Population genetics and demographic history of guava fruit fly Bactrocera correcta ( Diptera : Tephritidae ) in northeastern Thailand

Bactrocera correcta (Bezzi) is among the most destructive fruit fly pests of the genus Bactrocera. This species infests 62 spe­ cies of plants belonging 30 families, many of which are commercially important. In this study, the genetic structure, diversity and de­ mographic history of B. correcta in Thailand were inferred from mitochondrial cytochrome c oxidase I (COI) sequences. High genetic diversity was recorded in the 171 samples collected from 15 locations. This was due largely to the existence of two divergent lineages (I, II) revealed by median joining (MJ) network analysis. Genetic structure analysis revealed an overall low level of genetic differentia­ tion between populations suggesting that the flies can move freely across geographic regions. Because the host plants are commonly grown in Thailand, continuity of habitats is the factor most likely responsible for the genetic homogeneity. In addition, the recent popu­ lation history could also be a factor that contributed to the overall low level of the genetic structure. Mismatch distribution analysis as well as Tajima’s D and Fu’s FS tests detected evidence of recent demographic expansion dating back to the end of the last glaciations. * Corresponding author.

Information regarding genetic diversity, genetic struc ture and gene flow are crucial for pest control and manage ment (Roderick, 1996;Roderick & Navajas, 2003).For ex ample, the sterile insect technique (SIT) uses sterile males to compete for mating opportunities with the wild males, but this method requires a large number of sterile flies, which should be in the same proportions as the number of wild flies (Itô et al., 2003).Information about effective population size and individual movement between popula tions (i.e. gene flow), which can be deduced from popula tion genetic studies (Aketarawong et al., 2011;Karsten et al., 2013), is important for effective planning of releasing sterile insects.Despite it being an important pest species, which causes significant economic losses, the detailed ge netic structure of this species is poorly studied.
The population genetic structure and demographic his tory inferred from molecular markers could also contribute to our understanding of the effect of past environmental changes (e.g.Pleistocene glaciations) on current biodi versity.Previous studies on insects, such as mosquitoes (O'Loughlin et al., 2008;Morgan et al., 2011), black flies (Pramual et al., 2005(Pramual et al., , 2011) ) and another species of fruit fly, Bactrocera latifrons (Hendel) (Meeyen et al., 2014), indicate the important role of historical climatic changes in determining the genetic structure and diversity of South east Asian faunas.
In this study, we use the mitochondrial cytochrome c oxi dase I (COI) gene to infer the genetic structure, diversity and demographic history of B. correcta in Thailand, espe cially in the northeastern part.Several studies have shown that COI sequences can be used for population genetic studies of fruit flies (Mun et al., 2003;Nardi et al., 2005;Shi et al., 2005Shi et al., , 2010Shi et al., , 2012;;Hu et al., 2008;Prabhakar et al., 2012Prabhakar et al., , 2013;;Meeyen et al., 2014).We also investigated the genetic relationship between B. correcta in Thailand and samples from other geographic regions.The results of this study provide significant information on the genetic structure and diversity of this species of potential utility in pest management and control programs.In addition, the demographic history determined from this study will in July 2013 (Table 1, Fig. 1).The infested fruits were placed in plas tic boxes with a layer of sawdust in the bottom, covered by calico and kept at room temperature.After the adults emerged, flies were collected and stored in 80% ethanol at -20°C until processing.Species were identified using adult morphological characteristics following White & Elson-Harris (1992) and Plant Health Aus tralia (2011).
crease our fundamental understanding of the factors shap ing biodiversity in Thailand.Finally, the COI sequences used in this study enlarge the DNA barcode database that is crucial for the success of DNA base species identification.

