Sustained susceptibility of pink bollworm to Bt cotton in the United States

Abstract

Evolution of resistance by pests can reduce the benefits of transgenic crops that produce toxins from Bacillus thuringiensis (Bt) for insect control. One of the world’s most important cotton pests, pink bollworm (Pectinophora gossypiella), has been targeted for control by transgenic cotton producing Bt toxin Cry1Ac in several countries for more than a decade. In China, the frequency of resistance to Cry1Ac has increased, but control failures have not been reported. In western India, pink bollworm resistance to Cry1Ac has caused widespread control failures of Bt cotton. By contrast, in the state of Arizona in the southwestern United States, monitoring data from bioassays and DNA screening demonstrate sustained susceptibility to Cry1Ac for 16 y. From 1996–2005, the main factors that delayed resistance in Arizona appear to be abundant refuges of non-Bt cotton, recessive inheritance of resistance, fitness costs associated with resistance and incomplete resistance. From 2006–2011, refuge abundance was greatly reduced in Arizona, while mass releases of sterile pink bollworm moths were made to delay resistance as part of a multi-tactic eradication program. Sustained susceptibility of pink bollworm to Bt cotton in Arizona has provided a cornerstone for the pink bollworm eradication program and for integrated pest management in cotton. Reduced insecticide use against pink bollworm and other cotton pests has yielded economic benefits for growers, as well as broad environmental and health benefits. We encourage increased efforts to combine Bt crops with other tactics in integrated pest management programs.

Keywords: :

• GMO
• genetically engineered
• integrated pest management
• refuge
• resistance
• sterile insect release
• transgenic

Introduction

Crops genetically engineered to produce insecticidal proteins from the bacterium Bacillus thuringiensis (Bt) kill some key pests and help to reduce reliance on insecticide sprays.^1,2 Such Bt crops were commercialized in 1996 and covered more than 58 million hectares worldwide in 2010.³ The primary threat to the continued success of Bt crops is evolution of resistance by pests.^4,5 Field-evolved (= field-selected) resistance entails a genetically based decrease in susceptibility of a population to a toxin caused by exposure of the population to the toxin in the field.^6,7 Although many target pest populations remain susceptible, some degree of field-evolved resistance to Bt crops has been reported in at least seven species of major pests.^6-17 Field-evolved resistance to Bt toxins has caused field control failures in some, but not all cases.^6-17

Here we focus on resistance to Bt cotton in the pink bollworm (Pectinophora gossypiella), one of the world’s most devastating pests of cotton. First-generation Bt cotton produces only one Bt toxin called Cry1Ac.⁷ We will use the term “Bt cotton” to refer to such cotton, unless noted otherwise. Monitoring data from three countries reveals three different outcomes for pink bollworm resistance to Bt cotton. The worst outcome has occurred in western India, where pink bollworm evolved resistance resulting in widespread field control failures of Bt cotton in 2009.^10,13 The situation is more positive in China, where the frequency of resistance to Cry1Ac increased significantly in 2008–2010 compared with 2005–2007, but the highest percentage of resistant individuals detected in a pink bollworm population during 2010 was 6%.¹⁶ The most positive outcome has been documented in the state of Arizona in the southwestern United States, where pink bollworm susceptibility to Cry1Ac has not decreased after 16 y of exposure to Bt cotton. Here we review the success of Bt cotton against pink bollworm in Arizona.

Pink Bollworm and Bt Cotton in Arizona

In Arizona, Bt cotton kills essentially 100% of susceptible pink bollworm larvae.¹⁸ Nonetheless, our initial work indicated a high risk of pink bollworm evolving resistance to Bt cotton in Arizona.¹⁹ In the lab, selection with Cry1Ac in diet quickly yielded resistance to Bt cotton in several strains of pink bollworm derived from the field.^18,20 In Arizona, the larvae of pink bollworm feed almost exclusively on cotton and more than half of the state’s cotton has been planted to Bt cotton since 1997²¹ (Fig. 1). Thus, the pink bollworm has the genetic potential to evolve resistance and has been under intense selection for resistance in the field.

Figure 1. The percentage of cotton planted to non-Bt cotton refuges in Arizona from 1997 to 2009.²¹

Although pink bollworm evolved resistance to Bt cotton in about nine years in India,^13,17 it has remained susceptible in Arizona after 16 years. Below we review the monitoring data demonstrating sustained susceptibility and the key factors delaying pink bollworm resistance to Bt cotton in Arizona, which seem to be refuges of non-Bt cotton from 1996 to 2005, and mass releases of sterile pink bollworm moths from 2006 to 2011. Also, as summarized below, the pink bollworm eradication program using Bt cotton, sterile moth releases and other tactics has dramatically reduced pink bollworm populations and insecticide sprays against this pest in Arizona.

