Interactive effects of pesticides, drought, and pathogens on the common eastern bumble bee Bombus impatiens.

Progress report for GNE21-264

Project Type: Graduate Student
Funds awarded in 2021: $15,000.00
Projected End Date: 07/31/2023
Grant Recipient: University of Massachusetts
Region: Northeast
State: Massachusetts
Graduate Student:
Faculty Advisor:
Dr. Lynn Adler
University of Massachusetts Amherst
Expand All

Project Information

Project Objectives:

Objective 1. I will determine how drought stress affects pesticide concentration in pollen of three major US agricultural crops whose seeds are treated with systemic pesticides.

Hypothesis 1. Drought stress will increase pesticide concentrations in pollen from seed-treated crops.

This experiment has already been initiated, greatly increasing its chances of success on a short timeline. However, pesticide analysis is expensive, and funds from NE SARE would allow me to substantially increase the sample size and therefore my ability to detect treatment effects. I have already begun growing three crops that are seed-treated with pesticides: sunflower, cotton, and squash. Plants of each crop will be randomly assigned to one of three drought treatments (well-watered, moderate water, and drought).  I will collect pollen samples throughout the summer and send them to the Cornell Chemical Ecology Core Facility for pesticide testing. If I find that drought increases pesticide concentrations in pollen, this will shed light on how climate change can aggravate the risk of exposure for bees to pesticides used in major crops. Even if the drought does not affect pesticide concentrations, this project will quantify ranges of pesticide residues across different crops, providing more accurate assessments of potential pollinator exposure.

Progress for Objective 1: The pollen samples were successfully collected from the three crops grown in the greenhouse: cotton, sunflower, and squash. The collection was extended two months into the Fall semester, to gather enough material for pesticide analysis. Then, the samples were sorted, pooled, and sent to the Cornell Chemical Ecology Core Facility for pesticide testing. I expect the results over the next few weeks, and will then proceed with data analysis.

Objective 2. I will assess the effects of chronic exposure to neonicotinoids on pathogen (Crithidia bombi) transmission dynamics in bumble bee (Bombus impatiens) colonies. 

Hypothesis 2. I predict that bees exposed to pesticides will interact less in the colony, potentially leading to lower transmission rates of C. bombi. However, individuals who become infected will have a high infection intensity due to a weakened immune system in response to pesticides.

I will conduct this laboratory experiment during the 2021-22 academic year, using an automated tracking system that allows me to record interactions between bees in the colony, after which I will assess pathogen prevalence and infection intensity. If exposure to pesticides reduces interaction rates by disrupting social behavior35, bees may be less likely to acquire pathogens from their nestmates. However, if exposure to pesticides increases individual bee susceptibility to pathogens, the transmission may still increase even if contact rates are decreased or unaffected. Previous studies have shown that pesticides weaken bee immune systems12,36, making them more susceptible to pathogens37. However, there is little information about pathogen transmission patterns inside the colony, which is critical to address.

Progress in objective 2: We are testing the software and coordinating with our lab collaborators Dr. James Crall and his team, to assemble the tracking systems. I expect to perform the first assay this Spring semester.

Introduction:

The purpose of this project is to elucidate the implications of drought, pesticides, and pathogens for bees in agricultural landscapes. Pollinators are diminishing at concerning rates1, threatened by multiple stressors, including pesticides and pathogens. Worldwide pesticide use in agriculture has been expanding dramatically over the last 65 years2, particularly for neonicotinoid insecticides3. In the US, neonicotinoid use has increased by 80% since they were introduced in 19944; at least 300 million acres of US cropland are treated with neonicotinoids today4,5. Ninety percent of those treatments are systemic applications that coat seeds in major field crops, such as corn, soybean, wheat, cotton, and sorghum4. These chemicals are taken up by the plant after germination, persisting for months throughout all tissues. Unfortunately, neonicotinoids are highly prevalent in ecosystems, and accumulate in soils and waterways, making it possible for non-target wild plants to absorb them6.

