Resistance Management: Slowing Pest Adaptation to Transgenic Crops

Abstract

Transgenic crops that are genetically modified to produce insecticidal proteins from the common bacterium Bacillus thuringiensis (Bt) can help to control pests while reducing reliance on insecticide sprays. So far, no insects have evolved resistance in the field to Bt transgenic crops. However, diamondback moth populations have evolved resistance to Bt sprays in the field and many pests have evolved resistance to Bt toxins in the laboratory. To delay resistance, the refuge strategy provides host plants that do not produce Bt toxins, thereby promoting survival of susceptible pests. In Arizona, Bt cotton has been extremely effective in controlling the pink bollworm (Pectinophora gossypiella), a major pest. Despite a surprisingly high frequency of resistance in 1997, resistance did not increase in Arizona field populations of pink bollworm from 1997 to 1999. Nonetheless, pink bollworm and other insects will eventually evolve resistance, so any particular transgenic crop variety is not a permanent solution to pest problems. Instead, transgenic crops can be used in harmony with other tactics as part of integrated pest management. Evaluations of transgenic crops should consider their advantages and disadvantages compared with alternatives. If transgenic crops can greatly reduce use of hazardous insecticides, as achieved in Arizona cotton, great benefits may occur.

Keywords:

• Bacillus thuringiensis
• cotton
• evolution
• genetically modified crop
• genetic engineering
• pink bollworm

1 Introduction

Insects and mites have a remarkable ability to adapt, as demonstrated by evolution of resistance to insecticides in more than 500 species (Georghiou & Lagunes-Tejeda, 1991). Because of problems with resistance, as well as environmental hazards associated with conventional insecticides, scientists turned to Bacillus thuringiensis for pest control.

The commonly occurring soil bacterium Bacillus thuringiensis, or Bt, is a natural pathogen of some important insect pests (Charles et al., 2000). Bt has little or no toxicity to most non-target organisms, including humans, wildlife, and most beneficial insects (Charles et al., 2000). With genetic engineering, the Bt genes that encode insecticidal crystal proteins – called Cry proteins – have been incorporated into the genomes of some crop plants, enabling these transgenic plants to produce their own environmentally benign insecticide. Unlike most conventional neurotoxic insecticides, Bt proteins are toxic to insects only if ingested. They kill by binding to and disrupting midgut membranes (Charles et al., 2000). Although other genes could be used in transgenic plants to protect them from insect attack, Bt genes are currently the only ones used in large scale agriculture.

Bt crops are being grown on millions of hectares in the USA and elsewhere (James, 2000). Continuous production of Bt toxins in such transgenic crops exposes pest populations to intense selection to evolve resistance. Evolution of resistance is a genetically-based decrease in susceptibility of an insect population to a toxin caused by exposure of that population to the toxin. Note that this use of the word ‘resistance’ differs from another use of the same word, as in transgenic plants that have ‘resistance’ to insects.

In this article, we review pest resistance to Bt and resistance management with special emphasis on the refuge strategy and pink bollworm resistance to Bt cotton in Arizona, USA. We conclude by summarizing information about resistance and recommending that evaluations of transgenic crops should weigh their advantages and disadvantages relative to alternatives.

2 Pest resistance to Bt

The diamondback moth (Plutella xylostella) is the only insect with field populations known to have evolved resistance to Bt toxins (Tabashnik, 1994a; Frutos et al., 1999). In many parts of the world, the diamondback moth evolved resistance after repeated exposure to spray applications of Bt. Despite large scale use of Bt in crops since 1996, no cases have been reported of any pests evolving resistance to Bt crops in the field. However, laboratory selection results show that many pests have the genetic potential to evolve resistance to Bt toxins (Tabashnik, 1994a; Frutos et al., 1999).

Since 1994, the number of Bt toxin gene sequences reported has climbed from nil to more than 130. So, why worry about resistance? If so many Bt toxin genes are available, why not just switch to a new one when resistance occurs? Three factors make resistance a major concern, despite the availability of many Bt toxin genes: 1) Bt toxins are specific, so relatively few toxins will kill any particular pest, 2) the time and cost of introducing new toxin genes is substantial, and 3) cross-resistance.

