How well are we safeguarding all bee species with current safety testing of crop protection products? This question is at the very core of sustainable agriculture practices that seek to balance crop productivity and environmental considerations. Bayer is part of an international effort to provide answers and create guidelines to effectively assess the risk posed by pesticides to bee species other than honey bees. The process is challenging, and progress demands all the knowledge and creativity from experts around the world.
// Regulatory authorities, scientists and environmentalists are calling for expanded bee safety testing of crop protection products to cover bee species other than the honey bee.
// Efforts to create testing systems that account for the vast diversity of bee species are ongoing with bumble bees, mason bees and leafcutter bees.
// This global initiative will provide integral knowledge to support cross-species pollinator protection goals.
For more than 20 years, required safety testing of compounds destined to be marketed as crop protection products (pesticides) has included a battery of tests on the honey bee (Apis mellifera L.). In a tiered approach that already starts during early stages of development, a compound is first tested in the laboratory to determine its intrinsic toxicity to bees. If the compound exceeds a certain threshold, use patterns are designed to mitigate risk, or higher-tiered testing is conducted to assess bee safety in a more realistic semi-field or field setting.
The choice of the honey bee for testing makes sense. Our relationship to honey bees, and the clear benefits we derive from them, dates back several thousands of years (Roffet-Salque et al. 2016). Thus, society as a whole has a vested interest in protecting them. For the same reason, we also know how to breed honey bees so they are available, as needed, for testing. Yet, growers increasingly rely on other managed bee species to pollinate their crops; bumble bees are used in greenhouses, mason bees are used in orchards, and leafcutter bees are preferred for alfalfa. Unmanaged wild bees also provide pollination services, playing a significant role in certain landscapes.
Recognizing this diversity, regulatory authorities are calling for expanded safety testing that encompasses a broader range of bee pollinators. In response, scientists around the world are evaluating to what extent results from testing on honey bees are applicable to protect other pollinators. They are also working on developing standardized testing systems to assess risks posed by pesticides to non-Apis bee species, that is, bees other than honey bees.
The task has not been simple. Even with just a fraction of the over 20,000 species of bees buzzing worldwide, ensuring that the use of crop protection products is safe for bees requires knowledge and experience that, as we will see, are in the lengthy process of being gathered.
It stands to reason that tests cannot be developed for each individual bee species, and applying the same battery of tests to all bees is also not an option.
The situation is complicated by the incredible diversity in behaviors, feeding preferences, life cycles and physiologies of bee species. Current efforts on non-Apis bee testing focus on commercially available bumble bees (Bombus terrestris, B. impatiens), mason bees (Osmia cornuta, O. bicornis, O. lignaria) and leafcutter bees (Megachile rotundata). Stingless bees (Meliponini) are also slated for future development of standardized tests. Alone, these species already represent a broad spectrum of biologies: some form colonies, others are solitary; some build wax nests, others use mud or leaves; some keep reserves of nectar and pollen, while others don’t. To an important extent, the success of a testing system in determining the potential effects of a pesticide is governed by the biology of the tested bee.
Thus, researchers are using existing test systems for one species of bee and tweaking them to ensure the best possible testing performance for another species. If that does not work, they have to design completely new tests. Either way, establishing a standardized test for a given bee species takes years of iterative examination and discussion to arrive at a robust system that all can agree upon.
So far, the process has been a learning experience full of surprises and challenges. The following are some lessons learned.
To an important extent, the success of a testing system in determining the potential effects of a pesticide is governed by the biology of the tested bee.
One of four working groups of the International Commission for Plant-Pollinator Relationships (ICPPR), a non-profit organization of volunteer scientists, is the Bee Protection Group, a non-profit organization of volunteer scientists that develops, improves and harmonizes testing methods and risk evaluation procedures for the protection of bees within the context of sustainable agricultural practices.
The group has stimulated scientific debate and generated recommendations for regulatory decisions and guidelines for over 35 years.