Bees can be exposed to neonicotinoids and other systemic pesticides while feeding on pollen and nectar7–9 from crops and wild flowers10. Chronic exposures are likely to occur in agricultural fields, causing learning and memory impairment, and increasing susceptibility to several pathogens with impacts on productivity and nest performance11. For example, sublethal exposure to neonicotinoids increased the proliferation of Nosema ceranae12–14, black queen cell virus (BQCV)14, and deformed wing virus (DWV)15 in honeybees, compromising bee survival. The simultaneous presence of pathogens and pesticides in crops exposes pollinators to a complex source of potentially interactive effects that we do not fully understand.

The effects of pesticides and pathogens on colony dynamics have been assessed separately but not in combination. Recently, Crall et al.16 developed video software and ID tags to track individual bumble bee workers within the colony, demonstrating that exposure to field-realistic doses of neonicotinoids altered colony behavior, disrupting nursing and thermoregulatory efforts. Another recent study focused on pathogens showed that a higher proportion of infected individuals introduced to the colony increased infection intensity and prevalence in bumble bee (Bombus impatiens) colonies17. However, the mechanisms of how pesticide exposure influences pathogen transmission within colonies have not been studied.

Extreme climatic events such as drought are becoming more frequent and severe in the world18, and can further influence resource quality and pesticide exposure. Drought constrains the plant's physiological development, resulting in lower photosynthetic rates, reduced growth, and lower yields19,20. Drought may also affect pollinators by reducing the nutritional value of pollen and nectar 21, and by increasing pesticide concentrations in pollen22. Thus, increasing drought frequency could have negative effects on pollinators, but these effects are almost entirely unexplored.

Research

Materials and methods:

Objective 1. I will determine how drought stress affects pesticide concentration in pollen across three major US agricultural crops whose seeds are treated with systemic pesticides. Hypothesis 1. Drought stress will increase pesticide concentrations in pollen from seed-treated crops.

Crop choice and plant propagation

The crops selected for the study represent a wide variety of agricultural uses: sunflower (seeds, oil, and cut flowers), squash (fruits), and cotton (fiber). Crops were chosen using the following criteria:

  • Reliance on pollinators for yield: Yields in our focal crops are improved by pollination; sunflower38 and squash39 productivity are entirely reliant on pollinators. Although cotton is a self-pollinated crop, bee pollination enhances production by 12%40.
  • Production of enough pollen for chemical analysis: Sunflower and squash are exceptionally abundant pollen producers, and cotton should also produce enough pollen to permit pesticide analysis.
  • Use of seed-treated material: These three crops are commonly seed-treated with pesticides, including more than half of cotton plantations (62%)24, and nearly all commercial sunflower crops41. Treated squash seeds are readily available ( obs.), but there are fewer statistics about national use.
  • Regional importance: Two of the three crops chosen for this study are commonly grown in New England, and the third is abundantly grown in the southern US. All are economically important nationally (survey 2020): sunflower (1,558 million acres), squash (46,700 acres), and cotton (12 million acres)42.

This semester I started growing the three crops in the Research and Education greenhouses at UMass Amherst. Plants were provided with a 14-hour photoperiod and an ambient temperature of 70oF, to ensure optimum growth. Seeds were purchased from commercial providers used by farmers. Each crop has been treated with a mixture of pesticides, all including the neonicotinoid thiamethoxam (sunflower, squash) or imidacloprid (cotton). I have provided full irrigation, assuring plants are vigorous before starting drought stress treatments. 

I am growing 90 plants per crop that will be randomly assigned to one of three treatments. Crops were transplanted to 3 gallons (sunflower), 5 gallons (squash), or 2 gallons (cotton) pots, according to crop requirements for blooming. Following transplanting, all crops were supplemented with a controlled-release fertilizer (Osmocote, 14-14-14). The soil substrate is ProMix BX enhanced with mycorrhizae (75% Canadian sphagnum peat moss, perlite, vermiculite, and a pH adjuster). Cotton plants were attacked by aphids, requiring foliar applications of the insecticide soap, M-PEDE (twice per week for two weeks). 

