International Journal of Entomology and Nematology IJEN Vol. 2(1), pp. 027-041, September, 2016. © www.premierpublishers.org.ISSN:3241-6423

Re search Article

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA

*Francis A. Drummond

*School of Biology and Ecology, 305 Deering Hall, University of Maine, Orono, Maine USA 04469 Email: [email protected], Tel.: 207 581-2989, Fax: 207 581-2969

Greenhouse and field studies were conducted between 1996 and 2014 in Maine to assess the behavior of selected bee taxa that visit wild blueberry, (Ericaceae) Aiton. Some of my findings are as follows. When individual foraging efficiency was assessed on wild blueberry for four common bee pollinators, bumble bees were most efficient and honey bees were the least efficient in terms of the number of pollen grains deposited on a stigma in a single visit (P< 0.0001). However, I also found that the prior bumble bee visitation to flowers enhanced the pollination efficacy of honey bees. Field observations suggested that bumble bees recruit to plants with higher floral density and that bumble bees and andrenids forage for longer periods of time in the day than sweat bees and Osmia leaf cutting bees; honey bees showed intermediate foraging durations. Honey bees and solitary native bees were found to forage at an increasing rate with increasing air temperature, while bumble bee queens tended to forage independently of air temperature. Foraging patterns among the following bee taxa such as bumble bees, andrenids, megachilids, and honey bees also varied and the implications of these differing foraging patterns relative to pollination are discussed.

Keywords: Bees, wild blueberry, Vaccinium angustifolium, pollination efficiency, flower handling time, pollen deposition, foraging pattern

INTRODUCTION

Wild lowbush blueberry, Vaccinium angustifolium Hedrick,1919) and later introduced to European colonists (Ericaceae) Aiton, is the largest (in terms of land area) (Davis, 1993). This tradition of clearing forest followed by managed native wild fruit crop grown in North America frequent burning of the landscape to reduce plant (Jones et al.; 2014). Wild blueberry naturally occurs from competition with wild blueberry is still practiced by several the mountains of Virginia and West Virginia to the farmers today (Rose et al., 2014). Canadian Maritimes and Quebec west to Ontario and Michigan (Vander Kloet, 1988, Jones et al., 2014). This Currently, Maine is the largest producer of wild plant species generally grows in acidic, sandy soils and is blueberry in the world with more than 24,300 hectares an early colonizer of disturbed areas (Hall et al., 1979; under cultivation (Yarborough, 2009). pollination is Jones et al., 2014). It is a natural occurring understory critical for the set of wild blueberry flowers and fruit plant species in Northern Mixed Hardwood, Acadian, and development (Lee, 1958; Vander Kloet, 1988; Bell et al., Boreal Forests (Jones et al., 2014) and a valuable food 2009; Asare, 2013). The morphology of the flower with resource for wildlife (Martin et al., 1951; Eaton, 1957). poricidal anthers limits the species of that can Management or cultivation of wild blueberry was first effectively pollinate wild blueberry (Ritzinger and Lyrene, practiced by Native Americans (Munson, 1901; 1999; Bell et al., 2009). There are more than 120 native

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA Drummond FA. 028

bee species that are associated with wild blueberry range of both native and exotic bee species with regards landscapes and pollination (Boulanger et al., 1967; to wild blueberry was studied in Maine. We measured Finnamore and Neary, 1978; Bushmann and Drummond, and quantified: 1) selection choice of bumble bees when 2015). Many wild blueberry growers in Maine heavily rely foraging among wild blueberry flowers with regards to on honey bees for pollination (Hanes et al., 2013); flower age, size or previous visitation history, 2) the effect however, there are still several growers that rely solely of flower number on stems and if this affects the bumble upon native bees for their pollination needs (Rose et al.; bee visitation choice of those stems, 3) attractiveness of 2014). There has been a concerted effort to conserve five co-flowering species during wild blueberry bloom, 4) native bees in wild blueberry habitats by identifying their bee species-specific pollen deposition on wild blueberry floral resources, providing nests, and identifying threats flower stigmas during single visits to naive flowers, 5) bee from pesticide applications and pathogens (Stubbs et al., species-specific handling time of flowers, 6) effects of 1992; Stubbs et al., 1997; Stubbs and Drummond, 1998; sequential multiple flower visits on pollen deposition by 1999; Bushmann et al., 2012; Drummond 2012b; Groff et more than one species of bee, and 7) bee-specific spatial al., 2016). The importance of wild native bee pollinators foraging patterns in wild blueberry fields during bloom. in wild blueberry production reflects the development of field techniques for assessing bee populations by researchers and blueberry growers (Drummond and MATERIALS AND METHODS Stubbs, 1997b; Drummond et al., 2015). Research on the efficacy of honeybee pollination began at the This study was conducted over a period of twenty University of Maine in the 1950's (Boulanger unpublished years (1996-2015) in the greenhouse, laboratory, and data; Boulanger et al., 1967). The importation of wild blueberry fields in Maine, USA. All cage studies were commercial honeybee colonies has grown from conducted at the University of Maine in Orono, Maine approximately 500 hives per year in the mid-1960's to (Penobscot county). Field studies were conducted in approximately 60,000 hives imported for the blueberry fifteen managed wild blueberry fields in Hancock, Lincoln, blooming season in the year 2000 (Drummond, 2002) Waldo, and Washington counties, Maine. Seven and more than 85,000 in 2016 (Jadczak, Maine state experiments grouped under three studies were apiculture inspector, personal communication), an conducted during this period. Methods for each increase that reflects increasing numbers of honeybee experiment are described below and a schematic (Fig. 1) hives per acre (Yarborough, 2013). In addition, the use has been included to provide an outline of the studies of commercial bumblebees (Bombus impatiens Cresson) and experiments. has become popular over the past 15 years with approximately 2,300 colonies being purchased per year Bee Preference Studies for wild blueberry pollination in Maine (Stubbs and Drummond, 2001; Desjardins and Olivereira, 2006; Blueberry Flower Selection Choice By Bumble Drummond, 2012b). In addition, several other potential Bees, Flower Size And Age – During March – April 2007 commercial bees have been evaluated for wild blueberry and then in March – April 2008, the effects of wild pollination (Stubbs et al., 1994; Drummond and Stubbs, blueberry flower age, corolla length (mm) and corolla 1997a; Stubbs and Drummond, 1997a; 1997b; 1997c; diameter (mm) on bumble bee, Bombus impatiens (Say), 2000). Given this reliance on pollination by commercial choice measured by visitation frequency was studied in bees, the wild blueberry production is vulnerable to the greenhouse. This study was conducted in three disruption by lack of pollination services. However, replicate 2 m x 2 m x 3.5 m mesh cage in the University despite the increasing reliance on commercial bee of Maine Clapp greenhouse in Orono, Maine, USA. Three pollinators in wild blueberry, native bees are still bumble bee colonies (100 worker strength) were significant in determining the fruit set and yield (Asare, purchased from Koppert Inc. (Romulus, Michigan, USA) 2013; Yarborough et al., 2016). Indeed, U.S. consumers and placed in the flight cage for several days without are willing to pay more for blueberries pollinated by native flowers. The colonies derived their nutritional bees (Stevens et al., 2015). requirements on sugar syrup provided with the colonies However, there have been few studies on their upon purchase. During the previous autumn, 12 clones efficacy as pollinators of wild blueberry. Whidden (1996) (genets) of Vaccinium angustifolium Aiton blueberry sods and Drummond (2012b) both showed how bumble bees were cut and excavated from blueberry fields in are floral constant with respect to wild blueberries and Jonesboro, Maine, USA and placed in 50.8 x 40.6 cm only Javorek et al. (2002) addressed the performance of plastic tote boxes. The boxes were kept in a walk-in native bees as pollinators and showed that the efficacy of refrigerator (2-4ºC) over the winter. In late February, the native bees on a per bee basis is much greater than that blueberry boxes were brought into the greenhouse of honey bees. (maintained at 15-22º C) and allowed to mature flowers. Because of this lack of knowledge about native Upon the bloom initiation, the flowers were measured bee foraging in wild blueberry, the pollination efficacy of a with a micrometer, dated, marked, and coded. In the

