The debate about the environmental safety, or lack there of, for the neonicotinoid insecticides begs us to ask what would be the characteristics of an environmentally safe or benign pesticide? Despite a growing and flourishing organic agriculture movement, industrial agriculture is not going away any time soon. Hence it is important to understand how to assess the environmental risks of industrial chemicals that are deliberately spread across the landscape for agricultural purposes (EFSA 2012).
Ideally, pesticides for agriculture should effectively kill insect pest species yet have little or no effect on humans or beneficial insects that come in contact with the chemical or its residues. This is a tall order. All life forms share an astonishing amount of similar cellular function. Chemicals highly toxic to insects can also be toxic to humans. Even more difficult is finding chemicals toxic to pest insects that are not harmful to beneficial insects. There are several strategies to solve this problem. 1) Apply insecticides only in a time and place where there is minimal contact with beneficial species. 2) Use chemicals that are specifically toxic only to the pests you are trying to kill. 3) Deliver the chemicals in a way that kills only the pests that come in contact with the toxin. The degree of success at only targeting the pests while not doing harm to other creatures usually gets down to the specifics of the chemicals involved, application methods, and the consequences of residual contamination. It is worthwhile to examine the various classes of pesticides more carefully to understand the similarities and differences in the chemicals.
The table below lists several of the most important classes of pesticides and gives specific information about representative members of the class. I attempted to unify the example data by looking for specific properties of the chemicals for honeybees. The chemicals are ranked in order of most toxic, (lowest LD50) first. Also listed are soil persistence, metabolic half-life, receptor binding time, the toxicity time-scaling exponent, and the specific mode of action of the chemical.
Table 1: Representative characteristics for several pesticide families (click-it for easier reading)
A quick look at the numbers for DDT and you can quickly see why this chemical was banned. It persists for decades in the soil, and resides for years in living organisms. The organochlorine pesticides are the most environmentally persistent and still can be found in living organism around the globe, although they have largely banned now for more than 40 years. Curiously, one of the main reason for DDT’s removal was that it caused egg shell thinning in birds, an effect not at all predicted from its expected mode of action as a neuro-toxin. Chemical agents often have unexpected, unpredictable effects that only extensive testing or experience after-the-fact reveal.
One of the main problems with neonicotinoids is that they have a delayed toxic effect that renders small residual concentrations of the chemicals especially problematic for long-lived beneficial insects such as bees. The argument is simple enough once you grasp the implications of delayed toxicity. Allow me to digress and offer a comparison of toxic time dependence and example chemicals.
1) Threshold toxicity. Consider CO2 – usually not a poison unless there is so much of the gas that you cannot get enough oxygen to breathe. If the concentration of CO2 got high enough to deprive you of oxygen, it would kill you. However, even long exposure at levels 30% of the toxic threshold are likely to have little effect. Typical insecticides that work this way are dormant oil sprays that essentially suffocate the target insects by plugging up their tracheal breathing tubes.
2) Linear accumulation – Haber’s Rule. Some toxins exhibit “accumulation to a threshold” behavior. In the 1900’s Fritz Haber noticed this effect with the concentration of chlorine gas. The concentration of the gas multiplied by the length of exposure produced the same lethal effect. The lethal concentration varies as 1/t where t is the exposure time, or LC50 = k/t where LC50 is the concentration that kills half of the subjects and k is proportionality constant. Several insecticides have this behavior, including many of the old fashion organophosphate pesticides. This all makes a certain amount of sense, because if you add up the total exposure over time, you find it takes the same total amount of toxin to kill the organism, proportionally longer if you use proportionally less concentrated toxin.
3) Delayed and Enhanced Toxicity – We can write out a scaling law as LC50 = k/tp. When the time exponent p=0, then we have case 1) above where the toxic threshold is k. When p=1 we have Haber’s Rule, case 2) above. When p>1 we have a situation where the toxic concentration required to kill half of the target organisms is less if the exposure is longer, and furthermore, the total accumulated toxin required to kill the organism is less the longer you wait. Such enhanced toxicity can come about because several conditions about how the toxin operates are met. These included 1) strong binding for the toxic molecules to the sites of activity in the organism, 2) direct action of the toxin biologically (no threshold required to damage biological function) and 3) biological accumulated effects of the toxin. The neonicotinoids often fit exactly these criteria. It is not a standard requirement to evaluate the toxicity time-scaling of pesticides, although it should be. I looked carefully at the literature and found several studies that, taken together, point to approximately a 1/t2 scaling dependence for imidacloprid with bees.
The scaling laws discussed above are just a simple way to describe the results of toxicity tests for these chemicals (Sanchez-Bayo 2009). They are empirically derived, but such power-law scaling does a good job of describing what we see in practice. Hence, we can plot empirically derived toxicity curves and discover the scaling exponent p for a particular chemical and organism, and have a good way to estimate the toxic effect over the life time of the organism for lower exposure levels. See, for example, the work I have done with imidacloprid on bees and ants, and the references therein for other chemicals and organisms. One figure from that work is shown below. The best estimate for bees is the red line that represents results from many researchers. In all cases the toxicity scaling exponent is approximately 2 or greater, implying significant delayed toxicity for imidacloprid on bees and ants.
