Population genetics of Liriomyza trifolii (Diptera: Agromyzidae) and comparison with four Liriomyza species in China based on COI, EF-1a and microsatellites loci

Liriomyza is a large genus that includes polyphagous and invasive species (L. trifolii, L. sativae, and L. huidobrensis), and oligophagous species such as L. Chinensis in China. Effective control of these invasive and oligophagous species is not easy due to the fast invasion rate, interspecific competition, and pesticide resistance. In this study, we investigated population genetics of five Liriomyza species L. trifolii, L. sativae, L. huidobrensis, L. bryoniae, and L. chinensis based on COI and EF-1a genes, and microsatellite DNA. These five Liriomyza species revealed highly conservative characteristics in the COI gene among populations collected from different geographical regions and host plants. By contrast, the mutation rate of the EF-1a gene was higher than COI, and phylogenetic tree based on EF-1a showed that haplotypes of L. trifolii and L. sativae were not distinguished well. Genetic differentiation in microsatellite loci was obvious among the five species. Our results also indicated that geographic isolation had a greater impact on genetic differentiation in L. trifolii than the host plant. Populations of L. trifolii in China showed a high to moderate level of genetic differentiation and they had divided into two groups representing the coastal areas of southern China and northern regions. The genetic diversity of the southern group was higher than the northern group. We speculated that the invasion of L. trifolii likely occurred in southern regions of China and then spread northward. Bottleneck analyses revealed that the L. trifolii population in China was in a steady growth period.

therefore it has been well applied in studies on population genetic structure, genetic relationship identification, genetic map construction and gene mapping to explore the population genetics, molecular systematics and ecology 15,22 . But there were a few researches using microsatellite marker technology to unfold the population genetic structure in Liriomyza especially for these invaded species 14 .
Previous studies on the population genetic structure of Liriomyza have generally involved only a single species 14 , with only a few comparative studies on genetic relationships among species 21 . In this study, we investigated intraspecies genetic differentiation in L. trifolii and interspecies variations among five species in Liriomyza in order to understanding better the species diversity during the geographic isolation and population expansion.

