Comparative survival and fitness of bumble bee colonies in natural, suburban, and agricultural landscapes

Agriculture, Ecosystems and Environment 284 (2019) 106594

Contents lists available at ScienceDirect

Agriculture, Ecosystems and Environment journal homepage: www.elsevier.com/locate/agee

Comparative survival and fitness of bumble bee colonies in natural, suburban, and agricultural landscapes Nelson J. Milanoa,1, Aaron L. Iversona,1, Brian A. Naultb, Scott H. McArta, a b

T

⁎

Department of Entomology, Cornell University, Ithaca, NY 14853, United States Department of Entomology, Cornell University, Cornell AgriTech, 15 Castle Creek Drive, Geneva, NY 14456, United States

A R T I C LE I N FO

A B S T R A C T

Keywords: Bombus impatiens Pollinator health Landscape ecology Urban ecology Bee conservation

Pollinators such as bumble bees are in decline as a result of many factors, including loss of habitat. Initiatives to improve and restore pollinator habitat are increasingly popular. However, to most effectively conserve pollinators, we need a better understanding of which habitats limit their survival and fitness at the landscape scale. Our study examined performance and fitness of the common eastern bumble bee, Bombus impatiens (Cresson), in four common landscapes for bees (natural, suburban, conventional agriculture, and organic agriculture). In the summers of 2016 and 2017, 64 bumble bee colonies per year were deployed across 16 sites (4 sites in each landscape type in each year) and their growth (weight and bee abundance), fitness (caste production), and survival were monitored weekly. Colonies in suburban sites in 2016, but not 2017, were lighter, produced fewer worker and drone cells, and experienced queen death more quickly than colonies in natural and agricultural landscapes. The performance and fitness of colonies in natural, organic agricultural, and conventional agricultural landscapes in both years were similar. Additionally, across all landscape types, the proportion of developed land and impervious surface were significantly negatively associated with colony performance in 2016. Thus, our results suggested that suburban landscapes are suboptimal for B. impatiens compared to natural and agricultural landscapes, and that this effect differs across years, potentially due to climactic differences. Future research is needed to identify the mechanisms responsible for reduced performance of bumble bee colonies in suburban landscapes, especially regarding floral resources and pesticide and pathogen stress. Such information could direct specific actions to improve suburban habitat for pollinators.

1. Introduction Wild pollinators in Europe, North America and elsewhere have been experiencing population declines and range contractions (Biesmeijer et al., 2006; Potts et al., 2010; Cameron et al., 2011). This is concerning from both a conservation and food security standpoint, given that pollinators contribute to the reproduction of approximately 60–70% of wild and cultivated plant species (Klein et al., 2007; Garibaldi et al., 2011; Burkle et al., 2013; Vanbergen et al., 2013). Recent research has identified factors associated with declines in wild bee populations, including exposure to pesticides, parasites, and lack of floral resources (Kosior et al., 2007; Williams and Osborne, 2009; Cameron et al., 2011; Bartomeus et al., 2013; Goulson et al., 2015; McArt et al., 2017a). Habitat degradation and loss from urbanization and agricultural intensification are often cited as major drivers of declines (Williams and Kremen, 2007; Winfree et al., 2011; Williams et al., 2012); yet,

comparative studies assessing how bees perform across multiple landscape types show a high variability in their responses. Further studies in different geographic locations are critical to inform effective conservation efforts. The urban landscape is rapidly expanding. Fifty-four percent of the world population lives in urban areas, and it is projected to increase by 12% by the year 2030 (Seto et al., 2012). For pollinators, urban expansion results in habitat loss or fragmentation, which in turn can reduce the availability of forage and nesting substrates (McIntyre and Hostetler, 2001; Johnson and Klemens, 2005). Diversity and abundance of solitary bee species tends to decrease with increasing urbanization in both cities and suburbs (Hernandez et al., 2009; Bates et al., 2011; Fortel et al., 2014). However, not all urban landscapes are alike, and some taxa may respond differently to urbanization. For example, Bombus terrestris performs better in urban landscapes compared with agricultural landscapes in the United Kingdom (Samuelson et al., 2018).

⁎

Corresponding author at: Cornell University, 129 Garden Ave., Ithaca, NY 14853, United States. E-mail address: [email protected] (S.H. McArt). 1 Authors contributed equally. https://doi.org/10.1016/j.agee.2019.106594 Received 31 August 2018; Received in revised form 25 April 2019; Accepted 27 June 2019 Available online 04 July 2019 0167-8809/ © 2019 Elsevier B.V. All rights reserved.

Agriculture, Ecosystems and Environment 284 (2019) 106594

N.J. Milano, et al.

is stable, and it is not a species of conservation concern (IUCN Red List, Hatfield et al., 2014). However, because B. impatiens exists naturally in multiple habitats, it can be used as a model organism to determine how the surrounding environment may influence similar species. In late June 2016 and 2017, we obtained commercially-reared B. impatiens colonies from BioBest Canada Ltd. (Leamington, ON, Canada). Upon arrival, colonies were weighed (nearest 0.1 g) and assessed for numbers of workers, worker cells, and drone cells. Colonies were then sorted according to weight and grouped evenly in sets of four. We replaced the supplied colony covers with 203 x 254 mm clear polycarbonate sheets (Lexan; SABIC Innovative Plastics US; Houston, TX) that had drilled ventilation holes (2 mm diameter) and a covered access hole (50 mm diameter) in the center. We then placed colonies in corrugated plastic boxes (BioBest Canada Ltd.), with each box containing two colonies, and with entrances differentially painted (black, blue, pink or yellow) and facing different directions to reduce drift of bees among colonies (Birmingham et al., 2011; Zanette et al., 2014). Prior to deployment in the field, the sugar bladders were removed to ensure that bees would leave the colony and forage.

