POPULATION STRUCTURE OF Spodoptera frugiperda COLLECTED IN MAIZE FROM DIFFERENT BRAZILIAN GEOGRAPHIC REGIONS

- Spodoptera frugiperda is the most economically important maize pest in Brazil. There is little information about the genetic structure, using SSR markers, of S. frugiperda populations collected from maize crops. In this study, 21 SSR markers were used to evaluate the genetic diversity and population structure of S. frugiperda collected from distinct Brazilian geographical regions. Two hundred and twenty-seven alleles were recorded with an average of 10.76 per marker, and Polymorphic Information Content (PIC) values ranging from 0.242 to 0.933, with an average of 0.621, indicating a high discriminating power. The overall F ST , 0.061, indicated a moderate genetic differentiation among the S. frugiperda populations collected from maize, and the AMOVA showed that 87.36% of the genetic variation is within populations. The Mantel test showed significant correlation between genetic and geographic distances. The genetic data demonstrated that all individuals from the six sampling sites were structured as two sub-populations, being one of them composed only by the CL population, collected in the Rio Grande do Sul state. The knowledge about genetic diversity and population structure of S. frugiperda is important for the development of strategies for the insect pest management and monitoring systems, especially for the differentiated CL population.

In the larval stage, the fall armyworm feeds mainly from the whorl of young plants and may cause losses over 30% in the Brazilian maize production (Cruz, 1995). However, data about reductions in crop yield are still scarce, as these depend on factors such as environment, cultivar, agricultural practices, and especially the developmental stage and nutritional status of plants. Furthermore, the voracity of the fall armyworm strain is a factor that can influence the losses in production (Cruz, 1995;Busato et al., 2008), mainly considering the breakdown of resistance to Bt proteins in the country (Farias et al., 2014), when the maize crops were 80% Bt transgenic plants (Celeres, 2013). Nagoshi et al. (2008) mentioned that a more detailed understanding of fall armyworm population movements will facilitate efforts to find more accurate ways to predict the timing and severity of infestations. Some studies have shown the existence of two strains that inhabit the same area in the United States, occurring at the same time and tending to use different hosts (Pashley, 1986). The "rice strain" was found feeding on rice, Brachiaria and other grasses, while the "maize strain" was found feeding on maize and cotton (Pashley, 1993). These strains were also identified in other countries, including Brazil (Busato et al., 2008). These authors reported the importance of S. frugiperda strains to the economical entomology, as they may respond differentially to strategies of control. Salinas-Hernandez et al. (2011) found no evidence for strains of S. frugiperda due to the host under Colombia conditions. However, Juarez et al. (2012) working with individuals of this species collected in Brazil, Argentina and Paraguay, detected the existence of mitochondrial "rice" and "maize" host haplotypes and identified that the distribution of the rice haplotype was apparently more widespread, being found in other hosts such as alfalfa.
The identification and characterization of the genetic variation among insect populations is an important issue to the development of strategies for pest management. However there is a lack of knowledge about the genetic variation of S. frugiperda collected in maize from distant regions of Brazil using SSR (Simple Sequence Repeats) markers, which are highly polymorphic and abundant throughout all eukaryotic genomes (Goldstein and Scholötterer, 1999). SSR has been the most widely applied class of molecular markers used in genetic studies (Ellegren 2004), because of its codominance, multiallelism, high reproducibility and PCR-based reaction (Oliveira et al., 2006).
To have success using practices for the integrated pest management and control of S. frugiperda in maize, it is important to characterize populations and the dispersion pattern of this pest (Busato et al., 2004b). In this context, the objective of this study was to investigate the genetic diversity and population structure of fall armyworm from six different maize producing areas over a large geographic area of Brazil.

