Pest resistance to Bt crops

Written by Jonas Kathage

Cotton Bollworm. Source: USDA

The issue of pest resistance to insect resistant Bt crops receives regular media attention, partly because anti-biotechnology lobbies use it as an argument to vilify GM crops. The German NGO testbiotech, in a recently published report commissioned by Member of the European Parliament Martin Häuslingof the Green Party, argues that because of potential resistance development, GM crops should not be allowed for cultivation in the EU:

There must be no large-scale, commercial cultivation of GE herbicide-tolerant or insecticide-producing crops. Such crop cultivation is unsustainable and will lead to a ‘race’ to step up their cultivation.

The idea that a particular technology should be banned if it cannot be used forever is dangerously misguided. The benefits obtained by millions of farmers thanks to Bt crops have been very real, perhaps most notably India but also in the European Union. Without GM, food prices would have been higher, environmental externalities more severe and health problems more common.
Pest resistance is a not a black and white concept where all pests are either completely resistant or not. What does it mean if some Bt-resistant insects are discovered in a field? As long as not all relevant populations of all relevant pest species in the field are completely resistant to the Bt toxins, Bt technology reduces insecticide use and enhances effective yield. Up until that point, Bt technology remains above the agronomic baseline of non-Bt seeds.
There is the theoretical possibility that emerging secondary pests might end up posing a greater problem than the primary pests did before the Bt technology was introduced. While there is some evidence that secondary pests such as aphids have increased in cotton fields, I’m not aware of any case where the pest pressure on the whole has increased as a result of Bt. Indeed, in absence of the insecticides that are reduced by using Bt, there is evidence that beneficial insects can return to farms.
Moreover, resistance to particular Bt toxins (or “Cry proteins”, so named because they form a crystal) is not the same as resistance to Bt technology in general. There are hundreds of naturally occurring Bt proteins as well as engineered chimeric Cry proteins with specific modes of action and varying effectiveness against different insect species. Different combinations of Bt proteins have already been approved for cultivation, and resistance to one toxin does not necessarily imply resistance to another. For example, transgenic broccoli engineered to produce Cry1C controls diamondback moth that is resistant to Cry1Ac. It is also not obvious that resistance development is as likely with some sets of Bt genes as with others. In addition, the potential to develop resistance may vary by insect species.
It is important to put the resistance issue in perspective. Resistance development in agricultural pests is not a specific problem of Bt or other GM crops. In the history of US agriculture, biological innovation has been key in the struggle against pests, if only to avoid regress (also known as the Red Queen effect where you have to run fast just to stay in one spot). The same applies to non-breeding strategies including chemical insecticides and biological control agents (sometimes called the “pesticide treadmill”). This is not to say that resistance development is desirable; rather, that its occurrence is not restricted to transgenic crops. Applying the logic of testbiotech, no conventionally bred insect-resistant plant variety and no chemical or biological insecticide should be allowed as they all carry proven risks of resistance development.

Strategies to delay resistance development

In contrast to anti-biotech groups, farmers, technology providers and regulators pursue strategies to delay resistance development, because they see large benefits of Bt crops in farming.
One strategy mandated for years in several countries with Bt crops are refuge areas. Typically, a field is not exclusively sown with Bt seeds, but has a certain share of conventional plants on strips along its borders or inside the field. The rationale is that resistant insects selected under Bt exposure will mate with susceptible individuals found on nearby conventional plants. Consequently, resistant individuals may not take over the population as fast as they would without refuge, especially if resistance is a recessive trait.
Another strategy relies on Bt crops with multiple Bt genes that produce a variety of Bt toxins (called “pyramiding” or “stacking“). Insects resistant to one toxin may not be resistant to others, and it is less likely for simultaneous resistance to emerge. Like refuge areas, pyramiding is widely used and now mostly based on two toxins. Under certain assumptions pyramiding could dramatically cut the need for refuge. In 2007, the EPA approved “natural refuge” for a pyramided cotton, arguing that sufficient refuge is provided by other crops on neighboring fields. In Australia farmers may use pigeonpea, sorghum and corn as natural refuge.
Meanwhile, entomologists are working on improving their incomplete understanding the complex mechanisms involved in resistance evolution. Recently published research suggests that pyramiding might not work as well in delaying resistance as previously thought. In the laboratory, scientists selected cotton bollworm (Helicoverpa zea) for resistance against Cry1Ac. They exposed the resistant insects and a susceptible control group to Bt cotton expressing Cry1Ac/Cry2Ab and found that the group resistant to Cry1Ac exhibited a much higher survival rate than the control group, violating the assumption of redundant killing that is crucial to this strategy. So far, despite multiple reported instances of resistant insects, large-scale failure of Bt crops due to evolved resistance has not occurred, but it may come sooner than expected.