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA Int. J. Entomol. Nematol. 028

Figure 1. Outline of experiments conducted between the years 1996 and 2015, bee preference studies (A) and pollination efficacy and foraging studies (B). Cage studies are surrounded by dashed line boxes and open field studies are surrounded by solid line boxes.

early morning prior to an assay, boxes of blueberry in bees. The stems were placed in a test tube holder such bloom, not used for experimentation, were placed into the that stems were 23 cm apart. Flowers were observed cages, bumble bee hives were opened and access to until a bumble bee landed on the flower and began to floral resources were allowed. Immediately prior to the extract either nectar or pollen. At this point the visited start of a choice assay, all blueberry boxes were removed flower was recorded. and two stems with a single open flower on each Blueberry Flower Selection Choice By Bumble (standardized by either flower age (1-8 days old), corolla Bees, Previously Visited Flowers - In 2008, one choice length, or diameter) were presented to foraging bumble bioassay was conducted to determine if flowers that had

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA Drummond FA. 029

not previously been visited by a bee compared to similar each floral species in a five-minute period of observation. age and size flowers that had been visited by a bee were The pair of stems was removed from the cage after each differentiated by foragers. Sixteen replicates of paired assay and a new pair of stems was selected. At least ten flowers categorized by presence or absence of previous replicates for each pair of co-flowering and blueberry visitation were run in the exact same manner as stems were conducted. After the trials, the flowers that described above for the flower size and age assays. A had been presented in each vial were photographed and non-choice bioassay using visited flowers from the scanned into a computer to measure the amount of area previously described bioassays was used to determine if associated with each presented flower blossom. previously visited flowers (ranging from 1-6 visits) Analyses of covariance (ANCOVA, covariate was area of affected the flower choice by bumble bee foragers. Fifty- floral display; JMP®, 2015) was used to determine if the three replicate trials were used in this bioassay. In all total time bumble bees spent foraging on wild blueberry bioassays conducted over both years, if no visitation was significantly greater than the time spent on the other occurred in 15 minutes then the trial was terminated and flowering shrubs during the 5-minute assay. Nominal new pairs of stems were introduced. A total of 106 choice logistic regression (JMP®, 2015) was used to assess assays were conducted over the two-year study. Nominal differences in initial choice in visitation of blueberry logistic regression was used to test if differences among compared to other co-flowering plant species. years, flower age, corolla length, and previous visitation Recruitment of Foraging Honey Bees, influenced flower selection for first visit (JMP®, 2015). Commercial Bumble Bees And Native Wild Bees To Choice For Blueberry Flowers Compared To Five Clone Flower Density In The Field - Two studies were Co-Flowering Native Shrubs - This study was performed conducted at three locations in Maine in 2008. The first on 21 and 22 May, 2008. Flowering shrubs that bloom study was conducted in Union and Winterport, Maine. during wild blueberry bloom in Maine that are often found The fields were visited during bloom on 14 and 15 May. growing around the field edges of wild blueberry fields. Ten clones (genets) in a blueberry field in Union were We selected five of the most common co-flowering randomly located and flagged and 24 clones in a shrubs in Downeast Maine to test the preference of B. blueberry field in Winterport were located and flagged. In impatiens foragers for blueberry over the five co-flowering each clone 30 stems were arbitrarily selected and the shrubs. The species selected for study were shadbush, total number of flowers (pre-bloom and blooming) on Amelanchia spp.; apple, Malus pumila; rhodora, each stem was counted. This provided a mean floral Rhododendron canadense; choke cherry, Prunus density for each clone. A square meter quadrat was virginiana; and bunch berry, Cornus canadensis. Several marked out with string in each clone in both fields. Bee flowering stems of these species were collected and each observations were made in each clone for three minutes was tested against the wild blueberry stems in flower and the number of bumble bee queens, native wild bees collected at the same time. A paired choice bioassay in other than bumble bees, and honey bees visiting each the same manner as discussed in the former experiment quadrat during the three minute period were recorded in a 2 x 2 x 3 m mesh flight cage was setup in a over the two day period. Linear general models (JMP®, greenhouse at the University of Maine, Orono, Maine. 2015) were used to determine if the bees recruit more Commercial bumble bees, B. impatiens were purchased heavily to clones with higher floral density. prior to the experiment (Koppert, Inc.) and used in all The second study was designed to assess the assays. All paired choice bioassays consisted of a vase flowering stem selection by commercial bumble bees as of cut flowering wild blueberry stems and a vase of cut a function of total and open flowers. During bloom (May) flowering stems from one of the five co-flowering in 2008, at the University of Maine blueberry research shrubs.Pairs of cut stems were exposed to foraging farm, Blueberry Hill in Jonesboro, Maine a field study was commercial worker bumble bees (B. impatiens) for five conducted to determine the frequency of visitation by B. consecutive trials. For this experiment, since the age of impatiens workers to wild blueberry stems with opened the gathered blossoms could not be determined, the age and non-opened flowers. Prior to bloom 20 commercial of the blueberry blossoms was not taken into account. It bumble bee quads (Koppert, Inc.) were placed in one of was important to keep the floral displays between the the blueberry fields (ca. 8 ha). The density of bumble blueberries and the gathered blossoms in the same bees from this stocking density of commercial bumble volume to ensure bees would not be attracted to a larger bees is approximately 3,500-4,000 workers. Thirty-six blossom display, ensuring each flower species an equal stems were each marked by a thread tied about the stem opportunity to be chosen. For example, the apple base. The stems were apportioned to six clones (6 blossoms, being rather large were paired with several marked stems in each clone). During ten observation blueberry stem cuttings so the overall total blossom area times throughout 12-15 May, each stem was observed for was similar for each flower species. Data recorded for 2 minutes and the number of B. impatiens workers each assay was the first floral species visited in the visiting flowers on each stem was recorded. A general paired presentation and the amount of time spent on linear model was used to determine if B. impatiens

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA Int. J. Entomol. Nematol. 030

worker visitation to stems was related to clone, total floral hoc multiple comparison tests were used to compare all density, or solely open floral density (JMP®, 2015). means with one another (JMP®, 2015).

