Vital Surveillances: Variations in the Bacterial Ecosystems of Mosquito Populations — Haikou and Sanya Cities, Hainan Province, China, 2019

Adult female mosquitoes were collected over an eight-day period (July 12–19, 2019) in Haikou and Sanya cities, Hainan Province, China. Collection sites were selected in tropical and subtropical areas using CO₂-augmented light traps and human landing catches to ensure diverse sampling (10). The collection sites and their geographic coordinates are presented in Table 1.

Mosquitoes were identified morphologically, and their identities were confirmed via DNA barcoding using the cytochrome c oxidase subunit 1 (cox1) gene (11). All specimens were stored at −80 °C. To minimize the influence of host blood on microbiota analysis, blood-fed mosquitoes were maintained in the laboratory for 4 days to allow for blood digestion before midgut dissection.

Following molecular identification, mosquitoes were sorted by species, and ten midguts per species were dissected in triplicate to ensure data reliability. The samples were initially stored at −80 °C, surface-sterilized in 75% ethanol for 30 seconds, and this process was repeated twice. After rinsing twice in sterile PBS to remove residual ethanol, the midguts were pooled into sterile 1.5 mL tubes and stored at –20 °C until DNA extraction (12). Dissections were performed under a stereomicroscope to ensure sample integrity, with the foregut, hindgut, and Malpighian tubules carefully removed. Genomic DNA was extracted using the DNeasy Blood & Tissue Kit (Qiagen, USA) (13) according to the manufacturer’s instructions. The purified DNA was diluted in 50 μL ddH₂O for microbiome library construction at BGI Shenzhen.

A total of 281 midguts were collected from various mosquito species in Haikou and Sanya, using primers 338F (5'-ACTCCTACGGGAGGCAGCA-3') and 806R (5'-GGACTACHVGGGTWTCTAAT-3') to amplify the V3-V4 hypervariable regions of the bacterial 16S rRNA gene (14). The genomic DNA was sequenced using the Illumina HiSeq 4000 system. Following sequencing, we implemented quality control measures, including the removal of chimeric sequences, and identified Operational Taxonomic Units (OTUs) using the RDP Classifier (15). From 1,508,554 raw reads, 1,384,001 clean reads were generated (Supplementary Table S1) and clustered into OTUs at 97% similarity. Only OTUs with a relative abundance above 0.5% were analyzed to focus on the predominant bacterial communities (Supplementary Table S2).

Statistical evaluations were conducted using R (version 3.1.1; R Foundation, Austria) and Metastats (version 1.0; Whitehead Institute, USA) to assess differentiation among microbial communities (16). Rarefaction curves and box plots were generated to visualize biodiversity and analyze significant differences in microbial diversity.

Microbial diversity within and across mosquito samples were quantified using alpha and beta diversity indices. Alpha diversity, which measures diversity within a sample, included indices such as Simpson, Shannon, ACE, and Chao1, calculated using Mothur (version 1.31.2; Michigan State University, USA) at a 97% similarity threshold for OTUs. Beta diversity, which compares differences between samples, was evaluated using Bray-Curtis, weighted UniFrac, and unweighted UniFrac indices. Significant differences in alpha diversity were observed between mosquito populations in Haikou and Sanya, as visualized by matrix heatmaps of diversity metrics, highlighting the presence, abundance, and phylogenetic relationships of microbial communities.

Venn diagrams were used to illustrate shared and unique OTUs, effectively visualizing microbial overlap and distinctiveness across samples. For phylogenetic analysis, sequences were aligned using QIIME’s align_seqs.py, and phylogenetic trees were constructed using the Fasttree method (17). These trees, representing evolutionary relationships among bacterial species, were visualized using R software, providing comprehensive insights into the microbial community structure across different environmental contexts.

Cluster analysis was performed using QIIME (version 1.8.0; Knight Lab, University of Colorado, USA) with an iterative algorithm that selected 75% of sequence data from the least abundant samples. The Unweighted Pair Group Method with Arithmetic Mean (UPGMA) was employed for hierarchical clustering to reveal relationships among microbial communities. The resulting clustering trees were visualized using R software, effectively illustrating the structural organization of microbial communities across the studied samples (18).