Should refuge requirements be expanded?

This research finding is bad news because the potential of pyramided Bt crops might be lower than believed. (Actually, some scientists have been positively surprised at the long delays observed in resistance development.) Let’s assume the results also apply to other Bt pyramids and insect species (there is evidence to the contrary). What should be made of such a scenario? Should larger refuge areas be required?
Before answering that question, it must be recognized that the sustainable application of a particular technology is not a primary goal of farming. A much more important goal is efficiency. Efficiency means getting the most output (e.g. food) from a set of scarce inputs (natural resources, labor, capital). The technologies transforming inputs into outputs, be they biological, chemical, or mechanical, are valuable only insofar as they contribute towards efficiency.

Cotton field, by Kimberlykv

When deciding whether to expand refuge requirements, policymakers must take into account that there is a tradeoff between the size of the refuge area and productivity. If refuge area increases, more plants will get damaged by pests and hence reduce effective yield. The crucial question is whether the benefits of delaying resistance outweigh the costs of these yield losses and other potential drawbacks of refuges such as the need for additional land, sprays, separation costs, and sowing and harvest times. Costs of monitoring compliance with refuge requirements must also be considered, while pyramiding will incur more R&D expenditures. (In some developing countries with larger monitoring costs, refuge requirements may be less efficient also because of natural refuge in small-scale cropping systems.) The point here is not to question whether the optimal refuge requirement is 0%, 20% or 40%, but to realize that there are costs that have to be weighed against benefits. It is possible that an arms race based on adding more Bt genes is more efficient than slowing resistance development by expanding mandated refuges.
Besides Bt crops, there is a host of other pest management options including chemical control, biological control and cultural control such as ploughing and crop rotation. Like Bt, they all have their particular drawbacks, be it risk of resistance development, low effectiveness, or environmental and economic cost. The most efficient pest management strategy depends on local context, but will involve multiple instruments. For breeders, genes producing insect toxins, whether introduced using conventional or GM techniques, are not the only route towards pest protection. There are exciting possibilities on the horizon, including transgenic plants that emit volatile organic compounds to repel herbivores or attract their natural enemies. The use of nano-silica that kill pests by purely physical means are just one example of potential applications of nanotechnology in pest management. New approaches will have benefits and costs to be assessed against existing alternatives. As of today, there are no magic bullets protecting crops from pests. But there are excellent reasons that we should keep looking for them. Bt will not be the end of the road.

Written by Guest Expert

Jonas Kathage is an agricultural economist with a focus on quantitative empirical research. Since 2013, he has been working at the Joint Research Centre of the European Commission in Seville. His PhD in agricultural economics is from the University of Göttingen. Jonas is an expert on the socio-economics of genetically modified crops. He has also worked on plant protection, climate change mitigation, and other topics.


  1. For sure it is tricky to determine optimal refuges, but some sort of mandatory and enforced requirement is probably necessary because otherwise there is a “free rider problem”: If all other farmers put aside sufficient refuge areas and I don’t, I still get the benefit of longer lasting effectiveness of that particular Bt protein AND I get the benefit of growing crops with a useful trait on all my land. Unless other mechanisms like peer-pressure or widespread altruism ensure that enough farmers set aside large enough refuge areas, the outcome will suboptimal.

  2. One issue I’ve noticed in the public media, is the misleading use of the term “significant” summarizing a study, where the phrase “statistically significant” is used. The anti-GMO lobbies take advantage of the relatively innumerate journalists and the general public when they do their PR’s. Might be worth a blog post that can be linked to.

