Genetic structure and phenotypic diversity of two northern populations of Cheilosia aff. longula (Diptera: Syrphidae) has implications for evolution and conservation

The genetic structure and phenotypic diversity of two populations of Cheilosia aff. longula (Diptera: Syrphidae) in Lapland, Finland, were examined using DNA sequencing, protein electrophoresis, and geometric morphometrics. The morphological identification of the species were verified using partial sequences of mitochondrial cytochrome c oxidase subunit I (COI mtDNA), and the nuclear ribosomal internal transcribed region 2 (ITS2 rDNA), and comparing the corresponding sequences of Cheilosia aff. longula and the closely related C. longula. Two and three haplotypes of the genes COI mtDNA and ITS2 rDNA were identified in the two populations. Analysis of 12 isozyme loci showed an extremely low genetic variability in the populations originating from Utsjoki and Kevo. Discriminant analysis combined with canonical variate analysis revealed inter-population divergence in wing shape. Variation among genetically diverse individuals, both within- and among studied populations was studied, and directional (DA) and fluctuating asymmetry (FA) estimated using landmarks in the framework of geometric morphometrics. It is likely that the documented DA and FA asymmetry in both wing shape and size reflects the developmental instability of the individuals studied. By using Procrustes ANOVA the locations of particular landmarks responsible for the variation in shape were determined. The decomposition of the components of variance accorded to each landmark showed that the landmarks differed in the percentage of variation they accounted for (DA, FA and variation among individuals). In the discussion the implications of the reduced genetic diversity and asymmetry in wing traits for conservation is considered.


