A call for reform in pollinator pesticide risk assessment

 
Purpose of Risk Assessments
Risk assessments have two important functions: (1) To evaluate the likelihood that pollinators might be exposed to pesticides and (2) To determine potential negative impacts on pollinator health. Methods used for such risk assessments should consider all possible routes of exposure, if a pesticide is systemic, how it is applied, its persistence and its toxicity to different pollinator species.

Exposure to Pesticides
Bees can be exposed to pesticides orally via pollen or nectar or topically when their bodies come in contact with pesticides as they forage or make their nests.

Many modern pesticides are systemic, meaning they move from the point of application into all tissues of the plant, including pollen and nectar. Often systemic pesticides are applied via seed coatings or as drenches, leaving residues in soil.

Some pesticides, such as neonicotinoids, persist in the environment, while others break down quickly. Persistent pesticides can lead to both acute (one high dose) and chronic (repeated low dose with potential cumulative effects) exposure.

The Risk Assessment System

First Tier:
To address the negative impacts of pesticides on pollinators, risk assessments use a tiered system. At the first tier, individual honey bees are tested in a lab to establish the acute lethal dose (the dose that will kill 50% of exposed bees) for each pesticide. Those tests quickly weed out pesticides which are unlikely to have negative effects on pollinators and do not need further testing. Long-term chronic exposure is not tested.

Using honey bees for first tier tests makes sense because they are abundant and easy to maintain in a lab. However, differences in sensitivity to pesticides exist among bee species. To accommodate the differences among bee species, a safety factor of 10 can be applied to the honey bee lethal dose to better represent toxicity for all wild bees. The pesticide sensitivities of other pollinators, such as hoverflies, butterflies and beetles are largely unknown despite their important roles as pollinators.

Higher Tiers:
After first tier testing, if pesticides are found to be harmful to individual bees, risk assessment moves forward to more expensive higher tiers involving whole colonies of honey bees. Results from higher tier tests determine whether a pesticide is deemed safe for all pollinators.

At upper tiers, the problem of using honey bees to represent all pollinators is further amplified because honey bee colonies are so different to other bees. Unlike most pollinators, honey bees live in large, highly variable colonies containing over 10,000 individuals. Large colonies give honey bees unusual resilience not found in bumble bee colonies or among solitary bees. If individual honey bees die, there are others to replace them and carry on their work. However, after a certain point (i.e. at ~35% worker loss), even honey bee resilience has its limit and colonies can collapse.

Tracking changes in colony size for higher tier studies is difficult. Imagine counting 10,000 moving objects that all look the same. That often results in miscounts of the number of individuals in a colony. The variability in colonies, their inherent resilience, and the difficulty of making accurate estimates of colony size lead to uncertainty in the measurements taken to assess effects of pesticide exposure in higher tier studies.

Uncertainty
Uncertainty is a problem. Confidence in findings is so important that the European Foods Standards Agency has recommended that studies on the impacts of pesticides on bees should be able to detect a 7% change in honey bee colony size.

When testing pesticides, it is important to avoid saying that they are safe when they are not (known as a type two error). Type two errors are avoided by having a large enough sample size to be sure that what is measured is not being masked by the large variation between colonies. For honey bees, 68 pairs of non-treated and treated fields with 6 colonies at each would be needed to reach the European 7% target (68 x 2 x 6 = 816 colonies). Clearly, the expense of having a large enough sample size quickly becomes prohibitive for higher tier tests with honey bee colonies.

We have shown that many higher tier (field test) studies used to assess the impacts of neonicotinoids on honey bee colonies would not have been able to reliably detect critical individual bee losses (35%), and none to date have met the target of being able to detect a 7% change in colony size as recommended in Europe. Although individual honey bees may provide a useful model that can be adjusted for other pollinators, honey bee colonies are just not practical models, either financially or statistically for higher tier pesticide testing.

One solution is to use computer models of honey bee colonies, making it possible to test thousands of colonies and scenarios at a fraction of the time, cost, and loss of bee life. However, the accuracy of such models must be improved before they can be used to inform policy. Furthermore, a model approach may not be representative of the diversity of pollinators.

Considering Pollinator Diversity
Risk assessments cannot continue to ignore the large diversity of pollinators that may be at risk. For example, bumble bees that live in small annual colonies or bees that lead solitary lives (most bee species) exhibit different nesting behaviours to honey bees. Around 70% of solitary bees make their nests in the ground and others nest in plant stems or cavities like those in bee hotels. Many of these solitary bees provide important crop pollination services and live on agricultural lands. Differences in nesting location affect how different bee species may be exposed to pesticides in the environment. A risk assessment based on honey bee behaviour cannot model exposure to pesticides from soil simply because honey bees do not normally interact with soil. Our work on a solitary ground-nesting bee species has highlighted the risk of exposure to lethal doses of neonicotinoid insecticides for bees constructing their nests in agricultural soil.

Take Home Message
After they were registered for use under the present risk assessment system, neonicotinoids had about 20 years of ubiquitous, prophylactic, and damaging use in agriculture before much attention was paid to their negative impacts on pollinators. Do we want the next generation of pesticides to be subjected to the same flawed risk assessment process that has allowed this long-term damage?

To avoid repeating the mistakes of the past, we need to develop sustainable pesticide risk assessments that better protect all pollinators. Perhaps we will need many model species to fully represent pollinator diversity, perhaps only a few. To establish that, we need to know more about pollinating insects, how they are exposed, and their sensitivities to pesticides. Turning a blind eye to the inadequacies of the present risk assessment system opens the door to potentially damaging future pesticides.