The toxicity scaling exponent is just one of a combination of traits can lead to a prediction of trouble. The other main concern is the exposure duration and level. This brings us to why we have a particular problem with some of the neonicotinoids. These insecticides are designed to be used systemically, entering into the plant xylem and carried to all parts of the plant. The chemicals need to be relatively long-lasting because they are used as preventative seed treatments before target insects are necessarily even present.
Looking at the representative numbers in Table 1 above once again, the neonicotinoids have the longest lifetime in soil of any insecticide class that hasn’t already been banned as a persistent organic pollutant. The mode of delivery of the insecticide is systemic in the plant, so plant products collected by bees will chronically deliver some chemical to the insects. The chemical binding at the synaptic receptor sites is irreversible, meaning the chemical can build up at the site of toxic activity in individual insects when chronically exposed. Since the means and mechanism for chronic delivery of low level contamination is built into the way neonicotinoid insecticides are used, we must also consider the chronic time dependent toxicity scaling when assessing the likely hood of impact on pollinator populations.
This is in stark contrast to the situation with organo-phosphate and carbamate insecticides. These classes of chemicals usually are designed to kill on contact, and have a very short lifetime in the environment. Many also bind to their receptor sites in a reversible manner so they do not build up in the organism. Also, these are acetylcholinesterase (AChE) inhibitors which have a natural threshold requirement built into their mode of action. The toxic effect due to buildup of acetylcholine in the synapse only can happen after most of the AChE is disabled by pesticide molecules, so low levels of pesticide are likely to have little toxic effect below that threshold. The experimental evidence suggests that these chemicals have toxicity time-dependent scaling that is either threshold-like, t0 scaling, or at most a linear dependence like Haber’s rule, t1 scaling.
However, with the neonicotinoids, we have a toxin deliberately delivered in a way that can yield chronic exposure along with enhanced toxicity time-dependence. These are new circumstances for regulators. The usual approach is to compare exposure just to LD50 levels and not take exposure time into consideration. The problem with this is illustrated in Table 2 below, where we compare the level of protection necessary to avoid damage to a long-lived insect with the quantity of insecticide needed to kill target insects efficiently.
Table 2. Level of insecticide contamination compared to the application rate required to protect a long-lived pollinator (e.g. life-span 50 days) when the insect pest is targeted in 2 days (Ratio pollinator/pest = 50/2). Relative safety factors (x 3) required to protect the beneficial pollinator are indicated for each case.
Relative time-dependent toxicity
Include Safety Factor × 3
|t0||Threshold level only – doesn’t depend on time||
|t1||Accumulate to threshold with time – Haber’s rule||
|t2||Enhanced or delayed toxicity||
For honeybees, the LD50 for imidacloprid is about 40ng/bee, or nectar and pollen concentrations around 800 parts per billion (ppb). If we are trying not to kill bees that are 50 days old, then because of the t2 toxicity time dependence, we need to limit residual concentrations by a factor of almost 2000, or less than 0.4 ppb in nectar and pollen. Honeybees live for more than 150 days when going through the winter, so we have to extend the scaling further and require less than 0.05 ppb to ensure we aren’t killing these wintering bees prematurely. These are exceedingly small quantities of pesticide, well below the level of detection of many assay measurements. These facts demonstrate that it is impossible to effectively limit contamination levels of these chemicals to the low levels that are consistent with protecting long-lived insects while still killing target insects effectively. The burden of proof should fall on the chemical companies to show that the scaling law is actually different that what has been observed in tests on bees with up to 60 days exposure at concentrations of 4 ppb (Dechaume-Moncharmont 2003). The data available shows such levels are not safe chronically, and the toxicity scaling suggests that even much lower levels are problematic for long-lived winter bees.
So far we have just discussed direct toxicity of the pesticides. Pesticides can interact with other chemicals and pathogens as well. There is recent evidence that the neonicotinoids clothianidin and imidacloprid downgrade the immune systems of bees. This is one of those unexpected effects that is not obvious from the mode of action of the chemical, but could be responsible for some of the problems bees face with pathogens today. A paper out of Italy demonstrated a mechanism for immune suppression by the neonicotinoids clothianidin and imidacloprid in honeybees (Di Prisco 2013). The authors went on to show that for one of the common known bee viruses, deformed wing virus (DWV), bees exposed to the neonicotinoids had further replication of the virus in their bodies, whereas in unexposed insects the virus did not replicate in the adult bees. They showed that the effect was both proportional to dose and time. This mechanism, downgrading the immune system by the neonicotinoids, goes a long way to explaining the observed general bee malaise and problems with multiple pathogens that are ultimately killing our colonies (Cornman 2012). The Di Prisco paper looked at modest doses and time span of a just a few days. It seems likely the immune deficiency might scale with time much like the mortality effects we see. In fact it could be the immune system downgrade that is the reason for the observed scaling we see. More studies would be needed to understand this fully.
