Final Report for GNC13-177
This project explored the effects of pest management program intensity and wildflower plantings on native bee communities present on highbush blueberry farms in western Michigan, with the goal of identifying best practices for managing pests and conserving beneficial insects on Great Lakes fruit farms. We revisited 15 blueberry farms sampled for bee community composition from 2004-2006 to assess the short- and long-term effects of insecticide program intensity on the bee community foraging on highbush blueberry during bloom. Prior-year insecticide program risk at the field scale had a significant negative effect on wild bee species richness, but not bee abundance or diversity. Intensive sampling of soil-nesting bees on four farms with wildflower plantings adjacent to blueberry fields determined that wildflower plantings support increased nesting by soil-nesting bees over grassy field margins and wooded habitats. We are currently developing printed and electronic educational materials on how to minimize risk to beneficial insects from pest management practices in Great Lakes specialty crops. We surveyed growers attending the 2013 and 2014 Great Lakes Fruit, Vegetable, and Farm Market EXPOs to assess rates of adoption of best practices for beneficial insect conservation.
An estimated 80% of wild plant species and 75% of the leading global food crops depend on animal pollinators, primarily bees (Klein et al. 2007). Declines in global wild bee abundance and richness, especially in intensively managed agricultural areas, may be threatening the provision of pollination services to wild and domesticated plant species (Kremen et al. 2002; Biesmeijer et al. 2006; Potts et al. 2010). Understanding the non-target effects of pest management programs on native bees and identifying management practices to conserve and restore these beneficial insects will be important for mitigating pollinator declines and maintaining stable and sufficient crop pollination services (Allen-Wardell et al., 1998; Garibaldi et al., 2011).
Highbush blueberry (Vaccinium corymbosum L.) requires insect pollination for economically viable yields (Isaacs and Kirk, 2010). While most Michigan blueberry growers rely on rented honey bee colonies for pollination, the steep declines in managed honey bee colonies across North America over the last 20 years have led to sharply rising rental prices and some hive shortages, underscoring the importance of identifying and supporting wild pollinators to ensure the resilience of pollination services and crop yields (Allen-Wardell et al., 1998; Winfree et al. 2007). Over 100 wild bee species have been recorded in Michigan blueberry fields during bloom, including ~10 species exhibiting high abundance and/or fidelity to Vaccinium flowers (Tuell et al., 2009). However, management of insect pests following bloom generally necessitates several insecticide applications per season. While short-lived bees that are tightly linked with blueberry bloom may not be affected by post-bloom insecticides, the sprays may have fitness consequences for bees with longer life cycles, including bumble bees (Bombus sp.), some of which are showing significant population declines in the eastern US (Colla and Packer, 2008; Grixti et al., 2009; Cameron et al., 2011; Bartomeus et al., 2013).
For the past ten years, blueberry production in Michigan has trended toward greater adoption of integrated pest management (IPM) strategies, including reductions in the quantity and toxicity of insecticides applied per season (NASS, 2001-2011). The use of broad-spectrum insecticides, such as azinphos-methyl, malathion, and carbaryl, declined strongly from 2001-2011, in favor of reduced-risk insecticides such as imidacloprid, acetamiprid and methoxyfenozide. However, the recent arrival of spotted-wing drosophila (Drosophila suzukii), an invasive pest that can cause serious economic damage to stone and small fruits (Lee et al. 2011), threatens the continued viability of IPM strategies in many fruit crops, including the region’s nation-leading blueberry and cherry crops. Additionally, these reduced-risk insecticides are generally not effective on D. suzukii (Van Timmeren and Isaacs, 2011). Estimates for Michigan indicate 2012 losses due to D. suzukii worth nearly $27 million. This invasive pest, with its short generation time, high dispersal ability, and lack of natural enemies, is likely to reverse trends toward adoption of reduced-risk insecticides. Since its discovery in 2010, Michigan blueberry growers have switched to prophylactic use of broad-spectrum insecticides applied on a calendar schedule to control this insect prior to harvest, a trend that may be highly deleterious to the bees active in and around crop fields after bloom. It will be essential to develop strategies for controlling D. suzukii and other blueberry pests that minimize risk to these wild bees.