This study analyzed the V3–V4 regions of the 16S rRNA gene in 281 midguts from field-collected female mosquitoes (Aedes albopictus, Culex pipiens, Culex gelidus, and Armigeres subalbatus) from Haikou and Sanya. The majority of sequences belonged to the phylum Proteobacteria (85.28%), primarily Alphaproteobacteria (52.14%), Gammaproteobacteria (29.90%), and Betaproteobacteria (3.22%). Other significant phyla included Firmicutes (6.24%), Actinobacteria (3.81%), and smaller contributions from Thermi, Cyanobacteria, Bacteroidetes, Spirochaetes, TM7, and Chloroflexi. At the family level, Rickettsiaceae dominated (46.96%), followed by Enterobacteriaceae and Aeromonadaceae. The most abundant genera were Wolbachia (46.96%), Acinetobacter, and Escherichia (Figure 1A–B). Rickettsiaceae prevalence varied by location, being higher in Sanya (87.36% in Ae. albopictus and Ar. subalbatus) compared to Haikou (20.02%), with a much lower abundance in Ae. albopictus from XH (0.32%). This demonstrates significant variations in the gut microbiomes across mosquito species and different locations.

Figure 1. 

Mean relative abundances of (A) bacterial families and (B) genera associated with four mosquito species from Haikou and Sanya.

Note: Families and genera with abundance less than 0.5% were pooled together as “Other.”

Abbreviation: CW=Changwang, HK; SW=Shiwaitaoyuan, HK; XH=Xinghui, HK; XY=Xiyi, HK; CC=Chicao, SY; HK=Haikou; NB=Nanbin, SY; NY=Nanya, SY; SY=Sanya; AAL=Aedes albopictus, ASU=Armigeres subalbatus, CGE=Culex gelidus, CPI=Culex pipiens.

OTUs were defined at 97% similarity, and shared an unique taxa were visualized with Venn diagrams to identify core microorganisms across environments. Comparative analysis revealed that mosquito species within the same genus in Haikou and Sanya harbored similar bacteria, with 204 overlapping species, 473 unique species in Haikou, and 64 unique species in Sanya (Figure 2A). Phylum-level analysis showed distinct bacterial communities between the two locations. Haikou was dominated by Firmicutes (32.77%), followed by Proteobacteria (23.26%), and Bacteroidetes (16.28%), while Sanya had higher proportions of Proteobacteria (32.81%) and Firmicutes (26.56%). Haikou also contained unique phyla, including Thermi and Cyanobacteria, while Sanya exclusively harbored Chlamydiae. Statistical tests confirmed significant differences between locations, such as for Actinobacteria (Wilcoxon test: P=0.038).

Figure 2. 

Venn diagrams and heatmap illustrating bacterial composition overlap and gut microbiota of mosquito species across habitats. (A) Venn diagram showing overlaps of OTUs at 97% similarity between the HK and SY data sets; (B) Number of bacterial taxa specific and common to Aedes albopictus of XH, CW in Haikou; (C) Number of bacterial taxa specific and common to Ae. albopictus of CC, NB, SY (street) in Sanya; (D) Number of bacterial taxa specific and common to Aedes albopictus of XH, NB, and SY; (E) Heatmap in log scale depicting the gut bacterial community of mosquito midguts obtained with open reference OTU picking methods at the genus level.

Note: For (E), Green colors represent high abundance, and red colors represent low abundance; black indicates absence.

Abbreviation: OTUs=Operational Taxonomic Units; CW=Changwang, HK; SW=Shiwaitaoyuan, HK; XH=Xinghui, HK; XY=Xiyi, HK; CC=Chicao, SY; HK=Haikou; NB=Nanbin, SY; NY=Nanya, SY; SY=Sanya.

Further analysis of Ae. albopictus in Haikou revealed that XH and CW shared 118 species, with 76 and 39 unique species, respectively, indicating greater diversity in XH (Figure 2B). In contrast, three Sanya locations (SY, CC, NB) shared only 34 species, demonstrating lower diversity (Figure 2C). Between Haikou’s XH and Sanya’s SY and NB, 42 overlapping species were found (Figure 2D), with fewer unique species in Nanban farm and Sanya Street (30 and 36, respectively) compared to Haikou’s Xinghui Village (76 unique species). Across all sites, 32 species were consistently identified, representing a core microbial community. Key genera included Corynebacterium, Wolbachia and Cupriavidus.

The phylogenetic tree and heatmap visually represent the composition and abundance of bacterial communities in mosquito midguts, highlighting structural similarities and differences. Clustering analyses revealed three major groups: 1) Ae. albopictus and Cx. pipiens from XH Village in Haikou; 2) Cx. gelidus from XY Village, along with Ae. albopictus from NY Farm, CC Village, and Sanya Street; and 3) mosquitoes from CW, SW, and XY Villages in Haikou. Three primary genera were identified: Wolbachia, Escherichia, and Thermus, which were widespread across locations and species. Wolbachia was abundant in all mosquitoes except for those from XH and XY Villages. Acinetobacter and Comamonas were found predominantly in Ae. albopictus and Cx. pipiens from XH Village and Ae. albopictus and Ar. subalbatus from CW and SW Villages. Novispirillum and Thorsellia were more abundant in Cx. pipiens from XY Village and Ar. subalbatus from SW Village (Figure 2E).

