Frequency of Resistance to Bacillus thuringiensis Toxin ... - BioOne

INSECTICIDE RESISTANCE AND RESISTANCE MANAGEMENT

Frequency of Resistance to Bacillus thuringiensis Toxin Cry1Ab in Southern United States Corn Belt Population of European Corn Borer (Lepidoptera: Crambidae) T. J. STODOLA,1 D. A. ANDOW,1 A. R. HYDEN,1 J. L. HINTON,1 J. J. ROARK,1, L. L. BUSCHMAN,2 P. PORTER,3 AND G. B. CRONHOLM3

J. Econ. Entomol. 99(2): 502Ð507 (2006)

ABSTRACT The high-dose refuge resistance management strategy is the main approach used to delay resistance in targeted pests to Bacillus thuringiensis (Bt) toxins in transgenic crops. We used an F2 screen to test a critical assumption of the high-dose refuge strategy, which is that resistance allele (R) frequencies are initially rare (⬍10⫺3) in Ostrinia nubilalis (Hu¨ bner) (Lepidoptera: Crambidae) from the southern Corn Belt. We expanded the methodological scope of the F2 screen so that both males and females may be used to initiate a screen and determined how the results from both sexes may be combined. In total, 62 female and 131 male O. nubilalis lines from Kansas and 39 female and four male lines from Texas were screened. No major resistance alleles were found and estimated R frequency for the southern Corn Belt was updated to between 0 and 0.0044 with 95% credibility. The experiment-wise detection probability was 98.7%. These results suggest the frequency of resistance alleles is low enough that the high-dose refuge resistance management strategy may be effective for delaying resistance evolution in O. nubilalis to Bt corn in the southern Corn Belt. KEY WORDS resistance management, Ostrinia nubilalis, transgenic corn, F2 screen

The F2 screen is an economically efÞcient way to screen populations for rare recessive alleles (R) (Andow and Ives 2002), and it is a valuable tool for examining wild populations for resistance to high-dose genetically modiÞed crops. It has been used in the northern Corn Belt (Andow et al. 1998, 2000; Bourguet et al. 2003), the Philippines (Bentur et al. 2000), France (Bourguet et al. 2003, Ge´ nissel et al. 2003), and Australia (Akhurst et al. 2003). Major Bt resistance alleles in Chrysomela tremulae F. to Bt poplar (Ge´ nissel et al. 2003) and in Helicoverpa armigera (Hu¨ bner) to Bt cotton (Akhurst et al. 2003) have been found using the F2 screen. In the northern United States (Iowa and Minnesota) and France, resistance allele frequencies in European corn borer, Ostrinia nubilalis (Hu¨ bner) (Lepidoptera: Crambidae), to Bt corn have been found to be sufÞciently low to justify the use of the high-dose plus refuge strategy (Bourguet et al. 2003). We now use the F2 screen to improve estimated resistance allele frequencies in southwestern Kansas and northern Texas populations of European corn borer in the southern Corn Belt by adding to our screen in Bourguet et al. (2003). This region is separated geograph1 Department of Entomology, University of Minnesota, St. Paul, MN 55108. 2 Department of Entomology, Kansas State University, Garden City, KS 67846. 3 Department of Entomology, Texas A&M University, Lubbock, TX 79403.

ically from the main Corn Belt by wheat and rangeland, relies largely on irrigation, and uses high levels of insecticides on corn. For these reasons, resistance risk may be uniquely high to this region. To improve the estimated frequency, we address two signiÞcant issues. In the southern Corn Belt, because insecticide use is high, it can be difÞcult to Þnd any O. nubilalis individuals, and it has been difÞcult to collect mated O. nubilalis adult females, the preferred stage for initiating an F2 screen (Bourguet et al. 2003). In addition, as pointed out by Zhao et al. (2002), we need to take more care in controlling experimentally the possibility for false negatives by developing a more sensitive screening method. It is desirable to start an F2 screen with mated females because four genetic haplotypes are screened. However, there are times when it is easier to collect males, virgin females, or a nonadult life stage, in which case only two genetic haplotypes can be screened, a loss of 50% efÞciency. Although one can start an F2 screen with insects at any of these stages, there is no formal methodology to perform this type of screen or for combining the results from screens started from different life stages. Here, we derive the necessary statistics to do so. Because the F2 screen is implemented when alleles are expected to be rare, one would expect most collected lines not to have any resistance alleles. Thus, it is critically important to minimize the likelihood of false negatives from any source, especially considering