INTRODUCTION
Studies on species from Northern Europe that experienced a range of climatic changes during their history are of a special interest for studies of systematics and evolutionary and conservation biology.Until approximately 15000 years BP the whole of Northern Europe was covered with ice and both terrestrial and aquatic species experienced a hostile environment (Hewitt, 2000(Hewitt, , 2004)).The genetic consequences of repeated cycles of decreases and expansions of species' ranges, colonization and recolonization, is well documented for Fennoscandian biota (e.g.Vainio & Väinölä, 2003;Audzijonyte & Väinölä, 2006).Since genetic diversity is related to fitness and evolutionary capacity of populations (Hansson & Westerberg, 2002;Frankham, 2003;Grueber et al., 2008;Chapman et al., 2009), data on the genetic and phenotypic diversity of a Northern Boreal taxon of the family of Syrphidae (Insecta: Diptera) will provide information about the ability of the populations studied to respond to environmental changes.Hence, understanding the potential effects of selection and the possibilities for evolutionary changes has significant implications from both a fundamental and applied point of view.Therefore, this study of two northern populations of a taxon recently identified within the morphologically defined Cheilosia longula (Zetterstedt, 1838) species was undertaken.Molecular data indicates that C. longula and taxa close to C. longula belong to a monophyletic group, named the "longula" group (Claussen & Ståhls, 2007).This group comprise the species Cheilosia longula, C. soror (Zetterstedt, 1843), C. scutellata (Fallén, 1817), C. flavissima Becker, 1894, andC. thessala Claussen &Ståhls 2007 (Claussen & Ståhls, 2007).Based on the consistent haplotype difference at the 3' end of COI mtDNA (with uncorrected pairwise divergence of 2.88% from C. longula) and on distinguishing morphological characters, a new taxon provisionally named C. aff.longula was recognized (but not taxonomically formalized) within the C. longula group (Claussen & Ståhls, 2007).In this study molecular (DNA sequencing and protein electrophoresis) and morphometric data were used to determine the genetic and phenotypic intra-and inter-population diversity of two populations of C. aff.longula in Lapland, Finland.Two questions are addressed.
Firstly, do these populations differ both from a molecular and morphological point of view?To quantify intra-and inter-population phenotypic and genetic variation, the wing geometry, allozyme nuclear loci, mitochondrial cytochrome c oxidase subunit I (COI mtDNA), and nuclear ribosomal internal transcribed region 2 (ITS2 rDNA) were studied.Allozyme loci and DNA sequences of both COI mtDNA and ITS2 rDNA are widely used markers in population genetics, taxonomy, systematics, and conservation biology of hoverflies (e.g.Milankov et al., 2008Milankov et al., , 2009Milankov et al., , 2010;;Ståhls et al., 2009).In addition, landmark-based geometric morphometrics is used to quantify phenotypic variability (e.g.Hoffmann & Shirriffs, 2002;Ludoški et al., 2008;Olivieri et al., 2008;Francuski et al., 2009a, b).An investigation of wing geometry could be important assuming that wing shape is related to flight ability and thus to fitness (Kölliker-Ott et al., 2003).For example, Kölliker-Ott et al. (2003) report that both wing size and shape are adaptive, associated with differences in temperature and that size-related traits are under both directional and stabilizing selection.In addition, an adaptive cline in wing size in Drosophila is correlated with a polymorphism in a promoter region of the Dca (Drosophila cold acclimation) gene (McKechnie et al., 2010).
Secondly, are wing parameters of shape and size bilaterally symmetrical?Over the last three decades, three kinds of asymmetry have been subjected to genetical, development, and evolutionary studies (see review by Clarke, 1995;Van Dongen, 2006).The correlation between developmental instability and asymmetry of otherwise bilaterally symmetrical traits is well documented (e.g.Palmer & Strobeck, 1992;Clarke, 1993bClarke, , 1998b;;Leamy & Klingenberg, 2005;Stige et al., 2006;Van Dongen, 2006;Lens & Eggermont, 2008).However, the applicability of directional asymmetry (DA), in which one side is consistently larger than the other, antisymmetry (AS), deviations from symmetry occur on both sides, and fluctuating asymmetry (FA), nondirectional subtle differences in bilaterally symmetrical traits (Van Valen, 1962;Palmer, 1994) are subjects of intense debate (see review by Lens et al., 2002;Leamy & Klingenberg, 2005).Only FA is viewed as a reliable estimator and epigenetic measure of developmental homeostasis (Palmer & Strobeck, 1992;Stige et al., 2006;Van Dongen et al., 2009).It is suggested that FA is influenced by internal factors, including genetics (high level of homozygosity and disruption of co-adapted gene complexes), physiological changes caused by extreme environmental conditions and/or environmental deterioration (Palmer & Strobeck, 1992;Willmore et al., 2007).However, there are inconsistencies in the relationships between FA and inbreeding and some components of fitness (see review by Lens et al., 2002), which has provoked an intense debate on whether FA can be considered as a "biomarker" for evaluating environmental and genetic stresses (Clarke, 1995;1998a;Leamy & Klingenberg, 2005;Van Dongen, 2006).Still, the use of the FA of wings has proved to be an effective way of measuring developmental instability (e.g.Trotta et al., 2005;Van Dongen et al., 2009), which is important for predicting population persistence in the face of environmental and climatic change (Clarke, 1993b).It is therefore important to determine whether adaptive bilaterally symmetric traits are asymmetrical in natural populations.Hence, variation in bilaterally symmetrical traits that quantify the different components of asymmetry of wing traits (wing size and shape) were studied using a morphometric method.Finally, from a conservation point of view, it is of interest to know if this potential asymmetry in populations of C. aff.longula is linked to genetic heterogeneity.

Sample collection
Claussen & Ståhls' (2007) COI mtDNA sequence data indicates the existence of a new taxon closely related to Cheilosia longula (Zetterstedt), but it is not formally designated.This species is included in recent faunal taxonomic studies along with a description of its diagnostic morphological characters, as it is abundant in Northern Scandinavia (Haarto & Kerppola, 2007;Bartsch et al., 2009).To verify the identity of specimens identified using a morphological key for separating all longula group taxa (see Haarto & Kerppola, 2007) in northern Europe, the 3' end of COI mtDNA sequences of C. longula and C. aff.longula were compared.According to the identification based on both morphology and COI haplotypes the specimens belong to C. aff.longula.
Two populations of Cheilosia aff.longula, one at Kevo (69°45.515´N,26°59.428´E)and the other at Utsjoki (69°54.47´N,27°1.592´E)(Lapland, Finland; Fig. 1) were sampled, and 46 and 18 individuals, respectively, collected in July 2007 (leg. Milankov V.).Hoverflies were stored at -20°C until used in the allozyme analysis.Altogether, 41 and 64 specimens of C. aff.longula were included in the allozyme and wing morphometric analyses, respectively.Wing traits and allozyme variation of the same specimens were analyzed and the legs of 17 specimens used for DNA sequencing.