Bee Pollination Efficacy Studies Sequential Floral Visits - In May 2014, a similar flight cage study to that described above was used to During May and June of 1996 a preliminary study assess potential interactions in pollen deposition between was conducted to assess individual bee pollination honey bees and bumble bees when flowers were visited efficacy in flight cages. This study enabled us to refine twice by bees. The goal of this study was to determine if methods to repeat a study with higher replication, using pollination synergy occurs when bumble bees and honey the modified methods in May and June of 2006, and May bees both forage on blueberry flowers sequentially 2014.These studies were conducted using field cage compared to only one of the two bee species visiting studies in both the greenhouse and field to determine the flowers sequentially. efficacy of common bee pollinators of wild blueberry. This study was conducted using two mesh flight Single Floral Visits - The bee species used in the cages in the greenhouse (see above). The bee species studies were the bumble bee, Bombus impatiens used in the study were the bumble bee, Bombus (Apidae) Cresson; the leaf cutting bee, Osmia atriventris impatiens Cresson and the honey bee Apis mellifera (L.). (Megachilidae) Cresson; the digging bee, carlini Bumble bee colonies or honey bee colonies were placed () Cockerell, and the honey bee Apis mellifera in cages with 1-4 totes of flowering wild blueberries to (Apidae) (L.). Studies conducted during bloom (mid-May allow foraging to take place and pollen acquisition by to mid-June) in 1997 and 2006 were aimed at each of the bee species. Only a single bee was allowed determining flower handling time and the number of to forage at a time. Stems were protected from foraging pollen grains placed on a new previously non-visited bees with mesh bags as described above and marked stigma in a wild blueberry (V. angustifolium) flower. The when flowers opened. Once bees were observed actively study was conducted in mesh flight cages either placed in foraging and pollinating non-bagged flowers on stems in a non-flower bearing field over Andrena carlini nest the totes placed in the cages, bags were removed from aggregations or in the greenhouse (see study 1 above). stems with 1-5 day old flowers and the cages were Bumble bee colonies, honey bee colonies, or Osmia removed of all bees. Individual new bee visits of each atriventris nest blocks with active females provisioning bee species (only a single bee species per cage) to cells, were placed in cages with 1-4 totes of flowering newly un-bagged flowers were recorded. As soon as a wild blueberries to allow foraging to take place and pollen visiting bee left the first un-bagged stem (flower A), the acquisition by each of the bee species. Several of the bee was allowed to visit a second un-bagged stem stems in each tote (5-20) were covered with fine mesh (flower B). The second visited un-bagged stem (flower B) bags to prevent bee access to the flowers. Flowers on was excised and placed in a cooler until it was brought to each bagged stem were marked and the day that each the lab. The bee in the cage was discarded and the stem flower opened was recorded. Once bees were observed with the first visited flower (flower A) was excised and put actively foraging and pollinating non-bagged flowers on into a test tube filled with water and transported to the stems in the totes placed in the cages, the bags were flight cage with the other bee species. A new bee in this removed from stems with 1-5 day old flowers. Bee visits flight cage was allowed to visit the flower on the stem of each bee species (only a single bee species per cage) (flower A) that was previously visited in the other cage by to newly un-bagged flowers were recorded. During the other species. After the visit to flower A in the other visitation to each un-bagged flower the handling time cage the stem (flower A) was tagged and placed in a (from initial contact of the flower until the time when the cooler. Immediately after a bout of visitations, stems bee left the flower) was measured with a stopwatch. As placed in the cooler were brought immediately to the soon as the visiting bee left the flower, the stem was put laboratory where the stigma was stained and inspected in an ice filled cooler for temporary storage until the end under the microscope for V. angustifolium terad pollen of the day’s assay. All stems with marked visited flowers grains as described above. Twenty replicates of each of were taken immediately to the laboratory where the the four treatments were performed: 1. single visit by stigma was stained (Alexander stain, Alexander 1980) honey bee that had previously visited a flower; 2. single and inspected under the microscope for V. angustifolium visit by bumble bee that had previously visited a flower; 3. tetrad pollen grains. Twenty-five replicate recordings of A single visit by a bumble bee that had just previously flower handling time and pollen deposition were made for visited a flower visited by a honey bee; and 4. A single each species in 2006. visit by a honey bee that had just previously visited a One-way analyses of variance (Welch’s adjusted flower visited by a bumble bee. One-way analysis of F-test for unequal variances, complete randomized variance (Welch’s adjusted F-test for unequal variances, design; JMP®, 2015) were used to determine if complete randomized design; JMP®, 2015) was used to differences due to bee species existed when considering determine if differences in pollen deposition occurred handling time and pollen deposition per visit. Tukey post- between treatments. A Tukey post-hoc multiple

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA Drummond FA. 031

comparison test was used to compare all means with one distributions of distances among visited stems, cardinal another (JMP®, 2015). direction between stems, flowers, number of flowers visited per stem, and stems visited per bout were Foraging Studies constructed and fit to theoretical probability distributions for each bee taxon group observed (JMP®, 2015). Diurnal Foraging Patterns Of Bees In Wild Blueberry And Response To Air Temperature - On May 21-31, 2001, observations of bee activity were made in a RESULTS wild blueberry field in Winterport, Maine during peak bloom. Bees were sampled with a 30.5 cm diam. Study sweepnet. A total of 10-50 locations in the field were sampled. At each sampling location ten 180º sweeps in Bee Preference Studies the blooming crop were conducted. Sampling was conducted every 2-3 hrs starting at 6 AM, shortly after Blueberry Flower Selection Choice By Bumble sunrise, and ending at dusk, 8 PM. Sampling was Bees, Flower Size And Age - Over both years (2007 and conducted each day between 21 may and 31 May. On 2008) wild blueberry flower corolla length and diameter most sampling periods, the air temperature was recorded were highly correlated (r = +0.527, n=90, P< 0.0001). on-site with a handheld weather data logger (Kestrel®). Flower age was not correlated with either corolla length Bees were classified in the field into taxonomic groups or diameter (P> 0.05). Because of this only year, corolla and their numbers recorded. The groups were the length and flower age were modeled to determine if following: honey bees, bumble bees (Bombus spp.), bumble bees select flowers to visit in choice bioassays. digger bees (Andrena spp.), sweat bees (Halictidae), and Bumble bees appear to demonstrate preference for 2 leaf cutting and mason bees (Megachilidae, almost longer wild blueberry flowers (x (1) = 7.221, P = 0.007, entirely Osmia spp.). Sweepnet capture of each taxon Figure 2), but they did not discriminate for flower age (P> grouping was tallied for each sampling time and the 0.05) when the choice was made between flowers that percent of relative capture by sampling time was differed in age up to 6 days. However, a non-significant calculated and plotted. Visual inspection of the data in a trend was found, which suggests a tendency to select graph was used to draw conclusions about the diurnal younger flowers (P = 0.106). Year and the interaction activity of the different groups of bee foragers. Linear between corolla length and flower age was also not regression analysis was used to assess the relationship significant. The odds ratio for the significant corolla length between air temperature and the percent of the honey effect suggests that for every difference in 1.0 mm length bee, bumble bee queen, and native solitary bee (Andrena between flowers that it is 2.3 times more likely that the spp. + Megachilidae + Halictidae) population for each day larger flower will be first selected by a bumble bee that was foraging. forager. Bee Spatial Foraging Patterns Within And Among Blueberry Flower Selection Choice By Bumble Clones - During wild blueberry bloom (mid-May to mid- Bees, Previously Visited Flowers - The trials that focused June) in 2003, 2004, 2005, 2006, 2007, 2008, 2009, and on previously visited flowers compared to flowers that 2010; the spatial movement patterns of foraging honey had never been by a bumble bee showed no evidence of bees and native wild bees were delineated and recorded. choice (P = 0.715) and the trials assessing visitation to The blueberry fields were located in Union (Knox Co.), flowers that had been previously visited from 1 to 6 times Winterport (Waldo Co.), and Columbia, Jonesboro, T-18, showed no evidence of an effect on bumble bee foragers and T-19 (Washington Co.), Maine. Using a digital audio (P = 0.363). recorder, we assessed the distances between stems that Recruitment Of Foraging Honey Bees, were visited by foraging bees, distances between visited Commercial Bumble Bees And Native Wild bees To flowers within a stem, the number of flowers visited per Clone Flower Density In The Field - I found that floral stem by bees, and the cardinal direction bees flew when density per stem (both open and non-opened flowers) in moving from one stem to the next. The bees observed clones did determine the recruitment of honey bees, and mapped in this study were honey bees, (Apis bumble bee queens, and native bees other than bumble mellifera), Bombus spp. (mostly B. ternarius), Andrena bees, but this effect was dependent upon site (Table spp., and Osmia spp. Data was recorded on individual 1).With regards to the entire foraging bee community (all bees until the bee was lost among the blueberry foliage bee taxa), bees recruited to clones with higher total or it left on a long distance flight where it was not possible flower density. There was an effect of bee density due to to follow it any longer. At this point the foraging bout was site which is not unexpected as bee densities, especially marked as terminated and another bee was selected for native bees, vary by an order of magnitude from field to observation. Care was taken not to disturb the bee by field and geographic region to region. observers maintaining a one-meter distance from the In the field,the frequency at which stems were foraging bee, minimizing casting shadows on the foragers, selected by foraging bumble bees in Jonesboro, Maine and moving slowly while observing foragers. Frequency was not influenced by clone (genet; P = 0.178), nor the

