Lepidoptera: Noctuidae

POPULATION ECOLOGY

Temporal and Spatial Patterns of Allelic Frequencies in Cotton Bollworm (Lepidoptera: Noctuidae) QIFA HAN1

AND

M. A. CAPRIO2

Department of Entomology and Plant Pathology, Mississippi State University, Mississippi State, MS 39762

Environ. Entomol. 31(3): 462Ð468 (2002)

ABSTRACT Temporal variation in allozyme frequencies at 10 allozyme loci was assessed by sampling six local populations of the cotton bollworm, Helicoverpa zea (Boddie), near cotton Þelds over 3 yr (1995Ð1997) in Leßore County, MS. Population structure over local and regional areas and changes in population structure over generations were analyzed using F-statistic estimators. There was low population differentiation at both local and regional scales, suggesting that extensive gene ßow occurred within the spatial scale under investigation. Allele frequencies, heterozygosity, and the mean number of alleles per locus were stable over time and space. Measurements of population differentiation, FST, ranged from 0.0002 to 0.0072 over generations. However, there was no signiÞcant population subdivision when evaluated by bootstrapping over loci. Moreover, population differentiation as a whole and differentiation between regions was also very low, indicating that extensive gene ßow occurs at both local and regional scales. Egg populations were more differentiated relative to the corresponding male moth populations, suggesting that population differentiation is greatest at egg/ larval stages decreasing in adult populations due to movement. Almost all observed genetic variance was accounted for among traps within generations. Little variance was observed among generations or among years. These results further indicate that allele frequencies were stable over the duration of this study. KEY WORDS Helicoverpa zea, genetic structure, allozyme, population genetics, gene ßow

COTTON BOLLWORM, Helicoverpa zea (Boddie), and tobacco budworm, Heliothis virescens (F.), are the most economically important insect pests in cotton in the midsouth of the United States and account for more chemical control applications than any other pest complex (Luttrell 1994). Consequently, high selection pressure from frequent insecticide application has resulted in the evolution of resistance to many insecticides, especially in H. virescens. As a component of integrated pest management, a resistance management program was implemented to manage resistance to pyrethroids in the southeastern United States limiting the use of pyrethroids to the early season (Graves et al. 1989). The reduced selection pressure and the effect of mixing exposed and unexposed populations was expected to maintain resistant allele frequencies at sufÞciently low levels to allow satisfactory control. This practice probably contributed to the persistence of pyrethroids as a viable option for control of cotton bollworm. However, the severity of the resistance problem in H. virescens and subsequent ecological and environmental consequences have encouraged re1 Current address: Department of Entomology, Kansas State University, 123 Waters Hall, Manhattan, KS 66592 (e-mail: qhan@ ksu.edu). 2 E-mail: [email protected].

search entomologists to look for new strategies for cotton insect control. One of the new technologies to emerge was transgenic cotton into which an insecticidal crystal protein gene from Bacillus thuringiensis (Bt) was incorporated. Transgenic cotton was Þrst commercialized in the United States during 1996. Although transgenic cotton increases the potential for resistance management, it is not immune from the evolution of resistance (Caprio 1994, Gould 1994, Roush 1994). Development of resistance to transgenic cotton remains a primary concern with the bollwormtobacco budworm complex. In particular, because the high dose strategy of transgenic cotton was primarily designed for control of H. virescens, the lower susceptibility of H. zea to the toxin may promote resistance in this species. Principles of population genetics form the basis for the theory of resistance management. Although considerable progress has been made in the general biology and ecology of H. virescens and H. zea (Fitt 1989 and references therein), there is limited information on many aspects of population genetics (Sell et al. 1975, Sluss et al. 1978, Sluss and Graham 1979, Korman et al. 1993, Mallet et al. 1993). Fitt (1989) recognized that high mobility was one of four key characteristics responsible for the pest status of H. virescens and H. zea. Movement was also a key parameter when mod-

