Helicoverpa armigera is an important pest of cotton and other agricultural crops in the Old World. Its wide host range, high mobility and fecundity, and the ability to adapt and develop resistance against all common groups of insecticides used for its management have exacerbated its pest status. An understanding of the population genetic structure in H. armigera under Indian agricultural conditions will help ascertain gene flow patterns across different agricultural zones. This study inferred the population genetic structure of Indian H. armigera using five Exon-Primed Intron-Crossing (EPIC)-PCR markers. Nested alternative EPIC markers detected moderate null allele frequencies (4.3% to 9.4%) in loci used to infer population genetic structure but the apparently genome-wide heterozygote deficit suggests in-breeding or a Wahlund effect rather than a null allele effect. Population genetic analysis of the 26 populations suggested significant genetic differentiation within India but especially in cotton-feeding populations in the 2006–07 cropping season. In contrast, overall pair-wise FST estimates from populations feeding on food crops indicated no significant population substructure irrespective of cropping seasons. A Baysian cluster analysis was used to assign the genetic make-up of individuals to likely membership of population clusters. Some evidence was found for four major clusters with individuals in two populations from cotton in one year (from two populations in northern India) showing especially high homogeneity. Taken as a whole, this study found evidence of population substructure at host crop, temporal and spatial levels in Indian H. armigera, without, however, a clear biological rationale for these structures being evident.
Citation: Behere GT, Tay WT, Russell DA, Kranthi KR, Batterham P (2013) Population Genetic Structure of the Cotton Bollworm Helicoverpa armigera (Hübner) (Lepidoptera: Noctuidae) in India as Inferred from EPIC-PCR DNA Markers. PLoS ONE 8(1): e53448. https://doi.org/10.1371/journal.pone.0053448
Editor: Daniel Doucet, Natural Resources Canada, Canada
Received: August 29, 2012; Accepted: November 28, 2012; Published: January 11, 2013
Copyright: © 2013 Behere et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: GTB was supported by grants from the Melbourne International Research Scholarship (MIRS), the Melbourne International Fee Remission Scholarship (MIFRS), and the Albert Shimmins Post Graduate Writing-up Award (Faculty of Science, University of Melbourne). The Australian Research Council (ARC) provided funding to the Special Research Centre for Environmental Stress and Adaptation Research (CESAR). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
The polyphagous nature of the Old World cotton bollworm Helicoverpa armigera (Hübner) on a wide range of wild and crop hosts across different ecological zones, its highly variable life-history traits (e.g., number of generations, crop hosts, presence of summer/winter diapause) and seasonal abundance present a unique challenge for ecological and evolutionary studies. The number of generations possible per year is directly influenced by temperature, rainfall and presence of suitable hosts . In India, H. armigera is an important pest of cotton, legumes, cereals and vegetables, and presents a unique challenge to those studying its population genetic structure.
The farming landscape in India is predominantly characterised by small farms and mixed cropping systems. The cropping patterns in India normally ensures the presence of five to six different host crops in different proportions for H. armigera at any given time of the growing season , thereby creating a heterogeneous matrix of hosts which provide ideal platforms for H. armigera to move between hosts and geographic areas throughout the year. Furthermore, the presence of three major cropping scenarios in India (in the North, Centre and South) are influenced by the pattern of the monsoons  (i.e., southwest monsoons: June to September and northeast monsoons: October to December), and by the sub-tropical nature of the south that allows continuous cropping versus the more continental and temperate climate of the north. India's cropping scenarios therefore provide a range of hosts crops for H. armigera all year round in any given region, although cotton represents the main host crop on which this pest species completes three out of possible seven to eight generations annually in 11 states , , , see  for a map of cotton states. In the north, facultative pupal diapause is reported in the winter months following the cotton season , , , with synchronous emergence of large numbers of moths frequently triggered by the first heavy rainfall (after the arrival of the monsoon) after prolonged dry periods . The first post-diapause generations in the north are on crops and weeds other than cotton. In the mid-hill regions of Himachal Pradesh in northern India, chickpea is the first crop to be exploited by over-wintered H. armigera populations, between March and May . Windborne long-distance migration of H. armigera in central India is likely to occur at the end of the cropping season (December–January), while rains prolong the growing season in northern and southern India, with the resulting adult migration in these regions typically occurring around March–April . The temporal pattern of host availability and importance in the agricultural landscape therefore varies in a complex mosaic across India.
Over the past three decades, there has been speculation that Indian H. armigera could be categorized into races based on host-feeding preferences and limited inter-mating (e.g. , ). Such genetic diversity in connection with host plants has been previously shown in H. armigera in Australia  where there is, for example an identifiable lucerne-preferring ‘race’. Variable metabolic mechanisms mediating pyrethroid resistance have been reported with a shift from mixed-function oxidase-mediated pyrethroid resistance to an esterase-mediated mechanism during mid October in central Indian H. armigera populations , attributable to both the influx of moths from other populations ,  and the emergence from diapause of moth populations with genetic make-ups different from that of the non-diapausing population , . Differential responses to pheromones in different populations and variations in parasitoid responses have been reported , , and can possibly be interpreted as reflecting an influx of populations between different agricultural systems from different ecological zones, although this view has not yet been tested using population genetics data. Recently, genetically modified (GM) cotton varieties which expressed Bt-toxins Cry1Ac and Cry2Ab have made important contributions in reducing application frequencies and dosage of insecticides for the control of H. armigera. The intense selection with Bt proteins may contribute to population substructure, while evolutionary constraints to host crop preferences may further contribute to area-wide gene flow patterns . All these factors may result in genetic patterning in the species across the Indian agricultural landscape. Understanding the movements of H. armigera adults between GM and non-GM crops, or between sprayed and unsprayed crops will be crucial to the management of Bt and insecticide resistance in this pest.
The only India-wide major polyphagous crop pest thoroughly examined for genetic diversity is the whitefly Bemisia tabaci which comprises a polyphagous species complex with ecological niche separation with respect to host plant (and some geographic) preference . It has at least 6 biotypes in India and probably many more. In particular the older Asia I groupings had a preference for eggplant and Asia II for tobacco and cassava. The more recently introduced B-biotype does particularly well on Tomato which only 1 of the 14 Asia 1 ‘races’ does. This is a particularly complex example but it does show the potential for such separations in other widespread polyphagous species.