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA Int. J. Entomol. Nematol. 032

Figure 2. Modeled probability of a bumble bee forager selecting a flower when given a choice of two flowers. As number become more negative on x-axis the flower chosen was smaller compared to larger flower (mm), as number become more positive, the flower chosen was larger compared to smaller flower (mm).

Table 1. Recruitment of foraging bees to flower density / stem in a clone.

Bee taxa group Significant Factors F-test Proportion variance explained, r2 Total bee community site F(1,30) = 5.407, P=0.027 0.584* (0.418)** flower density F(1,30) = 12.628, P=0.001 Bumble bee queens site F(1,30) = 5.707, P=0.023 0.331 (0.224) flower density F(1,30) = 6.221, P=0.020 Native bees (other flower density F(1,30) = 6.101, P=0.019 0.364 than bumble bees) Honey bees site F(1,30) = 14.716, P=0.0006 0.512 (0.350) flower density F(1,30) = 19.864, P=0.0001 site x flower density F(1,30) = 17.484, P=0.0002

* proportion variance explained by entire model ** proportion variance explained by flower density, based upon sequential sums of squares number of open flowers (P = 0.417) on the stem. The 0.0002) and time spent foraging on flowers (F(4,111) = initial choice to land on the stem was significantly 2.568, P = 0.042; Figure 3). Comparing the ratio of initial determined by the number of total flowers (open and non- visitation to blueberry vs. the other species (t-test open flowers: F(1,28) = 4.292, P = 0.0476), but only 12% of comparing percent choice of blueberry to 50%, no the variance in the frequency of visitation to stems was difference in species choice) and relative time on flowers explained by the total number of flowers per stem. of blueberry compared to the other species (t-test Choice For Blueberry Flowers Compared To Five comparing relative time to 1.0, no difference in relative Co-Flowering Native Shrubs - Bumble bee foragers do time on two species), it can be seen that blueberry is appear to prefer wild blueberry flowers to most of the co- highly attractive to bumble bee foragers compared to the 2 flowering species both in % first visits (x (4) = 21.996, P = other five species (Figure 3). The only co-flowering

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA Drummond FA. 033

Figure 3. Preference of Bombus impatiens workers (stipled bar) for species other than blueberry (measured in paired bioassays as the percentage of first landings on blueberry vs the other species). In addition, the relative time spent on blueberry flowers relative to the co-flower (hatched bar). The red dashed line indicates a region where foragers spent the same time on blueberry and the co- flowering plant and the blue dotted line indicates the region of no-preference in initial floral choice between blueberry and a co- flowering plant species. Asterisks indicate significant differences in 1) time spent on the co-flower relative to blueberry and 2) preference for choice of initial flower selection.

species that appeared to be as attractive to bumble bees mixture of both bee species, we found that honey bees as blueberry in initial visitation were apple and choke visiting a flower previously visited by a bumble bee cherry, but even these two species had 3-4 times the significantly (F(3,33.8) = 19.654, P< 0.0001), increased the foraging effort on blueberry once the initial flower was amount of pollen deposited on the stigma of the selected. There were no visits to bunch berry for any of subsequent flower visited. This pattern was not seen for the assays. Rhodora was almost ignored when first initial bumble bees visiting previously visited flowers by bumble visits were considered, but the amount of time that bees or honey bees (Figure 5). bumble bees spent foraging on this species was equivalent to blueberry. Foraging Studies

Bee Pollination Efficacy Studies Diurnal Foraging Patterns Of Bees In Wild Blueberry - Figure 6 shows that during the peak bloom, Single Floral Visits - Efficacy in pollinating wild honey bees do not start foraging until mid-morning and blueberry, on a per bee basis, varied by species. This they terminate foraging by early evening. The was the case for pollen deposition on floral stigmas after megachilids (mostly Osmia spp. during this study) were single visits (F(3,43.8) = 26.977, P< 0.0001) and for similar to honey bees in their foraging periodicity. The handling time of flowers for each visit (F(3,50.2) = 9.377, P< halictids had a shorter foraging period, starting in the mid- 0.0001). Figure 4 shows that pollen deposition on the morning and ending by mid-afternoon. The bumble bees blueberry flower stigma was significantly greater after a started the earliest of the foraging bee community and single visit by B. impatiens and O. atriventris, and least continued until dark. The digger bees (Andrena spp.) also by the honey bee, A. mellifera. The amount of time spent started foraging around dawn, but they finished foraging handling the flower was greatest for the honey bee and by early evening. least for the bumble bee and leaf cutting bee. When one Response To Air Temperature - Air temperature considers both pollen deposition / visit and handling time, was often found to be a major determinant of bee the bumble bee and Osmia leaf cutting bee were the foraging, however, light intensity, precipitation and wind most efficient pollinators tested on wild blueberry. can also contribute. We found that honey bees and native Javorek et al. (2002) showed similar results of bee solitary bees as groups responded in their foraging pollination efficiency with several of the same taxa activity to increasing air temperatures (F(1,26) = 22.628, associated with wild blueberry. P<0.0001, and F(1,26) = 11.841, P< 0.0001; solitary native Sequential Floral Visits - When sequential visits bees and honey bees, respectively). The proportion of were assessed by honey bees, bumble bees, or a the variance in foraging activity for solitary native bees

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA Int. J. Entomol. Nematol. 034

Figure 4. Average pollen deposition on floral stigmas after a single visit and flower handling time by common wild blueberry pollinators: bumble bees, B. impatiens, digger bees, Andrena carlini, leaf cutting bees, O. atriventris, and honey bees, A. mellifera. Bars with the same letters (uppercase – handling time, lowercase – pollen deposition) are not significantly different, Tukey’s test, experiment-wise error rate at P< 0.05.

