Sample collection

Twenty specimens were collected from 17 sub-populations (locations) in eight states in the upper Midwest and one location from Louisiana (LA) (Table 1; Figure 1). Specimens were frozen in an ultra-low freezer to kill them quickly, maintain DNA integrity, and for long term storage. All samples, except from Louisiana, were over-wintering 5th instars collected in the fall from maize which allowed all ecotypes (univoltine, bivoltine, multivoltine) of O. nubilalis to be present. The LA adult samples were collected from rice in mid summer.

DNA isolation and quantification

DNA was isolated from 10 of the 20 individually frozen insect samples using a modified Black and Duteau (1997) CTAB extraction protocol. Prior to homogenization, the gut was removed from the larvae; and the head, wings, and abdomen were removed from adult insects from Mer Rouge, LA. The procedure was modified with the addition of a chloroform/phenol extraction step prior to ethanol precipitation. After removal of ethanol, the pelleted DNA was dissolved in 50 µl 1X TE buffer (10mM Tris-HCL pH 8.0; 0.1 mM EDTA).

Figure 1.

Sample distribution of O. nubilalis sub-populations across the upper Midwestern United States and in Louisiana.

Quantification to a known concentration of λ DNA (Life Technologies,  www.invitrogen.com) was performed by loading 1.0 µl of DNA mixed with 1.0 µl tracking dye to a 1% agarose gel that was electrophoresed for 30 min. Ethidium bromide stained gels were then visualized using Advanced Quantifier version 4.0 ( www.genomicsolutions.com). DNA samples were diluted with TE buffer if concentrations were >22.5 ng// µl.

AFLP-PCR

The AFLP procedure was achieved using 3 steps: 1) DNA template preparation; 2) DNA template preamplification and; 3) AFLP Selective amplification. Template preparation and AFLP assays were performed using a modified protocol of Vos et al. (1995). Infrared labeled (IRD-700) (LI-COR, Lincoln, NE, USA) EcoRI primers were used in the polymerase chain reaction.

Template preparation

Approximately 160 ng of genomic DNA in 7 µl of volume was incubated with restriction endonucleases EcoRI and MseI (New England Biolabs,  www.neb.com) for 2.5 hrs at 37°C in a total volume of 12.5 µl which contained 1.25 µl of 10X One-Phor-All buffer (Amersham Pharmacia biotech,  www.apbiotech.com), 0.125 µl of 10 U/µlMseI enzyme (1.25U/reaction), 0.0625 µl of 20 U/µlEcoRI enzyme (1.25 U/reaction), 0.125 µl of 10 mg/ml BSA (bovine serum album) (New England Biolabs), and autoclaved nanopure water to make up the volume to 12.5 µl. The restriction fragments were then incubated for 11.5 hrs at ∼ 25°C with 5 µl ligation mixture containing 0.15 µl of T4 DNA ligase, 0.5 µl of 10X T4 ligase buffer (New England Biolabs), 0.5 µl of 5 pmoles/ µlEcoRI adapter, 0.5 µl of 5 pmoles/µl MseI adapter (Operon Technologies,  www.operon.com), and 3.35 µl autoclaved nanopure water. After ligation, the template was diluted 8-fold. Double stranded adapters for ligation to the corresponding cut made by the restriction enzyme were prepared by incubating equimolar amounts of both adapter strands for 10 min at 65°C, 10 min at 37°C, 10 min at 25°C and stored at -20°C.

Preamplification of DNA template

PCR amplification of the ligated material consisted of 20 cycles (30s at 94°C, 1 min at 56°C, and 1 min 72°C) using 1.0 µl of diluted template from adapted DNA diluted 8-fold with IX TE buffer (10mM Tris-Cl, 0.1 mM EDTA [pH8.0]), 8 µl pre-amp primer mix II (contained two oligonucleotide primers, one corresponding to the EcoRI adapted ends and one corresponding to the MseI ends; Life technologies,  www.operon.com), 1.0 µ l of 10X PCR buffer containing 15mM MgCl₂ and 0.25 µl of 5 U/µl AmpliTaq DNA polymerase (1.25 U/reaction) (Applied Biosystems,  www.appliedbiosystems.com). The oligonucleotide primers in the pre-amp primer mix II are complementary to the adapter/restriction site with MseI primer containing one selective nucleotide M(N+1) primer and EcoRI primer containing no selective nucleotide E(N+O) primer (Table 2). Products from the preamplification reactions were diluted 20-fold with autoclaved nanopure water and were used as the template for selective amplification reactions.