0046-225X/02/0462Ð0468$02.00/0 䉷 2002 Entomological Society of America

June 2002

HAN AND CAPRIO: ALLELE FREQUENCIES IN H. zea

eling the evolution of resistance (Caprio and Tabashnik 1992, Caprio 1998). InterÞeld movement can either spread resistance rapidly among Þelds or retard resistance development at the global level (Comins 1977, Caprio and Tabashnik 1992, Peck et al. 1999). These model predictions are supported by the observation that H. armigera and H. virescens evolved resistance to insecticides, whereas H. punctigera and H. zea remained susceptible. It was hypothesized that both H. punctigera and H. zea maintained a large proportion of their populations on uncropped areas (Daly and Gregg 1985, Korman et al. 1993). Lopez et al. (1995) suggested that the spatial distribution of male populations between cropped and uncropped areas was different for H. virescens and H. zea. Using a pheromone trapping system, they found that the cropped/uncropped ratios of the mean number of males per trap per week during the cotton growing season were lower in H. zea than H. virescens. H. virescens was apparently more concentrated than H. zea in the cropped areas, and the highest ratios were all found in the midseason for both species (Lopez et al. 1995). We have found that the seasonal changes in population structure were accompanied by a concomitant shift in genetic structure in H. virescens(Han 1999). In this case, the midseason generation of tobacco budworm populations was more highly differentiated than other generations, suggesting reduced gene ßow during the midseason. Little genetic subdivision has been found among geographic populations of the bollworm (Sell et al. 1975, Mallet et al. 1993). Allele frequencies at the Est-II locus were temporally stable in most populations examined (Sell et al. 1975). A founder event was hypothesized to account for the reduction of heterozygosity in H. zea compared with its Old World counterpart H. armigera (Mallet et al. 1993). Studies of temporal genetic variation using allozyme and RAPD markers in H. virescens demonstrated a nonrandom pattern in changes of genetic structure over generations within a season (Han 1999). The relative uniform distribution between cropped and uncropped areas in H. zea populations (Lopez et al. 1995) suggests that genetic structure should not vary at a local scale. Allelic frequencies for H. zea collected in traps within 10 km of one another were assessed to characterize genetic variation at this local scale. The temporal pattern of genetic variation among local populations over generations was examined to determine the seasonality of genetic structure. Given variation in the attributes of different habitat patches and the propensity of adult moths to disperse, we hypothesized that estimates of genetic structure in terms of FST ßuctuate randomly on an ecological time scale. In addition, we surveyed differentiation between populations sampled at the egg stage and contrasted the results with those of male adult populations. Materials and Methods Six local subpopulations were sampled in each of two regions, one in the Mississippi Delta (Leßore

463

County) and one in northeastern Mississippi (Monroe County). Local subpopulations within a 10-km radius in each regional population were sampled with pheromone traps from May to September during 3 yr (1995Ð1997). Pheromone lures were replaced biweekly. Live males caught in the traps were brought to the laboratory on each sampling date (two to three times a week) and stored at ⫺80⬚C until electrophoresis. All captures in the pheromone traps, dead or alive, were counted on each sampling date. Generations were determined by the pattern generated by plotting the average catch per trap per night against time. Each trap was considered to sample a different subpopulation. For the Monroe County population, sufÞcient moths were obtained only in 1997 for genetic analysis. Larval populations tend to be more subdivided than adult populations, with ⬎20-fold higher FST estimates observed in larvae than in males over a decade (Sluss et al. 1978, Korman et al. 1993). This observation was supported by a simulation study (Han 1999). In July of 1996, egg populations were sampled randomly from two well-separated cotton Þelds near two pheromone traps. We assumed the eggs were progeny of Þrstgeneration adults. FST estimates (␪) for adult populations (both G1 and G2) were calculated from the corresponding trap populations so that the population structures were comparable. To eliminate potential allelic variation due to developmental stage, larvae from the eggs were maintained on artiÞcial diet until adult emergence. Male adults from these egg populations were used for genetic analysis. Starch gel electrophoresis (Paster et al. 1988, Mallet et al. 1993) was used to identify individual genotypes at 10 polymorphic allozyme loci (Acon, Ak, Aat, Gpi, Had, Idh-1, Idh-2, 3Pgd, 6Pgd, and Pgm). Allozyme extraction, separation, staining procedures, and genetic information were as described by Mallet et al. (1993). Allelic frequencies were calculated for each trap population. Genotypic proportions were examined for conformity to Hardy-Weinberg expectations within each subpopulation by a contingency chi-square test (Swofford and Selander 1981). Spatial heterogeneity of allele frequencies among trap sites within generations was Þrst examined using G-based exact tests, which detected departures from a random distribution of allele frequencies (Raymond and Rousset 1994, Goudet et al. 1996). Spatial structure was analyzed more formally by means of WrightÕs (Wright 1951) F-statistics. FST, the standardized genetic variance, measures the extent of population differentiation with respect to the spatial distribution of allele frequencies. The overall inbreeding coefÞcient, FIT, is the correlation of alleles within individuals among all trap populations. FIS is the nonrandom component of FIT that describes the correlation of alleles within individuals within a trap population. These three statistics are inter-related by the equation FIT ⫽ FST ⫹ (1 ⫺ FST) FIS. Weir and CockerhamÕs (Weird and Cockerham 1984) f, ␪, and F, unbiased estimators of FIS, FST, and FIT, were calculated using genotypic data. Hierarchical analysis was performed when genetic data were avail-

464

ENVIRONMENTAL ENTOMOLOGY

Vol. 31, no. 3

Seasonal trends were compared with random ßuctuations generated by a resampling technique.

