69%) Purple Wild red loosestrife Purple Canada Canada Canada Canada raspberry impatiens (Lythrum loosestrife goldenrod goldenrod goldenrod goldenrod (Rubus salicaria) (58%) (38%) (30%) (38%) (27%) idaeus) (49%) (32%) Northern bush Northern Red clover Wild red Common Common honeysuckle bush (Trifolium perplexus raspberry milkweed milkweed N/A (Diervilla honeysuckle pratense) (57%) (59%) (44%) lonicera) (71%) (67%) (43%) Hairy Common Cow vetch hedgenettle comfrey Red clover (Vicia Cow vetch Cow vetch Cow vetch bimaculatus (Stachys (Symphytum (61%) cracca) (52%) (55%) (38%) hispida) officinale) (55%) (32%) (60%) Cow vetch borealis N/A N/A N/A N/A N/A N/A (38%) Jewelweed Cow vetch/ Cow vetch Cow vetch (Impatiens fervidus N/A Red clover N/A N/A (64%) (100%) capensis) (50%/50%) (100%) Cow Common Common Red clover Red clover vetch/ Red Red clover Jewelweed vagans milkweed milkweed (46%) (28%) clover (43%) (29%) (52%) (35%) (17%/17%)

When the preference data for each species are pooled, it becomes clear that each species has a select few flower species that they will preferentially visit (Figure 4.3). For example, common milkweed makes up almost 80% of the flower visits for B. griseocollis during the full season. This milkweed preference also holds true when only examining the time of greatest diversity, which is the time when niche partitioning would be seen the clearest. There is more evidence for niche partitioning based on morphological differences, as only long tongued bees preferred red clover. However, this graph also shows differences in preferences between bumblebee species in the same tongue group. For example, B. perplexus is the only bumblebee species to prefer northern bush honeysuckle, and B. impatiens is the only bumblebee species to prefer purple loosestrife.

29

In addition, all of the observed flower visit proportions were significantly different than the proportions that would have been expected if visits were only dependent on floral abundance. This indicates that these displayed species‐specific preferences in Figure 4.3 cannot be explained by the flower species availability, and these are due to the bumblebee species showing different preferences for different flower species.

Short Medium Long * 1.0 * * Great burdock * 0.8 * * Hairy hedgenettle * * Purple loosestrife 0.6 * Red clover 0.4 Common milkweed Northern bush honeysuckle 0.2 Cow vetch

Proportion of observationsProportion of 0.0 Spreading dogbane s us a lis ns us s li s s ri ol l e x tu a du an a ic co ti la e i g n rr o a ple r rv r e r cu bo e va te te mp e a f ris i p m g bi

Short Medium Long 1.0

0.8 Purple loosestrife Goldenrod 0.6 Red clover 0.4 Common milkweed Northern bush honeysuckle 0.2 Cow vetch

Proportionobservations of 0.0 Early goldenrod s s s s s s s s iu ola lli n u u li u r o ie x at id an a ric c t le l rea v g rn r o a p u o r a e te e p er ac b fe v t is im r p im g b

Figure 4.3: Top preferences for workers of each bumblebee species at all sites. Top: time of greatest diversity; bottom: full season. Time of greatest diversity: late June to late July in all years. Species are separated by tongue length (short, medium, long). * indicates significant difference (P<0.001) in Chi‐ squared results for each bumblebee species. Expected: proportion of floral species abundance= proportion visited.

30

31

Figure 4.4: Bipartite plots of bumblebee workers during the time of greatest diversity. Time of greatest diversity: late June to late July in all years. Width of bars indicates relative abundance. Width of connections indicates strength of connection (wide connection=strong connection). Medium tongued bees at low elevation shown here. Short, medium, and long tongued bees at high elevation and long tongued bees at low elevation can be seen in Appendix I.

These data showed that while a bumblebee species was often connected to more than one flower species during the time of greatest species diversity, they did not visit these flower species equally despite the floral species all being available. Instead, one to a few flower species received the bulk of their attention and visitation. For example, common milkweed makes up the majority (>75%) of the B. griseocollis visits (Figure 4.3). The same is true for red clover, cow vetch, and B. bimaculatus. B. perplexus workers preferentially visits common milkweed and northern bush honeysuckle, while B. impatiens seems to prefer purple loosestrife and cow vetch. In the sample bipartite plot above (Figure 4.4), it can be seen that when comparing bumblebees of the same tongue group there are flower species that only one bumblebee species would visit, despite all bumblebees having the same tongue length. For example, B. impatiens is the only medium tongued species to visit alsike clover, and B. griseocollis is the only medium tongued species to visit swamp milkweed (Figure 4.4). In addition, these bipartite plots show the number of flower species that are utilized by different species. While B. impatiens is willing to visit 26 flower species, B. griseocollis (12) and B. perplexus (7) visit fewer flower species, despite all the flower species being available to them.

Bumblebee males show resource partitioning among morphologically similar species When the males are examined for species‐specific flower preferences, preferential visitation becomes even clearer. For example, common milkweed continues to make up the bulk of the B. griseocollis male visits. The B. bimaculatus males show a preference for purple loosestrife when available, hairy hedgenettle when available, and wild bergamot in general. B. perplexus workers preferentially visits common milkweed and northern bush honeysuckle like their worker counterparts (Table 4.3 and Figure 4.5). B. impatiens males prefer multiple species of goldenrod regardless of field site or year, unlike the B. impatiens workers who prefer purple loosestrife (Table 4.3 and Figure 4.3).

Table 4.3: Top preferences for each bumblebee species per field site per year (*no purple loosestrife in Breakneck 2017). Male data only included. Percentage of visits shown below flower species. Yellow color indicates short tongue, green for medium tongue, and blue for long tongue.

Bombus Breakneck Breakneck Breakneck 2017* Wachusett Wachusett Ashfield Heath 2015 2016 2016 2017 2017 2017

Wrinkleleaved Grassleaved Early goldenrod ternarius N/A N/A N/A goldenrod N/A goldenrod (Solidago (43%) (100%) rugosa) (30%) Common terricola N/A N/A N/A N/A N/A N/A milkweed (57%)

32

Spreading Common Purple dogbane Common Common Common griseocollis milkweed loosestrife (Apocynum milkweed milkweed N/A milkweed (38%) (51%) androsaemifolium) (68%) (49%) (100%) (39%) Tall Canada Canada Canada Canada goldenrod Wrinkleleaved impatiens goldenrod goldenrod goldenrod goldenrod (Solidago N/A goldenrod (31%) (30%) (51%) (66%) (35%) altissima) (32%) Common Common Common Cow Common Northern bush perplexus milkweed N/A milkweed milkweed vetch milkweed honeysuckle (40%) (67%) (73%) (61%) (100%) (100%) Purple Wild bergamot Hairy Wild Wild red Red clover Red clover bimaculatus loosestrife (Monarda hedgenettle bergamot raspberry (48%) (48%) (56%) fistulosa) (67%) (76%) (100%) (50%) Red clover borealis N/A N/A N/A N/A N/A N/A (60%) Toad flax Bull thistle (Linaria (Cirsium fervidus N/A N/A N/A N/A N/A vulgaris) vulgare) (100%) (100%) Purple Joe pye weed (Eutrochium purpureum) /Tall Purple Common Common Wild goldenrod/Lance‐ Jewelweed vagans N/A loosestrife milkweed milkweed bergamot leaved aster (100%) (100%) (75%) (60%) (100%) (Symphyotrichum lanceolatum) (33%/33%/33%)

33

Short Medium Long

Short Medium Long

Figure 4.5: Top preferences for males of each bumblebee species at all sites. Top: time of greatest diversity; bottom: full season. Time of greatest diversity: mid July to mid August in all years. Species are separated by tongue length (short, medium, long).

34

Figure 4.6: Bipartite plots of bumblebee males during the time of greatest. Time of greatest diversity: mid July to mid August in all years. Width of bars indicates relative abundance. Width of connections indicates strength of connection (wide connection=strong connection). Medium tongued bees at low elevation shown here. Short, medium, and long tongued bees at high elevation and long tongued bees at low elevation can be seen in Appendix J.

35

In the sample bipartite plot above (Figure 4.6), it can be seen that like the workers, there are flower species that only one bumblebee species would visit, despite all bumblebees having the same tongue length. For example, B. impatiens is the only medium tongued species to visit Joe pye weed and alsike clover, and B. griseocollis is the only medium tongued species to visit swamp milkweed and spearmint (Figure 4.6). In addition, these bipartite plots show the number of flower species that are utilized by different species. While B. impatiens is willing to visit 10 flower species, B. griseocollis (8) and B. perplexus (5) visit fewer flower species, despite all the flower species being available to them.

Bumblebee species show variance in number of flower species they will visit The next thing to examine is whether they are true “generalists”, willing to visit many different flower species, or if some bumblebee species are more specialist than others. Tables 4.4 and 4.5 below show the worker and male average visited flower species diversity for all bumblebee species at all field sites over all survey years. For the entire season, B. impatiens workers visited the most flower species on average with 25 different flower species visited. Comparatively, the species with the lowest flower species diversity was B. fervidus with only 2 flower species visited. The rest of the worker data range in between these two values, though closer to the lower end of the scale (5 species, 9 species, 8 species, etc.). While the workers are willing to visit several flower species (>3), the males tended to show a much more limited flower species diversity, ranging from 13 flower species (B. impatiens) to 1 flower species (B. fervidus).

Using two specialization indices associated with bipartite plots, d’ and “nested rank”, the bumblebee species were ranked for both workers and males by how specialized they are on average, and the averages were compared using a 1‐way ANOVA. “Nested rank” was determined by counting the number of interactions a species has; the greater number of interactions, the lower the nested rank and the more generalized a species is considered. d’ was determined using interaction frequencies, accounting for both the availability of partners and their diversity. d’ can also have larger values for pollinators that are found on otherwise rarely visited flower species. Seen below, the results show that there are significant differences between bumblebee species’ degree of specialization for workers and males, though the two specialization indices differ on their rankings of most specialized species (Tables 4.5 and 4.6).

Table 4.5: Specialization results for each bumblebee species over all field seasons. Worker data only included. Higher values= higher level of specialization

Species Avg. number d’ d’ rank (1= “nested rank” “nested rank” of flower (average) most (average) rank (1= most species visited specialized) specialized) B. impatiens 25 0.540556 2 0.083333 9 B. griseocollis 8 0.670376 1 0.459167 5 B. bimaculatus 5 0.516253 3 0.670833 4 B. vagans 9 0.350795 5 0.397143 6 B. perplexus 5 0.439441 4 0.790833 2 B. fervidus 2 0.262108 8 1 1 B. ternarius 9 0.339417 6 0.33 7 B. borealis 9 0.329685 7 0.125 8 B. terricola 4 0.232968 9 0.75 3

36

**P=0.0055 “nested rank” ****P<0.0001 d’ ANOVA: ANOVA:

Table 4.6: Specialization results for each bumblebee species over all field seasons. Male data only included. Higher values= higher level of specialization

Species Avg. number of d’ d’ rank “nested rank” “nested rank” flower species (average) (average) rank visited B. impatiens 13 0.801181 2 0.04 8 B. griseocollis 5 0.470907 6 0.546667 4 B. bimaculatus 4 0.800349 3 0.521429 5 B. vagans 2 0.369377 8 0.869444 2 B. perplexus 4 0.559799 4 0.45 6 B. fervidus 1 0.906545 1 1 1 B. ternarius 4 0.527214 5 0.555556 3 B. borealis 2 0.308883 9 0.333333 7 B. terricola 4 0.406598 7 0 9

*P=0.0333 “nested rank” ****P<0.0001 d’ ANOVA: ANOVA:

Flower species show variance in number of bumblebee species that visit them Previous results have shown preferences and partitioning from the bumblebee’s perspective. From the plant’s perspective, there were also uneven visitation. Seen below in Figure 4.7, 2 or 3 bumblebee species make up the majority of the representative flower species’ visitation. In a couple cases, only 1 bumblebee species makes up the majority of visits, such as B. griseocollis with common milkweed and B. impatiens with purple loosestrife (Figure 4.7). In these two cases, the bumblebee species visitation on the flowers and the flower species visitation by the bumblebees is reciprocal. Common milkweed makes up the majority of B. griseocollis worker visits (Figure 4.3), while B. griseocollis accounts for the majority of visits to common milkweed (Figure 4.7). The same is true for B. impatiens and purple loosestrife (Figures 4.3 and 4.7).

37

Proportion of observations

r d e d le ro tt ov trife l vetch s en ne ilkwee w d e m o dg Red c C loose e e h Wild bergamot ry mmon rpl nada gol o u a C P C Hai

Figure 4.7: Overall top bumblebee species workers visiting representative flower species at all sites. Top: time of greatest bumblebee diversity; bottom: full season. Time of greatest diversity: late June to late July in all years. Discussion These results show both a consistency with multiple prior studies regarding resource partitioning based on morphological differences and evidence for a pathway for resource partitioning between morphologically similar species. This data demonstrates that not only do bumblebee species

38 preferentially visit only a few flower species rather than visiting all equally, there are behavioral differences between bumblebees that are physically alike. This is different from the few studies that have proposed possible causes for niche partitioning among morphologically similar bumblebee species, because they focused on physical differences such as time of activity rather than behavior (Heinrich 1976, Goulson et al. 2008b, Scriven et al. 2016).

