Genetic diversity was studied at allozyme loci in two Palearctic and one Nearctic population of Aedimorphus (=Aedes) vexans, a species of public health and veterinary importance. The population from Serbia was the most polymorphic (P= 35%) with the highest observed heterozygosity (Ho= 0.027). The lowest observed heterozygosity (Ho= 0.010) was obtained for the Nearctic population. All analyses based on individual (STRUCTURE analysis) and population level (pairwise FST,Nm values, AMOVA, Nei's D value) revealed significant structuring between Nearctic and Palearctic populations, indicating a lack of gene flow and thus, the presence of independent gene pools. Taxon-specific alleles at the diagnostic Ao, Hk-2, Hk-3, Hk-4, Idh-1, and Idh-2 loci were used for identification and separation of Nearctic and Palearctic populations. Population genetics study provided valuable information on the correct distinction of Am. vexans populations and their adaptive potential that could find a future use in the studies of vector competence and development of vector-control strategies.
The study of intra- and inter-population genetic diversity provides insights into epidemiologically important species influenced by evolutionary factors, including gene flow and different selection regimes (Poolprasert et al. 2008). In addition, genetic variation provides raw material for change within populations that are subjected to new pressures, especially anthropogenic changes in environment such as insecticide application (Paupy et al. 2005). Since genetic variation can represent the potential to respond to the selection pressures of insecticides (Paupy et al. 2005), genetic diversity is thought to have implications for vector control programs (Tabachnick and Black 1995, Singh et al. 2004). Indeed, genetic diversity has been used for defining vector capacity of the epidemiologically important species (Poolprasert et al. 2008). Hence, the study of the adaptive capacity of a vector of human and animal disease to rapid ecological change is important to public health (Patz et al. 2000, Manguin and Boëte 2011, Tabachnick 2010). Furthermore, the consequence of global warming is the change in the geographical distribution of some vector species, and consequently, pathogens they transmit. Minimal changes in local climate could initiate the replacement of particular populations, which is, within the same species complex, very difficult to distinguish but may change the epidemiological situation in a given locality (Olejnicek and Gelbic 2000).
Moreover, understanding the biology of the vector species and their distribution and correct distinction is essential for maintaining successful vector control programs (Ndo et al. 2010). Taxon identification is of great importance in the case of cryptic species, i.e., groups of closely related but reproductively isolated species that are difficult or impossible to identify based on morphological characters (Collins and Paskewitz 1996). Cryptic taxa could vary in their vector potential, host preference, and/or response to insecticides (Mousson et al. 2002, Singh et al. 2004, Yébakima et al. 2004). Therefore, detection of species borders and intraspecific divergent units of taxa within vectors of pathogens are essential for both fundamental (taxonomy and systematics) and applied (biology, human welfare, ecology) research (e.g., Bickford et al. 2007).
The relationship between human health and mosquitoes that are medically important (200 of 3,523 currently recognized; Becker et al. 2010) has driven most mosquito research (Reidenbach et al. 2009). Given that the tribe Aedini is the largest and most polyphyletic group of mosquitoes, we agree with the work of Reinert (2000) and Reinert et al. (2004, 2008, 2009), in revising it and establishing monophyletic genera that is in congruence with recent molecular studies (Cook et al. 2005, Shepard et al. 2006). Phylogenetic grouping based on natural relationships will hopefully lead to a more effective means of identification and provide an evolutionary framework for biological studies. Even though this reclassification has generated disagreement among mosquito biologists and systematists (Black 2004, Savage and Strickman 2004, Savage 2005), and has lead several professional journals (Higgs 2005) to take a conservative position, the number of authors using the new classification is rising nowadays.