DNA extraction, amplification and sequencing
Genomic DNA was extracted from individual adult flies using the GF-1 Tissue DNA Extraction Kit (Vivantis, Selangor Darul Ehsan, Malaysia).A 584-bp fragment of the mitochondrial cy tochrome c oxidase I (COI) gene was amplified using the primers LCO1490 (5′-GGTCAACAAATCATAAAGATATTGG-3′) and HCO2198 (5′-TAAACTTCAGGGTGACCAAAAAATCA-3′) (Folmer et al., 1994).Polymerase chain reaction (PCR) ampli fications were carried out in a final volume of 50 µl with condi tions as described by Rivera & Currie (2009).PCR products were checked using a 1% agarose gel and cleaned using the HiYield TM Gel/PCR DNA Fragments Extraction Kit (RBC Bioscience).Pu rified PCR products were sequenced using the same primers as in the PCR by Macrogen sequencing service (Seoul, Korea).
The population genetic structure was estimated using popu lation pair wise F ST .The significance test statistic was obtained based on 1023 permutations.Sequential Bonferroni correction (Rice, 1989) was applied for the multiple tests.Analysis of mo lecular variance (AMOVA) was used to test the genetic differen tiation among groups of populations from different host-plants and geographic regions as described in Table 1.Both population pair wise F ST and AMOVA analyses were performed in Arlequin using the Kimura 2-parameter model (K2P).A Mantel test (Man tel, 1967) was used to determine the relationship between genetic distance (F ST from Arlequin) and geographic distance (km) to test an isolation-by-distance (IBD) model.The Mantel test was im plemented in IBD v1.52 (Bohonak, 2002) using 1000 randomiza tions.
Mismatch distribution was used to test the demographic history of the populations.A population that in the recent past has under gone a demographic expansion shows a unimodal mismatch dis tribution (Roger & Harpending, 1992).The sum-of-squares de viation and Harpending's raggedness index (Harpending, 1994) were used to test the deviation from the prediction of the sudden expansion model.Mismatch distribution was estimated using Ar lequin.Population expansion time was calculated using τ = 2ut (where u = m T μ, m T is the length of the nucleotide sequences un der study, μ is the mutation rate per nucleotide and t is the genera tion time (Roger & Harpending, 1992)), assuming a divergence rate of 2.3% per million years for insect mtDNA (Brower, 1994) and eight generations per year based on the rearing information for B. correcta (Liu & Ye, 2009).In addition, Fu's F S test (Fu, 1997) and Tajima's D (Tajima, 1989) statistical tests were also used to test for population equilibrium.The expectation is that these tests will yield large negative values during demographic population expansion.

Mitochondrial DNA sequence variation
A 584 bp fragment of the mitochondrial COI gene was sequenced from 171 specimens of B. correcta from 15 locations in Thailand.Sequences were deposited in Gen Bank with the accession numbers KJ879751-KJ879921.A  total of 83 haplotypes were identified.The most common haplotype was recorded at all the locations sampled except Phetchabun (PB) and Chanthaburi (CB).This haplotype of Bactrocera correcta also occurs in India, Sri Lanka, Laos, Viet Nam and China.Haplotype diversity in each popula tion ranged between 0.3455 in Ubon Ratchatani (UR) and 1.000 in Maha Sarakham (MS3) and Chanthaburi (CB) with an average of 0.9337 (Table 2).Nucleotide diversi ty in each population ranged between 0.0076 in Nakhon Ratchasima (NR) and 0.0325 in Chaiyaphum (CP) with an average of 0.0132 (Table 2).

Mitochondrial genealogy
The MJ network (Fig. 2  cluster in the MJ network was associated neither with hostplants nor geographic origins (Fig. 2).Overall, the network has a star-like shape with a central haplotype shared by the globally distributed populations (Thailand, India, Sri Lanka, Laos, Viet Nam and China), which is characteristic of a recent demographic population expansion (Slatkin & Hudson, 1991).Genetic relationships between Thai B. correcta and se quences from other geographic regions are as follows.Three mitochondrial COI sequences from India have the same central haplotype.Two sequences from Viet Nam were made up of two haplotypes, of which one was unique and linked by a short branch length to haplotype group I, and one sequence has the central haplotype.One sequence from Laos had the central haplotype.Seven sequences from China included six haplotypes with one haplotype unique and connected to the central haplotype, three se quences had the central haplotype and three sequences had the same haplotypes as the Thai specimens and clustered in haplotype group I.Ten sequences from Sri Lanka were included in the MJ network analysis.Three sequences had the central haplotype and seven sequences included five unique haplotypes that linked them with haplotype group I.One sequence from Myanmar represented a unique hap lotype and was directly connected to the central haplotype.