Based on the compilation of results and analysis, Tables 4.7 and 4.8 below show the apparent top three preferences for workers and males of each species, which can be used for personalized seed mixes. It is clear that there is a more or less unique set or order of flower preferences for each bumblebee species, and even between workers and males of some species. One limitation of these summaries is that these preferences are only ranked based on what was available at the field sites and on the sample sizes of the bumblebee species observations. It is entirely possible that there are other flowers that say, B. terricola workers prefer over the common milkweed, but they were not present at any field site and so cannot be determined without increasing the number of or floral diversity in the field sites. It is also possible that B. fervidus workers prefer jewelweed as a resource over the non‐native flowers like red clover and cow vetch, but there were so few workers seen that the data does not accurate represent the species.

Table 4.7: Recommendation of bumblebee species preferences (worker)

Species Top flower choice Second flower choice Third flower choice B. impatiens Purple loosestrife Canada goldenrod Goldenrod spp. B. griseocollis Common milkweed Spreading dogbane Joe pye weed B. bimaculatus Hairy hedgenettle Cow vetch Red clover B. vagans Red clover Cow vetch Jewelweed Northern bush Common milkweed Wild red raspberry B. perplexus honeysuckle B. fervidus Cow vetch Red clover Jewelweed B. ternarius Goldenrod spp. Common milkweed Aster spp. B. borealis Red clover Vetch Common milkweed Common milkweed Great burdock Jewelweed/Cow vetch B. terricola (for nectar robbing)

Table 4.8: Recommendation of bumblebee species preferences (male)

Species Top flower choice Second flower choice Third flower choice B. impatiens Canada goldenrod Goldenrod spp. Aster spp. B. griseocollis Purple loosestrife Common milkweed Joe pye weed B. bimaculatus Purple loosestrife Wild bergamot Cow vetch/Red clover B. vagans Purple loosestrife Common milkweed Wild bergamot Common milkweed Northern bush Wild bergamot B. perplexus honeysuckle Bull thistle Toad flax Very long tubular flowers (e.g. red B. fervidus clover/cow vetch) B. ternarius Goldenrod spp. Aster spp. Common milkweed

39

Red clover Common milkweed Very long tubular flowers (e.g. B. borealis jewelweed/cow vetch) B. terricola Common milkweed Goldenrod spp. Alsike clover/Alfalfa

Some flower preferences that were found in the current field studies were consistent with Otto Plath’s dissertation on Massachusetts bumblebees almost 100 years ago. For instance, B. griseocollis workers were found in the current field studies to have a very strong common milkweed preference, and his dissertation notes that they “are especially fond of milkweed” (Plath 1927). They also show consistency with red clover and vetch preferences. B. bimaculatus workers have consistency in their preference for red clover, vetch, and northern bush honeysuckle when available (Plath 1927). B. impatiens workers have the same observed preferences for purple loosestrife, vetch, jewelweed, and goldenrod (Plath 1927). B. perplexus workers show consistency between current and Plath’s studies of raspberry and northern bush honeysuckle (Plath 1927). B. vagans workers consistent preferences with red clover and vetch (Plath 1927). Finally, B. fervidus workers show consistency with preferences of red clover and vetch, though it should be noted that the sample size for B. fervidus in current field studies is quite small (Plath 1927).

The same consistencies between Plath’s dissertation and the current surveys were also found for many species for males. For example. B. bimaculatus males then and now show a preference for red clover, vetch, purple loosestrife, and honeysuckle (Plath 1927). B. ternarius males have the same strong preferences for goldenrod species (Plath 1927). Finally, B. impatiens males show similar preferences for goldenrod as B. ternarius males, in addition to aster species (Plath 1927).

The knowledge of what the bumblebee species are visiting and what they prefer could show what floral characteristics they prefer. For example, B. bimaculatus workers visited 16 different flower species for nectar across years and field sites. If they preferred a certain flower shape, then it would show as the majority for the flower species. However, there is no majority (highest is open tube, 7/16 species). Nor is there a clear majority for flower color which could explain their preference. The workers were shown to visit flowers that contained 5 different colors (Table 4.9). Based off the scientific families that the flowers belong to, B. bimaculatus workers were also not preferring flowers in a certain botanical family. While B. bimaculatus is just an example, it shows that determining why this particular species prefers these flowers cannot be determined from basic floral morphology. These preferences may be due to some characteristic that cannot be easily determined, such as flower odor.

Table 4.9: Flowers that B. bimaculatus workers visited, with scientific name and characteristics.

Common name Family Color Shape Birdsfoot trefoil (Lotus Fabaceae Yellow Closed tube corniculatus) Crown vetch (Securigera Fabaceae Purple Closed tube varia) Cow vetch (Vicia cracca) Fabaceae Purple Closed tube Great burdock (Acrtium Asteraceae Pink Narrow tube lappa)

40

Bull thistle (Cirsium Asteraceae Purple Narrow tube vulgare) Wild bergamot (Monarda Lamiaceae Purple Narrow tube fistulosa) Red clover (Trifolium Fabaceae Red Narrow tube pratense) Wild red raspberry (Rubus Rosaceae White No tube idaeus) Common blackberry Rosaceae White No tube (Rubus allegheniensis) Hedge false bindweed Convolvulaceae White/pink Open tube (Calystegia sepium) Northern bush Caprifoliaceae honeysuckle (Diervilla Yellow Open tube lonicera) Purple loosestrife Lythraceae Purple Open tube (Lythrum salicaria) Penstamon cultivar Planatinaceae Pink Open tube (Penstamon spp.) Hairy hedgenettle Lamiaceae Purple Open tube (Stachys hispida) Common comfrey Boraginaceae Purple Open tube (Symphytum officinale) Purple cultivar Unknown Purple Open tube

An interesting result is the refusal of B. impatiens to visit common milkweed. B. impatiens could be considered a strong “generalist” species, where they are visiting many more floral resources on average than any other bumblebee species (Table 4.6). However, they were not observed to visit common milkweed, when both morphologically similar and dissimilar bumblebee species show a willingness to visit it. Given B. impatiens dominance at some sites (e.g. Breakneck) with consistent avoidance of common milkweed, competition from more abundant species does not appear likely to be the cause. If the cause is not morphological mismatch and is not from being outcompeted, the most likely possible cause of this avoidance has something to do with common milkweed itself that the species does not like, such as odor or nectar taste. If that is indeed the case, this would indicate that the flowers have the ability to both attract some bumblebee species and deter others down to the species (as opposed to attracting/deterring all bumblebee species) (Barlow et al. 2017).

These observed behavioral differences between bumblebee species that are physically similar could help explain why some species such as B. vagans are declining and other like B. impatiens are increasing. There is always competition within a community, to gain the most high‐quality resources as possible. If flower species are not “one size fits all”, then some bumblebee species may benefit while others end up at a disadvantage because they cannot exploit those flowers. Possible decline causes such as climate change and habitat loss can cause changes in the flower part of the community such as a decrease in floral resources, as well as any introduction of exotic flowers. These impacts combined with the naturally competitive nature of a pollinator community would create a group of “winning” species and

41

“losing” species, where the “losing” species are the ones in decline. There are studies that surmise that the more specialized the bumblebee species is, the greater risk it has to be in decline and therefore become a “losing” species (Goulson and Darvill 2004, Goulson et al. 2008b, Williams et al. 2009). While this idea makes sense, one confounding variable is that rarer bumblebee species will naturally have fewer observations than common species. Because of the relatively small number of observations, the number of flower species that the rarer species are seen on could be fewer than is accurate for the species, leading to a false conclusion that the rarer species have more specialized diets.

On the opposite end of the scale, the “winning” species could be more likely to be willing to visit many flower species, which would make them more equipped to deal with a change in their available flower species. An example of this is B. impatiens, which is a more generalist bumblebee species, and has also significantly increased in abundance in Massachusetts (Chapter 3). It is also possible that the “winning” species may have more flexibility in adapting to non‐native flowers that invade the community. With the introduction of an exotic flower, the bumblebee species who can exploit that flower have an additional resource to use, giving them an advantage over the species that can’t. At this point though, there is little information on how exotic flowers can negatively impact native pollinator communities.

The results from this study have far reaching implications. Firstly, the evidence that there are species specific preferences among morphologically similar species refutes the general assumption that bumblebees are “generalists” and will forage on whatever flowers that are available for them to use. The evidence calls for an entire shift of how scientists examine bumblebee flower preferences, to distinct species that are filling distinct niches within the ecosystem.

Having this new information will help combat bumblebee species decline by providing the knowledge that “bumblebee” flowers are not one size fits all. Based off the recommendations of top flower choices in Tables 4.7 and 4.8 multiple flower species are necessary to provide for the full range of bumblebee species in a specific region. When preserving habitat for bumblebees or other pollinators, knowing the flower preferences of the bumblebee species in the region gives the person examining the the ability to evaluate which potential areas would benefit the greatest overall diversity of species or ones in particular decline. The flower recommendations in Tables 4.7 and 4.8 can also be used to create personalized seed mixes, which then could be utilized in habitat improvement/enhancement for bumblebee species. If someone is trying to make an old field more suitable to declining species such as B. terricola or B. fervidus, they could use the flower recommendations for each species as what to plant to make that field more able to provide floral resources to support B. terricola or B. fervidus populations.

42

Chapter 5: Influence of Exotic Plant Species in Bumblebee Community Dynamics: A Manipulative Study Introduction Biological invasions by exotic species are considered a high threat to biodiversity worldwide (Keane and Crawley 2002, Turbelin et al. 2017). These exotic species have the capability of dramatically changing native ecosystems, causing biodiversity loss (Levine et al. 2003, MacDougall and Turkington 2005, Traveset and Richardson 2006, Orrock et al. 2015, Gallardo et al. 2016). Exotic plant invasion into a community can directly impact the native flowers living there. The plants can compete for space in the area through faster growth rates and lack of herbivory, pushing out the native flowers (Ehrenfeld 2003). There is also good evidence showing that exotic plants can alter the soil chemistry in the community, such as changing nitrogen fixation rates (G. Ehrenfeld et al. 2001, Ehrenfeld 2003, Wolfe and Klironomos 2005). Soil microbes can also affect the balance, by being less harsh against the exotic plants compared to the native plants (Callaway et al. 2004, Wolfe and Klironomos 2005). Finally, exotic plants can directly impact native plants through allelopathy, or use of toxins by plants to negatively impact competitors (Ridenour and Callaway 2001, Bais et al. 2003, Hierro and Callaway 2003, Callaway and Maron 2006, Grove et al. 2012).

Exotic plants can also negatively impact native flowers in an indirect fashion, by manipulating pollinators. Exotic flower competition for pollinators can lead to lower seed set for the native flowers, resulting in an overall lower reproductive success. There is increasing evidence that shows exotic flowering plants such as purple loosestrife are very capable of “pulling” pollinators like bumblebees away from native flowers, decreasing visitation (Brown et al. 2002, King and Sargent 2012). This leads to the lower reproductive success for the native flowers, allowing the exotic flowers to become dominant in the area (Grabas and Laverty 1999, Brown and Mitchell 2001, Flanagan et al. 2009, Flanagan et al. 2010, 2011, King and Sargent 2012, Skurski 2014, Albrecht et al. 2016). They may also work with non‐ native pollinators such as honeybees to replicate the partnership that they had in their native area, putting competitive pressure on both the native plants and pollinators (Barthell et al. 2001). There remain some arguments however that non‐native plants can also help the native plants, by “pulling” pollinators to areas where native flowers have limited pollination (Vilxe et al. 2009, McKinney and Goodell 2011, Jakobsson and Padrón 2014, Skurski 2014).

Pollinators worldwide are in a general state of decline, and the bumblebees of North America are among them (Cox and Elmqvist 2000, Goulson et al. 2005, Goulson et al. 2008a, Potts et al. 2010, Goulson and Nicholls 2016). While many studies on population decline have focused on their cousins the honeybees due to their agricultural importance, bumblebees are an essential part of the ecosystems they live in, and their importance cannot be ignored. There have been several proposed causes for bumblebee decline, but no agreement as to the definitive cause. These proposed causes range from neonicotinoid pesticides that kill bumblebees at high doses and negatively impact foraging behavior and reproductive success at low doses (Feltham et al. 2014, Sanchez‐Bayo and Goka 2014), to habitat loss from increase agriculture and urban development (Osborne et al. 2008, Goulson and Nicholls 2016). Climate change which is shrinking or altering suitable habitat (Ploquin et al. 2013), new diseases transmitted by non‐ native pollinators such as honeybees (Graystock et al. 2013), and competition from exotic species such as honeybees (Thomson 2004) are also included.

43

While the effect of exotic plants on native plants has been well studied, we presently know little about effects on the structure and dynamics of pollinator communities and importantly, the potential role of exotic plant species in the differential pollinator declines observed worldwide over the past decade. From the pollinator’s perspective, exotic plants represent a new source of nectar and pollen and thus could be beneficial to some species. However, exotic floral resources can also have the potential to negatively impact bumblebee diversity in many ways. Exotic plants can competitively exclude native plants that serve as the primary food source for some species (King and Sargent 2012, Skurski 2014, Albrecht et al. 2016). Through mutualisms with honeybees, exotic flowers can help support the non‐ native honeybees as they compete against the native bumblebees (Barthell et al. 2001). Finally, exotics may substantially increase the competitive ability of some bumblebee species while at the same time reducing the competitive ability of others. With the introduction of a non‐native flower, it is possible that the bumblebee species who can best exploit that new flower will have an additional resource to use, eventually giving them an advantage over the species that preferentially avoid the exotic (Figure 5.1). Over time, this competitive imbalance would produce ecological ‘winners’ (stable species). The reverse would be true for the bumblebee species that can’t exploit that new resource, resulting in competitive ‘losers’. They would also be contending with the “winning” species, who may be able to outcompete them on floral resources available after the exotic goes out of bloom through greater numbers. This is similar to the concept of biotic homogenization due to global change (Baskin 1998, McKinney and Lockwood 1999).

