Solving the Problem of Sustainable Use of Bt Crops

Transgenic plants producing insecticidal protein derived from Bacillus thuringiensis (Bt) have been widely adopted since their commercial introduction in 1996. The widespread adoption of such plants has reduced use of conventional insecticides while attaining yield gains, thus providing economic, environmental and human health benefits. However, the benefits from Bt crops will be reduced or even eliminated if pests develop resistance to these toxins so that Bt crops are less or no longer effective. Although field resistance to Bt crops has not yet been found in the continental U.S., resistance to Bt sprays has been found in diamondback moth and greenhouse populations of cabbage looper. Hence, considerable attention has been devoted to developing management programs to prevent, delay or even reverse the spread of resistance. The existing literature on the economic management of pest resistance has generally assumed that pest susceptibility (the converse of pest resistance) is nonrenewable, which means resistance can only increase once it has happened. With the assumption of irreversible resistance, the optimal policy is not to exhaust susceptibility before the new technology, if any, becomes available. Following this logic, the U.S. Environmental Protection Agency (EPA) currently requires Bt crop growers to also plant non-Bt (conventional) crops on a minimum percentage of their total Bt crop acreage as a refuge for susceptible (Bt toxin sensitive) pests. Refuge allows susceptible pests to survive and mate with resistant adults surviving on Bt crops and so delays the development of resistance in the pest population. The cost of this resistance control method includes yield loss of conventional crops relative to Bt crops and sometimes conventional pesticide use. Instead of assuming susceptibility is nonrenewable, some have proposed an alternative source of susceptibility — mass-rearing and releasing harmless susceptible (toxin-sensitive) pests into the environment (Alphey et al. 2007). With this resistance control method in hand, pest susceptibility is now renewable and resistance could be reversed. Based on this method, it is plausible to predict that the optimal resistance management policy should take a cycling pattern, i.e., not planting refuge and allowing resistance to increase until a critical threshold is reached. Once resistance exceeds this threshold, it becomes economical to release susceptible pests to reverse the development of resistance. After resistance is reduced to a desired level, again no refuge needs to be planted and no alternative resistance management actions need to be used until the resistance reaches the threshold again. The cycling feature of the optimal path is due to the reversibility of resistance under this scenario. Also, there is no need to continuously use a resistance control method (as in the case of current refuge policy), since the desired level of resistance can be achieved, at a cost, when needed. However, the crucial consideration is that it is less costly to manage resistance when it is low rather than high. The task of dynamic programming is to find the optimal path where the marginal benefit of resistance control can just cover the marginal cost of resistance. We notice the cycling pattern of this resistance management model and introduce a real options approach originally used in financial analysis. We propose to view pest susceptibility as an investment and the possibility to release susceptible pests as a real option, which can be exercised to improve the expected value of this investment. We formulate the social planner’s decision of releasing susceptible pests as an optimal sequential stopping problem and solve it using stochastic dynamic programming. We first consider releasing pests with a constant ratio for released to natural pests, but then extend the model to allow for selection of a time-varying ratio. The cost of this resistance control technology includes only the cost of monitoring resistance and rearing and releasing susceptible pests. The cycling pattern of this resistance management method is attractive since cost tends to be lower than continuous refuge-based resistance management, as costs are targeted to areas and times when it is needed, rather than being annually implemented across all acres planted to Bt crops. We compare three resistance management programs: refuge only, releasing susceptible pests only, and using both refuge and released pests. The dynamic optimizations are solved numerically using parameters for Bt corn and western corn rootworm. Consistent with previous economic analyses, our dynamic refuge study finds that more refuge should be planted early in commercialization to prevent the rapid development of resistance and that refuge should be planted continuously so that the evolution of resistance is always under control. Once refuge is no longer planted, resistance develops rapidly. For the dynamic releasing of susceptible pests, there is a steady state (economic threshold) of the level of resistance. Whenever the threshold is crossed, resistance control is triggered. The optimal path of the pest population peaks whenever the resistance control (releasing susceptible pests) is exercised. Intuition might consider the pest population peaks as a failure of optimization, however, taking into account that those pests are harmless as long as Bt toxins stay effective, these “bumps” in pest population are fairly acceptable. Our analysis compare these three resistance management programs and find that the optimal resistance control strategy depends on many factors such as yield loss on refuge, the cost of monitoring resistance and rearing and releasing pests, as well as the natural growth rate of pests and the discount rate. We see high potential for generating discussion as the EPA and researchers have been examining changes in refuge requirements for Bt corn, including the recent approval use of mixed seed; however, mitigation once resistance has developed has received little attention. Although we discuss resistance control mostly in the scenario of agricultural pests, the same or similar methodology can also be used in a much broader context in public health economics such as antibiotic resistance or using genetically engineered mosquitoes to fight malaria.