Population genetic structure
Population pairwise F ST values revealed that most popu lations were not genetically significantly different (Table 3).The exception was the difference between Phetchabun (PB) and the other populations, where almost all F ST values were statistically significantly different.AMOVA analyses by grouping populations according to their host-plant spe cies and geographic regions found no significant genetic differentiation among the groups (Table 4).Mantel's test revealed no significant relationships (r 2 = 0.0074, P = 0.3030) between genetic (pairwise F ST ) and geographic distances.

Demographic history
Mismatch distribution analysis revealed a unimodal mis match graph (Fig. 3), a characteristic of a recent popula tion expansion.This is consistent with the star-like shape of the mtDNA genealogy.Both sum-of-squares deviation (SSD = 0.0049, P = 0.8300) and Harpending's raggedness index (0.0075,P = 0.9200) were not significantly different from the simulated data predicted by the sudden population expansion model (Fig. 3).Population expansion was also supported by highly significant negative values of both Tajima's D (−2.3708, P < 0.001) and Fu's F S (−24.7529,P < 0.001) tests.Time for which the population has been expanding is estimated to be approximately 15,000 years.

DISCUSSION
The genetic variation of B. correcta in Thailand, espe cially in the northeastern part of the country (ranges be tween 0.760% and 3.250% with an average of 1.312%) based on the COI sequence, which is higher than that re corded for other species of fruit fly, including Bactrocera cucurbitae (Coquillett) (0.1%-0.3%) (Hu et al., 2008), Bactrocera oleae (Rossi) (0.09%-0.48%) (Dogac et al., 2013) and B. latifrons (0.09-0.86%) (Meeyen et al., 2014).The level of genetic variation in B. correcta is also higher than that recorded for other species of fruit fly such as Bactrocera dorsalis (Hendel) (0.7%-2.0%) (Shi et al., 2012) and Bactrocera tryoni (Froggatt) (0.5%-1.8%) (Blacket et al., 2012).The high level of genetic variation in B. correcta  is due largely to the existence of the divergent lineages re vealed by the haplotype network analysis.These lineages are not associated with particular host plants or geographic regions.Although all individuals of lineage II were from northeastern Thailand, many other specimens from this re gion were clustered in lineage I.We have checked the spe cies identification using COI barcoding sequences in the BOLD systems and found that all members of lineage II were a 100% match with B. correcta in the database.Thus, the possibility that the divergent lineage is a consequence of misidentification is unlikely.Genetic divergence between the two lineages based on the K2P model is 2.08%, which falls in the range of the 3% cut-off value for DNA barcode sequences (Hebert et al., 2003).However, some studies have suggested that this cut-off value is not appropriate due to large variations in intraspecific genetic divergence (Meier et al., 2006).The values of interspecific genetic divergence estimated for 60 species of fruit flies range between 0.1% and 5.3% with a mean of 0.9% (Armstrong & Ball, 2005).Therefore, it is not possible to determine here if the two lineages of B. correcta recorded in this study represent different species or not.However, because the two lineages are genetically distinct with low genetic divergence within each lineage, the results indicate no genetic exchange (i.e. gene flow) between the two lineages.
A previous study has also recognized two divergent line ages of B. correcta in Thailand.Jamnongluk et al. (2003) recorded that genetic distance based on COI sequences between two specimens of B. correcta collected from the same host plant (Syzygium samarangense) as considerable and claims that these two specimens represent two sibling species of B. correcta.The genetic divergence of these specimens was due to a Wolbachia infection, as one speci men was found to be infected with Wolbachia and the other not (Jamnongluk et al., 2003).Unfortunately, we were un able to include these specimens in our analysis because the COI fragment used by Jamnongluk et al. (2003) did not overlap our sequence.We were also not able to test for Wolbachia infection, thus this would be an interesting topic for further investigation.
Population pairwise F ST values indicate an overall low level of genetic structuring between populations of B. correcta.The results are consistent with many other popula tion genetic studies on fruit flies, which also detected low genetic structuring (Hu et al., 2008;Meeyen et al., 2014).Two factors most likely account for the genetic homogene ity among populations of B. correcta in Thailand.First, B. correcta utilizes a wide host range with 62 plant species belonging to 30 families (Clarke et al., 2001).Many host plants (e.g.mango, rose apple and guava) are commonly grown in Thailand and fruit fly populations are likely to be geographically continuous.In addition, human mediated dispersal such as local fruit transportation and trade could also facilitate movement of flies as has been reported in other fruit fly species (Malacrida et al., 2007;Shi et al., 2012).This could promote genetic exchange (i.e. gene flow) between populations that counters the effect of ge netic drift or selection by lowering the level of genetic dif ferentiation.
The second factor that could contribute to the low level of genetic structuring in B. correcta is its recent popula tion history.Mismatch distribution analysis as well as the Fu's F s and Tajima's D tests indicate a recent demographic expansion in this species.The beginning of this expansion was estimated to be at the end of the Pleistocene glaciations (approximately 15,000 years ago).The results are consist ent with previous studies on another fruit fly species, B. latifrons (Meeyen et al., 2014), and other insects such as mosquitoes (O'Loughlin et al., 2008;Morgan et al., 2011) and black flies (Pramual et al., 2005(Pramual et al., , 2011)).Climatic con ditions during the Pleistocene glaciations in tropical Asia, including Thailand, were thought to be drier with lower temperatures (Penny, 2001).These conditions lead to the contraction of tropical forests and the expansion of sea sonally dry broad leaved dipterocarp forests (Penny, 2001).Climatic conditions returned to warm and humid about 18,000 years ago and this allowed the tropical forests to expand.Thus, the demographic expansion in B. correcta is most likely associated with an host-plant species expan sion.The population expansion detected in B. correcta after the return to warm and humid climatic conditions is consistent with the present-day seasonal abundance of this species.Clarke et al. (2001) found that B. correcta reaches its peak abundance between May and September, which is the middle of the rainy season and coincident with host plant fruiting time.
The exception to the overall genetic homogeneity is the significant differentiation of the Phetchabun (PB) popula tion.Most comparisons revealed significant F ST values, which indicate limited gene flow between PB and the other populations.Geographically, this population is iso lated from the others by a large mountain range (Phetch abun range).An ecological study of B. correcta indicated that mountain ranges are effective geographic barriers to dispersal because this species occupies low altitude areas (Liu et al., 2013).Mountain ranges are also important geo graphic barriers to gene flow in other fruit flies (Shi et al., 2005;Meeyen et al., 2014).
In conclusion, we found relatively high genetic diversity in B. correcta due to the existence of divergent lineages.Despite high genetic diversity, the overall genetic structur ing was low except for one population that was isolated by a large mountain range, which acted as a barrier to gene flow.Demographic history analysis revealed the recent population expansion began at the end of the Pleistocene.As this is also recorded for other co-geographically dis tributed species in Thailand, it highlights the importance of late Pleistocene historical events in determining the ge netic structure and diversity of species on tropical main land in Asia.