Studies of H. armigera population genetics based on different DNA markers such as random amplified polymorphic DNA , isozymes , mtDNA  and microsatellites (e.g., ; ) have been reported. These studies found little genetic variation between widely separated populations, supporting the idea that extensive long distance migration was occurring in H. armigera. In Australia, studies have revealed small genetic distances between widely separated populations based on isozymes , mitochondrial DNA polymorphisms , and sodium channel gene alleles . In contrast, studies of Scott et al. , , ,  based on microsatellites suggested substantial population substructure in Australian populations of H. armigera. Endersby et al.  applied markers developed by both Scott et al.  and Ji et al.  to study Australian H. armigera populations collected from the southern and western regions of Australia and found no significant patterns of population substructure. The conflicting findings of Scott et al. , , , , Endersby et al.  and Weeks et al.  were due, at least in part, to factors associated with allele drop-outs (ADO), null alleles caused by mutations at primer annealing sites , and microsatellite loci being associated with non-LTR RTE retrotransposable elements (TE's) in Scott et al.'s analyses .
Given the wide distribution and migratory ability of H. armigera, effective and reliable molecular genetic markers must demonstrate efficiency in PCR amplification in individuals from within and between populations within a country, and between populations from different countries. Although less likely to be affected by TE-induced PCR failures as seen in various lepidopteran microsatellite markers (including three for H. armigera, ), Exon-Primed Intron-Crossing (EPIC)-PCR markers ,  nevertheless are susceptible to null alleles if exon regions are variable at primer annealing sites, although this is yet to be demonstrated in population genetics studies. This study applies EPIC-PCR markers designed specifically for H. armigera  to generate data for testing the hypothesis that geographical and host plant components are significant factors underlying genetic variation in Indian H. armigera. In the absence of detailed knowledge of gene flow and for the purpose of this study we regard as ‘populations’, samples of H. armigera taken from different crops, areas and/or at different times.
Materials and Methods
Sampling and DNA extraction
A total of 786 H. armigera individuals were collected from India in the three cropping seasons 2004–5, 2005–6 and 2006–7, as larvae, or moths (Fig. 1 and Table 1). Collections were made from 14 populations on cotton (Gossypium hirsutum, Malvaceae), 5 populations of pigeonpea (Cajanus cajan, Fabaceae), 4 populations of chickpea (Cicer arietum, Fabaceae) and one of eggplant (Solanum melongena, Solanaceae). Larvae were collected by direct sampling from different host plants, either directly into ethanol until needed for gDNA extraction, or kept on artificial diet until the pupal stage. Some of these were taken as samples only after they had emerged into adult moths. Male moths from Nagpur_1 and Nagpur_2 were collected by pheromone traps (Table 1). All pupae and adult moth samples were also preserved in absolute ethanol at −20°C until required for DNA extraction. Only a small portion (5 mm of the posterior portion of larvae and pupae, or half the abdomen of adults) of each sample was used for genomic DNA (gDNA) extraction as previously reported  or using the method of Zraket et al.  with slight modifications. Absence of cross-contamination during the gDNA extraction process was confirmed by the inclusion of a blank extraction among each gDNA extraction batch. The PCR-RFLP (Restriction Fragment Length Polymorphism) Helicoverpa species diagnostic test of Behere et al.  was used to confirm that all larvae sampled for this study were H. armigera.
RpS2 EPIC marker allele characterisation
EPIC-PCR markers RpL3, RpL12, RpL29, RpS6 and RpS2  were utilised to infer population genetic structures in Indian H. armigera populations. Molecular characterisation of RpS2 EPIC-PCR marker allele polymorphisms has not been previously reported and is here investigated using the methods described in Tay et al. . Ten RpS2 EPIC maker alleles from Australian and Chinese H. armigera individuals (samples previously used for allele characterisation by Tay et al. ) were randomly chosen, cloned and sequenced to ascertain the presence of allele homoplasy and nucleotide insertions/deletions (Indels).
Fluorescence labelling of polymorphic EPIC markers and screening
The forward primer of each EPIC-PCR marker was labelled with a fluorescent dye (FAM, HEX, or TET) to allow detection during electrophoresis. All amplifications were performed in a 15 µL reaction volume consisting of 7.5 µL of 5× GoTaq® Green Master Mix (Promega # M7122); 1.8 mM MgCl2; 0.5 µM of each labelled forward and reverse primer and 50–75 ng of template DNA. The PCR amplification profile consisted of an initial template denaturation step of 5 min at 95°C (1 cycle); followed by 35 cycles that consisted of template denaturation (95°C, 1 min) – primer annealing at specific temperature for 1 minute  – template extension (72°C, 1 min); and a final 10 min template extension at 72°C (1 cycle). PCR amplicons (5 µL) of all five loci were loaded on 1% ethidium bromide-stained 1× tris-borate-EDTA (TBE) agarose gels, run at 90 V for 90 minutes and visualised over a UV-illuminator, prior to individually multiplexed in 96-well plates by pooling 1 µL of PCR product for each of the loci labelled with three different florescent dyes. For genotyping, DNA fragment sizes were determined by a MEGABASE 1000 automated sequencer (Amersham Biosciences) at the Genetic Analysis Facility (GAF, James Cook University, Queensland, Australia). A size standard (400-R) was co-loaded with every sample to allow accurate sizing of DNA fragments. The final volume was adjusted to 10 µL with dH2O; post-PCR desalting was undertaken according to the protocol provided by GAF. Genotypes were scored manually with the help of marker panel set options implemented in the program GENETIC PROFILER 2.2 (Amersham Biosciences). All genotypes were scored unambiguously and where necessary allele peaks were corrected manually. Individuals which amplified for one locus but failed to amplify in PCR for other loci, were repeat amplified for up to a further two times. If a sample failed to amplify across all loci in at least one of the three rounds of PCR, it was considered as a DNA-extraction failure and discarded from subsequent analyses.
Analysis of null alleles in EPIC-PCR markers
To estimate EPIC-PCR markers null allele frequencies we designed nested (alternative) EPIC-PCR forward and reverse primers for the EPIC-PCR primer pairs used (Table 2). For alt_RpS2 and alt_RpL29 EPIC-PCR primers, we tested 42 randomly selected individuals which were identified as homozygotes using the original RpS2 and RpL29 EPIC markers, as well as six heterozygotes as positive controls. For alt_RpS6, the total number of individuals re-tested was 46 (18 homozygotes, 28 heterozygotes), and for alt_RpL3 a total of 46 individuals (40 homozygotes, 6 heterozygotes) were re-tested. Null alleles were considered detected if during re-genotyping by nested EPIC-PCR markers individuals previously scored as homozygous were found to be heterozygous and vice versa.