One conservation practice that has been extensively studied in our lab over the past few years has been the establishment of native perennial wildflower plantings in blueberry field margins. These plantings can increase local bee and natural enemy abundance and richness, augmenting crop yields in adjacent blueberry fields through increased pollination services (Blaauw and Isaacs, 2014). However, it is not clear whether enhanced bee abundance and richness result from the addition of floral resources or from the provision of nesting habitat. For long-lived ground-nesting species like bumble bees (Bombus sp.), it may be both, as these species require floral resources to support their colonies after blueberries have finished blooming. However, for the species that exhibit tight linkage with the blueberry bloom period, when the wildflowers in the habitat planting have not yet begun blooming, nesting resources may be more important than available floral resources for the location of nest-site selection. Preferential nesting by ground-nesting bees in wildflower plantings would suggest that these restorations have the potential to function as source habitat for wild bees that forage on adjacent blueberry fields.
Over the past five years, several SARE projects have evaluated or encouraged farmer adoption of flowering conservation plantings to support pollinating insects and natural enemies in a variety of cropping systems (Surcica, 2009; Walton, 2009; Vaughan, 2010; Wilson, 2010; Blaauw and Isaacs, 2011; Blaauw, 2012; James, 2012). Others have investigated farm management practices or developed “best practice” guidelines to support native bees on-farm (Kuehn, 2009; Williams, 2009; Rao, 2011), as well as evaluated the contribution of the surrounding landscape context to the on-farm bee community (Alesch, 2012; Johnson and Sieving, 2012). Several projects have also evaluated methods to enhance nesting of Bombus and Osmia bees near crop fields (Bogash, 2009; Sidhu, 2011). This project complements previous work evaluating the use of conservation strips to enhance pollinator populations and the delivery of crop pollination services by investigating the mechanism by which these plantings support and enhance native bee populations. Similar to Sardinas (2012), who evaluated rates of nesting in sunflower fields adjacent to restored hedgerows, we assessed the soil-nesting bee community in established wildflower plantings and compared nest density within the plantings to the density of soil-nesting bees captured in the crop fields, adjacent wooded areas, and grassy field margins. Preferential nesting in the wildflower plantings would suggest that the plantings have the ability to enhance nest habitat for ground-nesting bees.
Our objectives were: 1) to assess how the intensity of insecticide applications affects the abundance, diversity, and richness of native bees foraging on highbush blueberry; 2) to determine the effects of wildflower plantings on soil-nesting bee communities; and 3) to develop educational resources on how to minimize the effects of pest management activities on beneficial insects.
Our audience is fruit growers – mainly blueberry, but also apple, cherry, pear, and grape producers – in western and northern Michigan. Our intermediate-term goals were to increase grower knowledge and awareness of the risks of insecticide use to beneficial insects, the benefits of using precision application equipment and IPM techniques to minimize drift and off-site movement, and the benefits of conservation plantings for maintaining abundant and diverse on-farm bee communities.
Our long-term goals included reducing the use of broad-spectrum insecticides during or near bloom, increasing adoption of bee-safe application practices (including the use of precision technology and evening/night application timing), and increasing the adoption of conservation plantings through the CRP-SAFE pollinator habitat restoration program. Adoption of practices that conserve wild bees, such as conservation plantings, can improve delivery of pollination services, increase farm profitability, and sustain and improve the environmental quality and natural resource base upon which these pollination-dependent crops depend.
For our first objective, we collected bees using pan traps and nets during bloom in 2013 and 2014 at the same 15 highbush blueberry fields sampled by Tuell et al. (2009), located in Allegan, Van Buren, and Ottawa counties in southwest Michigan (Figure 1). The lakeshore counties on the west coast of Michigan’s Lower Peninsula are characterized by sandy, acidic soils and a lake-moderated microclimate that allows for a longer growing season than other regions at the same latitude. This diverse and productive agricultural zone, sometimes termed Michigan’s “Fruit Belt,” leads the nation in the production of highbush blueberries (NASS 2011).
The 15 sampled fields span a gradient of management intensity from unmanaged fields to fields sprayed every 5-7 days for insect and disease control. All sites were more than 3 km apart. Grower cooperators provided pesticide application records for 2012-2013, the years prior to our collection years. Previous work by Tuell and Isaacs (2010) determined that the insecticide risk in the year prior to bee collections was a better predictor of bee community responses than in-year insecticide use.