Cross-resistance occurs when exposure to one or more toxins causes resistance to other toxins. Generally, if an insect evolves resistance to one Bt toxin, it will also be cross-resistant to closely related Bt toxins. For example, in one strain of diamondback moth from Hawaii, selection with Bt sprays containing a mixture of 3 Bt toxins (Cry1Aa, Cry1Ab, Cry1Ac) caused extremely high resistance to those 3 toxins. It also caused cross-resistance to Cry1F and Cry1J. In contrast, 6 less related toxins (Cry1B, Cry1C, Cry1D, Cry1I, Cry2A, Cry9C) were still effective against the resistant larvae (Tabashnik et al., 1996).

Based on studies of 3 major lepidopteran pests-diamondback moth, tobacco budworm (Heliothis virescens), and Indianmeal moth (Plodia interpunctella), key features of the most common mode of resistance to Bt have been identified. This type of resistance, named ‘mode 1’, entails extremely high resistance to at least one Cry1A toxin, recessive inheritance of resistance, a resistance mechanism of reduced binding of toxin to midgut membrane target sites, and a narrow spectrum of cross-resistance (Tabashnik et al., 1998).

3 Resistance management

The goal of resistance management is to delay the evolution of resistance in pests. To date, resistance management has produced much theory, but little rigorous evidence from the field. Most experiments have been in the laboratory or on a small scale in the field. Retrospective analyses of patterns of resistance evolution in the field and anecdotal reports have also provided some insights.

Heavy reliance on mathematical models and computer simulations has occurred because this approach is relatively fast, inexpensive, safe, and flexible. After a simulation model has been developed, it can be used to explore potential consequences of a wide variety of management options within days or even hours. In contrast, large scale field experiments are expensive and may require many years for completion. Perhaps more importantly, such experiments could accelerate the very resistance problems that one hopes to avoid or at least delay.

4 The refuge strategy

Although many resistance management tactics are potentially useful, most attention has focused on the refuge strategy for delaying resistance to Bt crops (Georghiou & Taylor 1977; Tabashnik, 1994b; Gould, 1998 ). In the refuge strategy, a segment of the population is intentionally left unexposed to toxin to enable survival of susceptible pests. In the USA, the refuge strategy is required for Bt cotton, maize, and potato. In Australia, the refuge strategy is in place for Bt cotton, and in Canada it is used for Bt maize.

Let's consider the key assumptions of the refuge strategy. The first assumption is that resistance is controlled primarily by one gene with two alternative forms, called alleles. The R allele confers resistance, the S allele susceptibility. Although this is somewhat of an over-simplification, it is a reasonable starting point. In diamondback moth, for example, a major gene is responsible for resistance to Bt toxins (Heckel et al., 1999). With the assumption of a single genetic locus with two alleles, 3 genotypes occur: SS are susceptible homozygotes, RS are heterozygotes, and RR are resistant homozygotes.

Before an insecticide is used widely, R alleles are usually rare. Further, RR homozygotes are extremely rare because they must receive one copy of the resistance allele from their mother and a second copy from their father. Thus, in the early stages of resistance evolution, most R alleles are carried by heterozygotes, and the mortality of heterozygotes is crucial. If heterozygotes (RS) are killed by transgenics, resistance to the transgenics is recessive. If they survive transgenics, resistance is dominant.

In principle, the refuge strategy works as follows: because resistant (RR) individuals are rare, only a few resistant adults will emerge from transgenic plants. In contrast, large numbers of susceptible adults (mostly SS) will emerge from the untransformed refuge plants. If resistant adults mate with susceptible adults from the refuge, most of their progeny will be heterozygotes. Results from models suggest that if such heterozygous progeny are killed by the transgenic crop, evolution of resistance will be substantially delayed.

In summary, the refuge strategy is expected to work best when resistance is recessive, extensive mating occurs between resistant and susceptible adults, and resistance alleles are rare. A high dose of toxin is important, because resistance is functionally recessive only if heterozygotes die. If heterozygotes survive, resistance can evolve much faster. This apparent paradox has been illustrated with many different models (Tabashnik & Croft, 1982; Gould, 1998).

If no refuge is present, resistance evolves faster as dose increases. The pattern with a refuge is very different because susceptible adults mate with resistant adults and, in effect, dilute resistance. With a refuge, a very low dose delays resistance because few susceptibles are killed. Moderate increases in dose accelerate resistance when a refuge is present. However, as dose increases from a moderate level to a high level, more heterozygotes are killed and resistance is delayed. At a dose high enough to kill >95% of heterozygotes, which renders resistance functionally recessive, resistance is delayed substantially.