Figure 5.1: Diagram detailing the competition imbalance hypothesis. When a new resource is introduced into the community, the bumblebees can either exploit the resource or they cannot. Those that can exploit the resource thus have more resources (“winners”) and remain stable. Those that cannot exploit the resource have fewer resources (“losers”) and decline through fewer resources and being outcompeted by the “winners”.

On the other hand, it is also possible that the exotic flower introduction could be a disadvantage for any bumblebee species that can exploit it. An ecological trap, or a poor‐quality resource that a species will choose over a high‐quality resource, can negatively impact that species (Schlaepfer et al. 2002, Battin 2004). The cause is speculated to be a mismatch in environmental cues used to evaluate resources and

44 the actual resource quality (Battin 2004). Evidence has shown that bumblebee species can partition floral resources in a community both physically and behaviorally (Chapter 4). If the behavioral cues lead a species to prefer poor‐quality exotic flowers instead of higher quality native flowers, the bumblebee species would have fewer high quality resources, which would lead to it becoming a competitive “loser”, the opposite scenario to the competition imbalance hypothesis mentioned previously. However, there is no information on whether pollinators in general can fall into ecological traps, with the majority of studies focusing on birds (Weldon and Haddad 2005, Bruce and Hutto 2007, Bonnington et al. 2015, Sherley et al. 2017). There is also little information on how exotic flowers can negatively impact the pollinators in the community, so there is no data to show if exotic flowers are an ecological trap for bumblebees.

While the competition imbalance through exotic resource availability could explain only some bumblebee species declining while others increase in population, this hypothesis has yet to be empirically tested. To fill this knowledge gap, we conducted intensive field surveys of bumblebee‐flower interactions at sites in Massachusetts over a two‐year period. The goal of these surveys was to determine if there were any variation in exotic flower preferences among bumblebee species. Our analysis of exotic floral resource use among bumblebee species revealed that purple loosestrife (Lythrum salicaria), a highly invasive exotic, was strongly preferred by B. impatiens, a species that has dramatically increased in abundance and distribution in MA as other species head for extirpation. To test for positive effects of loosestrife on B. impatiens abundance and negative effects of loosestrife on visitation to native plant species, as predicted by our competition imbalance hypothesis, we conducted a loosestrife removal experiment at Breakneck Hill during 2017 season. Collectively, our findings are consistent with the view that the introduction of exotic floral resources has a negative impact on the dynamics and diversity of bumblebee‐native plant pollination systems. Methods I Site‐level analysis of exotic floral resource usage The different bumblebee species seen in 2015 and 2016 were examined for preferences for native or non‐native flowers. Only Breakneck and Wachusett field sites were used, to compare two sites with similar flower species composition, with the exception of purple loosestrife (it was present at Breakneck and absent at Wachusett). The worker data for native and non‐native flower visits for each bumblebee species were pooled across the entirety of the field season, then compared as proportions visited. Native and non‐native flowers were classified as those flowers being native to Massachusetts and those not native to Massachusetts (Table 5.1).

Table 5.1: Flower species that bumblebee species were observed visiting during the 2016 field season, sorted by native and non‐native to Massachusetts.

Native Non‐native Spreading dogbane (Apocynum androsaemifolium) Cow vetch (Vicia cracca) Northern bush honeysuckle (Diervilla lonicera) Red clover (Trifolium pratense) Wild red raspberry (Rubus idaeus) Birdsfoot trefoil (Lotus corniculatus) Flowering raspberry (Rubus odoratus) Climbing nightshade ( dulcamara) Common milkweed (Asclepias syriaca) Purple loosestrife (Lythrum salicaria) Showy tick trefoil (Desmodium canadense) Common St. John's wort (Hypericum perforatum) Carolina nightshade (Solanum carolinense) Alsike clover (Trifolium hybridum)

45

Swamp milkweed (Asclepias incarnata) English plantain (Plantago lanceolata) Wild bergamot (Monarda fistulosa) Creeping thistle (Cirsium arvense) White meadowsweet (Spiraea alba) Queen Anne's lace (Daucus carota) Rosy meadowsweet (Spiraea tomentosa) Bull thistle (Cirsium vulgare) Blue vervain (Verbena hastata) Spearmint (Mentha spicata) Canada goldenrod (Solidago canadensis) Common chicory (Cichorium intybus) Allegheny monkeyflower (Mimulus ringens) Great burdock (Acrtium lappa) Purple Joe pye weed (Eutrochium purpureum) Spotted knapweed (Centaurea maculosa) Fleabane daisy (Erigeron strigosus) Toad flax (Linaria vulgaris) Jewelweed (Impatiens capensis) Grassleaved goldenrod (Euthamia graminifolia) Tall lettuce (Lactuca canadensis) Common evening primrose (Oenothera biennis) Boneset (Eupatorium perfoliatum) Common self‐heal (Prunella vulgaris) Tall goldenrod (Solidago altissima) Calico aster (Symphyotrichum lateriflorum) Great blue lobelia (Lobelia syphilitica) Arrowleaved tearthumb (Persicaria sagittata) American burnweed (Erechtites hieraciifolius) New England aster (Symphyotrichum novae‐

angliae) Purple‐stemmed aster (Symphyotrichum

puniceum) Late purple aster (Symphyotrichum patens) Lanceleaved aster (Symphyotrichum lanceolatum) Hairy hedgenettle (Stachys hispida) Tall white‐aster (Doellingeria umbellata)

Shannon’s diversity index (H’) was used to measure bumblebee species diversity for comparison between the two years due to it being a standard biodiversity test in literature (Morris et al. 2014). The higher the Shannon’s diversity index, the more diverse the surveyed community. Results I Bumblebee species show variance in willingness to utilize non‐native floral species The bumblebee workers showed a surprising variety of variation between species, ranging from completely preferring non‐native flowers to almost completely preferring native ones (Figure 5.2). Some of these preferences appear to be site‐specific, while others remain consistent regardless of site. For example, B. impatiens workers at Breakneck in 2015 and 2016 had a non‐native flower preference, but Wachusett 2016 shows a native flower preference. Species that had a consistent rather than site‐ specific non‐native flower preference were B. bimaculatus and B. fervidus. In contrast, B. griseocollis and B. perplexus had a consistent preference for native flower species across sites.

46

Bush honeysuckle

Red clover

Cow vetch

Common milkweed

Purple loosestrife

Common milkweed

Common milkweed

Cow vetch

Common milkweed

Canada goldenrod

Figure 5.2: Bumblebee species native vs. non‐native flower preferences. Proportion of observations on native flowers vs. proportion of observations on non‐native flowers. Values range from ‐1 (only non‐ native flower visits) to 1 (only native flower visits). Greatest flower preference for each is labelled and outlined in black. Only worker data included for clarity. From top to bottom: Breakneck 2016 and Wachusett 2016.

Data from the 2016 field season showed that in particular, purple loosestrife had a large influence on B. impatiens workers at the Breakneck field site, contributing to over 57% of their flower visits despite their willingness to visit 41 different flower species over the course of the season (Figure 5.2). The purple loosestrife presence at Breakneck appeared to also drive a higher population of B. impatiens workers compared to Wachusett, a similar field site that did not contain purple loosestrife (Figure 5.3). There was also an extreme amount of B. impatiens workers compared to workers of other Bombus

47 species at Breakneck (>90%), leading to a low Shannon’s diversity index of 0.27 (Table 5.2). In contrast, the 2016 B. impatiens populations at the Wachusett field site were not the most dominant species, leading to a Shannon’s diversity index of 1.29 (Table 5.2).

Table 5.2: Shannon diversity indices for Breakneck workers 2016 and 2017

Site Early season Midseason Late season Overall Early June‐mid Mid July‐mid Mid August‐end July August of season Breakneck 2016 1.26 0.12 0.004 0.27 Wachusett 2016 1.06 1.06 0.12 1.29

Figure 5.3: Relationship between purple loosestrife and B. impatiens worker population in 2016. The total number of B. impatiens worker observations at Breakneck is different than that of the number of B. impatiens worker observations on purple loosestrife. The blue arrow indicates the start of purple loosestrife bloom, while the black arrow indicates the end.

Effect of loosestrife removal on bumblebee‐native plant interactions Based on strong positive relationship between the purple loosestrife and B. impatiens abundance at Breakneck in 2016 and the relatively low abundance of other species at the site, we posited that purple loosestrife was inflating B. impatiens populations at Breakneck, thereby giving B. impatiens a competitive advantage over other species post‐loosestrife bloom. This could explain the large disparity in relative B. impatiens abundance between sites like Wachusett that lack loosestrife but have similar flower compositions otherwise.

To that end, a manipulative experiment was proposed for the field site, where as much purple loosestrife as possible was removed for the 2017 field season (with the ultimate goal of eliminating 100%) and the same type of bumblebee field survey was run to determine the differences in B. impatiens population and flower visitation compared to the baseline (2016). There were two hypotheses

48 to this study, to examine the impacts of purple loosestrife from both the native plant and native pollinator perspectives. The first hypothesis was that removing the purple loosestrife would lead to a lower relative abundance of B. impatiens, a higher relative abundance of other Bombus species at the field site, and a higher overall Shannon’s diversity index. The second hypothesis was that removing the purple loosestrife (a non‐native flower species) would lead to greater visitation of native flowers by B. impatiens and other bumblebee species that may use purple loosestrife, either through increased native flower species visited or through increased native flower visits overall. Methods II Loosestrife removal When the purple loosestrife began flowering in July, volunteers with hand shears and pruners removed as much as possible, concentrating on the plants with visible flowers or buds. When it became clear that the volunteers could not remove enough of the purple loosestrife in a timely enough manner to prevent bumblebee visitation, a landscaping company was called in to clear the rest. There were some places that were inaccessible due to a wetland, so the purple loosestrife was cleared as far as it was physically possible to remove the plants (Figure 5.4). The purple shape in Figure 3 indicates an extremely dense stand of purple loosestrife that was targeted for the removal. As there was no permit for long‐term removal, the rhizomes and root systems were untouched to allow the purple loosestrife to renew the following year.

Figure 5.4: Map of Breakneck Hill Conservation Land showing areas containing purple loosestrife in 2016 (left) and 2017 (right).

Loosestrife flower surveys In 2016 and 2017 at Breakneck Hill Conservation Land, purple loosestrife surveys were carried out to estimate its abundance in the field site for the same period of time as the bumblebee surveys. In order to make the surveys in as different places on the transects as possible, the survey was either done at the beginning, middle or end of the transect, following by the next location (e.g. after doing a survey at the beginning of the transect, the next was carried out in the middle) (See Chapter 2). Within a 2mx5m

49 rectangle along the transect line, the all purple loosestrife inflorescences were counted. Every other transect was surveyed in 2016 in order to gain a clear picture of the survey area while not detracting from the bumblebee survey, and every transect was surveyed in 2017, as the transects for Breakneck were halved for the 2017 season to allow the entire site to be covered in one day. The data were summarized in the total number of inflorescences for purple loosestrife divided by the number of transects surveyed, giving an average number of inflorescences per transect sample (10m2).

Data Analysis The bumblebee preference data was processed using the bipartite package of R. The function “plotweb” was used to create bipartite webs for all relevant data (F. Dormann et al. 2009).

Shannon’s diversity index (H’) was used to measure bumblebee species diversity for comparison between the two years due to it being a standard biodiversity test in literature (Morris et al. 2014). The higher the Shannon’s diversity index, the more diverse the surveyed community.

As two years (sets of data) were being compared, paired T‐tests were run for relative abundance of workers and males for each bumblebee species present at Breakneck, 2016 Shannon’s diversity index vs. 2017, and visited flower species diversity for males and workers of each bumblebee species present. Due to small sample size that would fail the normality test, a Wilcoxon matched pairs test was run for each of these comparisons.

In order to determine whether the removal of the purple loosestrife led to greater visitation of native flowers for B. impatiens, a Chi‐squared test was run for the native and non‐native proportions of flower visits for both males and workers. The expected proportions were the native and non‐native proportions of flower visits observed in 2016. A Chi‐squared test was also run comparing native vs. non‐native proportions of Breakneck 2016 and Breakneck 2017 workers to Wachusett 2016. The null hypothesis was that the proportions of native and non‐native flower visits are not significantly different between Breakneck and Wachusett, and the expected values were the proportions of native and non‐native flower visits from Wachusett 2016. Results II Removal of purple loosestrife from field site was successful

50

Figure 5.5: Loosestrife pictures 2016 (left) vs. 2017 (right) showing the lower South Meadow of Breakneck Hill Conservation Land. Note the purple loosestrife clumps visible in the left picture that are not on the right.

Based off the results that the purple loosestrife at Breakneck was helping to boost the B. impatiens population compared to a similar site without loosestrife, a manipulative experiment removing the purple loosestrife at Breakneck was run the following year. Given that correct interpretation of the manipulative experiment results depended on how much purple loosestrife had been removed from the site, the estimated purple loosestrife abundance in 2017 was compared to 2016 to see how much had been eliminated. Visuals of a section of the field site known to have purple loosestrife can be seen in Figures 5.4 and 5.5. The total amount of purple loosestrife inflorescences observed in flower surveys in 2017 was 56, compared to 1410 the year before (Figure 5.6). Based on these flower survey observations, it can be estimated that about 96% of the purple loosestrife inflorescences present at the field site in 2016 were absent in 2017. Because almost all of the purple loosestrife had been removed from the site in 2017, it can be said that the purple loosestrife removal was successful.

Figure 5.6: Purple loosestrife inflorescence abundance 2016 vs. 2017. Inflorescence is defined as a purple loosestrife flower cluster/spike. Total amount of purple loosestrife inflorescences observed in vegetation surveys over the course of the season.

Removal of purple loosestrife did not significantly alter B. impatiens population Before the field season in 2017, it had been hypothesized that the removal of the purple loosestrife would significantly alter the B. impatiens population at Breakneck Hill Conservation Land and cause a lower relative abundance. However, there was no significant difference between any bumblebee species populations between the two years, both for workers (Figure 5.7, P=0.8125) and for males (Figure 5.8, P=0.8438).

51

Figure 5.7: Relative species abundance at Breakneck Hill Conservation Land 2016 vs. 2017 starting at the start of purple loosestrife bloom. Worker observations only. Due to large comparative numbers, B. impatiens is in a separate graph for ease of viewing.

Figure 5.8: Relative species abundance at Breakneck Hill Conservation Land 2016 vs. 2017 starting at the start of purple loosestrife bloom. Male observations only. Due to large comparative numbers, B. impatiens is in a separate graph for ease of viewing.

Removal of purple loosestrife did not significantly alter overall bumblebee species diversity The second part of the first hypothesis stated that the removal of the purple loosestrife would negatively change the B. impatiens relative abundance, and this lower relative abundance would lead to a higher Shannon’s diversity index. With hypothetically fewer B. impatiens, that could open up more floral resources for other bumblebee species, increasing their relative diversity and thus increasing the Shannon’s diversity index. However, there was no significant difference between Shannon’s diversity index for 2016 and 2017 (Table 5.3, P=0.25).

Table 5.3: Shannon diversity indices for Breakneck workers 2016 and 2017

Year Early season Midseason Late season Overall

52

Early June‐mid Mid July‐mid Mid August‐end July August of season 2016 1.26 0.12 0.004 0.27 2017 1.32 0.14 0.004 0.30

Removal of purple loosestrife drove B. impatiens towards more native flower species To see if the B. impatiens workers were still visiting purple loosestrife despite its being all but removed from the site, the visits to purple loosestrife in 2016 and 2017 were examined. The successful removal of the purple loosestrife led to much fewer visits to purple loosestrife by B. impatiens workers (Figure 5.9). The total number of visits to purple loosestrife by B. impatiens workers dropped from almost 2500 in 2016 to less than 100 in 2017. These data show that 96% removal of the purple loosestrife was enough to almost remove it as a floral resource for B. impatiens workers completely in 2017. Comparatively, two flower species (Canada goldenrod and white meadowsweet) showed a strong increase in 2017, more than doubling the number of visits seen in 2016.

Figure 5.9: Purple loosestrife and four native flower visits by B. impatiens workers at Breakneck for 2016 and 2017.

With purple loosestrife removed as a floral resource and with no significant decrease in the B. impatiens worker population, the workers must then be getting their resources from other flowers to compensate, rather than a lack of B. impatiens workers leading to almost no purple loosestrife visits. During the field season with intact purple loosestrife (2016), B. impatiens workers visited 26 different native wildflower species to varying degrees. In the next field season with almost all of the purple loosestrife removed, B. impatiens workers visited 30 different native wildflower species (Table 5.3). In addition, proportion of visits to some native wildflowers increased strongly when purple loosestrife was no longer readily

53 available as a floral resource. Examples of this include Canada goldenrod and white meadowsweet, as seen in Figure 5.9.

Because it was seen that B. impatiens workers were visiting more native flower species in 2017, workers and males from all bumblebee species present in 2016 and 2017 were examined to see if there were significant differences in the native flower species diversity that they were visiting (Table 5.4). For all workers and males, the data between years was strong, but not statistically significant in favor of more native flower species (workers: P=0.125; males: P=0.1875). This then did not support the second hypothesis that purple loosestrife removal would lead to higher native flower visitation by bumblebee species.

Table 5.4: Bumblebee species observed at Breakneck Hill Conservation Land in 2016 and 2017 and number of native flower species they were observed visiting. Worker and male data included. Total native flower species available is in brackets.

Bumblebee species Native flower species Native flower species visited in 2016 visited in 2017

B. impatiens Worker 26 [31] 30 [33] Male 13 [31] 14 [33] B. griseocollis Worker 5 [31] 8 [33] Male 6 [31] 5 [33] B. bimaculatus Worker 3 [31] 4 [33] Male 1 [31] 4 [33] B. vagans Worker 2 [31] 2 [33] Male 0 [31] 3 [33] B. perplexus Worker 2 [31] 4 [33] Male 0 [31] 4 [33] B. fervidus Worker 0 [31] 0 [33] Male 0 [31] 0 [33]

Based off the previous 2 figures and the associated Wilcoxon matched‐pairs analysis, it is clear that the total visited native flower diversity did not significantly change for workers or males of any species when comparing 2016 and 2017. However, it has been established in previous chapters (Chapter 4) that bumblebee species do not visit flower species equally, they have distinct flower species preferences. Examining visited native flower species diversity is therefore not the most accurate way to determine whether there are more native flower visits. Because of this, the overall observations of native vs. non‐ native visits for workers and males of all species was analyzed using a Chi‐squared test (Figures 5.10 and 5.11). The null hypothesis is that the proportions of native and non‐native flower visits are not significantly different between 2016 and 2017, and the expected values were the proportions of native and non‐native flower visits from 2016. For workers, B. impatiens and B. bimaculatus were significantly different between years (P<0.0001) and B. griseocollis, B. vagans, and B. perplexus were not significant. For males, B. impatiens, B. griseocollis, B. bimaculatus, and B. vagans were significantly different between years (P<0.0001). B. perplexus males were not analyzed because they were not present at Breakneck in both years. B. fervidus was not analyzed for either workers or males because individuals

54 were not present at the site in both years. Based off Figure 5.10, both B. impatiens and B. bimaculatus workers showed a significant decrease in non‐native flower visits and a corresponding significant increase in native flower visits. While B. bimaculatus workers did not show a strong preference for purple loosestrife (Chapter 4), B. impatiens workers did, with purple loosestrife being their top preference when it was available to them (Chapter 4). Even though there was no significant difference in native flower species diversity with B. impatiens workers, there was a significant increase in overall native flower visits, supporting the second hypothesis. On the male side of the analysis, all species significantly decreased in non‐native flower visits and increased in native flower visits. In 2016, purple loosestrife was the top flower preference for B. griseocollis, B. bimaculatus, and B. vagans. The removal of their preferred floral resource in 2017 led to new top flower preferences that were native, such as spreading dogbane and wild bergamot. This also supported the second hypothesis that purple loosestrife removal would lead to higher native flower visitation.

Figure 5.10: Bumblebee species worker (left) and male (right) native vs. non‐native flower preferences at Breakneck 2016 starting at the start of purple loosestrife bloom. Proportion of observations on native flowers vs. proportion of observations on non‐native flowers. Values range from ‐1 (only non‐native flower visits) to 1 (only native flower visits). For Breakneck 2016 worker top flower preferences: B. impatiens, purple loosestrife (58%); B. griseocollis, common milkweed (54%); B. bimaculatus, wild bergamot (33%); B. vagans, red clover (27%). For Breakneck 2016 male top flower preferences: B. impatiens, Canada goldenrod (51%); B. griseocollis, purple loosestrife (51%); B. bimaculatus, purple loosestrife (57%); B. vagans, purple loosestrife (100%).

Figure 5.11: Bumblebee species worker (left) and male (right) native vs. non‐native flower preferences at Breakneck 2017 starting at the start of purple loosestrife bloom. Proportion of observations on native flowers vs. proportion of observations on non‐native flowers. Values range from ‐1 (only non‐native flower visits) to 1 (only native flower visits). For Breakneck 2017 worker top flower preferences: B.

55 impatiens, Canada goldenrod (38%); B. griseocollis, common milkweed (51%); B. bimaculatus, cow vetch (41%); B. vagans, cow vetch and red clover (17% each); B. perplexus, wild bergamot (100%). For Breakneck 2017 male top flower preferences: B. impatiens, wrinkleleaved goldenrod (31%); B. griseocollis, spreading dogbane (40%); B. bimaculatus, wild bergamot (77%); B. vagans, Purple Joe pye weed, tall goldenrod and lance leaved aster (33% each); B. perplexus, northern bush honeysuckle (50%).

In order to compare native and non‐native flower preferences at Breakneck to a field site that has no purple loosestrife, Wachusett was chosen as that representative. While at a higher elevation than Breakneck, Wachusett had a similar flower species composition compared to Breakneck, with the main exception of the purple loosestrife present at Breakneck. Figure 5.12 below shows the differences between Wachusett and Breakneck with purple loosestrife present. While some species like B. perplexus and B. bimaculatus show similar proportions between the two sites, B. impatiens and B. vagans at Breakneck 2016 both show a preference for non‐native flower species, but show a preference for native flower species at Wachusett 2016 (Figure 5.12). When purple loosestrife is absent however (Breakneck 2017), B. impatiens had a similar native flower preference to Wachusett 2016.

To compare B. impatiens native flower preferences of Breakneck of both years to the field site with no purple loosestrife (Wachusett), the overall observations of native vs. non‐native visits for workers of all species was analyzed using a Chi‐squared test (Figures 5.12 and 5.13). The null hypothesis is that the proportions of native and non‐native flower visits are not significantly different between Breakneck and Wachusett, and the expected values were the proportions of native and non‐native flower visits from Wachusett 2016. B. fervidus and B. ternarius were not analyzed for either workers or males because individuals were not present at both sites. When comparing Breakneck 2016 to Wachusett 2016, B. impatiens worker native flower preferences were significantly different (P<0.0001), with the highest flower visitation for purple loosestrife (non‐native) at Breakneck vs. Canada goldenrod (native) at Wachusett. When Breakneck with purple loosestrife removed (2017) and Wachusett 2016, B. impatiens workers had a significant difference again (P=0.0068), but the native proportion was 0.77 at Breakneck compared to 0.64 at Wachusett showing they visited significantly more native flowers.

Figure 5.12: Bumblebee species worker native vs. non‐native flower preferences at Wachusett 2016 (left) and Breakneck 2016 (right) starting at the start of purple loosestrife bloom. Proportion of observations on native flowers vs. proportion of observations on non‐native flowers. Values range from ‐1 (only non‐native flower visits) to 1 (only native flower visits). For Wachusett 2016 worker top flower preferences: B. impatiens, Canada goldenrod (28%); B. griseocollis, common milkweed (91%); B. bimaculatus, cow vetch (31%); B. vagans, common milkweed (64%); B. perplexus, common milkweed

56

(67%). For Breakneck 2016 worker top flower preferences: B. impatiens, purple loosestrife (58%); B. griseocollis, common milkweed (54%); B. bimaculatus, wild bergamot (33%); B. vagans, red clover (27%).

Figure 5.13: Bumblebee species worker native vs. non‐native flower preferences at Wachusett 2016 (left) and Breakneck 2017 (right) starting at the start of purple loosestrife bloom. Proportion of observations on native flowers vs. proportion of observations on non‐native flowers. Values range from ‐1 (only non‐native flower visits) to 1 (only native flower visits). For Wachusett 2016 worker top flower preferences: B. impatiens, Canada goldenrod (28%); B. griseocollis, common milkweed (91%); B. bimaculatus, cow vetch (31%); B. vagans, common milkweed (64%); B. perplexus, common milkweed (67%). For Breakneck 2017 worker top flower preferences: B. impatiens, Canada goldenrod (38%); B. griseocollis, common milkweed (51%); B. bimaculatus, cow vetch (41%); B. vagans, cow vetch and red clover (17% each); B. perplexus, wild bergamot (100%). Discussion The variation between bumblebee species regarding exotic flowers in Figure 5.2 shows support for the competition imbalance hypothesis. In order for the competition imbalance hypothesis to exist, there has to be some difference (physically or behaviorally) between bumblebee species in order to end with differential population trends. If all bumblebee species treated an exotic flower introduction in the exact same manner (all exploiting or all ignoring), then there would be no basis upon which the exotic flower would alter competition among bumblebee species in the community to the point where the competition become imbalanced in favor of a select few species. B. impatiens workers’ strong preference for purple loosestrife (57%), an exotic flower, coupled with their complete dominance at the field site (>90% of workers seen) also showed support for the competition imbalance hypothesis, and led to the manipulation experiment where the purple loosestrife was removed (Figures 5.2 and 5.3).

The results of the purple loosestrife removal did not significantly impact B. impatiens populations, either for workers or males. The relative abundance for both B. impatiens workers and males showed no significant decrease. The first hypothesis, that the purple loosestrife removal would negatively impact B. impatiens populations and show a lower relative abundance, is therefore not supported. One reason could be the presence of other flowers during what would have been the purple loosestrife bloom such as goldenrod species that B. impatiens was willing to visit but preferred less to purple loosestrife. In the absence of the purple loosestrife, B. impatiens appeared to simply switch over to the goldenrods without a delay. One interesting note is that B. impatiens workers also visited white meadowsweet much more in 2017 than 2016, a flower that was visited exclusively for pollen collection (Figure 5.9). One possibility is that the workers needed another pollen resource in 2017 in order to compensate from

57 the lack of purple loosestrife present. While purple loosestrife was visited for nectar collection, passive collection of pollen while visiting the large amounts of purple loosestrife present could have negated the need to seek out another pollen collection flower.