Basic statistics for the EPIC-PCR data (average number of alleles per locus, allelic richness averaged over loci, and Weir and Cockerham's measures of FIS ) were calculated using FSTAT version 2.9.3 . FIS, an inbreeding coefficient, measures the reduction in heterozygosity of an individual due to non-random mating within its sub-population. Observed (Ho) and expected (He) heterozygosity were estimated and departures from Hardy-Weinberg equilibrium (HWE) were tested using the probability test as implemented in GENEPOP version 3.2 . The tests for genotypic linkage disequilibrium among pairs of loci were performed in GENEPOP using Fisher's tests , with unbiased P values derived by a Markov chain method (10,000 de-memorisations, 1,000 batches and 10,000 iterations/batch). The significance values for multiple significance tests were set using the sequential Bonferroni procedure  within the population genetics software FSTAT. To investigate population differentiation, pair-wise FST estimates  (with 95% confidence limits) and significances (determined with 6,500 permutations) were calculated using FSTAT.
The geographic partitioning regime used by Kranthi et al. ,  was followed. Genetic diversity was partitioned into three model structures according to geographic regions (northern, central and southern India; Fig. 1), host crops (cotton, pigeonpea chickpea and egg plant), and cropping seasons (season 2004–05, 2005–06 and 2006–07). Within each model structure, the genetic variation was further partitioned into three levels: (1) among geographic region/host/cropping season; (2) among populations within geographic regions/hosts/cropping seasons, and (3) within populations. A hierarchical analysis of molecular variance (AMOVA) was carried out using pair-wise FST as the genetic distance measure using the population genetics software ARLEQUIN 3.1 , . In pair-wise FST estimates and Structure analysis, the two pheromone-trapped populations (Nagpur_1 and Nagpur_2) were excluded because of the unknown host crops. Erode, the only cotton population in season 1 (2004–05), was also excluded from these analyses.
The program Structure v2.3.2  that implements a Bayesian clustering method, was used to identify admixed individuals and for assignment to likely membership of population genetic clusters (‘K’) through the assumption of known source populations and HWE at all loci . To estimate the most likely K we evaluated all possible K's (i.e., K = 1 representing no genetic structure, to K = 23 representing each population being genetically distinct) using simulation of 20 iterations, with each iteration consisted of 50,000 ‘burnin’ followed by 500,000 Markov Chain Monte Carlo (MCMC) replications, with default settings for both the Ancestry Model (Admixture Model) and the Frequency Model (allele frequencies correlated among populations; assumed different FST values for subpopulations). The ΔK method of Evanno et al.  was used to ascertain the most likely K value, although the log probabilities of data (Ln P(D)) for K were also evaluated. The proportions of an individual's genome belonging to particular K population clusters are given a ‘Q’ score which enables Structure to assign individuals (or portions of an individual's genome) to a particular cluster .
Null alleles in EPIC-PCR markers
Of the five sets of nested EPIC-PCR markers, we failed to design an alternative RpL12 EPIC primer due to the short exon sequence available, and the null allele frequency for this marker was therefore not estimated. Based on the alternative EPIC-PCR markers alt_RpS2, alt_RpL29, alt_RpL3 and alt_RpS6, null allele frequencies for the original EPIC-PCR markers were estimated at 9.4%, 6.5%, 6.3%, and 4.3% respectively which were considered as being at a moderate level , and were within the null allele frequency range (i.e., 2.2%–10.3%) of microsatellite DNA markers used by Endersby et al.  for inference of Australian H. armigera population genetics structure. As we were unable to estimate the null allele frequency of RpL12, analyses of population substructure patterns of our Indian samples were performed by both including or excluding RpL12 (Table 3). FIS estimates from excluding the RpL12 locus remained unchanged for three populations (Abohar_1, Hisar and Karimnagar), reduced in seven populations (Nagpur_2, Yavatmal_2, Yavatmal_3, Warangal, Guntur_1, Nalgonda and Coimbatore), and increased in the remaining 16 populations. Taken as a whole, the inclusion of the RpL12 locus lead to lower and/or no change in FIS estimates in 19 of the 26 populations studied, and did not drastically lower FIS estimates in cotton populations, suggesting that this marker is unlikely to have harboured an excessively high frequency of null alleles.
EPIC marker variability
A total of 155 alleles were scored from the five loci (RpL3, RpL12, RpL29, RpS2 and RpS6) in 26 populations (n = 786) of H. armigera. The most polymorphic marker was RpS6 (55 alleles), followed by RpL29 (49 alleles), RpL3 (19 alleles), RpS2 (14 alleles) and RpL12 (15 alleles). The mean number of alleles and the mean observed (Ho) and expected (He) heterozygosities for each population are shown in Table 3. The average observed heterozygosity value for the five loci was 0.34 (range: 0.18–0.42). Estimates of observed heterozygosity were lower than expected in all populations, and levels of allele richness did not differ significantly between populations. Molecular characterisation of 10 randomly selected alleles (GenBank EU707432–EU707441) from the RpS2 EPIC marker showed no allele homoplasy, with allele length polymorphisms due to Indels within the intron.
Departures from Hardy-Weinberg equilibrium over all loci were found in all populations of H. armigera. This significant deviation from Hardy-Weinberg equilibrium was due to an excess of homozygotes at all loci, and is further reflected by the FIS values (Table 3). Genotypic linkage disequilibrium tests found no significant associations between pairs of loci for any populations or over all populations after Bonferroni corrections for multiple comparisons, indicating independent assortment for these Rp EPIC markers.
Analysis of Molecular Variance (AMOVA) and F statistics
AMOVA analysis (Table 4) indicated that as a whole, Indian populations showed similar levels of genetic variability regardless of geographic regions (−0.06%, FCT = −0.006, P = 0.36), host plants (−0.68%, FCT = −0.007, P = 0.98) or cropping seasons (−0.23%, FCT = −0.002, P = 0.65). However significant genetic structures were detected among populations within geographic regions (4.51%, FSC = 0.045, P<0.001), within host plants (i.e., within host plant species across sampling sites) (5.07%, FSC = 0.050, P<0.001) and within cropping seasons (4.63%, FSC = 0.046, P<0.001). Large variations (FST>95%) were found within populations in each of three model structures (geographic region, host and cropping season; Table 4). To better understand underlying factors that contributed to within population genetic variations, data from two cropping seasons (2005–06 and 2006–07) were partitioned individually according to geographic region or plant host. The overall trend remained similar (i.e. there remained significant variation between populations within geographic regions and within hosts (data not shown)). Therefore, values for these significant genetic variations were examined in detail by estimates of pair-wise FST values. The overall pair-wise FST values ranged from −0.001 to 0.431, with higher FST values in general being associated with cotton populations (Table 5). Cotton populations collected in the 2006–07 cropping season further differed significantly between northern and southern India (Table 5). Within cotton, the highest FST values were seen in the Bharuch population when compared with all other populations. Pair-wise FST values for populations from food crops (pigeonpea, chickpea and eggplant) contrasted with those observed in cotton populations and were generally non-significant (except Coimbatore) (range: −0.008 to 0.100; Table 6).
Based on the averaged Ln P(D) of 20 simulations for each K = 1 to K = 23, the best K was identified as K = 8 (−9912.76±536.43 s.d.). However, using the Evanno et al.  method identified the best ΔK value (614.37) at K = 2 followed by the second highest ΔK value of 4.06 at K = 4 (Fig. 2, Fig. 3). Although the ΔK value was largest for K = 2, this large change probably indicated a shift from the unlikely scenario of our samples showing no structure (i.e., K = 1) towards more plausible scenario of presence of population substructure (i.e., K>1). The large and positive ΔK value for K = 4 (Fig. 2) was therefore selected as the cluster number most likely to assist in visualising significant population substructure. The generalised patterns of population structure for K = 2 to K = 10 are presented (Fig. 4) to help visualise the selection of K = 4. A detailed Structure bar graph at K = 4 across all 23 populations was presented (Fig. 3). No obvious biologically relevant population structure patterns could be inferred from the Structure analysis, and setting of K>4 clusters progressively introduced greater population heterogeneity and further reduced the power of interpretation, even for the Ln P(D) optimum K = 8. Between K = 2 and K = 9, the Bharuch (pop. 4) and Surat (pop. 5) populations in central India appeared highly homogeneous, while all other populations showed higher levels of admixture (Fig. 4). Detailed examination of the Structure analysis at K = 4 (Fig. 3) showed that substantial substructure within and between populations existed, with populations which were geographically close to each other sorting into very different genetic clusters. Across the three sampling years (Fig. 3), reduced genetic diversity was seen only in some cotton populations (e.g., Year 2: Baruch (pop. 4), Surat (pop. 5), Karimnagar (pop. 9); Year 3: Abohar (pop. 1), Fatehebad (pop. 2)). Substantial substructure between populations can be seen, for example, in Year 2 cotton populations between Baruch (pop. 4)/Surat (pop. 5) and Guntur (pop 11)/Karimnagar (pop. 9)/Yvatmal (pop. 8). Substructure of populations is further seen between populations which are in close geographic proximity (e.g. in cotton Year 3: Abohar (pop. 1), Fatehbad (pop. 2) and Hissar (pop. 3); food crops Year 3: Medak (pop. 12) and Nalgonda (pop. 13)). This contrasted with the genetic patterns of food crop populations (Table 6), which did not show such strong substructure.
Locations and populations are as in Fig. 1 and Table 1: northern India (1) Abohar, (2) Fatehbad, (3) Hissar; central India (4) Bharuch, (5) Surat, (6) Jalgaon, (8) Yavatmal; southern India (9) Karimnagar, (10) Warangal, (11) Guntur, (12) Medak, (13) Nalgonda, (14) Haveri, (15) Tumkur, (16) Coimbatore, and (18) Dindigal. Average log likelihood of data Ln P(D) (primary axis, solid line) and ΔK (secondary axis, dashed line) are shown, with best K = 4 indicated an arrow.
Bar graph of K = 4 for all individuals from 23 populations collected across three sampling years from either cotton or food crop plant hosts are shown. Locations and populations are as in Fig. 1 and Table 1, Indian agricultural regions are northern Indian (NI), central India (CI) and southern Indian (SI).
H. armigera population genetics inferred from EPIC markers
Population genetic analysis of Indian H. armigera samples using EPIC markers clearly indicated no obvious population substructure with geographic region, year or crop, and indicated significant genetic differentiation between northern and central/southern Indian cotton-feeding populations in the cropping season of 2006–07. Within the populations HW non-equilibrium was detected, with a likely contributing factor being the presence of null alleles within our DNA marker system. Null allele frequency estimates for four of these EPIC-PCR markers were moderate (4.3% to 9.4%). The conclusion on population substructure patterns remained overall unaltered regardless of the inclusion or exclusion of the untested RpL12 locus from analysis. In the population genetic study involving predominantly Australian H. armigera populations, Endersby et al.  excluded 3 pairs of SSR markers with the greatest null allele frequencies (i.e., 19.2%, 31.6% and 47.4%), and demonstrated that DNA markers with low to moderate levels of null allele frequencies were sufficiently powerful to enable meaningful interpretation of gene flow patterns, a conclusion consistent with population genetics simulation studies , . Chapuis and Estoup  also concluded that the most accurate FST estimates were obtained when null alleles in marker systems were not excluded (i.e. by scoring only visible alleles within study populations), although within populations with substantial gene flow (i.e., non-cotton populations in this study), FST estimates were less biased than in populations with limited gene flow (i.e., cotton populations).
Our study further demonstrated that null allele events in H. armigera population genetic markers are relatively common, despite attempts to minimise their occurrence by designing markers at conserved gene coding regions. The effort in designing EPIC-PCR markers has other advantages such as enabling nested PCR markers to be developed for retesting of populations at the same loci, as well as the overall lower null allele frequencies as compared with null allele frequencies from random SSR markers which may be associated with TE's , . Null alleles or inbreeding may also contribute to heterozygote deficiency. A locus-specific heterozygote deficit is an indication of null alleles, rather than inbreeding or other population processes which are generally reflected across all loci. Significant heterozygote deficiencies were detected at all five loci in most H. armigera populations tested, further indicating only low to moderate levels of null alleles affecting our EPIC markers.
A Wahlund effect (i.e., fine-scale heterogeneity versus large-scale homogeneity; e.g., see ) as indicated in our high FIS, may further affect interpretations of population structure of H. armigera in India. The sample size of n = 786 allowed us to detect significant departures from HWE for all five EPIC markers. Although the Structure analysis assumed that all potential source populations were sampled when assigning genetic clusters to individuals, our study was aimed at understanding broader H. armigera population patterns in Indian's highly heterogeneous agricultural landscape rather than inferring individuals' specific origins. The lack of interpretable and biologically meaningful Structure results as well as the general patterns of low population substructure in our study could potentially be due to our markers being in non-HWE (i.e., Structure analysis assumes HWE in loci, ), although various authors (e.g., De Barro , Brown et al. ) reported no apparent significant effects to Structure results in loci that did not demonstrate HWE. Our finding that the Bharuch and Surat populations are generally highly homogeneous may therefore either be an indication of a lack of gene flow and/or due to selection.