Insect collection
The larvae were collected randomly in equidistant points inside the non-transgenic maize fields in six Brazilian regions ( Figure 1) (collection permit emitted by Brazilian Institute of Environment -IBAMA/SISBIO # 35100). One hundred individuals were collected from each region for later selection, aiming to maintain representation of the population.
For transportation each individual was placed in plastic cups of the 50 ml, containing adapted diet for rearing S. frugiperda (Kasten Jr et al., 1978).
In the lab, the larvae was stored in ethanol 99,5% (v/v) and kept at 4ºC. The ethanol was changed daily for three days to avoid DNA degradation due to dilution by water released from the larvae (Souza et al., 2013). After that, the fall armyworm larvae were stored in a -20ºC freezer up to the DNA extraction. The identification of the collected larvae was performed according to Luginbill (1928) description.

DNA extraction and SSR markers amplification
Larvae in the fourth instar or above were selected and the genomic DNA extracted according to Rogers and Bendich (1988) with some modifications (Souza et al., 2013). Twelve
PCR analyses were carried out with three primers: a sequence-specific forward primer with M13 (19 nt) tail at its 5' end, a sequence-specific reverse primer, and the universal fluorescent-labeled M13 primer with 6-carboxy-fluorescein (FAM). The M13labelled primer and the reverse primer were in excess over the forward primer, which is limited. This allows the forward M13-tailed primer and reverse primer to initiate the reaction and, when the limited primer is depleted, the labelled primer takes the place of the limited forward primer in the remaining PCR cycles (Barkley et al., 2007;Schuelke, 2000).

Descriptive statistics
Monomorphic primers and the ones exhibiting problems in the amplification were removed from data analysis. For each SSR marker, major allele frequency, allele number, variability, gene diversity, heterozygosity and PIC (Polymorphism Information Content) values were calculated using the PowerMarker software (Liu and Muse, 2005). PIC is defined as the measure of the discrimination power and informativeness for an SSR marker (Botstein et al., 1980).

Clustering and population structure analyses
Clustering analysis was performed through the Ward's method (Ward 1963), based on the Rogers' distance (Rogers, 1972), using the package ade4 (Dray et al. 2007) available in the R software (R Core Team, 2013). Population structure analysis was carried out through the software STRUCTURE (Pritchard et al., 2000), considering a burn-in period of 250,000 and 500,000 iterations with five replicates, assuming admixture model, noninformative priors and correlated allele frequencies, with a number of groups (k) ranging from 1 to 10. The putative number of subpopulations was determined based on the ΔK statistics, which corresponds to a measure of the second-order rate of change in the likelihood between successive K values (Evanno et al., 2005).

Mantel test for matrices correlation
The correlation between the genetic distance (Rogers' distance) and the geographic distance matrices was verified through the Mantel test (Mantel, 1967) implemented in the software Genes (Cruz, 2013), considering 10,000 simulations.

F-statistics
The Wright's F-statistics (F IS, F ST and F IT - Wright, 1951), as well as the expected and observed heterozygosity per population, were estimated using the package hierfstat (Goudet, 2005) available in the R software (R Core Team, 2013). The 95% confidence intervals for the F IS coefficient were calculated based on 1,000 bootstrap replications. The observed and expected heterozygosity were calculated based on the average of all markers in each population. The analysis of molecular variance (AMOVA) was performed through the package pegas for R (Paradis, 2010). Gene flow (N m ) was (Wright, 1951).

Polymorphism and allele frequency analysis of microsatellite loci
Of the 27 microsatellite (SSR) markers tested, four of them were monomorphic (Stv_Spf688, Stv_Spf1068, Stv_Spf1890 and Stv_Spf1856) and two (Stv_Spf1409 and Stv_Spf1552) did not have satisfactory PCR amplicons. In total, the 21 polymorphic markers (Table 1)  * The sequence GTTT at 5' end of reverse primers was added to promote non-template adenylation (Brownstein et al, 1996). The SSR markers were selected from Arias et al (2011).   (Table 3).  populations. An F ST index, ranging from 0 to 1, is an estimate of gene differentiation among populations, which represents a genetic variation among fall armyworm populations (Nei, 1973). Through the clustering and population structure analyses, based on the selected SSR markers, it was possible to distinguish the six populations and the two groups, respectively ( Figures 3A and 3B).
The major group consisted of TE, SG, VM, NP and CA populations, and the other one included only the CL population. The same result was also observed through the population structure analysis performed in the software STRUCTURE (Pritchard et al., 2000). Following the method proposed by Evanno et al. (2005), the maximum value of ΔK was observed for K= 2, i.e. the existence of two putative subpopulations ( Figure 3B), with one of them composed only by CL population, with influx of alleles from just one immigrant. In the major group the population structure analysis, quantifying how likely

Correlation between genetic distance and geographic distance
The correlation between the geographic and genetic distance matrices was significant at 5% of probability, accessed by the Mantel test, revealing a significant effect of the geographic isolation in the genetic variability observed among populations of S. frugiperda. The estimated genetic distance between the six populations presented slightly higher values for the pairs containing CL or TE (Table   5). These two populations, which were located at geographically distant areas in Brazil, South and Northeast, respectively, revealed a trend to increase the genetic differentiation as consequence of the increase in the distance of isolation. However, only for Our results suggested that for this population the geographical distance could be one of the limitations for the gene flow, as seen in the Figure 3B, with just one immigrant individual in population CL.

The intensification of production systems in
Brazil has allowed the cultivation of two and even three crops a year, favoring polyphagous pests like S. frugiperda to find suitable conditions for their maintenance in the field throughout the year. This sequential system allows S. frugiperda to have host plants over the year, reducing the adaptive value of migration (Farias et al., 2014). In face of an efficient management practices, it is necessary to know the genetic complexity and structure of the pest population, to delay the evolution of resistance to any control method (Tabashnik, 1991). Many studies have been conducted on genetic characterization of the fall armyworm, using strains from different crops, for example, rice, maize and cotton. Although these studies are commonly performed based on a limited sample size and a restricted number of molecular markers, small genetic variation among geographic regions has been found (Belay et al., 2012). It is important to know the genetic structure of fall armyworm from different maize producing areas before the use of large-scale efforts to control insect pests (Martinelli et al., 2007). Thus, the results of this study will contribute to improve the strategies for the management and control of pests in maize crops.
Here, in this study, 21 SSR markers selected from a first report of sequence-specific SSR for fall armyworm, developed by Arias et al (2011), were used for genotyping the individuals of six populations collected in distinct geographic Brazilian regions.
Microsatellites were selected due to its highly polymorphic content, abundance across the genome and codominance (Weber 1990). These markers were polymorphic and highly informative ( in both directions, they concluded that these two strains were in the initial process of speciation and represent strains associated with host. Besides that, they mentioned that was evident the nonpreference of these two strains for irrigated rice, barnyardgrass and sorghum in relation to maize (Busato et al., 2004b). The dispersion capability, the geographic barriers and other related process can also influence the population structure of a given species (Avise et al., 1987).
In our study, the majority of the genetic variability was observed within population and Our study is the first one to apply the SSR markers developed by Arias et al (2011) (Farias et al., 2014). Moreover, the frequent genetic admixture of immigrant individuals into populations of S. frugiperda has the potential to influence the strategies for the management of an insect resistance (Roush and Daly, 1990). Also, the regular application and heavy selection pressure of insecticides has conducted to the development of resistant fall armyworm in many regions (Yu 1991, Berta et al. 2000. In both situations, the knowledge about the genetic variability, structure and migration in armyworm populations will assist in the development of adequate management strategies, contributing to the preservation of these technologies. The identification of two populations can contribute to the IPM (Integrated Pest Management) strategies, with more specific control tools, especially for the Capão do Leão in the extreme south region.
In our study, the population structure of maize fall armyworm suggested the presence of two major groups showing significant correlation with Revista Brasileira de Milho e Sorgo, v.14, n.3, p. 300-315, 2015 Versão impressa ISSN 1676-689X / Versão on line ISSN 1980-6477 -http://www.abms.org.br their geographic sampling location in Brazil. These results reinforce the importance for differentiated management strategies in relation to these population groups and also, for further studies using SSR markers to evaluate the genetic variability and structure of the fall armyworm, from more maize regions over the crop seasons.