DNA extraction and amplification
DNA was extracted from legs of the flies using a Nucleospin Tissue DNA extraction kit (Machery-Nagel, Düren, Germany) following the manufacturer's protocols, and then the DNA was re-suspended in 50 µl of ultra-pure water.Remains of specimens used for the geometric morphometric analysis, allozyme study, and DNA extraction are deposited at the University of Novi Sad, Faculty of Sciences, Department of Biology and Ecology (Novi Sad, Serbia).PCR and sequencing followed the protocols described by Milankov et al. (2009).

Wing geometry
Geometric morphometric analysis was done on 46 (18%, 28&) and 18 (9%, 9&) specimens of C. aff.longula from Kevo and Utsjoki, respectively.The right and left wings of these flies were mounted in Hoyer's medium between microscope slides.A Leica DFC320 digital camera connected to a Leica MZ12.5 stereomicroscope was used to take pictures of the wings.The positions of 16 landmarks at vein intersections or terminations were assigned using TpsDig 1.40 and expressed as x, y coordinates in Cartesian space (Rohlf, 2004) (Fig. 2).
Geometric morphometrics provides a basis for understanding the variation in wing shape.It provides size-independent measures of shape, allowing unambiguous separation of size and shape.Variation in wing size was examined using centroid size (the square root of the sum of squared distance between each landmark and the wing centroid), which is an isometric estimator of size.One-way analysis of variance (ANOVA) was used to test differences in centroid size among populations and sexes.
For the analysis of wing shape the landmarks on each specimen were first aligned using a Generalized Procrustean Analysis procedure to remove the non-shape effects of translation, rotation, and scale (Rohlf & Slice, 1990) and then a thinplate spline analysis was done.The resulting matrix (w; "weight matrix" of Rohlf et al., 1996) was used for a discriminant analysis combined with canonical variate analysis (CVA) to examine a classification matrix and the pattern of within-species/populations variation in total shape space.To visualize the morphological variation obtained by CVA, individual canonical scores for each root were regressed against the x and y coordinates of the original landmark positions (Rohlf et al., 1996).Differences in wing shape and size were analyzed by comparing the phenotypic traits of females and males separately.
Procrustean superimpositions, calculation of the centroid size and w matrix were performed using TpsRelw 1.44 (Rohlf, 2006), multiple regressions and visualization of deformation grids were computed using TpsRegr 1.31 (Rohlf, 2005).All morphometric software of the Tps series are freely distributed at http://life.bio.sunysb.edu/morph/.ANOVA and CVA were done using Statistica for Windows (version 8.0).

Genetic analysis
Genotype and allelic frequencies were calculated directly from the observed banding patterns based on the genetic interpretation of zymograms.Calculated parameters of population genetic structure were corrected using Leven's (1949) formula for small samples.Genetic polymorphism in each population was measured using the mean number of alleles per locus (A), observed (Ho) and expected (He) heterozygosities.Allele frequencies and deviations from the Hardy-Weinberg equilibrium were calculated using BIOSYS-2 computer programme (Swofford & Selander, 1989).Assessment of genetic differentiation between populations was determined using Nei's (1978) genetic identity coefficient (I).
Uncorrected pairwise divergences (p-distances) were calculated based on the COI sequences of 17 specimens.