Resistance Monitoring: 1997–2011

We have used three methods for monitoring resistance of pink bollworm to Bt cotton and Bt toxin Cry1Ac: (1) diet bioassays, (2) DNA screening and (3) field efficacy tests. All three methods have demonstrated no net increase in resistance to Cry1Ac or Bt cotton producing Cry1Ac in Arizona field populations of pink bollworm.^18,21-23 Results from field efficacy tests conducted from 1997–2005 demonstrated sustained effectiveness of Bt cotton against pink bollworm.^18,24 Below we summarize results from our two primary resistance monitoring methods, diet bioassays and DNA screening.

Diet bioassays.

From 1997 to 2008, we monitored resistance to Cry1Ac in Arizona using diet bioassays as follows: Each year, we tested an average of 2,730 larvae that were the progeny of individuals collected in cotton fields from an average of 12 sites statewide. The progeny of field-collected pink bollworm from each site were reared and tested separately. Because pink bollworm larvae were scarce in most Bt cotton fields, nearly all of the sampled fields that provided enough individuals to produce sufficient progeny for testing in bioassays were non-Bt cotton fields. For example, non-Bt cotton fields accounted for 9 of 10 fields yielding progeny that were tested with bioassays in 1997, 10 of 10 fields in 1998, and 11 of 13 fields in 1999 (mean = 92% non-Bt cotton fields).¹⁸ Neonates were tested individually for 21 d on wheat germ diet without toxin (control) or on diet with a diagnostic concentration (10 ∝g Cry1Ac per ml diet), which kills susceptible homozygotes and heterozygotes, but not resistant homozygotes.¹⁸ Based on the observed recessive inheritance of resistance at the diagnostic concentration, the Cry1Ac resistance allele frequency for each site was estimated as the square root of the frequency of survivors after adjustment for control mortality.¹⁸ We calculated the 95% confidence interval for each yearly statewide mean resistance allele frequency using the bootstrap method with 10,000 repetitions.¹⁸

The results from diet bioassays show that the resistance allele frequency was 0.16 (95% confidence interval = 0.05 to 0.26) in 1997, the second year Bt cotton was grown widely in Arizona (Fig. 2). This turned out to be the highest resistance allele frequency measured in Arizona pink bollworm populations in 15 y of monitoring (1997–2011). Although the resistance allele frequency estimated from diet bioassays fluctuated from 1998 to 2008, it never reached the level detected in 1997 (Fig. 2). Moreover, no resistant individuals were detected with diet bioassays in 2007 or 2008 (n = 3,602 larvae tested),²¹ the most recent years in which we collected enough pink bollworm from field populations to enable testing of their progeny with diet bioassays.

Figure 2. Pink bollworm resistance allele frequency (with 95% confidence intervals) in Arizona from 1997 to 2008 estimated from laboratory diet bioassays with Cry1Ac.²¹

The estimated resistance allele frequency for pink bollworm in 1997 was stunning at the time, in part because it was so high compared with estimates from other pests targeted by Bt crops. This estimate for pink bollworm is 100 times higher than the frequency of resistance to Cry1Ac in field populations of Heliothis virescens estimated before Bt cotton was commercialized²⁵ and no major alleles for resistance to Cry1Ab in Bt corn were detected in extensive screening of field populations of Ostrinia nubilalis.^26,27 Nonetheless, the estimate for pink bollworm in 1997 does not provide evidence of field-evolved resistance, because it was the first estimate and thus did not demonstrate a decrease in susceptibility. Despite considerable effort, neither members of our research team nor others have identified reasons to discount the relatively high estimated resistance allele frequency for pink bollworm in 1997.