Results

Fig. 1. Male bollworm moths captured by pheromone traps near cotton, Leßore County, MS, 1995Ð1997 captured by pheromone traps

able from both regional populations (Weir 1996, Lewis and Zaykin 1999). Standard deviations of the mean for ␪ were obtained by jackkniÞng over loci, and were considered signiÞcant when the bootstrap-generated 95% CI did not overlap with zero. Temporal patterns in genetic structure were assessed considering population subdivision over time. Total genetic variance, as measured by F-statistics, was partitioned hierarchically among individuals, trap sites, generations, and years. In this way, the source of greatest genetic differentiation was determined (e.g., between generations versus between years). The null hypothesis, random ßuctuation of population structure over time, was tested by bootstrapping over loci. Differences between generation means were considered signiÞcant when the 95% CI did not overlap.

Seasonal Occurrence of Male Moths Near Cotton. Seasonal trends of moth abundance in cotton in the Mississippi Delta were essentially similar from year to year for H. zea (Fig. 1). Peak densities were variable and tended to decrease each year from 1995 to 1997, except for a peak late in 1996. Moth densities were low early in the season (before July), with peaks tending to occur in the midseason between 15 July and 24 August. Major peaks of abundance were broad and accompanied by minor peaks, which may reßect transient movement of moths across the area. General Patterns of Genetic Variation. Thirty-Þve alleles were detected at the 10 polymorphic allozyme loci investigated across a total of 39 populations. Frequencies of the most common alleles were extremely stable over time in both regional populations. In Leßore County, four of the 10 most common alleles were temporally homogeneous (Table 1). The observed genotypic frequencies were tested for deviation from the Hardy-Weinberg equilibrium using chi-square goodness-of-Þt tests. Thirty of 408 test cases (7.3%) were statistically signiÞcant in Leßore County populations over the 3 yr of data. The proportion of signiÞcant tests within generations ranged from 5 to 13.3% with the two highest proportions in 95G2 (second generation in 1995) and 96G1 (12.9 and 13.3%, respectively). The signiÞcant tests were not clustered about speciÞc loci and were also evenly distributed

Table 1. Frequencies of the most common allele at 10 polymorphic loci and contingency chi-square analysis of heterogeneity among generations in H. zea populations in Leflore County, MS Generationa Locus

95G1 (344)

95G2 (261)

95G3 (130)

96G0 (100)

96G1 (99)

96G2 (268)

96G3 (194)

97G2 (183)

97G3 (54)

Aconb

0.909b 0.032 0.919 0.018 0.982 0.006 0.915 0.025 0.982 0.016 0.981 0.012 0.966 0.021 0.793 0.083 0.962 0.020 0.963 0.018

0.909 0.044 0.888 0.059 0.984 0.009 0.942 0.029 0.975 0.018 0.965 0.016 0.977 0.021 0.839 0.040 0.973 0.012 0.975 0.012

0.957 0.041 0.938 0.036 0.966 0.020 0.929 0.042 0.974 0.023 0.970 0.019 0.965 0.025 0.829 0.052 0.969 0.027 0.958 0.021

0.867 0.040 0.907 0.062 0.988 0.023 0.922 0.034 0.964 0.040 0.975 0.035 0.948 0.046 0.886 0.077 0.970 0.022 0.967 0.026

0.917 0.038 0.895 0.033 1.0 0.0 0.929 0.040 0.963 0.020 0.966 0.043 0.969 0.034 0.829 0.059 0.971 0.025 0.960 0.035

0.890 0.028 0.897 0.053 1.0 0.0 0.945 0.013 0.974 0.025 0.970 0.028 0.954 0.021 0.847 0.020 0.973 0.006 0.976 0.014

0.900 0.031 0.899 0.036 0.988 0.014 0.945 0.025 0.969 0.025 0.982 0.006 0.959 0.012 0.848 0.036 0.973 0.009 0.975 0.020