Selective PCR amplifications

After screening 24 primer combinations, five primer combinations (Table 3) were selected for AFLP analysis. Reaction volumes of 10.5 µl containing 2.0 µl of diluted preamplified template, ∼ 5.0 µl of autoclaved nanopure water, 1.2 µl of 10X PCR buffer containing 15 mM MgCl2, and 0.06 µl of 5 U/ µl AmpliTaq polymerase (Applied Biosystems), 2.0 µl of MseI primer (M+1+2) (6.7 ng/ µl dNTPs) (Life Technologies), and 0.2–0.5 µl of 1.0 pmoles/ µl IRD-700 labeled EcoRI primer (LI-COR,  www.licor.com). Selective amplifications were conducted in a DNA thermal cycler 9600 (Applied Biosystems) using the following Touchdown PCR program: 1 cycle of 30s at 94° C, 30s at 65° C and 1 min at 72° C; 12 cycles of 30s at 94° C, 30s at 65°C, (subsequently lowering the annealing temperature by 0.7°C per cycle) and 1 min at 72° C; 23 cycles of 30s at 94° C, 30s at 56° C, 1 min at 72° C. After selective amplifications, reactions were stopped by adding 2.5 µl blue stop solution (LI-COR).

Table 2.

Oligonucleotide adapters and primers used for AFLP analysis.

Table 3.

Selective Primers used for AFLP analysis their associated amplified loci and range of fragment sizes for O. nubilalis.

Samples were then denatured at 95°C for 3 min and flash cooled on ice immediately before loading the polyacrylamide gel for electrophoresis. One µl of the sample was electrophoresed through KBplus 6.5% ready-to-use gel matrix (LI-COR), and the infrared fluorescent bands were then detected by a laser scanning system LI-COR Model 4200S-2.

Scoring AFLP data

Sizes of AFLP fragments were estimated using an IRD-700 labeled 50–700bp ladder. Fragments were scored using SAGA MX version 3.2 (LI-COR); a software program designed specifically for scoring AFLP data. The data were then converted to numerical data (1s for band presence, 0s for band absence) that allowed the identification of polymorphic markers. There were five gels for each primer pair containing four populations randomly picked for a total of 25 PAGE gels. All gels were viewed for clarity prior to scoring to establish scorable bands at a given fragment size for all populations. All bands consistently recognized by SAGA Generation 2 Software Version 3.2 (LI-COR), both monomorphic and polymorphic, were scored and included in the binary data matrix.

Coefficient of variation analysis

Analysis was conducted using DBOOT version 1.1 (Coelho 2001) to assess the appropriate number of polymorphic loci required for acceptable precision for genetic analyses. Bootstrapping analysis using the simple matching coefficient (Kosman and Leonard 2005) was conducted with 1000 permutations to assess the strength of the molecular markers. A comparison was then made between the coefficient of variation values against the number of markers scored and was then plotted.

Genetic diversity and gene flow of O. nubilalis

Data were analyzed using POPGENE version 1.32 (Yeh and Boyle 1997) using a dominant marker data set (164 markers) assuming Hardy-Weinberg equilibrium. Samples were grouped into three geographical regions due to proximity. Analyses were observed at three levels (1) individuals; (2) regions; (3) and whole populations. The percent (%) polymorphism, genetic diversity or heterozygosity (H), GST, and gene flow estimation (Nm) were then assessed within and between all populations. Individual populations were analyzed for genetic diversity (H) for each sub-population as per Nei (1973). GST values generated by POPGENE were given for the whole population and were expressed as the gene diversity of the single population subtracted from the gene diversity of the total population divided by the gene diversity of the total population (GST ⁼ HT - HS / HT). Gene flow was estimated from GST values and is expressed as (Nm) = 0.5(1 - GST)/GST (McDermott and McDonald 1993).

Mantel tests comparing genetic dissimilarity and geographical distance

A permutation test (Mantel 1967) was conducted to test the relationship between geographical distance and the measure of genetic dissimilarity (1-similarity), to further verify the estimated levels of gene flow found in the POPGENE analysis. The Mantel permutation tests were conducted using the MxCOMP module in the NTSYSpc ver. 2.1 (Rohlf 2002) software program with 1000 permutations. Tests were first conducted on the full data set and then separately with western samples, northern samples, and the southern samples as data sets.