Alpha diversity indices (observed species, Chao1, ACE, Simpson, and Shannon) were used to assess bacterial diversity and richness in mosquitoes from Haikou and Sanya, revealing differences between locations. Alpha diversity showed significantly higher bacterial richness in Cx. pipiens from Haikou compared to Ae. albopictus from Sanya, with Cx. pipiens and Ar. subalbatus in Haikou exhibiting notably higher diversity (Chao1: 491.54, ACE: 487.92; Chao1: 411.25, ACE: 405.33) versus Sanya (Chao1: 220.87, ACE: 261.89) (Figure 3A).

Figure 3. 

Boxplot and heatmap representations of bacterial diversity across mosquito species and habitats. (A) Boxplot representation of observed species, Chao1, and Shannon diversity indices. (B) Matrix heatmap of Bray-Curtis distances (left), Weighted UniFrac Beta diversity (middle), and Unweighted UniFrac Beta diversity (right) between microbial communities of four mosquito species from Haikou and Sanya.

Note: For (A), boxplots show the distribution of bacteria between mosquito samples categorized by location and species. Significant differences between the groups were investigated using pairwise comparisons of means (Dunn test; P value adjustment: Holm). Species richness is represented by the number of bands. Boxplots show median, minimum, and maximum values, with black lines indicating medians.

Beta diversity analysis revealed significant microbial differences between Haikou and Sanya populations, with Bray-Curtis, weighted UniFrac, and unweighted UniFrac metrics demonstrating distinct clustering patterns (Figure 3B). Bray-Curtis indicated dissimilarity between Ar. subalbatus (SY-ASU) from Sanya and Cx. gelidus (HK-CGE) and Ae. albopictus (HK-AAL) from Haikou, while Ae. albopictus from both regions overlapped. The unweighted UniFrac distance metric highlighted the microbial divergence between Cx. gelidus (HK-CGE) and Ar. subalbatus (SY-ASU), while Ae. albopictus populations showed greater similarity. The weighted UniFrac metric revealed microbial overlap between Ae. albopictus (SY-AAL) and Ar. subalbatus (SY-ASU) in Sanya, but distinctions from Cx. pipiens (HK-CPI) and Cx. gelidus (HK-CGE) in Haikou. These patterns indicate that microbial communities in Sanya were more similar within the region, while those in Haikou exhibited more variation. Overall, geographic and species-specific factors strongly influence microbial community composition.

This study investigated bacterial communities in the midguts of four mosquito species from Haikou and Sanya using high-throughput sequencing of the 16S rRNA gene’s V3-V4 regions. Proteobacteria, particularly Wolbachia, dominated the microbial composition, with significantly higher prevalence in Sanya than in Haikou, suggesting regional differences in microbial composition that could influence disease transmission dynamics (19).

Blood meals substantially alter gut microbiota, increasing bacteria such as Serratia and Elizabethkingia that participate in blood digestion, while reducing symbiotic bacteria like Wolbachia, which may affect mosquito immunity and vector competence (20). To minimize blood-feeding effects, mosquitoes were allowed to digest blood for 4 days before dissection, though residual effects on bacterial diversity cannot be ruled out. Regional and species-specific variations, such as the higher prevalence of Wolbachia in Sanya mosquitoes, may also be influenced by dietary and environmental factors. Future studies should include controlled feeding experiments to distinguish intrinsic microbiota from transient changes due to blood meals.

Venn diagrams revealed both shared and unique microbial communities between locations. Mosquitoes from Haikou displayed richer bacterial diversity, including Thermi and Cyanobacteria, indicating regional variations. Alpha diversity was higher Cx. pipiens from Haikou than in Ae. albopictus from Sanya, highlighting the importance of understanding local mosquito ecology for disease control. Phylogenetic and heatmap analyses showed distinct microbial clusters based on geographic location and mosquito species, reflecting the complex interplay between mosquitoes and their microbiomes and suggesting the influence of environmental factors. However, the analysis was limited to four mosquito species and two regions, and 16S rRNA sequencing offers limited taxonomic resolution.

In conclusion, significant variations in microbial ecosystems were found between mosquito populations from Haikou and Sanya, with Wolbachia’s higher prevalence in Sanya mosquitoes, indicating potential for biologically-based control strategies to enhance public health in tropical regions (21).