Genetic Diversity and Fine-Scale Genetic Structure of Spodoptera litura Fabricius (Lepidoptera: Noctuidae) in Southern China Based on Microsatellite Markers

Simple Summary Spodoptera litura is an important agricultural pest widely distributed around the world, which can damage numerous crops. Despite its economic importance, research on the population genetics of S. litura has been limited, preventing insight into the spread of the pest. In this study, we conducted a population genetic analysis of 24 populations of S. litura in Southern China using microsatellite markers, demonstrating the information on the genetic diversity and genetic structure of the tested populations. The results will provide pest control management strategies for S. litura in Southern China. Abstract Population genetic structure is strongly affected by dispersal events, especially for migratory species. The investigation of population structure is therefore conducive to increasing our understanding of species dispersal. Spodoptera litura (Fabricius) (Lepidoptera: Noctuidae) is an important tobacco pest in China causing serious damage to multiple crops. In this study, we explore its dispersal dynamics by clarifying the fine-scale population genetics using 545 S. litura samples collected from tobacco plantations at 24 locations (mainly in Baise, Hechi, and Hezhou, Southern China). We analyzed the genetic diversity, genetic structure, and gene flow of these populations using seven microsatellite loci. Our results revealed high genetic diversity and low population genetic structure among S. litura. The genetic distance was uncorrelated with geographical distance, indicating the complete randomness of dispersal among the local populations. Our results suggest that the movement scope of contemporary S. litura might be much higher than the local-level spatial scale, which will provide a theoretical basis for pest management.


Introduction
Population genetics is the study of changes in gene frequency and genotype frequency in a biological population under the influence of natural selection, genetic drift, mutation, and migration [1]. Population genetic structure is the main research subject of population genetics, which refers to the distribution pattern of genetic variations among and within populations [2]. Many insects exhibit unique dispersal patterns. Some, like Grapholita molesta [3] and Cydia pomonella [4], are weak flying insects that can only move in a small range, whereas other insects such as Mythimna seperata [5], Leptinotarsa decemlineata [6], and Spodoptera exigua [7] with extremely strong flying ability, are not affected by external environmental pressures and are often able to migrate long distances. However, the longdistance dispersal of insects promotes gene flow, evolutionary potential, and the spread of   [25] was used to evaluate null alleles and correct genotypic errors. Deviation from Hardy-Weinberg equilibrium (HWE) [26] at each locus in populations was calculated using the program GenAlEx 6.5 [27]. Linkage disequilibrium (LD) between pairs of loci was performed in GENEPOP 4.6 [28]. The number of alleles (N a ), polymorphism information content (PIC), and Shannon's index (I) were estimated using Microsatellite Tools (Trinity College Dublin, Ireland) and GenAlEx. Observed heterogeneity (H O ), expected heterozygosity (H E ), and allelic richness (A R ) were assessed using the hierfstat R package [29] [30] 30 was used to assess the genetic structure of populations based on the Bayesian model. The candidate number of genetic clusters (K) was set as 1 to 10 with 10 repeat runs and a burn-in period of 100,000 and 1,000,000 Monte Carlo Markov Chains. The results were submitted to Structure Harvester Web 0.6.93 [31] to confirm the optimal K value by the ∆K method [32]. The membership coefficient matrices (Qmatrices) of repeated runs associated with the optimal K were integrated using CLUMPP 1.12 [33], and the results were visualized using DISTRUCT 1.1 [34]. In addition, a Principal Coordinates Analysis (PCoA) was performed based on Codom−Genotypic genetic distance among 545 S. litura populations using GenAlEx.
The analysis of molecular variance (AMOVA) was calculated by ARLEQUIN 3.5 [35] to assess the distribution of genetic variance between populations, individuals, and groups. The F-statistics (F st ) and the Nei's genetic distance (D') per geographical population were estimated using Fstat [36] and Popgene32 [37], respectively.
The geographical distance among different populations was calculated based on latitude and longitude. A Mantel test, based on geographic distances and linearized pairwise F ST (i.e., F ST /1 − F ST ) [38], was conducted to explore the effect of isolation by distance (IBD). To test the effect of IBD in different spatial ranges, we also performed a spatial autocorrelation analysis using the PopGenReport R package [39]. We divided geographic distances into several bins with a step distance of 45 km and independently tested the correlations with corresponding genetic distances in the subset bins.

Gene Flow
BAYESASS 3.0 [40] with Markov chain Monte Carlo method was used to evaluate the modern migration rates among S. litura populations. The program was run with 10,000,000 iterations, discarding the first 1,000,000 iterations, and sampling every 1000 iterations from the remaining 9,000,000 iterations, producing a sample of 9000 observations from the chain that will be used to estimate parameters.

Genetic Diversity of Populations
In all loci-population pairs, only 42 of 168 have significantly deviated from Hardy-Weinberg equilibrium, which was caused by heterozygote deficiency. Linkage disequilibrium occurred in 3 of 21 locus-locus pairs, including CWM−1 versus CWM−4, CWM−1 versus CWM−8, and CWM−7 versus CWM−9. However, no loci had significant global deviation linkage disequilibrium, so the independence of the loci was sufficient. The genetic diversity of 24 populations of S. litura is displayed in Table 2. The number of alleles (N a ) ranged from 4.14 to 6.86 with an average of 5.73. The mean number of observed heterogeneity (H O ) and expected heterozygosity (H E ) were 0.4793 and 0.5970, respectively. The average of Shannon's index (I) was 1.2342. The highest allele richness (A R ) was 5.7783 in ZB population, while the lowest A R was 4.1429 in SC population. Polymorphism information contents (PIC) of all populations were greater than 0.5.

Population Genetic Structure
Based on the Bayesian cluster analysis using STRUCTURE software, the optimal genetic clusters K = 3 was detected ( Figure 2a). However, there were no clear geographical grouping patterns for all populations (Figure 2b). Such mixed patterns also occur when K equals values other than 2-10. Similarly, the principal coordinate analysis (PCoA) also had undefined grouping patterns, showing that different geographical populations were mostly mixed. These results indicated that the clustering pattern was not related to sampling locations ( Figure 3).

Population Differentiation
Based on the AMOVA, it showed a 1.92% of variation among populations, 20.02% of variation among individuals within populations, and 78.06% of variation among individuals (Table 3). These results suggest that genetic variation mainly occurred within individuals. Two genetic distances measured showed a low level of differentiation among all populations (Table 4)

Population Differentiation
Based on the AMOVA, it showed a 1.92% of variation among populations, 20.02% of variation among individuals within populations, and 78.06% of variation among individuals (Table 3). These results suggest that genetic variation mainly occurred within individuals. Two genetic distances measured showed a low level of differentiation among all populations (Table 4)  The Mantel test of F ST /(1 − F ST ) against natural log-transformed geographic distance (Ln) between populations revealed that genetic differentiation was weak-positively correlated with geographic distance (r = 0.0145, p = 0.4) (Figure 4). Spatial auto-correlation analysis showed a very weak positive correlation at 89 km and 622 km (Table 5).

Gene Flow
As shown (Table 6), we marked the migration rates of all populations by color shades. The results indicated that the range of recent migration rates for S. litura populations was 0.0061-0.1536. The self-assignment rate in the KX population was the highest (0.7868), and the two highest migration rates were calculated from LG to KX population. In addition, we found that most populations tend to migrate to Chengbei regions (CB).

Discussion
Owing to the neutral and high mutation rates of microsatellite markers, they are usually used to explore the level of genetic diversity and population differentiation within species [41,42]. In this study, the microsatellite molecular technique was used to explore 24 geographical populations of S. litura in Southern China, showing high genetic diversity and low genetic population structure.

Genetic Diversity
The higher the genetic diversity of the species, the stronger it's potential to adapt to environmental changes [43,44]. In this study, the mean H E for the S. litura population was 0.5970 and the average allelic richness (A R ) of the S. litura populations was 5.1439, showing a high level of population genetic diversity. For other S. litura populations in China, genetic diversity was found to be generally high (H E from 0.19 to 0.89) by Wu et al., which suggested that the rich genetic diversity was mainly due to frequent individual migration [23]. In addition, the excess of heterozygotes indicated that the strong ability of S. litura migration avoided severe bottlenecks and loss of rare alleles. We further suggested the importance of high-level gene flow in maintaining local population diversity, which may be one of the reasons for population outbreaks and insecticide resistance [45].