Asymmetry
Here, the levels of FA in multiple wing traits (size and shape) were assed within and between the two populations of C. aff.longula.For this purpose, wing asymmetry, based on the statistically significant right-left differences in wing size and shape, was studied.Asymmetry was measured using both the left and right wings of 44 (18% and 26&) and 12 (6% and 6&) specimens from Kevo and Utsjoki, respectively.
Asymmetry in wing shape and size was estimated using geometric morphometrics.Two-way ANOVA on repeated measurements (Palmer, 1994) was used to assess asymmetry in centroid size.In the model implemented herein, "Individuals" (I) refers to a random factor that assesses variation among individuals, "Sides" (S) is a fixed factor that assesses directional asymmetry (DA), the Individuals × Sides interaction assesses fluctuating asymmetry (FA), and the error assesses variation in the replicate measurements (Leamy, 1984;Palmer, 1994).The Procrustes analysis was used to compute best-fitting superimpositions of configurations of landmarks to the left and right sides of a single specimen.Asymmetry in wing shape was tested using Procrustes ANOVA (Klingenberg & McIntyre, 1998) with the landmark coordinates of the Procrustes-aligned configurations (all three replicates of each wing) as the data.The MS (mean square; obtained using regular ANOVAs and Procrustes ANO-VAs) related to the individual effect and to the individual × side interaction were used as an estimator of the amount of individual variation and FA for wing size and shape.To examine the effects of particular landmarks of shape changes, separate ANOVAs on landmark coordinates were calculated and then the 307 Fig. 2. The locations of 16 landmarks on wing of the C. aff.longula species selected for geometric morphometric analysis.
x-and y-mean squares of each landmark were summed (Klingenberg & McIntyre, 1998;Badyaev & Foresman, 2000).The multivariate patterns of landmark covariation were studied using principal component analysis (PCA) following Klingenberg & McIntyre (1998) and Klingenberg & Zaklan (2000).To study the developmental basis of FA and variation among individuals covariance matrices were computed and compared with multivariate patterns of landmark covariation corresponding to within-individual variability and variation among individuals.Analysis of asymmetry of wing size and shape was done using Statistica for Windows (version 8.0).

Mitochondrial COI and nuclear DNA diversity
A 764 bp fragment of the 3' end of COI mtDNA gene, corresponding to nucleotide positions 2207 to 2970 in the Drosophila yakuba sequence (Clary & Wolstenholme, 1985), was obtained from nine and eight specimens of the populations sampled at Kevo and Utsjoki, respectively.Additionally, a fragment of nuclear ITS2 rDNA was obtained from seven and eight specimens from Kevo and Utsjoki, respectively.All sequences are deposited in Gen-Bank and the accession numbers of the analysed specimens are FJ158617-FJ158648.
Sequencing of the 3' end of COI revealed three haplotypes, two of which were unique to these populations.A private haplotype of one specimen from Utsjoki (acc.no.FJ158624) was different in one and two nucleotides from the common and a private haplotype recorded in the Kevo population, respectively.Another unique haplotype observed in a specimen from the Kevo population (acc.no.FJ158619) also differed from the common haplotype at one nucleotide site.Thus, uncorrected pairwise divergences (p-distances in %) estimated among the common haplotype and private haplotypes, and between the two private haplotypes of the populations were 0.13% and 0.29%, respectively.The 14 sequences of ITS2 rDNA were identical, and only one specimen from Utsjoki was different (acc.no.FJ158618).However, this specimen possessed a common haplotype of COI mtDNA.