0.842 0.059 0.919 0.026 0.993 0.008 0.942 0.022 0.974 0.011 0.970 0.027 0.910 0.027 0.854 0.034 0.971 0.023 0.976 0.027

0.877 0.012 0.932 0.015 0.993 0.009 0.890 0.006 0.972 0.002 0.992 0.010 0.888 0.083 0.895 0.073 0.985 0.020 0.980 0.008

Ak Got Gpi Had Idh-1 Idh-2 Pgm 3Pgd 6Pgd NS

␹2 102.16* 277.01** 69.97NS 100.60* 148.53 84.59NS 294.78** 188.35* 75.73NS 102.72*

, *, ** Indicate nonsigniÞcant or signiÞcant at P ⱕ 0.05 or 0.01, respectively. Populations are coded as year followed by generation in that year, e.g., 96G0 represents the overwintering generation in 1996. Numbers in parentheses indicate the number of individuals analyzed. b Allele frequencies with standard error underneath. a

June 2002


465

Table 2. Temporal changes of genetic diversity in Leflore County populations of H. zea (mean ⴞ SE)

Table 3. County, M.

Generation

Ho

He

Alleles/Locus

Generation

f

␪

F

95G1 95G2 95G3 96G0 96G1 96G2 96G3 97G2 97G3 Mean

0.103 ⫾ 0.056 0.096 ⫾ 0.036 0.083 ⫾ 0.018 0.101 ⫾ 0.044 0.088 ⫾ 0.032 0.089 ⫾ 0.014 0.095 ⫾ 0.018 0.108 ⫾ 0.013 0.107 ⫾ 0.017 0.097 ⫾ 0.008

0.107 ⫾ 0.019 0.105 ⫾ 0.009 0.085 ⫾ 0.021 0.106 ⫾ 0.045 0.102 ⫾ 0.020 0.105 ⫾ 0.016 0.102 ⫾ 0.013 0.119 ⫾ 0.012 0.110 ⫾ 0.028 0.105 ⫾ 0.009

2.44 ⫾ 0.16 2.43 ⫾ 0.25 1.90 ⫾ 0.45 2.12 ⫾ 0.74 2.10 ⫾ 0.45 2.54 ⫾ 0.18 2.54 ⫾ 0.34 2.36 ⫾ 0.19 2.40 ⫾ 0.56 2.31 ⫾ 0.22

95G1 95G2 95G3 96G0 96G1 96G2 96G3 97G2 97G3

0.0611 ⫾ 0.0360 0.0909 ⫾ 0.0413* 0.0501 ⫾ 0.0402 0.0496 ⫾ 0.0365 0.1742 ⫾ 0.0761* 0.0841 ⫾ 0.0568* 0.0719 ⫾ 0.0199* 0.0959 ⫾ 0.0276* 0.0483 ⫾ 0.0519

0.0002 ⫾ 0.0026 0.0056 ⫾ 0.0042 0.0072 ⫾ 0.0036 0.0023 ⫾ 0.0037 0.0011 ⫾ 0.0032 0.0031 ⫾ 0.0022 0.0021 ⫾ 0.0016 0.0012 ⫾ 0.0025 0.0005 ⫾ 0.0092

0.0612 ⫾ 0.0355* 0.0961 ⫾ 0.0447* 0.0434 ⫾ 0.0433* 0.0474 ⫾ 0.0372 0.1752 ⫾ 0.0775 0.0812 ⫾ 0.0557* 0.0700 ⫾ 0.0196 0.0970 ⫾ 0.0277 0.0490 ⫾ 0.0566

Ho, observed heterozygosity; He, expected heterozygosity under Hardy-Weinberg equilibrium.