Figure 5. Pollen deposition on a floral stigma by a bumble bee visiting a flower previously visited by a bumble bee and then visiting a new flower (BB/BB), or by a bumble bee visiting a flower previously visited by a honey bee and then visiting a new flower (HB/BB), or by a honey bee visiting a flower previously visited by a honey bee and then visiting a new flower, or by a honey bee visiting a flower previously visited by a bumble bee and then visiting a new flower (BB/HB). Bars with the same letters are not significantly different, Tukey’s test, experiment-wise error rate at P< 0.05.

and honey bees was 44.1 and 79.1%, respectively. It can temperature (P> 0.05). This can be expected due to the be seen in Figure 7 that the threshold for foraging of thermoregulation exhibited by bumble bees, especially native solitary bees was approximately 8ºC and that for queens (Heinrich, 1972a; 1972b; 1972c). honey bees was approximately 11ºC. Bumble bee queen Bee Spatial Foraging Patterns Within And Among foraging did not appear to respond to increases in air Clones - The foraging bee taxa that we observed had

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA Drummond FA. 035

Figure 6. Diurnal foraging periodicity of honey bees and native bees, 20-31 May, 2001 in Winterport, Maine.

Figure 7. Relationship between air temperature at time of sampling and the percent of a specific taxon group’s population (native solitary bees, bumble bee queens, and honey bees) for a given day.

very different spatial patterns (Table 2). For instance, the visited per stem), Osmia spp. (mean = 2.2 flowers / stem, cardinal direction that a forager took upon leaving a stem range = 1-7 flowers visited / stem), and Andrena spp. to move to the subsequent stem was fairly random and (mean = 2.7 flowers / stem, range = 1-10 flowers visited / non-directional for bumble bees and honey bees, but stem). The average distance that honey bees moved almost unidirectional as if bees were following a trap line between stems for foraging was 0.1 m, while Bombus for Andrena spp. The number of flowers per stem visited spp. was 0.3 m, Osmia spp. was 0.4 m, and Andrena spp. were low for honey bees (mean = 1.6 flowers / stem, was 1.7 m. An example of the distribution of flowers range 1-16 flowers visited per stem) compared to bumble visited per stem between honey bees and Andrena spp. bees (mean = 2.5 flowers / stem, ra nge = 1-27 flowers is shown in Figure 8.

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA Int. J. Entomol. Nematol. 036

Table 2. Probability density function fit to the frequency distributions of the number of stems per bout, distance between stems (m), cardinal direction to the next stem, and flowers visited per stem for four taxa of common bee pollinators of wild blueberry in Maine.

bee taxon sample size measure distribution parameters goodness of group of measure fit* Osmia 47 stems / bout Gamma  0.118 (P) spp. Poisson  45 distance between stems Gamma  0.073 (CM) (m)  38 cardinal direction Gamma  0.708 (P) Poisson  87 flowers visited / stem Poisson  0.351 (P)  Andrena 15 stems / bout Exponential  0.150 (K) spp. 25 distance between stems Weibull  0.779 (LR) (m)  24 cardinal direction Gamma  0.364 (K) Poisson  81 flowers visited / stem Exponential  0.999 (LR) Bombus 173 stems / bout Log Normal u = 2.767 0.10 (D) spp.  = 1.061 2508 distance between stems Weibull  0.994 (LR) (m)  2508 cardinal direction uniform na** na 2536 flowers visited / stem Beta  0.204 (K)    Honey 610 stems / bout Beta  0.251 (K) bee, A.  mellifera   4509 distance between stems Exponential  0.298 (LR) (m) 4903 cardinal direction uniform na na 5333 flowers visited / stem Poisson  0.999 (P)

* probability value for goodness of fit to theoretical distribution, (test statistic) = P: Pearson chisquare, CM: Cramer von Mises, K: Kolmogorov’s D, LR: adequacy LR test. ** na: not applicable.

DISCUSSION level and they make choices between competing flowering shrub species and wild blueberry Over 120 species of bees have been found to be flowers.Bumble bees (B. impatiens) apparently choose associated with wild blueberry in Maine (Bushmann and flowers according to size, the longer (or wider, as the two Drummond, 2015). Little is known, however, about their traits are highly correlated)were preferred. Blueberry behavior and activity during the blooming season. The flowers reflect UV light (Schaefer et al., 2004) and is collection of studies, reported here, begin to shed light on attractive to several species of bees (Kevin and Baker, how individual bee species or groups of species play a 1983); thus, it is not surprising that such a preference role in wild blueberry pollination. takes place during foraging. It was surprising to see that There are several major areas that I would like to address flower age did not affect the choice in foraging bumble in this discussion. bees. Despite intact floral resources and most likely not First of all, it is apparent that bees, as shown by lower than younger flowers, it has been demonstrated our work with bumble bees, make choices both at the that wild blueberry flowers lose their receptivity to pollen blueberry flower level, but also at the stem floral density over a 8-10 day period with maximal receptivity only

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA Drummond FA. 037

Figure 8. Frequency distribution for distance between stems moved by honey bees (a) and Andrena spp. (b).