Allozyme variability
Data from 15 and 26 individuals from Utsjoki and Kevo, respectively, were obtained.Of the 12 isozyme loci analyzed, 10 were monomorphic for the same alleles in both populations.The Me locus was polymorphic in both populations and Pgm was variable in the population at Kevo (Table 1).The spatial distribution of the two alleles (Me b , Me c ) in both populations and one (Me a ) in the Kevo population, indicate genetic divergence among conspecific populations.Only at the Pgm locus in the population from Kevo were two heterozygotes, Pgm a/b (frequency 0.115) and Pgm b/c (frequency 0.077), detected.Thus, within the population from Kevo there are two unique alleles at Pgm (Pgm a , Pgm c ).The average number of alleles per locus (A) was 1.1 and 1.3 for the populations at Utsjoki and Kevo, respectively (Table 1).
Chi-squared tests for deviations of observed genotypes from expected revealed statistically significant departure from Hardy-Weinberg equilibrium at the Me locus in both populations, but not at Pgm for the population at Kevo.Consequently, the observed heterozygosity (Ho) was generally lower than expected heterozygosity (He), and the genotype fixation index, FIS, indicated excess homozygosity (FIS > 0) at all analysed loci in the populations, except Pgm in the population at Kevo.The value of genetic identity between populations was I = 0.998.

Asymmetry
ANOVA of the FA of both wing shape and size for both sexes from the Utsjoki and Kevo populations revealed that the main effect of Individuals and the Individual × Side interaction were highly significant.FA was also estimated based on the observed significant Individual × Side interaction.For both traits, ANOVA indicated significant differences among individuals and therefore significant FA and DA (Tables 2 and 3).
A more differentiated pattern emerged when variance components from the Procrustes ANOVA of wing shape apportioned by landmarks in both sexes were analysed separately.For DA (Side effect) the largest share of the variation is located at the landmarks defining outer margin (landmark 7 in males from both populations and 1 in females from Utsjoki) and the position of the crossveins (10, 11, 15, and 16 in females from Kevo).Thus, in contrast to landmark 7 in males, which has a proportional share in DA, there was a disproportionate variation of different landmarks in DA in females from these populations.The largest left-right differences within individuals (FA; S × I effect) were found at the basal landmark 16 in both sexes from both populations and at the anterior outer wing margin in males (1 and 2) and females (1) from Utsjoki.Unlike the population from Utsjoki, the amount of FA variation was largest at landmark 7 located in the posterior part of the wing.In contrast to landmark 16, which had large amount of shape variation in FA in both sexes from Kevo and Utsjoki, other landmarks had disproportionate amounts of variation in these populations.Finally, the effects of individual variation (I) were distributed mainly among landmarks defining the outer margin of the posterior part of the wing (7 and 8 in both sexes from Kevo, and 7 and 6, 8 in females and males from Utsjoki, respectively).Landmarks defining the anterior part of the wing (2, 4 in males and 2 in females) had the largest individual effect (variation among individuals) in specimens from Kevo and inner landmarks 12 and 13 the largest effect on variation in shape in both sexes from Utsjoki and males from Kevo, respectively (results not shown).
In order to determine whether within-individual variability (FA) and variation among individuals have similar patterns of landmark covariation the patterns in landmark positions were compared.PCA of the matrices of variance and covariance components for the different ANOVA effects revealed that most of the variance among individuals and most FA (individual × side interaction) variance in both sexes in the two populations was concentrated in the first few dimensions, with slightly more for individual variation than for FA for both females and males from Kevo.Unlike the population from Kevo, variation accounted for by FA was slightly more than variation among individuals in both sexes originating from Utsjoki.Similar principal components (PCs) for both within-individual variation (FA) and variation among individuals differ from those attributed to measurement error in both populations (Figs 5 and 6), which implies that similar developmental processes control fluctuating asymmetry and individual variation.In contrast to the amount of FA in shape, individual variation varies within samples, especially the female sample from Utsjoki (Fig. 7).