Refuges of untransformed host plants to enable survival of susceptible pests can be arranged in various ways. With external refuges, relatively large blocks of refuge plants are grown separately from blocks of transgenic plants. With row refuges, rows of refuge plants are grown within fields of transgenics. With a seed mixture refuge, non-transgenic seed is randomly mixed with transgenic seed before distribution so that refuge plants are randomly interspersed with transgenics within rows.

Optimal refuge placement depends on biological factors and practical factors, such as acceptability to growers. In terms of biology, refuges must be close enough to transgenics to promote extensive mating between resistant and susceptible adults. Seed mixtures can be implemented before seed distribution and thus compliance can be ensured. For other spatial arrangements, grower cooperation with the refuge strategy is required. Given the lack of a strong theoretical or empirical basis favoring any particular spatial arrangement, practical considerations such as compliance are paramount.

5 Pink bollworm resistance to Bt cotton in Arizona

The risk of resistance is high in the pink bollworm (Pectinophora gossypiella). It is a global pest of cotton and one of the most destructive cotton pests in the southwestern United States (Henneberry & Naranjo 1998; Ingram, 1994). In Arizona, so far, Bt cotton that produces Cry1Ac toxin has been extremely effective against pink bollworm (Patin et al., 1999; Tabashnik et al., 2000b), helping to greatly reduce insecticide use (Carrière et al., 2001a). However, selection for resistance is intense. For the past several years, Bt cotton has accounted for more than half of the cotton grown in Arizona (Carrière et al., 2001a). Alternative hosts are not common and the pink bollworm completes up to five generations per year in Arizona (Henneberry & Naranjo, 1998). Laboratory selection has readily produced resistance to Cry1Ac and enhanced survival of larvae on Bt cotton (Liu et al., 1999, 2001a,b; Patin et al., 1999; Tabashnik et al., 2000a,b, 2002a,b ).

A diverse working group that included scientists, representatives of industry, and cotton farmers made recommendations for non-Bt cotton refuges in Arizona. The option favored by university scientists is an internal refuge of at least 10% non-Bt cotton rows within fields. Scientists wanted growers to plant at least 10% non-Bt cotton refuges, but an option of a 5% unsprayed external refuge was included at the cotton farmers’ insistence. The external refuges must be less than one mile from Bt cotton. Current regulations in effect for Arizona are similar to the recommendations of the working group.

To test the key assumptions of the refuge strategy for pink bollworm and Bt cotton, a team of scientists at the University of Arizona is collaborating with scientists from the Arizona Cotton Research and Protection Council and the US Department of Agriculture. As outlined above, the key assumptions are: recessive inheritance, extensive mating between resistant and susceptible adults, and low frequency of resistance.

Greenhouse bioassays with larvae from resistant strains (RR), susceptible strains (SS), and their hybrid F1 progeny (putative RS) show that inheritance of resistance to Bt cotton is recessive (Liu et al., 1999, 2001a). Confirming this pattern, inheritance of resistance to high concentrations of Cry1Ac in artificial diet is also recessive (Tabashnik et al., 2000a; Liu et al., 2001a). These results support one of the key assumptions of the refuge strategy.

The extent of mating between resistant and susceptible adults in the field cannot be readily determined. Although adults may occasionally move long distances (>10km), results from field experiments suggest that adults typically move less than 2km (Tabashnik et al., 1999; Carrière et al., 2001a). Thus, we do not have a definitive test of the assumption about mating, but we know that as the distance between refuges and Bt cotton increases, the probability of mating between susceptible and resistant adults decreases.

To estimate the frequency of resistance to Bt cotton in field populations of the pink bollworm, we used two independent methods: a direct approach and an indirect approach. The direct approach was based on laboratory bioassays of susceptibility to Cry1Ac of field-derived strains of pink bollworm. The indirect approach was based on the relative abundance of live pink bollworm in field-collected bolls of Bt cotton and non-Bt cotton. Results from both approaches show that in Arizona field populations of pink bollworm, the frequency of resistance to Bt cotton was surprisingly high in 1997. Yet, both approaches also indicate that the frequency of resistance did not increase as expected in 1998 or 1999 (Tabashnik et al., 2000a).