lasting for 4-5 days (Drummond, unpublished data). It has an advantage to foraging bees as high floral density been shown that as many flowers age and start to should minimize the energy output per unit energy senesce they change in color and become less attractive acquired from flowers per unit of time (Waddington et al., to bees (Gori, 1989). This could be interpreted as a plant 1979). signaling phenomenon, beneficial to the plant and the The last of my findings relative to flower choice bees (Casper and La Pine, 1984). I found a non- by bumble bees was contrary to the findings of previous significant trend (P = 0.106), which suggests that studies (Goulson et al., 1998; Stout et al., 1998). It has rejection of older resource depleted flowers by bees been observed in other studies that bumble bees tend not might be occurring among a background of relatively rich to visit flowers that have previously been visited as nectar and pollen resources. In most bee pollinated crops frequently. It is often suggested that several species of the choice of visitation may not operate in the field since bumble bees mark flowers when visiting them and this most crops are monocultures of a single genotype, unlike chemical signal is thought to communicate a lower level wild blueberries (Bell et al., 2009). of floral resources to subsequent bees potentially Wild blueberry is a mass flowering crop that decreasing the chance that they will visit a flower produces up to 8,000 flowers / m2 (Jones et al., 2014). previously visited and marked (Cameron, 1981). In two Flower production per stem can vary greatly from clone different experiments we did not find any evidence that (genet) to clone (Bell, 2009). Therefore, despite a suggests this is happening with bumble bees visiting wild managed field being a continuous “carpet” of plants or blueberry flowers multiple times. Wild blueberry flowers clones, flower density varies spatially across fields. We each have on average 60-70 ovules (Bell et al., 2012a). found that bumble bees, and we hypothesize that this is Flowers may need to be visited several times by bees in the case with other bee taxa, recruit preferentially to order to result in enough ovules fertilized by deposited clones with a higher density of flowers. The proportion of pollen grains for fruit to be retained by the plant. Thus variance in bumble bee foragers recruiting to clones was mechanisms that would lower the chance of multiple high (r2 = 0.418), which suggests that there may be a visits might reduce fruit production. However, this is selective advantage to individual plants or genotypes to probably not the case with bumble bee queens, probably produce as many flowers per stem as possible as has the most efficient pollinator of wild blueberry (see below). been suggested in other mass flowering plant species Bushmann and Drummond (2015) demonstrated (Augspurger, 1980). One would also presume that this is that there is great floral constancy in native bees in wild

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA Int. J. Entomol. Nematol. 038

blueberry fields. We had hypothesized that this might be residue from the inside of a flower to the stigma. This is a due to the ratio of wild blueberry flowers in a location conjecture that will need to be tested. compared to other flowering plants that might compete for bees adjacent to where blueberries grow. However, Another aspect of pollination efficiency might be our cage study suggests that at least with five species of the spatial pattern of bee movement between stems and woody shrubs that overlap with blueberry bloom there is within a stem between flowers. Wild blueberry flowers, for a moderate degree of preference for visiting and foraging the most part are obligate out-crossers, except for 10- on wild blueberry (3 of 5 species preferred and 4 of 5 20 % of clones in a field that have the capacity to self species foraged on more intensely). The flowering shrubs fertilize (Bell et al., 2010; Bell et al., 2012b). The that are less preferred relative to blueberry might be good implications of this is that bees that forage with larger candidate species for pollinator plantings used to intra-stem movements and fewer visits of flowers per enhance densities of native bees for wild blueberry stem, and with directional movement in a blueberry field pollination (Venturini, 2015). However, these species (i.e., not randomly moving independent of cardinal would be important during the every other non-flowering direction) should be a better pollinator from the “plant’s year of blueberry production. Under typical wild blueberry perspective”. My data suggests that honey bees might management, flowers only occur every other year in the not be the best pollinator relative to its random movement two year production cycle, the other year being the prune and short distance moved between stems. The bumble year where plants are in a vegetative growth phase bee might not be the best pollinator relative to its random (Yarborough, 2009). directional movement. The andrenids might not be the Not all bees appear to be equal as pollinators of best pollinator relative to the number of flowers that they wild blueberry. Our field cage studies suggest that when visit on a stem, but this might be the best in terms of their measuring both pollen deposition per visit and flower long flights made between stems. These attributes along handling time, queen B. impatiens and O. atriventris are with the efficiency of pollen deposition and flower more efficient that A. carlini and A. mellifera. The non- handling time will have to be tested with simulation in native honey bee (A. mellifera) is the least efficient, but order to determine which bee species might be the best this bee is often placed in fields during bloom providing a pollinators in the wild blueberry ecosystem. foraging force of hundreds of thousands per hectare (up to 10 colonies per hectare, Drummond, 2012a). The last aspect of bee behavior that I looked at Therefore, in this case, the inefficiency of the species as was diurnal foraging and how temperature affects bee a pollinator of wild blueberry can be countered by its high activity. Certainly, different taxa of bees forage at density, a factor often overlooked by pollination different times of day, but all taxa that we observed ecologists. It is most likely the poricidal anthers (Bell et al., foraged during the middle of the day during what would 2009) that make wild blueberry a difficult flower for honey be suspected of being the warmest time of the day. Early bees to extract and then deposit pollen. Our observations morning after dawn and early and late evening had a suggest that honey bees exclusively nectar feed on wild lower percentage of each taxon actively foraging and blueberry flowers and lack a behavioral repertoire for some bees such as honey bees and megachilids were manipulating the anthers; thus, they only extract pollen not seen at these times. This is most likely due to air inadvertently while probing nectaries. The native bees we temperatures (Corbet et al., 1993), but also possibly due studied had behavioral mechanisms for extracting to light levels, although this is best documented for copious amounts of pollen. Bumble bees and andrenids crepuscular bees (Kelber et al., 2006). We observed that vibrated (buzz pollinated the flowers; Buchmann et al., the solitary bees and honey bees that we sampled 1983) wild blueberry flowers and Osmia leaf cutting bees increased in activity in relation to increasing temperature. crawled into the flowers and drummed the anthers with Bumble bee queens appear to be active, for the most part, their forelegs displacing pollen. independently of air temperature. This is most likely due to their ability to thermoregulate and increase their I found that the efficiency of the honey bee as a internal body temperature metabolically (Heinrich, 1972a). pollen vector can be enhanced if the honey bee visits a A question might be asked how climate change might flower previously visited by a bumble bee and then affect bee foraging in wild blueberry. The first response subsequently visits a new flower. We think that this might might be that warmer temperatures during bloom would be due to the following phenomenon. When a bumble increase the length of time during the days that bees visits a flower it either vibrates the flower releasing a forage on flowers, increasing pollination. However, during large amount of pollen that coats the interior of the flower our study we did not have exceptionally high or a bumble bee that has previously vibrated a flower temperatures during the day. From personal experience itself gets coated with pollen on its head and abdomen, during the summer when temperatures can occasionally then when nectar-feeding, the bumble bee will coat the reach > 30ºC, bumble bees will stop foraging at mid-day interior of the flower. Now when a honey bee visits a and then commence foraging in the evening. If air flower to nectar feed it might transfer much of the pollen temperatures during bloom (mid-May to mid-June) in

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA Drummond FA. 039

Maine increase dramatically, then, at least for some bee U.S. Department of Agriculture National Institute of Food taxa, foraging activity might decrease. This could be and Agriculture - Specialty Crops Research Initiative detrimental for the range of some species of bumble bees Contract/Grant/Agreement No. 2011-51181-30673. as has been documented by (Kerr et al., 2015).