0022-0493/06/0502Ð0507$04.00/0 䉷 2006 Entomological Society of America

April 2006

STODOLA ET AL.: F2 SCREEN FOR SOUTHERN U.S. O. nubilalis

that resistance in O. nubilalis has yet to be recovered from Bt corn (Siegfried et al. 1995; Andow et al. 1998, 2000; Març on et al. 1999, 2000; Chaufaux et al. 2001; Bourguet et al. 2003; Tabashnik et al. 2003). There are two important ways to control false negatives in an F2 screen: having sufÞciently large F1 populations and screening enough F2 individuals (Andow and Alstad 1998, Stodola and Andow 2004), and using a screening method that is sensitive enough to discriminate between lines with and without resistance. The results of Zhao et al. (2002) suggest that a high-dose Bt plant may be too insensitive to discriminate accurately lines with and without resistance. Thus, we use a diet-based screening protocol that allows greater survival of partially resistant O. nubilalis. Materials and Methods Moth Collections. O. nubilalis adults were collected near Garden City, KS, and near Edmonson, TX. Blacklight bucket traps were used to collect female O. nubilalis during 2000 and 2001 (Bourguet et al. 2003), but it was difÞcult to collect and identify O. nubilalis in large numbers because they were less abundant relative to a morphologically similar species, the southwestern corn borer, Diatraea grandiosella Dyer (Lepidoptera: Crambidae). Hand collecting male and female O. nubilalis by using blacklights was conducted at the Texas sites during 2002 and 2003. Pheromone traps also were used to collect male O. nubilalis in both states during 2002 and 2003. In Kansas, this resulted in much higher male moth capture rates. All O. nubilalis were shipped overnight to Minnesota for screening by using plastic soda bottles with some mesh inside. It should be noted that difÞculties in collecting individuals for monitoring is not unique to the F2 screen. Collecting costs per individual will be similar for all methods by using live insects (Andow and Ives 2002). F2 Screen. The traditional protocol for an F2 screen, from Andow and Alstad (1998), called for collecting wild mated females (P1), rearing these offspring in isofamily groups (F1), inbreeding these F1 families in mass mating cages, and screening the F2 generation for phenotype of the trait. This will screen the largest number of haplotypes per line, but in many systems it may not be practical, as it was in this case. The few Þeld-collected female lines were screened by this protocol. Male lines were started by enclosing the Þeld-caught male with one to three virgin females from a known susceptible laboratory colony. The rest of the protocol was the same for both types of lines. A mean of 8.7 ⫾ 7.3 egg masses were collected per female line, and 17.9 ⫾ 14.1 egg masses were collected per male line. A target of 150 eggs per line (actual number of eggs was 177 ⫾ 87 per line) were reared on meridic diet (Andow and Stodola 2001), and the surviving pupae (79.1 ⫾ 14.1) of each line were placed in a mass mating cage, where they were allowed to eclose and sib-mate at random. F2 egg masses from these cages were collected daily. These F2 eggs (2,910 ⫾ 2,140 eggs per line) were screened either on Bt corn (2000 and 2001) or on artiÞcial diet (2002 and 2003).