For our second objective, we used soil emergence traps to collect bees nesting belowground in different on-farm habitats on four blueberry farms located in Allegan and Van Buren counties in southwest Michigan with wildflower plantings sown in 2008 or 2009 (Figure 2).
Response of bee communities to insecticide risk on blueberry farms
To measure the abundance, richness, and community composition of wild bee communities in the Michigan highbush blueberry agroecosystem, we pan trapped and net-collected bees during blueberry bloom in 2013 and 2014 at the 15 sites previously sampled in 2004, 2005, and 2006 (reported in Tuell et al. 2009; Tuell and Isaacs 2010). In all years, bees were collected using five pairs of pan traps (355 ml white and yellow plastic bowls; Amscan, Inc., Elmsford, NY) placed 5 m apart along one transect at the field edge and a second transect 25 m into the field. The twenty traps were half filled with a 2% soap solution (Dawn Ultra dish soap, Proctor & Gamble, Cincinnati, OH) and mounted on 1.2 m PVC poles stabilized with rebar (Tuell et al. 2009). Traps were placed in the morning and collected after a minimum 6 h trapping period, 2-3 times during the bloom period depending on the duration of suitable weather conditions. Bees visiting blueberry flowers were net collected for 10 minutes per sampling day, for a total of 30 minutes of sampling per site per year. Specimens were placed in the freezer, then washed and dried as described in Tuell et al. (2009).
Growers provided field-level spray records for 2003-2006 and 2012-2013. Tuell and Isaacs (2010) developed a metric of insecticide risk based on the total amount of each product applied to a field (in kg/ha), divided by the LD50 for honey bees (Apis mellifera) for that product, summed across all applications to a field in a growing season. This insecticide program risk (IPR) score reflects both the toxicity of applications to bees and the relative exposure by site; a higher IPR score represents a greater risk to bees. Tuell and Isaacs (2010) determined that the prior-year insecticide risk was a better predictor of bee community responses during bloom than the current-year IPR score, so we used prior-year applications to calculate IPR scores for our collections. Two sites provided records for only 4 of the 5 years; these sites were removed before conducting repeated measures analyses.
Bee specimens (Hymenoptera: Apoidea) were identified to genus and species using several dichotomous keys (Mitchell 1960, 1962; Bouseman and LaBerge, 1978; LaBerge 1980; Michener et al. 1994; Gibbs, 2010, 2011; Rehan and Sheffield 2011; Gibbs et al. 2013) and the online keys available through www.discoverlife.org. Specimens were compared with voucher specimens from Tuell et al. (2009) for verification. Specimens from Tuell et al. (2009) in the Ceratina calcarata/dupla species complex were reclassified according to Rehan and Sheffield (2011). Lasioglossum (subgenus Dialictus) and Andrena species were identified by Jason Gibbs (Department of Entomology, Michigan State University). Voucher specimens from this study were deposited in the Albert J. Cook Arthropod Research Collection at Michigan State University.
Nesting suitability of on-farm habitats
Mesh emergence traps (60cm2; Bioquip Products, Inc., Compton, CA) were used to sample bees nesting in the soil on four blueberry farms with wildflower plantings located adjacent to the crop field. Bees were sampled in four different on-farm habitats: the blueberry crop field, an adjacent grassy field margin, a wooded area, and the wildflower planting. The crop field, field margin, and wooded areas selected for trapping were located at least 100m from the wildflower planting to reduce potential confounding effects due to spillover from the plantings. These habitats represent the dominant land use types in the Michigan highbush blueberry agroecosystem, and were present at all sites.
Ten traps were placed at 5m intervals along a haphazardly located 50m transect in the crop field, field margin, wooded area, and the wildflower planting for a total of 40 traps per site per sample round. Traps were placed at midday (between 0900 and 1500 hours) and left for 2-3 days prior to collection. Emerging bees were captured in a jar containing 2% soap solution at the apex of the trap (Dawn Ultra Original Scent, Procter & Gamble Co., Cincinnati, OH). Specimens were strained into ziplock bags and frozen before being washed, dried, and identified. This process was repeated three times per farm from late June to early September in 2013 and 2014, for a total of six sampling events per farm. Pest management in the crop fields interfered with trap placement for several sample rounds, resulting in uneven sampling in the crop field relative to the three other habitats.
