Quantifying the Risks of Pesticide Exposure to Squash Bee Behavior and Pollination Services

Progress report for GS22-260

Project Type: Graduate Student
Funds awarded in 2022: $16,500.00
Projected End Date: 08/31/2025
Grant Recipient: University of Texas at Austin
Region: Southern
State: Texas
Graduate Student:
Major Professor:
Dr. Shalene Jha
University of Texas at Austin
Expand All

Project Information


Pesticides are widely used to prevent the loss of crop yields to insect pests. However, pesticides can also have detrimental effects on beneficial insects that contribute to successful crop pollination. While there has been much recent interest in the effects of pesticides on bees, the majority of this research has focused on honeybees (Apis mellifera) and bumblebees (Bombus spp.), with experiments largely conducted under lab conditions. However, other bee species are also critical pollinators to many crops. The majority of the 4,000 bee species in North America are solitary and ground nesting, and in Cucurbita crops (pumpkin, squash, zucchini) the specialist ground nesting squash bee (Eucera pruinosa) provides the majority of the pollination services. Recent studies have shown that insecticide exposure in semi-field experiments can reduce the reproductive success of squash bees, but how pesticide exposure impacts their foraging behavior and ultimately, population numbers, is not known. If foraging behavior or abundance is negatively impacted by exposure this could have major consequences to pumpkin production. In this study I will assess the influence of pesticide exposure on squash bee communities across sites in central Texas. Specifically, I will address how pesticides impact squash bee abundance, the contributions that individual squash bees make to pollination services in terms of pollination efficiency, and if pesticide exposure predisposes bees to be increasingly susceptible to further exposure by the two commonly used insecticides: thiamethoxam and bifenthrin.

Project Objectives:

Aim 1: Assess how pesticide exposure and landscape composition impact squash bee abundance . It is critical to understand exposure across the community, but especially for the key pollinator, E. pruinosa; this will be done To understand variation in pesticide exposure across sites we will  compare the pesticide residues detected in the bees themselves to the residues detected in the pollen and nectar from the plants, and soil from each site. Texas pumpkin farms off a useful natural laboratory for addressing questions about pesticide exposure since there is a wide range of management practices employed, from small organic farms to very large scale farms with conventional management that includes the use of traditional insecticides. Pollinator surveys on cultivated pumpkins will be conducted at 12 farms across central Texas that vary in their usage of pesticides at three time points during the blooming period. Specifically, we will compare how bee abundance is predicted by a traditional hazard assessment that only considers the concentrations of pesticides in the crop nectar versus a hazard assessment that additionally considers exposure through pollen and soil (Chan et al., 2019). This will allow us to identify the most important routes of exposure by comparing these two different hazard assessments to our measures of pesticide residues in individual bees, to see what routes of exposure are most important to consider to prevent population declines.  


Aim 2: Determine to what extent pollination efficiency is affected by pesticide exposure and landscape composition. By directly measuring if the individual contribution a squash bee makes to fruit set correlates with pesticide exposure at a site we can have a more accurate representation of potential losses. 


Understanding the abundance of the pollinators across sites in addition how much each of those pollinators contributes to crop production allows us to quantify if the bees are negatively impacted by pesticide exposure at what point we would expect to see losses to yield.  


Aim 3: Experimentally test if exposure of two commonly used insecticides have lethal or sub-lethal effects on squash bees. The effects of insecticides on pollinators are typically tested using honey bees and bumblebees; we have adapted established protocols from those species to test how field realistic exposure of two commonly used insecticides affects squash bees. Specifically we will address is propensity to forage is negatively impacted by field realistic exposure to thiamethoxam and bifenthrin with a proboscis extension response assay that can be used on wild caught bees. This will allow us to test whether insecticide effects on these responses are generalizable beyond models species, and if prior pesticide exposure in the environment leads to greater susceptibility to these effects.


Materials and methods:

Study design: We will be working on 18 farms in west Texas that all grow pumpkin. For each year, every site will be visited 3 times during the blooming period from May to August. Because all three objectives of this study rely on variable pesticide exposure across sites these sites will span the available range of management practices (6 with organic management practices and 6 with conventional management involving the use of insecticides, with variation in practices within each group). During the second of the three visits we will collect two samples of each pollen & nectar, soil, and squash bees (E. pruinosa) which will be sent to the McArt Lab at Cornell University to be analyzed for pesticide residues. This will be 72 samples per year for a total of 144 samples over the two field seasons. Landcover surrounding each site will be measured with ArcGIS using CropScape data layers.