The second part of the first hypothesis predicted that with the purple loosestrife removal, there would be a higher Shannon’s diversity index for bumblebee species, and would also be demonstrated in higher relative abundances for non B. impatiens species. This however was not the case. The Shannon’s diversity index was not significantly different from 2016, and the relative abundance of non‐B. impatiens workers and males was not significantly different as well. The second part of the hypothesis is then not supported by the results. Given the bumblebee’s annual cycle, it is entirely possible that due to years of B. impatiens dominance of the site, there was no way for the other bumblebee species to gain a foothold despite the lack of purple loosestrife as a floral resource. However, more years of field surveys with loosestrife removal would be needed to see if there is a lag in bumblebee relative abundance changes.

While the other bumblebee species do not appear to have improved by the purple loosestrife removal, the native flowers did appear to benefit. Without purple loosestrife as a strong floral resource, B. impatiens workers visited 5 more native flower species than they had the year before, and flowers like Canada goldenrod and white meadowsweet showed much higher visit proportions. More visits to these native flowers increases their chance of successful pollination, leading to an overall improvement. While the native flower species diversity did not significantly change year to year with any species’ workers or males, the overall native flower visits did significantly increase for several species, including B. impatiens workers and B. griseocollis males. This increase in native flower visitation for B. impatiens workers also made the proportion visited more similar to a field site without purple loosestrife. The second hypothesis that removing the purple loosestrife would lead to greater visitation of native flowers by B. impatiens was therefore supported by the results.

One potential confounding factor to this conclusion is the presence of a severe drought in 2016. With the easing of the drought in 2017, the more readily available water could have enabled more native flowers to bloom or to produce more flowers. This would then increase their chances of getting pollinated by B. impatiens workers in 2017. Because of this potential confounding factor, more field seasons in both drought and non‐drought seasons will be needed to determine whether the increase in native flower species visitation was simply due to more native flower species being available.

In future field seasons, one way to more accurately determine native flower pollination success is to check their seed set in field seasons both before and after purple loosestrife removal. This would help show definitively whether the lack of purple loosestrife is allowing more successful pollination in native flowers, and has been used as a form of analysis in multiple studies (Flanagan et al. 2009, Flanagan et al. 2010, 2011). The greater the seed set, the more successful the pollination is. In addition, more years of data for both before and after purple loosestrife removal would be useful. Having more years would create a more robust study, and allow for “fluke” years to happen without skewing the overall data, such as the drought in 2016. Having more years of study after the purple loosestrife removal would also allow researchers to see if after years of lacking purple loosestrife as a floral resource the B. impatiens populations finally begin to be affected. It may be that it will take several annual cycles for other Bombus species to be able to compete effectively against B. impatiens numbers at the field site.

58

Purple loosestrife removal (and potentially other noxious invasive flower removal) can help improve habitat for native flowers in the immediate area that are competing for pollinators, and could then improve habitat for at‐risk bumblebee species that utilize native flower species. More work needs to be done on studying the long‐term effects of non‐native flower species on bumblebee‐plant communities, to determine which non‐native species are having a negative impact on the community overall by causing some bumblebee species to decline through competition.

59

Chapter 6: Conclusions The overall goals of this thesis were to examine bumblebee species diversity in Massachusetts, and determine if there were species specific flower preferences between bumblebee species. Using intensive field surveying over the course of 3 years (Chapter 2), it was found that there are species specific flower preferences, and these preferences maybe be driven by behavioral differences rather than physical ones (Chapter 4). It was also found that many bumblebee species in Massachusetts are in decline compared to what historical data is available (Chapter 3). Finally, a manipulative experiment at a field site removed almost all the purple loosestrife (an aggressive non‐native flower) from the site to see if there were any negative population impacts on the extremely abundant bumblebee species that had been shown to have a preference for it (Chapter 5). The results from this thesis have the potential to push studies of bumblebee resource partitioning into an entirely different direction. Up until now, scientists have been focusing on morphological differences between bumblebee species such as tongue length, and selecting one species from each group to compare. However, the results from this thesis show that there are species specific flower preferences, both between and within different morphological groups. This shows that the entire community has to be taken into account when looking at resource partitioning, rather than assigning a “representative” short tongued or long tongued bee, because the results show that different bumblebee species have different flower preferences (different resource partitioning) so one species cannot stand for all. In addition, the species‐specific flower preferences appear to be behaviorally based, not physical differences as a few researchers have suggested. Rather than focusing on phylogenetic or temporal differences between the bumblebee species to account for any preference difference, the viewpoint needs to shift towards the flowers and their differences. Questions like “are there color/shape similarities between flowers preferred by species X?” should be asked. The key is that if the bumblebees are behaving differently between species that are physically similar, there is something about the flowers that is causing the behavioral change, such as color, shape, sugar content, headspace volatiles, or any chemicals in the nectar. This opens up a novel avenue of bumblebee resource partitioning research. As a final implication, the evidence showing unequal preferences for non‐native flowers between bumblebee species and the potential dominance of one bumblebee species at a field site due to a preference for and a large amount of a non‐native flower species (purple loosestrife) can also provide some merit to the supposition that non‐native plants can indirectly cause the declines of bumblebee species, through giving those species that prefer them additional floral resources. This would negatively impact the bumblebee species who prefer native flowers through “pulling” the bumblebee species that prefer non‐natives off the natives, causing lower seed set in the native flowers which leads to fewer future resources for the bumblebees that prefer natives. Having more non‐native flowers available at a site also gives the resource advantage to the bumblebee species that prefer them, leading them to be able to outcompete other species on less‐preferred flowers (native) due to higher numbers. While there is no evidence that non‐native flowers could directly kill any bumblebee species, it is not outrageous to consider that indirectly, they can throw off the balance of the pollinator network within a community and lead species that prefer them to extreme high proportions, and those that don’t to declining and extreme lows. There are several different directions future field studies or experiments could take, given the cumulative results from the field surveys described in the previous chapters. The first would be to continue to survey the same field sites mentioned in this thesis for several more years. While the data were consistent year to year, more data means a more accurate picture of the bumblebee communities.

60

In addition, more field sites should be added. What was done in the thesis is a very small sampling of Massachusetts. Regions such as Cape Cod, western MA, and northeastern MA should have at least one field site in order to gain the most accurate picture of bumblebee species diversity and relative abundance throughout the state. There is a lack of baseline data to show bumblebee species status in Massachusetts, and having thorough surveys of the entire state will lead to the best understanding of which species have been locally extirpated, which species appear to be in most need of assistance, and which need no intervention at all. If the thorough surveys continue for several years, more information will become available such as rates of decline, so it can be seen which species are declining the fastest. It would also show if intervention was working, if a declining species received assistance such as habitat protection and showed a corresponding decrease in decline rate. The other direction that future studies could go into is the specifics of the species‐specific flower preferences. The results in the thesis established evidence that they existed, but there is no evidence as to why or how that caused the results. The most logical reasoning is that floral odor or nectar content has something to do with the behavioral differences. Unfortunately, there is little to no research on flower volatiles, and even less on nectar chemical components. Therefore, it would make the most sense to run studies on common wildflowers that bumblebee species forage on, and see if there are chemical similarities in flowers preferred by a certain bumblebee species. These studies could also show if there are chemical similarities to flowers that deter otherwise “generalist” bumblebee species, like B. impatiens avoidance of common milkweed. GC‐MS analysis on flower headspace volatiles could be carried out in the field thanks to portable machines, and nectar samples could be brought back to a lab for LC‐MS analysis.

61

References Ahrné, K., J. Bengtsson, and T. Elmqvist. 2009. Bumble Bees (Bombus spp) along a Gradient of Increasing Urbanization. PLoS ONE 4:e5574.

Alarcón, R. 2008. Year‐to‐year variation in the topology of a plant–pollinator interaction network. Oikos 117:1796‐1807.

Albrecht, M., M. R. Ramis, and A. Traveset. 2016. Pollinator‐mediated impacts of alien invasive plants on the pollination of native plants: the role of spatial scale and distinct behaviour among pollinator guilds. Biological Invasions 18:1801‐1812.

Arbulo, N., E. Santos, S. Salvarrey, and C. Invernizzi. 2011. Proboscis length and resource utilization in two Uruguayan bumblebees: Bombus atratus Franklin and Bombus bellicosus Smith (Hymenoptera: Apidae). Neotropical entomology 40:72‐77.

Arce, A. N., T. I. David, E. L. Randall, A. Ramos Rodrigues, T. J. Colgan, Y. Wurm, and R. J. Gill. 2017. Impact of controlled neonicotinoid exposure on bumblebees in a realistic field setting. Journal of Applied Ecology 54:1199‐1208.

Bais, H. P., R. Vepachedu, S. Gilroy, R. M. Callaway, and J. M. Vivanco. 2003. Allelopathy and Exotic Plant Invasion: From Molecules and Genes to Species Interactions. Science 301:1377‐1380.

Balfour, N. J., M. Garbuzov, and F. L. W. Ratnieks. 2013. Longer tongues and swifter handling: why do more bumble bees (Bombus spp.) than honey bees (Apis mellifera) forage on lavender (Lavandula spp.)? Ecological Entomology 38:323‐329.

Barlow, S. E., G. A. Wright, C. Ma, M. Barberis, I. W. Farrell, E. C. Marr, A. Brankin, B. M. Pavlik, and P. C. Stevenson. 2017. Distasteful Nectar Deters Floral Robbery. Current Biology 27:2552‐2558.e2553.

Barrow, D. A. 1984. Size‐related selection of food plants by bumblebees. Ecological Entomology 9:369‐ 373.

Barthell, J. F., J. M. Randall, R. W. Thorp, and A. M. Wenner. 2001. Promotion of Seed Set in Yellow Star‐ Thistle by Honey Bees: Evidence of an Invasive Mutualism. Ecological Applications 11:1870‐ 1883.

Bartomeus, I., J. S. Ascher, D. Wagner, B. N. Danforth, S. Colla, S. Kornbluth, and R. Winfree. 2011. Climate‐associated phenological advances in bee pollinators and bee‐pollinated plants. Proceedings of the National Academy of Sciences 108:20645‐20649.

Baskin, Y. 1998. Winners and Losers in a Changing World. BioScience 48:788‐792.

Battin, J. 2004. When Good Animals Love Bad Habitats: Ecological Traps and the Conservation of Animal Populations Cuando Animales Buenos Aman a Hábitats Malos: Trampas Ecológicas y la Conservación de Poblaciones Animales. Conservation Biology 18:1482‐1491.

62

Biesmeijer, J. C., S. P. M. Roberts, M. Reemer, R. Ohlemüller, M. Edwards, T. Peeters, A. P. Schaffers, S. G. Potts, R. Kleukers, C. D. Thomas, J. Settele, and W. E. Kunin. 2006. Parallel Declines in Pollinators and Insect‐Pollinated Plants in Britain and the Netherlands. Science 313:351‐354.

Blüthgen, N. 2010. Why network analysis is often disconnected from community ecology: A critique and an ecologist's guide.

Blüthgen, N., F. Menzel, and N. Blüthgen. 2006. Measuring specialization in species interaction networks. BMC Ecology 6:9.

Bonnington, C., K. J. Gaston, and K. L. Evans. 2015. Ecological traps and behavioural adjustments of urban songbirds to fine‐scale spatial variation in predator activity. Animal Conservation 18:529‐ 538.

Botías, C., A. David, J. Horwood, A. Abdul‐Sada, E. Nicholls, E. Hill, and D. Goulson. 2015. Neonicotinoid Residues in Wildflowers, a Potential Route of Chronic Exposure for Bees. Environmental Science & Technology 49:12731‐12740.

Brown, B. J. and R. J. Mitchell. 2001. Competition for pollination: effects of pollen of an invasive plant on seed set of a native congener. Oecologia 129:43‐49.

Brown, B. J., R. J. Mitchell, and S. A. Graham. 2002. COMPETITION FOR POLLINATION BETWEEN AN (PURPLE LOOSESTRIVE) AND A NATIVE CONGENER. Ecology 83:2328‐2336.

Bruce, A. R. and R. L. Hutto. 2007. Is Selectively Harvested Forest an Ecological Trap for Olive‐Sided Flycatchers? The Condor 109:109‐121.

Burkle, L. A. and J. B. Runyon. 2016. Drought and leaf herbivory influence floral volatiles and pollinator attraction. Global Change Biology 22:1644‐1654.

Callaway, R. M. and J. L. Maron. 2006. What have exotic plant invasions taught us over the past 20 years? Trends in Ecology & Evolution 21:369‐374.

Callaway, R. M., G. C. Thelen, A. Rodriguez, and W. E. Holben. 2004. Soil biota and exotic plant invasion. Nature 427:731.

Cameron, S. A., H. C. Lim, J. D. Lozier, M. A. Duennes, and R. Thorp. 2016. Test of the invasive pathogen hypothesis of bumble bee decline in North America. Proceedings of the National Academy of Sciences 113:4386‐4391.

Cameron, S. A., J. D. Lozier, J. P. Strange, J. B. Koch, N. Cordes, L. F. Solter, and T. L. Griswold. 2011. Patterns of widespread decline in North American bumble bees. Proceedings of the National Academy of Sciences 108:662‐667.

Carvell, C., W. R. Meek, R. F. Pywell, D. Goulson, and M. Nowakowski. 2007. Comparing the Efficacy of Agri‐Environment Schemes to Enhance Bumble Bee Abundance and Diversity on Arable Field Margins. Journal of Applied Ecology 44:29‐40.