Genus Aedimorphus (=Aedes) defined by morphological traits (Reinert et al. 2009) includes 67 species, seven of which have recognized subspecies. Most species of Aedimorphus occur in the Afrotropical region, some in the Oriental and Australasian regions, and one, Aedimorphus (=Aedes) vexans (Meigen), occurs in the Holarctic and Oriental regions, Central America, South Africa, and the Papuan area. Aedimorphus vexans includes three subspecies: Am. vexans vexans (Meigen 1830), Am. vexans arabiensis (Patton 1905), and Am. vexans nipponii (Theobald 1907) (Knight and Stone 1977, Reinert et al. 2004) defined by differences in adult traits including tergites, mid-tibia, and male palpi (White 1975), while a “barcoding” fragment was found to be partially useful (Cywinska et al. 2006). Aedimorphus vexans is a polycyclic species distributed almost worldwide (Becker et al. 2010). Predominantly breeding sites are inundated areas such as floodplains of rivers and lakes with fluctuating water levels (Becker et al. 2010). Species composition studies from Russia, Croatia, Serbia, Czech – Austrian border, and the United States confirm its abundance (Merdić and Lovaković 2001, Kent et al. 2003, Fyodorova et al. 2006, Vujić et al. 2010, Šebesta et al. 2012). Apart from being nuisance pests, Am. vexans can competently transmit more than 30 viruses (Horsfall and Novak 1991). In the United States, viral diseases of veterinary and public health importance, such as West Nile virus (Goddard et al. 2002) are known to be transmitted by Am. vexans. In Northern Africa and Saudi Arabia, the species has become an important vector in outbreaks of Rift Valley fever (Fontenille et al. 1998, Jupp et al. 2002, Diallo et al. 2005). In Europe, Tahyna virus is widespread in populations of this species (Lundström 1994). Also, this species can competently transmit the L3 stage of the nematode Dirofilaria immitis, causal agent of dog heartworm (Ludlam et al. 1970, Yildirim et al. 2011).
Biochemical genetics, particularly studies on electrophoretic variation in enzymes, have been widely applied to vectors and have revealed that a significant amount of genetic variation exists in most vector populations (Gooding 1996). Indeed, Solorzano et al. (2010) demonstrated that neither Hurricane Katrina nor intense control programs effected the genetic diversity of Am. vexans from New Orleans. Hence, genetic data provide knowledge on the adaptive capacity of a vector of human and animal disease to rapid ecological change and have public health importance (Patz et al. 2000, Manguin and Boëte 2011, Tabachnick 2010). Taking into account that population study is essential for understanding mosquito biology, ecology, and genetics as well as for identifying the status of epidemiologically important species, we dealt with populations of Am. vexans. The first objective of this study was to quantify genetic diversity of two natural populations from Europe (Serbia and Germany) and one population from the Nearctic (U.S.A.) using allozyme nuclear loci. The second aim was to analyze genetic relationships between Nearctic and Palearctic populations. The results reported herein provide an insight into the potential for the adaptive evolution of Am. vexans and a basis for forthcoming studies of these taxa.
MATERIALS AND METHODS
During the 1998 season adult females of natural populations of Am. vexans were collected from three different localities: one Nearctic (Coachella Valley CA, U.S.A.; collector and det. B. Božičić Lothrop) and two Palearctic – Serbia (Danube valley near Novi Sad, Serbia; collector and det. Krtinić, B.) and Germany (Upper Rhine valley; collector and det. Kaiser, A.). Dry ice-baited miniature traps were used for collecting mosquito sampling. All adults from each locality were collected at the same time.
A total of 295 specimens (U.S.A. – 107 specimens; Serbia – 81; Germany – 107) were included in the allozyme analysis. Allozyme variation was studied by vertical polyacrylamide gel electrophoresis (PAGE). Enzymes were extracted from the whole body and homogenates were centrifuged at 15,000 rpm for 5 min at 5° C. Allozyme polymorphism was studied at 17 different loci (Table 1). The Tris-Boric-EDTA buffer system (pH 8.9) was used to assay AO, EST, GPI, HK, ME, PGM, and SOD, while the Tris-Citric buffer system (pH 7.1) was used in the analysis of AAT, GPD, G6PDH, HBD, and IDH. The duration of electrophoretic runs at 90 mA (135–220 V) was three to four h.
Table 1. Enzymes and loci investigated (EC number and name) and buffer systems used in electrophoresis.
Specimens from all samples were run concurrently on all gels to facilitate comparison of electrophoretic mobility. Loci and alleles were numbered with respect to order of increasing anodal migration. The fastest allele at each locus was given a score 90. All other alleles were numbered relative to that allele from the point of application. Genotype and allele frequencies were calculated directly from the observed banding patterns based on the genetic interpretation of zymograms.
Population genetic structure and genetic differentiation
Calculated parameters of population genetic structure were corrected using Levene's (1949) formula for small samples using the computer program BIOSYS-2 (Swofford and Selander 1989). The analysis included determining genotypic and allelic frequencies, percentage of polymorphic loci (P), the mean observed (Ho) and expected (He) heterozygosity, and the presence of rare and major alleles. Rare alleles were defined as variants with frequencies of ≤ 0.05 (Ayala et al. 1974, Munstermann 1994). Although the distinction between commonness and rareness is arbitrary, in this study major alleles are defined as present at frequencies of ≥0.5 (Ghosh et al. 1999). The diagnostic value of allozymes was calculated after Ayala and Powell (1972). The deviation between Ho and He of separate variable loci was evaluated using Wright's inbreeding coefficient (Fis, Wright 1951) and mean F-statistics (Weir 1996). Theta (θH) was estimated from the expected homozygosity (Ewens 1972). The θ estimates, which are equal to 2Nu (N is the effective population size and u is the average mutation rate per locus per generation), were based on the infinite alleles model (IAM). Genetic diversity was assessed by calculating average gene diversity over loci using ARLEQUIN 3.11 software (Excoffier et al. 2005).
Additionally, we used STRUCTURE software (Pritchard and Wen 2003) to infer genetic population structure at both population and individual levels. All individuals were combined into one dataset for analysis, without any a priori population assignments and admixture was allowed with a single value of Δ inferred for all populations. We evaluated K values, the number of assumed populations, from 1–3 using a burn-in of 10,000 and 10,000 Markov chain Monte Carlo (MCMC) for each value of K. Each value of K was run five times to evaluate stability. The model choice criterion implemented in structure to detect the true K is an estimate of the posterior probability of the data for a given K, Pr (XK) (Pritchard and Wen 2003). This value, called ‘Ln P (D)’ in structure output, is obtained by first computing the log likelihood of the data at each step of the MCMC. Then the average of these values is computed and half their variance is subtracted from the mean. This gives ‘Ln P (D)’, the model choice criterion to which we refer to as L (K) afterwards. The true number of populations (K) is often identified by using the maximal value of L (K) returned by STRUCTURE (Zeisset and Beebee 2001, Ciofi et al. 2002, Vernesi et al. 2003, Hampton et al. 2004). However, we observed that in most cases, once the real K is reached, L (K) at larger Ks plateaus or continues increasing slightly (a phenomenon mentioned in the structure's manual, Pritchard and Wen 2003). Following Evanno et al. (2005), we calculated ΔK, which corresponds to the rate of change of the likelihood between successive K values, using the STRUCTURE HARVESTER v0.6.92 (Earl and vonHoldt 2012) to identify the best supported values of K. Still, it is important to take into consideration that estimating the most likely number of clusters needed to explain the observed data is challenging and the results may be sensitive to the number of loci used, the variation at these loci, the rate of gene flow, and the number of individuals typed (Evanno et al. 2005, Huelsenbeck and Andolfatto 2007).
We used two tests in Alleles in Space software (AIS) (Miller 2005) to examine the correlation between genetic and geographic distance: a Mantel test and spatial autocorrelation. The Mantel test determines correlation between genetic and geographic distances. Spatial autocorrelation compares the average genetic distance between pairs of individuals over distance classes and is summarized with the global statistic V. The spatial autocorrelation algorithm in AIS does not use allele specific patterns, but rather calculates the statistic Ay, the average genetic distance between individuals in distance class y (0 = all individuals in distance class are identical, 1 = all individuals are completely dissimilar), and the difference between the Ay for each distance class against the mean Ay is tested with a randomization test. For both analyses, Mantel test and spatial autocorrelation, significance was tested using 10,000 permutations.