Genetic and phenotypic diversity
Regarding the question whether there is genetic and phenotypic differentiation among northern hoverflies our results reveal inter-population divergence between populations of C. aff.longula.A comparison of the genetic diversity values of populations at Kevo and Utsjoki indicated differences in the spatial distributions of alleles at the Me and Pgm loci.Private alleles at the Pgm and Me loci in the population from Kevo and phenotypic divergence in wing shape (contrary to wing size) are good indicators of population structuring.Indeed, each of the populations studied has a distinct wing shape, suggesting clear inter-population phenotypic divergence.This finding is important since differences in wing shape and size are constrained genetically and regarded as adaptive.Although the genetic determination of morphometric characters are still relatively poorly known, Birdsall et al. (2000) suggest that genes that regulate wing shape are more tightly connected than those that determine wing size and the former are probably associated with determination of wing veins.Moreover, McKechnie et al. (2010) reports an adaptive variation in wing size in populations of Drosophila melanogaster across climatic gradients and an association between wing size and Dca alleles.Genetic basis of wing size polymorphism is additionally sup-  3. Two-way ANOVA of centroid size and Procrustes ANOVA of wing shape based on repeated measurements and used to asses directional asymmetry (sides effect) and fluctuating asymmetry (individuals × sides interaction effect) in the population of C. aff.longula from Kevo.
ported by the linkage between the Dca gene and genes located within the In(3R)P inversion (Kennington et al., 2007).Thus, it is likely this cline in size is a result of thermal selection acting on flight (Blanckenhorn & Demont, 2004).Hence, a lack of phenotypic divergence in wing size documented for C. aff.longula probably reflects similar thermal regimes at the localities of the populations studied.
In addition, slight variation in both mitochondrial (a common COI haplotype is dominant in both populations) and nuclear data (monomorphic allozyme genes and presence of identical ITS2 rDNA sequences in all specimens studied, except one originating from Utsjoki) was observed.This low level of intra-population genetic variation at allozyme loci is recorded in hoverfly populations in small, spatially fragmented and/or isolated populations of taxa within the genera Melanogaster, Merodon, and Cheilosia (Milankov et al., 2010).The low level of genetic diversity detected in this study is similar to that recorded for the northern hoverfly, Cheilosia naruska Haarto & Kerppola, 2007(Milankov et al., 2010).Although there is little information on the genetic diversity of northern populations [data only for C. naruska (Milankov et al., 2010) and that presented in this study] compared with southern taxa, it seems evident that less genetic variation might be expected.In comparison with Southern Europe, there are fewer species and less genetic diversity in Northern Europe (Hewitt, 2000).Thus a reduced genetic diversity might be characteristic of northern populations of both hoverfly species, C. naruska (Milankov et al., 2010) and C. aff.longula (reported herein).The similar levels of genetic variation in northern hoverfly species suggest that they have experienced similar ecological factors and evolutionary histories.
Here it is documented that homologous landmarks tend to have opposite displacement in females and males.For example, landmarks 4 and 5 are in a more distal position in males than females, landmark 2 is a more anterior in males, and landmarks 6, 7, and 8 are more posterior than those in females (Fig. 4).These results are similar to those of studies on variation in wing compartments in which functional constraints determine wing shape (Gilchrist et al., 2000).