Prior to laying down a trap, I recorded microhabitat characteristics in the 60cm2 quadrat according to Potts et al. (2005) and Sardiñas and Kremen (2014). Surface soil compaction was measured in each quadrat using a pocket soil penetrometer (Forestry Supplies, Inc., Jackson, MS). Surface vegetation was cut back to a height of 4-6 inches using manual hedge shears and the area was thoroughly checked to ensure no bees were visible above the surface prior to trap placement.
To measure the compaction and texture of the soil surface, I took 10 replicate soil samples per habitat per farm in July 2013 using a drop hammer soil core sampler to a depth of 7.5 cm (331.3 cm3 core volume). In order to calculate soil moisture and bulk density, fresh soil samples were placed in #5 paper bags and weighed, then oven dried at 105°C for 24 hours. Dried samples were allowed to cool for 15 minutes prior to measuring dry weights. Soil mass was determined by subtracting the mean weight of ten empty oven-dried #5 bags from the dried sample weights. Samples from each habitat were then aggregated and processed through a 2mm mesh sieve before being sub-sampled for texture. Particle size analysis for texture determination was conducted using the hydrometer method (KBS LTER, 2008).
To determine how the community of nesting bees compared with the flower-visiting bee community, I net collected bees visiting open flowers for 20 minutes along the 50m transect in each habitat type in suitable weather conditions (temperature >65F, wind speed <3.5m/s). Based on suitable weather for sampling bees, net collections were sometimes conducted 1-3 days after trap placement.
Response of bee communities to insecticide risk on blueberry farms
We collected a total of 1,030 bees during bloom in 2013 and 2014 in pan traps and nets. Compared with the previous collections using the same methods in 2004-2006 reported in Tuell et al. (2009), average bee abundance, richness, and diversity were lower in the 2013 and 2014 collections. The changes in bee abundance, richness, and diversity by site from the earlier collections to the recent collections were not correlated with the increases in insecticide risk at those sites. However, when collections were pooled across years, a repeated measures linear mixed effects model determined that prior-year insecticide program risk at the field scale had a significant negative effect on wild bee species richness during bloom (Figure 4; F1,51 = 7.40, p = 0.009). Insecticide risk was not correlated with bee abundance (F1,51 = 1.43, p = 0.24) or diversity (F1,51 = 1.42, p = 0.24).
These results suggest that the intensity of insecticide use in highbush blueberry can affect the biodiversity of wild bee communities living in and around blueberry fields. Additional functional trait analysis will help us to determine why certain bees may be declining in response to this disturbance, while others remain abundant. Further analyses on the subset of wild bee species that are known to provide pollination services to blueberry will also help determine whether pest management practices may affect the pollination of blueberry. It is possible that the bees driving the decline in species richness at sites with higher insecticide use are rare species that are not associated with blueberry pollination. Once we know more about which species are most at risk, we can develop better strategies for conserving bees on farms with intensive pest management practices.
Nesting suitability of on-farm habitats
Captures of bees using emergence traps were fairly low relative to other trapping methods. Out of 838 total trapping events over two years, we captured a total of 49 soil-nesting bees. A Kruskal-Wallis test found that the abundance of soil-nesting bees differed significantly among habitats (Figure 5; Chi-squared = 12.29, df = 3, p = 0.006). Soil-nesting bees were significantly more abundant in traps in the wildflower plantings than in the wooded areas (posthoc Mann-Whitney-Wilcox pairwise comparisons with Bonferroni correction, p = 0.03), and marginally more abundant than in the grassy field margins (p = 0.052). Crop fields had the second-highest abundance of soil-nesting bees, but did not have significantly fewer bees than wildflower plantings or more than the other two habitats.
This study is among the first to use emergence traps to directly measure the abundance of soil-nesting bees in different habitat types. Previous work has shown that planting wildflowers in resource-limited landscapes can attract abundant and diverse bee communities to forage on those patches. This study complements this previous work by showing that bees preferentially nest in these undisturbed wildflower patches, indicating that planting wildflowers may be providing bees with two limiting resources in these landscapes: food and nesting sites. The results are promising, though further work is needed to assess the true density of bees per square meter across the duration of the growing season (see “Areas Needing Additional Study,” below). Using this method to quantify belowground bee nesting in different habitats will allow better parameterization of models of pollinator abundance and the potential delivery of crop pollination services across different landscapes.