Similarly, related bumble bee species visiting suburban garden beds in the United Kingdom had longer lifespans (Goulson et al., 2010) and colonies grew faster and larger than in agricultural landscapes (Goulson et al., 2002). In North America, bumble bees are in greater abundance in urban parks containing natural land compared with surrounding natural habitats (McFrederick and LeBuhn, 2006). Conversely, Glaum et al. (2017) found that diversity and abundance of common northeastern U.S. bumble bee species decreased with increased urbanization. Furthermore, Vaidya et al. (2018) found equivocal evidence that urbanization impacted B. impatiens performance. Additional comparative studies of bee performance in urban and other landscapes are clearly needed to understand how and when urbanization impacts pollinators. Agricultural intensification generally has a negative impact on pollinator diversity and abundance via reduced and/or monotonous forage and reduction of nesting (Scheper et al., 2013; Connelly et al., 2015). Conversely, pollinator diversity and abundance can increase in fields where more of the surrounding landscape is composed of natural habitat (Holzschuh et al., 2012; Joshi et al., 2016; Kammerer et al., 2016). Farm practices in agricultural landscapes can have mixed effects on pollinators. For example, conventional and organic agricultural practices can expose pollinators to an array of pesticides that may elevate their susceptibility to pathogen infection (Baron et al., 2014; David et al., 2016; Botías et al., 2017), which was recently shown to be associated with range contractions in bumble bees in the U.S. (McArt et al., 2017b). Yet, in many cases, organic farms will have a more heterogenous landscape context, allowing for more diverse diets (Rundlöf et al., 2008; Tuck et al., 2014; Schellhorn et al., 2015). Crone and Williams (2016) showed that in relation to floral resources, bumble bee queen production was greatest in natural habitats, intermediate on organic farms, and lowest in conventional farms. However, a colony’s success in a natural habitat is more dependent on local floral abundance than natural landscape alone (Spiesman et al., 2017). In some instances, mass-flowering crops such as oilseed rape and red clover can be beneficial to pollinator populations and fitness (Westphal et al., 2009; Rundlöf et al., 2014; but see Rundlöf et al., 2015). Thus, like urbanization, further research is needed to elucidate how conventional and organic farm practices in agricultural landscapes impact bee populations. Here, we examined the effects of multiple commonly encountered landscapes on colony performance, survival, and fitness of the common eastern bumble bee, Bombus impatiens (Cresson). Experimental B. impatiens colonies were placed in natural, suburban, conventional agriculture, and organic agriculture landscapes in two successive years (2016 and 2017) to address the following questions about B. impatiens colony success: 1) How do landscapes and farm practices influence colony growth, survival, and fitness? and 2) Which specific types of land cover (i.e., agricultural land, natural land, developed land and impervious surface) are predictive of colony performance? We hypothesized that 1) survival and fitness of B. impatiens would be greatest in natural habitats compared with agricultural and suburban landscapes and 2) B. impatiens colony performance would be most negatively impacted by developed land and impervious surface.

2.2. Field sites and colony placement Colonies were randomly assigned to one of the following landscape types: natural, suburban, conventional agriculture, and organic agriculture. Each landscape type was replicated in regions in proximity to Rochester, Syracuse, Ithaca, and Geneva, NY, USA for a total of four replications and a total of 16 sites. Each site received two weather-proof corrugated plastic boxes, each containing two B. impatiens colonies, for a total of 64 colonies per year. To reduce drift between colonies, each box was separated by a minimum of 5 m and paired colonies within a box were positioned on opposite sides. Colonies placed in conventional agriculture and organic agriculture landscapes were positioned in the margins of cucurbit crop fields (e.g., squash and pumpkin) and deployed coinciding with bloom. All colonies were deployed over a fourday span (July 01-04, 2016 and July 03-06, 2017), with each region (consisting of one each of the four landscape types) being deployed on the same day between the hours of 09:00-18:00. In 2017, one natural landscape replicate in the Rochester region was lost due to flooding during the first week of assessments, and one natural landscape replicate in the Geneva region was moved to a more appropriate location (Table S1). Land cover composition around all sites was assessed using a geographic information system (ArcGIS version 10.5.1; ESRI, 2018) with data from the United States Department of Agriculture Cropland Data Layer (CDL; USDA National Agriculture Statistics Service, 2016)) and the National Land Cover Database (Xian et al., 2011). We determined the degree of agricultural, natural, developed, and impervious surface land cover at an 800 m buffer for each site (Fig. 1). This radius was chosen because it has been identified as the mean foraging radius for bumble bees (McArt et al., 2017b) and previous studies from the same region have shown 750 m to be the most important scale for predicting landscape impacts on wild bees (Park et al., 2015; Grab et al., 2018). We chose cucurbits as the consistent crop across all agricultural landscapes because B. impatiens regularly visits this crop and because farmers occasionally purchase B. impatiens colonies for supplemental pollination services (B. Nault, pers. comm.). Both organic and conventional agricultural habitats were selected to have a similar field size (range 0.83–3.0 ha) and crop production type (i.e., small to medium mixed vegetable farms). However, organic farms were on average smaller (mean 9.2 ha, range 3–16 ha) than conventional farms (mean 22.0 ha, 5–30 ha). Conventional farms often grew cucurbit crops, corn, and soybeans while organic farms were more diverse, growing cucurbit crops, leafy greens, solanaceous crops, carrots, and onions. The CDL classes for all agricultural crops were summed to determine percent agricultural land cover. The area of natural landscape was calculated by summing the area of all CDL classes that included landcovers relating to

2. Methods 2.1. Study system Bombus impatiens is a widely encountered bumble bee species across its native range of eastern North America, where it is active from April to October (Williams et al., 2014). B. impatiens nests underground, often using old rodent cavities for their hives (Michener, 2000). After an overwintering queen establishes a new colony, she will produce only workers during the first phase of development, then she will switch to producing a mix of workers and reproductive individuals (queens and drones) towards the end of the colony’s single-season lifespan (Cnaani et al., 2002). Unlike many bumble bee species, the status of B. impatiens 2

Agriculture, Ecosystems and Environment 284 (2019) 106594

N.J. Milano, et al.

Fig. 1. Location and land cover types in central New York State, USA, where colonies of B. impatiens were placed in conventional agricultural, organic agricultural, natural, and suburban landscapes.

decreased in weight across all assessment periods in 2017; therefore, in order to standardize our comparisons between years, we analyzed the same period of time as in 2016 (weeks one to four). All other colony performance variables included all assessment weeks in the analysis. Brood cell abundance included counts for worker cells, drone cells, and gyne cells, all of which can be differentiated by shape, size, and surface texture (Figure S1). Prior to placing the colonies in the field, foundress queens were marked on the thorax so we could differentiate them from emerging gynes; these gynes were removed from the colonies after each weekly count to avoid double-counting. To account for the challenge of accurately counting cells and adult bees, as the brood form a 3-dimensional formation in the colony, we removed each colony from its weather-proof box, observed the colony from all angles, separately counted brood cells in the four quadrants of a colony, and repeated counts a minimum of three times per visit. Although we separated paired colonies by at least 5 m, there was still a possibility that drifting between colonies occurred, as has been observed in other studies on B. impatiens and other Bombus spp. (Birmingham and Winston, 2004; Lefebvre and Pierre, 2007). Therefore, we averaged colony performance data across all four colonies at each site. We also recorded the date when the foundress queen died as a

grass/pasture, shrubland, wetlands, and forest. The area of suburban landscape was calculated by summing the area of CDL classes that were categorized as developed, including low, medium, and high intensity. Finally, the percent of impervious surfaces in suburban sites was calculated using the National Land Cover Database Percent Developed Imperviousness layer (Xian et al., 2011). 2.3. Colony performance Colony growth and reproductive success metrics were assessed weekly during the study period by recording colony weight, adult bee abundance (workers and drones combined), brood cell abundance, and gyne (virgin queen, not foundress queen) abundance during the daylight hours (09:00-18:30) to comply with home/landowner restrictions. To obtain colony weight, we subtracted the weight of an average plastic box housing (the same as what was used in the experiment) from the weight of the colony plus plastic box that we measured in the field. To assess relative colony growth, we restricted our analysis of colony weight change to the period of time when the colonies were actively growing. After this period, growth rates were negative. This period of growth occurred from assessment week one to four in 2016. Colonies 3

Agriculture, Ecosystems and Environment 284 (2019) 106594

N.J. Milano, et al.

measure of colony survival. Once the queen died, the colony was removed from the site, frozen at -20 °C and then dissected for final assessment. Old brood cells and honey pots are often recycled/destroyed by workers or consumed by various pests that eventually invade the colonies (i.e., Coleoptera: Nitidulidae, Staphylinidae, Sylphidae; Blattodea: Ectobiidae; Diptera: Calliphoridae, Drosophilidae, Muscidae, Phoridae; Lepidoptera: Pyralidae). We observed some of these pest effects during the course of our assessments, but impacts were relatively consistent across colonies and sites. However, by the time dead colonies were brought back to the lab and inspected, damage had accumulated to levels that precluded use of these data in the final performance analyses.