Aim 1: Assess how pesticide exposure and landscape composition impact squash bee abundance and identify probable routes of pesticide exposure. We will conduct visitation surveys during each of our three visits to the sites. Surveys will consist of walking 44m transects for 20min and recording the identity of each bee that is observed landing on a flower as has been done in other studies of the pollinators that visit pumpkin (Petersen and Nault, 2014). We will also net all bees that we see visiting flowers for 10min so that species level identifications can be verified.


We will collect our pollen and nectar from 2 flowers during this second site visit, when the plants are likely to be in the fullest bloom, and so are a major floral resource for the bees. Two 10g soil sample will be taken from the center of the field to be consistent in the location of sample collection from each site and since squash bees are known to nest in crop fields (soil sampling per site  as in Chan et al. 2019). All samples for pesticide residue analysis will be collected onto dry ice and then stored at -80 degrees until they are shipped to the McArt lab. We will run generalized linear models (GLMs) to assess the impact of surrounding agricultural landcover, floral density, and pesticide hazard on community level pollinator richness and pollinator abundance. We will calculate hazard quotient based on the residues detected to inform if the risk to bees is actually below the EPA level of concern at our sites (negligible risk to mortality as 5% of the honeybee LD50) when exposure only through the nectar is considered compared to when pollen and soil exposure are factored in as well.


Aim 2: Determine to what extent pollination efficiency is affected by pesticide exposure and landscape composition. To quantify how the contributions of individual bees to crop pollination are influenced by pesticide exposure we will conduct single visit pollination efficiency trials at each of the sites.To do this, we will put mesh drawstring bags over 20 female flowers the night before they bloom. The following morning, we will remove each bag and wait for a bee to visit the flower (Canto-Aguilar and Parra-Tabla, 2000). While the bee is in the flower we will record the duration of the visit and the amount of the contact it has with the stigma. As the bee is exiting the flower we will capture it to be definitively identified to species and to have intertegular distance (ITD) recorded as a measure of body size. This will be the only visit to this flower, and the bag will be placed back over each of the 10 single visit flowers. We will not allow any bees to visit the other 10 flowers, so that we can calculate the benefit to yield (difference between a single visit and no visitation) that a single visit causes. Once the flowers have senesced we will remove the bag and tag the developing fruit. Once the pumpkin has matured we will measure its mass (Pfister et al., 2017).


This will allow us to quantify the variation in pollinator efficiency in terms of contribution to yield within this species. We will run generalized linear models (GLMs) to assess the impact of natural habitat cover, floral density, stigmatic contact, visit duration, and pesticide hazard on the benefit a single visit has to yield.


Aim 3: Experimentally test if exposure of two commonly used insecticides have lethal or sub-lethal effects on squash bees. The average concentration of thiamethoxam detected in pumpkin nectar following seed treatments has been measured at 17.6 ppb, while bifenthrin has been detected at an average concentration of ppb in Cucurbita nectar (Dively and Kamel, 2012) and both of these concentrations are below the level of concern defined by 5% of the honeybee LD50, so we will feed bees 4 microliters of 10% sucrose solution at these concentrations, in addition to a set of control bees that are fed untreated sucrose solution as our "medium" dose group. One hour after dosing we will record if there is any mortality and then the bees will be restrained in tubes to undergo a sucrose responsiveness assay modified from Carlesso et al. 2020, which will allow us to determine if treated bees are less responsive to increasing sucrose concentrations, which is a proxy for foraging propensity (Scheiner et al., 2004). We additionally added in a "low" dose based on the lowest concentrations of thiamethoxam detected in pumpkin nectar following seed treatments (Dively & Kamel, 2012), and concentrations of bifenthrin detected in pumpkin nectar of similar toxicity relative to the honeybee LD50 (Frazier et al. 2015).

To better understand the relationship between the honeybee LD50 and toxicity to other bee species we also added in a "high" dose treatment that was not necessarily intended to be field realistic but to be the same in terms of relative toxicity compared to the honeybee LD50, both insecticides were fed to the squash bees at a dosage that was 20% of the honeybee LD50. If the honeybee LD50 can be scaled to accurately predict toxicity to other bee species we would have expected no difference between the two insecticide treatment groups. 

