CARPENTER BEE NECTAR ROBBERY EFFECTS ON FRUIT QUALITY AND NESTING PROVISIONS IN RHODE ISLAND

Eastern carpenter bees, Xylocopa virginica L., are among the most abundant native bee visitors to highbush blueberry flowers in Rhode Island, and they frequently slit corollas to rob nectar. My objective was to assess if nectar robbery offsets the possible value of X. virginica as a native pollinator of blueberries in Rhode Island. I studied foraging behavior of X. virginica at the Rhode Island Agricultural Experiment Station planting which consists of 14 highbush blueberry cultivars. I assessed plant and environmental factors related to slitting behavior, and the effects of slitting on fruit set and blueberry quality. The average number of flowers that X. virginica visited per minute was significantly influenced by time of day, temperature, and sex, but the time spent per flower was not. The frequency of corolla slitting by carpenter bees among 14 cultivars during bloom averaged 35% slit flowers (range 16 – 67% ‘Earliblue’ and ‘Lateblue’ respectively) in 2017, and 39% (range 20 – 62% ‘Bluecrop’ and ‘Collins’ respectively) in 2018. Plant and environmental factors that affected the proportion of corollas slit included cultivar, anther length, flower volume, and number of days in bloom at or above 15C. Corolla slitting did not affect fruit set. Average fruit weight and percent soluble solids resulting from slit and non-slit corollas did not differ significantly in two early(‘Bluehaven’, ‘Earliblue’), two mid(‘Collins’, ‘Bluecrop’), and two late-season (‘Herbert’, ‘Lateblue’) ripening cultivars in 2017. In 2018, average fruit weight and percent soluble solids resulting from slit and non-slit flowers did not differ significantly in most cultivars, but slit flowers resulted in berries with greater mass in two cultivars, ‘Bluehaven’ and ‘Collins’. ‘Collins’ fruit from non-slit corollas had a significantly higher percentage of soluble solids at maturity than fruit from slit corollas in 2018. Corolla slitting and nectar robbery by X. virginica did not have a significant negative effect on fruit quality under our growing conditions and pollinator community. Understanding the nesting and foraging habits of Xylocopa virginica can aid in efforts to recruit natural populations to crops for pollination services. Measurements of Xylocopa virginica nest tunnels and cells were similar to those reported in previous studies. Analysis of pollen loaves showed that X. virginica provisioned pollen loaves from 21 different genera of plants in 2016, 19 in 2017, and 39 in 2018. Antirrhinium majus (Garden snapdragon) made up the majority (21.4%) of pollen collected in all three years. Blueberry pollen was a minor component of pollen loaves (0.1%). Only two of 168 trap nests deployed in 2017 were occupied by a total of ten X. virginica bees. However, 33 nests (19.6%) hosted 230 Osmia taurus, 73 Osmia cornifrons, and 8 Osmia lignaria Thirty-four nests (20.2%) were occupied by 151 grass-carrying wasps, Isodontia sp. and 6 vespid wasps occupied two nests (1.2%) in 2017. In 2018, four of ninety-six trap nests were occupied by carpenter bees.


TABLE OR FIGURE PAGE
Chapter One Table 1 Table 6. Stepwise logistic regression model on proportion of flowers slit, 2018……34 Table 7. Comparison of slit and non-slit blueberry flowers that set fruit, 2018……..35    Two pieces of weathered pine boards as above were attached at right angles with a 15 mm overhang at the entrance end. The entrance was 12 mm in diam. and 3 cm deep.
C) Two pieces of weathered pine boards as above were attached with screws at right angles. D) Weathered pine boards as above were routered in each of two boards to create two halves of 12 mm tunnel widths 15 cm long on each side of a 12 mm diam.

