Population Structure of the Invasive Asian Tiger Mosquito, Aedes albopictus, in Europe

Abstract

The Asian tiger mosquito, Aedes albopictus, is currently the most widespread invasive mosquito species in the world. It poses a significant threat to human health, as it is a vector for several arboviruses. We used a SNP chip to genotype 748 Ae. albopictus mosquitoes from 41 localities across Europe, 28 localities in the native range in Asia, and 4 in the Americas. Using multiple algorithms, we examined population genetic structure and differentiation within Europe and across our global dataset to gain insight into the origin of the invasive European populations. We also compared results from our SNP data to those obtained using genotypes from 11 microsatellite loci (N = 637 mosquitoes from 25 European localities) to explore how sampling effort and the type of genetic marker used may influence conclusions about Ae. albopictus population structure. While some analyses detected more than 20 clusters worldwide, we found mosquitoes could be grouped into 7 distinct genetic clusters, with most European populations originating in East Asia (Japan or China). Interestingly, some populations in Eastern Europe did not share genetic ancestry with any populations from the native range or Americas, indicating that these populations originated from areas not sampled in this study. The SNP and microsatellite datasets found similar patterns of genetic differentiation in Europe, but the microsatellite dataset could not detect the more subtle genetic structure revealed using SNPs. Overall, data from the SNP chip offered a higher resolution for detecting the genetic structure and the potential origins of invasions.

Keywords: Aedes albopictus; SNP chip; disease vector; invasive species; microsatellites; population genomics; population structure; tiger mosquito.

FIGURE 1

Maps of sampling locations for all genotyped Aedes albopictus mosquitoes. Panel A shows the 74 locations, by region, for all samples genotyped in the Global SNP Chip dataset. Panel B shows a zoomed in view of the box outlined in gray in Panel A, to highlight the locations for the 41 European sampling sites in the SNP dataset (orange) and 25 in the microsatellite dataset. Locations with samples in both the SNP and microsatellite data sets are indicated in blue. Each locality is identified by a unique three‐letter abbreviation identifying the location. Details on the location, year of collection, and the number of samples per locality for both sets of samples are reported in Tables A2 and A3 in Appendix S1.

FIGURE 2

The population structure shows the best‐supported number of clusters (K) identified for 688 mosquitos sampled from American, European, and native locations (K = 20) (Panel A). Panel B contains the same data shown in A, but plotted for K = 7. In A and B each vertical bar on the x‐axis represents one mosquito, and the y‐axis shows the proportion of admixture for each individual ancestral genetic group. Panel C shows scatterplots of principal component analysis (PCA) showing the first PC on the x‐axis and PCs 2 and 3 on the y‐axes. Each symbol represents a mosquito, and the color and shape of the symbol indicates the country where they were sampled. Ellipses mark each region covering 80% of the samples. All plots were created in the R package LEA using data from SNP Set 3. Plots for the best K obtained from other structure algorithms (fastStructure and admixture) and using alternate SNP sets are shown in Figures A6–A9, Table A7 in Appendix S1, and Files S10 and S17.

FIGURE 3

Panel A shows population structure for the best‐supported number of clusters (K) identified for 410 mosquitos sampled from 41 European locations (K = 14). Each vertical bar on the x‐axis represents one mosquito, and the y‐axis shows the proportion of admixture for each individual ancestral genetic group. Panel B shows scatterplots of principal component analysis (PCA) showing the first PC on the x‐axis and PCs 2 and 3 on the y‐axes. Each symbol represents a mosquito, and the color and shape of the symbol indicates the country where they were sampled. Ellipses mark each region covering 80% of the samples. All plots were created using LEA with data from SNP Set 3. Plots for the best K obtained from other structure algorithms (fastStructure and admixture) and using alternate SNP sets are shown in Figures A10 and A11, Table A7 in Appendix S1, and Files S9 and S16.

FIGURE 4

Panel A Plots of the Mantel test for isolation‐by‐distance (IBD) using the 40 European populations with at least four mosquitoes (Table A3 in Appendix S1) for SNP Set 3 (20,968 SNPs). The histogram shows the distribution of correlation coefficients between genetic (Dgen) and geographic (Dgeo) distances obtained from 999 random permutations. The arrow indicates the observed correlation coefficient (−0.011) derived from the data, which suggests no significant deviation from random expectations (p < 0.519). The scatterplot shows geographic distance (Dgeo) against genetic distance (Dgen). The line from the linear regression model fit to the data had an R ² = 0.01. The density of overlapping points is represented by color, with warmer shades indicating more overlap. Panel B shows scatterplots of geographic distance (Dgeo) against genetic distance (Dgen) for subsets of European samples by region: (i) Italy (R ² = 0.0), (ii) Greece, Albania and Croatia (R ² = −0.25), (iii) Eastern Europe (R ² = 0.02), and (iv) Iberian Peninsula (R ² = 0.0). Additional plots for subsets are shown in Figures A18 and A19 in Appendix S1.

FIGURE 5

Maps of the ancestry matrices obtained using the 24 locations shared between the microsatellite and SNP datasets. The top panel shows results when the clustering analysis was run for 637 mosquitos in the microsatellite dataset using only genetic data for inferencing in STRUCTURE (best supported number of clusters: K = 3). The middle panel show the clusters obtained using LEA to analyze structure for the same 24 locations with SNP Set 3 dataset. For the SNP data, K = 3 is shown to facilitate comparison to microsatellites. K = 10, which was the best supported number of clusters for SNPs is shown in the bottom panel. For more detailed maps showing sampling locations in the microsatellite dataset see File S18.