Study design

Drought treatments will have volumetric water content (VWC) percentages of >30% (well-watered), 20-29% (moderate water), or < 20% (drought). Each pot will be regulated by an individual emitter, connected to the main water hose provider controlled by a timer (Melnor ref. 65038). Drought treatments will be set at the pre-blooming phase (50, 40 or 90 days after sowing for sunflower, squash, and cotton respectively). Plants will be arranged in randomized complete blocks on greenhouse benches. I will monitor and record soil water content two days per week using a moisture sensor, and record crop growth metrics including height, flower diameter, number of flowers, and first bloom date.

I will collect pollen samples until plants senesce or until I have 14 samples per treatment per crop; I expect to collect samples throughout the summer. I will tailor collection strategies per crop to maximize pollen collection.

  • Sunflower: Floral heads will be positioned face down over a piece of paper and gently shaken to discharge pollen. Since sunflowers are compound flowers with florets that bloom over time, this process will be repeated every two days as long as the inflorescence lasts (usually 2-3 weeks).  
  • Squash:Anthers will be carefully scraped to remove pollen. Plants produce male flowers for up to two months, but individual flowers only last for one morning. I will collect the samples every day in the morning. 
  • Cotton: I will collect complete anthers every two days since cotton does not produce abundant pollen; the plants bloom for 2-3 months.

All pollen samples will be stored on 1.5 ml Eppendorf vials, labeled, and placed immediately on dry ice in a cooler during collection. Then, I will transfer all the samples to a freezer to store at -20oC.

Pollen sample analysis

I will use the Cornell Chemical Ecology Core Facility services to quantify pesticide concentrations. The Core Facility team has extensive experience processing pollen with Liquid Chromatography/Mass Spectrometry (LC/MS).

The minimal amount of pollen required per sample is 0.5 g. Ideally, pollen from each plant will be a separate sample, but I will pool plants within treatments when necessary, to obtain sufficient material. I anticipate this might be more critical for cotton since sunflower and squash produce copious pollen. Thus, I am likely to have fewer pollen samples than the number of plants per treatment. Other lab funds will allow me to analyze 4 samples per treatment per crop, however, funds from a NE SARE grant would allow me to include 10 more replicates per treatment per crop, greatly increasing my power to detect effects.

Statistical Analysis

I will use software and packages in R to analyze the data. I will compare the effect of drought treatments separately for each crop on growth measures (height, flower diameter, number of flowers), and pollen pesticide concentration (ng/sample), using generalized linear models with Gaussian (height, flower diameter, pollen pesticide concentration), or Poisson (number of flowers) distributions. Results from this study will tell us if the presence of drought events increases the concentration of pesticides in pollen across the three crops and whether there is a corresponding response in plant development. This is important because it will assess whether drought increases the risk of chronic exposure to pesticides for pollinators.

Objective 2. I will determine the effects of chronic exposure to neonicotinoids on pathogen (Crithidia bombi) transmission dynamics in bumblebee (Bombus impatiens) colonies. Hypothesis 2. I predict that bees exposed to pesticides will interact less in the colony, potentially leading to lower transmission rates of C. bombi. However, individuals who become infected will have a high infection intensity due to a weakened immune system in response to pesticides.

Colony tracking system

I will use a specialized image-based tracking system recently developed by our collaborator, Dr. James Crall. The system uses BEEtag43 software (BEhavioral Ecology tag), to identify and track individuals within the colony. The tracking software works with unique tags (size: 3x4mm, weight: 2.4 mg) that I will glue to the top of each bee’s thorax while they are sedated. Tags will be printed on waterproof, tear-resistant paper.

My lab has recently acquired a tracking system prototype. Throughout the spring semester, I have been training to use this cutting-edge technology. We have funds to buy two more systems, allowing me to assess three microcolonies (a queenless group of 15 bumble bee workers) at a time, allowing one replicate of each pesticide treatment. However, I would need to conduct multiple repetitions to generate enough replicates for statistical analyses. The NE SARE grant would provide funds to buy nine more colony-tracking systems to allow me to conduct four replicates at a time, making the experiment much more time-efficient and reducing noise associated with variation across time, increasing my chances of identifying treatment effects.

Study design

I will make microcolonies of 15 bee workers using commercial bumblebee colonies (Bombus impatiens, Koppert®). Microcolonies can be used to estimate reproduction and mimic colony conditions while allowing for more replicates than could be conducted with full colonies44.