Asymmetry
The second aim, to study the left-right axis in two populations of a northern species of hoverfly involved the analysis of wing size and shape using a geometric morphometric method to quantify the left-right differences.This revealed the presence of both DA and FA in both the size and shape of a bilaterally symmetrical organ, the wings.In addition, there were no differences between sexes in the level of FA.Wing left-right asymmetry is thought to be due to the separate development of imaginal discs (Klingenberg et al., 1998) and that the same perturbations influence the developmental pathways of two traits.This asymmetry is likely to affect wing aerodynamic properties and hence impede flight performance.Asymmetry in wing shape is reported in number of insect species, for example, Apis mellifera (Smith et al., 1997), Drosophila melanogaster (Breuker et al., 2006;Klingenberg et al., 1998), Musca domestica and Glossina palpalis gambiensis (Klingenberg et al., 1998), and Carabus solieri (Garnier et al., 2006).Currently, there is little information on the variation between left-right wing traits in hoverflies.For instance, FA of wing size is recorded in a population of a northern species, Cheilosia naruska (Milankov et al., 2010) and populations on Aegean  islands of a southern species Merodon albifrons (V.Milankov, J. Ludoški, L. Francuski, G. Ståhls, A. Vuji , unpubl. data).However, unlike in the northern populations of both Cheilosia species (C.naruska, Milankov et al., 2010;C. aff. longula, reported herein) and the autumn generation of M. albifrons, FA of wing shape was not detected in the spring generation of M. albifrons (V.Milankov, J. Ludoški, L. Francuski, G. Ståhls, A. Vuji , unpubl. data).Similar relationships between individual and FA of wing shape and size in Drosophila are attributed to the different genetic properties of both traits (Breuker et al., 2006).
Furthermore, the variation among individuals and within-individual asymmetry (FA) between body sides were examined and the multivariate patterns of landmark covariation throughout the wing compared.There were differences in the degree of variability among landmarks in both populations when the sexes were considered separately.These findings revealed distinct patterns of landmark displacements at the different levels, which implied different amounts of variation in localized shape in the overall variability.For example, the significant components of variation (both within and among individuals) were confined to the outer margin of the wing.In addition, the PCA showed that FA and individual variation are concentrated within the first few principal components (PCs).Also contrary to the observed similar patterns in FA and individual variation, the distribution of eigen values for measurement error was markedly different for both sexes in the populations studied.Similar patterns of PCs for individual variation and FA in the Drosophila wing are linked to the effects of subtle perturbations during development of the wing (Klingenberg & Zaklan, 2000).In addition, concordance between individual variation and FA is thought to reflect a similar response of the developmental processes to perturbations during development, and, therefore, a lack of distinct developmental processes that generate variation within individuals (FA) (Klingenberg & McIntyre, 1998).

Conservation implications
The low genetic variation documented herein is possibly correlated with a lower evolutionary potential, which is already suggested for endangered and threatened species (e.g.Reed & Frankham, 2003;Frankham, 2005;Grueber et al., 2008).This relationship between genetic diversity and the evolutionary potential of a population is of great importance; low genetic variation influences short-term decreases in fitness through inbreeding depression and long-term lack of adaptive flexibility (e.g.David, 1998;Frankham et al., 2002;Hansson & Westerberg, 2002;Grueber et al., 2008;Malo & Coulson, 2009).Because of this, studies on heterozygosity-fitness correlations are highlighted in conservation genetics and management of threatened species (Hansson & Westerberg, 2002;Grueber et al., 2008;Chapman et al., 2009).As inbreeding depression and loss of genetic diversity increase the risk of extinction (Frankham, 2005;Grueber et al., 2008), populations of C. aff.longula with reduced genetic diversity are likely to have a low potential for adapting to a change in the environment.
Due to their functional importance and, therefore, high degree of developmental canalization (influenced by strong stabilizing selection) (Møller, 1993), wings can be used to study the ability of individuals to cope with environmental and internal stress (see review by Palmer & Strobeck, 1986).Both the FA and DA in wing shape and size recorded in both populations indicate that wing discs experience disturbance during early embryogenesis.It is likely that a shift in the developmental pathway might result from both exterior and interior factors, including nutritional stress and lack of heterozygosity.One of the possible factors influencing observed FA in C. aff.longula might be an unstable developmental environment caused by fast and unpredictable deterioration in the food Fig. 7. Levels of Individual variation and Fluctuating asymmetry for size (A) and shape (B).Each graphic shows MS value corresponding to male and female individuals from Utsjoki and Kevo.* P < 0.001 of the larvae.Recent re-examination of previously reared material of G. Ståhls (unpubl.data) revealed that C. aff.longula is only reared from Leccinum versipelle and L. scabrum, while the closely related Cheilosia longula is mainly reared from species of the genus Suillus.A study of C. longula reared from Leccinum spp. in Western Finnish Lapland showed that they are as abundant in sporophores as Pegomya spp.(Diptera: Anthomyiidae) (> 20 individuals/sporophore).In addition, the size of the individual flies (as measured by length of thorax) is negatively correlated with the number of flies emerging, which suggests an increase in competition with increasing numbers of larvae in individual sporophores (Ståhls et al., 1989).Therefore, the development of larvae exploiting an ephemeral food source like a fungus is subject to environmental and biotic factors affecting the larval life-cycle that are temporally and spatially highly variable and likely to affect the quality of the substrate.Likewise, stressful conditions, such as a poor or a lack of the preferred food source, are likely to affect wing development in phytophagous insects.For example, Soto et al. (2008) reveal that levels of FA in wing size in Drosophila buzzatii and D. koepferae depend on the nature of the breeding substrate.
On the other hand, the wing shape of heterozygote and homozygote individuals differ (Messier & Mitton, 1996).Our results accord with previous findings, as a correlation between heterozygosity based on variation in proteins and lower FA is documented in several studies (e.g.Palmer & Strobeck, 1986;Clarke 1993a, Hartl et al., 1995;Messier & Mitton, 1996;Smith et al., 1997;Vøllestad et al., 1999;Borrell et al., 2004;Milankov et al., 2010).It is argued that genomic heterozygosity in protein loci influences developmental stability in such way that increasing homozygosity affects the (in)ability of organisms to buffer against developmental instability (refers to the direct effect hypothesis of the multilocus heterozygosityfitness correlations) (Møller, 1993;Kark, 2001;Hansson & Westerberg, 2002;Grueber et al., 2008).In addition, the different internal environments an individual experiences during early embryonic development determine the effect that decreased heterozygosity has on developmental instability, which is reflected in a high level of FA, is more important in ectotherms than endotherms (Vøllestad et al., 1999).In addition, presence and strengths of relationships between FA, stress and fitness depend on trait type and species (see review Lens et al., 2002).Since the genetic basis of developmental stability might be character-, population-, and taxon-specific (Clarke, 1998b;Hansson & Westerday, 2002), and affected by a variety of external (abiotic and biotic) and internal (genetic) stresses (Trotta et al., 2005), FA should be correlated between traits within individuals.Given all the factors that influence development (mentioned above), the inconsistent results of a wide range of studies on the association between heterozygosity and developmental stability in diverse taxa is not surprising (Palmer & Strobeck, 1986;Clarke et al., 1992;Clarke, 1993a;Vøllestad et al., 1999).
However, as little is known about the evolutionary implications of developmental instability in hoverflies it is hoped that the results presented herein will stimulate further research on the processes affecting the development of wings.The next step in this study will be a simultaneous analysis of genetic (heterozygosity) and environmental (nutrition) factors, and the asymmetry of multiple traits in the species studied.Finally, integrating morphometric and genetic analyses could prove a promising tool for evaluating the evolutionary potential of natural populations and understanding the evolutionary processes involved.