Analysis of molecular variance

Analysis of molecular variance (AMOVA; Excoffier et al. 1992) was conducted with ARLEQUIN 2.0 (Schneider et al. 2000). In this analysis, total variance of the AFLP data set was partitioned at three hierarchical levels: (1) an among-population component; (2) a regional or six sub-population component; (3) and a within-population component. Unlike the calculations used for Nei's GST values, the AMOVA partitions the variation according to correlations among genotypes rather than gene frequencies due to the dominant expression of AFLP markers. The significance of the three variance components was tested using 1000 random permutations. A two-part AMOVA analysis was conducted to test genetic divergence (FST) as a factor of variation among individuals within a given population and between populations.

Results

Number and size of AFLP loci observed

The five primer combinations used with DNA from 18 sub-populations and 180 individual samples of O. nubilalis produced 164 amplified loci, averaged 33 loci per primer combination, and fragment size ranged from 57–341 bp (Table 3). Polymorphism within the 18 sub-populations was high, averaging 83% (136 out of 164 were polymorphic) and ranging from 72% to 94% (Table 4), indicating that AFLP provides polymorphic markers for studying genetic variation.

Table 4.

Heterozygosity and polymorphism (%) for 18 subpopulations of O. nubilalis. Individual heterozygosity (HS) indicates a lack of homogeneity.

Coefficient of variation of AFLP loci (markers) explored

The correlation of the coefficient of variation and the number of molecular markers examined for the sampled population is important for determining robustness in genetic variability studies because either a large number of loci or a large number of individuals should be examined (Hoelzel 1995). Analysis using the DBOOT program showed that the high number of markers used decreased the coefficient of variation to the point that all but 6.7% of the variation in the population was explained (Figure 2). Therefore, 164 markers were sufficient for unbiased genetic analyses.

Figure 2.

The coefficient of variation of O. nubilalis populations using AFLP molecular markers with 1000 bootstrap replications. Greater than 90% of the variation in the O. nubilalis population is explained by these AFLP markers.

Principal component analysis (PCOA)

Principal component analysis for these 18 sub-populations of O. nubilalis revealed four distinct clusters (Figure 3). Cluster 1 consisted of mixed populations from several states that may represent areas of transition from univoltine to bivoltine ecotypes, cluster 2 represented univoltine samples from South Dakota and North Dakota, cluster 3 represented samples typical of the univoltine ecotype, and cluster 4 represented samples that were less similar (more scattered on the graph) that are southern populations probably representing the transition from bivoltine to multivoltine ecotype.

Table 5.

Pair wise comparisons of Nei's coefficient of gene differentiation (GST) between three regions of O. nubilalis populations and estimates of gene flow (Nm).

UPGMA cluster analysis

Consensus averages yielded a dendrogram with similarity coefficients ranging from 65% to 74% for all 18 sub-populations of O. nubilalis, indicating a high degree of variability within locations (Figure 4). Bootstrap values ranged from 26 – 100% but a majority of nodes had bootstrap values exceeding 70%. Most sub-populations clustered with others in close geographic proximity or with similar climate. For example, samples from the western region (higher altitude) and from the northern region (higher latitude), with a similar level of growing degree days and suspected to consist primarily of a univoltine ecotype, were more genetically similar than samples from southern locations. Those nodes showing lower bootstrap values (AB, CC, CM and CO) indicate weak association and could represent areas of transition with a mixture of ecotypes (particularly bivoltine and multivoltine) as is also indicated with the PCOA analysis (Figure 3). As reported by Pornkulwat et al. (1998), univoltine ecotypes, using current techniques, may be more easily identified than bivoltine or multivoltine ecotypes, indicating that gene flow may be somewhat restricted for the univoltine ecotype.

Table 6.

Hierarchical analysis of molecular variance (AMOVA) for 18 sub-populations of O. nubilalis. The majority of the variation is within a given population.

Discussion

Molecular genetic tools have proven successful at detecting genetic variability. Pornkulwat et al. (1998) used RAPD markers that were able to distinguish multivoltine from univoltine and bivoltine ecotypes. Saldanha (2000) used RAPD-PCR to distinguish between local populations of O. nubilalisand found a large genetic group consisting of univoltine, bivoltine, and multivoltine ecotypes in Nebraska. However, disadvantages of RAPDs are apparent and therefore results can be arguable. Alamalakala (2002) showed that using AFLPs allows the analysis of large numbers of loci while repeatability is greatly increased over RAPDs. Our AFLP studies evaluated numerous loci showing a very high degree of polymorphism between the 18 sub-populations of O. nubilalis studied.