I will make 3 microcolonies from each of 4 parent colonies, assigning each microcolony randomly within parent colony to one of 3 pesticide treatments (4 replicates per treatment, blocked within parent colony). After completing the first round I will repeat this process for a total of 8 replicates per treatment; a similar recent experiment by our collaborator successfully found treatment effects of chronic imidacloprid exposure using 9 replicates per treatment35. I budgeted for 6 parent colonies because some may be used twice. To choose the pesticide doses, I will use concentrations detected in pollen from the three drought treatments in Obj 1, using the crop with the highest overall concentrations. If the drought treatments do not significantly affect pesticide concentrations, I will use the following concentrations: none (0 ng), low (0.1 ppm; most typically found in crops), and intermediate (1 ppm; tested on bumble bees previously) 45. I will use imidacloprid (PESTANAL®, Sigma-Aldrich) mixed with wildflower pollen (BioBest®) provided continuously to colonies for chronic exposure. In our experience, adult worker bees regularly consume pollen and survival is much lower for bees not provided pollen46, so we are confident that pollen will be consumed. I will prepare stock solutions of the pesticide by dissolving 5.0 mg of imidacloprid in 50 mL dH20 and achieve the specific concentrations for treatments using aliquots mixed with pollen to make a paste to feed the bees.  Then, I will inoculate two individuals per microcolony with the gut pathogen C. bombi following standard protocols in my lab47,48 to simulate conditions where newly-exposed foragers return to the colony. Interactions between bees in the microcolony will be recorded for two weeks, after which I will dissect all bees to count C. bombi cells, assessing pesticide effects on transmission of these pathogens.

Each microcolony will be equipped with a Raspberry Pi 4 Model B 8GB RAM that operates as a desktop computer; the BEEtag software will be installed in the Raspberry Pi and automated with Matlab for recordings. The colony will be configured to record contact-analysis videos in the nest using a camera positioned on top of each microcolony. Individual contacts between all workers in each microcolony will be recorded for five minutes every two hours for two weeks using the BEEtag software. Microcolonies will be under red lights because the visual system of bumble bees is less sensitive to this color49; as a precaution, the lights will be turned on for 10 minutes before recording to allow bees to habituate.

To assess whether pesticides affect pathogen infection independent of social interactions that affect the transmission, I will conduct separate individual trials feeding infected bumble bee workers to each concentration of imidacloprid using the same diet treatments. After two weeks I will dissect bees and assess infection intensity. This experiment will test the hypothesis that pesticides will increase individual infection intensity due to a weakened immune system, and these results will be used to inform our interpretation of how disease spreads in the microcolony experiment. 

Statistical analysis

To estimate physical contact between bee workers, I will follow a threshold method used by Dr. James Crall16. Workers will be considered interacting when the distance between them, measured from the tags’ centroids, is less than one cm. To obtain this information, I will set the software to identify pairs of bees (BEEtags) tracked, analyzing the images per segment; the software places the BEEtags to allow us to get tag locations. Effects of imidacloprid exposure on number of bee contacts and disease transmission metrics (prevalence: presence/ absence, infection intensity: cell counts) will be estimated using a hierarchical bootstrapping approach using generalized linear models in R. For each bootstrap replicate, colonies will be resampled with replacement, and the difference in each transmission metric (between treatment and control groups) will be averaged across all trial periods (measured continuously from the beginning of the experiment) to account for variation both within and across days. For the individual trials, I will use generalized linear models in R, using binomial distribution (presence/absence of infection), and Poisson (Crithidia cell counts for infection intensity).

The results from this study will tell us how chronic exposure to systemic pesticides affects behavior within the colony, individual susceptibility to infection, and transmission dynamics, allowing us to understand a potentially important but understudied consequence of pesticide exposure for the sustainability of agricultural landscapes.

Updated Timeline:

September – December 2021: I finished the greenhouse experiment and sent the samples for pesticide testing for objective 1.

I started growing the three crops in the Spring, before the official start date of the grant, and I successfully collected pollen over the summer and part of fall 2021. The drought treatments were implemented by hand instead of the proposed automatic system because the greenhouse presented issues with water pressure.