Results
Genetic differentiation of populations. Haplotype and nucleotide diversity of COI in L. trifolii populations was conserved with consistent characteristics among populations from different geographical regions and host plants. The maximum haplotype number among L. trifolii populations was three, and most haplotypes had only a single base difference. L. sativae populations showed slightly more diversity, and the maximum number of haplotypes among three L. sativae populations was six. The populations of L. huidobrensis, L. bryoniae and L. chinensis showed relatively low diversity (Table S1).
Haplotype and nucleotide diversity of EF-1a was relatively high among intraspecies (e.g. L. trifolii) as compared with COI. Ten haplotypes were found in the ZZJD, HLJD and HSFQ populations of L. trifolii, and the CSJD population had the highest nucleotide diversity. The other four Liriomyza spp. also showed relatively high diversity in EF-1a (Table S2).
The average observed number of alleles (Na) in L. trifolii populations ranged from 6.625 (DGJD) to 3.37 (HLJD). The average effective number of alleles (Ne) of L. trifolii populations ranged from 4.3885 (CXJD) to 1.9154 (SQNGMZ). The observed heterozygosity (Ho) values of ten L. trifolii populations were greater than 0.5, and the highest Ho was 0.6771 in the BLJD population; the remaining nine populations had Ho values less than 0.5 and the lowest was 0.3125 in the HLJD population. L. huidobrensis, L. bryoniae and L. chinensis had a low heterozygosity. Populations of different hosts in the same geographic region (DGQC and BLJD, NNQC and NNJD, SQJDMZ and SQNGMZ) showed a great degree of similarity in Na and Ho. Most populations were deviated from the Hardy-Weinberg equilibrium (Table S3). phylogenetic analyses. The phylogenetic tree based on COI haplotypes ( Fig. 1) showed that the five Liriomyza species had an obvious interspecific differentiation. The species relationship between L. trifolii and L. sativae were the most closest, and between L. bryoniae and L. huidobrensis was closer, while the relationships of L. chinensis with each of the other four Liriomyza species were distant. The phylogenetic tree based on EF-1a haplotypes (Fig. 2) was similar as the phylogenetic tree based on COI haplotypes, but haplotypes of L. trifolii and L. sativae were not distinguished well.
Genetic differentiation-pairwise FST analyses. Because of the obvious interspecific differentiation of COI and EF-1a in the five Liriomyza species, only the intraspecific genetic differentiations based on COI and EF-1a genes in L. trifolii populations was analyzed. The results based on COI showed that the HLJD population exhibited high genetic differentiations from other 19 populations, and the highest differentiation was found between showed that fifteen populations were basically clustered into two distinct main branches and four small scattered branches. Results of two population pairs NNQC/NNJD and DGQC/BLJD from different hosts in the same geographical region obviously converged to the nearest neighboring branch, which was consistent with pairwise F ST analysis. However, the HSFQ/HSJD population pair did not converge. STRUCTURE analyses of the fifteen populations showed that the highest ΔK value was obtained for K = 2 (Fig. 4). Populations from coastal areas of southern China (DGQC, BLJD, ZZJD, HZQC) were assigned to one group (red portion of Fig. 4). Populations from Jiangsu and Zhejiang provinces and northern regions (CXJD, CSJD, SQJD, HSJD, HSFQ) were assigned to another group (green portion of Fig. 4).
Bottleneck test. Bottleneck analysis with populations of L. trifolii across China showed that none of these 19 L. trifolii populations exhibited heterozygosity under the stepwise mutation model (SMM), and there were only six populations (CXJD, CSJD, HSJD, HSFQ, ZZJD, WZJD) and nine populations with a statistically significant heterozygotes under the two-phase model (TPM) and the infinite allele model (IAM) ( Table 3), respectively. These results indicated that the majority of L. trifolii populations did not undergo a genetic bottleneck and were in a steady growth period. www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
Population genetic structure and diversity are important factors affecting the survival and adaptability of invasive species. Population genetics in many pests were studied to find out their invasion and transmission routes 14,15,[19][20][21][22][23] . In this study, the phylogenetic tree, pairwise F ST , and STRUCTURE analysis indicated that the degree of differentiation and direction of nuclear and mitochondrial genes were not completely consistent. COI in the five species of Liriomyza showed very conservative characteristics, but the mutation rate of EF-1a gene was relatively higher, and phylogenetic tree results showed that haplotypes of L. trifolii and L. sativae were not distinguished well. The results of microsatellite analysis showed that genetic distances among the five species of Liriomyza were significantly much longer than those within L. trifolii populations. In short, the five Liriomyza species showed high levels of genetic differentiation in mitochondrial and nuclear genes, and the interspecies differentiation in nuclear genes was obvious. COI and EF-1a gene were suitable molecular markers for interspecies genetic differentiation analysis and not for intraspecies of Liriomyza species, because COI as a mitochondrial gene and EF-1a as a reference gene are highly conserved among intraspecific populations of the five Liriomyza species. Microsatellites    www.nature.com/scientificreports www.nature.com/scientificreports/ marker were suitable molecular markers for both interspecies and intraspecific genetic differentiation analysis of the five Liriomyza species, because microsatellite analysis showed both interspecies and intraspecific genetic differentiations among the five species of Liriomyza. Spencer (1964) suggested that host specialization caused the development of many new species 5 . We found that geographic isolation had a greater influence on genetic differentiation within L. trifolii, which is consistent with previous results for L. Sativae 14 , but we did not find obvious influence of host plants on genetic differentiation in these species. We hypothesize that host plants have not yet driven reproductive isolation among populations, so the gene exchange among populations on different hosts occurs frequently.
The results of genetic differentiation and structure analysis showed that most populations of L. trifolii in China were in a high or moderate degree of genetic differentiation. Populations of L. trifolii could be divided into two groups, one from coastal areas of southern China and the other from northern China including Jiangsu and Zhejiang provinces. The genetic diversity of the southern group was higher than the other group, so the invasion of L. trifolii likely occured in southern regions of China and then spread toward northward. Bottleneck test analysis showed that the L. trifolii population in China was in a steady growth period, which was similar as L. sativae 14 . Genetic variation may lead to the rapid adaptation of insects to new environments and contributes to population establishment and spread. Our study has produced information on the geographical distribution of genetic variation of five Liriomyza species in China that may also help in management programs of these important pests.

Materials and Methods
Sample collection and DnA extraction. Liriomyza individuals (n = 281; Table 4) were collected and preserved in 100% ethanol at −20 °C until DNA extractions were performed. Genomic DNA was extracted from samples using the LabServ Tissue DNA Kit (Thermo Fisher Scientific, Massachusetts, USA) and then used for PCR.
primers and microsatellite markers. The primers for mtDNA COI gene were referred to Simon et al. 24 .
Specific primers for EF-1a gene and eight microsatellite primers were designed in this study (Supplementary  Table S4). A fluorophore (FAM, ROX, HEX or TAMRA) was included at the 5' end of each pair of microsatellite primers (Supplementary Table S5) used for genotyping. All the primers used in this experiment were synthesized by GENEWIZ Inc (Suzhou, China), and microsatellite genotyping was performed by GENEWIZ Inc.
pcR amplification and sequencing. The COI (n = 268; Supplementary Table S1) and EF-1a PCR (n = 252; Supplementary Table S2) of Liriomyza individuals (Table 4) were successfully amplified and sequenced. The amplification conditions were as follows: initial denaturation for 4 min at 94 °C, followed by 35 cycles of denaturation for 30 s at 94 °C, annealing for 30 s at 58 °C, elongation for 50 s at 72 °C, and a final extension step of 72 °C for 5 min. The microsatellite amplification of Liriomyza individuals (n = 281; Table 4) conditions were as follows: