Bumble bee nest density in diversified farms to ensure pollination services

Final Report for GNE11-028

Project Type: Graduate Student
Funds awarded in 2011: $14,964.00
Projected End Date: 12/31/2012
Grant Recipient: Penn State
Region: Northeast
State: Pennsylvania
Graduate Student:
Faculty Advisor:
Expand All

Project Information

Summary:

The purpose of this study is to estimate nest density of the Common Eastern Bumble Bee (Bombus impatiens) on small, diversified farms in Pennsylvania. In order to promote wild bees within the agroecosystem for “free” pollination services, we must understand which type of landscape best supports wild bee populations (Kremen et al. 2002). We collected B. impatiens on 12 pumpkin fields on small, diversified farms in Pennsylvania. We also categorized land cover of the field sites surrounding the farm (such as agricultural, forest or developed habitat).

Using population genetics, we determined how many of the foragers collected within each field were related. Related bees mean the bees came from the same colony (i.e. nest). This allows us to meet three objectives: 1) estimate the number of nests foraging and pollinating a target crop; 2) estimate the number of nests within the agroecosystem surrounding the field; 3) relate the number of nests to variations in the surrounding landscape to understand which land cover best supports large populations of Bombus impatiens.

Introduction:

The purpose of this study was to estimate nest density of the Common Eastern Bumble Bee (Bombus impatiens) on small, diversified farms in Pennsylvania. It is possible that many small farms can get adequate pollination services from wild bees, which can supplement, substitute and enhance pollination by managed honey bees (Kremen et al. 2002; Greenleaf & Kremen 2006; Winfree et al. 2007; Issacs & Kirk 2010). Wild bees are unmanaged, thus they require nesting habitat to be within the farmscape and greater landscape in order to function as pollinators in agricultural systems.

B. impatiens is a particularly important wild bee species to examine because it can provide pollination services to many different crops on diversified farms. This bumble bee species emerges early in the growing season (mid-May), and established colonies which produce workers until late into the season (October). Bumble bees are also generalist foragers, meaning they visit a diversity of plants, and social, meaning they have many workers within each colony. As a native bee, this species has also adapted to fly in cool morning conditions and can be seen actively foraging in the rain.

If a goal is to support and promote wild bees within our agroecosystems, it is important to understand the resources they require, such as nesting habitat. This study aims to 1) estimate the number of bees foraging and pollinating a target crop; 2) estimate the number of nests within the agroecosystem surrounding the field; 3) relate the number of nests to variations in the surrounding landscape to understand which land cover best supports large populations of Bombus impatiens.

Project Objectives:
Original objectives proposed:
  1. Measure nest density by estimating how many nests are providing foragers to a specific field, from pumpkin fields in Pennsylvania that vary in surrounding landscape and farmscape patterns.

    Determine if colonies that are utilizing floral provisioning plots are also utilizing nearby horticultural crops.

Modified objectives:
  1. Estimate the number of bees foraging and pollinating a target crop

    Estimate the number of nests within the agroecosystem surrounding the field

    Relate the number of nests to variations in the surrounding landscape to understand which land cover best supports large populations of Bombus impatiens.

Reasoning for modifying objectives

The original proposed project aimed to also collect Bombus impatiens on floral enhancement strips located on the selected small, diversified farms. Floral enhancements, as promoted by NRCS, are a mix of perennial flowers that bloom throughout the season. We wanted to determine how many bees utilized the floral enhancements on the farms, and also to see how many nests foraged on both the target crop (pumpkins) and the floral enhancements. However, none of the floral enhancements, as established by the growers the fall before, were successful in overwintering. This is likely because NRCS promotes seed mixes (versus transplants) because they are less costly, however, they sometimes have low establishment rates (which was not known at the time of designing this project). We were unable to collect bees from the floral enhancements because no floral enhancement areas existed on any of the collaborating farms.

Instead, we further developed our analysis relating the number of nests to the surrounding landscape. Landscape provides nesting habitat and foraging resources. The purpose of this analysis was to see we could assess what time of land cover (e.g. forest, open-developed land or agricultural land), best supports an increased number of nest at farm sites. To do this, the graduate student had to learn ArcGIS geomapping skills to measure the proportion of land cover at different radii surrounding the farm (explained in the methods section).
Although the original second objective was not conducted, we were able to expand and create a more sophisticated analysis of land cover, as well as expand our study sites (with a second year of funding), which made this a more robust study.

  • Collaborate with growers, collect bees from pumpkin fields (summer, year1)

    Molecularly identify related bees from collected samples. Travel to collaborator at USDA-ARS Bee Lab in Logan, Utah (winter, year 1)

    Acquire funding for second year of project (spring year 1)

    Continue collaboration with previous growers and expand study to include more growers, collect bees from pumpkin fields (summer, year 2)

    Molecularly identify related bees from collected samples with collaborator in Logan, Utah (winter, year 2)

    Present preliminary findings at 4 extension and 2 scientific meetings (winter, year 2)

    Categorize landscape of farm sites of all farms (spring, year 2)

    Complete analysis for year 2 data (summer, year 3)

    Prepare manuscript for publication, with 2 years of data (fall, year 3)

Key Achievements
  • Novel results about nesting habitat for one of the most abundant pollinators in the northeast region

    Expanded project to include a second year of data to produce more robust results

    Established working, ongoing relationship with collaborative growers. Direct communication with 14 growers

    Scientific development: student acquired new skill sets in molecular biology and ArcGIS.

Cooperators

Click linked name(s) to expand/collapse or show everyone's info
  • Dr. Shelby Fleischer
  • Dr. James Strange

Research

Materials and methods:
Sample collection

Six diversified pumpkin field were sampled 2011 (from SARE funding), and and twelve farm (five repeated from 2011) were sampled in 2012 (from separate funding) from eleven different Pennsylvanian counties. At least 200 B. impatiens individuals were collected from each farm field during peak pumpkin bloom, which ranged between late July and mid-August. Field sizes sampled ranged from one-two 100m rows of Cucurbits to 8 acres. Samples were collected within one week in 2011 and within two days in 2012. Foraging bees were collected directly off pumpkin blossoms while collectors walked transects back and forth between edges of the field to ensure bees were collected at all points of the field. Bees were caught by placing a 20ml plastic scintillation vial over a bee feeding in the corolla of a flower, and then capped and placed on ice. In the lab, bees were identified to species, and pinned as vouchers. The mid tarsal leg of each bee was removed for DNA extraction and sequencing.

Molecular biology

Molecular work was conducted under the guidance of Dr. James Strange at the USDA Bee Lab in Logan, Utah. Dr. Strange has specializes in the methods of using microsatellite methods for bumble bee population genetics. DNA was extracted by cutting up the tarsal leg into 4-6 pieces in a solution of 5uL of 10mg/mL Proteinase K in dH2O and 150uL 5% Chelex solution. Samples were incubated for 60min at 55°C, 15min at 99°C, 1min at 37°C, and 15min at 99°C, with hold temperature of 4°C.

PCR protocol was optimized using the following master mix for a batch of 96 samples: 197uL H20, 220uL buffer solution, 61.6uL MgCl2, 22uL BSA, 66uL dNTP, 8.8uL Taq and the primers BTMS0062 (27.5uL), B124 (27.5uL), BL15 (23.0uL), B96 (20.9uL), BTMS0081 (16.5uL), BTMS0086 (16.5uL), BL11 (12.0), BT30 (10.0uL), Btern01 (6.0uL). Initial microsatellite selection for B. impatiens based on Lozier & Cameron (2009). For each sample 8.5uL master mix and 1.5uL DNA template was combined for PCR amplification using the following thermocycler program: 3:30min at 95°C followed 0:30 min at 95°C, 1:30min at 55°C, 0:45min at 72°C repeated 30x, with hold temperature of 15°C. In 2011, the three additional primers (BT10, BT28, and BTMS0065) were included in the master mix, but not all primers successfully sequenced, thus the best 9-12 were used to determine relatedness for each population at each sample site. In 2012, the method was improved and all 9 primers from the optimized master mix were used to determine relatedness for samples from all sites. In both years, not all samples from each site successfully sequenced, and only fully sequenced or sample missing 1-2 distinguishable microsatellite makers were used to determine relatedness (less than 5 samples per site). Samples denatured with 500 LIZ standard for 2min at 95°C before being sent to sequence. Samples sequence with ABI PRISMTM 3730 DNA Analyzer at the Utah State University Center for Integrated BioSystems.
Data Analysis: Genotypes examined using GeneMapper 4.3 (Applied Biosystems) (Lozier & Cameron 2009). Relatedness determined using Colony 2.0.4.5 (J. Wang, Institute of Zoology) (Chapman et al. 2003 v.1.0 and Lozier & Cameron 2009 used v.1.2). This software program accounts and corrects for the probability of false sisterhoods, and does so to a higher degree than the previously used Kinship software (Darvill et al. 2004). Allelic dropout rates were set at 0.001 for all loci and the typing error was set at 0.005 in 2011 (Rao & Strange 2012). In 2011 the typing error was set at 0.02 to account for some poor resolution of some sequencing outputs. Samples with more than two poor microsatellite sequences were removed from the field site sample.

Landscape metrics

Each farm site was surveyed to determine the surrounding land cover at radii of up to 2000m from the target crop field. Land cover was referenced using aerial images from Google maps (www.google.com/maps) using longitude and latitude points, and verified by ground-truthing to assure that the aerial images corresponded to actual land cover.

Proportions of land cover area were measured at 500, 1000, 1500, and 2000m radii using the National Land Cover Database (NLCD) layer, which is at 30m resolution, (http://www.mrlc.gov/nlcd2001.php) in ArcGIS 10.1 (ESRI® Redlands, CA). Land cover was categorized by the modified Anderson land-use categories as designated by the NLCD data (see detailed methods developed in Standard Operating Procedure methods) and then condensed onto the following categories: “agricultural” (cultivated and pasture land), “forest & shrub” (combining all forest types and shrub/scrub land cover), “open developed” (permeable land cover, including suburban developed land), “developed” (semi-permeable and impermeable land with roads and structures), “open water” (all water), “wetland & other” (all remaining categories). Only agricultural, forest, open developed and developed land cover proportions were considered in the linear model analysis to reduce correlation among sites. Farmscape was measured by the area of land cover as determined through hand-digitization using modified Anderson land cover data. Digitization was done at 1:1000 resolution of aerial maps. Maximum buffer evaluated was 500m surrounding the field site. We used a linear mix-effects model to examine how land cover affected the nest density.

To relate colony estimates to the surrounding landscape, I used a linear mix model to select the land cover variables that best explained the number of colonies. Analysis was conducted with lme4 package (R Development Core Team, 2013), run on (R Studio v0.97.331). Farm was used as the random effect, and Poisson link function for count data. Models were developed based on backwards elimination to construct the best-fitted model using proportion of land cover at each radii separately.

Nest Density

Nest density was calculated given a range of foraging distances. The estimate of colonies foraging at a field site is likely an underestimate given the number of colonies not sampled. Yet, this estimation can be used to give a conservative estimation of nest density in the areas surrounding the farm site. The foraging distance of B. impatiens is unknown, and while Bombus spp. are considered to have long foraging distances relative to smaller-sized solitary bees (Greenleaf et al. 2007), the foraging distance of specific species can vary greatly (Goulson 2010). In this study, short and long foraging ranges that corresponded to the land cover measures and which corresponded to previous Bombus spp. studies were used to estimate foraging distance; 0.5km based on Knight et al. (2005) 1km based on Herrmann et al. (2007) and 2km based on the large foraging range estimates of Rao & Strange (2012).

Research results and discussion:
Estimated number of colonies (samples & molecular biology)

An average of 182 bees were successfully sequenced in each year. In 2011 more than 100 bumble colonies we estimated foraging on the six pumpkin fields, with totals ranging from 115-167 colonies. In 2012, the estimated number of colonies per field ranged from 82-138. The average number of estimated colonies was 127. The effective population size (Ne), which is indicative of heterogeneity in the population, ranged from 78-1007, with a median of 370 colonies.

