Genetic Diversity in Apple Fruit Moth Indicate Different Clusters in the Two Most Important Apple Growing Regions of Norway

The apple fruit moth (Argyresthia conjugella (A. conjugella)) in Norway was first identified as a pest in apple production in 1899. We here report the first genetic analysis of A. conjugella using molecular markers. Amplified fragment length polymorphism (AFLP) analysis was applied to 95 individuals from six different locations in the two most important apple-growing regions of Norway. Five AFLP primer combinations gave 410 clear polymorphic bands that distinguished all the individuals. Further genetic analysis using the Dice coefficient, Principal Coordinate analysis (PCO) and Bayesian analyses suggested clustering of the individuals into two main groups showing substantial genetic distance. Analysis of molecular variance (AMOVA) revealed greater variation among populations (77.94%) than within populations (22.06%) and significant and high FST values were determined between the two major regions (Distance = 230 km, FST = 0.780). AFLP analysis revealed low to moderate genetic diversity in our population sample from Norway (Average: 0.31 expected heterozygosity). The positive significant correlation between the geographic and the molecular data (r2 = 0.6700) indicate that genetic differences between the two major regions may be due to geographical barriers such as high mountain plateaus (Hardangervidda) in addition to isolation by distance (IBD).


Introduction
The apple fruit moth (Argyresthia conjugella (A.conjugella)) was first described by Zeller in 1839.It is a small moth with a body length of 5-6 mm belonging to the order Lepidoptera (butterflies and moths), the group Ditrysia and the family Yponomeutidae.There are at present over 160,000 described species of butterflies and moths [1].The Ditrysia, a natural group of butterflies and moths, contain more than 98% of the described Lepidoptera species.The A. conjugella is a specialized seed predator of rowan (Sorbus aucuparia).Females are attracted to unripe rowan berries and lay their eggs on or near clusters of berries.The larvae bore into the fruits where they live and forage.In the late summer, the larvae drop to the ground and pupate.The pupae overwinter in the ground and the adults emerge in May-June the following year.A. conjugella has a wide distribution and is found all over Europe, in Asia [2] and have also been reported in North America [3].In Norway, the A. conjugella was first identified as a pest insect in 1899 after a devastating attack on apples [4].In years when rowan berries are scarce, the female moth must seek alternative hosts [5], and apple (Malus domestica) is such a forced alternative host.In years of low rowan fruit production, A. conjugella larvae can cause great damage to apple crops, and are therefore recognized as the most important insect pest on apples in Fennoscandia [6].
Genetic diversity is influenced by geographical barriers, dispersal and isolation by distance (IBD).Some insects move only short distances during their entire lifetime [7][8][9], whereas others engage in one or more bouts of long-distance movement as adults [10,11].Insect populations may become genetically distinct because of reproductive isolation or lack of gene flow between different regions [12].Further, reproductive isolation is influenced by several factors such as individual dispersal ability, ecological isolation, geographic barriers, and local adaptation after natural selection [13][14][15][16].Dispersal is a key factor in population dynamics [17].Continuous dispersal helps long-term survival of a species and can lower the genetic drift rate of local populations [18].Population genetic studies have shown how genetic differentiation can exist between insecticide-resistant populations and susceptible populations [19,20].Thus, this may be a key factor for successful and stable control of insect pests.Investigating the genetic diversity of A. conjugella may be helpful both in determining sources from which A. conjugella disperse and for monitoring insecticide resistance.
Amplified fragment length polymorphism (AFLP) analysis [21] has been widely used to study genetic diversity of plant populations, but only more recently been adopted for population genetic studies of insect species.As an example, AFLP analysis was used to study genetic structure of Spodoptera frugiperda populations associated with maize and cotton crops in Brazil [22].In USA, the technique was applied to detect; high gene flow in Ostrinia nubilalis collected from maize fields [23], significant genetic differentiation of Podisus maculiventris populations, which is an economical important biological control agent [24] and high genetic variation among regions of the important pest Pseudatomoscelis seriatus from cotton fields [25].In Spain, backcross hybrid between Drosophila buzzatii and Drosophila koepferae species was detected using AFLP [26].AFLP was used to detect the impacts of human activities on the Bamboo Locust Ceracris kiangsu genetic structure in China [13] and the population dynamics of Hydropsyche orientalis and Stenopsyche marmorata in Japan [27].
The greatest advantage of AFLP is probably that it does not need previous sequence information of the species.Additional arguments for the choice of AFLP, was the difficulties of the development of microsatellite DNA markers for identification and application of Lepidoptera species, due to the high similarity in flanking regions between different microsatellites within the same species [28][29][30] as well as the lack of conserved flanking regions leading to unrepeatable banding patterns [31].
AFLP analyses have been used to study the genetic diversity of various insect species, but until now population genetics on A. conjugella have not been investigated.Thus, we have used AFLP markers to investigate the genetic differentiation between A. conjugella populations in the most important apple growing regions of Southern Norway.The major aim of our study was to investigate whether or not (and to what degree) the potentially strong dispersal barriers between the regions in the form of the high mountain plateau of Hardangevidda, in addition to the geographical distance would limit the gene flow between populations.

