Feeding increases variability among individuals
We first tested whether there was significant variability in the number of observed microbial taxa, evenness (Pielou index) and Shannon diversity among sample groups. Generally, male and female adult mosquitoes varied similarly (P> 0.05) (Fig 1A). However, when feeding statuses were considered, the number of bacteria taxa observed was higher among blood-fed than sugar-fed and unfed groups (P<0.01) (Fig 1B). Despite this, the overall within-group diversity did not differ between feeding groups (Fig 1B). We further assessed alpha diversity among sexes based on their feeding history. Again, blood-fed females showed higher individual-to-individual variability in taxa richness and Shannon diversity than their sugar-fed counterparts (P<0.001) (Fig 1C). While sugar-fed and unfed females showed a difference in richness and within-group diversity, the males were largely the same whether fed or unfed (Fig 1C). All test groups had equal evenness.
Fig 1: Violin plots for alpha diversity indices observed. Density curves show the distribution of the data among samples of a group. Overlaid box plots summarize the data to show the upper and lower values, and short horizontal line within each box depicts the mean. Comparison of the index values are between males and females (A), feeding statuses (B) and sex*feeding groups (C).
Blood-feeding accounts for divergence in the microbiota between male and female mosquitoes
Beta diversity analyses based on weighted UniFrac distances were performed to assess the phylogenetic differences between feeding groups. Overall, male and female adult mosquitoes considering their feeding statuses were 34% diverse in their microbial composition (PERMANOVA: R2=0.340, F= 4.384, P=0.001) (Fig 2). We investigated the source of this diversity with pairwise comparisons between test groups and, observed that the divergence was as low as 5% between sugar-fed females and males (FS vs MS) (P-adj= 0.54) and as high as 39% between sugar-fed and unfed females (FS vs FU: P-adj= 0.01) (Fig 2). The difference in diversity between unfed and sugar-fed was significant in both sexes, but higher among females than males (FS vs FU= 39%, MS vs MU= 29%). Blood meals reduced the microbial distance between sugar fed and unfed mosquitoes (FB vs FU) to 25% (P-adj= 0.03). Blood fed females shared phylogenetically similar bacteria with both sugar-fed males and females (Fig 2).
Fig 2: Non-metric multidimensional scaling ordination of samples based on their sex*feeding status. Distances were estimated based on weighted UniFrac and compared with using PERMANOVA. P-values are shown in bold where the pairwise comparison is significant with P-adjusted value <0.05.
Elizabethkingia meningoseptica is a major discriminant species for fed mosquitoes
Five phyla accounted for majority (~99%) of the bacteria in our samples (Fig 3A). Pseudomonadota (formerly Proteobacteria) alone averaged 47% while Bacteroidota (formerly Bacteriodetes) and Actinomycetota (formerly Actinobacteria) both comprised 23% and 21%. The relative abundance of Bacteroidota and Actinomycetota differed between feeding statuses, the former increasing significantly among fed mosquitoes while latter was increased in newly unfed ones (Fig 3B). Out of 241 genera identified, only 10 including Asaia, Elizabethkingia, were above 1% average relative abundance (Fig 3C). Elizabethkingia, Gluconobacter and Chryseobacterium were the only high abundant genera with significantly increased abundance in the fed compared to unfed mosquitoes (Fig 3D).
Fig 3: Bacteria phyla and genera abundance. The distribution of taxa with average relative abundance >1% among test groups are shown as bar plots for phyla (A) and heatmap for genera (C). Relative abundances are compared between feeding statuses with LEfse (B and D). Significances shown are adjusted P-values.
Based on previous observations that the sex of the mosquito confers different degrees of divergence in the microbiome after feeding (Fig 2), LefSe (38) analyses were repeated between sex*feeding group pairs to detect bacteria species that drive these differences. To make our analyses more stringent, the discovery significant taxa were set at an LDA threshold of 4. In all the pairs that showed diverse microbiome (Fig 2), Elizabethkingia meningoseptica was a significant discriminant bacterium (LDA scores ~5) that increased in both sugar fed and blood fed mosquitoes (Fig 4). It was the only bacterium that differentiated unfed from sugar fed males, and males from females. Whether male or female, sugar feeding also significantly increased Asaia simensis and was the only species that distinguished the microbiome of sugar fed from unfed mosquitoes (Fig 4). Brevibacterium casei showed as a major bacterium in all unfed mosquitoes.
Fig 4: Differential abundance of bacteria species between sex*feeding groups. Plots show only significant species with discriminant scores above 4 (LDA >4).