INTRODUCTION
Much of modern agriculture relies on the critical activity of insect pollinators, namely bees, for pollination services that enhance crop production (Klein et al. 2007).
In the last century, the European honey bee, Apis mellifera, has been managed as the primary pollinator for cultivated crops (Southwick and Southwick Jr. 1992). Managed bees, primarily A. mellifera, contribute an estimated $11.53 billion to US agriculture each year (Koh et al. 2016). Widespread declines in A. mellifera populations could lead to future agricultural instability, particularly in agroecosystems with insufficient wild pollinators (Allen-Wardell et al. 1998). Therefore alternative, managed pollinator species are being explored for sustainable crop pollination services (Bosch and Kemp 2002, Javorek et al. 2002, Westerkamp and Gottsberger 2000. Highbush blueberries depend on pollinators for fruit production (Brewer and Dobson 1969, Dogterom et al. 2000, MacKenzie 1997) and many growers stock their blueberry plantings with Apis mellifera colonies to meet this need. At least 80% of highbush blueberry flowers must set fruit to result in a commercial crop (MacKenzie 1997) and berry production is known to benefit from sonication and cross-pollination (De Luca and Valleho-Marin 2013, Free 1993, Brewer and Dobson 1969. Sonication, also referred to as buzz pollination, is a pollination syndrome that allows a pollinator to effectively release pollen from the small pores in blueberry anthers (De Luca andValleho-Marin 2013, Free 1993). Cross-pollination enhances fruit set, seed number, and fruit mass (Brewer and Dobson 1969). Cross-pollination from more distantly related cultivars results in larger berries that ripen earlier (Dogterom et al. 2000).
Blueberry cultivars vary in flowering phenology, and cultivars must bloom at the same time for cross pollination (Eck et al. 1990, McGregor 1976.
The efficacy of Bombus spp. as lowbush blueberry pollinators has already been documented (Javorek et al. 2002, Drummond 2012. In Rhode Island, Scott et al. (2016) identified 41 species of native bees collected from highbush blueberry flowers during bloom throughout the state. Andrena spp., Bombus spp., and Xylocopa virginica were among the top ten most often collected bees. X. virginica was found to carry the third largest mean pollen grain load of the species sampled (233,500). Abundance at highbush blueberry plantings and sonication behavior suggest that X. virginica might be an effect pollinator of highbush blueberry.
Indeed, many of the roughly 400 species of Xylocopa around the world are already appreciated as agriculturally-significant pollinators of some plants and crops (Gikungu 2014).
However, the Eastern carpenter bee, Xylocopa virginica, is a known nectar robber of blueberry flowers (Sampson et al. 2004

Fruit Mass and Soluble Solids (% Brix) Resulting from Slit and Non-Slit
Flowers. In 2017, one hundred open flowers on each of 2 early-('Earliblue' and 'Bluehaven'), 2 mid-('Collins' and 'Bluecrop'), and 2 late-season ('Herbert' and 'Lateblue') cultivars were tagged with different colored thread indicating whether the flower was slit or non-slit (Fig. 2). We selected slit and non-slit flowers adjacent to each other and in the same cluster. Both slit and non-slit flowers had an equal opportunity of being pollinated prior to tagging and netting. Following tagging, each bush was covered with 80 g (1.0 x 0.6 mm mesh) ProtekNet (Tek-Knit Industries, Mont-Royal, QC, Canada) 2-10 days after first bloom to prevent any further pollinator visitation. In isolating each bush from further visitation, we were able to ensure that tagged non-slit flowers did not experience subsequent slitting and nectar robbery.
Exclusion netting was removed from bushes after fruit set. Tagged flowers from each bush were followed to fruit maturity. Berries were harvested when they were entirely blue and had no indications of immaturity. Berries from slit and non-slit flowers were  Average percentages of slit corolla ranged from 16% ('Earliblue') to 67%

Correlation of Slit Corollas to Flower Dimensions, Bloom Time, and
Weather. There were significant differences among cultivars in all of the flower dimensions measured and in bush height (Table 1). In 2017 we included cultivar, flower volume (cc), corolla opening width (mm), corolla length (mm), style length (mm), and anther length (mm) in a stepwise logistic regression of percent slit corollas, and cultivar was the only significant variable (Wald χ2 = 200.33, df = 13, P < 0.01). In 2018 we included variables related to weather and timing of flowering in the analysis (Table 3). The logistic regression model in 2018 that best fit slitting frequencies observed in the field included cultivar, anther length, flower volume, and the number of bloom days at or above 15ᵒC (Table 6). To assess the factors that influenced slitting differences among cultivars, we ran the stepwise logistic regression without cultivar as a class variable, and corolla opening width (

Fruit Mass and Soluble Solids (% Brix) Resulting from Slit and Non-Slit
Flowers. Fruit mass (Table 4) and sugar content (% Brix) ( Table 5) of berries from slit and non-slit corollas were not significantly different among any of the cultivars in 2017. In 2018, analysis of variance for fruit mass revealed a significant two-way interaction between class variables cultivar and the slit or non-slit condition (F = 3.21, df = 5, P = 0.0071). Thus, we analyzed differences in fruit mass separately for each cultivar. There were no significant differences in mass between fruit that resulted from slit and non-slit corollas in four of the six cultivars. Berries that resulted from slit corollas of 'Bluehaven' and 'Collins' had a higher average berry mass at the time of harvest (Table 4). A generalized linear model of fruit soluble solids (% Brix) revealed a significant two-way interaction between class variables cultivar and the slit or nonslit condition (F = 3.47, df = 5, P = 0.0041). Thus, we analyzed differences in fruit soluble solids (% Brix) separately for each cultivar. Berries that resulted from 'Collins' corollas that were non-slit had significantly higher average soluble solids (% Brix) (Table 5).

Percent Fruit Set.
The average percentage of slit blueberry flowers that set fruit was 88% and the percentage of non-slit blueberry flowers that set fruit was 82% (Fig.   18). The difference in average percent fruit set did not differ between slit and non-slit flowers on the same cluster overall (Wald χ 2 = 0.0292, df = 1, P = 0.864) or among cultivars (Wald χ 2 = 8.755, df = 5, P = 0.119).

DISCUSSION
Our results indicate that corolla slitting for nectar robbery by carpenter bees did not affect fruit set, berry size, or sugar content of most highbush blueberry cultivars in our planting. These results suggest that nectar robbery does not offset the possible value of X. virginica as a native pollinator of blueberries in southern New England.
Observations of X. virginica foraging on blueberry flowers suggest that the number of flowers visited per minute is affected more by movement between flowers than foraging time on individual flowers. The results of our logistic regression model suggest that X. virginica does not necessarily slit corollas and rob nectar from 13 blueberry flowers because of a physical barrier like a narrow corolla opening or a long distance to the nectary. X. virginica robs nectar from blueberry flowers due to a combination of cultivar, anther length, flower volume, and the number of days at or above 15ᵒC (Table 3). The consistent significance of cultivar in the frequency of nectar robbery suggest that there is a difference between cultivars that may extend outside of flower morphology alone. It is known that 'Earliblue' flowers are less attractive to honey bees (Pavlis 1991). It appears that carpenter bees are also less attracted to this cultivar. Isaacs et al. (2016)  Benjamin and Winfree (2014) studied honey and native bee pollination in commercial highbush blueberry in New Jersey. They found that the European honey bee, Apis mellifera L. deposited a median of 18.5 tetrads of pollen during a nectarcollecting visit, 24 tetrads during a pollen-collecting visit and 0.5 tetrads during a secondary nectar-robbing visit (through punctures made by X. virginica). They also found that pollen tetrads deposited by Bombus spp., large Andrena spp., medium Andrena spp. and Xylocopa virginica were 23.5, 9.0, 11.5, and 2.5 tetrads respectively.
All their study sites were stocked with domesticated honey bees at densities of 2.5-7.5 hives ha -1 . Honey bees provided 86% and native bees 14% of the pollination.
Conversely, Winfree et al. (2007) found that native bees were the most important pollinators and alone were sufficient to pollinate commercially grown watermelons in New Jersey and Pennsylvania.
Previous studies have shown that native bees contribute to crop pollination at farms near natural habitat, but not in more intensively used agricultural areas (Kremen et al. 2004, Klein et al. 2007 where native bee populations might provide sufficient pollination (Garibaldi et al. 2013, Benjamin et al. 2014. Xylocopa virginica has a long colony life cycle, with many females living two years (Gerling and Hermann 1978). In March and April males defend areas near the nest and mate with females. Females construct nests in unfinished wood, and nests can be reused for many generations (Gerling and Hermann 1978). Xylocopa virginica has nectar robbing tendencies, relatively low blueberry pollen loads, and pollen transfer efficiency is low (2.5 pollen tetrads deposited per visit, Benjamin and Winfree 2014).
Despite these shortcomings, the natural abundance of these pollinators and possible ease of increasing numbers by providing unfinished wood nesting sites around blueberry plantings, suggests more research on the importance of this bee as a blueberry pollinator is needed. The results of our study indicate that corolla slitting and nectar robbery by Eastern carpenter bees does not have a significant negative effect on fruit yield under the described growing conditions and pollinator community.