503

Variation in productivity among lines affects the method though the type II error rate for a line (PNo, Stodola and Andow 2004), which we have estimated. Bt corn (Pioneer 36F30, expressing the Mon 810 transformation event) plants were hand-planted both outside (when weather permitted) and in a greenhouse (when it did not). Only plants between the V4 and VT stages were used. Plants were a minimum of 1 m apart outside, to minimize the risk of larvae moving between plants within a row, with one isofemale line infested per plant. In the greenhouse, 5-gallon round pots were planted with four corn plants, with each plant being as far apart as the pot would allow, ⬇25 cm. Pots were separated by ⬇1 m. Diet screens were performed using meridic diet with 3.0 ␮g of Cry1Ab protoxin per gram of diet incorporated. This diet was poured as a thin Þlm (⬇2 mm) on the bottom of an 18-cm-diameter hard plastic dish. The thin Þlm diet allowed easier detection of surviving larvae and evidence of larval feeding. The dish was closed with a solid top over a paper towel to help seal the edge and minimize moisture loss from the diet. Plant and diet screens were evaluated after 10 d. Damage to corn plants was measured using the Bolin Scale (Bourguet et al. 2003). A fully positive screen required many larvae feeding and growing to at least the third instar on the screening media. To be considered partially positive, a corn screen required a rating ⱖ2 on the Bolin Scale, equivalent to a ⱖ2 shotholes on ⱖ2 leaves or Þnding live larvae larger than a neonate on the plant. A positive diet screen required evidence of feeding, tunneling, and larval growth to at least the second instar. Larval survival to maturity was not essential for a positive screen for either method. Any lines testing positive during the F2 generation were rescreened during the F3 generation. Rescreening. The F3 generation was reared for lines that either tested positive during the Þrst screen or when the F1 populations were too small to be able to sufÞciently screen a line (predicted or actual PNo ⬎ 0.05). Nonscreened F2 eggs were used to start the F3 generation, and these larvae were reared and screened as described above. Lines that met the criteria for a positive or partially positive screen for two generations were considered positive lines. Mating Frequency. When possible P1 females were dissected to determine the number of spermatophores in the bursa copulatrix (Hinton and Andow 2003). Statistical Analysis. There are two main statistical calculations necessary when doing an F2 screen: the expected frequency of resistance alleles in the wild population (E[pR]) with credibility intervals and the probability of missing a resistance allele that was present in a line (PNo). Methods for calculating PNo are given in Stodola and Andow (2004). To illustrate the robustness of the F2 screen, we derive here a way to combine results of both male and female P1 lines from data over multiple years. Combining these data starts by calculating the frequency of resistant families, E[P], and converting to alleles, E[pR], because a male line is not as informative as a

504 Table 1.

Females lines 2000 2001 2002 2003 Male lines 2002 2003

a b

JOURNAL OF ECONOMIC ENTOMOLOGY

Vol. 99, no. 2

Number of lines progressing through the F2 screen experiment Location

P1 lines started

P1 mateda

F1 larvae produced

F1 adults produced

Kansas Texas Kansas Texas Texas Texas

133 61 69 80 18 9

113 50 52 67 3 5

65 20 6 18 2 5

65 20 6 18 2 5

Kansas Texas Kansas Texas

109 36 511 4

10 5 180 1

6 4 149 0

6 4 149 0

Screened 56b 18b 6b 14b 2 5 6 4 125 0

Line was considered mated if a female laid fertile eggs. These data were reported in Bourguet et al. (2003).

female line. P is the phenotypic probability that a family has resistance, which is directly measured in the F2 screen. In a male line two genetic haplotypes are being screened instead of four in a female line. The following illustrates both this new calculation, and how multiple yearsÕ worth of data may be combined. Suppose there were three different screens conducted: the Þrst screen with Nf1 mated female lines with S1 positive lines, the second screen with Nm2 male lines with S2 positive lines, and the third with both male and female, Nf3 and Nm3, lines screened with Sf3 and Sm3 positive lines. Using the method described in Andow et al. (2000) with a Beta prior, we can consider the mated female lines separately from the male lines, and with Nf ⫽ Nf1 ⫹ Nf3, and Nm ⫽ Nm2 ⫹ Nm3, we note that for the female lines, recurrent use of the Beta prior gives E关Pf 兴 ⫽

S1 ⫹ Sf3 ⫹ 1 Nf1 ⫹ Nf3 ⫹ 2

[1]

and for the male lines, recurrent use of the Beta prior gives E关Pm 兴 ⫽

S2 ⫹ Sm3 ⫹ 1 . Nm2 ⫹ Nm3 ⫹ 2

[2]

This shows that we can take any combination of mixtures of female and male lines and treat them as if they were a single trial of female lines and another trial of male lines. Suppose there are Nf mated female lines screened and Nm male (including virgin female) lines screened. This means that there are N ⫽ Nf ⫹ Nm lines screened. Let P be the estimated probability that a line has a resistance allele. Let S be the number of positive lines among all of the screened lines. Then, E关P兴 ⫽

S⫹1 , N⫹2

[3]

which is typical for a binomial sampling problem. Similarly, Beta[S⫹1, N-S⫹1] is the distribution from which the 95% credibility intervals can be calculated. The probability that any allele is a resistance allele is the same, whether it is a mated female line or a male line.