The development of a standardized test simply takes time. Guidelines for testing are reviewed, approved and published by agencies like the European and Mediterranean Plant Protection Organization (EPPO), the U.S. Environmental Protection Agency (US EPA) or the Organization for Economic Cooperation and Development (OECD). These guidelines are based on a recommended protocol developed and tested by experts over several years. Consider, for example, the repeated exposure test for honey bee larvae – an important component of tier 1 testing because the physiology of developing larvae is vastly different from that of adults. Every detail of the test is precisely defined – from temperature to number of larvae used and amount of food fed to larvae. These specifications ensure that the test is conducted in the same way by any tester so test performance is comparable. Details are optimized in so-called ring tests, where researchers from various institutions and countries perform the same test in a series of iterative cycles. Results are examined and the test protocol and design are improved with each cycle, until everyone agrees that the test procedure is clear, doable and delivers consistent and reproducible data. All said, the repeated exposure larval test for honey bees took over ten years to develop and ring-test; and this with a bee species that is well-characterized.
In the case of non-Apis bees, the development process can either build on previously gained experience with other species or can take even longer because experts must discover all relevant details step by step. Dr. Ivo Roessink is a scientist at Wageningen Environmental Research and has been a member of the Bee Protection Group of the International Commission for Plant-Pollinator Relationships (ICPPR) since 2012 (see ‘An international Forum for Bees’). Discussing his work as chair of the subgroup developing tests for solitary bees, he explains: “Unlike bumble bees, the mason bee was rather unknown to participants and so, even with two to three years of ring-trial data on the acute contact test, we are still working out newly discovered issues with the acute oral test.” In contrast, guideline proposals for the acute oral and acute contact tests for bumble bees were submitted to the OECD for review and approval in 2017. The ring-trials for those tests took only three years to complete, “simply because we had more experience with bumble bees,” says Roessink.
Another reason for these lengthy timelines in working with mason bees is a narrow testing window imposed by the natural life cycle of these animals. Developing Osmia bicornis, O. cornuta and O. lignaria overwinter in cocoons before hatching as adults the following spring. “So, in one spring you can get some tests done, but if hatching success is what you are measuring, you will have to wait until the following year to collect data,” explains Roessink.
The less that is known about the bees subjected to testing, the longer the process to develop appropriate tests takes, simply because the information needed to make numerous critical decisions must be collected first. A new subgroup was officially established within the ICPPR Bee Protection Group in 2016 to address testing needs for stingless bees in the tropics, for example South America. The group is currently debating which bee species to focus on. “The group faces several challenges,” explains Roessink. “South America has so many ecoregions and consequently bees, that one country may prioritize a different species than another. It will take time, but we will establish a common working plan and then build the expertise and knowledge needed for test development. The rest of us in the Bee Protection Group hope to facilitate that process with our own experiences.”
Bumble bees held individually during testing. The hair roller-like chambers allow the bees to sense the presence of others, which appears to be important for this bee species to feed willingly.
Protocols as simple as an acute oral test can be thwarted by bee idiosyncrasies. To gain knowledge about whether and how much of a compound has an impact when consumed by honey bees, ten individuals are placed in a chamber attached to a syringe filled with a known mixture of test compound and sugar water. The bees lap the mixture from the syringe, and the tester then knows exactly how much was consumed by the group. Honey bees share food, going as far as feeding each other. Thus, in this particular test setup it is fair to assume that each bee in a group eats roughly the same amount of the mixture.
“As it turns out, however, bumble bees won’t eat if they feel, well, alone.”
Dr. Nina Exeler
It seemed straightforward to implement a similar testing system with bumble bees, which are also social. “Not so,” says Dr. Nina Exeler, who leads the Experimental Unit Bees at the Ecotoxicology Department of Bayer Crop Science in Monheim, Germany. Her team is heavily involved in developing and scrutinizing testing systems for non-Apis species, and Exeler chairs the bumble bee subgroup of the ICPPR Bee Protection Group. “Bumble bees don’t share food. So, to determine consumption of one individual, we had to put single bees into chambers with a feeding syringe. As it turns out, however, bumble bees won’t eat if they feel, well, alone.”
Dr. Nina Exeler (left), Laboratory Leader of the Experimental Unit Bees of the Bayer Ecotoxicology Department in Monheim, Germany, visiting an alfalfa seed grower with her colleague Dr. Ana Cabrera (right) from the Bayer Pollinator Safety Team in the USA.
