Ever since French beekeepers saw their bees dying as they collected pollen from treated sunflowers back in 1996, beekeepers have been concerned that their bees are being harmed the highly toxic neonicotinoid insecticides, with imidacloprid most widely used. The use of this class of insecticide has grown steadily ever since. Bee losses have become chronic as well. However, unlike the first case in France where bees were literally falling dead while gathering pollen, the widespread colony losses today are less explainable, often associated with outbreaks of a variety of diseases, and with very high winter colony mortality. So why blame the insecticides?
To see why the bees are dying, and why these pesticides are still being sold, we must examine the toxicology of the neonicotinoid chemicals as well as the history and science of pesticide regulation. The toxic nature of a chemical is characterized by its “LD50” level. This is the amount of chemical that will kill half of the test organisms in short order. For many traditional pesticides (organophosphates), the LD50 level provides an adequate overall characterization of toxic effect. These pesticides tend to be short-lived and generally do not bio-accumulate in the target organism. If the dose doesn’t kill the organism, the toxic compound will be metabolized and excreted. Since the organophospate pesticides — which for many years made up the majority of pesticides sold — could be characterized easily with the single LD50 number, the culture of pesticide regulation largely accepts the acute LD50 as determinative for all toxic effects.
The acute LD50 characterization works poorly for substances that bio-accumulate and/or have a relatively long time-to-effect characteristic. Substances that fall into this category are heavy metals that are known to accumulate in certain tissues, and some carcinogens where an initial single exposure can give rise to cancers much later in the life of the organism. Neonicotinoid insecticides also fit into this category. These insecticide molecules bond strongly and irreversibly at nicotine receptor sites in the central nervous system. There is also evidence of delayed time-to-effect of several days for exposures below the acute LD50 [Suchail et. al.(2001)].
Toxicologists attempt to model the time-dependent effects of chemicals at various dose levels [Tennekes, (2010)]. One of the simplest empirical models assumes the dose/effect relationship can be characterized by a simple “power law” where the effect is proportional to the dose multiplied by the time-of-exposure raised to a power, b.
Effect = Dose x Timeb
Instead of a single number, now we have two numbers to characterize toxicity, the dose and the time exponent. When a power law is plotted with logarithmic scales, one gets a straight line with slope equal to the exponent, b, and intercept equal to the Dose. The time-independent, acute LD50 model is the special case when b=0. For a simple bio-accumulation model, one would expect linearity in time with b=1. Time-to-effect mechanisms require b>0. Combinations of effects for a given organism and chemical will result in a Dose vs Effect curve with characteristic slope that includes all of the time-dependent mechanisms embedded in the value of the exponent.
The plot above shows some published data on imidacloprid toxicity on a log-log graph. Data for two species of small aquatic crustaceans, Daphnia magna (red squares) and Cypridopsis vidua (blue diamonds) illustrates the large range of toxicity difference between organisms, and also the large range of applicability of the model, spanning more than three orders of magnitude in concentration. [Sánchez-Bayo et. al. 2009] Honeybees (green triangles) also have high toxicity over a wide range of concentrations [Suchail et al.(2001)].
The best-fit power law curves are shown in the plot. In all cases, the fit parameter, R2, is greater than 0.85. The time exponent for Daphnia, 1.3, is barely more than linear, where as for Cypridopsis and honeybees the exponent is close to 5. This means that there are strong time-dependent delayed effects from the chemical for the Cypriopsis and honeybees. The time5 dependence of the toxicity of imidacloprid for honeybees is the big problem. Very low levels of exposure, with sufficient time, will be lethal. The power law model suggests that we should extend the time of exposure to the lifetime of the organism in order to determine the minimum dose that will have no effect. Honeybees, including the larval stage, live ~50 days in the summer and ~100 days for wintering bees. Extrapolating the green dashed line trying to reach 50 to 100 days would require reducing continuous exposure to less than 0.0001 parts per billion (ppb). This minuscule exposure level is far below the detectable limits of present technology (~1 ppb). What we can detect are residual levels in nectar and pollen on treated plants commonly in the 1 – 10 ppb range; even these small levels are more than 10000 times the level that the power law model suggests would cause no harm to adult bees. Looked at another way, if one bee in 10,000 returns to the colony with pollen gather from a treated plant, that would be enough toxin to begin to cause damage to the colony.
Collecting toxicity data about bees is a complex process, involving multiple trials, caged bees to limit their activity to the test sample, etc. Aquatic crustaceans provide an easier test subject where dosage can be controlled by dilution of the water they live in. The fact that the crustaceans conform to the power law model confirms that imidacloprid is bio-accumulative. The same finding by Suchail et.al. for bees should not be surprising, given that both organisms have central nervous systems with nicotine receptors.
What about mammals and humans? Very little time-dependent information is available. There are a couple of data points on the plot above for mice. One point was the LD50 for short time exposure and the other the threshold for immune system compromise exposed for 28 days. The data for mammals is just not available yet, but even much more modest sensitivity to the chemical could present problems for long-lived species such as humans. There is also fear that these biologically persistent chemicals could be fueling world-wide wild life declines in many species [Mason, R. et. al. 2012].
Our regulators, (EPA) are not adequately considering the time-dependent nature of the lethal effects of the neonicotinoid class of pesticides. The most toxic and long-lived chemicals, imidacloprid, clothanidin, and thiamethoxam should be removed from the market before more harm is done, as is happening in Europe. Other insecticides in this class should be subject to further scrutiny as well.
The social nature of bees naturally draws one to a human analogy. Imagine that a toxic chemical is slowly poisoning our brains. (Think lead pipes and the Romans.) Instead of healthy people living into their 70’s, the toxic effects are bringing on Alzheimer’s-like symptoms to folks in their 40’s and 50’s. The younger healthier part of the population has its hands full, providing for themselves and those that no longer support their own livelihood. Everyone is hungry. Now introduce a bad case of the flu, or the plague, and the already weakened population is devastated. That is what our bees are facing today. The levels of poison are rising as more and more of these pesticides are being used, building up in our soil and in treated plants. The bees are dying younger, and we are gradually eliminating a host of insects and creatures we don’t even know we are poisoning.
Tennekes, H, A. The significance of the Druckrey-Kupfmuller equation for the risk assessment — the toxciity of neonicotinoid insecticides to arthropods is reinforced by exposure time. Toxicology. 2010 Sep 30;276(1):1-4