Table 1 ANOVA tables from linear mixed effects models of (A) colony weight, (B) bee abundance, (C) worker cell abundance, and (D) drone cell abundance for B. impatiens colonies placed in conventional agricultural, organic agricultural, natural and suburban landscapes in 2016 and 2017. Asterisks indicate significant effects (p < 0.05). A. Colony weight

Landscape Year Period Landscape x Year Landscape x Period Year x Period Landscape x Year x Period

2.4. Statistical analysis All statistical analyses were conducted using R (R Core Team, 2017), using the lmer4, glmmTMB, survival, coxme, emmeans, ggplot2, ggpuber and survminer packages (Brooks et al., 2017; Therneau and Lumley, 2017; Lenth et al., 2018; Therneau, 2018). Models were selected using the lowest AIC scores from likelihood ratio tests (Johnson and Omland, 2004). To analyze mean colony weight change, mean bee abundance, mean worker cell abundance, and mean drone cell abundance throughout the duration of the study for both 2016 and 2017, linear mixed effects models were used with “landscape”, “year”, and “assessment period” as fixed effects and “region” and “site” as random effects. To analyze mean gyne cell abundance and mean gyne adult abundance, we used truncated negative binomial hurdle models with mixed effects because of the large quantity of zeros in the data. Mean gyne cell abundance and mean gyne adult abundance were each modeled as a binary process for presence/absence and the fitting of a truncated negative binomial model with mean counts equal to or greater than zero at each site. Hurdle models included “landscape” and “year” as fixed effects and “region” and “site” as random effects (“assessment period” was removed as a fixed effect due to insufficient replication). Queen death was analyzed using Kaplan-Meir survival curves and mixed effects Cox proportional hazard models with “landscape” and “year” as fixed effects and “region” and “site” as random effects. Post-hoc tests for linear mixed effects and survival analyses models were conducted using the emmeans package (Lenth et al., 2018) adjusted to Tukey’s HSD to determine differences between landscape types, year and assessment periods. We analyzed predictors of performance, including percent agricultural land, natural land, developed land, and impervious surface in an 800 m buffer, using linear regressions with Pearson correlation coefficients.

DF

F-value

p

3,9 1,121.1 5,120.1 3,121.1 15,121.1 5,121.1 15,120.1

2.2 27.4 3.8 5.4 1.7 9.1 1.4

0.15 < 0.0001* < 0.05* < 0.05* < 0.05* < 0.0001* 0.17

DF

F-value

p

3,9.2 1,121.5 5,120.1 3,121.4 15,120.1 5,120.1 15,120.1

2.3 136.3 12.6 3.1 1.0 0.9 0.5

0.15 < 0.0001* < 0.0001* < 0.05* 0.47 0.45 0.90

DF

F-value

p

3,9.1 1,122.5 5,120.1 3,122.4 15,120.0 5,120.04 15,120.0

5.6 31.8 4.5 2.4 1.5 8.2 1.2

< 0.05* < 0.0001* < 0.001* 0.08 0.12 < 0.0001* 0.3

DF

F-value

p

3,9.2 1,122.5 5,120.3 3,122.3 15,120.2 5,120.3 15,120.2

3.6 36.0 9.6 3.1 0.9 2.8 0.4

< 0.05* < 0.0001* < 0.0001* < 0.05 0.54 < 0.05 0.98

B. Bee abundance

Landscape Year Period Landscape x Year Landscape x Period Year x Period Landscape x Year x Period C. Worker cell abundance

Landscape Year Period Landscape x Year Landscape x Period Year x Period Landscape x Year x Period D. Drone cell abundance

Landscape Year Period Landscape x Year Landscape x Period Year x Period Landscape x Year x Period

Adult bee and worker cell abundance – There was a significant effect of year, assessment period and the interaction between landscape type and year on adult bee abundance, which included workers and drones (Table 1: B). Similarly, worker cell abundance was significantly impacted by landscape type, year, period, and the interaction between year and period (Table 1: C). In 2016, suburban landscapes had 34% fewer bees in their colonies than organic agriculture landscapes (Posthoc: p = 0.04; Fig. 3: A). Suburban colonies also produced over 50% fewer worker cells than those in the other landscape types in 2016 (Post-hoc: p < 0.05 for all; Fig. 3: B). These results were largely driven by differences in assessment periods four and five, when colonies obtained peak weight (Fig. 2) and worker numbers. We found no differences in bee abundance or worker cell abundance between each type of agricultural and natural landscapes in 2016 and 2017 (p > 0.1 for all).

3. Results 3.1. Colony performance Colony weight – We found no overall significant effect of landscape type on colony weight in a full model using data from both 2016 and 2017 (Table 1: A). However, there were significant effects of year, assessment period, the interaction of landscape and year, and the interaction of year and assessment period on colony weight (Table 1: A). In 2016, colony weight in suburban landscapes was 45 and 46% lighter than those in organic agriculture and natural landscapes, respectively (Post-hoc: p = 0.05). This was driven primarily by the fact that colonies in suburban landscapes were 52% lighter than those in natural landscapes in assessment period three, 61% lighter than both organic and natural landscapes in assessment period four, and 61–65% lighter than all three landscape types in assessment period five (Fig. 2: A). In 2017, the mean mass of colonies in suburban landscapes was 16–21% less than in the other three landscape types, but this difference was not statistically significant (Fig. 2: B). There was no difference in colony weight between either type of agricultural and natural landscape in 2016 or 2017 (p > 0.9 for all).

3.2. Colony fitness and queen death Drone cell abundance – We found a significant effect of landscape type, year, period, the interaction of landscape type and year, and the interaction of year and assessment period (Table 1: D) on drone cell abundance. In 2016, suburban landscapes colonies had 53–55% fewer drone cells than those in the other three landscape types (Post-hoc: p < 0.05 Fig. 3: C). Additionally, suburban landscape colonies had 4

Agriculture, Ecosystems and Environment 284 (2019) 106594

N.J. Milano, et al.