Genetic differentiation, measured as Wright's FST values (Weir and Cockerham 1984), was estimated by Arlequin software, version 3.11 (Excoffier et al. 2005) and the significance between each comparison pair was evaluated through 1,000 permutation procedures. We performed locus by locus analysis of molecular variance (AMOVA) using ARLEQUIN 3.11 software (Excoffier et al. 2005) separately at two levels: (1) among all populations (non-grouped) and (2) between groups (Nearctic vs Palearctic) as indicated by our STRUCTURE and FST pairwise results. Ten thousand permutations were used to determine significance of variance components. Genetic differentiation among populations was also assessed using the genetic distance coefficient (D) of Nei (1972).
Population genetic structure
Twelve enzyme systems representing products of 17 gene loci were assayed in three populations of Am. vexans. Analysis of allozyme variability revealed 41 alleles, from which 32, 29, and 28 were detected in Serbia, Germany, and U.S.A. populations, respectively. Six loci were monomorphic with a common allele in all populations analyzed (Aat, Gpd-2, G-6-pdh, Hbd, Me, and Sod). The Hk-2, Hk-3, Hk-4, and Idh-1 loci were monomorphic in all three populations but with different alleles. In all populations Ao, Est-1, Est-6, Gpi, and Pgm-2 loci were polymorphic. The Idh-2 locus was polymorphic only in Serbia while Pgm-1 was variable only in the U.S.A. The largest number of alleles and genotypes were identified at Est-6 locus (Table 2).
Table 2. Allelic frequencies at 11 loci (Aat, Gpd-2, G-6-pdh, Hbd, Me and Sod were monomorphic with common allele) and estimates of genetic variability parameters in populations of Am. vexans (unique alleles allowing U.S.A. delineation are underlined).
The difference between all genotypic classes based on Hardy-Weinberg values was statistically significant for all variable loci in all populations. The observed heterozygosity (Ho) was generally lower than expected (He) heterozygosity (Table 2), and the genotypic fixation index, Fis, indicated excess homozygosity (Fis > 0) in all populations at each variable locus except Idh-2 in the Serbia population. These results were in accordance with Selander's D coefficient, since D was near 0 for Idh-2. Thus, the observed population structure deviated from the one expected in a randomly mating population (Table 1).
Analysis of population genetic structure parameters showed slight differences among the studied populations. The mean number of alleles per locus (A), frequency of polymorphic loci (P0.95), the observed heterozygosity (Ho), and the mean locus θH were highest in the Serbia population, lower in Germany, and the smallest in the Nearctic population (Table 2).
Diagnostic allozyme loci
Genetic interpretation of zymograms of the studied populations revealed the presence of species-specific alleles at Ao, Hk-2, Hk-3, Hk-4, Idh-1, and Idh-2 loci. Using all diagnostic loci (except Ao) probability for correct identification of Nearctic and Palearctic populations was 100%. Cumulative frequency of overlapping allelic frequencies at Ao did not exceed 5%, thus, Ao is considered to be a diagnostic as well. On the basis of unique alleles at diagnostic loci, Nearctic and Palearctic populations were clearly separated (Table 2).
The same major (frequency ≥ 0.5) alleles at Est-1 and Gpi-1 were registered in all three populations, while the same major alleles at Ao and Pgm-2 were shared by Serbia and Germany populations. The rare allele Ao03 (≤0.05) found in the U.S.A. was frequent in two Palearctic populations. Different rare alleles in the studied populations were recorded at Est-6, while the same allele at Est-1 and Gpi-1 were shared by Serbia/Germany and Germany/U.S.A., respectively. A unique (present only in the particular population) but rare allele was detected at Idh-2 in Serbia (Table 2).