among years (6.9% in 1995, 8.8% in 1996, and 7.1% in 1997). Heterozygosity deÞciencies were associated with most signiÞcant tests. Overall, the proportion was not much greater than the 5% expected for a type I error. For Monroe County populations of 1996, only two signiÞcant deviations from Hardy-Weinberg expectation were detected from a total of 128 population ⫻ loci tests. Helicoverpa zea has a very low heterozygosity (Mallet et al. 1993). Temporally, we observed few changes in heterozygosity in H. zea populations, and the observed heterozygosities were consistently lower than expected (Table 2). The mean number of alleles per locus was also stable across generations, ranging from 1.9 to 2.5. G based exact tests revealed that only seven of 88 deviations from random allele distribution among trap sites were signiÞcant (␣ ⫽ 0.05), and these were all in the Þrst (two cases) or second (Þve cases) generations. Fisher combined probabilities for loci in the second generations of 1995 and 1996 were signiÞcant (␹2 ⫽ 37.2, df ⫽ 20; ␹2 ⫽ 37.2, df ⫽ 18, respectively), indicating differentiation in allele distribution in these particular generations. These results were very similar to those found for H. virescens in the same region (Han 1999). However, this pattern did not repeat itself in the Monroe County population. All loci were distributed randomly among trap populations throughout the entire 1996 season. Fisher combined probabilities were high for all generations (P0 ⫽ 0.99, P1 ⫽ 0.97, P2 ⫽ 0.74, and P3 ⫽ 0.69). Spatial Structure of Populations. Excess homozygosity responsible for the deviation of trap populations from Hardy-Weinberg expectations in most generations was reßected in signiÞcant and positive f-statistics, which measure inbreeding effects within a trap population (Table 3). Although the allele distributions of 95G2 and 96G2 were signiÞcantly different from random, no single estimate of genetic structure was signiÞcant in either Leßore or Monroe County populations (Table 3; Fig. 2). The temporal patterns of genetic variation among local populations suggest that high gene ßow occurred throughout the season. Unlike H. virescens (Han 1999), there was no signiÞcant temporal reduction in gene ßow level among local populations.

F-statistics (ⴞSD) for H. zea populations in Leflore

Standard deviations were calculated by jackkniÞng over loci. Population estimates were generated using the method of Weir and Cockerham (1984). *, Indicates that the 95% conÞdence does not include zero.

Similar to differentiation among local populations, there was little subdivision between regions (Fig. 2). Ruckelshaus (1998) suggested that estimates of Fstatistics are a function of the spatial scale from which samples are drawn. They found that increasing the size of the subpopulation results in an increase in the magnitude of f and a decrease in ␪. When the underlying parameter of population differentiation is very small, ␪ estimates could be negative (Weir 1996). Because population structure at all spatial scales was near zero (Fig. 2), the effects of spatial scale on estimates of FST were not apparent even at a regional. Meanwhile, low differentiation between regional populations suggests a high level of gene ßow occurred at this spatial scale. Genetic Structure of Moth Population from Collected Eggs. The F-statistics for the sampled egg populations were f ⫽ 0.12, ␪ ⫽ 0.01, and F ⫽ 0.12. The upper and lower bounds of the bootstrapped 95% CI of ␪ estimates were 0.0192 and 0.0005, indicating signiÞcant differentiation among egg populations. The average of F-statistics over all paired ancestral adult populations (G0) were f ⫽ 0.0963 ⫾ 0.048 (mean ⫾ SE), ␪ ⫽ ⫺0.0059 ⫾ 0.0056, and F ⫽ 0.0907 ⫾ 0.0436. For the G2 adult population, the average of F-statistics were f ⫽ 0.1721 ⫾ 0.0452, ␪ ⫽ 0.001 ⫾ 0.0051, and F ⫽ 0.1701 ⫾ 0.0404. Ninety-Þve percent conÞdence in-

Fig. 2. Population subdivision among regional populations (Leßore County and Monroe County) and effects of spatial scale on the estimates of F-statistics: ␪ with 95% CI.

466 Table 4. zero

95G1 95G2 95G3 96G0 96G1 96G2 96G3 97G2 97G3


Vol. 31, no. 3

FST estimates (below diagonal) between temporal populations, and the probability (above diagonal) that FST is not larger than 95G1