INTRODUCTION
The heavy reliance of modern agriculture on the European honey bee, Apis mellifera, for crop pollination poses a serious risk of food insecurity as honey bee populations decline at alarming rates (Allen-Wardell et al. 1998, Klein et al. 2007Southwick and Southwick Jr. 1992). As demand for pollination services increases, the need for alternative pollination strategies is evident. The annual benefit of bees to agricultural production in the U.S. alone is upwards of $14.6 billion, with native bees contributing at least 20% of this value (Koh et al. 2016). Some studies suggest that wild native bee populations may provide adequate pollination services where suitable habitat exists (Garibaldi et al. 2013, Winfree et al. 2007). The commercial application of native bees for crop pollination may depend on both the effectiveness of bees to pollinate crops and the potential of bees to be managed (Velthuis and van Doorn 2006). Recent studies have explored the effectiveness of native bees including Bombus, Osmia, and Andrena as crop pollinators (Morandin et al. 2001, Bosch and Kemp 2015, Park et al. 2016. In some agro-ecosystems, native bees may be even better pollinators than honey bees (Westerkamp 1991).
Honey bees may not provide sufficient pollination to crops in cold and rainy climates, or to crops requiring particular pollination syndromes (Willmer et al. 1994 (Javorek et al. 2002). All of the native bee genera that are known to perform buzz pollination have been identified throughout Rhode Island (Scott et al. 2016). Between 2014 and 2016 the Eastern carpenter bee, Xylocopa virginica, was among the top five most commonly collected bee species in highbush blueberry plantings (Scott et al. 2016).
The The natural abundance of X. virginica populations in Rhode Island presents an opportunity to explore the adaptability of carpenter bees for crop pollination. Because X. virginica is one of the most common bees visiting blueberry flowers and has the ability to buzz-pollinate, we focused on the potential application of X. virginica as a managed blueberry pollinator. Our objectives were to 1) determine the nesting habits of X. virginica based on field observations and nest dissections, 2) identify the forage requirements for X. virginica by analyzing pollen provisions, and 3) evaluate several nest designs to recruit and maintain X. virginica populations. removed from their cells and examined for any eggs or larvae accompanying them. If no larva was found in a cell with a loaf, it was considered to be a full loaf. Full loaves were weighed and analyzed for pollen cell composition via acetolysis processing (Faegri et al. 1989). Pollen cells were identified to the lowest taxon possible.

METHODS
The eight full loaves collected in 2016 were weighed and dissolved in a 20 ml dye solution (92.5% water, 5% Tween 20, 2.5% Gram fuchsin solution). One microliter of this solution was placed onto a hemocytometer and the pollen grains were counted under a microscope. This value was then extrapolated to find the number of pollen grains in the pollen loaf. We used the average number of pollen grains calculated from the top 10% of pollen loads carried by X. virginica bees collected in 2015 to calculate an average "full" pollen load of 1,207,333 pollen grains. We were then able to calculate the approximate number of foraging trips required for a bee to complete a full pollen loaf.
Discarded Pollen Composition. In July 2017, one female X. virginica was observed noticeably pushing pollen out of a nest with her head. A container was 59 placed directly below the entrance of the nest. After one week, the pollen contents from the container were collected and analyzed for pollen cell composition.

Manufactured nests attached to a barn, 2017.
Eighty-one pine boards (36.5 x 2 x 9 cm) were routered in each of two boards to create two halves of both 9 mm and 12 mm tunnel widths 15 cm long on each side of a 12 diam. entrance hole 15 mm deep (Fig. 1). The two halves of a nest were held together with four 3.8 cm screws. Twelve nests of each 9 mm and 12 mm tunnel widths were randomly assigned and attached to 3.6 x 8.8 cm boards 2.5 m from ground level where carpenter bees have been nesting for at least 30 years 54 cm from the edge of an aluminum roof (Fig. 2). Twelve nests of each of the tunnel widths again were randomly assigned and attached to 3.6 x 8.8 cm boards 2.3 m from ground level and 13 cm from the edge of the aluminum roof ( Fig. 2). All entrance holes were directed downward.
Manufactured nests attached to a lean too storage area, 2017. Twelve 9 mm and twelve 12 mm tunnel width pine nests as above were randomly assigned and attached to the fascia of a lean too storage area 2.1 m from ground level and 18 cm from the edge of the roof (Fig. 3). Carpenter bees had also been seen nesting in this area for at least 30 years. All entrance holes were directed downward.
Manufactured nests attached to posts, 2017. Fourty-eight 9 mm and fourtyeight 12 mm tunnel width pine nests as above were randomly assigned and attached to 10.2 x 10.2 cm pressure treated posts with nest entrances at 1.5 and 3 m above ground level (Fig. 4). Posts were 6 m from the edge of a 0.3 ha blueberry planting and 4.5 m between posts. There were 12 posts on each of the east (sunny) and west (shady) sides of a 0.15 ha blueberry planting (Fig. 4).