Landscape

Land cover was not significantly related to colony number at the greater radii (1000m, 1500m and 2000m), but all land cover types were significant at 500m. The proportion of land cover for each land-use category was relatively stable between 1000m-2000m at each site however proportion of land-use varied at the farm-scape level at the 500m radius. This suggests that immediate farm-scape has more explanatory value than the larger landscape. When analyzed by year, year 1 data (funded by SARE) suggests that forest habitat significantly related to nest number. However, when the data is combined with additional sites from year 2 (additional funding), then the farm-scape as a whole is more significant. Between year one and year two, the number of nests is consistent (average 182 in both years), but the robust data set provide the sampling size needed to relate it to the surrounding habitat directly.

Nest Density

When calculated with a shorter foraging range (500m) the estimated average nest density was 161 colonies per km2, ranging from 104-213 colonies per km2. Given a foraging range of 1km, the estimated average nest density was 40 colonies per km2, ranging from 26-53 colonies per km2. Given a long foraging range of 2km, the estimated average nest density was 10 colonies per km2, ranging from 7-13 colonies per km2.

Research conclusions:

This study results in novel information about the number of nests utilizing pumpkin fields on diversified farms and probable nest densities in the surrounding agroecosystem. I related the number of colonies to the surrounding landscape to see if certain land cover types were significantly and positively related and thus may be possible nesting habitat. The maximum radius of 2km was selected because some previous studies suggested that the foraging range of bumble bees is greater (>1km) than the typical solitary bee (<1km) and that bumble bees respond to the landscape at a larger scale (Steffan-Dewenter et al. 2002; Greenleaf et al. 2007; Winfree et al. 2009). However, I found that colony estimates were not significantly related to landscape at larger radii, but were significant at a smaller radius which corresponds to the farm-scape level. This suggests that B. impatiens foraging behavior may be short-ranged (<1km).

A shorter foraging range for B. impatiens is similar to other Bombus spp. that have been reported between 312-758m (Darvill, Kight, & Goulson 2004; Knight et al. 2005). It is notable that these studies were done in heterogeneous agricultural landscapes in the U.K. There are other studies that show foraging range to be much greater (<11km) in an agricultural-intensive landscape (Rao & Strange 2012), and some homing studies have measured bumble bees traveling up to 20km home when released far from their nest (Goulson 2003). Bumble bee foraging distance is plastic depending on resource richness, independent of landscape composition (Jha & Kremen 2012), thus a resource rich environment may foster shorter foraging ranges.

The diversified farms examined in this study have fine-scale heterogeneous land cover patterns due to their small field size, crop diversity, hedge rows, location to forest and developed land, and farming practices, thus the B. impatiens in this agricultural system may have shorter foraging ranges and have the greatest response to the landscape at the 500m radii. The relationship between the number of nests and land cover types varied in positive, neutral and negative responses, however the collective heterogeneity may best support the number of nests. Kells & Goulson (2003) found that edges are important nesting habitats for bumble bees in the U.K., which may explain why developed and open developed land cover are positively correlated with nest estimates. The heterogeneity between different land cover types may provide abundant resources for bumble bees within the landscape immediately surrounding the field site. Given the assumption of a shorter foraging range, the nest density at these sites is high and on par with previous studies that considered nest density in heterogeneous agricultural landscapes in the U.K. (Darvill, Kight, & Goulson 2004; Knight et al. 2005). This study is the first to measure B. impatiens nest density and the first measure nest density in the eastern U.S.

Future studies should consider examining the field site at a finer-scale and capture greater heterogeneity metrics. Previous bumble bee conservation studies have assigned values to floral resources when assessing the contribution of different land cover types to bumble bee populations (Hines & Hendrix 2005); this could be done for future studies examining the value of land cover for nesting habitat, but more information about species specific nest preference is needed.

Participation Summary

Education & Outreach Activities and Participation Summary

Participation Summary:

Education/outreach description:
  • “Over 100 bumble bees estimated to forage in pumpkin fields on diversified farms.” C. Sheena Sidhu, James P. Strange, Shelby J. Fleischer. International Conference on Pollinator Biology, Health and Policy. University Park, PA. August 15-17, 2013.