95G2

95G3

96G0

96G1

96G2

96G3

97G2

97G3

Ñ 0.0012 ⫺0.0002 0.0056 ⫺0.0008 0.0028 0.0010 0.0062 0.0096

0.025 Ñ 0.0023 0.0016 ⫺0.0015 ⫺0.0003 ⫺0.0010 0.0050 0.0119

0.520 0.002 Ñ 0.0051 0.0015 0.0035 0.0018 0.0078 0.0094

0.003 0.062 0.001 Ñ 0.0030 0.0009 ⫺0.0001 0.0012 0.0024

0.438 0.567 0.080 0.011 Ñ ⫺0.0006 ⫺0.0003 0.0039 0.0105

0.081 0.127 0.001 0.001 0.957 Ñ ⫺0.0014 0.0009 0.0070

0.020 0.250 0.007 0.098 0.435 0.171 Ñ 0.0016 0.0063

0.001 0.030 0.002 0.026 0.195 0.101 0.148 Ñ 0.0

0.002 0.002 0.013 0.079 0.020 0.014 0.065 0.191 Ñ

tervals of FST estimates for the adult populations of both G1 and G2 were ⫺0.0143Ð 0.0002 and ⫺0.0098 Ð 0.0112, respectively, indicating no signiÞcant difference from zero. Temporal Trends in Population Differentiation. When overall genetic variance was partitioned into hierarchical components of years, generations within years, and traps within generations, Fxy differed with respect to hierarchical levels. No substantial genetic subdivision was apparent between generations, generations within years, and between years (Famong generations ⫽ Fgenerations within years ⫽ Famong years ⫽ 0). Genetic subdivision was found entirely within the low hierchical levels (Famong traps ⫽ Ftraps within generations ⫽ Ftraps within years ⫽ 0.024). The lowest level within the hierarchy, traps within generations, accounted for 97.73% of overall genetic variance. Similar amounts of genetic variance (1.14%) resided in components of generations within years and between years. This pattern is similar to that found in H. virescens when allozyme markers were analyzed (unpublished data) and becomes more evident when temporal population subdivision is analyzed further. Population differentiation and measurement of subdivision were assessed after pooling all trap catches for each generation. Although about half of 36 comparisons showed signiÞcant differentiation (␣ ⫽ 0.05) (Table 4), there was no signiÞcant differentia-

tion when RiceÕs sequential Bonferroni test was applied (Rice 1989). Similarly, when samples were pooled to generate populations by years, no signiÞcant differentiation was detected between them (unpublished data). Data from allelic distributions and population differentiation suggest that populations of the second generation tend to be the most differentiated in a given year (Tables 2 and 3). However, the results from a bootstrap analysis do not support this hypothesis (Fig. 3). Ninety-Þve percent conÞdence intervals of all the generation means included zero and overlapped each other. Random ßuctuation in estimates of genetic structure over time may be characteristic of H. zea. Inbreeding coefÞcients (f) within trap populations, although very small, were signiÞcantly greater than zero except for G0.

Discussion Results obtained from current studies of genetic structure of bollworm populations can be summarized as follows: (1) Allele frequencies were remarkably stable over space and time. (2) Genetic diversity was low and stable over time. (3) There was little population subdivision over local and regional scales. (4) Estimates of genetic structure ßuctuated randomly

Fig. 3. Mean estimates of ␪ (left) and f (right) by generation obtained by bootstrapping over loci in H. zea (Leßore County population). The upper and lower bounds correspond to 95% CI around the mean.

June 2002


over time. No correlation was detected between allelic variation and generation. Genetic diversities estimated for H. zea were comparable to previous results (Mallet et al. 1993). For H. zea, the expected heterozygosity changed from 0.085 to 0.119 over time, averaging 0.105 ⫾ 0.009. Mallet et al. (1993) found that the average heterozygosity was 0.06 ⫾ 0.02, which included monomorphic loci in the calculation of He. When only the polymorphic loci in the Mallet et al. (1993) study were included, heterozygosity increased to 0.11 ⫾ 0.03. Population differentiation at both local and regional scales was very low in the bollworm, when compared with other noctuid species (reviewed by Pashley 1984). This was not unexpected for a species that is highly polyphagous (⬎70 species of host plants) (Fitt 1989) and migratory. Similarly, little population differentiation was detected in H. virescens (Korman et al. 1993, Roehrdanz et al. 1994, Han 1999). However, in H. virescens, the midseason generation (the second generation on cotton) was genetically more variable between subpopulations than other generations. The increase in population subdivision over the season was assumed to be caused by shifts in the behavior of this species (Han 1999). Population shifts for both species between cotton and noncotton habitat was also observed by Lopez et al. (1995). They found the ratios of the mean number of male moths (captured by pheromone traps) in the cropped area relative to the uncropped area changed from 3.1 to 10.3 in H. zea and from 2.3 to 50.7 in H. virescens (Table 1). It was suggested that a larger proportion of the bollworm population was maintained in noncotton host plants than for the tobacco budworm. Lopez et al. (1995) also found that the highest ratios in cropped versus uncropped areas occurred for both bollworm and budworm in the midseason (19 May to 31 August). The sharp increase in the ratio for H. virescens indicates that fruiting cotton is the most attractive host at that time. The small increase in the ratios for H. zea suggests that cotton was not the only attractive host available (Lopez et al. 1995). In addition, the inßux of bollworm populations into cotton from crops like corn during the middle of the season would eliminate any differentiation accumulated from earlier in the season. The direct observations of population shifts between cropped and uncropped areas are helpful in interpreting the contrasting temporal patterns of genetic variation in H. virescens and H. zea. The effects of habitat or host plant preference in general could affect genetic substructuring over a geographic range by reducing gene ßow. Movement in early generations that emerge when only wild hosts are available, and subsequent recolonization of cotton could lead to a seasonal breakdown of population substructure through extensive gene ßow. The relatively uniform spatial distribution of H. zea formed a single, large panmictic population on a local and regional scale greater than that for H. virescens (Han 1999). The low estimates of FST between the two regional populations further suggests that high rates of gene ßow occurred at the regional scale. The genetic conse-