Departures from HWE due to homozygous excess may represent true biological phenomena in H. armigera such as those due to strong inbreeding caused by frequent bottlenecks, the Wahlund effect (e.g., Nielsen et al. ), or be due to extrinsic factors such as insecticide selection pressure, Bt proteins and/or plant secondary chemicals, or other environmental selectors. In addition to possible effects of utilising microsatellite DNA families/TE-associated loci , , the generally small-scale heterogeneous Indian cropping landscape, intense selection pressure from heavy insecticide applications, exposure to Bt toxins from GM cotton, and/or exposure to host plant secondary compounds could cause H. armigera populations to deviate from HWE (although we should note that the later two factors may be similar in Australia).
The different cropping systems to which H. armigera is exposed to may also be important underlying factors that contributed to population substructure differences. Host crops with short flowering periods (e.g., food crops such as eggplant, chickpea and pigeonpea) generally support no more than one or two H. armigera generation; while hosts such as cotton with prolonged flowering periods are capable of supporting≥three consecutive generations , . Populations that feed on cotton are under tremendous selection pressure from insecticide applications , and from varying levels of the allelochemical gossypol associated with different life stages and specific cotton varieties . Although host crop species are generally the same across India, the temporal pattern of availability and of the size of the H. armigera populations varies greatly, with relatively small populations on only one or a few crops at some times of year in some regions. In northern India, H. armigera is known to occur initially on food crops (chickpea, sunflower, some vegetable crops) during February to July prior to feeding on cotton from August (see ). A large proportion of the population in the north may enter diapause, to emerge after the partial break between cropping seasons, and re-mix with the smaller, non-diapause population which has been subsisting for 1–2 generations on other, less intensively managed, hosts. The switch from food crops (i.e., with low insecticide exposure) each typically supporting a single H. armigera generation, to cotton hosts with increased insecticide and Bt toxin exposure capable of supporting multiple generations, can lead to intense selection on sedentary cotton populations. These cotton populations are accompanied by an increase in population densities, with peak infestations typically recorded during September to November in the north (see ). These peaks are accompanied by a significant increase in population density in non-GM cotton.
In contrast, central and southern Indian H. armigera populations initially feed on cotton (August to October in central India, September to December in southern India) prior to switching over to food crops. Central and southern Indian H. armigera populations are therefore likely to experience less consistent insecticide/Bt protein selection pressure approaching the end of each cropping season as host crops change from cotton to food crops. This may promote gene exchange between populations as they move between crops. This scenario is the reverse of that in northern India, where populations sampled near the end of the cropping season might be expected to show more differentiation, as selection by insecticides and Bt toxins in cotton on such populations potentially operates over several generations with reduced migration, creating a mosaic of genetically different populations across the various cotton types and management systems. This may be part of the explanation for the significant pairwise FST values from cotton populations (especially in northern India) (Table 5) suggesting raised levels of substructures in cotton populations.
H. armigera population structure on cotton
Population genetic analysis of Indian H. armigera populations showed a high degree of differentiation between collections within cropping seasons. Based on FST values the observed genetic variation was most marked with populations collected from cotton in cropping season 3 (2006–07). The genetic differentiation revealed by pair-wise estimates of FST values suggested that there existed seasonal and geographical variation within H. armigera populations collected from cotton (Table 5). Further, populations from northern India were significantly different from southern Indian populations in cropping season 3, although the underlying factors responsible for this remain unclear.
Overall estimates of pairwise FST values for cotton populations in cropping season 2 between central and southern India were non-significant (Table 5). The exception being the Bharuch population which showed significant FST values against all other populations. The distance between Bharuch and Surat is only 60 km and there are no geographical barriers, differences in cropping practices, or climatic conditions between these areas. Nevertheless these populations were significantly different from each other (FST = 0.192). Possible reasons contributing to such strong genetic differentiation may include different H. armigera generations being collected (see Table 1 for collection dates), and/or sampled populations feeding on different cotton types (e.g., Bt or non-Bt cotton; cotton varieties with different levels of gossypol contents). It is also possible that some of the diversity/structure observed may be associated with the mosaic of Bt versus non-Bt fields, and this would warrant further study. Pairwise FST values between seasons 2 and 3 in the cotton crop broadly indicated population substructure differences between northern and central/southern Indian, and may reflect underlying differences in cropping patterns (i.e., ‘cotton first’ or ‘cotton last’ in the cropping season).
Fluctuations in host availability may influence H. armigera populations and could result in genetic differentiation among local populations, but this assumption is only valid when there is substantial genetic isolation between populations. In India, H. armigera population substructure has been further suggested based on feeding preferences , insecticide resistance , differential response to pheromones  and to parasitoids . The abundance, movement and distribution of H. armigera were found to be associated with rainfall and humidity in Australia (e.g., ) and suggested for India . H. armigera is a facultative migrant , responding largely to local environmental conditions and host availability (i.e., moths remain sedentary where food resources such as flowering plant hosts are available). Cropping and landscape patterns, as well as insecticide application practices and resistance pest management strategies in Bt cotton differ greatly between Australia and India, which therefore limits meaningful comparison between findings from this study and Australian H. armigera population genetic structure. For example, the agricultural landscape in India is typically of low acreage, highly diverse and fragmented in the pattern of crop hosts growing at any given time. As such, it generally enables the presence of more than five alternate hosts of H. armigera at any given time of the cropping season , , while there would frequently be only a single major host over the corresponding period in Australia across comparatively large cropping areas within each production region.
H. armigera population structure on food crops
Pigeonpea is cultivated all over India, where it is commonly grown alongside cotton or cultivated as an inter-crop within the cotton agro-ecosystem . Furthermore, flowering periods of cotton and pigeonpea overlap which may facilitate population movements between cotton and pigeonpea, and may explain the high pairwise FST values associated with pigeonpea (Table 6). Pulse crops (i.e., pigeonpea and chickpea) are preferred hosts of H. armigera compared to cotton and are planted on larger areas than cotton (Directorate of Economics and Statistics, Department of Agriculture and Cooperation, Ministry of Agriculture, Government of India , , ). Pesticide applications on food crops are, however, less intense than on cotton and H. armigera populations are therefore expected to experience less selection for specific genotypes, which may result in a lower level of apparent genetic sub-structure than is seen in cotton. Overall estimates of pairwise FST values among food crops were by and large non-significant in all three cropping seasons (Table 6), although within cropping seasons the number of populations sampled was relatively low.