  3. I guess I agree with you on the free rider problem. But there are already certain refuge regulations in place and let’s assume they take externalities into account. What marginal adjustments should the regulator make when learning that resistance to current Bt genes is more likely to develop than previously thought? Seems to me that stacking up and looking for alternatives to Bt are not necessarily the worst options.

  4. I agree that the argument over resistance is misplaced, resistance is an issue with all pesticides and herbicides, whether conventional or GM.
    However, I wonder if it is a legitimate critique that GM could accelerate resistance because they constantly express the pesticide. Conventional spraying will result in periodic exposure, mostly during peak growing times. There is selective pressure for the pests to become resistance, but the pressure is not constantly applied. The GM crop, however, represents a constant selective pressure, as the plant continuously expresses the pesticide.
    I’m a medical doc/molecular physiologist, from my experience that would be the equivalent to me of having a patient on antibiotics continuously, or to use an agricultural example, the rearing of animals with sustained antibiotic treatment in their feed. Both practices are going to create a constant selective pressure that selects for resistant organisms. For our organisms, you treat with a high dose (above MIC) for a short period then stop. Granted, bacteria are different than insect pests, faster to mutate, can share resistance plasmids etc., but the fundamental biology is the same. Constant selective pressure will result in more resistance and more effective resistance from the pests.
    What is the current thinking on whether the different selective pressures are more or less effective on creating resistance? Is there any?

  5. Mark, you are correct that the continual expression of the Cry protein increases the risk of resistance development. While the mechanism by which the toxin works makes resistance development more difficult, it is not impossible, as we are starting to see.

  6. Interesting.
    Bt crops only exert a selection pressure if and when target insects actually feed on them, and that’s what you want to achieve: protect your crops from insect damage. As you state, conventional spraying may be less effective if it only protects crops during more narrow time windows. (But on the other hand, there may be drift and you “protect” plants outside the field, and it may be less pest-specific, which both adds to the selection pressure.)
    What you have in the case of intermittent spraying is a kind of “temporal” refuge: You spray the entire field but you leave unprotected time windows that lower selection pressure. Whereas with Bt crops you do not plant the entire field but you leave unprotected areas to reduce selection pressure.
    In both cases you accept a minor current reduction in yields compared to a situation where you protect all your crops at all times. Only in the case of Bt crops you do so consciously to be able to protect your crops for more growing cycles – and where this is regulated, society decides that it is willing to bear the costs of monitoring and enforcement to make sure you do it. That is, this becomes an economic question of optimization: The benefit of extending that particular crop protection option into the future comes at the cost of incremental yield reductions throughout, and it also imposes monitoring costs; at the same time you can do R&D to develop new crop protection options that offer new benefits, but this comes also at a cost.
    The challenge for society is then to legislate the refuge areas (and to do investments into R&D) that lead to optimal outcomes… I thought that it may make sense to ensure minimum refuge requirements, whereas Jonas is more fearful of legislative overshooting. In the case of conventional spraying the “temporal” refuge may be simply more implicit.
    (I’m not so sure your health example applies as antibiotics are not so specific, or are they? Is there an antibiotic that only targets a specific, harmful strain of bacteria? If such an antibiotic is given continuously, it doesn’t have any effect unless it encounters the harmful bacteria – but that’s exactly when you want that it protects your patient, won’t you?)