467

quences observed in these species were clearly driven by population biology. Results from this study and the previous one for H. virescens (unpublished data) represent the Þrst genetic evidence demonstrating different patterns of spatial distribution for H. virescens and H. zea. Genetic differentiation between populations from collected eggs during G1 was signiÞcantly greater than genetic differentiation between adult populations during G1 and G2. Egg production and oviposition in H. virescens and H. zea are complicated processes inßuenced by many factors (e.g., nutrition, mating, female longevity) (Proshold et al. 1982). In general, although females might live to be several weeks old, the reproductive lifetime is only 8 Ð10 d. They deposit most of their eggs within this short period, thus most of their reproductive output would be concentrated close to their larval sites relative to total adult dispersal distance. The movement of immature stages is extremely limited compared with adults. All these factors may contribute to the observed discontinuity in the population subdivision measured at different stages. These empirical data support a model prediction that larval populations tend to be more subdivided than adult populations in heliothine species (Han 1999). The current study provides important information on population genetics of cotton bollworm for designing and implementing sustainable pest management strategies. For example, the deployment of two or more transgenic plants would be less likely to result in subdivided populations within H. zea than within H. virescens. In addition, there was substantial inbreeding detected in both H. virescens (f ⫽ 0.0485) and H. zea (f ⫽ 0.0808) within each trap site. These inbreeding coefÞcients may have been inßuenced by the fact that only male moths were used in analysis if male dispersal behaviors differ from female behavior. This research was not designed to study mating structure in these species, but understanding mating patterns will be important for evaluating the efÞciency of the current refuge strategy implemented for delaying the evolution of Bt resistance. Effects of limited gene ßow and mating structure on the development of resistance to transgenic plants could be evaluated using simulation models incorporating detailed information on population biology. References Cited Caprio, M. A. 1994. Bacillus thuringiensis gene deployment and resistance management in single- and multi-tactic environments. Biocontrol Sci. Technol. 4: 487Ð 497. Caprio, M. A. 1998. Evaluating resistance management strategies for multiple toxins in the presence of external refuges. J. Econ. Entomol. 91: 1021Ð1031. Caprio, M. A., and B. E. Tabashnik. 1992. Gene ßow accelerates local adaptation among Þnite populations: simulating the evolution of insecticide resistance. J. Econ. Entomol. 16: 129 Ð148. Comins, H. N. 1977. The development of insecticide resistance in the presence of migration. J. Theor. Biol. 64: 177Ð197.