REFERENCES CONCLUSION Alexander MP (1980). A versatile stain for pollen, fungi, An understanding of crop pollination requires an yeast, and bacteria. Stain Technol. 55 (1): 13-18. understanding of the interactions between: bees and crop Asare E (2013). The Economic Impacts of Bee Pollination plant reproductive biology, as well as among morphology, on the Profitability of the Lowbush Blueberry Industry in bee species, the abiotic environment and bee behavior Maine. Electronic Theses and Dissertations. Paper and plant behavior and physiology. This work reports on 2038.http://digitalcommons.library.umaine.edu/ several isolated studies that begin to increase our etd/2038 understanding of several of these interactions. The Augspurger CK (1980). Mass-flowering of a tropical shrub contribution of these studies is in providing quantitative (Hybanthusprunifolius): influence on pollinator relationships that can be used to construct a computer attraction and movement. Evol.: 475-488. simulation model of bee pollination of wild blueberry. Bell DJ (2009). Spatial and genetic factors influencing Specifically, I have shown that bumble bees yield in lowbush blueberry (Vaccinium angustifolium discriminate and select larger flowers, but that flower age Ait.) in Maine. PhD dissertation, University of Maine, or whether flowers were previously visited did not affect Orono, Maine, 144 pp. flower selection. Flower density does affect long-distance Bell DJ, Rowland LJ, Smagula J, Drummond F (2009). recruitment to clones, with clones characterized by higher Recent Advances in the biology and genetics of flower density receiving higher bee visitation. I also found lowbush blueberry. Maine Agric. and Forest Exper.Stn. that blueberry flowers are much more attractive than five Tech. Bull. 203, Univ. Maine, Orono, ME. other native co-flowering shrubs. This is important Bell DJ, Rowland LJ, Stommel J, Drummond FA (2010). because highly attractive shrubs that flower at the same Yield variation among clones of lowbush blueberry as a time as blueberry might pull bees away from the function of kinship and self-compatibility. J. Hort Sci. blueberry crop during pollination. Another important 135 (3): 1-12. finding is that native bees are much more efficient, on a Bell DJ, Drummond FA, Rowland JL (2012a). Evidence of per bee basis, than the introduced honey bee at functional gender polymorphisms in a population of the pollinating blueberry, in terms of flower handling time and hermaphroditic lowbush blueberry (Vaccinium pollen deposition on the stigma of the flower. However, angustifolium Ait.).Botany 90(5): 393-399. the presence of bumble bees visiting flowers prior to Bell DJ, Rowland LJ, Drummond FA (2012b).Does pollen honey bees appears to increase the pollination efficacy of neighborhood affect berry yield in lowbush blueberry honey bees. The last important finding in this study that (Vaccinium angustifolium Ait.)? Intl. J. Fruit Sci. 12(1- has direct implications to modeling pollination is that I 3): 65-74. found differential lengths of time during the day that Boulanger LW, Wood GW, Osgood EA, Dirks CO (1967a). different taxa foraged on blueberry and that this was Native bees associated with the low-bush blueberry in partly explained by air temperature. Bee movement in Maine and eastern Canada. Maine Agric. Exper. Stn. blueberry fields was found to be dependent upon the Tech. Bull. 26, 22pp. taxon. This aspect of bee foraging has not been Boulanger LW, Wood GW, Osgood EA, Dirks CO (1967b). investigated in simulation models constructed to Native bees associated with the lowbush blueberry in investigate the dynamics of crop pollination. I believe that Maine and Eastern Canada. Bull. T26 Tech. Ser., all of my findings will allow the construction of a Maine Agric. Exper. Stn. Orono, ME and Can. Ag. Res. preliminary model. Stn. Fredericton, New Brunswick. Buchmann SL, Jones CE, Little RJ (1983). Buzz pollination in angiosperms. Handbook of Experimental ACKNOWLEDGEMENTS Pollination Biology, pp. 73-113. Bushmann S, Drummond FA(2015). Abundance and I would like to thank the assistance of Constance Stubbs, diversity of wild bees (: Apoidea) found in Stephanie L. Allard, Lisa Campbell, Judith Collins, and lowbush blueberry growing regions of Downeast Maine. several undergraduate research assistants that helped on Environ. Entomol. 43: 1-15. data collection for these studies over the past 20 years. Cameron SA (1981).Chemical signals in bumble bee This is Maine Agricultural and Forest Experiment Station foraging. Behav. Ecol. and Sociobiol. 9(4): 257-260. Publication3500.Financial support for some of the Casper BB, La Pine TR(1984). Changes in corolla color objectives in this project (2011-2015) was provided by the and other floral characteristics in Cryptantha humilis

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA Int. J. Entomol. Nematol. 040

(Boraginaceae): cues to discourage pollinators? Evol.: Drummond FA, Stubbs CS (1997a). Potential for 128-141. management of the blueberry bee, Osmia atriventris Corbet SA, Fussell M, Ake R, Fraser A, Gunson C, Cresson. Proc. Sixth Intl. Symp. Vaccinium Culture. Savage A, Smith K (1993). Temperature and the Acta Hort 446: 77-86. pollinating activity of social bees. Ecol. Entomol. 18 (1): Drummond FA, Stubbs CS (1997b). Sampling bee 17-30. populations in lowbush blueberry in Maine. Proc. Sixth Davis RB (1993). The Natural History of Maine. Jokers Intl. Symp. Vaccinium Culture. Acta Hort 446: 101-108. are Wild Publishing House, Orono, Maine. 157 pp. Eaton EL (1957). The spread of blueberry seed through DesjardinsEC,Olivereira DD(2006). Commercial bumble manure and by migrating robins. Proc. Amer. Soc. Hort. bee Bombus impatiens (Hymenoptera: Apidae) as a Sci. 69: 293–295. pollinator in lowbush blueberry (Ericales: Ericaceae) Finnamore AT, Neary ME (1978).Blueberry pollinators of fields. J. Econ. Entomol. 99 (2): 443–449. Nova Scotia, with a checklist of blueberry pollinators in Drummond FA (2002). Honeybees and lowbush eastern Canada and northeastern United States. Ann. blueberry pollination. Univ. Maine Coop. Ext. Fact Soc. Entomol. Que. 23: 168-181. Sheet No. Gori DF (1989). Floral color change in Lupinus argenteus 629.http://umaine.edu/blueberries/factsheets/bees/ (Fabaceae): why should plants advertise the location of Drummond FA (2012a). Commercial bumble bee unrewarding flowers to pollinators? Evol.: 870-881. pollination of lowbush blueberry. Intl. J. Fruit Sci. 12: Goulson D, Hawson SA, Stout JC (1998). Foraging 54-64. bumblebees avoid flowers already visited by Drummond FA (2012b).Effect of Imidacloprid on bumble conspecifics or by other bumblebee species. Anim. bees – cage trial, 2009. Arthro. Mngmt. Tests 37 Behav. 55(1): 199-206. (Electronic Journal): C8. Groff SC, Loftin CS, Drummond FA, Bushmann S, McGill Drummond F, Collum K, Hanes S, Wilson M, Skinner J, B (2016).Spatial prediction of lowbush blueberry native Collins J (2015).A pollination toolbox for wild blueberry bee pollinators in Maine, USA. Environ. Model. and growers. Proc. N. Amer. Blueberry Res. and Ext. Software 79: 1-9. Workers Conf. Retrieved from Hall IV, Aalders LE, Nickerson NL, Vander http://dx.doi.org/doi:10.7282/T3NZ8980. Kloet SP (1979). The biological flora of Canada. 1. foraging activity in nocturnal and crepuscular bees. Vaccinium angustifolium Ait., sweet lowbush blueberry. Behav. Ecol. 17(1):63-72. Can. Field-Nat.93: 415–430. Kerr JT, Pindar A, Galpern P, Packer L, Potts SG, Hanes SP, Collum K, Hoshide AK, Drummond FA, Asare Roberts SM, Rasmont P, Schweiger O, Colla SR, E (2013). Grower perceptions of native pollinators and Richardson LL, Wagner DL, Gall LF, Sikes DS, Pantoja pollination strategies in the lowbush blueberry A (2015). Climate change impacts on bumblebees industry.Renew.Agric. Food Syst.28(4): 1-8. converge across continents. Science 349(6244):177– Hedrick UP (1919).Sturtevant's notes on edible plants. 180. doi: 10.1126/science.aaa7031. New York Agric. Exper. Stn. Rep. Albany, NY. Kevan PG, Baker HG (1983). Insects as flower visitors Heinrich B (1972a). Temperature regulation in the and pollinators. Ann. Rev. Entomol. 28(1): 407-453. bumblebee, Bombus vagans: a field study. Science Lee WR(1958). Pollination studies on low bush 175:183–187. blueberries. J. Econ.Ent.51: 544-545. Heinrich B (1972b). Energetics of temperature regulation Martin AC, Zim HS, Nelson AL (1951). American Wildlife and foraging in a bumblebee, Bombusterricola Kirby. J. and Plants. Dover Publications, Inc., New York. Comp. Physiol. 77:49–64. Munson WM (1901). The horticultural status of the genus Heinrich B (1972c). Patterns of endothermy in bumblebee Vaccinium. Maine Agric. Exp. Stn. Rept. Orono, ME. queens, drones, and workers. J. Comp. Physiol. 77:65– Ritzinger R, Lyrene PM (1999). Flower morphology in 79. blueberry species and hybrids. Hort Science 34: 130- Javorek SK, Mackenzie KE, Vander Kloet SP (2002). 131. Comparative pollination effectiveness among bees Rose A, Drummond FA, Yarborough DE, Asare E (2013). (Hymenoptera: Apoidea) on lowbush blueberry Maine wild blueberry growers: a 2010 economic and (Ericaceae: Vacciniumangustifolium). Ann. Entomol. sociological analysis of a traditional Downeast crop in Soc. Am.95(3): 345–351. transition, Maine Agric. and Forest Exp. Stn. Misc. JMP® (2015). Version 12. SAS Institute Inc., Cary, NC, Rept. 445. 1989-2007. Schaefer HM, Schaefer V, Levey DJ (2004). How plant– Jones MS, Vanhanen H, Peltola R, Drummond interactions signal new insights in FA(2014).A global review of -mediated communication. Trends in Ecol. and Evol.19 (11): 577- ecosystem-services in Vaccinium berry 584. agroecosystems, Terr. Arth. Rev. 7: 41-78. Stevens T, Hoshide AK, Drummond FA (2015). Kelber A, Warrant EJ, Pfaff M, Wallén R, Theobald JC, Willingness to pay for native pollination of blueberries: Wcislo WT, Raguso RA (2006). Light intensity limits a conjoint analysis. Intl. J. Agric. Marketing 2(4): 68-77.