We do not know why the frequency of resistance was so high in 1997, but we have speculated about various possibilities.²³ We infer that one or more key factors influencing resistance changed between 1997 and 1998, causing the observed decline in resistance frequency between these two years (Fig. 2).²³ One factor that might have changed is the concentration of Cry1Ac in Bt cotton. In 1997, some problems with Bt cotton were observed in the field, but it is not clear if these were caused by moderate concentrations of Cry1Ac or putative Bt cotton bolls that produced no Cry1Ac. If Bt cotton grown during 1996 and 1997 produced enough Cry1Ac to kill homozygous susceptible larvae but not enough to kill most heterozygous larvae, non-recessive inheritance of resistance could have quickly raised resistance allele frequency.²³ If increased Cry1Ac concentration in 1998 and thereafter killed all or nearly all heterozygous larvae, this shift to recessive inheritance could have caused decreases in resistance.²³ Another possibility is that the effects of fitness costs and incomplete resistance, which varied substantially in experiments, increased after 1997. Such changes could be associated with changes in agronomic practices, weather or other environmental factors.²³ A third possibility is a decreased effect from 1997 to 1998 of one or more factors other than Bt cotton favoring the resistance alleles.²³

Because our relatively high initial estimate of resistance allele frequency in pink bollworm has produced some confusion in the literature, we review some additional details about the estimate and three resistant strains generated from the pink bollworm collected from the field in 1997. Five of the 10 Arizona cotton field populations sampled in 1997 yielded survivors of exposure to the diagnostic concentration of Cry1Ac.¹⁸ Of the five sample sites where resistance was detected, four were non-Bt cotton fields and one was a Bt cotton field. Excluding the data from the lone Bt cotton field, the estimated resistance allele frequency is 0.14, which does not differ significantly from the estimate of 0.16 noted above, which is based on data from all 10 sample sites.

From the initial bioassays of progeny from the 10 sample sites, we pooled 169 survivors of exposure to 1, 3.2 and 10 ∝g Cry1Ac per ml diet to start the Arizona pooled resistant strain (AZP-R). After four generations of rearing on diet without toxin followed by selection at the diagnostic concentration of Cry1Ac, the concentration of Cry1Ac killing 50% of larvae (LC₅₀) was 300 times higher for AZP-R compared with the susceptible APHIS-S strain.¹⁸ This rapid response to selection confirms that resistance alleles were not rare in the field populations of pink bollworm sampled in 1997. Moreover, two additional resistant strains (MOV97-R and SAF97-R) were started with pink bollworm collected in 1997 from Mohave in western Arizona and Safford in eastern Arizona, respectively.²⁰ As with AZP-R, selection of each of these strains quickly produced high levels of resistance to Cry1Ac.²⁰

We discovered that each of the three lab-selected resistant strains had a distinct profile in terms of the frequency of three mutant cadherin alleles (r1–r3) linked with resistance to Cry1Ac.^20,28 Consistent with their origins and supporting the hypothesis that resistance alleles were not rare in the field populations sampled in 1997, all three cadherin resistance alleles occurred in AZP-R, while only two cadherin resistance alleles were in each of the two strains started from single sample sites (r1 and r3 in MOV97-R and r1 and r2 in SAF97-R).^20,28

DNA screening.

In screening DNA from more than 9,000 pink bollworm larvae and adults collected from the field in the southwestern United States during 2001–2011, we did not detect any of the three mutant cadherin alleles that are linked with resistance to Cry1Ac in lab-selected strains of pink bollworm^21,22 (unpublished data). We used polymerase chain reaction (PCR) primers that specifically amplify the three mutant cadherin alleles linked with resistance to Cry1Ac in the three lab-selected resistant strains of pink bollworm from Arizona mentioned above (AZP-R, SAF97-R, MOV97-R) and in a fourth, independently derived lab-selected resistant strain (APHIS-98R, which had r1–r3) from Arizona.^28-30

Compared with our bioassays, this DNA screening has several advantages and an important disadvantage.^22,29 Unlike bioassays, which do not distinguish heterozygotes from homozygous susceptible pink bollworm larvae, DNA screening can detect single resistance alleles in heterozygotes. Also, whereas bioassays require live larvae for testing, DNA screening can be done with properly preserved dead individuals or groups of individuals of any life stage. However, while bioassays can detect resistance caused by any mechanism, our DNA screening could detect only the three known mutant cadherin alleles and thus would miss resistance caused by other cadherin mutations or mutations in other genes.