If the ideas presented above are correct, patterns of population structure analysed in one season should be reflected in the analysis of subsequent seasons. Consistent patterns observed between seasons, hosts and regions would thus support there being host- and/or region-associated micro-population structuring in Indian H. armigera. Such patterns have not been clearly seen in this study. Unsurprisingly, the situation is dynamic. For example population 8_3 was collected from Yvatmal on pigeonpea in years 2 and 3 (pop. 8_3a) and shows some shift of genetic profile (Fig. 3). A similar situation exists with cotton populations from Guntur (pop. 11) in years 2 and 3. Although the overall number of Indian populations sampled in this study is comparable to that in other lepidopteran population genetic studies (e.g., H. armigera, ; P. xylostella, ; C. pomonella, ), the complexity of Indian cropping systems nevertheless means that additional populations from northern, central and southern India over the season and on various hosts will be needed to enable more detailed interpretation. H. armigera populations analysed in this study do not cover all desired sampling locations and hosts for all seasons, and the patterns seen may also be influenced by factors such as sampling errors. In order to explain this intra-seasonal or host crop-associated genetic differentiation, further analysis of samples collected several times from the same host crops and sites over multiple cropping seasons will be needed.
Understanding the observed genetic structuring in Indian H. armigera populations from cotton may be further advanced with research on the detoxification capabilities and ecological aspects of this highly polyphagous pest insect species. Variables such as insecticide usage, Bt/non-Bt cotton, different hybrids of cotton, different hosts, climatic conditions, and diapause should be considered separately to better ascertain their importance to the population genetic structure of cotton-feeding H. armigera in India.
C. Robin and N. Endersby (Dept. Genetics, University of Melbourne, Australia), Gordon Luikart (Montana Conservation Genetics Laboratory, The University of Montana, USA), Paul Sunnucks (School of Biological Sciences, Monash University, Melbourne, Australia), and Karl Gordon (CSIRO Ecosystem Sciences, Canberra, Australia) provided helpful discussions during the course of the project and/or critically read earlier versions of this manuscript. Field populations of Helicoverpa armigera were collected by Sandhya Kranthi (Central Institute for Cotton Research, Nagpur, India), Rahul Wadaskar (International Crops Research Institute for the Semi-Arid Tropics, Patancheru, India) and Vilas Tajane (Jain Irrigation, Jalgaon, India).
Contributed overall knowledge of the Indian agricultrual landscape: KRK. Conceived and designed the experiments: DAR KRK PB. Performed the experiments: GTB WTT. Analyzed the data: GTB WTT. Contributed reagents/materials/analysis tools: PB KRK DAR WTT. Wrote the paper: GTB WTT DAR PB.
- 1. Fitt GP (1989) The ecology of Heliothis species in relation to agroecosystems. Ann Rev Entomol 34: 17–52.
- 2. Manjunath TM, Bhatnagar VS, Pawar CS, Sithanantham S (1989) Economic importance of Heliothis spp. in India and an assessment of their natural enemies and host plants; 11–15 November, 1989; New Delhi, India. pp. 197–228.
- 3. Singh J, Bains SS (1986) Role of plants in population build up of Heliothis armigera (Hübner) in Punjab. Indian J Ecol 13: 113–119.
- 4. Bhatnagar VS (1980) A report on research on the Heliothis complex at ICRISAT (India) 1974–79; 24–26 April, 1980; Udaipur, India.
- 5. Jayaraj S (1982) Biological and Ecological studies of Heliothis. In: Reed W, editor, 15–20 November 1981; ICRISAT Center, Patancheru, A.P., India. pp. 17–28.
- 6. Kranthi KR, Jadav DR, Wanjari RR, Shaki-Ali S, Russell DA (2001) Carbamate and organophosphate resistance in cotton pests in India, 1995–1999. Bull Entomol Res 91: 37–46.
- 7. Sachan JN (1987) Status of gram pod borer H. armigera (Hb.) in India and its management. In: Mathour YK, editor. Recent Advances in Entomology. Parade, Kanpur: Gopal Parkashan. pp. 91–110.
- 8. Singh J, Sandhu SS, Singla ML (1990) Ecology of Heliothis armigera (Hub.) on chickpea in Punjab. J Insect Sci 3: 47–52.
- 9. Singh J, Sidhu AS (1992) Present status of Helicoverpa armigera in cotton and strategies for its management in Punjab, Haryana and Rajasthan. In: Sachan JN, editor. Directorate of Pulse Research, Kanpur. pp. 92–98.
- 10. Durairaj C, Subbaratnam GV, Singh TVK, Shannower TG (2005) Helicoverpa in India: Spatial and temporal dynamics and management options. In: Sharma H, editor. Heliothis/Helicoverpa management – Emerging trends and strategies for future research Oxford&IBH Publishing Co, New Delhi. pp. 91–117.
- 11. Verma AK, Sankhyan S (1993) Pheromone monitoring of Heliothis armigera (Hubner) and relationship of moth activity with larval infestation on important cash crops in mid-hills of Himachal Pradesh. Pest Manag Eco Zool 1: 43–49.
- 12. Pedgley DE, Tucker MR, Pawar CS (1987) Windborne migration of Heliothis armigera (Hübner) (Lepidoptera: Noctuidae) in India. Int J Trop Insect Sci 8: 599–604.
- 13. Bhattacherjee NS (1972) Heliothis armigera (Hübner) a polytypic species. Ent Newsl 2: 3–4.
- 14. Reed W, Pawar CS (1982) Heliothis: A global problem. In: Reed W, editor. 15–20 November 1981; ICRISAT Center, Patancheru, A.P., India. pp. 9–14.
- 15. Jallow MFA, Cunningham JP, Zalucki MP (2004) Intra-specific variation for host plant use in Helicoverpa armigera (Hübner) (Lepidoptera: Noctuidae): implications for management. Crop Prot 23: 955–964.
- 16. Kranthi KR, Armes NJ, Rao NGV, Sheo R, Sundarmurthy VT (1997) Seasonal dynamics of metabolic mechanisms mediating pyrethroid resistance in Helicoverpa armigera in central India. Pestic Sci 50: 91–98.
- 17. Armes NJ, Jadhav DR, DeSouza KR (1996) A survey of insecticide resistance in Helicoverpa armigera in the Indian subcontinent. Bull Entomol Res 86: 499–514.
- 18. Madden AD, Holt J, Armes NJ (1995) The role of uncultivated hosts in the spread of pyrethroid resistance in Helicoverpa armigera populations in Andhra Pradesh, India: a simulation approach. Ecol Model 82: 61–74.
- 19. Armes NJ, Jadhav DR, Bond GS, King ABS (1992) Insecticide resistance in Helicoverpa armigera in South India. Pestic Sci 43: 355–364.