Following the methods in Andow and Alstad (1999), which rely on the theoretical relation between P and pR, we note that for mated females, 1 ⫺ P ⫽ (1 ⫺ pR)4, and for males or unmated females, 1 ⫺ P ⫽ (1 ⫺ pR)2. Hence, P⫽

Nf共1 ⫺ 共1 ⫺ pR兲4兲 ⫹ Nm共1 ⫺ 共1 ⫺ pR兲2兲 . Nf ⫹ Nm

[4]

We solve equation 4 for pR, and Þnd that the appropriate root is pR ⫽

冑冑

冑

2Nf ⫺ 2 ⫺N fN m ⫹ N f 共2N f ⫹ N m兲 2 ⫺ 4N fP共N f ⫹ N m兲 2N f

.

关5兴

When P is small, equation 5 simpliÞes to pR ⬵

Nf ⫹ Nm P, 2共2Nf ⫹ Nm 兲

[6]

which is the Þrst order Taylor approximation to equation 5. Equation 6 can also be used to transform the 95% credibility interval of P, associated with equation 1 to the pR scale. Results and Discussion In total, 202 female and 620 male O. nubilalis from Kansas and 168 female and 40 male O. nubilalis from Texas were started as F2 screen lines during 2000 Ð 2003. Table 1 lists a detailed breakdown of how many lines were started each year and the number of lines that progressed through a deÞned stage in the experiment. Of those started, only 30.7% of the female lines and 21.1% of the male lines from Kansas were successfully screened, and 23.2% of the female lines and 10.0% of the male lines from Texas were successfully screened. Most lines were lost during the P1 generation (Table 1). Whereas 78.4% of Þeld-collected females had mated, only 31.4% produced F1 larvae in the laboratory. Only 29.7% of the Þeld-collected males mated with at least one of the laboratory-reared females provided (typically, two to three females were provided per male, with a maximum of nine). We suspect

April 2006


505

Table 2. Proportion of male lines successfully screened as a function of the number of mated P1 females per line or P1 females provided per line (total number of male lines evaluated in parentheses) Yr

No. females/male line

Source

Mated females/male 2003 Females provided/male 2002 2003

Kansas

0 females

1 female

2 females

0.014 (214)

0.351 (57)

0.805 (41)

0.600 (5)

0.000 (0)

0.000 (0) 0.000 (0) 0.119 (67) 0.000 (0)

0.000 (0) 0.000 (0) 0.172 (116) 0.000 (0)

0.067 (90) 0.107 (28) 0.286 (298) 0.000 (4)

0.000 (19) 0.125 (8) 0.324 (68) 0.000 (0)

Kansas Texas Kansas Texas

3 females

4⫹ females

Results are based on female dissections and include only those lines where all known females caged with male were dissected.

that most losses were caused by stresses the adults were subjected to in traps before collection. Most males captured in pheromone traps died in the trap before being removed, possibly by the early morning heat, and those not killed could have been adversely affected by the trap conditions. Shipping mortality was minimized by using emptied plastic soda bottles with a strip of Þne mesh cloth. When F1 larvae were established in any line, 85.8% of the time that line was successfully screened. Although only 40.0% of the mated P1 Þeld-collected females produced F1 larvae in the laboratory, 81.1% of laboratory females that mated with a Þeld-collected male produced viable F1 larvae (Table 1). This indicates that eggs of Þeld-collected females have lower viability in the laboratory than those of laboratoryadapted females. It is possible that the laboratory environment (photoperiod of 16:8 [L:D] h at 27 to 18⬚C with 80% RH) is drier than the normal oviposition site in the Þeld. In addition, the results indicate that loss of male lines was largely because the Þeld-collected males did not mate with the laboratory females. Analysis of a subsample of the Kansas males in 2003 showed that many of the males did not mate with a female (Table 2, mated females per male) and that successful screening occurred 80.5% of the time when the male mated with two laboratory females (Table 2, mated females per male). This generally required that we provide three or more laboratory females per Þeld-

collected male (Table 2, females provided per male). Because laboratory females mate readily with laboratory males, these results suggest that Þeld-collected males may have had a low mating propensity and were sperm limited. False negative lines were minimized by screening sufÞcient numbers of F2 larvae and modifying the screening procedure. The number of lines testing positive after rescreening is shown in Table 3. Only three lines demonstrated partial resistance. These lines cannot survive on Bt corn, and their signiÞcance for resistance evolution is not clear. The cumulative distribution of PNo, the probability of a false negative, for all of the lines that were screened is plotted in Fig. 1. There was ⬍1% probability of having missed a resistance allele in 212 of the 236 lines. The experiment-wise detection probability was 98.7%, and this compares favorably with previous F2 screens. The likelihood of false negatives has been further reduced by switching from high-dose corn plants to a lower dose diet screen. This increases the likelihood that we have detected both completely resistance and partially resistant lines should they have occurred. We pooled the results from all years and all collecting sites to estimate the frequency of resistance in the southern Corn Belt. Although the genetic data to prove that these populations are samples of the same deme are not published, we note that the sample sites are only ⬇400 km apart, which is half the distance