63

Colla, S. R. and L. Packer. 2008. Evidence for decline in eastern North American bumblebees (Hymenoptera: Apidae), with special focus on Bombus affinis Cresson. Biodiversity and Conservation 17:1379.

Cox, P. A. and T. Elmqvist. 2000. Pollinator Extinction in the Pacific Islands. Conservation Biology 14:1237‐1239.

Daszak, P., A. A. Cunningham, and A. D. Hyatt. 2000. Emerging Infectious Diseases of Wildlife‐‐ Threats to Biodiversity and Human Health. Science 287:443‐449.

David, A., C. Botías, A. Abdul‐Sada, E. Nicholls, E. L. Rotheray, E. M. Hill, and D. Goulson. 2016. Widespread contamination of wildflower and bee‐collected pollen with complex mixtures of neonicotinoids and fungicides commonly applied to crops. Environment International 88:169‐ 178.

Dicks, L. V., S. A. Corbet, and R. F. Pywell. 2002. Compartmentalization in Plant‐Insect Flower Visitor Webs. Journal of Animal Ecology 71:32‐43.

Dormann, C. 2017. Package ‘bipartite’.

Dramstad, W. and G. Fry. 1995. Foraging activity of bumblebees (Bombus) in relation to flower resources on arable land. Agriculture, Ecosystems & Environment 53:123‐135.

Ehrenfeld, J. G. 2003. Effects of Exotic Plant Invasions on Soil Nutrient Cycling Processes. Ecosystems 6:503‐523.

Elias, M. A. S., F. J. A. Borges, L. L. Bergamini, E. V. Franceschinelli, and E. R. Sujii. 2017. Climate change threatens pollination services in tomato crops in Brazil. Agriculture, Ecosystems & Environment 239:257‐264.

F. Dormann, C., J. Fründ, N. Blüthgen, and B. Gruber. 2009. Indices, Graphs and Null Models: Analyzing Bipartite Ecological Networks.

Feltham, H., K. Park, and D. Goulson. 2014. Field realistic doses of pesticide imidacloprid reduce bumblebee pollen foraging efficiency. Ecotoxicology 23:317‐323.

Flanagan, R. J., R. J. Mitchell, and J. D. Karron. 2010. Increased relative abundance of an invasive competitor for pollination, Lythrum salicaria, reduces seed number in Mimulus ringens. Oecologia 164:445‐454.

Flanagan, R. J., R. J. Mitchell, and J. D. Karron. 2011. Effects of multiple competitors for pollination on bumblebee foraging patterns and Mimulus ringens reproductive success. Oikos 120:200‐207.

Flanagan, R. J., R. J. Mitchell, D. Knutowski, and J. D. Karron. 2009. Interspecific pollinator movements reduce pollen deposition and seed production in Mimulus ringens (Phrymaceae). American Journal of Botany 96:809‐815.

Forup, M. L. and J. Memmott. 2005. The relationship between the abundances of bumblebees and honeybees in a native habitat. Ecological Entomology 30:47‐57.

64

Fründ, J., K. E. Linsenmair, and N. Blüthgen. 2010. Pollinator diversity and specialization in relation to flower diversity. Oikos 119:1581‐1590.

G. Ehrenfeld, J., P. Kourtev, and A. Weize Huang. 2001. Changes in Soil Functions Following Invasions of Exotic Understory Plants in Deciduous Forests.

Gallardo, B., M. Clavero, M. I. Sánchez, and M. Vilà. 2016. Global ecological impacts of invasive species in aquatic ecosystems. Global Change Biology 22:151‐163.

Garbuzov, M. and F. L. W. Ratnieks. 2014. Quantifying variation among garden plants in attractiveness to bees and other flower‐visiting insects. Functional Ecology 28:364‐374.

Gillespie, S. 2010. Factors affecting parasite prevalence among wild bumblebees. Ecological Entomology 35:737‐747.

Gomez‐Moracho, T., P. Heeb, and M. Lihoreau. 2017. Effects of parasites and pathogens on bee cognition. Ecological Entomology 42:51‐64.

Goulson, D. 2003. Effects of Introduced Bees on Native Ecosystems. Annual Review of Ecology, Evolution, and Systematics 34:1‐26

Goulson, D. 2013. REVIEW: An overview of the environmental risks posed by neonicotinoid insecticides. Journal of Applied Ecology 50:977‐987.

Goulson, D. 2015. Bee declines driven by combined stress from parasites, pesticides, and lack of flowers. Science 347:6229.

Goulson, D. and B. Darvill. 2004. Niche overlap and diet breadth in bumblebees; are rare species more specialized in their choice of flowers? Apidologie 35:55‐63.

Goulson, D., M. E. Hanley, B. Darvill, J. S. Ellis, and M. E. Knight. 2005. Causes of rarity in bumblebees. Biological Conservation 122:1‐8.

Goulson, D. and W. O. H. Hughes. 2015. Mitigating the anthropogenic spread of bee parasites to protect wild pollinators. Biological Conservation 191:10‐19.

Goulson, D., G. C. Lye, and B. Darvill. 2008a. Decline and Conservation of Bumble Bees. Annual Review of Entomology 53:191‐208.

Goulson, D., G. C. Lye, and B. Darvill. 2008b. Diet breadth, coexistence and rarity in bumblebees. Biodiversity and Conservation 17:3269‐3288.

Goulson, D. and E. Nicholls. 2016. The canary in the coalmine; bee declines as an indicator of environmental health. Science Progress 99:312+.

Goulson, D. and K. R. Sparrow. 2009. Evidence for competition between honeybees and bumblebees; effects on bumblebee worker size. Journal of Insect Conservation 13:177‐181.

65

Grabas, G. and T. Laverty. 1999. The effect of purple loosestrife (Lythrum salicaria L.; Lythraceae) on the pollination and reproductive success of sympatric co‐flowering wetland plants. Écoscience 6:230‐242.

Graham, L. and K. N. Jones. 1996. Resource Partitioning and Per‐flower Foraging Efficiency in Two Bumble Bee Species. The American Midland Naturalist 136:401‐406.

Graystock, P., K. Yates, B. Darvill, D. Goulson, and W. O. H. Hughes. 2013. Emerging dangers: Deadly effects of an emergent parasite in a new pollinator host. Journal of Invertebrate Pathology 114:114‐119.

Grixti, J. C., L. T. Wong, S. A. Cameron, and C. Favret. 2009. Decline of bumble bees (Bombus) in the North American Midwest. Biological Conservation 142:75‐84.

Grove, S., K. A. Haubensak, and I. M. Parker. 2012. Direct and indirect effects of allelopathy in the soil legacy of an exotic plant invasion. Plant Ecology 213:1869‐1882.

Harder, L. D. 1983. Flower Handling Efficiency of Bumble Bees: Morphological Aspects of Probing Time. Oecologia 57:274‐280.

Harder, L. D. 1985. Morphology as a predictor of flower choice by bumblebees. Ecology 66:198‐210.

Heinrich, B. 1976. Resource Partitioning Among Some Eusocial Insects: Bumblebees. Ecology 57:874‐ 889.

Hierro, J. L. and R. M. Callaway. 2003. Allelopathy and exotic plant invasion. Plant and Soil 256:29‐39.

Inouye, D. W. 1978. Resource Partitioning in Bumblebees: Experimental Studies of Foraging Behavior. Ecology 59:672‐678.

Inouye, D. W. 1980. The Effect of Proboscis and Corolla Tube Lengths on Patterns and Rates of Flower Visitation by Bumblebees. Oecologia 45:197‐201.

Jakobsson, A. and B. Padrón. 2014. Does the invasive Lupinus polyphyllus increase pollinator visitation to a native herb through effects on pollinator population sizes? Oecologia 174:217‐226.

Javorek, S. K., K. E. Mackenzie, and S. P. Vander Kloet. 2002. Comparative Pollination Effectiveness Among Bees (Hymenoptera: Apoidea) on Lowbush Blueberry (Ericaceae: Vaccinium angustifolium). Annals of the Entomological Society of America 95:345‐351.

Johnson, R. A. 1986. Intraspecific resource partitioning in the bumble bees Bombus ternarius and B. pensylvanicus. Ecology 67:133‐138.

Kämper, W., P. K. Werner, A. Hilpert, C. Westphal, N. Blüthgen, T. Eltz, and S. D. Leonhardt. 2016. How landscape, pollen intake and pollen quality affect colony growth in Bombus terrestris. Landscape Ecology 31:2245‐2258.

Keane, R. M. and M. J. Crawley. 2002. Exotic plant invasions and the enemy release hypothesis. Trends in Ecology & Evolution 17:164‐170.

66

Kerr, J. T., A. Pindar, P. Galpern, L. Packer, S. G. Potts, S. M. Roberts, P. Rasmont, O. Schweiger, S. R. Colla, L. L. Richardson, D. L. Wagner, L. F. Gall, D. S. Sikes, and A. Pantoja. 2015. Climate change impacts on bumblebees converge across continents. Science 349:177‐180.

King, V. M. and R. D. Sargent. 2012. Presence of an invasive plant species alters pollinator visitation to a native. Biological Invasions 14:1809‐1818.

Klein, A.‐M., I. Steffan‐Dewenter, and T. Tscharntke. 2003. Fruit set of highland coffee increases with the diversity of pollinating bees. Proceedings of the Royal Society B: Biological Sciences 270:955‐ 961.

Kosior, A., W. Celary, P. Olejniczak, J. Fijał, W. Król, W. Solarz, and P. Płonka. 2007. The decline of the bumble bees and cuckoo bees (Hymenoptera: Apidae: Bombini) of Western and Central Europe. Oryx 41:79‐88.

Kriesell, L., A. Hilpert, and S. D. Leonhardt. 2017. Different but the same: bumblebee species collect pollen of different plant sources but similar amino acid profiles. Apidologie 48:102‐116.

Krupke, C. H., G. J. Hunt, B. D. Eitzer, G. Andino, and K. Given. 2012. Multiple Routes of Pesticide Exposure for Honey Bees Living Near Agricultural Fields. PLoS ONE 7:e29268.

Laverty, T. M. and L. D. Harder. 2012. THE BUMBLE BEES OF EASTERN CANADA. The Canadian Entomologist 120:965‐987.

Laverty, T. M. and R. C. Plowright. 1988. Flower handling by bumblebees: a comparison of specialists and generalists. Animal Behaviour 36:733‐740.

Lerman, S. B. and J. Milam. 2016. Bee Fauna and Floral Abundance Within Lawn‐Dominated Suburban Yards in Springfield, MA. Annals of the Entomological Society of America 109:713‐723.

Levine, J. M., M. Vilà, C. M. D. Antonio, J. S. Dukes, K. Grigulis, and S. Lavorel. 2003. Mechanisms underlying the impacts of exotic plant invasions. Proceedings of the Royal Society of London. Series B: Biological Sciences 270:775‐781.

Lye, G., K. Park, J. Osborne, J. Holland, and D. Goulson. 2009. Assessing the value of Rural Stewardship schemes for providing foraging resources and nesting habitat for bumblebee queens (Hymenoptera: Apidae). Biological Conservation 142:2023‐2032.

MacDougall, A. S. and R. Turkington. 2005. Are invasive species the drivers or passengers of change in degraded ecosystems? Ecology 86:42‐55.

Magrach, A., A. Holzschuh, I. Bartomeus, V. Riedinger, S. P. M. Roberts, M. Rundlöf, A. Vujić, J. B. Wickens, V. J. Wickens, R. Bommarco, J. P. González‐Varo, S. G. Potts, H. G. Smith, I. Steffan‐ Dewenter, and M. Vilà. 2018. Plant–pollinator networks in semi‐natural grasslands are resistant to the loss of pollinators during blooming of mass‐flowering crops. Ecography 41:62‐74.

Martins, A. C., D. P. Silva, P. De Marco, and G. A. R. Melo. 2015. Species conservation under future climate change: the case of Bombus bellicosus, a potentially threatened South American bumblebee species. Journal of Insect Conservation 19:33‐43.

67

McKinney, A. M. and K. Goodell. 2011. Plant–pollinator interactions between an invasive and native plant vary between sites with different flowering phenology. Plant Ecology 212:1025‐ 1035.

McKinney, M. L. and J. L. Lockwood. 1999. Biotic homogenization: a few winners replacing many losers in the next mass extinction. Trends in Ecology & Evolution 14:450‐453.

Meeus, I., M. J. F. Brown, D. C. De Graaf, and G. U. Y. Smagghe. 2011. Effects of Invasive Parasites on Bumble Bee Declines Efectos de Parásitos Invasores sobre Declinaciones de Abejorros. Conservation Biology 25:662‐671.

Memmott, J. 2004. Tolerance of pollination networks to species extinctions. Proc. R. Soc. London Sci. Ser. B 271:2605.

Miller‐Struttmann, N. E., J. C. Geib, J. D. Franklin, P. G. Kevan, R. M. Holdo, D. Ebert‐May, A. M. Lynn, J. A. Kettenbach, E. Hedrick, and C. Galen. 2015. Functional mismatch in a bumble bee pollination mutualism under climate change. Science 349:1541‐1544.

Mills, L. S., Soul, xe, M. E., and D. F. Doak. 1993. The Keystone‐Species Concept in Ecology and Conservation. BioScience 43:219‐224.

Mobley, M. 2016. Examining the Potential Threat of Pesticide and Pathogen Exposure on Wild Bumble Bees: Proposed Lethal and Sublethal Mechanisms Contributing to Pollinator Decline. Dissertation. Worcester Polytechnic Institute.

Mommaerts, V., S. Reynders, J. Boulet, L. Besard, G. Sterk, and G. Smagghe. 2009. Risk assessment for side‐effects of neonicotinoids against bumblebees with and without impairing foraging behavior. Ecotoxicology 19:207.

Morris, E. K., T. Caruso, F. Buscot, M. Fischer, C. Hancock, T. S. Maier, T. Meiners, C. Müller, E. Obermaier, D. Prati, S. A. Socher, I. Sonnemann, N. Wäschke, T. Wubet, S. Wurst, and M. C. Rillig. 2014. Choosing and using diversity indices: insights for ecological applications from the German Biodiversity Exploratories. Ecology and Evolution 4:3514‐3524.

Nazzi, F., S. P. Brown, D. Annoscia, F. Del Piccolo, G. Di Prisco, P. Varricchio, G. Della Vedova, F. Cattonaro, E. Caprio, and F. Pennacchio. 2012. Synergistic Parasite‐Pathogen Interactions Mediated by Host Immunity Can Drive the Collapse of Honeybee Colonies. PLOS Pathogens 8:e1002735.

Nicholls, E. and N. Hempel de Ibarra. 2017. Assessment of pollen rewards by foraging bees. Functional Ecology 31:76‐87.

Ogilvie, J. E., S. R. Griffin, Z. J. Gezon, B. D. Inouye, N. Underwood, D. W. Inouye, and R. E. Irwin. 2017. Interannual bumble bee abundance is driven by indirect climate effects on floral resource phenology. Ecology Letters 20:1507‐1515.

Orrock, J. L., H. P. Dutra, R. J. Marquis, and N. Barber. 2015. Apparent competition and native consumers exacerbate the strong competitive effect of an exotic plant species. Ecology 96:1052‐1061.

68

Osborne, J. L., A. P. Martin, C. R. Shortall, A. D. Todd, D. Goulson, M. E. Knight, R. J. Hale, and R. A. Sanderson. 2008. Quantifying and comparing bumblebee nest densities in gardens and countryside habitats. Journal of Applied Ecology 45:784‐792.

Papanikolaou, A. D., I. Kühn, M. Frenzel, and O. Schweiger. 2017. Semi‐natural habitats mitigate the effects of temperature rise on wild bees. Journal of Applied Ecology 54:527‐536.

Peat, J., J. Tucker, and D. Goulson. 2005. Does intraspecific size variation in bumblebees allow colonies to efficiently exploit different flowers? Ecological Entomology 30:176‐181.

Pellissier, L., J.‐N. Pradervand, P. H. Williams, G. Litsios, and D. Cherix. 2013. Phylogenetic relatedness and proboscis length contribute to structuring bumblebee communities in the extremes of abiotic and biotic gradients Assembly of bumblebee communities. Global ecology and biogeography 22:577‐585.

Piiroinen, S., C. Botías, E. Nicholls, and D. Goulson. 2016. No effect of low‐level chronic neonicotinoid exposure on bumblebee learning and fecundity. PeerJ 4:e1808.

Piiroinen, S. and D. Goulson. 2016. Chronic neonicotinoid pesticide exposure and parasite stress differentially affects learning in honeybees and bumblebees. Proceedings of the Royal Society B: Biological Sciences 283.

Pisa, L. W., V. Amaral‐Rogers, L. P. Belzunces, J. M. Bonmatin, C. A. Downs, D. Goulson, D. P. Kreutzweiser, C. Krupke, M. Liess, M. McField, C. A. Morrissey, D. A. Noome, J. Settele, N. Simon‐ Delso, J. D. Stark, J. P. Van der Sluijs, H. Van Dyck, and M. Wiemers. 2015. Effects of neonicotinoids and fipronil on non‐target invertebrates. Environmental Science and Pollution Research 22:68‐102.

Plath, O. 1927. Bumblebees, Their Life‐History, Habits, and Economic Importance. With a Detailed Account of the New England Species. Harvard, Cambridge, MA.

Ploquin, E. F., J. M. Herrera, and J. R. Obeso. 2013. Bumblebee community homogenization after uphill shifts in montane areas of northern Spain. Oecologia 173:1649‐1660.

Plowright, C. M. S. and R. C. Plowright. 1997. THE ADVANTAGE OF SHORT TONGUES IN BUMBLE BEES (BOMBUS) — ANALYSES OF SPECIES DISTRIBUTIONS ACCORDING TO FLOWER COROLLA DEPTH, AND OF WORKING SPEEDS ON WHITE CLOVER. The Canadian Entomologist 129:51‐59.

Plowright, R. C. and T. M. Laverty. 1984. The Ecology and Sociobiology of Bumble Bees. Annual Review of Entomology 29:175‐199.

Potts, S. G., J. C. Biesmeijer, C. Kremen, P. Neumann, O. Schweiger, and W. E. Kunin. 2010. Global pollinator declines: trends, impacts and drivers. Trends in Ecology & Evolution 25:345‐353.

Pyke, G. H. 1982. Local geographic distributions of bumblebees near Crested Butte, Colorado: competition and community structure. Ecology 63:555‐573.

Pyke, G. H., J. D. Thomson, D. W. Inouye, and T. J. Miller. 2016. Effects of climate change on phenologies and distributions of bumble bees and the plants they visit. Ecosphere 7:e01267‐n/a.

69

R Kells, A. and D. Goulson. 2003. Preferred nesting sites of bumblebee queens (Hymenoptera: Apidae) in agroecosystems in the UK.

Raine, N. E. and L. Chittka. 2005. Comparison of flower constancy and foraging performance in three bumblebee species (Hymenoptera : Apidae : Bombus). Entomologia generalis 28:81‐89.

Ranta, E. S. A., Ter, xc, I. S, and H. Lundberg. 1981. Phenological spread in flowering of bumblebee‐ pollinated plants. Annales Botanici Fennici 18:229‐236.

Rasmont, P. and P. Mersch. 1988. First estimation of faunistic drift by bumblebees of Belgium, (Hymenoptera: Apidae. Annales ‐ Societe Royale Zoologique de Belgique 118:141‐147.

Riaño Jiménez, D. and J. R. Cure. 2016. Acute lethal effect of the commercial formulation of the insecticides Imidacloprid, Spinosad y Thiocyclam hidrogenoxalate in Bombus atratus (Hymenoptera: Apidae) workers. 2016 64:9.

Ridenour, W. M. and R. M. Callaway. 2001. The relative importance of allelopathy in interference: the effects of an invasive weed on a native bunchgrass. Oecologia 126:444‐450.

Rodríguez‐Gironés, M. and L. Santamaría. 2010. How Foraging Behaviour and Resource Partitioning Can Drive the Evolution of Flowers and the Structure of Pollination Networks.

Rodríguez‐Gironés, M. A. and A. L. Llandres. 2008. Resource Competition Triggers the Co‐Evolution of Long Tongues and Deep Corolla Tubes. PLoS ONE 3:e2992.

Rosenkranz, P., P. Aumeier, and B. Ziegelmann. 2010. Biology and control of Varroa destructor. Journal of Invertebrate Pathology 103, Supplement:S96‐S119.

Russell, A. L., R. E. Golden, A. S. Leonard, and D. R. Papaj. 2016. Bees learn preferences for plant species that offer only pollen as a reward. Behavioral Ecology 27:731‐740.

Sanchez‐Bayo, F. and K. Goka. 2014. Pesticide Residues and Bees – A Risk Assessment. PLoS ONE 9:e94482.

Sárospataki, M., J. Novák, and V. Molnár. 2005. Assessing the Threatened Status of Bumble Bee Species (Hymenoptera: Apidae) in Hungary, Central Europe. Biodiversity & Conservation 14:2437‐2446.

Schlaepfer, M. A., M. C. Runge, and P. W. Sherman. 2002. Ecological and evolutionary traps. Trends in Ecology & Evolution 17:474‐480.

Schmid‐Hempel, R., M. Eckhardt, D. Goulson, D. Heinzmann, C. Lange, S. Plischuk, L. R. Escudero, R. Salathé, J. J. Scriven, and P. Schmid‐Hempel. 2014. The invasion of southern South America by imported bumblebees and associated parasites. Journal of Animal Ecology 83:823‐837.

Scriven, J. J., P. R. Whitehorn, D. Goulson, and M. C. Tinsley. 2016. Niche partitioning in a sympatric cryptic species complex. Ecology and Evolution 6:1328‐1339.

70

Sherley, R. B., K. Ludynia, B. M. Dyer, T. Lamont, A. B. Makhado, J.‐P. Roux, K. L. Scales, L. G. Underhill, and S. C. Votier. 2017. Metapopulation Tracking Juvenile Penguins Reveals an Ecosystem‐wide Ecological Trap. Current Biology 27:563‐568.

Sikora, A. and M. Kelm. 2012. Flower Preferences of the Wroclaw Botanical Garden Bumblebees (Bombus spp.). Journal of Apicultural Science 56:27‐36.

Skurski, T. C. 2014. Mechanisms underlying nonindigenous plant impacts: a review of recent experimental research. Invasive Plant Science and Management 7:432‐444.

Stanley, D. A. and N. E. Raine. 2016. Chronic exposure to a neonicotinoid pesticide alters the interactions between bumblebees and wild plants. Functional Ecology 30:1132‐1139.

Stout, J., C. and C. Morales, L. 2009. Ecological impacts of invasive alien species on bees. Apidologie 40:388‐409.

Stout, J. C., J. A. Allen, and D. Goulson. 2000. Nectar robbing, forager efficiency and seed set: Bumblebees foraging on the self incompatible plant Linaria vulgaris (Scrophulariaceae). Acta Oecologica 21:277‐283.

Szymanski, J., T. Smith, A. Horton, M. Parkin, L. Ragan, G. Masson, E. Olson, K. Gifford, and L. Hill. 2016. Rusty Patched Bumblebee (Bombus affinis) Species Status Assessment.in USFWS, editor.

Temeles, E. J., J. T. Newman, J. H. Newman, S. Y. Cho, A. R. Mazzotta, and W. J. Kress. 2016. Pollinator Competition as a Driver of Floral Divergence: An Experimental Test. PLoS ONE 11:e0146431.

Thomson, D. 2004. COMPETITIVE INTERACTIONS BETWEEN THE INVASIVE EUROPEAN HONEY BEE AND NATIVE BUMBLE BEES. Ecology 85:458‐470.

Thomson, D. M. 2016. Local bumble bee decline linked to recovery of honey bees, drought effects on floral resources. Ecology Letters 19:1247‐1255.

Tomizawa, M. and J. E. Casida. 2005. Neonicotinoid Insecticide Toxicology: Mechanisms of Selective Action. Annual Review of Pharmacology and Toxicology 45:247‐268.

Traveset, A. and D. M. Richardson. 2006. Biological invasions as disruptors of plant reproductive mutualisms. Trends in Ecology & Evolution 21:208‐216.

Tur, C., J. M. Olesen, and A. Traveset. 2015. Increasing modularity when downscaling networks from species to individuals. Oikos 124:581‐592.

Tur, C., B. Vigalondo, K. Trøjelsgaard, J. M. Olesen, and A. Traveset. 2014. Downscaling pollen–transport networks to the level of individuals. Journal of Animal Ecology 83:306‐317.

Turbelin, A. J., B. D. Malamud, and R. A. Francis. 2017. Mapping the global state of invasive alien species: patterns of invasion and policy responses. Global ecology and biogeography 26:78‐92.

USFWS. 2017. Environmental Conservation Online System: Listed Animals.

71

Velthuis, H. H. W. and A. v. Doorn. 2006. A century of advances in bumblebee domestication and the economic and environmental aspects of its commercialization for pollination. Apidologie 37:421‐ 451.

Vilxe, M., I. Bartomeus, A. C. Dietzsch, T. Petanidou, I. Steffan‐Dewenter, J. C. Stout, and T. Tscheulin. 2009. Invasive Plant Integration into Native Plant–Pollinator Networks Across Europe. Proceedings: Biological Sciences 276:3887‐3893.

Weldon, A. J. and N. M. Haddad. 2005. THE EFFECTS OF PATCH SHAPE ON INDIGO BUNTINGS: EVIDENCE FOR AN ECOLOGICAL TRAP. Ecology 86:1422‐1431.

Williams, P. 2005. Does specialization explain rarity and decline among British bumblebees? A response to Goulson et al. Biological Conservation 122:33‐43.

Williams, P., S. Colla, and Z. Xie. 2009. Bumblebee Vulnerability: Common Correlates of Winners and Losers across Three Continents Vulnerabilidad de Abejorros: Correlaciones Comunes de Ganadores y Perdedores en Tres Continentes. Conservation Biology 23:931‐940.

Williams, P., H. and J. Osborne, L. 2009. Bumblebee vulnerability and conservation world‐wide. Apidologie 40:367‐387.

Wolfe, B. E. and J. N. Klironomos. 2005. Breaking New Ground: Soil Communities and Exotic Plant Invasion. BioScience 55:477‐487.

Wood, T. J. and D. Goulson. 2017. The environmental risks of neonicotinoid pesticides: a review of the evidence post 2013. Environmental Science and Pollution Research 24:17285‐17325.

Yale. 2017. Yale Peabody Museum.

Zimmerman, M. and J. M. Pleasants. 1982. Competition among Pollinators: Quantification of Available Resources. Oikos 38:381‐383.