- 20. De Souza K, Holt J, Colvin J (1995) Diapause, migration and pyrethroid-resistance dynamics in the cotton bollworm, HeIicoverpa armigera (Lepidoptera: Noctuidae). Ecol Entomol 20: 333–342.
- 21. Tamhankar AJ, Rajendran TP, Hariprasad Rao N, Lavekar RC, Jeyakumar P, et al. (2003) Variability in response of Helicoverpa armigera males from different locations in India to varying blends of female sex pheromone suggests male sex pheromone response polymorphism. Curr Sci 84: 448–450.
- 22. Zalucki MP, Cunningham JP, Downes S, Ward P, Lange C, et al. (2012) No evidence for change in oviposition behaviour of Helicoverpa armigera (Hübner) (Lepidoptera: Noctuidae) after widespread adoption of transgenic insecticidal cotton. Bull Entomol Res 102: 468–476.
- 23. Chowda-Reddy RV, Kirankumar M, Seal SE, Muniyappa V, Valand GB, et al. (2012) Bemisia tabaci phylogenetic groups in India and the relative transmission efficacy of Tomato leaf curl Bangalore virus by an indigenous and an exotic population. J Integrative Agric 11 (2):235–248.
- 24. Zhou X, Faktor O, Applebaum SW, Coll M (2000) Population structure of the pestiferous moth Helicoverpa armigera in the eastern Mediterranean using RAPD analysis. Heredity 85: 251–256.
- 25. Nibouche S, Buès R, Toubon J-F, Poitout S (1998) Allozyme polymorphism in the cotton bollworm Helicoverpa armigera (Lepidoptera: Noctuidae): comparison of African and European populations. Heredity 80: 438–445.
- 26. Behere GT, Tay WT, Russell DA, Heckel DG, Appleton BR, et al. (2007) Mitochondrial DNA analysis of field populations of Helicoverpa armigera (Lepidoptera: Noctuidae) and of its relationship to H. zea. BMC evolutionary biology 7: 117.
- 27. Vassal JM, Brevault T, Achaleke J, Menozzi P (2008) Genetic structure of the polyphagous pest Helicoverpa armigera (Lepidoptera: Noctuidae) across the Sub-Saharan cotton belt. Commun Agric Appl Biol Sci 73: 433–437.
- 28. Endersby NM, Hoffmann AA, McKechnie SW, Weeks AR (2007) Is there genetic structure in populations of Helicoverpa armigera from Australia? Entomol Exp Appl 122: 253–263.
- 29. Daly JC, Gregg P (1985) Genetic variation in Heliothis in Australia: species identification and gene flow in the two pest species H. armigera (Hübner) and H. punctigera Wallengren (Lepidoptera: Noctuidae). Bull Entomol Res 75: 169–184.
- 30. McKechnie SW, Spackman ME, Naughton NE, Kovacs IV, Ghosn M, et al. (1993) Assessing budworm population structure in Australia using the A-T rich region of mitochondrial DNA; New Orleans. Proceedings of Beltwide Cotton Conference. pp. 838–840.
- 31. Stokes NH, McKechnie SW, Forrester NW (1997) Multiple allelic variation in sodium channel gene from populations of Australian Helicoverpa armigera (Hübner) (Lepidoptera: Noctuidae) detected via temperature gel electrophoresis. Aust J Entomol 36: 191–196.
- 32. Scott KD, Lawrence N, Lange CL, Scott LJ, Wilkinson KS, et al. (2005) Assessing moth migration and population structuring in Helicoverpa armigera (Lepidoptera: Noctuidae) at the regional scale: example from the Darling Downs, Australia. J Econ Entomol 98: 2210–2219.
- 33. Scott KD, Wilkinson KS, Lawrence N, Lange CL, Scott LJ, et al. (2005) Gene-flow between populations of cotton bollworm Helicoverpa armigera (Lepidoptera: Noctuidae) is highly variable between years. Bull Entomol Res 95: 381–392.
- 34. Scott KD, Wilkinson KS, Merritt MA, Scott LJ, Lange CL, et al. (2003) Genetic shifts in Helicoverpa armigera Hübner (Lepidoptera: Noctuidae) over a year in the Dawson/Callide Valleys. Aust J Agr Res 54: 739–744.
- 35. Scott LJ, Lawrence N, Lange CL, Graham GC, Hardwick S, et al. (2006) Population dynamics and gene flow of Helicoverpa armigera (Lepidoptera: Noctuidae) on cotton and grain crops in the Murrumbidgee Valley, Australia. J Econ Entomol 99: 155–163.
- 36. Scott KD, Lange CL, Scott LJ, Gahan LJ (2004) Isolation and characterization of microsatellite loci from Helicoverpa armigera Hübner (Lepidoptera: Noctuidae). Mol Ecol Notes 4: 204–205.
- 37. Ji Y-J, Zhang D-X, Hewitt GM, Kang L, Li D-M (2003) Polymorphic microsatellite loci for the cotton bollworm Helicoverpa armigera (Lepidoptera: Noctuidae) and some remarks on their isolation. Mol Ecol Notes 3: 102–104.
- 38. Weeks A, Endersby NM, Lange CL, Low A, Zalucki MP, et al. (2010) Genetic variation among Helicoverpa armigera populations as assessed by microsatellites: a cautionary tale about accurate allele scoring. Bull Entomol Res 100: 445–450.
- 39. Tay WT, Behere GT, Batterham P, Heckel DG (2010) Generation of microsatellite repeat families by RTE retrotransposons in lepidopteran genomes. BMC Evol Biol 10: 144.
- 40. Lessa EP (1992) Rapid survey of DNA sequence variation in natural populations. Mol Biol Evol 9: 323–330.
- 41. Palumbi SR, Baker CS (1994) Contrasting population structure from nuclear intron sequences and mtDNA of Humpback whales. Mol Biol Evol 11: 426–435.
- 42. Tay WT, Behere GT, Heckel DG, Lee SF, Batterham P (2008) Exon-Primed Intron-Crossing (EPIC) PCR markers of Helicoverpa armigera (Lepidoptera: Noctuidae). Bull Entomol Res 98: 509–518.
- 43. Zraket CA, Barth JL, Heckel DG, Abbott AG (1989) Genetic linkage mapping with restriction fragment length polymorphisms in the tobacco budworm, Heliothis virescens. In: Hagedorn HH, Hildebrand JG, Kidwell MG, Law JH, editors, 22–27 October 1989; Tucson, Arizona. Plenum Press. pp. 13–20.
- 44. Behere GT, Tay WT, Russell DA, Batterham P (2008) Molecular markers to discriminate among four pest species of Helicoverpa (Lepidoptera: Noctuidae). Bull Entomol Res 98: 599–603.