Fig. 2. Mean colony weight (g. +/- SE) in (A) 2016 and (B) 2017 of B. impatiens colonies placed in conventional agricultural, organic agricultural, natural and suburban landscapes. Asterisks indicate significant differences between landscapes based on Tukey’s post-hoc tests of lmer models (p < 0.05). Assessment weeks 0–9 in 2016 correspond to Julian dates 180–252 (June 28-August 8) and assessment weeks 0–8 in 2017 correspond to Julian dates 178–256 (June 27-August 12).

3.3. Predictors of colony performance and fitness

65% fewer drone cells than organic (p = 0.03) and natural (p = 0.02) landscapes in assessment period three and 82–84% fewer drone cells than all three landscapes types (conventional: p < 0.01; organic: p = 0.05; natural: p < 0.02) in assessment period five for that year. There were no significant differences in drone cell abundance between natural landscapes and either type of agricultural landscape in 2016 (p > 0.1 for all) and no significant differences between all landscape types in 2017 (p > 0.1 for all). Adult gyne and gyne cell abundance. B. impatiens colonies in conventional agriculture and natural landscapes produced significantly more gyne cells and adult gynes than in organic agriculture and suburban landscapes in 2016 (Table 2; Figure S2 and S3). Although colonies did not produce many gyne cells and adult gynes overall in 2017, a similar pattern was observed. Significant effects of natural and conventional agriculture landscape types across both years were driven by the abundance of gyne cells and gyne adults in the colonies as defined by the zero-inflated model, rather than the presence/absence of gynes, as defined by the conditional model (Table 2). Queen death. In both years, suburban landscape colonies were likely to experience foundress queen death at a rate approximately two and a half times greater than the colonies in the other landscape types (Hazard Ratio = 2.46, p = 0.001). In 2016, queens in all suburban landscape colonies died six days earlier than the overall average lifespan (Fig. 4: A). A post-hoc test showed that survival time of queens in suburban landscapes was significantly less than those in organic agriculture landscapes and natural landscapes (p = 0.001 and p < 0.001, respectively; Fig. 4: A). In 2017, a post-hoc test showed that queen survival in suburban landscapes was significantly less than in conventional and organic agricultural landscapes (p = 0.01 for both; Fig. 4: B).

To explore why colonies in some landscapes may have performed worse than others, we tested for quantitative relationships between specific land cover characteristics (calculated at an 800 m radius) and performance and fitness measures. Our landscape designations were appropriate, as agricultural sites were dominated by agricultural land cover, natural landscapes were dominated by natural land cover, and suburban sites were dominated by developed land cover (Figure S4:AC). Furthermore, impervious surface cover was highest in suburban landscapes (Figure S4: D). In 2016, there was a significant negative relationship between relative colony growth and both developed land and impervious surface (R = -0.71, p = 0.01 and R = -0.65, p = 0.005, respectively; Fig. 5: 2016). These results were similar for adult bee abundance, which was negatively correlated with developed land and impervious surface in 2016 (R = -0.54, p = 0.03; and R = -0.53, p = 0.036, respectively; Figure S5: A). Worker cell abundance was negatively correlated with developed land and impervious surface in both 2016 (R = -0.62 p = 0.01 and R = -0.6, p = 0.015, respectively; Figure S5: B) and 2017 (R = -0.56, p = 0.03 and R = -0.61, p = 0.016, respectively; Figure S5: B). Drone cell abundance was negatively correlated with developed land and impervious surface in 2016 (R = -0.67 p = 0.004 and R = -0.63, p = 0.008, respectively; Figure S6: A) and impervious surface in 2017 (R = -0.54, p = 0.036; Figure S6: A). We found no significant differences between relative colony growth, adult bee abundance, worker cell abundance, drone cell abundance, or gyne cell abundance and percent agricultural or natural land cover in either 2016 and 2017 (Figure S5-S7), aside from a strong positive correlation between percent agricultural land cover and gyne cell abundance in 2016 (R = 0.66, p = 0.005; Figure S7: B). Drone cell abundance and colony growth showed a marginally significant positive relationship with natural land cover in 2016 (R = 0.48, p = 0.06, Figure S7A and 5

Agriculture, Ecosystems and Environment 284 (2019) 106594

N.J. Milano, et al.

Fig. 3. (A) Mean bee abundance, (B) worker cell abundance, and (C) drone cell abundance of B. impatiens colonies placed in conventional agricultural, organic agricultural, natural and suburban landscapes in 2016 and 2017. Letters above error bars indicate significant differences between landscapes based on Tukey’s posthoc tests of lmer models (p < 0.05).

years of the study. It is plausible that the influence of the agricultural landscapes in our study was more beneficial than would be expected for naturally occurring bumble bee colonies given that colonies were placed next to a mass-flowering crop (squash/pumpkin) and that natural areas, which primarily consist of forest in the study region, provide the largest quantity of floral resources in the early part of the year before colonies were deployed (A. Iverson, unpublished data). Recent publications have demonstrated varying impacts of urban landscapes on bumble bee performance and diversity. For example, in contrast to our results, Samuelson et al. (2018) found that colonies of Bombus terrestris placed in suburban (city and village) areas in the United Kingdom had greater peak sizes, bee abundance, and production of sexual individuals than agricultural landscapes. These results are consistent with other studies highlighting the positive impacts of suburban gardens on bumble bee performance in the United Kingdom (Goulson et al., 2002, 2010) and abundance and diversity in North America (McFrederick and LeBuhn, 2006). However, other studies have found null or negative impacts of urbanization on bumble bee performance and diversity. Vaidya et al. (2018) found no differences in the performance of Bombus impatiens colonies placed along an urbanization

R = 0.44, p = 0.085, Figure S5, respectively).

4. Discussion Through a two-year study comparing how multiple landscapes and land management practices (i.e., conventional and organic farming) impact B. impatiens colony performance and fitness, we showed that B. impatiens was negatively affected by suburban development. We also showed that this effect was temporally variable, where although the patterns held in 2017, the only significant differences in colony performance were observed in 2016. Here, colonies in suburban landscapes experienced lower weight gain, had fewer adult bees present in colonies, and produced fewer worker cells, drone cells, and gyne cells than all other landscape types. Additionally, foundress queens in suburban landscapes had significantly shorter lifespans than those in organic agricultural landscapes and natural landscapes in 2016. Our results thus highlight the importance of considering suburban areas to address bumble bee conservation. Contrary to our expectations, we saw few differences in colony weight or fitness between agricultural (including both conventional and organic) and natural landscapes in both 6

Agriculture, Ecosystems and Environment 284 (2019) 106594

N.J. Milano, et al.

