FORUM Evaluating Resistance Management Strategies for M ultiple Toxins in the Presence of External Refuges MICHAEL A. CAPRIO Department of Entomology and Plant Pathology, Mail Stop 9775, Mississippi State University, Mississippi State, Mississippi 39762-9775 J. Econ. Entomol. 91(5): 1021-1031 (1998) ABSTRACT The use of5 resistance management strategies in the presence and absence of external refuges (refuges that are spatially isolated from treated fields) was examined in a stochastic model. The 5 strategies were sequential introduction of 2 toxins, rotations and mosaics of toxins, and half- and full-rate mixtures of the toxins. The ability of the strategies to delay resistance was examined first with scheduled sprays, then using an integrated pest management (!PM) component and treating only when populations exceeded an economic threshold. When sprays were scheduled and in the abscnce of an external refuge, 4 of the strategies resulted in similar rates of resistance evolution. The 5th strategy (full-rate mixtures) provided for better insect control but resulted in resistance to both toxins in approximately the same time it took resistance to develop to the 1st toxin in the sequential use strategy. When an economic threshold (!PM strategy) was adopted, all strategies perfornwd approximately the same in the absence of refuges. When a 5%spatially segregated refuge was included in the IPM simulations, resistance was delayed 4.4 times in the sequential introductions, 15-20 times for rotations, mosaics and half-rate mixtures, and 34.8 times for full-rate mixtures (('omplU"edwith sequential introductions in the absence of refuges). Simulations comparing mo- nogenic versus polygenic inheritance of resistance suggested that both types of resistance evolved at approximately the same rate in the absence of refuges, but resistance took 2.2times longer to evolve in the polygenic simulations in the presence of a refuge. Resistance also evolved more rapidly at low ratt'S of gene flow, indicating that the assumption of random mating in resistance models should be examined carefully when external refuges are present. Symmetrical, partial cross-resistance between toxins at 2 resistance loci had a negative impact on all strategies. However, at all levels of cross- resistance, full·rate mixtures continued to delay resistance longest. These simulations demonstrate that full-rate mixtures used in the presence of refuges have the potential to effectively delay rt'Sistan('e evolution. In the absence of refuges and an !PM strategy, however, the full-rate mixture strategy resulted in the most rapid resistance development. Half-rate mixtures were less effective than full-rate mixtures under ideal conditions, but they did not have a negative impact on resistance devt'lopment in the absence of refuges and an !PM strategy and may be less risky than the full-rate strategy over a broader range of likely scenarios. KEY WORDS Bacillus thwingiensis, insecticide resistance, resistance management, dispersal, cross-resistance TilE EVOLlITIONOFpest resistance to insecticides is a major problem facing agriculture today. For example, in tIlt' Midsouthern United States, an average of 9.7 applications of insecticide were made per year on cotton pests, including the heliothine moths, in 1992. As resistance to these chemical insecticides becomes more of a problem, and societal pressures increase to reduce pesticide use (Luttrell 1994), there is reinvig- orated interest in noninsecticidal strategies of cotton insect control. Strategies such as biological and cul- tural controls can reduce pesticide use and, at least in many selection models, reduce the rate of insect ad- aptation to pesticides. Use of resistance management strategies for insecticidal toxins also can decrease the rate of insect adaptation. In many situations, multiple toxins are available for control of pests, and questions arise over the optimal strategy to deploy and use those toxins. For example, many strains of lepidopteran- active Bacillus thuringiensis (Bt) have been identi6ed, and many possibilities exist for the use of those strai ns for control of such lepidopteran pests (McGaughey 1994). The strains could be used sequentially (utilize primarily 1 strain until resistance develops, then move on to a 2nd strain), or alternated over time (rotations) or over space (mosaics). The strains also could be used in mixtures, either as tank mixes of spores or insecti- cidal crystals from each strain or through the produc- tion of transconjugate strains that produce toxins from both of the original strains. Although each strain of B. thuringiensis produces a complex spectrum of toxins (McGaughey 1994), to reduce complexity of the model presented here, it is assumed that each pro- duces a single toxin. In at least some cases, many of the toxins produced by a strain may have low toxicity levels so that most of the mortality is caused by 1 or a few toxins. For example, McGaughey and Johnson 0022-0493/98/1021-1031$02.00/0 © 1998 Entomolo~ical Society of America