- 45. Weir BS, Cockerham CC (1984) Estimating F-statistics for the analysis of population structure. Evol 38: 1358–1370.
- 46. Goudet J (1995) FSTAT (version 1.2): A computer program to calculate F-statistics. J Hered 86: 485–486.
- 47. Raymond M, Rousset F (1995) GENEPOP (version 1.2): Population genetics software for exact tests and ecumenicism. J Hered 86: 248–249.
- 48. Rice WR (1989) Analyzing tables of statistical tests. Evol 43: 223–225.
- 49. Kranthi KR, Jadhav DR, Kranthi S, Wanjari RR, Ali SS, et al. (2002) Insecticide resistance in five major insect pests of cotton in India. Crop Protec 21: 449–460.
- 50. Kranthi S, Dhawad CS, Naidu S, Bharose A, Chaudhary A, et al. (2009) Susceptibility of the cotton bollworm, Helicoverpa armigera (Hubner) (Lepidoptera: Noctuidae) to the Bacillus thuringiensis toxinCry2Ab before and after the introduction of Bollgard-II. Crop Protec 28: 371–375.
- 51. Excoffier L, Smouse PE, Quattro JM (1992) Analysis of molecular variance inferred from metric distances among DNA haplotypes: application to human mitochondrial DNA restriction data. Genet 131: 479–491.
- 52. Schneider S, Roessli D, Excoffier L (2000) Arlequin: A Software for Population Genetics Data Analysis, Version 2.000. 2.000 ed. Geneva: Genetics and Biometry Laboratory, Dept. of Anthropology, University of Geneva, Switzerland.
- 53. Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus genotype data. Genet 155: 945–959.
- 54. Evanno G, Regnaut S, Goudet J (2005) Detecting the number of clusters of individuals using the software Structure: a simulation study. Mol Ecol 14: 2611–2620.
- 55. Chapuis M-P, Estoup A (2007) Microsatellite null alleles and estimation of population differentiation. Mol Biol Evol 24: 621–631.
- 56. Carlsson J (2008) Effects of microsatellite null alleles on assignment testing. J Hered 99: 616–623.
- 57. Gordon K, Tay WT, Collinge D, Williams A, Batterham P (2009) Genetics and molecular biology of the major crop pest genus Helicoverpa. In: Goldsmith MR, Marec F, editors. Molecular Biology and Genetics of the Lepidoptera: CRC Press. pp. 219–238.
- 58. Johnson MS, Black R (1984) The Wahlund effect and the geographical scale of variation in the intertidal limpet Siphonaria sp. Marine Biol 79: 295–302.
- 59. De Barro PJ (2005) Genetic structure of the whitefly Bemisia tabaci in the Asia-Pacific region revealed using microsatellite markers. Mol Ecol 14: 3695–3718.
- 60. Brown JE, McBride CS, Johnson P, Ritchie S, Paupy C, et al. (2011) Worldwide patterns of genetic differentiation imply multiple ‘domestications’ of Aedes aegypti, a major vector of human diseases. Proc Biol Sci 278: 2446–2454.
- 61. Nielsen EE, Hansen MM, Ruzzante DE, Meldrup D, Gronkjaer P (2003) Evidence of a hybrid-zone in Atlantic cod (Gadus morhua) in the Baltic and the Danish Belt Sea revealed by individual admixture analysis. Mol Ecol 12: 1497–1508.
- 62. Stipanovic RD, Lopez JD Jr, Dowd MK, Puckhaber LS, Duke SE (2006) Effect of racemic and (+)- and (−)-gossypol on the survival and development of Helicoverpa zea larvae. J Chem Ecol 32: 959–968.
- 63. Kranthi KR, Kranthi NR (2004) Modelling adaptability of cotton bollworm, Helicoverpa armigera (Hübner) to Bt-cotton in India. Curr Sci 87: 1096–1107.
- 64. Manjunath TH, Phalak VR, Subramanian S (1970) First records of egg parasites of Heliothis armigera (Hubner) (Lepidoptera: Noctuidae) in India. Comm Inst Biol Contr Tech Bull 13: 111–115.
- 65. Zalucki MP, Furlong MJ (2005) Forecasting Helicoverpa populations in Australia: A comparison of regression based models and a bio-climatic based modeling approach. Insect Sci 12: 45–56.
- 66. Fakrudin B, Prakash SH, Krishnareddy KB, Vijaykumar , Badari Prasad PR, et al. (2004) Genetic variation of cotton bollworm, Helicoverpa armigera (Hübner) of South Indian cotton ecosystem using RAPD markers. Curr Sci 87: 1654–1657.
- 67. Khadi BM, Kulkarni VN, Katageri IS, Vamadevaiah HM, Maralappanavar M (2003) Strengths, weaknesses, opportunities and threats (SWOT) analysis of Bt cotton in India. In: Swanepoel A, editor. 9–13 March 2003; Cape Town, South Africa. Agricultural Research Council - Institute for Industrial Crops. pp. 393–406.
- 68. Ravi KC, Mohan KS, Manjunath TM, Head G, Patil BV, et al. (2005) Relative abundance of Helicoverpa armigera (Lepidoptera: Noctuidae) on different host crops in India and the role of these crops as natural refuge for Bacillus thuringiensis cotton. Env Entomol 34: 59–69.
- 69. Directorate of Economics and Statistics, State Wise Land Use Statistics, Department of Agriculture and Cooperation, Ministry of Agriculture, Government of India (2007) Section 9. Area under crops (pulses, food grains & sugarcane) 1996–97 & 2005–06; Section 14. Area under crops (fibers) 1996–97 & 2005–06. Available: http://eands.dacnet.nic.in/state.htm.
- 70. Firempong S, Zalucki MP (1990) Host plant preferences of populations of Helicoverpa-Armigera (Hubner) (Lepidoptera, Noctuidae) from different geographic locations. Aust J Zool 37: 665–673.
- 71. Jallow MFA, Zalucki MP (1996) Within- and between-population variation in host-plant preference and specificity in Australian Helicoverpa Armigera (Hubner) (Lepidoptera: Noctuidae). Aust J Zool 44: 503–519.
- 72. Endersby NM, McKechnie SW, Ridland PM, Weeks AR (2006) Microsatellites reveal a lack of structure in Australian populations of the diamondback moth, Plutella xylostella (L.). Mol Ecol 15: 107–118.
- 73. Franck P, Reyes M, Olivares J, Sauphanor B (2007) Genetic architecture in codling moth populations: comparison between microsatellite and insecticide resistance markers. Mol Ecol 16: 3554–3564.