Genetic Structure and Population Differentiation
Based on the Bayesian clustering analysis, 24 populations in Southern China formed three clusters. However, there was no obvious geographic division pattern in the distribution of these three clusters, indicating that the geographic factors had little influence on the migration of S. litura in the study area. The PCoA results also provided similar evidence. Such patterns were observed in Helicoverpa gelotopoeon [46] and Sitobion avenae [47], where the population structure was unrelated to geographical distribution, which is speculated to be formed through extensive migration. Indeed, continued long-term dispersal can gradually homogenize populations, leading to a reduction in population structure [48]. This finding was inconsistent with Scott et al. [49] and Domingues et al. [50]. These results may be attributed to the long-distance dispersal and migration of insects leading to a low genetic divergence in population structure, forming a panmictic population.
Frequent migrations may offset the effects of genetic drift and mutation, thus reducing population genetic differentiation [51]. Our study showed that low genetic differentiation existed in the S. litura populations based on the local-level spatial scale. Among them, genetic differentiation did not exist in XH and PH populations, even though they were more than 600 km apart. Similarly, Gandhi and Patil [21] found high similarities between the Hyderabad and Indore populations of S. litura in India, which were about 670 km apart. Moreover, according to Wu [22], genetic undifferentiating was found among the 18 populations of S. litura in China, although the distance between the two farthest populations (Kunming and Shenyang populations) was about 2600 km. Therefore, we speculated that the genetic differentiation of S. litura may increase at a distance of more than 2600 km. How does our research on genetic differentiation measure up with other moths with high dispersal rates? Normally, moths with high dispersal rates exhibit lower F ST . We discovered that there was no genetic divergence between populations at similar or much larger geographic scales than we studied. For example, Franklin et al. [52] also observed that populations of Trichoplusia ni on the west coast of North America remained linked at distances greater than 600 km. However, some species began to diverge at scales greater than the national-level spatial scale. Lyons et al. [53] reported high genetic similarity between populations of the emperor butterfly in eastern and western North America, but significant genetic differentiation between them and populations in Hawaii and New Zealand, 4000 km away. According to Enderby et al. [48], it can be concluded that there was no significant genetic divergence between the Australian populations of Plutella xylostella with geographic distances of over 3600 km, however, significant divergence occurred in these populations with the Malaysia and Kenya populations, which were 5000 km away.
Understanding the degree of genetic differentiation of such moths with high migratory capabilities at the geographic scale will help to precisely track the approximate extent of population expansion and provide a basis for future pest management planning.

Gene Flow
The extensive gene flow revealed by our analysis may confirm the cause of weak genetic differentiation. Migration rates between populations were highly correlated with pairwise F ST [54], indicating the major role of gene flow in causing genetic differentiation. The high level of migration and broad environmental plasticity of S. litura populations that introduced genetic variations into partly local populations facilitated gene flow between populations. The high level of gene flow allowed resistance genes to disperse over long distances, leading to widespread dispersion of some chemical pesticide resistance genes, which may have resulted in adaptation to control strategies. The tobacco planting areas in Guangxi are mainly concentrated in Baise, Hechi, and Hezhou City with suitable geographical conditions for tobacco growth. These populations are mostly located in mountains, plateaus, and a few plains, among which Hechi and Baise Cities, near the Yunnan-Guizhou Plateau with rich Karst landscapes. Despite the complexity of geographical factors, the influence on the genetic differentiation of populations was minimal. We infer that the spread of these populations may be related to wind, climate, and human interference. It is noteworthy that high levels of gene flow can easily cause repeated outbreaks under appropriate conditions due to the transmission of rapidly resistant alleles among populations.

Conclusions
Our study showed high genetic diversity and low genetic differentiation of S. litura populations in Southern China. Overall, we consider that S. litura populations form a panmictic population on account of their long-distance migration, sufficient gene exchange, and low genetic differentiation. This helps explain the perennial susceptibility of S. litura to large outbreaks, resulting in extremely difficult prevention and control in local areas. Therefore, we suggest that, for migratory pests, the implementation & scope of management should be expanded to achieve the desired effect.