et al. (2018) demonstrated that bee abundance decreased by 41% with each increase in degree Celsius in urban areas, despite floral abundance. While we did not directly measure temperature at each of our sites, increases in impervious surface in our suburban sites could have created elevated temperatures compared to the other landscape types. The large size and hair density of bumble bees makes them susceptible to overheating (Oyen et al., 2016) and in some cases, workers may switch from foraging to fanning, resulting in reduced resource intake and increased stress (Vogt, 1986). In 2016, much of northeastern U.S., including New York experienced one of the warmest and driest summers in 60 years (Sweet et al., 2017), which contrasted with the wetter summer in 2017. In addition to directly impacting bumble bees via heat stress, the drought could have had indirect effects by reducing nectar flow and pollen production of nearby plants (Scaven and Rafferty, 2013). This impact may have been especially pronounced in suburban areas, as water stress in agricultural and natural landscapes may have been mitigated via irrigation systems for crops and small bodies of water (i.e., streams and creeks), respectively. However, any impacts that we observed on B. impatiens due to increased temperatures in suburban landscapes may be exaggerated compared to natural colonies, as bumble bees nest underground where the temperature would have been moderated compared to our above-ground experimental placement. Although 2017 was a non-drought year when we may have expected better overall colony performance, the colonies we received that year were already more advanced in their physiological state than the colonies we received in 2016. The state of the colonies likely explains why we only observed a decrease in colony growth in all landscapes in 2017. Another major stressor on bee health is pesticide exposure, which has been widely documented in agricultural landscapes (Mullin et al., 2010; Pettis et al., 2013; McArt et al., 2017a) and can impair foraging behavior (Gill and Raine, 2014), reduce colony growth, and reproduction in bumble bees (Gill et al., 2012; Rundlöf et al., 2015). Pesticide exposure in urban landscapes is less understood, but there is evidence of low pesticide exposure to bumble bees in suburban landscapes (David et al., 2016; Botías et al., 2017). Yet, colonies in conventional agricultural sites did not have decreased performance compared to organic agricultural sites and natural landscapes, suggesting that either pesticide exposure did not vary considerably between landscapes or that exposure was below thresholds that would have impacted colony performance. Although we intentionally selected relatively large organic farms (mean size 9.2 ha), it was not possible to locate large agricultural landscapes that were entirely organic. Therefore, given the ˜800 m foraging radius of B. impatiens, it is possible that the bees placed on organic farms foraged on neighboring conventional farms, thereby diluting any differences in pesticide exposure that may have occurred. Furthermore, organic farms may have also used organic pesticides that negatively affected bee performance. Our results of poorer colony performance in suburban sites may also reflect pathogen exposure, which can reduce overall colony performance, increase mortality, and reduce reproduction (Imhoof and Schmid-Hempel, 1999; Yourth et al., 2008; Graystock et al., 2016). Although we do not have pathogen data, there is evidence of higher pathogen loads in urban environments in both bumble bees (Theodorou et al., 2016) and honey bees (Youngsteadt et al., 2015). This pattern may occur because floral resources in urban environments are often found in concentrated patches where bees congregate and share resources, increasing the chances of pathogen transmission (Durrer and Schmid-Hempel, 1994; Goulson et al., 2012). In contrast, large amounts of floral resources like those found in suburban gardens in the United Kingdom (Goulson et al., 2002, 2010) or in natural landscapes can mitigate pathogen infection through nutrition (Tritschler et al., 2017) and/or plant secondary compounds (Biller et al., 2015; Richardson et al., 2015). Therefore, the negative physiological impacts of pathogens could potentially have been minimized in agricultural and natural landscapes in this study due to greater floral abundance or diversity

Table 2 Summary table of truncated negative binomial hurdle model with mixed effects of (A) gyne cell abundance and (B) adult gyne abundance for B. impatiens colonies placed in conventional agricultural, organic agricultural, natural, and suburban landscapes in 2016 and 2017, with suburban landscape as the reference. Asterisks indicate significant effects (p < 0.05). A. Gyne cell abundance Conditional model Landscape Intercept Conventional Organic Natural Zero-inflation model Landscape Intercept Conventional Organic Natural

z-value −0.4 0.7 −0.08 1.09

p 0.6 0.4 0.9 0.2

z-value 0.5 −2.2 0.2 −2.3

p 0.6 0.02* 0.7 0.01*

B. Adult gyne abundance Conditional model Landscape Intercept Conventional Organic Natural Zero-inflation model Landscape Intercept Conventional Organic Natural

z-value −0.4 0.7 −0.08 1.09

p 0.9 0.9 1 0.9

z-value 0.5 −2.2 0.2 −2.3

p 0.004* 0.02* 0.3 0.02*

gradient in Michigan, USA. In addition, Glaum et al. (2017) found negative impacts of urbanization on bumble bee abundance and diversity. Taken together, the evidence suggests that there are region-specific and perhaps species-specific responses of bumble bees to urbanization. Understanding the mechanisms driving performance and fitness across landscapes will likely provide insight into why different responses occur. Both percent developed land and impervious surface were negatively correlated with relative colony growth, bee abundance, and caste production in our study in 2016, primarily driven by the suburban sites (Fig. 5; Figures S6-S7). Greater amounts of impervious surface could impact bumble bees in multiple ways, including abundance of floral resources and increases in temperature (i.e., the urban “heat island” effect). While flowering ornamental plants were present in our suburban sites, the relative quantity of these resources compared with those in natural forage was difficult to assess, and the quality (e.g., nectar and pollen availability) of ornamental plants may also be poor. For example, Campbell et al. (2017) showed that 46% of homeowners choose pollinator-friendly plants primarily on attractiveness rather than their benefits for pollinators (but see Garbuzov and Ratnieks 2014). Furthermore, if local floral resources are lacking, bumble bees may fly to surrounding natural habitats to forage (Spiesman et al., 2017). Because our suburban sites had an average of only 6% natural land composition within an 800 m radius (Figure S4: B), foragers may have had to travel further than other landscape types to collect nectar and pollen, which could be detrimental to colony success (Osborne et al., 2008). Though results were not significant, relative colony growth, bee abundance, and worker, drone and queen cell abundance were positively associated with percent natural land area (Figures S5-S7), supporting the notion that natural areas may provide some important habitat for bumble bees. In terms of increased temperatures impacting bee health, Hamblin

7

Agriculture, Ecosystems and Environment 284 (2019) 106594

N.J. Milano, et al.

Fig. 4. Kaplan-Meir curves representing the number of days colonies spent in the field before queens died in (A) 2016 and (B) 2017. Each step in the Kaplan-Meier survival curves represents a queen death, resulting in removal of the colony from the experiment. Letters indicate significant differences between landscapes based on post-hoc tests of Cox proportional hazard models with mixed effects (p < 0.05).

Author’s contributions

(Alaux et al., 2010).

N.J.M., S.H.M., and B.A.N. designed the study. N.J.M. collected and analyzed the data. A.L.I. conducted G.I.S. analyses. N.J.M. and A.L.I. drafted the manuscript. N.J.M., A.L.I., S.H.M., and B.A.N. edited the manuscript.

5. Conclusion Multiple stressors, which are known to vary among different landscape contexts, can strongly impact pollinator health and abundance, especially when they are co-occurring (Winfree et al., 2011; Goulson et al., 2015). We demonstrated that suburban landscapes were suboptimal for B. impatiens colonies, while few differences in performance or fitness existed among other landscape types and farm practices. Such results warrant further exploration to determine the specific mechanisms influencing bumble bee colony dynamics. For example, if floral resources are found to be limiting in suburban areas, targeted efforts to provide additional habitat via urban gardens, wildflower plantings, or adjusted mowing schedules could be effective. More broadly, we suggest that further research on how different landscapes and farm practices shape pollinator survival and fitness is critically important for determining where to best focus conservation efforts.

Funding This study was funded by the USDA (Federal Capacity Fund #6217302) and the NYS Environmental Protection Fund. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the funding agencies. Acknowledgements We thank David Lewis, Trebor Hall, and Ashley Fersch with help in setting up the study and fieldwork. Heather Grab, Katja Poveda, and the 8

Agriculture, Ecosystems and Environment 284 (2019) 106594

N.J. Milano, et al.

Fig. 5. Relationship using Pearson’s correlation between daily relative growth of B. impatiens colonies through peak mass in 2016 (assessment period four) and percent developed land and impervious surface in 2016 and 2017.

Cornell Statistical Consulting Unit provided guidance with statistical analyses. We also thank Susan Earle, Kathy Albertini, Mellissa Fierke, Greg Loeb, Chuck Guard, Larry Smart, Brian Shiels, Bill Wickham, Andrew Fellenz, Jamie Eldenstein, David Fox, JoAnn Delaney, Ammie Chickering, Bob Tuori, Finger Lakes Land Trust, Genesee Land Trust, Central NY Land Trust, and the city of Geneva, NY for allowing use of their properties to conduct this study.

378–400. Burkle, L.A., Marlin, J.C., Knight, T.M., 2013. Plant-pollinator interactions over 120 years: loss of species, co-occurrence, and function. Science 339, 1611–1615. Cameron, S.A., Lozier, J.D., Strange, J.P., Koch, J.B., Cordes, N., Solter, L.F., Griswold, T.L., 2011. Patterns of widespread decline in North American bumble bees. Proc. Natl. Acad. Sci. U. S. A. 108, 662–667. Campbell, B., Khachatryan, H., Rihn, A., 2017. Pollinator-friendly plants: reasons for and barriers to purchase. HortTechnology 27, 831–839. Cnaani, J., Schmid-Hempel, R., Schmidt, J.O., 2002. Colony development, larval development and worker reproduction in Bombus impatiens Cresson. Insectes Soc. 49, 164–170. Connelly, H., Poveda, K., Loeb, G., 2015. Landscape simplification decreases wild bee pollination services to strawberry. Agric. Ecosyst. Environ. 211, 51–56. Crone, E.E., Williams, N.M., 2016. Bumble bee colony dynamics: quantifying the importance of land use and floral resources for colony growth and queen production. Ecol. Lett. 19, 460–468. David, A., Botías, C., Abdul-Sada, A., Nicholls, E., Rotheray, E.L., Hill, E.M., Goulson, D., 2016. Widespread contamination of wildflower and bee-collected pollen with complex mixtures of neonicotinoids and fungicides commonly applied to crops. Environ. Int. 88, 169–178. Durrer, S., Schmid-Hempel, P., 1994. Shared use of flowers leads to horizontal pathogen transmission. Proc. R. Soc. B Biol. Sci. 258, 299–302. ESRI, 2018. ArcGIS Desktop. Enviromental Systems Research Institute, Redlands, CA. Fortel, L., Henry, M., Guilbaud, L., Guirao, A.L., Kuhlmann, M., Mouret, H., Rollin, O., Vaissiere, B.E., 2014. Decreasing abundance, increasing diversity and changing structure of the wild bee community (Hymenoptera: Anthophila) along an urbanization gradient. PLoS One 9, e104679. Garibaldi, L.A., Aizen, M.A., Klein, A.M., Cunningham, S.A., Harder, L.D., 2011. Global growth and stability of agricultural yield decrease with pollinator dependence. Proc. Natl. Acad. Sci. U. S. A. 108, 5909–5914. Gill, R.J., Raine, N.E., 2014. Chronic impairment of bumblebee natural foraging behaviour induced by sublethal pesticide exposure. Funct. Ecol. 28, 1459–1471. Gill, R.J., Ramos-Rodriguez, O., Raine, N.E., 2012. Combined pesticide exposure severely affects individual- and colony-level traits in bees. Nature 491, 105–119. Glaum, P., Simao, M.-C., Vaidya, C., Fitch, G., Iulinao, B., 2017. Big city Bombus: using natural history and land-use history to find significant environmental drivers in bumble-bee declines in urban development. R. Soc. Open Sci. 4, 170156. Goulson, D., Hughes, W.O.H., Derwent, L.C., Stout, J.C., 2002. Colony growth of the bumblebee, Bombus terrestris, in improved and conventional agricultural and suburban habitats. Oecologia 130, 267–273. Goulson, D., Lepais, O., O’Connor, S., Osborne, J.L., Sanderson, R.A., Cussans, J., Goffe, L., Darvill, B., 2010. Effects of land use at a landscape scale on bumblebee nest density and survival. J. Appl. Ecol. 47, 1207–1215. Goulson, D., Nicholls, E., Botias, C., Rotheray, E.L., 2015. Bee declines driven by combined stress from parasites, pesticides, and lack of flowers. Science 347, 1–9. Goulson, D., Whitehorn, P., Fowley, M., 2012. Influence of urbanisation on the prevalence of protozoan parasites of bumblebees. Ecol. Entomol. 37, 83–89. Grab, H., Poveda, K., Danforth, B., Loeb, G., 2018. Landscape context shifts the balance of costs and benefits from wildflower borders on multiple ecosystem services. Proc. R.

Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.agee.2019.106594. References Alaux, C., Ducloz, F., Crauser, D., Le Conte, Y., 2010. Diet effects on honeybee immunocompetence. Biol. Lett. 6, 562–565. Baron, G.L., Raine, N.E., Brown, M.J.F., 2014. Impact of chronic exposure to a pyrethroid pesticide on bumblebees and interactions with a trypanosome parasite. J. Appl. Ecol. 51, 460–469. Bartomeus, I., Ascher, J.S., Gibbs, J., Danforth, B.N., Wagner, D.L., Hedtke, S.M., Winfree, R., 2013. Historical changes in northeastern US bee pollinators related to shared ecological traits. Proc. Natl. Acad. Sci. U. S. A. 110, 4656–4660. Bates, A.J., Sadler, J.P., Fairbrass, A.J., Falk, S.J., Hale, J.D., Matthews, T.J., 2011. Changing bee and hoverfly pollinator assemblages along an urban-rural gradient. PLoS One 6, e23459. Biesmeijer, J.C., Roberts, S.P.M., Reemer, M., Ohlemuller, R., Edwards, M., Peeters, T., Schaffers, A.P., Potts, S.G., Kleukers, R., Thomas, C.D., Settele, J., Kunin, W.E., 2006. Parallel declines in pollinators and insect-pollinated plants in Britain and the Netherlands. Science 313, 351–354. Biller, O.M., Adler, L.S., Irwin, R.E., McAllister, C., Palmer-Young, E.C., 2015. Possible synergistic effects of thymol and nicotine against Crithidia bombi parasitism in bumble bees. PLoS One 10, e0144668. Birmingham, A., E Hoover, S.L., Winston, M., Ydenberg, R., 2011. Drifting bumble bee (Hymenoptera: Apidae) workers in commercial greenhouses may be social parasites. Can. J. Zool. 82, 1843–1853. Birmingham, A., Winston, M., 2004. Orientation and drifting behaviour of bumblebees (Hymenoptera: Apidae) in commercial tomato greenhouses. Can. J. Zool. 82, 52–59. Botías, C., David, A., Hill, E.M., Goulson, D., 2017. Quantifying exposure of wild bumblebees to mixtures of agrochemicals in agricultural and urban landscapes. Environ. Pollut. 222, 73–82. Brooks, M.E., Kristensen, K., Benthem, K.Jv., Magnusson, A., Berg, C.W., Nielsen, A., Skaug, H.J., Maechler, M., Bolker, B.M., 2017. glmmTMB Balances speed and fexibility among packages for zero-inflated generalized linear mixed modeling. R J. 9,