Fig. 1 .
Fig. 1.Locations in Thailand of the 15 populations of Bactrocera correcta sampled.Details of the locations are given in Table1.
Fig. 2. Median Joining network of 209 COI sequences (171 sequences for Thailand and 38 sequences for other geographic regions) of Bactrocera correcta.Each circle represents a haplotype and sizes are relative to the number of individuals with a specific haplotype.Haplotypes are labelled according to country of origin.

Fig. 3 .
Fig. 3. Mismatch distribution of the 171 COI sequences of Bactrocera correcta for Thailand in terms of the recorded and expected pairwise differences based on the predictions of the sudden population expansion model.Mismatch distribution for B. correcta is consistent with the predictions of the sudden popu lation expansion model (SSD = 0.0049, P = 0.8300; Harpending's raggedness index = 0.0075, P = 0.9200).

table 1 .
Specimens were collected from natural habitats in Thailand, especially from the northeastern part, between March 2012 and Details of the sites sampled and host-plant species of Bactrocera correcta in Thailand.

Table 1 .table 2 .
Haplotype diversity (h) and nucleotide diversity (π) of 15 populations of Bactrocera correcta in Thailand.Details of the locations sampled are given in Table1.

table 3 .
Population pairwise F ST values between 15 populations of Bactrocera correcta in Thailand.Details of the locations sampled are given in Table 1.Bold characters indicate statistical significance at P < 0.05.

table 4 .
Results of the AMOVA analyses of 15 populations of Bactrocera correcta from Thailand, with grouping according to geo graphic regions and host-plants.*P < 0.05.