Collection of A. conjugella Materials
Rowan berries infested with A. conjugella larvae were collected in the field during August 2012.Six populations of A. conjugella were collected from rowan berries (the primary and preferred host) from two geographical regions (Figure 1).West Norway (1.Espe, 2. Sekse and 3. Stana) and East Norway (4.Fagerstrand, 5. Storsand and 6.Telemark), representing the most important apple growing regions of Norway.The distance between the two major apple growing regions is 230 km, separated by a high mountain plateau (Hardangervidda, see Figure 1).In the east, Telemark sampling location is separated by 90 km from Fagerstrand and 80 km from Storsand, while the distances between the three sampling locations in the west (Espe, Sekse and Stana) are not longer than 5 km (illustrated in Figure 1).For this reason, it was not possible to conduct continuous sampling between the two major regions.Collected A. conjugella individuals were stored in 95% ETOH and stored at ´20 ˝C to avoid DNA degradation.Initially, 16 individual insects (8 female and 8 male) were sampled from each population.One individual female from the location Espe was excluded from the analysis due to lack of PCR amplification.The total number of individuals analyzed was thus 95.Details of the A. conjugella individuals are presented in Table 1 and Supplementary Table S1.
Diversity 2016, 8, 10 3 of 12 from each population.One individual female from the location Espe was excluded from the analysis due to lack of PCR amplification.The total number of individuals analyzed was thus 95.Details of the A. conjugella individuals are presented in Table 1 and Supplementary Table S1.

DNA Extraction
DNA was extracted from pupal tissue using DNeasy 96 Blood and Tissue Kits (Qiagen, Tokyo) according to the manufacturer's instructions, and the only modification was the elution of 40 µL DNA instead of 50-100 µL.

AFLP Analysis
AFLP analysis [21] was performed as previously described [32], with modifications that included the use of fluorescently labeled primers instead of radioactive labeling.Genomic DNA, (approximately 300 ng) was double-digested using EcoRI and the MseI isoschizomer Tru1I, followed

DNA Extraction
DNA was extracted from pupal tissue using DNeasy 96 Blood and Tissue Kits (Qiagen, Tokyo) according to the manufacturer's instructions, and the only modification was the elution of 40 µL DNA instead of 50-100 µL.

AFLP Analysis
AFLP analysis [21] was performed as previously described [32], with modifications that included the use of fluorescently labeled primers instead of radioactive labeling.Genomic DNA, (approximately 300 ng) was double-digested using EcoRI and the MseI isoschizomer Tru1I, followed by ligation to the adaptors overnight at room temperature.A total volume of 25 µL pre-amplification containing 5 µL of fivefold diluted ligation product, was carried out with non-selective primers, mixed with; 1ˆPCR buffer (50 mM KCl, 10 mM Tris HCl pH 8.3, 1.5 mM MgCl 2 ), 600 µM of each dNTP, 50 ng of each 0-primer, and 1 U of Taq DNA polymerase.Pre-amplification was performed in a Gene Amp 9700 thermo cycler (Applied Biosystems, Foster City, CA, USA) as follows: 94 ˝C for 2 min followed by 45 cycles at 94 ˝C for 30 s, 56 ˝C for 30 s and 72 ˝C for 1 min, and finally a hold at 72 ˝C for 10 min.
Eight PCR amplification primer pairs were tested using 10 individuals of A. conjugella (both female and male), and of these five pairs with two selective bases (Table 1; Applied Biosystems, Foster City and Invitrogen, Carlsbad, CA, USA) were chosen based on number of amplified fragments in the range 50-500 base pairs, and amount of polymorphism among the included individuals.In the selective PCR amplification, the EcoRI primers are labeled with the fluorescent label 6FAM.The fluorescently labelled PCR products were analyzed using an ABI3730 DNA Analyzer.1 µL of PCR products were added to a loading buffer containing 8.75 µL Hi-Di formamide (Applied Biosystems), and 0.25 µL of GeneScan 500 LIZ size standard (Applied Biosystems).The data was collected using the software Data Collection v 2.0 (Applied Biosystems), while GeneMapper v 4.1 (Applied Biosystems) was used to derive the fragment length of the labeled DNA-fragments using the known fragment lengths of the LIZ-labeled marker peaks (Supplementary Figure S2).

Data Analyses
The number of amplified fragments in the range of 50-380 base pairs was recorded manually, and only clear polymorphic bands were scored for presence (1) or absence (0), and monomorphic bands were not scored.The results of the AFLP analysis were confirmed by repeating the analyses of 10 randomly selected A. conjugella (both female and male), using same primer combinations.The replicated profiles were compared, and markers with more than 5% errors were removed from the datasets.Genetic similarity (GS) using the Dice coefficient was calculated as GS xy = 2a/(2a + b + c), where GS xy is the measurement of the genetic similarity between insects x and y, a is the number of polymorphic bands present in both insects, b is the number of polymorphic bands present only in insect x, and c is the number of polymorphic bands present only in insect y [33].The genetic similarity between the different insects, based on presence or absence of the amplified fragments, was also calculated by Jaccard coefficients [34].
Principal Coordinate Analysis (PCO) analyses based on individual insects were performed using four different similarity or distance coefficients (Dice, Simple matching, Jaccard, and Euclidean distance).We used PCO to perform ordination analysis and to classify and detect structure in the relationships between the regions.Statistical analyses were performed using the NTSYS-pc software [35].
We investigated also the genetic structure of A. conjugella populations using the model-based Bayesian clustering approach of genetic mixture analysis in the software STRUCTURE 2.3.4 developed by [36], and further implemented by [37].These analyses assume that the clusters (K) are at Hardy-Weinberg equilibrium and in linkage equilibrium, but the approach has proven to be robust also to deviations from these assumptions [37].Simulations were performed with data set, from K = 1 to K = 6.Six independent runs were conducted to assess the consistency of the results across runs, and all runs were based on 500,000 iterations after a burn-in period of 100,000 iterations.We followed the method developed by [38], to identify the number of genetically homogeneous clusters (K).Plots of likelihoods, similarity coefficients and ∆Ks [38] were made with Structure Harvester [39].
Analysis of molecular variance (AMOVA) [40] for A. conjugella populations was used to estimate the partitioning of AFLP variation between the regions and among populations within different regions.We also performed a separate AMOVA on the Western populations (Supplementary Table S2A) and a separate one on the Eastern populations (Supplementary Table S2B).
The genetic distances among all A. conjugella populations were calculated using pair-wise genetic distance method [41].The F ST values and mean F ST was estimated in order to study genetic differentiation between regions and among local populations.The significance of F ST values was tested by 1000 permutation.These analyses were performed using Arlequin software, version 2.000 [42].Gene flow was estimated assuming Nm = (1/F ST ´1)/4 [43].
The geographic data (based on the distance between the Eastern and the Western populations) were compared and correlated with the molecular data (AFLP analyses), the correlations were estimated using Mantel's test [44].