The current risk assessment framework relies on defining what pesticide concentrations in nectar minimize acute lethality in honeybees, but these are not the only bees providing pollination services to agricultural systems. Pumpkin is a pollinator dependent crop that is primarily pollinated by the solitary squash bee (E. pruinosa), and it is unknown to what extent this risk assessment framework actually prevents negative effects on these bees. Additionally, it is critical to assess if a risk assessment framework based on acute lethality is actual represents a level of risk that minimizing losses to pollination services due to effects on the abundance of important pollinators and the pollinators’ foraging behavior.

Research results and discussion:

Aim 1 & 2 depend on the pesticide residue analysis, and we will be able to do this most effectively with an extension as the McArt lab at Cornell University is introducing a new pesticide residue analysis protocol that will allow us to detect all of the most commonly used pesticides at our sites. Currently we would not be able to detect the pyrethroids, and we know that bifenthrin (a pyrethroid) was the most commonly applied insecticide during the sampling period.

Aim 3: Experimentally test if exposure of two commonly used insecticides have lethal or sub-lethal effects on squash bees. We have conducted the sucrose responsiveness assays with our 'low' dose group (based on lowest concentrations of thiamethoxam and bifenthrin detected in pumpkin nectar (Dively & Kamel, 2012; Frazier et al. 2015)) at 0.032ng of thiamethoxam per bee, and 0.017ng of bifenthrin per bee. Our 'medium' dose group was based on the average concentration of thiamethoxam detected in pumpkin nectar (Dively & Kamel, 2012) and a dosage of bifenthrin that was matched to be the same percentage of the honeybee LD50 as thiamethoxam due to limited data about bifenthrin residues in pumpkin nectar.at 0.074ng per bee of thiamethoxam and 0.157ng per bee of bifenthrin.

To better understand the relationship between the honeybee LD50 and toxicity to other bee species we also added in a "high" dose treatment that was not necessarily intended to be field realistic but to be the same in terms of relative toxicity compared to the honeybee LD50, both insecticides were fed to the squash bees at a dosage that was 20% of the honeybee LD50. If the honeybee LD50 can be scaled to accurately predict toxicity to other bee species we would have expected no difference between the two insecticide treatment groups. 

Although our analyses are ongoing, preliminarily we find no differences across groups of bees from the 'low' dosing scheme (Fig 1A), we do find a potentially interesting difference in responsiveness only at 50% sucrose in our 'medium' dosing scheme (Fig 1B), and most surprisingly we found complete mortality in the thiamethoxam group at our 'high' dosing scheme (Fig 1C).

Figure 1: preliminary sucrose responsiveness data across the three dosing schemes. Sample sizes, doses (ng), and doses relative to the honeybee LD50 are reported in the right column. X-axis shows sucrose concentrations in the order they were presented, while the y-axis shows the proportion of bees that exhibited proboscis extension response (PER) to that concentration of sucrose. 

We expected all of our doses only elicit sublethal effects to sucrose responsiveness as none were approaching high levels of toxicity compared to the honeybee LD50. Most interestingly, we find a dramatic difference between exposure to thiamethoxam and bifenthrin at this 'high' dose as our bees dosed with bifenthrin did not differ from our control bees in terms of sucrose responsiveness (and mortality - as there was none). This could have implications for how reliably we can apply the honeybee LD50 across species even when we include scaling factors that are meant to be conservative. 

Participation Summary
2 Farmers participating in research

Educational & Outreach Activities

1 Webinars / talks / presentations
2 Other educational activities: High school research internship program participants

Participation Summary:

10 Ag professionals participated
Education/outreach description:

Talks: I gave an invited talk about the preliminary sucrose responsiveness data which was collected for aim 3 of this project at the Entomological Society of America conference. The number of 'agricultural professionals' that I listed as participants above is an estimate, but it was a full room for the talk and I know a subset were employed with Bayer and other similar companies that play a large role in agricultural practices. 

Other education activities: I run a program at UT Austin that brings high school students onto campus twice per week during the school year to get involved in research and generally demystify the research process, and during the course of the work I've been doing with squash bees I have had two high school interns help with sample processing and measuring bee body size.

We are also hoping to finalize the analyses for the sucrose responsiveness data and submit the manuscript over the upcoming months. 

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.