A solution to accommodate this apparent need for company came from discussions within the ICPPR bumble bee subgroup: perforated chambers that look like hair rollers. The chambers are set up together during testing so that the bumble bees have individual access to food, but sense each other’s presence and feed willingly.
Having learned from experience with bumble bees and observed that mason bees seem more at ease in groups, developers of the acute oral test for mason bees used the hair roller chambers from the start, but ran into another problem. They placed individual mason bees in chambers, set them up in rows, and sat back to record the bees’ consumption of sugar water. They waited and waited. Some of the mason bees simply would not eat. “They’re hunger artists!” exclaims Exeler. “These bees emerge and begin foraging mid-March, a time when weather and flowering is unpredictable. They also don’t build nectar and pollen reserves, so it makes sense that they can go for several days without eating.” The protocol for acute oral testing in mason bees is still under development. At this point, bees that refuse to eat within the first minutes are removed from the test to ensure that the duration of the test is standardized. The upshot is that the number of individual test organisms drops, an issue that needs to be carefully evaluated to ensure that the test still delivers robust results.
Male mason bees emerge first in early spring and wait for females to emerge in order to mate.
“We frequently have these moments where we discover something new about the bees and have to re-think our approach,” says Exeler. “And it can be the simplest of things. For example, to understand how mason bees use nesting blocks in semi-field experiments, we glued numbered markers on the back of their thorax. The tops of our nest boxes are transparent, so we can see bees sleeping in a tunnel and identify them. Well, the majority of mason bees sleep on their backs. We couldn’t see the numbers!”
Situations like the hunger-striking mason bees can make testing conditions suboptimal and lead to a less robust and standardized test protocol. Thus, developers have to explore alternative methods that eliminate sources of variability and deliver the needed endpoints or parameters which are being tested. In the case of mason bees, the ICPPR Bee Protection Group is exploring ways to stimulate feeding. Based on discussions, it was agreed that testing should be performed with females directly after mating. When mason bees emerge in the spring, males hatch first and wait for females, to mate. Mated females find a nest to lay eggs, which they provision with nectar and pollen. Perhaps females would consume more food after mating. Members of the group performed experiments in a flight cage to see if food consumption increases during those critical days after emergence. Alas, results showed no difference. The group will need to evaluate whether focusing on mated females is a good approach for testing, or if an alternative is necessary.
Even under best conditions, you may need to start from scratch. Take again the repeated exposure test with honey bee larvae. A queen is placed on a hive comb in a cage where she promptly lays eggs. First instars (the first development stage after egg hatching) are removed from the comb, placed in individual wells of a 48-well plate and fed until pupation. This way, the tester has control over what and how much food each larva eats.
The microcolony is being examined as a system to test the effect of substances on the development of bumble bee larvae. Five female bumble bees (Bombus Impatiens) rear larvae in a cluster of pods that they construct, and feed them pollen and sugar water provided by the tester.
Such a testing system seems ideal for bumble bees considering that any female can lay eggs if separated from her nest queen. However, unlike honey bee larvae, which develop in open cells of a hive comb, bumble bee larvae develop inside a sealed wax pod (Prys-Jones & Corbet 2011). Female workers open the pods each day to feed the larvae, a behavior that is poorly understood and would be very difficult to mimic; not to mention the variability that manual intervention would introduce into the system. Thus, for brood tests like the repeated exposure test, larvae need to be left in the pods and female workers need to do the feeding.
The problem is, an isolated female bumble bee will either not lay eggs or will not be able to rear her brood alone. The same need for company observed in the acute feeding tests with bumble bees forced test developers to come up with a new testing system. Brood tests with bumble bees are still in development and are currently performed in ‘microcolonies’. “The cage design and feeding protocol of this testing system took three years to optimize based on experiments and extensive review of the scientific literature,” emphasizes Exeler. In a microcolony, five females nurse the eggs laid by one of them, feeding larvae sugar water and pollen balls spiked with a test compound. The measured endpoint is adult hatching success. Some issues with the test system still need to be ironed out. For example, the worker bees do not strictly consume the provisioned pollen, but use the food to build nest structures, making it difficult to ascertain how much of the test compound is consumed by larvae. Also, hatching success has proven quite variable when working with the Buff-tailed Bumble Bee (Bombus terrestris), a common species in Europe. Nevertheless, initial results with the common Eastern Bumble Bee (Bombus impatiens) found in North America are promising, spurring further development of the system (see infobox below).