Results
AFLP analyses of all 95 A. conjugella using five primer combinations resulted in a total number of 410 clear polymorphic bands (Table 2).The number of polymorphic bands per primer combination ranged from 75 to 91 with an average of 82 polymorphic bands per primer pair.Using these five primer combinations, it was possible to distinguish between all 95 A. conjugella individuals.The results of the principal coordinate analysis (PCO) clustered the 95 A. conjugella into two main groups.PCO based on alternative coefficients (see Methods) gave very similar results as above (Results not shown).There is a clear distinguishable pattern of clustering without overlapping of A. conjugella from the two major regions (Figure 2).tested by 1000 permutation.These analyses were performed using Arlequin software, version 2.000 [42].Gene flow was estimated assuming Nm = (1/FST − 1)/4 [43].
The geographic data (based on the distance between the Eastern and the Western populations) were compared and correlated with the molecular data (AFLP analyses), the correlations were estimated using Mantel's test [44].

Results
AFLP analyses of all 95 A. conjugella using five primer combinations resulted in a total number of 410 clear polymorphic bands (Table 2).The number of polymorphic bands per primer combination ranged from 75 to 91 with an average of 82 polymorphic bands per primer pair.Using these five primer combinations, it was possible to distinguish between all 95 A. conjugella individuals.The results of the principal coordinate analysis (PCO) clustered the 95 A. conjugella into two main groups.PCO based on alternative coefficients (see Methods) gave very similar results as above (Results not shown).There is a clear distinguishable pattern of clustering without overlapping of A. conjugella from the two major regions (Figure 2).Structure analysis showed the maximum likelihood distribution L (K) at the real number of two groups (K = 2) (Figure 3 and Supplementary Figure S1).This value was obtained, using the value of Ad hoc quantity (∆K) rather than maximum likelihood value L (K) as published by [38].Structure analysis clustered the A. conjugella into two main clusters supported the results of the PCO analysis.Structure analysis showed the maximum likelihood distribution L (K) at the real number of two groups (K = 2) (Figure 3 and Supplementary Figure S1).This value was obtained, using the value of Ad hoc quantity (∆K) rather than maximum likelihood value L (K) as published by [38].Structure analysis clustered the A. conjugella into two main clusters supported the results of the PCO analysis.There is a clear distinguishable pattern of clustering of A. conjugella from the two major regions.Both clustering analyses showed that individual males and females from both regions were clustered randomly, and there was no distinguishable pattern of clustering of A. conjugella according to their sex.There is a clear distinguishable pattern of clustering of A. conjugella from the two major regions.Both clustering analyses showed that individual males and females from both regions were clustered randomly, and there was no distinguishable pattern of clustering of A. conjugella according to their sex.The AFLP data were also analyzed using AMOVA, and variance components were estimated (Table 3).Most of the total genetic variability in A. conjugella (77.94%) was among regions, while the genetic variability within regions and among populations within regions was low (21.20% and 4.73%) respectively.In contrast, AMOVA analysis of the Western populations (supplementary Table S2) and the Eastern populations (supplementary Table S3), showed high genetic variability within populations and low genetic variability among populations, as well as low FST values 0.0078 and 0.1053 respectively.The analysis of pair-wise distances showed that the genetic distances between A. conjugella locations in the Eastern and Western clusters were high (Table 4).The greatest genetic distance between populations was between Fagerstrand in the Eastern region and Sekse in the Western region (0.26121), while the smallest genetic distance was between Espe and Stana, and both are from the Western region (0.01469).The AFLP data were also analyzed using AMOVA, and variance components were estimated (Table 3).Most of the total genetic variability in A. conjugella (77.94%) was among regions, while the genetic variability within regions and among populations within regions was low (21.20% and 4.73%) respectively.In contrast, AMOVA analysis of the Western populations (Supplementary Table S2) and the Eastern populations (Supplementary Table S3), showed high genetic variability within populations and low genetic variability among populations, as well as low F ST values 0.0078 and 0.1053 respectively.The analysis of pair-wise distances showed that the genetic distances between A. conjugella locations in the Eastern and Western clusters were high (Table 4).The greatest genetic distance between populations was between Fagerstrand in the Eastern region and Sekse in the Western region (0.26121), while the smallest genetic distance was between Espe and Stana, and both are from the Western region (0.01469).The mean expected heterozygosity (0.306), and the level of expected heterozygosity in the Western populations (0.335) was higher than in the Eastern populations (0.276).The F ST value between the two major regions is high (0.780), while the F ST values between the three sampling locations within the Western and Eastern regions were very low to moderate, respectively (Table 5).The gene flow (Nm) was estimated to be very low (0.40) between the two major regions, while within Western and Eastern regions gene flow was found to be very high and moderate, respectively (Table 5).Mantel test showed a positive significant correlation (r 2 = 0.6700) between the geographic and the molecular data (p ď 0.001).