The face of chronic neonicotinoid poisoning looks quite different from the type of pesticide poisoning that beekeepers have become accustomed to dealing with. Delayed toxicity implies early death for older bees. The consequences are likely to include: 1) smaller honey crops because forgers are older bees, 2) poorer wintering and higher winter losses because winter bees need to live for several months, 3) more queen failures because queens live for several years, need to consume large quantities of food, and hence would be subject to longer and higher residual toxic exposure.
Acute poisoning by any neurotoxin usually produces tremors and twitching in the dying insect. Often the twitching is followed by paralysis as the toxin goes from over-stimulating nerves to killing neurons. The much slower onset of chronic poisoning will likely cause lethargy, behavioral changes and susceptibility to disease. Bee deaths will be spread out geographically, as affected individuals cannot navigate back to the hive. Low level poisoning may not produce any bee deaths near the source of the contaminated nectar or pollen.
The incidents with bumblebees in Wilsonville and Hillsborough provide two cases in point. In Wilsonville, the bees were attracted to flowers that had recently been sprayed with the neonicotinoid dinotefuran. The exposure levels were high and bees died en masse on the spot with typical insecticide poisoning symptoms. In Hillsborough, the trees had been sprayed several months earlier. The insecticide had translocated into the plant tissue and moved into nectar and pollen, where it could be ingested by bees. Dinotefuran is expected to photolyze and degrade in less than 2 days in direct sunlight. Hence it is unlikely that contact exposure from residue on leaves was the cause of the bumblebee deaths reported in Hillsborough. Although the concentration of bee deaths in the Hillsborough case were not nearly as large massive number of bees found under the trees in Wilsonville, the implications of the Hillsborough bee deaths are much worse. First, there is little reason to think that bees visiting the linden trees in Hillsborough died immediately. Most of the bees likely succumbed to the toxin after delivering many loads back to the nest. They probably died throughout their foraging range, perhaps several days after working the contaminated linden trees, so the hundreds of dead bees found at the tree site is likely just the tip of iceberg in that case. The trees were sprayed months before the trees bloomed, so we have direct evidence in this case, that even when applied according to label directions, this insecticide is lethal to bees. The contaminated nectar and pollen brought back to the colony will be consumed by developing larvae and house bees, so not only is the loss of foragers a problem, but entire colony is put in jeopardy by the contaminated food supply (Gill 2012).
Even worse than spraying pesticide is the practice of direct injection of the neonicotinoids into trees. This method of insect control is becoming more common, and is potentially deadly to bees. Flowering trees are especially attractive to bees because a tree represents a concentrated food source. Bees recruit additional foragers to attractive sources, so a single tree in bloom can be more important to the bees than a myriad of scattered flowers.
In light of the growing evidence against these chemicals, Oregon should follow the lead to regulators in the European Union and ban the use of imidacloprid, clothianidin, thiamethoxam, and also dinotefuran from use on plants visited by bees. Residential use of these chemicals is inappropriate considering their extraordinary toxicity to a wide variety of arthropods (Mason 2013). These insecticides should also be restricted so they cannot be used on trees that produce flowers or are otherwise visited by pollinators.
Cornman RS, Tarpy DR, Chen Y, Jeffreys L, Lopez D, Pettis JS, vanEngelsdorp D, Evans JD. 2012. Pathogen webs in collapsing honey bee colonies. PLoS One 7:e43562.
Dechaume-Moncharmont F-X, Decourtye A, Hennequet-Hantier C, Pons O, Pham-Delegue M-H. 2003. Statistical analysis of honeybee survival after chronic exposure to insecticides. Environ Toxicol Chem 22:3088-3094.
Di Prisco G, Cavaliere V, Annoscia D, Varricchio P, Caprio E, Nazzi F, Gargiulo G, Pennacchio F. 2013. Neonicotinoid clothianidin adversely affects insect immunity and promotes replication of a viral pathogen in honey bees. PNAS in press. 10.1073/pnas.1314923110.
EFSA. 2013. Conclusion on the peer review of the pesticide risk assessment for bees for the active substance imidacloprid. EFSA J 11:3068.
Gill RJ, Ramos-Rodriguez O, Raine NE. 2012. Combined pesticide exposure severely affects individual- and colony-level traits in bees. Nature 491:105-108.
Mason R, Tennekes H, Sánchez-Bayo F, Jepsen PU. 2013. Immune suppression by neonicotinoid insecticides at the root of global wildlife declines. Journal of Environmental Immunology and Toxicology 1:3-12.
Sánchez-Bayo F. 2009. From simple toxicological models to prediction of toxic effects in time. Ecotoxicology 18:343-354.