Educational & Outreach Activities
Following the publication of the Master’s thesis (see below), this research will be published in several peer-reviewed publications. In addition, we are currently developing an Extension bulletin on best management practices for bee pesticide safety intended for fruit growers in the North Central region. The bulletin has been drafted and will be sent out for review at the end of February.
The research has been shared at several scientific meetings, as well as at grower meetings and public presentations.
May, E., R. Isaacs, and J. Wilson. In prep. Best management practices to minimize pesticide exposure to bees in fruit crops. Extension bulletin.
May, E. In prep. Effects of pest management, landscape context, and conservation plantings on wild bee communities in Michigan highbush blueberry. M.S. Thesis. Michigan State University.
Isaacs, R., J. Gibbs, E. May, E. Hanson, and J. Hancock. 2014. Pollinating highbush blueberries: Investment in this critical component of blueberry production is essential for profitable yields. eXtension. Available: http://www.extension.org/pages/71756/pollinating-highbush-blueberries
Wilson, J., L. Gut, R. Isaacs, and E. May. 2014. Minimizing pesticide exposure to bees in fruit crops. Michigan State University Extension News, 29 April 2014. Available: http://msue.anr.msu.edu/news/minimizing_pesticide_exposure_to_bees_in_fruit_crops.
May, E., J. Wilson, J. Gibbs, and R. Isaacs. 2014. Species-specific responses to pesticide use and habitat quality in wild bee communities visiting blueberry fields. Entomological Society of America Annual Meeting, Portland, OR. November 17, 2014.
Isaacs, R., J. Gibbs, A. Bennett, B. Blaauw, and E. May. 2014. Sustaining specialty crop pollination through bee conservation on farms. Entomological Society of America: 2014 North Central Branch meeting. Des Moines, Iowa. March 9-12, 2014.
Isaacs, R., E. May, and J. Wilson. 2013. Does the intensity of insecticide use for pest control affect wild bee populations? 8th SETAC Europe Special Science Symposium. Brussels, Belgium. October 16-17, 2013.
Isaacs, R., B. Blaauw, and E. May. 2013. Pollinator habitat restoration in managed lands: does it increase bees and crop yield, and can that further encourage restoration? Fifth World Conference on Ecological Restoration. Madison, Wisconsin. October 8, 2013.
May, E., J. Wilson, and R. Isaacs. 2013. Effects of pest management intensity on wild bee communities in highbush blueberry. International Pollinator Biology, Health and Policy Conference, State College, PA. Aug 14-17, 2013.
May, E. 2014. Pollinator conservation tactics for organic fruit production. Great Lakes Fruit, Vegetable, and Farm Market EXPO, Grand Rapids, MI. December 11, 2014.
May, E. Bee biology & ecology. STEM Academy, Lansing, MI. October 30, 2014.
May, E. Planting flowers for bees: How to support bees at home and on farm. Michigan State University 3rd Annual Bee-Palooza, East Lansing, MI. June 23, 2014.
May, E. What is a bee? Sycamore Elementary School, Holt, MI. April 14, 2014.
May, E., and R. Isaacs. Providing backyard habitat for native bees: Native bee nest box workshop. 27th Annual Michigan Wildflower Conference. East Lansing, MI. March 3, 2014.
The results from the first objective of this study shed light on the effects of insecticide use for pest control on the bee communities living in and around fruit farms. The intensity of insecticide use was negatively correlated with bee species richness over five years of bee collections on fifteen blueberry farms. Insecticide risk was not correlated with bee abundance or Simpson’s (1/D) diversity, which is an index that reflects the combination of species richness and evenness. More work is needed to understand which bee species exhibit strong negative responses to insecticide use, and whether these responses are mediated by bee life-history traits like body size and diet breadth. Regardless, the results indicate that caution should be used in the selection and application of insecticides for pest control if growers are concerned about conserving wild bee communities.