DISCUSSION
Nest architecture. Manufactured wooden nests have been used to "trap" bees and wasps to study their nest provisions and structure as well as to provide nesting sites in addition to those available naturally (Krombein 1967 Krombein (1967) studied nests in the Plummers Island, MD and found that coniferous wood is preferred to deciduous wood, although bees will nest in either.
Cells were 21-22 mm long with partitions 3-4 mm thick. The partitions were made from tiny wood chips rasped from the tunnel walls and cemented together, presumably by a salivary secretion. Pollen masses were well saturated with nectar and were 14 mm long. Nest measurements from Rhode Island were comparable to those reported by Gerling and Hermann (1978) and Krombein (1967).  (Gerling and Hermann 1978). Xylocopa virginica has nectar robbing tendencies, relatively low blueberry pollen loads, and pollen transfer efficiency is low (2.5 pollen tetrads deposited per visit, Benjamin and Winfree 2014).
Despite these shortcomings, the large number of these pollinators and possible ease of increasing numbers by providing unfinished wood nesting sites around blueberry plantings, suggests more research on the importance of this bee as a blueberry pollinator is needed.
Discarded Pollen. Corbet and Willmer (1980)  Manufactured Nest Acceptance. We believe that our pre-routered tunnels and two piece manufactured nests did not attract more carpenter bees for two reasons.
The two boards warped somewhat upon deployment and there was a gap between the two pieces allowing moisture to enter the tunnels which may have deterred nesting by carpenter bees. Also, the ready-made tunnels used in 2017 were quickly occupied by sphecid wasps and megachilid bees (75% of all manufactured nests in 2017) and therefore were not available for carpenter bees.
Benjamin and Winfree (2014) studied honey and native bee pollination in commercial highbush blueberry in New Jersey. They found that the European honey bee, Apis mellifera L. deposited a median of 18.5 tetrads of pollen during a nectarcollecting visit, 24 tetrads during a pollen-collecting visit and 0.5 tetrads during a secondary nectar-robbing visit. They also found that pollen tetrads deposited by Bombus spp., large Andrena spp., medium Andrena spp. and Xylocopa virginica were 23.5, 9.0, 11.5, and 2.5 tetrads respectively. All of their study sites were stocked with domesticated honey bees at densities of 2.5-7.5 hives ha -1 . Honey bees provided 86% and native bees 14% of the pollination. Conversely, Winfree et al. (2007) found that native bees were the most important pollinators and alone were sufficient to pollinate commercially grown watermelons in New Jersey and Pennsylvania. Previous studies have shown that native bees contribute to crop pollination at farms near natural habitat, but not in more intensively used agricultural areas (Kremen et al. 2004, Klein et al. 2007).
Numerically, X. virginica was one of the top five bees collected at 15 commercial blueberry plantings in Rhode Island from 2014-16. Benjamin and Winfree (2014) found X. virginica deposited a median number of 2.5 pollen grains per blueberry flower visit. This is rather low compared to a median number of 23.5 pollen grains for the bumble bee Bombus bimaculatus. Sampson et al. (2004), however, found that carpenter bees are benign or even potentially beneficial floral visitors to rabbiteye blueberry.
Because each X. virginica tunnel system is only 15 cm in length, it is feasible to manufacture many nest structures and deploy them around blueberry or other crop plantings (Fig. 4). Future research will be aimed at resolving the issue of X.
virginica's value as a blueberry pollinator and evaluating artificial nests which may also provide clues as to how to prevent them from infesting homes, etc. We will also continue to identify other pollen cells collected by X. virginica to see what forage plants are important for this species.      A 12 mm. diam. entrance hole was drilled in the middle of the boards just below the top board at a 60 o angle and 2 cm deep. B) Two pieces of weathered pine boards as above were attached at right angles with a 15 mm overhang at the entrance end. The entrance was 12 mm in diam. and 3 cm deep. C) Two pieces of weathered pine boards as above were attached with screws at right angles. D) Weathered pine boards as above were routered in each of two boards to create two halves of 12 mm tunnel widths 15 cm long on each side of a 12 mm diam. entrance hole 15 mm deep.