9

Agriculture, Ecosystems and Environment 284 (2019) 106594

N.J. Milano, et al.

Rundlöf, M., Persson, A.S., Smith, H.G., Bommarco, R., 2014. Late-season mass-flowering red clover increases bumble bee queen and male densities. Biol. Conserv. 172, 138–145. Samuelson, A.E., Gill, R.J., Brown, M.J.F., Leadbeater, E., 2018. Lower bumblebee colony reproductive success in agricultural compared with urban environments. Proc. R. Soc. B Biol. Sci. 285 20180807. Scaven, V.L., Rafferty, N.E., 2013. Physiological effects of climate warming on flowering plants and insect pollinators and potential consequences for their interactions. Curr. Zool. 59, 418–426. Schellhorn, N.A., Gagic, V., Bommarco, R., 2015. Time will tell: resource continuity bolsters ecosystem services. Trends Ecol. Evol. 30, 524–530. Scheper, J., Holzschuh, A., Kuussaari, M., Potts, S.G., Rundlof, M., Smith, H.G., Kleijn, D., 2013. Environmental factors driving the effectiveness of European agri-environmental measures in mitigating pollinator loss - a meta-analysis. Ecol. Lett. 16, 912–920. Seto, K.C., Guneralp, B., Hutyra, L.R., 2012. Global forecasts of urban expansion to 2030 and direct impacts on biodiversity and carbon pools. Proc. Natl. Acad. Sci. U. S. A. 109, 16083–16088. Spiesman, B.J., Bennett, A., Isaacs, R., Gratton, C., 2017. Bumble bee colony growth and reproduction depend on local flower dominance and natural habitat area in the surrounding landscape. Biol. Conserv. 206, 217–223. Sweet, S.K., Wolfe, D.W., DeGaetano, A., Benner, R., 2017. Anatomy of the 2016 drought in the Northeastern United States: implications for agriculture and water resources in humid climates. Agric. For. Meteorol. 247, 571–581. Theodorou, P., Radzevičiūtė, R., Settele, J., Schweiger, O., Murray, T.E., Paxton, R.J., 2016. Pollination services enhanced with urbanization despite increasing pollinator parasitism. Proc. R. Soc. B Biol. Sci. 283 20160561. Therneau, T.M., 2018. Coxme: Mixed Effects Cox Models. R Package Version 2.2-10. https://CRAN.R-project.org/package=coxme. Therneau, T.M., Lumley, T., 2017. A Package for Survival Analysis in R. R Package Version 2.41-3. https://CRAN.R-project.org/package=survival. Tritschler, M., Vollmann, J.J., Yañez, O., Chejanovsky, N., Crailsheim, K., Neumann, P., 2017. Protein nutrition governs within-host race of honey bee pathogens. Sci. Rep. 7, 14988. Tuck, S.L., Winqvist, C., Mota, F., Ahnstrom, J., Turnbull, L.A., Bengtsson, J., 2014. Landuse intensity and the effects of organic farming on biodiversity: a hierarchical metaanalysis. J. Appl. Ecol. 51, 746–755. USDA National Agriculture Statistics Service, 2016. Crop Data Layer. United States Department of Agriculture. Vaidya, C., Fisher, K., Vandermeer, J., 2018. Colony development and reproductive success of bumblebees in an urban gradient. Sustainability 10, 1936. Vanbergen, A.J., Baude, M., Biesmeijer, J.C., Britton, N.F., Brown, M.J.F., Brown, M., Bryden, J., Budge, G.E., Bull, J.C., Carvell, C., Challinor, A.J., Connolly, C.N., Evans, D.J., Feil, E.J., Garratt, M.P., Greco, M.K., Heard, M.S., Jansen, V.A.A., Keeling, M.J., Kunis, W.E., Marris, G.C., Memmott, J., Murray, J.T., Nicolson, S.W., Osborne, J.L., Paxton, R.J., Pirk, C.W.W., Polce, C., Potts, S.G., Priest, N.K., Raine, N.E., Roberts, S., Ryabov, E.V., Shafir, S., Shirley, M.D.F., Simpson, S.J., Stevenson, P.C., Stone, G.N., Termansen, M., Wright, G.A., Insect Pollinators, I, 2013. Threats to an ecosystem service: pressures on pollinators. Front. Ecol. Environ. 11, 251–259. Vogt, F.D., 1986. Thermoregulation in bumblebee colonies. II. Behavioral and demographic variation throughout the colony cycle. Physiol. Zool. 59, 60–68. Westphal, C., Steffan‐Dewenter, I., Tscharntke, T., 2009. Mass flowering oilseed rape improves early colony growth but not sexual reproduction of bumblebees. J. Appl. Ecol. 46, 187–193. Williams, N.M., Kremen, C., 2007. Resource distributions among habitats determine solitary bee offspring production in a mosaic landscape. Ecol. Appl. 17, 910–921. Williams, N.M., Regetz, J., Kremen, C., 2012. Landscape-scale resources promote colony growth but not reproductive performance of bumble bees. Ecology 93, 1049–1058. Williams, P.H., Osborne, J.L., 2009. Bumblebee vulnerability and conservation worldwide. Apidologie 40, 367–387. Williams, P.H., Thorp, R.W., Richardson, L.L., Colla, S.R., 2014. Bumble Bees of North America: an Identification Guide. Princeton University Press. Winfree, R., Bartomeus, I., Cariveau, D.P., 2011. Native pollinators in anthropogenic habitats. In: In: Futuyma, D.J., Shaffer, H.B., Simberloff, D. (Eds.), Annual Review of Ecology, Evolution, and Systematics Vol 42. pp. 1–22. Xian, G., Homer, C., Dewitz, J., Fry, J., Hossain, N., Wickham, J., 2011. The change of impervious surface area between 2001 and 2006 in the conterminous United States. Photogramm. Eng. Remote Sensing 77, 758–762. Youngsteadt, E., Appler, R.H., López-Uribe, M.M., Tarpy, D.R., Frank, S.D., 2015. Urbanization increases pathogen pressure on feral and managed honey bees. PLoS One 10, e0142031. Yourth, C.P., Brown, M.J.F., Schmid-Hempel, P., 2008. Effects of natal and novel Crithidia bombi (Trypanosomatidae) infections on Bombus terrestris hosts. Insectes Soc. 55, 86–90. Zanette, L., Miller, S.D.L., Faria, C., Bourke, A.F.G., 2014. Bumble bee workers drift to conspecific nests at field scales. Ecol. Entomol. 39, 347–354.