The results from the second objective indicate that conservation plantings may be serving a dual role in sustaining wild bees by providing both food and nesting resources. Soil-nesting bees were found to be more abundant in wildflower plantings than in other on-farm habitats, indicating that this undisturbed habitat, with its proximity to flowering resources, also provides an attractive nesting resource for these bees. This study, with its relatively novel use of soil emergence traps, provides a basis for the continued quantification of soil-nesting bee density and abundance needed to parameterize models of pollinator abundance across the landscape. These models require information on the suitability of different land use types for bee nesting, which is relatively well understood for bees nesting in cavities and wood aboveground, but not for bees that nest in the soil.
No economic analysis was performed as part of this research.
To evaluate project outcomes, we conducted an anonymous survey of apple, blueberry, and cherry growers attending the 2013 and 2014 Great Lakes EXPOs of grower pollination practices, knowledge, and research interests. In 2013, we received questionnaire responses from 65 growers in 8 US states and 3 Canadian provinces, representing over 9,665 acres of crop production, including over 6,977 acres of pollinator-dependent crop production. In 2014, we received responses from 109 growers in 13 US states and 3 Canadian provinces, representing over 19,471 acres of crop production, including 16,189 acres of pollinator-dependent crop production.
The first part of the survey asked general questions about pollination practices and hive rentals. Most growers (64%) reported using only honey bees for pollination of their crops, with 77% of these growers renting hives, 19% managing their own hives, and the rest using a combination of rented and owned hives. Another 20% of growers reported using both honey bees and commercial bumble bees to pollinate their crops, 6% use managed bumble bees only, and 10% use no managed bees for pollination. On average, growers reported a $5 increase in per-hive rental prices from the year prior to the survey (from an average of $45 to an average of $50).
The second part of the survey asked about grower adoption of different practices to protect bees from insecticide sprays, enhance bee populations on farms, and improve pollination. The survey responses were fairly consistent across years. In both years, just over half of growers reported using only bee-safe insecticides during bloom (54% in 2013 and 61% in 2014) and spraying at night during bloom (52% in 2013 and 54% in 2014). In both years, 64% of growers reported that they took care to minimize spray drift. Around 25% of growers in both years reported enhancing flowering plants on farm to support bees. A smaller proportion of growers (25% in 2013 and 17% in 2014) reported providing nesting sites for bees on farm. Growers were provided with an open-ended question about future research needs related to bees and pollination. The most common responses indicated interest in research into the causes of colony collapse disorder (CCD) in honey bees, whether certain insecticides may be causing CCD, and how to reduce the impact of chemical sprays on bees. Other growers were interested in determining how much pollination was provided by native bees compared with honey bees, and whether practices to enhance native bees on farms could be used to reduce the number of honey bee hives rented at bloom. A few growers were interested in how to manage alternative bees, like blue orchard bees, for pollination. Other research interests included determining the optimal number of hives per acre per fruit crop variety, determining the effectiveness of pollen feeding for honey bee hives, and developing methods for evaluating the quality of on-farm habitat for bees and scouting for native bees in the spring.
The study duration was not long enough to determine the long-term impact of outreach efforts; however, we received positive responses from growers involved in the project and from several attendees at the Great Lakes EXPO.
Areas needing additional study
In order to sample multiple habitat types on four farms over two years with a limited number of soil emergence traps, we decided to move all forty traps for each sampling event to maximize the area we were able to cover in each sampling round. This provided an estimate of soil-nesting bee abundance for six distinct time windows at each farm. However, it would be helpful to supplement this work with measurements of soil-nesting bee density (e.g. the number of bees nesting per square meter of soil in wildflower plantings vs. wooded areas over an entire growing season). Nesting bee density could be assessed by placing traps at the earliest possible bee emergence date and returning at regular intervals to collect emerged bees, keeping the traps in a single location for the duration of the growing season. Information on bee nesting density in different habitats could be helpful in developing nesting suitability parameters for models of pollinator abundance based on land use/land cover maps.
In addition, more work is needed in determining the optimal placement of conservation plantings on farms to maximize spillover of pollinators into adjacent crops for the provision of pollination services, while also minimizing effects of spray drift from adjacent crop fields on bees foraging or nesting in those conservation plantings. While the benefits of conservation plantings to adjacent crops may be reduced if plantings are moved farther way from crop fields, the benefits of these plantings to beneficial insects may be lost if they are placed too close to intensively managed crops.