After concluding the greenhouse work in October, I sorted the pollen stored in the freezer over the season, pooling within each crop and treatment to make samples weighing 0.5 -1 g. This was a laborious process since I had thousands of pollen samples from single flowers that had to be sorted and combined. The SARE funds allowed me to increase the number of samples by over 12 replicates per crop. In December, I sent 97 samples to the Cornell Chemical Ecology Core Facility for pesticide testing. Details on the quantity are in the following table:

Crop

Drought Treatments

# Pollen Samples

Cotton

Full

7

Mid

6

Drought

4

Subtotal

17

Sunflower

Full

14

Mid

13

Drought

12

Subtotal

39

Squash

Full

16

Mid

14

Drought

11

Subtotal

41

Samples Total

97

In the next few weeks, I expect to receive the pollen pesticide results. I will then complete the statistical analysis and write up the data for publication.

February – August 2022: Conduct experiment for objective 2

I will order commercial bumblebees and first conduct a separate experiment, isolating the pesticide effect on pathogen infection without social interactions. While I work on that, I will coordinate with our collaborators to assemble 11 microcolony observation boxes (we have one already) and perform tests to make software adjustments by April.

I will conduct the trials in mid-May through June using the microcolonies equipped with the Raspberry Pi 4 Model B, recording bee network interactions. Then, I will dissect the bees to determine infection intensity.

I will spend July and August analyzing hours of video recordings to estimate the number and different levels of interaction for each bee worker within the colony.

September – December 2022: Complete the statistical analyses, and write up the data for publication.

 

Participation Summary

Education & Outreach Activities and Participation Summary

Participation Summary:

Education/outreach description:

I will disseminate my results to multiple audiences using oral and written strategies, as I have in previous outreach efforts. I worked in my home country of Colombia with cocoa farmers for two years; I learned the power of combining communication strategies such as posters, workshops, or short talks to create bonds with audiences that might feel intimidated and not encouraged to read scientific publications.

Farmers: Local farmers are my principal audience. I plan to disseminate my results via talks, fact sheets, and posters, as follows:

A) The UMass Extension website is well respected among New England farmers for its current and relevant information. In partnership with this effort, my lab runs an extension pollinator education program, led by Hannah Whitehead (https://ag.umass.edu/resources/pollinators). I plan to work with Hannah on creating fact sheets to share on the website. I will also take advantage of a solid relationship that my advisor Lynn Adler has built with local squash and sunflower farmers48,50. Such connections allowed me to conduct a study at 14 sunflower farms in 2019. I will plan to share my fact sheets through email and invite questions and feedback for future research directions.

B) I plan to present talks at popular annual local events such as the BeeFest at Greenfield, MA, and the Twilight Grower’s meeting at the UMass South Deerfield Agronomy Farm; these are excellent opportunities to interact with farmers and the general community, understanding their needs.

C) I will create short and engaging communications for local magazines and farmer's markets. I will also create posters for bulletin boards in coffee shops and libraries with QR codes linked to upcoming outreach events and the fact sheet I create.

Students: I will take advantage of my previous experiences to target students at different stages of their education, from K-12 to college and graduate school, using venues available through UMass.

 A) For the last two years I have been a member of the Fernald Club, an entomology-based group of undergraduate and graduate students at UMass. I am in charge of the live insect collection, which has allowed me to interact with students at all stages and support outreach activities. This is a fantastic mechanism to reach a broad audience of students across all academic phases with workshops and talks.

 B) I will organize a two-day workshop for Eureka!; this is an exciting program that I am proud to be part of, hosted by Girls, Inc. in partnership with UMass. Workshops hosted by research labs are offered each summer to middle school girls from underrepresented communities to support their involvement in STEM fields. I plan to use the novel software system to encourage participation and spark students’ natural curiosity.

Conservation agencies: I was an intern at the Massachusetts Department of Environmental Protection (MassDEP) during 2017-2018, and I am aware of opportunities for short talks on broad environmental topics. I presented my work there previously, finding an exciting audience because of their involvement with wider communities. I will present a talk during summer 2023.

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.