468


Daly, J. C., and P. Gregg. 1985. Genetic variation in Heliothis in Australia: species identiÞcation and gene ßow in the two pest species H. armigera (Hu¨ bner) and H. punctigera Wallengren (Lepidoptera: Noctuidae). Bull. Entomol. Res. 75: 169 Ð184. Fitt, G. P. 1989. The ecology of Heliothis species in relation to agroecosystems. Annu. Rev. Entomol. 34: 15Ð52. Goudet, J., M. Raymond, T. de Meeus, and F. Rousset. 1996. Testing differentiation in diploid populations. Genetics 144: 1933Ð1940. Gould, F. 1994. Potential and problems with high-dose strategies for pesticidal engineered crops. Biocontrol Sci. Technol. 4: 451Ð 461. Graves, J. B., B. R. Leonard, A. M. Pavloff, T. C. Sparks, S. Micinski, E. Burris, and K. J. Ratchford. 1989. Monitoring pyrethroid resistance in tobacco budworms. La. Agric. 32: 4 Ð5, 20. Han, Q. 1999. Temporal and spatial genetic differentiation among natural populations of tobacco budworm and cotton bollworm (Lepidoptera: Noctuidae). Ph.D. dissertation, Mississippi State University, Mississippi, MS. Korman, A. K., J. Mallet, J. L. Goodenough, J. B. Graves, J. L. Hayes, D. E. Hendricks, R. Luttrell, S. D. Pair, and M. Wall. 1993. Population structure in Heliothis virescens (Lepidoptera: Noctuidae): an estimate of gene ßow. Ann. Entomol. Soc. Am. 86: 182Ð188. Lewis, P. O., and D. Zaykin. 1999. Genetic data analysis: computer program for the analysis of allelic data, version 1.0 (d12). Free program distributed by the authors from the GDA Home Page (http://chee.unm.edu/gda/). Lopez, J. D., Jr., K. R. Beerwinkle, J. A. Witz, and J. L. Goodenough. 1995. Spatial and temporal patterns of catches in pheromone traps of Helicoverpa zea and Heliothis virescens in central Texas. Southwest. Entomol. 18(suppl.): 5Ð24. Luttrell, R. G. 1994. Cotton pest management: part 2. A U.S. perspective. Annu. Rev. Entomol. 39: 527Ð542. Mallet, J., A. Korman, D. G. Heckel, and P. King. 1993. Biochemical genetics of Heliothis and Helicoverpa (Lepidoptera: Noctuidae) and evidence for a founder event in Helicoverpa zea. Ann. Entomol. Soc. Am. 86: 189 Ð197. Pashley, D. P. 1984. The use of population genetics in migration studies: a comparison of three Noctuid species, pp. 305Ð324. In D. R. MacKenzie, C. S. BarÞeld, G. G. Kennedy, R. D. Berger, and D. J. Taranto (eds.), The movement and dispersal of agricultural important biotic agents. ClaitorÕs Publishing, Baton Rouge, LA.

Vol. 31, no. 3

Paster, N., G. Paster, F. Bonhomme, J. Catalan, and J. BrittonDavidian. 1988. Practical isozyme genetics. Ellis Horwood, Chichester, UK. Peck, S. L., F. Gould, and S. P. Ellner. 1999. Spread of resistance in spatially extended regions of transgenic cotton: implications for management of Heliothis virescens (Lepidoptera: Noctuidae). J. Econ. Entomol. 92: 1Ð16. Proshold, F. I., C. P. Karpenko, and C. K. Graham. 1982. Egg production and oviposition in the tobacco budworm: effect of age at mating. Ann. Entomol. Soc. Am. 75: 51Ð55. Raymond, M., and F. Rousset. 1994. An exact test for population differentiation. Evolution 49: 1280 Ð1283. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43: 223Ð225. Roehrdanz, R. L., J. D. Lopez, J. Loera, and D. E. Hendricks. 1994. Limited mitochondrial DNA polymorphism in North American populations of Heliothis virescens (Lepidoptera: Noctuidae). Ann. Entomol. Soc. Am. 87: 856 Ð 866. Roush, R. T. 1994. Managing pests and their resistance to Bacillus thuringiensis: can transgenic crops be better than sprays? Biocontrol Sci. Technol. 4: 501Ð516. Ruckelshaus, M. H. 1998. Spatial scale of genetic structure and an indirect estimate of gene ßow in eelgrass, Zostera marina. Evolution 52: 330 Ð343. Sell, D. K., G. S. Whitt, and W. H. Luckmann. 1975. Esterase polymorphism in the corn earworm, Heliothis zea (Boddie): a survey of temporal and spatial allelic variation in natural populations. Biocontrol Genet. 13: 885Ð 898. Sluss, T. P., and H. M. Graham. 1979. Allozyme variation in natural populations of Heliothis virescens. Ann. Entomol. Soc. Am. 72: 317Ð322. Sluss, T. P., E. S. Sluss, H. M. Graham, and M. DuBios. 1978. Allozyme differences between Heliothis virescens and Heliothis zea. Ann. Entomol. Soc. Am. 71: 191Ð195. Swofford, D. L., and R. B. Selander. 1981. Biosys-1: a FORTRAN program for the comprehensive analysis of electrophoretic data in population genetics and systematics. J. Hered. 72: 281Ð283. Weir, B. S. 1996. Genetic data analysis II: methods for discrete population genetic data. Sinauer, Sunderland, MA. Weir, B. S., and C. C. Cockerham. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38: 1358 Ð1378. Wright, S. 1951. The genetical structure of populations. Ann. Eugenics 15: 323Ð354. Received for publication 12 April 2001; accepted 28 November 2001.