The lack of resistance alleles detected by DNA screening could reflect three scenarios that are not mutually exclusive: (1) the three mutant cadherin alleles occurred in the samples screened but were not detected, (2) resistance in the samples was conferred by mutations other than the three known mutant cadherin alleles and (3) resistance was extremely rare. The first factor probably had little or no effect, because we included extensive controls to check our methods, including “blind” positive controls.²² For example, in tests conducted from 2001 to 2005, DNA screening detected 97% of positive controls (n = 100) that were not known to the person performing the PCR testing.²² The effect of the second factor is more difficult to exclude because we cannot rule out the possibility of alternative mechanisms of resistance. Indeed, we recently isolated a fourth mutant cadherin allele that is linked with resistance to Cry1Ac in a strain of pink bollworm that was selected from a susceptible strain from Arizona.^31,32 In addition, many cadherin resistance alleles have been found in Helicoverpa armigera.³³ Nonetheless, unlike the fourth cadherin resistance allele in pink bollworm and the first cadherin resistance allele identified from Heliothis virescens,^34,35 the three cadherin alleles for which we screened were not rare in field populations during 1997.^18,20,28 Most importantly, independent data from bioassays and field efficacy testing are consistent with the third interpretation: resistance of pink bollworm to Cry1Ac remained rare.

Refuges Delay Resistance: 1996–2005

The refuge strategy has been the predominant approach for delaying pest resistance to Bt crops worldwide.^5,7,8 The concept underlying this strategy is that refuges of non-Bt host plants near Bt crops will promote survival of susceptible pests. Ideally, most of the rare resistant pests emerging from the Bt crop will mate with the relatively abundant susceptible pests from the refuges. If resistance to the Bt crop is inherited as a recessive trait, the Bt crop will kill the heterozygous progeny produced by such matings between homozygous susceptible moths from the refuge and homozygous resistant moths emerging from the Bt crop. Retrospective analyses of global resistance monitoring data suggest that when the assumptions of the refuge strategy apply, refuges can substantially delay resistance to Bt crops.^6-8,36 A large-scale field test incorporating eight years of data also shows that refuges of unsprayed cotton delayed whitefly resistance to the insecticide pyriproxyfen.³⁷ We found four factors that probably delayed pink bollworm resistance to Bt cotton in Arizona from 1996 to 2005: abundant refuges of non-Bt cotton, recessive inheritance of resistance, fitness costs associated with resistance and incomplete resistance.²³

Refuges.

Growers planted abundant refuges of non-Bt cotton in Arizona from 1996 to 2005. The statewide percentage of cotton acreage planted to non-Bt cotton was > 50% in 1996, the first year Bt cotton was planted commercially, and > 25% every year from 1996 to 2005 (Fig. 1). A more detailed spatial analysis for 1998–2003 showed an average of > 86% compliance with the size and distance requirements for non-Bt cotton refuges in the six counties of Arizona that produced > 95% of the state’s cotton.³⁸ Furthermore, the abundance of susceptible moths was higher in locations compliant with the size and distance requirements than in non-compliant locations.³⁹

Recessive inheritance.

Results from many tests on Bt cotton plants showed substantially increased survival of larvae from lab-selected resistant strains compared with susceptible strains, yet the F1 hybrid larvae from crosses between resistant and susceptible strains did not survive on Bt cotton.^{18,20,28,40,41} Thus, inheritance of resistance to Bt cotton was recessive, which is expected to enhance the success of the refuge strategy.

Fitness costs.

Fitness costs occur when fitness in refuges is lower for insects with resistance alleles than insects without resistance alleles.^42,43 Thus, fitness costs select against resistance in refuges. A series of experiments on non-Bt cotton plants showed costs that lower the fitness of resistant individuals compared with susceptible individuals.^23,41 Like resistance to Bt cotton, these costs were generally recessive, which means they occurred in resistant individuals, but not in F1 hybrid progeny from crosses between resistant and susceptible adults.

Based on several experiments conducted on non-Bt cotton plants, the mean estimated fitness was 0.46 for resistant pink bollworm relative to susceptible pink bollworm (fitness defined as 1).²³ This indicates a 54% fitness cost, which reflects lower larval survival on non-Bt cotton plants for resistant than susceptible strains and lower overwintering survival for resistant than susceptible strains.^23,44 This estimate does not include fitness costs affecting other life history traits such as the ability of males to compete for mates or first-male paternity when females mate with more than one male.^45,46

Incomplete resistance.

Incomplete resistance occurs when resistant insects suffer a disadvantage on Bt crop plants relative to non-Bt crop plants, which reduces the selective advantage of resistant insects on Bt crop plants.^23,41 Analysis of data from experiments on plants indicated that the net reproductive rate of resistant pink bollworm was 0.35 on Bt cotton relative to non-Bt cotton (defined as 1).²³ This incorporates effects of incomplete resistance on larval survival, the proportion of adults that are female and fecundity; all were lower on Bt cotton than non-Bt cotton.²³ Egg hatching rate was similar for progeny of individuals that were reared on Bt and non-Bt cotton.⁴⁶ Taking into account the effects of fitness costs and incomplete resistance, the estimated mean fitness of resistant pink bollworm on Bt cotton was only 0.16 (0.46 × 0.35) relative to susceptible pink bollworm on non-Bt cotton.