Laboratory bioassays of individual neonates from strains derived from Arizona cotton fields in 1997 showed that 5 of 10 cotton fields surveyed contained at least one resistant (putative RR) individual. Larvae were considered resistant if they survived at a diagnostic concentration of Bt toxin in artificial diet (10 micrograms Cry1Ac per ml diet) that killed essentially all SS and hybrid (RS) larvae. Overall, 16 of 500 (3.2%) larvae survived the diagnostic concentration, which led to an estimated mean resistance allele frequency of 0.16 (95% confidence interval 0.05–0.26). This estimate is roughly 100-fold higher than previous estimates (e.g., Gould et al., 1997) and indicates that resistance was not rare in some pink bollworm populations in 1997.

Because these results refute one of the key assumptions of refuge strategy and use of Bt cotton remained high, we expected rapid increases in resistance in 1998 and 1999. The monitoring data, however, show that such increases did not occur.

In laboratory bioassays using precisely the same methods as in 1997, survival at the diagnostic concentration of Cry1Ac was only 1 of 1,100 larvae tested from strains derived from Arizona cotton fields in 1998 (estimated R allele frequency=0.007) and 0 of 1,179 in 1999. To expand the sample size, 5,549 additional neonates were tested in groups of up to 200 from the strains derived from the field in 1999. None survived.

To verify this result, we independently derived indirect estimates of resistance allele frequency based on the relative abundance of live pink bollworm larvae in bolls from paired Bt cotton and non-Bt cotton fields in Arizona. The indirect estimates also show that the frequency of resistance did not increase from 1997 to 1999. Unlike the direct estimates, however, the indirect estimates do not indicate a large decrease in resistance allele frequency from 1997 (0.13) to 1998 (0.05) or 1999 (0.11). As the indirect estimates require more assumptions, we have more confidence in the direct estimates.

Because the results contradicted expectations, we re-examined the relevant theory to determine what conditions, if any, could explain the observed patterns. Modeling efforts revealed several factors that, in conjunction with basic assumptions of the refuge strategy, arrest or reverse evolution of resistance: non-recessive fitness costs, incomplete resistance, and density-independent population growth in refuges (Carrière & Tabashnik, 2001).

With large refuges and recessive resistance, non-recessive fitness costs can greatly delay or reverse resistance. Fitness costs occur when resistant individuals are less fit than susceptible homozygotes on non-Bt cotton plants. If fitness costs are not recessive, RS individuals have a disadvantage relative to SS individuals on non-Bt cotton. If resistance is recessive, RS individuals have no advantage over SS individuals on Bt cotton. If RR individuals are rare (R alleles occur almost exclusively in RS individuals), a net disadvantage associated with the R allele can reverse evolution of resistance.

Experiments show major fitness costs associated with resistance to Bt cotton in pink bollworm, but a non-recessive cost was evident only in one case (Carrière et al., 2001b,c). Compared with their performance on non-Bt cotton, our laboratory-selected resistant strains of pink bollworm suffer disadvantages on Bt cotton including slower development, lower pupal weight, and lower fecundity (Liu et al., 1999, 2001b). We call this incomplete resistance because the resistance does not completely overcome the negative effects of the Bt toxin in Bt cotton. Modeling results suggest that with refuges, the observed fitness costs and incomplete resistance could stop or even reverse evolution of resistance to Bt cotton in pink bollworm (Carrière & Tabashnik, 2001).

6 Conclusions

In summary, the goal of resistance management is to delay evolution of pest resistance to transgenic crops. Refuges of non-transgenic hosts that enable survival of susceptible pests may help to achieve this goal. Although most attention has focused on spatial refuges, periods of time during which transgenics are not grown may also be useful. To delay resistance, larger refuges are better. Inevitably a trade-off occurs between increasing refuge size and decreasing short-term control of the pest population.

Resistance management can extend the usefulness of transgenic crops, but eventually insects will evolve resistance. Thus, any particular transgenic cultivar should not be considered a permanent solution to a pest problem. Instead, transgenic crops provide us with tools to be used in harmony with other tactics as part of integrated pest management.

Evaluations of transgenic crops should be based on their advantages and disadvantages compared with alternatives. If transgenics can greatly reduce use of hazardous insecticides, as achieved in Arizona cotton, great benefits may be realized.

We are grateful to our colleagues at the Extension Arthropod Research and Management Laboratory and the Arizona Cotton Research and Protection Council (ACRPC) for their contributions to this work, with special thanks to Larry Antilla. This work was supported by grants from the USDA NRI and IFAFS programs, the University of Arizona, and the ACRPC.