Fig. 1 .
Fig. 1.Map of population sampling locations from Finnish Lapland.The collecting sites Kevo and Utsjoki are at ca 20 km distance.

308n
Fig. 3. Boxplot of centroid size of populations of the C. aff.longula species with the mean, standard error, and standard deviation illustrating between sex and interpopulation variation in right wing size.Non-significant differences were calculated between the populations from Utsjoki and Kevo.

Fig. 4 .
Fig. 4. Scatterplot of individual scores from the CVA showing shape differentiation of right (A) and left (B) wing between populations and between sexes of C. aff.longula.The amount of variation explained by each axis is in parentheses.Thin-plate spline reconstruction representing negative (C) and positive (D) deformations of mean shape between genders along the CV1 axis.Deformation grids are exaggerated × 3. Numbers in the deformation grids refer to landmarks shown in Fig. 2.

Fig. 5 .
Fig. 5. Percentage of total shape variation taken up by the principal components of the matrices of variance and covariance components for individual variation, fluctuating asymmetry, and measurement error in population from Kevo.

Fig. 6 .
Fig. 6.Percentage of total shape variation taken up by the principal components of the matrices of variance and covariance components for individual variation, fluctuating asymmetry, and measurement error in population from Utsjoki.

TABLE 2 .
Two-way ANOVA of centroid size and Procrustes ANOVA of wing shape based on repeated measurements and used to asses directional asymmetry (sides effect) and fluctuating asymmetry (individuals × sides interaction effect) in the population of C. aff.longula from Utsjoki.