A model-based clustering was applied to resolve the population genetic structure. At K = 2, one cluster was included in two Palearctic samples, while the U.S.A. specimens formed the second cluster. Further clustering with K = 3 clustered individuals from the U.S.A. into one genetic group, while Germany and Serbia samples were composed of almost even portions of two other clusters. Average log-likelihoods across five replicate STRUCTURE runs reached a plateau at K = 3 (Figure 1), indicating that three is the most probable number of clusters (based on a declining rate of increase in Pr (X|K) as K increases rather than by the absolute maximum likelihood (Pritchard and Wen 2003, Evanno et al. 2005). Evanno's K coincides with log-likelihood estimates of structuring. Isolation-by-distance (IBD) was significant between Nearctic and Palearctic populations for isolates with recorded geographic locations based on both the Mantel test (r= 0.15247, P < 0.000) and spatial autocorrelation (V= 0.0368, P < 0.000).
Interpopulation genetic differentiation quantified by the FST value (Table 3) between Nearctic and Palearctic populations was significant and resulted in fixed allelic differences at diagnostic loci, while non-significant genetic differentiation (FST= 0.013) between Palearctic Serbia and Germany populations was mainly the result of differences in the allelic frequencies at Idh-2 (FST= 0.048), Ao (FST= 0.028), and Est-1 (FST= 0.001); pairwise FST values significantly differ only at Idh-2 (p = 0.035). Additionally, the number of migrants between more geographically distant populations (Serbia–U.S.A. and Germany–U.S.A.) was lower than the number registered for Serbia – Germany. FST values indicated the almost complete absence of migration (Nm = 0.2) between Palearctic and Nearctic population pairs, and almost 18 individuals per generation (Nm = 17.6) in the Palearctic population pair. Likewise, the greatest degree of genetic divergence according to Nei (1972) was found between Palearctic and Nearctic populations. Consequently, genetic differentiation quantified by the FST values was in association with geographic distance and genetic distance after Nei (1972) (Table 3).
Table 3. Genetic distance calculated after Nei (1972) (above diagonal) and FST value (below diagonal) among populations of Am. vexans (significant value of FST P is shown by ***).
Results from an analysis of molecular variance (AMOVA) considering all samples (non-grouped) suggested that a majority of the genetic variance was attributed to variation among populations (63.5%), and the remaining variation (36.5%) was within populations (Table 4). In addition, we determined genetic divergence between groups indicated by our STRUCTURE and FST pairwise results (Nearctic vs Palearctic). In this case, most (71.54%) of the overall population variation was explained by between-group genetic variation, while only a small portion of the variation (0.13%) was found between populations within the Palearctic group (Table 4).
Table 4. Global AMOVA results weighted over 11 variable loci in the Am. vexans samples.
Groups of samples
Source of variation
FST fixation index within populations; FSC fixation index among populations within groups; FCT fixation index between groups. *** P< 0.001.
Nearctic vs Palearctic
Between populations within groups
By quantifying allozyme diversity, we observed higher variation presented by the mean number of alleles per locus, average frequency of observed heterozygosity, and frequency of polymorphic loci in the Serbian population compared to populations from Germany and the U.S.A. The frequency of polymorphic loci was similar to corresponding values previously obtained in allozyme studies of this species (Milankov et al. 1999) However, an average observed heterozygosity and the mean locus θH in our study was lower in the Nearctic population in comparison with European Serbia and Germany populations, including previously published data of both Palearctic (Milankov et al. 1999) and Nearctic populations. We consider that differences in the amount of genetic variation of studied populations are likely to be influenced by effective population size obtained by θ estimate. Since the θ estimate under the assumption of equal mutation rates in closely related species and conspecific populations (Zuckerkandl and Pauling 1965) is directly related to effective population size (Ne), the smallest value of the mean locus θH observed in the U.S.A. suggests that small Ne characterized this population. In addition, the Nearctic population is located in Palm Desert and is possibly isolated from other populations, which consequently influences reduced migrants and gene flow. Therefore, both small effective population size and presumably reduced gene flow influence the lowest genetic variation in the studied Nearctic population. However, a high level of intra- and interpopulation genetic diversity was registered in Nearctic populations of Am. vexans in ND5 mitochondrial gene (mtDNA) analysis considered as a taxon-specific (“native”) established over a long period of time (Solorzano et al. 2010)
We observed that populations of Am. vexans from the Palearctic (Germany and Serbia) and from the Nearctic (U.S.A.) formed two genetically distinct populations that had no genetic exchange. This conclusion was supported by distinctly different Hardy-Weinberg values, by unique alleles at diagnostic loci (out of 17, six diagnostic allozyme loci were identified), analyses based on individual (STRUCTURE analysis), and population level (pairwise FST,Nm values, AMOVA, Nei's D value). Further evidence of a genetically independent Nearctic population was given by the almost complete absence of exchange genes (Nm= 0.2). In addition, the presence of strong relationships between geographic and genetic distance (IBD) showed that historical effect and reduced gene flow have been important mechanisms responsible for the observed genetic divergence in this study. It is likely that geographic barriers formed by Atlantic and Pacific Oceans prevented gene flow and caused genetic changes in two different evolutionary lineages. Since geographic variation of the available habitat and effects of the evolutionary mechanisms such as gene flow, natural selection, genetic drift, and historical effect caused a differentiation of the genotypes, differences between allelic frequencies and unique alleles in the spatial divergent Nearctic and Palearctic populations might be expected. Indeed, there is very little opportunity for gene flow between Palearctic and Nearctic populations due to specificity of the egg-laying habit. Namely, Am. vexans is a floodwater species which means that females lay their eggs exclusively on moist soil (Becker et al. 2010) and could not be disseminated through international trade and travel. Contrary to Am. vexans, eggs of some invasive species, such as Stegomyia albopicta, were laid in used tires and/or lucky bamboo containers by which they had been easily transported overseas (Manguin and Boëte 2011). Hence, a specific behavioral pattern is likely to be a limiting factor for gene flow between Nearctic and Palearctic populations.
Contrary to the discontinued spatial variation between Palearctic and Nearctic populations, non-significant genetic differentiation was observed between European populations. Since Am. vexans is a floodwater mosquito, there are many connected larval habitats across Europe. A sophisticated Am vexans adaptation to flooding of the breeding sites (in Becker et al. 2010, referred to as “hatching in installments”) provides an opportunity for temporal conservation of the gene pool. Furthermore, the host-seeking pattern of Am. vexans is characterized by range dispersion to habitats suitable for host-finding and/or egg-laying (Becker et al. 2010). For instance, based on non-oriented dispersal behaviors (a drift with the wind and an active dispersal), migration distances of marked females are remarkable; from 22 km (Gjulin and Stage 1950, Briegel et al. 2001) to 48 km (Morhrig 1969). Similarly, it was registered that females can fly from 1–2 km (Petrić et al. 1999) to 10–17 km per night (Briegel et al. 2001). Thus, the persistence of suitable larval habitats across Europe, a highly sophisticated mechanism that regulates the hatching process, and extensive flight activities, provide prospects for gene flow among populations of Am. vexans.
Finally, our hypothesis regarding closer relationships between European populations than those that are geographically distinct (U.S.A. vs Palearctic) was supported. Considering the above, it could be stated that investigated Nearctic and Palearctic populations do not share a common gene pool. Recognition of divergent evolutionary lineages within Am. vexans is also crucial for understanding mosquito-borne disease dynamics since it has been shown that each entity within a complex of cryptic mosquito taxa is characterized by its own biology, ecology, and vector capacity (Alquezar et al. 2010). We consider the present study as a base for future taxonomic research that should reveal the degree of genetic differentiation needed for the separation of geographically isolated mosquito populations at the specific/subspecific level.
The authors thank Branka Božičić Lothrop and Achim Kaiser for providing mosquito material. We also thank two anonymous referees who provided useful comments on the manuscript. This work was supported in part by the Ministry of Science of Serbia (Dynamics of gene pool, genetic and phenotypic variability of populations, determined by the environmental changes, No. 173012), and the Provincial Secretariat for Science and Technological Development (Molecular and phenotypic diversity of taxa of economical and epidemiological importance, and endangered and endemic species in Europe). B. K. is supported by Ciklonizacija d.o.o. Novi Sad.