Dr. Daniel Schmehl,
Pollinator Safety Scientist at Bayer Crop Science in the USA
Bumble bees used as crop pollinators are a selected few of the over 250 species in the genus Bombus. Most bumble bee species are social, though they build far smaller colonies and live in annual nests that appear rather disorganized compared to the highly structured, perennial honey bee nest. This is not the only way in which bumble bees differ from honey bees. Dr. Daniel Schmehl knows this all too well. As an ecotoxicologist in the Pollinator Safety Group at Bayer Crop Science in the USA, he is an expert on honey bees. Now he is collaborating on the development of testing methods for bumble bees. “It has been fun working with bumble bees,” says Schmehl. “They are so different and it has been exciting to uncover distinctive traits and contribute to overcoming hurdles with knowledge I bring from my experience with honey bees.”
For Schmehl, the work on non-Apis bee tests boils down to answering one essential question: “Is our current honey bee risk assessment protective of other bee species?” He explains that “the jury’s still out. We cannot speak to the necessary testing methods without the data to back up our assertions.” So, Schmehl and other collaborators around the world are examining different testing systems in bumble bees. “For some questions, bumble bees are better test models than honey bees,” he says. For example, you may need to test for chemical residues in nectar and pollen. “Bumble bees are optimal for collecting nectar and pollen from certain crops where honey bees or manual collection from the flowers may not work. Residues can be quantified from the collected samples, and provide us with accurate exposure data for use in our pollinator risk assessment.”
OECD guidelines for acute oral and acute contact tests with bumble bees were published in 2017. Tests to assess chemical effects on larval development are now in the works. Schmehl and his colleagues Annie Krueger and Morgan Scalici-Dunn examined the microcolony, a testing system designed by colleagues at Bayer in Germany, with the common Eastern Bumble Bee (Bombus impatiens). “Bumble bees cannot rear brood in isolation. Why this social aspect is needed, we don’t know, but to do a brood test, you put five female workers in a nest and one will express her dominance, develop ovaries and lay eggs.” Schmehl and his team have demonstrated that the system works well with this species, and that a pesticide homogeneously integrated into a pollen diet will be fed by the bees to the brood and exert an effect. “The compound we used to evaluate the system was an insect growth regulator that only impacts larvae. What remains to be answered,” he explains, “is whether the test will work with any compound, including those that impact adult stages.”
For now, however, he is pleased. “We’ve had consistently high hatch rates in the untreated control groups, which demonstrates that the test system is working.” Further investigations aim to understand variability in the test system, and whether it is due to differences among test species, suppliers, individual laboratories or other unknown factors.
Leafcutter bees are one of the species for which new testing systems are developed.
Very often, work on non-Apis bee testing requires a dose of creativity. Dr. Nina Exeler describes an elegant solution to a problem that arose in her laboratory. “We have started trying out testing systems with leafcutter bees and are allowing them to hatch in paper cups in the laboratory. We promptly learned that if the bees emerge and find nothing of interest around them – like food – they will crawl right back into their cocoons. Pulling them out of their cocoons is time-consuming and could damage them. So, our technician Anja Quambusch designed a two-chambered hatching container. Cocoons are held in the dark in one chamber. The second chamber receives light and is connected to the first via a tunnel. This tunnel is wide, starting at the dark chamber, but narrows as you go to the light chamber, where the opening is set high off the chamber floor. Emerging bees are attracted to the light and wander into the second chamber, but cannot find their way back to their cocoons because of the narrow and raised tunnel opening.”
Roadblocks are overcome as they present themselves “thanks to the resourcefulness and problem-solving sagacity of ICPPR members and of staff in our laboratories.”
Dr. Nina Exeler
Light lures leafcutter bees to leave the chamber where they emerged (blue box). The clever architecture of this two-chambered hatching container prevents them from returning to their cocoons.
While progress is rather paced at this point, Exeler is confident that every lesson learned in developing non-Apis bee tests will lead to improved understanding of these organisms and explicit knowledge about pesticide safety for new species in the focus.
Roadblocks are overcome as they present themselves “thanks to the resourcefulness and problem-solving sagacity of ICPPR members and of staff in our laboratories,” says Exeler, smiling. The ICPPR membership also includes experts from companies and institutions with broad research interests, which stimulates innovation and enables simply trying out new ideas. For example, Exeler and colleagues recently worked with the Animal Health team at Bayer. This division has radiology facilities and experts who helped x-ray solitary bee cocoons to count larvae and pupae without opening the nests.
It will be necessary to leverage knowledge gained from species that can be tested, to make inferences about others, and devise protection strategies that secure their health while keeping overall testing requirements manageable and meaningful.
The most pressing issue in non-Apis bee testing looking into the future is the impracticality of testing all non-Apis bee species. It will be necessary to leverage knowledge gained from species that can be tested to make inferences about others and devise protection strategies that secure their health while keeping overall testing requirements manageable and meaningful.
Dr. Daniel Schmehl of Bayer’s Pollinator Safety Group in the USA puts it quite succinctly: “It boils down to one essential question: Is our current honey bee risk assessment protective of other species?” For many authorities and researchers, comprehensive protection will require additional species testing, and understanding how to confidently ensure pollinator health beyond the honey bee will require information. As test development with bumble bees, mason bees and leafcutter bees progresses, data from ring-testing are being anonymized and deposited into a database.
Ultimately, these data can be used to uncover interspecies patterns that may help define toxicity threshold values and support cross-species protection strategies. At this point, data from acute contact tests that have been normalized for body mass suggest that honey bees are the most sensitive of the species tested to date, followed by mason and leafcutter bees, and bumble bees appear to be the hardiest of them all. In other words, from the perspective of intrinsic susceptibility to chemicals alone, testing honey bees may already prove to be a good way of protecting other species.
“But the big open question in non-Apis bee test development and a critical aspect of risk assessment is route and magnitude of exposure,” says Schmehl. “We need to generate more data, learn more about different bee species and validate methods while bearing in mind similarities and differences that might support interspecies extrapolation.”
In his own way, Dr. Ivo Roessink concurs with Schmehl’s assessment. “Biology likely trumps toxicity,” he says. This means that accounting for diversity in behavior, life cycle, food preferences and social structure among other traits will be paramount for any protection strategy moving forward.
Bumble bee microcolony set-up. Series of test cages to test the effect of substances on the development of bumble bee larvae.
A plan currently in discussion among ICPPR members and other stakeholders is the idea of profiling bee species according to a set of behavioral and ecological traits that align with a panel of ‘surrogate’ or ‘archetypal’ species that are biologically well-characterized, and for which standardized safety tests have been established. In this way, the risk to bee species relevant to the use of a crop protection product may be assessed based on the match of their biological profile with one of the archetypal species. If the fit is good, protection strategies would be planned based on knowledge from safety testing on the archetypal species. If the fit is less clear, additional testing with the bee species in question would serve to fill in knowledge gaps and establish appropriate protective measures.
Whether such cross-species extrapolations can be done is in debate, but the discussions have served to highlight gaps in our knowledge about bee biology, pollinator community composition and overall pollinator landscape interaction patterns. Other methodologies may help address such gaps, such as modeling bee behavior both in the colony and the landscape. Predictions from such models can serve to better characterize exposure risks to bees based on what we know about how they move, behave and interact within an ecosystem with external influences, like a pesticide.
Roessink emphasizes that whichever approach is taken, it must be consensus-driven. “All stakeholders must be involved in decisions and agree upon the approach,” he explains (see interview ‘Moving forward via consensus’). This consensus may also extend beyond manufacturers of crop protection products, testers and regulatory agencies.
While Dr. Nina Exeler believes that the quality of purchased mason bees is adequate to establish robust test populations, certain commercial bumble bees may exhibit signs of inbreeding depression. “Commercial stocks are somewhat of a genetic bottleneck with only a limited number of distributors,” says Schmehl. “A purchased bumble bee nest serves the primary purpose of pollinating a crop for a single season so typically the longterm viability of a colony is not known.”
Understanding sources of variability is essential to ensure the repeatability of test designs and requires knowledge from various perspectives and touchpoints. The multi-stakeholder collaborations embodied by the ICPPR working groups aim to capture that knowledge and facilitate consensus.
Dr. Ivo Roessink is Senior Scientist at Wageningen Environmental Research in the Netherlands. An expert in assessing environmental risks of chemicals, he has experience across a range of aquatic and terrestrial systems and a particular affinity for non-standard experiments to support risk evaluation. He joined the International Commission for Plant Pollinator Relationships (ICPPR) Bee Protection Group in 2012 and currently chairs the subgroup of members focusing on developing test systems for solitary bees.
Much of what I do revolves around verifying assumptions made in the development and evaluation of a test system. For the last three years, the solitary bee subgroup has been ring-testing the acute oral and acute contact tests for Osmia bicornis and O. cornuta. The process involves condensing a lot of knowledge over multiple review rounds and testing iterations, with the objective of generating test systems that everyone can understand and perform. Every aspect of these tests is scrutinized: Are the protocol instructions unequivocal? Are the endpoints meaningful and achievable across all testers? Is the relevance of these endpoints to the test objectives transparent? We started off with very little knowledge about these bees. With every iteration, we raise questions that require empirically determined answers, we improve the tests, and we examine their fit to current regulatory requirements. This takes a lot of time, but the group consists of scientists who are driven by curiosity, are passionate about bees, and are tenacious when told that something cannot be done.
It is impossible to develop tests for every single bee species. Our efforts with bumble bees and solitary bees serve two purposes. On the one hand, we are able to relatively quickly develop tests for these species because they are of economic interest and commercially available. On the other hand, what we learn will help us tackle other species. The format and success of these tests is largely determined by the biology of the animal. Take honey bees, for example. Typically restricted to the hive when not out foraging, honey bees are comfortable in dark, warm spaces. This is in stark contrast to solitary bees that live in the open, only sleeping and laying eggs in their nests. You cannot confine solitary bees to dark, warm spaces and expect them to behave normally. I think that the risk to a bee from a pesticide will prove to be governed by their biology. Discussions are ongoing about the potential to profile bee species according to a handful of species, for which toxicity and exposure potential are well-characterized. The idea is to safely predict the risk to a bee based on its similarities to one of these species in terms of feeding preference, sociality, life cycle, seasonality, stress susceptibility and more. Landscape and behavior modeling will also play an increasingly important role in assessing risk, determining worst case scenarios and narrowing testing scope to enable realistic, yet effective protection of non-Apis bees.
Consensus; whatever our approach looking into the future, it will be consensus-driven. We can generate data, and we can create highly detailed predictive models, but we will never cover 100 % of the variability, diversity and confounding factors relevant to the questions we are asking. All stakeholders – from manufacturers of crop protection products to academic researchers, environmental assessors and regulatory bodies – will need to agree on an approach and on the acceptability of any shortcomings the approach might have. Ideally, this consensus will also be coupled with a feedback mechanism that brings stakeholders back to the drawing board in the event that something does not work. In this way, neither knowledge, nor regulations, nor solutions will be static.
Dr. Nina Exeler and colleagues from the USA (Dr. Ana Cabrera, Kim Huntzinger) visiting a cactus farm to study the pollination system. The Cactus bee (Diadasia rinconis) is the main pollinator of this plant.
The road ahead is long but Bayer is committed to engaging with experts around the world and playing an enabling role in creating test systems and developing guideline recommendations for effective and meaningful safety testing in non-Apis bee species. Achieving that goal can only be done collaboratively, not only because of the sheer amount of work it entails, but also because diversity of expertise spawns the out-of-the-box thinking that gets the job done. Regardless of the outcome, every measure taken by ICPPR members to improve the adequacy of a testing system, every exchange with other stakeholders who are equally committed to bee health, and every avenue explored to ensure that crop protection products make a positive contribution to sustainable agriculture, is a step in the right direction.
Crop protection products are among the most heavily regulated goods in any industry. They are as strictly regulated as pharmaceutical products, requiring even more extensive environmental safety testing to ensure that they will not pose any unreasonable risks to wildlife, plants and the environment.
This includes bee safety testing. In fact, the focus on bee safety can continue throughout the entire life cycle of a pesticide product, as needed. And advances in bee risk assessment are taking the science to a new level of rigor and requirement.