Table 3. Results from F2 screen and subsequent rescreening of positive lines Yr 2000 2001 2002 2003

Screen generation

Screen method

Partial positives

Full positives

F2 F3 F2 F3 F2 F3 F2 F3

Br corn Bt corn Bt corn Bt corn Diet Diet Diet Diet

9a 1a 1a 1a 1 1 8 0

0a 0a 0a 0a 0 0 0 0

A full positive result required many larvae feeding and growing to at least the third instar on the screening media. A partial positive required a rating ⱖ2 on the Bolin Scale (corn), evidence of feeding and tunneling (diet), or larvae surviving to the second instar on either media. a These data were summarized in Bourguet et al. 2003.

Fig. 1. The cumulative probability density function (PDF) of detecting a resistance allele had one been present in a line, e.g., 1 ⫺ probability of a false negative (PNo).

506

JOURNAL OF ECONOMIC ENTOMOLOGY

Vol. 99, no. 2

Table 4. Expected frequencies of major and minor resistance alleles in the southern Corn Belt, with experiment-wise probability of a false negative Location

Female lines

Male lines

E[pR] (95% CL)

E[ppartial] (95% CL)

Experiment-wise pNo

Kansas Texas Southern Corn Belt

62 39 101

131 4 135

0.0019 (0Ð0.0058) 0.0058 (0Ð0.0173) 0.0015 (0Ð0.0044)

0.0078 (0.0021Ð0.0168) 0.0117 (0.0015Ð0.0315) 0.0074 (0.0024Ð0.014)

0.011 0.021 0.013

between panmictic populations in France (Bourguet et al. 2000) and only 80 km farther than the distance between populations in the northern Corn Belt that we have already pooled (Bourguet et al. 2003). In addition, it is reasonable to pool the data across years because in all years, R allele frequencies have been lower than our sample sensitivity, and theory predicts that R allele frequency will change very slowly during the initial stages of evolution. The frequency of resistance alleles (R) and conÞdence intervals for the Kansas and Texas O. nubilalis populations were low (Table 4). The estimated R frequency for the southern Corn Belt is between 0 and 0.0044 with 95% credibility. This is within the range necessary to fulÞll one of the assumptions of the highdose plus refuge strategy (Andow 2001), which is that the R allele frequency should be ⬍0.001 for a recessive R allele (Roush and Miller 1986). Hence, we conclude that the high-dose refuge resistance management strategy may be effective for Bt corn against O. nubilalis in the southern Corn Belt, and we should not be surprised that evidence for resistance has yet to be found for this species (Tabashnik et al. 2003). Acknowledgments We thank Lane Cavanaugh, Courtney Cook, Vanessa Carlson, Elise Rosenberg and Kathryn Smith for assistance in carrying out the screen and J. White, C. Zwahlen, Y. Hu, M. Savanick, and M. Ng for comments on the work in progress. This work was funded by the U.S. Environmental Protection Agency, Center for Environmental Assessment.

References Cited Akhurst, R. J., W. James, L. J. Bird, and C. Beard. 2003. Resistance to the Cry1Ac delta-endotoxin of Bacillus thuringiensis in the cotton bollworm, Helicoverpa armigera (Lepidoptera: Noctuidae). J. Econ. Entomol. 96: 1290Ð 1299. Andow, D. A. 2001. Resisting resistance to Bt corn, pp. 99Ð 124. In D. K. Letourneau and B. E. Burrows [eds.], Genetically engineered organisms: assessing environmental and human health effects. CRC, Boca Raton, FL. Andow, D. A., and D. N. Alstad. 1998. F2 Screen for rare resistance alleles. J. Econ. Entomol. 91: 572Ð578. Andow, D. A., and A. R. Ives. 2002. Monitoring and adaptive resistance management. Ecol. Appl. 12: 1378 Ð1390. Andow, D. A., and T. J. Stodola. 2001. Detecting subtle effects of diet preservatives on European corn borer. J. Agric. Entomol. 36: 285Ð296. Andow, D. A., D. N. Alstad, Y.-H. Pang, P. C. Bolin, and W. D. Hutchison. 1998. Using an F2 screen to search for resistance alleles to Bacillus thuringiensis toxin in European