Soc. B Biol. Sci. 285 20181102. Graystock, P., Meeus, I., Smagghe, G., Goulson, D., Hughes, W.O.H., 2016. The effects of single and mixed infections of Apicystis bombi and deformed wing virus in Bombus terrestris. Parasitology 143, 358–365. Hamblin, A.L., Youngsteadt, E., Frank, S.D., 2018. Wild bee abundance declines with urban warming, regardless of floral density. Urban Ecosyst. 21, 419–428. Hatfield, R., Jepsen, S., Thorp, R., Richardson, L., Colla, S.R., 2014. Bombus Impatiens. The IUCN Red List of Threatened Species 2014: e.T44937797A69003246. https:// doi.org/10.2305/IUCN.UK.2014-3.RLTS.T44937797A69003246.en. Hernandez, J., Frankie, W., Thorp, R., 2009. Ecology of urban bees: a Review of current knowledge and directions for future study. Cities Environ. 2, 1–15. Holzschuh, A., Dudenhoffer, J.H., Tscharntke, T., 2012. Landscapes with wild bee habitats enhance pollination, fruit set and yield of sweet cherry. Biol. Conserv. 153, 101–107. Imhoof, B., Schmid-Hempel, P., 1999. Colony success of the bumble bee, Bombus terrestris, in relation to infections by two protozoan parasites, Crithidia bombi and Nosema bombi. Insectes Soc. 46, 233–238. Johnson, E.A., Klemens, M.W., 2005. Nature in Fragments: The Legacy of Sprawl. Columbia University Press. Johnson, J.B., Omland, K.S., 2004. Model selection in ecology and evolution. Trends Ecol. Evol. 19, 101–108. Joshi, N.K., Otieno, M., Rajotte, E.G., Fleischer, S.J., Biddinger, D.J., 2016. Proximity to woodland and landscape structure drives pollinator visitation in apple orchard ecosystem. Front. Ecol. Evol. 4, 1–9. Kammerer, M.A., Biddinger, D.J., Joshi, N.K., Rajotte, E.G., Mortensen, D.A., 2016. Modeling local spatial patterns of wild bee diversity in Pennsylvania apple orchards. Landsc. Ecol. 31, 2459–2469. Klein, A.M., Vaissiere, B.E., Cane, J.H., Steffan-Dewenter, I., Cunningham, S.A., Kremen, C., Tscharntke, T., 2007. Importance of pollinators in changing landscapes for world crops. Proc. R. Soc. B Biol. Sci. 274, 303–313. Kosior, A., Celary, W., Olejniczak, P., Fijal, J., Krol, W., Solarz, W., Plonka, P., 2007. The decline of the bumble bees and cuckoo bees (Hymenoptera : Apidae : Bombini) of Western and Central Europe. Oryx 41, 79–88. Lefebvre, D., Pierre, J., 2007. Demographic consequences of drift in contiguous hives of Bombus terrestris. Apicult. Soc. Insects 100, 1756–1763. Lenth, R., Love, J., Buerkner, P., Herve, M., 2018. Emmeans: Eatimated Marginal Means, Aka Least-squares Means. R Package Version 1.2.2. https://CRAN.R-project.org/ package=emmeans. McArt, S.H., Fersch, A.A., Milano, N.J., Truitt, L.L., Boroczky, K., 2017a. High pesticide risk to honey bees despite low focal crop pollen collection during pollination of a mass blooming crop. Sci. Rep. 7, 46554. McArt, S.H., Urbanowicz, C., McCoshum, S., Irwin, R.E., Adler, L.S., 2017b. Landscape predictors of pathogen prevalence and range contractions in US bumblebees. Proc. R. Soc. B Biol. Sci. 284, 20172181. McFrederick, Q.S., LeBuhn, G., 2006. Are urban parks refuges for bumble bees Bombus spp. (Hymenoptera : Apidae)? Biol. Conserv. 129, 372–382. McIntyre, N.E., Hostetler, M.E., 2001. Effects of urban land use on pollinator (Hymenoptera : Apoidea) communities in a desert metropolis. Basic Appl. Ecol. 2, 209–218. Michener, C., 2000. Bees of the World. Johns Hopkins University Press. Mullin, C.A., Frazier, M., Frazier, J.L., Ashcraft, S., Simonds, R., vanEngelsdorp, D., Pettis, J.S., 2010. High levels of miticides and agrochemicals in North American apiaries: implications for honey bee health. PLoS One 5, e9754. Osborne, J.L., Martin, A.P., Carreck, N.L., Swain, J.L., Knight, M.E., Goulson, D., Hale, R.J., Sanderson, R.A., 2008. Bumblebee flight distances in relation to the forage landscape. J. Anim. Ecol. 77, 406–415. Oyen, K.J., Giri, S., Dillon, M.E., 2016. Altitudinal variation in bumble bee (Bombus) critical thermal limits. J. Therm. Biol. 59, 52–57. Park, M., Blitzer, E.J., Gibbs, J., Losey, J., Danforth, B., 2015. Negative effects of pesticides on wild bee communities can be buffered by landscape context. Proc. R. Soc. B Biol. Sci. 282. https://doi.org/10.1098/rspb.2015.0299. Pettis, J.S., Lichtenberg, E.M., Andree, M., Stitzinger, J., Rose, R., Vanengelsdorp, D., 2013. Crop pollination exposes honey bees to pesticides which alters their susceptibility to the gut pathogen Nosema ceranae. PLoS One 8, e70182. Potts, S.G., Biesmeijer, J.C., Kremen, C., Neumann, P., Schweiger, O., Kunin, W.E., 2010. Global pollinator declines: trends, impacts and drivers. Trends Ecol. Evol. 25, 345–353. R Core Team, 2017. R: a Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Richardson, L.L., Adler, L.S., Leonard, A.S., Andicoechea, J., Regan, K.H., Anthony, W.E., Manson, J.S., Irwin, R.E., 2015. Secondary metabolites in floral nectar reduce parasite infections in bumblebees. Proc. R. Soc. B Biol. Sci. 282, 20142471. Rundlöf, M., Andersson, G.K.S., Bommarco, R., Fries, I., Hederström, V., Herbertsson, L., Jonsson, O., Klatt, B.K., Pedersen, T.R., Yourstone, J., Smith, H.G., 2015. Seed coating with a neonicotinoid insecticide negatively affects wild bees. Nature 521, 77–U162. Rundlöf, M., Nilsson, H., Smith, H.G., 2008. Interacting effects of farming practice and landscape context on bumblebees. Biol. Conserv. 141, 417–426.

10