72

Appendix A: List and visuals of Bumblebee (Bombus) Species in Massachusetts

Short tongue

 B. affinis (not seen)

 B. ternarius (worker: left; male; right)

 B. terricola (male)

Medium tongue

73

 B. griseocollis (worker: left; male: right)

 B. impatiens (queen: left; male: right)

 B. perplexus (2 band morph worker: left; 3 band morph male: right)

74

Long tongue

 B. bimaculatus (worker: left; male: right)

 B. borealis (male)

75

 B. fervidus (male)

 B. pensylvanicus (not seen)

 B. vagans (worker)

76

Appendix B: List of bumblebee visited flower species in each site (2015‐ 2017) Seen at Seen at Seen at Seen at Common name Color Shape Breakneck Wachusett Heath Ashfield Yellow wild indigo (Baptisia Yellow Closed tube X tinctoria) Great blue lobelia (Lobelia Purple Closed tube X X syphilitica) Birdsfoot trefoil (Lotus Yellow Closed tube X corniculatus) Crown vetch (Securigera Purple Closed tube X varia) Cow vetch (Vicia Purple Closed tube X X X X cracca) English plantain (Plantago White N/A X X X lanceolata) Great burdock Pink Narrow tube X X X (Acrtium lappa) Spotted knapweed Purple Narrow tube X (Centaurea maculosa) Bull thistle (Cirsium Purple Narrow tube X X X vulgare) Creeping thistle (Cirsium Purple Narrow tube X X X arvense)

77

Pasture thistle (Cirsium Purple Narrow tube X pumlium) Wild bergamot (Monarda Purple Narrow tube X X fistulosa) Alsike clover (Trifolium White Narrow tube X X X X hybridum) Red clover (Trifolium Red Narrow tube X X X X pratense) Common chicory Blue No tube (Cichorium intybus) Virginia virgin's bower (Clematis White No tube X virginiana) Queen Anne's lace (Daucus White No tube X X X X carota) Tall white‐aster (Doellingeria White No tube X umbellata) American burnweed White No tube X (Erechtites hieraciifolius) Fleabane daisy (Erigeron White No tube X X X X strigosus) Grassleaved goldenrod Yellow No tube X X X X (Euthamia graminifolia) Common St. John's wort Yellow No tube X X X X (Hypericum perforatum) Tall lettuce (Lactuca Yellow No tube canadensis) Lady's thumb smartweed Pink No tube X (Persicaria maculosa)

78

Sulphur cinquefoil Yellow No tube X X X X (Pontentilla recta) Sour cherry White No tube X (Prunus cerasus) Tall buttercup (Ranunculus Yellow No tube X X X X acris) Multiflora rose white No tube X X (Rosa multiflora) Flowering raspberry Pink No tube X (Rubus odoratus) Wild red raspberry White No tube X X X X (Rubus idaeus) Common blackberry White No tube X X (Rubus allegheniensis) Carolina nightshade White No tube X (Solanum carolinense) Climbing nightshade Purple No tube X X (Solanum dulcamara) Canada goldenrod Yellow No tube X X X X (Solidago canadensis) Tall goldenrod (Solidago Yellow No tube X X X X altissima) Wrinkleleaved goldenrod Yellow No tube X X X X (Solidago rugosa) Early goldenrod (Solidago Yellow No tube X X X juncea) Calico aster (Symphyotrichu White No tube X X X X m lateriflorum)

79

New England aster (Symphyotrichu Purple No tube X X X m novae‐ angliae) Late purple aster Purple No tube X X (Symphyotrichu m patens) Lanceleaved aster White No tube X X (Symphyotrichu m lanceolatum) Purple‐stemmed aster Purple No tube X X X X (Symphyotrichu m puniceum) New York aster (Symphyotrichu Purple No tube X X m novi‐belgii) Dandelion (Taraxacum Yellow No tube X X X X officinale) Spreading dogbane (Apocynum White/pink Open tube X androsaemifoliu m) Common milkweed Pink Open tube X X X X (Asclepias syriaca) Swamp milkweed Pink Open tube X (Asclepias incarnata) Hedge false bindweed White/pink Open tube X X X (Calystegia sepium) Creeping bellflower Purple Open tube X (Campanula rapunculoides) Showy tick Pink Open tube X trefoil

80

(Desmodium canadense) Northern bush honeysuckle Yellow Open tube X (Diervilla lonicera) Boneset (Eupatorium White Open tube X X X perfoliatum) Purple Joe pye weed Pink Open tube X X X (Eutrochium purpureum) Split‐lipped hemp nettle Pink Open tube X (Galeopsis bifida) Amur honeysuckle White Open tube X (Lonicera maackii) Purple loosestrife Purple Open tube X (Lythrum salicaria) Alfalfa (Medicago Purple Open tube X sativa) Yellow sweet clover (Melilotus Yellow Open tube X officinalis) Spearmint (Mentha Purple Open tube X spicata) American wild mint (Mentha Purple Open tube X X canadensis) Allegheny monkeyflower Purple Open tube X (Mimulus ringens) Common evening primrose Yellow Open tube X X (Oenothera biennis)

81

Penstamon cultivar Pink Open tube X (Penstamon spp.) Common self‐ heal (Prunella Purple Open tube X X X vulgaris) White meadowsweet White Open tube X X X X (Spiraea alba) Hairy hedgenettle Purple Open tube X (Stachys hispida) Common comfrey Purple Open tube X (Symphytum officinale) Purple cultivar Purple Open tube X Unknown bush White Open tube X Blue vervain (Verbena Purple Open tube X hastata) Jewelweed (Impatiens Orange Spiked tube X X X X capensis) Toad flax Yellow Spiked tube X (Linaria vulgaris) Unknown tree Brown X

Appendix C: Screenshots of ArcGIS maps of field sites with transects drawn Breakneck Hill Conservation Land (Southborough, MA) for 2016

82

Breakneck Hill Conservation Land (Southborough, MA) for 2017

Crowningshield Conservation Area (Heath, MA)

83

Bullitt Reservation (Ashfield, MA)

Appendix D: Shannon’s diversity (H’) for all field sites/seasons

Site Early season Midseason Late season Overall Early June‐mid Mid July‐mid Mid August‐end July August of season Breakneck 2016 1.264741 0.120288 0.003904 0.265725 Wachusett 2016 1.058024 1.059509 0.124251 1.29434 Breakneck 2017 1.319359 0.104212 0.007117 0.30219 Wachusett 2017 0.925181 1.388614 0.277601 1.2959 Heath 2017 1.709992 1.697409 1.004242 1.75339 Ashfield 2017 1.192565 1.099771 0.43053 1.14525

84

Appendix E: B. vagans historical data high (>1000’) vs. low elevation (<1000’)

Appendix F: List of wildflowers visited by bumblebee species from 2015‐2017. *=indicates nectar robbing observations. Short tongued Medium tongued Long tongued Common name Color Shape bees visit bees visit bees visit Yellow wild indigo Yellow Closed tube X X (Baptisia tinctoria) Great blue lobelia (Lobelia Purple Closed tube X syphilitica) Birdsfoot trefoil (Lotus Yellow Closed tube X X corniculatus) Crown vetch Purple Closed tube X X (Securigera varia) Cow vetch (Vicia Purple Closed tube X* X X cracca) English plantain (Plantago White N/A X lanceolata) Great burdock Pink Narrow tube X X X (Acrtium lappa) Spotted knapweed (Centaurea Purple Narrow tube X X maculosa) Bull thistle Purple Narrow tube X X (Cirsium vulgare)

85

Creeping thistle Purple Narrow tube X X X (Cirsium arvense) Pasture thistle Purple Narrow tube X (Cirsium pumlium) Wild bergamot (Monarda Purple Narrow tube X X X fistulosa) Alsike clover (Trifolium White Narrow tube X X hybridum) Red clover (Trifolium Red Narrow tube X X X pratense) Common chicory (Cichorium Blue No tube X intybus) Virginia virgin's bower (Clematis White No tube X X virginiana) Queen Anne's lace White No tube X X (Daucus carota) Tall white‐aster (Doellingeria White No tube X X umbellata) American burnweed White No tube X (Erechtites hieraciifolius) Fleabane daisy (Erigeron White No tube X strigosus) Grassleaved goldenrod Yellow No tube X X X (Euthamia graminifolia) Common St. John's wort Yellow No tube X X X (Hypericum perforatum) Tall lettuce (Lactuca Yellow No tube X canadensis) Lady's thumb smartweed Pink No tube X (Persicaria maculosa)

86

Sulphur cinquefoil Yellow No tube X (Pontentilla recta) Sour cherry White No tube X (Prunus cerasus) Tall buttercup Yellow No tube X (Ranunculus acris) Multiflora rose white No tube X X (Rosa multiflora) Flowering raspberry (Rubus Pink No tube X odoratus) Wild red raspberry White No tube X X X (Rubus idaeus) Common blackberry (Rubus White No tube X allegheniensis) Carolina nightshade White No tube X (Solanum carolinense) Climbing nightshade Purple No tube X (Solanum dulcamara) Canada goldenrod (Solidago Yellow No tube X X X canadensis) Tall goldenrod (Solidago Yellow No tube X X X altissima) Wrinkleleaved goldenrod Yellow No tube X X X (Solidago rugosa) Early goldenrod Yellow No tube X X (Solidago juncea) Calico aster (Symphyotrichum White No tube X X X lateriflorum) New England aster (Symphyotrichum Purple No tube X X novae‐angliae) Late purple aster (Symphyotrichum Purple No tube X patens) Lanceleaved aster (Symphyotrichum White No tube X X lanceolatum)

87

Purple‐stemmed aster Purple No tube X (Symphyotrichum puniceum) New York aster (Symphyotrichum Purple No tube X novi‐belgii) Dandelion (Taraxacum Yellow No tube X officinale) Spreading dogbane White/pink Open tube X (Apocynum androsaemifolium) Common milkweed Pink Open tube X X X (Asclepias syriaca) Swamp milkweed (Asclepias Pink Open tube X incarnata) Hedge false bindweed White/pink Open tube X X (Calystegia sepium) Creeping bellflower Purple Open tube X (Campanula rapunculoides) Showy tick trefoil (Desmodium Pink Open tube X X canadense) Northern bush honeysuckle Yellow Open tube X X (Diervilla lonicera) Boneset (Eupatorium White Open tube X perfoliatum) Purple Joe pye weed (Eutrochium Pink Open tube X X purpureum) Split‐lipped hemp nettle (Galeopsis Pink Open tube X X bifida) Amur honeysuckle White Open tube X (Lonicera maackii) Purple loosestrife Purple Open tube X X (Lythrum salicaria)

88

Alfalfa (Medicago Purple Open tube X sativa) Yellow sweet clover (Melilotus Yellow Open tube X officinalis) Spearmint Purple Open tube X (Mentha spicata) American wild mint (Mentha Purple Open tube X canadensis) Allegheny monkeyflower Purple Open tube X (Mimulus ringens) Common evening primrose Yellow Open tube X (Oenothera biennis) Penstamon cultivar Pink Open tube X X (Penstamon spp.) Common selfheal Purple Open tube X X (Prunella vulgaris) White meadowsweet White Open tube X X X (Spiraea alba) Hairy hedgenettle Purple Open tube X X X (Stachys hispida) Common comfrey (Symphytum Purple Open tube X officinale) Purple cultivar Purple Open tube X Unknown bush White Open tube X Blue vervain Purple Open tube X (Verbena hastata) Jewelweed (Impatiens Orange Spiked tube X* X X capensis) Toad flax (Linaria Yellow Spiked tube X X vulgaris) Unknown tree Brown X

Appendix G: Conservation ranking definitions for bumblebee species. S Rank Definition SX Extirpated - Species is presumed to be extirpated from the state. Not located despite intensive searches of historical sites and other appropriate habitat, and virtually no likelihood that it will be rediscovered.

89

SH Historic (possibly extirpated) - Known from only historical records but some hope of rediscovery. There is evidence that the species may no longer be present in the state, but not enough to state this with certainty. Such evidence may include: (1) species has not been documented in many years despite searching; (2) significant habitat loss or degradation; (3) species has been searched for unsuccessfully, but not thoroughly enough to presume that it is no longer present in the state.

S1 Critically Imperiled - Extreme rarity or other factor(s) such as severe decline making it especially vulnerable to extirpation from the state. S2 Imperiled - Rarity due to very restricted range, very few populations or occurrences, severe decline, or other factor(s) making it vulnerable to extirpation from the state. S3 Vulnerable - Rarity due to restricted range, relatively few populations or occurrences, decline, or other factor(s) making it vulnerable to extirpation from the state. S4 Apparently Secure - Uncommon but not rare; some cause for long-term concern in the state due to decline or other factors. S5 Secure - Common, widespread, and abundant in the state. S#S# Range Rank (e.g., S2S3) - Indicates any range of uncertainty about the status of the species. Ranges cannot skip more than two ranks (e.g., SU is used rather than S1S4). SU Unrankable - Lack of information or substantially conflicting information about status in the state. SNR Not Ranked - Status not yet assessed in the state. SNA Not Applicable - No conservation status rank applicable because the species is not a suitable target for conservation activities in the state.

Appendix H: Long vs. medium vs. short tongued bees at early, mid and late season Mid season high elevation:

90

Late season high elevation:

91

Early season low elevation:

92

Mid season low elevation:

93

Late season low elevation:

94

Appendix I: Long vs. medium vs. short tongued workers (within tongue groups) at high and low elevation Short tongued at high elevation:

95

Medium tongued at high elevation:

96

Long tongued at high elevation:

97

Long tongued at low elevation:

98

Appendix J: Long vs. medium vs. short tongued males (within tongue groups) at high and low elevation Short tongued at high elevation:

99

Medium tongued at high elevation:

100

Long tongued at high elevation:

101

Long tongued at low elevation:

102

103