corn borer (Lepidoptera: Crambidae). J. Econ. Entomol. 91: 579 Ð584. Andow, D. A., D. M. Olson, R. L. Hellmich, D. N. Alstad, and W. D. Hutchison. 2000. Frequency of resistance to Bacillus thuringiensis toxin Cry1Ab in an Iowa population of European corn borer (Lepidoptera: Crambidae). J. Econ. Entomol. 93: 26 Ð30. Bentur, J. S., D. A. Andow, M. B. Cohen, A. M. Romena, and F. Gould. 2000. Frequency of alleles conferring resistance to a Bacillus thuringiensis toxin in a Philippine population of Scirpophaga incertulas (Lepidoptera: Pyralidae). J. Econ. Entomol. 93: 1515Ð1521. Bourguet, D., M. The´re`se Bethenod, N. Pasteur, and F. Viard. 2000. Gene ßow in the European corn borer Ostrinia nubilalis: implications for the sustainability of transgenic insecticidal maize. Proc. R. Soc. Lond. Ser. B: Biol. Sci. 267: 117Ð122. Bourguet, D., J. Chaufaux, M. Seguin, C. Buisson, J. L. Hinton, T. J. Stodola, P. Porter, G. Cronholm, L. L. Buschman, and D. A. Andow. 2003. Frequency of alleles conferring resistance to Bt maize in French and US corn belt populations of Ostrinia nubilalis. Theor. Appl. Genet. 106: 1225Ð1233. Chaufaux, J., M. Seguin, J. J. Swanson, D. Bourguet, and B. D. Siegfried. 2001. Cronic exposure of the European corn borer (Lepidoptera: Crambidae) to Cry1Ab Bacillus thuringiensis toxin. J. Econ. Entomol. 94: 1564 Ð1570. Geńissel, A., S. Augustin, C. Courtin, G. Pilate, P. Lorme, and D. Bourguet. 2003. Initial frequency of alleles conferring resistance to Bacillus thuringiensis poplar in a Þeld population of Chrysomela tremulae. Proc. R. Soc. Biol. Sci. Ser. B 270: 791Ð797. Hinton, J. L., and D. A. Andow. 2003. Mating frequency of European corn borer (Lepidoptera: Crambidae) in Minnesota, Kansas, and Texas. Great Lakes Entomol. 36: 156 Ð 159. Març on, P.C.R.G., L. J. Young, K. L. Steffey, and B. D. Siegfried. 1999. Baseline susceptibility of European corn borer (Lepidoptera: Crambidae) to Bacillus thuringiensis toxins. J. Econ. Entomol. 92: 279 Ð285. Març on, P.C.R.G., B. D. Siegfried, T. Spencer, and W. D. Hutchison. 2000. Development of diagnostic concentrations for monitoring Bacillus thuringiensis resistance in European corn borer (Lepidoptera: Crambidae). J. Econ. Entomol. 93: 925Ð930. Roush, R. T., and G. L. Miller. 1986. Considerations for designs of insecticide resistance monitoring programs. J. Econ. Entomol. 79: 293Ð298. Siegfried, B. D., P.C.R.G. Març on, J. F. Witkowski, R. J. Wright, and G. W. Warren. 1995. Susceptibility of Þeld populations of the European corn borer, Ostrinia nubilalis (Hu¨ bner) (Lepidoptera: Pyralidae), to Bacillus thuringiensis (Berliner). J. Agric. Entomol. 12: 267Ð273. Stodola, T. J., and D. A. Andow. 2004. F2 screen variations and associated statistics. J. Econ. Entomol. 97: 1756 Ð1764. Tabashnik, B. E., Y. Carrie`re, T. J. Dennehy, S. Morin, M. S. Sisterson, R. T. Roush, A. Shelton, and J.-Z. Zhao. 2003.

April 2006


Insect resistance to transgenic Bt crops: lessons from the laboratory and Þeld. J. Econ. Entomol. 96: 1031Ð1038. Zhao, J.-Z., Y. X. Li, H. L. Collins, and A. M. Shelton. 2002. Examination of the F2 screen for rare resistance alleles to

507

Bacillus thuringiensis toxins in the diamondback moth (Lepidoperta: Plutellidae). J. Econ. Entomol. 95: 14 Ð21. Received 26 August 2004; accepted 8 November 2005.