Modeling.

To project the potential rate of evolution of pink bollworm resistance to Bt cotton in Arizona under the refuge strategy, we incorporated into mathematical models the experimental results summarized above on refuge percentage, inheritance of resistance, fitness costs and incomplete resistance.²³ This modeling confirmed the importance of each of these four factors. The modeling results also show that for a substantial set of realistic values for the key factors, resistance was not expected to increase.

Sterile Insect Releases Delay Resistance and Suppress Populations: 2006–2011

A large collaborative team has developed and deployed a multi-tactic eradication program in Arizona that uses sterile insect releases instead of refuges to delay pink bollworm resistance to Bt cotton.²¹ This approach for delaying resistance relies on the idea that rare resistant pink bollworm moths emerging from Bt cotton will mate primarily with sterile moths, yielding no fertile progeny. Computer simulations show that this tactic works in principle, even if pest resistance is inherited as a dominant trait.²¹ The US Environmental Protection Agency (EPA) convened a Scientific Advisory Panel to review the proposed eradication program.⁴⁸ The EPA followed the Panel’s recommendation and decided to allow Arizona cotton growers to plant up to 100% Bt cotton producing either one toxin (Cry1Ac) or two toxins (Cry1Ac and Cry2Ab).⁴⁸

The eradication program was introduced in phases in Arizona, starting in 2006 and moving from east to west. During the nine years before the program began, the percentage of cotton acreage planted to non-Bt cotton in Arizona ranged from about 50% in 1997 to 27% in 2005 (mean = 37%). With the onset of sterile releases, the statewide percentage of non-Bt cotton dropped to 15% in 2006, 8% in 2007, 2% in 2008 and 3% in 2009 
(Fig. 1).²¹ The mean percentage of non-Bt cotton was 7% from 2006 to 2009 and large areas of the state had only Bt cotton with no refuges. The percentage of Arizona’s Bt cotton producing two toxins (Cry1Ac and Cry2Ab) increased from 11% in 2005 to 79% in 2009, and was close to 100% in 2010, after the registration for Bt cotton producing only Cry1Ac expired in 2009.^7,21

During May to October of each year from 2006 to 2009, about 2 billion sterile pink bollworm moths were released from airplanes into Arizona’s cotton fields. Releases were made two to three times per week in all cotton fields statewide. An average of approximately 25,000 sterile moths were released per ha of cotton each year. Because higher moth emergence was expected in non-Bt cotton fields, the release rate was about ten times higher in non-Bt cotton fields (4,000 moths per ha per week) than in Bt cotton fields (400 moths per ha per week).

Pink bollworm populations in Arizona have declined dramatically since the eradication program began in 2006 (Fig. 3). In 2009, only two pink bollworm larvae were found in 16,600 bolls of non-Bt cotton screened statewide. This yields an infestation rate of 0.00012, which is a 99.9% decline from the infestation rate of 0.15 in 2005. Likewise, the number of wild male pink bollworm moths caught per trap per week in Bt cotton fields dropped from 26.7 in 2005 to 0.0054 in 2009, a 99.98% decrease. The decrease in pink bollworm populations during the eradication program was much steeper than the decline observed with the planting of Bt cotton before the eradication program began in Arizona and the declines in other target pests associated with planting of Bt crops in other regions.^2,49-54

Figure 3. Pink bollworm abundance in Arizona before and during the eradication program.²¹ (A) Larval infestation of non-Bt cotton bolls from 1997 to 2009. Linear regression shows that the slope, which indicates the decrease in infestation per year, was 18 times steeper from 2006 to 2009 (-0.81, r² = 0.97, p = 0.012) than from 1997 to 2005 (-0.044, r² = 0.42, p = 0.059). (B) Wild male pink bollworm moths trapped in Bt cotton fields from 1998 to 2009. Linear regression shows that the slope, which indicates the change in moths trapped per year, was significantly negative from 2006 to 2009 (-1.0, r² = 0.92, p = 0.04), but did not differ significantly from zero from 1998 to 2005 (0.017, r² = 0.071, p = 0.52).

While pink bollworm populations declined dramatically, insecticide sprays to control this pest also plummeted.²¹ The mean number of sprays per field per year targeting pink bollworm in Arizona was 2.7 from 1990 to 1995, which fell to 0.64 from 1996 to 2005 with use of Bt cotton, before the eradication program. During the eradication program, this mean decreased to 0.14 in 2006, 0.013 in 2007, 0.0029 in 2008 and 0 in 2009.