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA Drummond FA. 041

Stout JC, Goulson D, Allen JA (1998). Repellent scent- (Hymenoptera: Apidae) for lowbush blueberry marking of flowers by a guild of foraging bumblebees pollination J. Econ. Entomol. 94(3): 609-616. (Bombus spp.). Behav. Ecol. and Sociobiol. 43(4-5): Vander Kloet SP (1988).The genus Vaccinium in North 317-326. America. Res. Branch Agric. Can. Publ. 1828. Stubbs CS, Jacobson HA, Osgood EA, Drummond FA Venturini E (2015). The Enhancement of Wild Bees (1992). Alternative forage plants for native (wild) bees (Hymenoptera: Apoidea) For Pollination Security.MS associated with lowbush blueberry, Vaccinium spp., in Thesis, Univ. Maine, Orono, Maine, 169 pp. Maine. Maine Agric. Exp. Stn. Tech. Bull. 148, 54 pp. Waddington KD, Holden LR (1979). Optimal foraging: on Stubbs CS, Drummond FA, Osgood EA (1994). Osmia flower selection by bees. Amer. Nat.: 179-196. ribifloris biedermannii and Megachile rotundata Whidden TL (1996). The fidelity of commercially reared (Hymenoptera: Megachilidae) introduced into the colonies of Bombus impatiens Cresson (Hymenoptera: lowbush blueberry agroecosystem in Maine. J. Kansas Apidae) to lowbush blueberry in southern New Ent. Soc. 67(2): 173-185. Brunswick. Can. Entomol. 128: 957-958. Stubbs CS, Drummond FA, Allard SL(1997). Bee Yarborough DE (2009). Wild blueberry fact sheet: Wild conservation and increasing Osmia spp. in Maine wild blueberry culture in Maine. Univ. Maine Coop. Ext. Fact blueberry fields. Northeast Nat. 4(3): 133-144. Sheet No. 220. Stubbs CS, Drummond FA (1997a). Blueberry and Yarborough DE (2013). Improving your wild blueberry Cranberry (Vaccinium spp.) pollination: A comparison yields. of managed and native bee foraging behavior. Proc.Intl. http://umaine.edu/blueberries/factsheets/production/imp Symp. Pollin. Acta Hort 437: 341-343. roving-your-wild-blueberry-yields/. Stubbs CS, Drummond FA (1997b). Pollination of wild Yarborough DE, Drummond FA, Annis S, Cote J lowbush blueberry, Vaccinium angustifolium by the (2016).Maine wild blueberry systems analysis. Acta alfalfa leafcutting bee, Megachile rotundata. Proc. Sixth Hort. In Press. Intl. Symp. Vaccinium Culture. Acta Hort 446: 189-196. Stubbs CS, Drummond FA (1997c). Management of the Accepted 29 August, 2016 alfalfa leafcutter bee, Megachile rotundata (Hymenoptera: Megachilidae), for pollination of wild Citation: Drummond FA (2016). Behavior of Bees lowbush blueberry. J. Kan. Ent. Soc. 70(2): 81-93. Associated with the Wild Blueberry Agro-ecosystem in Stubbs CS, Drummond FA(1998). Asana, impact on the USA. International Journal of Entomology and alfalfa leaf cutting bees and other pollinators. Arthro. Nematology, 2(1): 027-041. Mngt. Tests 23: 52. Stubbs CS, Drummond FA (1999). Effects of Asana XL on honey bees and Alfalfa leafcutting bees, pollinators of lowbush blueberry. Arthro. Mngt. Tests 24: 69. Stubbs CS, Drummond FA (2000). Pollination of lowbush Copyright: © 2016 Drummond FA. This is an open- blueberry by Anthophora pallipes villosula and Bombus access article distributed under the terms of the Creative impatiens (Hymenoptera: Anthophoridae and Apidae). Commons Attribution License, which permits unrestricted J. Kan. Entomol. 72 (3): 330-333. use, distribution, and reproduction in any medium, Stubbs CS, Drummond FA (2001).Bombus impatiens provided the original author and source are cited. (Hymenoptera: Apidae): An alternative to Apis mellifera

Behavior of Bees Associated with the Wild Blueberry Agro-ecosystem in the USA