Understanding population structure of O. nubilalis will provide critical base-line information for developing sustainable management strategies. Populations may either exist as a single panmictic unit, as if in Hardy-Weinberg equilibrium, where all individuals have the equal opportunity to randomly mate, they may occur as a series of small sub-populations which are isolated from one another (island model), or they may exist as a continuous population where genetic exchange takes place by geographically proximate individuals and there is some genetic isolation by distance (Slaktin and Barton, 1989). Frankham et al. (2002) indicated that a GST value of 0.15 indicates differentiation among fragments. At the population level, our study indicated a GST value of 0.1652. Therefore, O. nubilalis in the Midwest may not consist of a continuous population as some differentiation may be taking place, as discussed by Coates et al. (2004), but this is far removed from representing an island model. Studies by Bourguet et al. (2000) also concluded there is a high degree of gene flow and no differentiation between northern (univoltine) and southern (bivoltine) populations in France: observed and expected heterozygosities were almost identical 0.19 to 0.33 and 0.23 to 0.32 respectively based on six polymorphic allozyme markers.

Figure 3.

Principal Component Analysis (PCOA) of 18 sub-populations of O. nubilalis. Four distinct clusters were evident possibly indicating some genetic isolation from other clusters.

Like the GST analysis, AMOVA results indicated that a majority of the variation detected was within sub-populations. Only a small amount of variation was due to variation among the three regions studied (north, south, and west). However, the most striking and unexpected result was the low percentage of variation among groups, further enhancing the implication of a high degree of gene flow possibly through migration events. The degree of variation within or among populations was variable. For many species, variability may be greater between than within populations (Roderick 1996; Clark et al. 2006). Local selection pressures and barriers could provide for genetic isolation resulting in a high degree of genetic drift. However, previous studies showed a higher degree of variation within populations than between populations for many insects (Coates and Hellmich 2003, Juan et al. 2004, Timmermans et al. 2005). Results from our study examining O. nubilalis populations from the Midwestern U.S. using AFLPs show a larger degree of genetic variation than previously reported but is similar to results found by Bourguet et al. (2000) with O. nubilalis in Europe.

The numerical differences observed in Nm between respective regions (north vs. south) and respective UPGMA clustering may be explained by the voltinism patterns that prevail in those regions. The opportunity for multiple mating during the year in southern regions could result in high genetic variability. Another possible explanation of lower indications of gene flow in the northern region could be that O. nubilalis living on alternate hosts, such as Solanum and Amaranthus may be unable to mate with individuals from maize due to different emergence patterns of O. nubilalis from the alternate hosts compared to those from maize. Or, as suggested by Coats et al. (2004), there may be genetic differences between sympatric uni- and bivoltine ecotypes. Coats et al. (2004) evaluated individuals using a portion of mtDNA whereas we examined 164 loci from random locations across the genome of 180 individuals. It would be beneficial to conduct similar studies over a broader range using a greater number of individuals for genetic analyses using AFLP that provides the polymorphic markers necessary to better address the questions of genetic differentiation in O. nubilalis.

Figure 4.

UPGMA Dendrogram of 18 sub-populations of O. nubilalis. Bootstrap values are at each node.

Showers et al. (2001) conducted mark and recapture studies to track movement of O. nubilalis after release. They found 37–52% of O. nubilalis males recaptured flew 800 m or more, 8–11% flew 3.2 km or more, while some collections of marked males and females occurred at distances greater than 40 km. However, a very large percentage of the O. nubilalis released were never recaptured (over 99.6%). It is possible that some of these individuals traveled much further than placement of the recapture pheromone traps. Showers et al. (1995) reported similar findings while studying potential movement on surface airflow of O. nubilalis. Although indirect measures of gene flow are tenuous (Whitlock and McCauley 1998, Hey and Machado 2003) our results, along with previous results (Showers et al. 1995, 2001) illustrating that O. nubilalis were capable of dispersing at least 32 km/yr, would support the concept of gene flow within a continuous population. Results from these ecological and genetic studies provide support for current refuge requirements relative to the use and management of Bt-corn. Also, these results indicate that if insect resistance was to evolve in O. nubilalis, quick action will need to be taken to defer the spread over large geographic areas.