The average annual cost of pink bollworm to Arizona cotton growers, including yield losses and insecticide sprays, was $18 million for 1990 to 1995, $5.4 million from 1996 to 2005 and only $172,000 for 2006 to 2009.⁵⁵ By using Bt cotton as part of a comprehensive integrated pest management program, Arizona growers also greatly reduced insecticide sprays against all cotton pests, including those not killed by Bt cotton, saving a cumulative total of $200 million from 1996 to 2009.^55,56

Conclusions

The sustained susceptibility of pink bollworm to Cry1Ac in Arizona contrasts with less positive outcomes in India and China.^10,13,16 In India, refuges of non-Bt cotton have been required, but apparently compliance has been low.^17,57 In China, non-Bt cotton refuges have not been required.¹⁶ This lack of refuges may be a key factor promoting faster evolution of pink bollworm resistance in these two countries compared with the United States. Wan et al. suggest that another factor accelerating pink bollworm resistance in India and China may be a lower concentration of Cry1Ac in Bt cotton in these countries compared with the United States, which could increase survival of heterozygotes and thus increase the dominance of resistance, but this hypothesis has not been tested. Farmers planted Bt cotton in India before it was officially approved in 2002.⁵⁸ Monsanto⁵⁹ speculated that Cry1Ac concentration is lower in the unapproved Bt cotton than approved Bt cotton in India, and that early use of unapproved Bt cotton contributed to the resistance problem. However, the unapproved Bt cotton came to light in India in 2001 because of its efficacy against bollworms and remained in wide use after approved Bt cotton seed became available.⁵⁸

In Arizona, Bt cotton has had a superb positive impact. It has been a key tool for suppressing pink bollworm, a major cotton pest. The sustained susceptibility of pink bollworm to Bt cotton has provided a cornerstone for the pink bollworm eradication program and for integrated pest management in cotton. The reduction in insecticide use against pink bollworm and other cotton pests has provided economic benefits for growers, as well as environmental and health benefits. Elimination of sprays of broad-spectrum insecticides targeting pink bollworm adults, particularly chlorpyrifos and methyl parathion, has been especially beneficial.

The objective of the pink bollworm eradication program is to eliminate this pest from all cotton-growing regions of the US and adjacent areas of northern Mexico.^60,61 This binational program, funded by cotton growers and the US Department of Agriculture, has benefitted from a strong grower commitment, public investment in sterile insect technology, a well-developed infrastructure for monitoring pink bollworm resistance and population density, virtually 100% efficacy of Bt cotton against pink bollworm, and this pest’s nearly exclusive dependence on cotton in Arizona. While great progress has been achieved, eradication of pink bollworm remains challenging as this invasive pest is resilient and can move long distances.

Because several tactics were implemented simultaneously in the eradication program in Arizona, including reduction in non-Bt cotton refuge percentage, increased percentage of Bt cotton producing two toxins, sterile moth releases and early season pheromone treatments to disrupt mating in non-Bt cotton fields, we cannot conclusively attribute the suppression of pink bollworm to any single tactic. Moreover, we do not know if the success of this program can be replicated with other pests or even with pink bollworm in other parts of the world. Nonetheless, our results support the idea that Bt crops are likely to be most useful when combined with other tactics for integrated control of pests. Much research on delaying resistance to Bt crops has focused on the size and spatial configuration of refuges and the number and type of toxins used in plants. We encourage more effort to combine Bt crops with other tactics in integrated pest management programs.

Acknowledgements

We thank two anonymous reviewers for their comments that helped to improve the paper. Preparation of this paper was supported by USDA Agriculture and Food Research Initiative Grant 2008-35302-0390.

Disclosure of Potential Conflicts of Interest

Research reported in this article was supported by organizations that may gain or lose financially through its publication: Arizona Cotton Growers Association, Arizona Cotton Research and Protection Council (ACRPC), Bayer CropScience, Cotton Foundation, Cotton Inc. Dow AgroSciences, Monsanto and National Cotton Council. L.A. and R.T.S. are independent contractors paid by ACRPC. L.L. and M.W. are employees of ACRPC. T.J.D. is an employee of Bayer CropScience. B.E.T. is a coauthor of a patent application on engineering modified Bt toxins to counter pest resistance, which is related to research described by Tabashnik et al. (Nature Biotechnology 2011; 29:1128–31).