  7. Antibiotics are variably-specific to types of bacteria, but will virtually all suppress normal flora non-specifically. Even as I wrote it, I knew it was a bad example, because of horizontal gene sharing in antibiotics if you make any organisms resistant, you have a risk of making multiple organisms resistant. However, resistance genes come with an energy cost, so bacteria won’t tend to maintain them without consistent selective pressure.
    So, bacteria and animal pests are not a great comparison, I apologize. I was only trying to make the point that continuous selective pressure will more strongly select for resistance than periodic pressure. If you have a selection mechanism that say, kills a substantial portion of the population of a bug one week of every year, that would not create the same evolutionary pressure for mutation and resistance as a selection mechanism that continuously kills the bug population at a lower rate for 6 months of the year. A continuous selective pressure will select for a mutation gene and more rapidly disseminate it among the existing population, and prevent the gene from being diluted out by drift due to continuous selection. So spraying bt at high levels, as we used to, might result in less resistance, than producing it continuously at lower levels in their food source.
    I could imagine a better system would be expression of the bt protein in the plants for a limited period of time under the control of a promoter that is only expressed briefly during the plant life cycle or under the control of some external stimulus. However, whether such a promoter exists that is appropriately temporally selective, and may lead to constitutive expression of the protein in multiple cell types in a way that corresponds to peak predation seems doubtful.
    It’s a problem. I don’t have a good solution to it. But from a genetics view it seems that constitutive expression is less optimal for preventing resistance. Of course, this might be overcome by stacking as suggested, animal pests dont mutate as quickly as bacteria (who can beat stacking and do), so it may have a simple solution. Or, one could imagine rotating through cry genes as one rotates crops, to prevent a single resistance mechanism from becoming fixed in any given pest population.

  8. Hi Mark. You’re probably right that periodic exposure is preferable to constant exposure from a resistance management perspective, everything else equal. On the other hand, constant exposure is preferable to periodic exposure from a crop protection perspective. This is because with periodic exposure, some of the insects will cause damage to the crops before the next application. At least some of this damage can be prevented with constant exposure.
    As you point out, dose also matters. In Bt crops, resistance management strategies rely not only on refuge but also on high dose (“high-dose/refuge strategy”). The amount of toxins expressed can vary between different Bt cultivars. Cases of evolved pest resistance were in part due to low levels of toxins expressed in the Bt cultivars grown in those fields.
    So yes, Bt crops may be inferior to conventional sprays along the dimension of periodic vs. constant exposure with respect to resistance evolution. One question is how much more inferior. And of course farmers, policymakers and disinterested observers should compare Bt crops and conventional insecticides along all relevant dimensions, the whole set of economic and environmental costs and benefits.

  9. Oh, and I hate the nanotech idea. The problem is with that technology we create products that have no existing biological degradation mechanism. The ideal pesticide should be (1)powerful enough to kill the pest, (2) specific to the target with minimal human toxicity, and (3) rapidly degraded to prevent soil contamination, off target effects or concentration of the product in predators. My understanding is that nano fails badly on #3. Without an obvious biodegradable pathway, we shouldn’t be producing them en masse. Not trying to be a luddite or a downer, but it seems to me we should be moving towards biotech like Bt that create functional and specific biological proteins – known to degrade – and away from things like nanotech that suffer from the same flaws of conventional pesticides.

  10. I wonder though – the first recorded instances of Bt resistance occured in applied rather than engineered Bt applications. Always on in a large dose is probably a better approach than on for a little while in a dosage which will vary in many dimensions (off the top of my head I’m thinking how well the spray is applied, the timing of the spray, the presence of the insect at the timing of the spray (what if they turn up 3 days later as your spray is losing efficacy but hasn’t totally been removed)
    What would cause antibiotic resistance to rise given exactly the same approach – high (but not enough to kill the patient) dosage for as long as the bacteria were present, or intermittent dosage which may only partially overlap with the presence of the bacteria?
    The issue that does arise, and can arise with any single overused approach is of course use of a single mode of action everywhere all the time – which is a modality we’re getting out of now – possibly with the capacity to, later down the line, rotate in and out various approaches for even better management.
    There are exciting things coming down the pipeline – corn root worm III for instance – which uses a non-Bt methodology to target insects (Given that this as much as I can find in a quick google search that’s as much as I can disclose about how it works, but it is helluva cooler than just doing Bt again, and again, and again) so concerns about resistance are a temporal worry that need only play out if investment in keeping the technology moving apace is threatened.

  11. You are still on the right path about continual expression of the trait increasing the risk. That is part of why the use of management techniques like refugia and high dose are in place. However, as mentioned in the story, lower levels of refugia diminish the resistance management.
    As Jonas stated, from a plant protection viewpoint, the continuous expression is an advantage. That is why we will continue to work on balancing these different aspects to maintain the protective advantage of the trait and diminish the risks of resistance in the pest insects.

Comments are closed.