    “Pollinator conservation on small farms.” Presented by David Biddinger – included bumble bee nest density data. Pennsylvanian Association for Sustainable Agriculture- Farming for the Future Conference. State College, PA. February 2, 2013.

    “Does floral provisioning enhance pollination of Cucurbita and Cucumis crops by bee communities?” C. Sheena Sidhu, Shelby Fleischer, David Biddinger. Entomological Society of America- Eastern Branch Meeting. Lancaster, PA. March 17, 2013.

    “Pumpkin Pollinators.” C. Sheena Sidhu. Mid-Atlantic Fruit and Vegetable Convention. Hershey, PA. Jan 31, 2013.

    “Pollinator updates in cucurbits.” C. Sheena Sidhu, Shelby Fleischer. 2013 New Holland Vegetable Day. New Holland, PA. Jan. 21, 2013.

    “Nest Density of Bombus impatiens in heterogeneous landscapes surrounding vegetable farms.” C. Sheena Sidhu, James P. Strange, Shelby J. Fleischer. Entomological Society of America Annual Meeting. Knoxville, TN. November 11-14, 2012.

    “Bumble bee nest density and risk from soybean aphid and corn earworm in the northeast.” Presented by S.J. Fleischer. Shelby J. Fleischer, Amanda Bachmann, C. Sheena Sidhu, Eric Bohenblust. NCERA-213 Multistate Research – Migration and Dispersal of Agriculturally Important Biota. Fort Lauderdale, FL. Oct. 4-5, 2012.

Project Outcomes

Project outcomes:

Supporting and promoting wild bees has the potential to reduce pollination service costs to growers. Many farms either rent or purchase managed honey bees for pollination services, which is a cost incurred by growers. Wild bees within landscapes can supplement, substitute and enhance pollination services depending on the cropping system. Overall, wild bee pollinators can reduce pollination service cost to growers. Direct economic analysis was not included in this study, however, results from this study can be used by growers, land managers and policy makers to recommend practices which would support and promote wild be populations in agroecosystems in order to increase “free” pollination services provided by unmanaged wild bees.

Farmer Adoption

The collaborating growers are very interested in the project. We collected bees from six farms and repeated the six farms the following year (although we could not get an adequate number of bees from one of the farm sites). We also added an additional 7 farms the second year. Many of these growers also participated in a related, but separately funded, projects examining wild bee communities and species diversity. We collaborated with 7 of these growers over three years, and continue to do so. Over this period, the growers have become more knowledgeable about wild bee species that exist within their agroecosystem and are eager to adopt practices that may support wild bee populations. Some changes include: adding floral enhancements to supplement floral resources for wild bees, reduced tilling to minimize disturbance of ground-nesting bees (such as bumble bees), managing hedge rows which provide floral resources and nesting habitat, and crop placement close to potential bee habitat.

Assessment of Project Approach and Areas of Further Study:

Areas needing additional study

The original proposed project aimed to also collect Bombus impatiens on floral enhancement strips located on the selected small, diversified farms. Floral enhancements, as promoted by NRCS, are a mix of perennial flowers that bloom throughout the season. Their purpose is to attract wild pollinators from the surrounding landscape into the farm and supplement floral resources to promote wild bee populations (Shepherd 2003; Vaughan et al. 2007). We wanted to determine how many bees utilized the floral enhancements on the farms, and also to see how many nests foraged on both the target crop (pumpkins) and the floral enhancements. However, none of the floral enhancements, as established by the growers the fall before, were successful in overwintering. This is likely because NRCS promotes seed mixes (versus transplants) which sometimes have low establishment rates (which was not known at the time of designing this project). We were unable to collect bees from the floral enhancements because no floral enhancement areas existed on any of the collaborating farms. I strongly suggest that this is a topic for future studies. This will allow us to examine if floral enhancements support bumble bee populations by determining if the same nests that utilize the flowers also pollinate the crops.

Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture or SARE.