Discussion
To our knowledge, this the first study that investigates genetic diversity of A. conjugella using molecular markers, and it is the first evidence of geographic variation in A. conjugella populations in Norway.We detected low to moderate genetic diversity in A. conjugella populations, and found very high genetic differentiation between the two most important apple growing regions in Norway.Gene flow in insect species is expected to be high in populations that have no major geographical barriers and is not separated by long distances [45].Although the distance between the two major apple growing regions in Southern Norway is only 230 km, these two major areas are separated by a high mountain plateau (Hardangervidda), which is more than 1000 m above the sea level.For this reason, continuous sampling between the two regions was not possible.Thus, substantial genetic differentiation may be expected as a result of the long distance and physical barriers that both may strongly limit gene flow.
All our genetic analyses show that most of the genetic variation and structuring was between the Eastern and the Western regions in Southern Norway.The pairwise F ST analysis also indicated substantial geographical structuring, showing a significant and very high F ST value between these two major regions (0.780).Due to the lack of previous genetics studies in A. conjugella, we compared this F ST result to other studies in Lepidoptera species.A comparable high value of 0.708 in Busseola fusca was obtained in a geographical wide study among Eastern, Central and Western African countries [46] where the regions are separated by more than 10 times the distance of our study (3000 km).A more moderate value of 0.437 in Tuta absoluta [47] was found between two regions in Tunisia separated by 170 km that may be more comparable to our study.The very high value of 0.90 in Boloria eunomia [48] was detected among Western, Central and southern Europe, and this is not surprising since these regions are separated by more than 2500 km.Thus, our F ST value may be regarded as very high, which indicates that structuring and differentiation between regions may be influenced by more than IBD alone.Significant correlations between genetic and geographic distances using Mantel's test supported these results.Our results may be comparable to the ones obtained for other Lepidoptera species, e.g., "Sesamia nonagrioides" in Iran and Spain where subpopulations are separated by larger distances than in the present study.In Iran, genetic differentiation was found between two regions (Southern and Southwestern Iran) separated by 500 km [49], while in Spain, high genetic differentiation was observed between regions separated by distances of 1000 km [50].AFLP data analysis of individual male and female did not reveal any marker specific for sex determination, and clustering analyses of males and females did not show any particular pattern.This result is similar to the one obtained in the Lepidoptera species "Plutella xylostella" [51].For the same species, in Cameroon and China substantial genetic variation and high gene flow among regions have been reported [52,53].However, recent studies also provide contrasting results and have reported genetic differentiation and isolated populations in Lepidoptera species such as for Boloria eunomia, Sesamia nonagriodes and Diatraea saccharalis [48,49,54].Our results may be an example of a combination of geographical barriers and IBD causing very high genetic differentiation between regions.In addition, moderate to high gene flow was observed among populations within each of the two major region in Southern Norway, indicating that A. conjugella show medium to short distance migration.Similarly, the cotton flea-hopper's (Hemiptera: Miridae) western populations in USA are genetically distinct from eastern populations, but with considerable gene flow among the populations within each region [25].Gene flow in the west indicate a total contact between the three sampling locations probably as a result of very small geographical distances, while in the East the gene flow value (~2) may indicate some structuring among the three sampling locations maybe as a result of longer distances and IBD.
There are several other possible factors, which might be responsible for genetic differentiation among populations in insects species; low dispersal ability [15,55], habitat fragmentation [56][57][58] and host plant availability [6,55,59].A. conjugella in different regions could be genetically distinct, considering the differences in eco-geographic factors and variable presence of this insect in the rowan and apple plants in Norway.Furthermore, variable rates of natural selection may act on species with a wide geographic distribution and therefore, locally adapted populations will exist, which will result in high genetic variation among regions.Genetic differentiation between the two regions may also be due to two different introductions A. conjugella in Norway, but we do not have enough data to investigate this scenario and further investigation is therefore also required.
A. conjugella analyses in the study showed a lower expected heterozygosity (He = 0.31), than was found in various insect species such as; Metrioptera roeselii "0.61" [60]; Leptomyrmex pallens "0.51" [61] and Cerambyx cerdo "0.63" [62].However, the low expected heterozygosity was comparable to what was found in Plutella xylostella "0.35" [11], Anopheles nuneztovari (Culicidae) "0.34" [63], Glossina pallidipes "0.35" [64] and Bombus distinguendus "0.38" [65].The regionally synchronous and variable seed set of rowan, also known as masting, will have a great impact on A. conjugella as it is a seed predator.The masting in rowan may explain the low level of heterozygosity in A. conjugella.In Norway, it was suggested that the masting in rowan is an example of defence against seed predation by the process of predator satiation [6].In intermast years, the low berry production will inhibit and strongly influence seed predator population size.In the next masting year, rowan takes advantage of the reduced number of seed predators and produce a high crop with low losses to predators.The low level of expected heterozygosity may be caused by intermast years resulting in repetitive bottlenecks.Bottleneck processes may also have been caused by historically intensive use of insecticides to control A. conjugella, which have caused a selective pressure.
In conclusion, our study indicates substantial genetic differentiation and low genetic variation for the species in a restricted area in Southern Norway.However, we believe these findings should be investigated further in a much larger area, so we have identified multiple STRs (not shown), and we

Figure 1 .
Figure 1.The geographical locations of the sampling areas (blue dots; the size of the dot relative to the size of the sampling area).

Figure 1 .
Figure 1.The geographical locations of the sampling areas (blue dots; the size of the dot relative to the size of the sampling area).

Figure 2 .
Figure 2. PCO score plot of 95 A. conjugella individuals using 410 polymorphic AFLP bands.The geographical origin of each A. conjugella individual is indicated by the symbols listed at the right; red: Eastern region and blue: Western region (Eigenvalue for each axis is given in brackets).

Figure 2 .
Figure 2. PCO score plot of 95 A. conjugella individuals using 410 polymorphic AFLP bands.The geographical origin of each A. conjugella individual is indicated by the symbols listed at the right; red: Eastern region and blue: Western region (Eigenvalue for each axis is given in brackets).

Figure 3 .
Figure 3. Population genetics structure analyses of all 95 individuals in the study using STRUCTURE software program.The insects are sorted according to Id number as in Supplementary Table S1(blue: Western cluster and red: Eastern cluster).
Figure 3. Population genetics structure analyses of all 95 individuals in the study using STRUCTURE software program.The insects are sorted according to Id number as in Supplementary Table S1(blue: Western cluster and red: Eastern cluster).

Figure 3 .
Figure 3. Population genetics structure analyses of all 95 individuals in the study using STRUCTURE software program.The insects are sorted according to Id number as in Supplementary Table S1(blue: Western cluster and red: Eastern cluster).
Figure 3. Population genetics structure analyses of all 95 individuals in the study using STRUCTURE software program.The insects are sorted according to Id number as in Supplementary Table S1(blue: Western cluster and red: Eastern cluster).

Table 1 .
Overview of A. conjugella samples from Southern Norway, n = is the number individuals.

Table 1 .
Overview of A. conjugella samples from Southern Norway, n = is the number individuals.

Table 2 .
Sequences of the EcoRI and MseI selective primers used for AFLP analysis.

Table 2 .
Sequences of the EcoRI and MseI selective primers used for AFLP analysis.

Table 3 .
Analysis of molecular variance (AMOVA) of 95 A. conjugella between the two major (East and west) regions using 410 AFLP markers.

Table 3 .
Analysis of molecular variance (AMOVA) of 95 A. conjugella between the two major (East and west) regions using 410 AFLP markers.

Table 4 .
Pair-wise F ST values among the six sampling locations of A. conjugella in Southern Norway.

Table 5 .
Summary of the genetic diversity analyses of A. conjugella in Southern Norway based on 410 AFLP polymorphic bands.